From f106e780ba91ad8c2949596f506a7f762e3724c5 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 6 Aug 2024 19:41:12 -0700 Subject: [PATCH 0001/1701] downloader: use 1 session for all files for better speed --- download-model.py | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/download-model.py b/download-model.py index 0014b689c9..306784a355 100644 --- a/download-model.py +++ b/download-model.py @@ -29,6 +29,7 @@ class ModelDownloader: def __init__(self, max_retries=5): self.max_retries = max_retries + self.session = self.get_session() def get_session(self): session = requests.Session() @@ -72,7 +73,7 @@ def sanitize_model_and_branch_names(self, model, branch): return model, branch def get_download_links_from_huggingface(self, model, branch, text_only=False, specific_file=None): - session = self.get_session() + session = self.session page = f"/api/models/{model}/tree/{branch}" cursor = b"" @@ -192,7 +193,7 @@ def get_single_file(self, url, output_folder, start_from_scratch=False): attempt = 0 while attempt < max_retries: attempt += 1 - session = self.get_session() + session = self.session headers = {} mode = 'wb' From e926c03b3d028ebad854e753517894fa6c637c3e Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 6 Aug 2024 19:41:18 -0700 Subject: [PATCH 0002/1701] Add a --tokenizer-dir command-line flag for llamacpp_HF --- modules/models.py | 33 ++++++++++++++++++++++----------- modules/shared.py | 1 + 2 files changed, 23 insertions(+), 11 deletions(-) diff --git a/modules/models.py b/modules/models.py index ecef9060b9..b0e2346e18 100644 --- a/modules/models.py +++ b/modules/models.py @@ -98,7 +98,7 @@ def load_model(model_name, loader=None): if model is None: return None, None else: - tokenizer = load_tokenizer(model_name, model) + tokenizer = load_tokenizer(model_name) shared.settings.update({k: v for k, v in metadata.items() if k in shared.settings}) if loader.lower().startswith('exllama') or loader.lower().startswith('tensorrt'): @@ -113,9 +113,13 @@ def load_model(model_name, loader=None): return model, tokenizer -def load_tokenizer(model_name, model): +def load_tokenizer(model_name, tokenizer_dir=None): + if tokenizer_dir: + path_to_model = Path(tokenizer_dir) + else: + path_to_model = Path(f"{shared.args.model_dir}/{model_name}/") + tokenizer = None - path_to_model = Path(f"{shared.args.model_dir}/{model_name}/") if path_to_model.exists(): if shared.args.no_use_fast: logger.info('Loading the tokenizer with use_fast=False.') @@ -278,17 +282,24 @@ def llamacpp_loader(model_name): def llamacpp_HF_loader(model_name): from modules.llamacpp_hf import LlamacppHF - path = Path(f'{shared.args.model_dir}/{model_name}') - - # Check if a HF tokenizer is available for the model - if all((path / file).exists() for file in ['tokenizer_config.json']): - logger.info(f'Using tokenizer from: \"{path}\"') + if shared.args.tokenizer_dir: + logger.info(f'Using tokenizer from: \"{shared.args.tokenizer_dir}\"') else: - logger.error("Could not load the model because a tokenizer in Transformers format was not found.") - return None, None + path = Path(f'{shared.args.model_dir}/{model_name}') + # Check if a HF tokenizer is available for the model + if all((path / file).exists() for file in ['tokenizer_config.json']): + logger.info(f'Using tokenizer from: \"{path}\"') + else: + logger.error("Could not load the model because a tokenizer in Transformers format was not found.") + return None, None model = LlamacppHF.from_pretrained(model_name) - return model + + if shared.args.tokenizer_dir: + tokenizer = load_tokenizer(model_name, tokenizer_dir=shared.args.tokenizer_dir) + return model, tokenizer + else: + return model def AutoGPTQ_loader(model_name): diff --git a/modules/shared.py b/modules/shared.py index c27657ff6a..43533a1480 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -132,6 +132,7 @@ group.add_argument('--row_split', action='store_true', help='Split the model by rows across GPUs. This may improve multi-gpu performance.') group.add_argument('--streaming-llm', action='store_true', help='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.') group.add_argument('--attention-sink-size', type=int, default=5, help='StreamingLLM: number of sink tokens. Only used if the trimmed prompt does not share a prefix with the old prompt.') +group.add_argument('--tokenizer-dir', type=str, help='Load the tokenizer from this folder. Meant to be used with llamacpp_HF through the command-line.') # ExLlamaV2 group = parser.add_argument_group('ExLlamaV2') From 81773f7f3681c3f67e2b79aa2c0f91cb15cdac25 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 6 Aug 2024 20:07:05 -0700 Subject: [PATCH 0003/1701] Bump transformers to 4.44 --- requirements.txt | 2 +- requirements_amd.txt | 2 +- requirements_amd_noavx2.txt | 2 +- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_cpu_only.txt | 2 +- requirements_cpu_only_noavx2.txt | 2 +- requirements_noavx2.txt | 2 +- requirements_nowheels.txt | 2 +- 9 files changed, 9 insertions(+), 9 deletions(-) diff --git a/requirements.txt b/requirements.txt index b1c6891735..3f702ecdc4 100644 --- a/requirements.txt +++ b/requirements.txt @@ -24,7 +24,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.43.* +transformers==4.44.* tqdm wandb diff --git a/requirements_amd.txt b/requirements_amd.txt index 9cef52be16..c7611af816 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.43.* +transformers==4.44.* tqdm wandb diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index e6df644c2e..c0dda7e8e0 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.43.* +transformers==4.44.* tqdm wandb diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 35131b95eb..3153376d4a 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.43.* +transformers==4.44.* tqdm wandb diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index ee9876eef0..22b676c452 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.43.* +transformers==4.44.* tqdm wandb diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 87b1a95cd8..95169f3356 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.43.* +transformers==4.44.* tqdm wandb diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index 91c3035447..e854cca9dd 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.43.* +transformers==4.44.* tqdm wandb diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 1adcec6f4a..dde919879b 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -24,7 +24,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.43.* +transformers==4.44.* tqdm wandb diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index bc8a59aae1..a95098a5c5 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.43.* +transformers==4.44.* tqdm wandb From 089d5a94152ea7802a480bf5c227579d1be8c899 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 7 Aug 2024 20:36:28 -0700 Subject: [PATCH 0004/1701] Bump llama-cpp-python to 0.2.87 --- requirements.txt | 24 ++++++++++++------------ requirements_amd.txt | 12 ++++++------ requirements_amd_noavx2.txt | 8 ++++---- requirements_apple_intel.txt | 8 ++++---- requirements_apple_silicon.txt | 12 ++++++------ requirements_cpu_only.txt | 8 ++++---- requirements_cpu_only_noavx2.txt | 8 ++++---- requirements_noavx2.txt | 24 ++++++++++++------------ 8 files changed, 52 insertions(+), 52 deletions(-) diff --git a/requirements.txt b/requirements.txt index 3f702ecdc4..2cdb94fb50 100644 --- a/requirements.txt +++ b/requirements.txt @@ -35,22 +35,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.85+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.85+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.85+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.85+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.85+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.85+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.85+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.85+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index c7611af816..042e8a2eeb 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -32,14 +32,14 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.85+rocm5.6.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.85+rocm5.6.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.87+rocm5.6.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.87+rocm5.6.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index c0dda7e8e0..0448e54ab5 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -32,10 +32,10 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 3153376d4a..1dff0bb793 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -32,8 +32,8 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.85-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.85-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.85-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.85-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8-py3-none-any.whl diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 22b676c452..694f8a724c 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -32,10 +32,10 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.85-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.85-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.85-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.85-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.85-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.85-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8-py3-none-any.whl diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 95169f3356..00071cd1e4 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -32,7 +32,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index e854cca9dd..ecb82e90bf 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -32,7 +32,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index dde919879b..b1ba1ecae1 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -35,22 +35,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.85+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.85+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.85+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.85+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.85+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.85+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.85+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.85+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.85+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" From bf8187124dddf99443777aeba86cf682c74f6a07 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 13 Aug 2024 12:40:18 -0700 Subject: [PATCH 0005/1701] Bump llama-cpp-python to 0.2.88 --- requirements.txt | 24 ++++++++++++------------ requirements_amd.txt | 12 ++++++------ requirements_amd_noavx2.txt | 8 ++++---- requirements_apple_intel.txt | 8 ++++---- requirements_apple_silicon.txt | 12 ++++++------ requirements_cpu_only.txt | 8 ++++---- requirements_cpu_only_noavx2.txt | 8 ++++---- requirements_noavx2.txt | 24 ++++++++++++------------ 8 files changed, 52 insertions(+), 52 deletions(-) diff --git a/requirements.txt b/requirements.txt index 2cdb94fb50..45bdf25e1d 100644 --- a/requirements.txt +++ b/requirements.txt @@ -35,22 +35,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index 042e8a2eeb..2aa39dc64c 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -32,14 +32,14 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.87+rocm5.6.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.87+rocm5.6.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.88+rocm5.6.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.88+rocm5.6.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 0448e54ab5..bd915071c2 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -32,10 +32,10 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 1dff0bb793..066a5aec4d 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -32,8 +32,8 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8-py3-none-any.whl diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 694f8a724c..f025e8b73f 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -32,10 +32,10 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.87-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8-py3-none-any.whl diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 00071cd1e4..df5a7d608a 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -32,7 +32,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index ecb82e90bf..fd9548ce38 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -32,7 +32,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index b1ba1ecae1..539ace51dc 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -35,22 +35,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.87+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.87+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.87+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" From 4d8c1801c2313af70944daea503672ce6d3261dd Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 19 Aug 2024 17:45:01 -0700 Subject: [PATCH 0006/1701] Bump llama-cpp-python to 0.2.89 --- requirements.txt | 24 ++++++++++++------------ requirements_amd.txt | 12 ++++++------ requirements_amd_noavx2.txt | 8 ++++---- requirements_apple_intel.txt | 8 ++++---- requirements_apple_silicon.txt | 12 ++++++------ requirements_cpu_only.txt | 8 ++++---- requirements_cpu_only_noavx2.txt | 8 ++++---- requirements_noavx2.txt | 24 ++++++++++++------------ 8 files changed, 52 insertions(+), 52 deletions(-) diff --git a/requirements.txt b/requirements.txt index 45bdf25e1d..a4b864b139 100644 --- a/requirements.txt +++ b/requirements.txt @@ -35,22 +35,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index 2aa39dc64c..db765afe02 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -32,14 +32,14 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.88+rocm5.6.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.88+rocm5.6.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.89+rocm5.6.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.89+rocm5.6.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index bd915071c2..3751af0e01 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -32,10 +32,10 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 066a5aec4d..c38fad1cb1 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -32,8 +32,8 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8-py3-none-any.whl diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index f025e8b73f..0f2f965683 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -32,10 +32,10 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.88-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8-py3-none-any.whl diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index df5a7d608a..9a46e3665d 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -32,7 +32,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index fd9548ce38..e78ce3e04d 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -32,7 +32,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 539ace51dc..5cb0d19aa4 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -35,22 +35,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.88+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.88+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.88+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" From bb987ffe6611a13396848b927b34bbd9d97331c7 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 19 Aug 2024 23:06:52 -0300 Subject: [PATCH 0007/1701] Update README.md --- README.md | 25 ++++++++++++++----------- 1 file changed, 14 insertions(+), 11 deletions(-) diff --git a/README.md b/README.md index 40ae94d538..017fafcb5c 100644 --- a/README.md +++ b/README.md @@ -10,27 +10,30 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github. ## Features -* 3 interface modes: default (two columns), notebook, and chat. -* Multiple model backends: [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp) (through [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)), [ExLlamaV2](https://github.com/turboderp/exllamav2), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). -* Dropdown menu for quickly switching between different models. -* Large number of extensions (built-in and user-contributed), including Coqui TTS for realistic voice outputs, Whisper STT for voice inputs, translation, [multimodal pipelines](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/multimodal), vector databases, Stable Diffusion integration, and a lot more. See [the wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [the extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details. -* [Chat with custom characters](https://github.com/oobabooga/text-generation-webui/wiki/03-%E2%80%90-Parameters-Tab#character). -* Precise chat templates for instruction-following models, including Llama-2-chat, Alpaca, Vicuna, Mistral. -* LoRA: train new LoRAs with your own data, load/unload LoRAs on the fly for generation. -* Transformers library integration: load models in 4-bit or 8-bit precision through bitsandbytes, use llama.cpp with transformers samplers (`llamacpp_HF` loader), CPU inference in 32-bit precision using PyTorch. +* Multiple backends for text generation in a single UI and API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp) (through [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)), [ExLlamaV2](https://github.com/turboderp/exllamav2), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). * OpenAI-compatible API server with Chat and Completions endpoints -- see the [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples). +* Automatic prompt formatting for each model using the Jinja2 template in its metadata, ensuring high-quality outputs without manual intervention. +* Three chat modes: `instruct`, `chat-instruct`, and `chat`, allowing for both task-based interactions and casual conversations with characters. +* High-quality outputs in chat scenarios with the `chat-instruct` mode, which uses the model's template automatically. +* Easy switching between conversations and starting new ones through the "Past chats" menu in the main interface tab. +* Flexible text generation through autocompletion in the Default/Notebook tabs without being limited to chat turns. Send formatted chat conversations from the Chat tab to these tabs. +* Multiple sampling parameters and options for sophisticated text generation control. +* Quick downloading and loading of new models through the interface without restarting, using the "Model" tab. +* Simple LoRA fine-tuning tool to customize models with your data. +* Self-contained dependencies in the `installer_files` folder, avoiding interference with the system's Python environment. Precompiled Python wheels for the backends are in the `requirements.txt` and are transparently compiled using GitHub Actions. +* Extensions support, including numerous built-in and user-contributed extensions. See [the wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [the extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details. ## How to install 1) Clone or [download](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip) the repository. 2) Run the `start_linux.sh`, `start_windows.bat`, `start_macos.sh`, or `start_wsl.bat` script depending on your OS. 3) Select your GPU vendor when asked. -4) Once the installation ends, browse to `http://localhost:7860/?__theme=dark`. +4) Once the installation ends, browse to `http://localhost:7860`. 5) Have fun! -To restart the web UI in the future, just run the `start_` script again. This script creates an `installer_files` folder where it sets up the project's requirements. In case you need to reinstall the requirements, you can simply delete that folder and start the web UI again. +To restart the web UI in the future, just run the `start_` script again. This script creates an `installer_files` folder where it sets up the project's requirements. If you need to reinstall the requirements, you can simply delete that folder and start the web UI again. -The script accepts command-line flags. Alternatively, you can edit the `CMD_FLAGS.txt` file with a text editor and add your flags there. +The script accepts command-line flags. Alternatively, you can edit the `CMD_FLAGS.txt` file with a text editor and add your flags there, such as `--api` in case you need to use the API. To get updates in the future, run `update_wizard_linux.sh`, `update_wizard_windows.bat`, `update_wizard_macos.sh`, or `update_wizard_wsl.bat`. From 8bac1a93822885d1e53ce0dca5803f1ac19e3a55 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 19 Aug 2024 23:10:04 -0300 Subject: [PATCH 0008/1701] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 017fafcb5c..fe6a33cad6 100644 --- a/README.md +++ b/README.md @@ -404,7 +404,7 @@ https://colab.research.google.com/github/oobabooga/text-generation-webui/blob/ma ## Community -* Subreddit: https://www.reddit.com/r/oobabooga/ +* Subreddit: https://www.reddit.com/r/Oobabooga/ * Discord: https://discord.gg/jwZCF2dPQN ## Acknowledgment From 68f928b5e0f89b90b81f5b029f4e40c077df5e90 Mon Sep 17 00:00:00 2001 From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com> Date: Mon, 19 Aug 2024 23:33:56 -0300 Subject: [PATCH 0009/1701] Update peft requirement from ==0.8.* to ==0.12.* (#6292) --- requirements.txt | 2 +- requirements_amd.txt | 2 +- requirements_amd_noavx2.txt | 2 +- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_cpu_only.txt | 2 +- requirements_cpu_only_noavx2.txt | 2 +- requirements_noavx2.txt | 2 +- requirements_nowheels.txt | 2 +- 9 files changed, 9 insertions(+), 9 deletions(-) diff --git a/requirements.txt b/requirements.txt index a4b864b139..4cda446a8f 100644 --- a/requirements.txt +++ b/requirements.txt @@ -14,7 +14,7 @@ numba==0.59.* numpy==1.26.* optimum==1.17.* pandas -peft==0.8.* +peft==0.12.* Pillow>=9.5.0 psutil pyyaml diff --git a/requirements_amd.txt b/requirements_amd.txt index db765afe02..a4fcf73cce 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -11,7 +11,7 @@ numba==0.59.* numpy==1.26.* optimum==1.17.* pandas -peft==0.8.* +peft==0.12.* Pillow>=9.5.0 psutil pyyaml diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 3751af0e01..9ae58d17ee 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -11,7 +11,7 @@ numba==0.59.* numpy==1.26.* optimum==1.17.* pandas -peft==0.8.* +peft==0.12.* Pillow>=9.5.0 psutil pyyaml diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index c38fad1cb1..3b868edd41 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -11,7 +11,7 @@ numba==0.59.* numpy==1.26.* optimum==1.17.* pandas -peft==0.8.* +peft==0.12.* Pillow>=9.5.0 psutil pyyaml diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 0f2f965683..ba9c495afe 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -11,7 +11,7 @@ numba==0.59.* numpy==1.26.* optimum==1.17.* pandas -peft==0.8.* +peft==0.12.* Pillow>=9.5.0 psutil pyyaml diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 9a46e3665d..80895f17b6 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -11,7 +11,7 @@ numba==0.59.* numpy==1.26.* optimum==1.17.* pandas -peft==0.8.* +peft==0.12.* Pillow>=9.5.0 psutil pyyaml diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index e78ce3e04d..32a9ed20fc 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -11,7 +11,7 @@ numba==0.59.* numpy==1.26.* optimum==1.17.* pandas -peft==0.8.* +peft==0.12.* Pillow>=9.5.0 psutil pyyaml diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 5cb0d19aa4..4e33bb0b66 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -14,7 +14,7 @@ numba==0.59.* numpy==1.26.* optimum==1.17.* pandas -peft==0.8.* +peft==0.12.* Pillow>=9.5.0 psutil pyyaml diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index a95098a5c5..5ec157c5b0 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -11,7 +11,7 @@ numba==0.59.* numpy==1.26.* optimum==1.17.* pandas -peft==0.8.* +peft==0.12.* Pillow>=9.5.0 psutil pyyaml From 64e16e9a46d677e0b25632211efca64062980588 Mon Sep 17 00:00:00 2001 From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com> Date: Mon, 19 Aug 2024 23:34:10 -0300 Subject: [PATCH 0010/1701] Update accelerate requirement from ==0.32.* to ==0.33.* (#6291) --- requirements.txt | 2 +- requirements_amd.txt | 2 +- requirements_amd_noavx2.txt | 2 +- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_cpu_only.txt | 2 +- requirements_cpu_only_noavx2.txt | 2 +- requirements_noavx2.txt | 2 +- requirements_nowheels.txt | 2 +- 9 files changed, 9 insertions(+), 9 deletions(-) diff --git a/requirements.txt b/requirements.txt index 4cda446a8f..08b7d56d20 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,4 +1,4 @@ -accelerate==0.32.* +accelerate==0.33.* aqlm[gpu,cpu]==1.1.6; platform_system == "Linux" auto-gptq==0.7.1 bitsandbytes==0.43.* diff --git a/requirements_amd.txt b/requirements_amd.txt index a4fcf73cce..52e36510cf 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -1,4 +1,4 @@ -accelerate==0.32.* +accelerate==0.33.* colorama datasets einops diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 9ae58d17ee..18a81d040a 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==0.32.* +accelerate==0.33.* colorama datasets einops diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 3b868edd41..af02904b30 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -1,4 +1,4 @@ -accelerate==0.32.* +accelerate==0.33.* colorama datasets einops diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index ba9c495afe..8cdd8519c7 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -1,4 +1,4 @@ -accelerate==0.32.* +accelerate==0.33.* colorama datasets einops diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 80895f17b6..807c182a0d 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -1,4 +1,4 @@ -accelerate==0.32.* +accelerate==0.33.* colorama datasets einops diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index 32a9ed20fc..e2a8993646 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==0.32.* +accelerate==0.33.* colorama datasets einops diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 4e33bb0b66..d22eb72c2a 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==0.32.* +accelerate==0.33.* aqlm[gpu,cpu]==1.1.6; platform_system == "Linux" auto-gptq==0.7.1 bitsandbytes==0.43.* diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index 5ec157c5b0..ffb45fe39d 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -1,4 +1,4 @@ -accelerate==0.32.* +accelerate==0.33.* colorama datasets einops From fd9cb266196f9e90a850f37f5f33b69057329721 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 19 Aug 2024 19:38:43 -0700 Subject: [PATCH 0011/1701] UI: update the DRY parameters descriptions/order --- modules/ui_parameters.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py index 234e1af2a9..eff62c20e5 100644 --- a/modules/ui_parameters.py +++ b/modules/ui_parameters.py @@ -40,9 +40,9 @@ def create_ui(default_preset): shared.gradio['do_sample'] = gr.Checkbox(value=generate_params['do_sample'], label='do_sample') with gr.Blocks(): - shared.gradio['dry_multiplier'] = gr.Slider(0, 5, value=generate_params['dry_multiplier'], step=0.01, label='dry_multiplier', info='Set to value > 0 to enable DRY. Controls the magnitude of the penalty for the shortest penalized sequences.') - shared.gradio['dry_base'] = gr.Slider(1, 4, value=generate_params['dry_base'], step=0.01, label='dry_base', info='Controls how fast the penalty grows with increasing sequence length.') + shared.gradio['dry_multiplier'] = gr.Slider(0, 5, value=generate_params['dry_multiplier'], step=0.01, label='dry_multiplier', info='Set to greater than 0 to enable DRY. Recommended value: 0.8.') shared.gradio['dry_allowed_length'] = gr.Slider(1, 20, value=generate_params['dry_allowed_length'], step=1, label='dry_allowed_length', info='Longest sequence that can be repeated without being penalized.') + shared.gradio['dry_base'] = gr.Slider(1, 4, value=generate_params['dry_base'], step=0.01, label='dry_base', info='Controls how fast the penalty grows with increasing sequence length.') shared.gradio['dry_sequence_breakers'] = gr.Textbox(value=generate_params['dry_sequence_breakers'], label='dry_sequence_breakers', info='Tokens across which sequence matching is not continued. Specified as a comma-separated list of quoted strings.') gr.Markdown("[Learn more](https://github.com/oobabooga/text-generation-webui/wiki/03-%E2%80%90-Parameters-Tab)") From 1b1518aa6a702b3a494f712c3422fc5efcc0daf5 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 20 Aug 2024 00:36:18 -0300 Subject: [PATCH 0012/1701] Update README.md --- README.md | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/README.md b/README.md index fe6a33cad6..230dfccfba 100644 --- a/README.md +++ b/README.md @@ -11,10 +11,9 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github. ## Features * Multiple backends for text generation in a single UI and API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp) (through [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)), [ExLlamaV2](https://github.com/turboderp/exllamav2), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). -* OpenAI-compatible API server with Chat and Completions endpoints -- see the [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples). +* OpenAI-compatible API server with Chat and Completions endpoints – see the [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples). * Automatic prompt formatting for each model using the Jinja2 template in its metadata, ensuring high-quality outputs without manual intervention. -* Three chat modes: `instruct`, `chat-instruct`, and `chat`, allowing for both task-based interactions and casual conversations with characters. -* High-quality outputs in chat scenarios with the `chat-instruct` mode, which uses the model's template automatically. +* Three chat modes: `instruct`, `chat-instruct`, and `chat`, allowing for both task-based interactions and casual conversations with characters. `chat-instruct` mode automatically applies the model's template to the chat's prompt, leading to higher quality outputs. * Easy switching between conversations and starting new ones through the "Past chats" menu in the main interface tab. * Flexible text generation through autocompletion in the Default/Notebook tabs without being limited to chat turns. Send formatted chat conversations from the Chat tab to these tabs. * Multiple sampling parameters and options for sophisticated text generation control. From 406995f7224073a7bfbee6a2ce4e67f723a21e36 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 19 Aug 2024 21:24:01 -0700 Subject: [PATCH 0013/1701] Update README --- README.md | 22 ++++++++++------------ 1 file changed, 10 insertions(+), 12 deletions(-) diff --git a/README.md b/README.md index 230dfccfba..0e264c7551 100644 --- a/README.md +++ b/README.md @@ -10,7 +10,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github. ## Features -* Multiple backends for text generation in a single UI and API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp) (through [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)), [ExLlamaV2](https://github.com/turboderp/exllamav2), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). +* Multiple backends for text generation in a single UI and API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp) (through [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)), [ExLlamaV2](https://github.com/turboderp/exllamav2), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). The Transformers loader also supports models quantized through [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) and [HQQ](https://github.com/mobiusml/hqq). * OpenAI-compatible API server with Chat and Completions endpoints – see the [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples). * Automatic prompt formatting for each model using the Jinja2 template in its metadata, ensuring high-quality outputs without manual intervention. * Three chat modes: `instruct`, `chat-instruct`, and `chat`, allowing for both task-based interactions and casual conversations with characters. `chat-instruct` mode automatically applies the model's template to the chat's prompt, leading to higher quality outputs. @@ -209,13 +209,13 @@ usage: server.py [-h] [--multi-user] [--character CHARACTER] [--model MODEL] [-- [--force-safetensors] [--no_use_fast] [--use_flash_attention_2] [--use_eager_attention] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn] [--tensorcores] [--n_ctx N_CTX] [--threads THREADS] [--threads-batch THREADS_BATCH] [--no_mul_mat_q] [--n_batch N_BATCH] [--no-mmap] [--mlock] [--n-gpu-layers N_GPU_LAYERS] [--tensor_split TENSOR_SPLIT] [--numa] [--logits_all] [--no_offload_kqv] [--cache-capacity CACHE_CAPACITY] [--row_split] [--streaming-llm] - [--attention-sink-size ATTENTION_SINK_SIZE] [--gpu-split GPU_SPLIT] [--autosplit] [--max_seq_len MAX_SEQ_LEN] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa] - [--cache_8bit] [--cache_4bit] [--num_experts_per_token NUM_EXPERTS_PER_TOKEN] [--triton] [--no_inject_fused_mlp] [--no_use_cuda_fp16] [--desc_act] [--disable_exllama] - [--disable_exllamav2] [--wbits WBITS] [--groupsize GROUPSIZE] [--no_inject_fused_attention] [--hqq-backend HQQ_BACKEND] [--cpp-runner] [--deepspeed] - [--nvme-offload-dir NVME_OFFLOAD_DIR] [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB] [--listen] - [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] - [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--nowebui] - [--multimodal-pipeline MULTIMODAL_PIPELINE] [--model_type MODEL_TYPE] [--pre_layer PRE_LAYER [PRE_LAYER ...]] [--checkpoint CHECKPOINT] [--monkey-patch] + [--attention-sink-size ATTENTION_SINK_SIZE] [--tokenizer-dir TOKENIZER_DIR] [--gpu-split GPU_SPLIT] [--autosplit] [--max_seq_len MAX_SEQ_LEN] [--cfg-cache] [--no_flash_attn] + [--no_xformers] [--no_sdpa] [--cache_8bit] [--cache_4bit] [--num_experts_per_token NUM_EXPERTS_PER_TOKEN] [--triton] [--no_inject_fused_mlp] [--no_use_cuda_fp16] [--desc_act] + [--disable_exllama] [--disable_exllamav2] [--wbits WBITS] [--groupsize GROUPSIZE] [--hqq-backend HQQ_BACKEND] [--cpp-runner] [--deepspeed] [--nvme-offload-dir NVME_OFFLOAD_DIR] + [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT] + [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] + [--subpath SUBPATH] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--nowebui] + [--multimodal-pipeline MULTIMODAL_PIPELINE] [--model_type MODEL_TYPE] [--pre_layer PRE_LAYER [PRE_LAYER ...]] [--checkpoint CHECKPOINT] [--monkey-patch] [--no_inject_fused_attention] Text generation web UI @@ -239,7 +239,7 @@ Basic settings: Model loader: --loader LOADER Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav2_HF, ExLlamav2, - AutoGPTQ, AutoAWQ. + AutoGPTQ. Transformers/Accelerate: --cpu Use the CPU to generate text. Warning: Training on CPU is extremely slow. @@ -283,6 +283,7 @@ llama.cpp: --row_split Split the model by rows across GPUs. This may improve multi-gpu performance. --streaming-llm Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed. --attention-sink-size ATTENTION_SINK_SIZE StreamingLLM: number of sink tokens. Only used if the trimmed prompt does not share a prefix with the old prompt. + --tokenizer-dir TOKENIZER_DIR Load the tokenizer from this folder. Meant to be used with llamacpp_HF through the command-line. ExLlamaV2: --gpu-split GPU_SPLIT Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7. @@ -306,9 +307,6 @@ AutoGPTQ: --wbits WBITS Load a pre-quantized model with specified precision in bits. 2, 3, 4 and 8 are supported. --groupsize GROUPSIZE Group size. -AutoAWQ: - --no_inject_fused_attention Disable the use of fused attention, which will use less VRAM at the cost of slower inference. - HQQ: --hqq-backend HQQ_BACKEND Backend for the HQQ loader. Valid options: PYTORCH, PYTORCH_COMPILE, ATEN. From 9d99156ca395455f2c871a77c352bb9fa93e02ca Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 20 Aug 2024 01:27:02 -0300 Subject: [PATCH 0014/1701] Update README.md --- README.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index 0e264c7551..6f2d73964f 100644 --- a/README.md +++ b/README.md @@ -10,9 +10,9 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github. ## Features -* Multiple backends for text generation in a single UI and API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp) (through [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)), [ExLlamaV2](https://github.com/turboderp/exllamav2), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). The Transformers loader also supports models quantized through [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) and [HQQ](https://github.com/mobiusml/hqq). +* Multiple backends for text generation in a single UI and API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp) (through [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)), [ExLlamaV2](https://github.com/turboderp/exllamav2), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM) are also supported through the Transformers loader. * OpenAI-compatible API server with Chat and Completions endpoints – see the [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples). -* Automatic prompt formatting for each model using the Jinja2 template in its metadata, ensuring high-quality outputs without manual intervention. +* Automatic prompt formatting for each model using the Jinja2 template in its metadata, ensuring high-quality outputs without manual setup. * Three chat modes: `instruct`, `chat-instruct`, and `chat`, allowing for both task-based interactions and casual conversations with characters. `chat-instruct` mode automatically applies the model's template to the chat's prompt, leading to higher quality outputs. * Easy switching between conversations and starting new ones through the "Past chats" menu in the main interface tab. * Flexible text generation through autocompletion in the Default/Notebook tabs without being limited to chat turns. Send formatted chat conversations from the Chat tab to these tabs. From d9a031fcadbbb241be0d26a4041903a8cc659023 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 20 Aug 2024 01:52:30 -0300 Subject: [PATCH 0015/1701] Update README.md --- README.md | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/README.md b/README.md index 6f2d73964f..9d73f910ee 100644 --- a/README.md +++ b/README.md @@ -13,13 +13,13 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github. * Multiple backends for text generation in a single UI and API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp) (through [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)), [ExLlamaV2](https://github.com/turboderp/exllamav2), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM) are also supported through the Transformers loader. * OpenAI-compatible API server with Chat and Completions endpoints – see the [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples). * Automatic prompt formatting for each model using the Jinja2 template in its metadata, ensuring high-quality outputs without manual setup. -* Three chat modes: `instruct`, `chat-instruct`, and `chat`, allowing for both task-based interactions and casual conversations with characters. `chat-instruct` mode automatically applies the model's template to the chat's prompt, leading to higher quality outputs. -* Easy switching between conversations and starting new ones through the "Past chats" menu in the main interface tab. -* Flexible text generation through autocompletion in the Default/Notebook tabs without being limited to chat turns. Send formatted chat conversations from the Chat tab to these tabs. -* Multiple sampling parameters and options for sophisticated text generation control. -* Quick downloading and loading of new models through the interface without restarting, using the "Model" tab. +* Three chat modes: `instruct`, `chat-instruct`, and `chat`, allowing for both instruction-following and casual conversations with characters. `chat-instruct` mode automatically applies the model's template to the chat prompt, leading to higher quality outputs. +* "Past chats" menu to quickly switch between conversations and start new ones in the Chat tab. +* Free-form generation in the Default/Notebook tabs without being limited to chat turns. Send formatted chat conversations from the Chat tab to these tabs. +* Multiple sampling parameters and generation options for sophisticated text generation control. +* Easy switching between different models through the UI without restarting, using the "Model" tab. * Simple LoRA fine-tuning tool to customize models with your data. -* Self-contained dependencies in the `installer_files` folder, avoiding interference with the system's Python environment. Precompiled Python wheels for the backends are in the `requirements.txt` and are transparently compiled using GitHub Actions. +* All in one folder. The requirements are installed in a self-contained `installer_files` folder that doesn't interfere with the system's environment. * Extensions support, including numerous built-in and user-contributed extensions. See [the wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [the extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details. ## How to install From 1124f71cf3c4e5edb9c9a5e8e8c49314e4a226c7 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 20 Aug 2024 11:19:46 -0300 Subject: [PATCH 0016/1701] Update README.md --- README.md | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-) diff --git a/README.md b/README.md index 9d73f910ee..22ceeeb8ea 100644 --- a/README.md +++ b/README.md @@ -12,9 +12,9 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github. * Multiple backends for text generation in a single UI and API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp) (through [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)), [ExLlamaV2](https://github.com/turboderp/exllamav2), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM) are also supported through the Transformers loader. * OpenAI-compatible API server with Chat and Completions endpoints – see the [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples). -* Automatic prompt formatting for each model using the Jinja2 template in its metadata, ensuring high-quality outputs without manual setup. -* Three chat modes: `instruct`, `chat-instruct`, and `chat`, allowing for both instruction-following and casual conversations with characters. `chat-instruct` mode automatically applies the model's template to the chat prompt, leading to higher quality outputs. -* "Past chats" menu to quickly switch between conversations and start new ones in the Chat tab. +* Automatic prompt formatting for each model using the Jinja2 template in its metadata. +* Three chat modes: `instruct`, `chat-instruct`, and `chat`, allowing for both instruction-following and casual conversations with characters. `chat-instruct` mode automatically applies the model's template to the chat prompt, ensuring high-quality outputs without manual setup. +* "Past chats" menu to quickly switch between conversations and start new ones. * Free-form generation in the Default/Notebook tabs without being limited to chat turns. Send formatted chat conversations from the Chat tab to these tabs. * Multiple sampling parameters and generation options for sophisticated text generation control. * Easy switching between different models through the UI without restarting, using the "Model" tab. @@ -30,9 +30,11 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github. 4) Once the installation ends, browse to `http://localhost:7860`. 5) Have fun! -To restart the web UI in the future, just run the `start_` script again. This script creates an `installer_files` folder where it sets up the project's requirements. If you need to reinstall the requirements, you can simply delete that folder and start the web UI again. +To restart the web UI in the future, run the `start_` script again. -The script accepts command-line flags. Alternatively, you can edit the `CMD_FLAGS.txt` file with a text editor and add your flags there, such as `--api` in case you need to use the API. +This script creates an `installer_files` folder where it sets up the project's requirements. If you need to reinstall the requirements, just delete that folder and start the web UI again. + +The script accepts command-line flags, such as `./start_linux.sh --help`. Alternatively, you can edit the `CMD_FLAGS.txt` file with a text editor and add your flags there, such as `--api` in case you need to use the API. To get updates in the future, run `update_wizard_linux.sh`, `update_wizard_windows.bat`, `update_wizard_macos.sh`, or `update_wizard_wsl.bat`. From c24966c5914e974b509e611f01527e1a2854b4c3 Mon Sep 17 00:00:00 2001 From: joachimchauvet <13382894+joachimchauvet@users.noreply.github.com> Date: Wed, 21 Aug 2024 21:33:45 +0300 Subject: [PATCH 0017/1701] update API documentation with examples to list/load models (#5902) --- docs/12 - OpenAI API.md | 26 +++++++++++++++++++++++++- 1 file changed, 25 insertions(+), 1 deletion(-) diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md index b00a1f3447..9b4f89bf17 100644 --- a/docs/12 - OpenAI API.md +++ b/docs/12 - OpenAI API.md @@ -19,7 +19,7 @@ Add `--api` to your command-line flags. ### Examples -For the documentation with all the parameters and their types, consult `http://127.0.0.1:5000/docs` or the [typing.py](https://github.com/oobabooga/text-generation-webui/blob/main/extensions/openai/typing.py) file. +For the documentation with all the endpoints, parameters and their types, consult `http://127.0.0.1:5000/docs` or the [typing.py](https://github.com/oobabooga/text-generation-webui/blob/main/extensions/openai/typing.py) file. The official examples in the [OpenAI documentation](https://platform.openai.com/docs/api-reference) should also work, and the same parameters apply (although the API here has more optional parameters). @@ -114,6 +114,30 @@ curl -k http://127.0.0.1:5000/v1/internal/logits \ }' ``` +#### List models + +```shell +curl -k http://127.0.0.1:5000/v1/internal/model/list \ + -H "Content-Type: application/json" +``` + +#### Load model + +```shell +curl -k http://127.0.0.1:5000/v1/internal/model/load \ + -H "Content-Type: application/json" \ + -d '{ + "model_name": "model_name", + "args": { + "load_in_4bit": true, + "n_gpu_layers": 12 + }, + "settings": { + "instruction_template": "Alpaca" + } + }' +``` + #### Python chat example ```python From 1f288b40720bbef6ed32ad5dee1899b9e5949925 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Thu, 22 Aug 2024 12:40:15 -0700 Subject: [PATCH 0018/1701] Bump ExLlamaV2 to 0.1.9 --- requirements.txt | 10 +++++----- requirements_amd.txt | 6 +++--- requirements_amd_noavx2.txt | 6 +++--- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_noavx2.txt | 10 +++++----- 6 files changed, 18 insertions(+), 18 deletions(-) diff --git a/requirements.txt b/requirements.txt index 08b7d56d20..a39a7c707a 100644 --- a/requirements.txt +++ b/requirements.txt @@ -53,11 +53,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+cu121.torch2.2.2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" https://github.com/oobabooga/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu122torch2.2.2cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu122torch2.2.2cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu123torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index 52e36510cf..5ff8686326 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -40,9 +40,9 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp # AMD wheels https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.89+rocm5.6.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.89+rocm5.6.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 18a81d040a..c0d094e881 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -38,9 +38,9 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index af02904b30..cbcf117515 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -36,4 +36,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8-py3-none-any.whl +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9-py3-none-any.whl diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 8cdd8519c7..d5cb367570 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -38,4 +38,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8-py3-none-any.whl +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9-py3-none-any.whl diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index d22eb72c2a..bfe57329a8 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -53,11 +53,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+cu121.torch2.2.2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" https://github.com/oobabooga/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu122torch2.2.2cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu122torch2.2.2cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu123torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" From 41a8eb4eebf02cd5e7d75786229843748579c0b0 Mon Sep 17 00:00:00 2001 From: "FartyPants (FP HAM)" Date: Mon, 2 Sep 2024 22:00:15 -0400 Subject: [PATCH 0019/1701] Training pro update script.py (#6359) --- extensions/Training_PRO/script.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/extensions/Training_PRO/script.py b/extensions/Training_PRO/script.py index 8f29646232..5365154c35 100644 --- a/extensions/Training_PRO/script.py +++ b/extensions/Training_PRO/script.py @@ -241,7 +241,7 @@ def ui(): stride_length = gr.Slider(label='Stride', minimum=1, maximum=2048, value=512, step=1, info='Used to make the evaluation faster at the cost of accuracy. 1 = slowest but most accurate. 512 is a common value.') with gr.Column(): - max_length = gr.Slider(label='max_length', minimum=0, maximum=shared.settings['truncation_length_max'], value=0, step=1, info='The context for each evaluation. If set to 0, the maximum context length for the model will be used.') + max_length = gr.Number(label='max_length', precision=0, step=256, value=0, info='The context for each evaluation. If set to 0, the maximum context length for the model will be used.') with gr.Row(): start_current_evaluation = gr.Button("Evaluate loaded model") From 4c74c7a1167defef1f9ea217507990974c8fde3d Mon Sep 17 00:00:00 2001 From: GralchemOz <68577430+GralchemOz@users.noreply.github.com> Date: Tue, 3 Sep 2024 10:00:59 +0800 Subject: [PATCH 0020/1701] Fix UnicodeDecodeError for BPE-based Models (especially GLM-4) (#6357) --- modules/text_generation.py | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/modules/text_generation.py b/modules/text_generation.py index 75e5ef36ae..e7a2b43f01 100644 --- a/modules/text_generation.py +++ b/modules/text_generation.py @@ -274,7 +274,12 @@ def get_reply_from_output_ids(output_ids, state=None, starting_from=0): if (hasattr(shared.tokenizer, 'convert_ids_to_tokens') and len(output_ids) > starting_from) and not reply.startswith(' '): first_token = shared.tokenizer.convert_ids_to_tokens(int(output_ids[starting_from])) if isinstance(first_token, (bytes,)): - first_token = first_token.decode('utf8') + #try to decode the bytes to a string + try: + first_token = first_token.decode('utf8') + #if it fails, which means it's not a string in this turn, just ignore it + except UnicodeDecodeError: + first_token = '' if first_token.startswith('▁'): reply = ' ' + reply From 9a150c3368fb61cfd77f3286590625467fdc5413 Mon Sep 17 00:00:00 2001 From: Stefan Merettig Date: Tue, 3 Sep 2024 04:03:15 +0200 Subject: [PATCH 0021/1701] API: Relax multimodal format, fixes HuggingFace Chat UI (#6353) --- extensions/openai/completions.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py index 646dee2d91..362620fac8 100644 --- a/extensions/openai/completions.py +++ b/extensions/openai/completions.py @@ -154,8 +154,9 @@ def convert_history(history): elif item['type'] == 'text' and isinstance(item['text'], str): content = item['text'] - if image_url and content: + if image_url: new_history.append({"image_url": image_url, "role": "user"}) + if content: new_history.append({"content": content, "role": "user"}) else: new_history.append(entry) From d1168afa76900c8632a729984ed383ee9c83220c Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 2 Sep 2024 21:15:51 -0700 Subject: [PATCH 0022/1701] Bump ExLlamaV2 to 0.2.0 --- requirements.txt | 10 +++++----- requirements_amd.txt | 6 +++--- requirements_amd_noavx2.txt | 6 +++--- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_noavx2.txt | 10 +++++----- 6 files changed, 18 insertions(+), 18 deletions(-) diff --git a/requirements.txt b/requirements.txt index a39a7c707a..076d8dad50 100644 --- a/requirements.txt +++ b/requirements.txt @@ -53,11 +53,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" https://github.com/oobabooga/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu122torch2.2.2cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu122torch2.2.2cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu123torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index 5ff8686326..b05cbebe6f 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -40,9 +40,9 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp # AMD wheels https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.89+rocm5.6.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.89+rocm5.6.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index c0d094e881..f30543c135 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -38,9 +38,9 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index cbcf117515..33760ab445 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -36,4 +36,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9-py3-none-any.whl +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index d5cb367570..dda1e25b47 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -38,4 +38,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9-py3-none-any.whl +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index bfe57329a8..48b2e85811 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -53,11 +53,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" https://github.com/oobabooga/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu122torch2.2.2cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu122torch2.2.2cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu123torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" From 64919e0d69bb5e6a653f866cde313c1ae1b668c9 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 3 Sep 2024 05:51:46 -0700 Subject: [PATCH 0023/1701] Bump flash-attention to 2.6.3 --- requirements.txt | 8 ++++---- requirements_noavx2.txt | 8 ++++---- 2 files changed, 8 insertions(+), 8 deletions(-) diff --git a/requirements.txt b/requirements.txt index 076d8dad50..db6f006112 100644 --- a/requirements.txt +++ b/requirements.txt @@ -58,10 +58,10 @@ https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+ https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" -https://github.com/oobabooga/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu122torch2.2.2cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu122torch2.2.2cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu123torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu123torch2.2cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.2.2cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.2.2cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.2cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 48b2e85811..5d8ae64081 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -58,10 +58,10 @@ https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+ https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" -https://github.com/oobabooga/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu122torch2.2.2cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu122torch2.2.2cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu123torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.1/flash_attn-2.6.1+cu123torch2.2cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.2.2cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.2.2cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.2cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" From 2cb8d4c96ea3272a7e5e8a6c94c0541b8af5bfbc Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 3 Sep 2024 05:53:18 -0700 Subject: [PATCH 0024/1701] Bump llama-cpp-python to 0.2.90 --- requirements.txt | 24 ++++++++++++------------ requirements_amd.txt | 12 ++++++------ requirements_amd_noavx2.txt | 8 ++++---- requirements_apple_intel.txt | 8 ++++---- requirements_apple_silicon.txt | 12 ++++++------ requirements_cpu_only.txt | 8 ++++---- requirements_cpu_only_noavx2.txt | 8 ++++---- requirements_noavx2.txt | 24 ++++++++++++------------ 8 files changed, 52 insertions(+), 52 deletions(-) diff --git a/requirements.txt b/requirements.txt index db6f006112..42eb0231c6 100644 --- a/requirements.txt +++ b/requirements.txt @@ -35,22 +35,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index b05cbebe6f..05b60c4821 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -32,14 +32,14 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.89+rocm5.6.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.89+rocm5.6.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.90+rocm5.6.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.90+rocm5.6.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index f30543c135..5d885506ab 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -32,10 +32,10 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 33760ab445..674c2ffc29 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -32,8 +32,8 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index dda1e25b47..7a63a82b6a 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -32,10 +32,10 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.89-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 807c182a0d..42d56eb5c4 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -32,7 +32,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index e2a8993646..aacdc82dcb 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -32,7 +32,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 5d8ae64081..316576c2c7 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -35,22 +35,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.89+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.89+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.89+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" From c5b40eb5558230fe5894cfd9388809429f5c799f Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 3 Sep 2024 17:37:06 -0700 Subject: [PATCH 0025/1701] llama.cpp: prevent prompt evaluation progress bar with just 1 step --- modules/llama_cpp_python_hijack.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/llama_cpp_python_hijack.py b/modules/llama_cpp_python_hijack.py index 64280dc9a0..3d42b2d7d9 100644 --- a/modules/llama_cpp_python_hijack.py +++ b/modules/llama_cpp_python_hijack.py @@ -61,7 +61,7 @@ def eval_with_progress(self, tokens: Sequence[int]): assert self._batch.batch is not None self._ctx.kv_cache_seq_rm(-1, self.n_tokens, -1) - if len(tokens) > 1: + if len(tokens) > self.n_batch: progress_bar = tqdm(range(0, len(tokens), self.n_batch), desc="Prompt evaluation", leave=False) else: progress_bar = range(0, len(tokens), self.n_batch) From bba5b36d332b2b3c0297c85cd7d07e5a50e3f86c Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 3 Sep 2024 19:40:53 -0700 Subject: [PATCH 0026/1701] Don't import PEFT unless necessary --- modules/LoRA.py | 4 +++- modules/training.py | 18 ++++++++++-------- 2 files changed, 13 insertions(+), 9 deletions(-) diff --git a/modules/LoRA.py b/modules/LoRA.py index 117022cfc8..4fd144ba00 100644 --- a/modules/LoRA.py +++ b/modules/LoRA.py @@ -1,7 +1,6 @@ from pathlib import Path import torch -from peft import PeftModel from transformers import is_torch_xpu_available import modules.shared as shared @@ -85,6 +84,9 @@ def add_lora_autogptq(lora_names): def add_lora_transformers(lora_names): + + from peft import PeftModel + prior_set = set(shared.lora_names) added_set = set(lora_names) - prior_set removed_set = prior_set - set(lora_names) diff --git a/modules/training.py b/modules/training.py index b003fc8c28..11c4b8c5d7 100644 --- a/modules/training.py +++ b/modules/training.py @@ -18,14 +18,6 @@ import torch import transformers from datasets import Dataset, load_dataset -from peft import ( - LoraConfig, - get_peft_model, - prepare_model_for_kbit_training, - set_peft_model_state_dict -) -from peft.utils.other import \ - TRANSFORMERS_MODELS_TO_LORA_TARGET_MODULES_MAPPING as model_to_lora_modules from transformers import is_torch_xpu_available from transformers.models.auto.modeling_auto import ( MODEL_FOR_CAUSAL_LM_MAPPING_NAMES @@ -292,6 +284,16 @@ def calc_trainable_parameters(model): def do_train(lora_name: str, always_override: bool, q_proj_en: bool, v_proj_en: bool, k_proj_en: bool, o_proj_en: bool, gate_proj_en: bool, down_proj_en: bool, up_proj_en: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, raw_text_file: str, overlap_len: int, newline_favor_len: int, higher_rank_limit: bool, warmup_steps: int, optimizer: str, hard_cut_string: str, train_only_after: str, stop_at_loss: float, add_eos_token: bool, min_chars: int, report_to: str): + from peft import ( + LoraConfig, + get_peft_model, + prepare_model_for_kbit_training, + set_peft_model_state_dict + ) + from peft.utils.other import \ + TRANSFORMERS_MODELS_TO_LORA_TARGET_MODULES_MAPPING as \ + model_to_lora_modules + global WANT_INTERRUPT WANT_INTERRUPT = False From 4924ee2901ce9ad17259ae817bbb46885d1ba8ba Mon Sep 17 00:00:00 2001 From: Jean-Sylvain Boige Date: Fri, 6 Sep 2024 02:42:23 +0200 Subject: [PATCH 0027/1701] typo in OpenAI response format (#6365) --- extensions/openai/completions.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py index 362620fac8..6bd8f409a0 100644 --- a/extensions/openai/completions.py +++ b/extensions/openai/completions.py @@ -235,7 +235,7 @@ def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, p raise InvalidRequestError(message="messages: missing content", param='messages') # Chat Completions - object_type = 'chat.completions' if not stream else 'chat.completions.chunk' + object_type = 'chat.completion' if not stream else 'chat.completion.chunk' created_time = int(time.time()) cmpl_id = "chatcmpl-%d" % (int(time.time() * 1000000000)) resp_list = 'data' if is_legacy else 'choices' From 27797a92d051eef3289c510906e75827dbc4886c Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Fri, 6 Sep 2024 18:38:39 -0700 Subject: [PATCH 0028/1701] Pin fastapi/pydantic requirement versions --- requirements.txt | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/requirements.txt b/requirements.txt index 42eb0231c6..7164ac8072 100644 --- a/requirements.txt +++ b/requirements.txt @@ -5,6 +5,7 @@ bitsandbytes==0.43.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -17,6 +18,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich @@ -69,4 +71,4 @@ https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310 https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" \ No newline at end of file +https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" From ac30b004efd7b9e2511d0baf619e8025c745ba09 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Fri, 6 Sep 2024 18:38:39 -0700 Subject: [PATCH 0029/1701] Pin fastapi/pydantic requirement versions --- requirements.txt | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/requirements.txt b/requirements.txt index 08b7d56d20..51558a845a 100644 --- a/requirements.txt +++ b/requirements.txt @@ -5,6 +5,7 @@ bitsandbytes==0.43.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -17,6 +18,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich @@ -69,4 +71,4 @@ https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310 https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" \ No newline at end of file +https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" From a50477ec85a7b5367e1ebe3658bb270735715bea Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Fri, 6 Sep 2024 18:47:25 -0700 Subject: [PATCH 0030/1701] Apply the change to all requirements (oops) --- requirements_amd.txt | 4 +++- requirements_amd_noavx2.txt | 4 +++- requirements_apple_intel.txt | 2 ++ requirements_apple_silicon.txt | 2 ++ requirements_cpu_only.txt | 2 ++ requirements_cpu_only_noavx2.txt | 2 ++ requirements_noavx2.txt | 4 +++- requirements_nowheels.txt | 2 ++ 8 files changed, 19 insertions(+), 3 deletions(-) diff --git a/requirements_amd.txt b/requirements_amd.txt index 05b60c4821..20ba9c792a 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -2,6 +2,7 @@ accelerate==0.33.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -14,6 +15,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich @@ -46,4 +48,4 @@ https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0- https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" \ No newline at end of file +https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 5d885506ab..e5c18681bf 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -2,6 +2,7 @@ accelerate==0.33.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -14,6 +15,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich @@ -44,4 +46,4 @@ https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0- https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" \ No newline at end of file +https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 674c2ffc29..e60937c70e 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -2,6 +2,7 @@ accelerate==0.33.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -14,6 +15,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 7a63a82b6a..e1b726bbfe 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -2,6 +2,7 @@ accelerate==0.33.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -14,6 +15,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 42d56eb5c4..b0480c436b 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -2,6 +2,7 @@ accelerate==0.33.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -14,6 +15,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index aacdc82dcb..a64d68a743 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -2,6 +2,7 @@ accelerate==0.33.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -14,6 +15,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 316576c2c7..cae98bb2cd 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -5,6 +5,7 @@ bitsandbytes==0.43.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -17,6 +18,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich @@ -69,4 +71,4 @@ https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310 https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" \ No newline at end of file +https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index ffb45fe39d..ec18464d0f 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -2,6 +2,7 @@ accelerate==0.33.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -14,6 +15,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich From f98431c7448381bfa4e859ace70e0379f6431018 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Fri, 6 Sep 2024 18:47:25 -0700 Subject: [PATCH 0031/1701] Apply the change to all requirements (oops) --- requirements_amd.txt | 4 +++- requirements_amd_noavx2.txt | 4 +++- requirements_apple_intel.txt | 2 ++ requirements_apple_silicon.txt | 2 ++ requirements_cpu_only.txt | 2 ++ requirements_cpu_only_noavx2.txt | 2 ++ requirements_noavx2.txt | 4 +++- requirements_nowheels.txt | 2 ++ 8 files changed, 19 insertions(+), 3 deletions(-) diff --git a/requirements_amd.txt b/requirements_amd.txt index 52e36510cf..e9dfb1283e 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -2,6 +2,7 @@ accelerate==0.33.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -14,6 +15,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich @@ -46,4 +48,4 @@ https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8- https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" \ No newline at end of file +https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 18a81d040a..69a350ad54 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -2,6 +2,7 @@ accelerate==0.33.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -14,6 +15,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich @@ -44,4 +46,4 @@ https://github.com/oobabooga/exllamav2/releases/download/v0.1.8/exllamav2-0.1.8- https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" \ No newline at end of file +https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index af02904b30..0326916eee 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -2,6 +2,7 @@ accelerate==0.33.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -14,6 +15,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 8cdd8519c7..70ee3d27a1 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -2,6 +2,7 @@ accelerate==0.33.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -14,6 +15,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 807c182a0d..fe2d796ccd 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -2,6 +2,7 @@ accelerate==0.33.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -14,6 +15,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index e2a8993646..982a12a162 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -2,6 +2,7 @@ accelerate==0.33.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -14,6 +15,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index d22eb72c2a..fc253f5c27 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -5,6 +5,7 @@ bitsandbytes==0.43.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -17,6 +18,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich @@ -69,4 +71,4 @@ https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310 https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" \ No newline at end of file +https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index ffb45fe39d..ec18464d0f 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -2,6 +2,7 @@ accelerate==0.33.* colorama datasets einops +fastapi==0.112.4 gradio==4.26.* hqq==0.1.7.post3 jinja2==3.1.4 @@ -14,6 +15,7 @@ pandas peft==0.12.* Pillow>=9.5.0 psutil +pydantic==2.8.2 pyyaml requests rich From c497a3237229db0de4a4704e829d75a45973c91f Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Thu, 26 Sep 2024 11:55:51 -0700 Subject: [PATCH 0032/1701] Bump transformers to 4.45 --- requirements.txt | 2 +- requirements_amd.txt | 2 +- requirements_amd_noavx2.txt | 2 +- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_cpu_only.txt | 2 +- requirements_cpu_only_noavx2.txt | 2 +- requirements_noavx2.txt | 2 +- requirements_nowheels.txt | 2 +- 9 files changed, 9 insertions(+), 9 deletions(-) diff --git a/requirements.txt b/requirements.txt index 7164ac8072..913c246398 100644 --- a/requirements.txt +++ b/requirements.txt @@ -26,7 +26,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.44.* +transformers==4.45.* tqdm wandb diff --git a/requirements_amd.txt b/requirements_amd.txt index 20ba9c792a..b9d74b69a6 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -23,7 +23,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.44.* +transformers==4.45.* tqdm wandb diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index e5c18681bf..ccacab29df 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -23,7 +23,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.44.* +transformers==4.45.* tqdm wandb diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index e60937c70e..dc1757e264 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -23,7 +23,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.44.* +transformers==4.45.* tqdm wandb diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index e1b726bbfe..a913a63c25 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -23,7 +23,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.44.* +transformers==4.45.* tqdm wandb diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index b0480c436b..b0dbc91a86 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -23,7 +23,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.44.* +transformers==4.45.* tqdm wandb diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index a64d68a743..828d0c974e 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -23,7 +23,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.44.* +transformers==4.45.* tqdm wandb diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index cae98bb2cd..458c1410a1 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -26,7 +26,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.44.* +transformers==4.45.* tqdm wandb diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index ec18464d0f..e7cbc29cbe 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -23,7 +23,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.44.* +transformers==4.45.* tqdm wandb From 7424f789bf699c9b2efc874da2b84168bb9e7a6f Mon Sep 17 00:00:00 2001 From: oobabooga Date: Fri, 27 Sep 2024 19:03:25 -0300 Subject: [PATCH 0033/1701] Fix the sampling monkey patch (and add more options to sampler_priority) (#6411) --- modules/presets.py | 2 +- modules/sampler_hijack.py | 186 +++++++++++++++++++++++--------------- 2 files changed, 113 insertions(+), 75 deletions(-) diff --git a/modules/presets.py b/modules/presets.py index b00e829eb1..14c5a7776c 100644 --- a/modules/presets.py +++ b/modules/presets.py @@ -44,7 +44,7 @@ def default_preset(): 'dry_base': 1.75, 'dry_allowed_length': 2, 'dry_sequence_breakers': '"\\n", ":", "\\"", "*"', - 'sampler_priority': 'temperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nmirostat' + 'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nmirostat\nencoder_repetition_penalty\nno_repeat_ngram' } diff --git a/modules/sampler_hijack.py b/modules/sampler_hijack.py index 9fb661aecc..9e865bb341 100644 --- a/modules/sampler_hijack.py +++ b/modules/sampler_hijack.py @@ -323,62 +323,141 @@ def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> to class RepetitionPenaltyLogitsProcessorWithRange(LogitsProcessor): - ''' - Copied from the transformers library - ''' - - def __init__(self, penalty: float, presence_penalty: float, frequency_penalty: float, _range: int): + def __init__(self, penalty: float, _range: int): if not (penalty > 0): raise ValueError(f"`penalty` has to be strictly positive, but is {penalty}") - self.penalty = penalty - self.presence_penalty = presence_penalty - self.frequency_penalty = frequency_penalty self._range = _range + def apply_repetition_penalty(self, input_ids_row, scores_row): + unique_ids = torch.unique(input_ids_row) + score = torch.gather(scores_row, 0, unique_ids) + + # Apply multiplicative repetition penalty + score = torch.where(score < 0, score * self.penalty, score / self.penalty) + scores_row.scatter_(0, unique_ids, score) + return scores_row + def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> torch.FloatTensor: input_ids = input_ids[:, -self._range:] + for input_ids_row, scores_row in zip(input_ids, scores): + scores_row = self.apply_repetition_penalty(input_ids_row, scores_row) + + return scores + - # We loop here because torch.unique() needs to process each row separately in the - # case that batch_size > 1. +class PresencePenaltyLogitsProcessor(LogitsProcessor): + def __init__(self, presence_penalty: float, _range: int): + self.presence_penalty = presence_penalty + self._range = _range + + def apply_presence_penalty(self, input_ids_row, scores_row): + unique_ids, counts = torch.unique(input_ids_row, return_counts=True) + + # Apply presence penalty + raw_presence_penalty = (counts > 0).to(scores_row.dtype) + presence_penalty = raw_presence_penalty * self.presence_penalty + scores_row.scatter_add_(0, unique_ids, -presence_penalty) + return scores_row + + def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> torch.FloatTensor: + input_ids = input_ids[:, -self._range:] for input_ids_row, scores_row in zip(input_ids, scores): - unique_ids, counts = torch.unique(input_ids_row, return_counts=True) - score = torch.gather(scores_row, 0, unique_ids) + scores_row = self.apply_presence_penalty(input_ids_row, scores_row) + return scores + + +class FrequencyPenaltyLogitsProcessor(LogitsProcessor): + def __init__(self, frequency_penalty: float, _range: int): + self.frequency_penalty = frequency_penalty + self._range = _range - # multiplicative repetition penalty - # if score < 0 then repetition penalty has to be multiplied to reduce the previous token probability - score = torch.where(score < 0, score * self.penalty, score / self.penalty) - scores_row.scatter_(0, unique_ids, score) + def apply_frequency_penalty(self, input_ids_row, scores_row): + unique_ids, counts = torch.unique(input_ids_row, return_counts=True) - # presence_penalty and frequency_penalty - raw_presence_penalty = (counts > 0).to(scores.dtype) - raw_frequency_penalty = counts.to(scores.dtype) - additive_penalty = raw_presence_penalty * self.presence_penalty + raw_frequency_penalty * self.frequency_penalty - scores_row.scatter_add_(0, unique_ids, -additive_penalty) + # Apply frequency penalty + raw_frequency_penalty = counts.to(scores_row.dtype) + frequency_penalty = raw_frequency_penalty * self.frequency_penalty + scores_row.scatter_add_(0, unique_ids, -frequency_penalty) + return scores_row + def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> torch.FloatTensor: + input_ids = input_ids[:, -self._range:] + for input_ids_row, scores_row in zip(input_ids, scores): + scores_row = self.apply_frequency_penalty(input_ids_row, scores_row) return scores -def get_logits_warper_patch(self, generation_config, **kwargs): +def get_logits_processor_patch(self, **kwargs): + generation_config = kwargs['generation_config'] # Parameter sanitization if isinstance(generation_config.temperature, int): generation_config.temperature = float(generation_config.temperature) # Must be float # Get the original warpers - warpers = self._get_logits_warper_old(generation_config, **kwargs) + warpers = self._get_logits_processor_old(**kwargs) - # Replace temperature with our modified class. - # Currently, it behaves identically to the original. - for i in range(len(warpers)): + for i in range(len(warpers) - 1, -1, -1): + # Replace temperature with our modified class. if warpers[i].__class__.__name__ == 'TemperatureLogitsWarper': warpers[i] = TemperatureLogitsWarperCustom( generation_config.temperature, ) + # Stuff we don't need + elif warpers[i].__class__.__name__ in ['SuppressTokensLogitsProcessor', 'RepetitionPenaltyLogitsProcessor']: + del warpers[i] + # Add custom warpers warpers_to_add = LogitsProcessorList() min_tokens_to_keep = 2 if generation_config.num_beams > 1 else 1 + + if generation_config.repetition_penalty is not None and generation_config.repetition_penalty != 1.0: + warpers_to_add.append( + RepetitionPenaltyLogitsProcessorWithRange( + penalty=generation_config.repetition_penalty, + _range=generation_config.repetition_penalty_range + ) + ) + + if generation_config.presence_penalty is not None and generation_config.presence_penalty != 0.0: + warpers_to_add.append( + PresencePenaltyLogitsProcessor( + presence_penalty=generation_config.presence_penalty, + _range=generation_config.repetition_penalty_range + ) + ) + + if generation_config.frequency_penalty is not None and generation_config.frequency_penalty != 0.0: + warpers_to_add.append( + FrequencyPenaltyLogitsProcessor( + frequency_penalty=generation_config.frequency_penalty, + _range=generation_config.repetition_penalty_range + ) + ) + + if generation_config.dry_multiplier is not None and generation_config.dry_multiplier > 0.0: + dry_sequence_breakers = generation_config.dry_sequence_breakers + + # Support both JSON array notation and comma-separated strings. + if not dry_sequence_breakers.startswith("["): + dry_sequence_breakers = "[" + dry_sequence_breakers + "]" + + sequence_breaker_strings = json.loads(dry_sequence_breakers) + # Prefix with 'a' to get the correct encoding of the token at the end of a text. + sequence_breakers = {shared.tokenizer.encode(f'a{s}')[-1] for s in sequence_breaker_strings} + + warpers.append( + DRYLogitsProcessor( + multiplier=generation_config.dry_multiplier, + base=generation_config.dry_base, + allowed_length=generation_config.dry_allowed_length, + sequence_breakers=sequence_breakers, + _range=generation_config.repetition_penalty_range, + ) + ) + if generation_config.tfs is not None and 0.0 <= generation_config.tfs < 1.0: warpers_to_add.append( TailFreeLogitsWarper( @@ -454,7 +533,12 @@ def get_logits_warper_patch(self, generation_config, **kwargs): 'TopALogitsWarper': 'top_a', 'TopKLogitsWarper': 'top_k', 'TopPLogitsWarper': 'top_p', - 'TypicalLogitsWarper': 'typical_p' + 'TypicalLogitsWarper': 'typical_p', + 'RepetitionPenaltyLogitsProcessorWithRange': 'repetition_penalty', + 'PresencePenaltyLogitsProcessor': 'presence_penalty', + 'FrequencyPenaltyLogitsProcessor': 'frequency_penalty', + 'EncoderRepetitionPenaltyLogitsProcessor': 'encoder_repetition_penalty', + 'NoRepeatNGramLogitsProcessor': 'no_repeat_ngram', } def custom_sort_key(obj): @@ -482,49 +566,6 @@ def custom_sort_key(obj): return warpers -def get_logits_processor_patch(self, **kwargs): - generation_config = kwargs['generation_config'] - - do_rep_pen_hijack = (generation_config.repetition_penalty > 1) or (generation_config.presence_penalty != 0) or (generation_config.frequency_penalty != 0) - if do_rep_pen_hijack: - generation_config.repetition_penalty = 1.1 # Set to value > 1 to ensure RepetitionPenaltyLogitsProcessor is created - - result = self._get_logits_processor_old(**kwargs) - - if do_rep_pen_hijack: - for i in range(len(result)): - if result[i].__class__.__name__ == 'RepetitionPenaltyLogitsProcessor': - result[i] = RepetitionPenaltyLogitsProcessorWithRange( - generation_config.repetition_penalty, - generation_config.presence_penalty, - generation_config.frequency_penalty, - generation_config.repetition_penalty_range - ) - - if generation_config.dry_multiplier is not None and generation_config.dry_multiplier > 0.0: - dry_sequence_breakers = generation_config.dry_sequence_breakers - - # Support both JSON array notation and comma-separated strings. - if not dry_sequence_breakers.startswith("["): - dry_sequence_breakers = "[" + dry_sequence_breakers + "]" - - sequence_breaker_strings = json.loads(dry_sequence_breakers) - # Prefix with 'a' to get the correct encoding of the token at the end of a text. - sequence_breakers = {shared.tokenizer.encode(f'a{s}')[-1] for s in sequence_breaker_strings} - - result.append( - DRYLogitsProcessor( - multiplier=generation_config.dry_multiplier, - base=generation_config.dry_base, - allowed_length=generation_config.dry_allowed_length, - sequence_breakers=sequence_breakers, - _range=generation_config.repetition_penalty_range, - ) - ) - - return result - - def generation_config_init_patch(self, **kwargs): self.__init___old(**kwargs) self.min_p = kwargs.pop("min_p", 0.0) @@ -547,13 +588,10 @@ def generation_config_init_patch(self, **kwargs): self.dry_allowed_length = kwargs.pop("dry_allowed_length", 2) self.dry_sequence_breakers = kwargs.pop("dry_sequence_breakers", '"\\n", ":", "\\"", "*"') self.temperature_last = kwargs.pop("temperature_last", False) - self.sampler_priority = kwargs.pop("sampler_priority", ['temperature', 'dynamic_temperature', 'quadratic_sampling', 'top_k', 'top_p', 'typical_p', 'epsilon_cutoff', 'eta_cutoff', 'tfs', 'top_a', 'min_p', 'mirostat']) + self.sampler_priority = kwargs.pop("sampler_priority", ['repetition_penalty', 'presence_penalty', 'frequency_penalty', 'temperature', 'dynamic_temperature', 'quadratic_sampling', 'top_k', 'top_p', 'typical_p', 'epsilon_cutoff', 'eta_cutoff', 'tfs', 'top_a', 'min_p', 'mirostat', 'encoder_repetition_penalty', 'no_repeat_ngram']) def hijack_samplers(): - transformers.GenerationMixin._get_logits_warper_old = transformers.GenerationMixin._get_logits_warper - transformers.GenerationMixin._get_logits_warper = get_logits_warper_patch - transformers.GenerationMixin._get_logits_processor_old = transformers.GenerationMixin._get_logits_processor transformers.GenerationMixin._get_logits_processor = get_logits_processor_patch From c5f048e912ffbb1c7fd4b72609a67bec8ad07b5c Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Fri, 27 Sep 2024 15:04:08 -0700 Subject: [PATCH 0034/1701] Bump ExLlamaV2 to 0.2.2 --- requirements.txt | 10 +++++----- requirements_amd.txt | 6 +++--- requirements_amd_noavx2.txt | 6 +++--- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_noavx2.txt | 10 +++++----- 6 files changed, 18 insertions(+), 18 deletions(-) diff --git a/requirements.txt b/requirements.txt index 913c246398..f0360d5398 100644 --- a/requirements.txt +++ b/requirements.txt @@ -55,11 +55,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.2.2cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.2.2cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index b9d74b69a6..d6bf29e619 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -42,9 +42,9 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp # AMD wheels https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.90+rocm5.6.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.90+rocm5.6.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index ccacab29df..5ce468c34e 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -40,9 +40,9 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index dc1757e264..12786adebb 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -38,4 +38,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index a913a63c25..901c9f1033 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -40,4 +40,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 458c1410a1..87e1bd7fa9 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -55,11 +55,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.0/exllamav2-0.2.0-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.2.2cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.2.2cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" From 78b87054004a18f788d4a52d72a8ca9358aca0e9 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Fri, 27 Sep 2024 15:06:31 -0700 Subject: [PATCH 0035/1701] Bump llama-cpp-python to 0.3.0 (except for AMD) --- requirements.txt | 24 ++++++++++++------------ requirements_amd.txt | 8 ++++---- requirements_amd_noavx2.txt | 8 ++++---- requirements_apple_intel.txt | 8 ++++---- requirements_apple_silicon.txt | 12 ++++++------ requirements_cpu_only.txt | 8 ++++---- requirements_cpu_only_noavx2.txt | 8 ++++---- requirements_noavx2.txt | 24 ++++++++++++------------ 8 files changed, 50 insertions(+), 50 deletions(-) diff --git a/requirements.txt b/requirements.txt index f0360d5398..1041025cf3 100644 --- a/requirements.txt +++ b/requirements.txt @@ -37,22 +37,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index d6bf29e619..b224ce1488 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -34,10 +34,10 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.90+rocm5.6.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 5ce468c34e..3cdb8a7a58 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -34,10 +34,10 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 12786adebb..fd91464dd8 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -34,8 +34,8 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 901c9f1033..8a50c8b422 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -34,10 +34,10 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.90-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index b0dbc91a86..4e89b6ada3 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -34,7 +34,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index 828d0c974e..2c00ccf5d2 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -34,7 +34,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 87e1bd7fa9..69d880acb3 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -37,22 +37,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.90+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.90+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.90+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" From 5c918c5b2d870f41a5215cfd5c45c652ad6d9b3a Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Fri, 27 Sep 2024 15:40:48 -0700 Subject: [PATCH 0036/1701] Make it possible to sort DRY --- modules/presets.py | 2 +- modules/sampler_hijack.py | 3 ++- 2 files changed, 3 insertions(+), 2 deletions(-) diff --git a/modules/presets.py b/modules/presets.py index 14c5a7776c..06e2be7089 100644 --- a/modules/presets.py +++ b/modules/presets.py @@ -44,7 +44,7 @@ def default_preset(): 'dry_base': 1.75, 'dry_allowed_length': 2, 'dry_sequence_breakers': '"\\n", ":", "\\"", "*"', - 'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nmirostat\nencoder_repetition_penalty\nno_repeat_ngram' + 'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ndry\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nmirostat\nencoder_repetition_penalty\nno_repeat_ngram' } diff --git a/modules/sampler_hijack.py b/modules/sampler_hijack.py index 9e865bb341..65503c12e1 100644 --- a/modules/sampler_hijack.py +++ b/modules/sampler_hijack.py @@ -537,6 +537,7 @@ def get_logits_processor_patch(self, **kwargs): 'RepetitionPenaltyLogitsProcessorWithRange': 'repetition_penalty', 'PresencePenaltyLogitsProcessor': 'presence_penalty', 'FrequencyPenaltyLogitsProcessor': 'frequency_penalty', + 'DRYLogitsProcessor': 'dry', 'EncoderRepetitionPenaltyLogitsProcessor': 'encoder_repetition_penalty', 'NoRepeatNGramLogitsProcessor': 'no_repeat_ngram', } @@ -588,7 +589,7 @@ def generation_config_init_patch(self, **kwargs): self.dry_allowed_length = kwargs.pop("dry_allowed_length", 2) self.dry_sequence_breakers = kwargs.pop("dry_sequence_breakers", '"\\n", ":", "\\"", "*"') self.temperature_last = kwargs.pop("temperature_last", False) - self.sampler_priority = kwargs.pop("sampler_priority", ['repetition_penalty', 'presence_penalty', 'frequency_penalty', 'temperature', 'dynamic_temperature', 'quadratic_sampling', 'top_k', 'top_p', 'typical_p', 'epsilon_cutoff', 'eta_cutoff', 'tfs', 'top_a', 'min_p', 'mirostat', 'encoder_repetition_penalty', 'no_repeat_ngram']) + self.sampler_priority = kwargs.pop("sampler_priority", ['repetition_penalty', 'presence_penalty', 'frequency_penalty', 'dry', 'temperature', 'dynamic_temperature', 'quadratic_sampling', 'top_k', 'top_p', 'typical_p', 'epsilon_cutoff', 'eta_cutoff', 'tfs', 'top_a', 'min_p', 'mirostat', 'encoder_repetition_penalty', 'no_repeat_ngram']) def hijack_samplers(): From 626b0a043702c9586478d9969184718d24ebf935 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Thireus=20=E2=98=A0?= Date: Fri, 27 Sep 2024 23:47:04 +0100 Subject: [PATCH 0037/1701] Force /bin/bash shell for conda (#6386) --- one_click.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/one_click.py b/one_click.py index 0a0412ba07..d35ecc8923 100644 --- a/one_click.py +++ b/one_click.py @@ -190,7 +190,7 @@ def run_cmd(cmd, assert_success=False, environment=False, capture_output=False, cmd = f'. "{conda_sh_path}" && conda activate "{conda_env_path}" && {cmd}' # Run shell commands - result = subprocess.run(cmd, shell=True, capture_output=capture_output, env=env) + result = subprocess.run(cmd, shell=True, executable='/bin/bash', capture_output=capture_output, env=env) # Assert the command ran successfully if assert_success and result.returncode != 0: From 3492e33fd50bcc78a81850fdcc995e097a79b8ef Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Fri, 27 Sep 2024 16:59:30 -0700 Subject: [PATCH 0038/1701] Bump bitsandbytes to 0.44 --- requirements.txt | 2 +- requirements_noavx2.txt | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/requirements.txt b/requirements.txt index 1041025cf3..54bd414a57 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,7 +1,7 @@ accelerate==0.33.* aqlm[gpu,cpu]==1.1.6; platform_system == "Linux" auto-gptq==0.7.1 -bitsandbytes==0.43.* +bitsandbytes==0.44.* colorama datasets einops diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 69d880acb3..39bfb0bdf3 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -1,7 +1,7 @@ accelerate==0.33.* aqlm[gpu,cpu]==1.1.6; platform_system == "Linux" auto-gptq==0.7.1 -bitsandbytes==0.43.* +bitsandbytes==0.44.* colorama datasets einops From 301375834e55fcda6fda96332cad187aab2c6f77 Mon Sep 17 00:00:00 2001 From: Philipp Emanuel Weidmann Date: Sat, 28 Sep 2024 07:20:12 +0530 Subject: [PATCH 0039/1701] =?UTF-8?q?Exclude=20Top=20Choices=20(XTC):=20A?= =?UTF-8?q?=20sampler=20that=20boosts=20creativity,=20breaks=20writing=20c?= =?UTF-8?q?lich=C3=A9s,=20and=20inhibits=20non-verbatim=20repetition=20(#6?= =?UTF-8?q?335)?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- extensions/openai/typing.py | 2 ++ modules/loaders.py | 6 ++++ modules/presets.py | 4 ++- modules/sampler_hijack.py | 61 ++++++++++++++++++++++++++++++++++++- modules/text_generation.py | 2 +- modules/ui.py | 2 ++ modules/ui_parameters.py | 4 +++ 7 files changed, 78 insertions(+), 3 deletions(-) diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py index 4015f6a1ce..f63c1f3911 100644 --- a/extensions/openai/typing.py +++ b/extensions/openai/typing.py @@ -37,6 +37,8 @@ class GenerationOptions(BaseModel): dry_base: float = 1.75 dry_allowed_length: int = 2 dry_sequence_breakers: str = '"\\n", ":", "\\"", "*"' + xtc_threshold: float = 0.1 + xtc_probability: float = 0 truncation_length: int = 0 max_tokens_second: int = 0 prompt_lookup_num_tokens: int = 0 diff --git a/modules/loaders.py b/modules/loaders.py index 549de5fb02..16a3106e94 100644 --- a/modules/loaders.py +++ b/modules/loaders.py @@ -168,6 +168,8 @@ def transformers_samplers(): 'dry_base', 'dry_allowed_length', 'dry_sequence_breakers', + 'xtc_threshold', + 'xtc_probability', 'seed', 'do_sample', 'penalty_alpha', @@ -242,6 +244,8 @@ def transformers_samplers(): 'dry_base', 'dry_allowed_length', 'dry_sequence_breakers', + 'xtc_threshold', + 'xtc_probability', 'seed', 'do_sample', 'mirostat_mode', @@ -304,6 +308,8 @@ def transformers_samplers(): 'dry_base', 'dry_allowed_length', 'dry_sequence_breakers', + 'xtc_threshold', + 'xtc_probability', 'seed', 'do_sample', 'mirostat_mode', diff --git a/modules/presets.py b/modules/presets.py index 06e2be7089..c8118fb3b7 100644 --- a/modules/presets.py +++ b/modules/presets.py @@ -44,7 +44,9 @@ def default_preset(): 'dry_base': 1.75, 'dry_allowed_length': 2, 'dry_sequence_breakers': '"\\n", ":", "\\"", "*"', - 'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ndry\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nmirostat\nencoder_repetition_penalty\nno_repeat_ngram' + 'xtc_threshold': 0.1, + 'xtc_probability': 0, + 'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ndry\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nmirostat\nxtc\nencoder_repetition_penalty\nno_repeat_ngram' } diff --git a/modules/sampler_hijack.py b/modules/sampler_hijack.py index 65503c12e1..6d92978e6e 100644 --- a/modules/sampler_hijack.py +++ b/modules/sampler_hijack.py @@ -1,6 +1,7 @@ import json import math import pprint +import random import torch import transformers @@ -191,6 +192,53 @@ def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> to return scores +# Exclude Top Choices (XTC) +class XTCLogitsWarper(LogitsWarper): + def __init__(self, threshold: float, probability: float, filter_value: float = -float("Inf")): + self.threshold = threshold + self.probability = probability + self.filter_value = filter_value + self.special_token_ids = [ + shared.tokenizer.encode("\n")[-1], + ] + + if shared.tokenizer.eos_token_id is not None: + self.special_token_ids.append(shared.tokenizer.eos_token_id) + + def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> torch.FloatTensor: + # `random` returns values in the half-open range [0, 1), so setting `probability` + # to 0 means the sampler never takes action, while setting it to 1 means the sampler + # always takes action. + # + # Note that while XTC is most intuitively described as "if multiple tokens meet + # the threshold, then with probability...", reversing the two conditions is logically + # equivalent, and improves performance because processing can immediately be stopped + # if the random check fails. + if random.random() >= self.probability: + return scores + + sorted_logits, sorted_indices = torch.sort(scores, descending=True) + probs = sorted_logits.softmax(dim=-1) + + sorted_indices_to_remove = torch.full_like(probs, False, dtype=torch.bool) + + # This operation sets exactly those indices to `True` for which the next index has + # probability above the threshold. Since `probs` is sorted, those are the indices + # of all tokens that meet the threshold, *except* the least probable one. + sorted_indices_to_remove[..., :-1] = probs[..., 1:] >= self.threshold + + # Convert sorted_indices_to_remove to the original indices + indices_to_remove = sorted_indices_to_remove.scatter(1, sorted_indices, sorted_indices_to_remove) + + # If newline or EOS tokens would be removed, return the original scores + if indices_to_remove[:, self.special_token_ids].any(): + return scores + + # Otherwise, remove tokens with the mask + scores = scores.masked_fill(indices_to_remove, self.filter_value) + return scores + + class DRYLogitsProcessor(LogitsProcessor): def __init__(self, multiplier: float, base: float, allowed_length: int, sequence_breakers: set[int], _range: int): self.multiplier = multiplier @@ -474,6 +522,14 @@ def get_logits_processor_patch(self, **kwargs): ) ) + if generation_config.xtc_probability is not None and generation_config.xtc_probability > 0: + warpers_to_add.append( + XTCLogitsWarper( + threshold=generation_config.xtc_threshold, + probability=generation_config.xtc_probability, + ) + ) + if generation_config.dynamic_temperature: warpers_to_add.append( DynamicTemperatureLogitsWarper( @@ -534,6 +590,7 @@ def get_logits_processor_patch(self, **kwargs): 'TopKLogitsWarper': 'top_k', 'TopPLogitsWarper': 'top_p', 'TypicalLogitsWarper': 'typical_p', + 'XTCLogitsWarper': 'xtc', 'RepetitionPenaltyLogitsProcessorWithRange': 'repetition_penalty', 'PresencePenaltyLogitsProcessor': 'presence_penalty', 'FrequencyPenaltyLogitsProcessor': 'frequency_penalty', @@ -588,8 +645,10 @@ def generation_config_init_patch(self, **kwargs): self.dry_base = kwargs.pop("dry_base", 1.75) self.dry_allowed_length = kwargs.pop("dry_allowed_length", 2) self.dry_sequence_breakers = kwargs.pop("dry_sequence_breakers", '"\\n", ":", "\\"", "*"') + self.xtc_threshold = kwargs.pop("xtc_threshold", 0.1) + self.xtc_probability = kwargs.pop("xtc_probability", 0) self.temperature_last = kwargs.pop("temperature_last", False) - self.sampler_priority = kwargs.pop("sampler_priority", ['repetition_penalty', 'presence_penalty', 'frequency_penalty', 'dry', 'temperature', 'dynamic_temperature', 'quadratic_sampling', 'top_k', 'top_p', 'typical_p', 'epsilon_cutoff', 'eta_cutoff', 'tfs', 'top_a', 'min_p', 'mirostat', 'encoder_repetition_penalty', 'no_repeat_ngram']) + self.sampler_priority = kwargs.pop("sampler_priority", ['repetition_penalty', 'presence_penalty', 'frequency_penalty', 'dry', 'temperature', 'dynamic_temperature', 'quadratic_sampling', 'top_k', 'top_p', 'typical_p', 'epsilon_cutoff', 'eta_cutoff', 'tfs', 'top_a', 'min_p', 'mirostat', 'xtc', 'encoder_repetition_penalty', 'no_repeat_ngram']) def hijack_samplers(): diff --git a/modules/text_generation.py b/modules/text_generation.py index e7a2b43f01..654eff8940 100644 --- a/modules/text_generation.py +++ b/modules/text_generation.py @@ -289,7 +289,7 @@ def get_reply_from_output_ids(output_ids, state=None, starting_from=0): def generate_reply_HF(question, original_question, seed, state, stopping_strings=None, is_chat=False): generate_params = {} - for k in ['max_new_tokens', 'temperature', 'temperature_last', 'dynamic_temperature', 'dynatemp_low', 'dynatemp_high', 'dynatemp_exponent', 'smoothing_factor', 'smoothing_curve', 'top_p', 'min_p', 'top_k', 'repetition_penalty', 'presence_penalty', 'frequency_penalty', 'repetition_penalty_range', 'typical_p', 'tfs', 'top_a', 'guidance_scale', 'penalty_alpha', 'mirostat_mode', 'mirostat_tau', 'mirostat_eta', 'do_sample', 'encoder_repetition_penalty', 'no_repeat_ngram_size', 'dry_multiplier', 'dry_base', 'dry_allowed_length', 'dry_sequence_breakers']: + for k in ['max_new_tokens', 'temperature', 'temperature_last', 'dynamic_temperature', 'dynatemp_low', 'dynatemp_high', 'dynatemp_exponent', 'smoothing_factor', 'smoothing_curve', 'top_p', 'min_p', 'top_k', 'repetition_penalty', 'presence_penalty', 'frequency_penalty', 'repetition_penalty_range', 'typical_p', 'tfs', 'top_a', 'guidance_scale', 'penalty_alpha', 'mirostat_mode', 'mirostat_tau', 'mirostat_eta', 'do_sample', 'encoder_repetition_penalty', 'no_repeat_ngram_size', 'dry_multiplier', 'dry_base', 'dry_allowed_length', 'dry_sequence_breakers', 'xtc_threshold', 'xtc_probability']: if k in state: generate_params[k] = state[k] diff --git a/modules/ui.py b/modules/ui.py index 47f92cf0f9..76b1c009c4 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -158,6 +158,8 @@ def list_interface_input_elements(): 'dry_base', 'dry_allowed_length', 'dry_sequence_breakers', + 'xtc_threshold', + 'xtc_probability', 'do_sample', 'penalty_alpha', 'mirostat_mode', diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py index eff62c20e5..a2665e0d21 100644 --- a/modules/ui_parameters.py +++ b/modules/ui_parameters.py @@ -45,6 +45,10 @@ def create_ui(default_preset): shared.gradio['dry_base'] = gr.Slider(1, 4, value=generate_params['dry_base'], step=0.01, label='dry_base', info='Controls how fast the penalty grows with increasing sequence length.') shared.gradio['dry_sequence_breakers'] = gr.Textbox(value=generate_params['dry_sequence_breakers'], label='dry_sequence_breakers', info='Tokens across which sequence matching is not continued. Specified as a comma-separated list of quoted strings.') + with gr.Blocks(): + shared.gradio['xtc_threshold'] = gr.Slider(0, 0.5, value=generate_params['xtc_threshold'], step=0.01, label='xtc_threshold', info='If 2 or more tokens have probability above this threshold, consider removing all but the last one.') + shared.gradio['xtc_probability'] = gr.Slider(0, 1, value=generate_params['xtc_probability'], step=0.01, label='xtc_probability', info='Probability that the removal will actually happen. 0 disables the sampler. 1 makes it always happen.') + gr.Markdown("[Learn more](https://github.com/oobabooga/text-generation-webui/wiki/03-%E2%80%90-Parameters-Tab)") with gr.Column(): From 46996f6519d483841a49c8a857104a6bf6e2ac7a Mon Sep 17 00:00:00 2001 From: RandoInternetPreson Date: Fri, 27 Sep 2024 23:26:03 -0400 Subject: [PATCH 0040/1701] ExllamaV2 tensor parallelism to increase multi gpu inference speeds (#6356) --- modules/exllamav2.py | 26 ++++++++++++++++++-------- modules/exllamav2_hf.py | 28 +++++++++++++++++++--------- modules/shared.py | 1 + 3 files changed, 38 insertions(+), 17 deletions(-) diff --git a/modules/exllamav2.py b/modules/exllamav2.py index a770e34257..42b9ade145 100644 --- a/modules/exllamav2.py +++ b/modules/exllamav2.py @@ -7,6 +7,7 @@ ExLlamaV2Cache, ExLlamaV2Cache_8bit, ExLlamaV2Cache_Q4, + ExLlamaV2Cache_TP, ExLlamaV2Config, ExLlamaV2Tokenizer ) @@ -54,21 +55,30 @@ def from_pretrained(self, path_to_model): model = ExLlamaV2(config) - if not shared.args.autosplit: - split = None - if shared.args.gpu_split: - split = [float(alloc) for alloc in shared.args.gpu_split.split(",")] + split = None + if shared.args.gpu_split: + split = [float(alloc) for alloc in shared.args.gpu_split.split(",")] + if shared.args.enable_tp: + model.load_tp(split) + elif not shared.args.autosplit: model.load(split) + # Determine the correct cache type if shared.args.cache_8bit: - cache = ExLlamaV2Cache_8bit(model, lazy=shared.args.autosplit) + cache_type = ExLlamaV2Cache_8bit elif shared.args.cache_4bit: - cache = ExLlamaV2Cache_Q4(model, lazy=shared.args.autosplit) + cache_type = ExLlamaV2Cache_Q4 else: - cache = ExLlamaV2Cache(model, lazy=shared.args.autosplit) + cache_type = ExLlamaV2Cache - if shared.args.autosplit: + # Use TP if specified + if shared.args.enable_tp: + cache = ExLlamaV2Cache_TP(model, base=cache_type) + else: + cache = cache_type(model, lazy=shared.args.autosplit) + + if shared.args.autosplit and not shared.args.enable_tp: model.load_autosplit(cache) tokenizer = ExLlamaV2Tokenizer(config) diff --git a/modules/exllamav2_hf.py b/modules/exllamav2_hf.py index 53143d9a92..febb2c6495 100644 --- a/modules/exllamav2_hf.py +++ b/modules/exllamav2_hf.py @@ -9,6 +9,7 @@ ExLlamaV2Cache, ExLlamaV2Cache_8bit, ExLlamaV2Cache_Q4, + ExLlamaV2Cache_TP, ExLlamaV2Config ) from torch.nn import CrossEntropyLoss @@ -42,21 +43,30 @@ def __init__(self, config: ExLlamaV2Config): self.ex_model = ExLlamaV2(config) - if not shared.args.autosplit: - split = None - if shared.args.gpu_split: - split = [float(alloc) for alloc in shared.args.gpu_split.split(",")] + split = None + if shared.args.gpu_split: + split = [float(alloc) for alloc in shared.args.gpu_split.split(",")] - self.ex_model.load(split) + if shared.args.enable_tp: + model.load_tp(split) + elif not shared.args.autosplit: + model.load(split) + # Determine the correct cache type if shared.args.cache_8bit: - self.ex_cache = ExLlamaV2Cache_8bit(self.ex_model, lazy=shared.args.autosplit) + cache_type = ExLlamaV2Cache_8bit elif shared.args.cache_4bit: - self.ex_cache = ExLlamaV2Cache_Q4(self.ex_model, lazy=shared.args.autosplit) + cache_type = ExLlamaV2Cache_Q4 else: - self.ex_cache = ExLlamaV2Cache(self.ex_model, lazy=shared.args.autosplit) + cache_type = ExLlamaV2Cache - if shared.args.autosplit: + # Use TP if specified + if shared.args.enable_tp: + self.ex_cache = ExLlamaV2Cache_TP(self.ex_model, base=cache_type) + else: + self.ex_cache = cache_type(self.ex_model, lazy=shared.args.autosplit) + + if shared.args.autosplit and not shared.args.enable_tp: self.ex_model.load_autosplit(self.ex_cache) self.past_seq = None diff --git a/modules/shared.py b/modules/shared.py index 43533a1480..894ed6fe56 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -146,6 +146,7 @@ group.add_argument('--cache_8bit', action='store_true', help='Use 8-bit cache to save VRAM.') group.add_argument('--cache_4bit', action='store_true', help='Use Q4 cache to save VRAM.') group.add_argument('--num_experts_per_token', type=int, default=2, help='Number of experts to use for generation. Applies to MoE models like Mixtral.') +group.add_argument('--enable_tp', action='store_true', help='Enable Tensor Parallelism (TP) in ExLlamaV2.') # AutoGPTQ group = parser.add_argument_group('AutoGPTQ') From 7276dca933626eef587ac0adf53ab8ea06c2e9ac Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Fri, 27 Sep 2024 20:26:36 -0700 Subject: [PATCH 0041/1701] Fix a typo --- modules/exllamav2_hf.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/exllamav2_hf.py b/modules/exllamav2_hf.py index febb2c6495..96a89429ff 100644 --- a/modules/exllamav2_hf.py +++ b/modules/exllamav2_hf.py @@ -48,9 +48,9 @@ def __init__(self, config: ExLlamaV2Config): split = [float(alloc) for alloc in shared.args.gpu_split.split(",")] if shared.args.enable_tp: - model.load_tp(split) + self.ex_model.load_tp(split) elif not shared.args.autosplit: - model.load(split) + self.ex_model.load(split) # Determine the correct cache type if shared.args.cache_8bit: From ca5a2dba72817def91b616e9571d8490b9b64a13 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sat, 28 Sep 2024 09:39:37 -0700 Subject: [PATCH 0042/1701] Bump rocm to 6.1.2 --- README.md | 2 +- one_click.py | 6 +++--- requirements_amd.txt | 8 ++++---- requirements_amd_noavx2.txt | 4 ++-- 4 files changed, 10 insertions(+), 10 deletions(-) diff --git a/README.md b/README.md index 22ceeeb8ea..4449639da7 100644 --- a/README.md +++ b/README.md @@ -82,7 +82,7 @@ conda activate textgen |--------|---------|---------| | Linux/WSL | NVIDIA | `pip3 install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2 --index-url https://download.pytorch.org/whl/cu121` | | Linux/WSL | CPU only | `pip3 install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2 --index-url https://download.pytorch.org/whl/cpu` | -| Linux | AMD | `pip3 install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2 --index-url https://download.pytorch.org/whl/rocm5.6` | +| Linux | AMD | `pip3 install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2 --index-url https://download.pytorch.org/whl/rocm6.1` | | MacOS + MPS | Any | `pip3 install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2` | | Windows | NVIDIA | `pip3 install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2 --index-url https://download.pytorch.org/whl/cu121` | | Windows | CPU only | `pip3 install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2` | diff --git a/one_click.py b/one_click.py index d35ecc8923..9b813873fc 100644 --- a/one_click.py +++ b/one_click.py @@ -117,7 +117,7 @@ def update_pytorch(): elif is_cuda: install_pytorch += "--index-url https://download.pytorch.org/whl/cu121" elif is_rocm: - install_pytorch += "--index-url https://download.pytorch.org/whl/rocm5.6" + install_pytorch += "--index-url https://download.pytorch.org/whl/rocm6.1" elif is_cpu: install_pytorch += "--index-url https://download.pytorch.org/whl/cpu" elif is_intel: @@ -239,7 +239,7 @@ def install_webui(): "What is your GPU?", { 'A': 'NVIDIA', - 'B': 'AMD (Linux/MacOS only. Requires ROCm SDK 5.6 on Linux)', + 'B': 'AMD (Linux/MacOS only. Requires ROCm SDK 6.1 on Linux)', 'C': 'Apple M Series', 'D': 'Intel Arc (IPEX)', 'N': 'None (I want to run models in CPU mode)' @@ -294,7 +294,7 @@ def install_webui(): else: install_pytorch += "--index-url https://download.pytorch.org/whl/cu121" elif selected_gpu == "AMD": - install_pytorch += "--index-url https://download.pytorch.org/whl/rocm5.6" + install_pytorch += "--index-url https://download.pytorch.org/whl/rocm6.1" elif selected_gpu in ["APPLE", "NONE"]: install_pytorch += "--index-url https://download.pytorch.org/whl/cpu" elif selected_gpu == "INTEL": diff --git a/requirements_amd.txt b/requirements_amd.txt index b224ce1488..9f46ed4f32 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -40,10 +40,10 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.90+rocm5.6.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.90+rocm5.6.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.0+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.0+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 3cdb8a7a58..9c5fb5f265 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -40,8 +40,8 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm5.6.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm5.6.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" From 85994e3ef0f75b408ef5b9a6995902eb600be970 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sat, 28 Sep 2024 09:44:08 -0700 Subject: [PATCH 0043/1701] Bump pytorch to 2.4.1 --- README.md | 14 +++++++------- one_click.py | 6 +++--- requirements.txt | 16 ++++++++-------- requirements_amd.txt | 4 ++-- requirements_amd_noavx2.txt | 4 ++-- requirements_noavx2.txt | 16 ++++++++-------- 6 files changed, 30 insertions(+), 30 deletions(-) diff --git a/README.md b/README.md index 4449639da7..350dd09062 100644 --- a/README.md +++ b/README.md @@ -80,12 +80,12 @@ conda activate textgen | System | GPU | Command | |--------|---------|---------| -| Linux/WSL | NVIDIA | `pip3 install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2 --index-url https://download.pytorch.org/whl/cu121` | -| Linux/WSL | CPU only | `pip3 install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2 --index-url https://download.pytorch.org/whl/cpu` | -| Linux | AMD | `pip3 install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2 --index-url https://download.pytorch.org/whl/rocm6.1` | -| MacOS + MPS | Any | `pip3 install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2` | -| Windows | NVIDIA | `pip3 install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2 --index-url https://download.pytorch.org/whl/cu121` | -| Windows | CPU only | `pip3 install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2` | +| Linux/WSL | NVIDIA | `pip3 install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1 --index-url https://download.pytorch.org/whl/cu121` | +| Linux/WSL | CPU only | `pip3 install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1 --index-url https://download.pytorch.org/whl/cpu` | +| Linux | AMD | `pip3 install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1 --index-url https://download.pytorch.org/whl/rocm6.1` | +| MacOS + MPS | Any | `pip3 install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1` | +| Windows | NVIDIA | `pip3 install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1 --index-url https://download.pytorch.org/whl/cu121` | +| Windows | CPU only | `pip3 install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1` | The up-to-date commands can be found here: https://pytorch.org/get-started/locally/. @@ -150,7 +150,7 @@ Then browse to 1) For Kepler GPUs and older, you will need to install CUDA 11.8 instead of 12: ``` -pip3 install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2 --index-url https://download.pytorch.org/whl/cu118 +pip3 install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1 --index-url https://download.pytorch.org/whl/cu118 conda install -y -c "nvidia/label/cuda-11.8.0" cuda-runtime ``` diff --git a/one_click.py b/one_click.py index 9b813873fc..0f9c703ddd 100644 --- a/one_click.py +++ b/one_click.py @@ -16,9 +16,9 @@ # Define the required PyTorch version -TORCH_VERSION = "2.2.2" -TORCHVISION_VERSION = "0.17.2" -TORCHAUDIO_VERSION = "2.2.2" +TORCH_VERSION = "2.4.1" +TORCHVISION_VERSION = "0.19.1" +TORCHAUDIO_VERSION = "2.4.1" # Environment script_dir = os.getcwd() diff --git a/requirements.txt b/requirements.txt index 54bd414a57..25b4b49d40 100644 --- a/requirements.txt +++ b/requirements.txt @@ -55,15 +55,15 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" -https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.2.2cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.2.2cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.2cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index 9f46ed4f32..cd17a600b9 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -42,8 +42,8 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp # AMD wheels https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.0+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.0+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 9c5fb5f265..dd9bd893f6 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -40,8 +40,8 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 39bfb0bdf3..ec0bbb1d71 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -55,15 +55,15 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.2.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" -https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.2.2cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.2.2cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.2cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" From 1a870b3ea7f0b3337a0c0202fe4b4dddb4e559a0 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sat, 28 Sep 2024 19:38:56 -0700 Subject: [PATCH 0044/1701] Remove AutoAWQ and AutoGPTQ from requirements (no wheels available) --- requirements.txt | 9 --------- requirements_amd.txt | 4 ---- requirements_amd_noavx2.txt | 4 ---- requirements_noavx2.txt | 9 --------- 4 files changed, 26 deletions(-) diff --git a/requirements.txt b/requirements.txt index 25b4b49d40..f42247a634 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,6 +1,5 @@ accelerate==0.33.* aqlm[gpu,cpu]==1.1.6; platform_system == "Linux" -auto-gptq==0.7.1 bitsandbytes==0.44.* colorama datasets @@ -64,11 +63,3 @@ https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements_amd.txt b/requirements_amd.txt index cd17a600b9..88fdcad7f3 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -45,7 +45,3 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/ro https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" -https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index dd9bd893f6..0ad440e3b6 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -43,7 +43,3 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" -https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7+rocm561-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index ec0bbb1d71..61981ca5df 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -1,6 +1,5 @@ accelerate==0.33.* aqlm[gpu,cpu]==1.1.6; platform_system == "Linux" -auto-gptq==0.7.1 bitsandbytes==0.44.* colorama datasets @@ -64,11 +63,3 @@ https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ/releases/download/0.2.6/autoawq-0.2.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/AutoAWQ_kernels/releases/download/0.0.7/autoawq_kernels-0.0.7-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" From 65e58640846d65200364b4a65a7975ce48eb4618 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sat, 28 Sep 2024 20:25:26 -0700 Subject: [PATCH 0045/1701] Update README --- README.md | 22 +++++++++++----------- 1 file changed, 11 insertions(+), 11 deletions(-) diff --git a/README.md b/README.md index 350dd09062..fdd9c46c78 100644 --- a/README.md +++ b/README.md @@ -10,17 +10,17 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github. ## Features -* Multiple backends for text generation in a single UI and API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp) (through [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)), [ExLlamaV2](https://github.com/turboderp/exllamav2), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM) are also supported through the Transformers loader. -* OpenAI-compatible API server with Chat and Completions endpoints – see the [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples). -* Automatic prompt formatting for each model using the Jinja2 template in its metadata. -* Three chat modes: `instruct`, `chat-instruct`, and `chat`, allowing for both instruction-following and casual conversations with characters. `chat-instruct` mode automatically applies the model's template to the chat prompt, ensuring high-quality outputs without manual setup. -* "Past chats" menu to quickly switch between conversations and start new ones. -* Free-form generation in the Default/Notebook tabs without being limited to chat turns. Send formatted chat conversations from the Chat tab to these tabs. -* Multiple sampling parameters and generation options for sophisticated text generation control. -* Easy switching between different models through the UI without restarting, using the "Model" tab. -* Simple LoRA fine-tuning tool to customize models with your data. -* All in one folder. The requirements are installed in a self-contained `installer_files` folder that doesn't interfere with the system's environment. -* Extensions support, including numerous built-in and user-contributed extensions. See [the wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [the extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details. +- Supports multiple text generation backends in one UI/API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp), [ExLlamaV2](https://github.com/turboderp/exllamav2), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). Also supports [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM) through the Transformers loader. +- OpenAI-compatible API with Chat and Completions endpoints – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples). +- Automatic prompt formatting using Jinja2 templates. +- Three chat modes: `instruct`, `chat-instruct`, and `chat`, with auto-prompt templates in `chat-instruct`. +- "Past chats" menu for easy conversation switching. +- Free-form text generation in the Default/Notebook tabs without being limited to chat turns. You can send formatted conversations from the Chat tab to these. +- Multiple sampling parameters and generation options for sophisticated text generation control. +- Switch models easily in the UI without restarting. +- Simple LoRA fine-tuning tool. +- Requirements installed in a self-contained `installer_files` directory that doesn't interfere with the system environment. +- Extension support, with numerous built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details. ## How to install From b92d7fd43e6f39410d4fc486e6168160b449aacc Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sat, 28 Sep 2024 20:30:24 -0700 Subject: [PATCH 0046/1701] Add warnings for when AutoGPTQ, TensorRT-LLM, or HQQ are missing --- modules/models.py | 29 +++++++++++++++++++---------- 1 file changed, 19 insertions(+), 10 deletions(-) diff --git a/modules/models.py b/modules/models.py index b0e2346e18..3e8ae3f227 100644 --- a/modules/models.py +++ b/modules/models.py @@ -70,11 +70,11 @@ def load_model(model_name, loader=None): shared.model_name = model_name load_func_map = { 'Transformers': huggingface_loader, - 'AutoGPTQ': AutoGPTQ_loader, 'llama.cpp': llamacpp_loader, 'llamacpp_HF': llamacpp_HF_loader, 'ExLlamav2': ExLlamav2_loader, 'ExLlamav2_HF': ExLlamav2_HF_loader, + 'AutoGPTQ': AutoGPTQ_loader, 'HQQ': HQQ_loader, 'TensorRT-LLM': TensorRT_LLM_loader, } @@ -302,12 +302,6 @@ def llamacpp_HF_loader(model_name): return model -def AutoGPTQ_loader(model_name): - import modules.AutoGPTQ_loader - - return modules.AutoGPTQ_loader.load_quantized(model_name) - - def ExLlamav2_loader(model_name): from modules.exllamav2 import Exllamav2Model @@ -321,9 +315,21 @@ def ExLlamav2_HF_loader(model_name): return Exllamav2HF.from_pretrained(model_name) +def AutoGPTQ_loader(model_name): + try: + import modules.AutoGPTQ_loader + except ModuleNotFoundError: + raise ModuleNotFoundError("Failed to import 'autogptq'. Please install it manually following the instructions in the AutoGPTQ GitHub repository.") + + return modules.AutoGPTQ_loader.load_quantized(model_name) + + def HQQ_loader(model_name): - from hqq.core.quantize import HQQBackend, HQQLinear - from hqq.models.hf.base import AutoHQQHFModel + try: + from hqq.core.quantize import HQQBackend, HQQLinear + from hqq.models.hf.base import AutoHQQHFModel + except ModuleNotFoundError: + raise ModuleNotFoundError("Failed to import 'hqq'. Please install it manually following the instructions in the HQQ GitHub repository.") logger.info(f"Loading HQQ model with backend: \"{shared.args.hqq_backend}\"") @@ -334,7 +340,10 @@ def HQQ_loader(model_name): def TensorRT_LLM_loader(model_name): - from modules.tensorrt_llm import TensorRTLLMModel + try: + from modules.tensorrt_llm import TensorRTLLMModel + except ModuleNotFoundError: + raise ModuleNotFoundError("Failed to import 'tensorrt_llm'. Please install it manually following the instructions in the TensorRT-LLM GitHub repository.") model = TensorRTLLMModel.from_pretrained(model_name) return model From c61b29b9ce143e778fc717accd70e0bbabede720 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sat, 28 Sep 2024 20:33:17 -0700 Subject: [PATCH 0047/1701] Simplify the warning when flash-attn fails to import --- modules/exllamav2.py | 8 -------- modules/exllamav2_hf.py | 8 -------- 2 files changed, 16 deletions(-) diff --git a/modules/exllamav2.py b/modules/exllamav2.py index 42b9ade145..0498c4882e 100644 --- a/modules/exllamav2.py +++ b/modules/exllamav2.py @@ -19,14 +19,6 @@ try: import flash_attn -except ModuleNotFoundError: - logger.warning( - 'You are running ExLlamaV2 without flash-attention. This will cause the VRAM usage ' - 'to be a lot higher than it could be.\n' - 'Try installing flash-attention following the instructions here: ' - 'https://github.com/Dao-AILab/flash-attention#installation-and-features' - ) - pass except Exception: logger.warning('Failed to load flash-attention due to the following error:\n') traceback.print_exc() diff --git a/modules/exllamav2_hf.py b/modules/exllamav2_hf.py index 96a89429ff..320a8d2467 100644 --- a/modules/exllamav2_hf.py +++ b/modules/exllamav2_hf.py @@ -21,14 +21,6 @@ try: import flash_attn -except ModuleNotFoundError: - logger.warning( - 'You are running ExLlamaV2 without flash-attention. This will cause the VRAM usage ' - 'to be a lot higher than it could be.\n' - 'Try installing flash-attention following the instructions here: ' - 'https://github.com/Dao-AILab/flash-attention#installation-and-features' - ) - pass except Exception: logger.warning('Failed to load flash-attention due to the following error:\n') traceback.print_exc() From 3b99532e0268e7c516e58c1c83931c6843114cc2 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sat, 28 Sep 2024 20:34:59 -0700 Subject: [PATCH 0048/1701] Remove HQQ and AQLM from requirements --- requirements.txt | 2 -- requirements_amd.txt | 1 - requirements_amd_noavx2.txt | 1 - requirements_apple_intel.txt | 1 - requirements_apple_silicon.txt | 1 - requirements_cpu_only.txt | 1 - requirements_cpu_only_noavx2.txt | 1 - requirements_noavx2.txt | 2 -- requirements_nowheels.txt | 1 - 9 files changed, 11 deletions(-) diff --git a/requirements.txt b/requirements.txt index f42247a634..d3531b9cd6 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,12 +1,10 @@ accelerate==0.33.* -aqlm[gpu,cpu]==1.1.6; platform_system == "Linux" bitsandbytes==0.44.* colorama datasets einops fastapi==0.112.4 gradio==4.26.* -hqq==0.1.7.post3 jinja2==3.1.4 lm_eval==0.3.0 markdown diff --git a/requirements_amd.txt b/requirements_amd.txt index 88fdcad7f3..1be85205d3 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -4,7 +4,6 @@ datasets einops fastapi==0.112.4 gradio==4.26.* -hqq==0.1.7.post3 jinja2==3.1.4 lm_eval==0.3.0 markdown diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 0ad440e3b6..5026fa6d46 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -4,7 +4,6 @@ datasets einops fastapi==0.112.4 gradio==4.26.* -hqq==0.1.7.post3 jinja2==3.1.4 lm_eval==0.3.0 markdown diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index fd91464dd8..eee8112d80 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -4,7 +4,6 @@ datasets einops fastapi==0.112.4 gradio==4.26.* -hqq==0.1.7.post3 jinja2==3.1.4 lm_eval==0.3.0 markdown diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 8a50c8b422..fa244a935e 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -4,7 +4,6 @@ datasets einops fastapi==0.112.4 gradio==4.26.* -hqq==0.1.7.post3 jinja2==3.1.4 lm_eval==0.3.0 markdown diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 4e89b6ada3..4c970b9122 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -4,7 +4,6 @@ datasets einops fastapi==0.112.4 gradio==4.26.* -hqq==0.1.7.post3 jinja2==3.1.4 lm_eval==0.3.0 markdown diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index 2c00ccf5d2..4b22cb867f 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -4,7 +4,6 @@ datasets einops fastapi==0.112.4 gradio==4.26.* -hqq==0.1.7.post3 jinja2==3.1.4 lm_eval==0.3.0 markdown diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 61981ca5df..fe6adb46da 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -1,12 +1,10 @@ accelerate==0.33.* -aqlm[gpu,cpu]==1.1.6; platform_system == "Linux" bitsandbytes==0.44.* colorama datasets einops fastapi==0.112.4 gradio==4.26.* -hqq==0.1.7.post3 jinja2==3.1.4 lm_eval==0.3.0 markdown diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index e7cbc29cbe..5c840f56de 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -4,7 +4,6 @@ datasets einops fastapi==0.112.4 gradio==4.26.* -hqq==0.1.7.post3 jinja2==3.1.4 lm_eval==0.3.0 markdown From 3fb02f43f648c7735f7ed8b340c3d36d344aa7b6 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sat, 28 Sep 2024 20:38:43 -0700 Subject: [PATCH 0049/1701] Update README --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index fdd9c46c78..59cfb5fcf3 100644 --- a/README.md +++ b/README.md @@ -10,7 +10,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github. ## Features -- Supports multiple text generation backends in one UI/API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp), [ExLlamaV2](https://github.com/turboderp/exllamav2), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). Also supports [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM) through the Transformers loader. +- Supports multiple text generation backends in one UI/API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp), and [ExLlamaV2](https://github.com/turboderp/exllamav2).[TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM) are also supported but you need to install them manually. - OpenAI-compatible API with Chat and Completions endpoints – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples). - Automatic prompt formatting using Jinja2 templates. - Three chat modes: `instruct`, `chat-instruct`, and `chat`, with auto-prompt templates in `chat-instruct`. From 3f0571b62bb020e19c122d0cd30d5c5df4aa2917 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sat, 28 Sep 2024 20:48:30 -0700 Subject: [PATCH 0050/1701] Update README --- README.md | 20 ++++++++------------ 1 file changed, 8 insertions(+), 12 deletions(-) diff --git a/README.md b/README.md index 59cfb5fcf3..954d7a7da5 100644 --- a/README.md +++ b/README.md @@ -10,33 +10,29 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github. ## Features -- Supports multiple text generation backends in one UI/API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp), and [ExLlamaV2](https://github.com/turboderp/exllamav2).[TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM) are also supported but you need to install them manually. +- Supports multiple text generation backends in one UI/API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp), and [ExLlamaV2](https://github.com/turboderp/exllamav2). [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM) are also supported but you need to install them manually. - OpenAI-compatible API with Chat and Completions endpoints – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples). - Automatic prompt formatting using Jinja2 templates. -- Three chat modes: `instruct`, `chat-instruct`, and `chat`, with auto-prompt templates in `chat-instruct`. -- "Past chats" menu for easy conversation switching. +- Three chat modes: `instruct`, `chat-instruct`, and `chat`, with automatic prompt templates in `chat-instruct`. +- "Past chats" menu to easily switch between conversations. - Free-form text generation in the Default/Notebook tabs without being limited to chat turns. You can send formatted conversations from the Chat tab to these. - Multiple sampling parameters and generation options for sophisticated text generation control. -- Switch models easily in the UI without restarting. +- Switch between different models easily in the UI without restarting. - Simple LoRA fine-tuning tool. - Requirements installed in a self-contained `installer_files` directory that doesn't interfere with the system environment. - Extension support, with numerous built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details. ## How to install -1) Clone or [download](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip) the repository. -2) Run the `start_linux.sh`, `start_windows.bat`, `start_macos.sh`, or `start_wsl.bat` script depending on your OS. +1) Clone or [download the repository](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip). +2) Run the script that matches your OS: `start_linux.sh`, `start_windows.bat`, `start_macos.sh`, or `start_wsl.bat`. 3) Select your GPU vendor when asked. 4) Once the installation ends, browse to `http://localhost:7860`. 5) Have fun! -To restart the web UI in the future, run the `start_` script again. +To restart the web UI later, just run the same `start_` script. If you need to reinstall, delete the `installer_files` folder created during setup and run the script again. -This script creates an `installer_files` folder where it sets up the project's requirements. If you need to reinstall the requirements, just delete that folder and start the web UI again. - -The script accepts command-line flags, such as `./start_linux.sh --help`. Alternatively, you can edit the `CMD_FLAGS.txt` file with a text editor and add your flags there, such as `--api` in case you need to use the API. - -To get updates in the future, run `update_wizard_linux.sh`, `update_wizard_windows.bat`, `update_wizard_macos.sh`, or `update_wizard_wsl.bat`. +You can use command-line flags, like `./start_linux.sh --help`, or add them to `CMD_FLAGS.txt` (such as `--api` to enable API use). To update the project, run `update_wizard_linux.sh`, `update_wizard_windows.bat`, `update_wizard_macos.sh`, or `update_wizard_wsl.bat`.
From 57160cd6fa37c8367df3c674256af970b31e0add Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sat, 28 Sep 2024 20:50:41 -0700 Subject: [PATCH 0051/1701] Update README --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 954d7a7da5..fb22d09945 100644 --- a/README.md +++ b/README.md @@ -14,7 +14,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github. - OpenAI-compatible API with Chat and Completions endpoints – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples). - Automatic prompt formatting using Jinja2 templates. - Three chat modes: `instruct`, `chat-instruct`, and `chat`, with automatic prompt templates in `chat-instruct`. -- "Past chats" menu to easily switch between conversations. +- "Past chats" menu to quickly switch between conversations. - Free-form text generation in the Default/Notebook tabs without being limited to chat turns. You can send formatted conversations from the Chat tab to these. - Multiple sampling parameters and generation options for sophisticated text generation control. - Switch between different models easily in the UI without restarting. From 055f3f5632f99f99409fd85d4e06c4c70acf8b33 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sat, 28 Sep 2024 20:55:26 -0700 Subject: [PATCH 0052/1701] Fix after #6386 (thanks @Touch-Night) --- one_click.py | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/one_click.py b/one_click.py index 0f9c703ddd..9bad25d1be 100644 --- a/one_click.py +++ b/one_click.py @@ -189,8 +189,11 @@ def run_cmd(cmd, assert_success=False, environment=False, capture_output=False, conda_sh_path = os.path.join(script_dir, "installer_files", "conda", "etc", "profile.d", "conda.sh") cmd = f'. "{conda_sh_path}" && conda activate "{conda_env_path}" && {cmd}' + # Set executable to None for Windows, /bin/bash for everything else + executable = None if is_windows() else '/bin/bash' + # Run shell commands - result = subprocess.run(cmd, shell=True, executable='/bin/bash', capture_output=capture_output, env=env) + result = subprocess.run(cmd, shell=True, capture_output=capture_output, env=env, executable=executable) # Assert the command ran successfully if assert_success and result.returncode != 0: From 0f90a1b50fde3bc102bef1a034942ebb4b155845 Mon Sep 17 00:00:00 2001 From: Manuel Schmid <9307310+mashb1t@users.noreply.github.com> Date: Sun, 29 Sep 2024 06:08:55 +0200 Subject: [PATCH 0053/1701] Do not set value for histories in chat when --multi-user is used (#6317) --- modules/chat.py | 28 ++++++++++++++++++++++++---- 1 file changed, 24 insertions(+), 4 deletions(-) diff --git a/modules/chat.py b/modules/chat.py index 00c4ffa956..b81cfea6ee 100644 --- a/modules/chat.py +++ b/modules/chat.py @@ -1059,7 +1059,12 @@ def handle_start_new_chat_click(state): convert_to_markdown.cache_clear() - return [history, html, gr.update(choices=histories, value=histories[0][1])] + if len(histories) > 0: + past_chats_update = gr.update(choices=histories, value=histories[0][1]) + else: + past_chats_update = gr.update(choices=histories) + + return [history, html, past_chats_update] def handle_delete_chat_confirm_click(state): @@ -1110,10 +1115,15 @@ def handle_upload_chat_history(load_chat_history, state): convert_to_markdown.cache_clear() + if len(histories) > 0: + past_chats_update = gr.update(choices=histories, value=histories[0][1]) + else: + past_chats_update = gr.update(choices=histories) + return [ history, html, - gr.update(choices=histories, value=histories[0][1]) + past_chats_update ] @@ -1132,6 +1142,11 @@ def handle_character_menu_change(state): convert_to_markdown.cache_clear() + if len(histories) > 0: + past_chats_update = gr.update(choices=histories, value=histories[0][1]) + else: + past_chats_update = gr.update(choices=histories) + return [ history, html, @@ -1140,7 +1155,7 @@ def handle_character_menu_change(state): picture, greeting, context, - gr.update(choices=histories, value=histories[0][1]), + past_chats_update, ] @@ -1151,12 +1166,17 @@ def handle_mode_change(state): convert_to_markdown.cache_clear() + if len(histories) > 0: + past_chats_update = gr.update(choices=histories, value=histories[0][1]) + else: + past_chats_update = gr.update(choices=histories) + return [ history, html, gr.update(visible=state['mode'] != 'instruct'), gr.update(visible=state['mode'] == 'chat-instruct'), - gr.update(choices=histories, value=histories[0][1]) + past_chats_update ] From e4b0467f9f4b054723528c93c4cda56a266f8687 Mon Sep 17 00:00:00 2001 From: Hanusz Leszek Date: Sun, 29 Sep 2024 06:14:19 +0200 Subject: [PATCH 0054/1701] Add beforeunload event to add confirmation dialog when leaving page (#6279) --- js/main.js | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/js/main.js b/js/main.js index 899bd8f02d..71489f14e3 100644 --- a/js/main.js +++ b/js/main.js @@ -600,4 +600,15 @@ headerBar.addEventListener("click", (e) => { } }); +//------------------------------------------------ +// Add a confirmation dialog when leaving the page +// Useful to avoid data loss +//------------------------------------------------ +window.addEventListener('beforeunload', function (event) { + // Cancel the event + event.preventDefault(); + // Chrome requires returnValue to be set + event.returnValue = ''; +}); + moveToChatTab(); From 01362681f2f9e17b6e28ee99b5cf8551a5e62161 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sun, 29 Sep 2024 07:42:44 -0700 Subject: [PATCH 0055/1701] Bump exllamav2 to 0.2.4 --- requirements.txt | 10 +++++----- requirements_amd.txt | 6 +++--- requirements_amd_noavx2.txt | 6 +++--- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_noavx2.txt | 10 +++++----- 6 files changed, 18 insertions(+), 18 deletions(-) diff --git a/requirements.txt b/requirements.txt index d3531b9cd6..c7ba213f0d 100644 --- a/requirements.txt +++ b/requirements.txt @@ -52,11 +52,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index 1be85205d3..6ce5eb62c7 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -41,6 +41,6 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp # AMD wheels https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.0+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.0+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 5026fa6d46..81ad6b23f3 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -39,6 +39,6 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index eee8112d80..6c3a696acb 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -37,4 +37,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index fa244a935e..86ef71975c 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -39,4 +39,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index fe6adb46da..516b0639a8 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -52,11 +52,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.2/exllamav2-0.2.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" From bbdeed3cf44d413eadb09a8f443b9d43ef0ce261 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sun, 29 Sep 2024 20:45:27 -0700 Subject: [PATCH 0056/1701] Make sampler priority high if unspecified --- modules/sampler_hijack.py | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/modules/sampler_hijack.py b/modules/sampler_hijack.py index 6d92978e6e..87f0b25e8e 100644 --- a/modules/sampler_hijack.py +++ b/modules/sampler_hijack.py @@ -602,11 +602,10 @@ def get_logits_processor_patch(self, **kwargs): def custom_sort_key(obj): class_name = obj.__class__.__name__ - # Return a large value if class name is not mapped or if the mapped nickname is not in priority + # Return -1 if class_name is not mapped if class_name not in class_name_to_nickname or class_name_to_nickname[class_name] not in sampler_priority: - return float('inf') + return -1 - # Return the index of the nickname in the priority list for sorting return sampler_priority.index(class_name_to_nickname[class_name]) # Sort the list using the custom key function From 9ca0cd7749f734bd5e7083d672dee8d400229257 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sun, 29 Sep 2024 20:47:04 -0700 Subject: [PATCH 0057/1701] Bump llama-cpp-python to 0.3.1 --- requirements.txt | 24 ++++++++++++------------ requirements_amd.txt | 12 ++++++------ requirements_amd_noavx2.txt | 8 ++++---- requirements_apple_intel.txt | 8 ++++---- requirements_apple_silicon.txt | 12 ++++++------ requirements_cpu_only.txt | 8 ++++---- requirements_cpu_only_noavx2.txt | 8 ++++---- requirements_noavx2.txt | 24 ++++++++++++------------ 8 files changed, 52 insertions(+), 52 deletions(-) diff --git a/requirements.txt b/requirements.txt index c7ba213f0d..5623f1f7ac 100644 --- a/requirements.txt +++ b/requirements.txt @@ -34,22 +34,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index 6ce5eb62c7..d203c157c2 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -33,14 +33,14 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.0+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.0+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.1+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.1+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 81ad6b23f3..3a7a950d54 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -33,10 +33,10 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 6c3a696acb..61c5367ffb 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -33,8 +33,8 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 86ef71975c..494c127afe 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -33,10 +33,10 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.0-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 4c970b9122..5838ac47c2 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -33,7 +33,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index 4b22cb867f..97d4eb9063 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -33,7 +33,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 516b0639a8..d34a598d65 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -34,22 +34,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.0+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.0+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.0+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" From 4d9ce586d31450d2ed692b1e7a72dbcdfd56e670 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 30 Sep 2024 14:04:21 -0700 Subject: [PATCH 0058/1701] Update llama_cpp_python_hijack.py, fix llamacpp_hf --- modules/llama_cpp_python_hijack.py | 19 ++++++++++++------- modules/llamacpp_hf.py | 3 ++- 2 files changed, 14 insertions(+), 8 deletions(-) diff --git a/modules/llama_cpp_python_hijack.py b/modules/llama_cpp_python_hijack.py index 3d42b2d7d9..2a9c10da2e 100644 --- a/modules/llama_cpp_python_hijack.py +++ b/modules/llama_cpp_python_hijack.py @@ -2,12 +2,12 @@ import platform from typing import Sequence +import numpy as np from tqdm import tqdm from modules import shared from modules.cache_utils import process_llamacpp_cache - imported_module = None @@ -57,8 +57,6 @@ def eval_with_progress(self, tokens: Sequence[int]): with tqdm to show prompt processing progress. """ - assert self._ctx.ctx is not None - assert self._batch.batch is not None self._ctx.kv_cache_seq_rm(-1, self.n_tokens, -1) if len(tokens) > self.n_batch: @@ -80,13 +78,20 @@ def eval_with_progress(self, tokens: Sequence[int]): if self.context_params.logits_all: rows = n_tokens cols = self._n_vocab - logits = self._ctx.get_logits()[: rows * cols] - self.scores[n_past : n_past + n_tokens, :].reshape(-1)[: :] = logits + logits = np.ctypeslib.as_array( + self._ctx.get_logits(), shape=(rows * cols,) + ) + self.scores[n_past : n_past + n_tokens, :].reshape(-1)[::] = logits + self.last_updated_index = n_past + n_tokens - 1 else: rows = 1 cols = self._n_vocab - logits = self._ctx.get_logits()[: rows * cols] - self.scores[n_past + n_tokens - 1, :].reshape(-1)[: :] = logits + logits = np.ctypeslib.as_array( + self._ctx.get_logits(), shape=(rows * cols,) + ) + last_token_index = min(n_past + n_tokens - 1, self.scores.shape[0] - 1) + self.scores[last_token_index, :] = logits.reshape(-1) + self.last_updated_index = last_token_index # Update n_tokens self.n_tokens += n_tokens diff --git a/modules/llamacpp_hf.py b/modules/llamacpp_hf.py index 327e3a7b43..6611a7c1a8 100644 --- a/modules/llamacpp_hf.py +++ b/modules/llamacpp_hf.py @@ -127,7 +127,7 @@ def __call__(self, *args, **kwargs): self.model.reset() self.model.eval(seq) - logits = torch.tensor(self.model.scores[self.model.n_tokens - 1, :]).view(1, 1, -1).to(input_ids.device) + logits = torch.tensor(self.model.scores[self.model.last_updated_index, :]).view(1, 1, -1).to(input_ids.device) else: self.model.reset() self.model.eval(seq) @@ -205,5 +205,6 @@ def from_pretrained(cls, pretrained_model_name_or_path: Optional[Union[str, os.P Llama = llama_cpp_lib().Llama model = Llama(**params) + model.last_updated_index = -1 return LlamacppHF(model, model_file) From 6063a6641444da58dc6f38626bbd639cd114eeb8 Mon Sep 17 00:00:00 2001 From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com> Date: Mon, 30 Sep 2024 18:50:38 -0300 Subject: [PATCH 0059/1701] Update accelerate requirement from ==0.33.* to ==0.34.* (#6416) --- requirements.txt | 2 +- requirements_amd.txt | 2 +- requirements_amd_noavx2.txt | 2 +- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_cpu_only.txt | 2 +- requirements_cpu_only_noavx2.txt | 2 +- requirements_noavx2.txt | 2 +- requirements_nowheels.txt | 2 +- 9 files changed, 9 insertions(+), 9 deletions(-) diff --git a/requirements.txt b/requirements.txt index 5623f1f7ac..f623372f70 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==0.34.* bitsandbytes==0.44.* colorama datasets diff --git a/requirements_amd.txt b/requirements_amd.txt index d203c157c2..67e6363238 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==0.34.* colorama datasets einops diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 3a7a950d54..9d1828d6fc 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==0.34.* colorama datasets einops diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 61c5367ffb..7e279163bd 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==0.34.* colorama datasets einops diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 494c127afe..1f47665727 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==0.34.* colorama datasets einops diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 5838ac47c2..fe67192069 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==0.34.* colorama datasets einops diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index 97d4eb9063..32bfe9c5df 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==0.34.* colorama datasets einops diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index d34a598d65..85f94eef62 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==0.34.* bitsandbytes==0.44.* colorama datasets diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index 5c840f56de..5e566798cc 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==0.34.* colorama datasets einops From 617cd7b70514e97a96c365bb5eda481594d48028 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 1 Oct 2024 09:06:25 -0700 Subject: [PATCH 0060/1701] Revert "Update accelerate requirement from ==0.33.* to ==0.34.* (#6416)" This reverts commit 6063a6641444da58dc6f38626bbd639cd114eeb8. --- requirements.txt | 2 +- requirements_amd.txt | 2 +- requirements_amd_noavx2.txt | 2 +- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_cpu_only.txt | 2 +- requirements_cpu_only_noavx2.txt | 2 +- requirements_noavx2.txt | 2 +- requirements_nowheels.txt | 2 +- 9 files changed, 9 insertions(+), 9 deletions(-) diff --git a/requirements.txt b/requirements.txt index f623372f70..5623f1f7ac 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,4 +1,4 @@ -accelerate==0.34.* +accelerate==0.33.* bitsandbytes==0.44.* colorama datasets diff --git a/requirements_amd.txt b/requirements_amd.txt index 67e6363238..d203c157c2 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -1,4 +1,4 @@ -accelerate==0.34.* +accelerate==0.33.* colorama datasets einops diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 9d1828d6fc..3a7a950d54 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==0.34.* +accelerate==0.33.* colorama datasets einops diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 7e279163bd..61c5367ffb 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -1,4 +1,4 @@ -accelerate==0.34.* +accelerate==0.33.* colorama datasets einops diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 1f47665727..494c127afe 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -1,4 +1,4 @@ -accelerate==0.34.* +accelerate==0.33.* colorama datasets einops diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index fe67192069..5838ac47c2 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -1,4 +1,4 @@ -accelerate==0.34.* +accelerate==0.33.* colorama datasets einops diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index 32bfe9c5df..97d4eb9063 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==0.34.* +accelerate==0.33.* colorama datasets einops diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 85f94eef62..d34a598d65 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==0.34.* +accelerate==0.33.* bitsandbytes==0.44.* colorama datasets diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index 5e566798cc..5c840f56de 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -1,4 +1,4 @@ -accelerate==0.34.* +accelerate==0.33.* colorama datasets einops From c6b50f88dab6435a0c27e5836b33f5b790c072f0 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 1 Oct 2024 10:19:28 -0700 Subject: [PATCH 0061/1701] Lint --- requirements.txt | 2 +- requirements_amd.txt | 2 +- requirements_amd_noavx2.txt | 2 +- requirements_noavx2.txt | 2 +- 4 files changed, 4 insertions(+), 4 deletions(-) diff --git a/requirements.txt b/requirements.txt index 6d939275ed..5623f1f7ac 100644 --- a/requirements.txt +++ b/requirements.txt @@ -60,4 +60,4 @@ https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3- https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" \ No newline at end of file +https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements_amd.txt b/requirements_amd.txt index 7741849bec..d203c157c2 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -43,4 +43,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/ro https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.1+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" \ No newline at end of file +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index e4765488e6..3a7a950d54 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -41,4 +41,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp # AMD wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" \ No newline at end of file +https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index edeae78d07..d34a598d65 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -60,4 +60,4 @@ https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3- https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" \ No newline at end of file +https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" From cca9d6e22d6cb3e2909121f6d6077af36b2116f5 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 1 Oct 2024 10:21:06 -0700 Subject: [PATCH 0062/1701] Lint --- modules/text_generation.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/text_generation.py b/modules/text_generation.py index 654eff8940..86245098ab 100644 --- a/modules/text_generation.py +++ b/modules/text_generation.py @@ -274,10 +274,10 @@ def get_reply_from_output_ids(output_ids, state=None, starting_from=0): if (hasattr(shared.tokenizer, 'convert_ids_to_tokens') and len(output_ids) > starting_from) and not reply.startswith(' '): first_token = shared.tokenizer.convert_ids_to_tokens(int(output_ids[starting_from])) if isinstance(first_token, (bytes,)): - #try to decode the bytes to a string + # try to decode the bytes to a string + # if it fails, which means it's not a string in this turn, just ignore it try: first_token = first_token.decode('utf8') - #if it fails, which means it's not a string in this turn, just ignore it except UnicodeDecodeError: first_token = '' From d364aa0a3c0993b09d53966b53eae862f4deaae0 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 1 Oct 2024 10:22:57 -0700 Subject: [PATCH 0063/1701] Lint --- js/main.js | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/js/main.js b/js/main.js index 71489f14e3..3028afac1f 100644 --- a/js/main.js +++ b/js/main.js @@ -604,11 +604,11 @@ headerBar.addEventListener("click", (e) => { // Add a confirmation dialog when leaving the page // Useful to avoid data loss //------------------------------------------------ -window.addEventListener('beforeunload', function (event) { +window.addEventListener("beforeunload", function (event) { // Cancel the event event.preventDefault(); // Chrome requires returnValue to be set - event.returnValue = ''; + event.returnValue = ""; }); moveToChatTab(); From 93c250b9b6b7d625c9cd698f99400d4fe650779d Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 1 Oct 2024 11:16:15 -0700 Subject: [PATCH 0064/1701] Add a UI element for enable_tp --- modules/loaders.py | 2 ++ modules/ui.py | 1 + modules/ui_model_menu.py | 1 + 3 files changed, 4 insertions(+) diff --git a/modules/loaders.py b/modules/loaders.py index 16a3106e94..deee00a7f9 100644 --- a/modules/loaders.py +++ b/modules/loaders.py @@ -90,6 +90,7 @@ 'cache_8bit', 'cache_4bit', 'autosplit', + 'enable_tp', 'alpha_value', 'compress_pos_emb', 'trust_remote_code', @@ -105,6 +106,7 @@ 'cache_8bit', 'cache_4bit', 'autosplit', + 'enable_tp', 'alpha_value', 'compress_pos_emb', 'exllamav2_info', diff --git a/modules/ui.py b/modules/ui.py index 76b1c009c4..c07beeb466 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -90,6 +90,7 @@ def list_model_elements(): 'cache_8bit', 'cache_4bit', 'autosplit', + 'enable_tp', 'threads', 'threads_batch', 'n_batch', diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py index 1883fdca4e..f87b680aeb 100644 --- a/modules/ui_model_menu.py +++ b/modules/ui_model_menu.py @@ -136,6 +136,7 @@ def create_ui(): shared.gradio['disk'] = gr.Checkbox(label="disk", value=shared.args.disk) shared.gradio['bf16'] = gr.Checkbox(label="bf16", value=shared.args.bf16) shared.gradio['autosplit'] = gr.Checkbox(label="autosplit", value=shared.args.autosplit, info='Automatically split the model tensors across the available GPUs.') + shared.gradio['enable_tp'] = gr.Checkbox(label="enable_tp", value=shared.args.enable_tp, info='Enable Tensor Parallelism (TP).') shared.gradio['no_flash_attn'] = gr.Checkbox(label="no_flash_attn", value=shared.args.no_flash_attn) shared.gradio['no_xformers'] = gr.Checkbox(label="no_xformers", value=shared.args.no_xformers) shared.gradio['no_sdpa'] = gr.Checkbox(label="no_sdpa", value=shared.args.no_sdpa) From 49dfa0adaf52bcff76b3ab9c33326812b281c2cf Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 1 Oct 2024 11:20:48 -0700 Subject: [PATCH 0065/1701] Fix the "save preset" event --- modules/ui_file_saving.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/ui_file_saving.py b/modules/ui_file_saving.py index ac72c62316..3a27e1b957 100644 --- a/modules/ui_file_saving.py +++ b/modules/ui_file_saving.py @@ -74,7 +74,7 @@ def handle_save_preset_confirm_click(filename, contents): try: utils.save_file(f"presets/{filename}.yaml", contents) available_presets = utils.get_available_presets() - output = gr.update(choices=available_presets, value=filename), + output = gr.update(choices=available_presets, value=filename) except Exception: output = gr.update() traceback.print_exc() From e1338a1804d51d4e74146f2490586985ba499ce0 Mon Sep 17 00:00:00 2001 From: SeanScripts <64337075+SeanScripts@users.noreply.github.com> Date: Tue, 1 Oct 2024 13:49:35 -0700 Subject: [PATCH 0066/1701] Add whisper turbo (#6423) --- extensions/whisper_stt/script.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/extensions/whisper_stt/script.py b/extensions/whisper_stt/script.py index e45c8b1e7c..d949e93f73 100644 --- a/extensions/whisper_stt/script.py +++ b/extensions/whisper_stt/script.py @@ -96,7 +96,7 @@ def ui(): with gr.Accordion("Settings", open=False): auto_submit = gr.Checkbox(label='Submit the transcribed audio automatically', value=params['auto_submit']) device_dropd = gr.Dropdown(label='Device', value=str(startup_device), choices=["cuda", "cpu", "none"]) - whisper_model_dropd = gr.Dropdown(label='Whisper Model', value=params['whipser_model'], choices=["tiny.en", "base.en", "small.en", "medium.en", "tiny", "base", "small", "medium", "large"]) + whisper_model_dropd = gr.Dropdown(label='Whisper Model', value=params['whipser_model'], choices=["tiny.en", "base.en", "small.en", "medium.en", "tiny", "base", "small", "medium", "large", "turbo"]) whisper_language = gr.Dropdown(label='Whisper Language', value=params['whipser_language'], choices=["english", "chinese", "german", "spanish", "russian", "korean", "french", "japanese", "portuguese", "turkish", "polish", "catalan", "dutch", "arabic", "swedish", "italian", "indonesian", "hindi", "finnish", "vietnamese", "hebrew", "ukrainian", "greek", "malay", "czech", "romanian", "danish", "hungarian", "tamil", "norwegian", "thai", "urdu", "croatian", "bulgarian", "lithuanian", "latin", "maori", "malayalam", "welsh", "slovak", "telugu", "persian", "latvian", "bengali", "serbian", "azerbaijani", "slovenian", "kannada", "estonian", "macedonian", "breton", "basque", "icelandic", "armenian", "nepali", "mongolian", "bosnian", "kazakh", "albanian", "swahili", "galician", "marathi", "punjabi", "sinhala", "khmer", "shona", "yoruba", "somali", "afrikaans", "occitan", "georgian", "belarusian", "tajik", "sindhi", "gujarati", "amharic", "yiddish", "lao", "uzbek", "faroese", "haitian creole", "pashto", "turkmen", "nynorsk", "maltese", "sanskrit", "luxembourgish", "myanmar", "tibetan", "tagalog", "malagasy", "assamese", "tatar", "hawaiian", "lingala", "hausa", "bashkir", "javanese", "sundanese"]) audio.change( From 22baa5378f27de6427640a4ce3ba3a54bb3a9722 Mon Sep 17 00:00:00 2001 From: Luana Date: Thu, 3 Oct 2024 00:35:13 -0300 Subject: [PATCH 0067/1701] Fix for systems that have bash in a non-standard directory (#6428) --- one_click.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/one_click.py b/one_click.py index 9bad25d1be..f1c91f478c 100644 --- a/one_click.py +++ b/one_click.py @@ -189,8 +189,8 @@ def run_cmd(cmd, assert_success=False, environment=False, capture_output=False, conda_sh_path = os.path.join(script_dir, "installer_files", "conda", "etc", "profile.d", "conda.sh") cmd = f'. "{conda_sh_path}" && conda activate "{conda_env_path}" && {cmd}' - # Set executable to None for Windows, /bin/bash for everything else - executable = None if is_windows() else '/bin/bash' + # Set executable to None for Windows, bash for everything else + executable = None if is_windows() else 'bash' # Run shell commands result = subprocess.run(cmd, shell=True, capture_output=capture_output, env=env, executable=executable) From 9d8b1c5fd94d360da4ab55334442cfa64aaf40d0 Mon Sep 17 00:00:00 2001 From: Grzegorz Lippe Date: Sat, 5 Oct 2024 16:58:17 +0200 Subject: [PATCH 0068/1701] Fix intel bug described in #6253 (#6433) --- one_click.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/one_click.py b/one_click.py index f1c91f478c..8fc1edf0cf 100644 --- a/one_click.py +++ b/one_click.py @@ -313,7 +313,7 @@ def install_webui(): if selected_gpu == "INTEL": # Install oneAPI dependencies via conda print_big_message("Installing Intel oneAPI runtime libraries.") - run_cmd("conda install -y -c intel dpcpp-cpp-rt=2024.0 mkl-dpcpp=2024.0") + run_cmd("conda install -y -c https://software.repos.intel.com/python/conda/ -c conda-forge dpcpp-cpp-rt=2024.0 mkl-dpcpp=2024.0") # Install libuv required by Intel-patched torch run_cmd("conda install -y libuv") @@ -329,7 +329,7 @@ def install_extensions_requirements(): print_big_message("Installing extensions requirements.\nSome of these may fail on Windows.\nDon\'t worry if you see error messages, as they will not affect the main program.") extensions = get_extensions_names() for i, extension in enumerate(extensions): - print(f"\n\n--- [{i+1}/{len(extensions)}]: {extension}\n\n") + print(f"\n\n--- [{i + 1}/{len(extensions)}]: {extension}\n\n") extension_req_path = os.path.join("extensions", extension, "requirements.txt") run_cmd(f"python -m pip install -r {extension_req_path} --upgrade", assert_success=False, environment=True) From 03a2e7005412efb7765ce3f0bfccd6b94e79a843 Mon Sep 17 00:00:00 2001 From: PIRI <34787507+ThisIsPIRI@users.noreply.github.com> Date: Wed, 9 Oct 2024 14:25:14 +0000 Subject: [PATCH 0069/1701] Fix temperature_last when temperature not in sampler priority (#6439) --- modules/sampler_hijack.py | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-) diff --git a/modules/sampler_hijack.py b/modules/sampler_hijack.py index 87f0b25e8e..97951dd278 100644 --- a/modules/sampler_hijack.py +++ b/modules/sampler_hijack.py @@ -571,11 +571,10 @@ def get_logits_processor_patch(self, **kwargs): if generation_config.temperature_last: for param_name in ['temperature', 'dynamic_temperature', 'quadratic_sampling']: if param_name in sampler_priority: - if param_name in sampler_priority: - index = sampler_priority.index(param_name) - sampler_priority.append(sampler_priority.pop(index)) - else: - sampler_priority.append(param_name) + index = sampler_priority.index(param_name) + sampler_priority.append(sampler_priority.pop(index)) + else: + sampler_priority.append(param_name) class_name_to_nickname = { 'DynamicTemperatureLogitsWarper': 'dynamic_temperature', From c9a9f63d1b9c4bd8a08dcef519728cd82b5dc4da Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 14 Oct 2024 13:05:51 -0700 Subject: [PATCH 0070/1701] Fix llama.cpp loader not being random (thanks @reydeljuego12345) --- modules/llamacpp_model.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/llamacpp_model.py b/modules/llamacpp_model.py index a16230caf3..96f7ed56b5 100644 --- a/modules/llamacpp_model.py +++ b/modules/llamacpp_model.py @@ -136,7 +136,7 @@ def generate(self, prompt, state, callback=None): prompt=prompt, max_tokens=state['max_new_tokens'], temperature=state['temperature'], - top_p=state['top_p'], + top_p=state['top_p'] if state['top_p'] < 1 else 0.999, min_p=state['min_p'], typical_p=state['typical_p'], frequency_penalty=state['frequency_penalty'], From bb62e796eba0bfe4610ac879bf7b2f390d96cf92 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 14 Oct 2024 13:24:13 -0700 Subject: [PATCH 0071/1701] Fix locally compiled llama-cpp-python failing to import --- modules/llama_cpp_python_hijack.py | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/modules/llama_cpp_python_hijack.py b/modules/llama_cpp_python_hijack.py index 2a9c10da2e..f3872a7446 100644 --- a/modules/llama_cpp_python_hijack.py +++ b/modules/llama_cpp_python_hijack.py @@ -9,10 +9,11 @@ from modules.cache_utils import process_llamacpp_cache imported_module = None +not_available_modules = set() def llama_cpp_lib(): - global imported_module + global imported_module, not_available_modules # Determine the platform is_macos = platform.system() == 'Darwin' @@ -31,6 +32,9 @@ def llama_cpp_lib(): ] for arg, lib_name in lib_names: + if lib_name in not_available_modules: + continue + should_import = (arg is None or getattr(shared.args, arg)) if should_import: @@ -44,6 +48,7 @@ def llama_cpp_lib(): monkey_patch_llama_cpp_python(return_lib) return return_lib except ImportError: + not_available_modules.add(lib_name) continue return None From f1a8eae04dec1c70f66fc2fd662011463003ffca Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 14 Oct 2024 13:30:45 -0700 Subject: [PATCH 0072/1701] Remove optimum from requirements --- requirements.txt | 1 - requirements_amd.txt | 1 - requirements_amd_noavx2.txt | 1 - requirements_apple_intel.txt | 1 - requirements_apple_silicon.txt | 1 - requirements_cpu_only.txt | 1 - requirements_cpu_only_noavx2.txt | 1 - requirements_noavx2.txt | 1 - requirements_nowheels.txt | 1 - 9 files changed, 9 deletions(-) diff --git a/requirements.txt b/requirements.txt index 5623f1f7ac..0d75c5f329 100644 --- a/requirements.txt +++ b/requirements.txt @@ -10,7 +10,6 @@ lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* -optimum==1.17.* pandas peft==0.12.* Pillow>=9.5.0 diff --git a/requirements_amd.txt b/requirements_amd.txt index d203c157c2..00bd81a1f8 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -9,7 +9,6 @@ lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* -optimum==1.17.* pandas peft==0.12.* Pillow>=9.5.0 diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 3a7a950d54..36ab65fe6f 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -9,7 +9,6 @@ lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* -optimum==1.17.* pandas peft==0.12.* Pillow>=9.5.0 diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 61c5367ffb..2de71a73e9 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -9,7 +9,6 @@ lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* -optimum==1.17.* pandas peft==0.12.* Pillow>=9.5.0 diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 494c127afe..03cbf62256 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -9,7 +9,6 @@ lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* -optimum==1.17.* pandas peft==0.12.* Pillow>=9.5.0 diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 5838ac47c2..39c11b757e 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -9,7 +9,6 @@ lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* -optimum==1.17.* pandas peft==0.12.* Pillow>=9.5.0 diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index 97d4eb9063..224a020734 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -9,7 +9,6 @@ lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* -optimum==1.17.* pandas peft==0.12.* Pillow>=9.5.0 diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index d34a598d65..247b83ff80 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -10,7 +10,6 @@ lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* -optimum==1.17.* pandas peft==0.12.* Pillow>=9.5.0 diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index 5c840f56de..c0ab9eeda7 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -9,7 +9,6 @@ lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* -optimum==1.17.* pandas peft==0.12.* Pillow>=9.5.0 From e784938654f31a1b5b8342d49e5a125cd19899f9 Mon Sep 17 00:00:00 2001 From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com> Date: Mon, 14 Oct 2024 17:32:53 -0300 Subject: [PATCH 0073/1701] Update accelerate requirement from ==0.33.* to ==1.0.* (#6441) --- requirements.txt | 2 +- requirements_amd.txt | 2 +- requirements_amd_noavx2.txt | 2 +- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_cpu_only.txt | 2 +- requirements_cpu_only_noavx2.txt | 2 +- requirements_noavx2.txt | 2 +- requirements_nowheels.txt | 2 +- 9 files changed, 9 insertions(+), 9 deletions(-) diff --git a/requirements.txt b/requirements.txt index 0d75c5f329..b218ac8f90 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==1.0.* bitsandbytes==0.44.* colorama datasets diff --git a/requirements_amd.txt b/requirements_amd.txt index 00bd81a1f8..3baafb3913 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==1.0.* colorama datasets einops diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 36ab65fe6f..dfdcc86b9e 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==1.0.* colorama datasets einops diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 2de71a73e9..59afc45e0a 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==1.0.* colorama datasets einops diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 03cbf62256..2df573c8df 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==1.0.* colorama datasets einops diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 39c11b757e..2653ddbd19 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==1.0.* colorama datasets einops diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index 224a020734..35092d2247 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==1.0.* colorama datasets einops diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 247b83ff80..b5e0204c80 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==1.0.* bitsandbytes==0.44.* colorama datasets diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index c0ab9eeda7..47b0dcda8d 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -1,4 +1,4 @@ -accelerate==0.33.* +accelerate==1.0.* colorama datasets einops From 18f836b28050da93b054547b6d48214b52e56727 Mon Sep 17 00:00:00 2001 From: Molly Sophia Date: Tue, 15 Oct 2024 04:51:20 +0800 Subject: [PATCH 0074/1701] Add RWKV-World instruction template (#6456) --- instruction-templates/RWKV-World.yaml | 25 +++++++++++++++++++++++++ 1 file changed, 25 insertions(+) create mode 100644 instruction-templates/RWKV-World.yaml diff --git a/instruction-templates/RWKV-World.yaml b/instruction-templates/RWKV-World.yaml new file mode 100644 index 0000000000..bf65511b8e --- /dev/null +++ b/instruction-templates/RWKV-World.yaml @@ -0,0 +1,25 @@ +instruction_template: |- + {%- set ns = namespace(found=false) -%} + {%- for message in messages -%} + {%- if message['role'] == 'system' -%} + {%- set ns.found = true -%} + {%- endif -%} + {%- endfor -%} + {%- if not ns.found -%} + {{- '' + '' + '' -}} + {%- endif %} + {%- for message in messages %} + {%- if message['role'] == 'system' -%} + {{- '' + message['content'] + '' -}} + {%- else -%} + {%- if message['role'] == 'user' -%} + {{-'User: ' + message['content'] + '\n\n'-}} + {%- else -%} + {{-'Assistant: ' + message['content'] + '\n\n' -}} + {%- endif -%} + {%- endif -%} + {%- endfor -%} + {%- if add_generation_prompt -%} + {{-'Assistant:'-}} + {%- endif -%} + From 6a0837451e1f88ad2db1de5aafda2b7155aab245 Mon Sep 17 00:00:00 2001 From: Paul Richardson Date: Tue, 15 Oct 2024 08:39:00 -0500 Subject: [PATCH 0075/1701] Minor Documentation update - query cuda compute for docker .env (#6469) --- docker/.env.example | 1 + 1 file changed, 1 insertion(+) diff --git a/docker/.env.example b/docker/.env.example index 2de9f0ab6f..bd0f8bcc5e 100644 --- a/docker/.env.example +++ b/docker/.env.example @@ -1,6 +1,7 @@ # by default the Dockerfile specifies these versions: 3.5;5.0;6.0;6.1;7.0;7.5;8.0;8.6+PTX # however for me to work i had to specify the exact version for my card ( 2060 ) it was 7.5 # https://developer.nvidia.com/cuda-gpus you can find the version for your card here +# Or for a programatic approach run `nvidia-smi --query-gpu=name,compute_cap --format=csv` TORCH_CUDA_ARCH_LIST=7.5 # the port the webui binds to on the host HOST_PORT=7860 From 386c0d82890c258b3552b0440befb1f744a99c22 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Thu, 24 Oct 2024 11:09:09 -0700 Subject: [PATCH 0076/1701] Bump transformers to 4.46 --- requirements.txt | 2 +- requirements_amd.txt | 2 +- requirements_amd_noavx2.txt | 2 +- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_cpu_only.txt | 2 +- requirements_cpu_only_noavx2.txt | 2 +- requirements_noavx2.txt | 2 +- requirements_nowheels.txt | 2 +- 9 files changed, 9 insertions(+), 9 deletions(-) diff --git a/requirements.txt b/requirements.txt index b218ac8f90..5e5c1381d2 100644 --- a/requirements.txt +++ b/requirements.txt @@ -22,7 +22,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.45.* +transformers==4.46.* tqdm wandb diff --git a/requirements_amd.txt b/requirements_amd.txt index 3baafb3913..b2a2b50c3d 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.45.* +transformers==4.46.* tqdm wandb diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index dfdcc86b9e..5dc0de683b 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.45.* +transformers==4.46.* tqdm wandb diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 59afc45e0a..cc88d62bf0 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.45.* +transformers==4.46.* tqdm wandb diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 2df573c8df..3ddfcd2822 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.45.* +transformers==4.46.* tqdm wandb diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 2653ddbd19..5d1ea2cea6 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.45.* +transformers==4.46.* tqdm wandb diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index 35092d2247..5eb08db352 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.45.* +transformers==4.46.* tqdm wandb diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index b5e0204c80..6ee676b653 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -22,7 +22,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.45.* +transformers==4.46.* tqdm wandb diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index 47b0dcda8d..782242f945 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.45.* +transformers==4.46.* tqdm wandb From e1061ba7e320727a3eab90d2c73c06e8eb740ec7 Mon Sep 17 00:00:00 2001 From: PIRI <34787507+ThisIsPIRI@users.noreply.github.com> Date: Thu, 24 Oct 2024 18:24:02 +0000 Subject: [PATCH 0077/1701] Make token bans work again on HF loaders (#6488) --- modules/sampler_hijack.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/sampler_hijack.py b/modules/sampler_hijack.py index 97951dd278..24dbcf2ee1 100644 --- a/modules/sampler_hijack.py +++ b/modules/sampler_hijack.py @@ -454,7 +454,7 @@ def get_logits_processor_patch(self, **kwargs): ) # Stuff we don't need - elif warpers[i].__class__.__name__ in ['SuppressTokensLogitsProcessor', 'RepetitionPenaltyLogitsProcessor']: + elif warpers[i].__class__.__name__ in ['RepetitionPenaltyLogitsProcessor']: del warpers[i] # Add custom warpers From 8deea2936d303e5543237004648e2d06b09478ae Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Thu, 24 Oct 2024 11:25:42 -0700 Subject: [PATCH 0078/1701] Remove lm_eval from requirements --- requirements.txt | 1 - requirements_amd.txt | 1 - requirements_amd_noavx2.txt | 1 - requirements_apple_intel.txt | 1 - requirements_apple_silicon.txt | 1 - requirements_cpu_only.txt | 1 - requirements_cpu_only_noavx2.txt | 1 - requirements_noavx2.txt | 1 - requirements_nowheels.txt | 1 - 9 files changed, 9 deletions(-) diff --git a/requirements.txt b/requirements.txt index 5e5c1381d2..2549c64864 100644 --- a/requirements.txt +++ b/requirements.txt @@ -6,7 +6,6 @@ einops fastapi==0.112.4 gradio==4.26.* jinja2==3.1.4 -lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* diff --git a/requirements_amd.txt b/requirements_amd.txt index b2a2b50c3d..78bdd3ca84 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -5,7 +5,6 @@ einops fastapi==0.112.4 gradio==4.26.* jinja2==3.1.4 -lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 5dc0de683b..9420e861f8 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -5,7 +5,6 @@ einops fastapi==0.112.4 gradio==4.26.* jinja2==3.1.4 -lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index cc88d62bf0..625021ee1f 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -5,7 +5,6 @@ einops fastapi==0.112.4 gradio==4.26.* jinja2==3.1.4 -lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 3ddfcd2822..3cb66cbc81 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -5,7 +5,6 @@ einops fastapi==0.112.4 gradio==4.26.* jinja2==3.1.4 -lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 5d1ea2cea6..fbd6447bdf 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -5,7 +5,6 @@ einops fastapi==0.112.4 gradio==4.26.* jinja2==3.1.4 -lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index 5eb08db352..e9ab0fbad3 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -5,7 +5,6 @@ einops fastapi==0.112.4 gradio==4.26.* jinja2==3.1.4 -lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 6ee676b653..99791ea9b4 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -6,7 +6,6 @@ einops fastapi==0.112.4 gradio==4.26.* jinja2==3.1.4 -lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index 782242f945..f5c3966eb3 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -5,7 +5,6 @@ einops fastapi==0.112.4 gradio==4.26.* jinja2==3.1.4 -lm_eval==0.3.0 markdown numba==0.59.* numpy==1.26.* From 8d5cf7b134e227cedde9afc744c06ba94cffe86a Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 18 Nov 2024 06:51:06 -0800 Subject: [PATCH 0079/1701] Bump llama-cpp-python to 0.3.2 --- requirements.txt | 24 ++++++++++++------------ requirements_amd.txt | 12 ++++++------ requirements_amd_noavx2.txt | 8 ++++---- requirements_apple_intel.txt | 8 ++++---- requirements_apple_silicon.txt | 12 ++++++------ requirements_cpu_only.txt | 8 ++++---- requirements_cpu_only_noavx2.txt | 8 ++++---- requirements_noavx2.txt | 24 ++++++++++++------------ 8 files changed, 52 insertions(+), 52 deletions(-) diff --git a/requirements.txt b/requirements.txt index 2549c64864..fadbb96115 100644 --- a/requirements.txt +++ b/requirements.txt @@ -32,22 +32,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index 78bdd3ca84..9f44f53edc 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -31,14 +31,14 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.1+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.1+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.2+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.2+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 9420e861f8..5e552d290c 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -31,10 +31,10 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 625021ee1f..301e2fb01a 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -31,8 +31,8 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 3cb66cbc81..eb2ffed0de 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -31,10 +31,10 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.1-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index fbd6447bdf..9a1ccc8d58 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -31,7 +31,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index e9ab0fbad3..7ffdc71858 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -31,7 +31,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 99791ea9b4..3333b81d93 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -32,22 +32,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.1+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.1+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.1+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" From 0c48ecf3599786a58a1b8360fcd398e382049823 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 18 Nov 2024 06:51:56 -0800 Subject: [PATCH 0080/1701] Bump exllamav2 to 0.2.4 --- requirements.txt | 10 +++++----- requirements_amd.txt | 6 +++--- requirements_amd_noavx2.txt | 6 +++--- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_noavx2.txt | 10 +++++----- 6 files changed, 18 insertions(+), 18 deletions(-) diff --git a/requirements.txt b/requirements.txt index fadbb96115..127f63bad3 100644 --- a/requirements.txt +++ b/requirements.txt @@ -50,11 +50,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index 9f44f53edc..f898019169 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -39,6 +39,6 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp # AMD wheels https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.2+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.2+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 5e552d290c..4c5d757427 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -37,6 +37,6 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 301e2fb01a..aad1f59eeb 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -35,4 +35,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4-py3-none-any.whl diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index eb2ffed0de..52d5c7bc20 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -37,4 +37,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4-py3-none-any.whl diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 3333b81d93..c503061788 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -50,11 +50,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.3/exllamav2-0.2.3-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" From 5fa9336dab1666d125d1892b5d4f266444c57f56 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 18 Nov 2024 06:55:29 -0800 Subject: [PATCH 0081/1701] Bump flash-attention to 2.7.0.post2 --- requirements.txt | 8 ++++---- requirements_noavx2.txt | 8 ++++---- 2 files changed, 8 insertions(+), 8 deletions(-) diff --git a/requirements.txt b/requirements.txt index 127f63bad3..eb2c8ea8bf 100644 --- a/requirements.txt +++ b/requirements.txt @@ -55,7 +55,7 @@ https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+ https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" -https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu12torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index c503061788..57fa41f4c7 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -55,7 +55,7 @@ https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+ https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" -https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu12torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" From 9b3a3d8f12d3d4f2157965a203e3d0469cf90ec9 Mon Sep 17 00:00:00 2001 From: hronoas Date: Mon, 18 Nov 2024 17:59:52 +0300 Subject: [PATCH 0082/1701] openai extension fix: Handle Multiple Content Items in Messages (#6528) --- extensions/openai/completions.py | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-) diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py index 6bd8f409a0..2cefc22bde 100644 --- a/extensions/openai/completions.py +++ b/extensions/openai/completions.py @@ -143,21 +143,20 @@ def convert_history(history): new_history = [] for entry in history: if isinstance(entry['content'], list): - image_url = None - content = None for item in entry['content']: if not isinstance(item, dict): continue - + + image_url = None + content = None if item['type'] == 'image_url' and isinstance(item['image_url'], dict): image_url = item['image_url']['url'] elif item['type'] == 'text' and isinstance(item['text'], str): content = item['text'] - - if image_url: - new_history.append({"image_url": image_url, "role": "user"}) - if content: - new_history.append({"content": content, "role": "user"}) + if image_url: + new_history.append({"image_url": image_url, "role": "user"}) + if content: + new_history.append({"content": content, "role": "user"}) else: new_history.append(entry) From f93196e3065a997fafad7fff503ddf0c96169798 Mon Sep 17 00:00:00 2001 From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com> Date: Mon, 18 Nov 2024 12:00:24 -0300 Subject: [PATCH 0083/1701] Update accelerate requirement from ==1.0.* to ==1.1.* (#6515) --- requirements.txt | 2 +- requirements_amd.txt | 2 +- requirements_amd_noavx2.txt | 2 +- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_cpu_only.txt | 2 +- requirements_cpu_only_noavx2.txt | 2 +- requirements_noavx2.txt | 2 +- requirements_nowheels.txt | 2 +- 9 files changed, 9 insertions(+), 9 deletions(-) diff --git a/requirements.txt b/requirements.txt index eb2c8ea8bf..27aad7a62f 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,4 +1,4 @@ -accelerate==1.0.* +accelerate==1.1.* bitsandbytes==0.44.* colorama datasets diff --git a/requirements_amd.txt b/requirements_amd.txt index f898019169..63b00a38f1 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -1,4 +1,4 @@ -accelerate==1.0.* +accelerate==1.1.* colorama datasets einops diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 4c5d757427..1a16ed0e4b 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==1.0.* +accelerate==1.1.* colorama datasets einops diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index aad1f59eeb..74e3041ad9 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -1,4 +1,4 @@ -accelerate==1.0.* +accelerate==1.1.* colorama datasets einops diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 52d5c7bc20..42975d20bf 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -1,4 +1,4 @@ -accelerate==1.0.* +accelerate==1.1.* colorama datasets einops diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 9a1ccc8d58..dd7039e28e 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -1,4 +1,4 @@ -accelerate==1.0.* +accelerate==1.1.* colorama datasets einops diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index 7ffdc71858..c39d788f86 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==1.0.* +accelerate==1.1.* colorama datasets einops diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 57fa41f4c7..140d92f66a 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==1.0.* +accelerate==1.1.* bitsandbytes==0.44.* colorama datasets diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index f5c3966eb3..44d5899534 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -1,4 +1,4 @@ -accelerate==1.0.* +accelerate==1.1.* colorama datasets einops From 1c937dad72dfcd176bdc1f25938585e91a48d3d3 Mon Sep 17 00:00:00 2001 From: mefich Date: Mon, 18 Nov 2024 20:01:40 +0500 Subject: [PATCH 0084/1701] Filter whitespaces in downloader fields in model tab (#6518) --- modules/ui_model_menu.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py index f87b680aeb..34d581776b 100644 --- a/modules/ui_model_menu.py +++ b/modules/ui_model_menu.py @@ -260,6 +260,8 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur yield ("Please enter a model path") return + repo_id = repo_id.strip() + specific_file = specific_file.strip() downloader = importlib.import_module("download-model").ModelDownloader() progress(0.0) From 3d19746a5dd0d14768035b7287814ca6504988ce Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 18 Nov 2024 10:14:09 -0800 Subject: [PATCH 0085/1701] UI: improve HTML rendering for lists with sub-lists --- css/main.css | 6 +++++- modules/html_generator.py | 26 ++++++++++++++++++-------- 2 files changed, 23 insertions(+), 9 deletions(-) diff --git a/css/main.css b/css/main.css index cf3babdba6..26e58b5cd1 100644 --- a/css/main.css +++ b/css/main.css @@ -404,7 +404,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { .message-body h3, .message-body h4 { color: var(--body-text-color); - margin: 20px 0 10px 0; + margin: 20px 0 10px; } .dark .message q { @@ -456,6 +456,10 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { overflow: scroll; } +.prose ul ul { + margin: 0; +} + .message-body code { white-space: pre-wrap !important; word-wrap: break-word !important; diff --git a/modules/html_generator.py b/modules/html_generator.py index d0afd6b213..01b2086610 100644 --- a/modules/html_generator.py +++ b/modules/html_generator.py @@ -104,6 +104,8 @@ def convert_to_markdown(string): result = '' is_code = False is_latex = False + previous_line_empty = True + for line in string.split('\n'): stripped_line = line.strip() @@ -120,13 +122,20 @@ def convert_to_markdown(string): elif stripped_line.endswith('\\\\]'): is_latex = False - result += line - - # Don't add an extra \n for tables, code, or LaTeX - if is_code or is_latex or line.startswith('|'): - result += '\n' + # Preserve indentation for lists and code blocks + if stripped_line.startswith('-') or stripped_line.startswith('*') or stripped_line.startswith('+') or stripped_line.startswith('>') or re.match(r'\d+\.', stripped_line): + result += line + '\n' + previous_line_empty = False + elif is_code or is_latex or line.startswith('|'): + result += line + '\n' + previous_line_empty = False else: - result += '\n\n' + if previous_line_empty: + result += line.strip() + '\n' + else: + result += line.strip() + '\n\n' + + previous_line_empty = stripped_line == '' result = result.strip() if is_code: @@ -145,14 +154,15 @@ def convert_to_markdown(string): result = re.sub(list_item_pattern, r'\g<1> ' + delete_str, result) # Convert to HTML using markdown - html_output = markdown.markdown(result, extensions=['fenced_code', 'tables']) + html_output = markdown.markdown(result, extensions=['fenced_code', 'tables'], tab_length=2) # Remove the delete string from the HTML output pos = html_output.rfind(delete_str) if pos > -1: html_output = html_output[:pos] + html_output[pos + len(delete_str):] else: - html_output = markdown.markdown(result, extensions=['fenced_code', 'tables']) + # Convert to HTML using markdown + html_output = markdown.markdown(result, extensions=['fenced_code', 'tables'], tab_length=2) # Unescape code blocks pattern = re.compile(r']*>(.*?)', re.DOTALL) From 350758f81cfd27b32abe8ab6a8837f5bf0253479 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 19 Nov 2024 20:34:36 -0800 Subject: [PATCH 0086/1701] UI: Fix the history upload event --- modules/chat.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/chat.py b/modules/chat.py index b81cfea6ee..787277b997 100644 --- a/modules/chat.py +++ b/modules/chat.py @@ -1108,8 +1108,8 @@ def handle_rename_chat_confirm(rename_to, state): def handle_upload_chat_history(load_chat_history, state): history = start_new_chat(state) history = load_history_json(load_chat_history, history) - histories = find_all_histories_with_first_prompts(state) save_history(history, state['unique_id'], state['character_menu'], state['mode']) + histories = find_all_histories_with_first_prompts(state) html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu']) From aa629e2809dd89b704ce37e1574ae6f0bd7b8f52 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sun, 1 Dec 2024 12:00:28 -0800 Subject: [PATCH 0087/1701] Bump exllamav2 to 0.2.5 --- requirements.txt | 10 +++++----- requirements_amd.txt | 6 +++--- requirements_amd_noavx2.txt | 6 +++--- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_noavx2.txt | 10 +++++----- 6 files changed, 18 insertions(+), 18 deletions(-) diff --git a/requirements.txt b/requirements.txt index 27aad7a62f..9be94f62ab 100644 --- a/requirements.txt +++ b/requirements.txt @@ -50,11 +50,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index 63b00a38f1..f03bfcd2fa 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -39,6 +39,6 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp # AMD wheels https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.2+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.2+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 1a16ed0e4b..06416fe631 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -37,6 +37,6 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 74e3041ad9..66bc6c1a3d 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -35,4 +35,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4-py3-none-any.whl +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5-py3-none-any.whl diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 42975d20bf..f76222d682 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -37,4 +37,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4-py3-none-any.whl +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5-py3-none-any.whl diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 140d92f66a..43e9274299 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -50,11 +50,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.4/exllamav2-0.2.4-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" From f7836c4bd844beae8f1158b205d4dbffd4d1f526 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 9 Dec 2024 07:00:15 -0800 Subject: [PATCH 0088/1701] Bump transformers to 4.47 --- requirements.txt | 2 +- requirements_amd.txt | 2 +- requirements_amd_noavx2.txt | 2 +- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_cpu_only.txt | 2 +- requirements_cpu_only_noavx2.txt | 2 +- requirements_noavx2.txt | 2 +- requirements_nowheels.txt | 2 +- 9 files changed, 9 insertions(+), 9 deletions(-) diff --git a/requirements.txt b/requirements.txt index 9be94f62ab..825b2c78f1 100644 --- a/requirements.txt +++ b/requirements.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.46.* +transformers==4.47.* tqdm wandb diff --git a/requirements_amd.txt b/requirements_amd.txt index f03bfcd2fa..7ceba1cdc9 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -20,7 +20,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.46.* +transformers==4.47.* tqdm wandb diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 06416fe631..275ab52ff5 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -20,7 +20,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.46.* +transformers==4.47.* tqdm wandb diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 66bc6c1a3d..9abe4cdf77 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -20,7 +20,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.46.* +transformers==4.47.* tqdm wandb diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index f76222d682..0e7ad13865 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -20,7 +20,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.46.* +transformers==4.47.* tqdm wandb diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index dd7039e28e..a847838187 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -20,7 +20,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.46.* +transformers==4.47.* tqdm wandb diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index c39d788f86..b3d54717ec 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -20,7 +20,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.46.* +transformers==4.47.* tqdm wandb diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 43e9274299..090abbad2b 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -21,7 +21,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.46.* +transformers==4.47.* tqdm wandb diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index 44d5899534..28144afec0 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -20,7 +20,7 @@ safetensors==0.4.* scipy sentencepiece tensorboard -transformers==4.46.* +transformers==4.47.* tqdm wandb From baa566b0c6693d655717905cf82ad6c3af481ba1 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 9 Dec 2024 10:16:33 -0800 Subject: [PATCH 0089/1701] Bump exllamav2 to 0.2.6 --- requirements.txt | 10 +++++----- requirements_amd.txt | 6 +++--- requirements_amd_noavx2.txt | 6 +++--- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_noavx2.txt | 10 +++++----- 6 files changed, 18 insertions(+), 18 deletions(-) diff --git a/requirements.txt b/requirements.txt index 825b2c78f1..5010813956 100644 --- a/requirements.txt +++ b/requirements.txt @@ -50,11 +50,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index 7ceba1cdc9..5ab892208c 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -39,6 +39,6 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp # AMD wheels https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.2+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.2+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 275ab52ff5..3b9e7ab752 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -37,6 +37,6 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 9abe4cdf77..98545639ac 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -35,4 +35,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5-py3-none-any.whl +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 0e7ad13865..2cecee5a5b 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -37,4 +37,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5-py3-none-any.whl +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 090abbad2b..dc8409b4b5 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -50,11 +50,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/exllamav2/releases/download/v0.2.5/exllamav2-0.2.5-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" From 27398428f606022fc5d1b454ff8795e999e9fab2 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 9 Dec 2024 10:17:17 -0800 Subject: [PATCH 0090/1701] Bump flash-attention to v2.7.2.post1 --- requirements.txt | 8 ++++---- requirements_noavx2.txt | 8 ++++---- 2 files changed, 8 insertions(+), 8 deletions(-) diff --git a/requirements.txt b/requirements.txt index 5010813956..7b88f62676 100644 --- a/requirements.txt +++ b/requirements.txt @@ -55,7 +55,7 @@ https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+ https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" -https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu12torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu12torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index dc8409b4b5..96437260ed 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -55,7 +55,7 @@ https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+ https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64" -https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.0.post2/flash_attn-2.7.0.post2+cu12torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu12torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" From 97f56156618fb2b15a87fcc1e324599ad975f6aa Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 11 Dec 2024 07:14:59 -0800 Subject: [PATCH 0091/1701] Bump llama-cpp-python to 0.3.5, remove macos 12 wheels (workflow is failing) --- requirements.txt | 24 ++++++++++++------------ requirements_amd.txt | 12 ++++++------ requirements_amd_noavx2.txt | 8 ++++---- requirements_apple_intel.txt | 6 ++---- requirements_apple_silicon.txt | 10 ++++------ requirements_cpu_only.txt | 8 ++++---- requirements_cpu_only_noavx2.txt | 8 ++++---- requirements_noavx2.txt | 24 ++++++++++++------------ 8 files changed, 48 insertions(+), 52 deletions(-) diff --git a/requirements.txt b/requirements.txt index 7b88f62676..ea4de887cd 100644 --- a/requirements.txt +++ b/requirements.txt @@ -32,22 +32,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" diff --git a/requirements_amd.txt b/requirements_amd.txt index 5ab892208c..ddb6531fe5 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -31,14 +31,14 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.2+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.2+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.5+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.5+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64" diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 3b9e7ab752..90944e236a 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -31,10 +31,10 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # AMD wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 98545639ac..9e72cfd7c8 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -31,8 +31,6 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 2cecee5a5b..c487ce5064 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -31,10 +31,8 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.2-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10" https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index a847838187..51280779dd 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -31,7 +31,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index b3d54717ec..f416b0cb82 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -31,7 +31,7 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 96437260ed..48f1bf74fd 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -32,22 +32,22 @@ sse-starlette==1.6.5 tiktoken # llama-cpp-python (CPU only, no AVX2) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.2+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # llama-cpp-python (CUDA, no tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.2+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # llama-cpp-python (CUDA, tensor cores) -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.2+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # CUDA wheels https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" From 25c640ec0ced52b6fba316ba3582a757537b04d6 Mon Sep 17 00:00:00 2001 From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com> Date: Mon, 16 Dec 2024 18:58:50 -0300 Subject: [PATCH 0092/1701] Update accelerate requirement from ==1.1.* to ==1.2.* (#6583) --- requirements.txt | 2 +- requirements_amd.txt | 2 +- requirements_amd_noavx2.txt | 2 +- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_cpu_only.txt | 2 +- requirements_cpu_only_noavx2.txt | 2 +- requirements_noavx2.txt | 2 +- requirements_nowheels.txt | 2 +- 9 files changed, 9 insertions(+), 9 deletions(-) diff --git a/requirements.txt b/requirements.txt index ea4de887cd..663edb5794 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,4 +1,4 @@ -accelerate==1.1.* +accelerate==1.2.* bitsandbytes==0.44.* colorama datasets diff --git a/requirements_amd.txt b/requirements_amd.txt index ddb6531fe5..dd537a66ca 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -1,4 +1,4 @@ -accelerate==1.1.* +accelerate==1.2.* colorama datasets einops diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 90944e236a..275bbb22f1 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==1.1.* +accelerate==1.2.* colorama datasets einops diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index 9e72cfd7c8..daeab92a2c 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -1,4 +1,4 @@ -accelerate==1.1.* +accelerate==1.2.* colorama datasets einops diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index c487ce5064..99b5f3c244 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -1,4 +1,4 @@ -accelerate==1.1.* +accelerate==1.2.* colorama datasets einops diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 51280779dd..446a47cefd 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -1,4 +1,4 @@ -accelerate==1.1.* +accelerate==1.2.* colorama datasets einops diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index f416b0cb82..73b9f764c1 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==1.1.* +accelerate==1.2.* colorama datasets einops diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 48f1bf74fd..0b811b0fd9 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -1,4 +1,4 @@ -accelerate==1.1.* +accelerate==1.2.* bitsandbytes==0.44.* colorama datasets diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index 28144afec0..f6ba93bdfd 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -1,4 +1,4 @@ -accelerate==1.1.* +accelerate==1.2.* colorama datasets einops From dc56fcff125fa0fff2be7e5c046bac51c24a534c Mon Sep 17 00:00:00 2001 From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com> Date: Mon, 16 Dec 2024 19:48:51 -0300 Subject: [PATCH 0093/1701] Update bitsandbytes requirement from ==0.44.* to ==0.45.* (#6584) --- requirements.txt | 2 +- requirements_noavx2.txt | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/requirements.txt b/requirements.txt index 663edb5794..4e976d8eb3 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,5 +1,5 @@ accelerate==1.2.* -bitsandbytes==0.44.* +bitsandbytes==0.45.* colorama datasets einops diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 0b811b0fd9..5ea4f65416 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -1,5 +1,5 @@ accelerate==1.2.* -bitsandbytes==0.44.* +bitsandbytes==0.45.* colorama datasets einops From d76961859141762a067605447d0e389b96bf5fee Mon Sep 17 00:00:00 2001 From: oobabooga Date: Tue, 17 Dec 2024 00:47:41 -0300 Subject: [PATCH 0094/1701] Improved UI (#6575) --- .../Inter-Italic-VariableFont_opsz,wght.ttf | Bin 0 -> 904532 bytes css/Inter/Inter-VariableFont_opsz,wght.ttf | Bin 0 -> 874708 bytes css/chat_style-cai-chat-square.css | 2 +- css/chat_style-cai-chat.css | 2 +- css/html_instruct_style.css | 103 +-- css/main.css | 585 ++++++++++++++---- js/main.js | 244 +++++++- js/show_controls.js | 4 +- modules/chat.py | 24 +- modules/shared.py | 3 +- modules/ui.py | 47 +- modules/ui_chat.py | 66 +- modules/ui_default.py | 12 +- modules/ui_notebook.py | 12 +- requirements.txt | 2 +- requirements_amd.txt | 2 +- requirements_amd_noavx2.txt | 2 +- requirements_apple_intel.txt | 2 +- requirements_apple_silicon.txt | 2 +- requirements_cpu_only.txt | 2 +- requirements_cpu_only_noavx2.txt | 2 +- requirements_noavx2.txt | 2 +- requirements_nowheels.txt | 2 +- 23 files changed, 876 insertions(+), 246 deletions(-) create mode 100644 css/Inter/Inter-Italic-VariableFont_opsz,wght.ttf create mode 100644 css/Inter/Inter-VariableFont_opsz,wght.ttf diff --git a/css/Inter/Inter-Italic-VariableFont_opsz,wght.ttf b/css/Inter/Inter-Italic-VariableFont_opsz,wght.ttf new file mode 100644 index 0000000000000000000000000000000000000000..43ed4f5ee6cb01173b448af26edb9d7459f9d365 GIT binary patch literal 904532 zcmd>{cU)9Q|L?!2?AZb=#ia_kq9S(1uBh01H)>3x(O47PV~M7lX8MzuW@3yn8jU8# zSYqrQ3o0lAvUJ#`vvgS6?wBZ>?RoWbJb4!?zLtNyJy4;^eoFh#$dAoGg<& zTymK}Z;>nggOtqI;ve%PM0v^Xg$~)+;A%Wuy>{f?h!R4s2q11pMU3Lroi3R;A&fm{L;~u z0%*!bRD^J!kb+3jTa1zOX$Ny@6CXn#^53CXNFm^Mp2RHTWiR?(b|eciW9HAq$4%Cu35YOMbaXSp$CYtY1#4(gx*FuNPy)) zffYuU7@2#SsZpBH_)k9Lm$+n^K+pOcIxM#M>Ofoa=9T+o#n^M z2wOud^jW`Q-5M4GcRgHwqRA{`f+MRRX8`J1jX z*&O~T1pUUb6d+m8i1HP7kDS_c=bk4ikUos-zK7Wbhf=cs&=E!qbtRs)CpVNFZOGyJ ziVX`_qFs0jq3$$@#?TC!PtVfx^alNtcF<4sJ6)kzN})WYh5B*B9IfnV0WOk_J4ZNi z!yP#^g^wWEt22Bg@^02gk>5joG)l?BaR)bs^_o*3OXxMHK92Po%8z<-<5{oq=oRO+ z2^@OPsZV4*=hP>)h%;HxKMbGZuoTu}NCt76{zVEBOJ7vO`-xKP)H%{9MW@b_t5oOI z1?n!{a_S=aNynVJg8E4>J9P=+i=DcXW=S4ST}3L1^##jIO?pLvQ+I)W)2X|*(A~HY zalbQMOZ~*PPTik;#Cc6RD&Z%NaOxbniv66r4srBO-J1eMXI(fS>MW!5H!>I>A@9xxFQ-Gjz>VXu%H#qem^5UI!;eyG7H#o!FApE*h*Hc%%Ij^?l z$G_nW51|nLX{X+fy7S>qy*&l;K2E&@g>vOiJ(Rq-45!`^`W>epM&Vp@`hCfVd&?Q# zk3zUbPQ5?*aTA>SfJc3bc_4u6=?ou4!JL;cgRb?$k#>|In$Agx*}vQP7?C!;OZ%!5L=^^p#G1EXL7Xr#_A}G{vcp z$9Rl%>Ju=3_HpVHF*e&f^+`x4(5X+x*wi@nDbR>UAdiLg3_VAqX*2R(2j2)kt!hJPbMXFy*Djis`VCLwegEk`S3sNFiW*g}V>ef_8J*`KJvUXqV zXpNERw^h(te{Mj?2KsZzQiLo-j7KSqMZ8T8|8i*Sacv{g8{_ED4b%;3tZ}5i0XeNh zX_nAq@UL~0tDB=d>!ItB!XMLN91ngInOKc&bhIn08PnIG9hRVV z{&gydn#agr}B{m+`;YufrOjgg;j@E2h2R(&4O2uqt71CT!|1PITEz3sy zGOz)6%t0>ABmK{#lC?K0#dzf1WWO!PB%2#Hptsgzz0m{znRT(AW4*%SH;r5t!tAO$ zLf0YJrV+9p(*drJj1BWDI?alKnAXHX>MQ7GgEkWTOkwkF392j+2G(je#Xj zcf{O)HqoPe|2r-8=ewAdtpeR4m!O?zIC^>&EP|D4IBL<Ej*c zYObSVA}GkLKq;w`w1hwkQ7gqQOo}>qh+? z|NG#H@C>dj!vB`#Xqj5eyJw((CxgK)V*fGkmUli%k?R%*}C<3O0g34L%o<4 zvi^G%Ur*BzcMaCeOHe*G=MF~;GaF-e%UU|@U;49Y3@t~zMX2vHh`9>=)7<_&kaP36 z>4cT4;8;!LK`rbDd04%E1%APmU$LU&I6v-v$Pc-%A-~}akSSa}WFwEe`7ZoY$mRT7 zkX!k!kniyaAP@1sLLTLhL7wEVLEaDs5+_6mV<5)~vmqZB=0QFwJPA2p*a*2vxC(hq z$bifi@*xX_LdasF7_vfCV128^3YQaIMK?%y(F@XB^n(l#yFhjmyFvC4dqegWr$9~< zr$NpTXF@(EB42T?xBzmYxEb6n}*LS-b*yUAzf-OS}siC&od>iaDjAFctCn7P(p>T!WYtC5eONq&_jkP!XP^< zIzx6>bcgJz=mj}IF#vLqVi4q5#W={x3e-aJs^WFXcNFhJ)+*|VlXwaH3`vsOLbjJe zA;TrqTIw$igB&i+fP74v1NjVg44kw=dKPk(vk_PO5Z_#FMSWWSNZ|+ zC+R21!_r~MW76-C=cIFxH>JBINZC>@amv2RzK~Ot%ORgtJ_or@xejuJ5@!?250v{M z4=4{p9#I~FJf=Jb8Ldo$G$^r(SLUcj6Q>%dLXN7Fs#B0>RhMv1a8(6sQQcDAf=pJy z22=$q>;P1isw(I;DzvkzUeyR`!?uc3^J?sR)UImme$^gow7J?x?E~qr_J?e(Mw_dH z)xnS<>JZ3KHQGuYrVfMbtVa8*`>6Xsj#E#BoUMjUs-ILp4Y^3Y2=ZAqY+L=28ttRr zr-p^7e^O&~sE?^pBlYj<3y@dT*~Do&YbKGPnW~u$xm0r+@{Z;Xq(PI6s~MUM;x#3j z5`>g$N)b|~DTCgi!CptxsId|U!-0jns9j*;E}kx)keysQId-0GU&+o9*dE_oik5Ck zx1~E$4BG4*@>-@@u34dZR`Z-@rDm(0XVa0A3)U3 z3a)?y;8oCvDC9Iz+bZw{ksfC&dYrVi=|~iO0&K&f1Wp?y#Fs|kROKhk7bq9@W$awU z6K4i3PYt@s`{e%g1jWnH5QAwE@;WM&U!*hgQ2BXi|A2M|XA1@jCW9O!M{+0SujJR{ zLU|%xmb2w&a35iBL~?tka5!bKmpOJ7Bpc;d$)7UmYkQsi3U!jdGh%ZALPN1ny2+=k+4>sW`9 zNLxoj31=b09LfWFCsOW@IQQx8k@5h!8+M6r^_2(90@c&^ljRZi46Z%B-&r1Mzsb#_ zgM;Le@<7^6?@e$h-!o;b97-qY?Y0i(EWI5hk3+~QdMijCFZaP=%$x4=M0*tX7=7IZ z$~Epa+7=>Dk~L`M-5upwvP4<*i60atr=fS*QlMQ`ckEwY;OX-^;dU0Fc|^l1ku61w31 zKzYBsAN~C$Ohk@CSN$|iPLk=E)1cg=9dh1n(%`-Uv^$ty zlr!lodE;*Cjptko>qNJG&dqwEuRR-2|aI7B@?vG_7hB5{?5W(;bzX4GxLNBIXyrF4&FU>rkzb(K>_geaSw{Aa9U|(ZPX(<;}2$ z?=~pl4xs3jR(fClg|a`pg#9@?y=*#1V`q<}xUdgYD)~9O^@(DZ{2b@?iD!4Y12<=z z$y*-6J-#j3TYiiiv90G&*^?XgOqY((Y4BLe>wBS}ypeLh7dpwy=-QqN zUVehs>@)S1*VFgESbX8^Va_HxdbFsDIme2s?3LVz6M<1^;>6Sb7v$A+@&ftDld1G_ ziV42c(UnSh1n~yU*c7O!MTc2} zBg}Y{AmBeHeDulFm{)c~&jDL7csJ4(481M%ru+&DyIo#^%AJrWVh%r#$!8?p#C(Nb z#`M*n8e|DWyT3i!9?k8rFSRe_zOgT{FX67*huDYE1oXTD7J#Xq$)OI!MC`(;flirKNfV;8#j!*Q#E5+FI~*R3{6{NcLIPI3E= zH*$1!|K)}%syn#d*M;6W_>1pI`tE3qtfJixnR(`;&hGT)_52)}KECm7u9a@wI3lPi z1yV)7-MDObqjzs4X}i$-(Vtf5P(t+A_1&lnvJIz(BsvzoD^*F?ASL=V`nG#V`t??H zqlUKK-ddAR$8Y~uucVJ+zYLmBAKv@A=qvgvE=FigpWja{FQG5u-x3Pw>-cvI6X~4? zZ{>VXpBPTny2+~z#;`z2PyN6aK#8fJS{vzD>W(5m+Ml`y@@(oJgv6!pQ^eC7X+IQ& zQcT)GEihIj4U=hg)s|{XXs}mT(P`^3PEAMc z=g@jP)c5l5&x`Mg|UVDoyOR(F*^&xykRR~f(aHHi-<_UZ67=1rF{YsJqm+GhdwzOWm%JyYt1x9xOgAPrxQFeD+2|g& z|G?v|-1Z+_Z|}DM;3j*g{pZ%(JM2HVSq|NQbDSl3|Lv9Cy6=yRsO|jY!40-fKOUcv z>h&XUL2z={qgSdPCb5PR}bubMyEb-a;dfTfm3s=?S8^-oqvj%7INX|>CuFyDtuDfAn5Qw}@lLpwNQjJ2%SrjvqKRAPiz) zCnmZb|7Bj#o#Q7*D*GM3I)dwY{OU-En3rh%@ygN@ce=R^K5?&8+u{?c z{l%OU#?BS|iOQj6BTv*0EggBHabk)6gk3rs`TGyEI%od=;|uu#zh4jY>HGVQQ18LN z=Zq?`{%#pmJnH0)88xmaV`hfzIhj6E=ymFRh<3~=7IrFrl=r<;hAE|Wr&7G!dYnq{ zuhE~%=&#nF${A&DI8`*Ga-G+talJ z6g|&;^`dY5nVa3cx}CY(S=;N3VM>YROvbUW@H0ibAcD8{hIm(>#rM~u}^JQ)8x}UEf%7tBs>aOj4A*z?Q+l8B> z^5Z_tT?Se6!&%2Q4@9K4-aF&#Rp`?vH@Iq-D8^kO{{>9&>tJ+*V`9ed` z#e{y^+>6D*r7jmsTEkm1jW4)Z@(iDUk?U{mdeJS{=5Z;;uIPIyWtP!=39XiSDXX6< z^OAXdR^LnJ30ZwEl?K)aUb2K2PrCg380o>~><*3om$T;x*_X5D2|1Tdfz}R}3+JhG zE|&+_`d==eDd$|S=vW~dYZ^3Ip7qp}BH76;XbUHS5b!68>tW@Q&$DHtfU zyHYriZ+FEUU~O}yrhQeXEA>n?w6E-RMV2}4YW&DX&#NheJW{W+?!H>kS8ji`u)V$N zYHh2crYc-uMq1S4ex2|`M=QV3PbLcg>hI6^5jj*^}zuu1zxqiJr-}d^= zk!|l>PwcKKyeZN6cgS8uxk18IAs zv~?xKxO&^ooE|FE&DQd$wZCX&99Bq2K)(~B2 zZI}{W*r6dPy12a}C%QbiJSe&%xGX5Tc5Fr9t@t5sez%g^XgqIadDMs9%IPXq-zwKt zi?@{3y3*S*%iYD>@nPC_x8nzRW!}#7wU^&6Y^@2rZ5@_9^tM|^p~IbDRw}#Qx$4!c z_nljzAsz0-trWW5N$_oGdndDxBKJ$GjBcfHVi9w zjcFKOu8CLW^A znSeH=#3h#5Mr@BWW#M;EjIIp5dtshUe>b^bdBEMww5pDG^SxC5cMAqu^>>T)nxea9 zqf6CyYkKB*+^rp5t-fm?TBN?~>SAes?{p6-?%ur-)m`ox^sb}sC2#Z(x|h2UqGVV; zf3IXjsdTSoWU1m_$*5BCUJdTOS2MO+dC!Uj?pcSWLW?X?-g8w}cDTo^B<`#J9S6mI z_nfzD+`&mzZQ@S8Q0*I+(WfdX4wj>eGj%TTiZgX7@Qf?e+biOVrxm%zm9$0_7?LW^ zIygxcXX{?1ifbF4;d6iQGEcYrsSC;j?x((5?s`9Mh}r9Yp%7Ygzpz(m-u>c$TE+d+ zz$%~nwIi$4@h4xX^o~C@L@0>A-%B>eC-&r9$EQY^z2dV56??}U2Nd|km$XI{X1nps zcHPWRxYgy>1`^$Bs25D(gv`X^)#EcQ;w?#HPUvPTwqCYYEvr(eEL zLe9VTI{V)Que-9%<=fSnH9+eNSP4K9Bz)bXkVMu}dgVZ(ws$g8jJ@!Pc|6MW5NwaO@zi^Y|RUXf)TZDhT>_t8bhg~x~-vX zxKeK@n_#pX$|f=~$!Imyw=%UdG>p&I7__E*hLHtFS$l?&+_MQQ7dV{L8LjI;fEWMo`jVROld zT4wdkxc+QiZN`nMVo}D8X<}gpv*!$E&lz#gTB|Z1bT9DBNEw}`%t#wj?3s}_B+osg z)U(<EYfA>Pb}1B76<0} zXO?)@cxRThl6^AkCg!of2OuJ>&7yOn3WKT zJ#xLXl6&RqvXa?_*8lvvLO)w#v%uUC=r!Z!9k456<<>${(7e%_;~g^vf#H=k?1f9FIH- zryvhA3o8jKfS8?Z&#LQNsLb-nKonmKUz)|nR@P`u%A%|>A7*)EpL?>^&Fss|tqs|i zS6J(_Z@yMvoy`iKo!q*{JKHok*F8J8M{Zy?D@Jxc6GcHex@@y|Zcw(_H@9`R*+17m zyQEcao9t>e!m8cit?q8tW!H|uy>)Vq&Uoh8x^m;0RdrRy^FcLMG__)*9Oz%|1E)#?tP{vQo;?8|x%^8(i}>#zxmXjnT^Ef|#4%!>D@l z&P}6Raejd5#9CXW>BM?ljVXF%on(qxUZgU`FfDdvtz?QD#2HP==n7Nvz&t-wT5zSO zDWiR^n<-15=Vi)jhYO~UYHbQO$*LU87RsEM71dHs>X;NoPU_ec zF(+eiEuWL;ksFv(F(KQUQ=`rf%#jt@9y#i{H*VyF6`mZQdwpJ(S8iIZ$tAbcJx9zf z^~zD^R<|?ga=D5e-(0_T3A$W;x%BqVol*Q)@j6(jtpa7i9I6a|=wZ()$;f zeAD|CHjlN%ke! z%+qv6bII`$a_^F~u8k!n<-P`MNx7e)u|%jfxRhSqR9IJfb#q~DX>4aNvDD;|Qd?>Y zGIT7>afMc?eo$MM?q_XPmKBoPp)A|?LAUY;y>qL|Gg}u3Au_C2eUW|mY{XX+B`tMgmeYO9OfD%`3|T<%v@ zv)ig+2hiN&>Z>bNxKOJSE33JN%FflQ+62#o@x!2_TLa^Iv*c~-BWiBo? zH8n0SmMG1bFiWf@0zx+?++y^)Z?+h{@0VF}ePW9(`Mw3RrO4Y*W-+5rEZCLnAaoW> zx%O_Er9vB9W~m-}w%CHsGXjmqVnLr92_T6DNtdt+9;wKlbFwM%VL zK)Fk8ndV-7ZG~HrTx%<86;i9rb5Yf$wk>R|D{=KGt*fe1y47*Dr9pKrm2I^3_j^dL z^|g(S?dn}BgWVd~=C+}REAww)^=+uv2o()%n+p+8-yrC3yEd?GZsXC>oT~BqSgvQ| zjWvPAjWOXiFYECo?w-~gYg!jsvjd__tZWZ!t?hV6VdZ+dCtHPt;Cd?#OY5x~`KH?H zQW_X+JHABgVLLlSR@&|jlY7`Q^;#dBIpDU*RxaHvv(+2?ax$mt!do>d^zR`Jh%a~tH_Qn$TO1sT&3AS@g z@b;QuIlUv!8OzZUa>ei}QLb)vJzuU4vMJ^Idd^L*cjM~ihRWWha$||MLYCdR3K@Og zMizV&jk2;qZs2|$+JWN=f^L|&619s5SE?=LxpEhcCs!Vzv2itAl^<6lR<-6V6&ej^ zsnn=BOO-~&S*kTk&Z0)-TA!;Xt|7mJAJ@>(tC)!ruCWeL8m$_UvlXiSIGa^~Cf2v+ zL|a*FPFw1t zG`aDOb!sKwXjKcmtx)9)q4wi#jqOF=W^2dub{j5eOEs-{e=l_zA8bes<-?6}UP9E^ zD>=fA$4>t&qy+jF3Tf65UZ~4gss)ycVB>`VW^Kzsgz#c@iON;1ml~_Z`W}~TqEKg3h*AxQPlcLPO0j#D zwNde4L0!EfX+cAcg59R5E>XHD#D+$tl=_6lB5}L|ouCjJlpHVGRlhz~5};~O@b#)* zWf~640TZt-YIinCU8~ObP}gem-PH>J+8WK?Crevt_AV&((d;*dhiQyDa}>r;7$!oE z1wx0hC$~GpX*g?+heoNaFL8`EjV9P))VL|D8#Owt*EN_iEgGF$WtAqF4ao5Qu{Sm0 z;n4;x;=|1@ag)x*yWC$JQ|FTTba|OeL!H*@;^J9TaE8I#a+%9*k9Cf=|%N~$gzRgLywr`sUMeY|SRR_2$YZ@xu!?cYB9_I$zNZ_qR~THQ(pnmQ0bv_Gv|BsZ<|o)d1^>q0fbQC86R4&@>F#Bt-+JU^9*aS-18g| z5?|o8Z?VzEYu^&1#w$LwPT8v7Pnp$x_or;9#S% zO?r5dv5k9{pHiP5UYMF}*BnjG0h7i!nmSp0^saF2Y=1;^J=Esf)0fEre-2tRX+tzo9a`q@f|) z3`(^&-+V3aSKyWG(n^Pi0iFb;!`3j^VlanW48{m^lo20CVmY!8aO35}&LfLA@4Kog zIbsYCJ~epUxKkU)jXPrq)|@epi8y<9^yu>gZJo~#vUR;AT;ZY&8e%n#vP5X2KkVN= z_HOv#yVbpX5+qw?;+tXJlF~bMGwcoPYcPlD3_4?MT8hq|btlUs>yCa_R+dI@EX)Zu z)ddCReI5{)mla^mE6wyMd{eigup%_SxVV$IITl065~efP=`@vUZ7txc`=6<-DUkCp zP}stQ8a|iiHspEg8}mHc*lbm8Y&uNg6+GEdKD*k7qr{|Al6A=>73#>WcJK0C=D~4 z4Rpke*{!m00?RD8{#Fkk>*+b!6pzn97{g-89A>2QNMa^PurX@DC!i6Ng5e)ywkS?_ z6D1p2S!EnwTwvDbu<+RZ!DH@sxmTX5jkue2CZNt}jwmwXyww;StjWn3Tc3A%O^u_1 z1~XY;g%%^aU1v1%zYcAW!#c-Qq6yZ;>I`Gdp#hn8!j~|bU)a67dGiMwH||)vbmtk( z&c(rdpL*(tC!hRjw&94ujB$Kqn(oN7X-B6{{e8lOQ{%=(MHpjs#=8bhf=dw=u?DnX zm?hb0NaMJykTA?yC{b?hk43yttnn+>`I#}ASoi2kU8M?7-)h%_8jaRsFkoEPs?{=w zS5>a$7?hk#5t_oax}g;e$bd%#Y+8qZ0UQ_PYHptv7G|_`N-Y?2)c{KwS6R7Y`*!N! ztOe^gVd%RC+Pn9uz4R2u-`SE@XKC~$Vdw+uY%qiwF#as&ybAXk(l(O9di{>oCfZU2 znY3IX&ctO-lKFB04`p7k^Gd?2Prj91XeEnEe59bURFS!1rbTIMC! zc432uj2=7fsl|Bd>v$2n_iU1}M%HM9yY~toJ^d+srqE%u?|ty$!L!${-%m`>Fy$2# znoDXMtaeGIc5!w0)cFSKL)vxk(|6G5G1KQfwS2=y6c29$Bk{lGzvgcrHH9@@WwJT^ zQ3zfpM$(Vs5x8&5N&n9|K}rRX;weWeqGTxS1MEb5}^Npl$Dz5lE9VfiVFDMso-g;F0% zR<5QTG;l51CinnM#yZ$fbI*oFkmPd;Ym)a|8(H?(~e;}R%iRd2;=5MJxu2-U5 zl_Ub@{tMDTdX0tSds&WaN+ZotIpZnaC`LJ%3YD`dS+$LFR0rU0hr0{)s^D4yPvs77 z2aCtj`0wJfID9DTJE_TKIzDgPjSrO{bLhq<4|D(22WfByh>U@j}+Je%*8ZgSquaTFQ~{aXZ5 zC_ZIB@&7R+TIR1DNgZ1v=F{+pWYCC*It$N}t>R~xlRhIar_1#JntWVqhSeLo)RE>Z zUTzWId>yXpBk-Qn6&!Im@f)}fWaA^fIqW~~ZsH5&5xiBN$$xLJ<3E-=^H0HrF0Fw3 zI$+nCi#XC*d8SxE{(p){me5n~#v_fU z>+%kMEz{+EC*$DyI-c*(P$++xy6|i8UG16FOK6X0K6~!>kyH6g4m}KhAioFZvh;+0 z)LO~Ya3{~gWiD;im!$Q{IU7#Gfd`mbE+XBw_(g?KTr(TB)(@p?=De^_UE z{h1h);bD3He*Q`m#%&x;RJX@H`IrZ?0CWEUy9O}U1^h+^@DN1#M{tk9U!k@9Wwg5_ z2^* zjFCUWH;3StOZe}pweuRg9`s0;Z{zw7hyQhYoPV7<1LgwR?$jqjKLlQ4e%QsDhtxnn zCfo3=_IMc9^7{V?o#pXo&eB@gJ~T=E9(J@| zPQse)L&b3F$@VB%lXO;UXp_=~DkMGWr8?~EWULp5VZAsGYlb(mPQ_@P)^hz--X+(xQccCZ0rkm&^^T|z}y`)SLs7We4}lS5^<&X zu@7Z?yk6Mf6%k&#K~JOws$jUGayGujH%FBLJ&FFI%%*kt7Ts_~4eX_cW-9hjEWS@? z6c5l?#Wbu>|1Th3lWh)LJA`>+BK30c9z{CbNi^j)N`WD2;%vKQ4YFpK@?oJ|%TVd4c#lNWk^yoI3NL zK>BBCw&FFcO*>-_+6rzog=38ELOFIZSNZ~JusmMC{bBUDA{Tlo@&U?aa;&l|{GU*N z=@Q)K@@?@k8lhYO{Zq7A1dUgW#d!IE+{7C4Mw*|oeH+%zui~2oD@7xD3BmHK;$=Kv zzoQVvY&^3b!ycq7{ib%qIG#?66lZ9PayNysxQ??n4_YpTAuc=rX+9H+$DEJ7ol6@E zkuoU~<9M#(9u2~`O6MqIVc!Yzxrb+IF0@t{LC3^wJWCmD?&wBS#q%^v%%tbVDwOLz zT7>WkVl^$nw|>_O;gpK<&veAwhj>5G5$qMe0&jz-!3r=9Yz6Ou-@$V52562q6Z%xA zzlIN$6LGFG3`Bwj%rEFES7;-f;LZ~C4)<%JE&5zfH3+YQyBOijJ&5pXMNe)DVELa0 z&w;g0euT^T8Jq#zobDR92LZ~-jfI=gq|-Za8CRTKhnolvIo)^RMuFYnCdy0A;LL8Y z=uOx7$!NnJG!cU70$+r>7h{Yw%yI=oc#b|5 zo|Vg$Cy_=m#Uf5F++{qzlZ|V8glF;NWG88Z!jmEtcktUlM)@#*kmm87s1Mf23BnQj zMC5Uf+>@55;4&~S>Oq~jh3gLV=;JpqcAh|3DdO)zd@sb`gt%U~&hP|d!9c|Q25~n) zSMtj!-N90>y>JF3QMwZ~!T~o(Na6-@(R4ugfDR}o(*dPB9pIvI=>Gw}{4^Pd^<|ud zI1)cuE@W*bb$lak}jKr{HsN75oOia1sZXkpN8Kg45j&_ZKI*!hZQ(d`thLP(&MqYZ$LQ z9mCn(HN`CYNI9L(H6!Lwh8RoVE54u%MIPpieAY&CK821bF_-9bg=aL6XLzmS2m3q(i97{Q^a0=-vA?w9ce)Zt!=EBQ z!G)CY2f*J~@TbQFElm>wXcC7p&za?TE{VEwNpdP&51>W*uQ8CvHRN#?d2Ga26Sp9b z*OC594CL_@(tH)!_3Uj0i+9IXF{|jwYzC>HF9(+lPhJOovB~{_N0s1C{e+>V7SX<&6 zd$uT3x&E48x&Cf%a{V#hXQ?DQtGo`9>5%dS>~cTt7V5b+{Ab)?zBA`5p5?xg=5yZ& zQ!!>-xTl49ZjPX#M&T=J#QebVi|D8@mD4Dm;lANJ$o1%pXnY6r0uk*d$FB@>_@zN^ z4tSa?lXHlyF-#el*6&VTxrgI5XE@$_BgZ2w7uR#8zH|Za$3B<( za-nd)W%p>Q>0ahahp>)6L=UizSPl10{8mFUsKfmQh+Ba43z~JL?}_w1k-jIws&SoB zfOH&jnTtG*Ag{4-FCgy}Wh<&w`fyLH@OcQ852>-ve?$2gjdJ=`5B+^)lX4mzcFbMV zXc2VA;$~FKX-f-USA9rP4!Tnzo@MzCp5#<2#xOEGgrjrP2u_ej&~<4z-I3sS zAXN{Vk2;h#qdr4-zFJJ`&PifA z?Q#hwF9)vj7k_Y@!~cJAS^fV%#Ie?8?=hY4veJ1cjbk+(QQb)H4Db7vq|JIG1e~!-#FdJg$EbP38?bjUk!p?VaHsd(k!dVVG zw_*K*Hr~Yc$`AL+tbMRQWqVn+cV%ZuI7f(V&YSOxH5twzn*GehSYd1Q*`QOii}m_! zz6*_H`x=Cu>~~n47g4?uz`KRJ6yduRbGW_D2(NQ4c=GNA_-zvh-Qfb9k)GxD zauuA1w1C^o-XpMgDrmbZmN&Z}W!j2$(RV@wCj#vCxm_SlJOcL!+-Cn-_|I}_SXX=k zX5e#{PcZ(*{BZ?+BOo6(e_VCWAEVJ%*y&;IiQm7%FKsjKf;jl^f;i5O{pu~sRYcNU z&0X{nK3jrc#+uAr^wT7TE8f*wXqneVNTeiUW4`XH?)p7(mLT=+5tP9;W#H{XQfTJIM#M- zPG<8co5R_B$Y+Z{o~TkKgy2%jSROYv_}B8m{hy`xju}jWb>a&NnsR(g%+7 zlIHuJ^Cde6MV@om_+)3Vj`P(P;}!K?fU{P|7#;KeX2;Bw}qoLhJ@V&f(ZOl4E>ho@H&~Ja2!Goinkq?>M_cpa08T!18FG zFI&tBjyddYoO547+u#|iNI-d69pSP*XYVq^WAu_^9(k3Xa=2p<#_YidXIeUX5_4jt z;ud|0K6@U0(UG11I?i2@C+|G_bmZ+kbChtF=vd=posT~ImbD-I%#XAV?f5gwz@AY* zQ>;3UVqI#Xe}gdASKr8K>dTl1x5Dpy7PIFlA42^EBgzwx_CwmRt9KOF=^fQV`b;we zbM{mgCL5%wa*ire&SCEWVW)&KI)gn^uooYqh@m~wHu{up3@`?~uphjFakX6d0ej7F zkfr1#`loDcMpw#c`{8C(Fy=^=z2!n@5Eg+!duWWd&e z@SQwg*hOjVy)GZZ%|-eK*s~Gsm4Y_-3M_@bh|k3yFN-9;6YZ4%pY6HwMme0$8(9ms$--p#j{cMCh^SYf+- z2m4+F(pfkuV2{R}aXJ~Q4 zVz?_Qjvs^b6aYK-;7&qRrPsRQc&!%qxK1@m|8;4IkQzJMEEbzbDggq9| zhkC&FsU?Vl{2O&B2f`bFLpe4>XB+@%A8P`_(Kc2+(!oC~t9Kn>_po>yz{`kp4i^ss zRK_|9{(Vrt0xUjg%trlh;@TDXk*5`TacD~p{lQ#PBWxpB20llexqu}X2|x0)W&u0W z!LQ;qvUCb??H#xSKsa=4pKW6S(sT5~NVsPJ%3!^Yd#*KwGo7X5u#H}D5zh+yVEx{R z_Os%XtqrVBb;3Pf;4`^C%uZozDx7PE@|k$=(UCGh7~hfW3VXdGe8qVQU(;O{k9&^p z2=SQXn{59e_Z<6OJ?o2n1Q$tiby+xV0GXi@1MdU(Ghf`UdA9qxmm6KR!&3#;=Oa6MD-(L%+k1k)!!B z_M5yQpXA5d(H{*-u%~gvSy!SR_v7rZ7-wf_bE^veG_VV8jo%7wjDZ^kK4&(J^~)|JkRXZCShD)>{>6wWqtYgL<)luvZ3tZ;-gz0CNVn z73>1P1B^Q^4<|mTHy;f8fhk}qcolpJ4uk6;16XhZs0E>5D3}dagRNjcxBwo2QalkP zfOuj*fH5g90-M1n-~hM;44@oO74(xr4+elK;4p{+3BZhBniN3@hy+iAEnqje1hTN= zAm9gP0hB|Dawt&_WgbqoFs4N|@n9i%0ib<+(Qdw}0PW+4_VH^627{U4Ya)N#>yLZ= zaj*YofVT8U`TK=%?2YG-n5D8G%pw(b2*adzEu^mUOlZW}EK1w+AXfVgcCx6M%y4YENkk=`A21n4V0`bv+!(xb2R-+`0h zF31Nmex=nH^Z;YQ0i9bN8XN`D0A=fl>tQHcSSK(XU`%#e4sL)Yx zI`|qK1<@cI)Dw001f9Te@Hlt@Yy$_uWsnRih=qqrUXfWzOcmkLQekO_t1{gaL7&{Rd4-ps-5g0oW z7&{TyK?cCs8G^AhBoqt<7&}8&gRNi}_#MQ8JYXjp>I1riFjyVEu0?cD55{>l(u=}yF`>{`h&EOM&cw-T7EQ?o1G)@cJ zgFyf`F%C8`?hUX5zy`*_2F4jd1JQWczNTY8t@L- zPBaO5ObP{ui6)~glTnt*xNq_SqA8sK+GHxqFcmgF6*fM#o@g3uWg4zc8x9tNe}LCP z4bgNr@EiW-*K$xoH1irr2e>v1*JgD9L%?HT713j`)5kCw%vOOgfP81e7UoO`D*)1+ zgE2VgS8xMlf?A@vsMlPiH5X~kMOt&81Bg2papxWdaiEasaSr%`9srI1IPBr^O#pR& z{6_$LcsvoHtn*fbc>EmpIDj%d84eKVNyK?F3FHGCewx}FbOW%7r;x||DaL1aX%l?oz~EinvP=cj-#-7T5#MfP0{TXc+<6-m>ms0)U+?dl7sJ(C*8w zgAAhOXxrr-0qVE>DX<=V0MM?>VVlb>L@O487r-`f5TK1#pdDA#5Is8-Am3;2gCe5m zcn|=>!Al?tWP@6wmB?!)`f=qy0K!(m23NrbSHT8X?FXpmss{jhuGWH3@C-oSt3L+& zh}Lukqrj739e5w0?rW}q6i`Vx4hA9M5WYSJyZQ&xTlYDDe;vxRE{$k?4}dzaze%)V zJdlYtz79~wji}?s%ODw45IwI3ZNWe=9iaTr#}I8oe{Pxx76RDWW>3%w3;-jAvZ z+JGToCV-qr9mY2civff;x+OrKkH&SQaorfCW2^%9fRo@JcuN%0lW1I40C|ni08r+} zF9r9BCLm4|)&PVtu`Fl|;D6#cun6n`$G~myooEu$GYPVpgs>)eBbpLS6q*V60LU&B zvJ1ueP{=NH1%Nz4&w@upVH!X@!jR^v1wbOvG{}A03~-BRdM~0GB|r#3-!KDlpJ@-= zK}mr8o7o*8ZZi?LnTY=^l*3slhqF))XF)!*VgO`0>jlwl732o}z+`|l&v62X+Z>el zaFn%hl=pCy_wecfVTB{CaL76wVTB{CaMay!_?=4t`7*ZAe9I+Y11Ne=2Mc6_CAcKX-mxT+!R&a!9Q7~u%E&!x& zF=VkAvRDjREUp1MfJAT$yd!MS0;54PK)x)U2EGz4L*6e#-Y-MmFT*{S^#@^K6*vHp z_sfu`Wgm%_qwFou3y`Mei@|Pi0^9{(h*ls@D+U6Dz2X|SVIXZQA=j1X0OX1wX%*zU z$`2sFR;>d`0P$Hp3hV>8*BZoq&1-ymv?)LxSc`bBLp;|Zp6e=smH_uzhx@EU*F2M7c>zY%p~QxQ-TK*pPKelyN*_5q#2 z9-=Lf@0K>8510%f+pRsp1aO{c8_LEul#OjD8{0M!ZFc}DYui!Qw#NXJ>mA-;2hq;l z0O{XV4y*)_(=NzqcQc|r@UsVg_u{&}y})J=PqeQifLtQ&0LoS*@*%Q5K-r2M2Nr={ z;5fKLw4VX^+g|~+2jjph5C`sn&qPs<0O^jZ2D$>2k*Fm^2arDpP#zCLJ_j#>XGDjl z0_6MQQs6pKG-MKe65In{iDEJV$SbB2XbBz>#g+yQ0pt`5ImIIESX>v2>tbKj;Kde&WLcWEF1#kY7C7 zkOb791jsre2-E^7TM5Ge%0R+K0KW;>z$>C7{fUk`1B8F{F%hZ{CAtDXu!-oHJ3u;) zjRbSR1`rFbf|o=|8ps2R1H>~4X-OIjBESxC6eNSUM91yHGNKb+U>4CyPf!Ll0?6}| z6Twoj7n}n3!8f8)nL$wif2Rh4d0;QN1l|#y&J2nIr1dnyIgM~muO&K@1so(g>kI0G zaRBl=3)!ASJv-MIApG-1KwS_9(2iX|oxjiotO5@J{9i;^7xRI#pf>0Xb`o964$%Hw zY6WobOSt|r|+sBB*<3~h)AWcuY0LbSl?)@|tT*b0(Z!n4IInw{D5#jOar*)!xnG|%b*nj()%$2Aif{>gJa-#fV6#LAO|Q0s({7-ar-l> z@%elJ90$1WGp_sm8EeY+AQvbMkk4OggVq3P{xTZO0L#EO5DiX)Wbhp5Sl@L9c>(VI zwHjy+kdCjo_AAo(brnE)F#a3(3R!-|eZS(qzw*t3wP3{eyB7!sW55EeNf!X+K{N0K zYv0p};mf;>27*K2J~5UHR0U&+VMUtpHN+%*$4N>ASHMH?ff&|lnd}1aEhKp;;j2`j z8t4M%0DR{N|LQU9hF~f-CaVZn#rL(;ZUAAacY&3dRuxPG8v(xLq$PoC;4#2Ahy=b` zgePN|@Bj6T=Igcz)CDDOjpFqbri8|IRX4;!#%Sj zjoA@)_NByfAg(zPpIqg^OmLW3?%bd)xJN8cArL{#O$CsJ+ZAH&H307IQ48RDj~m3W z2FUUb2MdUKB7Dzw0Dki22Z(RJYs9=z-n zyvu^4#C#C8&kztvtdJMjL#(h5XhO^v@$#KdtO)+EDS|MIOa>@}MJ^FTH^lsqZoldP z-{|s#eEg6He#iqqq}gvTNCb$RAM(KO8!><6fxjC-9{3{<{BaNeZUFh^EnCy7L?IqlO0R0=UT7v-m)keH)FC|t7 z@vHNUSY61rF4A8wGe92HgIoQ@#2TO;HHZR9;2J=FG=R(-Laq&6K><(#R0qvKH!uu@ zf`wo+K$&ZJ20*5b5Whz7-w5%-IvQ()ve5{6-3aF!eIVA@9YCIq7Xaj6la>H-ZL*43 zQ{-V&gxgF8o?rynO00P;VlB!Owwi%?;0}0AtW`k(|E+cdgx$I=u{Nk*Z4ho-l>N3< z0OZmR;k7Fbju2}P`L>TD)*&P41K_{I6Ji|$K`*ctARV1@f+hg@-3f8;oE@ONcZR*o zY+_v{kO|}k2)irN+qEY^UUXdo_JEV%9{5VE8~k>Ith-eLEx|xA4Xgo&z(w$sSa%8J z0D+(;=mUDP0$gH1arYA5DTt?=fno1%npX%!TtdC8EbZI@D^f2@`BQ!Ay^O4#ty9m5dP5b z#D--B#X(ch7fb;wz%4C zmjTMiM3j+<`9N9F81x1c0Pa0;7dQrz0n#%G>6wJ|OezWLgYIA)fZs`o=Op-@1izEs z6Pug?Al%6acQV4AjBqDU0?R-&!2KuV{!?)ODLH^Y_zfWKQwD=+U=26~&I8CP6mkm9 z2H-aoenZ=X;b0Ef1meIQ@R?YcBghYM|1gyEFr+^W_YYeF5N6m#@RZn8ggG@UC<3a1 z)?gsO{ihaS6yCXmtv$ug*@SfNl4d5PgP_O5p4Vse(Zh;TP!X1DY2nJ06;vPO3ECuM#!qIPq zqaOh7V(W?n7kJ&Jx@B8)y$;--LYM1pk}hXR{lCthb;pY&lG9Yj5y7v28s-6tV4H!DC`OkggpG z#CCQfw#$#$Zlq<83E=#mp#bsRI}AjF)5P{cHv5ndkw|;wXkz;d1IRQgC%`pPtBD=( z1<11l$fE;@%Yl!?4mtvq--ELN?#>WyMtAuj^vO0o%I^qf-rz6!tJ8+%YQKb7Q{3JF92=^GuMiS(jGz3fs z7&9dy50Wkcl(nR9#ExeMg+OJ1x_P_@Ks=8x1lz$8fUu8$Aa=qGJU~T&_?$$!JXs6u zC3eaOR01sl%KoX5;3TorD5Ixw{TY;*GlpzK`O3UIFr zH^Dn%7fk^7x>yP{0Np?cI0O*xr78er?-I(>r5ONWU)l#yHZDCUb~y_u3{Vy?!{6n> zU=o-Q;P>)AfN(CqCUzx1KzLW+?+W}~!TBo>h+UOHCJ+cZf`I_>y*d}HA$H9P%mGNB z!S5*Rzt<*q-3PP*C}YL$umayfu9myG;NMjj`B zB6drFq5$!{?FCQ{ZbKHg_kc&lQu2VNU^+kcyT<|Q;@!8z?jis0^#O}P0)PzfLoW9lf*BwZq!4?6GVq`T=mchiIARYu=m4$& zo!FxV#2zE9kHd*!`HTHg4y*t#h&^cpP7r(Q0wBkymx(*xp!3n>f}4b?R{JDhSBf*ar|aV0-E4?YlA1*ikM13ZDFUL~&O0?ok$ zupOKOZ-`^av7;6Cx(9v}ojK6%Wb zAHZ=Qlz}|(p9eC@V*%$t3V21_4Ki}W7n9u@g8|?mad(sv_d(z=aSz7yBQp%yJedCnrazOir22#m&{y!*fZq{oQ$0wkfIc4<+W> zY<;D_Q?6pP7zyp<4&b8d4UHKUrBA2e6u`~SgpKVNJTd@)}vLJj?& zs!DvZxg7kes=GO>Iiop;Ij8EN<~ICGS&TC{@A+>i$t=u@naucKz$}|J;|izTX4NqH zQodBmXXv8hPbFuUv&n@;Makc${L)P0sF%r^SN@A?!T)R3jJHtjrB-RGQ@ZVk%fDlG zl71-o#;0jA%2~|?%=yed=0akg;qo(7d7~6;Q-#G8DVRIqObe;o->5pGoK!<9_fJ%5 zlRaPXXC*uHhG{Nccz@pIKUDp3jZI08a#5QqA^)oUWhmhx=D;j1O3J1GL3NT#o6F0k zRS$C}b1pTHabAotT;wwUsQ4mY`2Ua!=BcVUH&i2=UYwWZ`TkZF<%TM3=wh(VZgWvH zh!8acUjr9Er~sR8*#2WB6*5=g!YLPT%Y`}ElvOHZ%4)7)4&c6jqdM{qyiS^`#cS~q zypG&KZkF!$<6Ip+gU?VM4gKFxv-m726Q5v$t&S#rlH$Zgj7xT`&A(gKUK|N>^4``#PNM5o662h zrP)%C|6RQlu1i(f|HbwG@^XA#DM-r6gVIznUJSYJ_(Mrmr7AqbkE37AXHre>!d>Jp zhWv0?jn6|P8Aye_YcYoY%$ zC6D~6{?U)@n5s+}|EgrKRK>kzC#VoPqs=tzf3JLHLpgCDlk#_phw(O|s?>_hJkn(1 zaeSs!M3{tx?|NxwH**%VyV*m{CM%NtKe!a(`$Q&T2G#jS-cG76+X*+3U8*e?k_+&W zyg5SMEc;4z<#FD zzsR@p0sjp(T6W~&oN;E7pmUQfvKdFesK&hJf2gkW^8d9`J-MB{ikC=J6?k8<;2-n# zA6?|U!dcGmApK`bwBhA=SE;R`r8#k1-M}&KH==@F^mhZrH}Dm3AB6COi6LGLcll4qGRnX-UPCMT1VoKwytuTFDGJL2c% zx_q@9AnNiM{!I$Oy+WiAnK_WmWX3=8575)x|LkfEYUNCV%o)s%W=9-JW(T>Fxr*7# zDb3_N%$3zFY8EH%WHwAFn<+U7JD~|7RDSZmqomPt8@ajYBDdxD_!g|*MgXfSMeyzU8H(aA1O>;DNU2cN(#Pn zh>UW;YkvOoE6hP=Z>JpQVuG2g3NB=kMPv{zW&H(Ts%C*}ZX&)!DrU+cD$2#AR&q)|4`3oI_vdFN zZ~mUY|2zLb{H0!nE2H=vxuM)Zt}i$JVM6!J7bo`jk!rVj1(g<+N*_8l8czddl03z7iK53vzk-QWiDd2hwhv{ul zW6TXO?#69ZGKJo@m0dx;>E!UE3@$I~;2o_yMXE9Xd5w-Xz zeod|?YKRW91I9P*Vm^PUx`+Xi#uv&}_#=LU$Meg)O{$U$@ko@>zI*{j!v4Zt@)c!K z{@e43@?buY&y9#nu;55&bz1PtJL4fQQ2SWD*2|4xKc-2qOj~Km6gYe zMPf3qF6YM>sRYj_<{Kk-FwtrU#2&3|kqLH6RU_;=K&0T?R{kiF%a+#O}4 zA4rzH#czD22ol}o>ihx5zYJqlqr_x0%8rA?|A0Dbto%4-F_)B!noF9qJK32_$o}RM z=4@D5=_|SN%2Gzv(aBj=uw*mLYTS*!CanX}MlIzHIAmd&9jcCDl zqGxfC28rVQHGjh&^T+ZM$xhZpWqw5Tl^e^AMPKCAAkl~KYeBjpuSaSv?2 z`QQ6VL!br=C!vVU!d_k^FLKH&Tpb)mbJ0w~9_ar!e&(|Lvt%bKi4yz^Pl7tlpNLVS zzhvTfQO6;DytRtt8BQKzhSWwH#Y5!zVu^_0Rk$0*`7=Zh(OT|}5yLpSGFp$`(in4T zuFHMoKE@G76TNZ7(XwntC|&<7FTPteLmy(N3VuTBD0MJplV&=(%RWwd4B1KJNs~k6 zP}0cAMopRk-7H*rEtpIiNs@6)j(C@hYA#I~NWnFO;Hr|0YmCn_n#BsE^}_d{%~Ccg z8-Gjq&qZpEe{Ae=tqCZ2GuH^QP%M=f%00w#d4s%0UhkkfWN^^r={!YrG}fwJUKsDI zalxOv4*1i|bdSjH94>8u6ir9UiXu8|5mq7cM0kh=t}*^ilE2o2o#dVJPHQQ>8Z*;Q z(yVRu2%bT|sAsUY({tk8#|rsaJLwZ>8%?I%)~>o8#o?9?)^MvG%fX7UlGa0dN%p&5 zpTew%tt06)J!g*A>$*E@uh*tI*84hV1t^ldtPicvX)JA_2i7P0LDojUtGBV)M!?p< z`WfHFcBd&cMK@c|vTn4L8tA#KS6CfdLWOh>y$!3Ox1nLWr|w7(X(jd0^I0#jx-^Y) z!FG`irwNo<&!>mdRQOx17uQV`g;470rS+_MTkS$>qL}`PTX*WCsC)>g=?Hbi0*Pkmhx8J7 zB}^}TjnikJA7bB2*OLxX7rnLq_{t?ZuaDKcq_|UpK2jfbKOZe4Ss(KvjFyw4k9rvj zQ_@Gi3}fZtUY{1zcRE7Z@ve_&da&-3*BvP;uiq;lOWk4aFK?mWZ1G7i2iE@9t<1gG zQ0r9d4klVwgt?OiG%RU7rm~SbO%=6<%;#4o{?3|cu7~RCt2C0R_*^y;yz-W~CZ z4fcXrjYc;3!I(+BHgS8O& z)nM*mGhmLeexcZ=B~ed0vaT7OP%5iYXi^=*zl^r_MNVyD^nx^TZ{0(BGphB>cObPu5qA>XL3PxN6_XR)t2w)MKBkEJThN9kxYW+XSU>W66a(I-c3|E*&W0n&m7 z8#SsOs}k}eP9H#tA>|9{S!mn1QC-l)+#HwuQFo@@6a9pqi|$V>>Sg_%c}?;uk5G5R zFpegB+l~yQGu;cl@(Pz92rKeQFUmZpMrP0p(aLF!i|N17qnR0s>)DxU_P0w2?9%KK z-}MIcX7;OIdIfqg$NCWhnlbNwEj_1YUAdTEN-u(z1?37fg`S_x}U-f`2_;(8Od zbd$10FTjd!z7nJlWzJi&eztbtBeq^Ds%K@Lw?4F@IMm$gQAaP!Xj`!y$m}!Q+;ixS z>BY7WfqG-saNFA))*xPe`-K2pHy(zbo7FOmMlAdG^M$ROn6~4*zx6Ad0t4S&9%&dw z>Hf}he)?z{v(xo5DtiB2XNw>yeRrLyOi$?Pu2bIDFKqGdGrqbL>t+}(tc77zWe%xE zM#Jbyr+44;N4vRX_gR0G1B>D4Ob_;)F0AKg84Safb=Z5V5acq*FuKv6z4smTY^>wH zlRkP~dbiIbKp)Ji?K_zbW#o^2uV3nd4U0VCt((~>!x%sl4Py-TgJF$jllC7kXnoH{ z8^(9G3P%fzjWvcbmSSO`!3>E?DgdA1hT+Sq8pa5kXc!usa^P5g zJr^?{IOYPw;lMFhJuho(IJT!A2c#!@dp6EE>dWdHMt@2+jErp5!9*{;BFk(T9jKFG zbfTGtF_JnSOe}y7J~IapcWP8Ac6yV~KOq>(O}QXegbq#1%yE&?wGAsYY(SA4SFO zaMfGTx}!}^dJmQ%@v|Qa-<{)Vp_YRe>{D}i0*_le^}7_hh56(h2}Wy`o{BNWM%h}QjI6|-x>j@`=XNI7MSodqSB1u?ftCf8<6y)^RI;EDe_t0}5+a#?!}+M* zl)}URD5W>2l^cD%A+VcpZ!` z=H|S4^$cV2{Dx?W(~RZ&U%DasJNCZ0idwca^3@d>JCa}TfDxVidXwRJ>(zD`yKcSO z1_AAe`d~E3Q=+FggpvI6{$^zBjtkE>p-}C-{A2?p^T&!ix%Ck&aK+nqINGs4{22yB z)92r-3ggQB_c>u4-+!A>=axRnzKu3P6vpt$7_gv0rjU@$Uc0P2Lnx#?#g0pUu!OdZ zyL&5`rjM5rXVUdaQ4b$eNLUmrL7S#8IgpDDoVjGlLW-RI^!O_pAHFJgIEBtza#QTpBD869*wx}>*&Kc) zh*oX3T=Au;TN3e5@QN*$FK442TW(w~#`10TzSNNRZeRMWAT8a#^;$6+zjJl&SCqJG z-nr_Ov}@G^M?A;A{agS&+I8w21_isP=37a7_C&~zv~SP((}n5F-q2UCXy)GarwY;3 zz0Xd$(dB*PKH~+A`zD?Ap-ub3pkLj$=;ceBZl-In+tc`cdrlUj#K`F{UZ9m*dBU3x zMy`dP7`YoRlOoTY^rmh5C*a!gQK3l%&|FQ2zAI`j^vO}Hp+`m$^n?Qwj^(Ev2PQ#3 za9|4b2?xTU&o~ePJ@jCBq8Cj$xB~X^2RFjL;-H>v!dnU=j^?9rhc+F}$3`Fe3j6fK z>yLQS#KRk*huZXshquB$*`_ZzY;VoT`bN)B$V-c&S3nPo-UR(j^xjW`wTu3g)QHx_ zM8xN%c`>WwJ!opoX6O@Rw#R$0_A#H2HKdy{Ry1IJVj95xT^=xIM9XXkHvU;-D9fGsWbg`SBZKbJMAWnb&X9 z-Xp6EJY@XH-VYsV*0GC^{3tGI?T0t0s?jeBP|C^G?>*?+$@neVXyK_hm!8qK)2rU) zrI6F>p8C@LGY22%rHHdPZXAN)&Z%dgf5Nl)O@DLhF*ZB|jqT zt;r)i$dckVc_!X_FvsfyMOliun|Mdi(T7X1(Ju5!=yMmg_UU=bLh;XHADAfg z)$H%>Sgkj^9n2K+E_wCu&~L52PB9;DnoM-KXk&+^3HIRyb) zQc@_f+`-xB7#@+f4KC9ruj68y+T9u#Ut4K5KCX`3c*2?i+51d*JNQ}6iQ7kb?4NwJ z<(DQ?wvT@76c#@CLya)m`Ca*`ah)?IPL1!9@#wUNm0mucxnSt)8nX@#`o42^#Nb!8 z=1ds$d%-zTqn!Qc#P&;k7yc#thdT2PG|hiv-l0bMPs~3w!rp)WnX20L1si(jJ-*;_ z;FGEgt_1#3H6pZ2*;5gR8hV{sc(_5nvkR}4%wK)c;rgEE7F{j&N2NtB-JfS!$ohce3sJ7 zVk+baSQb+aTEc+eD=s@z>_MevUtU6E#qL#Fez1Z2+2yg7aukP_v-omL#T+FKt@LtB zrR*h_pQIt+#}W?%SKcd~QeoAav9G*V?W^Qk ze3hlrZNJsV;p!K?Zs%EZr~K_wYqQuVmxFe@+**%1hsv(o(dvf%y4OQ*WH+=N>)s5# z4sF==?CTA;^=}4W&$(e|C;L4ccJ+4Nu_3ad^SKQP4NleCkW|Cjf5ZLOmrHI885CD! zpV8_HE zTE!hV%U|%=IZ2yeWar8O=lpiws(!+6*N#pf19$z={EWwL=W7dm_N?yzyy%`??N1lp zv%l$Sr#<^Sil{w@swMdCT{iS|;k~;CpFXj7@1RqK_8uB|vf$oC-wTEJp6P+Zr-2uW z>|5Pay}IvEw*YwD0^-*seD_QDEPVoDY5XJsEf+-@cb0&U^2BIT|NeCTJ$} z$!w8p$DHtvJlq6YbU(H(($Xl&JMu)I69pnq7dsIUd9Ke%pUC^|PIyIr_y8{->YVW2 zKdsf#y!$tdK3;IYr9Z9RpU}YByx+KD|EWH3yEFt|uKAt_+J7S-4j&c0QFQ;?w<&|8 zrnOG=jGEd1O@^qABa#Y4ZEXf^Zx3iuC7xxAIyelrv!2f~MVj~{ZcW4LuK#OhI^QpeE@g2=>vGJYk^TZx+9_Jo=A!odM z>~*)#)notAarl`fxy3Rbn>*I0P3#BD0yngi+xr;W5NLZoxqDdl4z;*iB1hoxU_WS< zZidzaT6`;LM>;@DY6tDqeQ4Lq8(LLpw@MjWSxauqonf%u4S;sv1KMNb#1l_wFL_)+ z%V&pUSuCH69(!r|!V>dXzNr=u3$4Xjx?pHV!wt>dk})G}x#1Q!ZfI3ClaUUna4gPGJTO75t)Q+Ft>WF82OuOiS_){%n+~Uvs$7YIu5ER4XUlF?P zFLbd%!WQM2lyFvxUkIs~Mt=(gXV-I^qJ0xA|vf=6R zBLR_Zj$iAs|KjnNwfAH@5uG>Q^JHSt$W~_#x7E*|J=k79bM{aN>*=#Mz4h`J&-c2Y z`{MPU+nQXw(_?GnOV~=v)^aZygV?KDqVA{)Nl!&HYY2y}iBNs#Uim%Vj-#`)~i;Y#%28O>rBf( z-w)}Ov+VtGopO|Yc(?MR$j6sGKfQQ-!y&uFIn?K$i7n>SuFKfHaLf5!U{S3Tc8|8S3Tvrngk#g|X# zD~gYwK7A76^Y$v)lRxjQ;+p(*Sr5hc+t!}W?%$^eNz(VTwYdDfSrtjr?__)PO}AP( z({;?|b#kkfLa?EOrVn_r$ZM-32J1MF2N$gJxuW>01GF1jw6Dtt`{ZozzB=scwCKI# zoA-<20vF~!{RQizSf+-OVOZ5K;*-N0*T>l(=DzEe;+0Y$rBI4*N3~RKNDl~GVgSG3{ za-(hy>ehB67ecbg%XXD*#T)|=6>Ikb>+L42(VMZV?|>D4Cvv8YlnMT`VsMiU>tH#s zLY#|oQyy|7ck-aTSbNWhkn;bBA!hkM9%ANyBShqbEuQ)Q?Ueu5=`PT|eXl|MR!2j> z(6Rl%&iquzj=k~!X~!Xc{(K|!06q_TAP<3Fl=p&Oj5mWG#4AHD&i$a5;5ng}`8XFUq1i_3GXgj{^4T+na6fGN64&HoaHhj=k7qn;v4*NAwxodjK2M zcR;5;tXDrAur31;E7opc_dZ=%i-80E{8>Hd0jw(YKvo8NQ5FEb7%Kohh~19k|*4r%Q7;?ok_V*awvUGu-fR9o~%e^Jlo#gj>#^POzzvhVhS`QTI(aW>gru zOukfv0w@r6ZgXwc*rDEuy6x(g3Vs-TBzSY_TBZF;yOanjk-J3h;x~&QF1|GAR!~Tg zci^kQyMgh6;equ)0LT~^9`FjB0_(vT&hA1DM?z4Jt{$x z%rD?S=*(!9r?VE!i^+JM+is-9NR0rhh9?;DekeT0m*&RFjWXcc?${5}Q|g7CVSVr~ zdq1f^{(&7W#vsiS{;h0*W%70sp31@g$PSRs&#SQk5IbLc@!kT@wZYyEJNE~OGQar9 ziIN$C|K=9qU#lhfw{98!%UVGzMFE7(X$IllR@yu5-Je2r!54zOl~Ssg+EneUhH8T} zBNak%Q~cF@Y7@1OIz{WL8H+}#DdVVZ9>$U~J)`IJf?lF4evSJ1mfq2ObXXth6Md#H z^c9~@`HoMapd(`p+h!$fUR3akD~$@Zqgt4($pvI6S;{n ze5u|&{hp8}>`)S(vP_RhGb$*}m6k|5rM1$0X|*&@S}1Lm7Dzj!&C+%$LRu*;m6l5@ zkY8J*WzueGk+e(NCasdzNb95x(ne{Mv|fxAAz~bMxMOb&cF0T?Q*5~~KpKo(7%GL6 zEINvw$hV(&^(04rj-N;FU&LPH%lrzz%CBLM<8|y5xhX~nJe!5QiQ`95h!c4t>`DAA zDcE6B6Ya#$a16N@R?4bH)OPAHb&fVv+lRA4$*&Ywebu(=P<6J}U)%AQD+(!PR3EjK zI!K+N4bk@6uJBZfs@`f#b)Y(3>!WS`%N0IKX|;gbT6CuswTS2ih|vBI!T!83S2edB4$N7_T}k@ko7K)bI!(VlA0wCCCj?UnXgd!xP7-fHi} z@1lUXu6@uxYVT1RK5JjJuP6@%g}3m*9^tv-hW1U|)V^z0O&27R1+(pgH1@Ab*uiQN zw}d_Rl4lUNu~*7TI0$Ev5j)E>i)`2{nH75-T|`ciUE~nCL~iT=a>LUt?%18;De?&~ z?0+tdor6V$pYRs}*hN%S6vG~x;-aJ|B}$7DVwgB8%80U}oG6c7L>2IiQYGwbsVu6B zBO*mq7r%)+Mowd=%Uz=^h+3kyxF_m}y4c~=L^Q%K!^WbixG$Pv-$)DbK(rQZuv@wn z_Rl;N?L`OF`A0@ufEJ;n=q5UeF1DRYJ+U{c$3K=;(M$9eeMDc;U-T0Luy1OR7%YZ} zWnzU`E_RDuVvpD-_KGcHtJomci_K!A*eN!NZDO4|PF*bGM1qJH`^96Ch@D5r#C8!W zqQn7lP#hA6MYM?v>=27( z7M93Pu(RwOJI^lSD>*mWEtbOGu=h;oD)tSV@!W?4c9s_4{&-3#m{-R>s203E+S7h$ zM~Ct0*r7KQJO1Zm&*3V*9(#;-^Mlw~mc&o+Q`jeS8TH(#<@d3h>V<8G(kIC*IZ9cu zFU}8p!YX3NV`Zr>cKX)GuJnf3`P%`zTsmQ|S87Y&AA5qwN#msn(nRd`oF+||!qIju zMayKgQ7h3dtw-w=DMd-qQj8QQ#Y>6ODe0_qPkM;m_Yv|!C0Hq^lvgS!6_o+X5aqBE zt;8r6B~D3DjwnafvFbwgka}2+SC6Vm>KVJ`cF*lzVW(X?Q+rbfQx{WL(-c#82^gbjNhh^uYAc^vLws^u#`+eJ1;Y_TKhB_P+Lb#yY))T|m{fY1(vchBi~1rOnpn zXyMviZJst?TcAZ~3$;bsVr_}GR9mJk*H&mNwN=_`ZH=~8Tc@qpHfS5QP1aDZm!nwisFtY3YYEyBo9&qK z)c=7L1W@3&k8}np^SOqqejbR~d2EN&}ktMN{>;k*SUMV@1 zTuN^AYwn7N;;H0QywKJbPzs{=^HB<+EjFY&T?to~D<_px%30;SazVMMTvDznzbiMD zTgq)EMY*GPQy;0%)X(Y{^{c9DvgWL1{6%8(sG3$stE<)1>T3oLYqAV zeZw$qxHdu?2{lR^t&P#fY9ZQqZNguU#%WWuP;H_%Ntt)0=% zYUi}`+6C>Rc1gReUD2*;x3t^ZO)Xiwr`^?3v^(1G+I2NVU8G&p@J<5E0p`*&v}(qV zuJJ4!?Z^qV9xu>J)J3b%%jg@?Gj3OQC_9y1%5K!<1Y7-0vU~MQEP7xp{J$t2l>HJY zV|cb3qsD(2hy61tp%&#NUj>SapwxCFKh*X0&tCCxKohqf2R?1+M%?M+RZs6RJ*CXJGX6qnxL;4+zc8R#QDuNBxxmr6@*4`7mbj#@MI{MlLNe zLh6I@%Rr2irl?`+bc~ja+S3R1#%Oi6{yK8PyBt>j-MILFDqTCr)baSgDuaJF-v6JD zT7PM?N-94joHt#@jnbvuD_zc$(k1<~O-XN!G2SYOk&Q84vCTyOW~6~Pgg}Nf@b}X! zyfMT0D&<_f#(q9#6i016`f>afb&^hDKj0ZUOXui3`uK}^z}C>8SmD(O(}GT z?$SNHKlcH49X_JR^aop%JXsMch2cpYqsN^g&6H+Iv!yvw zIQjykCosxak@OPgkzU3crkApw>E&!qu}#xY}spr`uaSP7GrkIE+lxZ;5)^lM-FN6dtb z*|8_)a1z{$0%LW-xUM{O_<|qf4BpZ5=QB8mcQufN8jM*xQttsf(r=qv{kW?!D?z)U zx~mP;9_j>J`(VsUaw@*6n_6G(u8v2oUvImgkrGZJ_)@7aWawv;Vjya2F(pVTj&fN^ zsjO5}sw-8Ls>*Ll4W+hHN2#gQQtB%8ei_TuR~je{l}1Wq1z%TInkmhd7D`K{mC{;i ztF%+vD;<=MN++eW(naa2bW^%3J(QkG8>N@hTj`_p#r>WCF+FH~yvd}DQpS+I5~75V zgEC&3K#s~JWh(rKE8&zCz5H@=HAYdC%@{@D17`_J0_8&Ae;?mTexf|366!hiJe5*! zsyFG^S+|$iY%dDWozOQn3G-A6`!6qpGT= z3hcBtsrGojF9Y7v;HWyO&UgnzCOnyuMRhTr6;ZRP+0`7zlRRo}H4k#vL-n+k|AJ^` z3#ogx`sgz?1iuj;EGa;vZ-LstKi)_u#JFiK)T!TbhjZk z+W=({##n}HYDgUO11Yr){pp4^IJ|KO@7($63+o!Z$KSQ2hBVNAbi-`jSUF*6CAy;Y zs|a7FYuG34jTjoWAsGCqCnaDvuEOex?VQIi=YkNTv1dGe6~-uGc=iXYjhHo5AeC3; zRS}zRygQkVwHR`+h2w;0)nf3CUJJiT*`#7paeV)`v{Z#k+QO=gc=e~M(jI9KHIw#9 zQPf;Ih&P3|#%j_P>LJ~bZqjgDTsj~wM$I+a2YeI^I^*HcIquj4K8$*5)brFaXAo>z zFlU*|=kbMj|Hcx$Ib#)`i(ZeXs5kM=d<&j^-p+T}<}&+uBz!Z-$_P2N?)Je~yHjlC z(Gc7^Pm#0PN`^VW@+?Y2XE8z5Y;5a+JJMo+*!#yWwfVQnH(DmOe5b_P&aA{4 z*liL|Ju|{|&1B4(C*$quhM9T=Bct~8f$t6`OS&t^cRkzS2;*3}mML|W#t4PWMX4&> zQ4UN-*+>n8n?{*N;m8}VhWR7pw9;g}zuo3$tfE%rIO6v!>tshV@t|&})Xv>SE z#Vo0m#fr8>n&$FT(RjR#-dM~15x%N^`k|;3a@G3FImKu>ZLMc@oVokU85y%wD2)D@ zlJtwaq;6I><9{Ef>`Zq>e3dCiJk*&#l`S5|vur=((GR0Ro3h2@)i381tYz60W{ym` z^vfApEoCTMJP!WiE}<>-u*G9_x~mb7Fk3v(y#Dyv;t`bY2Q82C_=hs$p}PKZPQk3w zrfl)}_RAR=a{wq?Jbsmvq%2kz+v0H|-M0~sNr;E74Um3@Z{+ACWg2pHxw4To)Yv%6 zh+2CewdpB(GdIjuov@~g^2y$@ces+_o3rV5gl{XMm?>Z5-*7G-;{}^7{Y(st!P7yE zZv-&Y+&Np%PrwB=f~`-rr=OY5cjC+tgk-a&yAR?E;f}Ixm}#zzqipf&2&D~cW7yN( zd%)cfVN^v6_|vr!Tv4_;%ru`U4O}G!t39a-_2cIm_Kalq%uw9xPwtm+$4l%IH+r&E z86X@MgZss>Pk(aT4EOu#E)45fj4gz}h3o>Z#4{;arLrmXJB*=*GmFh#W_=8WwRi@d zv5huYi8W)*C=|!dK)B6~v&u$G0G+WhHWx-6U=IMjQjb|H*gatX5nIM;z{Pkb#tW)^ znlGbg_=Z}Z5&g6=USt?O+)g`ZvpKW*4eq>rs2~KD-Y; z`PPq*qF#8;ZY~YS+So#xiGF4UK8=MnGJKA#}nA^DS>~K3et5vbzO{QkcvyCnX6P*s={)it)JclI0l z%B!rilq@}BQ}Mj-XEtB{Du3lI)d%VW-U@52k9cc4C%bCAv)x#`)qI)VM!Q%Zhm_jc z`WLeGFOsc)k!}5pX6s+D*1x#f`WJUx>1JFF8QEi2BaAgA%y5ojo$>(I zmcr;J{^wc4TCsKP1slX(u`et}x+Yy`$EC+I{x6lkDcQIy#)V~hkkU`-$D3h232!b@ zW+}6HN9C*Xm3LB?s7rZgb-B8ncU4!at9UncgSvtDP&cVtc~5nR8p->rDQXHIs%6q# z_%JPpmWz+l@@hVOtmdl)@=z^E3+6MiLVg=>>UnMFz?a)Ku^Yn=+pV$Nz#pItXsp{$ zmDZ~uNjRgjIzJSiq_X{!XmeLsP^!Y(EyFlB)wj&t429CeP^yfd(Qe?CT^REjEsyat zTA~n}KE}8+{tNo4Qu{N*moXZiS*1q2Ir?1&8;A}V%7REna4VsF?G z8zP8eLu`nMy?5+1!Cnz|X`Vw?t-9j(tLVPXvT>&A}yKY*a3cHPM#JF@%vy<7%bmS`J81aq@tWy|%kZ$K>KCFO< zk8I}FyT>O77w{`0~)+-U|9AD`Tn`$;JE%fqqX%uiezj$@Sl z2*(&@M^l}afkuUboN2_Wl!UFw(jiKW=qo)w^`I;H(UCjW*4S({u~Z=xn?zrTjS4}8 zRp>$fJPkiD;%Qyd%{^{_$`Q!{+NXJ3yXVt3E$6x|W8+!Q7%4gO+)Z9Ytmzp}Ot*|w zmP4Gcc}Pxt!tZ0#JLkakqBpTReHxPM!8VjV*`uN7PD+d8_uVs}E}2g&t`L@dk#OvX z=Vd7+?`OYOr_U*R4YAVhrQs~clTfo)rDyf_rlQHjCe=-<8_#GwwDHi&>81r!C?-{2 zQ+avi`IV<-wJ`DtD>ergG!Tj+MpA=G!L$GS^c1`lK1omUyYt^lw_ROf~$?S@@en4S#bM{^n4_Pt3y8{-yEg z3(8c(Pt3wk%)*b)!jI3wkI%x7&-On)3qL*!KQ0SDE(<>{3s0+?9%oz@eq0uQY!-fO z7M{A6hU2+7Q$1gLt_sr2f_;dQyKXMKCYD{}%dT-{R~}carEzcOc)2T&D>lph^0;C* z^%KkGt~{<-FZaviiUo7OJg!(#{lt#BD~~I-%>DAXVo&uGi{`F8u2?nq%j1e=bH6;U zSXceTzPT%pD>lyk^0>L3L3 z3000OWmM89etF!9<>TgdmS6MZ=62=?<+xH}buYP{<=6bUxt&Q#`M6ScHAZe{`87Y^ z+|J^cA2+u%XHkwTC0X~9+gX0ikDJ?>v(1hxE^pm)c5P8dqFuE6%YTH@CC+ z<#BU6i(ei$w=-5L$Ib05etF#7&f=HH&FzdG%W-o%i(ei$x3l==adSK4Q;wV4S^V<2 zxt+x?kDJ?>eU#(ob{4-pZfdO#V?PW+gbebxVfDHU)Dyz6;lzm)#bGkCr}hF+|LnXl6`TFzbH_;fptt>+x&FVXz( zvFYEONv7)0S)_lH%1qUt6h@zNb(GpazU&%TcICgy{xlqxDZBDtWv}`x=U#T{uf)uL z6_@@huKd@0&uQ6CeV3Zl-a?mH(RWInA%^ zSIMu=%s%VEOp{HyU$17h_(t5V|HO>7CETmO&glBn+^J7-6B#kO)QxbbFoxCN$;#6^ zagW}Gm04T6HCzQFAK%d{I>)|bpSBO%X||2&%su~CZW(KNzR3vvv+iN;_a`w|?sD$- zPouB!DDL%pF*>#lclsS!k+r!?oUuR9_BQ1SqMEgt8#!lxq8Dz7{gByy zuaLqAnddc;QJPDc^>qsS?oZ!pPrDO67@IO5tetIbo6~m{P)<)XCl+}5o#*v0*v{eT z=eXfKsq|xQ>^*r->B@?7?RZLIAwByWvj>+mi|}p6fTlCkaGJf9zTVOHLVJcCY6p=1 z-nNI`j$?GP)wTskaLj#fV0OcY%z${s@FFg&7EPw#_Zs>aX_sq&1#YNN&ugpGA6Sw6BM8+GNH5cjL75FAV=b9;eMy zZRr7Poc2A*Ie5N)J%+QPH`3mko|*LTGAi-s5Yo>Y9G~?jDC%(fiNob%(-IXEXW z+w0`$6lQ##9-R@L8J!iK9i79xxf}kp4=ffi_k)`M_y5zRxhHLk?aMS1ke0J4;E4{Y{TgdpeS_1VW`FleC-&Ukys{a}9#*|+D)#;78FdCW8 za=MJZsnOAu(U@p#G%mV28c%Q4wb6t$Y+5GW^nVjxLQWv;)e`fsM`VmJ9knDCjPa8D zgUb7}ZYx^NJblisM+^TI<8O|ztC=m>l@~j%`gvhW@F3<`jY!| zchQ4h&9qd>zSIBu#Tj}p`AhbsD)}M*Wcd>bC%@#bRlf!4Z~E&h_p0AzO)LE`@&EkS zsxVPGjPB>Jf3FHx_)|Cy`FEdHVWq76`M(_fU;S2vkN*-{VoHpE@?8}t`Ik5v>!1Bs z#glnR|N38!_wRnuxHcS}#&hXRv!b!->@=IsN?^Q&u@gcN#-;PMtn`!P!~ASLS{hqR zwlf_mm;cmeNNaH}?ZuhWdJs?Mjs5FqQ1qGuA=IFfaKmNC=m-%dJn^oftwAyJ23@T(i z;wApFym|Q#zq=Xv?~K&ub=@+P(J|&GdvSIxexbhzyQh0f|6+a2Kl`z7$(N?-R*x+2 z6H=S)7}nnHx9Z4YgW1;2WlrgnobhDlMvZl&{`cMvg_iq>w>cD{IY9A*>!H%A(atWr7_H3wn5CDw&MUi^ zub27pow>|)TG@4K*~J`=?AMdauHj|ZNoCiGW!JE>YiQZUe9dex%yr6Kd`B#E4Jo_$ zu3YAKOxZQK>>5;d4J^9`lwC)cT}PE&N0wdv%dR8Ju6|_~-#W{V%lqt^>#(xx(6Z~0 zvg_co>!7lW`C3_AzSowy`jlOKpDpv-uk6~l?BYvp*{}PQUA@Y#y;E1YX89K=S$1DX8 zlwG`Ik@?+McHLWc-BWf=E4%J4yQY?1ca>dJ%C0-huE}NB9c9<;W!G(G*R5sOEoIlF zvg_uuYhu}TQ`vQ6*>ywNb$!`&UD-9E?7Ft>x~A+JUv^zxcJVfFW}&Oft}$iTm1P(A z09m*z%C5^Z*S7y_Z#j+T8SO9cg#Gy~u=EYEa_us5zpr)jvNTFb2L& zyeyu^yRHvc+{)XZeJdIjJ}h))mQd@0W3Jc4Yz%pBcy)M2*ov{?`N8$UnBbIPy>x81 z4`aH{O?T(eFLOP2);;YuyqR%-cI~2OhF$r#=y0i{VaL>!&Nv8|!H^6HDwx4Ale3Vy zGBOWz0Gy*qd6n=?zI6Ynt;O)LKz zVUG24(vW;-GMeMxVEdYX9iJh7Ub@e@Y-jmd_zd#22FU>b5!+#YCjKDvA!rZ9pLQV` z$Wz#+R;a4T=TUc%UmQbAOP0 z?k~Z1P5mggwfo@+kmej3|Zl-a-JiCM!BOhXP} zZ%zHtY^(fHgq-V-#D5_7SuPps`{OgiAAx)DKkR|)T4g7{Pa3ON8mp&-^n0hV_Dc7# zC)+`O4}50+!ya}{kGo^KhwkYfcEBh0-SC<1w@>%BUAnhz(>-W8=#qxqF!lL!-mQ(# zT)!6E*L*9&4Dl^-&+;wOkd$>+@|Scp|DWZa>Es)cnhkv=`}*7`Y(Mon0IUk`_wdvK+*?A1OvE9#+)DiCxSwzja6jXI&HaIUkoz6? zOy=fFZpF;X+K!n<8?UN^?mPC;lvVG^DfbPb=DM$mH_&}0G2EB92fGH`Gu#(!XF6&u zW!tB1QU$XIC5F`5fo`$&#uh`$eIo|w!}`rA1>t|`fKcRTxw z-EG94?QTuu-NAMUvsA^J6Y-zpxDH`Qt~}WECbomzjrh!TH?W;mw!sAURO=?=H`m=l zs9}F^Md-nhv|q`#(oy5t8}*OS)IZMqGPZ-A)agNgZ5I2v|b^8%(sFQjZJMJa1$f0asbBEwR#2t)# zuA@G3{9bH_xxLvAbW$sianwiB*aPn(?Q!&xt^o#NRD;V0t|3#Y`vGdYUb0waAE#2!Jd_BY!;%n=6L=hNd%$7im62=_4iAYnrL0PbV${cMl1_wnl- zdoSD9>^=Alu+!KMvUlS%(@teO%RXm(G>2cFMPI}BapRK#_&+AfaUMd%Osr*j|XwYSsQTWWYN3wm*_9x5`dj#%TwqF{OmdhTUy4GPFxX!%K_}2_G zoj0TI=U!+M{oMa0x9z^^aeJjH+?zekwzNj%=ALW^TiPv-co6<`c{hUg!qPI4%d|}7 z?{4_dvE6YGush;@jZsCh)ULQ^+FjVrvIk23^49^W&Z6(0d?x(^_>$wl6w}Jsu`X8V*hWo@Eb7b*A};>#*I7v^BMxvaPb4aP+x$W70FwZj`2^3+};o zL)QWg(@r+0LTO zNUi(ZlLaM+t1TblKKI^oG_LpTjao&xakytt4kf14 ztz)=5m*aEhtWhJrEg7|0HDdc;@7>J_sU41I+teJ(9>iv1lffKmHmlIfnVyLM5Z1ww zRMUQvj)824@hl)mq|T5IsatbcFGtQ#>ee75b!#S13u2Sw(h@t9=Ladlv&^Y!$$O1F zN=u|UIsJPUt*DgLzuS)MKB=SraPHKf$|`xAkhE~O!14RA{o8fgQ2X%fnf@+ya1i+~ zso4kjaI**7@^zZ~G1;Tkxw$;k$sV0tWwJLMMpuFigEnl%D)r1UAHqr=D8IK(btdPKwg;^Ar`B(k{yj_H zluWH}#SW&i`Oke3JK3FCqb6UW=B`z1)X-PooG+oS>aRaO>#-lloDJ!-e_ea+nTzo& zJ@&{1C^^)GJ3#U1B@I+|>`&gqzJ>Txi6?7F@!mBdSGSWO@P9B5{~N!;FY|T&T|bvMlV4yq{-eB~Je8I0Ze&LOXx6bfpEWK{Va32> zcw@OQGxK}%wb@;KH)iK=z});6e8ue#=F%)d$Ok>W@B>IT1X6DN%dp>8; z_phtl{G0KmCn>G6>Wz#tHI&DhwqvAeZ^oMX^KQX^8*f^*-j$5X$;gwes4(^)R?m4x z-g;vF2I-^wH!CZY*Vl1{k1Fn^XYT!q<<$J^qU+g?h|XiY>GQBJ+a=*)Y!`=zvRxF? zeuniS?I-WaAH*9s0~ifIBU~7579Pl7M>EEKdbl9kG^9Oc#fCok4P@l|Oy1ntJlr3@ zkHY=fei-h{_JgoD+xJ86EyC`F4=e8CuN^Ax=db6(ZWU8_U+3EBoN)UJN**KV=Z4!A zKB%~pl-A3NdWE`*FWD}t_>}G9iqF`7T(N@f{EE-nF05!^yP)C=_Bk=64d-3{UD#d{ z?#zCV3P(koggtN_8D1Q19PUaw`!i0@`(V*V;ZFQ@1S9-sg%?F#LR$0i#fpd7K40+& z+vyb#v3;iE0k*s!L)s2!R=}x|tU(bn;&>%9GS&|_$9)(x22PIF3pc}k2=fO{iaLdx z;yyUKEIKh-H{1mGfzc(rrQIRi7>jO!`<;rV zxOa^OIh$lD3Q@M69uDT}I)t#f};ZOYjR>fle+KF*qt~ZRla=j_MS+R)i z8x{5VbdR|96ke~W!@WbqHG?tQrWLzj*X_ffqb3zQ;J;nOb%+sMu0xFE%5`Y#i0dCC zyIhSJ;pJ+?NN<6-TSQU#Bm0|Iv5@WDiUn+6tN57hoQnBuXIFg0c2>oQY-d({K+VhA zbb4*K34i>r>nL{o_iHE?ek-q^Xe(U!6W@_uwSHo{dZHcszrK26zINhE>AH!sa?a`H zH5327Vq&^pqThSfdWpV&wqD}@E2|~4mdZcP2IwSn5B_)7ajCdFSy3^K^)~*zGR{A( zywQtw8(FdOKdgo`q`Z>jE4nI9wDP~N$C0l1_)C%^@$tt0u-ap@@=A}g z&f^Y$S?N*Nc^vbYu3MCl_+0xa;~mVl)kb0Y4ln2Iq$fB5j8|#^5)~$(Kpez z(Rb1J(GSs&(ND5A<^R#jlr(_pJt%imJ+IE(?I?@1jQs6+&M=JBG?HLFhsu_rwo$6- zR$9<%C3+pLQ9^kGyp0l0L)$9h2D>6jre@ zMpgoc@C{VfWPtE#R93zQW^NnCt5R5#%CHi43UiwbBV~r3eiODvWp(V$a95+U^2nCB zC5>At;dE5iMhDijGK}h^@J6oTsU#(whi<2YbI|RT@D#M05?+YzkRf-s-7~~*#|&0m z=NVAh58XNAaCDc9OVC|2u0Z8Idf^t3_`=JuyJ95|_fTw8bWep@KSov@Pgfn6^X;iv zvE|;1m9+Iz?5QX#o(sIqZs^BIu_I7cOc$)AN74=U29y=o(^c1}pnViO17*c_!HNwJ zRP3ARL5h7JWyL7LeuW;QgmT57-~bB5wSM$c5j^U<@E@JjS-g|*-eqp~UCd*~!3dIP;hkr~j8;-&<@ zqqix+O7wOml=5?j68wQqR>Cppok}Qirzp`==v_+q1v*vXF3|8L=ag_JI!y_eqW3D{ z3iLiDdJdI*g763Q0VR9~eGsr&Bz5FrB@(+mqIj|Eqe>{{N`8azB~E76o|5dmtLjohu##SrziW^&_FfMIu8-;!zW7{h3bhMo!ZK17I80|K;y&`R! z?Vz|z(2k1KPrI(-u0<&ag49>rSz(;s$hADZp1+Q+uP};l>;_6&hd0Cx^wJsIMUgtq z=t_!{G)taAg_N016n8JWX@=zaW(so*jNLqAHB|Bsn2lg;S0$D-ZkZvr-%4SIgRxs@ zNIpoJ0Omg!yKM&fVYgG53t{Z`3Zpf~c2k%&VdUN=Me>0*NMKGy+BQl1l-gh?6qvtZ z?9Lftk6jdIcNj*!6tPVYg}EQb?w%p%w};|BMWxONyMdHZU9+7GR{X2P`udbz>EvfgB0fZ7{&mV(dZ$Hmoyxj zaV2_~;^ln$W{g1(SG=5GzYM7hM<`y-wSUHq=#h$-uu_MGn?ULPnk^2jKtzu@P6Ea$&*D3A^^m@g;h~A*Y>!CL)afC`4 z6~sOh6}JMtIYY|Dq>P=>TNL*zdaDv!^fo2l0KGj!^7{_Oi~nRL=7jB?8Is;9itmHo zrBwWgPF3Po=-rAx7@d|O$Gk_0H%0HwkYnDLA^Veh0$$4U14E=tCJFpbsnl zRrHYz$;U?(zW{wqsrVUvT&WP7h~1y0ju+9Vl(2+8t%PmSXOws|^jRgYMW0jRUD4_A z0{iKSO8!B-7y7ay&w|XJNpU-&l2)kr1D%l}$CB~^PSQI|iKV>F&gh2DQCM@)Fn1?~ zu~K8_D$b+x6!RAPy5c1N-^kb#eN*ucm2iUC@@>VBK;Ov_yS}ShP@{Q0Ql zhal?evl16`i6I{K;Nq%3@vA^G^Z;>ETrG9<6RP`s4M2BqQ`^h+gf zjeey#vF+DN#W!e%ISlu=itCPkr!cp{*zXm0Ci;US&x`g)B`%;+)`g8g$~QPUrlbQ5 z1xWfA1lJP%O)=Y}zbmc=m9i)t4J#GD4sqCpV6cy)?MR`mb9f4j*&8WyDdNT!>2uFu zG|cFYMvC-9q>F=*cDFB7DDuqaNNdV%s2l_2`AwdeQf@~n6KT2F8*Qw3IZhKr?q8&Q z2wsj!xe??##Rx+rp=D9>Pmm%QCrkv70>qWIb9rWvoIn`L~4Zm#%4(JeBjpmNSzQp`cF@q&@G zh>bw{2jy8VMbf^lV#R*jDRQl2mTQXN9qp$0I#kL8#FFOjN-X={Q87ECJ1NqpJNX;p zN^}=R`mvH4N9|Q*zW-as(!1lrZ=wV7AVf!k9 z#6LXaF0^09aP$Zz7Mu6aI2%1u30_2xQsOnxqccXL17ILr0E05FM+Ymx8R#(?_o2sT zyo3(Pcm+K!<7M=CC8q4U6JRLR!>|m=_Y*TdM^DOFj>`E6w8wH?5xm&*6vcN!PgUX# z(bJTeaxc%=f)_hWx*?W&aHit7M9)#Yl(BOazZE(H&Li$F==qA5awB$uxCMHl5=$CJ zDsfBnV#P~djZ(bi=Ov1l^j->=u^%bFQqI6Xj9#JmC(zM~e-FJ<@$=C!iWeJ>RpNEf zaY|f;UaiD!(eX;$5xquOjhD8&^wh__A^ECQa|p>ka9FtiL24O zVH$Sc7QIJ_YtVZ^%CF?r{R%S;jC(+d<$UFQAZm&}q(m)I$tQ@VZa$*K`=XC3ac}f7 zC2Ea6u0%yt^7To&Y(egOgyco^X(g1jKcgfspwBAFEL3a?$xL)QJdgV&RKlSw5M-Rx zH%Kh{iW15GUR5FqKSN2xPxdvN@RFW6O2XgnH6@X-bCpE)H&002Pl-iKOdI zg+*tLdrL_Y^lc@nLf=u65PerkD$w_oq=3G!M6%Bh;6w64&h;ZDlJw74BC*rQN+kA{ zG6zXJbfJ<+U8qyy-_Uv`S{+@aM6J-p3UgwNTcRYj=u#yqqRW(|8vR5`+Mvsoqz3&| zNlNHv@Hy#`yjTHW;BJIAD4~?cFO?)hzk(kL(**rVNn-S8ScyOFk2gw0*^oXXAtFEJ z8AD)BrjdJSAtFEJzFCOKXSr7snA@4&XUM$x_+f45vD7*f0CGFc9jT_{d$zPfA zEjmUCB#l=oGOq5&Dy9cI4z7kzVZ34_{56U^llW^Db00b(gYxIEQw;UVU$02NxW7S> z@oj&jV%J4)Qly{NPs~_>-mKUG=%ftFw7*43`><}qEi}OGiu7&xI}~|#^pi8bMCCV- z=UqPq?gHw8pQ@OTP{}ip=Sx3L2{uRXQG$KZdzFCl;pO}xkbIH+fIw_RJ1+#Ap%20% zxc5dMg~xFBLm!7HaG#D!Ie=Kw^^_7wem)J<=~&YGtP)&-K9|uNl`{7{`;k0<0Vpql z>Y72_*gRDzVr@@&ID7-+U#Oy!=>+#l~O3*MxZp{YH`J3IDAk*GB)HB7Ls@ zdnFeCAC$zPKPyJ^?ia;&LVr_&vFPuxl4DVJ0^$U;b!H^$GB!cI5+8sPC}n*#RHQFZ z`nKe}Wvnl#P;3)41`-prhsum>`jj5Fd^<&1i|PrkF11>KU@XH59We z+DwW2pv^O6e`_k{BD6(@>|6FL$o{CSf{djEYbi2*60Dse$5|&s_DA^>WLzkqtOzpZ z926Dl^A1WtdlnypwuN>e$Ej7~eyG?*kmGd7kbQTA^?)`i5L<5u;wR@1^6VXK1d=~5 z0p%k_%J*i9Jflb*PML*n0djtn@jz@1@%gBnr!Wt;&UhQ$CgVMH+l&v;?G(AT1lwmU zK)Wek(zSyk&%QzTjD_fqiqw^0r;IvuXT`sW?vk+>m3$SJLJ!ytmc#C_2Ye2DW_*V3 z1$)ES&~&=mfr^| z@x|yMC6=&*m3S0-j1o({W0m+4bchnmK8{o3OVQ(%SoU{<5?_W6RbtuqFeSbmJu#yN zdXnO6(cu{zqbFzlik_mxlCD#g_zLtiC6+XvuK3RA2qiun6}v-xD|)^X|Bi~?u|Xo` zK+c)-OQc-L`GAah267IR^JjM?Y}B_4uar^J$$>lGP;4{peK1HCb$8G2L3o9HCPu8H1~ zA>~l)1?Fz_wu}$a+ZA&UdPl}b=;Vy`&^t4}N2g?LfZmnyBRW-yk3;1c5R3h$De>{B z90Ovp<-JNQ3KS1L-d)9pV4QPco_Pe5=;J1hvzxR6VVrxSjxhS8O_m`GG0et zR^s94D;aW*QntW|U1wy>M`tQV>^dvsV^qoz#3!S3lz2S)nj-TZ(sDl+eIMp2@|+h) zI>2m%zM)9}ZXo3eWPWELWeB9NH+Wky-B3v@NS|)-u41JAy{AZjSb97;W*7KCiKQM$ z`GfcrRLUI0QYYj*AwCtA^g%oUU7*CLp$nB*>PlUPq`yAn7j%(gw?-E$@%88uMdm}K z*O;a3{|4MrKEO)eNtpn7j!LgR%kh^y`c#Q;L_brk*!*)Pz6o8S#HXWQDDgzJLGfaX zFO~G(@hjZm#Wr6naxWQtqj<5?w~E|X2Hz?18R+*)d^7rk5+8{EsKg`CpOpAa^k*fW zgo=%X#_(D>g}%CUaW5WD`N#Al%^m6$drq;3e3|4p`6!nKX5GD_S7vf(OT#sN+iec zuSBG`>PRJ$V{%QcqRd8;Hu6}ADp2etM3v}MN>oChQ)23T)pU4~>&`VObzI;rLQ_Q@ zucF>2*Q3-~A=v_*4R7MU8KrJjVZWp+`ks>9iM|h%!(d zxeNUbR^ood7^-TD!6rqk$T&~YDW*H>71sd`6hodAL&bGO$s57!gpxOcjI|W8i(try zVywt`NHI|iwl88YLB=kM1Qo&s>nRT;%18UxfVB9WZqwK3q|@zi(M5f_S;gCzS81WiWOUKtw_IV zaT~>oUAI-F54E_RV#Q`sZb14Si`^6}_LK4f(g#_Tat~H)DP&(tN149eu_N}m1BUH{FMEIJsp*OftS3M_+Uq(QjWokZKS?|y%?1` z2GUnpJXEox(8CljHtVa%oTlR8iWmF!Q)JFj@d(AsIrUfU)##Cmm-9PHvE$LB6))#H zK(W`L0~IglJxH1ZY(yv(@qS#63af+8?90MadhGKF;D9ip5%*!cCxDNwKe|xytS*|o#HaZK97E)_$$!mik1BQRFSct;%ADLy!~8}v7q7##Y#SZp~$#Uu|ctt=U*x^ zMpXPtv6BB^D>7bG{6?{IuHPy$K2ZElv2xzuD>7D4{6Vo|gC7+cHz@w3$Q;1p&x(vC z6n|0TM(D3fyaW215;sPFSK{vIA4=Q=U8%%78mWe+L^~nr<7vJyXzo}vUJ(NmT93G_6@w?j`?;-}Cv6zMN3ovFl6qh~47&r}lILHsN_ z0xl-~Vw+KLr7>m`bPQaJ{}6Nn+>F2MXA<0rznseyxR3Mw9lakO#=i$Dzj02|w_TER zc#^oCQAr;d@~I^I0<$&xjFKFPN}3>k4$Vjoz&%|_jz+0hLed|70jM9+Ut6NwmtMs! zcA5dSapnS)wyi|FX2foD6hoZSYf5r3N_`NLL(zFk(ieRl-ejNHsPvW+b3Uba;XV8> zMc;>y*#8prVTHLrD%Yrkc~PuB30_ZFDtQoxEe;)oUo>)o3#%9EUas>T)<9ZK1@ZtGcBU zbNuR7uohwNLf3|MaNm!%h9Yk2T6IYYA4Ew+bq)U5tGbO6UW>L>!t2p?N{r2_Yn71l zRoz|*u~l^kCC0whlu;pi6YT;U;g&RP44ZPShtSPn3*3*PU6q*jp?XW$3jbHpt(BPk zsoq8jPeQj5nM!P97`Ch$)64T~XcUNM{MfHwKcpbWv5)DFkR>E`9 zU6k-VbXVv>+D=7xQ^J$c-Ib8Gs(KG4B=4*DguMuV1KLxGzd`p_;;+zNun+nEE!ta2 z)$aiqPNhYN+fwG$A+j0I$Vh)Pb7V(;4k}>bf1oUJydLW7H;yR zn)*R`l{vpA(C{R4;SDjebpB#(ShhiFcSZR(Tm{{+=rpm^XkiS zAB|oCqj4XBj!~j5(5sY4@_eikNj{8IBC+4qN-TCAuSAmn*8q7K?SNjdM3Qe}_Z#t- zyqcsW9no8quqk@060VNkri5#uw<}?S-l2re(aB2K7`;;oBXo)qwnXnz!Wf;ZglnRA zD^AijO$j}Ej}lg(_riUYnKjS{l(2w4sDzTg55dEPq3x)CLFs2}Rv0OO`s2(MKLc}@LU^^k=oNBO_5G_KRDM@>jv#B{;up{hlz0(Jof6`m(2HOsZb|1QN^}!C zMu|Q_uTr9)QSwHJR-ogQ=oggy6Qb`?@=1t(L#aD8*WnhMNct$R(eLQ}@BsPS9eqeC z?0`z1K|#{?h*Ic=QpReY!hcKjbtU=?eN%}fzus1&A5qFq%|gQOiY|erxOYb7e4wx) zD)FJv#h5lu30I<_5(Q|ac*3=5szerDUEycbW(_4IEp66TBFbVL%8wBKfl?NPNRG3) z625@;P{Man@=FLgzcxp~(YRkjv1J?VSRgOkT&on$M<*zS^SBk~ctT-glynP)O;OH8 zC~SgK_k_YmXsuEpEp6LF2mH4|NuyApT(%vd6gERIRtj67mnns==v<|+JvvV*Y-dcn zwo1Z&+i_0q2C>h<=%YZmLQfQ%3I%?vodmby#$L5km5_2+`=C;2j#5s90_RZ6euP2` z^f9G??Q1C~Lh4U>sg*J#{8F_=H4&=Dnm`-t%Cq2z~9z=rLQPzvj! zM=OO+D0UMH>#!sb=Oz>=n;keep+MQ}&_^k>LyuAl;#GjzwE75ZF z2POIf{ZWa;?#tN;Ez@k!N(s+2X7bHSvY#<`5=MykG3G9gC&bH) zsWVEUJxZOdtHk44v=Lw%^BGDy1a}>}r{X4|w9|sS6Qvvr?rwB%#XW%bfdEXD5UnJ-H zAUuToNc3SPqAb-tqC|2|kHTYwISzeXiH4$20QEFF5v7jSQ75C5(HTH~MyI3mm58>k z?qemo5M2NZ2{RI{Q=&`IdRT=2W$0p9ihDFF`3;fS?OULHCm*mpNu(qkr#^-Re~wd6 z8K`fHn`6|k4wTP?{nfXGR=C+$eGRn1y&P?)B$S!@P6}(cn)=SL1!b6XtjE6fTjD1D z^;^NVxXI&s@~^%R?#bu@a4_!a=poP-_sb~u6`~gCi%LlT)xQa12g-Ln=Po2GP;9Vh zIB}Pt)Q3f{@^0O!D7IbnK5kLYe-Ytj+(_&s`TiV45tHQHV&Ci)cZonj)s|GVbpo<30Wx@Xb1lIIDm#;Ei?r-t$`H%e% z{+HnN;HF?@xM7p+tC-PIwNBNJRl8N~Rn@Dich!Mahg2P1bz{|}s;8=ERlQmDUeza6 zpH_WewX%3aab>AcYFb*eR8{Iw+OV`mY5UU7rM*fAlnyE#QaY@3cxhnigwolii%VCQ zt}WeMy0dg&>EY5#rMF7&l|CwcS^BQ@M|EX&hwA;S52!x8dRX=7>Km)?tA3>V$?E5- z7gR5={;I~-_?oCDu4z`&vZhT<`tZNs*$+IDE$ zx$TIy7q`8v?Ywqv+YM?ruia<0x7JRreW>=a+GlE?uYI+4cI~{{Wwl?jEOpd=oAyVv zKf3*-4$V5W?9ivfQ5{Bg__D(<>o)7uwR6Z11-&2K0)Bn@jcG;e6xP}{I`!>$d(8ZKiw$a zRi9P;R%8sVM89-twNk56b*W>iOQ~zATd7B>SGgSYD-9|QEsZFRDvc>kC`~F&DcxUs zwDd~p-O~G|k4xW_eyFC8ta`(=91N`06`+!aybb9r5uc)9DJXZgEjuK9Jsnhb$itHsymEwaBSUq zb))LW)?Hh7d)`E}n?4&wSo^{dyntglJSLFf9e_1o1SP=82T4&JP%AS^m_(TqhO zG}wl?p;1~6sv6o;4tg}4NI4kQFotq)6XoFThJ{%F5RMjou#|gAH8(P`b(DftiOHnf<@I!cV9Gm(G`m>Uv$~hy_V2!Ety_- z_mcVb?UuG!vj36^5^KrBa?M{d$Czd8kG&kejO)W+|9yJTr!$v6vXnM>=?9-SliwE2 zUi!q+#}^*FHH1n*M9KQJ7>;+?Y$$9eB_AU z{T}bozIwd$rL-9V4eg$(4QjcQsNDe`;0}G+y6>+0_PTGa`{ugy)_rx|SJr*LJ=fFr zgW3;ne@wfF+AghKf8EOVuJ*U~ruK79OY;%U?^<)uHIHlYOAD@2tp~2PPwT_i+NO2i z)?w?zj9D$@{)jT*0yo{gA2bU7h+4CT_=&7`J(WD)s?a+>BT;z{%>Na76!t9i!LMIo zWTBz(6)QIXQSp+$Gw$YvZl>CSv!-JN-T0pF^b~H~72H-~Vphiv6W^ zF%6&n)%Z`juqwts`x3u<8kegSl6EQdW|zXnBB~N=5TX~xx(CHUbcyCYS*&u?0T%De5yUC z{7s1&cD8-fzHh%`-TbRqIsemeS@gueS*ia&e+l~h@Jd$Qe=m9?93K46J;@cpAK{qr z_voqci}1&AX1Fw55j_|F5WW+&i)y3C{3+qD;Y-mgtl$1qI4hbI-5A{xo^C?!j0z?& zo_nNi%^s$oIl}ZeN1EAYj^TSS72n(S`3A+V+&ArLyN7Sto$OS5H{Tw($A0LS`vcr+ z_BZO#+OE>I;QJUG@b!W{+@7w7yMeDd-{i&zo4QZjZ|--#fAF*WDSX@g;!pFZb2UB5 z&vc#qD}2SFnQ3HyH;r9$v!~n8?B%+co^B(vx7*nCaobdU>$WgQxxLKMuBRE`_BI1u zFEhyPW5&2s%vJ7GGuEAE#<|nYZSGQYm%GHwb=R7CZlZbJ-E7`)x0pBGxQgH1y|&Ul zWLI-f#+~ijZl+zw&9bfC?246chOKcg*bUqlc0<=-ySOjy7VcBKn_FpjXZhkiytSwJ zb?te6TYJ9W&R*cRx1;?Y_Da8}eAC)q<$Ky2eLs7XKO%PiaC@&GZtwGF+WY-k_5pvk zebAp{AM)qgIsQrek>?v=_G`b$e&ZM0Z~YQ?ylZE6srcH}#-5e4dYj$d`sPrxf5mrh zD^|KXqv9vlvbr=h=E@*(7n%3XOxw~u;a-f_izByu zTbT{aK5jkR#!mLT`?@$`MUb8CtE~BPy4%e*a#QWOep7q7-@-2TGu%4yYVqo>ecXkw zacmi{8Lt(u9qb?U2@VVnaCf=kaT|Ab@M63McaRtGJaK(+hP#h%n!n;+j@NhfZjq~V zpS$Jm2jAXziZ_fm@?GPN{g!@fz6!F7-_`FHw}?B%E%{!^A^y~O=Xe+Yuz!N*jQPGk zUMH@K+r}N^4so++X}nFapF1bsjc=Ab=N9-);&yR+x1-xR-Xh+_ALdSstK)6s?cy4D zu^SaNj`xe2MlIw0<9*}aT+@o-li^dzi{aGdg=AJTldsgioVdg%L1Lq4qL-3alP8m> zlBbi|$&BQci5DJ9P&&nAtM#^Jq5#OmQKl9oxUWUXZFWSyjSG>fmG@trQd;dV?ikZ;cnNsf)? zB}0>8@gMO@yInjzzB`^4-xI!K8%Iw@FVn64vN_A#Z$7k3{kOckTw{9pPwCe_)BbGk z^Pig!Y%8;kUDs~!SJ(mm3wxGtuy@)e{!9BF_vK%kkL)tO;<9hhKR7G6HrOoKJLnbc z6Z8)H1xN7gc1AoUxIDfqo*G=?TL+_qD}!;t)xr4SnqZ25DVQ4E9h@CZ@KwQe!L<1H z`2P5Tpb%69ad21sV0>SETYPW)Q2eml)IDNPh#yIe+1GXo#>RKVljA$hNON(pNqj)C zzMWu}n@??%pncFGsE!Y|TgHd;)@2vo2HiF8!@Hgr1x5g*e$L9t=`2G0; z&OPqY;LG4EcV2v9e1U%}SP~x*UljKb8sa14k?~RS#qrVcsCYnpNjxyVG#(UR77vau z4_f%zVB_G|_?Y;L_*maLK0CfO{+;g%^-A_hdMEoPdnG;Xw!z%^oZt(4hF_ois#oKK z;`4&e!FR!G!R&ZQJleJh-UylnZ`w6^qvmejfVn65C|ED}KG-1mF+MK7GCn>Y6Q2-Y z6%UQa+INCQ!8*ZDK~=DfZ}NN+l-yOphQZJAuy|a2Vti73bMdHHTqGJ)?VjZ3aiZb)uSZb~L5Tks90Et9R1t&=;G zDZIDP+^uE0^WMRZykoEv?-TTLo0tQ5PvAheojJ&DZw_|d%pqlwGsRtLX1NJwhP%eD;hwh5+%vYhd)D$L zSX*?j*^--UtKB?X>lWMgZi(&SmfDTo*LD;4jos9JYd3S>+5LT`?c*ES1AJq9pl@PN z^BdaJeHS~@cefY&9qlN;lfA_6Y{&Ut_G-V69q)VFYy7_UTECyY*&k^q`J?PD{%Cuv zA7G#I7u%=(DEo}R#6IgUwa@v>>`ecNo#h|3ulc9!T>rG4=O4GP`)BMM{#pB$f8M_B zU$6`O9Q&PLYQOi(><|7EH^vY2&-vHfnr>*ky}dr}X4Z5~Obcd^wsfIsbD}%Moa7EQ!P3afh2zT|aZ0JHoun zEXMcTWP6ZrY7h3S*+cy5_E5iuJQ?siuidmH;;uhSot?UfyroxFqGhrJnIlQ+lcF?w-J_#&JWJyAJH zxnFredEA@n9pW8koND~hIL&>)o8=wq%{Hzwt~Rc5AHtpD=NQ)-*QsBrU#s8X?x^!*7}3B zy*1HFTNPHN6}J*rIc^!+A2$#kh}(%~;^v}5ajVg6-0d_McRAJLo}>(J@X1;GxO?Gl zpBmf&RA*&zZ_pHL4{vL;)EsPXVtVEfv)HuE;ihX2)qGPo9ka|VH*K@TjG0B|Ak#Dr zQ#FT~8t!5r<(==H3+--i>n`g~>wWWEb(wm)`ET<-=D)1#t(&b|aC`gZ)|J*(*45TE z*0t7k)(zHQtsAYItV`Vw-K_hO`!R0W_!;+Wtiz2O>+KplYfrKa{Yy(R*VxC|OYIZv zyV<+iL+lauFngr@mD|JZq4p;7H`X3!nf9jE=eVuoN8H-+FY71m zVr_-IEADo^Qr{W3Q_)RawYVYcXWWgHv{Jais2VpK?TmYlCgT>Sskob|!D_@!P4jV! zQ#)>W>cZ_$y|@`_5$?Pu&~?dR<0?HBCT_KWsQ_RID^ z>{skp?KSpm_6MZmSss+d}&7ON%Dtp=%s)iSkQ9ik3ZhpEHW5zw|qs-x7=@(!EL)h*O9kn+Yt z0~@D~S0|`jLuS}k-46QL_Uc4+2Q{uHRB_8qTCGsdRx8zvTBVFrtJNLVoz$JxUDRFG z-IVd_B&AX56F1=~8=(J)+W~!Lg4Bklz<#)=x|h1Qx{o?l-B;Nf8dH<LKc(>S5|^^>DRbouf9WjcSuRS8Y~X)K+z#I$v#57hqO) zsGaIUwF{DckJ_tjqaFd>aj~+kxvHuB+y{4-dbWCwdak^!=K}Rl>V@h>>czM#?o##7>SgL*)XQ;m+!gAT>Q(C1 z&=anO1a-Z7L!j9ye^75xo>y;GZ&PnqUJ!a9ZkxMPy-U4YS*PBk-izDk?pGgBA5{OQ zK7_mI9#&VWkEoBTkExHVPpD6-f5)wfPphk;=RT`Gr#`Q~puVWS1fBO~^&jdh>Z|H& z(0<>*J$7$F8+u25SA7q++I^sYsD7k=tbU??s(z+^u706@DfpJMz2scqs{c|Zs{dC1 zqke~5@P1H#RM)CMK_1)zx8nVxu2=iiyt)Ad_MxVN-^Vo_8mS4pzpXh~6B5uHJ?N)J zS~2d@E7b;RgS9fP9QQd6h0Z!$8=-9i{dJT!T1hI+%B|2UH&tF%UWE3z88q50v@zP2 zxc6}@Z5;07o1ks2ZKG|gZKwS~+aB8P4q9AG;HJJ5WWoxqQp*TULEBN=N!wZ51-JO^ zrcKgnv|32(SuLme+GJ?PQ?xxG-=?IVytlTGHWeCe8X9edHci_fzKQ9uO3%;^(hk;U zYO`=B;Gs&Tb{N*GUzCrv!?k*Cj+Eq@w7FWd)}mx0GqPRl&^on+S{Lr7>`|(qmA$I` zLwN=I@)6o1ZLzjQI})CcqqSpT^*&BpsvWPLAT^hhp}|0R&`#4%*OqC^wKKFcAxTuj zLvoI?Bdp`+Y3FO=UZx9`op6id#kfZi+P$)~cA4@OH0=kq%fXed(5}?3f=A^V*wU}V zEtWUne#0B3oOuhhqTAqY-Ua&0oyw=$UE=0N?OxoPc|X2imWB5BAZ|o_NLvZ*W|j5` zw75sL$Dl_&0Zr=fkbj=ip2mHJ&p<;Iw?RFxy#Q}@yY`Z{Mtd2O(Lc0Tv{$v)loi_R z8t#cwa@t$k+uA$YySU%*edV8uFElZwL&<9&LE=%gPjKtuXUb$%g=Y4J_9Zm5ueEP* z58}7*0EqjgzSF+fe$akYc89F6R#~WYLT>m8whP?r$##qlx}vMPrt9z=neZ9dx}&@B zA;sW5DAr5hNgAXN*2|PB@F)#|PidGwTpyutqK|}kX|%G3zNx;MzPa+VzJ)#pcSDZV zx6;SyGR{aEhV`zed{6Z8||GdWp5ML!icWS#~; z+A?@h&VUc)ESy1@1}*OxNn{(>$m8);x5kH^*eCW>RtNXxS8`_{XXc? z59kjnx8tVHhm`&Gm5?4+>5u4-DhKF~>5nVZ^(XWv^}pjT&!_d(`ZM~o`g6G1^9B7y z{Uv>k{xW3$SM*nv1EGt*4lUqKWxetV*1)&n0en|~5BGq6pns^GrhlZ&&_C8c!Oftb z!4md`{-ypE?g{-y|EK<~&_r=-=y&?}`Va6LuEiaqKkMuCUzCHC;~{tT;a1TN%5m^V zE;STGH8evv48t@m);2vO2F^Zxrl>}Dx=!iQJHD%WbAD0V(e<{W=t|_ zj9O)uQD8iyE% z8iyIPg_f;+2Hm>>UvK%wXfoy+%|?sSYRog{8*Rn{quuB*I*o-!m(g92XN6vF90h&- z7~@#uIQZ3$H%>54G)^*3Hclbg);Qf*W-K?(kQ%#?YtIe!_6v=RpsinGTx$Fop4q<` zmm4dLD~u~i3KsIOalLVa)a-9EZZ>W)ZdJZCZZmE-?lA5&?lSH+?lJB)?lbN;9xxs> z{$@O6tTY}rRtXu|c#QS!zZ*{(PebE=M(Ete^U%3pga*6Dcp19)E5@tPzh5`rFy4d~ z`?m3p@viZn@xJka@uBgN@v-p<^w!U$CjO<=#lJEBDYS0V$e|PeVEkyTg?{_9vCjC# zSPwgH9+q4M5~&7R)qr$rK~{B?OCi5{QerKJliWXLA>pDQnDHv(C(# zx&JTs+;u`5=z_M;gPr3xw3$CCCo9*&k}fobCFYUlQPKu~EVPHE&>v2Ko%^JIcj7HG zmz!rOYhjx|ODR`|C_~M&m0`+oWrTT-vWYTM8KsPd?fyJ@bKakRb$8z7<_hx)^Gfq7 z^J?=N^IG#d^LnZC+z2ng%~IzP+Rq)V;oL(yj*gDZYrctt5N|10nQz0&^^W#7Rp&liSn)TFXcYvJmos2-YSOPRto#aU|2xP zts$_246}yAHn@p$wNk5`pgannNU=3iIbB(1jj~27%dJhV&8*F#?~Q>5dztbVYfEda z@<(ecSkBH>E>bR5E>Unhr8Q2u-x?2%aBHa(ZpT{T4pJjbLc<~bFvFV8PEyy|71ozY zp$?dX7C0H!m?^kv78V)kLHkHeXg_Nj^vMIHMmYmI<-t;`JcRYi!=+ByU^QAz)?BOE zYJt`?4|-FZ)SNn?J1vyjQxEG;i>)Qrk=9Yx(bh56vAD%}sdYSbsuQ7Aoh zpz;ft+HtV`jExlC%7E37N}X_wcB`sK~gJ#Q6S zw{-_J)4Qy@t$VC{t^2I|tp}9vtp}~YSq~{cC_lnV`!FoFk4T&Clxfr|D5%_^@8;xyj1Tgb=FJDnbsP37T!?aRNk^)R^C?L!G^PcDDNuoTd!EJ zTCZ8JTW?rzT5nlzE4uZL^{(}v^}h9i^`Z5V^|AGd_37`qRT3KUI_no}z13&sH+loA zYkPLgF0za561&tMWDmB>;Bg-UEqEC8;Stb>N7|$8(e|eHX7=Xx7WNo>OM9%nl|9ZL zZ%?qdwzsjjwYRhXU~g|vw0E%Mb^l$!P_fv#;YvzOav*k{^j*=O75K&w8_KHt8;{*!&7eUW{!eTjXk{b&0! z`!Dw8_6qw7`%3#N`)d0d`&#=t`+EBZ`>*zm_D%N9_AU0U_HFj<_8s<}_FeYf_C5B! z_I>vK_5=2V_TTJ>?3MPz_A2`k`%(Kb=xiiEW>I`#+J0qM; zoRQ8bXSB1avzfEGvxPIp+0q&7Y~_q|#yb<7t(|S0ZJq6$KRDYv6P+ELxRY>_PRdC; z6;7p-ajKkZXGdozXJ=;@XIE!8XOdIn)H-!e*2y`(GdaBDcrRyfN8E0_ud|;s&Dq~M zz?tqG=*)2B4ad$b=MZ_v@oeXCdB1Uk)95rgbDd_V#c6fsIrE)1XMxl1bU2;PLZ{2= zc6yv%=Llz!vzTvBKH3pCA|K~0b&hvVa87hia!z(maZYvq=$z)9?ksbbJ7+j&I%hd& zJLfp(I_Ej(I~O>AaxQc(!i~U}IF~wqb}n=N;#}^maISE!bgpu)cCK-*b*^)+cW%H< z#5X!O;Wpx1oLimSoZFo{oI9PnoV%TSoO_-7oco;zoClr1IS)B2orj%OxRv-(=P~DT z=LzRY=kLx_&eP6n=Nac&=Q-zj=LP3Q=Ot&2^Rn{~=N0Ew=QZbb=MCpg=Pl=L=N;!= z=RN0r=L6?M=OgE1=M(2s=QHPX=L_dc=PT!H=Nspr&bQ9LoPRt2alUiDcYbhwbk@R) z`Lnam`Ndi9^f`HFgR3ZCz-Ot!=V`d6Yq_@Tz>DX(F<3~8VI3_MUN^VQEq8~&=Qhk8 z?v4u%+abH}?A;3wM#p0w@Y7v0{S2v1wwO}I%n<)-1~ ztb~`WN_hUYz5k8W+@Q<{>H!{zi5AVhTcrQBOxmYN@6Fu-o907mB zV)zt}g#X}Z_)Cs;kApAac=rVNM0j3KhOgyR_*qVak7b#=9A1?(;Q=`t-j8$P;W*#D z06vQg;jg&Zy~MrL{WH81e}Ok*g?k13E?2=rB|T5>^}?e?el&QfZWVqj_^aTla_@HU zaqo5SbMJQ_a36I4=04=EbRUM#?Gg7;_c8Zz_X+n&_wVjg?$hpS_ZjzD_c`}@_XYPw z_a%3Y`?C8F_Z9b5_cix*_YL<=_bvBr_Z|0L_dWN0_XF5FVeNE3aX)oGb3b>#fW7xC z_iOhXSbx8TefQt6^1{YT7GBttVPSTEao4+j!t3Y3tjIn{6COhwenuA_$C&UTdL>?| zH^>|8m3igf5N~LIZz8;so5A0>1-y=1dSl^_9OsSqCU{$W+j!f0+j)QRw)ZA_J9u#~ z;U&G4m-Z^W%82J>XKxp8S8q3Ol2_x^dUamb%Xz*x+1uTl;_cz>>Fp)_R}pX40p4`) zKv>@o@(zZjeHLu(hkA#3vpr!)pW`)njb4*C*K77#yjFNO=fk(Tz-#w9gm=^H^18hq zuh%=mTjVYFmUu^cM|nqk$9Ttj$9YS={@8Qz)R zS>D;+Io`SO9-Z%95c-k|o}|mY72XxzmEKj})!sGUwbGw-gZEeOM(-x?X73j7R_`|N zcJB`FPVX-7ZtouNUhh8de)jSTPwm6-@IC@>@MF>!{3JZUPr)y|8Xn+hz305=y%)R} zy_dW--pk%UyjQ$ez1O_gy*Io!y|=u#y?4BKz4yHLy$`$(y^p+)y-&PPz0bVQy)V2k zy|28ly>GmKdf$5g^8W4p$NSFv-uuD((Oc{N3qzaI7p=9vc!H8XFcH9vczcBsMZODmFT{X>7CD=CLhe zV`5vz#>Tdajf;(sO^9tB+a|VcY`fSWV%x_i#&(FsV~JQYmWrig6|u@#CRP=zj_nxR zDYmn)M8|fEO^VgTYGZY=Y%CY^W0PaM$EL*gi0v8ME4FuRpV-vczOnsc(_;I_4)AKa zTHBlJ8+v=1V)C`4xud;lzFo7RzM-q5-Kv?>)pSIYUBlqs_SSf!E^EtYHb>7~Tt1WW z;F+XwvQj(~If5s0H8LJg8INb0;%hivO%;c0_*utKSx+KY%lT_Le=X;)tu$-eI$P>J z`M0N|y`x)Z^y5^ZAFt*x73{}rIZPG!aoM4SAE%1^xKHu5oWGXy*UJ3yWQxnFmF2|K z3F>${8$79jeii4h;`~*duZruh;`*z&{wl7&DyYBC+N*xSoW^?5oOmLav}u z<^IPRAMqe8@flC1QKa4czU)doQB`AQ8|pzJoSb@=i}R2%CUb12kYg1QKcPlf@6n84coeCHoae;)N;!dTwQz`CLiaV9% z{8`SQWu#;oc|2{?@r*uYVj?yrs68Z>TtGn8nGjY7>YCErRewa2c#-@M)H$U$#G%e2YG|@50u*%%CzO%C)E07QlWF?7It+iLkK@vGC8_%F@CO#q) zEFO|fATR{+L@v2Bw;?UtfQg*d_U@H6qo-caEpArV5Ui&M_KjiH@f-mD)bkq(ohN+8il?^pLUQWrUC>tF+e7%t@&cEmUVwrmJW0zUW+%<@Br&0Ql9&&k3=ak; zP4g_5W`tER!YX*s>NviR<4N#HY~GYT58c#78?pT&qY@qeJd(wG|Zc_LK+ znVc$klq-mo(Jfu(O^XmtMWr>ZKsZU3)RGBCR0SiZ zBB+WHQy~$XOeB15+F}~Rbe*-oEI79Rz_}Ml8|#duDfQ{UZWA3&I(hNp7+0rkD+ zbV0+`bXroTGbvA(q-;(XNsneRE)%t>*ujPJ8Nb}es+@JOtdEKh#4De> z#8V)NIX#_)M}EqARmy>WZS*U6epT?is^Ixl!3!?q$-F+_hT4cUa;(JWSb5DQ z7+(o$Eo2TNSH@qWuF`B=($U;vGTu+LR*`do|?kCv7XUbCxafd zkewqP2099lA?rdp)`4;?MCVws&XFPw+F=^0Wg4sFKGZRW>zI}TO&~?}r)hnRrz?n7 z(%FpJBnG)?c4vK8&&0X21y|JPPE6p}YU0NdSuO!N%$nQO-aIj3LY-^wj&;jNPglqM zrV{yHzJ;0ZEJR>tv4 zL_{b=S7=(KX~{Dt;-bf>MUySuf&vf^8KeCgL2DynvMoC4X{Cns2%pEqCxs4{ zC2D}r;+f9~_A@L1XHrH-Fl*CfM1Z)TF*}+9N=RpDUZu$v3elFws3LgMd`puq8Baog zUC0xtD>{+VS+h&>N%@yzin`{ucDD#V3FEdTlw>C7bTPSET@i|@$XHzkih)8v$!Sr= zNHC?ae3oPRELY(hU7{-xM2-;8?r4)zzSS+Oh;Y?eT6!CNJ)hYl6TpkyMji=fD0?X3*#BDu=I8j{N-f>>byvU(+FG<)T| z6H_bB(q&SZ)OjBE%6Vw^N*$)R|7`5H%#ehe2n?yD6l3|3ItZRphk))Pag8U>nhcLY zhQ}Ziz&dt$ikLd+*_cO z07Drzk%<<=BGemb4q?8)vXskUK*+EgSe8U@Ll|3@&_a~s^s=c<8n1FWRANi~2i4{R zGgZzH2P+p2YAzhq90s+t-*9k+K&*J9(F;n=$P#F{M2%uFMPnoZgo(ncGia%Kgdk$# z*crxc#&;J5crX{qdB3QDIcpK=gNwom4(*Rei+waXw(I8DN0TGp45ac3y&=>BSyxK? zu_cAli8BQ2J^OHS)z#LLs8+~C;&ZiuA%uN3X)-*38A>97@+oH;o+)!l&|SnzH5gD5 zPSQ{vAj%8Jov{S$#yB#V*hv!b;0ub|M-nfvj+DLkj>P&N_4ovSg=!{u2$+-W7c8ik z-AYxb%>!DX5^4vvYWudD`?oePsMq&xY1Q}bCe=8_`ke3AdQ%rPHAijmWcPr5o@R{C z<_w?Ni_c~azmi7LXY+>7?8YafV8Ukyp&wLLL)HVIR7k{=6~Jdz)2|_GfzK+d@7Fo3 zWs|Cv;)OcJYF{dwwq&0f>CElZq$>m6#3|Bj8H&@ih59@C;;cqOvq8!mvjKa9%{&8<{3?8Tl9HxcG?NIyqdHW*Kub3Pk< ze4g??`;&dPVEF7;_Sur*vwztq#3%gPj4=mOz-$)N!fb46gD2(NWSK?|I}w?-e=QeovnjZacs|>Wd}bFu+mC!^8$MADL~BMm%?*Tk z@Yu8Jvr6x?ht($!Yr?O~m_35g7_DGWgTU&6Eu?>$pNtE(ELc% zCCzThE_>>G?Oq~=A^hyux6%Q?A`VAJksSd(YXUwy1bo&Ad^VN)EG7G-{KDuEOdaN! zZ1DG4;`hl70yDbgF?eS5&X!gOzh>by)X{3QLvVFZ{*APVZQedhtUlYmeU@B(wt@RB z!TM|q_gRwl*>dai<`AFEecDm11`x#5l zTW0M6qD*qX%Jbe;^*waKmiH`N^SzbX-4qvJmEW5_|K$r(84fl-tKtVM0fOcM>IP>%1eIx;iFG?c0GsCk!PCyQc-S5K{2mj$WvhfH>W)i{*>k-Wnv5 z8GGhhA{QP#i#qUdC~MDLTjmwdRyfiTFG@gsH@0`Ri
=9V73*s{c)xy4)_wIxQS z$~rx64iXn}sS+NeBISD#mm&jBLkCt&BzHTTx>`FLMRgLncp)rLY=D7Ql-Jz^V}roE z2*7yT)ZW%KS2U|w2D-SBMRTB}Hw95L5JZ)T(EKJCEoHt^4hGSK3x#r>Ws#B@G`LWF zkSJ0IgUX{7F>FY`+8F4zEa_~KNG{rbc-9sRM}$ucNc6eNJy%TT>7BDOJ+b)mq=)+}6|& z%$$rEda|OXCKe&)7|gUNR9046s6>WiO$)-f7!^tB3N_2HFiiu6DH$m5--c*{x@Z6n z2lP*7MU!NpcwsMQhuF5-1@U{33}M6qfZ5sAx}YgWTd(9h^_y0DxU@2kgd%-OB#{WB zO`UB`L9ECYPx6A9tgR^m$CVz5`o^X%kC>Q5zUWsnRq56X(r;^Ol`9XnnB-hJ|78aP z4j=+0{pZ!-Urj2nFr$k5O(!>;OI~3j75AG(vB(rE=|6u44V*ls17;5R1ak(_Osh8m zeS{Xi1R2Y)uOhCaj}oKUE##oiMx+wP% z;eOml0Qz$uS+pcO@y28)Ux`S?s~puiG^%)rQN=@yDjuRMc!*KOLyRgOVpQ=Eql$+Z zRXoI~;vq&A4>78Eh*8Bupc4-)kPHdf_(sZQf{c~e zXvpn}$4lxP8?nc_8*^q}Q$vp_jVF*~Y2)=oS$8q4#*htL8_1^|M22>jcq$3`L2SeY zvgWwbp&;I*43c6?aw?v7rN2bPI1okVG_`dsf`SF*OTIR62FSOx^785(XXM7oG7F)X zHDTHzpFAm!u|=qFu5#G+sk_!Mk*6rh56)F3@`m<46BIio;nK7c#5a!_Lv z+uAyM7Q^}7 zTHg%($!Bk;9N-F^@R8vzqLT_TFeNLMGqUxrOxvlMD+IIKAAWbEw z$mxE%; z0|HMb_60B-CtnQ0#CZ~Ij>A@dSzLmQ>d6G{SjUr!ttGcbvV#1jKsKwXw6z60Zpc8# z&a-5ihc!(H!4RgIoaXLe!<(#+om+?}ww@-#9l}&^nhuI3(@ZXDa%&(={YjJk7hxKj zG}(C(CbDAtEy7$s*>Mr(`e}zh!dySugAwNXX`^p4P3Ctz32$jKzaz};U=m4_`5kZ? zo-~=?5w0pDl|{(46N%Cr2T5fG(nkPbi6aaT4h@k(ia}CVjbs^2;%Ed?LdKIY$HHF0 zykY6%TFbEF&wCd`-X4( zg>TcsxBbJn1H!lI;oE`X+l=t-pz!VB@NH)BRzze;otD;TMnFtzSp+X;?TSKVjOJuT z(#{>^k<`g~SUcvB6z;-GiD)S*Ix)DM!o*-eI59XJnHWKqaAE{W3KN4N!Ndr1gcE}S z(TNdcVcQ!AX<~37oERMDi4i0vt&V%oiXDXp&4uM$&PYx>D3~JweggYzP?KEfF=aR$ znKD6&V9NAMA*T#OqEjXSMWzgc`cD}K3Ml~ltCLb|CZrN(<-BBc;?z65vu2)W*`tx#ytG*zg|hDN!P4 zOHwmS3TIp+xwy#4dDKE=U9=0SXwQk7xqO^ME#yf@v6YVINmDZj2SlM<01lpt0!pU# zLkJH=p$+12Vx*?JC{Alatw6w}7F2#umrVMODjn-2cEv+(I z8tq)qluZA1j}loqAGtx9LV4H-DJQ3lK|oHR1XJXBG6B?tei8`!NFMKTWA3mlaej&?;VtEN5E!r7fj%6@H5Mw^+WW(qU4m`Sbu zs--63z)OG%agc=qeELNOgmIe*9lJE6jY?5f;Xw>0L|T%9Kr}--ick2_f^wlmMJYKQ z9RsHRl8#@s9TUHRXmK_nQeIsY>$!e4MU&AY$W5(?HnpNqG`K*tw905xE2B-VjCK!u z%nRMCj1~>!K_n(x4>p{On3BRy=^!DfG&&uzUpf*~nBGWMn5YOu=QH+B2YC`=XLKQe zt<8}D;in|rr5uT?inbgZ!3(*nq8O^8@=BKcYRJGJqUeWBNEA_8*y$T75})`GkwTb} zgelH#NeHn6aWIa8;W?m4dZIz*FtP-E3<1Rvgh~(__#*|1egDxQ_KC~jpu%Vr;^lZ0 zf(f~UJk0#p%~AzgM4&)^ z2t@NReQ>n}dW&Npw5+t>_DcyFxNlJcH-U>!d<-X`M`^VVTa8a#6qZ{A3OL0!!)Q#D zk6;6$ETAOxnag0{K@gN%5R%2K*fAua{pICMaTAQ{Xk9mDXA6W#h+P6~eh{H@a_0yu zG|W?tGPS@*H(gTa9G}YsoqoZck&_HbLIKn2Y9_&GsPf@x%r+&^7xp&)X=_+}_69a_KvUBXL}4le*7j4wV~ zQ^7|AE3#RqnfuKJq)VFNW9q4I14rrveg^js;@IZ^2-*_OG9UqB0im)J+1NaB0tAc= z1sBa1?$P!(9PMmvaOv0qz-F8Hc7x22sKIevu{WO%l29QPX|Z{~t-hzVJtj}S0-=1A z8fRr`q!P3t8r_yo29SvyA;e230Y*?d5pW2d$%IarV976@#ITFp*}0 zN%M!C()1k#oS7n8Pt&Oygb6cg{?JpJkKeJICeH2|Jf&L;PcA?BG7z1N160fadpxz(_(pTp#~!tnuO5bT~pSh}5rkSnnS`IeuY zk3VFUrc)Kb3sFSc=XmlKV*e}Wqum)uN0gT436^FGO4B)M92W@s$?YLG75dHjf^*Jv zQV?*W)HIz81Rgjaoyo;<3!1=brpPo?c$)U-0}q^@_uaGG4*d`M%lU&3){z@gh}VP< zd>2MA&NQF#Nz)l$A!IY0+ecsdK>xWsqIlqk>Q6C#**l7}LX6H!}WAi9AI7J6vA+d7-`0|N3qZU*XJS#0(T*AOAQb_S9owBvQ(}hnZv_qJq7f8^agh<)Ym9N5& z$ZuFvC>L=Pdhc0;jUr8yYi{e@o+U*x366zh z6HenGsuguHpjZO%#faW+a9Dw398ad*F$uy9!g5mMDSM13`=t3SN`fXO;F4(Zl#>Hb zIrwlLeF>Tt~J$PHY+~ ziyUD*<9zY}>v>UBrsm{SZJfRrA~YR_2gPzcojJqCXfBp_Ud4ImRlK621Y50pS{v%c zdIdJ@kaKGG>|%Op7hlS#pI?k2fP1&01LFW*Em!mi>#a%_!2#UVRo^BSy!oZk00ex| zHp~;^mq~J~$CKoC$CKt|l8;>_`8#FFV1eiFl_mM?Ka?t2UXl;~Cix(6GWcL$@Wqbc zJ7atdJIUW@N(LWOVt$z9`JN2g7sx_9|C4-jILXJ#l6*EW$@4zR^FGOZBAHFohk>BU zH?;FBw!Q?f*!U8>Vs}jNid{2uQwI*N68lVMbEHF7R}{8eC$Jwy?&6Z$?|Jh|xFd`| zHAYid6C%jZwN_}o9AKP~|SfFNw-my;D-T25w^%jK~E z0CNCMDE@L+igx}t)J6zgxq>axmZj}@uxCUe6&5qN7Yk&6g|a9 z&Qp9^GR1;qiccY?m`+oC+AzhZ4O4u|FvX_~Q+)a`#it2Ve3~%DrwLR1{jU_UsCX)% zVirSEe7Z2jpG!&cH=$DeX|)u8)G5V!LW=c-6n}&(#bQc|#g-I*r6EQ8CgUmEg@Y%r zt-K2d))YA!S>6o^g(_}g$UD{ zoaOI=W|?}jOf^}imMkA^&ho+LEFWyn@&|&lLA&@ff?57_NtQpqm}M%bSy0XM3 z;p*Xj@L}96AI8n{Vcaas99dTVvP_j(rur-^2JwLH@<+Cke9AS+Vrr7w7f-U-m8AZ~ zll&RKWbj2XR&0`dyfVqupJeejS(kI#AWn$o%O#sgu+k)1LQJw)ljIFqXd-`9u4cB_ zF2Mo%YU(ih+?@Qn7?#4>u({w^SmY*6ZG~-a3BG;T#bwY(ok}{Jg5AbKH=-2+URB`%5CMG64;#HW zfCElR85k1SgrWwrB5B`Z^P4y=By{)LVx<;K55x{aDok8MgX{!h8&+6o!RFQ0P|{Z4 zg-wFu@S5n%(2W|ailv=Owgsh!m;L{Dsc-HVg{eFIyW)RcLK*!{9lapHyG)}-Uf+lD322b;$s3&V6^yEhaHgi&Drg8V+{l*aB@F4Q35Fh z&s!t}7eFLb*Doc!645x+DK18!Qs6C!gz%*v8FJ-cymyKwgR4{mwZ*ct0Vx+t1O*|7 zzt>X}yi5&VrUx(e!An!{(hb2Sg;*$XSRwiVxS<8*2ZR$lmW#A;81A;#Q(1PIj!xDy`rqaVv``tg^6 zU|k&o5ECE3;QU1a8xs+wL4ffYhBv8YmDXEa(Ix4&KO!iXb4)VqjdnloEx6cKw)~ z>cB5(wD688CYmABA|evN^xc?gdn3T!hB9ieya7ejElQFHV!16PV5#U0{5j3CjRC_i z3tnamgv}Q5#k5zAyITr|JYVeckA_8B4i6IX60r;@767={$iXp4wo5D#+>lsMA$ZH+ z;C*&CK9|-52r(X*&M}MSxwRmoWVS3wW`NGnAGHnbIN*nQBylKvA3zFhFMv=n0z(b$ z!wNGdnl%EB<{w%}E$Eu-i6o8|U5@Al_4KhSi5*loBnlJXglPg0cR^m4Kiqi1o}}nX zF>ZYvin|>r;C}21WhZ5_qNro=&mNN*o5_r^Q@TE;+#Z{e&#`t2>iyWv1QClX_Ye|E z>~h3Q7{xGDpV7Ckx>&ny{oS5A$$RbVSbkOP^^3MuCvAK9s-dgy-cFsg-Kt@eVv~lh z+G4wyI%(Xbkzd7DjausES9wbZ#V3_c>K&&}iY*;^qsV^tptD8(&O!OZo%erICQUl7 zW|F#e1MZ4nwNzato=Yb!UA1)5`zw}CS}|$T;p!HLPkMh60+SBM|BANZHAOMLH`mAx zD`Rov`&fOvHX&tMwpxx(Y@-i(Tit%u4aHjz-)g@l+vayXag^3K%3QN?5@C;x>E5_U0;Asy!jRHGo^s;hyvY{PYmM8aKah4F0rs{Knv3 z3vQ1uq+b`5<9(x^qbNbW&N~4-w!Qi?!Sj0=KiNV036vh~?b9zP2_eJ#EVn{=FTRV?Co>5rJP7)Ejfv zSrPc60PZzw4#E5WC5AeL4_2>We9YX8w6*k143 zLV8nk?g-P@1@Hm&%W)o9Kk#S?IuGy^;??qg{`zklHsIb~#oP+F%wjgg6~*!nO1~XK z-wvS%j}vJ>MuJ-0dpkkV)O|2_EN`YO;BVmdV1_qI_$((#zdOSZknlso^ivr=OTrJ8 z=}$&muC&LB@|9a8{IFT8Q9i-TBk4I7O*37VmF(xgqVV$Dr8(!N` zYi6u7!A%7(RaAAXs}1BYW@dfrTK&0S{EhM%>&&0-viJJAYoq*zF}x+ZMRMY^cSR@845x-g<28w)HEY8KvEVmFBv9`Stn1*O?R6y-`?aa9gV2&%2;?w)L!jVldAE z7krZ7PaEG9;BvkSxUS3q6@y=+J)45^l~nndW5;Ch zCSJ#mnXy*${`}ySvXX<3o$Mb|TV8g;m$&BcP!GQSmO(|;#~hMBZ=_l`^~}cNLG#be z-)yc~aBs(~qxUTLj9J$m(Rpjrb>}tiafH93cG13v9*<#^4X_1+t`QyZn$@R8LzNj| zJhYH7TscxOV!)S0;nuSoPL>S%Kh%-1oEJ9SlYdAtlnXJUgV4bkuBYBcNhwNsrm_+r zDI2?`W!vSyr@I#xjkxVEhmJm^ZjYj(@&A{u8#DJf|B~x!cHjM%YkwCOFv*} zTiw(B$XQd>m3ro|qjz8a?xoY_+|;?(ImgWz{qTfS?tF2`*=q5Fx2b!p?w@Oqy}M_x zo_)7j`g;C{)0eNR-)7&Qy}IsNS_ck^ig6qB1)$grh`$`DOAZ@ugC2lBUHzX;F|E3x zx1Kg8Q!=7-i`a|ODl0X7eBq?=j z+J+wlTp&-*1qrvF)}EInHUN&b^_mD=kR+AUtWA%=#hfB|Ymi>l8QhC{3z_doIWs{YKi>nqeDDBNq=aF0w6LNvjeAub#+3Jfx~ z^QX*p+;s#Mt<>K!p9r{-yUrRWQUgAL;G!4S?g*=rhrcQDEEbS(QJb+=Up~4oxvyBu z=IpP%CKs?xuFHGZ5k`zRB}Qh8x~$de<^e|f!JpQ4-54B`b!Y@GL>E~O6s_42xL6pe z{AZPaMBri)Q~G&oRRk_3J;9sRCkPIO27L?R3$?QXoJ8A+5skJdEFTKWVUhBMf+EX@ zf--{3KdfK*ooa8Pd@0ydd)6th6v~Of&kNxB-TRf(vf&Y?jOHlZdTv8K!%=UDe=sdk zFUB;)2iD(cznc}NG_13Qd^QxJI|}Wxg(4^03&r-{0B#;rXfG6`O#`^`W(1y#z`u#W z{RsR`sUQ*lCI@iQ&PYEbT?jcsv=_{G#)ewsP04&qnfXR2^htf%wd-%me>IR$B}-i? zw>_7Z|? zmrHo4B-v*({2mDpl_dK(hF>G$p^{{u!tk3UoP_6V(atk5gn2vqxz1SFUCo^uJ33F*tz_4 zL!>Sh%Zi>6GErrwJf&OK*e%Da;mAKYa?s|n19MJ%Wpe0^XN<|HM!y2{hh45i@KDF* zUpeMPQ-$zs+mJIteiI!&dA+-n1+zQ=aLCq}Chp}Sq`lCi`f1loW%qZony$?IdIn!p8$xinjr#e@h$?OLAs7D9wgYx9Zlu)7qX@$KJE}$+MIiWduG?eKZ0W?4Hs$ zYf}lHFBiiSz^$j$FC{q-ET?tjk?RU?nlZ@lr6YryvNA(Pyk(IbScCzjE!$h3o&_X|CxzT03doMD5zXnHWDK zf0)RFvS7x&x&v#=#ly};ceRa z>$Phyn0#)#c68t3tKXPU%a|M?ON<=C&tYV~L!}d7P9Gqb*eNNOTrp5Cv3Hho!xaPN z5@!*?2g)VRnG7EwmpI2Te1KfyoW$^mT%y8kX1^wlw^yheB&AE^tvXo3g)uLHOCz3o z3E)em9CL+wy-eS*-+$|`oEcJzA-sH%FYDKTVPu>q%VGHp@lia8_-+}`a$3HuA3hLo zkkL6bg7(9Uh=Gm@3SU^tkki zNw_yR|6Rb2^c|(hPw=<%KT9}PNbPAV#}{3G;ol=*)1IDh7FDB6nc8};?}vaHX@e!R zBlzkKcLt?vKSbb9D{BI{9xK418s8AY%cUa2Rq<$j^UVAYZBbm+;Aj2?jB=|svVLZz40Sv>qo#EF=LJk!f<35HDU>HVBGK?n%G7P{4 z;RkRgc(i{bqbUep-^l2zU4So_4CV=SsccWfHva=}{qVa{Ji^NW=B7UZyh8{WOWLMX zebmva`*42kXzhj3`JWzEozd4DCtQF1FUMXFf)iOjkkbnyD^~lt8?F{ZFX3Ow=|%9j zH~ck(qY99afMa?cBBs}de6y&Sp|)%(nbtHSh8$O$IlF;_V>0uR%BZOU41&#Zp!Y2I+BoM4y`W{aEoJDsB|OyZtveY0jf962!n&W~-%B{J2pH-j zIU&B(CQEp@f&wlkL;xo@m>9>6;$IM`WQYC6OF0aR@De#8z6>n~aCl-|AY+(YOIGrw z+8}F|whtOH!GgpH@Cg2jj9|^V zW9K!)Z(n))*!(v`?!BROTy0gldehk4w&lx=t$%s_A77$YlxHpy6rp-(4Wwg@EO-L0 zp`HN#g0?M`Wtm>ouUR7fz%aSe%vH$eLUkS@BvW`}uuDdYfx4}aQ8TK##faUGI<`hD zR<*wJSl?M|AD&)abg1@GzVR6KU+Q0<@s}RCv-&~bf$I9#%r$u}|L~!GtF<4`Uw;Iu zn*p^0#iO}m3AUkjfd2_N(6LH(L&9N*JWvqiKseNn;|LC6K2i==pj&Ar$S;e+VOoAb z8qna6w*@+MbV~~D<&Zy9@M5TAu|GyvNA=}L-43z#MQz%reTS-#f1F?ZW=AO5F49ix zo7kslSN9zv`>@KqL7>(slPG`Lz`~B9JqUgH*8EC%MTN?fQY$Nk%>nyvP>HP{^yA7U zW!mj^=Ps*m*m`blzD)i6(|p%k`$4!JtDUf0McKsVo0jaH-);R1+D$Lz#|532V-?CP z`bVhPWSz=}HuaWv$7skbuYz|<_WqGXBJ4)ObrT2)`ErpDNGJGH>b?Qe^^dF(0X(+5 zkS=8hXGPNkxicBLfQ#88%b7rM)DsF3`DM|3fH!fz-J@^_DK`@wvoF%F*sjW{8$L(< z5VZ8XK(M+A#CX{CKcH3}W}Rp00*c;E(5^R5k?3zWPL=2f3E5j9*B=CS#*}ZS4~P2) zvqD-^Z(zOpRcZO}Ui8pA>LtNgQ1TCmmnoAv!CUb*xe`MUP>IYwA&JM;- z!YxrwP@klgjf6?oH6bDiZ4<<)3a2Y(Hpq0WDsiF-oV=;7DoIq29BP&gaV}iB_N{#9 zCKb8#=+fbXZGnKvn=QQXpaU;GDXyQ3?(bW9)Ga40dElh2*WW_*o8Jh5Qw9Ff-{;hw z!wLM2!~?;f)*j>jmW%d;>7P=u#)RbvekRj{5nQs-a#pKSE|3i38;MU!->mHs;4@l2 zX!SeB=dvgqG@1`+Gz!N`?+eBuY_S3kT_4`2U)y3gj>(N!X8#*%f33p*Hmccl5H@Ke zrh|N`hJ*F(sgo|AzW@0>WA*#{QAL%c zuGQ+b1E@m6VQ+sp0vDW$(mzWZ>6ltz>mfM&Y@afn3j_XTr zTl%^hTeke+dSNKXa}F9NEWR%aaaygB4VtU`OLmyxV!BE=rt9t!-o0Us`4zUo1?itx z9z{M~lw-}buEg3lPVS+|B!#a%1M{wMN=(!un<##>Px|>oci1_z_px)!^v(aeS;?FO zC)6BvY5v;(+EH`f(S~IiWh?KQKD}sA(Yh1F4uRj_aWSIQ!PV+*k&YMOPirr13=S(J zQ5YdgP!g5%tlG5E^z(Sf52Y`Zf8TGUZw<=N?ss#7mSAHy{FTi7QxLkzbMg0)#lfO103Msz4_-*WPMy2aa?T6j5cr~aYF0K#zC8fnUYk}Ilw*|?+5?a6 zEdktoF#^y12Hful$Cg&c=i~q`+9NpzjgO=kG2;vEvS!NooS1kJTthnR2anKY6yC2N zGv&HUw7kZ6~LhxN({9BCI#;C}T>JX)eZ80Y6Ov9ADUntD7@E&Ke&KzT*h4O_y-i_yFIT;cXJm8Xw@>F?^PUv&IK_lHqeC zoHRZc^lO99|H^vwsue5N-}QU> z&2QmR(x=!m#P}w9{SAgJ({;#kDddZUB_l@|bS4=&Ru*kDk4Hvnd024?L(&efI-(>H{mz z&HsG+E%~3%TM^w!AoMk{A#u1AwrRIzVRt1qKnV#ufAh0@4jX#TkGJG69Ieh?a@mZM zl5J01oL~Kxx#pNxPd@pLl!UuTbPy9Jv{(Wz7?Fg-6avd9 zIVxlebm>4*p=1TyAp5sNpQ~ES^XvMK`7iC!SLWjfsxPj;=vNIxMHhnDgvK8eYxhR? zUG}dqy7g-Of2lOG=jvD02KbO72KtTg8)(J_!W5z|Qcso>025EH%sp}f*w`!p5mikM zA6`Co!UVhEj{nUCS^M~&6SjEto@2-S^S@36bXP3z7to|*^3yE?b6D)P zg8T&AdS$rCNN&MfvY&}b zlVLJNm0_TVDPU|+AqP@5hh_XqWfZ^LUm?`d-TP9Vja0CTazJ6Z|RlbP_2sA0p|Y z2#t{o*o^*gEd28#aIx^qaM0qs--#*WCcSwh$H$ke36&1UvfFL4L6qF*0 z(gX!W6zmOqLu0{)JsM+)m{?Jh7^Bf>dNFy^OpkA3@NT}B0V`pbC9y##bfSNl+mTzcAtVx-8xWvkG;H=?? zYfid4?EG`xy3eNAoL29y99KQw(kgCt_tia@l=ZiB*wo;bb8dE0VRt0CM~&|79Mb93 zwmEfqIVVaAX7n$N7@pbPwykYw+SFVjPvrsqP(#l!johhmA~d!jy6di#K`2)r&( ze@@VET6#f24u<2j%ymBURl0*+LDklwiz4 zPh6X6seAE-low|^zTvq~?^gQejNCP8J?+}s_Q)I_HK}iC_K@Nexq0X3Y(Cp1#Hsg~ z=<+q)3zKG@%az@e!@;X_d~{Q);~*y}?a%7bE#K z?~ZeFthu-5uKV3J=N+9^tvTgF{uD&}Nek^x34`9aQ6t*hWSw3x?{d-pRSV{?uj1My z=1ryh=;JU!L7I#U%@(^5pc=RE<0Rd(xDl9KSG;hR9#$ zh8n+Mpbf^9X@iS8h(Gv%U+BG59JoV6dBgWsJ6LB%_HY~DFF(fH(&3?z!ZVV)ksAM) zDbW>*{+8kj*94YHil7`#<2Q7(l)tsVq4O}~w^{b6`e)N3Nf?9ADRGHvedAB+&p-^I z#Z(boJqBe&f7J-wz7zkOi9#2 z{G|rp>bsSgDLT}=0vCfh+djaNvUdS?K>*8slvSP7yV}g9*5F5I_@w2L)PvSC$HAlh zEe#3AV{>2(M$;;2`nKe3Y!1K5oo0;G3=UnkRsp9*J?jU?`R58aJ#k@HhpXPjs-iV!!{C?=ZY4`&nMGkNe~bnDS`JNMRO4^- z3OhA+Xeh04=j%z@R$AiX(Ys%3vh33ot%Vuq?$M`jYqH=?eesuB=B-Fza9pqm4c;(0 zdehc~gsq#R`6}h_h+Z5^|7hK^k0v&ZkPtSDV$1jrMHn&EC2V;y5hYx6ja%0u{3?G; z0jFv}>L*(^wWWUOz9hkO5SF4oHV}kP+yp0IQ-Uk8Z@-Kc!nRiiuV%VjL1wI7?wUR zob+w@Wo5%DQqh0(=>Gl2jP5U1B3jKxWJb>Wi~>2Fief0k>5L?Jwu&jp#xIaPNHH|P zKM;mf1p|0e6Zl8`c?mAf2%87(M`EuJ@qHUCxvmL+@uYsDIQiXA{JJ|2>zmz|l>Mw9 zI*}XC7m2Oo*B6z%5KF#kSgYRwAwW->HtHWC1lWxUwnfUF{W&S(3R8o7L(@jk|N1HE zLiUHPSvN4c%fzl?tBG~lTtk~S+{gOGR1jn5V**Ma&9{dO)(X59}NCw$6S_bOLr zSXs3K4#gjJRO9T=EUVKWUaovp&eD+*}5i(x9YqjZ{Tcr$|Ai$5w~Hl+i{q zALFMqC5rkOuRnEO*k+ZhQ%utEEkfhh>NcQ5D_GhHK*LTWKH30h{=0_HJOnG}1`j^a zqf*#+U}=3iyA)9Fg>a}97X5zsNh zb5v8<6+%`HNo{?ie)r2h=|grE z)Xkc8I%mq6snaeNW6jy_99H6}!j9u&KB;WfUME2($Y-knf~|o`=d{-nJeni{A!Qpw2divQGBB1V2apWRyE#g$JA*9dJ7K4Mm*;TvTUHA!7odvqm6GUOwSwm3C%Jk8#DZmc#c)Pp-DM zNuQXvCc$#BbUsYtl2LOvrn-LT7H>3x%}dUw7s9rDX2*qcU^q_KQ6D` zUpHV|cGz`wE211;dLbE&5m>ngKwER4lpQ>O-2daE!`Y_^f zBMK_|+Sq#?c~F~X<2GV(-ptzZu~i@Jn6M?Ud-?o{nxD#5z_b>*9k_E0Ua7kYLCq(5IgpBAVp?{ z^AWE4-=2w@y4RL1kB&YAKj1M4WLdBNnn_1bC1FA(!Z2xRzAM4M$oNVwXko)~eJ7QL zT03~`tvVmqF3hXRb#!buN8j4X;pKI0LTy}ay{zJs zTRC~VM5lWl=4&iHQ=?Xi`eVn%`qkA_U!Pu9R|J>hoJ)ng=Y@skmF!w$=k5zH7$~Ya zZ=fi&NP8V1)DSLZ16a-@o{c&xn{{Mfx49T*2tEiFyLSP;?;Kdv(_ zvDz+qJuh+FSE~|NO$xU-ruOTb*du$}q#oT%QhVo&4Abpzle=J}OYytQ7eBf(t$jji zbYT0)5Rc3q=hG9H#(DRO?a+H@X?%O=6MIa04<^l1rh`FFDAPS{WP3rgqW%$?`W)kn z8QRi^E6MWbK!Z;Gp4uJuUMS`L02ZIqGJmzR^RrOmYyI1-+t{X~m5XCX+wNVg0`%vf z13TXN+~I9%egn@BjJVP8>a&mpLuG(gSRXJHXoc;kh+fQ~aJp_x<1o19D*3S`IP&@G z6mV?oaz8l)`_kk6z`&8)({YXc(+UF^mm2BHap;ah`!n=o_#10X@uNAr2F8|_ms*+d zW6D;hQqhs$Ui&0={Rfp(%6nN`rk#9n)Jx}*xGZYo=Im}2vnOoM>0a(~@7AJ*cVbSUq<8`LPRV9x%pKXEEWUrzfa6%$BOT4BRWHxY$6*@ClYy z@~g_$H2!`Tc7krjT*tmA`n4JyFGX7Qr=EqbICB16QXp>V(TT-b$@-_G4>V!3s$Yi| zv{MDCpan^D{RTXr!&v7(cuc+N$2_0?&hKoZ<{F}(Vb4?2%F9dxf$e}Z&MD7B&O*J5 zD<7e6^(-P{19{~1Ou|7%)JS1~->8QwlC7^|fqMi522clDpg-F!aC&+k1UpWh>2KNd zWM9k3?pQM5-KNZQ1(*k;L~v)n)brO*a)#+bk?32+O< z*t){A6-q+cc~$G9g^lw}*&5`+tAvIx+ys@A-j?HLxjpLtEYdFEJ6qN)Z}`-*=DD%RTOJ?@E|FdVr|7spv z9+SSKz{aY3_3|Z1{`P)v-@wCf@`bR6aI2hQc(OjQpkmz7j?>LWrt>}gB z+H6!N>8w2tr&u-v+=6hBJDHp@ihlBxqZ;9G3up|j{WLgoOTFXBeNpNK?qY(3 zq!@vlL1GwRh>Z%VKs-G^8p=AbLyWQDpfwlm*w){^gLSx7RHF4~4!L^^0wevp+70y@ zk!JUq{rIh7n4m>ct~PFz+PiE|lBHR-AY@K>^q;~FF;EJL7-S9x;b|rZ4gNiKDBjXV zKhfpj(+3nXs;d6^_9=sk4>72(a!?o#Po5ul+(q2xB6|OSj@sA`qrT4QiK#wynl!&; zT=U_qT`hH}*eFvq-IX8W+KjY>+ee5@TJ_>BTT~pI>DuQS6MLd|{nN)JUtm}!?~Zy5h^dn<F_c_2F=vV~{H0bV#maaHA+6)79I z@WVH8-H^BEAIWvuYS(k}xaHTJ{cT+nD)XkF%pShs{Y8ld1AWaMy^daV+4g;(oK0Ds zD`t<|oY!NH7&&-SpCqrC&?U=07koyAWz{X6_xD?6!Gi`lsZZ3;U3(&lD;V>_$+US} z#%wy3K6@i%0XWKshF-KdV@rny@aw-oyQTBPz<)iHW=Dxf2mal}pzP?x2D5`NQoXT= zH@QiTF%n$z{BsgKSA{ih0iI7YhmCN$g-Y#nNVWowD3b)Yh?eu=*rGF)EVa*Txqlpl zzcTg1Hgwt^8~QY?f@*BzJz=C!Xz=P7Y+y%UWMmrjF^o*9JQ&C6mb{slxc$rJ6=(BL zx9XD>F|N9x=eikb8^&9=6W?+r=bT+Oer;2DZDr-7x2C4=pAwedHFe9m%(WYbR;E~2 zEm;L0Ge?J1w->&}MPJ#S@dSScHi(w7;P3j_cCJYG=?TY}r`4!%J1u+~&&G?ZKQwA> zGINd<@95uf|EXvaS}0hc>3Gf{6Qb2{2C!YiNsB)u4>FlBrS?JXq}JVLS0xtLM7Wfl zEOja=jaxB}+o2y1N9XLk*`euYSL=IouPs~~xflZcGYl5=iM5o%phbVsm6u17q15a7 zspR#9JAh(f1E&|97trX24T}Z)o`4n27vpGIG*@eH%j8%y&ss$P#L6e2+ThGpIwPJa zY~*m9cnP6EA-W~x$%o6AW9=xav1;k)j|{tJj!mA?DNypFo9J}=v4ej5DYa)_4t9&V zKD)GLu*HQm=_Fl&hVxXW6dtmWV-h?)+_1bl zmP}`_!%Cxi?!bq&skYsZFD%(M(8cvtC&ww%S`T}FUdh%0&Jot5!p$9ZT33h8 ze*Lo}-4RhKm_A~~INK?;qL{%Q-?Av<+)A=Hxj)Boo3wmjY;lY@Y3_!|%1q#b;V||{ zHa1Qf;0ny**~2qL=te2@UNPS>`Z-Yfs8#p~3vXK&^EPdn^X$x1(RvWt5so!Z%{@ud z4RD%a!1MwVoElfNi~*(@6!5$zaJnZ@Jwl_&{iXIW>rMKF!iTZLhtv-}hDq(SMX7|^ zKry^Y`~1d>|3p)K;D}N514m464q36U-WpyVi+il>Rp zA^f1zwWo!Q`rT4{z6PI2?X}!`a*y?cjez@%ovvv=xI86pV{iyJ_Z>UdNboPj-Ntdx zu|pGsqkIgCk3hfnMb#z~J^>6LTy@v8EJXS##Y@K41V4sRVvVzqzXdI9+n6umOwO=< zRvPPCBZ7Jti z^Mc>$$iJYn(_@r1&k02$a zCkXW_Q~Q1s zI+l*_KQ*-**)54&;{*)#8=KW}QBoCOGX|rSKEalmQ5uXIZ-jrPfafXT%_h?F^g!W( zproFSc?pxIH2ecUDGeztv46vgn2(K6gIscJ5H*w*dco5as6UAGCDOs)hwJo}evgCL zS-z9n!to1;b-NK^&>LKB%3|s`YcKBrsqZkd-dM`N*<|NV3UZrG5KQTb551HSu%$ zTU(7;moX%#zk@~p+Ay=={;j+Fsx{rZkq2oHUwyzOg_S2omq!Q36%G_vlcgP_yhFT# z`g5O4<281GobAGRt?BJl;BGRlVDmMP_^`QzJ~O1#md@^}K^rYHz18na+{p77ULwC% zeV%OLrbuv(6*Ondf^ezCjg8eJ*3Y;w8qM2Z$UM#WVU7p7h+2y%TBQ%m^&|AYsWsL6 zKTzJKaP9|sN0Id&D8XqON+V0lZHy9ED8HwHK14)4~!W#!h>)7#42b%@!La^ds(#Ml9Xuvu%4sRlGY!4;oT&Y>Ttiu#(8QXiICYgSlC!{vD&uw4vR|4VSCUkmgdBEsHB15VIh$ zUEfYa26t*7TcxjQJ`qrX7bN89Dqt833&2_OW zTgw!zr2>w%MD{Aim$e0KNXkouCXaIeff|CMpw^;~*8OyC4Y}_sWHj{IB54=R7NLZm zfbqFj&f0P|W^d8cxRj*@i8UOf2A(+@WD_uLS;VN4UYt{#iXEdSY>af8RXfMJLu$_v zbK)##)XuZbP6=O-VN)t(JW{7k>p66GoMqzlu;iB)=pDHK#b>r1TpXzn=G9~K2h19y zA3_hmxHe5Ou~GG5Y|!*H{@l0;+nXKE>)rnawoOi+FQuXK|k$--a)MgP!B8bJmzza6$ zZj7gTKI|k8)ofuV&{vW?$w_kHT)^pRhrw~WSSIcGxCI2hFu@7(w_K|W zGeSK`EjdEC%`JI8B-DghcJM*;4kCRch3=2kACixOXx<3ZPs;s~Zpk71V?*;lG}|)u z=`G{ZWE&L*w}hCC0-dXbf*e<&XlUS0c!FODCarB-bDE%jQc6&-egX#cp#(WkUFiQr z1TxeGK!D~}P#P$7@#FjK6KEoE&)j+&K7JC8Ps^bYdbf;qqd)az8OVkWq^U}2dF7@x zQy*l+6Hsd6)oXF#z)?8$2yrwYN*HG}Ut<~bQx2nfRA6?>d7_SD3MQA}Zxh&$297kf zM_!iX=7z8c%6_h?E-KnnA!7aDOpq=2pQ`NNg*w4MysO1Vevch>(I}|QfqB%k$>X! zFXHX?1Kcb%)^?WpLWX|+`2BIc^NR;@b8#4$KX(7f&XdX`^>ZNvG&);|)wH9LKPgQ? zvJ@2gve>GmgjtrPxc#e@@LVVD{Ax+g_WT|_O472o<@cyCANzwpGzCL zbOg;%oO2j7XX7w;$pg#@Y4T=tB^Z?H1_ssK$ON=eZ2UmDA*=wsrM90CdICdRHBonD z_0GD1ETM*KVQgCfBO0?|T0;GZE;uzkt-U?(UE|)WYeKKl7a!1^!j>n)-!J3T0&icp zqt`Hh+xkD{T*9Z%jbY+}g40k-3%wz?L2Fz!=7wSTBe&s_ z5u-wST|r*AudPk3jZllu*(NBJDTxNkoJmt3^vc6hQ{T+3#%N=H$Jom(ZF|Fr79EYy zLPgt0bBrS@pz9?l+>Z*nK!%f;Og39P;Q%k1)kUw;a{o%o$mjlpdw14(_-N+nvf={!ckk@#*4lEWF6ZBfpRn20 z$2G(wt<=rOEwJ4D9#s@z3PvKsl1ggDNZJ5%s#cmN-BL!-P=++9CSz-@`9HKm zk`eB0WZV?8qeb{=TDzl#_G9F`Iue7K`bjk~09D8u6Fg0YO`(N;(wnq5DjD-E;%RDc z6nKsK$AYXxnYyQ?%>uCK@;z^34Apc=!l3l|2d)kGU5W3LwIx@+VS|3k^XSOO$jccd z*R&sK3y^IiYXxoy9|bPQ#X9UwzhVcpU=_&}wa0uUX9%(e*rgX*EdtS}g45hvZu;MQ zkl$QIT`1?|+Hj@nospg%`|fY5b*;_ZQ{?2>ZBb^eyO2>Wp4%iI+p@I&YX*3^qm?75*|yX_RgiECrlhGvTG{MTYw&up4pfgeOgM6@jik zU3eC@uqYVqs4(-vlGfx`zgE+oNW+3hmHDYr5>tR&EWzhTi)f)*D!=~^4ahFbx&;` zWM5x0QTee96j0bY8({QAeP z1UxIho|)BrH(|0|TU{$fw&jdE4K=BccvuD>pWx8L*2}V=&xqxzR`9zV9+|_DlbqS$ zX~Xp4&`h@CNY;ss>?>{J=Vj`F1x$29+M`+f^iQ(nFDt)Zg?sa)ET_~GW{IWDlOsamjvH!9d?i`{$ zx5SaGA6xo+pHw+S||MeojgA;kbhYVmxa7 zWW9sr#^8QOFZT4rBq>PK3Jjd(vg&M_cPDSHhizB?KDp60v$)^Z?T_Sma**TNq$TT} zK>?@?z6ZF2r-=qwNWP`T6e15tPH1WUnE!?bw2XwM{DN%-Cs2L_(?o3Ha8~WWG|pDk z(>EoX{LxTt8+N$G}$}lCulM4_%2%={#Gz%+hyi$mO$dT@*4#@0^sgZPb== z8CkgvD2gX)q@IQC=-B}QGv;ok#wL2iHHfbCL3V^vLdc-gCw4ZhU1Mugutq%TN|L`k zR8B z%$;%GmTs=mNT)OYM4XenQFI{7oRA}CCFA1fjW9C{i#t&@*THt)8>PDry7OwX&nh`L zqAb?Td}zk*syPmp3G?^6gjJ+yt+bwLBf2JZ={`gk?6;#LeRbTzY;#8oufAy=hj;BA z?i}Q|WBQO8DV@QAA40WI{OnW_EaGbP)FXxjA*8m(zd`xyk52}fv#2wDH{W~J&UrkiF(&+s%=PooK>(_yJZ|7ud7hR27r_YBJP z=;%MIPgq*G*9J2E|1hhA?@bOOot#R229X|l(WKzoznK)|l(7<%NNTUrLq!df$fj_O zdG2&7zx|+Eu&|IY)*AWv&{I+`g4);%q+M>UQ?^FX7C}qnm{hK%UUEpF1c&@8d(Tl> zu@fzD5iGS`4ec%3k_Co-gjU*MBfOnzzNsIkH)#LR1GFC&5tql2#0NGK{hX*e;_o|P zS9t4Xl?zX*mj06iUy{uPJCWLoRDyASJRDeYxMaqWL|ff~eV0a#HMg8Sbz)V(e&1ym zN*A0OWve?_b2Ka0!fM9&DHYy)dVEEA;`+>@tpUNh*xp03Mx_R1E(%Yn9$&o4Kgc<1 zKvdeqUO*%nZoKulg#%O1p!^voRB+{V4mLbX>@Xn0C!ca$@zMO*r&AmpcTVP3ZE$y3 z`SHSio_k|wC0IJlt(@EgOt$L=8?q+G;b5JjHyOPm4O9lV> z+k*dwE7`+#%o>_GB)mhn*r?=+)CsF&y0qTr)M?bPKFPxpCJh@9(>Hd|q^!kIt0|3t zawpWe*x#tFfO8Bf zJ99JJ@Gip_hT94mTecAMN1Y-fUA0_|#wKh*CXQLF8h_&nI;6pGFgi-5B zkWfQ5J<|^_&e>DDNAJo-yE;70@lJO0pEuw(D!n;?Rr(2qqI2=YEH+r+Vl<((zH$W7 zP#{qbBg#T%#F%j%_?})Y;O}g|F~Y<1#fHr@j%NxdwS^_)_Z2xhhAhe&SQg*PtaIN! z#U%pTWfJK_o@9O39v2Co*QlyHQj&oNC-vGLVf8vD-{ zf56LIoN6=(vw;yt7HPh@NN;cvPr3L%Cel?^%|*H$M4EQ>G?8xJd%MR)x~1WoiFC8E zL8j4I1dNU8<;sh&lq~|L#|N`TU~nvgC0&F@5sP4Ik3|ThOU8)77D4JIN7aEX0&4gw zTGK@s%7}d!i(u*nhONl1l{L=9A}HWkgzan*C^iCH1gZaQ@d_4!sHUJA4NXsWYNk7I z`eMn-NU}Ivv~{5KwAvEK!m_v(<83FOJbk2tOYDc%Y3EkcoaA}yBPx-PU%#8!Y@dgj zupVjXgf9U-5ISMhU{D%5k)U!|6;O4&dK8k>e4q|nDtHR(*@O!J&|yp)jc<()UQW@+ zw~QA5!;Cf`zcq4XgJnzj;moOrbA-;=A4dSDca$-CT3MDRKUt&X1uD!!S&FqqcHN4H z4xc~TuBdiRtDt_KZhp(AJKD9iA2p?@Lx+ip0rqX@QID8@O#Z^Y0~Y3R`x|zKw%PWC; z)S3i}}~=x+=qvhACv|7e?qp1@wC( zv_b*>#soEvhfqCOsb}cIZ8OLunv^rQA#*Mr@JIL zl{N(~cQTBl(y)*`oA1gPwAq+Bs^O+SjF@W?B1tq*P#w?U)7a`OR=EjW!4*n)(|(ww z(tcP11YJ!;DIlr9k@b#E0@y%*6v1jN2V29OG1(qQiNZ!Mt~wR%Tv}UV+qQ62^85tb zlG>@x%RY0M_{M7PoIZBk?0)@bkK<10!&knUjbT`KY!d5DlVH;Gn=C!9&80^%m?S%# zd5c3b(TRP&agEH~3?Cy~HisKTrwB1~kOO35_+#N~*rGLUZ(uhGN>wFoZBUgA%9uog zO3YmZc2M@5Eq-V0Sz>yrwZ!zz`;eGk8jlK-)K(Iv<~`@AVvRlH1yI9WjOs+{#R!)e z2WteltjxWipv)CZ)kKwVXmP=enq2ebMUy#ruZdYG6B&7x!NIG>2{mbNkOYP}o2b;@ zAW_nUD7h%yAPIE(l3rs(l=T{gmNU&kp<1KnU~uN1l3Gi;ltG!qOA8@Md=Ye`Vu>wK zUV%xr)Hl_ztZ$=+mHIX=C*4yGBBWp+NyijHQz`m}rV3;7iyjdw$Y52DQF)jqYhc5r zbru%?d2YUggVmr8qe4mA4gEIcW#r}D$jrQ+%WbOPjg1uQpZ`tzD$&vM#DT0lu`l5s znZ7FSy}?ajsmz;luhTE<^Mnjw$;M?`=7dDs-W2!sU~*uCnNldM9aSfW4ke^`1JQ}j&VT93tl5i4T$WIgNcL68S z1(<3gcOFiqIFWyhI8gzVaLOe+8J|eF8sVtrey~OORqkyCoVsMBesZ`t365+#P}>A2 zJ(xm}`loy-wV%X6>w^||z(YBpg}X{B*6b{=528#c)(i(3V+^ea=jS56c`dhX_tqIb zGQ(f_SpUG0#9p5n)UEZ{Zt3ASxrNKQiJv7^g@;!qeWqW(SiihJVb#Eas}kyAHM238 zd*>osha3`0ah2gzg;F?F=zB&eW}snEuJJ|Kd^-htK7gIcvz0;M6{a4Xc!cmBB|i($ zrq0E*&iP zdnJt^FX#_)?{IB;cd^{>Jvv#O!3}SC2c)@+(v(!ZVDs^z&Q8>sWa3|p#m>LFD7(<~ z#Gp?1AaFWK;L>mC!#Yo0G9oxaoU~1_vWkgI7bZyZcO!?}^r${%YENrSpY_k!^RhJk zZxN87Q1u=4*D8Azy!Cj$81EFCoWgkX;z@|`UqyNKiOd`v5awuQ?P=C0vYmZQef-m~ z6EaryN)7W-3#&Qe*4by_BdaH3E8)ec!CsA_hu|mEL>TVW(9vbx)5hk#Dl6+_pE#b8 zI#eJib*P{75@Q;PYDPE=-_0gCQ`U@{n(N7wnWWZKOjs|7x<;8=)3Ts4d=kM;8)JK8 z@7OnYo8U~WjQvotlKVHd$9~$~)SjtG37Z>KMM|w1YXfTdn{8fFulcI8CioNPLdAvk zX`cP<<9)zXPo`oG{Vr5#%%t-QyGNmIqUM>eMyPshiNrlOxQks%mc|7~s{fLgr+yn} z)c( zCuQgl?{i=Rvmku9)9_e{^|EO3!z&w{eQkTRjSM2%FZ2(h$!Tt%xGl6`#E61WGC`jz z4nZH#7eaURVFB-~u>_Ql*2kIqgVQ5)GQ>si$8iPjm3)B3fi(@^SnkyEA429Gri^NN_!xnOkcP^SaIg%gGkceSux;N-@}9Vzr?RE5588f%5KKtUt^ z)n=jNhM~==)+}&r7}`Wc#@LLA5Z=R-u_fZNoj@XP4hd1fsTfJ{Jk`P$;kgDlYz@%Q z*gi+)z;-zwq{AxUxKYy@HVCI022R+AyCKqx6L5|YH7e{-A3Y82u@C{Gd}PqYhjqlA z%-eM}$H z7@fXK#;dfF1i|x6z0dg9*7Z`k6&FZ5{G=^?my95}#j$G_DF4oLJ+e}XS!sx%mZ znd;P^+^@Uc@F$}Qu3oigWFf3v*=n>zdKiQNVF@iB^QHs%AlEw(78nwbkaA4(eSEa- z+{wMRu@QZotn3HI`itkCgcuL~p|o}VESv_9>_jx?;cyvH^oR`0vKY+PSZ@i$nrO{!#^-~3@RQ|>%D7y;w z;19t!ka@A_B^uv00u3-ZQ9FbC{ib+(k&n+WWbCaaw}|f|Zzu_e^UtC`IeRlTub)@W3N34)HmX^MP+LqGY z2RHCTd)TAI6hwl~S*yhY=;-T)2EV_(GwpNL55Q|tl{+r><8L? zS_E)_4|Z?BGg7ZLHquvT0mrKjM!4}+2hC>GF{JZFH6*mp+DcX9&!T;kc?Gt*!1rgL z%k#tM%K6dHhp^A(d5ZQ;<{4m)d8)LnE(i;lD*X(fE9Z;v3tQOt(W62LjRi`V!C#lCJU-xKxg)_t?CJ9ps*={DB#z(;Kl{IqP#{dMd9*}C%ky@4Mc zuuK)Nf4D>Z=Td(DCDIDWG(d1RtAC;bnZ}h7QKE+?9f6iIn(DY}O#UAGmU-JY zR;|ri`FHn^8sB<3>Djv^!#}+BAhoBpcZjn%YAn}}d&$pRt=?~E!||CVVk?dxn{8%& zNUiqv(~s1@A<2Q3$B==7mP}&wl;ne{yv?jrDih}{(5c-!ge8PMh?x!}+8-Lv_o+pVL&N?~Ftrk%y@_^Xm&Ra_^Pujn_ zvV57;qgso}@i3EA^cq~Kw7Xiz9*%7-@339TyJFWGw#~wx{!vU4{s$ZaT8)Bz6JXKj z8_ItljiTqO55$WaUwmdub9)dxz;1)BiN;Cjo0s@c!)ZE1+x(|+bomz2M_4(XJ zefNfcxDEOO%n5H5v2Ftdn8H}+4>W6;Jdc3W>ImdTqXsi#-9dl#K@4w6sgL40(!joW z$hdCOEiT~)eL8vbjyU6Uyl|tR&Mnh-)T_9?`ZUlv1>dB_%X!d-?m2R)0SR%Q3)@w= zOI&=vZZb(Dm%q?Al7LUdMsdp|Aw-`eW}~-ku3GP^cjOl8XVc!q4;la2va?5Mh_{Ld z2ZlN^7NDz0F+<|=dv2c)S7O)6bH}}E*Hv$}_IdS9{qg(f^Y(GMcN=csE9PqUidV(E zFFjbapWHb4`gi(F3Gs$^^_iHi8Ib-SNZZpbObRqnL&Jwh^vBFSXvo+o|KIt|59)5~ z-@oEByx@hL1HE@_;4e_)qv6vj3&w)` znvE3-Q8N06m>*!Ddv0X5cYs*?=FB$jTHAH+mRS&?|3u%hZC2IPgTcMSg}GvQ`Hc$` zCe~fAAm{4+h7=6|xo0%KsF|Q?$1X-d*)_F2=+3XSJy>#!WP7}c&9lVY`xY5|eZdW4 zRovS9eewLY^pfqrtaovZy|SaE`nL_PF7a=7Ck}PuZ+5$tsQZHZ?u$C{mF_X_L=~^@T2u4rT`)Et$$a=<61HEY2asD>HrUbV z3GOF)+>xSa)Va*66`*5o&8Ax^9>eN}*$-GbZq6@xt#ki)8M4C#kAMF~a!m{^SQ?fT zex}bbJ>D*254PRHve7;R1lUYeLWD3TA@W$zkP`eo@}I^RWPUd@;y}*wQTp{6D) zZ4kZrri{ai@eGH1M}PyrfBbs~#UA9U0S8*boY4k5on|(EvxoMhF~VlO_UCA?YS2!2 zbKCWJ_aSvd?E5XtobzLzSHz}+zkOZzwHW9j9QNQ>O?-c!7)OpxUNQia4D9DiE3b4qtc9xN>TR)F9mkw*hhA;=F%WFIp-Ek*xyZezBw%#1=g>Oc)HSJ zQo)rat8Nxy><^+AjZU65h_Vz$xDR{-1b;fR>ay?)eo6ioJn{aZu=6E@opB($_}jdWe#hkwH8$q#C?_!75<)fGbTrwj_Eccez|-QsX# zaa7z5lhmIYM{2~p14HOVMZ>$~*zA3&`w#~iC=;4F#N|%CuzVP)<9m;v7hU<^L#@3N zFI7gbnw9BU7r5-z4~j1mi;}>A@5I->cIRq69Kz?GEFLy(M9>nVy>O9emITF3i!D7l zXV4k(;IF@mFP`O2gDnM`e5SnxH0VK5dc>WcS;6m+?`NJ{x!dkyru%zTxC)FD*Lj!rA@A1C62`s?J&&&9Ut_E8H{7j8N@VyxIn(~9}ggBQ+SG}h1&Sqsuj^6a4wWHHc#Fn9EV_*Y3_ z;2jcuvH1MT;p75ayT$)KP2$?YzNk$%v|?@mFmU@EAn-hPs>#H9KGg39z9VVyx(}0VO}# zNg}^)1dErdY}7|kH&*^UFmqt2ij(AV}e$m(BXEOpkt3DvY8qdn# z4|#iy`r^<=;{RZulYQ6nwA78+I$eiZ<2ueu?KUmaD=0nRaoE7<>-Q`7-e0!lZuL&F z;c{j8S(5VeapG|$|JYLT(S=ju!&OVc~skZ*@3dC9;jK2 z83n1}fK?kOK%Zy=H~MHL{f={kJrKv0W=daLCdWyNC4};Mb>jV^K+iSr6V1j;|I29O zUSH+T-|g+%>%hvT_g?b$PPn*e_=>Vj-!}u6z4~cG<#9_3O|P(Ej}F@xF{=Of7;(Fj ze?)xeojv<4HvUpsDb`;A;g_|GpBi0oV*U`Jdu-%DyZyWA!_tWTq1Y?Ml~NyXA%h`9 zr2`EP-5k3Prx0C~YM+1Ue`!AM!eoyI9m8nZy_X=nb&E!*?IDcc6axHSwd@f3OoyuibuR*wFe6oDO`^+qaDeH1zQR z!{gQ8axg2Q5{4iZI9<~Lzj@Ir@Pq`iXJTZ2a}I&h3WeLD#V2xvk@~3~+zS1` z+`8k>rM*VAPkaC7sM&olh)@zlYw@_4h0KNy^6Eq-C(3jMoW*X}D8j>e!&T7caKlwl z2G@=@;*%lQJ7^9cO9nGrOzOupFT>T*7@YcJ=@nEqU-$$_0h_mReoy$^IL{}1Zk+EE zJ~z($37;G1|AfztbfBLb>0rX&NRKCc{v^5>Zrn3^(FhQui{XyEL@fgqO`|oTCe8Mj zxJRNUXKDfTTR)G!)SDKhDKnx^8-hq~~j6$5!=$~nBX zhLm%Jj&VPg?Hj(z1bAF{gRKiq8voJ>OR-)Ys&Me+>zQZsV-~!aw#{Z%^y0)h#c7LE zd${zPo)$H8Sa!h@-MXsnePb5AG_If~uUk;^@;4lYEEttEeVAWT*R-&KqoOO;ZwyPB zIk01L|DMssgK~Q%FB{vn=cMX!6K-ADlw!adqNNci)|{#ru;rCxHdJS)h6d$!=6S>I zMQSF=spk^C>6NZi6$er@r_@1d@e67!%{u0esI=NPZ~JfuGrP67xox_|(<^ZirKoVe z@7{n}Co1OEjcEs?i z?Dk{hLxA5f;5PvHxyft*i$}6BC^9YLhYAbyw7|k*b7Q+{rvL9<`%hAN_J;Co^Z^lx-w_($x-%J6V5oSAr_~XzMh*gEIuF1 zFuudss(9cxGJNQ`aVIL~zq>AW`1-sNGX}vL0R7a{fS(;U-XU0i8u`?QqOz)FS0Wlo zA#%n0=b@?aM?$;@f7+FFBHEzNZk;SNUf~V<{!er!UkoUWm{`4M)cP^j!VD{&X#2c` zH!up@(Ioaxksn5t*q-im5_~j)fz(3#F{)iH!c$f2jBqDK|7oh@MmVQ{rz_fX3USL) z{lvtrxA=_^1~nE8;yLgyuGv7zx0LSmNsF_Geg$Qbn+3ux#-u;q>P?(a0L ze>cb8f=12lS|<9=+*%$zrl8C8`ENq>TLRY-kZaMuiY{3++>BBuj=TDEd!R;Rx75`*cP7JZ3>i_+*f1WBYssd@LDigvTl7 zl||s`NB^xvdm+v^Za}LGaOP^w4YO1Oz}2{o7{L_gI~Y8YwM{5unff-{)C!@i@rkz@Rh z5h{%X8z_e!ZYIMNxTgLEN{rPnl@4?4L{E>MaJmF`MU8rU6mZ~pEoaYMFmBtgM*3O! zy+YKxcD)l3WEu3DsGqy{({@hn%J=U1tesQ)IhC$FadoLB(cgxDMTWk5QhfF6@bAS} zkL@C!L{&*t#8dpPQB^56h(B_U8#h9JXaUemQnCzi2!#L{uh+2xpleSqvvTX@+1kp& zX^2^61)tq;Wmp&rfNCs2lo-^q^NH3Mi{Z zF1I$+BB!l8GWL^Nr*x&QDYU%;I@*UqBLLlDglf&lH0HArY0=UB3V3Sc7zQ5+_*V*e zS|i3lKeYW71w37DPe;AaL`T9U<6{FZ4;0X`pU6W?pchFMy8=phO9o@8BXweHHF#`EGQy-gx6=M_+DBv4DY#(ktuYRcOU}HR zASCQx9ByO3^5R%A-;MWhUipjI=%ib_?t8a~r7!2&IBwiOdii`OZ}FQ;er@f>zA|&x zmUvG`hdxs?iWl_NsO`3_HM5#IZ*qA}|FV%4D@rQgWD@sKZN@lVM|X?n-0lz7r&7k> znw#ygq;{T>>l;=cJ8|@?t1#b{@_z@tYS=(Q*rc&AqqkZ4%k0u1t=v)#lSD_8vUmiW z8Rbncd>alkH0sNI0>W<3z4tf0a3#UrWAA@g=1;?OzT?OS>G|IB{Olb$c&2zJQcZ6i zC1mw7{gGPneidF()*t8E7hRk_?My+=shP7cP6Ge1p&2LBkUCY{Z!x}V4j)ZUGI*+l zD)@)Bw+Io>Hl$v((_RxXhMd^xlfuo;kLG0gfYH4Lcu~P>%;zA{}m{V`Lk>ufds9`f#g&lgA zov7IPa5d9>f5HrBMMxMgP@@H*T!RtGjb5|TxkOy+##gzCRg*0{MM>9-tv9XU4({5e zpBO!a-7SVPVF~#xHxsFc^55&INK5xEG?xz>pye!Q7p{^;iKhqIKzRmf!ux!b;=oVQ@~T@*L|C}PgDJ;#rEmS_EZJR{bvxU9Yg;L_*l}% z1gBeo+&+sSYG-J#fE%upF??FQFkzyk);v>n&eRLtSy}JOD;nigHOga-SE1+`W9e~!AF}|c*^bjb{T;DnxVs5A;pMMJdZ*N;xQ1?=T08Y* zxhsD6J}mz4*nQHjR{YpuIkCRXIqH9{zMP(Zxtjj=y7=cRu|#YB@IKaHP~#sOXGmKY z>4;9rPD_GH;%2O>3qeO9N=`h%ueeCIx9mWKhg;doKG%kBoa*9{T3oo)z3tJq?h!@P zH+#4Qjp=@^V}hTn$GV#pnM$Vh& z(z~CRP}U$M=MD1q3+~XS!@>&w@A?%z#<-W$I=WFX5^27UztO{oEgr+l(uk@bAqGzq zCp#dGCJQI5y_M#v@r*Cs@m^*4FkelLb^Ml!sV|l%`>5NkKRv$uq{4+0{RB>VQ^e?Z+Oh9A25Y>fYuZ4OKfbj+DzT!3Ozs-TZDw$&|o-C27^YH^nx)ZjAma`L&XQ>ny#KY<7~SuC#;~S zwRQ0;CE}}O;)qf!>$uT)jFYq;ztGYmCrkWR$kaW@*-;cc=+p z0w!vb)`e<(#Dnt|cYZ{CM#t60C6n`?=zk^SJ`-<`99}zoq}Y$^#rcc=_QG;% zC3oSL$w$Ump3=-1GpI1i%(}cd+A(_kDocyeBa_xxAGMobmp^$^xQ*SqG%~Z$-gePd zt0D7KdJXTC&@FL7WmeqK^rVO}OJWjMrc9V=XXdaXKVxlzw3_rDHqCbNp|hoaArOJT zbh6S)pZn73+QpV!^&vi>;d2z-`h;)GMEs(_9-R{4i$NVBW|u%UzR$|DG6oSSMJ=W5 z3}cQQ-r$OxLjH3{|1+6zN4!%#93CYPa!3E2{sQ?>45oY|Hn4B%5buC#?IYQ-+Ca!0 zXhIoDPHVGpBd*g=&kC`#pMR;;ZTNyw77jLTFnpOsn*hP&c z{XYGht~RRi$?XIbWJ9}XdE>lD|Awy!f1}+8dCkEe?%-IlZEB|M6_9V<%PY>9&+fEkk&$HJB%`Q4Q<;8rRecQQ? zr0+@DoIOg^YuAePM-Ph)Yd4j@KDp@1oH;|4Li>a3$4yxJn&~DM%Ztfqc}e zkKKmq`xKZ$f}CVT8#&3sZL}UVWrEclsgcXkO@r*8esmZ z%r&wZiS~daSyY05Nao{nj-FP8v1m`r*)W3uvLzYFKv;)AIRDo#x9M}Tuj7pTf|D5@ z{mSPpJbm&s$+&TYWQ*0arw6s(@91qWGN!`#J1`FIFF^QS9zUJoC69xhNrJyeVny6q zrQEDZM0==B493#x=+FAv9c0H2>A%yDUWCewqMX2=l3pIfhCsg*Akg3dCa;diPl=ld z;Mjtir?IhJ`~NZb9Z*qSTf=kiy;B%qC^K}VN)x2_-h1!8iS#DD3n&PR4GVTrBR1?E zqb6z+lc+Ijs%bB#nDR{37-cU1KKITbro6mw{pIl@vL}6jNqXkR>8-<0t zDB}cx{tRe?SH<7di}&5s^Ht+n0-69}&oF>*@c@W5-R94}u`~FZb-B!Q&Th5g_L7?UC+4kPFadNXlZd64h~*UUe|y>{1pka;_5-3sD;B)d(^7LO#zCjId8}SH% z_)g!O{1-232t2$0tc4!hpoh}`{>E`;%%6aRtMEVFIu5+xn}B|Ez&vwzC}voIQG$IB z>L^Aia5BcORAiOx?$D|gC`yOk_wi{R_12qf(fP@Sx6sh*} zRu9Kc;#-#F$OYz0&WC}2JF7A5iV9SN$s8UAeJ%NEeIbClyH@wli>R;X5(P<8u{+xG{83(Yq$bw;Tm>W=C@m3|VA2V|% z-q*|Rk5g2O-cU$?tBW)@e^R2O5KzDSbNN|Af4)U4y(O%0dCVxPe|N3>@};HRGWqcX z0|SjAqpFmx9r<;0f6!B6jzS9z<}jPYwB96PX<4vLPl?qf>K0u!?(-6TB`=tk_iv=P z!^+WPG$;6|nG;tZ!Kf=6FVt3_Z!mC+p6d|Rm}Cu02kyb(M1bD%9IPEbhT~=RH_x<| zvcb!=m|D}eX;7+QCDNoP{YM8_~-4NFIU&T)LM6DDAyhqJot7~zrbXXK~L_#kSSX!lzDQ4{x;RvDzt9tE9JR+KVDclnqF{?e)(hV z(~nmBRirK0US#LmUR<%qdXF#t5PkRhSm6?BbLAV$*C6th+Ni;t+`eRs0S(7wKmVAj z0~eDL`(ktA?G|qY`mRB@z$FmC)^O)cU7~?YU;@Oy1PJhJ_*xOC89s}s;a{(8Rzexh z+iIh9!!5nFXe$`&GGb40>2}jOlPj3jXU-*goO@@t`ARv!JLR_QV?er5Xn@j7LtfR5Pl(UUro?LV(x{Xvb&S?ckl^n0DZ1JrMiOqLW zwNiDTdvuZp z;*30E=6XgpBv^wuT?sw-n&|<{Aw$d~67z9(<2pZ z?D)cc0w91HsE{neaI|6AF9E3G^(m?d4umeBzPr_1)s)t@90CW?88QDSPFJcAwBY zt}dzCmNTb7FI3r5FQr^no$}l`sawA}j;pYfM-`-!Uu~>v+Lf%LxP#*5l-bD5D`Ga{k-3Eb74a9iEM$Xz zqICzF-soz1rBS*c;tdY+FcMAxo{Oe84J(bw5J48ce&ID@jo#Budel1jKZ`p4G=@69 zXAd%DA}Zi2R>5s%ilR7i9Pma!j2{_0O{Vw+mbq!M2*q{}mM)Qud6@N@l2I|apPw1D z4R<0wbFgA;gM{_J^gpZV3sO|pF0Vdu_rfF8uxG+YQ_^#fhxU=0-o9~VihPly{eV}( zT#5nd>VaV4et~(Uw{ND5`%3B5N@k1HDbndzG< zD`|!}Msx?O5;1$?d&9zelkrMvdb7LtdLvS`)CqJ*@UsZ8SFuv}j>~ZhY{_v+Tomc) z6lZH0ZfoOiqG+p~-5H*o+{KIGHstqUKGS3<0C|&P;PLD+gExoRCg6BcgBEp7>&c`xbxr%p zTcqS8H20-_%9dKQ6I$ocp2R1hSvUe}nL11Z$%Zh*xp0zBoE(C&mjZdc4#<7LsC8~xkuoUL*k6APqI zcG0%Gq<+Wu%}I^Z*LBMGO3BTu>KRLb#Q+qV=PAPkxFX0G18{Z_&;9I$d^p|tx5X^IyE*aw@t*fqX}ZuM`h>T>R>#D* z3@h~ND)1RSieiy)Ygo_m%J8~G>+bjIM@w(tt-ji?Hnwl+$6J#DC(EV+xhHr-@Qcs9 z6&0i@OtHzo>le!mL6i^mv-G8W!h0w4|Hi(_rPVjk>{InGuZ)8@IJex6@S#sOCl-#^ zMz3C2sDII-|M*Sb(Blby=;H~BANqR{4mCT1VvxcX98wkC5N`#&d$@Gyr2?z?1}3CR zhOQF~x`0EIag*}Mx98C{-7O-7!4>)6bV!b;|Dq>gf#)89WiG*P)M%P{0^K=d6w=g$ zD^}o;ke|UUzbK3MVgk{)PFYIUvolv^DO(+HJ8me{nYS+uHIcNGxPRU?#|vp+Br(ng=-(k4 z>d6wTxW;I*?qnxed)N~?sOtnw>F0no`B91WuQEml6K_V=jMkMZ=@OUVzox%FRjg0X zy?fy<<-3LccCBO+eSakA=bI z1JoNk40bz(>h!|Q% z(?^fcPgX_b7&6{R7n|$PRhOLYXuntq%h!LZm)8hJGm>$xfPIC9e3{4tzvM6e(vq7M znN%zhana-#$4SGghSRMYIaz*PxhgfMCQcbDYR_97R@R}dq*JjvK6j~)R_m!YkyEz) zf>m73lU{C@^t?2Pc)zPWtT#&v_e)OLXL~m!*=KpD9a|=~BtA)=XFUwPN(U-v4+5St z_G-Y|M>b3wv#E?$6AKBz&+ts)I)&gU5i%bkoCsw(E(&RaRF&;*hYW!=8kvDN^k90`$ez1I_MO^D~Ec3d&?OFjKj@m zQ{D6eI6Bj&LaE0L>D2-yl#Ife$$3z|#&afJMpYzduF<>5xrM$)EBr0>D4=pj`lkIrm*jV=m`TT!H}S}{x?WN7aM_g2;Enyii}v5?Bz(^PwS z{+=~OkxikaATsCl0NS2)iWNysVePuXI)Nnt3=`~M_WQr;CZS{9Vtoic(hQpxn?vxC z?)#tgnA?wd^b`E~9m1s9U;i)w=EhwJ9me67VCP_*cH|3=@f(&0akG?~Fdw@fWNCQ{ zDZKE)^$<(T6ZCJ(s*iLS8k%(1R2=CtGBoXxoIt5R|AJCao}dr>{E*&vd~79BK0kq! zRxP7{x%49a5WCAAgH3J;Q!KL%nIbnc((<>74~2Av34#>~%&7_e@Y)ZM)dyC?4(Oi| zu-0%l7q)Y<>u_4@NMa)y-$#oGo7W$usO(r3o4l$(bW(HVv%dZd84{sO!R7^YH$fcI z8zgkHmbQn))p&<=q%3}|w)T9ZPhCv$s%12NB zhU_*|f(Mhd(jNM4loLV0-4QUzWEkWos0O_F|6p^+=;m86H|el1jAKk*4;Nj`6P@Ar z)Msq3QBn2i&I;P*bm%szsxMvDA#h$BGw^bnj`r?{63`|UE=pQgP`EC|+EkPHd=DLC z?$Go3@QTm6q4Q19>Wk1SBY8l#-LRH#hQ0qU+a^r(>5uf0&VSbfxX@?M6W#yZjY{ao zHJD)&`J1sA4Xi68lbg!coBY%f&2VO6+-o!mmz)!eolT5e54_jab-s+4P^ho1yU=f- zS5-`O zogXZW`7n@IxZgaR5jQ7B!RzW~%4otcJEdhnFnke@#0p0_UBF#e>Uyb+cu`PWzu@8$ z10|=>3>5??*b_WeE~;=b+X`mD$~%w(Npr4 zl6Lv(WL;j|nkLhozfoTbO?}L_dK3pR`&|}x-4sB;D{8`&yzW2`L%F=xpK%}ST3 zz^>#>*zO{m6$5K^(948$a2DCFEq$AM;ckJ3#^zrnxyL#pOPsx1v*HI+M$l=cJ(BU5N{5<5;|GUW)v zW!^>05Gt5@ESk%FfoASbpuw0u{&nfgDOyVY9rMtx>(cdNz4U7D!PC)NTHC&d}IDX~%Q*}D9LyE*Y-*~=-~iUI9Kbk2369C}dY!}j3YYTN^8IVELv5%i>) z`&HK`t3pRkHPI@>d$KOP-~PugYyc#@16>efq81=A-9vM~{~ z>_VYU{-(RDmtHKi&fWOYin6WMA%+V(nXhh1$Af(hNQ49*F6#PW{{s3U{rK+Uv4$5v zyPJ08yzC3c-3Q}#XGpaoK8nFx2N+U(a`)+y8+FsBPw&yuXXiH@s`*gt3iOr7H890~ z3}w*w*O+G`CXff5Myb6}EstxAaC$$aUt^w4n9!X;c>adTKYJEVKLsNOEk5KoOUBPw zp@K;_k85s#pL^l?b08c%t_7H#Nsb&>ButnUTyDi~(|^DiYGgL*Z~qZ*vQ7Fw#U3c^ zF!r9`tPBpNKR>@?M@MJwSs0sq&oWnKz_WZi@j&FnW5;8~*S+w#Ho^EATydy73~uIz z!1L5Q^hT5ccVF?j`{~C|XvpC=^$W^GAmMP&|AMUHKVXrVZXDf;7@RO9`t#;DeO0_R&vpbgchaA` z=Y*I|&YPRNuH=ix5|n2ioa;e)=4@?M>8KT(sRop9B|459 z93%S8LUP=ZL1!}>vCUgs{8iKZNorlwLZt7J8*HA3N*lf?UXyQ2hNDvm=40bD4!Z&| z4!?jdjxjjx{sqt1!>KT`g0~Fg*bOATsb8i-i3HwKjAH{V`4vHWIg=iu%BF`AzcbIF z_?feBx_nn6PXBc(j5x)lca!jpVhJnqG4U46>M$F-f%1v9l#ujfAU8P>Gz71=v7z3i z3Zcz6)o`CG;JK4|zIF>g*K$Irx2a}Gzy7oQc@y(|gL&SA^r$stJ*2<&6Hed5v8C34 zEX?EHWS$Rjil`xK4m`ij}v_oH>xF;WSh-^2M)7%@bxhF!#m0JwF`zPiJC zi~5So1<0Ear@RZ70mudha&7A03}gxCZBXojB+b1JtoRV*!H}a2tYKbA^NK9R#P6OmNN66;k zd0EYASvBkBb87$EOJo)IW@1;tUnQhfk<7wqb|o>9Vqx<>8=$a~QG7Jyvk zOi{b3+klJjNRVPVUO-VUazA^Hmt=S?Mmqtgs}`Jma?=i$3;102)MaeISS<`&1^ux8 zYN}Xu{??-34I_OO6r*Ci3jKu&MnQ84%XesB{}Klk)tZBq^&0|Jg+avu^pVneN{S9O z@tG$#SBnp77nB&=rhOr)z1WnutTkYl#3RuyDA^&+!=+4;jsnIqW& z3-YZ$4sMAIE%%IE`cmWim#X#W#zn#~UIAO`epphVTtbZh4{Agu(r@&LH|R#n|oZGyBqE)^hn^w^t6e=KobQb9U)HyJMkYYT7&Z9ZSNAUv@r6a|%C4n;UQO8=XY_{OYU&=3l!#xbNKZCWBDyoj z`x|Diqn+x(gWXQPOXkN{xFY0Gp4xsSr~E|iqSt+$BU3}w@x=mkf_#<1DfO6cA@isT zz!}@a|BN;8$QD0`Io!i}3D83&mr_6PrfkU>-8Hy}5n||JM$&W-BNY{r9bu&ll86`8 zgO)CI_8FKTSLtG3k(z)M-(`RgYsG( zgn0s6+SRp8&56(TI>+-4m+q_6))aVq9L{ki@^_;7D5`xseULu2YBBu-ee?Lo=n}X; zyaU{V6L1Sq4T-C9CBV$2$z0IIx31qYPg9%g?tH}6hp#vL%@^qVN5JP?(l?AEP}25x z`Xs%1*9WxEhtQog=#CC_2UJaBWDL3yCw>}oMk{nar|Hji=x_MJ8-IU*-#aSLh<%tq z^$(5*vh2i=4t-Te`m7M~-i8g6>av1p>L)mpkFzb1@VXjPFx#oI)1Z} zr@rwqe#5*x!v1H4*cN03NzO2gVFR@AKjM@|{}(wWh|MP+J?qea$tivP zzs4zH;!QX5=X1AlBCxM5GMUXv;-I_6))lx$1k1Rlpryy(tmh2eq04$igCY-`V<=Wg zXDu`!Nt1t2h>(4s(r4Zbo^5&%2~qBy8_4{azmA$)-loQ=N)IHMy`^FCPERe(W!IVu zJH$fK_G+PRU_?Qnp2pn4e;+(RfA-0+fq}zTq%aY;bCIFge&vq1g1!W?#H=zSVNt3` zMLVBvZ2JzSee^C${C7v+J^G6cW$&M?wu*N0o!9%Xf?4K}Vg9}Z<}iq54jJaJ z1k$aT^e*NZ0rPj6W&W7woshnldG2MNIq=M5o)mY_;ttmoK_Mtp!o8|j}X-_WD7XHB{x)e-t<<~NoZ(dPB*>G_P!eEWtC z93IC8tca(muVLNemm=7JPRkj~yO`gA8-lPFiv&scIn@ucaMQvcE_}rBA$r@d_Pc6o z^j!S9z3ug79-5kqFWoQ5*gKT3s+Dx4x%vI|nwklRi5-RC7t^nf4efopsNnk|Vr=PN zN#21k`})4vpO?Gui~jyE59I9(@7|l=d#9`OcJIPF3rUrZ8y!#kZ?s+O=(yHKd~&1X z1`w@7Q$`@tXu~-Idgh-Lg^@?>HeUAYylZFRjxj-WnT&G!9`!!EbpQrLRB9k*CckLaFbG@2Atg*L%8F zkN93*4*a^N%L(-LAG76PnL(X64K8p5ZW-G69jTLR$YIPxG1}Z|%?zQJHQ=YBQn@KP zb4h@TsA3aR&)SfgzBD0wBr|PU0$|`V6kW#@#kRKTt%?~-(4~Peh*i-Liz-U%^%1F5 zjizTT@>3Dj?g)tMQ5NOJ#CD5B*+@NgbwgWb~lst3M6^yn?=s7Uw7@`>epi!KZdoS!!;eK1NUGTsbXP){P%ql|rI%|MGQ zh7WKxBElHzAq?m+T{J(P-mHhtzH`e|bNMH9>AB_S_Kp(jqaR=IyD*=`;b;$#Ejdg! zFoVF?#>NuC`O+832I&M$#ZF=^84bc2k2|c6MKg5}9c-EBHOzA}ag?GVe2K?hD^C}& z>Fe+_QU_&oO25gobYK=WJngC<^o@w4C!chCs>`EGAEAT_Ed_Dd|KI zgMK$~hoC!odNJ8;9;f z61kMA;xtp_G}|7F-4nRi0+5MuEaGVb1;-Q%zS*;hO0-V#&Xs{YNHwg_ckI{OW!+l#gh?xXccwwj+cOdXtP~YAGPF!1x)NgPua1E6%f;EN4;hSET_JRh(6% zF6EC!STpR4hR_jsZN;ZIXiMb13cZisr!8qq$fpGK|16si984jf5+q%@fr4n`>1@z6 zR0lCo{(!32m^pY&b|nZDr{Wm+RR(?)!}0nArOpH>brcxAJvhejEf9aNCOK&^`II7f z#E3B==t3V|p${nO3+V&E!E-i3kppZI-~q6%4E|H=CMjc*fFjbWS!4krpl4q~DMEcE zlyZfh4aBw!v^AfCc8CNr$q-3glnc#q`PAIW1U>Tb60h#C>tu> zIz=Cc3QmDOYA5KUuvf&i``?rvIiNQQ?%V@<#PzW;Iu=e15|q)l6u4M0S+Udr+`rt@ zlkD9o9o=dAtQbV_+$6TkiV~nTD*}zp9JUo43x?Swf;+lzlvdOyU(qijU!C+s%g4t7 zlUG6GwE{F=vJ;(JF>5kSkJt)n1iTlpmO2C4FQ&IB%Quiu&G9pF z3Fq_DVlBK4c!E?H*KnaAk=U|-zjQp+!^F)f*4HhBcw`=6=@gda?S$*cGr1r1Zmf3+ zIC%mah(9n^W+En!NwDeXCY?qUiw4gp;QqDH4G}cOQ53D!@d+^mR)D#J6NTF z1!I%tlYg7CUM+cXM)1-I7dMlD6y=>NNqw0?ZShtiHvWmCy=qB|GlK`C96U^!w8X_( z{w=9iA=ds$B4T;PhD5IjXXBjY@(qbT@m3}o>2!D*z#<(?vQx@7CV8Y-8KtMfq^t$C z;Y*-41Wx7ed$?9yx*|HVHq_SF+%HXer&`I1n5gPt+dvD?bQN^EeSNTBsI7TsTI-si zpdeergnZbUmqE4oj!fC8$EfK}0OVCx*$Ay+#oank8$OaL8?_i^BZgd*DH~bH+n{Vb zFH<(MrQU`-=Vi)97IFuajVEQwMiz1pIEH;Bt88Q;7mzk}hk{o)fwEDbvlm*w2g=61 zGG!y1X9Sdumu1REhHu~;1ZCsLGG!y1=O8f`)UGc;J@3iR`!A6f6-0c)Xyq^~M3PB! zrSo;k;4tKcRJPN=?o16#Ws*3u7?wV8TyA89-ejg8>^*{W5%rV!U3OedlMp^MnJt8N z&0cV9K?!|OhZO3dezb?!4X=rn7&xSmlpQ zUFn;M%K6Ef7(zY=x`v0l48lAP!QH*{2iUQ>!9J$<(&yP&O}bes@X&8Lrf`b?p-8b1$S^Sn9mPy?N% z!!#Za-1NLQ__GGQx02#4N15CJ(1l?$qZJNKW90n55WTS8PE*s&!buBQT3IM9V~55Z zyD3TZw_NFWlSn$xecsY~rO#a=>6-YwAp5}bJQYdmiMFf*EAWq#NG+o`i+0LDswUyq zLSbmS*=5mo9Y2Ax``Lfb6L(<^CC zhMm2=DUIK3^WxMuyjx%lGz7nn5a_>56p_1hK&QbSQ?6*lqXR=KegPf6$``KK#lj_f z+bNfcz296qnn@fX8jxCmPVn+*dO%7 zx}{HFsq4A7X6Yn+eL>uBxzdoap&&9ptR*76IV>-#U{glp6=_Ss;co{Hqkzfbf(6b8QtFiWw{NUSfp78vdSc`)2s$-4Ji zN;V`BM~LF&h(%e7%1#v#DeH8P7z~{+X*rrAz4DKPX};2cP{ziB$O07F5+2cl$Fng5 zp~fqdga4BeVa1jQ-WFIK0wv9i#UuXbS!wG=k!$-U;xp^t)4Z6Htso3e87#N@ht8vH zlE0>*et1#1r)Q|Nn_m9U{pE5#At$&H*aWHxAtMcHLZ3nW2A4_36%;lf{>Cf-xo)_y z@LnPPB7HqGKONP;zsc&vyu{NYv|EXsAhhXLv_yKJ&}KjQ{o&|Jax(0?9povl5NISe ziaO%3Kzuy2H=tl^C>(u^nqM9b2s=+p3qoRfJnZjsd6~-Td|NN~06uTF zgJ(dvP^mjEEmM@vxAbuK;!-9So^fGH$_uZj=cR97!}RvT){Ojg@>FV~wS|eP(QJWG zG1@;q459dBtJ#L8`i6X=a#B!KEX8*F*${vTW7IHKGa!rj+6f(IIw5e)2~6EAPPD*& zi%FgnpIDcPBGvAD-4{vI;`8D$QbhGCpQ)6j7es%}peLDLl`Kcf$H~cG1H!zMp_y&u z1aA(^fC8{DI6Mu4gT>&QX37V*H$EP?DWD(luLZmqRz7XZL~bhYX-}w~HF+Dkq%F+L z|HfE^>{Xi4tL3yOQlRfwA`Ltr$y*ZN#lu3o>8#G^=uY{}ux)bnXUw|ApJpB7n_aUE z#RHc>xYZE+LD06fg+=$^U2F7>%)E3|n_iHf4IekrCXY--RmEWUZ7SOeYgJVg;(78PW>J7nbL}8rbJ{LQf&A2OFlW@=3kd?I|Cuc)u=7t>SBuB?2XQvb=rxemB zYeROZ#1e80OtGf08{|46yaQNm5RY}C#~uQv zSig%Kb#s#&l2A`dw~w-F%UDWV6(1=^rFv%isc*7bv?1E1p}<}xdrYigveGGXL5M(I zl$LK0U#}=kBbIp8CFT{+RZte&P{GkgP6igyIdgRU>N2Ib>m8H&T~h4J71b3fp@*-h z5A+X0@5m?6ziAw>ytGVbC8x`*E7D}=zAz2VB{w=d_l0X}3|^m~)DoqvY>=0fh`$YU ziGiw1J&P_^R$gAzbGdS5+@iR+#qs!WQ5>|?OWa3y_*!rhU{6n2ax?9Sy}jFeLo99h z0zWe&J1*CkxKC@ANOZ-Nw8-$rdUB&wAi%<(06?FpM_l z9wWlIhkij19T|NsR#S8RkK3j99-VzeSd!FO`<=dlJ8jbUv7#7iiJ)&$&cLPka}wG0 zArF!Thd>Jp#x^~}|4|+=|ldhV6WeWqViMx)|PGVOn)gx3Yq{ zpW&2eoQY5vSszAuIw-5wAFrtzj8xjL9#x#*q9ia2Fr>X>jD^CO`XFkKy^27k^<0~N zZjvHj;%G%nXQf9g^30B*eKzI*VU0FK+VoSrAQeJ?i4}2BRXD&9^Ej zNsVxca@H8LD5>-Gs|n}vXUC>H;`bN~sp}*LT7~-Pydc(33a||G*N3H0#QB16LMelG z+#g@1gW&9k<3I}!23xKHpKLHDAcHYlvNJ`>127>Uq_8fo+-Vm|9v`YOfY20ISHdi2 zVX2k&Y&DUsN$~mB>O*-7{K)2!~>Xh>RB?L$}lm-+eH8~W+%d6+o(_?GzjmG~Ms7?_5Nbo`7A zoD6--VHGR{M}?!X3Je*mEyP8$W8t${U~a)?Z9lL(wV22XmiYr>VM`E~l1FR>QOh!7 zhr)ybs~Rr6<2^@cwD;l2<||Qpl-A8|U3vr~eQ=FSfk9+I)bRO|L&D#qSlpk$P*72OMuund_1xhBe z@@0awh=?}y(z6i-se~tqK9`j4sGX%(Xa$zA9LryKBnMQ#!4Hk1`&(l=!ZyoKB#(zuTS0^4o8WCNYB4(;S&p7<;sN0qh+vws(~0`$5K8+T)|2VHw(bzV2ds?ttJ!)ZRqNo{G}mi z{rigjNKg83DRH0pc0_tmI=*%-kxV3wNHwL}P|-oMjoici1*ohMOOIxl8|6n7dv*bp zK}VC?UDd>8!?F2UN__En@qTp${RXw93T075QhJF<6i&8{%~A>fz=GFrG5?V|sxvz(o&N9>_etD05*5wS&}mwTUZohz_-la#32ZnKM8(uUVKi zR_2oFZen1k7hCHb7&6PzV75oLD~#wYjObMukp%lC&tNFDSPn~ez%zG#FfF2qh=$*lI)PrXK0n;g<6C!opSdrHaav`Lt?4X)v zL)Q8nShB}MVA+!%SM8)2IPRkm78VNc#H5E|aqf*w@+^jcLR9S<7zjgz>d<}Oy}xFP z#*4>G3@z>Xe19`5TRz{9>?JCs>xlD?CZ=xi2<^Ig9<2EV{Gd-DJMYls9nywP+s;Zdx zN2DeRB>JgkZWRrx!Q%l6evSdT0-vz}fro2guD}=LXcsp#Dz8v6C~kIRK0Ochy4d+S z(#y~y2gg818V7LU@eRVlf&h8kFL~q7nni2850s2*@XUKN6RI2(1IB$7LW3hqsc-2G zM8ya%G{L7dh;Bv=uHNA=2@^nB7DIJxjHQWuSenS%-d56|Mr&jB-Ppbha@a;@DGD}`-yGj2Bw=pATf?+kVaVk5KJq6ZprsA->1pf ztIrLs54B3}NiWJ(5_)_zO&_7MxK_J(=h`fv#dTrv*7QDTy>z&rj= zn+!eI9H~3*AvCN?V`*4PzEGgCkB*xrU$8P0oB~-AmS)pQt35}(U^2Gqd9d72){ zG!4@Vc9v1Vj4`8a02=>BT1%)hLfeEv0Vz=kSv$5T&nL;$RT=GrDu_TMiXQHEAz-Sq zcI-*dCy4ybWV2nQLaNE74cbREZA^%(6?1Ad#AhzSvkpJWHS1`l$jGo0S;-iFfaDJJ_HbjXYTyBHg|N`DVg>I5 z#iN>(#Q@pSRfJa3!)^fqZm3$iZX_ZZ>bec9|1#7C%nla6fma3^22~=|bLeo2suI^^ zmW#>73AzuXoIdZ}5f;l9ptEO{rDT=V53Un=24s)-v0ko7fa20On!~eih2ozH=f--$XUg z&TUS~8G2m%v*QW+CTbxD0oH9mYnN_BZDrOac7e&(Wpw`t-O@d`$YE~j+-_*2nb=6a zCTqhD2864B4nII#P0X9axkjG$CIaPS$KZWpq^Y5+>8oex_*wBa(I$b!kv7C5lyo&s zh-$!GcW_J##$1wZ^jsxqVHb=d6n{@kUC&d{4R2~Tx8AEA0pvNsR) zaQn;YrQCj4OQ*>e@;lxcm@llAnznZR!Kp;~_%373mWk$LJB>73&gAEHYHG&i<#oY- zvgPiT#pen?dS%f>;fkTY{uL|x7B2^yJhNVGOZ%Ie!Bnb6{C(}gKIp@MzieX8EFxP6 zM*S;H@G&?AFvAbj)|Bt&MjY!uSS|QT_Z~hCh>blPIw0RKknaV^hu0+=3ucPd+a#(j zQ_zafvDQLy>rK}zfn4pBskLNTy!YfCD0Uc-dI->ib!oB?TkL1Up35LVsu0Hf!xGfv z8ez#*ND2unQWz88ST(X;jb{+llzjl9lUihNJ5o0HbTM%C^K-2m>?<`6iq7_~;V*;g zCjk2sP`wRyzU@Btb3NSVhL(Hx6$ED~@{|i>LMnZr zALpPScK|ywRn;?{D)hsQ4XShl=^0c41W~j;M&T$~5D{OgAjmID6(a$n6SN4RFiUvV zTceeg>JO1E9%W%xj{a_-KRU)8YNIDDL>_ff(KX(%;*JAKPeNON2+HMx8?zI69Nxg# zViZLmMTCbIeQlN~e`h}>=qHR`rpEvm-$D69-bjt+!hqj`0iT9W$XMwa1BTgU0(#}=A&ePKeQhExwh(_8xKmfpN2R%+ZdayT-PGjj@Rf@hN^8+Per7 zZN&Ja1Gg#HgvsqKveOq@oA&0yCkmku(v@u?)@*;VIcIS&agd4#2rdm0@h#RTC%1Sg ziK_M`$2JLtS#d5|9we1awpeCwTCnI+Wk6N%9Q%NlB#)55`pBg2@Wn|!K8cnAmR?EF znPYIYb;97~vJs;-$C4>?q_JIamFo~e5c?);b(Bak!XG&tpoeT#gEH-VR!cei?mYxFfHcZK$u`CIzLoB3J!#4||mCC8y%*sI`p zz!0dn2o*%@X%U;Xqz_>4ZQ(9kE-e7^aae{gL(uetog1IPfUz0MDrPq@N~&pm0lI)? znU)_3vTSRtP!L?=jl$v;g{7mpxkKSX#kwfunhYX)-O-eUMu9?pTwIfaAbUk{8Mt$Y zgp>#biDZjUiNAfarvhJ89Ozr%;Zf)lSS;cTJV7^8;?ox5>>M2I>>Ps1{q5|0ee7)m z0Hv>iyz~<9f_#3Vw_tl>WIh?AgLNy+`$lE1vzUqiYQM|Ivr5MJ29*Sh_?GLFl3MYs z9$66HmSm)+*?!L2DchSMlVjoL?YMVfwB8sM1w6 zD8r$5l^D;kW2!q}m>LZe4KvI;>)VgC${rG z$9e*M9|8EcLnCmEo0}s-C8ErCTW*I$iElA`tG^9(JME?8AwBf>ZQ;XXB~91|w2JtW zP~zRi@wy=Mfm^$U6U;Dlv?{w^qG@8c+suNes@arFeA)MPM;{V51s;7lr}t466g@{g zAO?6}FiQ*DJiw50ZZ9&E^mg@X5D&;lT~ctyMVvm!mMF{S1_lu@toLc)JjiSh@t})% z(lsdr^!-fUCFk=LU>)LP1;`T6Kt4K^vm%>5s6}LJk#`AO>8+b9FuVT?b%315^j^9^N5)^J~x)r0H6ABi>*q$ORgNd#+i?T3cw$3qYjHT4z!2JsoZ ziW$Z`GxaFRAYanAh|5ep%FiIkm~etpA{iuD`alLzd6tKGn<=IG3<9y< zxX)_OAXE$r28jAI2yuyNT`YtA0^>T*KqNBAMd<^m8Dm332KgBvpE3~5XZ4^TP$=Zl zl0iO&dRmyav}KTc^s8tEdK2>K$RJPXYs5#$2OzpEgrfvwpJX69_#5+w}>wph>a}I{pnJ3WqCNT4u2;^Fm_uRL<9Qs8KD3W zI~l}+_D1g$jsUTj)iXxBpb)|uAPzFfue2{j(>nviQ3kmHkYxniIF_~&1_9F19<+)-A2hkurk;g%;iT!8b^rOYpZ~LAq$V?|HQ@;^! zlP;|8B`=>tJZ0h0ATC_!d$#OVG0t#t~o zYn{dV6XHPsP&?#{f_zuWxtJq%5`zX06tEzH153QSR-m`YSwqn#H(tA^pV)zd{Cdm- zEGBP5A#kxIwgOt9_fCJu4|CN~grkoUCGdB4Pf}Q?KGxc`O3c;wDobi3P`*YL(zOos zG;sCub*$*@E!6Xlsf?;oz|>h4aJ~;Jw`NfD-*Lkrotnq?gei=YF>apeLdBS{0QErc zY(7{w+%g2~m2w7S+ZL-S7Du4$c#--A1;@MqBRfB5_kaa~&UW6eHbF)1eT6}pN&@Bl z*wRJN2{q`%>wrFBg<*c4>x3V{@GKTACyeGK^o9y{ph#E0Oa*~&Y?As%h#u7-jMeEE z1=&jil$G+=C^+N>nav6Ga89UrnZK}%{>D_rt1&XH!V3-ukkg3O&=A81yG>tC!!ogo z5;R_G7)pPS&`k~ck)BH7swG`*g!4Iiov^Fx1s5WTVT0v$Fl;cc1EhcO2*sQ<)X=ns zUW3qkxV*akKoTx5{Sb=paz!Q#7i>B8h&T*g0B!sq#`*^?7!D<5nb7R6xRLMa;#)3M zjPTVD(d}J;-qxV^z#B;OJxiLzlJi2ZVn17tAV=GGb8i(!u(U8<62OEfFC6~@Z*jUu$I&yzl=FybNtev7eW|s8H`r~5rXW9 zVZmd@mr=9+k@rUM0wZ6M(8mQSuBKmi*J$-Ct(S-jhP=G@NH2sD|E8ibQ3?LzG1cXUhf z<%xVi&=F>1Sx#T31f6qXTv`}oCGbIDdYsuwpg=PjUSyxP6dYMP<9x!E)yA9R+XIA( zWdSH6L!_9yIwhb;AP5ccnXe!S?Dk9tg|Dkynu3Clf_I5uM4hrgR2-7g=9%vmQY7Ll zH%59F`_x2wx`ao#J4d7HNCyXhe@BN1K%wr`ufzceaga^qG@~mEAp09O*C4jZpV~@z z*u~Xo@r?(==2fVRibk^Ri(-t#8fBX-Eu)<%uCKrUJOx33Qb0|xo$KZy-{|Evsf)t6 zqzk%1{e5Sv@WXI2hSw^ za5ZxEkhBJrEf9+eR(Se=x<0_$y--Ohh;;R7O7(5@rLUj`x>h~)NgW~0$$^21aY4b! z&|V6sn^;BcWm!4wlgzM$-0bdb`lf`ikaQE4()ZtM#fs^86i+NgcksD_m2t#U4e4DC zbZ7Bms3w|ogV;pi8$@A6$nQ^W)*wV0(%0~RbZg;4da}D4s!5wViXMbamL^rwwMLwhQl9)e}_HWbcg3-h7yGgf+TXr`l3GHcGT zu{mEw+>{qryKm6H5g9KDJa;v{@nH)zY%}!;slcnRfw;_!ELdmjCEy2J+Pm|4fx=*i zIqo60R$kE5UHV^0je}jjIV8r*>fKiX@(K-UK+%zSY^)1@wua=wfZp zKvpcZBjeSO3S~VD+aki8++CE!i++uqoms#oNIgeS5U~=%>Q+KsqA7Ev2sapjbO7E` zAn3Vifp!wAFcm3lnmGvg$_ot6nW&V-d>G?rH&--qJF^5Fau>pJ6~Mg<$9@^7mKb{u zQZ9qMLW~jN*uKfDkU`D?#GQK^AeAymBQZuQGxb!-AaKtDsl!03Wsn|#crf+U$RO=d z3Net{KS37AARFNFKSL%@oeVMn_2@G-*UKPn#1`TMCQrk3o-q=e&H21W8KfCX88M}r zWRRuACQ^_4Ipk@UL3TrbDDFo9X^}x*A5xH!h%|JKfplUB%-05@8?{3~6)YpH{bNZioFiz)IwHECEhJ7f zT84Vbc)XB6YfnbQLIMaZB$%`gv>6RRWqe*Yr;=C+kT!G#HIrDb=mE%5fTRIU{|!)L z!A>KjloG*A5DTR6%4eiluqlwEioDL`SUB|?KLU_v!7dZ{E+xTx7;;ooqih{hzwz(| zGV%|W9nd*2vY|bftWhB$@}qK51-XhO9n7Gx}m4*nL3z7h3g>hF?H0E zpUL|1h_~u5`%y>!i^rH+Cg0501hM-I~pvv8TQfO9sC zrH(oz>&I`r*1zmWy$l~403Y^D4rXC9V|mKdQBN&q>wwYXW&LNIHd0>Gt#tlnjwV?v z-Ovi&EEc>!VyuaRmm?vL8*deh6T0ag>||yP0rL(L<)EO=Op>AHVW&Y}%iD z+AQlw+SG4?BY%+tO5hxTxiapj@XbYTQ}=lP0qQh&f#F171aN!pn|$ znIa2|8JG|K+f;m+G2xcqI7eAol({;$pkaWOTfuvZ75)L^6EMC2Rj7!&91( zen`LMr3tYKOOukAujEB#L9xN5rNOa5Ws}B#n^epQO`zSFx~Fh*#>f&ld7OFwm!tcC zFm_TAkoKRj_xqgx3*;>N2ec_*v?&e#0c{I73ptDaS5f!>17baibY1e%WEYeEkRSOM z=@N)t!|CHwU-PTD-$8la)MH#f9B!CM2q?~Ett{Q|FVeMT>7!hNFy(1-c)Zu49Dxj} zA1MJmGmItpn1ON=jIPJlj9sl-`@!qfkI0<<5Sc@SsTcJl9NoC!8M491_7Z_cM4_glHPm=Z|vZ8OQp}-`rRSPtc<~ICuyI36o1^JfQj+ z4_u8<2ZxXl2Zzu7`-(>w@gBPB~1EJYB%vNZwowEF!c+ey%kLPGQcMuwl?)8?>A=HU>Bar z!OR>gmb9rcz&IZ zMg`zB%o8qvE}*FkG+3jjmdY9Hz*T3@1EiTA1m;RU{jlgTuQWRG`j1=!N$Br6XnKYr z4V&Q6o2y>WWETLe6PSMT5{#2wCy;CmAJVqvYA|g!{(7@~q0n$Y%ruU1MOd}HQCs@1 zHupAGdn5UL0F#8Br>2dfjAtx%8X$RQL7w(vePdpicwU@^uZz8ouAx%9xP)u0W31z( zpWUEqtZOl+paw?%1)atDVJZSHv@W6V{T|9eli|`k(q&WC5K96cq2Snvg$Fn3GLg_N z#l6HfTuC$>=BuiR0#ee7mDN>2V0gNk zs)2WR6hGaBUWOPrt<+5=1*Mx(Q#X|slx#{JuE;D%FRMr`$SH;Oe1GbDg+Oq^H3zp8 z*yo%Gas{&v??ca9AWT3GA0)wl9P$G-;J(DYMeslr!;9lR@WK3*|Aa6HAri_9|BY5N z`If;wAUC;R5#|UONE|IqBoK9w2VdK}n2H6ThS}hDilroc0W0G*3K0krxS@fuPk}cb z9|CjC&B-@JIadpZ-!0IUv>$oBVDLmbHA%bM_eJMj9*Eo=U`H6(+jPHL>gv4_aW*x) zu_MnAb@DpQ%w7^Kt$GnR%e#;(sog0u^ zTEy>QfOuT+Xl`X~g|)GDYKj#_=s%^c6638jO%hV$jgAmQ#8AU6^Q5H6Sw;#Xg)s#c zfzhnUr2oU(cYsxOH2vQ_=iCBFFBiCgAWg7Kvw?t$poj=aQ;HQ(R76Cvg2t|>*id6b z#jc6H5H)Yq#NLvqDJE~KNz^1JHn{ij|LvZ80ri@E-}C?SMBvPsGc&t8J3BkOJDccp zOX6U1=YPvV>Fo02r~fgB`dYgRxvLDeU5+vk2v0as_a!gO)BSBT*!9s&vOvLAaLcFT z3RBS!z3EQkuphqsvrLRmlUunh1@a0r;R@}BoZmy|o^P!NdkqGGS;YwrQ>S~h7 z_WOd4$>0AYIx0`i3-yP}8OG=k0xeNIKpnw$dHAB;L(q%Zy*pTrQ@7=OeX)P{4}A-? z@J~+iuM=Ga&VN9BOLd#^1+Ctj$GIEoUuB8d>#I*r`?Y~4+bFlnBAI_JYVjM9x2bq!SyGAPtr z$SAo{+-$K2$UOT|qw+rc;)F)IWWk$ckmO57-Me*m;xZTNwmSNI+t?7LnWag`bNsq$Q!5iQb%#JVU8`d5bGPs) zPyhClN~Z^lpQ8h;1dG)Y3vW*(-J~0fmBg1Wx#BO{%!01Eqlt5KcT2&wTX#?Gp`m=7 zXG}lemJz%5MT?(n_;u%qfAI9uN$veTqr%;E{>9v9Zh;-tW+qmqH06sAG8ciT8BTr;h2Ym|N3v2lwhwaU zJ}cJwyGHi(l1f4RQHl*z=@2D>ZHAN=%P*bEvdc_pc;^;d8Y3v;TV4mbt zUeR6ME4p1f_w8n-X}{4GHR(~W8#MD?{}*zq&i;d(&%^)!m%Yh#%!WT947~NQAKVLC_jDvaaf8<~qW8m=g zSW^zBH$Qx4`Ju_&f961BS{OzJr^AEI2Pju4|PS3-|O3#CXnZ*Ej zJ7D&#Ko?3|u&aAWc^P=HLm3KJ@Ze?cbS(}eO%w9i?zOchl%rmQ|N29U2&9 zOYGfyM0QbA{?*moC0?a{(S(Oy1Xd_21Ijg9_fpYY*_`&LCJg9BkW$4#8=6$8p{#0d ztM1!*2i$u`bXBwZd?3|k+OVN*o(`dPJr8xADAn}?rP>8GExpsdq?OPhfgieIV?CpS zUTFq&!^R52AVqt>>b62AYx{cdhc8KwK+#jFY(d5IpF9)(o3xGj7M9&+rNwGu7hcLz zJM8>>q1NN=T`6fGVyEOxc(7=6_jMJ&;P2)YKGFhg6oKxF(;7 z3DnP)8Pc=KPwe@D{yD{Ni+l>+|6+JnJYvtk8lFvg_WT=t#u7k=;yL^Nq2bxol084t zKM%6IBuA=a#bfsU-?CKTPsQ`^hG*c@=ReppVQq&v#aRlrm1Sn7)Bus&KB&Ra`mAS0Qx>k@dtoYv~=je@_x9 zUyFNsw1XjBBQKPKc$n~vU9vccT;W3~Q~AQWGXmYm28UOfz*qwl%)LE`O4!S7tQF+E z7)Yh47U&%mje-DAZo=UvP9}^v;Ip~}Ceh35a-u5*YNK30RhtM*vRX@FkJgvTrz_`5 zxiXq#9K6@C0?Xt)Z$bLX|aIwOxkWap54M|7QZa|pnpi` z`1sBt{mDT>ug2Mr>=xf5y9M@m$n^;?Y4Qr1@DjO^1U_MlSbMOq#(C=#;Rw z|0RB*F1bTATQ(4z8xu?Jh(B)GAU?h^5rr^h%g|pd@Nd~L_~?I0v7r96J={= z;xV7~;D^4D-@^awWbaVamGlbgMqhd(y$D07{Zp8Gwr&c=s}|-$^x;5NgNg;#Y-JuT zLg;ZSSQw;@S9p5Uo9F2XXAdm+W7j{4ec-dxc1eLCD;#FK({}NAt(>pz7+|tfZtkL0 z1hhBVC0IYe%gK`qwX*S(s(nLvZF6(~V68%94b?ZWqW1;2pqhsK2LHk^UEayD_7l^8 zXDc{XXs7(&^0s2Ulm4AI7h54ly_GkSmDW3|S}0DU24p6@szqIe_KGltd!TL4-{9%l z&&K!q+@B_vigeLNTuol^Ek!GFGkL6Y6IDo-_cz}b@JjMb-b_p;yUBiSTQPy`C%eUD z={KHJ$xgB#4cnBe%tYD1H&VAtJDbZg@YnX>ZtLW_Ht5@BkkgyVDam1rm>xTAXTr<9 zedW?mZxfC_=zo!4#WmzP$r1gC*G=(DZWe@lN`{E-h}#wMD?UKnK>lQp^sV?C($vwn z?TIVik}i_ub@)%)o=PnyJbhYMH>D9gg_co*Vn%tcK1n4gW*~^u9I-~4<)a<4(}c*n zN*Q*r3W9-B8nan4gN8%d4h}R5M$>z-X&*JccxGl8o)I=16QOPT+;9uag6)IT=R{ao z6mIVskRE1Y>fN_P$MjGWRf|4p(dB)J^8hj?Nj%_4PCAHbVr?-|XfFY-n>&DvP1KzN z+;N4d1(9Xlj+;L1sj*z<$oq|0#K(DHKKV$ucq zQ^gO-Nipi)NIZ}(<>iFF7B9fmH6_+JzP8(r%k zD1Sk;+m@bDmJ{mi1_xt~#eP!nn&9qtA2`9-!@_ayk>Q9aEn%PUg1cAQexH z)2$?@#mIRF$q90pv(Y_9VQT8{E9XG7y2_$uNw{Zwj10kXxwL-_9)Z+D)O1H4BGh3V z`wJluYyg++1fmu+=ZytNg9{;^QUVLx089Xs(%&9zm%jcHS@Y-{PrB$zOYmvxtS}3U zF&opW+f}E`4zsizw`pMNtnL;r#8qQ4c2jCvS*V4@xQ!|F^{g;Ui?JJ1Q)h)*I*Z?H z0FG^&7OrxsY=@twJ-;}rG3yc4IxQSmbo0F{+oVRQ%z8$)O$P==GYju^*u}i37(-mz zi9fU;1DBq9YRMF`tnC?%pruw1zGRQzop(S z;ztNkk-~A@N%D7zn9n;Fa>=^gh2%Tv%muPd(TuqwP803 zi~Pz{xZ@gKbe*W-BFKNpR~p?JFFX(r4ztm%dBAzgY3U<&QXE)J*)FIGZ$xWr|O;1s{Vbr6V*_I<~ z=W=#)b)^G2;j(63?VjFFPMd!h@2{tSI#=6wnsaW_FyP-9&&Re? z5ykumrs7g^n9!7@l=WQ$YdQ}{F=?JTWebE43GAwZ@M*Uo%rNR5r2x$6Z^t~ zVcF1xF7Qg3Ik2~d*)Qrw({2#L!q~ zQd}?>TUR5V9?v}=udSp%K@ZwsLVa@q9#=D((h&YR0%V{^)ta)&d?p{ujnZz`{+YoS z5B?1#xv6yng$3Luv3VUf)zEYqB#4sJnXn)AGA@$q35u!EG@6QoPBUo?uEF<5Jcgw= zO?{^L;n~Zt<&|e^bNM_*S^IPlKM|XGn6z4n;CLo0hrB@FC}v^d3=5+yT+N$s3GL;d zx38VXkL3!!!9L;poCz<}Vf9J&h3t;e`b@;1xKxR+bjR;}6WD%WW4@T0`O11QU?DlF!DVr-m+0_qu>a8zvSA#%&b z3L<_@KM_xy+eS~$KrM0c`&O+&*ibhNJjdbNNsZsq4CwffZrzlZK-XTpIJfPbnFa*7 zHB__ej?nLTPypY)uzr^s<&_f;Dlavk{55d}n&*d)w|}C2so{@m9LbTn!m+v-!HG&N zt3Z~OrNV{=iK+9)zMD=3*3MbBt%B5QgeVO*|Db}km3*$Yk__Fp4c}hDx0f2fg~W1w zafSYnxOS#Sh;}7&mu_8#v*Wa$wrvwP((h#gK|zI5g+*fe0*{nwwwY;jvX4XZRV!Xon9SowDJ@{+M@SG>G%H-;dCrLwX4ABphuRkDy|5T}|J zEyDJQ&&1squZhKCmH6{15BIRWI7*xRZ$j>w@gur=(bdaj3~Bdu$;_WHw^(rfqPSZ8 z^sC9ZNoAF;%It`$Et^BpEGi-JyxDx880Py#Ki?+`hpAKY?MS}e>M5$qDWmMfm3Awq zJRhyP=;SVtV!_e*d1t2hIU#S9q0C`1QBo-+T-HEsBqURZky(X%Lpu-wU4?p$ui_y&WDKsf=4@r1-o#hh0FLtlhEb67p+gWGa&A&N-+)JOkBq`& zxDuoABwR6Mj7}Ec!YF(J4pO#VGzHTo7Z7&J3=Y4n=R!Ddk=m#a|+ssyFc0bH?S| z$t^QV=M&DTK42(1VHE$CWD4Kldzbs#DE=SP(EB<*h9aoro8^_e^6X^B8~AR#vr&9^ zuGT2rgZs=VyakU4j1j(GyqQtBw{d;IP=uJF5jwtaq0{m$_29>!GfKyw``swK75Bg> zyfxp^D7+0118qdkLLu2GyqKS96u$T^?Nz~lNQa^~+Uqi0BHTjp2K)+FZxnw0E%-Oz zf`5zRW3ANS`Q8{Jf8{lhRfu{gY_v-r4y}AtleTEM>z03)4Cp~U;FvEkd}h34 zl#UI9m+IlG8Gc{DdA7&bAdfWvh|o2$>WCP+#z`` z;As-}Xo}ql(3K3DD?wv1r>=hpIh^Xp(=(yAbXzy_YQAM86SZ6r9jJBqAs($S&TM$`UDs3}ON*h=VUfsw-rn0C zH#iDzo&Bx5bVoLxPQks*Ow+~B+=i5ArR2tjhW6_2PFlC+1EK;Q`JWPkgX7}51ktO1 zpYyl)mvWuVM|q#s+I=`Rs2{@d#{>cAGkuibV1k7bAR>wyqWD}C@E!3$cXmQB{RwqU z;UQfZ$F+0|9T;SRP?RFbE1;W+Nm_WX9+p-EOXRx2Zaqc}P;he3(CDsR;zp!pl@F6) zdkHQr>kn3xBwY!4z0#57%FQh-$p}g0$&yI&LW{7NE?xUkl0!k#0wfbb61t6ztqoO1 z1#GBG@^P0mRN%1Q;bA5wi9z}b?Bd5Ca~)cqm73G9bLZIZ9;A&_X;>r!QNf%Mh5ddj zS-SpXXrfp6@ny#Oi6+M3#^o4nOgOY-d+6Fsy{@%qy4F#=&IbbG3tj72=SbU7_9J~$ ze@9Hhe*J~8A!u5eujwcr5{(Aaik|JGE1vxiWQ#uU(8?##y{A&H~2Kf0V_%*~tf{RJ~ zdCFH$r>1^C;|o0%cHo$)52*uvjE6Ez^r$|{QjEUOr3ZvMJw_O^b4Y=R|50SyC4AgsNGBomz0G0WQ#qfbKm7+Y!c^SPjI0E?dDIEh` zHlN9w34H9;1-z9We+;8f;ombl@WUCMd`2e|bl5zA2ESODf+RYCd(k@KdIJvIDEmPF zKE8oZAmc;dqs%H=3d+T+0ZOIKm$*}ELb0A2Eghp$pD2d&Gn8e3=StLxA2|RzPGXc) zquc~Nd<3HdJ1oS}mc=RomG+bd`!o$ zrS-%()%U{j+NX>YwJ3lxdQoB~HxNs*K1c?KIyVcbcdXxH_=dH3lNlAhX~)Wfvh%2O zeDxudu9}?YCPKPDAzkB43w193m*P`6_Y<}L4BuCRvr)Gn>z$ydSj%MlhS96x=1X-C ztF~4s#CLi)k$Nrgg?A4FQihj`eblg zM+_g_DV&u?`YyHp32&?T3h(Lsm}>pc#1!-Z$9=nv@ETcP2@d^foKK^a^}`vTarI9G zZ0!b}LeN3Gu|CD1wD0JlR~1y&acG|@LZ%){c}gQE#h%>+`?`cpDK~-MY=lyc&{M{7 z8Y7fzWHFL-X!!y1WMEH}+R^~0v=wy?R6Z}JRNy%X)l5FLEYlNzMrA7sww0Cfm{eDE zBmw^GaCq(#9DQahqa)$hkPehi{SW9fIgHL(&_P|@0+rXGs97S#Odg5JLN&uaFJXUA zv0sH!Y*=CjH^@%0DZB@h-2*&>7!`zt8JvYDWgaMN5R*|4msv6ysdYO*;d-lABejwE z0yf)|+3Y-tj_ew<*(voecyDHhXbeZGswQWs9g=kfoyOP>)t44b9>BSunH{3|z7m|- zA<5fKgZ}-^^iLWaPT`gfs+SbLvI)FK)|$Zwf?i`phw|yg_!Qu@Irc4~g(X`r=_P|Q zosyuEUKXG(N?2V`DAh|G`3FymE$O9%E$QX!G$g%bP^Om>uB4Z*Q?8LkOOzY*a+|(3 z^>8XR(@Tkpq?ajZ!?(;pjn3>Ki~K)yycF6?<&j-yo)A1j<*y9RqJrqF$C>Sq2iN@+ zE_c`C4`Oie)z7Rt>ES;!IGYJkd@@cCmu5&=;9JG$x75RTG5qbIGlszt?4k85OJi_o zfV`B!UG#7kQJ|3k#WQ7;%t!DB+*fuU*B7(zNN@TbTE?hnD4PV^8A{un8wq^$ub=h( zOM$yCQ5%=^T>i^?sVDB2|AfJ*UkB`Wfh~iTm+YBvRe~~mMm?Oy(aFd3bAf%q#;6dy z>&fW+DSdI1(%A?)Sq%E41bwQfE{7#S$!LjuxfT51_)TkM8Iwti8mVM8qSV7HSWSi3 zzmR+sWI<#2dLM;p9p&}AUgB6d?1F3sUyMj%GJLR+IO@mKZK(mu zWS?Js9y|ctPyC(1Ef5E7%%w+91o>IHv<)>iv0CEuYb)E?+2Zd*`giGU^FVw#z6=~F zy~hGLg9ELZYT0}~1aL}Eg70_*4t#|S{3U!}RW;X7dM~WQZ;qzE;_T2u?9?1R!hzd> zlzZR3azy+AqLOd;zqIC}2Sny%`_aViNTA_CgSP}RP(%RPF;04FQp#MKl`rW+rxj@BrmI#>JA4c_bl5iHC5p5GM>J{#|H}ZICt)hXx^W zg^nlgj&cZ~dBp>WtxI;AZL42>mN=N0dbV}h{;Th=;_qeSiSmI>=Pv#EpD9^X4}Ykk zdO)0358zW|_c$8?X@S7b3)>|r+r5Gn-k`&f1bm@TRX?Pd_7FpQ-oc*f`!66CD*U0; zKt-&9D4qld9-}A0N6PLIQP0Urae+DHy@!Dt#PLo@=`RE8VqNIcmpw-7!-Y8+_~ zD%t|q*y0~eib=M5obw>l@Iy?exo<4a;lGb)z8fRo(Tl-RQ*!*{7flcTkXW4aIH%i$ zAM%^{M2$}oB-eA}6sluT8J?KQE%*;{2AOhiPBj5H7vUdoGzkqcgDBtPU(wr=(J%=~ z=&LDd5QwI1=_X#LnFc1Tc$Az}Fa*J^BFvC_YU;QCh(oXtE)Te+i2OY{a> zp?7iEAHR7Y>9@QHzfr-;|Hu(LrkIK^j9>zTVxrlt#8mmo`fy64{-l^ne{)Lw28NI% zrs}JW$`4CRRmva@o*|9Y7>Jii<--!Np@g(58%o(&QrZ&r!!dP22m+I!_eH8hH8M<6 z>gx^83g9_MB{*|d>i-gxWOwUn!D{qFJT)IMN2n5*5517kB&lstthsQnE!h{$DL^;d zB;4T^D_cm-00TiU9aL`m`=xj?6*Vo~A%5Z=GVk0uw3r|~n5}yo_u!> zzm84VbBGgv-gBD!1s^G72M(BsPl=gHB2%zPMK*sywFqKfmW3fU$&-y}`v@i7{7W}k zH`EQ?_e(Z{0<9PEI8C`EX-@=AyoBO>rI;nmRNVokt4%0P2PG<#RijiPv=p?exxEdd zW#|bjhZ)4v?V-I#Pxlm~d%X!=-zE~$G*1SS$5&j;5EJ#(6#c1e;+ZCtMleb@n@}2{ zc-l}_n&(MRNfNyM~b?p zlzsJ-`IVHiLQh$Kth~IG5=PYzk}K$h%y2Hzcxov_QF-!~UM+?gNGo3L*rcDn0D3;q zygSU|F4C=Qy5y^RK;T|LBgZ$u*RX-h08atD$&xSd_cDCIUkSrMASnj;E8+16FeOjl!px-py`+!Vi_z}8wO_zMZw*!UO|3|m3NgzU@dO)WyG9J*zX#t>K6iR6{ z4H`TkSkeKFk05*F)Ed}7Q0!1Zn@)Kj5K|^eFWtH(Q&eD&q}V=yHck)##e9j>f1_I` z)7B56X9l2LO;h)JSUCHIn5ZmfdfxJ}E=KL8Mg;{}-F!5X#xMy6tuQv&?YK*c;~6^x zsW*XVXxa{fIU(TJ`Fs2wYM4}ySR*~6v01=x0*=@;tm6*ZP7Gc6f}4Qf^!lHBbajOa zL}=axoJz##O;1L1&0ckYzb%g^&l*|DFNsgWL(UT;yAN3GQ>Qy&@oTM&5>tB8K zE&zY8Vfcq|_>RZI6@-=op}&(W5MH|ZnZhmTS}jhFAX6932SL*}(bL-0=kfGNea0_x z(IGW=!PaX@4i2lHY!Ua6g5i}}&1@I05s@BthuC%0+1ax`F4*3GW^{ff_(2a%f2RCo zY(b9vAwSP52_a1RJ_)XYw`SJuCn^Q}5P4hqVMz)PdM<|^y~W`e9B#6uhsEg?!9u9+ z)6Lfh+B>ZNyNV?B)WvJ~d17*kxVv(*wV7|>@Npl{n*GrjQu%)EOU`Q5nCPi}A%`)3 zL>3G8_YAzdK=75AHNES{v&lidX5sTP2CmGpKI>TZm)*FtQ!-Z%wQSbny&0p977RZ) zF=k{)=-58}3cGKQnmV9gaYRT?@8W~QA+sB#nuR)2@Q*QrAhH}f<%6`Ccp|jLuI{BihF)LDE{!t zD2Ty{(Q-tC=msEqft7UMkd)cPaQ_=UNhaNLZF|-D%i#gmlELMh=)DvW|G>_Rue9Gd@TO7 zW$u+(?HpWdh+{P;#;FB=H|>{Xjnnw@KQ=qsMN~~4eP*7z$Lb$T2k-b{-T`8<8J(c{ zqCL$-)%wEQ6*DgtCsz)Gu67e+g@Ljnu3j~262^Y3f2!hgpQlmLW zTrL({JIr1*cu9)c9lK4>Rxkg2q_g7z5qEKtO_LLIJsp}K6@L;}e&%|JG`~Gd?GQ)? zyc<+FZ`;V(%evrFg~GQ?Z1%6ue8(9CxHHu3SjFZk=?qVLI6PCoY1rQtaj zmWmMqS$_2-+9A!`uIvMk{4Pn}#ztNMK2k$pQ z@>QQ+dv-oO(`o9}6=RnNYxZB;Z{r@@A#YKFZf2(AjO#1&D}r41-`H>K(XVx0dEy&KNVgE%e~DV>F=@wK$rH{y zI;TwtJ;=(E9nk@o-eQiNEI51io%qRJ61-EhhqU|sT_-ae`@W^=d22^`MaJ$QbM(;= zH&fS(;uR6S;KgEP(aVy3nw9s7?V|&u%EyrTK>yc*pjxp!rT{@Poh+BNWK%=dFaqR^@+^}tJ_rP85 z!R}+es3M=$wq*+H1vP5NC^ly^adyU>8awnkx@pWyO*zmI083ost3O)qtG7HrgHc4} z0%nNzPwt@O+6NOI)Q82tHaV{Uwb0q=2r;X4eD}JedgBJi^*7Y&sttab>+>uvqSG~F ztK;n)4~g29e&f!}3$?YM_t98Q))*_Rm}UILAs1r>*Q!mKQcN zchZT`Yo0G(x%}Dc!(uSG1ZIem$Q4XKO=L0;-Z#L6!*z6m(+%@#;w28#t^AF*+%Ek1 z2>DchtkcMEQJaTP#^V!%jdUXAuz5rl4PQYaKLk$(Q!{o2Agfo{wqP&hKSTmI$@a1E z5SNrPd2;a$H`iV#>HHn-=qUcnTIWIU4$0h*HGE_0kaq{2^oj~h7~mZp93O|uz(68b zp%F3k1t1R_)krmteA9-*CC(wzAQ0S^M>7&S4A6Y9DcfC_d|Ns0z@pCLxQbJ;_Ufv? zDq`kGr1<3ZUi`j;g=h4@zP3x9NlvSxLE{$0SXT32?|D$%ePpQE-(<|GnM=-%baLKq z?OstjAg7CyViQ@?qU*@2bgd_vjIGCGO;EvB*g07wqP6RhhTnlU7PY*-p7gk|GJr|S ze52PBxo`tVPw#GF<#O=EL02osghly_&cwUAZv1h5-sW7~^*eQR&W4fpc79VGqcfZD z6EY7qAJlu1Sby-aShr+L!I|QsQ)6>aPMh|@Xz1K=)Xf%fca?>rnXuQC)eiFlJVP9q zdLaED>cm^AA0g>xONS68+hZvu%OxIp-PWg#m@&MoeLH6}>+Q2$e0&6@Q+TjvU}$@f zNFjTbeW1Hn_kLp%gHuA3a%WGMX=T=CF`0hxi)PP{O`FOmm3GaT_i%O4qKe3zX}d{F zC$I5G7Nm_0Xq{2&+1ev<`M8*}94{9?vb$td#Ii-5$DLjhmx(FU3#`ujK=u}pT<_*G zBcpF)+(4^uOz1i}5nh;R@sZuCPm8M0hC3!Wt^9M_?DzUxe5A_FAG;{lD#_A1M|>EV z9+JI%7eS<`Y^nC@Ws}|yXp#%7N(3^D#hAHGlgt0PMOXw3d zT5i1Bwjt<>s!~f%C-;Z0qboUX+Nqb@;CH%N^UfRoShfD6evZ8LqHmVkG;8M7zh7_H z?!2{I{_r7-ldXB1_-?k|PPr@7#raz-Bl@}yI&yBj-EQ5Ya_x5eNtfr@a7bN|kdWZV zd)gHq%E>uYXg7IZYo&b1W+;41^nMKlWxh8QN>XuThc*lq{RO+wAwk0;k1dmx3$PxGNAv~ZXeD0$+mr{Z|hK%X-LPQtY-J^a<}Ds z_w?;x73&&5*m`(eMP@~D!la1a(Z5c@^~Ii+3T2npf#H2(&=!5g}IMg_$W@Qe;gpff(UP=t*s3T`B7jHA!;vqCXE)^3tf2nxi|>20qN*bZ!( z>LMtR*(#*{X6~xJ9jcpVL-JJyc}vEfUUyEu%-@`vdX=OT{xk8xrpVDL<2Ma&7d$0r z)Xr8r!iZV*!V7tsJHO`so6z5xpF&KD;M3l@oteU|r(@p*8JP=*$GB8z0=p-75AQc- z&zZ!8IdK{B?cQ0Nf3bXps5@UkQsSofvRdq&JTJ~dm~3u~qS%X*g-vvYMebk!9}(Af zN)XhzuLPkJBH2wgOz~oduo$?(Xa{NBXXZz(^wry8RuH<5K~qs;mP8NOqlZzQotnIK zMB8TO*_A_wtRC4PtaKXB)&g*;z2nfUDsmbBCtr{Y9yt z8rI(${lyJM`HSLgOl%xRjmdmhecoZ!?FqZjg}Ir!Eq%DN=xOGi!JKdYp3HWcd-BI8 zWmk`=e72o8m$g6JKS!^7e%4|3tc& zXeu7skMh|5)A~;~-sT3~npXti1|1L2{_r zb&#wx&kRv};c1sNArF8r?2e;@vJH zuybrdB5n=76BZxRF+8j9kd;Zf+XuxYd;9j_!`H4wbtAwl45vGoP1t(AS%Az|_cP38 z03Gn4h*GtCN=a**h|^FnwrF=Qy@PAJ9j7B4E!E4P>@3?KZ+>1mEI+?I-qO<2E;l*h zMgsp$_3Gyp8RgF_^A3+4v~cX+jEszZ^K+UV(MD)c-yQ4@s{V!a!2Pc=x1wcV>2p!@U_95XZ4!ryZQj?O- ztBGxhFX1;JsVyf<6QdC-qp=oh>|rd_l4XB&2r|qbNZgUfxYpT>ba9 zxapa0u{V1IQG4y?pRI^ng`*wG6?59g_`}U@xtZ{lEu%Y#Asktu zb{uznY|F_zQu!;jGU3?_LgqoRftbsALof|Bs6;6_LbOIguu!FdglxL*r4B*jSZ3K< zz$3;R$?K(#Fu&dP+_PB@7N&jU<5#*hU*5ds&+m;})XUt`uJ|(X-RXLs@a6t?mNAoe zjXO}$viVYXJNJan@na$_9J*f@@~fxboHcNJe#Oz6laEC)qw;84$qz?HRacKb^8MuW zW0ivm_xNPZ#O0Brwfzf)l4G~|xL;lL79qQEOywHAF@4N;5tW0ej za`TFoX+CmZd}~hC=}2c2$2GsdJM&Nq85A-&xNu);$go2N72{`|9lb`kHh9^yH97R2 z%h_U)Ktq?dGCwJ>N}Weco|x%D#WZB+l00t<u15jyz_{dn8|~~EbwQi;eBFC28Y>`;A^+;?^g}(*G&@2c!2Rle{5feOneF|cIRF5hvb%o(P;@)32#np>y z$&r)umfDx?F0tqmXYH*vbLuVbbmSr(#jNHIcKepM2(fgsax)*0V&dR#vAL92>U>iM z2*NrOGyaQ;iZ*9vm&}z5t2nNG(Cjm){X*C=7om%OVDzO6Q%V|WXDd^*qhJAq3d8I- z9UV#Wd%L)Ot|E!>oYCdbK3D6>*VjZZDG0H+WIlRG(L0uw8TmL| zBx25v9@f@VFU%HwPaNJzyhkoGx5ytMp5_W;$!2lf;LSNBw`B{N+UU^(lE;Gh91#CT zo{L2BlC!5*1NeHN+UV-h8!#|7H8^?*U{RO-?1JSEM-K0^=oCxtOj+5?v{%BRH4|;E zW?!7VZ-@16m48Z~;j7I|;xe|bnrvr1wjqF>?1>jb)H3|M~wAHg2x!x+$S(AHdwZ5DeJlwk`Jbta$QeDQ&967O8`^@7=(yPG%jUw*M{-oEhPc`N(j!%H`J z^d@a{Qid%|;sWiq)i3X}_)=Nu-oIy+BU!v%!J)i;Q@$K9CqDP!eL<{k%emX_CbIHb zPxk<&YQ4 zD=DhUrtEK23Ag%-;JwygYDHF%(#<_AUM5t(Ik1-f5_ovhApr z_9F+PS`hGVbMr>+E&ePk(agWzYf_>eGrvrn8#ns!n1%}3{S=E^4Hc3!r(fC^Z3bF>}v(TZS)wqqyR+Qp2`n`Iw+v(GK= zgM}Xs&#qZKe*dttITMQ3ET3E}hQd3JgxwFuu3?2h7t#b-IQr}X=E+d!%o+m%H92$^ z8Q@6`IUESNyGHwyE5BSjajcqOnwO%{s$3U7JNlx!x|(mLZMTZQSdFuYEVQqp;UvLS zeh_VC3L9$=m%tsv0v4;W>qiM7S(FSny?SQ9qy`Auae@xCT5W91I#BOux4{}a-ls%A-=GA^q&dw1- zxo9zbKK((?>W;Gq^3fb>k5Svh5O`o45Iq`4DcWyBXkA3!IPdzy-sTa_ef)!x3#@)~ z*6nZ>cD#RDrKna7&+4{r9ctAb@!1ZjMTIcnKIpMcmUPl<*!cuH-qh{IKVg$b>+ej~ zJUjY~^Lughg}h@`)n;0GtOBK|yr6LUA4cU9Q;N3qa*(3(X^6@@7uOK7y~D)wmAOu4 z_U%UXNSvNvsgKI9x-+}{Qzvd`R)fk_G;B9P-W3Cj77C#-LlFfEiH@pWoMW7 zv@G!@U;L^oZaKajZhZ@JpNC*oyVJXOKteoXTX6$UW^Blk{_nJvzhC>ufC*v$2R_=O z+l@Y2>A_cgwB>~}&gAQTwA>xwh3G%-eaxqUG;b+GEFva6lBd#;9$f|CB_9?GZ?LE# z)P2MH2;Ok9!{!d~Q>x6Ioh_7tTWrN-C%fM(?~ORSVQmsFHk=9l5?v5;V{k(O z;!RJSroR{KIQ-gh>z0E&oI>;ZT8Ef=wRQ>d=`^MD(7wD`C&$HCD)(kxsp7Wl1{Kd} zvv+U1=@ZG+o?T5uTn~2A%Rp^h@{uMC?NZOOAw{f&*@*7-q z>s=<%`Dh7#aLZxVuF{z({Rbt%r}78*3&sp?lPzNS*qTD4ZS3kW#Mat~N)L^BSFHi4UG@+=g?l@$@4p4!IEZDukHWOiY6KT0o zT&CeGG-Aa@{&;n@HX2d&S_B4UWIo;6f{8M_6Iq%sQnyndB}8LUV~2GevSY&LD0_ut z`SJAPCGBmkxzWyI9A_0ds7pq;x!ijA;2uL-4{yJZ%pbO|z41;` zf(tD5F#Rng>=5m%#aUmF8A6d*_9>aco!u=u?%>yovpyp;mn@NcF0GA|A7uZ&G(6E^ zfm_U9meYG^;3$Jo0HM1m^aHSGh?-U=EMLuM#QL^qA+DW!B7aHpK&3LI@5fF*WR-Mi z)+}p{kU-M4-}3Id>5=VQcVACpOL|9@#8BcShN}c|3)+WlL!`w;DOd}0${NTpg9<71 zP`7dEKiImxP1@8bY&iKv6VTB)Jj|@wsuRnVZVq<-=G@#5%$iNgEI2S+5bw%WLf6r~ z`C{#wmd$YoE+<=uf-F;7U}&d0Ky{U^UFat_`o2p?W~K!S^6RQo>zcLdAFk=xUhQJq z($3B|zk_YoqJc%lE(1@++jQ&JDIri{r=85Nh#uWdu2Ssg6fU|tVf=d|7FUY`xicyu zG#6|WxEb6uRz%y^i%6#ho(Zg2nU<_r_Fnb5r{*qgQjnuuk@j0^n4#x%>!BR7L&q_E z-gD6fe0lmgw&vA-W?0@9?$OK$Cg>D6lXa2n%1)o8h-c}i2Ds^MdJ-BMXC{%-jA)}m#x{R8?9Ne;_3Zt2 zd_RWKb)Z1F|C#j{7m z1npvSAE}}Rs%nbIdY{dn^?FLbk%xZ3p>tKP+4IxS`O6t6GIZm4`!6L%YA?n}EO3^4 zBF;=@)@HoI4|iHNGL+L9r~(_g;3PX?XC%F?MA~}5X5`>YXX3ESxKJX1=TT0jC+6$O zPwt_N2z4y~zLNa1`on&X-A;$9r_K|d=Iv33k^^NmW!=Te_eq#vl#+|Qr%M-8HT-G~ z8N2LaUd|^gko3A>enrK6+_z<>eaemDM`(YJP2#P!>$wk+^IAzOL3L6u6W!0^gVC-X z9b;fMjLAiZU+XvI<0`v3F8g6M`K9v1zK)?EbW=~8BRbC6r|zaBsJQML{(a4>g5HIo zX_BBTKFP_uxQq(AaM`kjbPp~jhzqN$b;U+S#-JjG3p14y5zh&dg`q1M0vZj|CDb3X zQ_%?&Y7;R>8h@Y>9y%qaV;_RZAp!PKqY#onS}(bjjdi)-SFW&IzHz;^UC{#ZR`9aA z$8PrT{<=6ICe}pNGSV%l``9k#_McVyC(cPTw_CDyDe1iI>I(5cx!-c~TbcVNP1u(; zWAU(p>AJr^`C#6MBfAB4jB0B+*ClOM%KYiwCnuN96Rz&v+YGz1t(&1E{s}$!lbIN4 zFZ(q-n=zBok(r{)GAR|%A~D(Y96F{xl?#7ff&H}jJ{~LQ`(c9g2O&)h$YRMfiHXcK z`B2jY(^J54^-qESop`qC213)@+yI6TzAQC{f?y;Dp5fVtK4a+ZDmx^u2RuSNtC&Oi z3D6AC{ghtm4mfm&rP@)@)I5@j!@*5zQY{(YC~g@usVeaQkr^KBC&Tx*1WX$7{SClN zp?f}p3xnSl!WkS#`fxB52^pk=p{65oh*xhyL7&1U`Xk9PMn4|T@qP-IyE8f?W$6qq z9rUF5Y+(p=Sjs2a4V1f-(UJI$lvOc&Y7m|&i(fy`X*!Zb;dl#YFySzn7+GmYpk7A+ z)*|2+;sVwHtZwN2L3oDe9Rv%9Iuq`OTFkm#<<-W^-lJ8|xUR(NTV0Bp9J_tNy_Kgu zZe;BfB%b6>&7G_3#YZgR-NbXfvIj?mXAJ2{x@mu1tUXSqgr%j0g{G&4iS$ykg^+y~ zZsGKz6K`)CKGd6JwP#xzWMc}?6dhhEjsI}-NvZw@1X!&~Oxr>PT!3$zs!7xe>XqzYmUO{I;t(8-xRL6Bs<#xU(WEa>-p}-Z-Ok74;6v%n7vVB4f7@H&%H6o zVsR1^rgb@syV=Y*GSezGGjp2N%tJG*QU(v6ZUuN+F%#JJ5Z?xHAcJ8CXGI+*YE7 zQG}hu?pnGtneV!*{f2s7rC+na9?lW@_U&dwIJfEUqw>>La^u!-7L#$2NZ-1ur_nw6 z^@08S2d?LuhqZqyWQs?$kKnJNedJpiBYJ3yj&F<^bED~92)0I<3g3V!k=1TGarA_Z z6M5k@xGDSA-F;Opb!)hB?`|=$r?WHi&ieJ@v?hYMg>`5YM1UXx5Txlai%BXS&XjZ~ zHzxELW)URSkF$tC8OB*b<43GNBRxO8>g#-1W7DoMQLIUtoob^YgR%}~m55*5(rLPx zk_-2Ae~a%pbAM>6{hFn}{7(p$g~-C8$ofc7!=y<6hX=donhm(pB74Koi!&#-`U8a_#H*mPhoxoy zo-ut!mzkT79Ie$n99@b(Mv0ZqB)4R1A1lkLCkq~`v;{i`SXoZ0De04zZ)%#B(boB+ zhQoehX4I^-v{_M!vHcP^C-&=?u({9tjLeBa?ITLFvKGho2n*}6S$W)W5D->lHdd=% zAeE*<*v+g35*7$QGHX!>E1^(aX2;+?QOsl@Nr40PnANtW<+CPvysmZ}9Y0179A z7(5wM4a~+drRdAxw+ZM1e+BSqY7T`E&z=|PpMB^vo6P>Ar8{)ZA(b6XIV8N4=yJU& zK25TrkZjo{1|KD5RhDf8zbt5wkMN_!kI!KAU)doBy6joO8BzSDxgwo0gQv+ zvae}sfx8r&m0Dpsay=~rv$HfIjFehHab;oyl;KjV3|vNmB^+o-Wuu@82S9;AnH0}~ zt4~8R>#)=#WA)A86>KUGlXIPirIS~JeeQ9O4k*Dt7fW^i`TgP-f3+mn7*rnU@h zo)wtU^E!Dpcdls4&0ffjxR+GXvu8!pJ>7~qx`lNKOClncB-CL9j>R#BKFqqiVg!cU zu6~{&H(RCh8539*`=%%6kDPkNwRS9e> zn0ahpl>8B2LTw0^O>pNJV*KtNVN+*Ba^^2}wx+uB=cfmDYL>}$noXLELzj|2$nMKg zrBk|*|LT&77BgUx6VGJmE{SoYNHq6srIw~epdQs0c%*Dh#a*reZZtw-?C6%jEXiIiT(ZN5hV_6 zDd)G0FX}hp$jr!-;;u&z#a-0pFwD92j;;IyE`}QixLzD|!|hNACE#WZPE{BiP&kDN zx-Ghu4&9Ju^id$U19c6i;{82gBt{4fQ5X)`-XOKDUUBH0>-brd*6o~}+(u8c8fItL zETYZ85R!3D_bxxf-*f$dBvFxbJu~}i4!5SR3iOy#LnZXpg7rgMEp)DdKpVzaUxb$F zaSa8c2$vHUjp|`~SneI3)~R6Y2pe19X$2ue;zRAWd*;2fW^VYu`JkZZ?JV9!T$`_mJKjDTF{MA%s8zp_e4I5CQ2` z5kisPL_kEWD0VJ3^jZ-W6&1T;Lm}C7zW%YL#F95-h05Hm~q=0Vxp%HPpBK5YO~+7e&I%PxF{y3 zh+fiwX=hj5e(hhgg34y*`2}T1x>j$0V&tG@#qRM%{S)$MmIeVaLvffYE33L825IT< z6b@qo#7rtqwMtfI5=4llIAEjC7(F$IY(7^lR$UM&ABfJ69Sw zk(82iuwzoFl?$+I(zc>t?gXQ<0P5)e&}r(#vx<-hCII^7WpobhGLfJHI<%gc7zZ6) zuWwV#Z?`>E;^egD`_0>)O!71HJxa_T7_l_nLQ}cBIcw*L!9|VzOG`y{Soz>p6ASGN z{uP?HqSn%B)Vi7#gDo8=v-eJTWy7YA=iN;bzsviajC(m{c4c}?sbkP)cc%=`fWoSS zA1%xub|7QX=(`>oys2jGU1;AUfcIf^d)m%)C>l)LsHy=N8`)IoRS`NJb27snw(%** z^*Hw1ENjMm+OhH5H@VGpyyKmP%TE^DN52wdKe2hjU5+!Y;xZCPTCjc6_AYXah>B*- z_L^I&S9k2Up0<6^>0{dLJ71YF{p<>oEj_kq>+1E)-#1Q+<#A@Dr>H9V+;%dCGjbPt|od-ufxNzu!CxPZh45L-b6k?|cfzXot^yQ{> zVrZLKp7M4!Mt0~M+3(JGSl7<6Dk-t>v2bzBcFCKP<76>{S(Z07+sUHpt2vgX11-G! zxH#m{Ru{y4Qd`WiJ4MrwVGGN(Z~qoLY+)G_dQxjQwvgp^YSh4^QSybQHC<&AtVmMm za$zi9Kznqcx-N&lEG7AREN4*Go#=)(N<>4x!p=El5?PDWp9I%XlKT;&O3kFxb3MtN7i=T#mzOV%FS3y*ZMIv>xNpgW*!qI->7)cJ`OXV zA~p{XV=MQ*zr4W4(ld1yxqrmAQLbU3j<$!SRco=vVbzBV_Yc2oh~3oV_4UW6+6}pD zIC_^wIf-zS&K@|G>>$9%b3ci6lU|#&s;G}( z=`^1ByJ$aj-nSnsJp4BPv~K^ndushBAF3wj_wL0+L(iitQm97QJAn?YCATZ|PdGXT zqqvvp-F0+dIwqWA#eKBiT_>TzHe^hu;ywrNf~MAq(m4fieO#;t#YJ63&1LD!F6xSg zR<;%FN|kPTG@H8XwlM-W^Of+_eSO!4O2YGp6)H-;EI24>-mAlf8Hy*7uS(X~TVMqp z*0BCcXP!q1dQ2c22IcnYx4WNc?b65E)XFkKvK}@3)Z+o(3{xivm{0XlF~2?uP^{{1 zW)AzRa8@z1p-Kls1?Uw}XfaelNTqRTE_96yd2ei6nwx{$(uc>^Z%T2sce(x1;+iSW zZq6xHBd5Febx9}tv`;uPI22|4`Fob!^H%NHw+=2j@ZN;GH|#mFXWgm;CwAXO(zl`L zdWfJDvMGX!lytjAtDZM38%Ajl?R3XpUDO?{+4EeXQgnc$w(9g3$r*h-DiWi+-wNc6 z5)R+udK!*?)A+ouA;8__;ZAqJr|#5WM41%l>W z2m5G!DE@LUq5O2Al)c6->SD2AhIFunO+8|X>;w93MwWt-+k(C?un#CK1HSo7G&KEI zpedqEO&xn)fd)2B-JeRLNhkzhJxWrN`8qRe`}(J+Ik-AUY`uNky`iu9?Ypw;>cLl~R}bFy9)|JVQX4-VM24qg3FFX{ zGBT{&(-9URuEBY6VZ|})kB@uo>q>VQ_a{lf6MJ5*GC50{?q2fDY)2P+kJT$H@0k3< z($|)fwA)guqYshrR}Q^MA`Z8Fu=cL|7Jael(2B3t#FQTHc!TYaQS1)nJ^&hKdi<<8 zN{%={t*oenU>rr#$Okq%By2l&{@?F14{7`EDrTNw+M9fE$4}bf_hUO+ z9J)z<;G)m*JxJ+QbZ;m|r@tD%*K~K-0$iM%Qd_DkVR7WW{@V zCuF*(&#sD%t%AuimPxH)JIAjyRJ^3l65O0x9iPw(@s=NC%fUHwYB z3~^h%Y9SxZ`y&@!>snmE7E>jYqR^qEF3jhYb6`zO z73l-37Nljg4o7!7h9eGMV)K=QQ${oSaDin6nSlL91pW!71D_ZC*{zr;MYQuWJOwP+ zS;oD?y{X`7g9;0Rs7-e&5#`twdmc>4=wQb0y5;}ZyA#)`?1l#Ens(MUW?4s?I&(l` zlv>PBj1MQ1)!B)O5o&c_VtiPeIx8Uo_DOm1iD6_C^^aDIc}ejR7nZQD%Q`Nx&dWMJ zB?H5{H*lDBWVjNi098c3Hx%k9#XX#mGfsP=lhcueX;cr&KB1Ga+`>X%clYe_-K@lC<-(^x=?}loP}v z;zp?pmZW6Sz=6SqA=TLf%NG{>X4V56;w&w}Y9!3S{8#npxlq;Aw_ij1I?Oon)$B4D zc`9eqQ1dcM5TT7xHXTDN4xaIw*qU3s;jsr--x&MaM2P?Aww zoi(s@VIg|b9MvrL5__g|#84SV1eOkY^w8~BI*vH8gRV(^W|JTGuDx@ebapUvau9iI zom9qDkJIjA4s{%12X%a3$BY_>Y8WQDvk!~Mk@2;365`jAU}e(rX9AelnO9W;e@7dg zbh>?`vZ|zNHUlx}(4|u7bkORoi*v_FXZ9s}m*gEg2-&d%A>LQL#a`eKD+2+bL%fA$ z1~6G=LBY~>g_MfzWc;WJdZYuj++!GNxw-kiR<=%#HhqUyHofH> z;%k{u?2ozyLCw9-CqpBRp@tVB6eK>eJiUM*FS3==6(Ww5u6%ivcuCq}*zsSGu8tj z>{yF_k~_IwJJ^<(ll|H%PUK@8sD>J&5+@gO+EI!p_fQGm zOw|T-4oZ8_rjP2ROYq54nOig4$ToXvG}B+IBbyhmL$BFPjxzBe#r!anYhtGXLSJy}*~dX(cJ@V{_Ql5?wG*6}DfZfWM}b|w{PUG7(byhAHXk#T z1N2ldTcLAuF6mb>@Lrqi$1;NPYwHFu@`B7cTv9K}Xd7oGZ-qVzE zx0a3)d#Rt<6QsX%uSf5yteDH*H&Jv>Pf7_Ch ze`%zH!84{mDSbM9a>zhL+N!-w_R0@pXmn9{Tf>TDbO6eV>Nruses*?UxT)#jd0E-x z@GPGf5mBmE4+sk{RhuMeFDKQdB-bS+)g`CYB~4Ec3rk503C%!<+^V|5oMK#|;C@k6 z-2rZxG;x<%2|6+;lLlabBl$eTUZE@7p#pz{4u!%CDkz`Od7qMUnJ45C)Ygbs2weRf zpcB<%4q)OJn-xt-j!X`T4hTq9FE=lqk{p;AmEaeUx!0^nfhqx~8UoVfG17>eQa7hcQ@e#^ zG0(>{56O2cAWcyeXjYVf_rS%lw$fO;#Kf%$%+Z9c3DUzYEtjNiq?TSt<7LTm(!>#Z!X;z=_c}{8;3I%6My>$aY8RllZOi_pV(_hDry~ zm%-UGKk3a-l`b2JM3|p0zy9Mc7pc-sXIQ#_KR{xBVtzCZ$ls#oOT+U0q~ZN^a69@aVcU9!)xiZ0(Oq!0vz4&l zaksM;0@Wtb!H5z7lmbmJCofjqBRw&T_JvjNjI*K~rKi%yx6LtjS5V^}ty@5r6iGyQB_cKj&4wZG5pUq3$I-oiOLr@&@`h3$~RqnsaombrH7q2g_`c1qVy z)om$lJGZPdZPW~{FnFZ*7NjXaQ+rvpik6|$A<&G3!s8KaO8C)yOz?z+$X4( zdB<6M#Oc|?Cl53EPTF=SX0wC#kTWyp&Y7>S*tp`g87CiWoLIW5a!ci!K~mUVcd@hE zf8;ErA$Q(M?orI;Gch!M3VOpFV<)t#5asD=0%ND~J5QA#qtXXz)UhSO=>tv0*b>&f z+h=<#_v=0FChR*^>A`tH%B`cl>-}|V=BHw`#oR%8g76V|WlGQKYeXT6>GA@>!3%Vq zNC{=V?Yf=Pz^Ig5IuNM#3=U2ZT1;1u9W}|ymlMS03=?BMH8zdsy@UFug|POc^UDWo zOxHE?-dBSAdwZDVjviO@iG8k*bKDQ*5pHgt&OVldJhF!tlnhf5Z6&u`o(fzj(_mB+ zC~#^pLy^#7L0h?$J9KhDAI_$7PR9HyTeURQg7_1Y;G~2C3)VC-BQo19!DAY?J7-RT z?Z^e`J7$Cq7GlyXhS?Sj@^a7i$wtnvamOXCcmVwkBs`t`JN0+yLDKnQONMEL2rn-W zfe-Ta@ew@UGi&3264!N#U0JPy@n z7MU{^;TiN{9?9+A(qh)KZ5wV|CRY`jl@1?OWL8=U_(M=JbA1g(nVRbH{HuI(ruCR+)5@+fYo}SM~kb~NDAbT1&PFz58q=(ZJOHsCEKzAI_ z)wL;zn{}*h52+3{#FFAhjQ6nM9NKQ1TBBaThC~D;`-rvSQYQ+i;=(#6y7%^c=s z&e+{Ku3?C&sZDmhb6k;=B}t7QK+7nEcV>=ud}W!}Qi_tgZBAC%w8Y72u~U1M;sP`- zYvbjI4bxbuK4spnw-;{DTe#!h*~0llhotYMuZ|o6kAfXEJCm2fMp_M*8PP<Rh*JF z7k$>ERhWEkC38xzP^JlkPEtQcYp+_Xy#p~d%r{-7WNoVX)QD`J^9l;6u`NjZ?#^dG0~!~6N%Z2BZw+T>2A{1F;qr(&Uk*ZQ*OFn!WJ zPCZp&K&H#;0+t;x81O@h7-&$#tf3nUK?7)iO=xm2b{^=HQ>w}KY8+-!bzAtDX|{|- zMRR&?SwB&*%^UBSSzvEVqVh6IF^7b)=RzVYg52X$$i6w+&&tP!ytZSuS^0vTthr?_ zgXU&T%!r*Dk%#%DbU4N`KNQm#QLS=U@Y16pf)dgrtZ!c&X zFV5xqCl?f(o0*0WjL5PypKDE+K!?rT!P_UbGF7%)M;-4SsJB04EVZRffanj30_4Ap`_?_ffx@vndl(CoZ? z6=9uJchY)CK(rtP`z$O$R!tg4o0|+RR*M!^VsuJYfs>~fzMOGCqeiaM7A@E)=y!r(5R{KSvM*{ zz2so0{kFerMy^Lmjish!TyRQ$v5DZ}=8;@8Y5fGDtkABnWmS26MG}bTK3ukKRAldarL$~_0Nnh99);{7UbsE&(+)|FR!d3w!D9s zXI~_K4mq1Fo>NHZuBH4+N0CM>uCZJ^$jZ~%UE^Tx5v%5oGu1V4=*L zl0o14Y5VNj2jK^6sll2&Ee$HPBquWZTT6?rmG(MJimM@~Cr?Iv2V}eo8Mg#cs~bXY zJYwjq9`r=Bfe^CMem1VY4>2vCU65VqjP{UKW0zL$Dv-pKfaG9v#xjgJ&dKDhta*{) zZGxQ=f-FL~gLgKohb}G5om13r=)p#n=i1poK}Mj7CNolG!L%tB);niLjgB>qEGSGb z(3qu0Cr=qY$D+@sG5yVg5^&O0e0abJ^XA2k7NZsjL#9y^3bpJBY(l_$! z9`g6TxLM`i147X&FhB-^V_{cyeCfZwTz0u z3(h!E@pjV3!Z2m&Fg-BGL0YuP_DPoj-tL8YYi1yTKW`@&=>o8HmWw>|Vs?q`)BJ@O zGH0v-9*2R)H60$<3c{eFqnLpC>B}70<#=CYURG6Z@GuK*nP`1*VR3yy8ozxBk`Ifq z^sSlc9qOgxv?^}5$`;fgo%N{p9ej|NBJBi`(K$j~$M}M62hLKZ?E|(6BNh)MYiG}K z;;e?W7Q~K@(KIpPp;(He;WQReAz{g;jplbFVhDgvEjAoaTLma9_IHTIb&xG9SJ??X8 z#NDYdkDkJ&A0X&*4vZULn46hL#tVE%q>r^v(|B|9Mcn7P-roKCH53+=l>Ft`|K>ht zASP5)?6 zTgzUL*pE2=tp{E>1Wt$9Bm=PPF{8k4fZ0|BS~tWLXpaCjcd-&hPUVi*Rk)2_JFSe% zI>{~=SrTzwJ9w~{gQSdeK(=>3OA49kp8R#i+)MK9*R_O-UX$<2@^#6xs+GJW`VKh*ZmVex8pahP&vJdHkn!NGgHV`f-brt$SC@ufN<-Vuw8 za6B8#cR)9)B<`wm^d3!*&c;S*7^ipRL%Udu@C3EGEIb^DluCg{A7pHA*}s|{9-fsI z9-eJ@lCmCkKu8{U8NJRP=%E3iS+~Op}hEMPc(tFW%Y3NyjpuzCz&+qrfO2lhJilur6p*=LztD{YE3Uu>G66*4edTx>pO9oJsi zTAWqkmmN}kTLBvR{oFq0ZDAW|Q=Npi7d0f=k~cBR)NI(ITs+N8hb@eX9;{ZUL_`fX zF-hX~WzNXTo|c(8Ejw#QW=no_bWUzmQ~^-Cf;x%;YSdDVE+5n_VzI{#Igc$JX<=f! z(q`p=0vB5^?(%Xw>qhtZf?(vmh3sS6+0VdZVN%jF0I^KFyJ_aQ)cW{5OAq$5N7L6W ziYUryt2}w@;ia?ed#W&gF|9WnYW0SNFR@|7KWkWIUEeStR}+45?fO;zMLvb)qvD`J zx(Z{4R?yC{n*gEGIp>HX*dkD?qE%yTm>_t^mt~GpTZ}3(x0o|msv7U+?cMiMSOr&5 zQdtflFW@Nq(WoH+p?1t*X&{~lK~3S|ZSUwgmS$6iA1#`p5xo6fBW!&=s-(i&s+`5K zmf_XOzW!FGw^@0>xbj;lXQx3qi;c+Xsf*ac=CC|lKEORndkBjiVJbUT+|D*~k6wF< zPld0o{ETXmU`c&#;Jc{Zq?!9wu)KO)7zp0+EcdefxKIRN-L^yMMutV59cQ6+wYMYN z>=IJ9rgATn+oT1AZE2CT^i7uAPiQAf39!}_g_vtEUCqP{-YhxDljQqhsQRyt!&lpY z$b=NA(4Rr>FT9WMyCW0h=%?-OI&0&aLN}$f2!bqa`kmh4HfcdrYkRIh0~G- zC6oMQ4~c1`rJ~J{E5jCDPhrAsV^=v$!$oE&CieV zO9@({&YPro1o&nMOHFedlHv;Tq61R+W9s<9p?0ozVP5K(if|9lkY7z>%R_BlY(u<- zxFHc6-g_}9es}G8$zI;o+2|ku;bVOqK={XM zL&heJlTO$YoAB)pQnvOynR;d3zcW|QBYdtbcU<6P$TvPj!(uy{nZw|$f8#2pU--|J z@kLPRAkmk0t61rQp<~P~)|6ZL=a{I&gScU2(%{M&RU|1iC>7#xtu41+UMcJVUjktR zR2K>}uG^K`*oJZWUQJw1VX4Vpi>kY-)-IR&tFz-0Vqlb*fRk-kCzlOQESa0LRnms0 zr-kC8O{SEY;d4~*ipS`Q@{qtN}qk2dy0#*XG(r|gbW$*(4+A~Le@lWT&LpN_embo zzvbsTnP!#lEDF8QRvG5>LL9-tAlRYe+-^SYCrk?_Cyi{d)Sx+6O`Ib}M0>{ja6)LL zSEBDF_zkFfoZvEgk zw<5kep1Qe5_Y2MRh(?QPL$P4lbX5n3LU-g*Ed*#)Vd zt(MyuQL{>=x_SoDE^s^9YHa*PiFb@ie#)9sDCe<%MZ>~(_D^%$>sYgF@33L_ZE5t_ z8{YQp=iJVyArs?WlcPL5qLW+`CJuo)Wc$Z_ylN}-)^s$6AZM#n`zR;=b-F+NwKLKt zd7&JJs$DI8eia+Lbv(me;70JjbxFg_7{heoU1E|?ZHsN+zP9nm&SCCuHiSQ`Fta_3zuAzz zj@3)|jTmuk^GuJuk(;0UtYzMt2b;OOU6NxwJ!6tw2TUFu5nc}2QpXAHo6J+41B5a- zV2;6{0C7-7pKFu)a7T}mdxlCm$B8&q`efut=?CJJ@ZpD1Akh4H-XvJCR!qgt@yZkD zB`Z&UCY~@I!+FWje-V1p?3?=2aFsMD7oj_2_H%klFV2^3(;v|bpu&Y400l4#6ph|Q zD&SAaI8vE^LU*(>mg6c%EAvm#qm_}VS==`H6~zWgM-pLIKfQQXnG1~P1u68-%wR4;NH8yg1-WWO8O`Xj1Chh}5i*kmNL^P=^%uAO%xB z(PcEdP*coqDcoz46Q3NeRu{x2hf4SUdPRCvnyb!AOrRFi2?^o6X=oY+n`uIiO5PNb zmI@m%9EsPlLHnlmJ^29L@|f4bw_xdNIc0-L-CjeWdSa*v3yb4eC?ngkA*>-HD_Sjt zCk7J|k{BomnC4g}J*{t6T@=qpri8*dA&}#u(;}>`J#*bxJ3A-)_VchbHPM)uJL6}n zHWAJFfxJtgo3j%>OiWC1YKEJwwVA1jy(Na-q1;3A&w`7N5MUj6Ka4YCMyymEd!qL9 zYQbgE#YI+&HY`F5RdMggn@~tR`Y9Xf0CM5bq0fBDDn|~5ddDV&NJWS@^&w(CFe_$gSY>L zEelTb?@_t0jfc{D^s?nxkE7b=s10RnR@&?}i6|~AGGPgyk(AKZ6kkz+{@H#8F}LYr zvQYZKwnIlMg1>#XWMMz;JV!=ShZhtUo9wrFf_ph3sV%-@a6(fXaQuQhz_#<>Q6|HN z7ey$iVtw9Y-MP=TjoR>6Id%&F-DSs=*Ir9;{tkt>lyjB)@gFI3yTVR(iQDO$lvy*o zCt*5@|EMQnibfr$eN*D)HHKN8n*GvC#4MuCZ@r_H6wSYD-dQWH<#g#QV!vDZf;i5c z37JkugybZ8%FASnr@g=6n;v}^y_F}xgxO8**8AM%RSL*l3DDKq}U+?14D*J zRHR>XbaHq15Aq7DO7`{jbOZyT<+%AWJ*?Chm8#5%P;2Ulb(Ez}K?Q-o2v$ky8V>gM z(g8Ogo_CC}3@Hj!YicWUGAu+#S2tfl9q3|}Jyc`bke8QXYSNc`->08TB0H1>6vbwY z%`f)#bLki8$ySmPIXOd0lfz+3tIgqBPgn6CDT%B~kIGF190E)@0eGubyIKnVXS1($qX5wO^RMsar08M`-<^fzy-s-8p&N_!R2~+w2u{ zYsL&r$;~h^DYmvnzba#2RSZ}9=;Y`+*2GVD3YuT6VkV>;1+J6m!IXt5SG$F%)uV<8 z?*2UQ>s9ebrY%QZ1dsmVDI}LzI@z5pvgkE>4EwOkwFb5v9kbWwRDOk+Oi=tg!~q{cbYxYMosovNbmYZ zr{tQmrtxXvx$g1qGb+j&Ld;n6p$oX(+h&DPqk!RqsZoGOrf(KC78ur=iIRT|`kC?2 z?I>n7iuSOFhUar?OR{eHSgYI1EZv=4&FxJ+F~-7}MEg2ZJ!VXorl$>%0#R6Ykj7H^ z3)n)cj-o@e9$(}@yrC-=w0KEB$C%`Wc=>v`+WGP^Cb^-m0shW*zVR!74c4(b1#TA|%Jb!Q0k61Ye$HFLa}9$7%K_dAuSe zrkm4)b#q|RnhbVPV1K(?adn)2z3TOLy(L$3BP_WN)MXWRS#m{iA;c_6x^mx*?YY?Y z=h;(|zg{;=O`|Y^L>lyulwL$@9N_^fZkcjYcyPb?DBXQtFVNtER$Q4c(}K@qw0o&m$qE5nXj@V}ZI{yKJ20;tFdo z+rAombN4t!gV$NF{k1!5#tK{=;e*2Uejp#QdVA$xKa@$m022qnfB^x_0)=zu)t_k(q zvb-F`!71dsB};x8tntr?@Z#r@s!yfI6>}we>$F`Wnb21&W{|UB3`m7&{S`YH-JqhI z+Kk>oSs1ljsyzN36G`pIVw8uUs~`mVc=@&S&=tE2LVquB??7r;5|r3!Itk)#8DV40 zr$uGJC_hj!og9;^7y!)@tKI$jySe%MySfJjy1V=MXg`OMiI=C&ma<=(mviio<>7t% zDn=%rS&A(srwzwIF>U%>Hbn$iekJ{5h zgr}D`&xiPVqIAQ2ecgCI(9h4qCHwZ_wa%e7^QY+i4q@p?l^u&CF#<Im`D#3{m%&|k#Z z!QozB;a(aDLm(XqM^2A%TjjUJxyUKg;mG8wxN9@DAF7h&$2!%T;FdwPLxkajGq>q# zD#Zw}XyV6vr4{XmZOPlj-i(p1NCRz|6VJW3Z*G+K;sZDoH-Mc#)pMYk$FzA*N}rQ= zr1u{{DgMS?l4Z;$)Uq)a+P`jWxb~~auM)pI)9|?W-q6nzzIuMbDG%<_mDW{FGuJNs z?YBkiX3kjGdPSBQme1m{_%|S#ds%gu$`42s&oB=|3`-YzY*1XbLNI#@E`so4wF+Wj zP>W0*yGlox)58{Q(3F@uH!0rbwYIZuI8=eT)G-qQu+bHQE;9tB){{!mWp2`Rw+(Ht zwJ~n{_me3!EZD6eDASM1Z_|Y3l~Z9iHqt@ym%ce-TZKR!`!8OEyf>KrhiAWgE;Ois)5*#@#HlK6QLz0p!W=J({uh_As41nIYD& z_PJ*zn+#vH$v{t*j+&Y}Vwo!g$F=T8tdOiiBQALe#n*^NyYh(i!^e zXFHH7)Lw~wdM}ADYN)odGG)_BJ)FIdZ}POVp7P+F^=q}?Jev9_|7~La>bkLMEoE&N z&x~z&Xz~H=E@?4U@<_*+Q%VJ3A&^o5FLMjXJvhk$6+ov~a!=;wbiqj)N9c(W3DQbt zp6o=Mj!wL6Sh#h)gpD+pO|a;*w#*_ZOD%>4aYM=Yp(CeNQwduV0VsoHO3*B!k}7c9 zwyLc%-MU=mfONNJZ_O^3>IuZe9do$jsvu=n6tNtF?xO63(|S|{bQ`5?V8!p{hfXDW ziInrFu8Ou>N5bjU3Ye(1j%1UCR2Q{m*b>%lgYZHTZin{R3wl(Q$k$b$c(j#MfZx;x{EpY%@b84@DB&iL0!DW%h9`PB|DEiDT7 zfvvf!oULPdWzhjuKVx;b7kCL}hi8$JnB}DGjaA1JEG?iFPVrkwY{_g}wvK7g-X-Vl zDjTvrN1or*gmSoFew_~h4@t`b>sP#I@WS+vVf>r0sZ$ML_A}?$V5M^a(FFmFDRTl#G$OvH(UEo4grClMuLx_v2RGKsk_|~Z>QsLPu zwIcR?bNc7OEPyMFNw{?1WONfj1WMK0|Du1`zlqKLz63y0M2{pHa8 zK<_X)$k{OuQKg)X&f=?Uv{yRW<;<|^+L+9`ih#tbD4d>Dn~+&II51=!IhYht5fvI5 zog5oq5f&O62VAuB8{Es>N{S~f*~$^zj1m2ZaaC9~K{x;4W?-WV3@wq!g9om3oF2vW z+5X_8^sFfMk@nqCuN=eJR>Wb;MCML3TDq^lr91)AhcOXQAr7LTl}@vD$;l=BppFk& z52X7u$s^zLM-=le7*Gl5x?sJOjF9L@mfJ6Bu?@F!GXr9rr*SMnO`5j{Y27-o99N{8Lwn|5*)#9-X@+SCf2RII1Y2z?Vf1qsRAM3t{{ zkhiGu=4|96XySKqSDAN&Uw{on7RI>HVV(>z(rc1JNA25=G*O*)lXSGU>a$a-08n(+ z#vqKdL&5c)9j2cFb`SM%R~m6QQDhh~dQW4?oiB{+t;kT{-tqyqgKeWUI&aYVVTjQw z2&meR^CI@H4V&L`a^BW<_ggNNZp+?Wz4e`HC#S=bw6*fs;f-4#IJNiKH%$vK-F4T; zEhrf(#gs%C%Ik%GC;=a%3N|hsootg~daG_K@my-qLmQqDln&h9eehcWJ*8>#kQ^m973*wgb1}4GtTg^mp>!yeMkU0@`Z)S>LWob% z$r;r++)}XuPg&zcM+cpTR%cZ-?x6?nu8U1K(Ts?VOV^l;V5JU*`=B^3Zs5SUxMIZq zM&MZs@tVF`I|V*;Yj@*yyyQaRv}iNa^7%PMQ)A3bE9MV~sL*Hz#83tL0D;eK&Pkt` zn$wh%IVq(jJ1V*W3iNz5M7q_{5llK5w>;>eg^PQreb=w2-e*2}{}h&|y~PIsiC?)7 z*<93)CAKuC9oJyQsu%%aR0nB6&1dYX=KV@flK|iDs`w9$3pe_8KQHfp?u>qy5Sc=mz zES8Qj8|Y~DHO7-pO$6X%f#PleeXUFw*aYSr^=Fj6M2I6ix9UAnr^JzVJiVu4DL^Z- z7fXSCRV`E?d#re>sp#Y1H^SQ2W0X`_GdgEctYt)1l3#$e29^Stm4@Mrol)#sN?OwP zoXlH8_{>;vT}B5n)$^?TOy2d_BirgshM10CcX0pe;cA#4C_rI;Ft(<2X7$*G4-@Xe zGtvhyt*FY&3G?+gz`qyp12BJJcX|cB0L?t#1$@bXjLq7%)7GV`N+Ix&A)s=%V^UQ1yA=-GYDrVlxqKGzh;x`o-%5~!#h znLjtmGGKU`Pk@!CRVe{#bMYqODNCxix$vP%5FLe~x49TUki-R!ACo2U?m>op*H-1O zE_-e0vWX*ty@K@FmCnF?o@e~2l?)XIBx71I&hE(CCQGxLHQP5#vYYQXe%acV!8Er9 zLymDHbLuN=794L`_uS&jocw6LJqK#(VdNEz51ns;qc8Rw=)`Y&QMt2*h$UV&QGFwk z$)Qq3LT&Qd4Qa{x3{srP_hk3);xu~YmNoU(^KEMvt(rTiJ5ITy^YW`#Pnf%FTv}|p z0Vk*;-{EoaJenuA257GaZeZT7VZR@cN4Y$sdI>&o!8&uebWD`?YQKll7%uHEU)>EB ze!sLsdJ~EOWuY0bv!$miCP5WpotaYjy07*}Mw_NYq~&XBOmn6dRE;H*Ycw%wk!7Z% z&GILwO&K{oE+Zo@Yhv=!dczCG!>?f%HCzZ)=o2cz?b6&c1BcF!vsi3inU&RKzSv^i zCLy$Bd3pZGk@@AzOMv_Hk{uH*r-P_7kv6lVK$l2ZhkcndiM7vAL%)H zIb57OEM!c3?c1gP(h~VjrOQ)0FLYN*>{CXhefxl|1FY=HHrv<%n-iq|WSq3+ixo&W4Keyo1~d+i(29nux0?qBR^L*4&5b1$tR6-n1rJJ8`(K9pH+|3ie;zU?WXG%rbpiYTQyCr2ibd)y$J-%|5 zYrso^8q3%)b1y3wmmI0+zQe!%npo^d_}!uiam&-%AGOc})7Hf6mp(Y?Fo^At+JXgJ zCsSw9TJ(3%E3?d&e)jQW$vDX&pxe_ka{OJ+eZQuib zf`*kGq>o`3rG&lOhQjQ@XQNWpd|v9OJtrSU0xxv*y?O?%{dVnhs;6b_z@k|qy??bI!UReE zq!KK&5H$_$bFR)oPu(l6l)jhef#rtot=qZ;!`fY?cWGzUTWuUaq26k|CSgFUw0OMr zT@xnMS=DRelUk*f6K}tL;(*k74NK?)$!sTY$O3Z!1Hvoq$sNa$v#+H(X^@;qU9PR9 zO{u+JIwJ?jlXZE;lWC8TOk^iTH~>NiXC5ed{oJ`@xC=Xn?ygk6d$%OxJS0ZDoqJb) zKV=N?f#@1ZhfBsE6IYOPC z5ElideOz=5)!#yG|5g7?Y$8U@vIxh0!!8qpAt9qn%&AojwZ$~1-O0n96fMay;*kWE zftXD;uu14?=|VMoTD=qFP#~p8f!AKnowXFs;*IXD2EiXB_Px`Q2ai2+e&_GelFlbi zNJ8fHOl%C3IV~ftZa{N-M0jo{H9=)nhd`~0>JM%*tzE|+I=}vR(ozSUIH7(2#bkD_hi)b%`_ZMusO!Uyl9DEfJ6c+N3v0xK>Qwl7+tnXij?f+eHa<|_P4dm zF!YpRTP!S{%qqM=6QiF3p|4dNso4BCM%#Sh~AUM{e4tlsS%M zl|yGxBy>$0B;7CHi-e#x&>qAzb&#u*-WD1$ETMoDRlTAw7r|>|9LG?sJ!tRIpXHn# zM0LuzR9hEcjntC%CqY4$cMz`Ak!X1HFsZRSXKP8*Gp5eBmTMUWi|c|$n^1C@X|k!*Y9 z-45v-3=k7T1u;7@AygPhMCmFC3rf5EufGC>GN?}9;G zyR+a1a0ffNZ-!=r-7cme8}3FRV{bU98lkthw^B2t@pg{_gN+xropA>>i3U&UT@sRf zVmh~-%)!xKgxGVhF&RRWVw~TpOCk1lnuH?nG2zLkcHC>XB~KpSFFR>sYBPpd^br*> zPCx{Wssv}K9L7e3)z}JlDs~lZg*J-I%q}cfO&>FR`b1S@-Sn{ys>X?n}-S+!BQADuLMdV^}`q=rcis&xuv0?9Q)Qz70^$kEhx>Ysb{&x86W-H(y_ zA5)#cvj*2=xCFFGqIwP&J(Yx2oyI2EZ^NCTUnE)e1w*0hazXE~6GJEl7tI$Bm^@&J zs1*uJ=dSk!nMde9rvzavQ3fQN83b8$w)3bc70EhXhygI`<$*Q0pUXOp`$t(1;r>*~ z4N975=!lPQZrNfJX7Bu<_J?5#@_y2WEYQa6_{?kRGEVVFQ(?>RqHjzv~+T z@noOQW^c&}&OVfLG3TM|>DhDeW|=dO{<9b6q~h`&!=Ayft{z|6%?AMdkTo zer{gh+(+`Ab6@QCKb04k`yBpw;_Wy5&*yH)eJ;-s`*7~3osg)T4%(Ld6}da{{|^5b z5&IEb_{wAQEOYng{oEPjcjE=3ngEe^?Gl9#_sZ$O_ZupesWvayTPrur#ybvIr;Tj7kcxj?()f5 zpZ=bIZ+T`JdDhJ*EAP4au)h~lKGq1v@9+7^XPXz_2$G!Ld{b@05LXsv zePwvR@pIN?gS)Jnb*1zDV$*TM``__2^ozd()}MhP-*@w|{`0ubPQH2U{(+DqXLYFh zHtJ@)e*paNf&9Dr-tVPxBR{$IMy@w@`_H$B8{=`kYa^vU{=%;ji+=vfuMsT5t^efx z-Q&osW|vSGd1#|^cb~2=xvNiabxg}eY;%c7aFln;`*Qb zdLu-8z5i%?kUwec(Px{Fc8}3W>8t;PzsCH3I#UoDcK+nfx4-ssEA#+ONj7R}hLnx* zGPo&Wjqo*g@9kY_KYIE|iI4WAJRfU(NcX1yhV%^Ma)}Hbrt}MSYpjr~n!roze)4$W z*-GD=3Uw~m-<7ls@V7ONkWV(&7>AOlbbZO*jURP=>E7g-jrF>>W=De`EgSh<<7D|t z<4l7aeadBxbBw>Qd-rc_HFlF*8<+H?59zmZMK^Y%9HlQB?=ZyG<=ODn^)#MKde)U( zrVB^gf&67xoLhO5-)dB9ts5=j*0|jO!`Qty9BEkTf+5BqKQSimH4a7Fg5f7$Y&>Lm z@9p_}c=Q(MzrnpHEp3=H3}r0q`^v4(&2n7h<2RE(S?O=RheSVmrgz`&KJv4TXS%y| z|1#vMyX%d=(8tbFAM~9!m1kmCIF#A>uCE)v_0}GGbH6c^f5x?EpS_h-jK?mtjZ3!@ z_h!S&XPdsD*Ug64g=_rg*24e!(DK>FAJHy_71~yBp2=Po(g7!TjgT z;_t##2Z!O4%9H4L6hT|0`h(vZb!} zpyR3kr|1ItnJAx{`!9MPh$SpV0Y1ee+|x!VjIfApoh3om|JnX8JV#$ z?P^Td8*u3@J*sQ^qq6uT{~N(JPS+Tp8-@L++#qv&fxFWUbD)0PX8-u^rRo; z@4KMM-!uzdU%IzjneXM9&1OC0-~4y8_3wxMbNV9S)c5z{{{6o4m1b|1syXo1AvbUA z8AtZgxfpCmZe*X-Gmi21b#^uF3^TLh`fC18+7g4JY5%Tz*sGvFC;1*K>R`xcBzF`SlvTLHHf_ky1;CjW}cZNn@W|dza7a!pq;@IG%j@#_ru*?Yn*b zkABG8W@rA7B4Cs_=JJ~)6=cDZ~3v=YV=o&-u(Z6beGT0_Wi5*maojV z{i`r^@n<{#J#cOY%53+Zq5eu)80aViOuz~@y#LQUXGi?cr2E%WHKry08l?aIv5jeo z|NXi7>p76WzRCIRuP1fu5J(c;HHK)W;;z!Yn_4L@(-ar%;-X?SA(}Akz@kce1lKcg z!CB_QS>aXjocaq*qXy@>3ol~D3iby}I)n?{P0}e`XK=la3!0y#ONh@2F9@%wRO*xJ z$5p&$oMr;H(E48UJvKBlPkJcnsiYT^-b}iX^hMH-Ngc^TvSqS!vUhR_O^tsXC*PeC z-&OBdA5ddo7L^?!y(qjSyezyzp#fH%rd~5%GeOg!nW$N!S*^KK^Qh)qK)*)!O-)9* zBx`VCyie|n3)FTp#+hWuEt9ixmEanNs}|Q}T+O%^;#!3Zvg_oXxRA4CPyw5L_?OV-Y2Wb9x{ULBj1sk;H zoz|;-GN~SC7}EPGdKcfNclCbden7b&qIZ=!4{2ypIv=3ir{SK9*e@iNBR0LCQtn^E zeIWM0rgzYDNdo~7zJN#X2bB9EdPf`bWxwKC-i`O+~WF@`x4$9qG zxw|O$waR@P?vL>u^j-Lf-o?Jky`OS-SMDCl-BYaz`U*K0 z(ayd>EBl5Il3M|N51_9CoNWG*bW){8KF0GtxYpyk2WcLW6KS6W^h7{URGIK=kmlpS z>Iqz@a9u*W@5-~#I%c75%u*jh-c1m$)?T#R_@airiqcpky^C35!_u5a;v1z`3F0D+VkjZukdI2SNYfYH~6>sxB2t@d;I(S2mD9;Mg9~1GxWsI`7ih{ z`LFn|`EU4d`S1D5{1v`Km?bO~?i0Qez7`pgRr;o@=!QP|zW9N70X=h{zHe$YrWk4S z@E?THc9d!vM%-PhBdTcCeb_NAoivh0)ljmUY*tm0t=NWo9LC%Os>w_c6Rn!Y#Hc)kl&1EZ8huKPYnCc{3#ZFT_!C7%u*!?|{i&VXi zQ#@W)y@8W4uBzVTIykNBEn0S}cX(?Y2>UMYfKlSSpcd4s_XLfgQN1si3+Ad11V?PX zdqId7;#D6Bi9(|4V*!()>Y|V*p>IZ zMT#_~5s@OLNYjXjlu}A*jEIp^L`ow?M2eJBO1YGBDIy|r5hF!PDMgADF(M*TiWvXD zway?~?fu_c_Gi|7v-acc$6D*`Gjqr(4drJr?~b(wOj^)J@HDE+OkTmPmEut}R#vTQfoZdS5wU$HeQIksuG z=}Mk$mTi_&Xj^Q1SaI1NwLPj7*_PRUtQ6ZGvpuG`ZBN>sQarYw+n!PUwwG)#DIwcm zY=2S0N+eeO&nY}TRvq2@c~W>*bZ>ZHcz(D&ye!-qULW2P-WlHaQA_KhpGe`Z8-It- zgfE1zMzly$BqNd)DT;)`$HQl$>;73AT^A{j9{TJQsfk|stQ1T9tQ6fG-5eQ1^7Eva zGrGe}k*4c(<2^DZGBbMYU#G~N=;`R`$im1HdjCW^AN^E{$>^1sE0!KT6j?!Q6}=K! z6WJKq7TFy+5IGV#895iZ6zPdtqbbpTNO@6DG#0IlHblooo1@dBtF6D$Lq3 z9P9kYR@YpQ>+^}(izR+E8|L_KjKb`v*}J_l8FR*5*IPtAC-nCmA}EPy@- zViRIzcvZ(nBUXGJuR-xx5yQUhIWY-4vscBP;*W|=v?@apF`iQ9Ds#o3m37MV;)HTk zIVyUTW6Im&L*;$hUrCcu8C9YbJ(QU2kdG?k<&Wf3$^^MV{zCbt{H5Hid`tdaEl|F# zx>T3)lp0V2$~twX`fcTD^i z_7r~9xm)X&soF*DqI6K4k!kwh^(!*nz(1KmQAOTtBpFGvui-Ep@)o17(O32}`s064 zzSRgDVcDM|j2vJLF^1r`GKU+(Wu{SU)XFTQ(P)&}6m4XVaoTuY7FvRqpd4svur$ab ziaD~_GS)IymRKfOCP=rX$rm@(`E~0k>sUG6dWZFH`EBb=Ym1y~z1MoL{I2yo)*s6G*2UJv@*!)7b*Wrz zU2a`2AF)1eeOz`}S6P25AGNNwu8}{o?y>HX%dH<+Kah{v>^8exK~YOSZmYIc%ayio z+Gfa~*k;;h%AeZqx7{y0DQd|lC~C>oww1P@$ThZ3Tc=!W`>pMF@+sSH+itnu_MGiG z`E%Rzw&&#r+kV@A`Hby=?SR~9>#}vpU)cU&`-9wMd)4-;{H5)f?U>wbJ8nBJf0gjZ zgg?qHu)6Qyg=-QqR8+$P$A}4H7|vMFh&r6Bo)eFY=V5=Fm75fYcu&cI6~0easEk+U zDQ(K_xMFxrnM~&}TroTeo16oS{D-oME%K1emW9eCoYjWn?_o$Oa=4rxj+u0^PH^5dG1l!0IOXmzf4vuzd=@uQooVUq1^Loft*jaD^IIMYE-_ima0|q zT{UkTwt5q+`B^oetXXY=HJ?y_puVR54Zm&-n^rs2cQjpnR7=y+)peQ^ zzaYMz>|6c0R-hHA8#I^ZQlHU^wPJOn=FvRrFX+swZqkBUgZfMDYuYUJMe+pouiDSG z-RgVVVeQXaZ|xLPAMyNo4R zXj$YxS}xpYfR+yr8KMm}h8m-^8sn7lx;CDCKx={zG-_Y9e8DnS`wDJs^d{>EHlcr& z&bay=WWo9*vS9sATb3yg^w>_hO$F|wF zS^u7Gt8J^kz_!h{P5(YTVW+;3JV9@>?X~UIAGEz_dr@z<9kd^?b?J+3M{Gy* zM{Iwx{Ymezov^*8KT6iG{}`6vf!73%yU&Q*#V?d}u^v`FS4m;}zK`vD0o!*6+xH5# z?}M=CLghSJw!D=sJHVD5WXrxyPKV7-V9TD!mOT@ey$*kqO+U&u{bRQ2$JnOVkxi=! zWYcOdvT5~ZvS~GkZQ9Q^9b}uXhfSYQSFkm2QQtuNE!*@Cw&~w#7R{!1X^C2I^$*%j zT8jEATlt@0lNsutwSKVm6Kv~$Wm`WDTOX*t&bIzH*t#Fr`DEqY+Gy>InxTD38>?Bh zaoSh41hRWAMVqZXh&m5z9a<^*fL2aEfYN)lecDuQzxI;$4ee#5Z)vZ?h0MC;hY^>-}}7 z_BeL`1==dzrMtAB>Lq%K)~S1Rul9r<(IeW=^q3yg*04uC36ILuo-%TbJndQbtG)26 z(b`MKmyFxA*Nm?lQ?=8^H;iv;XK=;+J?#x#buZH1G*%ibwe!X*W0m%{@r3b&cER|W zu||8xSZ}P?-ZeHF8?{Tu4r7P*9(Kg9YnRzS--mxr(XLshTJF~SSY}#g>Ni_jEG>Fp z_~%;v7Rx%z(|Xvl$+Ah0S~goY>oM|AeURm}bvQh&8veGy`Wx#u!)ATf`kc{=yv|4> zuQP6i*IhRHTmNBw-*8$#w0>x0lkXWh(Foe6+NK&Y?2cy`gW!!180EI_*cKRdwufv>jM4DDCycSQFE;KZ ze>0|#zZuiuZ##`|kdGOy5A9ce>l@cDV!YGIzCmw0pw!_S}gkeMWxF|2dL-vOnhD>Ym}A z?Oxzs>|X9(?cU(t`cX@J{YdUzB>#2tEwA6p)-%H+6 z-tqo%{&keRlg#(eFaNg|ywfqpJTrM`U8jHZeY1bFcfKF;b0&`;$=goyUnTD{l<4%X zhYij9KOJS2cZ$LLz3S6^NxlqUmaoVc@|F8)d}Dk~zA3&b@cK&M zOy3;eLf;bK3f~&vM&CByZr=gl5#LGQIo~B;&-Dnw@7E*2b+7%T-+bD_O`E;$rysY; z>(+F`I(WrDTIu+@d~7w$`ML5LGY~^8pt))AFGc-TA1g5qInB^cW8c|}GHv*aN>RCW zT$0AJ1Fu8=W3bfIc%AoO(U*r0{->#=vt)hAmXe(%`%1b>j+dOd-d@QvlRh*5XgHoq zE)d1qO0LG*Zlq{=WZCCQZY{FRJ;t5n&TwbBi`*f1xx2w0h=w+Ng$Ssb>|}JRbR2 zip~4yk7QBg`y`Dp!W1iEV^PVeSNtzoi zdegyYKApTyuPbulvy(RvUGZ5deEB~~-m=&=^+9ST`lusw`R`1qGshOfl(IoG#>wn#NkDdNkDN+@w`gopqzxkKhdg1gFaNg|d}AY{d=n$1%;cNun~HVMpyX>onrrgMrEgI*>t82dN3nUb6Q)Ct;-%<1Z zv!(A8N_0o&|8w$P_D}K)B!|DhzrgSFm-?&xqx|Fj37^tTif)Xo_fL-=i0<~!^3RWM zi>`=yK1YiEoB#eV^ENK_ZzX>_rRWl*h1XNe8tw5f^LIwiMK1W)`?vUaM)pNl`1jE{ z+~4It?my$d;J+Hs0!e|4KvtkA5DJtBY64>dO@S$anSnWhg@Gl36@fK@je%`}-GKvv zBY~5FbAd~N9(=u%66_bu3wnaFU}dl&I4;;6oEB^i&I`5$mj+h_*9A8RcLeta4+W0} zPe+c24@TAp&j+uBWGFF|9&(0Up+KlCR2>=}nh=^Cnh}~ES`bg~x^`hNp&G!gIrm!X4q2;kDsS;qBo)*d=sD zj>qQV>~;p{m8-EGhy_XTp)B}A==wf~-+2!cYrC=2i0m`>3fIr~)Q)+6y>Yhu=$vMr z&En@SI!}H2taSY>L+2LroWkdduIqb(kDnzd#z#+b{JtKm5pm5ZdOb!WUSd3);xX)# zaVat?9xHCd2#x}u;n(bSk?G`Tale8G#ZE`szzY69Ep!*jinSWi}k}R4?Y@$mi%ZOTAhY@o`?PD(%7oly4YsS)85!2MEldm4@a!= zRJ(V$_qq>_SU6%0B~JkVGt(OPF=F>=_j&h~xL$e^nHGA|k(?eEl|%W_o(Z1Go*ACm zo&}!8p5-Ivpas&GLC*11e=48ywt7~3Hh8v<*yh;F(_Z8sX=PUJ9)zwh) zjq;86O~U(h-z?w!>VDPzeC@tvzD~Ta$7>5|%;ekW>+&7DB2*U|8=6QO zyjrL%jxx7?T4+(IBeXKK7PP6lxo%2mJ6?N22SZ0gr|PCu&%Ag3&da7Pmy$;`Rw}-33qwp>F`0ymsKsVid2h9r4zs~Wz_V6<3 zJMmiI*b?5-*iyZhz8epB(f8cddmHD{H{FeMDb=sy@4-Wnne<(F{VM!j#ByOI6e*9? zM8-s#=o{w9oXEoZRU>6&Nn{0mBaHvKES~BvMYfG}MRwE6{Em2JAaV|`OLa*2uDG)9 zQq)QuO~DJ_853Vm^!;(&rI9Xthumy_i#&1yCHgLz-Wx`n3EwEkzgecgN6w~%di)+; zXTEKhM^yhysbIfD)kIy&gopl;CRx;d0;!0_pVrJ}LBvAZ+@$x+(B zvAb?zX#w?^Qe&jlhrdfpt4c>**OiWMtZUd+I;nJeV_j)i>Ao6!X?y7`@O-@5OP7^) zHr9dGlf02ix0LRrzfgK#X&3U3<8|gn{wE3TUl^QGdUbHdU~O;`WY*v;*u)Y_gNu+t zgUdlRgU1YR8a$e$KMasZM>ca zA2Dh0$(pHy&y_VmUMj0Biw*86vzDdQOr=!T&wK~vm3gjnJTF#O2|emHjxC#4Hg2e+ zthsC&xOHek+3^6m9Y8*~0IseDiQ!H<d-SoFATj}p;aVRWK?8T6jg*O z$}4Ir##A&_OsSYzF{ff7(vpf56>BOsR&1--U2&k|NX5yDa}}2=dWKntr3~wblsC*X zEH3xuJ3^-ih~D9)yhL;SF46hhoH+<~yiNmK3Zy7#!_@dz*!&eSpJABjd?ZfvBKZyTt z^wjX~;g_pKRYH}cs()2Mm9MI_s;X*K)%dDORnx0xRn4zzuUb~siT~eoOV!S*eN|mm z$E(g%U8uS`LK~4ZB4b3>h@ugp5#=LlMvNKJg#Qn8CU#ggw5z~gU`pLgGfE=HA#NZt zY{b5D;fN*JJ+8oOBW8T}hy#r)M;xh57;zGNxl1E@s;$*2)%|K6jZ>@huscFa)t>5D z@6BRbagN$Ba_h)lBXO1) zd3NN9#>b4>6t~*e7r0!(hx%#Dbml_t-_0(JIQ*cJktM|aiDjO!$kE?IMYg~PE z{WS8R`c~|0=GC_`V^7mKwSHawX2h|*^@r+@)t|0EUw@@>Zi8${WG{9$xEcZtWewHX zTTEz}+%N;ax}afk!}5mJ4I3J^(%GUdTrVkg~#oz)_`qQIVYQ{(nVI)fi=JcYCH(v_N3<)yfJ_p7AbWGKdh=22^2gBh5_U-ZN7axPp+O36+2E4+pN65a2n$Oc(hwx= zVlt${xps}lVwcw-Ej%u}Tm;7dP0W(|oXBJB%_Fj_mC$Iy0cqE025cfkEbfHFtUy}1 z4|cT*+?#uD6B=_3BxZ=^t&mn8OK)x=AwE|$cKqMUkO^F}x2RxE5u{zc4btHL*hCsQ zh5Km}VanBLq|zLhLMxRdGDxrNgtT#=Z9GOB_fzNglZ6Ypc=iU;D$-ffy4y8ck=|ku zWD@r?iD%NTz7LH>+ze?G#gKMo6J&2*U%NVnxi{oZaVw%#=q(x`6Zq)3sSs>e@z4N{ zLTnRWNaRD}|BPl%Wl2ZD{~yhIS_{d2wyUGT3EV3w=va1nfi+|ay?As&^aTqZ9gYC* z&9kcV{MdMnY;56nt)6qKt|sWXHW7hTdAtccE}h3>SIJH^?q@HaE4y$*W7lF_D~~w{ z*;|mbt6zlFxrLipCh@A>BsA#pf7C&0+0M1S02KTQnKTlH2kP`^)~D~j}aFjt>GU;n=F>p#$cAj0|& z^&g6ezF2=)L~-wZhZuzW?^lY!`YL^u7)E#Ci%Q&sze!Z#?)zVh2Hb(aQ#9(k^j%`K z{;d9-_@e&2zF&;R-T1GH@wf~Bw73&@-oGVg&>i>UUfgqkS=^_;kEd5=8&1P5<`@AZ zDB6uL7|mj_ai=j=bm9)4Z-}+V4C5a0l+j{*TRd&tYs?ZqH|E0Fo-q~}-xr&Vhm42B zX5(SwVeuAVhitW~h2w^>L!Mr1hoAF%Gt$4DjKUQc6R>&#(=$Tlj0a&M8%ypig z2MwNnK*>TplN6wPpr9G34@8+Fy$G{kj1~3qiduO^t-PXEtmPy;k8_uP7oM)c^Ag}` zSY4G@Hv_ACpUBZ?>$63Me!qS{c#b{?mhpi8fJoKnV(n9S?NhP#^TFTKzX$ySeF2`3 z`aZ0|%GO|IYp}u^7K<`G+4L~R@`(NjZ0^VU3LNWk{c%{*O4x_WmZ8Ejo)$@XLSsEF z5>IJ}L_DYQ3^W_{jnMo8c4K3^>BV+qVY|^`H_r-Pe-74@2Xz+{&7*BYJbb7`^B;zTMH=*y=yF~$>_IL~9 zdRu=Rd;wON&Q_QVD|{a#x&~Ve!WJ`yVPqNEA|Lmc<%pq1u8{{#zEL1DjY7i(9%u{{ zgYi69k%$?^MhP@-Sg@ZhI2RT?7SG7tX51$F8{>>GgU1`=!M7W?gC`ghM5*x=<0~R+ zG{M#-Y#mS7;n}b$qJ-{T6Mfn4%h~Rou=^I_!qX|=77s?KNk4U29aUi2rts%MdyUg`lj_w zag(*%+ARvLZ&}|0pSPX|zYQO|iG56GA5+=KH2Bz$*~e5_C9A|u>}f&vv_$r_685y4 zIObW{yHxfrJbMW5N@MSGvUioUcNMaCW#CD?9OROZMcBvObcZ1{G~9X3YdIM>UZmRgUQo${T}@ucqTQ?qBslTjup(Od0t{>J9gUP$3{wm_P>0?qqq5lO; zo~G-s>92vw-=uz8e;rI-m&#seXRp)Q>vGuZEbMhT>~*Q^b$0f;0qk|EA@QVGfPK%# zzL)(m-y6ujcMJQTjeW0-eQyx^o{fDkWZ}&vhP{g_f$*B5)xtdK$Shv${uL^uO67d9@v{b(8?ZY zum@V%10B|9tj`FA{m`&(vThOvdt(B7V=jARK6_&>dt*L(V;*~>!QQB|H>yJInQkl% zT@$oBc}9rP6(Z#;pi4wOk6ZeNE--DjoCs!3YhJU=YIr@s=~6)pljslOB2fqtp- z?|N5gTbSxT3fh8RtJwO<(57$#bSn%sv>}Wu6IqW|)`hy_v?jD0G#tDKBDnTA2Qf+!>)c@p98b zGetU&)XMu^gRim-fejY^(Uk^f0@AMC4r%Z%wzsGPYrIF-xJ+-}@Aei^=&f8=Eeg#AO$9B$3ad(9PgfA% z41sb$O=w#cgF)j%W1*9vv7tKXeg+yHsz6V3KKuv1xgUdEI@+ip%HbK_i}-G<3G@Pp zbB)0tzK2|Ro$dwIVWw4FQ;1+S^;G>LsFFwzfy$X;pi-s+)QALkAQxqV0n({8AP?sj zgNm5EpaLSX4U|J9X?C5g`w0lE2wfH^g9&W~9Yo51gX~OWK#5F2kd@1!A3?N)x=|1S zgq*-Nrc0nJOh49|0+;*~(B4YuIL^thfB0UYJ3wA*f$l7q`vr8T1Mpg9J#=hg%1a=& zF#0tFI*Hkh`0Y zF$%=?XW&{fu$^=U)n)rL9s#la8L+7U+n?46Lch^J8qr!J<#rHiVU#mKtC;Qstsqjj zfR-`62ONCcd$i-6d6wSgeET@7{`z%uH=)T!O$J|VoqfR=%^QSDrfg}`PX<+Lw@hWd;FdLwHW0R=vI3%r}Dpqj(eQf zfLfvJ^kPnx`#~$co6w%F-sNBJUnNwFAAMdyeYR97e%!-^NT`;pv&~%XmvWzfg?BD; z52D;MF4qUd)~VxfKU=5uix_VkW=pj;sxRW+f~~0WlJcH^zJEVv3+m8YZY3V>YssD_M(+({sm}-jy8pAXtW-axGrqNKhZzgTvgob zO=E+n_}O-OjZE9^hHfm^fW`Vp`^RBkXhdU3ryK^s!%_DkQ*Gb#L)MEZx& zl@n=~6&-iGRba+{q<-07hI~~!_falv$jpVUtu}KF%&}Qlfp7T{D`6Y(XFseKtx(+n zw@kextA~A(tbKx*xQfb^irg#k32=E3M*;9l>m( z1`)5NS|x<<5bN9^LX$$`A_yb&USvYd_MT^I24S60_t&5^ zAJHi$^v!#MNO>7_j7UuoBD{eK^Bi7Bq&0xnFx?62WWt<;S2E#f;pIf?M$poaXbF+L z1-f=7tZ#S`Qz4Evlj$kYbSBt=cWM~kqwWDsCY|~}AdCR>X9YDe4FHX2N&<~#f^~XF z6RGn+=q+dgsD_C4X4rXyWHof9OlZN&RxEv>0P7H`y==u=8K{VKI>z8FU|I;uA<|%B zUMG=y0K_e;?}0K%hy9xHI+&;xyEh#=8z_Zz;vfk924U{Jm}|^Eti)?@4O}mHr3(>L z^yvwEdb}EPvGQJIf?9e4o@-3!gz#M98V_hqo=dJh$bAF43#1eO0CjV@DCjH`*3xsD ziOQX1`X=Z&k=h@0lt^OFBYyrp-A2?jTbB2Hhy84?uN9T%(%HVOQ^|bR9-} z@F;T!g|lCL2PNQ*9=1Q_5ArcjLGf9%*8|EaJ^{L@&i53#a2}MsK?Sa*&|#16$#J!T z-T^sX^WwUGu2#?>=rUZ>K);uddmJ9R^11|Fx(nwN1-m3qimL(Z`&-$Gdm#@%cN99! z6$2?C;qu^UzgFkFuetKfy6!8kejr?bxG%YIH7icYweAb zu&2viP`CnA1U+p+gZ2}#S8||W8SHro&Jb*WbTv`3lXB^-R>Gb~S7W{^u7RjfvVll1 z1Fd5UA-9An3?2KHJ{CGB>%IbAJ5vm_h{=!KBqr1?F}=qpN=sUOF0?!nI`$qN-d!@C z%OMVxu=mjZw1mBf`iwmv^#w=s!b?G=$ZhbHgUUfQtn-1m_f*$w_8A^p!qHp9SW6H) zP@@vLONi9JLYK!n>OmG0Mp)9nIn5g$jOmwsaCd_2~Agg(0$ zGNDY-RU&17&xoSSMTm7uC3M^?{6dc?y2QC?ujm5T*aqrm8U#Wg(eilEX{NoPlT44m z>e`v^2DK9DoygtCy8A(Un9!r5T}(K~7wsTY&VaTRAzEl?rHE}v=>ox0a5P-^6m4d* zfHpF{4qDH3(et9UM6??zTFtq1j8#kuXa&=85L+18Ur`6^V4WpwFA_1LXhBh%Icr7c zT+wmoQ#rL1G?z$F|7H^@uY!1#3eCYx&c%_6rgJX(Ry39A6lgM$3VSYU=5nz1q9)e; z)N*I>6(7z$>XVkX*u*ki&!}rHZN=xx5T8j`d}eTGTz9Gr=Ph-ejuh`J#uX`AnO?lV_z>t1*7d~)iV>;R6_ydj2TQG> zcD-xhQSQrXOKb72vh=uaKkFzLCJ?Wd_8SndmX-kG)nek+G7KDT z+8~_ij1-J$%HS&4#y-8v^fL7hjALT)WE>+1U2~L1bP?s6qSWWNkvrZ;b3nSWr5&Jm zL8AwwS8^I`yn*|2E9%w|sSU`jDy7+|McoSDB+wnum6g(L+=jly1|@;Mirf&+DyrIy zTwiG0}G z0JMup-v(kE$M5!ax!_CC?F4OMdJeRSNc|ts1|ls0TE{x{s%USi54m5pv=(hG#a@7V zTC};e6qKxYnZAG#-CMMtJ}c0MK)06bcB9#2GM0IBmB1@#YtEKItknXYCo#E}nz8Ks;;OWf0Gr;RW%m8Py=3 zH6sk-Su@H&JZt#Iu?xE^nrEU;BDobdd5K7Q7sPQ*xs0Ax#STOFcT3yAQ?bn;=mwq$ zj|W|_+&Sv&R z;Ii06jPfb;en$+}Dc6A(MNh_c3+NMd`2c;I<7teG@hvzG?octD;Q0rJqa@Xl8F71``0fz564)E z5oLrCZPgUy`lyDAs8Ucug#NV-6w$?K1=jAm7TgcI8D4RP%l!qEO)w&;>@6`XOVk2LN@s~&>?)n zVbX!fHqdF**oXa>s?ssoPT>n}t^@c)M^#T)M!2@p_g3l|ORH;h%v1_WlTQ@#kj^dQ6m(NpxktjFGO*ObS2?Z&^?#~my2T9-DoebXeB5Yb+foGeBafd z>07ALH*^uXGePONE;lJ9G#T_w)J-ay2pWhQ30&iTlrusU|7Iaq7FB@$Y-ua(p-%(U zW0uy!tKlS2mt{oZWqj|!xh{K%R*TA=4N?3%0Xj|JTd5~4j~AXSgw?U`6d#w$ouKcn z)K@Kc79MkzfR-X69pN6}ykFSm>WkSujB*EksLQlJguZBhfNpQd33?fQ*&UkCbZ_C# zfrvRcci&sMeIWb;)LOVTGzs)LYHSXkXI)$2M&B~fk5O*@K$?RU$UPK13tEcft}I;5 zxvhn(LIqesoON*(id8j!09wX`Ggx5<5q@p9t8g**nObQJ?n2#%F)s`06K3sMHhi)dm1u|(mKJ8Q0zD9cC)C^jMQI_Vy$%WHEH|tjm zr}Uw<9BBy^PAVj8x*6pr_kpdU+#Q9@1F82^W8ACnD8WcCxR!pwELU(Py$96W$S=4Q+z!(8 zcM2|~9{>%t+)>a?pPYc23r@Lcyj1rz=UNO|a5(*NTz4oPkwMNidKVm^D|-1Q&_24N z$FErSF4#j?^z!SL%>}y(FsH2BL09zh8_3;8SM;2_yA_pG|;;AF{pb*f2CkeI((n*FevEs!{6jq(8_d-O~F2*V7dEH zoR<3F`wI3E1xx6&0#Iu~hkqJs3`UK`>5HIy471UehW;r(#W9wrVRn^H(3~`kjdioq zHh`AmxGib0V%E(_+Z3m1Y19hUn36`vSdDp}gm2sU&W3`C={ca4n8^ugm}ih$FfMH& zhjo>p3 zRrw{(g8}j*R22njdN+;gkwEwoY9){ucf(=tLyLNUvW%?ZZ~w7 z@~=V%yDunm%rGgx+ffY?`YZWo9kk}K`~1@mvWO%jKmVlP3W9&+A9rA`)C$WT`6u#^ zGBxKP&OZjC8b=&-HbBeyT@E{lj(aelY?pEmp26`Fgc2ck8B+z!pI8Xv;tD^xl8g)0P!Xo4sKpRu%#dYgbF)wlq zdcW3J18P9at5Z8b-$dP2o(Z7mP-8{v0?;1NvQ#QZ)R7tk{T?+Ir_!uVL5;T5lUOa> z6O+F%72{Bz(5K)R+SY*1SbFEr%EvfBiTN%0?V#tOn~~qjl$bv=e-_4u`+xFVQWMbL zTWDoQ>JsR%g89=@VD02l`BPG0?W~)Vf@3g^PuU29=jD%0*#W{GpZTM`)E?;?QsBv4 zt|ny_T7i}1SEb;%e2j|JqIhmu3hbH7#Zo54>xNP=yWEN|1+zxE1uIi9yL^lse1~AR z=S;~1{SOALtFt zsYppgD=!;r-nA6TB=fE$Uj(6*yh|w=Hr=UqrX2YMaUoxBNj3N_9qAByYFC1YNw zFL}q3;mKU?NOC9WnEpy$SMo~GDMRKROkR%T-fUR&_UEB4&OIaY4ip^#jkC1o?JGHr zazDUv_u$h$>Q#E)uH>1Z|H09AB$L%0f^J)KL7cWE=b$cjL3x|VleO=HHZY9?@l}k5 z-EbaX#c2CMos?@}Z zUe-<5K#1dc{cpO0a_3O4uYCd4MY;5wu;yfmc_}whw1rpXCEY~pwgTl6>=U5F9cp=o zeKhEYAZf2gZbWO!?Xj0Z_af-3JpkQPpv!g_Xe8(&_X>Bd<({{vqg)&4oE?^^{Tg(} zPL{X}bc%b`4mxqudFW2*{c?}xoqzuk!%iRq+r97UqFb5v5EQW4=E@Fi8Rp{pC zz`~S=p_`Yx0`zs6l`|)2K4=Mab8`_v#7?CrXLhaz+AsY%tvPc*zlUyC&IKH8EXuXe zah0W@8GPJ!(6n5{4P`rMN^Uu58E8^&2-F`mF}Dab88jg`3+?SyUd$PnbG?=Exkab*!`bk0@i?g2H>ag_%_H92QM-vCwR90wgluPSoJgMJ4p%NYgQ1B&HTft~|} za!NtJQj2mTIT#yLfJUUOh0c?MImK_U7v&V?U{00IpaSZZ%+|Yda>zQFoJ_d0E2kfk z{wOd110{HX!%AMeV(L^X^`y{pePCd9SuGfYv?DJJ+9 zt|h1(cKX?ei7dlGhgesS+yg|~HRyJ;4iU4@P9_&@eiG$LF}P+#b{)0q6ooy~;V$ZlnFgJv=nf_j*!6^_qF zAJ9bBQMn0BxVFz8$AofyGMEs3vTJxIv0B+xq|-NmDmZtL-j#lt3D@!20Ve9Rhl!3` z#FT?_JZlE7)U$a#@QcD-*?l<|bDzG0NP8DLUTLEM#A|6}B6rR8R`mFUhdz-WtBc`kEs;t$y)p1&teIA6!pLUXK+>YEO zS<9fKadc$S-r{f2EzRBw8b-OQXE6eJEZU!ipg$Rt# zP^z**If&@==|`52`l7u6T}e(}oLpJ(3hhC?%fVizb|AMO{2RRoWniZ16NjoS2Rzs$ zdln+F+5}x<))Ek{n;8#iHnc26DSXCQl_jz$@;9O7IXr9lTp;tJWhZJ7owuw6VSO{t zS!flA&RA$p#~JOJr!o;m)xLU0=7~JSUF=J`GV#d}+WQc>M>3CtevjOvd5F7owVip` zQb)NccgTX(XF6cPoYM7v=03|Q&<@LYGxuaq2VFtAT^7uNf_2N>k)08zZJFqwf_2Me z`@=q`D|4%H1v;Y52Ic@YMr3Zx!dU~KR&-^qH%`XeTWcIaU$B?TTy5-%_j#4E8-%!# zxx(0na#f&Z2Ids>N@j<#74%=wEykU4s&c|Op4papElvwFv6kG^`38*~`=-pfmcAhD zvomK~5S7)}P`A|>7pIxV7|_pGb5QDpiWRAa}j71$dfr21fR|<%4|U^ zWuSu0snlmsPUb|=3G~^iQxA@U`sr91`6wtuF9*>|JMhiFX+w6s2s(T^otdc9$h$#S zJp;Kv0O9FZ&=1iv9@xP+Xw9W|)ojJtqg{Xw$8cWNx}n>otOna1l~$)?Az0@bh%5#70m z3Gv>!i-?|0adJc_+i-5v$X=|VEx1=ymG^@-;r>i~uA)zIZg6e`eHXOOxe>>O*E;zu zN#B?_*K*7E8tu;2OuMX=&Q;DeDEDpXI$4*ZzUW+ud|XkXui6kdHgSjP!W! zLCz)Lr(Uhlrr@}L!ZDUPvC?uLbjzKqK)>N!td@EYXqkr9QYTv4oE>~z>`R=B`M8fj z*XG30=<{r6JL_(TZjo~{biW2IaJGRQ$X&>}*qJ-$GkpOxmq^==aLgY{z5-pfJODZZs+5>#iBF@P<@m0fKjU$hsy&&Kya-Itg;h3D7+b>L*8o za5d)4kQKnP&c0fch=g-wK&51ha>v@@tmcNXl&H za+WW#=0_~=BdNZ@@+?X5J(9{-m}jzVflT6-lgwW&5cjbDn=FTrRPZ$-WHZY}kiGbK z5A6I41NdKANiPDNyNqQa%XXGDcDwR(lJW_b9VFGam|tPOljZ#^UnePoB&C%#B_y?* zSWk1P36(WG8)^lye3o;+LsFsgX*}m?A_wdcG-D3og-jPokg1|Sq+N-Tlr#gWBAGZo z7Fvrm{zZs1);pLT{HqJeLPKsc|C&Tv{5ikpp|Q)SNs2?v>$yxd^GcHHmq{x3vYzKh zoMO%QA$#%Da&{$!**v<9a#gky#ZRmagG}SGrwM~KI!W`vQW8il2=?lMY*zpOY)kk3B<~;SkEgc8ktjB&*N3T2u>Aj z+k)6G|3XsZF)CjrRvoP28CL3um4_j1@l}2SoE)Fqo02$?R(eo{py%OaLE##ElAG(kN~6Et22mvjg_=Sr4r zhYDF{D$l%w?Zm-7NfR3MY5a6!8fzTvO%B%J|Di>0dR)(TtNsxhyGnhw3yKEG>`e~- z#TW-$rh{+0bMTs`vS+0V8oNXIsFw2Ikf~fNjqT7OXm0Tf9+1f#H&TU#Oa7H5=^cV* zBb7bEE+&v(Zih^c_hB&SlARsGyP6prWW|A&;&EWC|5|MdPr(_FwbRn zvm6DPz#eT^zsApgHJ$aVSik8Lo<`gHgA`kz=>>WcKI5(S3H`TXA@KQhTg&B za`17J*-Dd5YmwB?RE`*Stq<6)>?Nu4nI&#HH0KVs3J1@;gCnhj`;aWCWU3%9w##=| zzRi-`QO*#vr>6-!*uf=}c|PsZJXez5F3H-{c%&)jD3o=iSIKAX@>e7kj#o-w)?{+- z4@vTjrSrJ#lA?!QyuoECCfL>Qu$%{(%)gbCEGkK(oF^&Cv+N2*47)NFQWqLzvKS6& zmn9^{n~+K7IY?@xQGN=U5cjx|V4I+LWmk7X_7WWZl=n!ZXpqV5%XZ~S)=VU+Qv6Ic zEkXX0G^YLYlhA4GM`@hvV0OgUg8Vd@%dYV|PB&oKs+|80!$FA;S z$@ZomWj;(&bFoZi%|(_xGb)c!UC)}|vb;c2t6=^vOSU)lHDXCCYggG0)mNFHC#lg$ z?Hb##N|v7#k4`j8c69@5l9}1w)UOcBF(eh!O1Cl(V#yIzS;NeIQ|KtY_;q{+V-MB&iOC#4exNJQvVx%eW_+Gm}KQvW_K1OS`g< zIm+xKsj;2QFN2dgZrkOUEw{R9Kg(WB8!M6uZ!XVrwX!ayH>_! zrm!Z+OrC4u^GK>WcI89XSF@Z2nJRLL+L3#=K=@{26VPc5e1%B8FsNK)Zh zlDtMr18aIX_Y2HlV_C|Y@hr)1lg;BQG!J%#^l5w^Pm7Y=yGC6KAYL2Rxdmrskw)fBEv>*C-zlgwv?WZo}1 z*vB35UXj12any3ecQv$L4nDIxc%IYay9C0UVOP9mxNgZUX|jvjI@xOaS)b(Qq$y_`$FY*%kmr2k#I086C zA8>Nq>d02@lE*7aZiUi zgI&#J*^6a0WUu&M?j^9*9ECcHa|>D4vTP)&c}SX9wW1E};2o?Z?!S4g$%jl9vQU1X49hQ4o9%lI*OY%XxOkjSOWf#kfEdNMSvV}{YD`{g*Gs_7qzru13%llbQ zV`*o37t5(6H7E0AlJV<(>aBz4Cz&ImUGjLP;F2`bRMWcTH(7Hx%bQuUe~IrB3mR!} z^L%Q?O#Up$9@nsMnvqRVJh2NNwV-vki-|1fkyO0Q4=|5q*~XHOtGdC-@pWHFda;7# z_gV7Fis{7ix2)k~EOSXJ4>Geoh!>du%JKlqvn+SA`~Wf`zIzXoMg~}ZkEBL6Y?t#%3XVleC$Yjk zS4_>Ftht9}3rk*ch33kxEMU2klBWnl4o8eAF->{ zezF+Kau{jkH%Ka%SpJ12&$+UNnOjqML`pJi%y`116aP&b!F9z6aAG`C9bx^mEcdcB z`}{Vz7hezB1@*@o=gVA%{Y#j>!yY02PI~2skZGpX$=#%p6o>7Sk0SRFi?>+Pd7*ba zmOcsYB{bHsuZbqGU92Xl%w&F!<G;#+?^(sp~=PJD7l2>2(8<$}V zRA@f!Y6eNAKTGyLr5`i6ce=EzISXPo0uQOM$q>{bj=Y`ui zcMVH!hp!6lVmxc;EMr#|vwVm(%z~{>{D3uVOJWuC0+#o&d>k^J{auS+xrd=i=4(2; z;8BZdT#|iRjA8x~Np&D(s^AgH&8&B@j6tUHrvYi4>tJ@6`_K3ly&2`UQF)~pGS%D{ zDYc~0QXrG#S7G&}k?b$>c4i+*h1Wu5E0wvd$zf^s;djhSSWafibrtphvG*qMQB~*v z|J{bU6T-fVhzN+NC@$c_CMpSQQ^zmGl@T{2uBZgmTANbERwFL8HAO@vZl&r__cB(k z0wO{ZLReJXaYIGi*E0Y2^O+1JAz{(>_x*lg;UmOM z59TVIX~UZpww7t5^ZKSvW)xgbOsksp=oU-kd?Xf#(ivBp=FBA~mT8kfZRJKRjenrg zwNca3ze<$;`;S>TkUn^?SlE*m`DZI!XmqN@$Hhy;O-6UU!VinqYb~88#i>Sq)PkO7 z3FprWuhG^aqH5|aP*^SOJge{zqDm6%ZedsXIGS6WxVB3FofQi2Eo#njFH`stqx&oI zbfdS}=qpcut-@`@yT!iZt74{o4kz4Jb6wjSd$OIjKB#!>ucnU0w8{J);htw(og2kG zqkpQxDo>CZtqH}eeS%?xn`hw52AgSRY>_#O@ulbf>q`Gu zaX!p-q|d%;8XT+mdB$K*aiC&s)HK&QOUocr&Oa$$E$OS>{LdA0mna?fA65A8#$Y$t zOxHql9h*0ri!J>srFlRxR+1K)nVZRuYU#gVTLYVmn(GX!WzgGVJoS-RY+=8N7*!he zcyN|tMncyK6doZ)MCp(JwZhMdQ$&p~Kd`X(fv8dD%~n`_c0D-rEYV6RX9Q;!EOTM*mBr&RbhLs+XhjUZa=k zyPgWqGzLc~e5=BmH~g;@qm|LOvbA#hTa15$;@?pCTG9HYl_Smf-qY4oj2hQ1-ANX9 z*D3xrv5#WZR_+;!v372ivGCnzEGC$scv&J50=I+Ib4AmE3Z| zt#sX^rIU&O1u=mWz?QCN0Ow+bA22%7Sl7zwXvJ*lq!n-TUQ73C@mQmGhd5Qd(ij}5 zuyXSjfH>{%;ktMzlGwrJIS7=D5==#yiUBWXAYzgxh57I@e~EmhNAn7F}R5!5&8EW$_lH8^Si3mUsIpUV7*((bgixysGeC;-8J4^-fDq zW%I@={I#et<*v4{tFhp$Rrq{uJ;vz#UOXL!nU%%Ol%!bkrk^cqe%m_8!rpMjw1Vz_ zGn}EQMU!yCf6MzmyzAC`=S&IDw|CBzhTh@KaAs_5Sk=(BZN&|~zxlLnZQp9!;&95= z@7RL&#jyn)I_l4Y`5VdBVTXA+v$oo|!IC$SYF7Qa|L4BP#&*4-?ULBojs;C;@oGH7`vzml6bFYYnTk52^p0=mJzY#hh? zB5col8*SOpwph1@VjGQ1)Y+1VU1M)6Lnc(8+4}6@XI1!9--$L|!$eLH-j_3_`B&k6 zZSFViZ~23bZ`z#L*d(Ks7=B@*_kEk|HG#6$v|;Qn>b4^#B7IFu^i+S$tf75M)JBAw zI^N zoMWidG3`$7ygrkn)wFEd;X`xw$=Qc`!^}IjE_@?4@IUd+9@9Q;j*G3^@(!oO7GyNu z^u=_d+x$9fk*yE4Rs&0)-GiR0vwB#+*34$iVww3%t)88m)Q1{XHNDB0*ESoO*|Etu z)kxdCXEtrkXHBnWM&(wTKu!rAA+!CyklInoFUZaIx1-hULu<_1y7u9HUFNjgAl;at z5k#F-FUmyiEVnPLeTbWM;6r>{K<+kYXSL0aU@7=s>zJ9cXSSVhdcWmud)L&sDz+|X zisQy6rH>u`C&vvwOphLYyyHgpa-7j)`M!Y9F^=n%Iq}h#>vQ2~{;R=zg!mIlw;-Jw z-QRKDm{T(PG~)k6x&w%Rk?rK8{dy`sZ*&j-XXoN{>FCSZD=~VS<9W9^d7}^W=D@*h zJDGhB@yEe|ju*5bWq#0|{QEkddw05U^kDZMBinXi+dh%i>G`9(IG*zYd$bQfH}Z7G zu1>qry_@U?dp7A}VeU)3J3W4M7xso$(@{@2fmcfUzFs-(?^VDdk#}Lf6Ikj|8!h#y zjg)`fsIRooHBR7fOm`ghMbJ5&9JN6u>k;!PQAf5=uTgjT?^4>pf6r)j95t8YG)>PM z^_ABqednkjyiVy0Mtk0@^rF!rTgn_~)D!+hcyr`M@|+oYC!HE~zZ21x+1hfMwiGux zo-I!{Il<^e+}HSg#(hJ4Q+&%9eLX#I)FOVDe8I^ZHPdQ1>UgW+sD#yU)ETr9@jPBl zX-9r#r5!caN;_(-m3HJ09HnGb2}glW6KBtK;iz++e()ma5O|4m7`(zc9A4@4hu1hw z+*#~D$-T`;`y@tPPS158b#~B-7MVsX6^)ugURI*eN;Gnjm1yKrE78a*Yx9w7R4#tc z@opj2aP9>*TBkVhYHAL3IY{V=kJ^~1N|q`DBp|2dr>|}`HX)N z=ZO!CkBEtq5sFNGp!$ zWvw{;AZx`Dy{r{S^s-hQL9YI*>D2H;d`203U1Y2gW~@QR8hl$^C!#OJFrFaeY4{%LdrUPnw1q&R!CVPWgR|E`AxU74xeUa9X`#< zI(!<}*sgKn!zS4LoBuxTN|<&v{zaT8J}f>WJ}N#YerTj!A?*rjSLn2`ejm2P`W-SV zA-xajeOTs99=6tWfG|1$(Sc#@O$UZOVLCAEA=3fE=)kaUrUS#4m<|kE!kIEUI&8Yl z2>CylrVg8K^Fsc7(^taiD@0!*`U=rkh`vJf6>_xvXKV(^U*w^?!=~GD@|T(h51Vcp z44Lcle=t4fSnVRmq?7r5BWH>e#EIfL;!Wagae=s8tP-om)#7X7>*AZ@I`JK2laIuY z#SO;r0i)B-yNS}2dVdga7E8rHi?@oiMAk%7-XY#8vOW@Xw|I|uuXw-sppmC>V0ffB zRLm2n8l86T;`FTi9_|a`67j`!Nq!f1sklrm7c0a{ak(+bvGhSl@gQTko7ly+hP#Vh zjj=@*AA4TZ5o0={)86!R=uxyneCRW@g7IdtRQ$7et2j%%O}sw1>$nCN~{)Fi?4~Vi*JhS z#CMG7g7IT}QkHh?(E_4>m@zVU8LfwieqM&3c0g@qDAok~%tBd^5jGk&C)kHrl}v=%xY zO<#^2ZTbSy7l^(<^aY|X5PgB@3q)Ta`U24xh`vDd1)?tyeSzo;L|-8Ka^z^UV~;%6 zwC2dsrZo_)IdZgV4Mb~>9PMBJFtp*zM|;%y>* zM9l5t9path9I;HiTf9fS*J$H@$Rr!_L#Cid#GpsU?&85lG|1S-hz1!G;`v6EX#5d9 zD3NiNc$+v|yj{FQyi=SbmWiAPu@~n-@LnUOFm@HYiw7IgW=L6#ZH$z~m=MnwIqw?$ zl(UQJ$Ps~Q2c&Hw+Hpic-<2Nm1bt_mAWjs|5pNP_iwnf%VwG4ez9zmdzA3H~-!al_ z#*f7fM$@t*a_Bq4nQ{$2*UAMc7o=Q+&!t?+K{F|r5z7H$IY2B2h~)sW93Ykh#BzXG z4iL)$VlzN&28hi7u^AvX!=Raz%ZSYYu^9%r;~Fxq2j!YoJm@~NiXm1p#3~+ipIOBNA2O?W(0yhVLoDt=xn^+>$~6ml(0xvK zvycZ4H49|GBD0VO4mCRjVuwKN5QrTDu|pts2*eJ7*dY)*1Y(Ck>=1|@0G1rD*mAr?5q0*6@O1BaU3GH|Hb;RA=7^#ZY8Al3`SGKbhO1BaS@ zK5(ehof9Tn?)@4f%}-n4zbuFRyAAyEZ!>45^oc+=ShP-5ARUSo#Gr3+n)Hl z6~0HjSG-@u=4b1J3S)~8m_iLWmf8OUJDUAJu(jF$gt5y9OruV0wX=YM``B3kR_^tSzF}lha4ik0UP{$2*+|b%)UTjyf zvp7opjd+8RKcc$Y>C*oWTE$3@9$rG*Sa`1ZS8Md7HmAFEzHquT; z{xcVzB%Ume6BFVo;&|~?@ig&t@eJ`yae{c3k#;ivQCuwQxU`eSe<*GczY)I`zccbL zuSVMA@Y%E(VcN{7GSFrgRvBnB3xDb4^}mOfBOL22juL+(-e7b_aAnXrwA%5-Ktwf( zX(C3&P()XW&k>u7xneW1x!6K%DYg<@i*3ZVVmpy@PIBuYb`*CJJBjo+Tj_7Oo7hF% zUF<6EA$AkHi#^0W#h&6`VlQ!TaUXGCvA4LNxWCv(JU~29JV-oP>?`&Y4-pR)4-*d; z`-=m_f#M)>uy}+x#OSX!M%Rh&i1aNn?}_h=9~fi9MaCZCtBuaFj+ZVu6kicyAV$O{ zVpI&pn3y9r6?4UAVso*D*ivjIwiernZN+wCd$EJqQQSrBBbOV(~ZPH1W6MmEu*R%HvR5Sys)60?aQHW9=og0ujn1t2yN#3q8+M36p#^hw_(W)tR{U(@Q*l&*6g>Zt1T}T+a5L&6`m|X}Lh}eaMu?rz~A;d0(lnqigX#FzB z>_SLM`pq%B5MmcX?81I?!qZ_@c!p?4uL|wxRpD8pm7yvu6wel|JXL19^_yU}8>D~x zE-~8;Vqf-~V-{S$2|1j3_N#Jk@^($HJorKHb@2`HZE>CWj`*(lp7_4_fw*4$Nc>pb zAbuil6h9R|6F(Qf5Wh6~y~X{+{lz}w0pfvTU$LKfhsUvp{SXh|L1ASs)e) z#6p2sC=d$;Vxd4R6nLa~lz6l_RLm3qU<`gPo+@Io6LY$VwL@m29P zQTi50-vVh@AngjIU3en1+@wE&G$)YeL{!J<3fmg3Hio0cpNR#cdNw>x;p4><#1loe zVW<`h)k>jSDOAgZYLif93stt*2-_?6Yq40ICjM5uQk*WTFJj6$rZ$f~sQ9GOxzEPj zz6PQ z0pAhd72gwSb>eAtNUOt-#E-=d;wK`#!PZYj`h@W3B0WQxp4oSSjcK@_xWCv(JU~29 z>?<;+Ny(Uohl&Hlfg)p?c*Zn5LL4G8rinR9JX#zo=83fDz6)$jL&h{bO*~yZLp)QQ zAf6>Mrb)?|hKy;*n1<(y=ZWWw7l@O@3&o4Xi^a(zW14*#({QSIiFmzugLtEe4v>a1 z4bex44nTAOqDK%-fM^0l6Cm0O(N>67K(qp)6|hXaOGHD6K|>%K0vC#l#OK9jV!2o$ zR*K8TDsi>AMr2Hr4`Uj>BCZt~)5I{Qq4dqhG+}9%jcLNtF1w1mXvP0bED+V>Hr7dhyvSH5%vgu?0(+^2ZNw9!Hnb5> zSnXybp0G-8BYxinHsay0#bR-q_*?Nxak{9UvJp>u^@xpl!bu}nDCQw(HD|rYIs=?B z&dGecjCY>%d#A+7ckXl+IV1V4e5o^ue>7d;Om^ArAL z?{I&L{|E0F|9OA0cfP;cU+qouKkz^BF7!Y2KlCp0KlV3z7YDgPGjD3pB52`V8nh1D zc$Wq3f?d4JgU&%`?>E8j!5$uWA_cv?tAgG^Z~XWj9Q5@{f-%7uZ$?lOlz7(%GlHA& zhIe!D7w?b3qTmJZuffuw+)DY-FiHwSj^43Ov7WtX? zYUG&6G5GB~GctjH0GJq==)D;^CvpzH+Aoh>?!6tkB65YdE;21L&3h+uW#lUF-N@CE ztG)LlzmNRh`yg_Ci-iFA6$O7+^$kNDCZ)0S6q{{m= zQXQ$rSNZD5YVV84YmwKzFC%Y8-t@kXtc$F}7vj5-_q}iV2c?g_A0itg8~N{ruOr`k zo0@cJ(gClHhc+3@e=8i{v@q0!f)ulxDYbs}MY8~ePDjX4>=wBU<56|Hr@1};A z_;-ig!|vZ3UK3vH-ydEV&hQ@$Zw_zv=Y_X~xB7nzXNR}@kA-u>IsOyjUEy8+-@^yO z2mJYV$D^MLUkP9JpA6p&*ZEI}?}qRB&xK!xU-}DUUd;0s#deA9;y)kj66@kGj_nuQ z&wn8{JT}5#5*rm8&&C$` z-^8AeE%U#REsre^yqq;TYXU##qnwX|AZJ6)h9KfR<9?q$y~kpY`v|)~V)qetpKtdO zc7Mj(SMj~Y{lxvnKH>r5f#N~p!D3&rpLmFPsCbxoxY%DDAPy7#3AC5;!)yI zF;C1Fhl#_*5#mU3lsH=anOGqHTs%gMi^q!pC5{n)A&wP~6OR{95Kk0O5>FP#i3#x( zalCk{c$#>+c!qeUI6=HvoGccJQ^cv_CE}&xW#TWz%SG<;?7qnRTH$ZRZ^iG#OrQRs z@V~@OV%q3CqAPl$F9u>nY$8U*P>hK=VpB0!Y$i4rTZk>iR$^?U>>dx(3AJ;lAmUgF-OwAz=(`qJ1y8XHJs18HnZKpGoJV*_byaI{iNZv*LVAiWKww}JFFklqH;+dz67NN)q_ zZ6Li3ZdN|`iGLB7if@RYieHJ+hd@0Zsz#w|6sks{Y80wQp=uPWMxp8wYFvaG7oo;Q zsBsaFv7EzSh+{>SCoEO?&&HVgE~dVVU9XrwiMNQe#M{K#;vDfYqg&{3U)k>S9Dcup z{C)?y#|(0h8O#@liNnP);xELp;&I~f;tAr3;z{Dk;y5uOo+6GHPZduSPZ!S+&lD$! zXNeQVLh)?z9PwQ7Jn?+-0&$Xfp?Hyau{c>Q5~qmVM@LOB5ib>gDPAdFB~BNw7Jn!H zUc6SkQ=B7~iFb*2i`>^o{@m9G?-Tzb&K3VEa=#y2`Rx^Q*B?xZ^aJ6)i4Tbni;swp zijRqpi%*Du7w3y9@k#L?;#1<&;xpnu#b?Fm#0BC)agn%0d{JB~E*GoB72+y!wYWxn zNqk>iZ**ITEk!&H64P32BeoUWiyg#{;x1w*aaXakxSQBT++FM{?jd#)yNf-J(bf_IX#urQ+Yg<$5VN{ z>$L9-@p|zF@ka4+@d@$o;(Srs?n&D{X}c$F_ny&K>A5F8_nuRX^xRvh@bltg@da^- z_@cN}Tqc%_6=J2hQk4FCs})`&z9haZz9Oy_Up4y2iN}i*MYXDbw!&&#Uv2BFZGE+^ ze}T5Dh5ZW^eqa1R{MZ;Y6>~+6uR!A~(D({ED87@ZQ4?sq1R5`aMoJ(Z2-N$*-b%Tz z*xMLU4iV)L*;O&>qsR{yZt{jPs!utERoh0@wo$chRBanot45b= zFU>GfwQ6*|Vm=gA)940;KM^;IpNgM}pNn6LUx{Cf--+LgDs%K-3U3nA#?TR6(Gzn- zwQZ=j4Vx)OGeD>rAk>Hsk5K#&akw}_94U?xHO51Y@$fRm{8GGJ{FQix_&f1BQMwvR zmqO`MC|wF)(^j>7_@=^diSHR>EkyM~Y(Is^ipPm3iYJLDi>HdGi)V^wiPF;8MGBXR zcZtu68dEWishG+cTcUXNW6nnw&QX84e>S_L!(6jF;L+kxF;C1Fhl#WwX~u}Z5Ggw` zlpQj9;0Yr2AWS_V^?>8Vgm{WLUZj3(Jx!#RgwGJEE8z(uH6~1rA>$IBEuJHuE1oAZ zF4@Ysgp5nbxP-J5yjYwp7Ku~DsUliH8ngh?|L{ujDv|k{n5#v!fbj3dYsEXoIU>Em z*1N>J#e2ki#rwp+h;zliiua2Th!2WMah~`$@gebH@e%P+@iFmn@d@$o;(ReBJ}Le~ zd`f&;d`A4Ii0)8wbO$aF7mABSG>G^Y#ib(JL`;>qLR=-T7T1U`iSLW+jb?YimSQWh zwb({%E4CLqh#kdU#7^R_VrOwTv5UC7$k-zPJ;ZKecd>`Kr`S{6OY9}?E$$=kEA|!{ zqvW=~*hf4-JWxDHJXq{2_7e{k4-*d;`-=m_f#M)>u*fXMQHF?9#Y@CX#mhwLq}d&$ z{FPWN{zjZ8{#KOUn%zN4>8;rvgr&D;cM!ftl=hn4LCj2}*&R@=WOfH(wUXH#gw;xB zcMvWV&k@fRH3rS@V5>%-*&T#cPP01*tDI(c5LP+O?jWpkdMcaQAjGI_W`hvEPMjfL zFWw;DD3*zji%*Du7w3!8f3rl$NBVD;2;pZ$X}(z^#7O7O1|j^sxLABaTq3?GE)|!F z9ER;>+SI;#%=l@lEk9quC?ycyXer);4>Dc(u9NBZSrFW{(iQ zKvc_{JwnVy;>DuoJzw*lKSeRpG+&zLOVfO5nlDZBrD?u2&A&qF)tkP0(^qf$>P=t0 z>8m$=^`@`h^rdyaw9Z%0`s!Ipctx45s^+h|r5R1RiU5mwpEs_HP;tf~%k z&8mXR->fRaI*M6Uga?T)ic3Xli&<4{m7bVYMYu{_Ev^?o6jf8Rsz~#RxKaF6{7n2@ z{8Ic%{961@{9aU<&8i}wO=8+;RuyzbPs|b3wq{kawV9~-&8#Y728&0CL&V|Y2yvu1 zN<3XWLp)QQAf6>ktHVNt&lb-SrHy8baV%+LC~XX-jiIzLls1OatMEFd)R;9}jJ?*1 zuZpjU>Myg!*!q_Eo~U*+i>$-kn0nAGGQwlUVyFRR7J(;-)P^v%fz$?$6BFVo;&_o-vh_5Px)P=*AvK2dB&6PudP8~=(vy&$ zgy)LqiRX(Kh|EuHy-=jJ2s1xH<|oMf1eu@UR1u9Jj7C6o0bVIyB{I_xbG3*@5dOV* zt$3$6N2E{KdY5>&c#n9mc%S$eajy7R@qY0E@j)>u&J+J8J|sRYJ|aFUJ|;dcJ|X^H zoG+%tC&hn=Pl->9&xrpN(I85W2Ehg5LUECZ9ufbdxKu=^h^Z1+h^xfaBD%)bm&EtQ z^+vPHU`w%;*jj8OwiVlp9mI~}E@CHfSFy9Wo7hF%U1Wrj{~lsDvAftq+*9l+?j`mT z_ZIgN_Z54Kj8t;lU+g0uARZ_lBpxjG75j;Yiie4Zi~Yp`;y`hbI9Oz!<0wPKsp2K# zrQ&6xbkgiHQvOOT7Jnm76Mrj8Z_O?vrS#V9GQ!eZv&#sVh%=4csgl>cz$`MtY9F)6 z2&;X}A|tH!F^i0Fp?HpXuBg#x78zSL?#v=1tP+|JXcie^mB(x`!q?H0k?jd#)dx(3A8ee7=vhTj4wc8l(YvQ=v*JM0RJY76P zJX4$?o+Vlx3%IYzXmu>$z9ys9u^_NI76ewug23un5Lg`x0;^*|aFKYiI9V(br-)O< zOT_EN8^jw$8-oSh*JQj!EER1G7I0sa(MDnc_ca-9Bo=UAlks-ZMr8r_H5uoKW#V1p z-J*@q0`6-vE)*Au&x>4Pvl8-TK9a@>kn6(bA%wxDe%F;&Gyl=z{PB@kDVd93Gl}jxk#vUMBuhyj=X1 zc!l^o@jCGqSipT^#@V8DlKaFgMj9BJzKt<^ADTXlv1_mJHSrDcP4O*J8p3^Lmcx4% zV^%-xD()v*tqaWdC*JBcCT4XR6FW&fSv*y=Q9s6Pe`q6pjM@J1BJtO-AXY3+6MrjS zDNYw>imD@btXcXpIL6Kx;Dd@uiq9(MIdOrgy4%?UDW#Kk#(+iF(s#H`&T|47Iynl5 zVr$r1e&=J(Ai};pUz3(_GyUI>blSpZ@*VFg z=cv=1n3%$zL-h%F5%+|F(^O&GHzu#mp3{P`BR{sj(tFCm6`Sc_D_o};X`=dvA>RoI zb4;-(Y#AI3TRFy-fgKAk;;@-&(O#u(=j^BW-mtZDZml$}l}~FYSMgTb)=m@1AArNC zV=WL>8}b*ngt?IrHk0q;P@a4}<%3sq;ywA4Z>Cm_)f~mzzrB-be=eEOVl5S8$MV(k zq5co4eNBs5+d581(bBjoTcEAEig#5bPy0q|X!$o&pN38kg}cI-O5&*&_&_5jR4sx` z5AHXF{&0f9In)0;-eLKUy<-Qod@HuD>5|yG@Y(R$RPnRm0&oTcHDym#=ZKQ&!!Rqc0J-fDL9n_JCpHM>Qx-Fvi~+&Ra> z9oBYO+iLcfdhB>8w|VoH&8lLvYrmV`pMAHutnDQ&F3Y^zl(x96O{u-JFTRZ(AIrS^^VvW5p4xS2+Nt)pOjtVcWbNECPwhlY?*|d|r zW79I9KZ*{2*yzOWulrWq{M-ZVv$1zMi_4l$3a6|7)3YsOGeB%;ZgcxC-=g*Wlx{B$kGMC5CG1<$(7X9-IqhSM z8+$kXDBOg0)bw-PufpmbX+xFtCXTvO?`8+o>iuWYVsuyU{?^xMgv7MNLQQAHhIZ(t zmf{<6=&RoSjkNsGu$1>~KJSW!c2{uta`^Hd8$0z6=Z154?P}lb&7PUh&N=VKket!H z-_7Z-cOoaT%ek@Eu^UXfp4g%?e7WU5J#*P(Ugq6uNKP-_vvS8;c_}|7Wl#3AypT!C zuhzPjna}aR%09Vc>(1`6Rni8nC0piTdS3IZ&8nF>Q?q345nIzgH7zjT+QV9;`LUa4 zh)jEC_NkkCXw4Y28RgjqdbdXRGg@wXOiRzN8GYqV?>F7#XqKntzSwdwZL-g78(NEm zPu6QC+L^U;F)PxLwoT}zYHPt9bU&=NR|2Y=nt66yYY*Gc z-a2+cY-rAO^?bFJwpOV!HMa5%wU%dFL2mQh=9U8`qyL+B!h7wz7+Egmr5o6wlV6BwYmYM25%PTbwxF6XV!rFF1MTqtNlp z!>c`2jR#zg$g!`*!|nKVX>cjIT3j*v77^E+xEInT?7@@U>{TqkxJf)TB?-O6)*Q>l zLf%7^-EuyboQtsPI2v~|#dSnKTM9fl#xvV2-n?qyD;Ljec)%?r*FwkV==tV}S0(T% z`8aA^O1^9(RWYfGiAgy={vqS$5t6oe+wrq?;v3s61zuKDk*CrF_?8JBt8GE#FX^Jl z1L>mZTI4m#8u&aQ%ss-(iF=RZ!(yXE(-pb9)Pc@61?{Eyt;WwG9O9i zAxUZa;H%eKuQb3blsx;YwQAb5)QQlhN!qz6xDKIhgU{JB!uHL%M? z$hFQbjMtJTZ?LVYo14zHKJ)SHLA{G@%uw&r$Rfforu#%z&>mUsGv9<;^kqldqFIBb zIi%qd;k7DHrE0FWnM)5mh-}e#>ya%jxkblQQT$urDJ+F2+4_C>BspYefUWwIo=eel zDS9qN&!zb165E^BsG|itmBS9#mn$ATPn*XRw0YJWo9o_MW#~jXhBL2zgM6DV^`Af% zF{&<2Czv}-hx-KAZ~Bt?sE=m58T8mw%)bAW#wO5+5~np9Ul!n@B={AAFOmRHBmsU% z0=$p}=yz}xg6EL{zas%&M}lh*JdW6!5Z*?c+=1LlyR>bvOq0_|yrcB_QDti=Q46G zBj+-5E+gkMaxNq1GIA~>=Q46GBj+-5E+gkMaxNq1GCU%dk$V}rmyvrJxtEc9nQ~8( zdy?FfBYC9lEfSdP9rKwyLi`^h3is^rMercH0ggr4x1e z$(mZ*mS|LKcl}YbL*040!4<6jjIUXvvzwF7b(EyXIcuV=1oda(dNbaS*1g?1 zVv-{!IbxC{COKl#+u7OhN#^>D#glAo#iqVVR=6Z9T#^+osrv3NI}5)}=C#RuHkl_+ z^Vbx?x0HEoGJj2VES&{>$8VE)Z8D!tvOBSV&0CZCYBEnv*a)d6*E{2yFrv**lX+>X zZyA}FC1;NvYFXwGS~ul$1V(p?(Vg(Soa%}h@pYzZa&Pts9-5H;Le1z-g zUIb}^@M{5gYdKyF$w8VTxkxjlInn}YiL^pmBW;kjNIRrG(gEp+?1FScIwQLwU69?8 zuE-upH?G;gi@e9>{QZ#qkv_-)$brZ~6!O3wYS&j;Ijn8*mYZEg(6c0ZmPF5z=vfjy zOQL7nUR}1|?j?2Fy@)oRg4{#fZ>#+eGB2g(qtrZ~1 zDY6VHM=Fp?1W(&OezyG;2%fh6mB=ateb704J?k`kJ|8?szke?aNtSAOF5!!ii;>Cn zZ7s_+b{*euL^xBA&BY2IQfIc9o=%x>Nb?N2`3&qLRvWxS`uK+QSw+k*qw6i<+b@|7fX_=oy^O6{{)|-b!^N$#^-kWd4@N9&2-#i{VM|zaTqcq-3 zWG?bo1dZ@m`8`&CZw12Y@1KC2h@6C+jEqAP$SKHpwj&V;g5X(6Jl3kx3(YR zn$K@%TU|TCu4~q{Bk=XjnMYjq!~=YqO@W_p^YU#zzRkn8`S&*O-odU;UeFckjvS1b z-GQg?C}%ZcYorZgHpo$2ojn@K+9G)UHhUz-8H{=TjwSd!AN%BI|C>gJ|DUoOcVHA{ zZA*=&GnqLPtPcs+hXm_Gg7qQ6`jB9KNU%O6SRWFs4+++X1nWbB^&!FfP{{gF$of#| za@~=C%PDl1A>~K~Qi&`_s*;rsT7x*;rsT7x z*;rsT7x*;rsT7x*`{088DihPEAj(mZ9iF}28jeLWA zi+q&Gw$QoA2s$IgWUBaqe;#^Fx?DX{^dZn0NDW+G7>6IkClB8FX^h%Om zNzyAxdL>Eg7t;ELw0lf1cg|vPltzSs% z7t;ELte+{?&lKxt%6}Yr0-2Ab5YBP^e;`jGPa_MEg$VOB>uZYjHO2axVtq}qzNT1T zQ>?Ek*4Gs4Yl`(X#rm3JeNC~xrdVH7tgk87*A(k(iry@wH;d`PVtR1Pb+Lo^&b2Yt z*c5AQD)Kew6W<`;BHv+K)m$^9M~msvBt4p>N0anqAw5}0PZqKsr^54)^AWD`ur{Yy zn^UaKDc0r^dREuk=+#1cwU9MB#TuPrjZU#fr|8*YdbXIJEv9FSoj*9N=#GyBNCatu zM3E4QAy|P9XW0&C*$!vf4rkd;3#28|3gMcT!!<3ZErO-vv`0E1SdGpu2E&?CDFIiPV{Yrv#=;Eqj9Co5lQ97NIU#)#2+`JRl~8~O}ozEldHM0tNFwt zMVm^|rc$)2)LG|@pk+rQqma?a&yWJ-=g2Wg961*GFXS3$p>rLdGmz_%8;~23n~*;s zHzR*U{)F5@i`l$$h_lc=lJHT;(a2CF56MS{A(Yak3@&AGDT7NHT*}~52A49pl)m)*M^Xsa#N2-%Pa#hu&mdd_ai2w=Llz(lkwwV!$YNIA7my|C z1lM;LVr4DFvRa5`wGhi{A(qubEUSfBRtuSz6U@sA=H&$QZi0C?!MvMb-c2y?CYW~< z%)1HZ-30S)f_XQ=yqjR&O)&2!n0FJ*yZCcM__r+P-30S)f_XRLf5=GsM5Ac{qbU(I zcP8^sOl3hkKHGB++mX0k__Pt$Io+L=)qbxi4SKS@7xDY)JU5{^+J57Fj`aLK%3Pg@ zzQBn22>F5$;v)ePL7E^@Bt&9Jb`Eb#xE;cA!wyJCgky)D5JqaqY{)#H;2QTr=JW*f zc!GI6!91P_uVEWwB7BJvW+Sh?voMAZYV@H~G4x3z5S?LuPcXkHnBNo3?+Hd^f)SZu zL?#%K3Fm!0u&2D~{1(I+f`?VlIavwkWF_9)gx4YOAXxO?d&v992grKlBjjU*GYan$ zWFzt^@)`0u@&)oGt6gtoKV*NT4{`u;;G>j|)41J1+(&cuQV$XUolgmb{)Y~&n-)j42w4p^N7R_9<6av{Ph z9k5CVtkOXdG6mthH@F00g$}MqZa`S616JzbPY7pg0cUFgXKMjxYr(C^EaWz1HgY?1 z2XZIESzJ(t+=bkY+=JYUEI<|_ST+HcO|T3pM=Fp?WI0lWtVY%#FCi}@uOMp?&IE$j z5OgJ2hoB#LPfy`HJ%#7=RDdqvH9dvT^i-rXvIqAb^x(4>vM*uk&-L)42<2wo%4OZk zW!)OUx;21xYXDbQi@Ca5%+=LmuC5ky7CJk!IK755(bbN3R?-Ag!kFX7z1 zgmd>2&fQBmcP|NPYy7pR@YbHfS9=Oi?J4}U+wa-~SPKVmWw)3syTz=D1GutV%$40@ zJh-Ru-=1QP%w>(t#fy6iAMUAGF)|JLEpjC?9ijj5+MdE^dkT;3saO)>{D89@?hJAs zMFw(pZv=N74B&awYq`VcuiSAkhC6^>;Cb5>PPKD_o8xwOPI8C1L!Bb|ioML8>R#c@ z#aHZo&O`28_n&y3d>)UnYur^__kCBsV)w;YY(IB@?@&C(9*M`;!`-7i-tN)fP%jT3 zvgS87Uw&gJ$Zu?+{KlT+P4*_c=ixo}D))T6$KL4{d1d&Joq-S8XWW_Iv);4rU-2Hh z$i3fN?7iSV;4Sl(xk+z1{$%ILlk6jSl6}v8T%Kf~lqcDz-|W~T%>K@PrUbMY{HnLNy15ws85d)VbcCwZ9d>RlOh z3%YyLgFS=2z2C{t>~+CGK>;3Sj|t-Vll`w?y!SwGX>h5R3Vs><(t9%aRq!kCA3n?spBo3cm8Y z2j2$Y`aOd0gYW%4%?qvH)4b67y&|zl%-`F*(E9sCI!8MD`$qPR?CJN8>=o(d?-w~M za+cpGa&F{Y|A5G($Rz*3$mGam|DedFkxTu9BfpIN((j9J*&F?Sky|3O{6ivhB4z%- z$X_CJ{lSq3BJ=zqkw+ts`a>g6MxOHXBF{vg@rOsAi!AU*M4pc<@kg2eS^wve6_FMG zF_AYSZ}{=ZJCS$&V@nVfH)!7m-boP5uedw$Zl!iP2r7yZR^LH+HFia%Sj97(VEKEFZEP@FBa_|0H}Pe8c}Vd>c=)pUIQ# z7vTrtdjG5Nv+y(j8}7sX(*HJ0hiUw}ns3=3hKJelL2m4{*qK4I*m*Jj1u}MV?BbxEd6*5_n}^w;BOYe240bUO zvq9I`b+PM$ZuptKDd=u~W`n)V&up-d`I!w4jLnVB4f@8Cv1HIMwlKCRI3%_t_F`~Y zY-4O=&_5?9r)e-io@WQS5AbY%oG1I^Jl7xR>RX&EPH~58QO+=AI5GzL1u_;n4mln< z0XY#l2{{=Vha`|wknzZ=$Z5#w$Qj6)$OPmpWFk_CoQ<4=oQs@?oR3_9OhPV1EwM1R$P>ulk@-jp zc@p^t@)Ytk@(l7%UPP86%aJN%1+ofRjjTalLf%K#a~-q=(h_Nf zv_{$>ZIO1)IJZ5Y9gvR5E=VV2SEMtt8`1^Y9qEeffpkNWFKT- zq&Ko3vOm%XIRH5jIS4rz>5KG3>~4)i`8*6c9O;h?Kn5a%kip0i$Pi>IatU%Nav4G& z^0a!KXVv38bQE`ggG@tyi_o_`pC0Gw^f=F^$9Xb6&U5K;yamPGYmpLUCUPrnFbla2 znT_0z+=1MQ%t82Fm}|u2xJEpVYsBMthCS|4Z=PR|^YnV0XV>FAxgO`a^*B$h$Gvlq za}jisr`6*;s~+b`^|(j*cuGCaGwN}kP>=I`dYq@zTW>Y826+j28F>X+i@b_(XAXXZ;yiC1_a`E>vwt>1Tl184+^4O5+S};7kX^ao75(O#`#5~-#PO^X=i2+Y z0R0Z==YT%u`RRDDFVY)1m~C8>$FEKtuR8I_t_Xb<`N4_T+&vV>pH3WaI`L>%ggm3< z$@Tkjc+!dUWNw`2a^pOe8|RtaI8Wrpc^)^;)41^{Z4h0Le27r)=mz8yWFzt^@)`0u z@+I;W@-^}u@;$=wqyIuSA!&XKa1a;qkQ{_Iz!Od!KR9uIM;OQN2;)M=A--_pc*2SE zG;N$`Y2!7&E$Cj&oivv$ zZ&K&`PyVu^|4+^$n?YVBcM@#pHE!7?rKsI**us2SnW$PhK?oO21Xb9W$9%x@tv`E?vPLF^-ddWkpNE-0iacSGs%^ zXM4pBZwW$iMBybz~>~iS@xU?gZNUwWDRNyJz*4b-$DUPi?>XRp-~w)4r$YRZY&u zqtlnApQ|d)e$RfU=T}Y7hV8TVKDFbuEmc*rC9dkGEn)V%C-cqPtp1B*Y*oIU9XIL2 zmfcZ5-`d~P{C`kmFX~>a+>NEGA752VD>K^~YCqf3P@T8rwp{(KHt4ea8l_*e-0seD z(rcFgiKn4zUdv}m-wKz{NguoXKDAHNpX8k` zS^kWL?bT2}WVS3{l&u4eAE zebTd*ag}k|{pkw`_o$ke&8PNfrrb68OWDs$pAX;L5Q-Ht%ij$Gw>iz3R7rLuqQqrHd%;nby(Q0OYnhAd+9Ns37C26M?&dVTmMyU`=_>X4#xj0q=lh!F zUu{maIk0StooM@futNK2bKCOP*<|UD%9|+MW!YPs!`Y1ue5T8nVM8w4(7-0m6Tc1E_eXZ8|Dm6^{0}YHzoS@r zEAIbyWK%m2TaRjg|L+KAbuGIN)h@&Tg!0uc&klum;{3_hl%0E(_t?4gKQZ+*ZhWuv zD(_h5>rQ_u@3Pao{9ATtpxyGy_xZ_LcP$^Dey#jmr)&8|mRDv?ET9 z%G0%+dr-2ewtQNg>aZoM!BBSJbZN!24JO!`$eQ-ryvw*6JF|AXZQfXZ#pb%##5Fc< zu!kF1^Vun<{7<#7H<%5c{;K>Mr%So)=go&wU}d#^XgxMNB3YiXL2I`i`PJQLN57TN z%BD=;Q$EMS>Zf8p%eWrWrTji@n+La?BVM_b-F@r&SovzF&xX^~AsgP-r`2IoX?c~U zv2&gV%kasza>#5`d(A3;ZTl%n5wDv{`a%2EjotpY^!SRN>5__r($7_N()rifEiPB`YbHpsB_AiW26hqe^5MQ3;o+<)a!!^FB{D2O}eZiv^d+J^Oxop z*0?OLXlr49m(xpSP=0LZ%piOI=r|Q!ErFF!3G2b?u<8AZ-u0sOb#ob}v%k$3PQ_!k zL1p1Qa7SL*JS?rsZS8ea#bI^oZQE$KY<@Z~=ZhPu7`i!*x@FrnbAHZu=68FgXH^`N z4QFDuY|DOEx?0zU>crQZ&oxeJ>RLP9_QMrVX49mbEI(j#cx>epo5PKiA-nzG@tH2K z7~5Dr>HLb58{1~_wadM&bV^&N-L}+S+v>Gn`mGgfx0m0K-QMW+h##BbcK2J^si7o| zo!d8*_CFW5^)IiN zCf1JsclVQBFK_HsyWEbGv8(E)t=(?fEn8oA%3hnyrucv1vmsmH|Mq$}tlJgm*RHqh zyuymfwc{+Do?r29{kVp{r?Iy9y>0V#KkVeo8cJtzb@jKgRR3q<>s+h+KXXVYUA3}d z+o%Eir~Y|{+2IYf-;T~dvX`*CvakOa$3wR6KmN1&nTGeR);U#nTl%Z&tT~pqnS7YVdp`Dc|J-MQI$5a1jQr2z1>NSlYWn0@C-z&Q%U9#e# zZ1~^unfK#!;e3oUqyJs6AJcxeUTpitZ#MN_y6#q1QBp7ZKlHWYrtM|C{W@fFs6AhA zd!GK0*NwJU(3l+1|GFYb)-l zlV<+Yk?CQq?cRw)U;Pt+6uJPE$YJ*mz36RX=6J-_wUx+?8&m z-gWkK|ZBl372EH_Qe^B=Na(0*hWoSG_!&|m_FYf#mmuL5?JwMhC zXSdsD`m@Zp^qLhzGNHQP(>3{Vjs7F6cl~s=_ffdsyk~LNep{v6PDJf{ino(=^X9Gl zeeNsf&fla5RCY@rTe0Og+(zo& z*!WHtt*mwJwfd3l?zPL1jmv(f=T!z4&i=-m{k=bZY&J#aH`?sC%=XHh?DzVgo6iYq zHl#;awzTy$eG)RSa;tNDOIEwwn_jEzoXvAnc8#yf#%%l3+Hcz_w!O8ncCEda>e9eD zz|PnVjjiFekMsYBe|ih-tSy{5Vr8$Ne&qkHIn=HDk88gzb?5hvo?E^(9J1d;($B3t zsQ%WSS)a&b^Zg#mXGhnk9VlP@a{W8MZ&mIc%6#SK`&hEc8*0Dow(WiH$edR@j|R_k zYbV&5@aF4AJG0YIOS`l6t$(z1dFB4~;~M}9{O}Az~-*whnd#&I4t>4~z?cZ;obN0!d%RPtO^@ZgN*Xt+lr@{uz zwe4K&7w+ag7l8AX&;O+G6u(sqg5y?B4{r;<2Kmft-g(h(mbcwoOZY5X4KL08=k>{5 zJ~h9RC%@*@)LA+IsgZ(Xmna#2WDSi-6%;oO;D zSL*vUp1ndgJAK}#$vCI`Mk~`#>t9$!?&nqfE}!#J*yPS}+u1zlqcD-@@IAG{#1QxN z!};h-0s|4gs<<}NDrS|VJRW~!lje1p*fr^7nbAm zhsO%resZjLT(_#!DQ(Vlw|Z_)>%89s49m-X|NCUy_!DDj zkLyH9jN4Ylo-E~)JEyfHpZ;fCb*A_G!g}YmS6F_2cJt~6tvWe(uWi^SdGlU5_g}7d zp=ZIY;%TKRBfRG$1jntal)I^y_r9l)_x>I2s_AV>UV8eU?o-hBpW;_F4_()C`sK`D z>kz&^apatKKIJ@jBJXJ49I~@k)z4e+m3h9+318Q{#%3>D)kxQm2)90m>yxLqmWRDE zSKSttQ0VJf9%I1?=_)v3Rr^9^o#V#5-_tqA?tl2^|1JCF_5IG(@TZ(h=i2!X-ahgg3+zN&|Qox7^{udQeBn^paO?J<6;{QtRQ;4H6C zKh<7m*DrNKzU}h;e8Z}N`L_MD@+Vm!@9})%zCn&OeV-u5C;LeQ&?)@hx!3=U*QzA1 z)mi-I=M$erw_mGCUc3GLaemr!-SoIT*DW-D>VmV;<4n(WD_F*=C3&}n?*rzoFQ^=j zp9|W4_Vu}+>sD~DGwr9m=R?7gLR*6Gu38&hw2J6It&e^$Va2KxZ%bHC`1MS33fq46 zGSDF>=F-n4$aCK{_j#CmJZ}!U%Z25GBMMln3+t3;TgYB{Dhn#+c@9f6-N=$zB z=Tet@oD==xQ_K_joQ&5lJYM)Ucbo7W^4*r1QYPFMIdS29-q(C}$=ir>Fh?F^CU~1E zSN2}Lzq6GQYJF;oygH}WoVgTSo6;cMD!3%&>TuqB&GtUycG;aZOg9*KRsWlyxe`xc->CQ z-TJ5J^WNwDsgh3Ew-5oj?Zx@lIo!1H`%WqUbxo-&)yv@8wR zkzeTLxsJ^{S@&~txj)O3J71X0-1-#O?$2#!vcMU?|9B<=IopFNDJ#6}nf$l>#BnU8 z=2xFz-5+ADoSTx7-`s!Gx;iQAe$(;KVQ|*4jB{vyT2uV6^Y(Xz@+$jVpASFhcZKrK z*-Z6i@{kFPZ-m(7s<<$*C&z1WHe}Al?Ni;mW_dW{h z{=2O|(`U_ow-ftcQ#}9Y+MeqybFUW##gXSK^OtIWzUlKz&Fi<6^Go^VzvUc$FJu3& z$MnCK{g9uY!9J@y28XTg9^U5VgNs&o3D@!3%-a?&hufXYYwo_}TZa3RZxhZJ_KH63 z3f;~e3fulmZ_7QO+_F8pHRP7>-2?n7pIyi;ewt$DZ1ykY^BuKl5@+>YKVx zxa<}BcnQz{_#kz%gtvQDx&Fex+mNg3WY+s`Ma7fF0&{ui@@(n{1-A8M=4~IgpVz6d z?{gHkjjq3^YeV?nyEo{j|M>0I0-`DHu zza#nk(*IO#gPTt+|C*)XC#SCc{QhQ({KrdcoqzrLU+aB7sL*=|jndkl|Fbw34by(H z=e*>-w{tF9rleh1m<@3t|=Y?N>mL|z>9V^TiFKpXc4*82W3+_CJXSx?p8?6;6o;L26P=xyq z6Z4f*JZ(z8^@UXyPn&&WbMdrkx$-kl{p;aXxjyjiJh5Mm`&Vt*R=&mf)W* zH~7)1_eRe~(ogAgHgf-4ZGP?h@xi%i**|A4-fxipoc@25zQLL45&6wExFS7}U;Qs% z7y1p8klft)Q%Cwe`P`>!o~v|AdbwPs-+eZCPgzH|63bj@wI-(vj#!kEGB>D~Ve!~8kK&$K{rb9%3zDdH!0|0TX- zdE#>?Wzq-zWJSD=1@sRtN+0&Kg0R0!@qt7d<^My{AiA_2LNR3w(v zl?zA%Rx>X(W|BsW=$hNS)XXA{k&?RJAdQp&sZSKCp<6~taqVw zEG;uOM_O5LtNEke>gicRA2}iylP@mb9*Na8k$R5B^&XGZ@}u=ukp$WVjLM{@-WQIx zq0+}N#}xHG-8Y~@_OIf#SKdAfs5^9eJE>;dvVeT)T+P_Ih5O=DdjvcH$|M;ayGK8ngu!7)pHM6*XnxDO3`|DZjR->BOzdV zw>|>ua=rCdoo_)JnbXG8MD#f;Asbp>P!lbl%!r(|BD9itc8dtj*J?}N)<ZH>-3xDTpseY8jGv)(U?tv;<` zbWXju#)^1nicecCU>+s!if7ws@hpiYNFBYWe7Yr{CZ#5zyt6W(?jeoRYgdfkg8_A) zuJ?|#fa*x<*LIGUCgfxE$`+wz1}yI!@#|wFN?YF1bM)(T*~vNIJ^T2@+rM~fq|th{ zbVKW~@>UPgvx^q*ytB1rQ(MEY_4I3N__Rz{&%)3a@XIB%FRIUufbwkamrA4o>s3;p z9xF!czSI`R;HG3 z>iSqMS+q4u=p)~svkG~uilxsfrO!3?1~o-=O+?OE?Wi$Dk&2!X=OTGmUu#s(aqb;O zzMNwwQE&C^8L+l{sj8%FywsngiDwgLtW)o}bwrUGdc+c1rkk_2>!VuTz}A3y%u6kA zg%Q1yN9&e;y;dNb`e=Qm`St2bTWX?0voJHsyLz2;wUq+qZr$EnKcn?A@9Oo+udS-I z4gGrWDDl<}GiNNXy?TTSM!KmoFlURP?y9y1lE9b5~0aD3h%L zb(NP|uaUa?SPEF)ksnarxUSYGph7E>ccqTXSSYcvuQZ=DXPC{+nP#fl&rCPJHvevJGIyA7n7ho~=DX$|^9S=i^RQLboNHCL zt~2*o&8-(L*Lt7-+pG_*Io9pgTx*SWr?sB{N3E}|?N&c)r?tl#Z0+NJn2JzEtY=hF zRmOT=l~d)cS5yU6)f%sAsG8Or>QdFl`lq^0-DZ8J9#@ZB%hVI<3F~u}q@J>ttEbgy zD@Bb_pI95!VwI_4)L!+ys;_=fM^z(rT>YpT+c9=PHMNV`#nkn7NxP(KZkMsks~hYD z`vP^7eUV*7-E3F0YpPrA8|)j@9rlg(jjEk}lYNuA({63IR_*P3?H=kbyRY3>-ETi; z4^-XkBs)p6?|X$HrX|NTilE7x^7LkpZ%EI-yLW#c89so+RNPM+>!PQcZ@s6 zUg^H|L#t<)7>fVG<$2~{MG%{9ot{iU(<2?&HXnxKL3sW z8y&yDt-qra;qTwnj2;$P_h!nxMJ%D>9F z*}vbv-)S8Y5fSO!60tU7t#fNcW<;iQTg2Xoz0U1o8oM~(61h?|mgcNxH^{A=o43hh z2Utb$+P?$@*L-<;{04f{!`A$$ud$$*+>tYSR--9T4S5U7`u&3(tXB0(*4FUDPtTrj$@T9vy3=qS+lHIW;wGW`3ubp#V{{2 ztC3bWn~2G(a~)|jvzgTORv+@+%x+TMe87BwHr>tcw!Uiqo%}@eAJk7WCzF5O zd|ld`Q_LyUykY*6nm5h2q?|d;e4F;u&FR#DbbscGQtF1&?cUbMjU=_Vzd{#H>Ve-ALzlm%0wfag&*45GE zU$kDNJu9z7u=36({g*XIE@1urH|YZF6N$GLS?ScTvDS#~t;pnAk)^D)!`g|=UDhtC zY3;Uli(-Y|Lk(**`R}an~Emf@I2vo&Yaq3H` z5~5T|Rg#)gsuaCStJ2hzQDvwptIA3dR{09#6I25E3sgn&7pe=%S5g;|udFH~=VEm+ z`6{XkeXFXfzlj)Y7Aq&q+X(atokSQZ>qPbnWm;mC-t^kDADQ@^@&8NMQV|>;>z?XZI-H~wEs+f zCe_q3^|{2W2GgYP(<@&dm{64i0Is4UqwtlC+L&5=dfc*FBdn|BJaqOu>>M-?3 z)G^v0SI5ydOJ&I|Tr+=^y53b&F7mFIQk?5$S<-TLIcaE@x68|QTssq_xqX3sfm~%* zvMW)4k$sVLwkz9}rHx(1u0l;!yDBZK+0|%S-L6i{8g>ooV%M~5N>jUq-9j$5TiPv2 zTiLBhxzawePdLlE2@+UrO8E>~7>Au)9(snuza@3-Y4$X! zVZUupms<8a_B+(fuxC*7u02z#cvpS$T=k_gSN%^U!Cqu9Lc(HuG3`IKKb1s#sl8Mh z+n?E=kzZynBmafHl3A~^Q)s!`Ud>pkb}IQaJDvJ9_8Ri*?61jhv^SB@urp}0+1^Zk zi~SAtTkUVjZ?|_y1AC{vOD?l_+q-F#X=h3^dyl;b3+%J^k>79cr_CYz5N!_I$E3D> z+&<2p%CfVhxMMn|+~x2mmGDA*a*5+Q5#%GCNNMXtIZ@=JooMP~oEW*&DdH574o<)+ zN=>X&OfJP=iK9(%r#Sf%P6_fQos#5BIi=)Yr?gXAx;kZ@vT}n{&M7DNJLR47(#?r? z;;FgVsUp`nmpYfql}=r!E@?gIGP%~N@6?yuoXefo!qwSnE2W%v&Yg0#bC+`$`MaIF z?cAw42jS?(jSzxz)*VGMHDUvsVJnK4-tgdj62q#2+e4TFzHqF7n0uDoSyD zqxy1>=Npm7Hxh+!v_cHum%cBhm~W*oh5Tw?s#v}>Uz$Yu(tYXT^R4l%q5do1S5n!x z&bM9?d>eclq>k@v-`7&Zx6!we{3hQf@>_i0NEP2!-%hc8yL`LpmFdf5ls&%h$shC` zB>#i&F!>|Cqf*j$%r(VuEtd$~jc}vM$GAnMm>cUBlOk@MTTc9LdAGb&aO2$sspwwd zR+KpRLiZwxaVxtQOBp=5I^wu>-Fo!B%)L@fw~^aOs=HUYSCPNk#b|m0+6XUlKZ&>kv7a7M*58VjFfbTyTh5?v+lEK z{+#<9^&{Pp)Zl9}*30h8j5W@Eh5W1TtK|Rg{#{DA6WxE1pX9zqezH56{1kTz=^O4F zQp=s{{!=QuZ@O@EJHwqp{Y-Zz>0*~_y63Yp;(B)@`AzO7 zES%wPA^(lbI_!Sye#=a^x!aiOc6U4VJKbGU#m#gx<$8CoyO$InS6X=9t>D}GiLw19 z{3WCuK5iK?{bl{-#OL|A)Z^WfHupCd+kb=q2C+OJm#uC6cTnHn-+}aQ|J|be9sM1V z*2&*V;{5me@1_1P{=blR@pqx`eg6B%-|z26`k?5~N2mOt5_%;F z+WsZ}&!mEXnSUAi&;4JZ3H~>E{BKF{{BJ4k`QPGt{>+{2sh>AHF$WBfj~wQV)-OEYD|ong5dVwoK>V z#5?>)%MAW)?W331-g$ZLmzT#MUoYjgM_yk0;^nm$?&5`)l2S%#qawcYg~o;W&58Km ze*EvIVj0&N*U4Q*GvjaM`{I%Nv`1bIk9;6CPvVzXGCaS$r1r~i(0+MI{PJ;l+pplA z7uVi-9NzgqB^tl{O>vF4jJKqNG0pf7`Gv+3scn3UzwT;(y@vMJtKhGvN?9Wvue~f@ z`#y=%p8AExaeQ=B`{*_C(c{Tiz*Db@r%wFf-6xPr+#zTroy;cq>*c)r1f-s)Zg}?z zhk!B@wFe_j0Acdw4`{wDdi@Y)^iwfnW# zUR`_be!TX7v2_mqdI{~XoA~Ps$S=lgucWucZL(O$cW*S;AIx8S>9 zfbYH?op+c!Xt@hN-qC)%$^DOm)c=4t?>7&d$FR(C^SC@{X5rZz+Oxk%d-evNeI2~; zy7>1O;osMjO4em~`4LtFs{tPW71kBhG_)F$_x%0{?e}v}6TiPWet&y$th=ncXw$*! zK>lv)ZanplL;?94(iM|=MUe*a5&$)3k=YmeVhp2z=J z?eQ1Y9)Bdh{$lRpEy351(7t|s?dw;@*Z+bUtxzlQ+P_p^5?!oRE2&w9-(Ld1Kb3qM z-oJzQzlQu;wU+!>>MQc=)H?F(715g7pf*tRwfdTxjl>D2juRquoM7lUp)zs8H&S11 zB~CDj6SmQEyV^m1r`k#VE~16Yb+q8u(Skv=u%DXmh!6}NA(-4h`+=H6>JT-Di4$rP zCmf;gQFW9WFItGy(L!e(EhG{x#K`@25n_g$ygP09uU_Qvkd7Q~)R9AT9XV9hk;8R5 za%in%hKF^`aGiZIF++15EnKIgh1NP+ctl4F-Mu?=a;bM`j+8rdq+YZTtD}WtI$DU; z(L#cb7Gia@P)tV)u{v5v&~ZXD9Vgtb#;S{)}e(Q(4H zI!n!T1e2* zLL(h5G|u}I zXrp6>TXoD(OvendI%Y`FF+;458H(wcAy&r>#dORNt7C=)9W%u0n4uUk!&YqT#SBe# z%+OfJ3<)}BXs2U_hB{_wr(=euI%a6BV}>hr%urRw3_XYwT)E8gJAP^9L=Y`prlW=X zbhK~_aY7M%_ka_?%P&ft(25A5IBmSx;1(Sl^whCIFC82F%_-xQk-j=YXs;uLemX*U zOh*XqoeE9`dEC1jihs}DQ0eX64V9MO-B405T4<}Ig)TZ;h|$qPY3B~-4!K#!45gj+ zPJ8lRyxZ@F)|RPe1JekiHqhdAPgl~RKEVYQgPRHBFo9Yq9)BG!nb zd{t)^SB;9alu^xT2hnEBwS2{p2Ed z0MUh`qYIyoF3RiZ!qU-2SshncI<7EuTw&_C!f?kESC~4kD5K*FTgMe;bX;NUxWd+P z#T7cPxJ1VlwRBu@iH<9p>9`_R#}&1>3%-asdU3_|I)0b&#~yLsopt4#UUP}2p z5P?+F5lFm_K%#X7Qr6$u-x--+1QJgKav!k-cj6hhhrfqhtz(aP9eYIU*dt0u9c6XY zVdkk}&`-3Ivw6BP_0_9&ra4@1WuO2;0NI`%Mh>=B`3k4XQE{uiYt z_wrw&mluZ^Iu0>)9AfzYK^$W0IHa;Q~eCLUVDAU-5T+xSdATvY|S0qrZ$ zK7n$=0FjD}Td6&a;6p7Rh+I4zmWotypedktmAPi++3*omETh9@> zB?i#{meGJc+mLTF5wb*X?Etfc3p+G{5x^X8$7Z*;fN_u_(l!>_z-XX+2jx2`-!T#x zuN`BwYXy|sQNA+@t_F0tbBRcMlM9U1X$-6rxiV;$if7n6V$332g6!j(xhrR9GwWw+LXY zzLfiJ6?wE0bb?8cA<{1a+QS%FA<}=B$P*$ma5yX%d6M}&iB3xp6UfNVZX@G+CX_I<)M_Hru=jdp!_uDVU&mUgc-0`pD`PpM4BhX<)f0zrDpQ{H+FiT{lg4Qq|80-0FFaZvTjKZFy%K>AJ zo&*^pFUCSs7$Wi#a>m{UV<1&z9Bp6OCNjPhbP}1+QDh?Zlj7hu7zHc&07d{>1LfD2 ziA<*7WX`qM7m2*lL1gM|k+-nrTiEg~Z28twk!db4?li`oHea|I1&J^k=r@Bg-eted ztOvbe7GRtAu+4ke;yuoR_cn;kIwJD^P>~Oq?}tsGkI3w?B6BK2XQ17jEh7I%`Tr>Y zALaj}JU0QD)7(jrA@Wf?pzBA-`DmTUJmxTu@;qeCV~!tlEPdPpI5s{;zxfK9!VsYR zZ_5A1z6%D4{HG0!0s4KiP-GF?7O@`|4Fl}67+WlEF0uqYKCJ{@fw_LVO=M{q$QD^v z1-ip@VBViI@6USycKDp~^4fr1zF@vzya`z%D;R$T`mNX~@+I59OajJVX#(S|oB+&i z)eu-BlHvgMDfCI9yqb2Y?Eo93uHZS)i6U#!eXaL80Op8%(49=(cVF zWQnYA0O+-TCS;0ir~+MJA~62fB>)?JJrc0TM)cU&1O~!}}|?EfEI543Qn9V5P{e43RxUfq7sa+1pTLKX(3Zo5=TUJ2VIu05T4v-(mDSjDCj? ziX5p7?L>~1gAO8D%stqWPX{#M9sut~t8uVS44e0T?ba{?_)LR?>*Lmkequ!OK6z9? zjOZg`#MBWZ&;a_tOvn_YXcg!JlYsl|vAjPV+ZILx@1+-ug61#?=7|x9?6`)|7iIzX zfQ$2qg5unhEIt{yhf$&ubb|4aE=H+^Vw7nmM!9lgl;?e$3dp_yeJ>aelq)*W48}mJ z7#Efh<0A4^QpBi=?p2v{RotlhrTu-~})5T~`xjA|@A14NW zfpG)&yOF*(b^z*b#1<`?e@ptbMAl8{)GAwyn;SxZm6{U4aDeLM~rSwAPE?!8{>4Z0nD!lvU?)07rOQC36p>^9w`O4 z!6;ZRMjr=|(PscG6r(R=JURn*iP5hzbQ7aL`}lF%Jx;sF2f|!9B*p;7AAsxuv>$Lx zj3?MvPxJ=LPtboL{Rei2iLg$LCrdyZzz$Dhha|>ILeC_|PMRY|vI|Xt`6SO1bL(yRQj*G1llz zFpcRY;+I~v; z)0Qw&jHTFo89FXQr)Bj2JRUm21V|TSc>r3A!L`HqVj9rq%bsGaswD>30)uOT!L`6x zeKnx-YINo|g$;gt%;0yD41PDj;2FV2S^{){agZto-|aWjTfi__2-#w+X$07F&1^u% z+B(n+rbDI}>!P5!80)dadhE9WT{p1rHjIIlVtnmD6F`rz(c^1mY($TZ*lr_zHqs}f z8<5W+zZso2GuO@Ifw^vph1P&wwyY52n+RwMNibK8t$oG#wjuO`55(9$5ax@qlR4~M zD#k7sngiusOT^fXKABx%3NYTD5@PJ7et#KgC&qz6Vtjv0jDrmT*#~C>I{(1De_-A} z%!IvS9I64GU?OZ3<8V1>FUAqZK9ViQ(L_Mj(FI~0n;^#V4nW&1T(#+t!l~OQOOg{5rI?Z9em@c2fiJ&erQ_Scoa7fG;w#AGE?ywiB1H*tj z?*Z}wWCv2jEXuoQMSH+pF=LS%+ZU)WRvUPStJqpG7UtS$mq8bxMd?59-$v^K$gOd=~5#^9nw_cvX~`*US*J z@i?GQlR0ouOs*s5_053!UXPuwKPG1L`p_HDvH1=$Z>R*FU_4--7V$v21-iD_D&~!q zp&OvXjoZX*SrNLxB-kS6O|-v>_BYeEHSdSqT1U*=M*@1ZYXd{XY|pm4TZ`EV{W_(J zd2aw(0`tFjk(iwY8j0DZg_!rD>wV~YAG&r$*RJT=mAQ4@CFcF;LL6q^KM~f6*^T|u zts|gIH*|TR5_E+(f%aTC%0%DWe#2^sIeZY%XLz=l&o+Vq zFc*%9IRY6YkTK!|I3VV8^?`XlM|30?DCVd#fK6YZ-J%sF;%&>$P5DPVNGefWEKC zLTfnK{iaO@`n{bY=Ja^LZtp|@y3FV(=DW=8 z-GgGjhd#5g)BEhZ4~B?28(DLDi1~kv`%yJ9=dmB=qtAc(iMg<)n8Yc(bqpKCT-sjD zWo^X#ysemDbQAN-Mq;j{UrH%4S04~F4O!{6#ay>s%ni)#YjoH+UCfMaVs2*4t;pV1 zMa&)8X&3X!q`WTyc8Pg#w3vs+iFqVe%%f~SIzYUyN}0!z_2WD-gM9AK=nKchvPQyE zv7BmR`5dv_Wnx8qAXao0m@QU-PcFtD6f17JSjG9gV@c{u^3Fi%wPKZ>BUX9j#K%EP z;PW2w^MLjhYCsp50I6bKFkh^SZ$gGx7nXw7FbrsaA^KFJeWl*On3cAQbx}F7E{+nb zN|snv(XDD{K$og3;E-6=;-M8FquK(os+-UdkWqaaY!RzQ8E6B;VIeTznsuN%Oonx0 z)rtk=)S^wTC1PEIoJ$%3^SWdv>=LVXW#|MGAYH6W>2qm|Q>lmk^(KmS8TwqdOsx9T z0Xtp6T(9T=^t}?@xld;$CWzG-8I7~WYBEBsYnuZ$YKjicCW+OYaa)c8+TTRGn;7e+ zL4dt)LT0NvkR{eFRiHOu+cvFW9AKMUZ-ZH4-4=i*V%^?AthUVe4)*UIt;K4GUD|CE z>(2Vn6Q;sOvD%k_X3!sI0CjgU{$14F)fYC1)d6`OXy1W(+|8Wt?giMtqXPDK$2Y~g zhjH&|1I*2%J)fIj3uLw)T>PDLfdO)^V-9HeEdtlas6U6FKTdcn_UQguoVw~Qz@7*3o1N-I? z6R^o6$a-W7a2)iB1@=RqB$y3*#rhk4{)XJXygrH^k8TjFAM*M;VhzA11G|fr#C!+Q ze=vQXS}NAiBVr9h)-waddX{!0`iM1>Hlq??hgdIc5oBZ0YP91v@BC1?k<-%P*F*m_F>Aa~0Um;<}S`lcMT0_OeA zEMT0iae(c&GOw-Ef%$$*-M7?z+aJ*HTk5u<$F`=>7qIg-^xYl>iO?Ic@%B`)b~w-g zy2AumA$&*$>OfauE<35)RYt7cEyT*C-<|}({`=~Q^&R;Gqs98ZqgV%N^TQai4kw9q zGy+D5bv#9^9~X!fYywL~Nky0}3ZGJ$twmXUlg;WbN{tm|_l9Y(MU+zp+5n$qbQZ!f zQNB7rTOa*>>qNP+&;ka*TsR=g&!^q}onRcS6cypZJW-MH&>5z}E>TepVH9i;6S7b>19Q80BBY4oJsVXO9jo?)NpMIM>%FQrRaEtQFhW!f z#;LJPR88t@Vo%m|RTFtNvqjZnj9Rq{pcD1zTWdb-2yrPoTzWMOgjuj(RGn&~>N3a6 zkX3(*sLPv*YCygLeHwNXbtSKjXn)mhqOPXxHR#s3vZyB5zA3NGR*Jfzzo-_>^TxTN zZmKV;)k0CNhlsirIk$HZb;l@J0F2j;KJD7VP?!zdMBNz+O`tbS1m=6^F;VRquYC`g z0rb7A4onl(p%Em(B2jm{&=Qu2>PUUZnWFBY-#sfub)ql+uex`usLt%mzaYO$swn(k z)wL_2)BP<(bsH|K`vFlsW{K+AOVmTuJ=7P*^44Q(zy=Sm74-;UCKGer$X$HDX;f*qf#4C$hVjuiEDYf-~!H;i$g87*o! zGM;5WKi5vw$Z@a~GDSU4pXWOPHhv!cN5w#M7zFd+kf;}`0Q$T@n-^#^nl__pGnzJ| z=fe?EFQWg8*yP2jkRfVJIbggo!(bs~iF&C%pvOzo0K1Ki2lN?>eq)*Y%jo-ZBJ_t3 zV6Uigj6IIA$4vm{_DTR+!U&+x_y|Ck@k2#TK$i*GqW<0p`oeVBCTb$%aBWZ%$3lvz ze<-LA-GDj$13CYo-=x~WSd+%U5@24hRfG1xc&{PifQhhD)SCk6 z{U&v9P6GDDTeO?j2atu&t=>i!*EscdFQCoa^qX!1`)7Jjm;rl5y;B9c0J7g%Cu&9s zXagf*nW%RiXaa*^zNnel?Y&aaUDW$cVK^ZB!*Zf#cN6umrK0AJhP9$TstwrTBjz!W zKJ(fG`)?lGKBoLJ<9tlpkC8b)3OWLM%-<;L-)#SPI~WIRMJ=Gbfc6V$zkv4ti3M!> zAIAR=`)DD0FKi3!vxV5>lNjg(%>9!LQHv6Q@fWdu5!)BX!EL}i7qfi{+n02L@sKI% z(;CnfrovWHOXHyfuzl$|QJ>NOvsOU=&#==n^j}7K8Tu?+F6#3bXbtH5`3h0XDKBpU zY+sJOzChnE(B})}eX&c_iiXe^X2D)jU$S4m#GYSH2JEpC{Z`U`^ z>7r7Qm(mtS!7@>+*>9`aZ>t%9HS$Y`TFt zZD39tHj4VX9JB*;`g*OXjb(uH#<76To7l&jTEPfFhYS-M0Wvd?xfvZcHv`7pya3Q) zOKs>5Q(=p!Z{h)&-;9A2QCpF@wGE60#=z%M-`)mn|CaJL?69pZpyRe3qPACp&M*`|~pR1g`#nPLmyWsvcZ zDz-6BY>Vd^SWO`b7Kp6`8bV)~1$)J|t3g+o3>(G9|FWI7Fcwn8_VMalF18zhRxnj; z{{UDdb_8QaAS0?WAR`JH(V1e$R1rHc5@-|qf!J|;(mIaU679t{iUJRTHsqP8a)@ zQDWbEHK5n+rr2%Ah~2ImAmh$1V&Bz4?7J(9eNQ{FJJ%BXFX(EGDE*1OndSVYi_JD0&?EQtXk8{k(|%LS?{)FE#?^^kS;mV}m9$#yqAmhqqm5 z27_Qe91wdtHkjT6Cc+l6-${TDK;L&ZiajG9D9@lgV~5x?ne!~#&SK2>(FOIw%D7{Z!>n;ioL#N{BLQ$4c)i(0%Wi@*{n_WcE;Jx z_}kZsy@TUm2lcy{=PugpLXX|-pWTeRoAU1cVrQa9<~X1|>yW*NvG+`Zm16HTfid=W zg~@u2Sg#3|% zVjnvu_HlH{Y9#iLjP)bB{x}2pHiz)Jd}E$Cyhq@eTf|XSpo=*6BypSp;`m0waz20Q zKqBxRp$Ld<1ic_doalPMr}krpic^Gq5%PiCV81v;`OZdB%CXgeeC#N3iZMp9iQ>dH z1U~s+yr(!N7Qr!bO47e%A8|_chiq|5)4mMfqA!DfWt+nWamq1HIkuJW3rEC>?+3hA z;59)YQ=Ez|#i`eQ=^cJTJHo0$xIQP@<0pxUFA|@|K9X@q_sGGm|Gv__BZtI+ff|u#XFCVhXdmDL;rpo#d(Z5KgPI^%@(KsesP{? zF3ywamQ)egUmEI@=ZQ0z`oU9SB^(oH2sRmVNSvpb&r{fCSS*Z&`Qkj2F3xad3~wdQ zv)E_^@}I*#&mnVU3z!J#F{*<&FEFkN04P9UyFz-#Y;j=f+rm5m&Fh2LfoD7bwEsi+fpu<-7 z%{JPwE;>7CvvY_zyJLVpyOFbdk~oa; za4Jc*l#90v1(l~oX@LFr_mMvd4 z4mQYMk(3x5XgCh(;9>`M9{ur${ofxrvg5m5`(*oe$B}&~b|B+;FsODPkfXuiQCb~~ zEo1W?e8cwra9H*oC>|?Ox>eemH5+M_wr5Y;U_JYY1QDKjDSXp|R|9w-%P>r%#D~V! z?1EIL^Hh>iLo4=FlIdeOUXCO0sgV5&ln@h=us$T=uX!YFIb8yGY){qUs*r?iPr@M2 zGK~CrNP)Fp_rhneJJj0>jekXW{DUC@-<=%a71Q1_kT?(c`XA3CAG%N}qOc(xd`k4VtRs<#8M{Sr?Cud6%}vIheg z<%Huncu$fi#K?*=E(z@eZwAD9oGm>0kiE!zr;=pt{}BO0ldPEf2pF0qI}>GSl4nNh z-d~FdUJreVB11z`ytBpA#M=)pnz#w3J`^tX42d2Wd3rRKnMj~C&(nd|O($kj;E}K~ zv?mUPOc5P6MRv#(MNhY6M8e4dT$12Iu?zP28%-) zoYqwGXh_0=lM?bjD>6d+QlHXE z#-4EZPeR>)^t$V_I<#_!&I)e@_vajo)`tckB8q;+s3cOvJ0m#UWA+&vL4AUNEVm+nlZ>qmJNfGBrzGQB(rl0U$MUm zS$1#8vToS2Ipri1*G+|I$~;WyZ!E69I7*veWWyUE23wuSaY0SYPN#&#N57Cnqm`DJDASF-$bXd*QWGjljXY9gu+>SacW9~J`r}EhZgyc56K-%Py>s`T5%I&4)UX&yE3axF5mTUI&w7oe;>I5zI z*^tzt6{UP0si$gts?>&B+tpfXqDK9(bQjFvk%B-gz_;&SAAM{ZP}eQwTPgJmD2 z?d5~bWu?jgh5YWp=6x&Cv=RAjgRQ#zr6Rf5!B(XUq&B(3Ir}{<(bF`16q$=XiT!DH z4HC=dNIX6`QAQcdk$FE`(g#}?@{}fW5rZvtkp#&3!pBm=>3*F)u+Q2b`MDmE_U&@c zEYG7N4)GMwUK2h|V_CwvbwnBmIm<9L_1Zf%8b=ukp|h(Rmc9ZBd>TWlaTN3EGbx$D zU%nB_usjh;<6zfd*O+zd3JO$)6&R=l`XtJ7v)ft4E5izoSG216#w^3~#4IKE^OtYj z;|n7tpcjkaU~Ah} zIiL?0eO0oNc&;mFfmismKEeu=aMtlxe%EKU_gilPPG$jCG2eSDa55{fb+}|`2~K7S zR#~OIH8`0ySQU%)7U5(TVQq=L61`}`*!!)w2q$wYBGFrglUXyZ67@sNa5Bp<61{ae znH5vn%3FwY;)(#44Y02Q?5hCBuqUr^uv_q+oVC~!+BkSsuwu??>`9hK6Xj(iv?6=L z8_RTAs#l{|Ln91zHqYNiy@L1WTmeEeXc(-Jvs#B{Fj*GntkWr?>^1MOtm;L3h_$TryE$v@?jgo~D--1@bk_^_5LI{o&7qZh zh^lnKg`uT;h}JwT*W0@%8t!3|q*6w6Ej@3t@V`%kGX7yGVUgIs_$t>SW zS>F216UK7Z@BAZ(!=VV`+mL&ez*Sn`7bq8UuY#V7r}Zk&e5-KOU4mZaIqJ%D)D;P> znp<=3>v(>VFMLmjyE}Pq=OlS9(`hT@kOFVwuxC%ddBeRKeaq%x=$4K4y?o)@HEh&( zYWQAxXtEO-DG6oDByhSmobY1ocs+OBP79WUz~ zIg*tn1|uImnw@>F@ z!tStyBOwX9PD;rC+VPC%WzQk;j6~M(k_KU~yAAu*dnz=aPn0(7*IU`IRkwY+f7cP8 zQh~V0$dYjpk^W;lzg@R$i}FWCWbIzRUSf*+{gDTcZ;(yNgQNZaEaeLxF%BP#DCT4* z`;Q(yy2p))G);4mlYRVX1P8<5s`cs}UAOS#k3L$mV%f4U{`2AN4?o%*7!~nj-A0YC zxUj@KuU2b(L*r|2D(%jZEl3QMFHuHi?)iTIww;F$9!)Ox^wUqj`;A=OuwlccOP79Q z$K3qbV~-6kHhp@0jhi}m>U?L@x^-LLbML*KZcMx>Q6?CNrZou;3_c!w)O!Bo4UZ?v zaPynb#x)7{2|g5jKs~j*o0Ll)92F5`#})HOx_--a)u1Nws`nq=wQWOV-MV%4>eVZ= zZ}odkf)9BAgR5p_q>K3cqZe(Ih*`}Xe0*u6gc`+a-UK3lhI*W!OI&&=Gl zY1vNByTQ?i538)>$~C!1eBALR`&~?RC zRz;;qTU0=#cS47R8c1)G-g}+NWS;MTpF2Ym$Uw6DzTfZrGQZ68Ofvs-?{n|D=bn4c zx#wOjC@64tc2-#1t9YKb$tt_1s0d+eD7)<}xRBY-BGA5IH*jAdEO*jOu8rVRHHztL z@1X0p-fJ-LLTc|AQ_MvWZciY!c1SW|?ch2s+{~WUn+$w(EAswAZa(225s>o*S<|;<2Mnk^eyiz|MKI*r!SoNWobY^`F!%1lnGlqp!78}|GV$GaA1CFc6MLs5*xI}Iq_Vxcu^5V2fV;+tttGLnH zI8v$;ZP`?ypGoB_UT+`}X3OA)(1g5{i#b)rMOEcF8I^+=PNKN9GQAW-rs782z=oDs zY--*a$3Hk)<6Q3%JL;Z6j5N_yOSWEGQ&ZE>I(1N^&MtSQ;6o|+kXbtlshlmQZ-Mkp zuy|N>Oj{W|Vm^2su~Fc0F~NX>tj78+x}~b3qN3g;63d;OoSnTq+})hS-AxrmMMZ^h za=Ey=`?$I~Da52v7iF=?)k8D1t}R-$`0?oO+Or!!|NQg+TrO5koBzTKS8|&dQ8l2v54@LJ(=UxY`+!G9&a@W+qdjw3^LI;mBx_VaPpOFJu4_U>Ihi~ke< z4F9O`dPDrZQ8j%Q`2-#T`r1tXDShq12cLf^B5Bx}v^+@3%!{W^9y@gCz`>oncJDcK z@(%IGymB0d5p^J;c+JRqt)Ej zZq^5Mn@#Wvb~MxFWs8ZoNHiJ^nqDd`&V)NwkY?|rhyRKmMxK1~K6?01=;0||K0az! zjm*WxxuV3ZadjaB@9{}3*3OQO4rn+$TA#DF444C^JofaHZl%{woz1+SdLgSI``REl z!c{!lTPXFlS6+JQrAJ4L2g@P}{3{TozYo~ash<4AjM0Abi1>tA{~g=7?p>SV>guW> zNy|R@Z5h;{PrG)UDSS;p6E(7K> z_wu5W?5tAKSk_`$P8$`nHbZ!WMkanQihRo ziw=AnUcvC?48Mcn-5GxT0Pvm+KZ47o`0 zU@hkaKdv~8Hi4EnC1JjQP-qA(aZA#QdD<9S(uUh=+933nwdpgU`6A>3b`_ClAQ!Ns zlZ-6G&md%hVc_C_WZ=T`t%Dfw%LX|GxF;A@Z~S-T`Q!g?8W>3pm4g_vhJnuoGKg`_ zz{lmsgP9NPGydD?cRJ$>D>Jjq&BMdPnd|ImX=rGuFRQ4mYV7D%dieSI1-P19>uM@X zpuaSC&>0_+rn#fFjm>CIH)ph#c5`dr8Eu%#TA7`lT~N_tu4+&W3HjoSm1{4wN{7#d z1b-mhwPNr0Uw#?p%{O*{b@dXto0pqHFZUZ6ifJW%lWmTdt~%Q4HTie8`CtBQ45Xbg zqcTbD6%qgBlTXf#8x!qXla*;qAVnZ zKkX?KCVi2$xuU97FID(WeDJ~7W+4;$X4@qHwfL2g_LH;t4{S=nGq5A9-n)0hp1h=C zsmXhG?b>zl^rg#b85x4X}}t6VW1tDT*Kg2N($JYBi=diDw3d&7@@=0WR(;fi@^<*7FXxelIEuBlF~9`PF`MlIZ4*(B%Pg|=5C!>qEIOG z5}5uaq7?z>ORnZL4}bc_R~Ozhl&|qr)MR92q@3KeY5aKVFjXyEwroSTDEhgjOPA^; zB8f!AuaHJp$;Z`vmI&SJDgJSOiPJB|munQB(Gw<2932x+^*-^oZP?qdOJLn3dBf9E z`eg#^BZG*ya-RgWmz37mmK7Ehr0hRbTyi5lucWxBy0yC6Xe1|4QB+VS)`=~suwKIR zR*Q8-!0doP|L|~sZx2#i?jInL`b0-ZPrffQQllZmP*Fg9d`Mlb05O}aBFPG2E_x8{ zv~G2zv}9JzXO-_|YP%)~SP zo?igPWvP63QDuGcsqM>Q+v0xs;y$RV&kL`4SRIbwM~IFbf!XVWK1;oRTf5i7?*5*I z9cd)Lkt}#maS4kfJ*NvBzbdqN70qz0OyYIqSl+w@Y?G8N@ zm2*BNtHM}VQJQt5{EofVdvVrp&)L^vVYao{9Xqo3>Z{*QWwW6@N~*Vppa+k$C@Cp3 zzv2K0s#pYorEr-CHt9#dkb)P5<$}@02!(bp`HSdk?ZCe?eOf4jn1@s3L&8qLuEW&E0C3X-Vc{33s)D3nknYm_-Q}c!E1exL1eS zqq+JkmBi>sweDkanJc-0%X1y%1zaji$WORr^FkeHG@}w@kH)guYpg^&Yq?}FKCPFi zf$X6d5qX4OPLRX6aKpf>Nw}eS74S5YjK%_>LEz;DUX=W`KmhS#2)o}yKqmlnqCh7AbOJ!frMx^mB0oPMurSxppIm`Sm;R{LCR38j z`0*z*4t;ktZRb}w^!D9GNg-`SK*mwr%q!zaB?E7d{_F1l}Br3Zb>WM6tMB0C8LzL8u?z?xTCl7=9ANr!ssq!z%}X4`BFC zhTp*O!x?_T0PtfNUcvCk7+%Wo9s|HX#PClud;-HCV)%{$;7tr4$?)qK-oo%;y1TYt z6vIzv_>Bzj%J8Zle1GFbE1C)pLqC1Uz;z6q!@%VfeCKsjAS>I%;I_7pVet7qW!rh@ zw)ojMd$(O2|3@vWKsKX*sIj{~y)UD1PoLJSWcX)#@b;(gT?4HskUjkv0UXEAvK8;b z;G(-ff4_ZykinaKO1D41UEbW3>vzJp_uw7LX7?*pN_ON?DfxX2m6Fjsm6ET;*`=f- zWXKgRig2!Z9OeW|43+94o=YX1B90r!<^w02j6)^|LB>&kv{oEwLX+ue$gg+|<=ACf zG@FPrEySD)=t{6ZDx}fqgapI{|Lq=YtvDbU z{cj6t?2Jbf2qBH)4L_n8SU}@&XOI#C8pR8I(Hu0Ok#W=zaMTdYcF~}S^DFTyB5q}+ z#bvk5c*}ZbJhu^}5A3^T)?3yy>lyl=_Lki`?YSA@HoIl&Th=r6xzPpCEz{q!p6QR( z-ac1{qV_lE>Xw#$do7k@N6psc<7gNs zrl$V+&q*$S`&(A?$sexQ?)hg*8QM)R%+2CGdmemHsCbggkRgpt-X02-x2L5Wtw~jY zjt;ApssN)$2Zx6Qc>9NVIcdJP{S{pjdW62k&^MmG#o5+tRF|O< zfPG?wj+@En&=6-LPVW!geNb!e$0r&Ma+LoQ6qpKpLXePJLtJo&AfXO7G6{F~m;_3b zMmtj3p(3_u$35wU#>gSs@tm&^frMC_IjQ6w{0*^=0ZqgocZhY2jUaF0?oP3e;xpM~ z6~2S8iK9V?KM*1lpGO2CS%G`RI{MWUU)&)oQ4>lY)v;O39_vV-v?qG-RUvld9)VgM zKb1z;@IC@X8s~@D8gfXW8AsL&Q6t|nx{@$5|BiH@6zE3S=#HZiDZHzV?hzV~LdtD) ze-!AhXLMsi$s2d1`=&tmUK`zsLVamuABTv)jY54d+30>D&|S&swuO-igP=P}sIP~O z?j@nV9X7fv1iH%^-P+KbQJ+++wsR8I@TgWCM?S*|XV2(9rg09Gkro2hX^7+0n0ey3 z$@o-n5UW%~NxzbSnM992h1ev<{|(-hqLws*t*QBg@w zsw&AzqZv0cBzjiD&`9zg{#;n7R=2b;Sv48;nv8lqih3cpDSlxo%FoXi%bPGI~KGiOQ4thgM12yK3?1h@93n9qnewW zO}8{dKvh-MspIKIMq_nVWl2$Sai3<<%ESbb79w6J<-4dG>Sj9y_ylOoa%XQZb;R(P zsG)xHKJDhp`}qe21x1B4Hir4C)oS_8hGcV-LtQ0gzpyr&j{bs{zn)#3&_Vd5K!@1Y zX1zHCY8m5@+}!>9N(+Dbsjy7o5R=Jh%+5}7I(|IT{hs-w0umk(IArU`?(cy{M(E}Xq*$*&_K5$ElbiVOUpV;a)r5FU=TST(Z?sr>6KR!<~;fE z=y8ueA}~m3XmhiRi_QAD5FA2%3vYo#kja)j0S-|#H`l_e+|(`Bg@`0liF-hxhm2U7 zD&SQ%HaEg_?5e`%M;Ym8B1O7k7K_4NT zf9h;XdI9{)B{wq8A3V7K$k7uglFwhyBI|X-B$Ceh#-?@?Jj}e*$<@_aWJU|MwYIl5 zpvB}8nbit2tJ%cE$J`Fa4{I$eEUj(pvY7MIvx{qLs+xHm`r|Gpb%g~rg~jb{ZRjGH zTy)a5`o@;d?xX;V$z*PCfswhsuDP=j8kdrEx3$3!VCv#+J??0|pef=w!kulxY&(5lRm~@B$WO7t1rL& z@}pxoy=1TqHzaM>man##n#1FU2TEJiv9b63_Z%qBUzc2?=S)pA`S;-0fXL^UidL+w z_^&u~TF0Rs2hJQjdM-KnM0s`k(PJkD#i~>9;hdadR~PeEIvwzw>mYd7OeGJFBULtVKm-k;&48GbjzH!*x=58iG?>21fLcI+Zr&lUz&GjJ^f zmr<}VJ{>;0D=VAI)ShQ|X7KSnWjn;17(SHYb6N2>7`|cv_!$hJ$nf(R{s)Fn@4+vE zd|3kdvJ~=V5#-AYkT0VoT@97xWmU}fz_(DiJjp3yZL-ZK^h_U5Cex2y*5S zujA|_mm}P21fpD=NM}bkosTT-_W8(!c}qs)dGMo_SrOyYP3NVgQC40Z=AmaoocVpNi%0orQ2fOl&;ue zBtgGRgcoNIqMTQW9~~#z9nXKrFHxVsq;fj5_{OO!>yxu-bh;E#q*8>d$}Fvzp?l9! z?}<){rY9o!Pl$n%zo$5SIOlRktxhtLPF6{yPNZby7aH@fUpRf{RPwRpUE6l-J8>$b zth}_S#CRh&b^pHazFEEX;OT>ia5;4B7+Iqmfk~r{AUu{%RuYwcvXW~=t!5!?%*CbI z46h&&(@867R6%7s5lOn6D@vZ z#+puAogE?ZX0#l4`6CCQMtdN{OVasi8wB5!305X|s-IF1O;InTvC$Ggs@H zzB5-6I14$v$VPBh5;#i_&YE#e?XMB5oH&I-UR==Pq>>Mk)xc}__A`AAhWTdPHiHe4 z8^MmZPhVQy`+wPc^6;V4M~@yW zsYp9?_}Cy>8@zpQKeO9ln0>}=6Wk!VAMD@HG&k7s;80!%Fkfa3sigA$Fquj(Jo6TY zkdV91RVIe_WB3CMf0f~z27r$$7UtS9Ox_=X}=qwt8FUQaJ0~{{8$1VY{N)P@4Oo0~k7? z2i?1Gq-fil79}TEI(F{R>i$Mc*E4i`58Ba8BRWWZDpMOd_;h~7vWIZp=c!Lc;(J%0 ziWe)FW943CPur|#z78#s3DD92;sJ>IUF`qd7e8?xCmcv8f!1Q&nyhwGzQ#c z-)H7_9c$?uH&RkEGA>@s$Vg2A4=%HFl-t*v{MEj`pK2pVuU|iU%BoMvDV=z|Lb3iP zYthY)o5VS#Sy}O92R0a-IS@}iC!cc%evT(Uk{w!Dmc`GGAOGx4PIIemXsT^&G@I+| zJ3H&^O=gmf@@(AZ78X7+dc3rB0+_P zg;He;BAW^djb#<(_~|o3Zns$3A28^?@sr{b#!pXt=*ZrKox|>Z;FckDyUofzrsDX& zYrfuipoljr{9Ig+9R`v!9 zGIclSRn>QPG|F6@mGZ8tTZZfHHmkm^+5h-0CBIV67r|RW$M1()>QK{)ib)#1Sg^a2 zjbA&u?@FYEjsCwgHcn@3T*BD6o3gPWnJF)Zq8~@}=0zH@+NbY!_Tt53X?b({bWed! z%f0oM!QLa%cb2+^!gKX&@`zHC!cFy}jI_U!pPx3K3twD;F*T^{4^d+4bP$B$jE zBego2rMosSZ`w5a&^oSkUjqMk`ckYqnim{gT3FX@khEUewCdmZJC*Tv!LM5V*^95f zGtqQo(+VAuf5~P)@x)Aj;*F|!kDjn|Cv6rT7qujjnhNpyLbiCq)vHkvZb`virkaAh z+zRZ+sjS49s;G9hwskAzVlV$7jg^>nGKCzORJyth;7FZICya?cUdqnmV&dr`Hv|U< zDVmz=!w`iR8Qt2fQac5OMUNOcYW&P;_aFiPLwFoZmxWYU8^D!17k3` zS!~{i{yhZ;bxAt#7v9wL?L2++)v+}I^_ki|Aa=i(2);BhGG>OEH@#`k;Zu({Eft@b zH;v}lVBR$9DYARhDAilNX(D*j;G4s@m_Kbl?wCJqKJJ)5tps<>pEetJc7NKf)P%VQ z{xmpf@Ga(1+m1WzTZcz&3GSFbtq6C_pB8N!N4@=N*qbZNVsSk4r_pRBG|LC`r^N{L z?EbW!0?jzGUQ|zBVN~Hw!;U{;boNh`dDGN3s?3`fZllV)X(=|UKZ5G-7*+Vv67EQq z`O>0nRGBX=(ngi}(tfj1{Q^`MFsksRxeS6T^Q6%-IC$uoCoR`Tbp@zC%BaGRCb3bq z_fT(-tY{D2Do*=*68xuru*lP_A#`^!@`jML_=LP6QH=<;6Y_>23rPs>dgl$H9Au|R zV>v{pD8-NPE%f=h0D&US8giKKJ0}hfXwvieW)Xe$JN*VOqieo$(???QZw*-D%v?-T8E1 zMDJJ!@t&XA5EJr4oXhq_3(@=#k~xZ0xPFD~5CM@*qy)Ff4iP&hg3QLPEjz@m6lsf# z&m3mMjPCcB9K8}mwgbQ!B+m!afhJ^!@QbL2HBHD%pz#PL(e_9|d*%b0mw@`NDLrbS zc~cHNKk2E>Ndh&}fm>;9XK`WM($PlupDy{(gu!M@aKW`&-Xxt01P{(hnEv@|MtFOh!=KJ8o^`4!|V zVJr&dYa2o_7<*-epuEv1BgCGc*>KchI7-fC)A6WdFeC*jXXg()RT_aRK6Q`{Nr6AkWz#W9vq4}?ivGd|rH~B5Cyb6tnhZizSu@n2 z!$cv_>|ZBpXF8EP(}|LqPUOUNBJoX~i2Cl4Jrq8B#14bMXnmwYeExpja;Xq?pQT%V zGl)JGe-wh^Q*a}Mx=++CXJeXD>K$84=NB!vAF}369aOk1a3e0U zS(kY*Pi70rqJVr7OD~;ZK3rVLMEkm|_sWV&rkg7(8N*u{K8xYo7+%>6kJ{6mO?@I) zR3<{_Xx<-d%nmD{C!ot1f-txY#x1K)>pL+EC8hm1UeirVSo<9hbTNDqAVovkbD8f|F3sIKoy6 z)i$bhr9fK-b;U~I#XHuOtrUjh+bnk>jS^&%KF$_(cg;qXt`umH8>nKXFnkbH*-F74 zBgj??T{fz8r7(|C#Y(|$1+n+IU9YEECD}@WJ{Mal&>rtuDICKmwo;(me34-gD~0K} zbF@+rC<-furT7+GDNu?%D}^e2Vk-r@LPUPS%Aq6}cX#FL0ZL+rqhf$FfupqwO!i_$ zG$}<@u~un~<5gNAcNbf$*bVy6g>WHvTdVvpV_w>?w7Qe*ao8RMloF9BF%lY1EXK9n z5-T-#uHV>YMutT`fm{7`JU@ev*T(Uqby{&}r$V63SgO2<{iSXFz1rqeC7g_^~BUgQqq*|jY-mWz|e2#Yeb0A+UdyZWDa~x&Q(cJSK4ykOiQ)#hN@w8JZ zXH<}T;jV3Z&`xEKoeIqvCba1gMrGjusGPG?Nwrg%W2a*4UrP@a+ODdQ+j6#(YDNTH z;>D=MBgS(%&D*|%)&y4*B_i~V!&dpmUXx;bO)Bj*X=Kkcyr(8~^wM?Kt^68~#6=ix zlu23tJZ7hn!f3Sg&~R9jGxnNLi=)8vwRS3EM#XsmR1Vpx?6p(5*+0RI%7OtANB$6z%ZgV0JLwdh`2|M;6aD4myY%lx|@<|HfCXrc_Fv)CEK$Kf6spCiG+ zumlO(n17S7D*)Qq)Clqy{H?b()|;ZBjr|65lC=omLgwEibbrA_+(RFmM$Iy(a1VXV zw_cdypp68ElILu^+`r~DkJ@8|tejNGVR@8y3l!rpLST&HRvnO0~QtjOxTt^1&UcE)=MKYNI-VT8^y0+NkP; zx+-i`SAyy@j4Bs)7khADsw_kGP8-!xVwZ;I+S=OUVq;NVT}jD) zy$Z3hX5TO4&xNM&19l2OvlA|B)K@VN}`HXDRVBWK;V8;OL=)?LuIZ`fw*d!94i*F&QGBZ z4h|j?8t&%{H=d7=r&=oJf&v5GP=Y%`{CH00i5weVKE5HLp`pIcI*|pb;DmYEp*}2O z)(Q=t658HbP{JjGy;o317*F4FZFfR#2akwF8FZJ-Ec8BtEbuEL-UU^a6{W@1IXTFnlA8lJX>m?Q zHgXPVC81u(YO3QTV!f!t!s$3Umn5r8Nz2YJ4oQH@zLft!F@F5U^*?Of*mWbG|A>DM z1MGF3^v&y+&zwA`<2)65QSDK)j+>d1B#VDw=IHT{??^yo;s){wJi0TH<7PW{Dtrp- zdFo5?geS+eab!1!z%!C^tJkE(I_{T4`Z#ion^%oLE<%zh@5=mw!s60Aq#!jKYicX2 zYf5Vy>*$;$DXcMzq&m*htrsaJypA*RN#2_1k@5HVy9WmPYBT|XA^riLfdN6VsY?A@MiASg@o$-((t&S^WG@R`;csq?sMgs%cPJ>Dqdrpa7s|T)o57j!n z>7fkJI@0kNRPz^=QnQzv8Y+kI6pqlN_aJ)*8E-=)UATZ=UrIB0nADst$EiQcjs}m7 z2aoxK$A*H(+;elUU(d@+yLLJA%B3q+#n;o2MdP0w&nKzcm7@a3kKgoN&ZL#_Ny
  • YkcmBK%&UtRej3lY2>A>}$5sjk-yG+@aKn{^gazvCd>b*Fk1>gL3 zGc@9{#{|s2eMzqENLhpv*|>;+@S)*K4X?wt(0@XTB)JFr1cwI(j-4P58D<&g+7dgK z1fa}L;2bGVK~}i zBGC7tEfB1WUu2RlE6dHvDJd(-%gWEm&c2ael$)Ike$qRcOkxS?=HyC+0)tH_bNjb%`N;N{`A(;R2?_gDT(`8 z*njptStH8h;^L7a^`Q9u=YRiUJSQQYBFnR{&`o!7$i%IWCu_(W$+;qqYe>YJ;T8Tq zem*2zJpT~?N|G0%cWUdeU8^!Tl-1Q$wKo>j)Yg(qsHa@s)W%DAlc}>)tm8UONnY+6 z|G+?v+TB0cHy|h=EX)&WjL1mgtKHp1%$qvQofwpz9k#iQN-n!qR*H5JWaH(wsZx?6>TX;lub^1Cpn3LJ{T3`Xs<8N=YeJP9IOYn zu#q3QnK1@^ykH`b(}$`oC)e=}g(3;t#v(^d9t(-A=fc8RIrcd~igM&A=cd=E@}v;W z7Y_0tYe}gA3{VC_xG1Se?>htZM$<^olk|Ho4BAAYJcA+=wT)Noa|g{{V=q@LvNGgp z!Q&Q3zN1hDIYy~D_mpbaDC8l1tOcCtB*r^;znOz?2(MK{!F6A!gIzPP&-#M;Bq@hG z;Y&_vSEOBtMZsFmnRpbgpj)=^z1 zoP_k?UbEL_BI@$}bt*bK=RtJzNFK@;w?0SvNe3slzgh)}1$SiSuAX5xC3V`n_Yo#4!@YDj@^V5V%N@0f?h( z9&u$c&EW&p?xsDKYMxZuQY^$Nera^?$>R{zAL^EA#pW{(Jc3!?QEA3j^wZe1fk>u_A)k~K$uiwZi zG+r&r$;rBrh5)eAva%Ag0i&&~qQ1*LVswD0t*O!A6Rx z;Y?_fHQY1YaW2P@Xsxr>^8v_`zbz%@w?yPbpJ3&!b-Dy=9WUn>ig%=>>`a6-4Yf8} z>yl!PMeW#1SPf06rm?A^v6#2E;&7CN&`sq9s(%kQNFN#6#?QMt@^YE4l<$MhzFSVkfa;O{`et z`oRC_h>>E%+^#JPp1FEDYb>E}LR#t}I@V7AZy=_iR`i5{$I%v3Fpj5S90!$`U&+bM zymt922Jz+0OP4ANDoTx2#l?u-%q3fOdSyes$#0TJb6L?K@Md&=H420(>qS%d*o>pSy7J%2)5b_wA&p@Cgqrcz)6I z&tQ~kNvKYLKlc`wzy%=lUN&AgkYC|x?k&Up)~nVIJ_zJ}@QUKytR>dVNkK+qYkgf) zXIE{_mu4rKbGMW8Gb9npJsUD0Pq~cKf>?_8D7Qk<{rGG z@!q+Wj`w*(>3H8CO2@lAn~rydpMAVLLJr?bHG{VVnTJV$~)B-_H&|UxUVnk{&A{T;B7Ggw>yge_(9Y^TxarcgfY~byg zBL*;pZ;#F&%$U9{>x7hLofyDCcNoz>7y&Ty_AC{56bQHP0&pZDM&6#s;*JvI_FV#w zL`YxW_)BlzICbh&Rb6K4#fv9ToxYHol5!X6JqX@-`}sw0EP4L*KfGUFSQ7U5Ywr$7 zy#H^!;kJD3FQ+pOZ~f1zH~#YWu#!s~2PFUh&#;A9EY8nv=rWrnVo`UO#d4QYQZW_EUZTH4VomoDa~!%TKPBQw9s z)L_iGUT_IgU#C>N$VK8)#_~$5L3y>bvc=<(iC$vZGok`3ex zxj@c|wXmGU(@hWZq_2MbdDGXcf1x3^O%T4&mwmVh9RKpGFn1r}IB3g$hD1!ckvtCf z5$5^#h0|8%e$2i1)l8V;Uc!k#f998p^{ang9gmH1bj!p8FtDYNbJ!=no@~ZSe-qyK zlXUT#_aNGJ_ggPpOL-SQRG7^}d4Fr0^{Rmm{AZEvKAq0|ST2Qg_i=B19As*!t7~j- zR&|=qRZTFxwKO)=6kf`#Xlku8VS>;pn_EoQ?t4T0eMJW4=s;hWy5CZ2y19>olzewr zYkmEsXm_{JP*~x@Lr0E@(6|Hz(#TLh3M|RYNbhbfucg4~u_Ilcn?5$2>#?TYYI_kh z+bL|OiWg?8IR1UAv|!>r%H}G0&s^njuFPQ+Ca}5A!0?R>A2$H}NQQr%;UgJ-2g9oe zfOltjABNw+@U$uM7%@F~C-5R&o7;243M)%-P61dBn*)=yWI`O>oHmp$Ck=W~!L_jx zWN?j}2RgR6W2*b!Zg1JDOF{@OvQ>!^$iq}$79!MeT$V5nDCTwz8yFqR!&@kQ%cgIB zHiUg0Of3mL>)@@Mu?~Ko)@bO%a8WIzorUEa{X=#h*+j2?f$)#8mGE4Y`V^k{HPDz# zIApUHk)bVhl`!_#x0%{4V&poov^2D|n33p9B{H?Q)mIc16;(7M(mfpc5H#4`F2i>^ z8=6}>%ob!pXr|}AU>`rugOPEjR>RQpMg=|dMQLfTNk4u1bV_Tt3+SOm?AVuOqIiT$Bg=u*Ex^wOr)8Pz zN-J6k%e?cA`1SDi6EpeuZCQC<7k`tdoDolteK8D8%gU$cyj;0><-*AmCr(^Q%ScO0 zJ$d->;Zql{T+O8EcnZ>SAY3?-8Z@@Go6JaRpm0;WC}k2XGu8BL7o>)Orf%lt&IS?A zXVDKaHPtuQmX(y0R5!P`HB}TKKYmedV~0qAQ(kbYBsL{T+Z)?jJG%6~P9Cn#G7&-A z0vEZs3rpn2Rx=bm7u!w+hvO}h<;7?dw&LOZc)EQ~REm1yQh^fgrk4jZyoKR4eejoi z@Ki6Sdr9b8)*ff56@3fd_*l$$0nC}r2CXQW;>A!M*XcN!41aivk`TE5T~JsG3ZH_) z`{2Ldga4+qnY!ejfj(#sjfUoXl8Z%M1LDbYowOE*EQ&b8$dC|lr9{%*Qad0n)lP0| z?pC_HDH9)_dH(8+TDj=bmcs+$++rP>znt8olaG6N@#4kLKmFt&8F`p%n74y?MbMrT!{TICMdgbE9)le$?f-B$kN;TQuJi`dD1Kx4w&n45k1#KEIDfA9QxWVa`jry zK*q^F9SQkVUP*@+=MnP1kC}%V&ktcdzmf6$RyJl%_wc;KG1tZL5e%Ql@YfhVe*k!- z$wJJVsEpxvGrWr7M+^WT&+xMu{&9w1&G4xMz}GXp2gC1V_*{nX9svGbhPUOiKF#pi z3|}$;yc@%p_j!IL!^iaC`?m!IQ<n`!&ux{)>@n=u{s|4V(^U(2d+fK?&4r~VG{E({;tr~S_{{HVTo zswdoLKlGz`JwvCk(u){+Ag%AqN^fU)mTW;(#qc-!mVT?pUCo#7|+)ZAhD|DVLW z-d2Aan?%&C!!n+hL1z zl?yY0aH6VKlm}%UGVPjh95i;%?Ypb?_AtW_K@1sy8yS8p!=D}i-d27f!`sSFW%!B# z;DcHD7KYcb@=F>1au41iFP96vES&4QYo@@(Tn&=ZmsPV7ho<^u`5pr&q;S(a%{506lNgbhr#B@7~h+&(39M5xW zmH_dtbn-8J8+(6Be93X#VeikFRPq+?u=fYMm(Su3(Wa9kNGtB_(WZ1&NNM%n`y)P; zEoAM+9+2!HL?Y0-iSDD3rps`TNK@=wrZK083H1vl>EIgb^XWg*l)yIXpjI>g$3ta|IMcb*6#c|WcE{BuLkkr|4{mi%)Uua>qn#*weZAKq~P zAd&1N2N0Y0E%$&b9+u~?xjA$0C-v|lOSl-)LUK5FZVIuGB5fR|3!S8_&>HBgPoPPS z_=S=m8?B0k31ka>Ap^xC1aI^lH>LO`rNCI2UzB+~9u~FN6uYk%Wzsdgr znTg)!)~}=aWI*D9vX=zWA@)om2q3m^3N4N?tBkC4;Mzpq! z3U`${YrMRDy4u&T-?e?`-b4F#9zK(dz4$uGkO*sDu^vGk_4!5C@F;h8cSFyC9>UBt z7X39A{WSyqH5UDa6UvBRIig1D2;W^-ZTan2S({0zax%Awk?Sr+XK;6Ncb56a#>PfP zMFqHd`MKd#jwHYF<7FI6o4uGq{?I&u^!6 z$BrHAZoIHNDFOZhypsHkMxs%5n!21^yL|&%Wn4pjUALqKsq;>vz%G+jYO1WPL}sUU zb6ayOP9sV3(`Y(8edJ1~0DlrcCw5|_M(q?lDjGTUB7iV;Sj3Gu9R-JFibzWtjnKZ` zTTf@0o8J1W6hX6$#uGv_%tUQ*DP=KlD$u-}T~NjFqZs~IhUXaGsRut1{c{ic=OOgZ zJt&_>y$5w=Tt@V`gFZT`Ti$OU1zFCHSR5N$zVDmWt5>f%(R_4kMZP5>dZb5}WY=0m z?29i>eP7joXL;;Ut9R_#GkMIMMbADp&eNCktK7dL5&5tG#J|P=N%Zacx8SnXV!tm5 zi^rzOcS(ffo(n22uPQAlu&Nd99VVqqm$zTDgs-cu=@KDlEQ;N;>uBv>-*p|MkAJl$Qx#*K)KK{WdEe!G$Bg%6+|pGP~=^LZaYJ3fqd9NPPw zNr&x;lNpm^1LzS8a=#JD{qw&|?_(yB{t+=-Gx?!#d(qc${t;fwKgdr+%JARIKHatN z@Q}I9jpiPJ$g1h*d`ZWhG_?2el6M$ty$_e{zt6<-?)E)#{39Wr znr5G|<7V7p6!tz_@(v@i|2a)aOmLL0EyC&BN%|I{vyEZ&eeX2^9mC?Y*}|Ftdi>cS zVdD()mOOTaysNYC7o$61plymnZ=TZl6zIGL9{MYI=qd0JJ?nQQ%~dV@49wHBd|+DQ z$T{ao<+(ZI9vYygsN=pnuj4{m2dOLuZBUiG@XEF0%l-?6k()mseqXUDIM|&oeg^Ll z8VRS9@H0T|arwZsN0qUtsHm#3(>O2<@{@sTkZ+^+|AyYDk*;r}_uoeEPnO9vZUI3m zVri}_uc&H*M@j0BnZ{hvs)jwIxx1q(FE_Wi!Ntkl%}pkkH{=)MOb$+s)wLd)pG+Sr zQK%F$3H1$Hd7iYkHZ<{`Uc&jhoY>N;R3e^U2|WfoIjlsYG)&&GAv%8cxFD&*#ltTk zILKGy9PH<*j&80Zp)up5y!<NwXLsI z_Y2M(nHk0f4aPc8EBm9HTE6-T(dHZjF``e4rq+lZ)gJPw!>WWTzW!Dcs{P5kk zwDiyqK3M(rH)~IwJc}(+m#^GNzjWgG>a{!5h)TI}jJv^}G4SCTe?}k{5 zKTkj$KVFiLv&g8wBXgIyxjAOBPPzDv4?g%{=|`NY*)aL|@x6O~-g@}Vh3k2R1$j48 z&zw&FW!v7psRh^e@BQt}_4AkWs;a9?G7cQty=TwyW5+To%afCjpWYAqgI>h*W__ro zv%S-7>4tv=JwcqE-CVia8XPrY?&iBC&K(^H`&Z&@V=E3Tv6F* z<_X_gQB;2AJmMnD+N!TzNy)D+uCb_`oh2=2&!=Zx%`Pfx>yR~;v5JRIUn+!+eFxOPq10wOp>FJ65n;rS{-= z>(;IR@pM~lR!OckCU(4w`+-f5J@lU~&}j{B>fo^GiE(jpXU?1{kcTBgPs7eBxuB~> zPG2~G-bl53)QsZY2B)zAZoh{PR8c+~Z%F zw#BBe>5ytODbQ%FBj&R5if&m;Sw&HE9fn4ev8wvuiNZ`|mNK}tH+AyZn_&XCTUr}i za0eBhOA6FzN+mdu~E-iQpe)Ti56oK;g( zgI!d2i!ruT3Hn_P@V&fvotoSq?YH{J1NhM&RkpD}z6!*>h-uV(mAhCjmaRSf^&0PsN! zAHnd+46k8$a}VB;Rnn=7S|vlza)`G)GL~8;MXA&(sdTqnCHuD!J0+mmZ96+c>B9>< zJH$s`Kn2*|4&+g%W+=M1Ls@}}+`hNt)>!Z9RWyQghESL1g}SUyr4gJ>?)C`I>EM7_ zko!+!1llvrh7}qMjQRO@8i#qhPaKWIFw18b=S>fWF1~x!>eb(Uf2^TAEkC_8B05&? zyld@)vwvKV%^T8UUzKxgt;0b{PRC`tUyz5K#3V z&Y)T@=;czqcER(=(I)k+%yX`p;p)W*h&&Dh}8`3wt49%T~pWNKsqQbl)oQGCXoS$2imy0~V*;%=HxjA$f znwaNBoKB6&va8F&V>^(W8plTHA^GU9K|a3u*v4(!fBsZR80q%OxnDbLlw&7N59b>s z`!|g@S0^B75%d1|yVn;j!D-eme{c$#Yu>)JKSNwye{N+0O$Ea*(sM(G&|gEivAnFZs-m(QlUQkGO_|YH zR8U-5TV7LZtgftYs&B&PQEX@vbDWjesp&ZlQVBK$wbnObKG8#J(qBWk#@oxs&)?r4 zzwSQ%UK)3IwTEAzw;#5a`UMRM48U3#d);oO($>)2n~Gp3n1+6dN54FXewl`TnTCD| zDJh{R(>b1ta&*fbje(FXjrD9qMBY!Iu3Y)WzxI^noxh%1?;jk|E!w&&e%d!*!{H&G zHZ3K^A^&^LsV2{!JUM&X6K^efJuxEs-X}XRtX>W~`bVOFXZ$UWd@KFp^H-OU4`D<- zKXc|Cxjw{LOmluao}9MeIUMkOh?|?Mo2O@`Le5IPv_9~I*|adr?< z0DDJ{UxJo>3oT3Ic9x)JY242E#^UUf#*(Zvm#f&h2dRYlVPY7n?E+;yq({OLA2ntxd|9RKOrynTYviF3-@71IT9p!&bbfc zYZq;_;Yb91=#--GYm-qZ~{-_Cu-!wIpXXn>8RT-}q6xNx|?Y_ZvSwDThKO^nB zu?#-9u~uvK=`C9hl%Z9{gm-clcO!EYPRBv3%fjR0BB8igWiIViZJb0iG&V@0lv)kr zJv1RvJ|m~hogEbvHge=BPx%n9>7z%*PMk7&*ytIv6DNm^o;Y#L*x1LO9TV8x;5BLN zxCzln4(ij{S(p6jr(5&jfHQS;NL5@HX>)gX66<|jqNYv_!zdTaRn6r%2heTQy~E@x zuJbPW0P0f(y(fUp-SsRVKmm0nYDIoD9VwH;4s{nX!{;&l8HV>^_}6>zQgHRH5wLou zA$SN-Y~#~1FxRVaLDo1dTf)lD?^CungTLBScKAPiE)lWR^*N++0L5 z?5ey#rD0c>q&la#op3k+(QFYz{IzE!Cl_?lk;_@F-R(Frv%!oIe~a(Tg$wWT5*fx< zD8iy5Ls~DUHp-{X89R1Tq_i<}-==lLV*@Q!nftf=wC!?A<}Y9W@cpVCIo&~#(O-S@ z^F?x+mepQadSc_IeJ3v!hy(q`ErRG<2!n-80$C{cj=VQv+C3u_RmDdkDmSNQWtE#m z1{ZHd@%f7ve*PhuFK$eR+u*PKhthAZ*2`xtg#lcB;LzzZdDIhcy#DgT6Bo??C;{^C zS8`CgI5wgq^$fls`TNTgq4~Nh6jphdt0rpJ+!@nHNwRm%pO2nQGleXkZ4;K8AvhsEMfZ6?5*DOom6cXHN4OVV`sSM>$8Z$knX9?RqU;-) z8JA8bpFVRYxwA?puC17BxF;TIj#xH#Tjp>mc-0+bjBg2PJnKE)j%M)3FXdJi_=#YnGxcxxOVEzy}c1yW_bq$<76buUo-^1~(?HNcg( z=oL#F+S*#7zqRRI!3!rELmq$p@%rgYmo5!YOHysy_*1U;2QNgcC;ovR7pwF5==bAV zCeC%LUif@$(A{8y#*d^hTXpF~ldj96JaE(T-_MbQJgAs|3Ke^!Pg$oz@2R!~Ts{Y86K|{nvetu9f z33htwmdP7mAJ^5Ygn7vYXk>)INgg4Ulm_b&r12o$=#e1w2t*=&MMRl?qZa0XvU23M zB`{XNV1b-N$}vHk4{rN5dK%pI1f~d>Bgmf(l0!E>TKpkS?Gqo_fbILpuY$A>aJbPs zgR-`6XoQOz^RkYW{|ndY+=g}ohcAN)c8D#i)Zi|6wcbB|Cvg{N-tsH;S4;Kft&AM zwt{8+31slA3?9kgEj?vBnyL8xbfy||io@#s@;o|Ig}c$2ig&ipRF05i_tUlXy#n$r z0eQlWuB9uS?Q3aA$d>(dH)Q%LE|hTR=h5AeGu-HI$hpq;-H;=}H6y__bp0_BTr(P6 z<5kkuQkIclkda+hm06g7_1e`d**S>m5d1Z;wo-r1O=j{MHGPux_|?*klp}|JDZBdB zf4*Dh9fLp;k>uG}y$q(VH=dW=DQcX%pIj#pu8aJC8!6k6d zl4aZy62}EHzshXNfq1mNQ+N*z4jLNf?H&;rJUnLjh_F!NqZ3>wG6>9y z8*w)oRqP_RAo1Xn8h`K3pvhBadH8#4)SlK(q+AxfBQn^w8d|5`WZ^rEsG6$0d|^cW zr}v2JDDH1WH7M>dqOM&x78;8(GvFfZHKK-%9uXN4lD?}=FvAd`sQrY*XX#MfG11DliA}3d~$O0uu=K? z_3LTrIR&s0lpY;S$6jp1OU}x{v5q_DxphNB9;hBVa?62ytPiEEl$IZI` zAofgLfPn+b6L0R+DVo!N)M~XqT&&WGIh?+2kTx5Wf88Y{3;Kk7A$fO<<>Ds^7=8=+ z$HY-FQNB8)Ns(aoaSzvV!ycnfQ{hx`1a&+=Q78765t+e%`D05iv=vzzJxf2hc(w0k zx>-81UXML8it^*@*WbCF!%|X&{V3oO$-l-~IUl3(uzc>n<-@!@UFnYshao!X zgyCqej}?`zFm&9ZFC4Bc$8Q%_#XjIUOw?Mi4=!g|mb@DDux-9}h%aXNK!!ia@bwIj zDUdS5o+$*Y?k_(kHUg ze`KY%Gq|d!bVUf;i>e4=`C=3yLOw?-gYNOM(er`mGM zB@Wv1j_Xug?#iXwvhSF?Y0KT)xhR~1Eg-7}q&1i3F&B@q=P`FwX2W)>E$3b0pe=ub zR8@qF$fery;xTq@+1_5aJ*}<1>)F#*veuHZr|s%_T8H?)<9rmu+s1h>yrY_T39S_= zJo8H7nH557dE9I*hb7x2tS76vO~O{Pnh!+674H%xd>~r~9VEg5O7-MxDmf%HAsW+w z9F{+x;e8lh%kXCyzGeV;Tlt=S%CBU2^#JfvR(>nP*Rk?BhHvY^J8bz~luskwD4(v( zqI|0On)0c~a66xlfqaM~Y=t!mddei|DFNkh@HkLVl6A{&*XEQtrE1xMw92}=dTM)k zmVb(W%jJh%XIlM-gn#wbSF1LX9G${>{(NT#%7`){wzPe8u(htSI*}wJ1;k~t%jJ(# zRp-`yyYiLSUztB02ceQ;wx7ND&WcBXvAEO-Ykq%9Ub4>MNXrXwb=SB$P%=UQbD&-@ z?g>I0jzSyuKHt~1qU>6FT4ur3?40z1oGhBXQr?Wb#8y#P^D@p?4MUUk$dLoTo%$i3 zCc@xC{{bJC7F&Gu3D#z-g@2gahf_u$wYFOE^%V0=Vj=5c3fw^cEqY<$pW!6oOGrNF zqm3h(I(<^54~KCgu^1MR89YBZa48< zLv>SI-OYcx`1`By&ptEeZE{z&R~SvX!cHXaZ7@zI9Y`vcO*D;OSc};C@s_h3IDRkX zN0-&$N8$lAxFDx18jytu@7GQc^E3BcGvH_z->!zd&qQ zv&N9XFTj0>uAjdd63|WSh+g#Y!%O3lEO>nq*YN*P_a5+3RoffrIn#T3@4b*ffY4h) z02NTNf?W~4R;*yX*PKiMyH^BMnu$)rqq-<~r=OE44f|NVaN z&B&BEYoC4gUVH7eSNYb!d`w;fbr$Hs+o^5z3vb>9?8NP`$MO-j`4yhA!gt=Ceb=(r zR;NZ&sZ^1u)!TE*gjG-9J0*%L#spu%)#)X47gwVXb%h?#BHm&E9&xasrXOkMy;^N&Pfzb4)z{Ob1$JGW*xE{@ z>+aTxtgM-vJ}ZSCjlF(;`pyp0n1IWbJ34Rx5x^OQHCF0ihh_~_9~x`zs;uZ}QahO& zi?}x?=H3XlG4p379o>rb9QxN^5otzuN3+&{v-BOpZ`zYaemjc^SQAT2!EeT}od_0c z&-iw${2tQ6G`0l|+rpr^g><%sB=R;w3-xRZTDFBqa|;@_g(?eLuw`3FV_OI^w;*9# zXttmQ5!(WnZ9!&k!JciQ`bGM;o%<>|V3rA~tjCjDxO6u+|v$6vG zF;~lo1&H%*Xz0`Q3gWrq>YlzR!(;N(zwbMG^xUa4$F~K?f4>%q8xp6~f`W{UKhK}L zn1__VSQwH~vt~SOx~~Y=ZrpzC*iRb)C;SRS2c2catDk-JzUkBNee~JY@m%$6U+eim z3;csRfkNm&=;q%uFJ}JSYcW(Jj6vtAN>O28$fTGB4?puRi+s`@8-=vISo*$QjQGR% zmqy)xfArG#K4j>Vix|$@PuCJ0!YA;2o3m!!+ISxgkb(#G5<$PFN7Gc_*Mb-ePVYdM z+CUq0LW!0O+e$3WAHZF9cJ?+js5?6a46Wm16eMDSy}eK@8Ho4k0+x%We-J421G@h1 z?gs6^psq(V(BIqN-`lO$(vX(1G>;6Gz;}|rf26N~G(v{5qqe@UuR+y0U_z+9q{?!}_oVUy#&8ICT5TTkQ>be8(2-*cOu57Q)RfSg|d%-e|$>r@VQEk<-WTzogpS-j)rf~}Z4MFyv2$}%U1c1gyRi`R0F9A$zRYPq9 zf{2VxOdcaNh@=xI5-4_x7tisXKsIRkv16AqB?24y13P!Dr;akQsFNmar|e>4Y_#nR z^=v$c?i!d&X~2S8=}+E{qLkcM=8sLC3n6}m$^*w;0m)8A=rwWv^UueO^J1u-@itw+ zN$JnSaM)ZV2>WU$@|QvvuXDR?Snoak95yS#r1R-*Y4yo7_E|qIyit0)|@%>|RD#`NU+g zM!V{RXbk;#)H4kAjB){@evMCqmdXdf`kZI)hQCrzdsFq)B^1E=6Nw5ooLQ4Fu?mL! zC>{*|(LZVpgMBfEe;Hmzl0*zV7Su(hLwkE`du@-pOO4V_y*(Wbog_0PogqQeJVqjd z@0;T21~jd(AL^72j(AwRx_Wxx;pXBBuzWik6e~3S8o?J>vk1uG8h;&2IOvXRz$D?V zFhXF#e!m^+T6X|I*AE&R=1`}ez)%TXG;l$B|hvNi5fo!57%J6_#1 zi$eD5*JrWQc;_5Tc3BC#%ht0kyaZWj)HY1}bT&JbWDG+v2N76vv$5tvu;xM_SAdd_ z$L+G^JVeoLvPH0IhrVYx`SK|cc%q_uZ2&_blG>NAvQ37EOx$ghQXS8{@4_tpk+a#mLw)uv!%>YxI zPZ`^^V%z+Lv`I!0yfDwU@#t!CLU~t+7pdq z>re60s_L4`s;0(P%rkZ~PCVJhJGzJ{E>5S$0AhbYv1`nnVC)RQE- zVj-c8rq;8xKNxcrIzr9y6$AlxCivu0@QE44-3_3NJW*eLeSKMZWkVhIrd#+%prhck zMBu!b7Kz1T9(A3M9A}RiJ7wDV$_lB?^lr|)@Y=?&S#<% zsYPA9_JX{`xzrD^R{hL>!e%h*PRw^JZ0|>JGs5MeU0q%MxAUYMa=H`Gr6bj+v7xS< zu+ZS*3gywyn-Tju zz~u`ci<#MtJWbXvshfH0iN_c-Yh*0vbW)3hmYX@MmgNWi4SpbU=?=SOx^b6WLv~4H zTx6(umz)I}f~4>m=8{CN%>o~Llvk7&V2Aa{9op zr_UEtxi=cv3yHyB`on=^yT6@BuzvCUx8WnsfUZv|{?GGs$NAbDT5H^E!F*wpJK?UyY)+DFdt( z>hdbppq0(MhbDRlhs~HVAu{l;XQJiyp8f$rK|U_xZq!`rVo;4r&g1oV(bi%FOlDth zY*gCI*Hk|c=z27&Iqe; zc6PGobL6%TQV~L2_lBt*&U?39VR7#UB%)nTD)>YKREcJbHO>g_(# zSk6UE4>lVCTYjhy1AuRG^c8PzxB#nRnfK_My=JkTlPs;8rA=jNomko#BhVgUX**fk zt1PVtOFQ>EtrUDW6c6oWJ%|~~Q5nlshY?l)szNL`H3f~ZaA;08Xo!VFn=uQGF@Vha zn=vrmNAGa=G4py~h93jV?qhh`qwE-T+%g94mW^K#OPj>fX0YQopQU|iggzItv{!GT zUBJ@5G6HQTOPj#bUSY3$A4~hrb=u+Qt?N% zT5moh(AKiFa+Wrer441T>wcZqtep|-qX2t^0NN8#PV2W3yL#9;8vfOkXroo#jp%i= zCYbfQ;d@-kUbTn4YBhV+vFugF*RMJ}tpiI-v$P2;Z7oa78G+U$5%nysNg@g@8|6Iq zx)PSwBoPtV$UvIDKEidyEUk$%-eCKjXW9MMu(T+wNv`C}(wZc3!3ftq&(fMCi5E+& zV6QuM1lnwt)`q?AUY0hMrS-Z_OE|-Pb|v{cX3p5YjZUOsfy*W2^!{kt9p7MEjU^4x zffLLz8#hmWc~15a*6G|2iyBS zZ11nIy`O%)_rufHv9wU^$cTrrw5GKvyH0D?o8G@!Z@PLp;m>pDQ*v^(y_r{X@>M0- zxfcq1TT?UBl2cN03UV%J#qz zWQ>WRw^yVv)HTI%yL)YHlsvUsF6oUXI90rF^6AjmOyz)iE4=CQjYtYn2216EGpA4W zmP+!I8V9@ED=Qo8)wzEaD+D?XEZ+SbyO8nY#)gnhQm)lX6oyU>VjafLa&l5arA#kE zaR5C>*ULdjq^(5P)!CyA7P$L+`ve3z)AfCTM{_`YvYoH2slmgQ#HhNud5(?n_x1=H zgE^Bc6neG6)gB%I{y;&2l4oVrqfs!j!G0YOKg41eCtZE3$x>%Nx9~Bu#GXRV;tZxN zp1~O7C7Qgrk(UX~-|1ifiw72?69Cdp&)ins`uC1vi zLAND^Wfe85+QOVGdD&SR8HgLm%gIHgGD$8k%c^JwWLR}pVnSk4T3TANXL4R0d?tnY z`T1pKDM{H`ATKOD`{$*U%Ng0Fu#GWc0~!g1%o$WPEGxt5v|p_ut`$91m5Pk4%jBs4I0^LJk!%xm+AUN~!(m~J@n-FI$IvVq2eOXm_wTC{*r zaSE9>ZQ7*qkuxG(Q8dCzAlJ9%3GYMV*n0A`fA3z=;q|{BJag{cxjd2Qn8)%@sVNN; z3cpelmuf|rOiRn%cQ0JH(6tR2r=K4e-yJaVU;p~od^swp5G)R@z`BWBFI`|h?jHODz{@hh*q;`2kaVT$2>X@aWzc%_rm!x z%b$Pmxe$+@zm8|u>x43EzekWKwOF2?R@_vdck$vyRiD#}1z4dw;pW(fS3Jcpj29>E zf9JiYsi};cUa#&@_yz=rj0qnL1iq4@YT4vxpM7?kpjUltE_gua&Dl;=`5%OLGj;pV4ZEbFAtgor6ZtJExd(;|DTU&c)cNg%YnwpxM z8`bqC1qB7QYMzj%#|Y~@b-+VuY-t0KNn@i{-=hY3Tbo$Q(bAfpCaMnnCJ^^hs0{8E z80hB*v<0D5OuS+o$}19xD>G-$iH?ejj*gxi9VHS9xN#w3VRx@0boT7ov!d^co*Wh! z7#a$~Ab(UspEz-*Oe2s=AXzj$R~zEzNRX%{8|amYCHPM&l{+~)+Pm0WNrWP62R~~Y z2M;e9-O|eUcH@bFn&aZ{C?eC(cMJ#)@B+pFDwo^R5{cXj#ULdT0JS(O5CmW~V_c+% zuQM{vCoz;1sJUh?1TuD9%PO}ME7?zwe$FVzE~!2R%zr2+2J}Cb$0dx*Q8V zcsqar#ak_U_IKyMVwQFWOWViNs#x0I>$Jm5N+Z~@hR8?MreQ#Xmx(4xNp?+wbU|u( zJq(FOqMJf;`v*&|m6q43N(z#T`g_u@U4rzua5+6cI~^6hQqHDj;ls6S1*pS{q{_N7 zy9KKRHpyhrbL4h5BG?;#yEb4m8}^1L zCh$5_zJjZFGYrSauy%K2;qAd@xd5McP+PDR4`9>$g7@xAufP8KTq+ipp1dgfm?7uK!A%( z;N$A*9T*rm0XdT7;fYKNepeSy1QZo{SQxJtT-cmnoRM7(TSR|Xr@9tWM8`qE9lIBr zZQ(cD6O4xi)pj}`8es971aAhF7Yx zvq!-t6b%=@g`5(u4*vGnQE&_ODue$lV;Qb{lA+=Mif^I}s}LEq>L0LAxT>vxwtAvRj(>6EYGgv zYSOP=y^?gTu(l{A@k&xkLPBN%YAz#3vanR8szx1q=r2bOoi8fRN~d;{1yY^Y)5Am@ zK3rJX+Xx|Yt)RA>=5e`Ah1G495I9+-)s21JBZ>V3u*Yi#wF8j&{r!nYl+&xbOHQ4< zbTl}4|K|0(@*4B5o!Wo!@WtZtMoD1Uct0seoq6=&zLV*}!Q;Z3Pwv>TY4=W~SRdH) z%ML0DVulWX?KS1t4I6S1iVE>f$t2W<4F}8hUJtHZxpI+{UUhr}YR)?wFS`8Mr=NcM z?lEwaE}(t{O79==C+x)_9HFl!l4SP1)GrA2`U8LYk^j+~@4x%@yYQ^=BIy&KBB$(Q z-q%UGajRnB)7l09`BveI#H17FMGK#a0oL$Nn5mVVmmdSv$hT0DexrVl1E=m$OC^5 zIiPrQ*L_1(DW~^dK2_8Tw5W=T(wfT$(>Z+4i4oy;{am}?2@&Ih?C9S1ww9*$!HUMp z%9_T;Zs=k5j{KJ9o<;$z;m0O-+}wAizX&s#A@fQYzu}sWB5glo8YM4ReG7 zcrClgoSmH9 zP!mh8<64J#IQmQt39_|8`P%VB#7Pt#9gHvXwnGH{5OOVb2)V?bt#IVFv|_{7>bYEf zzZKQm9VZc5scI@KEAtB|E7tmGJ_n1uQA^GH#yPlsJ8e%v>e&;ihxhSB?BRV3k$be` z44^C&5(`TpYhjsTwy@lZ(sw)YoJ}@TzHFqt6SEAcFOUc zyY`&ER#Ybq88<0FN_8aR9J~Lfjo5wT#vJ0FQ+Md)QtZQfOLaa=u@5hC)>R!O`|!6o zvK|0$|BMx|%d`*wihcMO?9T_N4L0wud}rmWD?f=aoc?NE+;`XX;cu{7_ixk&&NDy| z0IzGhrBbp3ixqKv2N$Y)z~l#b)z46|#k~J+$Tx%_tZf^77pbYueJxGce>=2|$yxma zG<-7sogF$_JD}kSG~K=0!NIN;AY1|j!aYPTSGorUNu9AtPYL&w>9stbTd0?{8x)UF zKQA{0PZ=z1>UQvPaz$#NPneIkUE~$y7d$4!B_Lu#XpryN2tO|`A74KoA8H!mOp&a! zQz;7x5eBoM=~%5bY|funeqPUXq=X zRgjl-=3HV*MtX8a9%N!+NpW^gdeX(q>4nwRMaUs%{Iy+8Re7l6m3scrwoPX*Bxm51 zSk}eU}UJIR5k4C&5K%xuzI$#fm1 zrR`D|TSdRBs*~sB$n7k;n%|{T4R}RPacIb|K)FQ5-|OMktL}4yhGp&6d2mY+mxB|t zV%AIVJ{Jy0T>rqB*Y4#Psp3N%?Co%zaQF9h5R0rm{M_9TR^;gB;bdoLXJhM(Y!EL; zKR+kAoSMZ5wc0^Ft*{ZtxeFC`Ha?1hPPEyh=Zoy%I)bZ6+tN_i0!@~c56dvKPh(~Y zk~7hYo+4V&-jd>dqnO<_7S3+CmCdOWqnP2?(6M7Fi)Y#Oxw68?|Jh6{|JhWBPc0~z zJc{X#rL(i+ESz%AibQD?_4{aMUTN{v8+V3Pn0X~;{%6em`kz}WrNhzrh8 zpF98Osei^(!u@g)PT@Qbso(iN?ha4Mk4L`!M?z(qV9~PsAAV%%z4zVo>@%-Czu09I zTu1Fd1e*NB@#Du2?caas=+UG5|9jX{!81+&*ZPe=U+Qk{Y;VYDZ0_?Lg&?AOqYQ62 z{4oglNPR7=g%y`hpbq0`WRuzYaBHMU4fYcp@I0^utbyf!1q$G@-(AR?7$WfR>dcNWDs3A8D&xQxNoyr9f40}@s<;6sXxj>4=Ru0TI;Xom$vl_<(e+3{O4(55e5N=qtE+<7P`?HW-O zlJJ^~M->b;h=(wAG!&&nF(h#zO8btE+!|h+Ork=@^PwH? zO4UdriIrwN-F)FIrO){S7_L*#?*HwZ|8C#9dHY8nZT)%I_VvHab}BoybLY--&zOJD z!=dOe<1?1vNg*mk4K+>|$CYpz8c0+~Tery^6U)6%)(D#^8y2$WaI@6{qdI6WtEs8! z;Mm&Vy>wboSOmgCCWkCsKHt$RFbI|tPdhmdIPLAkIf!BClQ=t*u#l{a_4P`7YwO7& zFvb8ZQox5XMk}^=bV682xE;!rxcHBquxN&-kDq5ykhi;I_bJo)?}N|21P{DTnb6-;pg&$af9cZio43NEf9_hI#XR9577OY2c5?b< zk=-rh5#fniC%ZAbzWE{s(OB=|`R}mU_22HWoO7(i;)aapp3699#&oxgmH1u-za857 z38~Zb73ZzTUm-f@djJ1CJ89G`=GfI0vfRsS`Kx9m_ffOl)Jc;f#z)SXG2`JEo}V>q zW^|{KKZ*=P#T-VR?L-&rmatr4eh> zJv{t5kC{7hyt|ukfaPIxT%(1?`6dgE<#BY6gZq|&`&NSc9tZb5 z2JQ<>%Fn-Oxr`&gUKZ|sVQq85zqee9(eP1Nv;_wAhtKW*9a(BxHu= z(g-QomTrn+ZW39L*9S4z-(s%+i@Bb2H6<-QJ>wD{S76Xe zxp?LB<>d5??9{YtNf+S1P8^93L^l+mIIdpH;V}Lvu4SN*gbp7FpM%ek!W;w2npmVB z@l^^hJ9{K?d?5Yys{^?MlV;4-=KcjR_jv|6t(y{M_$cD|nWPg34U=YuNW00FQZTIU4V`=OR`42zAK!lTkA!y zpLy!X>}it%P##UDX`@EHm(H0wX~OvN(As9qm^Nj~kyxTvgr zBxzw57Ke=EYsqOaQf6giYpsw;#4?F!q_uBe`p#(H^3^Q{*NDJ1#0r5-68uAq8~pQx z2(ZZZ_$Im3E7gbkR z)fVId^R%=CR)L&~s_H6u0xJ;qlbM{H`FH#%yV;0vn(?rfaUoO1!0&Xxd6#$X0(lJZ zIDDy#)5abF$2)?)4D6Jp5DwWU|Rqis5xIM#x85t5X}e5QRQlK7Q*hgW;)~wk|D~tL5|KXrah5<`%s7=KcQi zmV+cX>jT5PLMBmSH|D^86aind%X31b*oLJ%EJ^EcWYW~cKasF;@ z?k<7Buvs`cd*k8h?&j`+ASvI#z>x5Wv46K#WH(z*S{TL+dniZK)>Yd~f(tcJtZ7=H z)eovWyLz>8);gV~yX(}L*}*P$PCh=Kfqu4JaZ-{qYQrcLaw}Uqdk0S+XIt+v9)GvW zWH%dNTo_45uL0YAUrTdYQ=B5Cqq`d&L*)-cPitGh(pulwVbL;VEi5D(d1w|)+7)a! zv2=b5a$llyiS3JO#>YZ$>^lqHTOJhN zq_otOl=RF@qp$EvGOT%(gFLlP4|^70IN2{gtckWqu&te3ZY$}5QS93B!?Wgp_k8kfP5V}P~M zKQU33eDJjomQj-!#)%^k8+v<85sYV2N=j~xV3d8nw^;A}7#e&i@}9BrVQnRdb1f?? z$;&EhXhQY8!tC7K{Ni%tS!QQu=9N_=jJC0zE2kvxJW9nNrB1@B%&m{}tVzo+ZyTV2 zw%*xLQIUxeDNinwQwD?<2H!W+D?TiU)6nNQX2E?k{Qcp{^$GBIk;$Z1_6T5fKr~{I zpPRM7ZcKzSHTr*Xc@>pkhP1_4|8!~g`kD8>C?K^kRv8_UsZWWg+o%(QA zVl9HWl%wJo&Wi_oqNrW}jBVD8l5N7v5m?kzkPSrRqJpfnB9w%xtH{Ml$tJPQWmsrg zfI2R(P&FIGT7fe|)i7%;W}S%XGpnmd$3AN;XCGGb46o<96aI$K)`5Qz-BId|&wZOEU-sb?u6$NK_8Pv4`W zUq#eT3~l{)vsB4mxs$!JX|#gaD?fMr%EQmngj+^yBirkImezIzTCzD{mL}dZS}KS|b zm_*KQzAAEddo||fvzq|&^!DtyzHM|6MfVfz0@`;HIj3q>1nX87W5&82-WRALObI32 zpKvAd1-=GQ?n&e?G~rCV3kYkPe1UiBcds!wZ;bKkSB+N}s7P*}dyM&%JUoAxR&fwJ z$EH=>$d2@lJ!E*=R(6hsENuWg!>0X*e|@Bfr!8P<3FZ}8EP$JL?L*p1lBl zFflVZF#*a$a&BH~Ms8M0BF0vjn3$QJn3&rd=Y<5LzP_HWuBIjwGS>D(Rja<5kapz~ z)Y)@?e))5v(y=!Ej}s?O{CVNRUzdteZiMUVDe?7EZH*z7qVfF1X5X<&8#qC1+S=>R zZ1;&!V8)O{q@nF_az+5EwzY1cUzM2|8ag4uuJzANn>Ky*L$>|IWzd6{J}}3nPZ{Vq zW-*d*o_lGjtCj>P$*YT3imJu#+(hq1j z5*hSCYoCai1(O2;va%`-;~)ChTq>SCWviD4a;kR6$HyN?J(V3Frjb~qCnAZpy-BA7D=-*@LMvw(n9`BuPeqSLl=}QzOTX88c>_x3{~yPjKX{2xnU_ zcgKJL!lywsr+4nVS`7ufS0_T*KAP6)2722X)Hue~CGFaJfke}CgziozFZErR>qVIB zWti)E6peoz%2r_$He;MeXv=`MuC~1)|BoFT<6-TVZTJQ`9L1GIrD_G0$lzegTjqFUY?5(7%d}t>z#MseRsu+za&()Ds98OIdx?QK}&Z_ zP1U)J4gU8&gCJ-n@1e!7J^%4Mm_mM|E+VL4142xmW$gO;dVAF${~LwHSH1lGA9c84 zT1z5~Oiy}8Cd$Wi9(U_Q7d4gHPcVi=YEQxptYuO=1^?snrIKN0_(5t*p7VAg{9M>b~#R5{J=i(lxQ$b{@Hwm6BFxfbr-(#+SHBMw%9+ zG1#vW^zQiW@R@|&`#xL0t)xt^X}q}Qn;65}s3-QZ;oqFS2lvlIQ2BeP(eZ@gc{&!2 zKgz@kSFAwNo<;K&jjUFsT|G5@taVSb+!g?7V-pP9$n?#4F%zWo_>D5{tgP)3C}?SO-L68L}2n zUM#CAt?09+KC^5p{f}HwtIDfddIxLjE~ghcxlErJU?{kDaQo>$R80-Xj;;Uxr@yH6 zqnbI0K)011lLK4^EAsmcsuopg@u|P6JQgf_U;HSO;N$sx|*iOj$W$Qk{O#i#oob1QgN*^?~<#X z6s3`DIDp?N+#R6IJE#iP3Y)i=Uvb}9x$@=tR2-m@ey5Uw#)|mt(5|ji>AD+h2h=UyH5`M0 zFS98wd0+|Z^mq+jZ&I87p6g_OpTZ{DH0I}VdG;KKXCs&i=rdP zg-1kAY4_R@04j(^w z__TSsl@uYdo)Zf^VT-2Gd&r8b?a^v_`^+m&*Bq@sqEC#a{yVw}9J1z?BL5+lxBR*1 zZd-Ozzi3)OwZ~HT{*(C|vh)hllFExR%}cMcdQv1Z94wkY{~@bS<{1L2KCB_XSC z(v$%Wpn5Hzy*r(!DnFyS6z#Jfr*^t^6%CkPaW7)-Xw^cb2^pm=@u`qnw#4IU_Nm0u zz@tQ51^jnk;ln3v8QW+ofu0N$K{2rk5o^nQvp0{N2+e0n95wNoL*XifG9KQ??Rf2{ zJ|?=o0Uz}sykQMABR~-x4&u*=16)>E3|ioEPgIe%{7n~|0W zBV9>edh+i-A4xB`c>Yw<)yrwAmoBDcXJ=ncxt5U)|5XVN6H}msM?MD!8?^+qEDuzPqcTrcnd+ji*W3ypiFG0rDA5GV^5{&vl4MBQmXOh0|<` zP9RFQ!ZZ4#7$~&HC|St4*GFtb2stC@PrawLk%N| znrKgb_~D9~0+l&pmLI#joErb3a++0LEzktAFXUNyjvF^^+>+PswX+$32cm;xWjplS}zN4#X@dM)k?qI_W54nF1hE6VV@NgF4 zd{S}cT5A_$*Dtr>dw4k8h(#o(mSgAR72u(q7Qk}~x3TrIw?TG>qie*xXO~6#JA>LM zeB!vdQDF}D-Y5aYnHK6~#YI^OscYDzg)t`n-wN*g1>8q+R=0rrHiP@-oJ+iR zEh#baG>)m?Z`_q|{?zf&@uTz>hMY4ru5olMD!z$F=M0T>939t6Z(-ZvVjurEPIlb7 zU6mHLU%X`!(UZ-81C6B#-2}#==QsuNzKiFH{$ZU9V%sv8&~!A%+&#JZh?)(YBla-Mo&X$V!9@GBx8{!r25?q52)rQ}LpPDwrB zycxb>pu4@ZFV3~8p3k?rYxU~YPfZrKUjro2!45Bf|8wVDoTTw`Z&@>3=5+%KdzYCH z-yMy-kX5|x)wUDgMnNiI1mVeBF-~>?mZ}fLpZCk3Tl&cSaKNJ@;8o$^6!5eI7 zI7f_4_ZXkAaCbCWFOM-5kx+`f0FCt6W6mN&d_ZqU9bC^<)paP9p`*H68k>z^MJ+Ar zp1}bams(n(GVu`@KYq1b?N=tH?H9r8#;d<4)41k}JFM;UCONLg~> zV0UkSU$2I**J}IK-NaGO;|h3ufmjWQ9t5wT%&jL)4|by!Jwh934woxN)n97~XQ00k zr>8EIdf}_pgM&KSwBLPz8GR2kx(PE%auin+ZR^KUwG)c2%V2RD=uc!q(0PXgUu>QD=fuES>ggRzBf=eOc!Y9l3b^#EFEWYuT3aK|Z74`&$(pIB?+T zNmSKvahV(iBNA@b66Wyi5Tb^F^Vmntl`L(>^pRm#7-~e|iUoLa9v9_K%Jyb8_(@K@bizm^Ht@U6@8X6ABP#oy;7zFwu6I5cPA9yv@NK#H5zV=1UE*4 z8}9`-MuHn7!Hq6Cr3gpMO*r@GAAcM>otStnKP^2cGdnkr+Qm?*N+?5Di8lZ--;5wF z2LdHOU+Hp)fEMgNbKvmVgbSBaGLn}pdooPae)SO5$xus_VlOWlfho|-gBih;#Sj@U zJfI{n@&uA#5g2DPwG02&0diw2wZjM+bci}6$i8x}*={`S{6EA}?=hsbKxb!JeH^E= zjlUOO9mKLRpqqDWOub09paKeUWlK&^ zG^^Rg#o5Y`DT|gp5_9*G2jCMh4lT42BEj_gfi2 zVq$=?+1X>2(s3@$88uxJH$!br_I&CvHipyG2czH&eo@`Z=P+0jz`@fWe~7HHe=TRY zfQP>$x9|=W1qBr;!`nL=+B@2kuatGPcMSAvTJOX`Rh4ywizEav9=UiVCxPJtK>k*) zl9;#(d7C_CxP!ReARJJ2_XCMVVkJjz;_#d%6g>G9=2iksc|&Lza=>OzWw^kYzm@wW zChjw~ZREgZ-~xhP@d>y9&?QFniqL|R+S;m;k}N>R9X^;)Sd^KTU0PS1lXorw)gw<{ zxNt4+T547XLbj;i7~7#|qEtqbn3$D`GTb+>?E)pj>LbEiPW`&^`#ryH*|PKC-W|Ul z{xu|c?}0z}C8Z=MrDmlpS^7k{pdEDo563=<%aX1;)@4%BvtO=GEdlr!1lzr8q6t%Pvm(>lC)6!(p8{Wr^lWf18 zm~pd}X8P^xel#3#ix6;t5U`vhrcYh`*rW6AdEmjA`3oPob}3mp8V)gD*DM_zARR2_ zm|H}}Tvk*FKls06qgy1101{*v%%+jpYI<9HeOr5L(&f_jwsvTaEu&$xTO^DFB#fnO zH~n^TGa5F$MfM0FdyGkeBeUgoIrId$lGqMO7T_{S$gy>`#U+5uC@O;8ueerKS5sMr z>UqWGwWPF2aS6)Im6lgx-H~{}Qe@%BK{goTpgqS4xdNd`gbc7a4!ppOh#u_g2b|cT zo}u(>$3nyS<@+BGZr{4)r{6ZXM@$qp7MweLct>TuILJR@&-N{AzfI_J@(yrbzh%n> z74;_*PwC@41|`Ag&ri7r-s$B_mXIxvV~C0}G@V+%etnjI6vK^(bmJ)LTmEY%u~R{erdjt04L8$!b~Ht8)>!{}c5af4;aN^=fhq?1aA}M{A?tmE}+R*5+r- z1IXA>)V8=pZH}eh!5x*Bw#IQfZq-v()BE$|I93WZL#r3h$MyCYpGT1^9ma6NX86XB zg${#9E3iMCn;I)BFiiE04UJ9p73CFGRn3j~V^h7Vs-n83xuuCz#xagl9HP?WggWRp zC`GBqXwn8oWDo(DQ;4x-VE*LBkF~OKadL8TcXK0PL{`=c8+#`gXID2UyErwVES{^A zlY>2wIq}iWg$lVt%h>uv20q#a#wLfMIBb5yWRn&d11`9sMW&oTpORhQ(9+i0(^gws zl37%go1dGV4gJHZmw3JbuV_-q5);*JCY5aN{mIGqFPt(jI&#AJ@Ic?-Uw+(lxsek( zm*|m)mEx+Z9^($&UC0QMkKxF<<)D%-%!&pC^C2nldCNN6OY$?XB_$D)>=Eh=^(BAB zUD66{5>%*?Yh~rbRI=Y=bt35t|(Ru`H=2eW={a7=^ZyD6qLeo5Dv}yK>>|S*Y6=&z%Ai%ISopbak_F6fwH-&UT7H zIdm0uqiSH#2Vv9-k*2q$AbF=U#_%Rem_BJ(Ci&{%l{SIE?yIjzX;z=Assa3-?karv zF`)Xs^wP`JATA^FiGDcwuyLRduXPL=;}tQ($HUdxLChVL2y|Tyl=aMK|8qAuzm#gD z>QK#5fGRJ?sEypm!UMgcDIKMYjiN5Ye`948630_m>kof1pT&q^UunHIjxP~b9*zO< z7?}=C2>np#+90_Vbof?P8^GkWwc7q(&A{NG-iEIS76hWoIeeM5Lhj@Q%+**QD=Si~ z8-hcN=oui-4pMx4=hFi{QU~K)N!A9ncX2hf#W{_90goGW*L)uXSKu&xVuY`~!d_u5 z<@0InU_TryHcYIoy*)0e?d|UCX;m8^9A2EXt+wBbwldD5wJDnLHO%7&n8yv6$Ja2A zuVNk}n*cc5OxGpfEwG^`q zqeSPOXIZ>v*SRJ#%WIfVTdBf21=4g8DEz7GWgIUt=*M2Gw24NE;iO z+-;wLL>iHV>1LMxt)%rGF#{f-hbjtYggdPL@8B+mW-m-Y2F@TCfq?)Sf*8ShB1 z+hwepUCN$03)nNKCwu06s!6_YDZyJ0aIYBmL~146xrM*Y9L_ z+5whU&HlQFrPZ+C1S8NISlUkZ*VAvgZs&DcDeielW*UWh!Y3Jq3POCcH)(*4S}vb~ zM%buj9_a+Cy5R@(AQ}VoH>)9tF(YQ5{!Muh2n#|TM>I%xrVy5&v68U7CWQpb4+bMp zp5?q@d&AW=8+&6Qy%8I)+l`OqdSh>Z%*y3C8vja@-nbik)A@JSuCeUM*nzGgw*?`|Dzs*2ERwBhZ>^crIdT?_+5b*>CCBX>X1L!K|99y9|;gyf&7* zh}=|3lw?v9%aF@Pui=me#rvbndjHtmm!s-)cd zdK?EC&b&ZL91jjC-%zE(&OIn#a?}KCJ4c0w7d3|b-|4P#JSaaC?>1=`ce>}MlVne` z&ZL{*I9o}Cjng$6;Z;6k@M@Iv!;L=A(5MBtMJ%O{=jj!JR24op#q$ija_W72^pEFh z?F_~raN>Dfe*pk6Vkr*0gZ=ZK49~lk?R~@3TDIGe5g;*a(wp|pSg!8mG2>phE|%w= z(nW2?w#Y4Cd9QK5!{2!D%WlNqcy#a>>LJsf8Hknw&{70i@&NyMfPeUdea8=Wbw=Z) z_B8$L-e|)L!;7&|hNVobSglS^PfSctSF7XsuA+XAFjo>Q_SyGyajpZvXe@~#7O!9b z`s)V6>#wh0AI~q63O`SzJ3!XP$-g2P>Nu?VeoE&E)on@$9_@4$`WH~ z-dGM7%*pB^*u#u5z0YAJW@049V z23I_T4ATt{(m($kO}=4}DTZ;@)DZsDg9||6uqCZ zZ)j*zb9hqtd(WRg-_fTJ2@ZiBnYUz)o3>${Pc8DOYf0(aN>o^`sHjv`*Hu=YJPk`) zO|4PuLLm;G!2nChI*J#E2#;uk%pi-!EPUo;YimOs(1@9&)xHFoPgBHeNPZiG;>}SDJ8Zn7F^4T*s+75 zJ)Gj{`8Dy>oAa^Wj~LQnk7=a7Fe~r$sRyGe2&P!114J2C)2mi7baHY${r%*4>$Wbz z(G$I0gvZizE6Ona9LKOQhW7G`r#%AWt)1Nh;L0L(P@ENZ&MwqghND*~7?@q+DIL3l z@4_7|#vQ$cJDP_(nu|N~9qO0J&pwxYB|W>gIJ3CCFfD`H!?+FgKbY>CLZw`C;|@b> zMeP1B!;dz+WLPaw1_k}{3pX5~scQ=Qb0vDCZ6`kah`aJLPqZ+>M#K`{mtA)HdVwx+tdvb|e)?4+g} z_T1)-yvlOw0%Jea(0}u=``TN}6?&dcxTllvw9sHXTNig1_dwWhoosAeT&$4oM2$0f zljJvflWw1x;klg;_QFao0{7r;BACLsuCegR@*ayfaz>PZQEy+}#+YXPKBEPpj3uvN z$vueNAi-iFXQjg(=^I#jHY2T4h#(?C4Ut;5gE=C^-M7q z>KS|WJLwh^_mQ*sL?h|@M$&EP$w|C-tobDF4{i&k(XoGoOAM4BZ${ z|LGCW;g?*BLjAzRFPoyF@g1NuGEl#g=cgl4iP#TN`cWlB8#a!R~)cKR8GG$kF? zH1fMzs%psk#2R5}lVq@Vv=h5IJ34!MIy!p0IXaN_$?581Xf8LN;<7X8H=M*N$Vo`K znvp{`(~6?>OlmJ9zHROOe^%(QGi=T$kYS$8HVl3#;#g4gd>wxWf>WUMuQM7ixJgM+ zKGyM@5HLdi$ajW^7UzI*l&cG9oykoSPM&Q#s1pB(mJr; zJVv0USy~)k9ZpZ&(aP8(|C(rLK+P{`a+$Xth-s)qXd(z3G4$4-=#lor&QMgt`r zMKmKusG8A`%oDG3xX>2h6S8B=_n*hce)i?|Z3j)mgowyhU_IQSxG# zfKH*D&UQHEH^K384yw(U)N{lz#Evbut1EOlBVPWB8Nzp+LyZc+IZcg?w4Dt+TkX}R z!RKl;eFjEsAV`nsvsSatfiJXoumT=Z; z;&6pxRE-nx`S{194FF=a5ns)yY^bfuIr87wd9ay34c+l6!y9~cRejo_?J-EWdJ^S* zme9(*bM;yPyFJV)Ey2dWfKe=2vgE#*L7r0F5D&q1(eNd1lR8DA*x6IFLxm;9#p|Ld zgUF@1psKREKVtcO>?^fYH`PqlQ#z@ai+kk#k3Rd-tFJG+e=aSd1+;(@Hin9#Z5bg@ z?$2Jjkl7kX7jKUS7}b2~*|R_G-tgUq-_B>M44?V-R9021>grp&Iy!s0JK8nvb@lai z&8;o%?aeqpb~QA@$}WUqd{Af5>f#i7UB4#4ZxBwE!G1(MFcfE~03NmQ5I#|QwtVk0&ow11lG(mr6+$qTZ{}^V3WGLT- z8MzBH5>SDz)zz2cQI5dX>Jl;)b@fd^-m5e^iA}s|!U1&{9tZSO+YXmdPcLcw)*nt+ z@S_&oE3CK*1kU>n&eo#2YZET#od4s`vl%Al9Sv~jI{r&j+ENQH%mYjRh6s~ys938v z9({1hqPrI_u**CCZJ*wE(E>!IRKm$s3}(O1=#2Z{!fk#UP+wbBSy@3`x((#q(%jJ4 zsH&_et*)x6Gv*VUmVpU~(qVXH%ufRxp}AXIIiu*kBXX8)f%xlWXQhxr$G0c$TvyW? z8KIdQdm+(V&30*_H-TH_(25JR0_TV~472I`u#CQ>AEvLutazU8!6ed@42sa-YXS|Y zjGBNOE1X2Es8X@tjK-qj&oVp;>S zI&Jz@)i_7z=2tf6rX}qY#xFS(oFC#8dp_GxjQ==CtZL9*_-GYAk>hqM^^u51s zJK910%5Vi8f1N@_5oGNNnGKu1{dxds9K5F`skvA0!Ad%eP(HFb`~qeAJH&8pB@%ab>rK`zaz>NkZi!Za zv#v2ykL#wR$}tTFB`2*9GW%urvlSl}nji7mPRVuaN9Da(l#}CjbE!7tCll|ZJ;{*v zMChk6+poF}Ii$vxZLHDSr}#`-8wXE-u{F1;$SNeCZf#A0-jNd|$s^=CyCO>6*qXHj zxRT^k6t#Rjr7ZT(N!!phfN4L&JR`OehS2?lgFwCzo9*Y;p zc-k52?^wLXr2O3a4Z}D(!Bo;d83>@#6Maw@1bcFDu)e-aN-SZSmWq%H~ z`I`RETQT$u8YV`P*uHe(LK^-L{2`hSr~PSn+L3mqohdzxEpkS3{(NJfWK2|46n`J7 z%!`m=tKxjbEw^d~;bUi?zY{->sp#)R=3Z5C?xjEfJc~dDU>Wug3=ULqxhIQwdaXci zU7lRprIzZ_seX)JH==Fg1Oi(ZZ*M1Y9B075R}N)gObIOz4E70qyyY?iSBHm0B9n?; z9l8V}Az%LgF!$YoQB~>N_fBugr1yk`R1!iBy`>R40wPv)#V)S8x^`D{XChcu0V#^0 zB1jcP>Ai%OMo1^UOxmPOdM`8IbLJ$GGAa1mKfa7+CLzx~_nz~f_q_eJv62|%mX3N| z88e0x*45QY6cQEk3f9Ub77iZn7Ipp5I^hX-^7USRVs_BT*Izf-8~OTI>PG(^KFq1TUA6>b}PJ-e~InI3c@}c!o#qVxbhKe@b|$zK$q`Wd#4Ot|GRO!8yUI|&p{YEy<&g&610RU>v*qb?OCuC8uvGgB~P{2?LiD6yU0Ft4$r(Zrbi!p^Q|ZdPK{Cot+) z7EoRp`l(P%PnC86Htt?Uv~CNMocEjbxUpE=nX>G%s0CKzcG z9bHnAejO@y%UuOlneE zdPc_WyyWPsSLe@Lv~aA2u<-oJqf;h^j4+g3I(T3wQ_ZO?8){oRA=?^SwS`rrM|*R# zO44P&=&iTjdSRvmC$1WWPmfO@9|b$wa!kedU_{>{hL2d(x8eH^WG!sQp?4Cm!|1?w z2}j>$CR$vQeyhqMV9|d_0O60!Z=!<4IKBIXS3~i~+nGHu!+*)CtP;{HNu{NZ%)Q4o z-FVkkwsY+6&%$B!f5Q0s7rW&*lwl4>8!`Hef3iPbj|wbAH}Jaha`Xz%PfE_qDM45u z`9d<328MF-d&r+frDdd8CzHUb_$|$?-72BJy{WmIT`TG8B<)(;I=VPw#vOv*fQV0h%RAdTsrIoF%wVAmwdeGUK7?V{9ha+`3Wiki<0B7=1 zZf4F&@PQOqk}5c{oP5G>DgFHilt8l0kAM>10VRA4`kPT&QVq+#+`P4f1c>I9KOkk? z70T>@P-a*RhfFCm#?D^$(2UUe3#QKun}79uwBP|L=dO5W9^#pYPD=8SfbODZflHue zOLIvWs(qxQxDGMY{*MXluE=H{BAfp<#my+vWEW`)&z`yk?@M(RBIX~A0`CfO<{`v+ zC?q+mKJC=Z|(SiBzjkbGY=8Y4@$G7!W!XTBAig;}38Bw<@Y!gJvu+^*8gaYD#hKtX7ch=;+Qad_vdzXm#M#tF!0LpEq;N$WdO7E=aVuwN*NgfT@4iPdm=$Y89g-+cuS3cgY?Zh1WB&vFCBYH7j>)2Nx;h5LLA&Jsoe9JyRH7c;Wy0{oE zreTYz>Flhnt0+R0EtpfUZd|M>0`ICSFN5z(V`Nn_xsj zz=?%W?crA zA7J^%E?JV4^nfI=#&W16z=bi1c>N#gV9np9gRz;Jg@q4D3-DA9g%-x5PBuax9*7>+ zSUng$^jdiDuGgA}{S7%c5PR{nNY>uCQ}A;M=ooKpeHIAQXgx&-QFLuDnx*LWK6J1B z;vS0u_8DX&;scpf&B9-yBTfV21xX+RFAOEU77UW>IS#ZuAGACUv^);9?9hLP7=|~U z59ZK+_7`J1*DK_K(z%!c(zYD>Ptsycr)0q$(q94=aym*n%%haSSg#3C5 z3}P<~p+CcfbA!b5qsrj}rQ{i10J6Va00XS4Ut@DkZEZ(;yQQZ`XJ>~2w(9;8@uFi%SN6R-S2iIbp`o=L)feIu z5;M|L2Uyui5y?)5xfKZ^5~S=4`d!&aLT881nH4@4s~aOj0&f2+8!1}%Evhv9c7qc^ z^u#qyudF1YxXxhJU0#|JYFyou{I~q9v=6v1!AZV?*0Ywp*pJ@gNRAw`ZUfFd6KpD#0}Bk(nccG2u>5d0lC&~=*`*H&(a1>nvA4Nv6v^fFn4e=_v(LX#i$pB z-Wu-i9LZ`!hOOzPHCh8m_f8IbmQ-bU6MObe?AghQ$*BF55sxo4f=EK$&V+=d^o*?3 zw3J&hvH0qL5BE4S>`gl|`sJ)v4k61T=8Qjh^-`fY@a5Ov87;0o9|9{{u;-Xpe=pZ{ zYKw|XTl=Slq4+$+JWxpSu;%VP)W-DoA#f}aZ4zA7~{jcv`sw7R+Hzh(Cd5US<+-_o#yC;jo zxyLqlr_3SllR46;%rTbA915DMqN7qrYoF9%8rw~=dWxMzu@Z`H9|T)Tv5gd4OtIb+ z8$Jluhhj%kY%0ZGrP%yIu+0=ZnPN*QRz|UHec1l2t!=xB<#%cWMB$#CAW=3EghkH1 zmfv3FJ#sV~u$$OvA0x=V%$#4diKVv2c`!?D^=@LRt!rRmsm*5bYA~O60P{;(F zQ_iJ1my(mxbB>)jeeU#$Lno6jT|9d{H#(p`mIl+9AaYTosUUL`QzyEQ)fO~wk+YXh2aq|zwq3%&jd4@B9WG)n#+c; zqmc0@!OIHSexunC^<4c$eX71$--uVCK23j9e=*9ru&~2M(ON}H63Wc%?DS=Id}VgN zR;Q~(Sz(o=w$mvv&_`TTtCm>B`co(TbEn&-V#Kz{;cd|W}XyPX$71W=@43etB3owIT$r%U@DJslO zjk|X4J+m)z>~h-pQboe%j;I!?rJbX-R3++YsI6_*3qm2puS4ne7siN6cZSlc z?E>|$xuS`|%sRDKL#3t=5eh|>ORSy8rP|Bg3(I2R=;iC~He$kr38Q`7{XK{K1_lQD zx!F3Sj2@~wXuu>n<$u@)FIDc#Wy%XCPQUTgPN@2%S<# z^7I7GH&`G*o<#*U60XoY`oBSAZ{P}RaD~@#g||Rsd=yGK5P}L-AC1|)d-wkRzyA6w z-mVRW@_ynh&m<%!qPtjJoJJEJO?(_@hC+-JJ8N}ic^T;zP8L7y+H)i(JHK@(G#BP$Xmidy z@zm}eJ9eypK$@P|SW{&aG6w-mq){y5nCx4JLggAxB4~jzh!_IZR$ZvwAsPzR^WBDx zdODavZsh!{0pU<}k_@ruT+B zrEfP%-$yBZ$I-pf&_|u7v56GxNwL)w+eoqUL9oRXTSBo{D0U>pJ~0T^hGJ(^>~j=* zmSWon!Ct0VD~i2DvE39KFbH-I#m=DEG>Yw{Sj$1MB@|nF*YgHY>|=dcB4pS#0*>{k{?ATjAU3v z@>h~cgMbi1W)1HGGHsDeSyUwQHIO75fggC(j%|9)C(`HWnfvv04zlTUNc*0{G`5gp zed!$Z%zXvLt{engO|b@w&8Fk>DR!{AucFvyigl(~ImPk@8Ml^V{VDb;#ad9TVi0W4 zxRw-~L9sRzI~bk*U+(#Ctj}G~s~qH>P42BT^q!mk57@RoY_BYRZ_Z@H79xcQrn8X3 zqsJ3-v{@yQ!kr>|rEt?!)x(9H=`8nfAsxNr!A|!ay1Ka(dxp+b5XC;%cgLp3{r?I( z_TGCo6?+)TFK)mC(@fWrKs+?fbiERd2d9~?Bh7Fj+7;^wCrm(JrmN{WNJQ0o2vMFi zF~1tQefX)7(U3PeGZSz9*FJQoST~A2O0jVi+b{^WXI$sI#??`*Z69_wxakhaR3{-* z{S0on>n@pU)&t5^H()i);tG03K*v&n<16ON8v zW2aA_9^~R8sHmtYNWs0PYBXJr>i@IU=9rUJ#kaL}bhI~Dy4l&v+R6$vg}LSBS)hXd zt2Ad1OK>YBGJabdO6T!Wc~>MtcdoWp^{l@o=D~W{)Qs+C4Yl>=_|)8j|5{o)+FL+w zgUNsXCbS-f!i+eaVG~5@EM!_j0m8dNq3)FnXFaG~_{h?Q^XJbE4-K6#C2a1(#ml4* zKfL1c$6k2uaqNVF1jf<*X;Y~1TmjXpT%&2i3>vy1Z|Zrx+QJ?!t6vWOlaYhx8aen> zBL~+QIrz+84lcotB+<3Rj-tShgnR?}1nGubxHST{&I_6n@pd3IT2C1H8iB#00m>38 zyFptbx%Yim4I!>{e-zsmVkt`}-^P<~mE>ClH;_D2Lr;J?^aS{vo&Y`P`(S5fk4z_~ zSVxLYr<{1O^ZhCvS4y!xG6`Y6;G7P9Cl7hbdvoYi6uE#RE9j`lDRQv$j-}Ye6q`)3 zmnl}zH>zn~-$t?1?i#n}Y5y8`fID76k+F2tXgaFvAfxu^-|tfFPKs4htn(mPJ-uTa zifyOZN{W^AVTl!eIcSp9RwHqSBnsN^W@23IQ1t9A9KTSbDIAJs?PuwEc?>2!{N zbNu+5Ls6d{t1IWue^7;aJSP`+b|U!af$Tnh=teRmt+8k%}TfcZ)!GZ-egU z0yP&X4^fv?-q0#nv~`#(n@bCd3i7kEi!3awbJMf3b23Z1t?ks2(&*^eSPcsCcXX(% zS}Qeqx2|4^hU}1(UsBT!J6sgU`|iNk!G?eFHtt4QgI66l?CRiIyEr*n3)_>634*&0T85{q5WG?Vb!>PSUA;`}U!w zXKYT){(VPxYU<=3NQRWPRV5zXwQE-la-dX-&6|^mO<7MGno3kQ2EP|ygk^bme23*s z7?z(LC#aA6T8#kH2xcF%6R-X7^6q2468!sRWGK!=n&B}@ih2!r>=1cyw?u{2i zX^lt0+7m4<%Y&I8)guZEwZ-M}N0S9o&&iV~k8)Q^GGj_xt1Crv2X9!4odgKGhCtR0 zH;>Aztn6wke7;O~L#3`A6XZWZE@u=9en(wRC2C@J2*ieNg?faAh25}Ww$0TdXFn9q zvlHosTAjIzy@Q97le>?P&p5PUC0}mL6i#LB+8#_E582Qwr$%kFURB#CiU@gc=8X_qK^7uT~#>UCh)3Loy zHf$ssGC0Y2(&~CEcV`!GH#aYTfB%Uf3Gy|Z8P4&EMg0M2;zQ8H`=E(GK@%Tw|0CvNs;0ZKzQ*66 zxoP(o;T|#tI&+ry7v~r%Jx;s74)+5vD_S4^&R$VNBh`VtXSLK$V?`2XBYBqK;>Od> znY`@+gB-}lV?nWw6k9>DVxvQz_iG>4w6un1j2v+I7D~P$;DG~za@t3*tBB10B6bzY zTr{4gR}y10qj&$2o{(*`Gx^??UGwt(E!pE2I7WSWTa86@;KqfMh zH2qeHM77O084T@gS4X#suMnep2t7AZ;ZIS~+Y?am*&kKP2;;dqeeA@M^PZh;Yt@EQ zShYMW+xbt;2pBbfy4^5eKO74lj@H696qaL22PFi&kM+iM=ovFdd(!#u777h>N`*qq z=Zi_xzb@3Kw#C8ei(Y=C7R?Vt?=&kbN3T)i76f}ig!e^>9|ug312TWemynr+i`Lc2 z#5gazB|@1(f~&W5IXI{lD4Ei7e!u;u++a>$E5&CShwc+tV{zx?lyKdfKBb=Obdeg6PtV1i2yl`?|9 z|LLnO`R)^*`S8WZXN`_{>E(~!eQ;Xomks_6BAVtq*3U0s?AU2=&Oi0^qM-59$36gA znP7uMrnG6p-JD$%R>P)^^d2*3@~InTKzI8vlhedvZ+7 z&4kzs(b4 zEhVo2YTz(Zsk^q8R8JvwxCO|zVH%3l;=pyZ%~@Fpu(4@uyj_A$Ee&Nwc^P0nWpzdA zrCn+t|omw`{Q=G1;rR@N!C< zt&^K&YkBUe?>2n3m2@~&l3Lx*K6~qEZ#@=HE|nYVTgaWHZr-f6S-pCU0Mh>Nnr??B z;82fwbm#2>w^0C^{kVf zif=BhtyLrIsIgT>Vv2thyyb7m*wicjZ!a(mUrpE;2ye{Tpjp$VEq~#;88fDb1X-9j z=NFZBDlNkw9`Eiqa@w?UK|XVy2{E&A^YSKKP1;ylU4wEC&YUYP)J+P-&`L4jKn<*g z6znbKT1AE0#>{M-uK>zH7hh@skE?Crn^`E-lPxTUxjT3S1O$gHnHuOl$z`Oyxs|1v z+-dCmsSXZauEU3W472c=JJ;3$K|A1jHbS16m91QY27IpO=61vD>nBg{tvJ)rz>~A( z)!=;{k`W`+g8F)eq=8ciMEq{T0?RA;4%YNU3k1!L2F;Ov2mzqE0MMLmY0j-%H?Hs6 zxqI`v4I8&^{w)Um^hmt`OQf^LkX+VF7vqp1kdT#|NGhG}+xyFJ7j8y>`_<0AhSKs% zWIC&q<>mD-z#AKC)zlQ_S4Y`eSU92O+o-UxhgLkca>c__-K=dKn1%RLG|a;h(IvD% z1FnwB0yv_HuDl4du>`a6CT4>~T#+8%-Ws&pDa*;ty?W$Wc6Lq%>Q9)Q9hwZ`@>enh zq~rT$%lqB}M*Ym*eFcsBnNC{fXdX`O`^p^go*mj$-&X=@pt<+Hw>nZkYjIz-Bom9{ zeQ*7weirM#dP@DR-o5r?nLDifyed#GL`E~vB;LAd**05}+iwhUa9PNOx% zO$-{=SyFt1c2~cD$QZ-rdqv)7ow|dg|mL;})o^YllCG*ICyIPnF$pryiMz z^toS!S$GMvuo!!Q^!)I-dIe^o3m4A+w&}a`=P#VOaIK$dp_@R?!fn&j`@z&a3AS8q zs%u+>iFnhn8j0hN7*?Q6)GBm)c-!!S;fwxfu{SyBa9@W96*4?c%&XuD%aWS>MKA_l zhhgvtY(;yRi^wHB!|Y~0L&3*>XPxc_*{YZN`ZcJ;{jY-$%R^ZzRB~||I|^Bzaxv4- z!t`bX5swn)6RjCkh`qdLseDo`MtF@FKWS?4^!anAO$fEi?Pu+=T4JK$voHb&xXSSM_wu>3T4`a@Ym@(`BqVk`L!d5 zF69Rq-IKi(P zzsMBec#Dm)84>93;4sRjDmkgN+CE5n^S8ab5ZoOpf$ok_8U6zfQ__I+3p=+Q*i2-dKkc=CO} zV@XB?Es$!gg(F+o>&Z7AUt~~VSR@uB($JSYGA)Ip`4q$V)Hb_{TZm*xDI6KL0X~LP ziLdug?#-|mx1MBJ%qDk2GAwAtRbwrjhK0Qu79_)A3U*ZpcGVQ@swvo2$Z*Jtk4sKT ziHpn5&92KWEiNd~C1>YUC0#-t>PTw{&++M~^N0xAyu7^PqGXtLIaZ^Ia`yEtyu5Gs z-hGGTs)~}djU&g5cQ8BpOL)kkgG>y7=rS#__wXG*b?Q{qgAxp%_S$Q&k?`&)>l-%~ zE_m_NkKS7mG&=lgVZx@bKii#8WbR*un>N1lA+sDz1mG@h@`*ir_wJoFYgT76!s|&l z+NU@cWzwCCibNJDxoz#_C@<5thz)ooSduET8d2sFBjc1X=(|}4E!#FaE zFcx>~YisMe+VbrkZ0vBf+1NTc+f0~f%~((%l%n>=va&M5po=wVQN^+C?R4fRWA-Lu zs8Gz_M9khq%$`$ePD)OPt_0N zMJrnL;XmSB3W2h{OE#hQkNrCn%qPvCyCj4`u!;0P&n;N6;k$pYCK(zuEn%JF$#GJc z&F0OUH*=(um}8pppV|pEMXJulCf>Sv^J=;FN;Y~2W+j}7Psqx`iK2oQEJg1*e97f< zoEB*7&BK3cC!`)Yb)soEdbK$^30+)WU7ZyjR_2awj2%#@45u`cAvOoy59DSRQVwbL zj5}_NXeh5?)>dQIdgJw9$E;1r&W9haqN=(q^JYe3Lj0Ap$F8SWmZe0WI(7VT%%KCj zcKm#1|M5SMUA&Q;c;x6WzZ}0DpO}gQ-`p5hZtv{rE0drXTSFVKv#Pu}KYIU(Yf$Zu zZ`$z9SNmgNzl^_hAr1m+k-$~SH$+XyEMsjQ9Go1tZaJD!Wf$n}%5M|E3}D05Rm7e? z^2dP#!-ut0-Q4}%&->4wi~fGi4||SXOhAJ})o6W3SEVlF?E3ZVf78gE$4{O-)~uqo z1Ct|Z>acPMd>UT8#SX1i331nNwK`acc~Rndu0?TZ7oQdaH{TDIFF!T1IV(9fyH=x# zJF(|dr}H!6@Ca^)aIuNmD9k8m@tpTgCbuI-^7W*kFE-F zmI~C7L5(`Jw5zPD(;D~+v?la$7E~wQ$f@dJh3@9I_QQs`dyJ7-3Tw)8b>{Y?r};Wt zDW!~@8!wZY%M5xOYkyMU!$HXtcXTsCv74nHlie&4=ql3E5bM-!rr;?=V!mG6j_O0| z33g5{wq^>MbhvMjt+%VvewhDoD}%tn&(md8fX&F!K@&zg2Dm!=1bREWj}8c(?B(DZ z;LnWZ#<#b(RB7_FGHz=Lqu}Z4J5c-7Y=oo@yVfmix)*A;LDB7&+Dk<|fuSt1M!;{U zJ#)SVMQs8_k!ZzD*oT`yQL`#5>g!tTDk|d6o=Yy)kk)SxMq}bWBC{k`sF5RyyvYL} zVbGQB@6es(qRw-e*ZyC3tRaahpH7eZ;>+#Zb{_eg)QFqX-m0rJc8c4!^?Pzte}g~; z%cL2F1@obkZG~VH1(D_pP~17@8YphN^gsXjU(>IDS()pKg$q}#TD^MpYj3>STw780 zcPM(6rZ^|J01X;=mKL2Al@&T|5%WMaF76}pS;W+3VPWhzXFxIl^qL1FdtndV&)QQg z;p_R9*0vo2hvCk)!)p&e*iI1lZG~A*PLi7BWUo=+_hSM_$}R1k9&B?6@7x`0u{$>OeXj*^Y^yIx%21Gox61Q#~*?XpUj9l z5mPSr95eiH?TCY%W3n8+UJ-kHov z^ADz5mZYXN{SWs7`#54<{>zD{hCTAan{U4UZZH$c26?Jn0+uaq`CI#;wyD0dyt24X z#H*@;3Rz#9|M&L8C2m%Qwz>f($hww}I#kJPu5Zx&&HZ3&Zy^RJH&-gm%q`3m=4Oh& zw;$Y5>=UI2{u|W#V4L91J#epGjb<6fL4tq+qN2pDYlkkU+u%1*2d6Pb~7|AqK)~HN2b~s+J}6IkEGJH9eH%BjX9FX z@6*PF;I{WiS8paRc=X8TZ~=UilW$`=szs34rg^Zi6(q#R#bjg@Bo#3mxPY}kpM_H_F*6_MVsTlm zt_C}gFB<7coX1TiB?Y-TIT@Mh$w-)pJFE_Lbhd8EjybUPhwZzsCLQ0j{)_!LwB|NH zZ2LVvr_RjT$w8{E>FLe4rt z=BFoIk4s`5Lm>XXFjg(RaVp1m#V7L-*K`_Z4l!JQiHdIPjd4yV1Ru?EtIY|8NE^c( zw0L=}qb}`sRs%c3xitT_*|axE+0!2)nU6W%<}(#4Z+&g{^>bI^Q`)U36BpaK$0CN| zdF1XBLuO8z6rilQ3Ssx;d39h}MRB#z%6rV{(UZJn9bKKBJf+j zOC4zC=_Hn#lgj+wKDN+%oZS4JZCve@3Of&T3F-<91qPl}CYSI!tb|zAfVqL@P&hjp zo4ZI3VwXS!<9buowVYZUxMFLBT?%-D_O1>*NuN%%5;X8SXn;gGlK8ikpn*|ksBl(FK&`(fMnpU*bD1WVrAhX0`8>8AqKXKSzj z8Ekl4_(L{NWPUu@@DiJHKG^WSVt=ED-x%ArR#2?*66CM zsH`7IP*vS`WI7#?NkI zKj`j?8U_6q{Tu9@q5dNO*~8rpYOpi0*P)Svje|XU2-w@ei(zZ$L~12Wc%X|Lb>-JV z7w>^Cz64#6+;*ZXPt3^&WuSCpE>YIA(RAbX?W}Bk%FZS2%L?*PCKN3>N>SvIbT%tN zGYCy?E_0M)Sk6cx=)u8*DFhwOy3obGy}e82z~gmwMTA2ykYpbhT z$aqWg~B-Q>V71x}hxP^5x{jGL?g!%l3^RnLRgCt5|bULun=7 zZ`!Q)KX^YxEqVXrg9i^@@AiM`Auy2bu)Xgm_6#;)~bWSliIebhR70JK8%8>WTJtXg_6UVdH>46OK+Ut}v}Q6Y6nvu(l)ZL2Rt8 z3DtqBEUjRDK`xZFgFT9u8%G1>qEUz*+nr^20yJsw8BM^Kkk9yEfO~-{%s^Hf-D!zE zhWHFaKw}Z0u|Uum@i5rJ1K6zK@kQP3`0MPP(z;FyGlXS};oXSN%*?M*N!pq^;KVVn z%7)MIb;GATaZ5uO+MPdxc=eCjZ@&nGcK}UW49^O;{FzjxMXdK?E|PuiH5GewYp~%{ z*_Q7&Z`2Fr(wg+k>JUUZujN%RWpfb(e~Y<|QqI4^%XJ)F;b+lXGXn+M8pz9=xX2L0 zyM}jF?Cdv4umpLnXV#0P25>gzD77t*?d;Mb)uK$J?UuURwH2erDWk-wZEY$uTSr%s z0ISg1+-l|QsJ5LO&J)?WjqnT%4qLin&f70PG{(`HnTtQJu2z}NK$m;8pKm2z9m`(3 z74a$(yA`P!^$_OiAiko`{&Klkqw8nl99W}^DC?M&)zS0i2uBl| znn_7+Tw-SC!S4^GhGXh{47P?q!+&_^lM9AC#n$4Ve?ffu`p?m=@TYtv$jKSgb1pfT zl;0-FIClC%D8d+iV73b0nq<#r4s|Ik4IjRbBYEjasWg082Rt^dNT(lQaeQoSZOTd< zLdKiBgsymeioJtbKdWWk+1&}+^qgB+Id7IN0vuWy6U*l_Gh(Vsz@j%_SvYsqQ!l*y z=z>xGtfw_Ql4#M(UpE+nCwX8#q>PCLHtqxBVVa8dU4-==i}fYlFzrzwg5klp*xB8I zErWUy-8=!frvZc_7ORxq9pIlRDvTY4ZPnF{z3`IZ1D=HJg;%f_o-w>`zV^EfUvJ&~ z&E5+!MGdutirH6QQ7I1}{ej~+)f()4AZzom3BHG`zk;opj}V|FCIiRhKCH_I9GTm} z{5FW*T}22r ztN=WZHFO}g6S*{zuwH{D2_+&S_yY=;qO&EBRolyD5EqmRbD3BsHkXUdrD8rgCPX~| zKH4RKe@JjH;BXM|du+KRn)F@F8IfH{H0cV=*@&td)E;bvdq7)RU9GDrM<*&>Lt_iH z(rTPVWn~RjMcS6;>o>qpaPYw7B7plF_rdZ=pD2V)qKyWtl9PMqWoGhXk+h>0o+)nB zR-}g-788f7M795x%)&l-|AtUZv7=tYrYM39OQc_{`|(TlXKS}MG=?A<;x$&>Le}H` zNM;9zD;Hn76k>=lL@3fOBB=j-zBs_kt#MPZA&jlt5MofGt;(rWrxNq31zjB5wVGpA zqZ!Cr!3wjgmcrT0d9bE9lv?|l`NsU6f4%YQORHZP*OdB)y=%y;;V_sPkl@qK#Ioh( zAuyg@V=KT?zhfdHxTJCJ21BdEqc?AMy9EaayGgq>IqA1fe4jiy$P1JyZ_%NpD}(-Z zq(S66NU$(NB{&HQ1x0@-b2B18n8Q0QmnoFCl90Fp#w|bbz_)^&QA_A>vRsAsCJ|=y0mi)8i{pz|Ww7Oi;jA zpn$JH0kgD?s6g3VUr|z`$uH1WL5MA_s@7E$mlW5uceHo5wxBp%V@)+Y#Px(2%Bw4D znC{M&rpES~`g(0~i-1taRgUv)uP!lGfVy_^(j^%3Q!{dlDx0z6I$1GfT@WMooQgP> zGBfq0yB6}IaKG^K@HxJbZme>(4e$)qmn2`<^W!gmV;Tx#wuB)%V>QU=Bl*7T8ikkd zsN!S4e60$F@cXjiBhlgb^fH~ck%ZH&GCXVX*^x^pa+8k#wDsEq2M?c5?xUz<95+I* z=XH@V5x&&IVh*_BLym^x!%#`KpSaO9apFWDb5UglOxCRh7dgudry=Nk3+MbblBb?t z!x#x{A&StS0-=0vd~Jd^AcfJhBUpcQRC)CI74QX|Wh&T-?2GJb_I37Eb_x2f*nl&j zeI9g$Z(*gqiq}OZp1H*A61@^6)K*rN&jtm?;lhVu!P?Hg`R{pP9~u+}ld`_-V!R$= zXW*wk3T1hm2-VCD*wrdQ(SfKaY$|XU!#jeX3s~EyNbdx^EdM=8UXhfPlvyXURN$HA zc8;8d`*J+>BlDkpe8GYROC!QO1-(=}%US|1A(C0y+dI0tSSuZzUF9{;a2#&kvs+v2fAs>9ZCro-r*9 z3Ey+)KQ!ITHFV{wRnPz9=~q5{|AiM<|MQg>!w>~IdeX$OB@17Ca(dX)&piI{lBLhQ zfXBeTc#acccj4l!%2{gX=GI9{aka2c4!9MWTu$n3*_o4D;WgH?oaMMcVD_CvVm0%C*Tk*|!TzAieW>Z92`(B=E;btF|_#va%!{ zPS@DPn2X1b9lNbnvxT=p4RZ{S8J?1BY8z9|UDLMeb*FwL6pWyZNGfV=+N5F^y%fxF zvR7Yz^`Az$Ct#MEZ7~Z)&e103cX`sa^Qm>U1(nUiSG+{x7JozBB9*eBZhWa?&u$1u zMe(TGSpMcpRxt$MVc6w{> zIkt5$`hEl}rVWYurevuQ2&12};-(>sPNKHwpKt3&S_;7y_=?rF4plr7B>wlmi33YS zj(irvglUnLk)S7-DY?fnERly`0F0L>G8+Y@uPdkU(!3i1sz}1H@?jM{`)uHz?eLk79qCxUP9k{2^qbFkX(Z9 zyxwQHH&S90xdfxfz9Dzrh)qB&l;%oep|qacYoY89{cIlMEEpGxCWK-{A5^laW!X8m zH`1a%^zb~?vSf}Cw1=^r{3cS*(qGA~!ef2EwczXOZDjPs7CDT8ZKK+T_r+1`J zucFUx{@1e;EpH6gk3>n2z!i*I-sQ{1#rbLJnYZGTlWrY6bTay;rhqvNEe~J!YI(N5 z9_!Zm?cBWKNX*6UU$6Ntp6I+`e5{SyP*zUD(VznAhdNoWY2>&+gOaLT5=877m`S zzEfu`SQ7HjRrCC3hqs={x`FIrb@hawn#7PrvPIl_#sWR=cueM(J<4dB= zyzRL0__2%qOp*zW&cv)qtr&s1xhAntx`qU-EHFIhc(hU6-&~r|-b_!fSdk$d}7+~a9MDCc3$*qBrie4Lkn8b`NGeX}pLVpz#KGqz1Cm%C~Y&uM8@>WK%9_XI- zF5zLr#e|1F6?!KRGx@{{`oz`ri96{Nx6vn-{Pn~^*s&zGJs3N-ucL;p5uI*wl1b6n z^wbnY`oyK&DzB(PEnkZcbPsHcj&80-w2DsGf@C3neLd5mc9=SOw6}-f&hIzHRPqBt zLzl$H&I_NsU?zev1$sTdDU%b&{`uvXUv5AAhninhg!(zc)SGc}H;hetNSDoZ!WZT$ z3NlkdAYmUCmE@q~aIznZ6121Yq#retX}g_STt*VHvM zHq>=?Hn!yFl87*Y2zs~#ZM$p~3Wc?`OwCs+8IjtdIKSCkse}dD#@q}(99AZ)>r&_| zQ0o`TPEGY~oS232&D_e`Le00a>FDaw$Ay?9;(ZchjtD)Ib3%p50L|7D0%A#hXBEaZ z5|b0&$S&yJLwBM`ClzD&4KYQA-hnK`ZBXQVj5i14t;BdUF+j#+lKY{)+-H*NHVNv6^uO}NI!WZ_e_w}C45V=ld`kYQ_<8fB{lY9 z3>x>v#-${srr;|TP5R(b%FfQpX3iJ`6ZLvxcx&s_qdOLTrZ65lh$a%&(Eb%+H*Pbze??MBbR2r+yo%osffbya2V9TE}} zJS}LHQNBp@adNh|G*?Q&wB-sk;FHT>Z6kjmLfLwAD|`IA6C)eS z#f*D!I@T`?>-P}WZz|SrD%Q`X8wMQ z7QYD6Dl0At$I>J)X;2e6!SjBdO*J8~I>bi`I%-fWH5h&#{-kl^CJnGtT<`=P)nz(3 zLs2V-NBk2afl?`!HPlrxt(>G0T+uW)bUAfcFFt)N9xdN=HMNx`#c;peE-bvBqDi~R zoJ0G=)KtYld*e?2Chp;ikvw0@-~1_mODFs-l6RHz#@s#*XBwMMv7;&0oMLqp+dK%i zr_z5N#gPNmq36nl_j+Xulerr3ED+d{D@j6`m|a}aFLY`sOXk5X(Z#a5I=H0)nyPJz-eATST2wr3a zx8$(#4BO(wj%NmzbsLObBx>qeNUYVyE)p&|L}Go!saIk(z4v=&nWaxVgU*zYK5cd1 z)AnKq+`mzG@87@Bcg~g6oyELSYm44ESCfs8+dv){(jC$Ym|V;OnJ&{>36TOxnJuF{ z=S%Oqhi8@3`!)=6-|5UPvROwP{~9f>iF|~gbxZ+yYi9iLcBdqH_fz{)v_D0s_oA;- zbnYPNp0R!J8M~aK4JOgOvM-Thjdw1l=ukSgiek(9u)Xt5EJJtA_bn2)HaQ2h$u!`h z>5Lcj%(!3QjE@Fw6WwGs@FY8J9A?};v#_*81KlJWF&aqKONqaHBQ_?H*~6jCUuR8p zw62x30q$(oMMu}Pe`e3R1bDeaE&1-x+hxrjvm=(oUA}g+6n$AppipHZC%JZc_cz~s zv-|Qj2&GzsKrNi+rc1olqn4ax4v0Q|baZ!7c2bzJozzk08{VVH9)jd)>+8?4Gtj_u_y~er@_*7bvBZ%k)p%o{wmI_HauGE@i zt;C<%D+EGw3#Glit+`ZgE`hCGZYH;}LnRae&h<6{CqWG!ssCPHq zeIBJ_E{s%V?C+_ZyQA;2@#D3UI%9)9kFN?-3sZ5_(BCj zp;%yUWnn3YaKtCVk;w&dCbE$xJEx?qRHv&dDJUw*zm4Rq;-bpxUeTzZg&Bxf#L$&- zqg{b|a^MJ2V7ctE*mv*tyH~Y}V(Tfkn67GHC!>L}l@wcd7uJW48__pz?{2-v_Q6)~ zAvTD5+85{@qYc8?7wBP^fjRmNI_gq7>L2uu)99$peWMCNhxhB&pkNin83?YYKP4#* zG|(J1(h=np*+EC_rXzOuVf%Yd;U2=3Bn{w7ENjf`Yd3aXlDhPsn5Ore$x*xR9o00p z!PFh=DOTTi$EL9+#~t$1wi%yxn(=8h#;2t{rUZjLtuMV}e>%shbdLMFVhy}LJ)Fsx zVtY7KIepr{);6MtGd&B=M07S{{TP0S^`oBy;09?Hhc9aUFv+Ke-(mUa=Uhm=)9}#` zpAz0{`55fPU>^m)TGwOxkPy?yz+7wi9oCP24w7>2zorh7;dfX*`Z-e&)YfD87!n7& zTjoe-ZV+ZQ8d;3+YblJ6g`agyIr()fBIfA+?A>Gc`h?nUkUcgnm7M$<%!$#AT2ACr z!I*n@+1(@i(vkh?$f>;}U!^1G_l?}USMEml+_&#N_gzlWsD99M-*=+#ly;0yEvD#T zI<|^pwfAA~=G%+uewjyUzKQOa=Am(ia*7>Ku|4NSF~!cik2~D`#Lg5;xi3oTMADIu zQ*`Nrp;dRIYx+e?u~#E`sB1vT;Z@^(2Q%w?&@INXgY`L)_u=oJvl5D*N}u~Yz4L?g z&O1$F?_4Y6*kXzeCfL62#shPw9$w5+Dr?FTc>}EVYYISawId`%v-r~ zo^5+{bd++`s2#D_lhS2ULF=4DhahjO#9#b+yQ=E;?l0D`r!#_?KP0cc_~ep@A9-YH zMhI+tX_180s6r6nwvPYC?AZwZU&m}+qiDQchTzj?O+^FW%!OkbaUF5>$z8inR*Bar z#S#mJg^f!^O$*N;RUnoh|5%@#lapL8USqs>KU4R9>EyK=C<*fCsdI6KHOaVlCYobo z4~3yr%L{mYz#fXluhhyzho)KC$wkSjw@WsS8fAQh_nsVKYdhk}_tu0UB=5AOz|%iy z*7RvpX9hE;c^PMo8%#xl%--a0IC~ahWdAa}v?kKr#f*=rEK3t0gn78zUMcK$ZynP?2GIirUg39^UsID``*Y#urEcidrzSTn&h?T zR?G{V89cKZNd%HrPd)kYlJLpnQbU+a%uQKIQwuATdySqs|6$|Kf1gvKnY-}a7gw)- z{S$mbcXqc|H#7PY`ZDjXenSa-!AVWm&IGni?7$5W4#P6H^fxS`=Sm>#epHZA43x zoSeFXi@$7q+(-xP3U(rMhF!TblsUqVW>;{b_|1f#-yB8j%N6P{{SW$62ASa%cKh}) z#P^x$&*-MEbt$Q{b8+?d8y_4IjB@LzehWn~R1~92n}7> zLkQU^rCnV*dLN z#rslx3dKiLyzQX)^S$_y6fdXvDvBR9C_cXzKc3=?DPBVHqwm1ONOqUqv8OW}QvgCG z9C2bbWIQ(OeInrrlfe&}+1O^i9MSaCQ+|gHG>MLWj*fnSjy~>=(PvkKG6M*P}+ptUZ73+&i}!`$3s**j7EjI>j!!OR}XYyViWc*KaQZQ{9bH}4xk;x#l zt|P_g z-si2mc$V_HJLTZNJRM!emI6rP-AVXb1=d4F*xg9n65dLT#eDj=SBfY)lcGt)fe}5N zqIG@fUW)*cu!LA!68*se8^r+`h4Vv2KDv=_bzIMGoCkWC2YMiJNArPM0*q&2)~$@( ztkm?>^xH+1<%PG?np>J$T44%ms4T0f&OW;L50d|ea*zC$TI5=&Y`Y9HJ#*vA4N?ge zLF*STpKC>@Qnlw&o6^Ga2F6T2%*8)+$x_!V!4t-MTZtr1t<|@$U+GwORYNNAhTIa_fElsq(HBsQhqMBa*{L{YgqjyO<@Hd+a(~IF*|(kxX1)@=8{yTK zn_`db1F2_SQ>VnfmOuW{Q2fC@<{-POG8k1$J(v%0?N2^g{q{#6fBdnEr|=jRY?u{E z0_?wA8)o>3ES%weA%>~Sg*bQ&d^Jq%*(oITJftEdBpBM-x&C?w0s(@97r@=ea(Q^j%rP!OW8A0Bi{#+Es(gSt!NL zSh_4V<(|{AUX!t2z1GKRSTC2Zn_lLq6_cyUNK4AjuGMN%&;EA&Vr+V8s{}%tQ=_h4 zuV~Lu_R7aLkdKR7(cx6mp?2v^xO%g+LqBfRFEQ7WlEngEYs8_=KcC8OpQ)(@ljS6PE_1*h<{3 zXt=GZLvmVywt-c+Sah@e%FI#-mEv}x&_-_N--?^6i~jS^=sGdSbL(zg>*Cz3fVb95 ziJ6x>c@%0Yn{}+bQ#AIWU~klc5Z8(=ZRPgU8h~rMm6LO;5xDNW5*1=^h9KWIQA@pz z@lmi@ZK-^li!0OfXcbn3$f%^oFv*wx2UcYCIpjj-qo@*!m0#b#=Tuz6^^*fFQUVtk zO)3WAtF^P4gm2WAZ?4Xgp zvzc$OA}9fb*H^MfJ(&JhSzLPXOnQg^q8CCTlWb)^7ku#S!>is8B}>U94w(eL=OXzV z4rey1*p`|I6sUX)Whu$)J*4SH4V5}ZE39SB#d#%JH{-Px)~;fKqmSi41cP=kj)KPA zWK>>C%gRi}TOQ-g^hZOZo3z@3#OTXOO;v?DbGDxC8b)c2c;pDPat7QEx zF^L-6FWbAne$3Lxp9!(bs|ZCL1@bvWuRQbEQ_rnx4@Zmz5<;0`W+(rJg%Got1DxQ* z`t>JFQ@n(4`zGs=))*H^}f;X*4)!6niv!yjX?6IPkx!AGX_CB zP_!FGLxZE}EQ&tehwioa^jb2dJrVj^qu!ry)cYR?n8>-!MDK_1SPBU(d_b<;vue(C z^uAT=qND%Cl})c&1-}o&y(9KmNgDb_>_z`?>7>U-Gw2!CJNn&pGK=2(40`X<-g}qPxz6mn_u1fUB!lP? z@U>|Si~o_F_hSN*uN?pTM^fSWlUJ_AUc7kV$f@i3H6>X|SB~vmk1LCBr6*rK5nFxq zKGT<|ApYE$GiPq7T?&=<<}F2e`wvaEwv)?Hnxka%sEFyFj5hNYZc5z#&wu{s->d$C zc-8lw8Sm*i{+ahUUZ9g$#TeAEhY3D>eA&WRL$S4PGM9;^EM&p5#dGEcGk=OQ4*h;Z zGW%)tJBuXb%hZM;z3x3!<^O>9?S)Q`7|`D0uRD7d(*2u;mpIi>Un{Sf18O$b+k31MQ2uZi1b9=AHQ7z1 zpE!~2CGKorzRT!i>`VrhV=v}&?z0f6OdTA4I!|;0yx;884SM6d?_6QRc~~2Afk<8| z!AG*Qu*e-mCy^MDNP;(Bs5ds9#7p0Ig;n$lF7ygPy;ta_S1A4K3Z|*7lv0@;rLy_G zR2ECA?B9J<=8YNf#`vQ!1KyxAZ%~1J|TMnYoTh2L+)6TQSt3$spXou#F& zBp1qRl;zYpfm@T)N^7lcJiMO2a`_nS)RAUO4Tjn_*2TrpTx&2ajpAj-q=z8Y|8wz& zD;Ga9!$I^*C{&4E%+W~ZQ~dRh27`Ox)aA>kjYK}`k5QJ@Rhi{AkkKpn4NaYj_I7lT zjWla#*{;Ugv@}s0a$eh`ETv+pRAFwew6HT*NF*W&l3*guT3Ob`ePTesWDjSSZSB#* zhT+*f@N9v2wq6UjqOA1yLvb;iw_LjX^*UG=UO|7muFT|QWZt0xT9nw`a@Di*CT{yJ z1gUp(`QPqlu0bZ*_sItzaQtUfYu84xO}r4;1()(O>znnlzc+g~ zX`1eqZn`LYj}8hV!!0U^BH-)5t*?r+H!1XWfTA)42O#1rLsT|YHiZ^SDQ)R)lD6rd zY11ZY#{YYATSQ-9{QZ9aiTh$lJJa-aa#D$Yk^9QH$}PF-VnV zxOxS!PCc3p;jUY&-4DhGQJ%9d?ws)AG~6J$w&EUE(=B$o7I;gG@#&#JB34FY1ky1A zLs^JEtqmp=3b;8824wHnc}9X01ZWBQ@1xOerGj*_DWk_nM1`pyTfz>@JokKos{Y`y z^XQl}+OE;Eqa!4vlXbpjv`$dp(`)Lxb(m(8VWJj6zx@Wr|1}x7RnEDHXzi^-Hj^DP z)YF)2J*U%@nd%Hph1WG2GJ1A(f z8-RHD=+lj9kcJN<3+xN#F83t&NcTARSa&2aAjTr){YgM+EO5`nZSL#mxmSQ+m%$Q~tu+&>_b znY{N?hdGfdmoB>Sx{r@%n#_)VK!)}^kr~JVHi_FgFyL?^bcly|B_d$}=|28s`?g>HLTs=# z02BqkvNZre;1)5$bsf}xjKm|^BhNl}@8mm|y}TsZ=hh{2i)3Mu{-w*F9V!uX>#`Y= zG$G-Uz>q=Rs^oj`j*E*O5%wp^!{YiYFwO)_00pL~TO?7)q$zjINt<{7{ZBpf=CTJD zJf41Y8d%B17Pe4Knmkmth5ECk`m_Di6N^Zn_N*3cduk%2GB`lpAQu6<7fLaae{Fka zTN{U32ePfJ*w&G3YuVqlzA@Cggl%2Pww}SZe)w-%mkhNo{zL0zw)LW$t^aIw9+J?_ z5EI1etiC|Q>Ri8oXye1}(8g)iVmh7W&7rnKn*2~(N8I%eNiF}7%2fViQjl-LsDP%k*$1ytzTVV(Xg?7U z%0b+AU?v%tFJG}}tk{tK>#x7+f{>|UFd1v=+S=Re4W_1gV2#%s_4Q5l4OAYQArnKc z7Q=Z((x1|?QYM~?XPa>aVlj~^016NY zxD>G<4$6HmjG7ll%?G3Ag;Db&@<4B>u4-t&0i2+{tPz$pAKf`_9Q33Ndey2moXNwU zhKZNEcQ4U1QVHVN8ewBYbBn=%@C$;ovPS5q3X4JS6Y;BFKj__6qbK5B4@FOE^rS{l z;)bTO%8DkVkvhcaxHg+eq@xwSX>d5MbWh_3U#kGutzvX^Ru+Lp|Cf%opHBl1bkgx$ z>!L-N>SjhalcdLxB11Z{xy$8twp%O+_0<8!%WC!f$?59CpE`&B#PS-+gI$WdA@zdp z$K4QnbR0}$z2-Kk0uAlGR?7g%d4ZwWY_@TL2Tofpz3mOaRcq)(B@JDq!y!~kCCKIA z@vNQZo_@Qdx3|aK+wT-0rA;Ok2@ut&iyNK3ds9P~2!P;fv7@iG1V|WpB}StyY&yW= zQi2_IC%!>&>(*0L6~nzadGbYuPY-mpby~FoH{VO65+j-ZJ@l4 z{j7^at+B$Qa$<5UyOw_~?;PT}@`_4}imsf?&dxrcf32`g4-7W_4P+|mB6&Ppm!;1J zKx>CcrI3pSsGTKO$i!Sb8EK0Z1^rMVum`VRJD8fbn>%~j2|`_KPk*1eNl(zm+Pke> zF$R=BK-Dp;`F&lM-u?ltzZ6w31soJ@@sbO<1EhQiAt9722l+Y_^Bao!9fkP~#r%e1 ze&y%S<>uvFxPrPFd6y|2v=%7T3ZYOTWN`eE3wz62<<>fDkg(NFm|d8aMUBtS9>0pA zR_W;crc~-I^#gsylnRCY6Z!g(vXG z6R7Y6DzaYmK#TXR&0~y~tWEKW<8vSLtj-SnfvnDD%RWwXzk)~x&k9vAh|(Y{RNUB9 zQ|(!!m+(cbQL#{DZu6{CKm6h5DkaE91bJ^Ydc7CDlG^sePyZGw)SuqH>xX~mbn#A| zJ^@jtuWW89-uroW_UfHQZ5>tm!e2b9;EPO#%(7rg5Z`WfbJ{NKsV8&t!mz?6@=)7>1`MeY>?5biQ$TjKm(z*q7Q^=dC$!3Y7o`1;FBmZB2fVaiCi z`K7tT+;iOT3%>pSlMkUPX419&sW1#bFZkC%eQV)q9K4@%zsS8%yI^b@K=!JrwVWrH zy+WYDj#K;T)Uk7@5~`6Z2gLeMfI_OIL+AkRiw{y0m<+^5NO&%3)U@QC`>xbhrl2H4 zIvHuamji?GQzi}S>rxq{*PYP%$EHG5JN_iYzhD>@O$21@i(Jr5<4>%bDYw4iLAq0j2w|Paly!u|40w>QTeEZ!-oY>Nf~K zhvba~@HPU8`ia=+18So{Dwl8ExbaW}9X$V8yzUI(G#%Nf>g-hMhEs96v{r)w0mhWmP0}o(T@$QA z;92S7z10$$rtCdtp+MmWV8qCn;WjUk1PEE`pfHJAC6UU-F!lKQB9rAmN%914>i{9? z8gTJAEJ=3%n~IqwwlspqGh|B(H33kyq`0OY&}B^xCfqb$B_%bD4W_Cp6BH&!f?X9l z1YaVTav3pCA|}6xCA_TA$S>D#*tFw7P}ImVqY}sdxNZBHc6!sE*5JT|9Xk@Jd^9a+ za*=QIoec$y_X#Ji1L?b~<|NFzTv2%Fn{9g_3cOx_?-9xMw^AGS z?mYbd`-iFVAaG>sCmZnkV$%;t81dfHV5nT_!KHh%LZLP_cXc-!0RjvVP~1CSU0sbe zdP8%w0h$vd!I>2L-9TTD#mQhB?kB(W_c^jcl@Sr)N#hfuv2f&G%82OU<7Z4($yFM) zEIL{SUO-bPaCHX;to=Y%68nVvlF{@H_sQ~D6B5HH)NJRWEE1QR9YzCL0}+D+1s5+s zr;<->FAWXGdgv4l4fRHRpsbpq%cHkP^D8J5Of&1&>Qn>* zzQkK!CK&e~iG(;tU7)_;-aD3vH&5jb;p?w09t9?QF zPJfc&ph5{Fa6?AW@(?Z8g;mES(7d&u<_};cdJ?n&lSOkBGv9j~G{HqYuFGn(ySO|) zk58jGO9rsWV&emN&~2SsF+M0>KzyuH$?*OCgS=%tEuSa%4)XUOv?GE0WtcYw=1l^* zCc(UMFWO;QBz^%-#vRZ*pTy5J;0f2bUxl!*!*5MGuAh>+fuGzuu0{oeBjP*;cWcRi zp~&DnLlhsw&%aZw!fz}^;9U(;?8i?-6lI_&!97ZDxkpYmIfns127Z5Tg+DN=GpJ@A zS9O_s3O`-eaihAa7X0L+z>i1F#Yn z_}!rT5=(VXIF3BGyVKV_R7VX`z0-52!#z|v&yhPlJ4o|O(0rVwsg6L&pg+?5r|UuWA(pBn+!I1YEZ##iO~!PH8#&pfcwREo6EV{Po|$gd zaUQvbd)$w|Ht4vcx{&jgL3!)AG2uYY%Aoph;YBjnf8g%RrPOMC2XkE!%4n8M23lnN-0BS0CKb0}a-I;0b1yr#6Vf^ml?@aE>{23TR z4})i>Q9rsnsGl=f9?#@OLiRPVO3<&Uu(+ zb#mvaKnDKB!|a}$pOZ$oosIv#_1=G{{<{0cL4Iec-|2o}kmCt;?sLg3%d@n5!OLy^Tq@?~&rV1HQfq}ZEr zDZCDW*bJZ)ngEbB2 z0WD|1T(3*0t;QilZvc+H2_@880cc-S2@STYrmhhq*^25u4Ry8kh8le>Qj_@ZE`syI zD1irTbGSUoy--V>Hb8LT?|le8LEQijOk=tPrI$nwbr5(hD2f*n9O&x<$Z>^Cu2lQ^ z`fCDUVDt0E+09#}q=tjAm4G4t6=8wR)(>=A=NfG99v->_eDD;X}Q=4^jAz|)j>aq=Y9$DYb80u z9-{Ag9oARRo1rol5=UoupL&ccCJ;p%9D;Nz6BFY3Z~CA2;Om*{n)789C52ayohm7) z>l*&*E7MxJ&&+{sXdMpRN2nd(y^jdi9sS08sQ{(fg+A6Y8MAQ~eiOj?Z+YJDDl>UD zJVo=7`mN=}d`4x?#?7(+`2xPUGBcvRtsONB+5x28X14SV*n7>rh>o<8d}k!dSo->@ zdJKxqlXPvjQ*J&EmM$lUPtH^n$tkv5?Y7K_@UZa6i16^Ru<)>msOb2($neOxxP%dj z2?=q4ABu}lAQdHVA%g+|jsd$H;KBesy-CJ1cYmqj721(7;nDC`6P$C!BGVN$W!y#R9;z)@)_kdwFZNy&Q(L5zOk~}0JCltDkd0eYmGHE zV8G? z`RT4EQ_0DlA7ymS0mja2Fm*hDYS0hB6a2XQ6`}U4ZQH*4b`u^z#=sI13@gmt+z-CZ z>C$oxKWt3L_wPeCiP|k-6isKgeD>w%h{|2JW&Nfc#7f?CCw(6PHD}O^=_T|ce2$^o zs8fQsqhg0gB|UH-4(uINJe`Cjgj(dk*eNRoEEiz(9D{v@;DPS}C;o=ve+NC&S~`l_ z%uOCK!dJ_SnLJ_QJ!w$M0_aReG48P!ADMU8qD6~l&wu2F$HwVmDq5OMrl#g517vY| zc|}=yMRjdOJw^%#_U1+tOdU0_Y2Y}?_YP3HP88X63+w-{>msCi*DRMs{E%cBk7!n_iH~(Rcp7=HVVKYaak%sz4xL%MRgt9zUm9@rZKvZZUd&%Jtcgm1}W)9_X zY4|IZzG}Z9FS(z_CpbDL#8-tApC1M+0WI3wVU1^70CCFY=x=Ll8`OOLFh+hDqbQ6K z@iF*8zDkYtO{PJQ#vV8%e0&&UIh8*5NM*;MX9E>MUtzefP_Un-y6<*B!%L3--O1q% zd+s9I;lSshw?k?&8I6M;k3uGsSW~53v06Up^_aj2P~ESAx@l7-el3C_OZ-}JdwKLq zUcAE(80l+xCZP{ihU*eKfqD+trj`o;m0H6q=Ux9@GaU;GGeChvK8*#1v3Rb)qU+a) zHNLR$(!~oog{X0Kp|Hr4l0+)%@@z1>qfja=t*~;#_B>#<>6AetJ5TNda^E#5s7`z(>qaI&{D8G@{=ud}aSyTrink_jvLs#TZv?%t&n6?G`Qr2K*2 z-kE?YD+2Pn0Rx!^^VvR3A$=ts(mxmI4^>nrq&`F*F{8q{&V6;Z#sQN7v3#=5nx*cg z9EbPP6zUsYU}sk?++cXMv{qLY7TRnbMtut@TSMjIDNvCJW>9pK2^|O4dcH`g3yh3K z{T8wdxVR9ta755CmTz!C@@*z+aNo<($4vcvo_+OHVS8SE;*LONp~ zVdFy2(`ASY6;qL9fB6aSxYg|b!UqLDDDW|6e2f_%V}|1blgUt}?;|@IKO^m&&s_NYtYb&3I8&odxA2nk}&czp?e@%IU#RYZ zeo)?38gB)%w|rDy{())<@J0iv5o}8+J~#t#I0y{5rR7a-Nx(Qx!Z^l)TVla2J|0xb z>xISn`K4v~xp`Nv6%|weW-|CV5+GpH``2fb3ARf)Y_y*P28EY#ljHasU z(&|<~SX#PGP30u%#Q-+V;GIl>eZ_^9m-{9x+>f<_k8!eki#=^Qksuc z+6sn~eY$S7&}ZzzrFTvUwa=oVfx{tKilfeV{N;0}B&YYyq$#Q!^(!`UUVRHI8ev_t z@W15!3$=|}gN^S)YBm3f2VZ^lfyanA`;g(>6T@E})YIE%aXS0V19obly{Wd&*wE7n zwHrqqvW@Yf31O4T_)zvwBcYOE#fqY~3m2!Ukh*41rj=AnAN|kTi8y~%43-uc_T%k~V(LD&a zFq_-E9SR?X0yRYrx6{7!Ud3yUn$vxJt+K02iCPSuMOX3*%6e6*5%cc9=k7U)@~9~K zP#VsVP-6HSZ`kcYPtT;b1A%=Hl_^Nywd>6{sSrkmM|048xC`MyDqrYczv(uL?c`~BP-6^>2V-Yen+l0L!#AP_%%4YpcZM(hI9M9ob52&^hovf=61xq z3>ONW{cZu`S8O((h-Y;Og^0c3+TC0S)?}Ak;OcWLyOhc@N`QN85vaIvq~u0P?q$%+A{r%;rRtHvdr$3)iP^k5e)_%ILh{CFxX0=opqzJ2+-@78?}5L^0XER+nQSCfA_lBh>g z*4I#LGS$dN>~Qq-I9)wvaA3FD<$y|>MKnON`KUu9pMQE}Ow6d!G10(|h>9II5_<+L zcCZI}z7rV8?(8G*L`is}ad@JkIA6b3WKdyeudFODy-{LpXl!V0Y^)}SB}7(XdzX;- zOuR@X$t*ty9Xv8VcJr1^pX|736!|4aaAwT-=_dqW^QKLk8Y^QYQtOQ~XU-Hf{A2#K zEPj7`$%Tu#M7`Y!BW;HK{`)4j{jzVbe#@4ZUIHKr@ABpArandzlJe-XWy_L%UAdRD z{3@zp;DhtBrLCj0t6pEDuc)bMY9cy;3m!17gvNf37pcjZH;y{XPxrDY&M(T3=I*K>*xxs}ZVKO$|^X8sKo(V<->~ zfThl$_#i^`a1bWql@C=@NDR=yrTM?^*!9E236p1zmUI=I-gDr5(cUlPri~8Se-5W; z-h~UfRTf6pCyGx=Nf{fob>FEA&olJ%p2#J^h%xgXU;HRew(q&$6n}TUux-Yh@4mNu zdMfnz&xOD3IbOVQ8WqLxLPG;ZJ&f$izdwdXvb(6&D>8H)L#-PMXbL81&r%t=FGMGZ z0%0Bh}pZN~+9O$k2tJU?>k)1z{C-29k;V%i@pQxt{BbsPI>56O{935KJ_V`WpYAl|}w#QJz!9*K`l6{%xmD z&|Ve~q%&tPAWhk0D*KrwhZ0^_+uClbYi~C|_7S@Q`IjYE#w5r^(Ft-PQu?7dGB|CL ze}9t3w`QwWL}n1)AzBd0atbLbLLp;g zEo@Loa|G{PFD$F7MW9`6b$Kb6M<>LX!*b$uXG(|!b9ll>XkEz2;J^{bIE|daEpM>Wq}Bu|8J-+5mAadIceTzvYy-Ix;06eDJ~ciRYVc^xf!Vs>Ir zz+F$Tcp^>KuxAh9m#tE(;lLN4e7G&I)(5a!g92Rm=9_OmvtZJ21&;TSTh;ME0ve*WopXyMu9;*tPO{Rp%0G=l#$gK+r zjSTXQ8VyhdxjcCKqM)GB<6{El_$F848wP9vhus0(?aFo)!!VYb_)3 z*OIV5W)Wr3=W*nHduTI!(R~M#L6ay_l9Gh5+Dqi(PedXs%ZSxwEM{UFW@5*Nez0vI>V-#6IimOw=jnzYKOjK#ac_TAOOj-PmH{ zgS(+p#9Aq!0=yrCDdQWWB$va3X3{B`}!TiL+MZymr={% zWcu1?iGTSyQGq2sJ!N`m)7h-7C2tgGe=-xB z+=~GGuviXUkK6EmJXqt8FEbodIM~iIwF~o`vywg=2k?Sbq>HV zhCAr$fPnNKim{5YJ&o>>!RA z-Tec=mGSlS5db)ii&Vlm1VwR0QmMdYMewvh?(ViDELub%5VF@s!BAtfxWobn&*|jp z61Y89c|^qLpRe6;!emW)bosLnjSaJ${N}SSBK@3g-Q0l&6jg7u7oU!q3d)SoFMb#T z5KF&j-M;VfwidgGFH0jj6r~j|jzh?szPL#%wpkIv=ZT5d@xwYVT_`mWyWCM~Gw+4j zQ~XY)A)M?L_cQ!&I=upC0(4^w^}QlDC*LRti<|xEBMZ|Ld7-ITSNCO5F>sTF@VuA9wY>e%p-ZNItyn5^nJv`5^z?nYgezB(hIs?F zD}8mz$=(0ny?5vCGr2`D0#%fk7GKRf@ypNqfBNOriR|-dsdY>o4`Gj1i^a|nVX*8y zw3h&bXQjl@UxXCAd_ojfxAPoaF5JG-h`12BG$=46)K}b(8m=8iLodfmJ=Q`N)}a?8EgH`4_WyA$oG_u3vx6>sAaKHcZu?fAHgU6ged2dLMT` z$N2<@L*e-e6V8K=UCjS34C544~XOQH5_VU$&8`TiE)g>@jlORs+|1dv60tuiZ3V)B?tsQ3T?l89@ zrKqmO+-K=-tSJZRbX!+H2Mbfmv%?Gq+X+qo&+`T zuoyIc7O}IT3FTDlNo8&>mxTAJVA6wO29{YDexf-NZ&@a>r~5}tsBgLQ!*^d_&#mli z2$(Qt%$TuW6`y?;7EXjPg0pha76I~)jSaG(94n3rW5R<2&=!Yl?#~-%JeQ7n`W4>O zZ0g_A6@m_~sPR&Mp+(N|9L?#fhEznwAI!ihF(VB@?szL!A1&&Gwzzxu?p;TUJ84Aq z%QJ(GhUQ8=RJ`IAQ(eiK{Q6qswQCKnt-WpiZftZJDws*2@59E*!B1{iUq^2y#w;_q zt+}t;t@PG(^tHFd2?YUuzG9yV98Q0|nM<&cRIRP8wT5P5Q-Rp-b#}FwiSLa)@Bj5_ z*8o=m&_xoTmq>Sxm$yBz260UT7aOd(PlMWyS8-c@`p$9yH9QW1g`aR?xY7Ff2^Rg zPb|tk6hun2Feo)c+3Z?jQ?VZW9=CWYbB3C_#;M@mFd}6$`Hs?Tb7q|`oytg zzaBVn;OObHCd5mdsxF*7a_HBe4jnpjBr zo(+*>w|FvcS()DI-~d#f?rX)FyRD<9y1dfVBl4C>#40amm&stL>Cj(3SJc$(Kpks5 zKtBkJw8IeY4LRP8@RK%<7stVah*vA6e12_d^DQ!c9oFs^tlbNk`*m2mYq55daXRMn zD2LNwg-L;6YayV<#iPtkO^v3;wri_jzECpH_4vrG@HB3 zBs`|i&cS(57Z*0=nP(oGIeOgKF-h^T9>yl7J-qC`2_Zf~fdLAtyNzf3`LlH&Y$*?% z`A#~F)dvwnaf;fhdG3y|@_+C6>gUVXs(HM?sNn$$nIAQYQ9B%_k_-8*P8$h$!Xow> z7AV*9g<`J5N*ugU7I+V69L8OPtQD~SV4W-Mk+;QF?d>yQwZ9g5@9vWG^yyx@i^FOYNcG#`&%hr6)q;5+XIq z&{2~oPaYTL9Tpt~KDC%BiA)H6o+P@h;^er;5Osv)!2L{pE?=JLZQ8qj{rZ2`NPUxE zoQ-{K-QZ64HMNt&?XzYk=UmJ=eda>$rQCd2w2)$2dZW1D;+YFY)zvkIn)2e@yi#oO zwN+(h#RZr1@-F3^&n<)@^+u&0N3^oC%Im!A1$i)iCqqkdbGYsS2ZwlTEx;;JKJBQ+Xt z|>qP>^$M&yLOOKUuef2(jKtR6k2%6jS-o@<+j{@q5jtlKktC z-HlLXh=ZBqL<9_BN}nmWJfuKD^*Zp7Lbru|NFEPbjRX&^iw}(*Icjvs!j+eIef^7p z8ov0Ax8Hwv@mOWqFW>D=oe(0G_$r;9UEOw(FL@@hGGWTOb5o~Iq4y3Q%l0-?0SgD& zP5Ago0Zp5#+O>R`Y6N0_rr6q7TTpQ0@Q#UnjhZeyAPS=|Fi+P15*A0gcw$#)b%dh921SZI_GuMj^UDh6mJXESHym;=+>2sIz zsBMHJ+H!7LuXZyf=j>@e+H#+ocE#2?`7i@K*C}0}eMW3(y1( zPl}hJ)^S)=s7fs0OOz_aaWKJkC-#aELRQ>PJZ|<5hm3=HH(Rhdxw#xIU+KfO_6}-@ ze9ZKJMzUNY&g)|G@elIx1NmrWSe-HoEl9-O6!1nQMN<*r4V9&*qtyf(0&yOhs@rQ| z4-tD51w4=0tfO;sbmFB`dw$K>urvGYHG;Lvfe-p^?w(CJj;25oF4YbN+{IT zRLY)EZ>t4gMTh`&=4fgLx$xL9R3RK&tQNciBSqqxiFIQIMrs8{YE)fYld-O@o`n3F zOt2{-BeE4TzO}JdUtLvGTVtrh=GagV9SjjS6;&`}C((Q^>+tj#4H3}mbYP{?E?XaT z1~(0r29|9arDFwL0;Zyi^qcX(NcphH;NalM(bJ~f_1vnLpMPM%ijWDTlzjsZ8`UpN zdlQQxOZr%Bh_ZO=hINNpC|@QiVS!gWMT_Jp@k9`FP;Fav=R^02l`7D-LY)xrLvrlOMTx%v9@KmC04`1dmr@Hz#N)swu}xE8$j z(u+$bO%Uf3R2-vn%9}F*mQA8jH=)WxG85YR(HS^a9RqY1#7C=a(yUpFCe`ik^d2{D zR8)+b_MEU}D<;MUlx*F)^|ClL;ZZ=&=tddJs_SYE0E#fc><{z5USCsLUR6<9Q-?!g zV{N0U4*yewf?EwRNVm7bgq;LwN(_Q7w}beP5k$)6aa;~JM+nGRKJ6ADu#`i?ieX1a z9sF*(QQl%4QM^?Ei~#(M!V5};Tq1_AN1@jE`TME;G-`q?q|#{o0)i0Hijemt_W1XFTWdKO&rG`M6WYK7ssfFQ8@ zi=DX@H!5?B^3LzydEnZ$>!{RPT4}hr@5dAQxu+@*|NPC`jVBTJ#mKEKb>-I(8E;Km zkeqB}G)C_bJABnQY8u^R#nM;CZK>;W59E}t zlZ>4ZE$2vkPHz5u?e_eRilTa9u;kOyd~56F>>{8TGccpk*fDti@d2HkT6)UcB!2Nz z2?7!B4dt8mqsVI-f&yo|S5nzpo>ZFqYjLJR)>~C_qtrOy9qh~N?HQShoq4VBw<>*E z6=(Eav#>+|>d~(cQ0oQ%oTqH7s=7dcX?F?(uaxxp1XC}gS6e?V6li&hsHwAN={U-! zooN7!!<)CEN<8MDY1rxUMmUF62cKU-y(M#Gbp!001k@y8VzjHnT<;vJq(oZf(QnQIw5cCj6S5!V z^WS+)H`Al<=2E-q2SvJA|6#%2fLoBM!qXP7cuaFSaZ*i12bbIK^wRkH%gxno1GeVM zNr}HrOr<`fKHxnu!H*h;*^+tze+Z@u)5$M%x*3eO`_zw@$q~g4#9f?P;^$9wGn6|O z>d8F?*tSPVTP@pCa63C&cJm$VGUgM zWVCQiCCo9H!}nh(kQ*ly2!9X?gh2=e!XbnLVUgQYAVW?l5H2AU2%8WJgiiC z>yUdOyhHARFwf9EfM@?46OXwd?1OKn5dI+)2m=ucgo6kL!a}#FFfnum+T&7!=D2jAwdaqwU$4Pz+XZafH`%s(*|xvSm_Kk)7@H4( zXSWSLJK}mgu=@D_7nZY%WJ6b7hsjt}Lgt5|&q@f7TiF zhI-^fHJECRP(dpztMw#?lmoR$3m`a36DZ@-4g^<1ebUOUwU-+xDA7(g1@%TN=OWJ% zhewfs=dkr#b(3Qg;=`miV91;~XN-L0z65i@sWWFzpS^hg%;6uu*^%E6lsbC>=W4I9 zjfzd5JvUJ?aCP_h>p$I_M=~rX_{d3$8CMk;AL}bezGs8kiSS7z^fh0sr#}2}PmLpP z`t<3ja!@A?Pka6K*B?z%Zu;cI4|T#zdyie~^_j8~DV56i-g#|F+MN5Bzx2i%ufOuz zd&^&*2|K`g?Ac$a=S0+JAKq8wn2;7HD%|(Ow>QKIPoyJ&V>R;gb|Dk@1*BjiYvCb! zXPeg7T#0UcP-*z)EijwWRP7-6+H)Z#Eui%cp# zw123uEy#dH@EPjW;>@Ce!rxlo<#6#-{@y@)(8`5=(S5id0hgv=Bhbp7c32(B$;QDG zibcA~N-w3KzmJbE3I6~pphKn!@DK3yRs~b7E)CK)RU~WEOQP~e0;f{t9|YTZU_gMs zFEx^M$>nQ-)j_%Yd!23`YQ2&!Lv4M|T675$ZZBLII8{mjS3|mV^T?Ojvl7W0hyf3h zNN|!j5CjThQj{uReABnhD^0u!iF6ZOwIh#QxQklFUPeJRra~V)W?CSIU}g z&fb1)7;KF~k;K~7#}_LSQ>O&^Nhq6*uax`vB5EJtb~fDvr^q)rI7lOQKlRk2c_}H= z0^2KpI^%0AI+cCM6a+}PCUjL@+f{2H`}mVDPE}h^ZTx)QG5Bh|L?XFD!sW~5Egk(_ zmA_aQ21Wj{^QT1 z$F_t0)5YCmM$LU{<@3+4eC~w@AHX8akiGNP^6AO)-0!~Gd`X#-js^A;9B5~#tPI2v zF$$q~U?{htp{2zL$Ah5*2l$sg?{^fN_8q=h+SKI?8J(z6@TpXSAj*AXidPu{;l!J} zv)ykYK%;dNOn?N0GqAe40|k|8YfYRWABk8kv$eFeSL9u75=bouQv;xv$k)oNYJtSp z+}J_#6cs&)Q>rN_7(Xr{#7pWU*M{MA1h0b9S3b~hE1EUgR0v9eFFV*-4`1_HsiDlltSWteUB$3h6 z8ypbTHI1D;W`V@ho88vO_fm?4mJ=I)tpCs6SY4l3iAfGC@r@HJ@k`*ai9esdQCHVU zdGm}1_Fv4ucCDb`=#M*g|MKr$zwG|5T_WwRHdGg!JoL+fA3oc0rTh0& z3^U&0uymQ_aN+m&BOI*T+C{?6I?a|4(SW0_uH@nsePJF>kEedznwJmvZB=6%qC@I* z6YHA!!GWPu=1-KKx=?xjr*$7>es{5s9}vFd`_t9Zq~wXQ32`cO{pCGdw+suF^cU^l z`PEN(2Tu`WYq;CmCCBN=+k4WaNipu469`=Y?vfq|PL#K|&?N|vakL6}oW?AA`Q?}I zjCKI6&s(R6kr^)LUTNmgO^+>1O8rL~#B(O~5r0l%0H@1D3<4SQ=jSvDIfm2c&((Uz z&3GP$48i;PH#&VLzLko=bshhyDa#%MiYF6}xbxshUqsSXRaN!)#>_!COJ|Lb^9ELj zWT2xLzG9uctGxIZ+}A;)J18hfqMIn7ds|u?FaCV0nI}?HR;0ZM70RISps>ggWy#UQ zd4{T*HYe=`dv9p4+HC_;56y+OmCuaFYEt`H^mBPo5~MPJ z-@wTDgqWn6bBFuJj!7OndTfx(OHMtE2bB6Fla{=|!QI`+^XBt8{Y|~Tey~FHx0dDC zShNaYAt;o%A-Tf=u|1e^|JOQR1^5KI$^=XvE?=@psKfDW_{|?3vYOI+;LBCuOC9*~ z9q{EU@a5F~r%DjHWD=uPHu(U3?6F?ZF29V_*9$Y(Mj z!l1aQvHE&GIQ3ddMMYIzd!Iu$t*nuUP-I_kh5XE=vh2+uyf3d_6dVKBUWn9P`|IY- zn~&!oJ|jpR6BdR{TubYX6MJ^<-1*aQS4!b$ViIkw_4PeGmQgjem-m1nc3rWCC*AoE zu*||}K?7k;Z*Ly0l5%OSkS|wzM+^^KNe!DH6qGeCp>N_FB3ffJ^j`j*gCs2=>OzP7#T`{QdldeJt0i!aZCo zbF>+a=3a1VQ%idvVN+Cg=eBh8_G8QEND;w{Knsy>+T^5R8jU7u`a?^XFOh~$KJs%k4Q9zJ^f@Szh$r9YC=?;)T6`2DVf zS2h$~--FnxeZO70np0YO@qAv+iQU__e~p~yXMtvbnqCmBPDzNx+zEbd?zWjBT*1R4 zjl8#~AA*&H*wrI(zWiF=)*nxSZ7*N1$vIP4ryFair9;CuZTj|TaZl{X@!?#%QgZSO z?dLyyx9fD(Nr|hOj*L&zbQ{kfN*WQ^Uvu@BZ@=A&=1qS;YZr6qN4?-ujZ z_dfVCNlee8)@WXs^|aX%+II2Wi7xfjrHFtwdPPlnE)5RewJ1KcEG?aE-mzStxbd@= zg8HoCe!Y;a;8{&XumGU~5~N)P4b&VY^0c_TO9zL9sH3T0qRW_I?`Sl^Gv=nGO6Z=v zZiiJpbCWa~fG~S^T@0`29 zpQc%VZ6Rc`q-oH5L{E+>WqWQEho;olae8QTX|N z@$u%v7dv7SM#*{w)R9lYMG@<^ef9M(`9~;Atu;DfjM!9qxh8$pV8UT|EOVll6to1sgmzysOK>2b3IF5v=wC>d**vbK!MT zFK!*B4P7!dsyIKVQ#Aj9CFygXeE*%do|v)Zfrr)l6Gwk7wZ;A;6^!?h@Op8*BOq?l zGUCM3Q4II-dC2-j@4zb*&O{e;6XYzT`9=owtTvX>Y*mMFV|y#mJF=7zb0|a17g&vF z*4MYT&?*s9+7vvqv05`__Lz|H@L0;Y4SY&kqPI^<%6$)|$Bh~_ z`i>Q9-@s7|9)I%5Cl}ojpw_HfHZyhRq&x1KJ2S<{ckF_DC#TK7dmgfQhyaYIMN057 zIE}KlvYe(;X#xVIET?g38ScV6RH+OPS84gZRe2X{-Gg3(Z@_E2Au zn*tBGZM@I6Z`-+f9wMM-;XpJ8KG+3_FqndWDF8HD{(lvT^hW3qqZrkcDe)l@!YyBY znGIc}ydq}Al>eKYOarLxadPf>bLAUzfwF>AA7XGifLCyXG6Lh@LY4i0iBc{RrKq&P zW^OAo4|h=MQ(evQ23YN1VDEYJ|0HCSeJH0;wD>WJ6p5C@hWTlN{iy$y*cIQDyD}!( z&o6P5Z%EX{N7AtzLW7|e{O=^O@TMr{&lsMVJpC=?lO!)#FlWxak1Tm;=JYZDC;5zc z!&!_fHg?=_F6A?M>0L<9j)_tS$NqoF?7!B8ud$kNAFBz+vzl-ls|j;%YQlfr_Gh-O zm2I2IwvA%jhW}05HnuIY0Eo_QXWL$3+lKy4+jh3?bhd3R+qR5tOO6Y-^8(egZR7vY z@8I+PUB83(Jd15x!S-9u_8apz{Z_O6KFPK{z_yKH+eY1NOQOV|fUG0?Dv1(b0$G=Y zB#~OU(Hk4^0h3u%#f2lk9?Z_ZRMFU6kBYh#)kvy=|JFdR zD+2_-<#El~AFa`SeS>xuA32!^uRqtXsJdIHt~h!!$3A-cvb!dY9y|FT&%E#o%qa{7 zfUqptYvH+jzg&l~x%EF3_yGppZ0Tv%Uo9vsE-g9oQ3{Zww5tzx$BgQ?RqR9l*c;r! z=H>U!hSB6l*cyKTvieV`*?Cv&WSlU*#J2jc3~C8Xbw&04I)#&xIomI9-?)C`x7#-V zYfU=c6-|U@obvH5!2c2itIURduTyaLL&HTJ_aN zL`@*Uh2&y3n;MXV(M-@1XjH0!$=wYYtbqZE2B5qGt_wy-WEKu^7%^PuWE_QjPG*wY z-+%bHu|fVg7pUbtj*x4*bU546*unMpRjK^cA^tcB`1vTX4F~uJP?PZ#kx||{wMY}7 zmWz<*=AtA%K#J}m#N`Fe?cH7NT_m1cXtfObKqr732wu$saKkuo!#Hq5!1YQ~WpT;L z-;N(W`si}D`9(ON-ngL)*uNt-_Q!q4 z_Z_~JcO@^cAore!AB{x}-F`#~Q}^h^!-mO_;mdK!V;RB3`LH)Xd9RK*ymtYOWgZHJ ze~HQ20*gczKDXiLJ%~`*FSwp_vROU)xfE)122v*pglKnnb)%Nk-N9k#-`36q)F64k zLG9B85J;2827p|1I3gZ$6}vh+Fy@Vo#Q52#6L)tHAh6MntTRSn=|m_EAe{h{rBucZ zlWF27PE1XZN~Fj^Mns=XgBV4sODAr#isdkw^N?pMaQ6|^fY#eZvb14@AZqTyBjd6a zcZ`b~{o}#~$=gYdTDC1hX34h`*|ueD+xEX{dzNjhWZNdOZIx_W+26Evv2E{T+d9~` zN7=UGziI1c+djy)t!LYwXWOET)oq`*k8PXwhkgg2_wV{0yyqos+bXu-3v9otzv;Jv z?bpS&Jhoz6+(eb+U+S2b|Dl9zYq$9VF(4nF@yqPncGt!OhYITt|1hL?qO&g{{9|@o`UcW zp+J}iS2^JxTxEoPZhsGif$&Wf7how679tb~4-pE4iEd9}Xf6mJp+zTQBtn635}`m? z>Gl)|GvS*~!cBw%VJAX?@Drgx7;2EhU+YypAOJDrM0;F1(Hs{!TI2G&#P~0~M8q8F zoya-TI}voGcOvOX??lwyw)ercz+rnI%l0m2d+$PHR!;tx-iMx@v?jfi=A?Jhp7c%# z4E8<`yW3sZ-JZtoHV?ZSsW=x+f&~$$*wR#QfYZoW4|w>h=GK7tv>0>7Z3Ft475igSk!9y*j)f3VTEwk_vFnhYm*enyPBg;07yK-fm z&PQ&myqI&Nq|PZ1opSdBv*xBo$C$JvLi_{w+nfWv}bMYd!~`@f^LNpdgu=6n9 zN>f_R&BofIvWlj*()^qqzt#h428V6E7KD(SP{c68$s{_ZXJlYN)Q=y1kZUT~v;Es& zyQ!ZTzNjzjC_}xU!oDJAHhuZ&M>bZX6FlgaU!JoJG3~#(BIWj=Ik&?6_=5V3_d4|s z!Q1d(!O%-zOoyM2TpxlR(ZijmQ*O;_YiYB%qIaG8CWnoJN#lrlyd5dh_pNc)F?cj`i%xczifyxPIdt&`x7TC;A91VBP-%$J2_b!Cu`$mWq%{v&&fJD*bHY-=ojSbmeNPG)@+0d!&hW|DWBlf|Fg$ z$#!wFhdEj6-_$GQWRw0<@A%XHT}zGM^Ua*BiK~~W5^i1fH}$r2_4aVG*_`YJPImTn zWrY|+T?64M+i~v%S9yZ@NY#^n z_(X}->s-%>p&Gs zeD{uX*@dkmeXV5;{gUXp3w;7L%9=()*LNffI9zb|eb2tUW=%gis&aTn?Ar0gN8fyd z*;gXr_13<&I@v{3t#_-qRNg6&@w+NV(^oIY{`eqhs!QQb-H-j<`#2%`7QgvKzBJK$ zprO$_Ff49*;`|f@pCHNv9LKzo{G9|G_7%dJCG90xy_V`T^B-9R@AY~V_MTWt@-W2_ zsf5e=dHyFZF6)=rC-hO37TXYCC3EX>_a~nG{BgqH$6+q6yRYxk2>|k}B6N+T~lt~_HA7wAevbB#mtR4yA zzLt+?hLDIda$2?yefJGkd+s%s47t9t8VpaFUU{aVdVVPBba63kBE;8@q& zQ&-p8R$X0lIls29r^`@TSyfPw1IV;eVEklI^g0T7P9+`I;7@5OE3Zz+&Q^4nme3gDS3se@<6K4K|2Q$9k;)YDhqHg{6Y^7|io zhVgp*>6aD+Isg^CXwjlbYtvEUp4kTXjMujBe%yCp=WoYOojh^;Lglbn=BPgM^_I`U z!>ao5Y{7=_6a0FE?|43X;|(zQ=de1W_jIAU@{P*8w> zP}tng{m5e+uZpuJh#zM@^oyW3&w_rJI?c1H}V;I2siWXsZQb1RZdQSP;%Hp-M}sbJ^MFbV9pED)Ie16;XJv zD~}bQ{OZ&9zo`+suV*}*>o-H`M8v8J4JB5Ud*&BkefBY|Dk?M-D81A(I>BB1n&y&a zu&8#yeM1tBzH)JtZlvDlz0hJzMgGx#x@iP>k@Q*$D6Nc`m|Ok*3!h4d>xW$HfTjp@ zi|X&k+_o0(yS142*7|Hbmv#JnZvLi>jk|Vc9sQuXvdP%nbM_qYt~55!aJSyx$RMy%W8w7Z74^P-+ z6#PWM=pG)o%?Rc_!3li=?RhiW^Cq-s1metUO7gCpKXdxb<>E5SNOgX7ZQg~8KxDm| zUs+q4a|HkpKz_&wGu0NtJ9f3Ws-5JKwp0%e0tckCzNw`FP zH}0vk2i=M6py18NZfbbrs5l92+r6um+bLPwrcbC*%F(9nV(|2-wTx^ikX&D?H8FtDDOI)ZF4mg_>AqHeN zoSZ0^Q-F5EE@}kh&?Xc*`R=+1rA!qL4GoKljf?S>^bC82dize9mJky;DLl+a>lYdp z8WBUBn#xLhz$UoiAcajBwXK~ZbOg2+~9MF48C}3 z{M3b;#M=;x|6lfW4ks(*Wc@kWa86czO;!>@0lg5|>meMHp(F%eL>hJo%17O_$64A= z;iq;YPs{wEqWEd@HTheiYqvrEQRv#O&^5vn8?|@u-e2|>SAt%lp|bEwab@ZGqh}HM zQ(oNASW#YDRepKfwx71|*g^eBRsnsGqrau904LRF_HOw!`w|d>OG{f2>tiPKz73Hr zQK9~UezGn~8{qG&rn^VGix2LdK6Cz@Nn$#G^JgFI$}jFM-}}`!+knc=^M-Z?%M@ag zRm$-Sv-r%Q?0u*KBIYKPg^&ekO0+-1x7HY zXnl}vbd2Y)SS+d_%b~3mG%%riTiP9-Dl$`wgtR_t_MEv%iC%%e2-Nip^in7^VUvA* zRWb>9fs|T5jW+DjNAJ2}X$s&tG5}bjgAO}|!b~p_0z?BmoI@zDq1ox{kA!%)pu|{C z>qzh2g5G-uy+_zgSEBbKiz<&~GMeth=G;LUwlc6@1m0>MKZ>wN=+QS6;bdu?!9L^^ZFBkzKSFF_0>%>-Y1m zA}a68Pu|;cwYX9`Z_%=Ncafu~^75H8r_Wb{wTlr5`1aw^OuuKIdHLm4;bTQV;8=fK z$ynGuFTeckv$s$4@m7lUa(`zdEQor$-*b1}%bTAL8|@&Xeos=n`OhW}0io$lET^x? zH(cV${H85ldHdbB&!QeEE4gNp8AIFp1!Vr%F_eU)&gfwj}yQng!dSL@*;@@S$%S8ra#eH z%a~Z>g_J6O4OyTA28ypokb5?Rv;*?O| zAg_SvX^C;+L5Mwv(KtLz%VnAV2=nM}AK)pqC{U%~4Ix;wre@H?M>;w@!BI^xMG1cS zjp)TV^kN)(QQKfSXU^pcs)un?D0*tHTp?Z7(19%2Asb(zlb)>| z8&OLv8@6stMK;}2ypMm}ya+b!@7wp?WsD4Q!GFf9o(Gp<>oIwcM9Ef zqAAYSuDmu>y$j=J4_HP@scdQ+qUaCeH{bGJn1_|}XM)V6igPtDPP);^^x|g-<1)EG z2+otT4xM%w`Gi3HKsu&WHaaw_@=uBp!*CIf0LmRlKyFs6HYh-ZxOU#yuui*Rl32z^ z#Il>4uf{ii`n1I}ZdeL5J|Um42R6`DcV^rT>f)+sWt|Kz!lb zTJo)id<$Uy*PJY%c9QA$4oKmx5psC5LlSSAYZ<>2wUND@km7c-nM5*^lVgcoCMQK- zle+#YFwE6>J14n|t5L?uj{S{nCnu}sWM^}-g`BMUZ)9^hSuH0!my-n+6?sy@-^hwN z*}FMeDJT2u|Jn0OIN5tS*?vy;Bq!VVH}$fd?9#u~JN~?X*Usbj{0Jx8&eeO0t5^Cr z^%}W)2@oAx=Q1aIm6J{Q8`<%8mT|ITPFBRpdjCas!dvi%hhx-LFRalFPCA>DmU7a6 zWT0PPvvRn*4&r2+IN6cw)l58Ci1vi9V!|uq&&jTI$$~PKFn#0E4|CFg=+=Mnatvi~ zch07d-?_I-K7|H=7SXU2sx0F#&pzQpn;1<Ohb9WwiP4+>|7*F82Kf!Z9i06I)^J=`Is;aW2F#l?9VQGGDxxoyqvlccwT{~s~ zj%Z_ZV_8mNeNA=#<@`dh&s0-;VFTx0y?p7?rOQ>lJotK%mM>Ebz{2V6?CQ6W<${q6 zbyVjQ9_A8#*s3*;+%+#UL<7U#H!=Z9*?s{*A(O(@{U^Wr{G(r*#evhGUIYjF7s#H^ zl-!yWYHzG5MRq~Pl*QAO*v47MND>ad_(IPBmBaSL!zrGYx?c$|@QGf@lvU3^|NLW1 zqV)n_iCgsRHvkN1kWQdi;JOX3WLhf=8L6qPsA_S#-|;kATEBy@`2h79efO+IbR5lx zU9^t(#>+6$zsASCa6_NQ%4-HwO9Y;C>(4bXb=fhiT%mK9d5Aqc+?6{0j}`W`#gv}p zZ@8{^`O6hXcnn;5+W?hOu>yTOkPb{OCS&Iy7#a}gZN~9NZI^At-=EjfMpyv*hkB40 z)z)qXuxwWs04XwL2oS~tvtt8&7F__}+5)ydr-M(!E$>cjCTD+VTNm=EN9|61e7Mpl zA~AW!g0uy5QqyKljrMhOmrq`Lk3|^~6B!v57dJa~W@L19Vsc#ay!rFyQMaS`jagkN zo^LZ28Fc>bmCb!4L%od+-6Or`K~j82VR1z(il@b#4Sol7~3s;M-WF0@1b^7R$>?`Fp73Jl1wZPxI3PK`AK4AstLs1yW^mVXoQ~YVE@Wr1vk^GAkiDIKlC1*b!u`SzPiE)r#_sc9 zQs+p{=FdLdQ{35^ivN8wBZa*UHA&qBqc)hwfG$EN7o&ldwY%AE+QTrD*RBe+75})= zRa1b|Wga{Hmji%jgx1o*bPz5z=)|gl`_W4s5{L9QtLo7IFP!`t$;KdgJ*KGM8b|a5#)d`N`lA{M5uXqXyj}k zK_Bkz8lV&!5SG&V=#cOL@R|fggx;7AYO{M*q%RA<`Hs8qzVr5#E0zZNsMJW^2@VZf zdLytFi3w(LO2s;Vi5ptVZby+?xdW^-_-gHTDzqL{XD4SV5Vr7i^#1MW{h8?fndtq% zj=s*u+S=mM%8JtJ!tz=G4I@>dq^ha8wz!H|&uY8ra&~rMr=3g!_RhlW?8^pgCS6sf zca!@~3h|5Dw&~05xr1Q;5qz1~Y936-ka-ZE&}Z1!MeF7A_3OcVh{m**?qj6;vi6b9 z4c-{Rz9Nf2uwH02=AAvAlZJH1&!}&B|C*&rrCO+S{O6{GPFYN`2wfZjrV?Pg0TGxA zmuEh@>Jr27`pPdeyvxEz$!6Pi@g5bD-;{K}7g$?f)6&t@G0@XH&}HfCXzw=nT89UL z{|x|estCe@fZ2>RQ43lsFm7rS8Q$z*@q2-4xx&lOGYqgu3E?whCQpftn3+0%Mr?$? zKej(NToVhH6ES)4xESXfdIBx`5?b~nwCq!8*{9L6(>gj@u(hnMt#x5w5NJGOb3=J` zO;Z!tdJJ{dwIxNR6_r(uh+8r>)Ya7+JFqomB5XaTx&mxhu9P*k_Q7}B+f`N7)r$|H z7pZ_re5s_-)NKV`P`Am{W7GQytkNKVf8UVUZ99Mb^79Y(nzT2syeYRRuP^Y{#c;uW z4?A}Y4%t3Z@nsqxZ@<{tNrBpkWdG=dAYgw35zNaUY*izSRuSkUto!V<&rVmLJzISm zA8JH?!3xI1N=ZfjUg*NfH{N{n&Cvp#L<#oKiju;z+9fD?19DY&VUj&ey)Rf5X|AbB z!!MA&DtY?`W#OglqPp@Wa1(AuFA;~}7An*2iHDOz)7HH5+VhK`zl=~Q;k9?NjKEww zC{H6{*tqEVh&|%XAOA^^qOdfyGD5ny3t7II)J==1N_sjy9eAU+=#wrL)Bstfw!Esb z2H>Hc4aFCa96E9)>&$_jU%$U$!>3!m{PxGKpKsctz?mJzEHixHfSQY#N1=+WUJE3mhQ0b=LO6t>s2wpO?4rwKs7I8q(xsgAG){U3Crx;tT_%xNw4Lk1w@+o za0vVCTDZxiMErj}xhE(8Iw#-7)&AZ@?W5PN{f{RXbJA=6RC@+}?a5u0XH0OgOxUk~ zNOSj|LSq!d(Hv3+Q5{kSr#qw!qCD3rgYd|Y-v{XeQU=icKsPd z14${Q8?RH!pPq&&5h;ae5h;bJ5xEmik4PCrk^XlXM3qPxM3+bz+@1WnMgHwR2wOT~ zjr`Mna8EK*&beQsdnw%*fUeEG3num9+)O_f4M@Stt)=3>2MZ9;vssYmGB zy%Y}2E!1|zc>hf8#G3zs`0ZzIyy>jN@@e!5vEklDpFD#;A?L?4 z%gY;Ev3YLo8Wp$GX!O|Df?h{&V>u-^U5z?xB`N2kh_rl{<<;(miR8v zT&3>V59lU=q+Co30_V<6lmpgq2$q~3dqO173=Iz9gN{W@rEdep+Z9)9v4FFpVC)6YG<=CL)8-I*LKys-Jx556lH z4tj2pO9A#%n|Q0JxeSPoG*iRNzu#Lc^H9oojOOAuUv2v0^RJIrR~MW=efo5j#c8wS zybCnKf=9!w#h<0I&yq{9rGXdxKnnG~Zf0L!O9Rfl>TAjhiAx_}hU%iig8ZVgvf^e+ z#dp9)7#pN{AhI4C05S>UFWLv}fb3i+2FbS_Jl_cRVwn(t4)!C5Vi*~Z?!?RjmjZ&+ z-Q@(t6Vz8?fo^7EVz7r2A#)-Ybj9vSlhjU43RbxL#!gR+MGCqo5b1Ddl=i9da0kLhA zQKFQ4X37z4fq~eTy)7dnLl@rE*az&_;m&$vFHR9bW!W*%3W~I$;jXTBGeL70YR5dG zlVJiAc`C(-pl~osWam;s@H)tmAZ&H=@zA<(%>)^hmmdJ$|iaJ^>^tT5(&!RPJT`fz%vmH0h;@F#RL+hsmb*> zHru;mFSQoA4uGc2pj%qTXL>)>M!2j4QJXhvb8%UhSE9!!Y7}Lclw^UPJ<@~S`8 z+}Fp@s9pHsCpX>j*E^v7{BTG??!v7F2}OUp6C#7VakG!Ay#8`W%pZ3(v4*a)50_OlS-X;A@kn^@9U#w&v(F5^rk?wiZzp*?$}ql1Au$oN@pSi~oEa#`6%zsE znn6GV1GTWT6TU5Ak_>es2BaG$$Z=<5mae9{>}(qVhZ!+`&dI6kwCOxG6way>r^heZ z{pGuW=GX|zve(7_sZIO#e?k~7#2YqjK2ex+fZ=c0fD|b5^Xso4IIveIad-heUSw&# zv9>Py9SmH8zXFXiS{^cYJk4fP$ZWP)&O%+9W_=U$BrZ8VaX zo;SAEk-|R86W5w=gd65gHiWHVmpXG1`P7QE)u-79dFH-JDeMzi`F(|B>1}DK0MVeS z5et0-&^m{DdWrYX47_?K&f#zi2ni8F_j^0q%_hXao6YsrRe-c9IDh`a6-@TU)du)* z;3)>bj6;{eii4Aq_k4SQ=AAr`b!a9DjIx^lIKH+9KNi%1+cUOIa1tBcSY z8>siF_xP_Yrye3s#>j`(u6;8NQQWxP-?%wJ2GEAEu&}mEKcpe35SQlbW5=#^5xdGh zIARmIX%%Xqqv;Y_+KP_+o}E+I0_2l1TjRyj(sRwA#~o;IDk{2oA-BfZ z#EL)O%I-2TrC{O&g5L7HydhGB94Tqw zn07csIEx+`p*0#0{@BWb}%y6MzEh@q$ zgrRY=E6mQWtjxwTvOD4S`+km!0$M%t9QiT3LDz>lr;4hAyC;T?S(ifnDlFN)_mP~O zbeLadaB;d(zmh~eCe_*F?9rvtfpp+Hd1q5?88Cu7`appJ_7#S%uh-F`q3h%@aW$Z} z@l+v80{aUxLUgdi*KwYdiFuyv*T_82?WKzvxxJF&S=X2ygUX-fOLAg9vO+m+}uyURR=VE+8a3eCAX%+LAVn$%GXRLot(^2tqi?? zzl#;VtKHDm-_?ZW7Cf1maL&7?4X2Zk!;^5@fbalM1$a~w@t;0E40v~G{3q82Dd@8s z(PzX8OTiN?!V?5HHgv-23g2v7Pb)C&fMMNO=X$F8LXbe9c@*GMZvb8bkMHAWXQkpy znaKpMGxC&pM7rQ)3j3Hx#+GllY}ved>o=z>9bN$u-N=M>J^bDPf`su;J|6raqfW;Z z(A4j64)eojm7dLNaK@&myV`;@1i#fj{qCi}+xZefVc$_dV0X2f`d%94Jycs=E)BWX zOpS}E3rOYX8Yq)d?~jX%(+<}+w09NmQEJ6qHK+FPceNU6vf!51ntDSkA~EYK8UT~n zY-nm}Z|VS-c$)=;Cg^2F00ytO6wHflEdqMPnkfqipcX6$3tO-tAV5zmRagQ8URx21 z-IRS)qnI4&rB;W*5JO`t0$@y0QyDduI>fy?MLlKZ?b48#08dY(vXJKgqwUE~|3ByE zW~dCNCiZxNVF7TUu+xh_{!r#b(J9k1AMd|63$uw)04*Ix+ z=p&_P%~T_uOx4o$4=p0i4eK|P_Z$hOJaSDfMPLw)i#u>Yry`Bo+c`4QZRqOlKoko( zV8{@04H?Xc(5PWy0RcLdr>9>~fHE*5Q0WVFGJJu4DU(*Ih^m4V1o$NG?mA9iBk{~6 zUw8(dc_KqcH4Kt*cDS>rf2h5EkR9q9=CXvdv&VIEeB_BNo3@vg-3V;?bcF71c6Q+! zWq(5Y?kaW#mi$l#QAX5b6OT_6P}oK}TPiA+fHvk7^#R*KmCzAX0o??a^PSWoj7D-s z_Yd;=^Pp6Ta7sqgZgLq12ejFv}crWC2VS}14P{!66^~4TMee{Y(@ZV z%{I$A&2V!`hl65y{=0W4EZYC!`+EzEFt=rhHf-4N$+vs<>jbaA-q~eYCtSM5vQ7hFH7r6-C;Sgq zD_<(10Vrxh1^Dvt4GtiIY=J(qbsCX4z+dAg_)S>5hICFvb|sI zh+lN+_ifnvUA}zt&ES;9fDt^FHGO*42h@wHFo`}Eyq4g9>l!jTNM3y^4!#caD&gkD z-;NIy{*tq2mwW)C`k^5+5dO&?v9r@)z|yDVBZ{5(fzlm82VS0PUmxrj{r%NyY^pFU z1Vh5kPT>a`vnh!(O7Y1aJRG3-HkbZ!xlg0W1CH8?U|g+7pj3qDNu|Tn!Nj zk4}@J@x;*rLzAJiqpjgDjgrgAz#@m%$P@}O8#6UNK7QUTMid70&1-EpvFebhwm-%dF?vYA&ifhAIH*h%ZWwRG=D`~= zdAO0~g_c*M<^3S7gytx5Z&G*%>WXXNAvi+N+R*Q$MWv@srQLbxM|cTLn9Q+HCkD)B z`ovGfuHHdh1rITq+$LQLB|tDhQxgz9Gs*T`>gA=7X?@`?n2cY*UEu0D+^Iykh z4~7K^Zp3#)DPoAQ0QVW;|*GsxZ zz5)Jzo{k<8&~Ym}pPkFj!}ldvSe|6-*t^*W1Rowhd+BFl=e{m=SgwBi)hGMPy1OSL zI_~5mI&7|p4x3HjCS{agef8Cz8o`u19>v!kQv@}8zFIGt7u1Vr=ib1%x6wZIOnMFd zj_X=O&!T;)Tj^Nn_a#Jd{*38+FI7tO=!n}%8Yzjz|Bd>Y_x9YmsSxx*tk3P^5gjxl zI#`&D2<{k1HWaM%5)J-F;-X!b;I~1M*UA)%Ssy)Y!YRF>rKjI6*Q%7BNV6WMJc0fJ z_DFWf)Bu216QVh=Dh-m1V?=WZT+tk?z&Xak%L^bQg|lyvP-+bJITfDb_0baG-~zMQ zORCa%B7c|WE9G8FsgQO8Sxzhmr(i&UMk-aQupQ+>IV578k67XaKbD;!8ENe;W^8zd z%`J9qkoS6*mMlh(lhcD+(Bn(c8OkHVZm9e2JzpOMT|7rz* zvs+sc<0Vw7+ys4=Oi^oDK~+E7&d8gwP07w~rh1w?MrA6Uhq+Ir>g~%vpNqY&`RZ>+ z0CkC=(J|P%^fw(#5WI6c&J9+>wtWTb;-+&j;GSSSfCj4B^u^Y#Tleid)HLz3)D~XFuJ#maFYl%!YZqf{`VC?OGLVP% zF8_%*@H6UUZ@%^R+i%~N7_3&h!Iz;&8XM^8SWAEheEk5C7J9k+X+3#Dk<`;ytC1=+ z!9Kp>!C_u{lpa*3x@;g_;>;lk*|n3(iu0UxMy3kn1+P2|+%ST(&;PaV|> zrmdJ9>+?U?#3t$jnNpk-!JelNU}-o+eipcvUtT3h&;QpNZr~;)9w)sQ((rX>@Jw*f z;LRNm_*dzFpXH8_KU>S$3t5$(o_ST42seWSauSnVr*`P(W&>c@?#>Re*AvxZbLzhd9+r*Z3I z6#g9a+)lLp`=sGa2<5|@JB)U&?L~i@r$uN#5f)w86?Da6x`Co6*U-Zht-gnbd%688 z&7)|qDB6#!Hvp0$XnC@qzUD!wEv_-v6je7@V@C~LkrMAH_gW{6+V(Nu3lNBJ(ftB4 zxxS5sYbW;k8#a*Fddjtr&Z5UmshAee(QI2PCdf;44(|OU;%Y#~7x#U=96a*PhE8M) zbagdh|G>~_8obuaJ$-@bGXQFKQt$q{yN`mI;P`b3C_=&XO((ZLBF5MynA@D}ETjt|?J!o<5DV5%sQc zWs(vKuJdkgvkbL&bauA`!58s+CIFvTW8uzc#AG5Ku~9s|1X@K8bi*U1^Om-@7alxx zINxZ*_Sm<(^x&_1f4wS(o{BGCyjaxKiWn{O=c9%qgx8Mm@n)k{2x`W1vI((7ko4v~bPq-Wl^33stISo*StLU;a z!ZkJ*%k(?^(--r$q{1h>8vDT2#H^J(cD=?EwNav;88;(akOa>j2HVlc{QKrtRHR~q zm@By>zPrt+o*9b%)=ikZ|FtFkiQ~>GPl-ib-t)`(>S}BrJ&ad&?*MwY0~!V+Su%9JW_9V>M3xw_o+Kggc3sO7t3mAQx>VRL zI5eTYp=HQt1va0ZS%>$%uE9-oH5I*bCwhg%M=nIK%tx;T4-exg-e_!UHkFmaCDhP5 z+|g9g*kHu*z}T=EXYwATXR@oR{!8x&MV5V8RaMJfAX+!Ex4;WO7mH{L-j@JSay23; z)4<-ekiAi$%f3?dZ7P^7A4V5F%x%?m*VS-(2sc|yU8J+JQZcLU-~}e%N_@6IU@_ey zcr`7Y{Qhg|cj;WOvcjvjAPMy}gW*LT(*4)Y3Kofif)L8)?-MjNCO9N?Qn1>~Ds)czNOc!h>w@7eqxZxZW@bb_zrN5)$L%qQ%nQ(z5byKGuO*v*m-Og?WV?kw|sN z(m>Vdt~)4Pz4zZ+AGY&#eU;{`b}wxLTi)T)*{MVW@vX zLfrrV)}NT=|7=b0<#OsAvyLxGAt*xpkuqu--VhOnsh)2WU{>Ph_!NvAlG#OOGvZ;L zL&i<^S;CAwY8^U%-ZqGta@2vjz81_bP2?SZiVOuNctrRjT@~|Mzhx{F>mD`O)ip@W zi{Sj@=j?M*m-^IIU%GPX=&|FaWtjT6T{;*(bHxhmtg0Z78ULV3j(qf!Cr?(^G_T`N zKXq#QdiST^ndVthbj5;gK>X8b)HXO)H}h8oaBIL$;mRk%+{Zs>U|3{~PycQY%+umN zfZn?@!FwJ5((>h()(7*2gz+~&ub9WnEvc%k%+GOz(^Z&z;Mz@4_14AYSa9WB)|qoT#Z~$FRn)f(6DbdhPog=b$X4v>%m~vj%Da^P>E>SuxVLM= zM>RJ$HR(jTXRn%hQXDG3!@e1$`}WxB-q5Hxg|R;SyN@>g6fUYccP97fPv3t3!;Yij z$QOSeYy2@LS|SOVY&-Jxx!m88yRo}goJhj+@1DQt@tb5#9dJb0K}Vnf6!V1(I$@-+ zzNoe{4SVFzsEoj;R-`D4_U_(s*sKYgf9Dg=ojZ22OfWh9k!PQM_L0T!ksb6V-p5GS zmk0#fX{%pd`zkn1GqFp+CAbUR=?BsfIEqW0xnbi+A8s$`$&_-tyq+FUPu-NT>9fK` zIAYdJj`p9LG$kr>+U%$qvnJV%jr_RzNx{)c)2Ag(i}jy8*&ipCOr)`zf8&}+D@bR> zp7ODdR>b4ebj8*~Bf%;}&^W-eXG69DP3xwFM@0mvlwdVTiVg@+O9ix4s|`(@y)e~3 zbZTO3L||x4bW}*7RKS-YJ|@6!9|7Eb_o!F}^$7B^o9h7w-aRVScxe@)Od)u|K|{>I zBgLHo-;W+&g&u!(yvOfBk54t%)|#7|032{BAGz&CdFM|Y-uv^wyzzFXaI5xB#{u@r0vCo12*sr=_>8u7qstOKO?{FJZL~bQqgk8lVHU=FWDLRR@fU zy?YBzoGL1BXOSc-6Isl~NB80D(1nsZdth>MviD%S@%(o>o$f4k5$>xwIXQEf5Twu8 ztPXUq1Lu))F_PtfJl`vxazFb1iDiN8<&DIjo=OO4J??r=*s~%51UoGj{G z7vui=Blpf2&IYZ)snPhAi?I%=9+FINp^%WR zBQ?X>R%b%=es}kn+(+Z@F>VdcM6VN%KRNlIKnTt7l~M-8%~(cydHfy`Qo8BECM0?Y zMuEY`sF2BAU0H@`Ae`}(<0uZRAVW7d>-@eBNiw(Y+4)0$zG1KDzO!H?CMM?SQE;}jSoJjMXZ;W#5uy^u69?X}hM#`iccG{nG*t#u zd-r&}&1pubwY67O)*6sAfU~3gQhcZ`t7^4g3;mfiFQ#9dwDOL-7f(`H^Dss2$f~Y2 zG*nhqw^;e(!8sNtNDBAWZa~C zEAEywN{@)i{qzejFU9G?mpD|?BLrnT7N;Ks|4I-4Q*6&0jQna^TG|v*Px<~WTekde zVB9?jX^Pz0(~MJbit{H?cJ$ok#rQ@t`3o6BGtwjQBuL z{vfn$fTvK&K*4MWceSPTg zH-`#qYKsq$#bx#%&eY?K*4$olBB!IvSVg`n>&zVoKG{)n&BH3mDl-TsMNJ7T0}2!E z?8Jega?_rl_w3lZdDk`P>p4bqo2j69)GKOQLPEmn(^*xtW@J#c(B+cVye?=ux@-5& zlQpgPz06L&f9;DeJpba_|8Uf%G8*?g9=$m-fSm!~(#^up@2 zxajEkTh^{xwf3=Fm)&^7G|k{O_wPfDX4hzYtEsHUf+@z&&wr%GRCVCvjV`A!r|ZNH zxEC$_()g-yH?ae*W$l$zl@Z_y1emyz^$>+xR9cvmot4+sX|~Iw?L#PQeUl)=lZA#}O>7JtYLeVs5fOJBkFo5B8MS4EsifsXza% z-ZHxLvuR7QqksVbKfJ>QW6>m=fU$TuQCzzl2Ep&S=e|B-n;as+`aesj^>>UTS465r_Gv8%_Dx@QJYgB z%nTAKv|bwNXg6$tdb8at2sVLAJlfrb=wh6J@VIlYb?EI6(c72N+v_l%)}yx%kLU_T`qZ6C%*%+lmH z1EY362PQMv*V)}aV%Ph7dU^->`zoF7?C9mMKHd0z{!rMmTb3+Yo;X(VE6J24JKH%p z7PvQ<(eL>A#E~nfPFy(G?iSd6?!txaGg&82p3XW`(k)aIs*ZU|Z}Hd|j|X<{Xu9k0 zM<0Fkdy{}4e)lV{ymChfZ7<#N6-baafAu44$Gy}S6(d=Y4tzp2?y9P& z=<%BsdfyF*F((Do}vmLq6t5>fKD9<8^WnbFv-`fHw%cSDN7fU zLAxx72qxS7fMWEX@@zEdNdTZal&A%eVf&Lfz`67e3-WS*xbS|gqI@fo-nuM4W&y^j4 z4CJGPV0*m_x9K{(^lBl#00$zrkJ3m{9komNlpO!<=eG_oJvV>f59FMQSkKStZk4oUC(t?Nz>?){~FH^%&B?b~uS65AC zB|{JNF_U`A(FQsB0D~_mDypigttq*51(D%RV}fD8l-O)S@r+4&ESkJ6XOG{@E@PL% z1$G}B?>e@7fF;*G@IBpz8GE5`N_Dh{9Q1fcK>>=RiabPQFjeLx_af(mmi!L9UUDNf8*o-Oo!{X$aQUVe>8Sd zLdyIpFyiN~czP*3y3l!YQD%huuID`z$V&!_fgqyRciTvf#;CwnC(GTz9)U@b$? zt-hwBtP@vjYvbcX=%RgF2=)#x_xE?#I73#w`R1FqCohZBCxaDMB$CR+{t`ZLx@5@D zmT7SS=PAU5&leNbWig~==oe3~fP^`*GnXae2v!^u6RN=;Gnx@fWW1)b*32MdenmqM zcAlWPYin#QLN7F2sh1N`aM`8dc5M@XVAc^(hSt!>DHUpK?Gt}dZo+rc6GPOX*fY_E>haw;s zF+LK*AQuKXWk+*)8z7u33Q9^ZUCu42rp~xh6-Gvi^P!2@m0T^u%lM8D`~AFQveUkvIsr>dj-z*g!kWn=0dj|JtufEI{mq~Q>e=s)C2fy zl0!@%Yrrc1GP&sYcj=YYm8LpFeS@*HwPkRix3dFvX97#Oe<%|nIo6Ruy;3Cc5KCB0 zZrvR!xxlFeX=o-NDQMn@>Ym=HaL{Az)%9>zLM2zs1GObv%W{c7{K2XR6* zBc!PbTk)~hs;aBo;S2jd$IvdBGH>AxX_4Ll{p0a*;++8fesrc^ps*qD%z^K|KWpl+ zB7qyQnF%wsjk)zCJ!^Bot1Hq&iog8$y`OU~8O|R)dAev=JNv;W9^CY$NiqfI*>)_5 zzfn5gJx{Di!>P({>J!lmFF*g_Qjj@|$t_4E8lW1HGn6mki{9h9Hn`mwY%M*XwP(k% z+}Z&&|D~NfbrVz)Zz+z{2on}8T(KxZt_YJ-65TAza6Wud-NuUME`hVDB0o2;;LtBQ zSym8C^!K(@*LU?JUbvyL8#ypOLETlmzdl+~QrcujIK~Wt-3}P-jte_?9y6o0`4sEs z&UXlHz_cRSIig7^DU+cePLbTwK1>TdBjN%DQlV2fOX0I3UF{hX=^ZAEMf&-9 z$pvG*?JX@WHCK)vsH`=%!>usFlZt4mTCGN!lh>^`&588W#frse&X}zpm_Yr!MMG?V z*^&LpGiD`dM%ym_j6m`O<%Y3<=rGOa2WtBIVbSX%k+UG-IqU=nv(u=!_0g1hE@-+G zwY78?jQZcMp51!f9vLjr`9(#=#Du%G<0P-5-r|{f|4Yx^|LK>%l{-{P&!uCZ@-&G+isu#mWIQI$U+3q>EB2*nx(Ie!0KKbF9@kDDw3Y4%S|&8ftB|ht8Po z-BD!#h@dW#rbF(0?bTZX;9yk-4E*{{jerl&kjH|j);b!F#&~HV+Pw zYXtDI#xyjOidU-26ynIy(M0oz`Z^_@e{+6?ZR-+9bKpW`M2CJd_tBLNnw6uao2FaDC zep`1{w-}8k!~wJ(K7u>JRty7Mz@sz$+Ar+CYQoGa67a2~ z7R%@;T%6W1m}_<#DD$XV7b}(egN!!ZS4Qy%%CZ+fm@2ni`0@K4KVKd43e?a|l7yMl zPnEqok{SBCKSBR$&W7t&oD+EY;fPo#GS*qR*y&gOaqy6?jL_`Ud! z;M#V+1!%5|;Qsj$7vpx@o=O5^I%X>TeN!3^0CDj$c4sX{WXw8{)~>@Ad0WOJ*B}2# zeIR(|p@(iwV)#{2QB}HFqp`ULgsw<8sVc1N?*KPpOGj6y$yjvi(8XpWnhwxUE#^k} z!#?>KaiL;_DFMjY$!7X_&%13#09wojFTDgNj|2n+V)&%t-eve;schB7I%%rO(As7} zz@&#l#6va&h}c1(*w=(EwCKV~ZKQNo)>L1kbdU+a!t?t4h;!%4^wEc&l_wHX0#?qHz%cgyceDu(k-QVR3Gq$MiDPoiEe(+^Q}i#ta;+j2cMrm z+b?e6)bPj~Zo;Pi0+0ufQ#tsa4c*v7U8EXe#&#lDO~YgeHf~&=3<7tSNyk#U0iCX= z*72WQzHuYNGj!WBB|-@fIzIUDut6V*Eb;a>q$QhhE{%8`I6G_W8|%O_VQOh@Z8NnZ z!V(`~Bk%>lSk`$UHbo>9i;x8are^`m>ZGg_3FBkPpaFoS7kJr`FO5|Lc_jEkj*L5Y zgMfx4nbbI{(0F_Mc*6+I;9;wyL&g?NWmjn=C5vU(iI6Qr${M$?NLC;@N52hyN^piq zR-m5&uIg6!Vw((2pfhj8y|)^Xw`gj{DHXHsUgv=1RkqQv1$@H!P99&_0hn+j`_+eQ(_WvgM1U^HAe{B76Lf-za>?SXPJt|2;#Ux{T zJcA;Y)Z)d1!O#Y#Th^;7)B&28N6I>*^J{L#8VwCabZ%pV zm=nvC6DhHXa)>09QzRWDhzmAqN1YR zYEFN*2oR}HkV`KbZM^X9#xLtqv4vem)#(DwhUOMftRQW?!Q9c_&}3@s?1FkB%hb>e z8^)b@maqpx))>@joy2K(+Q*zDBTikQ25PQ=YAY2U8gE~Pr^ed{DO5gQBoZ6JvfI+N$cR8Uznv{=$FY zH(0(%LZAtpHyGaTEI_`nAB6dVJatCOf+QMIG1xL@lDP{@tVfksP$wugu>Y&7*SdpasK)qFMB&!#RS8K)1%Uw8qn+p{K3SzWPoYrY2m1TUAsK zPDv3m#=b3JHg1(lG6imOYS@xVjp-B%lH7eG6f*Dohuvak&sen>0kplCz3ku&-K2T( z#Ya*?SKe~>?Wqx3txr_io%h^Lr8AT_Bl4!qW-3^p4npAN2r;^ey;-0ycZR>d7}>PH z<8=`I`Yzo*oGBQ!IK`q&%0m}zG&Gu!$lhdVFf`TGH4s+X-o9=G5*dw62DFjEK-#L= zfS4)FE$(O=w2%#r0;B{+LIXcRg4vwak*NkIQ8Yqn)yNT5s+B6WN~!S;3<&i1Mi8n> zMScU@7ood2F;P5aS3~yO|=E;tR35>V?<=;wuyBt#H2l)8~L7?}4ty!{ySL zxJ2Qou+aS5C!cTn{DUtKrLe*LpMLrvJC*$hn}B8KpNvQMv(I)@DKL51m__%!`LBOv zN>njP@#**9e?N|i-+E2&c1ya#ug7PKyyM_Upk94jc4mN{BKMhd`->W(z2Lg~kI1jMNX&tl7}mSXJLtS6hYP z&yspjP-2!gG&LAnyINX}uv~CuS(WO>-DBvT!Rt8Ps1Q5duenA_CrmFii zt!rtmt23Eu>r4g`e^0W%5mkU>Vc;z9`FhH{gES8UR`T|j#>Hy6&JT+{rgacU!8OT8 zP^!ZY?20kW7d#&0jhSc6;ed~6%z=0Y%#s4AEdB@@LJ5VLGN+yFXKhv-j5DyHojULQ z(|b>x9E_ha&!fKf==S8~OW%IFZTnA~etN}Oi$(QjEDTCk32)(P=Mm>2XSTD-X>gu* zp24SFXR+(b!}m5;z)r?;63ZqzkKymGI?c{D=K#EjE0LvsH=D-#IBo1CHogRj?0D z{MeN*lC?ny_F;tZo!T>y zbIfn{4NI2IjZKX3;QxQDy$4`aRrWvr-kV-ClSwb6cR~W87fA?31zc>byx&Z^9XAk zW$t@s+`@i}oyZkKby;mnq8 zB?Br}EwyP=CKG@Ee`aG%w9@@z!z)Fk?Mm1BW@WDI+Zf?ZK{R2E@Ufx)#1^9X*#1|P6 z5fu{?4Nj7TgfZZZNgfj)7abKA5*i&H8y_1D402RdR77-m#HfJa;E=GG#N?1*Utb^8 zBmr3o9ssHP!Taw=&HUvqTo|pT0Z#W#xP~aw+np@1`C}L(4bp*|a8el8xCe}D*X(bD z8pV$?S&kbw0VngVH{qVy(lE27V`h64x+x9Kkp|rq+S}XF4Cn(fuTZV82hH8r+tu9E z)6?2YUeJK50Knx*i=i={G{;pk7g0*63_MicGMQ%Ah8_ElR?t6v(bXZII3dN?W=GAL zj|ING{ki`919{X|WtRWcCGP@t0{2VMj43G(y#N0D)H6=!Gp~z8ud5h|fL`$Wvgu9GX01{P_j{HOGSFc`;K#9EX z+}d=^o}~8Da~V9~bDK8fp$Z)u(!nDDc7Lr_r`4!I$2&NrF`3nB1Mnn3`WSF~@bLno zXhoOPPAgd{B4mLF<}{ls6wFNI6Iq6sQFMbGX9NPP2Y8vhKp6YS2XM^_B!4*-EuV&# zznRtQ)rjomhK8=L>s9p)b#<-qDOVvSnFUUHYZfXHW7bj9EN51llkXfwJ9^W=#qkcB ze-?6KgPd{nNwo3%^x>v?LE!4jW4-G&H2F6Kf z7&0Y(^E+3IkbD5L$|8(0H(5^nk%r&ZP~Y9t*xcON+|<(9*wEF{(EAePiKlW#kIn&tCdL_qP%gHrGzy}yEDfyDqv?h7)95?!eBz0_s5TdW%o3^#R1Fo> z^R$=_VeWeM^;D!7$4-YTovWM&{EqtxHZStba8WfV(`bL%UzO(A@7TZ0zo);cv8TVe zf#iU8^>(*_+YPta)P&I`#3F`AC@v_{UlNIm9v)ui9}p<^2n>=+uxOHj#7yQF=pmEB zn}}4F-`&C0rq7{GpGTX1iZ*=?tz!?5YpKb72etrHtN4TZ53vyufOJL<4ME9q(+g*s2c0`wzYTnbXJ%C zk+-d_b?w*PUEe{u-SoK8%pZH)cyO*TVgCMMs_da56KO$n@B)29>L+i z^14}_<)sc*)wN$dZ{#F@kmx!mmbzL6Lo;cvPVk`o2ViGt+#Vn%R1y420} z?L8e`EyLYIy){S#`;;dQ<5s0T8Tuuu8+n(om1sO((3B#%Aa& zl(1z`($6^JC80wl=NJuKF95O>^kG=ml8eZG=@*4T3Eo#ON40j*h0Z zl!5csyZ7F6=LdfWSI z4|F&eaY%CHOa%jJz!DYNi+0#r?sKW6xDXm%1QsgD@etZx_RIQ>FjU>Tev9gKdB%xx z31nsUHJFU$*Gj8PO1hi6J6d|%2>!Jm2JuRPddTcQNulj*EmH}kR*UvJi!TyOvjm=@ zp`KhnKkk?@a3hGP+_mU$_dW8&Ba0V**x!$$k>bPqcb_QDy--|QyaA*WJpL=M@Jq*y zE6pScWs?E*dXxV8>vY8CXToTUQKe(n^QGBTF8u}~#>W7;ilBY71U-G^+1GRUa4PkM zZ+TaLbIs+FtL2s7efZ(V8HsVz7CrIo3(r2CN_~}qg~iId=r*?ut-M`$E z*IhIy4Gc+E3A7xpG|>)vZ*5kTObS*ODRFrF`>?sbA<(`8A%M&@c!bS^p`n3(0l`!# z8c;rYR{GnhLg98zbV9%|}JDsHTzrk5`q*oaKJmKFzN}lWoBH5>)Y~B%O_Sqj? zI6rkqwf066?8tbPX=~J^vrT}$N_)%zMZSDQ{ zhk0W4n^e(!VsAx7Ef(~ZMdu3&E}T1;e{}cW6Xy#qS7QC#(GFs??CcY#&zw7#S9rY! zjH`()p`k5owrOY>1xqk!og8H4h#A7lsUIPZKWbZa;&6(jN zoy{EsdT_cnRMvvCuvhB@a>J8F4T2@AxeeembRxwlS$V0cc}k9cXjo^mX9a?I*J98i z;yW~`H4TG;OF-H6dO{C_VsNf?8JSbwglv2a*?1AM@eE{R9+iQ`=>lj)@;;wZX2j|c z+3X%NV~P-o#9{FePk06R7Q6|Y^po;cAu#W@0pW^puuaL)u;NMp-;^QjU4e6AE8!A z=0z&Hv)5!j`TD=#Or@ffpP+J~)ZbT}$uV_bK6dQ8InXhW-Vm3US?gPbcgz6`i%Vdr zZPaI+7v|4@VnHPQth2$uK@0-Czn|(%kGoiofLK#?@%jARoSc){2M-@PnuB^K2x@n9 z)n3d!dHl%!M|Uef$4{KqL@Q(QV5F z_jsOR;P%CUMbJYCRo=ENyaQRtf-HOkS$GGs@DB9I^tRrf-kz=wxaqqAGi=536ONpY z!Jz?YlTP^Z>k$klb%^>ArNEz{Vh}|qC=0}IE-K^g7}J<%9N@LG+u;WRAGO1d_yrF* zNIsWEAukON3*5%eEci#3GVl?rrk`W1!$T|Y%Hi{OMGMIUFYbt&cwT;NrzkJvMvOrmI)$>M%Hq zOKK?=9Sf3HM6BRMeI|~oTz^xyY^%W~MU3&!WFv1JAqS?nvRUQrTTpo7R9C`-kKT9e z*nN*b{=}?c52=6TEclS`b3V)d2wUueVCP2uqO*NIV;lO9C}hQO6JfF4Rqt2 zS8KZlh74?9FlzYDICkvt;lsnJk4(bj-FH8Hc-**@X^t~nNb%#xh^8h1t||n$WS60j3z;V5b;YxR47^Ae}|R zRt(&JsxfSIkCVplg-$1y6I?cyr!20?Xu-;r;R;;^VaZW!uD>u!krVq;3{Jtu?EJr- z)V(GpypYk5$uW@05XdCKY0H5@X(#HnvSMHm@s2Kp_Ln1EgZMR~HY$0;&`?9&z);iJ z;BDD^zl9&=yprECcWz5vYHHp4O8WixmDKwxaddQKcw~5VbWX*;m;5W0`jT~e3w588 zcLD4PCD!Fno|I*c}<~D@iOD-2)>uzt4(K5W_hY#-F zvf{ZHEpFYno~Ywz}rAlTaQ^;a8F_51tvTX(@V^@dUzqz5Y&Xpic#48d6R zh>Gg$*4iNjup~)AD$8(hPgIOorWCb-YRb#Y>-vlswN*t*R(r>gNg3WzR|3>`aZQ_= z6eI7bMvAu0Hqg}8*FuURV?C^~sf3Lq-ZBrS_58uL0ONiD0QAz;dwPSS0>Y@PJ9dCz ztFjSs{xtijf2+M#lLOq2f!N*-3QRyzMz`NYobnWkfq zC-Qf!7Z)j&fvx9`uLr_r)fa2O1N&HxJ!LT9{x?4S@WZ#?{1;al2w@e;INFpbKZJN? zd|;3ahVt|KR;^lfTB2mt=3Qt57mN-xt&CpdK#bdqPIT9AZnTFH;Nl`bd+YwsRG~8+ zwe?NLsCf^*{NBI+_0r?1o?!XtZqX#a^N_Hlph*762X_L_^BP`Pg6I@_ES}?F$2>^C zMZZfwPd`V$O22|KRu|bH-M{e8zuyT6?Kx@~W$YeDZ0H~KBXlIpU!hVd@QFyCw|Md5 z$6tCU9c~a;c8T!Gq^Qo?Q=3rxOht$+-(`CBsO)Pu4)sLInM^5*EB1f^Gb&~@GWAmC zE7PdW@cBL*p=8?IQNh`1?CY~&Evwi3Qgu3tZvElss?#-e0^*|^R5J@3Fn_dks7(Y_ z!0_mS`g&v$YEih(g!QkPn5lZCEj2a_GD2b(vz$gXQE?q6t9_)Ytu6NY`P_{gHto+V zDDFZOr=_UiyQLcs@7Z~=6x4V9ZRG_hCU*>FnW$<=yU52c9Lb|mL0%+RMN?lN=r1Kq z?@0foXcd<$^YyE*9mte&J%dp7ZcL1i05T9iNy!Qh^5kcR3Bk2(;Yk?13Bn$m0+IqO z(AbQPFbRO3(a9F8X7Ym+04_u(O`bV(+LY;O)8Yb=>K&InW%}d^r_Y`~b>^M-%uio9Cp~>u{DjFf#*YdQk5#Z7L0-Yh6Jq1WlBSJNm>i*r5s`@# zCZYK4gvp8V@uL$HQBOYo_wPqwLds+p!YH@i4_oZ%;{NoJBQPFx<9?L#YLP!~MCR}7 z$Aw3WFDA9OMN(>L06`OjmEm#RD?*Y5NN~d>i{vI7KvI2U8>+W;w6}M*wKUe%)pvK* z)q;h;wyw6J0fFNhe5yzMyc2!Z-qrvg9zky*bdPptIUOn%<{U>RFk*nRX2HZk(F%mY zxExi|cb|zBySK!R8JnmpJ9_W{_z<`Jv~kv<>eLNk;yi+wi;QPjKSukuba z+OncLd(aDnqhS}dg)S(#9}sfF*>e(;_ZGQ3j9YuHSQ8Zb zH?WCiU?x)$_VTJ9XF;#w$sFd6Q?Z_S;=ZU%_LvznrUhp*Ttk(TDcy~dD6U^nS5%|n zkHc(<7?9Rr#A*oD9Dt017GXd|NMeV?bH~txnnAAvW$6{11f1>Qm$q2!G@~NRCnl3j z*K8&Wg}k^9gv}98QY9f5*Owq;JpqpOLv2iCbs}L3k3fRKCdD%GF%X5sGbyQJTwc+4n36# zF6~*a)PiN4yMk@D;LxSPEy%mjY7qwRgF*sMWngF++yMO}MiZvJ8eAEQ7){W1tKzP*E0hQJPqzbT0uADApF5)BP35zm@ zULLqi;S(g$b(QV?V&x~mpV9S~_J6hFljY07va=aGc}#u=S&&O6dMNpF1%uGH(?h-- za6?{yS7X0?^vqXh0kQZ!wTatQTVsq(yLUEL&bu)@z#PqVbs^rEg@g3wt;xv9Sd-VQ zHn!g%!JGq6`mKJYw5x&i&bN4x=3;#?jWfYLe& z1o`{}g94>eNUe__Bp3lI8IUkHce2^ZPVi|u$;QoJ!a<86GUQ_B=b*(nXt7M{Yvs}m zri@yJH~Jo2ZJ}~$COq=kjKx=khlyyqe_+3V-lFLeux2J`f|l;ChDOiINw{q+tmvXJe;fKmFa7-Af$_+HAR>QVB(;%Cq^t8uH;VfP_e$^T9y-f zHFX=xA<=9#Dv~O`C~VT`wL{fsj&A)dYo))RsQcn(Hz4w5=X31kxz%rB+Eg+hP+OGX z?j*0w&=o5vFb_1fq5q+N0Wh(rJo50vXfMy5+WmI&1cS>=?NY^{RQG4ai^wYt4h!eQ?y3I(5T@w(A&|~JBTDx z8?w_z`iD±)A*UftV6K-K68$R~MahH?p`9pWkm#4Z3hKm$S`U~E|IBuNnuRRStB zC=}ja3Moj7ycJk!BV!RjVqYJ4th~JesPqM10_7htmBs#iU|-jPU44QByOgc#z?gb& z9@uSP2q6PP$N+HruDWSL$POz$0KU=;imhUGPQy^opi)(=2__sqgW9ZORpnyLWl%O1 z>tr*P==$V77Rk8zJC5}ljzuzV9>cK`YdXYKSLbk_3I0}@mq_FbEUv)P+gaDw#ukW# zw*LBh3iT%hq7iEYECyCXWMqR%A`sZvJg12^T7(L&1N9=!=0U!c6cES+)wP+#XjO?~ zaen^z@#ebxGp9}!l!8C#+_B@wPPNLu5ZY6wr%j3Q*|$DE?XI-+h5r1LsJH_DyG6=@nR{q+u= zdZ@CfuCMRniQ<8QzUIrlSk)Dn^ZhS~%2&4~C9IAPA0H)W%ox{u!k96CL1MnT zJz;Tchq;iJ*^rheAuV$uEwdplVY(4@Pgm!_h~8kujA7C0j39LP;F$D7t;lw38$#i# z5Sxv|qs*?n9iyFx!p{ba-A;4C{{d7CWi_b6YKEm)?TLnJ^0$2T)!HBPE>xO*g2H{Q z{ndxoty#71FkxS2BL-x(WqMo9F!-qzE4J6$BOd{S_Jg5x(~cD?!Ts@SphO2^@Be5v z<_r`LfpvOP_@CDn`sw;R)IO8{J_ljY!5a&yjR36}l8i!@M5xS1JKQ5zfxhrZU-+XhfONnM+}oEgS5-7tU1#IY zGFb+L$OvfEht3aqyujf3<07a7%1rtrz(UqjJYb#>7~oZ9XLnc`dR+xPF(D;_{@F&rHxw-H?YT| zhR>I%QuaCZ8T0YuGtx;RbZR^GX$G}G3BmY|yC9^mDElYn4fCI~bkDhSb^1)6W1#B# z<+H~R&q8hsW`wNp?k;Vw20*yJhL*V~qbl`Xv*I~yd->Z*%v7y(#fWsI`$QT&4o z!7v(GKEc7D53<>GYNM03^fscneOCC`BwwpT;U$vZ@%u1aS>jQK?Pcjz% zR;0Pi#+_mH*&9nKNhNCr#|&agPxV z5r&nl6TSj(xzu4}%-Z20qa%x*DD9{~uHj*r%IkbyUX;*#`uM?fBVmkJivLjApk8#1-Y z^;|A}XtHcOmZcT$1#KkQA}*~No!I%yhP)C`*|_3@LjvnF0Re`Mp@ zjk^je3NK&AYUbkE6Icu#J9+xdnG?sd&*dIFe(=!Av$=&u7Yokkoz1y;^*Vs@mr=H2 zJe%EjcI(!H5wNx)P!7y4=oiR3lyaJ(TzFB_rp1dPafT7wh@B11f^{U~a}#YXgYb{p zxeU?Z-Q8NS&DssR9weA_Yi%BWo}L1fLbN+Ho!#xj3~d1$a(llC952c+jv25X4OS=y zgAr~m*Xe9#I|ckL$0IB$X%s(GM2mbuw!jy`FJN8jmaqs&SUe;w84?x&35$S)$uAc< zOm

    Wp5w2-fb{UFDmINb^#iv&?cqb16a#&HJSe%>GI{b?bKjuW&$k_E z?CK%NQ3U(}TqR!&upMwIG}~Q|;2zxDy|}lRaBn0#^8wsjw4Br5&{S75K%*YI+%q67 zr2F)l+}#G_{^JL)c7b&O{v!`xESZ&_vbf2U1JR%qU+m!no3U;C7fNV`>S{X@s=Ykn z%+2ByREGQc>Wkp#*@7QzL70dXv;CHNi-H%P?K$C|d!8YvB0Tx8PWCfh*t&J= z8lYnU&n2|H&j?FP2jIH%50CcaXGIqkUaPLCscUIy9UReVy02fXIeWfd9T730R$r`c z(_!+p_g*DD+n&QUg@s*M3ey(jAW3J|vZVz0?WCIE9PLJBG<#NbeK$W)A(dIxT`f%w z^%bS%j<{q+gwZhbo_i*YWM8NU^sK28^HZy8HwsKp`sHWp^wCqfCu5T&k&?z zkZ>Hsh-BZ$6kLoxB%G3Opbr1B10u z(T$zW4)3w!#!r|q{*JkT7G^*zXWmX<3R?AJgG(B%UjD0LP=~C^sip%%qrDAX*Gevo ziQb!c;qyKmFKSbD_Jw!s`cE((9>Dks0xqz8qdgJ zzW|PTxDN}=rkVzR%GBUQ)P{ZNp_z{JrJXvBwx=21%nV-)o7?Yd*If4+dirDZ^jGNV z*KW*p6PojKEBia!YsxCyySrK&yOEgDQq$O6SzJ=p(AL_}P+wc$g6J@)2O3)OxuL$U ztqTkUDVXXKLM$4M74sSL4Oj$c;(*5;8CyWbTWsK=((6q&q!r=86>#iW?5V^jk5^N{ zell%UVcEHF0TKgtPbV@~oZb3)264T<07kl#Z@-CJ)4cU>6|6@_J z1sraZqWP^?HZNVS68$aS+Kjrv4SL^%`-n?rJ++pTSJ35?^2P#8X}hr)`VLdwHfYh~ zm>`MMY$vr^^x~s4c}+FfJ7r_i-<*xVAHiR9P)y|C)Z4fWX=p;MACJytcV5WO{wnf3hMoRV zL4wX64bWO;)vanP#FOTJ9U8Ir11Ib`W}gof`e5ztq2Jb z0V?7gDJg5WAwZ8g@Dd5ZeZXSbO%{eP^-x9fz+H+UzCDfMNre(URhCc^)mIXGK2ASMSrsd z@h`nNHaMqj6>;1!A0yZc4;o6163VUDSveKY%tuL(3~GuJg7N}Uwo2sbWeGx}V+RJN zO-UlO5{%9j)y)zFOJ$?NVxtqNJMksbNfGLGuM=0|YL??_w&H45;%X4npk_9=A^rj1 zY$wv9@eV*QDg1{FI>bim8j#2d>;&+F-JNZ9Rrsp84emwYD)E0(@uIPjX~J@MCaQnI zwFzt{LL~?cU{ql7h42{AI|$ojnP)Ttbq3Bh7T{Nini<^_DTtheYL(+HhFDh;U(P8RjaCwpr9Y~jhU)9B?68?{6Bfts zF~+`R{y}|&Yvc%cY&(z(dY#!~&m7|bKan~4K>VlG0GaIVQN=r%`26Tps($dna*UBsFh{L+#Oiq49YZKPzO^tO` z)m7I@$|~z>!OUIP)Y4FmxCinkD{5-0YMDkX7-j;>!e&b(a6s7++QIbzO5yScvxHKf z)eb^b37^H~5UMovAtR6wS0@`wY2gQOJ^!vF!LOn9)AVzY_nz(#SxEJv25oGX9`As>y$ zV>=8cBwn&I{mB9s$`syK!1;J^(&7Ic5f6jeZ5tEqPU_Q>EDxgHUxREK`AN-Wmfql>UvG=Ak#aG*CZ*QKyrMB+Up6|Z?;>SEviexN-ycaE7w(LyY2Oqrl zKw4Vb^rXlD9`0G&t@D^ZCyFX6YOqC4e(pc;c0M&NM$l5EVm~^;RMSN8p?j#W6-(xF z$^ieodg}1L?4nCo`n_WpemDnmw-L`Dq3c(QUz?po*PYEdR%q}Y|BrM`gx?ZkQRZ2! zlR@6+KP^4|#dJ_RlILe3bNB!gz?4vTHkmoNaZ_hib(g_rd-~~x3G~2qM9E$B(wucA zYT?YMRTHevk(SC94MiKo@rkivVqIr*dEw=vvK~jTP8J^t{vVyjBJ%e4_4MGHhMFoW zsz4X2C7Rsc-_~XV#Lp>6Ne-dGXROxhdj_p0qa@Tz+c#w8!;b<6GjLsj`4K}~uUAb7 zjR*u5SZGsl1wOtCmSW1hsi;UsLy+Kdl86MDccA*Q*dvoKR{%)2i+V5x0dPp4t)ru( z=xA>4#cJTahU|QU9!d!TK&wt?b%lGarePFRgp02JedhLAe`{fX1eZGncN-@?#uqtS0+=r>}4$~wwx&eu8Y zHd(NEunjd*k0>*Qt5w_E!S=y0rjcbbk8nR{ztx||LfrwoN=1qp2w(ZPg-vHB47$E74DCx$6lHkCqa4VpG<|Bt&%`lJ)*-g!?* zU*D8i;vwBEoj55ZM}Q{z9!gMCk1 zS?nzKG-3U9_*)KQQTeUMYxj*C$~k%<$27|#+`Q@JwNz>?>%+GeKa~zTpUqHH>scS( zOFa);Gbr_*dq9%5#^qGmM19VCK0R{a(xpqWzpQ><;I@UvrzD1nmqs2udf>;kTMp!1 zIDI7Na`Bbx1wZxmU8~rCU|_hry|)$2 z+OfB}zp1qrJl>6$5WP|Id1ZNL&Q@tzT=$BL-~b6 z;5U6A+?$>cPPF2kt_)bz#P1=IaU?(cJ$`n1u&(y~@C-q5?!4jf;4^6miJrLW@NoM& zdPgtwP!AWJR}YOF>mgSNJb(6d-0*X>v}AcCkL3iX&mw+~z8((erk}&LIAeh~(0v!g zMkZ`J#6~6&!o|3YIF5Y?kXZvq%wqNFM!Ngij1(mheLd_fMQB)jgN=)=8g~^Be^$vzw zwQ6_$h+xX2k3Kq?J6yYal}b2og6T?GdG+}%$KOe%vYGVsN9JJ}w;gSG1Z%d_0M&de zdj8Q_epZ9V88-Dn1et%pv(@QoEyulJ$LS04Lfr7KS{9#;Z>BhFD~n_v3G4FaE;fU{|VevTvSnm>Oa>?>zZp( zA-?!>VL=gyEXu18DJN_u|y5au-{UKFLyjlu{#W>K`q6|{2 z2O5ERtgLU=vEYPtYKMk85U%OeXh)m^_bQ6?^h4-rQd5O^LtH&=raeG(#W!iKRxgpz zS7EUa15wqjrftaIZR{GcP#*AJd2&flv)Qy|WEk*Jaym5uST^(x8Yni8=g^{PA}ot8 zt#-u0Q^gH8TrMcQ+#33n%1bw7;)$_V8iiV>w8(V+`>)shbfIQ2?9K&?QbVXF=mT05 z5c~HuNV`=Tbp3iah|D-LZxK@Uz&{ryHeTAcZCjy57PI&r6+LA>>DQysIQ!tfI!!fE zmt~?$Gwz5s)zy}qzfjl{5H%CaoTadUR#QsBlXFMY9nH1f!W*3|<_$DeY5kSrBS&&C zoIig)H?OG8Oc7iCP4TCvYRob4GRN8)&U^|ugetD08sz{h&hFe~am0_0HdWV^Ub%X; zq`a!={AKiW6(E&$C~sa~O8VCX`gD%E_~4Jqphq8>J1!w1AmYm9_f~VfsV(r45~-CDBdS>m~`zZrqJAvno$^9gM(b5 zmwRY=M__1;jtr9!dyAHM%7bEJVt`74L)P8DgrAAzmoGv8-mHeIj3OE2W(tgO975nL zpa2VD2|ay5ARuBWlcpc&cmripMu9FAB{_X$0+@F;%}{U8kO5&m%|LfgzY#$liKnlR zf^Sg87^I_NUrn75=gn1@=HiiG+M%Tal4hVLNQBVZa``mMvQ*=UFs$_hJ$)KWrawpK zAK)z^A-?ZUHn0?_@Zdi`FT&P;c7@81+j?T>&Ye4UpRCsTr!88v=-vf)gh)mTw<2JN zNSE_N(eZx%#E0Md&wu`t>N+UT(?7rm1!QFe&*TH#udA|Xag*g14&w#Y*)k*Z=4^nk4^@K8uF6bVa&6)-xW;{ZLUg)pO7 z7!o`bVM7~04GO$H1)2W99U72Sgz4U_8&>OKSb?+#7NO)u{Lc%$N2)eQqW63;I(?ys zRf67Lz0RiBM&Q<$xdVU_f8M4;YuAR3pZM(j>G(Q>{`_;*&rv|`&ht;tefw=xy93WN zdq?g?@(WiKRP}Q}5SAW?ob`JE^f z#f-zjvly*w+v?iJV!te=iERI1KM&V$?XK+=*Y+;gex$|zXukm0Z^yc}XIxvSYd_1M z?C0b9ZGvlCF*nhI$ZCvLgkS^j0C){n1K)OghE!eee=C&0Ctej|AIk%-G5-#y38g5%r zutYuKb8b&t1o3JOk$+8NxJ@4JUsf9lHVaR9+o~h3t(Bs0Q+*^bz*XF~3UM1lx35RS z(bIA0>8a@HDD-p;(IYZLM_=Ut7=6N%Mma}NN)`k6S0<4c(JIkXl6)A?C_Q1T*bhv%<|m7~IP zeQC1;WX(jde`PFr5hWy3@VzB8m!0g#XVZw`;Ya^vFu8pGzpIl{T;pbnYun-4f@B*y zsqIh3P@wC#ZjF}f`fPCRhXCFm>Lj-7w^G;Ez27+3e%^nwAM2L=+`7!YpU*G*vC)$^ zBmacT3AjWgMyj%?WEEAYqJB_Oq4*D9Ma7U$ZhcGkVY{9zY(dRO_L+yDN%7i-7Z zyj(5F$FBwW4Ii?CDG@J-=Dc)O2Gy;irLn3EKuc8gwl(-7OGR^OymzYTm@vHG$RHv| zxQSK#`3Q>^sOa&5Dq5bPqNFG>p`y2bp(3TUKUUE!hlyw&BDn&R5YJlhz$R4{-Znp2zUAnUgz=3$Llg)|F`T2FxIjVXJ@dL zA`p?qOafdsi|M62Zpy%8xZlUnV$Y++9z%;Zbp; z6Go3tfc1^P#>d5lg^f;#icuxhfvg;55Y)pOjX@6t>QHw}okpV@(G8>euO5y$vlc#X zR5utL)TK+`Y3^vnaO+uo&iI!)b|4%~I`oY@LRa73_l@8)m6opa{Q92NVU z?Hj&A_R~jES;q$CLK7VjHg#H(m$Bec$Y9tG7i+Rs2Et z0q$xKNErnS_H7UcuQ{ANE|@i_^k~1@9h`k&-(H2n(SEi3p0`|f6Kmi7_wU2~qSP4v z)d8a1)P#@K^@Pc|iMX(J;w6<2&R9#o`eMxiyfGS$ax7y4$1|EK2r8ij1 zXr%JldVAF-C+Z2YSR}s}2w$`!T2iYwnhY!&K|`Av42?#B;z6s1e@5a~cJeIhVLn&I zC-ee0j*S9sHm%M=^L{zD+3e(dNZFuqb{*R-C%@a@Rdw(tle1s=}9y(N$k9J+}XJ zLCrO4*6S~R@b3@aG79O=rov0tv?(t>`rIpTfX$Eh!Cju}M;o+n!C)U`1&>QUjOgTB z&bO8PQ7HlKcR>W)X0g{>_mp34cPJEl_Z`d*?kaAh_U^m?0ae22SP;Y^u^_5<-A3$q7gjYFU1$gD*^?z#-)H{LBaBp$t_?+U>D>|}u z@5lQcEuBhbsG!*XTV4IEr}@x2q`LDB3>d|T$jZP&#UPxbM5g2ojWzkdH5QpA8SIPo z9?|Z2VMbucsj_Z7ez}+1{oQY7n18!2{de1l==Phl*w5GQfp#I+{)esfyYqX9Yu@y7 zZHHakFxPym`DNZDXC$q4>ukbA;YMO?5F(;`!(INpUVdMrLTX} zCvGdk*R@BpYa8OazrjB_-~YN_sOubV+w9i+`?KqGw~KqfM%OXi67XmHxv$f`&U4>i zr0f3Fe{z5S>-pUG=k7oEb-LU07D@eGd9k?e$-Sb?c72X>U8mPyxX!v;+C_I;`@t6k zO|{0g9pyUTe_iLz5%u#J{$)HM)04@^o>Y&D`Q0(`n|*&76J%fYulD_~aq*jd-Tgt% z4qO3KL*Bb@p7+1T#-HpY z&GoCZ-a2BAkt0A3e|OCM=3alfUW%n+h}V)DRx$2&`m=E}3ct=`+}@~5uFny!aWnGE zcw2^#el5|!{wk75kCiSxVfJt^$qpH3KzvoD7z>u2+3qj z!jwS3vNHpCVu8$42uA}~B;_EO#gQUsYs{=rhQ-hFtFtIjbwOR#P%n%Q4UKiQmhJg` z)vDzyzl4<*8&KCldwEC&#aNR2kLH|i;Q0Eaz5VvvPtHsn4UCVH0C}^*pP3w_5F%&JD>^&i(VSP(!J)JnQQ39e$5Y}%kb{?wSk5j)1+q9(huTnXMJYu*Uhd5rZn;i# zjHUZNPbUm}cz($Fe&?Brx=N)_^Zrezu2ks|nHjl$E-&Zs!6T&>?`iMN0ZfiOHv&od zDg7>LT`cE3U@0$dwW(%yfrY%e%QvL|>gIjTot>x>)UDADvIK+uUENZ4dv6c;;!%^K zw>{@XzmDneLD;scqSr=iE#{_cJsky19zf21#J{fd8yjTlD?U^(I1Chq$;tFwD7C-` zpS!1)vJsTwP;1A9V-V``Uzf3rO`T4{()$fu z4C@F~6`iw;$3{JDDM?YyU`?8I#~oQTpQmK;c_=nWkfv^2b}Y`AfHRURnbA07G|ni? zMP|y`bA^`*a&xmmwr~pEBJU^xvlSl#N?tSA3%hYxXH&@&Q@8Bw5l$lyaxc7{*nHWLP2mP zrBjNL0~9gu$&WtzXi=<(UPTWMDJ5n-EK*X%1!z%t5U|V@0|CdAkXVC?Q1_~z9}28uD9NK{P>2;- z=u`$^UOCF8Y1A3|1M)u&(kYjMeUk91+^2MB6G!6+;W)x1v{e+^Dhh2SKUK}X$*z+zw^o1n^EX+gQ zH@@i18Buq)Nk5#)8b<1pJLV<89g;k2VJHjlG+qpqfzHnW>}n}#qV>w9u0}&My$!^q zhJ2C1>;gO!jYTr9m*5C5;|LjOqa`@P5*#5ZZhT_$n-! zi=Px9HF#xL=JHQ3P`OHl`2mmCo)Q1>fUyswmgGDIi-nRV9z3t$$@fnl6E*1p@cULA zS#wj6*Bq%}t5D}f=IpAfufAT=;W=(DTr=NO+d0D;R>af??uH)C2Se#;1kk>rwgDP; z0GV`$xDO|Lcn8jXoz(Kmagk0CPvCK-P#)_CyrrzK>_Em%h5AqS4;L?XJ2QQRTs10g z_3MPOa=l(Yk$^5AnJCw*l1?8wd2HXlb$bsUJb0#5jT~G%Woa+Day}>f`00YnMR}*r z=bb)r;^f)$mr9C@F95ojC@=2hdhj&;6o6Y+t0E;WEk!|)*G6NXnI{ow8sq@+iaf}W z67gwOlD==C#}Y^*!5xV!-l2g$V6wD=uN;L-0>&iBMuu%R7Kjb_vH#8cYJ|CONE_Q|=9?e4ZT(^0 z#vi^z9_IRuKknGReG|w*(a2pK31_&M6b?WmkH68#sOQ*c;)wa$205E8l*myBRxS~; zRY^a7x#?tn&N8eKzCBaz9DmRJQ~9O!ttOeTx10)&hz$0P?#N%id0V41^v?7>Fr#eh-)zzog8FG2?TuF zZb8`wJ84n%_yCK>ECmdl#DoWnZ>Ru(%6)leolzPISFBmD0mMwBH#=2H2@@wwj^+f1 zg@#5%j1p2@p|^KPM0i-jgp~1otzI@dIWjUZDmvOHBnGt;MpL26#A`kQGB!IZCcs;S zGZ;W_$MJxY@%9&Rzz=Ve23<26QNIv0SwL{#Km|MiiFpSSL$UxLfYg#arW73#=76}= z0(7at>T(#Fe&m4)6xa?UZ&imcbj78W71ys^DL9{ht`M_J25ZD%b(JFel5v|er$}>;D;`foO9yRBq-_Fuf z|KO-7AI?y7by<&J%G|fpuxi`GQ4dR|-22MikcP{ks=5FMma|kNIIT)Bp~;D1&~pJE zk`b+^XH4?=)a2xyJClF41V|ayju?UVhD3}2V64X$B^bycp{1A|X9% z*B-B~Dk$x!KJv}FZ_c3@Br(swkSK{i!K$vdcXgO(PD76G;7ql zS=h-*?U8(}nEcq^?@#m?yik12JMh#DPgQW_y>&w-vT%n{F+1*H=S+B%erMtYpe?%6d zuv{zd?HypLxZ8JN%6blkMKht|KY_GnaNl@t`}VCcVOUNMpQGe*MSKV4Ay?AFT9v1O zC-(9W5As&{_ymMT`pYG-24%h;BAJg5pDpzZiH-E}iHlKrMn{JS`Xo&pKRO|H?1Utr zz-Vfrl7(6$w35pb@)+YFh_9F%{(5r9vAE;0H|{u=+;M$HNpVRjiWFZfExK?%@8UJw z@drwow!5=mV`e$ET9e(ZZEwRR#ruh+a<8D6gnirAqM+tCdr@TsizuXmhR8++2CnzW zeYnGYDt_VNAHMl+!`9=4vyjTShFZ#e@E$Z;2DL5&%3r1(Kot_cA786=2wB?73LJPq z2f`+^eyCq>vct@$4HQQy^9hRx_VwThe0*eBBSl5~s60cX$0Q|IgrlAY0A_1kL`U70Z9%Foxwe`>xg>etuEO3lklz2!nBL4hRJ`18dg z1Io_2amA8plOr)P|9sUf6tCgkxOBIEB`q0%YY)7&C95jWo~y3PCCkNsDS2270BNSt zs5s&BUQtT83$~R(eUD~byY^)S^_(hB&2aAOb2UT`GBT(!$_!U2PSTdk1c=Wu1A`on zOX0&hw=NTS`$mO&dM71O_h1fCClsZHW4HxLC=U1mA#LE|^c6 zkaAtLy9Ego^v#;dgbDfyTy&*Ywl;(051~~&2;wThtDcEq}^q~WrH*7r8z2SmR=pl-XjEfE)IBOs;~T3D$Ip~lF% zAGqMbmevR&hQAh(PXv*KCtido%`ZACA~ZB45b5uJUSL>}q3D38e^_KxWO#g#+;en1 zw8H-qlfNkygj@b+sVFMBhO&WIuU6GmBNL9mvr9{=tE!McS5{tGSzcOpwY0dn^jc+U z`Lz<*n13xQpZHS+*Eb+|^q80l5!Nfm4jubp{Wt3)JUJfnu*h&p?XE)yjvf7e!|nph zp~6G!VG8`UxIB1k#7~Gj>*_A#Ud*p;3VVAF%$*I8jD0|_pA^(;^zBWTFV_p6oeKk< z@KvmZq->{tV4m=9gg1t^_<(XA0`a5re+t+qn)M_ZPOl#T%Zo;bMFgHOkH8C}15QCh zR&nw!TvtM z9PTKnx4#y&cX@D5Lra{6ZmWSjWJ61wf|i(B({BXFZI{_2z?1J8JvKHAQ!ec!N&FZWRx<*>{Y@}RiU zaibI-VlO|CZq9_F^73TNHa@b1{lxX#zuw;jFLy{h5XCUagnTPalEy)-2`C*4X%UX% z4dpBk;Hbg%CiIZ-{IS|cJ|x9*nvs))+874nnZbx7yNb4N-+r_#zPhle zs8tx*-B1zW%z5tl=hx1-Nz=UR>n|=0P$`Jw3`V7EhK-F5Nc24oyYk;Y)`JlL_$hbY zm4>`azq_XkMc#ZTLt*ig`24z^Cr+Fw+Q%N3z6O{twv$lkKR@2OBug0l+ux2` z>}_K~YOtxK^h$ki+KS&S1tZs^=usCcuN|lI)sH^;yQiOj_Sq+H zU$!7i6_YULmOJlybi>9c7Gn)OjlXa4mES+Q_V&Bi+_U!4`_?@4w1Qv2gCRxp!@Rb_vBoO!^z=gDKA|rZ*dI_t7qMH*If_X znP4m^(C01i`+`9yc3FpGC^GYA91_XIiOP(}9(!!j&@rQ{>d>J>XX@P2q+8aY7Z2S3 znA1>k>EVZOi|su3^_O3MdD_P9=FEDST_A8)Rsy!X^w_a0W^2`jvhr?)Iw0vRE-9{Z z>T4P;HoSSY{>o{9xtuDrc_?#9RY^r-!=-bVsu3*I&=TLx^LPfVwN*8)#?F>Ah3Bu7 zo<4TIwxRiI2?WCFN+oa59U+&}kl=g#6E;4zDj7EUHv8}}5(Hq^l!;KRCx{?^44WL< zgKzTFb4C8V7`{+Ixjm@dBNFq51C&T96f&aNF|m1!kEcnG6S50#iVAoTY#tnJw-2>f zoW69~PIGKN^RW}hFElsTmR+r=XsR!{Sbq8P`O=FehPtW?XNpfAIeqzxq0i&>`2)dt znlsq%C^wqBBWv`v#U-WZPMiP;R8?L16{yR)1)c%FC|(7KIcu{yYUSE#KCHuscOnbG zWHQ;^LL~>Qi7zGw7ME`1W{PfqbxMx~*d4Gl^bHZsGREz5BP7mC@pv48GC>X|92k}- zt9W6FRD}6WJa?Ea_#ej7q&Xo`l*Ev5A$3w}R)^grA4wwu9o2&ux9dh^_tCS7$B>*I zig6@o!)hKndmA|$QGeRdYvetX3*bWQ$p5e$LaxD#w1D}i7U72EqTCguboAxOVB1q>B9+PjRS@}VvACA@QlK=N@f=|$l?>w%TMTa zN=J!C=rbPeC0Up$xJC+vldu9qzNOdPjzk*nf#P#Ng52~y7}Kj;Q_P*UzmF3%2a?{UcNk?RWvj_`lyY2@4SY(jgQ*syXR1I@r7f{ zmK|fcCzdTc!77YE9d$IdB9jWa9`(%*pboKIU>o~K9#J$~qmD>SiIB?F8eV(~pUSxL z3*nL8fS$aHo@_!-NT&G)^kjzFjOacBj6l-KO^`TEX7DkB|4c^mrkT8f&rJr4#bC19 z@ENM2!~OLRaHdhJK#T_*rcC7NsE9Xv~K7FCA z_GEP+dD`q-CPg?(J|g9&o<{MZSHxe`CsBtb+QXlJYC1!dFOwS~oKU@XBH09WsS!KJ z$zzaEkF`RlefQmM1rAQu{m(!5{F*eXVduNI4Kz)a4Sae)cI1nJJEk}EvC~&BojV0f z;?#M|R^9#J!y7kkdief3*W9ye#gZjCIg6JsyZzq#)~>ntsa0!!f9LY&URj_Wvotks zF~P2VceH!jNt1vH#2GeKlvfyssEDLVT`~po!|qZP%+MAvZ8E<1zGxb zpvT$m7QAOOwt{8^gvAcxqJ{z_QIiAJ1Z)=S9LXW_2ar1-$YVHME@-6#-Y9tuxCu~I zr$leb+rc29Cqf68%g%rR3IJ*%ZOUb`Xf@oSsAh{`K?TAEWm2R|M&TQ%m9D}s3IpJM znL!o~40^8aVw6Y&Q~z$?TbAOx>4EzK5!GO&gD@^ht< zI;h){g9w0kxq3bIJG|*=Gm*k;TNsT75W-`0Tf$sNyjbK)k=I|TxzTC|YWWuCy>~o5 zxY}{0BI+!sx)#L$5N|u%4SGOxZ(-o|jRchT$hAlw5z*gCO^0Z-DjKa4TAN#J&3d~P zna=1*9w?SiIYNcNgAVZ3^PDyZChWaBB87bsTLap823myHTK$jH}*|>ecvorqY2*2EsQey%P*;) zbX(|VJu&jWnTrhJEsV)vvxUf{gi)CquM1-;N1|tuBUF-VX*>tL^{D?syp2Y?-N-Wg ziZT?WPV2=53of$E7^Tez^jB`|U;i4*ashI|;g}4CoMDs_26rg1wsBFKP!Jx?U%NUn zWs(rik*f~~g#&O-k6ayA3gW6By*h=-2lo;uh-5qR1WVYzxOyZ;MXH6Lk_V6s1dfE4 zDaP2ZVaTh(IdckGW=3)m34-A+I&oqV%kUXoFhV!7W+H@>%Tf14aCm_*jsM@aodMl{ zxr-4iZ$1LBu00@To|q@YdkXP8&%j7c$C!`$=oLqf0XpyW*$bC1oi8dXJbeslw=##r zjIuB&Rp{u1wY&pyJL0L&kr|WEymn_)&t*g&zZ`sxlegQ!hI|0L-^BG@1-Ner$)2~v zCY}6Sz(ccl1aAx8K_DdYLmt<`kN+@AXWfrV>GuN%a}|Kw5i`T{pi-VEkH<))VqU(Y ztn_@*`3vXI6r6&^v#9tAX;4f|Vp>K<642_Sq+&@_46B%YGo12SncDb-l(f{eF%zh4 zI4v|u&|;H9CFsP`Xib=d=Azekq1UU?>sv6!)4-pJp$hlsF5o`r&z(Pa;o|uVq<{)) zx)m0k#|`rkwT=mi#S6p}6VOsD-~d+i`w)hd^Ws8~Vhg3?-6^u3{WGD(f5>$QyuyL1 z@>087eDkE~i3t7;Q)c5$e8ZQcl=-aIm+}xOBc422Nw-5Bb9fY!+53WPLqFvIeW4%n znmikQd>AD*Q&4bFgL)>n2RCxw{p-hMa|uz83cf83QnTl%Y1{ep=Pd#C95&(L@l&Ty zojQB&T+!)*Q>Ts}LQKJl6UR@Xb7#+BL3s@{2f&Okn5WhA#aEud1pNO?U&^B?KVd8{sP@ptX~6 zrJHdjV#AEra;Ru*c#xM-oT>v*{^ZYWCbBfv(!X);eH2d7x3C=MMLvDk(&JZOdF05Y z6?68Vrf$(G$B!SQqT1?_c)WD!any9Kt{%@ulv_MNy&W`_=YZiKpi3_2fcYO_N=l|o zA#~&kgw)V5Dy5@D&YRKt)>dLQGawg(A$esKVBNlQoT3apX$Xgtgntd}W8l@=6A2CJ& zsvU9wl{)g-&-+rO7UQeEZuf>DV#NS%Aq#j;@F{4PP%CpGRdKirv8t>O?9XG2C^<&H z!^i;!4DycbyJVFlo_(TcFc@V9v*tNEuHo|n z;qyk9*l_!Q{XB#u>acBu&%=LmzmRTmEsO2f&-?G)kK_kjuaoFN2P6^b$ohc$;)MELj!yLIscP($w4n7-9({XqiY)P?5@v9Jl8U!8&a!S zCd$kB6c|iswf< z`lXjPXH(zO<3(XZ^cfw!ZQEuhL7p&;TCAfB3pX>h$8G(fJl)o|In>5cjpv5Az0p30 zX@AQo?LVj2hv*g;rd#VM?FpADelcSht3}@e3{Q%SUzS#i5evxqfx@hwh+n*sm_zNL zJ^`Z@=|~i30m~STb|-ZaMh#-uU?LSUYAS2wRhUmi+MhF+NE~K_HWY`6F(SP3N8RR} z;P}8kT@F=3zlX(%lMO2nSPX2k3@s)jLhMImgo&nzv@NV}7-7ztT%0~z_XeV0#|IwB z&GxsD=eWuTx!Iii@SMudrr`5ihNoO@HuYQlCs&i=YNL7(St-QiYV?`h>|i#;kYU}H z+=b+V{x@(*A$dhE$yuM9%}v28Dmw>)3%*Tl!7E(%VJh@A+*b*%O=x=kxHGLbYb$6< zbGe(bVADoPXxv#HJ#iw&{S!JWDVy4f3l{0%pi)Kkdr(QQ$qXy z7cECuDl+#*S6#+z4f7CBNAXz3r`=LWOmFw#mt`)TduSn$l%TUxKVgA2gn7rOL+uqt zNdTw8JE2!UuQ4Jmj}qYe*H)=fz6nVg0Yl&|u=hAv4Y4eQcxQxl!dDSBar=zqiXOKi4V>WBUufx{>(Wx}FA``7h0>>nQV5qzAI0%!@2?t5nK@|Aa*yw#a@eM&yzl@2#F}z87Lo77Ty&+I3#505^ z2?faPA(A**i!f&p1r2t8h)j-_uJ;rVEA&e`I)Te_gRx{(1E^ZS=F+av$kLoQu-^MHwlv3% zZwzB}>7m;2djG$DJwoqawvQ8G0|g$|5o7?ah!4w8I47Ue-iQW%j?E{Z)rx#CCo&WYau-UC$WeK;;D>@q6`!i%f)J(t<4v}r-;Ifkfh%VwV2#P0gjbje~;!P zmBo82VHO;@2g9RiqPIoIM2CXz^Pz1Mi`3v?4t0Ut#h1dD-^7QVaP%&47(wN_b{8VA zla)#cx{S)jdFaDt!rRa*0}Z-d653B^l6qz1$V2w%PWU^YM@3X~q4p#5kUFV!)S9ZRnLkSo~+xu1J zbWH_aWvBq(YBT8#dNfmSKs%*0?e%hj=Sa$&4hLAl?s2M*(pB4!E&|BqwBV%R1aAI6 z&}wmP>=$1Q>~cPN>2hUFc5o_nkd-}B+wn@}z_SmpyywnUuRe{Y^PJA-pTCq%ea9@b z&YVAGCbE90F{}(!u^JGPZnHQ6MU8GqgF%MzM{r@Z=WqoaR;H3G)H0<;rHMf%Qxv-M z%Ws6JKH8&k*Y^k=YBhjS%y79_T0%YYdVw-BjQ*fkZkCS@?fxp3S%gSmQXB4F-i)-5 z&?5#NiXAwJzVK@*FJF4nxr>#cLyxR8_5cx_Q*F&AvRAf_L7$wL#*hrAnN#M^v@S#c zSQ)xxw!w&EG?-AAFVrW%fcZHB^a%ZmU;#v+R;gqPcpah@%9v1}Zu~~5PilIcf)O32-+_L?U$iV3qkvZ;NfI_i`7=EH<}S< zh9bzQ>rvZ4rm4f}Xf=dZNVyMjFFYP#r|OllQdhTO()O_qtXq(4YTq zVemi65BFf>P%@kR!n#Ul?C_P^rfzBC(kCBzVCJ}}9;3BAYTR8fKKINL5J0`Qx_a%G z+bf46(xy(C>TABZ-Q!WtS^c|(h|zn8dYgOyjKpC!*-o-OG3()^v{>N9G}*E67_Bhl zw%g2vR&p37QLPUV`Iv5WaB#q<6Y+#nF=97@ZXeAiYho30u}BdqRm3H$Rq^o>p;#)5 zlp>icK0bzQGb2|R>Kj6lU0M>!!4>>N13p|qDi_luQcx18O>hDU2aL+{QnCtK%=%V~ zt%Yy_(x1FEL3%_T79T53mF+p5B>wJLLi)E~fo|euSFS7@^TQ8g8vXu8R!SDV#-7(U zH8;NW3dq4D>m8ryh`Bues+luau~L)KhHA4myHRf^OPv_2caY*E4V%kFaYC}tDdZ}( zQl=r?apR|94tWV}hAxHl8!q)SIOHX8$ZUa7ds9MM1TMv({v{`9Z)!Jpx(9+hfEN{> zYj)JPo;zJoaQ1ZJ*@ENSKltX;9S8THI8}7=B>E?(Wg@`11Dq6vhle~KS5ICVcNlhG zj)>+)N#VK$@DIrt^ZPscup$cqUC;0(Ty}O+iZ+TPwiF#IZn8yccYXQg(Td8xq-43r z{Ns1285tVB%z5GDmDV<8`VU|4I9^@ZlbROow;kF8vT?+%Uy|w_p`VOhg+TEC>Fd9r z7Xafav(E|qratC3Q~Qy7QyGkgY{egHk9lArAtRM5czx--6)RROyN{JhqQ|d%c*EkU z=@V~CjEKkqkL3wB<=i}R=HexfJo5@ErSWd!$#eRkQth@i8k$N^o}4#L-QC<`?j2Ml zj-5LXBP19|Ub$}FgO8(P)%%ElB0qWKUU}v26y%3^tj(wyU>PWIM5m{xPg=6({eR&CpVh1=Do(T*XD= zl1D(3+1Wvh143T+K%3Lj){eoy+3V~Eoh3ZfMC0-~9F+{NcsKaXZ>~XfOmwWe=ssusEF_e2fAH-{nMw!N)ptQX3U&D zW7d?Zv*+IP$es7!wP?xQnN#M>B~v5VGsuJAjvqvkZXwdiWqD~n0tFNp2>Ltw(I6o= z$m8(@T&_}roqvc=BZH-H6y5z6@ZT-qzgxh6#4~pb_%G3Bb+kF0MEJA7d0=R53e^p7 zG$Q?;NJjBo)ZOwS6$WWAU?t>0V=jTAa2F02zN-|;Ku^E@QsLE>HhJovua7oZJGrT2 z)L#8hyTB~*lLwFQ-1*I}DyzO#-&AtSJ8RYLs}?0nPaVyx4NsFcu774P=~MhUeYaPSxq1x3#ptm7n>%g-ia(UC+-Wzxt{ zpsq?Cmq4Qb;+4uMwF;z;C&k^!*?gP}cMBH>h8jy75`#w8AR_ONYQUqqf(|sx*P-h$ z(L?cq5aNKF<=tfH+o;7;hP6S7wLyTod5FRUN9V%m`G#QPtm$dHiNKi4ojtp_IA2h5-sF4j zx40ft{KF4RmK3vsit>xaC^Z0?398f^0)+rIp*7%q&Md!QN6iY`9*E^u3i=bDvK04{ z;$8@B67XSs&ZYVMe5TA4O;crf@yaXtoRza&I#R!qe)G+IhRZ>5Iu2fVJoy~J*kL~n zzX**G(MU(6t7=S`TCuPNH^XDBsR+%JiXrqQlKbcom zHIL=Iyz;(3sNuenob&SWly&Q;o4|v0nX#qcrr0L|Y|dL=lfcNfGVaY$mnrfb0Su!Y;`r1(S#YJ)bH0 z>1eq~oD^u_6TIkeJx>QBOiDmCK@3Qn5@1=DXp;&7M<&CdCt>YI}j}O$2TaeG>N7Bm2$_^m-n8JrBK}fnLu* zuVXFd8i(!jWhmMY?6BCufiWU?H%b|CJ$`w_;9!K@@8JeGbhlf?%EbP< zVQuF?dgGk^Tq>blnmDNW5TZ zd0=F&lIpP}_v%H^>_yP*InZo@@#h1!4;%>W#4qxN5Q2eMIPk`m@|LEC`kyMz5L8TM z`>oDe8&FD(S50;rp^;k=<3fyzv@QujPgIaIGz9sL!$BUs+Y9XvEX?#<5E8~=x8Bsv z9lLDZLmO9*lX>fo@7(dl{+c>dHlYef#$PB`zv% z$wLo4bmxpH%64KqfRBhA#LzrhLS{xr27>3sG8KRw7KX3=M;2*2ih-xavP~_8% z4^Ct>I8k?^7okvg17cNClM&AVlxyomE?h^4#pp0J3-sW!whl1;b`&%~(7HsB7YiAk z8X~%&XRzN*`JtKeMN&jE1wCwXWF*k1`kR|Op15~zDpmB!2Y=aF?}=LYfWF$qQ>zuy ztz0v20aSgK{4h`5lHfi2(L!Lc;D`R;EW0^6e(}0>PoV(g18}i_iwa&JeD&1{+usGb zxqmP6CCqz#M8V@fKA&ek_i-+(O@=dwFgHwYp-1unZJw*ah{5hp)5)n(ZfkS>M;~|0 zS^wy7?~ZIc{zEsa_@-@t=N-XAFTcEQVs4VPxfT3h{Mk_(C-K&e8)vJLot!D2DJXxp z*gf%q7Z(@|WO`W0Z2QI0Twh-gTX9`=11gHx+8nL+HY=XhSJxWqt7=I3cB|gjCUAiE z%?()OxATFjOAYo9qdoz{P(leG)DuaFe29@8>>Hv2G@ng~r?diYs3-sQz~javv|2u9 z!8|R~URm3(@H;Gh+}QDHGQb-|C*`YM#h>av#iZo(J?5exFOsAqkCzvg2S$11cOyLV zI9X#zU>uArXd9jMh9jGy1}(dwbPuAJxXonLvN@&wQd4XJlri#H(j*+IXAo=x3i@1W)Ku#dPpG zPQhwSBBZe_haSl4j3iIEDS*X{b^CvtQx?b(G$@lH=1w8Qe;FPKcaTFCNx*r-&qI&D z$}$3g`lzFoRFZCY?5}c800o-)VQKo0SYb9`g}DdZ`p1zSYOb~24%?+!@2IUCZfk3A zce>mDEyrA2RHAy(GL?*^?qOX-^YNq?!o+`LrHPt5cewdVd8I=zXqP1{xpmc~wCeN! zhI69Gwx`fem^W`8xI02Sd;R+LoUQ+IZHbcl7{tw+%z?Nh+r_WG+;*&fT+UWSl=~uh z;+?O{U4mIyK(Kt_mF<=?#f?z>+VyQ8p_ zuyZRbx`0m7+v`HLAE&Fmy-i;s3$HIybHI!eIfh^3mOf5#oym?MgmKOgo{$XmrIMfiHrQ^`C3mSZx&F=N0n)l{|aJ3n}#vZMXjF_-K3 zaf(+}#;3}vki(Ei?R|#}OpA0KmCn+Op2()&V>YBOe`e959N0TIOUI9Y<{2uGv&c;? znoT`}uP^vBws5MP)mUs0ydlh)a67XV>ojI*D`fBU_!9TekgS1kszD;PDwMFIR!2n< zhf=f>^NYA1e*RKOvVRzTdmMdx7JYjJeR~9bn`J?Q4Ma0+7om03=&)Mq%yuL0h;Uhw z1#Y$&4MsC`xmHX9*pN+_Ja9pr&}}Bw(BVnI+63igC{O4oTN(`891fz@X)d#c^1n?7 zIJ#-I)op`JidGvbz`LB_kau{0Q}(_0_`KrGC3mm6Z^_J=Icx5_|MrzDLyj+c)eG6w$HJ#4-1YLz8LzxTy`sxP zDL&MI$AYO6xSb%oXONgN!M1Hs);dwvS0YDgQfc4~YQfvHh$suApb6>-hQwgeo6MbN zJ(7|@F=8&XSd2*XGPNL=$!;;*iT(hQ(FU}d&9S&Dq$C_r%OZ$)3lWW^e37taG~2s{ z;=Vl#>!Vz$R4JumC^0fw6mFgr6^W!2P$DJ@f|*i|!B$A4fMNnHM{odKq$*;b+bqEi zAc>CLKqm^_0F(>g0PuO_1|g1^3XULI7{o6F2y*;!sCckByfFqq4?^Dq-?i#tplk(d zP%}jHJOmJsa)caGA`%HtdAMkPJedB>nF9xo5vJ!+Wi_SekDRKh`}*_0Y^Ap1l=Sq* zMqPLY@X}-Z`eO6w#>RYl(V`{b|6J;I)Yp9vaSF^LG6m+YfG_YZ{N$0z0bGM}q-*~O zeeg$Isl0sBq+b4O%-1Y3FeAF;#dV z0FA^Zyn-b&S^?Pu>$j536{v>*T2jZvs5OvdfAy5GG%^MBngDvu2fe0%UQz2i8LsB3lf>_ z4niM{pALVUVW1)Oc?TQ;|@Y4Jj z^Z|&{OJu7rY~Rl;7hN*{>0k57tekE506t%5Td^w zO9WkV3|OrVRh3l^N(~Ks+?X-&YL5FQ@nPSxf}aRanGl-;pE036M1SOif^_e4{2V4v zj%AgDg_RhlU|z#-{z{xA`QMBmvQlalP9{f4D}#DT{(}svK}%C(lVvatmX0{8v#67jkDo7>sSceg zYqa$F`YerQr&a34&p!QZPotVjabW8*8X)=U$spLBD9+MsFgZFqyE+^;6Ewskn0CQB z*ae~q6F~+*W-RfEg>V6AnoSXlB4tt0KzY+3*G3_eBqVA`0S@^8q)}>3LK0$S|IH;r zeE%e<{S^5ADe(Q1;QOb*_tOvz0H@nvPhU5ge!!`5b_{e>6+UY3f~rwYdT2P*L{z7THJ1Z~@}IVS^_eO$McRF_Vi4#x@na`wPJHt5 z-}X_3x+xHMdpoaOXwWy?U8+fI)~vaE$@F9)F7NJa3nmQ>Wo1PSZQFKYcPB&QrewdFHBhH-cF2xz$Bhv08z*lBPTLR{n_4lyz^ zQW_JNkd~Pel{IIIR-s7B)Mj7;1cwzuAzY6#9#=uUk1Bu=OmVroLbZ^(yLvj>T^Q%K zJ`WQY3HJ+6>aOlZ9aD!3n2GInxE$JDu5KXRjaZnLk{Jq~Sqh$63Z7XCo=IsjIq|JF zOKVegMMXt*Lo*<*fQ4FTaJIo}XSE`uI=Q!Z07i%a9Ggf^^b7(8)76DyZ%~zld{1Y4 z51Y~vRARwU1YM060giX6so4=dFK6la2!HRn&p!U!u5zkLmpnM=4FH=3W8K?39PkZp z-MYJRShE-yDT`u;8+UJIQxtR$uw=R^dCDXJ2~SQ<^j8&}I96!krQg2svB!Ti9|>W} zh!g+>A`l$$2Lun%z4y-1h=Yv<1qBTOv1ZP_tZ1;)j&0rU8t~G(Xqxs9(HwxjQNez@ z16b287iH5Wi^Y5*d}E-vu#xk{5=o?p>qUhdfQfTOG6|caQcDZWBTTnwrer2tyg6dm89Yd>#ao zEe&)}1Kky&?H{5V?7&bU5L;NLV-e|19Or&N+~z>O)@t|eb@38n6deth_Ut=Q3oQel zsC3HJfZo=1aMM{uvwqn!)DShBd9l-OUA8cpdK6zV8iyp>?3`?^)MEtDgAoEaOc&S` zOfV4IApEhwL&V@=Jo#90S)}$yMLTk==^n=KI6VQ+G=*M{U zLv1u7Vhq+2fI2{cL5Cp97^$~)6ePT`I;3MbxEYUT)d%)!wFeF!ICyB^-XHeu)n<;B z^fw*GvOtYzF`x`)gFvQZ#@+JZx(Da6>P5F13_00LmSitkzGTrdeQkA51o9>xBZ0{R z52_piy|_!qG+*AkXYYAdOf2C&!e@FW~IVgSnd@l914oLZjTe{TulQ2crOQ`K$&Eb7%MQ`^ZDkq&>(w#94~x-gQsd$h z6lE==62ECC<5``Vq<^nEgzY|X!(d6+L2exhC(At5nkdiIPNQ(@L)dMu3>j~>jNpmf)M`oY%swjaUbG0ILd0htLwD?MGeRb*aveM9TYMZc9>sC>?8)<`wUt*Y@iE@P<7NDv zIaYY)av7eLL|+P>M;@+_PsxEi8R+)?tyIEW65>08P=~S)G>^;j6scM|lBMX)(jX1h zN+6(o?jhjFe6#?oE1wD~@c{Xc957nL2h`VW2;{W1|`4Mv;^bdPbG9I!mnj@wI9*!GvVuLh;AVbn5 zPfw@A(gX3+|50cSIE1-Wfw^SDTsn-obcoC);L8Vx0A1%r7!$Tsf4i}%>_TDDnZnBz z)wRuam6c^Bh^4BuxcUMDkwuSXt_IF#M|)@QfS2-m2Q+OcF$g#0P)~>5Vl+0^*AY3Q z-VC{63S=GBDDc4g6b=W2dW8Tf9s?X^gr(t0NG{?b^3d<|Q7{1c5cn{@MnX~}-q#czjbZg*4R))jyWx)*>4 zd8nTK0{HWm;OgL${4JmFE$g7((g}+%{`kIb%ljXEke?lVS^6F1$lA`JIC;{XIdkqN zez5PTe{lZeDP-Rx(RlhZOY!zNnXlJSUhHctJ%0SSa7u=R6YRA~mM>pE4esR`_@B{L zW99e+<8O(q#y75Zu@oQ--=@9>vi5%JOKJ=6Pp__DvSvBn`T}T{dJI?%RSE=dUCh1N zH$_DVqQ=iz@bGhwuNoUY+*o616HkAbm_pu^zp#4V#H3_(u*>I9%)DjAL$4#CfH!vh zqKz-zM(I#q<|9NFoS-fM7N-ES>Jz-)!o2jm`(}@!P;^wsU$f@c`SVwzq+#xYKio5K z;eBiFn-@*@)|^5if)nR29D6_aokQm@oWQH%cAwP2yRMOF3H%a@LvxLRD=WNja!7$z`0(AU#t zclGqrGO^d)WooP{K3!mG8*ujxz@VsaY5;UcQBg^yrn$B5;`y?x)s6Lf?C~94cC%}s z*JTBw&jJ7lN)TMh$Lt#5@(_Ior3=d=SQ^5n$Lsft_yn69q34AC1blqQ3s5Ai;vfZutXJ~HF3)1X;a3frzC3; zq>MSY-`vkAg^YqOJ@U=Ae|+)D50@%Bc<@_bud9QI?sLxC*l`v zQzveGc@rxiGd?ONNXI12{>`S@8=qcz=aZxa4d9chcO`$gWBSA>Ux4Dul4n1-1O~!y zu&;bg0TVRhm3t-!^|j`@D&)d#(TU^Zr!7(z{`Iea{phbwmQjz*%#;N3=;0xqNER3W z;YY`fR%gWOg}5+&_{%zqRHD+4*r}~8unL5{gl1%6!88C<3~C!0TIy;W8(JZgS^%_* zln{tmO^wYhO--!`1e1`pAHg{Y1@;DMHe>~1HQ<5nVUlD73#5=NlzfzUp*cY>I~8I& z1SMjF)TrW+_>+*JiGf)X@4;Xrk5)sPih=l-6sL-kMFODqCp4f=Abm#Q74Ulrwx1(ND9~@Egj+#MoGXrpCp`YSl1yM-$}K>%T&W zGsRWtbu4;K5U62y!XL7W%obOBi{54>F!v*mF6-mBGbaOA};-TxQm+ zmtUSWE1&zs1N(R70589pGhsfK)o<|8{{29--=8nEn3|hOS`ajzTn3sZ=6Jp^Dq0kw z)+iOQ`&@f7G=mqSEep|>6==&sv?T{^Nv;QUOnFOrGfYs8jrFzFS1z4zDy^xiEGxsl zTt~I)5<5F>&CS4ygANS9XQ(FTb|kRr^(_tdwvJ9V+2ZifZ5>ytTN9ISnK_o%Tz+83 z$y1K5jw46@zU`YrKt?2jA%|HV^pEZPo?8(MB~c)F`H!!?_S#cV|8Tj(%f@J>`U*>X zv)xcvbNGZIap~G;U)!*5)4G=yQy&7n^a!@FH#xUb59<^hlzd} z`4%}2nO?%n>;U*7-vSXpI5-PH<17p@?M zAz1@V<>X{3b@x1Vp2UynDID(pf@*K{h<>gGrKf|^Q$gub?M`NDYj1(ekJ!pA1|v~Y zIV|G5bv!Q6*oAC%@G1ItWs8FggS!JB02`JU?X3AyyV86~!pkaD*u%Wyr8f`o-tr*}NJ(d;&du5Z zytEp;G}dmzTnQZ(YzrnBkjY`O!fJ%@p?3HwjCy@@ixHW=`Ft$*5PgGr8nT)|UPUAr zDh5#y<)FCpIF#4T#=4Z56l!kZ7Zo=m1pEk*P$U#OnR!@=-vR9G$5bBwjx0&-cUwTd zSGj*V2lZSmGW2SQBM6rd=8>Hl5-*h#d?ok_^WFhVF#XYmp)ABVnHN?iahkUQNr8=x zfaNCwwPZrYJW$Ys+(;x!O;CVN#v3if z9{v_%Sd!q#gin+B zH4!_D1OLP#kW(auDjX9Nt5H!}G#P#E^Rt65fO&JVFy~@v zf8k925&0V5SMDE|tiMql@Fm|vHJ zUx`;|Ir_33{JI?cn%dZ8?{qmD%Bp_cfBewCoqG=MKeF#w>6xQP4jn$<1joDy!K7Mv zm7hd=zXo|p@s#Zlo7xLwImMIk z2V8xf?H%ja+HFpod458-%PGiO^}@@qK8zq}t;f^b&m^T1XI5lnObiUOJuc7ksX{s5=clAm!juUK38`bpre$VKPKnYaC#NJQsss!#61%t-uJ=BlOav** z+}Id^85ZmVKb6~ujl_o_f&iP^Z#BExY^~P*VQX7gTa(_^;q2$O7~0z$_SVY!#-?%% zgtoie2R*a}Xu_!NMG6-T4*I*>+6MbwFd*C9fISM2z$%QuDvZD?jKC_4z$%PDMq#V5 z-ePyP+ngqo{&>NGZ@>L^=dMFn&m4eGaICnpv8AP@mimV-&0@JyQv(~UqrKgVkvMwf zTw!SgMxnm6ycEN5siwhbXEXNw^v#z$cORapNL7 z>JEMN)mP`>pi7fVqmq)8N@Cb?U_9(?!$W>P#^Qp5lZ?TbEFEq+#Ad92@`=YcZ5;6Y z?y=uK`PgsnSh;%j>IXJ%%82fTF?Kj^(w!SNY*<9C)}?t!`KkVXrBW)5#i(dgqcsX1 zMr3vzuMun+kw$>y1MKb9=Cg-e{_6))s|mej~I})`l4fp z4t)RJ?w?Mats)lR`jWzfyLTNaIDH1V6UjDPb9qI{xpU`E|Mc~R!c#ad*Vh{n4rDX5 za^3AVHpNPVw-g+g?egJ65s4RFnY4S}+>jk#kng zn6Y^A;u*0#E6mbYt{y4smPfHEW0hF7{KE8Ux88lb-RV}Q&wTL37hk;N=9s|X@Y=gp z&ZF+qB_m@*ERtaqWbT&Y`Z}0fng${?x7>2e)XCYCRK5W=h3V*}L_!|p=g2fM(fpu5 zrjFzd_3?a^Ix9_%)i6bu?DIt@#lRx$;kg@&;bEa^MC1yw$@effr_lXAHpScyUhE$v zR&WD6Nj8hkh&URPwR51;-0HLu+dI4yJzYZs*!bEy+|mIT3@NBKG=u^2_BA;BhX+UI z#Y)VJm6#W|VP34nyjY2Ok>1dN!R=@_R3G^9B#hhp5A8d)|9I)yqsVwHxKOFLSVA&l zD)L~>b+yoDi@*S6dR)Cq)~06QYGPJguBmIbv*|pJ5Wy~rNC{VotcFi_BqpcD@vfK| z2th-Q=XbpS{`)^-eM*%|l~UTStqH$SclCh62kyLlv3G`8eE;25YtrLHZmaw z1j)2W?`k3a%0+n#$CrayafGpfqDJ40=5t zu}mp39bnxd`EdTAAJZXypv?w14;&|MYgdQMWFjQ@>07M`2)DHAA=oxnQ&)AV-Q7dp zP(_8#kOLCjKQM&JKx#o@GN1s%@CZg5sn~W8#)QN*j#g|NbFtil2)mBv^QRC0_~Va7 z#(^PM)oJ7do;g=$2GvT6FCW=|=vWD+M5f(tYHD;g6&4m=xL9=R^tp3auU6I8lwCM; z=1g<9$%QH8>g4v@*)h%`5kkOad{0*`CQ!Mp4}s@#Q>RW%iWoeHl20Z2PG#2NZwrz4 zlo=hZ5b}fp+-W#Mrh>x=Au-*;#L3y&*^4HMojAR+v7slL9g~_QCazauTlh9!(H|s6w2jr`T*sR+De!gZbTg7W|W_aWFq1KSEh@^OTS-l z!bo*GdWM2J)$nki3z1G3EptO3ar=p4;}l+YOn2>NeJ>AS`+X?7)!~G*ySl}NyYvjU zw{T~;v?=YwwD@Es^r*#dF%qX&w zkr_qwoF$l1A+Kb7OjcHOOoWsN>!P#Zn;knpIehf!*(<1BV~9R{h~}cAb237&+=#a3 z^ZVK)j-yA9p1yecSk3uMMkg#zWi`EhTyOpN+uvuu*s*)x?tEPi z%*}sfK7HpO{%#6>8eMWyQgTu;D!N{(tequ+uxGRYP-e!A%;+j_?2eoBeywl%6^qTHh-4*q9q3drj)<>hlT|Ls&n0 z-Jbe}O32NwPD4u>7Ce|k3JSb-lwdQ~RuO}88AA5k*eQm7X>6=&|Ir5HAo6N&(o%dL znPN(rlrPt))SrI-WpTMJ0{ON>P3L#ODsu?1YLk+aV?@I}-L~rLYG*`xdJJS|m69JE zK>xbz2*#kK$)`_O_A)8A-g@iwXxe(3jY>-r5BGZRUzMF2-QV5SSZ^dtWBgEeZ~H(` zckd&=SvP;Cmcr@sn-`LG@jmK4-6SWXCwTG*G1SgzkHu<2)H?(vpBM{elvENE!=s$j zre#G2?8H9a?pID@r$`xy<;ZO55GSGfQT*5xK5Acr>jlEp2?^Sav14vZj*d-)HX5&1 zNnwhV5(5opAT3rX{n)YKK1<=@K6Yz&KgqNKHXl;+MP4+ef0!6EQ7*a7$xi92Dz^{! zwDq-jV=uG&7|~F>1G8!ng>y{^l5H~RhhT8R4uc>kJp`b7h<=ki9!5$#dxF^0201`{ z$IMZx#e6r$wZ|W5YvT@hL?W)wO(s1h5GqAsy@EvfO#y$9`q86Nele|fL_Sz9mt8n} zXy@nKc6|TUSBK7CA#PvQQL?cW6cl36i;IgKj-qoAG0w6vUw@UP{(Q&wFZO-EbLY3; zexEt^Wfdehy%S-0`2r=IY* z2O`tvK1^-WsUX2AQMCj3>M9h#?(c`lQ&iN##&D6-=MO|n8I+8HOBZ}0iGb%NN~a3t z-e3Vl*9Q9W;2DP1%g4qTP1x{lbui?6>g!DcQ~Lcx&B2aKhM4g*W)-fs zR)j0VlmR;=eDel)_hqO?=dP=3Xlz1=T3vm0-AKP?pyBF7(}0hQ5~?2jpzr@Z-PTuuw$>#S*z^ zXnZ1qJjTsh_TU3+=Ti^qG%+!f9vfI4yKVv?E#!$42{lrqd~BQ_R+2tbooCqT_IY3} z=OHE+)=g{{L7fIv3GHTJ{s9-F9Z&<80j#+qnvLr~?59U8wA(sNJ$>k&)nT?e5gTfy zDs-B*Hld6UP9Whmin_YS7DBNaIQ~Z{mJW)I1H}l&QaUJ>4vNX`wuVE8YMNaT9SWOj zPf>exkxdm91qBtAWO6tvuQ*uVfo<xT|0lpH@rib`X=p(F>IS~Jc3mtx+NXz#rsb z9pRWduA-_F71k<=q(%_m?*g^%0=4b}wK826JM4e8C1=ZuOU~~5^0W8edh^d)-+TA( zpMQVo(DBp9kDtBVNFbvK_+19(Q%lPguvl4VXWa$-zlbH~(glc97cMv)rNwxHMTgBw z((Kyt-8Xx7e}DAAiGzD~?fq{1_ODY?Go>9hd%xca0hZrh`Q6uFA^I=_b1#A+ep7FZ z+)rxz`us{|{M0+{xMNm|Qkh7eOixy_Sp&ZNHa@uFkw+f7>!H=_pL+VYPe1?sQ+}UY zJtq5+jhj$vTbViM;SHORlbV5fI^c=K9F!OxJcLz8M~M3S9hFdys%`!K)+#)yarLuV z{=hKYCIJqw$31{0JVGjn5W?HZ0d@o)VLAeSAuMeXx(v+e?yg?!D5J6WRfG>_1aF zx&xq#C$S_K6<;iLI0`R@cYvKcl4HN!x#Oz?`*wf#{r9`F#@;OHYuN?Msw!QHQQ!vr zXorU{QYw?CV+WWU+5vnIKJxTqYaU#;iR=I#H{2yffO;@h^rHm=pUs{g;G-RTXBh>Y5+v@_}th5V9hp|Ks8%!OXq;Q zjqCs>g9G9I2nHfMKwF!|jM%`=o*_`B*LmFz@T&+Zr7c8%Y`l1Q=gysbPF55hKpN4Z z!b*~t*#h%!yv^29Q+E|ST3S_$Y;9LVN$Ew9>S{}q!Nn$8DWdlUU1g~HP*`aiNJg5I zw99x10U(7~YLPY=5f~(6=H`gx@)IY{8{8@=-gC8#{REqsMe-atk*Ra>)}mQsRh?B; zRo>(&YhQTbh1;m5y75ESm(4o=JhHj#sfGr!xzizN0( zp$`ECTyCTgbq6@G+zE8?!Jywe1mGYK!JdLt$${||pAB^(2boQ5>xX#R?`|iGmDAdb z%2Zyj(_**49@K@@m3Y+K9q zs5RW>X!pXthV*JB)PrBzoBF*^{`QH-h@}aY)7Cxq_#@9g{@BKinW{dd$H3mS8oJwJ z*f`Q$F(LUBrOLGmwH#dN=^h+({nFk9+y1bxGZ+gvavUh0LHh(eO00-WN=lN8_+&UD zs0>)I0{@4-_W+NpxcbKL-QIV#Dy!a$dzEc$TqGNJY+ArTsHQ{Ug%BW+yrA_J5E38> zAr#YWW3Vx9;4WLPvL&nc-lbJrX{GIb@AsSC71&@qU zW={DXUt?nnDiH!UN(6r!4k9-2$DqCmCv9L7rVS6ZLqu2CL+9Jo)l^wgUswvwt%{6@{4AjNur(DBy~p=N z?~#y@k`jv{2?=UH%m{fl`1Kw{7dTr*^d6$yu?Nj&)zJ3KZr1RMM%dXGAy_W&a{TgmbWYCj;D zp`N>f-h=pKPJuKf(Rq_0O^H8d*f8+-n(K}nJhk_`14oYTE9!>mJfu})iIkKS;J}u5 z^1@j#nDZwFn6QMXt=qSKi81_Q^X~6z8wukLQ?+;B-aQ8kY8&B<)KGlp z;K8r9Z`<<8w|kEj0;;R4;r!u)+rQp+zDh5S5a4aHT1UadLC_@lUsH z+ph53_$!X~s@e0a>l_!F~ zZCFg&_19lNC5&h70!UTokVC9uX3t%k>?RhUdmgxHg%sXnLY0)YTa2QVjN6`h_@M_M zK*;{I+a7&z<3qP)ZFu~-=U><`hk9B&({8tNW0MnN(x%LuIaAIB*c&uxK7|H+pW}3U zT$T~Arx9AH0(59ZteUplnb~qaUj)D^KI`OB8K)`iP^!2#hczeST?y$O6oNpIC^{8fLbx zsA_m{Xs}-o^BErq4x_^eQ5YWWYBhAU4-5_s8T#x!4M>x0;JiNEqR%#~@F zE1_zcI@HG#C__y?Rv?ZD=fDMz2`woxm>`^n208~!W=nT_{?ViP4CPu1@VY;;FRT8v z<>R;hwCUs14MVK^-A}xu*o^)`8;T&ivP0PEoREh`MJl`lhU{zu6V7}7`IjENb^Tp; zK6qnloPy@vy^FhyUK*%EDtic5eA}+ZWr8A32M|dsugO zOHY1&H&8U-T}7;CBs`9Zh))8-&Zi%LaHL8PmGAalF)@kBG1$0;At#&LU}!CdhPAYd zSjoWaRE(6%g{||cr=PuZ*+iaJ^z4%ozJrxv+ZJZC2?QEV(rmEm={U++07NIYF;GIl zrwG0e%7$!3k<-&ctX_7z8~)o>r8u#`EZf%&47cw3rrOq# z{swH>!Ymg3FgeO#qBy}q9Egy$jz>QvgA&98IvJEm1|^a~31!K!uKZ$Y*@aX2N8Wqq z1eL3mlM}m%7a2)I@0dTpD3diGojFxh0i5DS&3pfVe}o+DarcRKr?c(&9(0*eTIlE2 zuU)!eN!F4q@oTSAuW039ifgkcN+cOsH=>h_QsSnG#X2M|A~GqB4mIh5zl;>9gl6O3oZP5WX93)OyQUS4-xe15?Oe z(5)Y~dVOCaWRe4N5S-oTd7o5Q=e?iJzQjI78w^?OQ@or5)z#l+V+lRQC37PiS!X}s zzx?YB8-D%!Z1^5-rqoROfaS3p7GFPWcILcAi>ceDXG}~9naBfgZc3qhn;WPv`T&BGl1Q(2YBnD+&h(WFZb! zMY@=DkHvruFW=7N*sMZcWPD6~Xqa3Pr%*tBE947kA7805;f$7w@#^je_X`yd_29N~ z6oFDY+&LWSDKwHUb^21$f-;Pp9kMZsSs2B|7)3w2qPetgbhrVo$qgk1d7o}AE7-Z~ zLK&0>Cok5z$=PAx8+U$wV;fH{w|6x{m-;QVY~TNN)22Uvhgr(yIwf8>>PH^jckoDe zYfXvb?_1vaI09DjoNn6X@lXq0*uHs-sq@zMz)%X6@*fv z7jCeO(FoLZ`laivyc%G)Xn3)_s4 z61Az=!iHbBVnuei(8a{H7u9z+G}gf!)_3mYk)sX8C(f0Yw{;#kUsKZqgK~`H;4Dzm zmpp6M47Yt~XvRK=qoH=od+!~pX}pA{1hI1Yr8E^cBq~R$&YY_l?k+2;sx3No;_%yV z9XNQV%O_C5V^F;|qP%3y*HpD*>(j-=iJGfx8#+c@iJA{i=3lJr z9cpU~c~_hJ&yzZn02$F^A}XdK? z{rlT)UNG3ruD;XZNW7`=6$3%|@mLNCbk)u65MiksFNPd{bNd`DZmXAAd~2DSf~RGje|Zet41&FhZr^rkj$( zzCD}`N1D61I}gKSNvdvMxpJk!Fb&ojH6s+O&YY#>H?7R3_ELMfx37Y(U1~`vDK6G% zdb5`t4DI{cUEHX+gLDEUFR_YqWS>5N8xnLJz~s@HC^t8s{^-4Tc3^Twsnr6H zWe7HWqlc$Z@893*iCwW`MVzN)KOS|0D{$^$*&^mjFS^64@2FQ6&^sWo#J-Q}li+|}tyAhel;YtU92 zp}H1*0Z(p2f`x>uQ@NUIgwP&h_l2O|fQ~l#k~ROsKB@L%aX|~zjhgVb zvhs530QO1SKRG$*^}<*M9~118l#yzMB1|KccpR2Lzejd3>J_VQ`qdqGU}hpP5Bns( zmxn-1pNy5oB){?k^$_++XU}~9^;chEho=q=#XgA#ZxB9oo2_ju90ws)<*u|zN>hJ7 zPQwtb*e5mAgNKzj)H*wAE*6&nPD1W58+icz0D~BY>N41eEc9UQd!?rQ+MIa$^647+Bj9!-+oK|MH@RX+zKEt0x#27bNX~GctL+lRY~m# zAIFgaCNfT8Ri>n;PY7}MoIaU3T|8Xy)y{8r9jel2vk$Qkar251SQ@L&tJDj_qB52& zS(2z$+fkKMDho@PlKK6?gK8$y%864t;oQoqQdg|*(or!>@B8(y?_NJyKtLZibMLu{ zTBeP)ESnU|AJa8LQ3#`>5KKW}HTAVN_pDyI(!#`BZOBh4k;;AMQEbzD3??zC7ojlL zSgr807}dtQyvC8?5j_b*=8MH#OusRs!#UPofrxJx6Yp{iG;s_i1DLN5V7}fD3O@h}Pirh| z9yJbhl~$KuyjWa`K!VmLL{e4P)Dl}mQ)ShKi$z5cMJ4&CPM)vAq@JSFIS^s%gtnxs z5Bj=(99wj~#7^B^U*FtIoJ2-PM}`bzJwTn-G1EE>!tk(v{&i1@J}NfJ*&?SahQg!c zsGf^Ew?{{Eduon+_Qe-FcJ2D)!;e1RQwkS>DPX6bv9UTFPipl7z~sPoW0EFkXJ;n^ z{cjY5V}qvM?QvS|v|P&a*kY86I=M<}8F(?)qb2EdRR zp;$uXA;dD4s@r7;dv|YALLX~4kD1Lj8z8+A2*OP3sc1DDo7;x_Ac=dMicI#h7NA@Y zm|X@e4FgWyb&d56Jw3J6mDP=%Si@5gjX#QnprG@JKWNmDFZ!f==fQ ziHPuGsE0-g_|*Wh8V!qQst)t$Jw`XT9z5UULH?Exl`*DKBB8Z5O^qsHo@y1g@?l!?2I0Y#Kz0~*qF3f zE)x?^8!wa8)zWV#Iw^mp|0P<~;zw0lrM8 z98ju2VMZ7&NiS}bz5jj|`zrOWmX8B5wjx^P%qpGk&O58>>Sn6L^@Kr(c=*H1XY-%` z?aR+8Ax8fA5>D(&J-FCcUnR!H=`A{V9$A;Lglm7U7(<2rNA#z;KF}>Z6z9gij#>Sy9a?%%nPux&# zK;$@FTy}KqL|#ict+zipf_R5ywOVK$ z9&W%+yTcxunVC6nagt7V!wmv~n^Zq$_DR)D^o`d=NtFNn>eF{Fjg1PQvu53fdzJDR z9$SC^gZKVnIa2t3fB>mY_-HvV+y~LQNtPF2jfhXnvt-f^!$)oeez^)uZ_5h zs^Af4fr}s?b_fAqA`QnGJ|RS57~xCi&h^^tL(&N|rT|GSfxr|dLmlWG9VI3r*z>@{ z1{>jU;&z`rSxgK>b{9YxnAD@EDv@*HOy1GMd3ne8e}AT-wYvUT9-z59_WtF~%~<~5 z9Uzpm*-WEAoP;KD%n0GC8Sy zAK{pK1GG#C!SGU87JL|13;>SnCAxgbDue_e+G~)RURFOU3|A`+rA0?Jzxg^>YHups zv-$mh{Oz5;N^KjNAiO8b8hRQrfvt&+!oXZ!)&RulVnT!X-z2)V#zbNK{R zw05f3Yc-gWj|Fnk?qOL#8R>dzh+`J~7{M_;ULs)#+67pK)yYh+tFCWBphZ(%U3Ga$ zDd40^PoKn&p?2?{edwWwCj+72=S-Xj*^zhg#FR!R? zYpz3Fcu940JMmZT!k#t-TT!dQV8w16<1>t&^ux=lAI3+lT+Bjp8D!GBMud^!YTwA* zq7$P-b;rNk@#WXw9IdSGq9USVUA?u3w|?>IaRiK{sMX;TYw!8*b+CpGBeJf&6@j(t zW5+Ib8-?&Xi>6Gi#~8`{Obx|@UzFv>TS-LYTNqYt#M4i|@ceTRu3otyS*?;L&0T)) z)6YJCFZG-@1-5XDC^S7ot&WY2O@NsyHA>{LWM|KgQxdC`o>r!2Gbvt<$850+B_oXm z`;OGRMSQ6gUXdDpl*|}xc@e8L;pSe(8Qe1T5LJ`@mSCp3*BdR3D>E!XO=FU=b>a^%V zzurER3#jgFw1pAJrDss=#ED4=IiX#&7HSe?n-ik} z7{|e0q(c-s;5ZBS`|)$ci{Kh0iuPs9l@~BqUIf>eUsT?X@kRi=aR3Xxxvc2?vAuhC z@7}s?=dPVQzW)AP(aF90_MIxK?(XiYxp)q|3W3j9y;am-wexhk?w)}$)96TFZ(IKm z;6wEz!~MV(M9>>xw7W?lF6<)c89@6GdNOSAKQTDU%VKM3-^EZ z^^R|M?b`Fr_U+pbR@4|%GbV9+3;zD@Uk*2z5|Y!XzLL*2f0z#koq4cL^0DANUKi_j zn24%E4}VwPI%yUnuQ5Wl8<#G zqlCk0vl(4Z2Rz#X$l$PoAl4KpN(dGs2wRww;2@|B>LIXvSW~p~U{FDHl%0rZ9?@+; zGXmrxhl{WYn#yr$x0y35jtPCuN)DIXRi_;2+P^aBw zu@56@kAtuxixFjw2W5!jtX_u%uws)-E)5EKyO$wktyHMF%6p9>i| zM;MLp&cdp8`d#Mn2OL1s${Nbg6r5`{6dn7!)x&f4HFb4$G*#6#z|PiETU%FKg@9>9 zXNTMEdcE7LO7%y=@g?v!6$4y{5kb0^v``C~yWD=wm`+V^loO`yYPx*{7d=_NgQy7Ne!4T|GdT?*m_S_*5z>hY2y6 zJ$&Iq4?gn9V;k?imvR~rnWL14CIWyUP9_eCk3>+ZhzSu1M2gT5sf+}SsxjlC)l<5l z1#mjYkPL#w!GyzOLID2XRNn|!wzkfW#)i&e#FiH#o{;R2w za3l{-DkPlWe--+dt}Zz1>pI(zig5@=PX#ODd%OfamIs7;z%r^5Q<4(n;x%7x-MVf2 zj&F7&I9{PqsKpMzg0>7&QcbLeQCY|O2lS)vjrZSs_njN=x`$e+RXF%QIwUPFE-gjP zcNrb<6|n*30yq+W#7zWriM0yTfZl8c#I_N>VxvYAqcRU$tQMS;NA!r^8qfnwPT?3e z<7nh`nh;rH`@vIjT#I_yzM-MvVqsHrQ4!IiLZ%pvSpAGRQL|;wzE3yr#Apa3J3BKo z3ko6`vB`1AnmZo<^-Z9@(3g_Z+PZwXkAYT4EENhQQldizo}LX3vtx`{uLr^yQi*}O zLRcVRbwpdQv4837udC7ZH8xN?(GIs8@ZVOnBSIvMik5%yId+%A$by2|vpYM{4z7}G z^obuMA*4dz^5w0qDJecC2WYT-juzrng?8w1ZZ(>=Ft#f#gSY`}fJ&XLDZ z2%gt+$y*s@Kg1;AX}O3|A7*%8@91d9i(b`o`AQ}K66%v2u>|Xk=+gbYw%ef1M1T_% zq(UN6a4fMhbgV2J=sz#fX%SiMa~vQoYlX?@&L!XFa@~b!|Ilmzg?vbR0LS$i^|wIp zxf?cct1~jHF-AP(W#GD90KO6;CJl>f*uaQ-I<{=<>Dju??;;|UbGh;mp)e%aGCo!t z$qU9u%lK#+AN|)NbS5{L4DMItE`wZ(wE@0!$Xsh<|lz9U~8JAeUc2tv{3J&XCI^x$}`r*)>amGQknK zpWLzJQgzJ|yU8P;k;`w%#X>H>pjG}ce2_fsf5-TU2(Cmc;xdm1^U38&auJaWI&v-g z`l)r6JoqTNsLAF0HGA?;@`#@RE6AOYT$-<0;v@2ifByPsV;cGOe|t6_B9(3m6}O;;K7-;rmE$mM2o(UA)V{_-&wT*iU?BSNWSOzzr$GA5U|>>Kh3VRytzj@vkL+49r3{NI$hyk-7jN9%)Q zA42Z-{iJ1=mpDKk!BA*n$;8FKw2of0#9zpBedMx?Tt>-d;F=};ts+bk#wJ{(*EDSFc&d|DHe@|9iOpGJgn`kz!qmu~Q*Ay`&V4 zZ>X_ioBrEdZo%z`+BX9;dpw?U z9_sobo{|`ISusxlBR&AiMNnExc&uJO>_Tc$j>jX`hQg%iLau1oG7!Iy5CfXxIhg21 zGZ2eMtG<8#pC5c(P}0p8%dBn1Uv2sHDC9#V^x#9CQ2TTbjXK%Ad#m+SYHF&WzjQAX zoh*Ypq&g$(?mO>Xn>jVg>#|E?XRf>Xwk%A0ZG=z}g&31CIec)JE=`iVppEk=lb15l zatXqAxKedea!Q<9EJIiZ0+4_KD8f=i%&9iaPShwqA$-UAfIWgD&PH+gO!Sxm2MRq< zZ~(!Lcyk=%%s_2`z7%Uq!LmLM@+{B-ocDm34wn?D!R#!@9XKCkf%aLT{SweC3$*w9 zZCsUJLLNYFgEZ4ziDt;T>}2AbEO37}S08!*gO9$sP}<3t$ZhRKUv2&T@agu(JS^W> zLOUI3*D`4Gu)BAc>1ntMaQdru+bk_@O#D^pH#0VIE}>!cRcI&{Ya*C2#U;vkSCZECz&TnyES z27nH^N?ImWjaaO;4kf%Tpa91Mi&wxV4t&;RqgJebBHI zg&=fufz~fU>z4)RIG6)D9=5p8yayUWd)Eec}dwrFbZKylM^i*=>-s$ z5jtw*Ozpz#WcBSeZTyMAcFc{8Q97hC35h9T979#v^vU6#_QS`IpDgb65O3+%xaX^p zC^IMe=(%RZ3kks>*s(Dvt!jHLV`eIP3euYA#`hX{?qTTlj4okfa&3LLH+t^8>lV(5 z@jdcL=0xlia-tVch>@}`GoG>0w2~}xp#UKG9a&43I)}tU_08hlRcjB%Q%~4WBq6|43;sHEHUMC`U}l2rq5o)G&6W zq;TQ{4QuK+U2wjlvhw`#6DON-QkV?otkY<$gkkhzivhY~L`!y9#4vIKqEy#RhQn$b z8`SrA_8D1bs(?4z(KCNRYB<%=+F%e2TNMuIh=(a@#-!G^L9Zxs_Uz2bDJj$E&6^XA z1G{ztPZb|O9R{mu2_k50z;Qw-3)Hp-w_6ZF1`GKqFcrhpqTHz?uFu8lDHgP zq1!m@<3gJ$;&!wdMu@u9O(T~!vJ6SULqa^Bh6XfedIuaF3h2+2 zgGhCAYeT~boUDXG4l>Rvf$XOh%ft%gJ85ts$$#GDp=VWR{DJc_Y&YVNdp7TaXv^=@8Tl&p6pX}uFoVLwdu@6-wRo2uj zSWsS5lSGO7YHD&r`i#2tNZSp|Zdu{QEV+3tpi~w8Y4P!;r5R~y{gg8tyf?&Vlcyv{ zQFK}w;>_YABES>?qO28RO9OaOz#+Z?L1-Kt0rE(D$CznE4;g^02YjJcq#~J^_>@Qx zz-0ZY^dOlFluqZ$yG3Hs6NNyj81IR{ADMHA^oW+PHXJzc$xa^MZr{2c{g9D<_rfg_qX1K2u%nm z6<-~pA*~-Auvmkw7x>#Q4z_)HaORKW7%oTKlP#lwVr**GjoF~%X5HH&u%6V&$jC4RoiNE29UiG@+ZW$o=t3@n z{)UR8GeZ8>Jr@oh$lv|(ho5}%*=k$4dF!ukT~000CepMtIw@%)LKIVw z;AMEYyWSih-_bET%7KT0CI%^~lEr)6#kOI-3bLLDbJoiW8#?BQ8DrOj^IZ({X6+(~QWp;Koj=&^|4XF_5qaFx2*q&)>OdAh;k>q$DJTxImFbFD8 zKLLG#`r?k&K%2;zmyF}GJ&l1)2I0TLjZDD^|I7S^|Q8@=q|!!_Mb7|c0? zO$CRUsVOmHhA~*REN%{`OnftXZ{qT6(m{X%ok!FI=-~#e8ax zmZUiZRynpo@!{vxb#UMI>Ne~mQz**$FR5cG&cPqk@uVYjawahzaHoAdrEQM7` z?9@hE%{G1L(;oT@A-weC+6V2alX@X&gqfVUNC}@X+^r&*EIGO%V8o zn)CP8)b#ekAG5Zseb`MMKHMms-YO-*+1CwWoq*Io!T)Sq~{M7Uq z4(pLcCC|=Ux;&H0)F#0s%n41HH76cG$Eoo1nv{&(ZS&_(ik3j|_+*-l`Amw7=L6-0 zw5jpjH{aXkeBj1%TpS_98=v-Ajb75Z^A<1^ANL(>M{GV7U8SF zGAUMITUaff7Y*F_j1;O1xxwJp1SF%*&c0!!c(B?mR4b`TEcUqj%Cs6T*9$kj5u*>_ zIyADuW9tG>3mY96i7<>rI7T8IBSGv}#Ao za`;J!iI6ThJ`@-0Ml<3T!?`wxkC^QE_|9B$M0|qV86OvcQ)dnbvJpr8T(JPLyFeHg zktaE@>T4+^we-)0|3V9%MGHt)*Jsg!|H53DYcyD~skNZ3aH4j&y$-mQJ8gCc{Mp9< zlml66vzrlPW3eN5n$=^s!sWzGm}BFUc>40~+rBv4 zG#nlogLF>&_UtZ(ra2RSD`p&=f#ubYOdEA|ZMyz31HzF8N5>rOsZ$L-Bzdp7x9U_` zS+_-qpnJZtwTuxYg!*85WZ`jVvzWNyiF21N&Ax8VoLS3mSUx`^Re`7{PuPST*Wa`# zQ93qipw(&F%P`euX|=Ni;0b_=hu=9MF5;sALM2nmm0?N=z_4d0sl_-ci6Rnb!|#H& z!e_+B3rS(-@`WtSsEW`qKFEY+h-4>xE5dFGa)^*aGUK>0**u_+Uo(Kz3W&?b=?e+Z zgvdfkA(00UxT;VdIKjSIoEZNvr;uP#;<{F0$7ONE~H>S}s zxY%Indx4i{qa0SV!)XBm3A}7DQQ@;eU=Q&qIS)`nS`mDG0KMQtqJjvH2;4*@6Cnp8 zO?hE(h8%Z$9Gifh;l)8g0xbrQCqe>+z&ySuC_U~AN{{<6i?k_lzaJSHwey6408gD7 z4?f`-8RV&iNL?b5_$Z_o5VKkiva)C)JW_Fb07N4Y?}mCxN*Xc6XsLijDn?uXF!Gt8 zvy{ilLS_jm?7{1FRA|_@U+pi~aV0WoVtN|TpS{h67mt7Q`PMJgQ8PDr;p!F3uS-t|a~sC2bZBZ;cGfHikI9H3L~?;C>(;HCha`WX z4D=FmO=9|#jBqZ_2s3A4PtA}TU@w6Oywb}BqxAyG$b zqU01$f^-E+g}`NkFD8*55SYNmt0lh9NObBskcF@=cTaAOjNwhhs33=+HwKANC$lOLNkt~Wy?rJ`F;7DFs-4?5(sDcfRjM9Yh4Cb!< zyhAum!XH#b%=U2LZyNSQrD-Do`++xc?fs3lv9Ywd3xE-A245_boRECo>Xj?z&ACM( zoH0de0rswu3R5S|Td;6?LI_|UxR4z}c(EtQIU@tw2yiYeUaLnDo6y^9=ZE5a86mXy zG6_5vFn5I(k1Tz`9c#Fwb~xF1cy8OMVa(z6K5XQ%dwL`qxamtI!)JA}7 z0@pDlgsgKg3^DdlIK+>S0FfKx`l<2gH#iEyRgfbTFvMHJ2B#heBaW4yEDIG}tf#uQ z!iZC+B0|oe4`IYZrkNARTvL*vO%s|@Q%xh2CXFzdsZN)c3r1no3d0f59imj?Su*q# z>v97Fog`zV;pa2nic1mCl!-Jnz`!IRvQjLSa=b2z;-9Sy7eeL6JsTN$mZA95Qb(Os zl$JZeV3>ekNfZw3xLo9aQ865!$IU^KI=@s3NxZy&ME?e9jH=>Ls9WN4k7cTVT;Mu|Ch~UgXk#Cb$D3eNMWVgaV z%i@A63j14+mVpA!hddwt_y5#;DGl!r>D6`AmKE2vwYD}?Tr92ahDWjzn_?^usew2CI(LL z<5AbL{ofrqa(Mr~-Eh3z4NJuBS}C*u7C`F)JjCWRiQWjs`rBHm-5xi8-6M}Yl9ed4Gqi}W71PKN9wbkE#ok*JyBNA?^M zB4UX<$>*CCct<8EGY`AF#gNOH*xk*=J0ee=I9CFn#Il;|a*`wQVr5ljLtRY)Abui% zHD)r8VYg=_yu?j1bpn9j!bFBp^%s!5u?Y!rU9zP37*}V}{%=3;K;x)?WU;q|z0uj& zf~*cE6dS6_+q`-6@jAmrMkozk`@r4zJ^VPVJ>u7Xb3=w)tyz4>V@s*G5p40gKa~4z zZA2TzPv(R!zU7t)7?FfIHzGmbJ=s%COr*s=M#2XAI*~*OzTZPV-90VsNK*61iN$XuA6=kCRxGU3>PPEN`l>IeT#T=Fhej`fX7WOERa$Mn>X* zp@FIo(0$n8+60>U0a~SkR>ZS!GH8_wT1^J6R8`2ph5SzKbqyU&4V9&^-zx~TrJkUj zLS*6xqmuIMHV^pjtdz55CBr@(lLQyHf3@rE08h)`vSp~bs+bX5>4=$2Z&J*~_9%@HJeRbVHZ*?<3l;BdNfNB#cx@dFY z2P1B8D&6<-Ki_$$?Br7R0VHAjBj=lXbn!Z^;!e1t&y7W7dQnl>qC4*lt`I)@CPLqc zjgF7L@zFOeR%b3p1Y;OAmBBuJxt8i?f_CFet-4n$xBv2TepjSxmxIL=?}Mqt|2`N; z$onM33VEqj|L5=1`^!}X%Vh-0`AE5upbhfM{cHD^GX={@1Lgb%(1>f6v-!(41k0rc z%9;Q5`_K#YukP}44pI&;^{@FWw^QgZmme&LL`#G&XrdN6vj0T64*&ZO2j90O@IDQB zU&S@wH|8&g6juK6n-eIfB<1?ASq?U4(4{h1E;CS$CFNSLSuQ}&mS8zWpq$^YL2=D; zfpNbWEH^7q&Ocv~8fhd%ab4<#@$=OH zrxp}!3l@wD6ts|n?rZiA1V_29%gdQaInOoA!9fJ&4g||VbwFlk5-FE_&2lIG<;J-L z&UmDp-$yJpSS}G%A#$7KdHz9e@7sU)@S%NszeTQ;Z?7%4Gg1xUar)cR*?v7LJJHkRDq#)>_EO#R9nUiin^NkrSznY+W)HHW&y( z<}&i05u`;M?rDXj>mX7anKmSdFXr%(bR#IW(PkpGRcJHu$R$#nfADzTVL0k=)1d!G zr0u+dIwG}2z=>mceU{f&1w{5yk?1JH7n{riBC@S1DOQFb)0Zb9B70Ft$f6*x76#{v z1aBZX!V*8P{s6vw)9B7BXMyizI(2$0dn1LFOcw11aKr6}4oXi|jhcH+N9^D1E zD^dwhOU-0xD=K*ht}_^x<60JSI(5TYG9G=Xw^Li;*LbOJXk z(U@GWdoK0g7~R|)PC0eE_VpajX?Ux>?&o*1Ahl${zTkvF5GpRC=8xA%Ja)(&zI;?{m_Rwe zCqj1{ve-M(-EdMN*PZ$rm4!sq{sx*KrT$2&8+n#`6hDby9qH@OYU7T6Mdb$D_G?rX z5~ynK^Qh|kAl;&<8UB{dhcDny=yqAlxPlW1q#|_t3UvE*pr%Eypr&I%s>MJ#7oeKr z(wctWD%);&A_Y45B3gwlh`)mmqOL8$4lc$VCLMefik(YZ_4B%f`S7+wUHMt;J*Z0^ zsB0R*C&D@-qZ#oa^?Op6?Z3zjxqPl&*2D1?aaWLHSP?#7kYZ#-j8p7(xHY0atcai0 zclnCw$YSq8n^u$h4rj61s1J;rtO%f?zmbzg7Cc!I8hDQ3w>_lRibtt+|FV^2ojvkn zE7{~rTKShiugj;x%Zs#fKKc0zdVMJ$P+d~rzF>W~Be*2c>+Gx5M|{q5D9qqz(Z2Ph zeVD=2VG8~E1yuayb7#qT zeG0Tr1wNnzmqH?KeEYd`Hv0s718BEM`zH8vHoKCQWBy>qOnrj`4vo~um@)=*{EZx5 z`R(^pFB2!f(;JphvA7|$+=qG(q28}h@5iWj7wS#;?!dR(zWVHo-G>f+cl6Au!h)mw zzT0!I==6pChYoG~bo&>F9zq^zoM71L$b>qHox;vXs>r4I+j4e2(uV#DIi{Dh3-S8` z_GV8Pc#jA zBa#ynX0q)DHaR zeeO%oKmY7XAn(0TeMz083h^nSj`3q-A9{#-O)LE3i*L`Am$#WW3CwNf<)?Oip`ENN zFF(9x%a#M>Tvb!FA^wGPPWoD+S_>VriHZwC8UYSvunQI??bJF!w(kDSV zl$bVi>FOI+tXQ>f)tq&!moLv+H#a49;>4*5FFn2yz~B$v^Wf93u0tzw{LmV?2;|N*LU3E_aXgOD~O4Sk5YyT8JUl(l!nK~#KvYQL*1KF z?Mi7%EOJUaN6q~()OEJ^jOwk>J$i(r<`(?kFskQz-b`3`%es}TSFgS4)|+m;?XC?Q zZePD<{puS5tiSrE4Y%ET>&-Xck8}I``2I807jFYp;5|_DJ>V05fXt+u@%ac@O5efv z+xXkx{hzn-f08x)pZNTP`UmgXXaDdAEODJyh(z$fXB7%I$pldP+H7bHewS>L$>D7P zquZ7L%k%#?;44dqP!tWI1Dc%aN#X;t7y6g0?6v>#Dhp#ZtOC?-<6UN4{ zsmNhG37L#zST%lEvFYq&HU(=l)c*t@T5KVnk6d=iGCgtiWxC`SEz@f(W8N=V$FKcw ztz%!Yug_=q+3;=hnSGL|6Fpj9HbaRJpunP8@hrzg=9 z@JXjbDJx~7y1{SG;B%6y!M%|Vp~LZsqJ=y^r^!Ot)fITnl7ENST!|M+|1-=;5M)O8 zQ^%-pz(sdbBv0;F)aT%#`>504p=Uvf0_p;I=>??iF2P^U;d2sq2dS;pHt^mZ$i4d+ zFUZbDuFB4q{33RC4R*)-h5YWd|0%zt@{p-H5B-C46=uW<>J+|@ff^V5v^Yy0hx|E? z{~g1Xd=BIP1^E9deDbOMAit!Q|F`_|-{7Xtft$VrZu$hprSAbZeeC}UZu*dK$T#fk z^|kogef7RZpU!8%XUJzniaZB#(ZbyTe(wY4A-fLNH|T$&1stdi&l@h`#_#{kQKi-+JFHU!pJBm*h+LrTH>^i+pQ*xA^|- z`#bpi`@V;K_xf)2#reX0(Y{dMvP-zV_#nmXU#Hii2NGOb*Fra3Uv*o`7EwWksbUEg$p<@VcgGAazvS5ACUz z3gTYiY(v2l%CBWtp?(>pF(pK<1m8itV1br1jf!M)IL=K_q^v|S19Cgg&!Tpsewmgt zK*<=I=g6YIzzg$1hk0w|7o7G~fYkiVE)_`fZ{)8-E8ad?U+OgI7Epdq*{#cYLko=&4ZrcYPMsi!SlC93JIjcy#f-)MWA`H=v=v z+B=^0HQ*hO{p20u%ieKj{2g~>QLT6fz$*jqs9p{y(I4NzM?VPO!*~?w50+ysblJIx zB7K`!f~JwnecGT;X&lNBJdrq1cHBFOyZxIzI{-XF=O9G&6?A!qR(SO2i4)|T;qH{D zzXz2#-PjDJCI0!CRu~$}vgDiLKJOcQ2HHD1X+jnS;vv)s5ZsZG+`AVqe{v~w8Aqw_q=F%1 z#EOUSoe>t3_3TpULEfacg%|gSOrIXofB6qe4csK7$p~amBq2v)MSy6xr~whhfjjvn zSH#{(-D4ITFD!NHpKa`0e&36QC}no9By zUhV!8pNF5%+$utH67iiy!pgPO-^k^!<1=@BwqHJKeq;=P5H#%ZewY$}@VkGVllkfS!=wz@A|8je+JE`6%+KBgAGBPKgM`>~ zWP$j1{6R}4UNUokP6IM?sn>DI3Lqc&CvL{I+XkDu7R-*6@mtbgA-}@_{A1gG{+^%G z8f3nN)@h*JQq1?PL@qTYk*4M)auAsWTPtz%!v)Sj9L4zy=g(iLtFNo8|L(|^-Jgk} z`wV_9WA7(0^6dRNA;{xOmWO}Z5*Ip}lf^zm|Lu*XYy!?bbA(gY-_Vw4E-hOMmGSfJ z8rAZ+-s8Q?1J~m_b{wbtHQ=peo#Bhrzb}60lbp-}w*v|$TxZFb!wEoJf@+JefkCk-F$_lj52=^3;Y;=x>Ls}Ap?1V- zptU4ogS<-xNLD1w90V#=|De+(_6@-a=*KUGL;_8{f3}f%^Yb~7Plg4UPPo$k$##wq z8YcQ5%KV%Ykum{}HB9c#lFQ@462ullte3<_MXanu`f{NrZD!!>N`WKKSbU^t{C>-> zyHXi2F)D_^GlP8$MH7k6RjHyPg7<;8kJn194Ww4GXMi@$?qs+PBtLKtN1pQuLtMRS zhIZ3g&^|oNaQh-7`|z)c(ZacKN!FB%Adi|~4x$Da>87`483|AQc z_iYT9h!cjM5}6G>%T|U<1PH@TUA=w+!zE&a;b!T0%i#nIVG)oNQJ_L3s1SqphM~P- zXzw!iHJBJ)VqfNN-}U7-Mi>-eM;ZD;9ns`Jz;(yJ^G`+?lwEk?L(gSXqf`TZ9bSBd zI>-ovlFPve1=FvSrmns@jS&Xr)(9iKYtoGSV|Byv3?Lf)QfoXui}Z5b2KmF7?A`a> zeujJK$dN+~%-G?pG=_@^ ziSf+b$*6ZuA`jnODjB^@rZ_4kg(cu$23Sp4RVonCOWE%ii&Ew zyFF11rQsL`jkb}#`l`47(KtrS5z85wmhzj25*bZlVRaWlY%D2}BhYJ7ES#Y;X3lgJ zGD>)8i~==Sk0>Yr;hM*kG6be6VBQD;vNqiBWy+MKB;>tB*hq3pa!N*eiV_KX$B+mX z&W%XG=J=u75^q^+%^v!kuEtGlz4K+r*%^p>Qsu(-n|T(frVjW;qmJhx!dlFUCL@>lF} ziOVmZDY(E$1R4p4C*ZlpcmhV!)6>=2)zjI5X^tvrga*GQfj77+BoyH%DmZxr{+7eT ztPJo3E`HMZCKkNYS_XTlJ8|{jLK@(m|MruY%D`Yn+b^y1M~?JU?h<^un>;=1(jqw= zto-q(bAZ`~2ky{%FAbMpJNjr%!eLrW_#v!DDh$ze}nxq^d z_ zZP!GKU754fH0^cg&Yf#$*G!nrfOvgCiSF<3Lf9S9T7jN{ul{~=N8~N}0jyrQ43q4b zLxUp-65B+J#bEaizv$A&foLQq11pL}JQAGJK}Te%z?YBq(#%X131bP-XhOom5z7%3 z>Hk88G28*w6>*)BQ4tyqkfkG|5F8p6M@{@U-{7C2Ld*;yW`+Awj$pjKm-&Km#*io%?@$154DT4k zI~vIE2LkN}^nLJ+m+p}f34rxUgbRmW*hNEmAZ?{@r0D!ex`3jqzMzd5j~==|u!6GD zSCXY}8R(RalG&hBdV4!i?>oCXJHa1;D`0W9<0tk5t^QvLAT|DsFa|9}Z$fM>3+QXy zk>FU0vx#EgPA>w^dQC$Ic32(FI6t+uojrHvG;)1a)|6j_yLVe_aZyPHwN3l_l}e}G zbI(1$TDxMw%vm$%EL^_!mgioDv3ldeh1sjtuD@gLs%487-g)P3>p_e6(E#q18vqw? z08t1s73oeB~4 zsu0Z95RYArpDIMy`hUg05Qh+g?fKOz2vC>gRlEn(B~g1nAhlre0H}M6l1YaK`*Z+7 z#(L`M5sN&-U0vNkh1?VcAAb=5V!{;&H1mS%c#L|F4cX9;hhh_TVvTPHOq~|^aXiAl z2=|CJaDPZ*BWRms4eY~n-~zFM-N?SoK8fv8D9jPf*o|$XewEEW%Dxo3wODUyt1l@i zC?Fe#j3h~D=6%_S75JFi%h1mu{lMqYgb}OJc39Tmg~fdn^(w4JZdyu9txI87)KS~% z|N7kuY~25fLf^vrbc|}j=IHl1P}M|D<55n#n>%^Zyk*OlAzfTyO^u7gMQ~F!6OX~@ z?&+>OpLe1ig9Gs6j*hP0-qWW`ntFP>z}PoM0XQ891dcHqNnS&`jx5)z5QKYin0Smt zSYe$zA#=v{%cdg00CA@i7R{NJnaI&~gi#5kbl!yy3+oBf$U#52e;|yTN2}5oFXsiN z#s5DV|Jl*V93PFuA4elJLKYZHi+<$t(TK#_h{xJU#@dJk=h1-kNV@vkstbDi^u6%? z6bUG|h|GTl!$ol1zATnc+htkoJkB=AY4YD+ojZ5^^(Pq4)6|RER3UWea@+qQ?>*q7 ztg^rH`%KSFdasa#gx;%2O9Ch;C}LULs;qtOt0Isr>Xh+@IE7Nl4p5Snxd z3F&Pny%*9m&;NVwok=5;uVnHOoeS@7XgE$7R$As226@EGnA57gAj`@EZ;M$Xg{E z_+j79ya>MnTQ>N%FssOl!RkRs`nhWn1V!rTfK*19CMdTkA1i-Vr*?Yf#3{?AdYw$q zUE}rZ4OZ(7!$1FgILCP*j;Ob?qwHh#99{j?IF=&$bYdj^!)&_Ka4vmgNXW+0RjW#6 zFG9Y=T2YjrZ>=oM$4K#@7Sy*QWK;)YuPCw?H6wIHsGp~oUnqo&;q*jQv`N+d1Yeg1 zZBa={VV14W`qL1dHk0L1JP4k|9oLFeti_P_fpv6 z#YitHG_eMsMobH}sy#LJAZAp|tFy+AQ@1rMnu(Zkll&+mdS!=BgTGG5H{(h6kTg;n~<+*IL(z2+^DCCLHd|F~mHVtk zkoCYt=3vd@`6In0O-;QvtYKb->vIp@&QbLwJiUO}-9wd+;z_=IPhDyhjWUWrszOPPwSKQo;;EoXd^G55^Dd@S2})E2$nNb|8lm1O*U zi@&#T%P)s~pas)!{+PbL;zozAlI9u83^%d&%vs%5QG%py#brqBmzPWPjJCO^2@{!f z=9F|RRj;8{N1d&@l0m;VP^(U5 z@>b;_Ka6d_R#jGN6;CK^Jm3Qy5KJ@ZK&_%Z(RJvn>(N&f50q#A{+++FE38>*#||F& z;@cxf_W!mc_4q-Ei7ANZ8-@@7P%)+K{PEP${mIGJvVHrKxBvLr$6L1H2O0kD&29hK_Y zm-S>uWeFR*^5*%y_M=yU_LQ-7j&t=o2*2S&Am^Er(EZgk96Gdb*UwuvZ#_saIB5{= z_io$r@n=77Pu{n$%!)9uM^F8@GX?zt+It>Oc=GYP|Fn4SyoJlIyX~`@1enOZ&o-zZ zy>sEs3G6#;;>T-pp-W7*SQbjlrRmtu|0q3q_af-c-v-h5u;1Osr9F249k<+c+ue_# zU)Hh?BGj|Wv$_}F`|OjGXTFWc!u5!>$|nIL#{}hm?MLr_@Xe;R3HWxQa-%8?7U~OO zvppBIU9DaB_6M)7e~~sru}Y{^wgmm4OZxio!MEPt;OL>X5w%#Bs~g+xMa9T7cqj$o zZ<<;>{Ip2CRa*-w*AgVDyuxFKn+(iu&&o#>%+^Mumco&)jX1LJ_k%k>fA_sFH~)~5 zlApb6_vV-W_4%g5dIZVZgBFI&pMUepv8<#r1;N5|4PvR zPY(KX-lrVgr%vtuZhP|nV`t7ZR3Xb!UeUp$siZjq?D6~XiHXRo^X>2eUmvf~@US2+Qv&Dr_3PKa^6p2UZhHTX z7ssV7!c?|ESqTgG$EA0kTL`mzll`du$0TV^SiEwh@`(PcAAdiw``hoIi&w5wmV33` z``ZCqN9(}@yS9Gw%{F3pXfzVA9l@XOiF@vyJ@cBUu`7^v1Si0dFiA>Z*{iqqLNBJUCy~(d=PVEB2_72+GuUt}xqH zZce%jsM9dYB%yd2;j5*6-c#7PDiKm-`=5k+qc=@&{z?CgP7m`qx~p+2bN3Ey|W0O1NWc@7AUja)5g8{#`~K- z{pj6S&`E1q)FO}>3Y;T{2Qt8=;o>_?Dvyo-8 zy-;>x1aRZa_3U}=Q}^8d(3-~**eB?U5-E5Gx<>c(H8W@5d(RV&PFrgZB8nqKmWLaZ zA*04ai_+0rlwE-{8qWWUy&m$$xi*A285xC$M2HpKK@phOMj$Lyn?7V@Y+xATNE8;U zr10tELWZB)@$tTW3i1Emy_>&HXCVQjCx@$4<+d)no3Cf|hya2W{Qm%Y<8$=JN9c_Q z&>Ih9_dK(|cA~hXydq^^Mr9=>X+!efoTJHmtUO;=Z>$nCv7?Wb`G?4xn%kqi=p`UnlZ6)`ho zyvI%It%wnOR^znz)-La+dXguq>WcY(wnsrT2HTJ6?R= zrEz0?HTv3W5;W}Rs%!Nc-!XBQR*Un7|9d&r)L)SNU*J^p|Gk`A)h)>W&v5GG|7}j4 zh&lCQ%&7^OQzv3horpO#pyFJGwan(!9hFoSW>pkphV~N@P(KZlq=TULu?})jFIAG1 zI&~r2t>i$;`Gn1t{e(!=Pp8Dz3&Cir=5@^HtGm@p(ws93m8EPdOHuA+hkL|{^Hog& zS*LLZUEowW<)&FH({c$@Au{z-I|Wo$=^$b_l}|c1zfhwdBi&C38vRsIV$Jz~iazC_ z4`LrUlyetEpQ+%V>m56!ncyG3Cxw8C!?x;5YkOlE4lZyaSCoD-J1spMAJ99A@U7I; z<>b7e%gHHg#d|pjhcbV8@7rw2*xFcGTE2hYs#~vFG&gembqVY_Xt0hmSOmfe^0KAU zENM7g6W_#{@cY;c?1VNl!*Us16_4Wm=lITyz0YpQV6oCHcA3m76QC4pW3Oo*y6^74 zKDroNhmXLJpR$kH5s!tT5i^!9`Rikk-FL@bGg+JjBS*8rycm9JS73Yffbt9khnHSW zzPweEAM+HPH~$Qo{t9KJ9D(I5I|q4A^G>GcAyzy>16!*b>YAIX zZB~9xpoiCm&j>^fehwjqqDO=vf|^~9@bdODr53asbrHiu{rm#FJUr}8EuN}U8IAj5e7*zUKZGM36MRp8XWQg> zX#`s<1w69|JIznfH=nTgv=2UT-(T-vxC|=FCOCX$V&8cH2RGy@`3Y2)U$6`_b;A;7 zR$9@MS?mY)4diB-y&^sOU;>^j!vTK+YsG!D^x&hH;7Dg3jDkLaR{95)#XF+d2G%R8 zt1O~kKwq3We)90q)SO&WLnj&#jjlE~4+dRp-EhgvEp5ROxf}sAih$*DBZETKYMqCN z$;Zzxdg2)008R1Xy&lvRXukWAY}14>lP?;9_^Bp?VZ=p~$Baixj3e$2Nnr`b!Ce>! z_n|MAU>q#LJQ`hWD<~`}L3fnYwNy7EC~R@D;E(L={NiJ&`6ckWHzS8K;;n?BD>^$P zBmFh#cf7+@;~yDGU&@|JRauAM>^<*&`1wz}vW{;~z`iX(nc$rOUGV$2T{YRerB0>u z2}HCPvugc%9Q%f&Tf*UaIU5{$J>TTbuq$vd{7v&E%KLJu{WpA2aP)RNe!LzJ%~nRq zIwkPIKMgN^C;@%-DSI(^^_puhzjXPZZW{66Z2zZMUUkj2S1*ek9ua}y5oS7oTaT`~ z+HwOtFBUQdN|58QN-spmv`9v{1Z3kJ?rLm7vvJ7bg}9!T_^krZ{Empm3+3p-qLSj& zf|AUPGsg}d+Pm)vvWX&CYgSG!Ok}dkZJDPL{=%$=PGhSjWb0Oh-+t}Y_r5~(HduIW z+rIs)_g;Ms-))tnM~)gdHa2$LxN#R>nlNTUq^>kGE@8BP=v}L>m^)?quwhZL>`GCN zWC=kO_$gzeV|^M=?u!VPLL&AZsq&1C9y29e5~GRYU2j4Uet{mm8$EcxkS;J~<>wcb zoysVJ(*dkO!@pPhKyQK0%f5^2SOWyV`4Tq=Q!u2 zx4}8^k}Q6k?zncEe@Bzf7hI)RDax>iE=flM|lEL+x>!ddGbJByXOPU}pU@H^Z%4?(r z(kNC0p>RD$6Dd&OE)FH?I6$sI32RW&Y$*9t?R(g_NuwsZJ+kD|rI#;>4}eaEZ{hN*LT+a_S>*0>gD4VNfuRHxag#lYec2p)HWH}93! zZdi;X2rt-7F*rqhEv6ZoN%Bmbnq$~kRoEJ;Y#8^>>9?fVN>gE8 z&ROTwY|TqMc^a0|IoLE{La-p*BLaoej36cT%*~126Lyjttlk{cMAFQWenEfv^X0QA zO$`r=jKM3wgoC)5S3kVQ%QSNKjLKHu$ndGtC&BrC>_v;>MvaS5S<~jlBZP(-v0zgU zS9-)mjh-AP2Na*o$j@mEm^lx6&~n?n30>Lw87GTnW7y=;Q8C^PCsL63m1}TUfg=;a zk?X;ct3ZHN;K<4SY&CAOEO3tJ1=mFaYe zzAzioXrQIu;FatX3Q1G(yW(W-TnT6&Jx)2G%Ap?I+X{t>mr<$lG#Og54j#$N`u+1)Ufl5Z zkNcLw;rk#CE=l(*V|1UEJf&#OpAl@d3$utgtc>ZU?wAy$Yt9D)fsJMc>zXrgdf8V; zvX;G-07cr<%9Fb1H-7Qq)*a?}L@W#M_;~UA&v1(>U}c>=lJ$H%=~57vAmkDL-b#C_h)NZX}aVD@y@!`>HQG5x`kW^hdvTqDOgL z^qxb|Nq$~8Uo*+Awy~_BnM{2sPmICG*WWKBECgn0;lslG{X0r(>T5h_E%Ng7Lx452 zZ=YjP;*<6cUVM88B1L0%=zmVC1I+^9 z0v&uYf_%YMdYi4LCO;bn>sm?O)@szt`p{o&ekt{c>#7}d_VbeyZ;wNEhTYjZj<%n;uB_19s}EUZKIpM)7~ZaB+mTsQK;AH z$>)|)TAB_Uqddw4uSX#DuGUVt+||W-{u&W)-Hn)~`WxDgPq-@P3^<@6PPt#T_YxeZ z-l|rv#c!xyWx8%VT>Mlgud7BV^1S@oQW(L&7+l|`H{x_tqsVUKr_7ojF)3~`a*XP` zS~?|lS*IiZqLY@+=!g{cMpIm9g_Z0&P1BBa_BA5K1wOop;V=W=6xfPz1hr1%jf5mI zmUTkQGhi7dc#@Zqj@Tts#-ayNMiP}#WD{lh`VUb?PC*VYV^>`=`vPU$5qt;ASdTJn z1)>ap@1e@T78co;)@dp#w!Zx97{rWf>uz`ynj+2I|J|D#ioG3l>Oq>#iN{eh&*@e7Po$eee_4N-o`S42V zv*N*_g?)KWxR>PO>(+>G)ZbXB6N838y| z=&yL?GWLn;HFO*H45N`+ZojNT_a(4H%H1pp^=`lq&~q4-*=TKnwd|}79h#^zvKF+~ z+b>w}kB$|eMa|^lJ0=p5%_G&!2d(vwR2lS4ujZa=dYCdax1MV$Fl%7rtSA=dUx&1m z1>1l4=18%9%*1Jap}+hzffZO1J?6{}Qf#|+?XoFBb2Bs7vP3VH&I5@7VHMckW-p40 zD3%kE&k#}1BM}iT$QKdv*TTuRwY9merOu4N`_LL$bXrkyA~KS~o{;+>a84lW@EBkj z4J>^lB{nQl1)ZOcF0gScPEIlz$yq^yFA@HX$Zmp^Ca}|yx{!0Ev#kXov)xoi*s;Tn z4}Zu5gDixq&SoginAxD#d#PAkeN($$ry-9ytwsl9GW;d4b3+;2@lUNa!06fA)63f@ z(hLu>L=++@xB|G%M&AtwZYw~+OTjOrN^2XcaiWZ2QktIwcY@NIhPslPMhXsZD=fp; zuv4q3M&DR;YAvFuy6Y1|kPeWwz#^!<9k$L4Nmz6`qX%$S+U3{PR?Pyq$IA4ly?=gq z$DTLI<^X@v8++u}^=9+iDV^bS?@54~3xB$M=4q_^-ZIN>1{e@%&}>BS+VsK;`de48 zzIl>imj2|6hgv7y_}HlYvu9uylwVR&mS0+d{3ZDrXVMEYQqwXsGmDDz@HJFSIhkbY zu0xI%*4)&V7=oNS%nb%aK+m0)O>8Th8(R=MLnn`d1mb*+UNAROTX{OY$)~H#)~F5{ zKf=p50B#L|m?(!?UI?i=^q%Ft!&NAz-J0_K!7>s%I_wOY3gLXjJ=+q#Ve1)lxF>E{+_+A zJRm)js4ghY#1p!9ruGhUi{$TknPks512tZRW;yLCBO$hNUjVw}m0bv$z$2tMKQB8w zD=QN#d=hv4SnGurkd>fTr_*YXH@IEK24#b`Lz?v9<4?q)R~BoP)Yo2kK2CWS`b;O# z`t=)9sv0Ifv3m6rae>~)H*g@$S8=@CRNhwEEvM_0_7g!7I-w9E#qZ0`0 zVp$@y({9xc)ELG-WZCw7Ni?-NnkP03`k*UZ%7~fhbg>w{M?g+Qh4ju;9w_l&f`?| zT?L0$3_;;|Xz0F{=FfK)az2IO%=lkX*l6^I$KwA0g*D(|4gbSD(r@Nzt*}VGF@0ws zT9y1GHCU0leXKB-L;BTXCsk08oMb@8F)XB5YV=Lb$cWU`T=AS<$F#JjXm!5c`16gS zZ5KU^U6;3aq>z6pJ9Id32*+sZ&JInFl@(^t$pbu<4HCSIsmuguIn>GuRVlprWmca6 zg|mlpy{v{cvu}d(7nHD96@~4F%sdiszEsbCklm=LjQp}fUX;{|6lq8vjb!3TLyD0u ztNB-aT$9BTt!{LlOBPGBF|MO90T$aTN(x-vsaRD9(`k)rTkJ>8oM zS7|pmKT{J_ud;`dR4=hdlJM@nB-JZ;rz28DZ|XwWbse=U=2Hc4J^8mAyPl=5Mtq8S zM>bPK;Scdj5-u;%1@G=-uO~%OlM8W7;_A+E#+h@Hs_!L=2;*b&3UGBNg7t&pAH}GB z3iKln%|}7Mr$E1_K)(fqgLq+P%!h|rKJ+%w?O+(^!Dt0W?Dz^MyYvjb7oWx4PVXfk zyCDF6XK>7fLI>PLVFlDGhHqpO`M2}`xb*w90n^k}Y~{+;v;T7Ey*JJbwIAI0?!P`g zGHKF@Q^nz7g-B><7&Rs{ZPKJu_#9ejY;HBuXW3cXyYFo;RfjLV^Nu?&k6^{y-;+%v z@^ho6oH{inHYz(`HpS*=^G_q9a*+}VliFGYqg7ERPq%h!+Non1)$O`r6DCd?>0epx z7hq~?G-3JatgZF&5ucmfkTTj{U+dvTRibSQn>Z=jOWl}@P$cJ>umSeOt00CMPq64F}4i3!Izz9I?2;t|LENuLF2iJlVa zwGl;xu)sDE-^eBhVX+lP44A&377%}k7uHc&Q^Q4z+{QhztRl@2Y#_y&inF9DKGCe0Dwf?0WDSoLQYO!DoVxT=NyN6^!se^`xP= zC>mTeG}kx_8;VIrI+$c=rhp*A(Hn#ZdNo_{0QKu&ea@TB37^!rPy2p?=dSZ5dU&WF zr4EI2B|3D_a(Vx)ME_lj{<{JFw-Wug68$#@y_ZkyOA|pJrUm5fL1^K!QmB6Oa&wD{ zt>ogJnVpRWB$CIUU6I3xMKRo1YbVpW84TEo8gMuR5dhiGWb;To17BpE-ZGpYkYBgL z>Mpi{F`-X+O-@ku5>6K4psC+P+Aa657GmsowmFop;pO}#`!Su?;v*Ghjzq%=F*n(T*nFYej zBQp7`QV$|a2FDZevkWUPYT5qLi{^iQ7pJT=eE5R-jc+_dXMsqj6v^> zLGJ}{>^bM*ofUnCZUm+l^0$>F5I{i=Qzj)Pf)sdR7YsEUPGFR^f8-HV#~q-!-*SE~ zNCc|!>M&cOA9WxL^5%L6UP9qqNCWBNSE`O>MmA56zgy4Uk1Hj`A` zPhumF$MKOG78NNx9)p<*a=VeY9s;V$2tSh2%94_NYeBNv4f(58DxC_+aba>(m4I`f zhonnY*UvhXV9$i_n7530-EsC7RpwoH#Ysz4ciokVPcqw}a@(64r#z|6stA5^#vSqo z&GZL+($iskuTD?*c|bPO32qid-lAeuf(SWCYHCMvbECJn2M(aU0(}D_dwJ(ZJ-01s*d@}*J?4;Bt{SNnd%LCsVhtz>cbEtWX;wVCgM>7C5B zB18d#YNxS!m2V4-g!C9$SKv753WRdPJK0cBQD0qav$9hbbt|Jhx~;OoJ2Wa{xW6Bp z((|3*yJ+A=)^@XiS2XyJ`)E4wf(=AQX4ctm^eCZfXD3c>dt)gbt-J(bwI6|z?o6DE z|C3_H^+#H}a^)9atXzq-CcV*@d|P17k7N2BntLvv`oQ^3(SG#NHP?Lf5h9<-UXDgp zRaTXimXvq5O%KZyLaJ5vax@Vk)q{fqL%Lhq^91h)vQWAS?OlQPPDOjCVw?r#ALP`LR<4A?zool} zZIa@ifDo2BY@>Ab<>Vr{MSb(-U#Dy?|9MuhOd)ya*}@k{a^pGNJO@q zag~E?R-#>v)mB?mSz20I3Hp^)Rf2+^FzEq*bVB1m33xj!L`O2e^7kWJ4)gOL4riR8 zurSawI6Mp#bp^pGn+3{NAv!Ugn`D8qS)l9!>`ZAAE5N){bCYfusnH99R? z6`etyRW~s#k1b_IS;r3?{AKjCX`@5D z4V^75u$KxuvCANxtF5c=H2H>&8b9{pW667tWoK4eS|Ml97+lykR%yg?v z`{BKJ-q^d#J9g==HI^emmSCR%e#yoUZ&*F-h- z!)f^q*oQO~rQ7Z4MNN2nt|;yB&-e)sESWWKxCgYC?!(5aRAYy^V~j1CF$M1>)3C9# zR|4Vy9{lORnIhU$oU^4L#K)aS(m7-Vwq9c|=GCCbiGRV**hA4an44mNU9DCaxqw~y zZ{TfguxPi1C-DL18DLc_$qmHIMc9kjY$YfGi!A+MwPHlzp`2{mq~Qf+5n_Z8ye%i& zit{zu17ZYN5Nt;rC?#@+MM7dF(GwS9bK-M~DThNrusNkOh6({@gh@DdM1Dukvi9fb zXNt~Qz%PKPK}$M>VJ=l*V}tz-qFMP-7FrLeM(`V=SK-1Hd|qDofNYk2 z{{DU>mv~PD1olM!{(&d~y8`;b&krDY$lU}PhBgbHcu#?$5QW0sz-Z9BHXNMfw__44sUTHBcJZ}^-@-o=rQRp04dY=W3xgv_fI zwbiIy*3OwVW7eE`vdPD8_t^s_#l0iQ*5G0ML-HuqefQoT&vr_XQcLVNFLI!BGthY$ zW+6z|(u0%<%!KWHUO{#)PKso$wY0p{S}2>SAF(aj7M%LyP*^2U@6Xp03Wt`_@ z-DkGeVw`dnkRhlBO>4Ky`cCQNqWDf&bE*uw@8gu)rR_-IJk{{v67N;)28)@EClkSK z_&;9OTMI$h5+tTGYd}?<3W4Ul1N?jfSZwzuujJ%Sf^1@1cLj|G_ws>HB?N7VX7>c+JW3z(PUw6Q()>c~j+izuM zCr%u}yX>^%d-m)}J$brRw#4jcxQN9Ji=8lbWN6^<@Uc@SPyOkanX~6ijVfZ}r_Ks) z&G}}_<`g`)&%BFycm#MWr8c-uR?yM4uWwXTpvO-?6?FwKUAi=kmHdR~B8_G2WNtxr zUQvZD_3){qzaN=9bN{}B`}TyjA3uDsSUdizYp$P<=fW=z$H_)RlXh5iKtKRuWCR5@ zAyaKt`;sN$nrb+sHmO3F$fL2F%*-wD3UpVgb$%gUqozkkkDfGX%=mG$MuiNEjU5#` zBFw{N6iw~~jgX(bvD3?llLBiN9BSL!F`_UXw03GSx>ZVxT^@~ZFw&ZetyL8z`Nd6b z#TB&`*?Cpfbdz6-UcKR#Z($ADm z_GabyarNre796iyaH@;U>lI~X%97wfWm^fv77O{v;?KhEc){Ixgl1b{l0Z4h>=qP|!YV80v7jIe@Wp-{_Qn{2-8d(CI0~i;2w$_KV5%U=12Urr zrba&si3EK6{K)a0iqUXcB;s&HcJE7rF=OCGat+-j9UVL3bWx=a%FoQT7UmQ}n5Cu^ zWWi3k6k=?$POq}VVnj!h7({P`8-T|ffDoN-!iiH;LKp%VGruJKilWzgu3!7o=HIQI z6R&@K%`=bQFiBQV%E>?guLPIjA6ZMYEZ|4_vl=44Prf7HXWDal>5|2ryl=|`jiXqcv=;4CC;Ki zK(=1@=oJx??T44B$Abz>t-1L&IyNO57U1Z|nhE7ZvaZdJ%CCfNS6?t*5g6z1)%jx(0V+0c|21EKql))Wum+Pc_c0> zEllpAuhBtYz1og4{hZ|RS-cf?0+655x^n2hm1xC8QZ#4?0gz}x_Lzo{4A$oEZRF2 zG)w>uV?o1MrtmA$*^$B*_`0s)7Wnu1|XT%hYP^v>P8=lcWml%8;M;T)q0P#bHwUuJ<_$maZjYHO0&9d3CvN z&fN(t2~6~o?jN^K3F_GYE^QD2*%|Sa{5&6;x|e?}k%p~+`!HuV)uUI0uSGC(DJDe* zV)?Kwh4o`$PF5DMECFKB@<5%U!9cU9+hDszyCPrz0L0SsamN{vw@+Z?`45TJY$WIr zi`6Waso6--gYwpsS1DwYlG41ha0tuJ$}1*cqa^qCb{*C(1w7m?>EtBo*T29gNU8vc|Ji|#m2Iodfj`vN1mGeY91bX=9$Dq z_Dn>gHR2g-BKKm&zA-DasJs9!XeCA2dAvTHkGC_Ogqc7GO8`I2oY;dI-F!p5P2K^4 zz9vsB1WE3FpBk!F*;tg(rUKMuBkf_5J2eJv!Yqvgc{xe{eM9RGwf$2#?L2`~s86Lu z8-Xx225m}*Q^@Moc<6!XbthkHy|X=mZBo*e#nQKr;~{Dc+Vspb&oCTK3n~^DW#y4u z9cYQ zKCh~#pt7W}q|jDbMq;FidWzb_JIZJJ^z`&Cb5@<2IR$=WQDe)I7hE<`KT(T#H)+25 z#$t(EsD#?wopSV_nx z>IZ5U?<^jsE!b!Xq*et`S7G1W_Yu)1in!>5@#ul^=*W|ZL8gMj4lI}qO{oRvffg3N z4uM56cjCH^sY!LRPF)=mQoT+*rbre9MfQ%wPS_&KjX}ATaefqN6$4tufL7i)h!2{R z4WE-7tU=IU?X~D}nQyp+SDrsCxNLKlbt_#NgL2XW!4>zkl{@ zi!}Rj3wvDlzK#83=CXTN9L~69WjVX$Qg$=)pRiRs7cbt42)8>a8Y=$F-Zn(kv=!U( z@@!ZfpxD}n+!9QS%DTENQdb+2#LI9I48%yqCd(gF85CVH=Rd>et#Gu*c`h1`_JpH7 z?nNaqCM$$w1Lf*-FxBR4|mzU7t(zYMiOK9UdZ-A6u$|oJFy%N zs&`{qh~bu>nMKEa2)L@&b|BF*#5_=4c0cmn3r6LIH;j$z-a+hJl)44r)T&EkXp;LJT4|H!U+z=07N`-Ag@A_$#k{s?z-P zWuS=^#nN(|xvrP|{BSknmgK&o?%`4MZ%W8ayWv(YcFmJEV%KuLR9s96e{oB4x8@f^ z0c0h90BPM?I#{a$_dKYm-Tk}*eEt0Whz&s5-amEsK8<|h=5D&eAfC2R&@L6Q9TwPl zX@M)pS$0p~{WjCQ{Jfv)_0I&tX3cU26ZUSTWzF&0WvnwEZZYdr*7gt!`!kq!o>{Yz z;Z&g;aZct!m~nA3AFvo(;G2wH7$MHd?1Ol&oXjzOf9v4~ikvjQH{O{!so6PcX$ARd z>BtjcR&{jHaKQdpCto66B>kP)q`$#)1-^VpXBW@fB@c5Pj>pyR+1YLu$qmV!|0a_& z>cw!!x^huGJiIcc3W@I9@L(lMeB_aZ7Pb(gFPgMQb}zM67uzZ;3koYM*irnf)zWas zZogf25Au%)@DB^~_6ZAxL2RGjx+zbR@R;5$NXtwu$jg8QRAv?!(ug@k+tH!r9Gw7f zs~42zZpLQwn{Qa;H({4(ypsL6V9%ZfbN26_1H`RtEq23o-iTra9Oq( zF^OzkwQA!pix>YQyH}Q07ZjDjda1Mo#>7U<935J1hv4t9pa^f@U|1&w1+r0nf9uAY zW((r)Zhk4IEh_e?`59TnFEkD@TnKa7eO`=WY&qv{Y)O0*@`3Tn9~ogwSYy}rto~!A z$>i2t=M%@?lvV@RMBuuL5uSvv?4C_~dTf0Q%SzZG;DfPe?k0B%9ga;%a3CAq z_cx9!&D0Zs>jdE1JyYjrrDOM%m79Au9if~vF>iQknHEMRi0BI~5p2z5GuwiE$=bBN zFK(==BK^kWnlmXceF)yY6Az!!&N^{o7Cn5?$&(k!(yl3S%;KT=UpAwv?6D_|!>^5F zcUxd#@d{>YslFapv#z?DZJj)MtLzP@By6$Z`c+$6r9b=&C1S4u~RG_)jlbJ&G?3 zZg?5pj0Qa%oncdp=BPDHWwB_f%_`|RanhJb(S4x^^#YmTGYB(8y+H@jqDvm@iXLqq zldrtz$!DHjbNk{5UE|3g-rWF+;92#$cXk%`AyuT4R7{p%{$zJivuWxz5B~k{e}DAW zOGf&4rGERmEKhMoLX%XLb#!m?u9Rar)y-1im|1c0@pGrdzznMoX`{MH%jV6SH7-=& z*;thNJ5*VR&sIpD0V5_|G*9*;K^=pIh>rjCf>DDIWQr04)zG3U!oTZz=JR_ z`9~aB^?@r!yCJ(u|2pDvGA|n*zIu9SfM?|%crkLzV#6z69n3AP@ScA+re&c=iU%i< zmFG-w0!0Y82%IonOE2_2e`&N@{6oJZgq)UmR7G=?1S3I+*YFUP$h8ltBGD?mB&D-U zeo;qlI~?|S0?>Gpt`H;<2dOdFaf)PtEs9NrZ4EN$b5L znyp6;?*8qB)etsocKqTg{h;Q}n>WI*Ivd@Sl3ndLVjfx1uH!htK?N^#3lUl^0S?kW z_S*`4EERH<(?~pcJJ~j9)PERG%o9-MfMOb^1^uv3!G9IZB2<>7q$H5Xf{Q{IyT#!b zpmIDU3aR(OoBT7jZ&w z?t;TIK-h4)w9uuFE=O>2%>fRY#^NE>Zh_@U(Z0;?x`=m(llyWVhg=J==01(9Be21^ zb+n-9Fsd;_fJTTKt$k`dpU)xDQeV^(WTb$M5|FFC9nAs`*S7*2JT z2}lQhdPay2^wC8^nTwID!^ky6V&Q{+EgXhejG0-8ycXMa3AqSH!!QTm;N*I6PIuG_ zz-|t3jLXU)ARTi-XfH^|TySOx$S6VGFoH}KEgdbWTRa3LETE`gnh6rxS-e+G6OcoV z%l~BN>YFVBry5k}Jk$wQD^?w<)zLH9PvH~s{J^6m5A|Trd^0!8fQ*v=A=G;iRzlGD^N!t#fJgl3+w9ZisD1*E^IvwR51dI3qDqBaAZ*_dbzUUuj2)s*>^=yRuX`T~$oT@&X9&m{0&))3=~$&5 zv>j@#b?hl<{Zj1^b=pLoO+(ZP_bcPGB(HVWUhcB?^PwGUX`lldbrZCs?g4fyf2{7iK=lWJIc4X%3#3|u9g}&wR?Gf? z?94UA7c4+eJUM#w$v+}Qt0$zUPUx+4xBy9d#;m=2XZ-@U)uf8+FZa5XtzFl0LGty0m0gIC-NW(ck2ub=Vyt#>o(I?3z_pG|Am2Llz@N|a zj(OsOMhxxj`|hoH?bzAFvf8n+8hEr2B<*>jp~QdiJ4O{ox`3o9Dw0pW$m})v`W7u_ zH5Gw4b{)gS)NsBGLQcp%_3?JDU7Luh5SYq+r=;!~jqv;%&x*XyhJ}mkYd++pAt{{^ zFNt&o!7kXz=MGp?#$zPy~b^;T+K=>$w|`8T!CM+srC!dWU^Nd99qQe3y47fRAM^QX&nx(q&|I9BuNxpj2m1@xawXl8fn zEzOLR)*)>Z2m`u-+G|i6Fg%gjq;I1KN(Q||bOQ#EONA92yG4veg^Gtf@Ll;tX zxvqn=Auo9^J#42-^?_|Ca_=q5Q9fR`~CnpUdPNBZzoMvl8R=}T_tu;7X#cUmm1d5Pc zSu2js!&nhAlC+DtM;F=iye_hPcISOY5-UPQLs+FcaK)G$0%{nZAgBxu6~&>jwdp(0 z=_7m3jl4COJ_*fCR?h1Z!@oDp`oi#)1Pnkh83cox&L3NSVFs^fDd@HYR9gVL#iP`C z?8rwb8V#~0;-%wPeqeXrnH{&|z?j1>+SR$8nuN@u4)3%iQ1&V3yaPNn$&W5h@w6l( z_5nZdBu@?ZSzVH=<2-I?Hu#a=e_g;3+=Fffmv*2_A+FJOsyEL$_f!8#uM&(?&gxl> z!3o_`R^QR{N1i`_#}-Y@sG`dmx?oJ0Stc${es-~o#8Y&r%K5Z=q&YF_yC>u=81Q*b zuklag*)jehT9VWSfpY|C3UGvUhR8`6^@gDRodY5YZy;qcEsJEs@vi%M(v|A*W8d>E zA696A09uPo?zn}+MjW?#_7UC6xr9n}^@lY^lHA+Hih99@F<<(mW*=CD#;REbUrT{2 z!MmaEz7lsUf|ai|OX)Cmxrg4w!Q$vgD#MN0=w?9fY(afz{Y1?kIk*ePko@U(+-MVP zR%uu&eWhnB`$6_?bt9tM;W1wpOSfniMK@2DNjHUs;5JaJ=`0(Ov|tF2#|qg2x{YE} z>E_0cSRjxS1qOl?dIJ8CSydn)r?cq<*@+O4xCOxOaWL2p_9Z=*#w>J;XOGiO2Sa1r zv}_`7uIuf9*%2NqMSMuyma+%wrW2nuu+U&(cjR@}q0m$U$Kn z*{Ae$J8PiZa`pt>JXjXpx)3gOuzK6!<$=da*;{m5?rbm5%Pz5HgTj`xe-Tg_T*Pp@ zoIORi2xg+2H{67K>OG$)$FN-dF6{zOUO{he0Z+a`fC0>lZbr5Qx4~%rZ+fhWCDAR8 zt)`m?)6)&!2)Gd}dDBeAc$f-)c6xaSBGo|=`3obIS#bsYovKRmVM<#d2Zo1+goIMe zhG4|)9}_(+I5-%7&!KYQ(gki6O-<*TyOe;aIX7IhB3jA)W#g3d*~rKu(;maTW9d}89gFgYACI!>eT2$fuGtFLWpXlg>ry}ZKG%KG~H znudm2IglI@f?T-4R zcT`dd0R&)W4JR+C$Z}!j`Bxdq5nvo{VNc1>Kh+~%qpb3w&y$4xjyp0tRrR~ zQ?-=vVx)>vS$#%_g5s6GviDFx7B5SBDoIv*X;GTwUo5MA+)$o0Dc^zDH7M~Glz2Z% zr09nfw`An*l>PhmA2{^e&TTt(rtIIpYgbCj!6S!%`(fKIc&=rhJ$?MxzJrI3qt@Fj zu~qPBE2}{lh|E`|B_$hLuj#?z4<0k+-psGSX$+FNP)Dz3Qe|wROkt{T(pl(;hCfm+gBOp|9xA>z_a6z(A-dBjZeSNRVeP zzF*BYv(1J_u9^blq$$@vMxLT^>^0rekch~Ue@0x_ee|+GP1W$;F~S%+Fc5xGfj(~FK`?f0%H9IS)4#B=vbq)yezoOgh4QG}%=EOhGl+j+ zD=J9OL%ekaK{%C~UtW@zvU}&xzaBVrF!kh#{fHwPJ7MC;;lo2ljT$v96h3%9Q>RUv zX3WcN2^urWYHet2gy~pgZLQVOv*eRtLeH9dttzUj&(Uj`D%hDgWgLPdoW|RBkOtul zFnZM!m$KV&3!TIlOd87rmQwMlqm@T>5Pw6h@V&HNS zaJdS&IPK0tl9TtO?Ay10|Ni9NyOV$UY1_7)yAK|KW9B{-{)8pC6}~J8V&ADKo#3hR zx^o!1ZHSc6CWn6T#mRPhPn<~k z^wUp|H~24Fw8+bvyk(ah&t792OgBy$J8ArgVXZ9ma^yC71t!=@hO1``uWxSCB=7;F zE7`ei;|JR}CBP;ATIE{J`kgy>zJ4i8b8l2G;(trvb^ny~MVcJi($d`2*x1m}fJikh zozNcJTkGl(Vzm)rqJt?3D=bA}2dD`R4Gr=(!Sh>>I2L+2G%YRl^yyP4Po7Lmx8-NS zTt748x1n1@^;9_E8Vnag^gbLG*r(VnfM9X;4a!GGmrljmiPTsu>BX!^Y zltagl9Xq`5x8HvK?bqF)<&y#xgp7n=7@-pAP07L?s4Tq5s{~&1#U(X$Zh_OTx$?50 zx+9-{^5H>$mS!>D(i{oY|ovOzJ-y6x`2 z-1Xq26d~hP!tOG`6>Ow0FP>RxksCbHdJ#n4zMsxrZ4} zpGK~jbYwR@eQM94WBU&rIDR5MEj{gQI{Tc%wKNMjxc29il%cN>g`=TO(30YXUJY6< z0xd7b>~}F{ztHa~!OV`G|K74?%a{M&w(Z~l-nwd2>xPgL3 zPd`wTeE-M&>H3Sq+;y_yx#WP0Mgg;R*MVlL{L0t;Wk+$WF z1I114a#%$vmhFOq3S_sfEvu}pswyo-{+p`irb?Dgko6T6#h@wXdXU!vxwim29U_YyP1D*BAqH~WjMRBuR?a2uLbJ*U7++HJ*$Z9I`M9fy^AvE@U zB~FR8*VqqUNFUpeTxcJw(Z~M*H{2k3u)XXJaKd|R6FbW)B{yl3G**gXEr?0pwv^hm#j3>F<7248ReAne*gwYg$f*S`$5 zf9vG4EA8J(F8V_))BReI_vSv~zqEUZ#ckgSfGP5ZeiQykdQzL3^>50q=dID^#< zGpqCT>?ey^nja{e)eS!E9gBs<;5^mC7dW292LryhyHS>2mV>uGcIJ+^iyUQ|1}M`4 z*|WRIeLY2b_bzfmPmyuXBKd4!j1)3@Z#kKtxCZ`&XiRiY{Eh`e$#{+)BnmxH!>t`? zIdpX{OE8EMj*Aj*a+VM%O4xS+C3uPwOriwQ9#rlrN~rHCflh&Spf#O7>p*L#Esp*t z2?-E<-oaR8z929;-MQ6DIZU#2=MT+r!Nu2g~iyN7yC3y&KIyerJr$%Of^TEaX-MVNQ&AtdE;;k>Q>cr*Op1ZR&U3pK&n*9WpVyj|qrY?ri{a)P(T zfZ~AtE)>NZ!3`I|5ze-_gdZc|7YKMl1@I07GPfRhSG0&eL9|$$L$ugDmS~|@h!%ly zPFlFEIeI(Cd;x2uDmRInffdjvJ*gKQQmA7Dp4W@vg&cm2;DW6}QC1@qWu?@oL%;{@ zqWbMDcu05;0e_Kz|3Sca3HWn8@UAGQ+ewt0Uq_Vt!ILQ0dWa~e z3LcVjeu8q1$Que=bGfof)hg6S?ri3em}>=mmw+JGFdr2DN>nQtdxSQ6H!>iIBPwXCFA}*mnl>4u*2uvAP#e?VbUD{0YNF%z(xcwh?b> zZ69C3tav-*3^-EE1LwpHc7jG85p*NnN>GHOEiP#hEZ}PeJRkXHRil9a?gHS43Hb2> zo|=M+D+PRR54+V30cA+`3m~eA`bZuhur3bblK)0 zfpfFK*(7kT5I9%&;Or9K{V?aB$lYv)MzX#Xh+M!(Za>1?5+vI4syIYAE)EeMr$dAxW&$4pKSIDC6!5JAKBNcU6&Ls( zAzH+6S{$8Dw5Z!gva9SWh16nHPv>2*f(Bdc|AZSrXNC;Yl6CxKt3!_*WB3KIb zFVB%y5vZYZ+I_Mvtv}(*W%Pk zg20bY9(2TTelKXT2+%nY-%~0AzD2-i6MUk8KHCHBst*j`P#-iNBfP&&CA>Sb3Gd)3 z&OV^*T>a1b64^JPDb^IM-5l%R1lC=m-lm>J~)pOEuJrM zzC%!Ko51;CPn(@J_m_Coy(FGMhs5K~C7znv{&k-x^Z9P$Ja!CWT_E#~g2cv89F1ZP zub^9BE<3LTqbNZmO7L)&;3G;X>?y$|{@aLh#uq_3=7IW&cJ4JqyPBFI@%Iz*;vs%y z3CD{#XaF!gyhvm+tExyG8UlZlC_xb=P*f9MLZB$YdI2RkXjs60_%CS~BuerIx zXb7p7)6j#{uy#Ni_8D=G9*kkf3IBRfU@grQeC1L|dJ+_L)qifh|0W-&x*GvY{dbP9 zUAnRT`|mu7XG{_2zo|mv`Ar<}7trxO&9&VUk2@r${^zJP-r0#bWipi@(QyTx!6K+f z9D9zy+Yp}xH-Co15YYmcrR;H*LZ@lGl$p*_T16>&1DDd9j_jbbln!Sp)uNO{QHH6f z3}QxT8{B%nd`tVUb-+IBckZog`puhJY;*^?HhWPYKvj0yeB0*2&-f zSMbUYXnvZ!3;iJ-i|4ab81=@Ssm|UYNT)u@&`YAJ{}75uXpRENgaY@ebT*IsXsTKY zm?MqtJF@!M-1s5Y9G*-yA5Nf}qeRVfoHY+j=W--o1xd$o(-+l#HlIHEbCVOOA?D!| z(b%HzpL?&;FeEF zYQ{>CCTHM+o>#0j~&nT@Sn~s&#%w+&-_Ks8;+0QBB47 zUfNJ6)%v$u*h(ovu80!KntWk-B+;kDA?%Ly0&)uQi3Ye|K}5iPFHvC~)=9N=e6;0!&T+0@WRJ!T#Ky3b#482V8pj4e#m3<)>i7O_ zq0)zapHsaWEws06P)5pn+=@07 z2YNM$eSAKO!6zMxz|9}vgU$Pu0=7oLdJ5P80h>1jY`uUTCtwMEROlgKP2I4rxUcz5 zqTZA=qTc>#)MthYqF(SNPU^XYpXsS~Cm87#w#4suqK^_*qo%7^M4vf=~@1>D4m4T_onI zfhllA*ndU}`_D#U|Jf<*KO;$K=QGB@u;~IeRKRW#u(bmAuS39A3)qnYHebMo2w2Z< zSXb0*_!m)c{4Yd3>r=!lJ|?1GP`7?#VCpqcwSs!%scJzzD^)M3=R+t6>ID%R-PAKi z_DF8d^@eL|F3triW@#H_r7J)!331uU2sTlezJC@r9!yGv&q5^1hX6k8pMW=e-@ z%}yJgLHfQy%&|rRIl0d!=}kTHD^IwO*p$paqKPX)sB4XRZaJ5?rd%EwU?$eZcly{#~8^+%z{=w37mIyx7St8P4CbMo62i$ zK#~Be*|U*qR(zcNMAn?WHK%U0`XcTJ1)0UH+QnCXEcpDB9a*X{JZ6Ep$=sV(;DO-_ zobb~HyiLGs1^mQb@U)9`uHbHvR}F}fP0AE25FqR?roDrdvj({14DJ+-Ynd%X{b)en zU1VR{>0n<7{4=;YKb-L?wktkIBT+eSReXvk{0FAIT|j#Y=n5zF7y)e^0=k}Cj9}Hz zMqS+gha2?=h%@+X=7z6b*Ih24#|h|m)Xky81@!4|Xji>n#e2Po_xjBaeyiiXZg+Fl z>wIKNR`lAvM4m7Np2jVgEvFl$9l*_m@XrG8-hI~V1Q;43Q!g!bsS*@xPrYxx=R)K93b8P8H%+xY5l8tH87 zzLckXactyb5d*EI50SA{ZO5(IwO|+2o)bFMQCoQz- zgWl^F$;*kbQ^i`vy=4^IdnMBTK9MY)`izghnv@2DBF7XCH&N`_6d@TL?uK=Rsg_Ij z1}+&$e9zX#B?Gr`C#KFBp>M5?qE?fr)mzlsDQYe4u61D8`av*0G6bxnwGN*mM{B(< zw6#Kyu|UW{ytP6OQgyf1NvHm+R_8$)$qQ1^qpaAmc1Wevr>e>QDb;u84Hd1uic>0% zZ8r;A{UmVyxx2kCdGWusx%zMIcVY^-?kNoEB*14|5jhCA5@GpdV!Gb@IAf%cJxON? z{Rbb4o?8IU%o5p#UuRhx&t8?@hr$E$9ty(!Ap6R}-!3UIGMi(dhfOA`?qK)P?U;~r zikZ!cflGK_VSf@N>`y)x_9tz^{$xy#{fSHX|JLSXd;`bdF+^?LZBCWPUZ!R!thqn_ zz4?ce(0?n3PesUfET9Xp1o9Qgz}H>)xxp*0XFopReCW4>mryS}HVZk^L8rj2&-A1c z6Yw6GmB@Rlw-1(Cwa?;mo6=96l#%?3XPF6;@_96UL4X_NxV|BFaOB8$8R&kVP=+@+&r$xlN z0P>RJ1myK6%?a*pq&mU9f^;Xi8%TMwuJ8V@?|j`MC{}`u06?LbOf6|A6F|2JHV3!f zJFvd6WYLVrn#7D%!>*@K4g)|#w?36HC*C2?J1ls#a|PyE9gywUfo)bu?||Z=^q##> z4lk0CFrTW2=`QYHpJiv9EueAOyWhH3b?FV_V(C{DL{oqSi&6FxA*22OSbGn^rpmT` z{G5~BjHYS2cj>0=y_=Q|A_yWbhKPt;UlkY9lT#ECMNv^u6p)SVA$yd)N-3o~ZPLA) z?j}ug{@0VV$m^@G-}nE1Nt3Re@jUna+~c~(Lxeyw<3}Kh|27L^YBTvZ)e7 zx|WgbK{f0DJKfywlKs^W<82Lc*|23!8x$bqC>t3(8?p5Hr*U=B|^{BQUDww&S4pn9n&!H0+k@+NSBZAl z>=7=bJHQaA63yTkA5tYseaX=kV2%AKICzm`;Vz<&4O}-=iJDz8RdpAKc_F^0aCun* zSdM7DaqY?x0($^?qRr+MX3&?0Wb8?o3>VQ)M_r}%f_wL~$vTvm{E&jK^(e4xb2LH; zGMUEx$+Q}Zj;x~|SPM~D3uFhUFsuc#gOmHUt5;IbpTBq!mpFX*1gSg2Rmi@n4Q9On zC2@@_)V`|D#+rtv`oSS5#Kx_JM7&aI}luCbw^*4)}^X9d<4FhMdZvxFXM zy+ABhyRT4Z<>X$!el<1y-hJ#Ho0U!NV@2w~z`&lqdJ*Q$5Dl;%=-4zrMDl>>{ zD!@Q!4@yA$lKV}dRwFc0<-8wE-M=_gqL6D+HvH6ylP6PUoNJbd#30&A1u77~N$#>3 zx=%>^6yjuvNjT4j9c0MaXx@wn3`I^w(wWNss2U+2O}Fe(5ZQyhG#u3^6`4omi{T{P z#!o=F{q)@A1V%~z)>w>uo==jg{Ruq&#^Z0U`y#udlKu+8pq^JKjU7lCN_zIj@jW-L z9Xk%Py~|=h8^J*e3?dsearc3Q*!2gHi-MEU_YmU%p;BlKqoWF`o3w9#9c84hU|I|; zJXQfhcS}>mK&r^b0iudZYkga(0G)rutju`RP?DZekcUct>Ty=^IMuJ83+0@F`*{NQ zLn@GGV4h~+exiE}?9+ef&_{<28Z;=zr>!8*P+3+(`}+5YkM9!`5)cv<7dU7@SPvLC z5qAdqcQ$u=_+ZBfJnC+rJ9oPd41f|!Z)s;shbrO@8Hp-(XuMXY9y?~VuAj2`<~kjA zCs^+7F72o^-8gpm=!vVpZQrr)@YVr=tA9ld)6+6bWvXoP3-<2y#G7xvIl)sPXl>KW zCEB3jkB{gZHF%b#Ya&!?tp`<}m7jg@cB!WS__uT@>f0_dH!6ordvP-LGfK72Qm3gc zNRQq|?V}E$tnVQ07NMgusNc{+bRv+MBblms%Wo2XiTGiy7TsU0q=fKtFTZZfCqg z1H9#O5q2fCpoZPQcTZ)r*-=%~+6>x945=x&`3J$;mSS5jjzwIn^Kf(cSf(UCR-yy=V6{Aze^u$^E z<19(widb+(PjH3jv2&;vh6p=#>iF@K$Bv%3d5uW3`K(?*if;8%a|3S5!vlLUYcFLx z1`P;Vv%MY_VflAXo_S=@HnK&GfR?tKolLA3n(yDfc2!5MrdHDq2dASpQELQ4JH`Gw zh@E{x-NO0LzMTll9RP?|MdM~cv7cGt{{8!lqn#3n6~zUKAd+<0cK%wb~JSl`^*Xe!Q0%P2thuh=D! z)Ox9JPl%|pvewqF*>GYSp?`QEh~^U0*(n2vw_%`865j)>pZw;ugJDVR$K=d~Mc}ko zF`vYKdARZ{8Bhu#3(c!JmwclbqF$VcAOtSenZUo+CE>794M$ADeJm}m>AEW+&iSXV zP82nsOF=iQ6lxb<61-yYa^At;embn{(~5^bO{8bHmDFu-(c>p^q&F7r83$7xl9=AT zme}l&n_?0EW8>)P9{B#S1peD|gbn@kITZ50IfolY`ET5e`yg{h?t=nne8JJXzUExB z_25ooB5+sTGexkW|LN>AJkjo${Z^viUHXM~SV18>nSHsrYxap2j>etEVTFWagGE+Mb8m-e0jzf(e%77Fn#iJ#{ztewY+CJZ}z@y9Q8ybM5C#yuUMgpRPpca{5H zyXHq;r*`Z(#nGmAWPmYmE_`1HL$sOhJ3t3%_6?@p#LDyL&(9+-COn+sl%<8mmE@(6 z_wb>zSZU$$Dl~sNXS?h3_TMQszzP_*}zbFb?Nmz-Ye8i&L=CX9Lnr@t%#BGRg#$p61u-Rslcbqtd#bn{_66?}~ z<8jP*%)ldY9v{46XH9vd!%?4mXiulbQr*;0UR-JM z>yZuFbB+y($?(VQqXXLdp@lRzbVy%g)E z_6^c1?HvbB@;)&G2lVbUX6j@es?%376X2kR6o!Wvl9y8H=@%Rv9IjRmNlHpeKuPcr zU1DMnB^UbBx(%B)ZTNZXH*3HCdhOS1)~s3mC4N}@4Zgnl=KFQufQT5HXI-5qd7U|P z_UyTHq`yHb9v4zCT)2SmXU=e8SFYKt_2!6hnN%W?N?-vnLi`|=;Y%h{DrBg?2+0Ty z%^)w02JJrr12Kc)y?XVEjfsqmjERpA)VOGdgt-U%)N3*STFk$Mtfkwx&Yt5QtRzlC z7SNE`ZHGBY_ln^}ruKK{$-{KiFW7!U(!D%530W4E#pSh}qzhJLvBY`YUjKlCdMGtjNtH*QSIVEw#cpqtMKST$ZD&c(PgG@0@^KjuRB-6cpnf z>kbAgtV$G}e!Iyfc^Qlh88dkO{VqhX?VS=z-F0<$e44T*@~=6H9kX#>JUD@(^3ogVI623-POT%S<`{BbhHot4u%jGgy9GPtJ;^cTl&*38z4i^yCG z3(pO2OUYXvc?;-Tx7|mO7>9`JUhBV;k8pTPfF+L~Z<4p3Y?m$!##MrGm1rD88u&ku z-GtY6?cDGqa>Bbf&LzJ&WjEJF`+fH-aIOc^+Q|%}+V*!|@^~rN-{S| zM4ZSq_=zMLgAs1TniaUZqv@y%^d45w-a%@b^aOcwQH% z1Xba%u}HTwfu)}M2tUCtgrP$UVQEU>D5u8ZZ}Ki$5gXypmL?-aP81T6k#_^%7&@4y z2nCSN0{{zM8Iai~2OX%fUFRT{6`}XPQI8l!AeX^OzWwCKM)HQAl8G%P!ZH6_>E3;> z1bOeaXK=6Ey6&}!+$+N%aO%3#Jm;OdA?W+yPLZ7La}!64ymi#C)LJ^XK}VgTc7RWQ zWl_pSS4ni#7kHx2jFiUSe~ef`Nm&$83;BT7)&NdMGcqYHa91KI>HMorDz&-zg79+; zW-%W37K(d|!o4Y4T3YHFn%eE14fRG7r?{MU`SR`T?Aw>moH%iYLnkG{dF}cQxI>|@ z)uKdqjSrd6s12+_TcXwasMS7N;o-wVZXDX8{6zgo52OcCH-&Qr%yVz(Xc5(hrU(hF zLK&e{dbG89D790kYB`ls>gK5l>Ok(2#v2=lvI#8{_eWy;zbQSo zxu$mQy2Gx9E(#mE$#j&Jb=XP+155vgFHq$V*5v>av$)9=mRuezAtXOydjGVpBPiFp z4#QE(mgchZ^g9<)Gc#}B!Ag;L>+8+G{IZ!-j_J4eRG+YodqR6{I)w^Cd2CKjEO}v% zN178AG`;bZIFPuNDMCwbk6~bD7R$Cvy?QW=2dAvBtZizlHMiAz`M05NRj#S6)sPpK z4A05RDgykJ?n$}<*g)_s^B95z@@{?T-mjm#i++SV|GWwib30E!ad09TB(96`7Lm8e zu5}R8MfJWIZ49W6UBD>K_3PJDFI~Nrk>1=?T6QPx`~{pv{mF_?KKXLx$}d0Z9y_P` z(a`VM@9PHnc#rP;+sWRcKkde4)O;GxV<%mP@16zDZOQ=k;uoj-v@~0>*X)WMe3U77C=qtu+?mV!K(IGqXAQJu#bQD+;E!V z{rmUnJ4kescE|JtS%VX?`}5sWFi^N!R>zOL?+l zA4vK8S$ERAIf8Hl;erRVq6``K+sST@AlyK>;K5wT9I82E9^weh0hoa>!Gqx|Yp&f! zTd}(4=DNmql>f^K6FeBSa>#wJ+1(65IDv4$gP}TQ+y9soj1O}{_ejGZ9_hWztosi} zx{Z}@*}8ShA0yp+#!K|5?#xXxTBug}VUg(YA0wSSLK*n*NTGM-k9c}M^2b2QCcwnz zDK0RULm|uUot^eS20EaB-+m7U8oA~}m$b{$rQ`x(YBIw4S zWj{?KNUu7NBMD(xEc|O^cHSayGrMI&D2^l^PSitlq&Rr!$l=SEj{UxQ(^_KkzRJqa zUAS8)G0wvrb|Z}}z(ZqVE4 zmGq{#A$P@(e#EtZjg^!3y**!ly(f8b%3(u?4vPtoijR+q>xROy_kCBD7ZZ22teUOhYlS+d>Hu2{sa4A+Tr_=BSaTC^+8xmgHF9St>&OnAI8zij5X3wM$R9XyAsxDa389(_lQ7%?CqK!ApWg?G^`F2iK%^c5XB zA|fv?NYGH-)Lvhaow4fM+(rfpLFN|~@#^zo*>x{JPWvQK-!aQkk{fd4#{JTUhK8!N zH1*)YgL{RbMPJ;&ft@$pXU=pdFD}g7P+#9rUt3#SRf8WI%=j{!Tbj*znK^}o?Xa13 zh(e)M!bbuG1ZND}Lj%6ScZH%YJ-v;*xG)X==Ap)C=?4*o7!mKxGDFbIll*qy74G&RWp7R?@aJQRFG3<|#Sb0Z>?)b7AfY#>jTZb* z9dUhK9dXOSnJ?l*K-9ZB;?}tWTz&r`!WiVLiT88HhpB81cU}ot8sGZ-Lq}XsS4Z4% z$UJ=7>I#uJxkBX8&hn(cj}JzMj#aqI(T%Q> zhf$;){V!*mLu1^D>oS-7jo>-2f4}Q*ls+KcrUK;PJ*2e zW+Y+%Y2=5&;7VstT#GB2xBox%#0`L7fXfJ}6!Oj2l}yzLBLKg}JM+^Cox*YLa7cfm zAqiR+4jQY__4Cf{+vozBaq+^1tG78-3>rTA`J+o8I?Cc3r-}{5j`2dVU~Hd#=gznK za;i5+jZU1h5J8E?#)V7Xdy5n171!B$RwB(?{d&B&?Ra1ON=c>lnG{#08Vr>cMm zgB+`CY($U0wgygBQc+RS)QJAHRaK4n#;M9Gstbz@x!6a&p%Jb8IMqFa!BAOn?b@~M z?5u)&_a3r|AJ%=uCWfwquY^>zV~{-rwuA&S;}2Q_ZaKM|cvyzY7MCEx~V1OO=^f(Xug(4jNzm+0p9K9zu5(UCrcz%)W}KeQDHtcN?!iO6Z!eDuJ1 zqd-&A=Qs>B4a0o6A3t&O?D_L&&Ky5^>-O!d7r<5Tu+EUA8X-w&2=xl}q0@JAi}G^{ zzTGn-I1HZ0Nu^wcdGsf!eKu=Mt%d)Mjd>aM6rMCJ!bK+h>CO2J55*Y7@-9 z18lNn%a$#>)~?vUUx!wOKJ?1lE7aI{9_F5q>g;SbqeG@M77FA9kr9E_YBsM>*Voo` zTJ20rV?$YKNkbh-lq`XlR?&gm(o^lOxqdhX-R-G%1q!eU2>ARM=-d^9J^=E+2s{Os zB0*>moDo52m!%Ym+3AJZ8G+e}=$@SusbH#W$4;C*d+OAg6C~XDHcMuO)=ThC@d|~4 zQ(wzxh7Ji|ySu5rrlK(8EVKsc*t>>WN3EbfW)lgGs4s;dFMN9T<3M&-3YUhK)YFK5 zQ0qjoQznjkX1TMWEtX{xnHoGNQ)^trq}Q%ULSxBy$dglxY>1Av+B+#BadsuJ`Gn{> zD|2Hd_~s@jP?p-<&{$bsZmu&s1Y)|e3AnAElIkTe2BZPxW?f_bTmJmm(S$2M#jC)d zU&CbtcSlg(1P*};&g$asL71t*n5oH_sX>@2?6yhymz5P73-Zts7pykg3$IKW zx)Vl%Ubb}UmMy>TO}&>>&&odgY}Hr$?ont9$_e-E9UlB*0<}SUKF57PzuxraT_lMm zb>~?ABjD{%^RxM7f*s3|$$~ij0)Bzyz4zXG@sSa+?jXf_@)UpY=%}FTsw&RkXsjr! zsxg(6m6w;57>zY1(iPNTFjm7f(Mv5Bbi@^+U#eavkw`?qb4DE^9Pm;)MIwj8LSwf$G_$o>IL)42X-#&UrR2m9 zV_ymPI5hf+y}u-(>*-_sWPY9?=TZ`?=_=3nH&!E8`a5MwyJTF-Ji zL>5@_qQ34L?U)asx&-B^MhmL%n9AGB##1|}Q`A|s0=)%+=z;ra zoRM}o=?wN#q(Dso?wmhD6B@PJ-9ziGQmEX$RZ1^kG`#ln^M=Lnw?{dt_zI-{)DwoS8wx+gD2n2yZAW-rm8nz2ZY)-pkvHG7fzAwQp#+ygqx+=PS7%uNo8hWu9L(7av}| zGG+CS^ZC-q$Zvkwc@eUNRoGkW>YMG*q$-WKLLbJ{0zv3(w6%Yxe;{MHapT6AF=OJz zHTQR*qwnq_n~kPZJVyGKr={P?YLH4AuU@&{5Hb3BA`&;C1^VSnCS_Q{Yf029%*c1t zC!z)87dWEzZxO`Ndq z_K6ceCm}!LRq|A&97wAV=-mFvsD6#&~jz#)^jyNZm`vcmYRHLeR)N@ zNG7Vw&Mhb`tt~O!$tZ8IiA2KACUaGlxuy}VqwSro%`t|G`5ZLe9mi1e%H+YSk``qIBBQy=izx!-$bfB=Q(cV^?ot{NNXK(9v7m)2>|4X|RiXP8>IGc$5+jCMy$q z#mt+`e8eu~;@2C;3@=36knb#5hT9(?t4#$bVPXQsJqlnLA;NtNeATT zUF8oqV92)%WMY|8i`-YRwrQJQumAo^RYz#jai6&4XD%s~q| zswfeW@#EANf(fBbzb5jp18jOu_*FwcQ;UwD#m^G_o-ut0iSB-_T>ji+vmc+(-`}IX zARF!#Gj801B&dxo?zU8cNUMIgs{ZY0{lo=t?Kv@A4|+T-_y~!3AK<5f85~=s!0UyG4fz zs7atdc9r+>rZH6PZ$-xH95Q|&=k(6Q)+Z$!@{t@q_H12zH{$e7oBaw@Vwo|9XG1 zJ!-Kth?DH3zYvK{4gq7%%PlG^FI>ce`rrYTVe;g}z{aA~AHW>jOd@odP8>T(+u3Be zc|P^r_7ysGLw{fJ>*=AvI`sJGgrTzI))iv05rJ<&w zxwW;esiCH#!XUylFrd21Vxgg-vC%n*DMAb^*x%3B%f~l3Ff7nFFf=?oB0SJns~z*7 zpBFhl@1oKIQ0Yai(iE)H1(3=MuuA(Fi}Le}N=iWxm~Oen#W1=H^PyG>a`Vb7OAPr1 zB>YVbaZ^c2aRopIHQi}}8bZn;ws^o2g-I)=g)%YVdNIOqA}MOroL&SzWD{`mrlL~W zz*)1N2(@OdA@1rEY>1`Wka>LT?tSMntKu>*@A+lxZB<+ddieY>XUt@j`mvHFix(C*7alV zrCo3HAN@W_Xxk=kYZr(1d*Xw!PSKtMH*ydC>Q4#$8}#`xNz_U_zoc0Ca|6741<(zq z+*MVvReE}9fw5|UwyNFXM#!kh|Lp~wtGU~^__taC`giyiPAn~lv?xku zdQ;DlZ#P`9*I??UF3+O-XcAi>NRzA1HWf#|Pd;xYAgtbn2Hrzb7f`3iF!7&s9Zl?w z?kIV;y`d+-p8)KHBxLFqCyz%2NP}?&`4)y0Tzt!Q zHT>{%R(=v0_8hL9g4MAO_x&91`#G$R0ol2^c?AX|dE`J;H&2r8y5|1s+vqtKL&Avh3PsDcS`Yx8tNo9cVi z6=HFPp+-+Dg7nxvgMOkY1=%drhZHAS`be~<$yi*F?GLYp8N=Y5PjfB*9v=AnzH3yE5)>0c|Dr8>vfI}$9n4flmS&0 zmF1=Ir17Y%tgfys1t4ImD2MhaE-oprsI01lt5#$x#!54rE6a;a)o7$6kfAtM?*~PL z*=K|>C`CdU-QI4)emYoT4xVK|l~0+LhR&1$e#Gncgmr^tcwc`%AFUQ^$H!acuGVNg zv1ySXQsmv;J&1cA8tUuq?%_}M{a;<-KM|6Ls)*^r-x1BHJuZ^>;>YM;&^5<{?nG^` z0`Jb1I08I513WngJUIe9IRZQxR8U@3wEyI(6Tf5Epi?K09^JG1;L%e^4?A`2&;jae zmX?Z~;SJz2lo7slE9BlPT+Mc9bL{!NXprj+!|Yo+h?DBw)tv)jW2>x9J8deb~5}a z6EtKR8Xg7Feki7FhtK^ZJ%4@@IAc1sm-DNCS~C65e^ zR;47lpcE=uJbQGc15gha)Y(|yf!%)sf)Kd&(1e7<#s&Czf$#l-!y~XMJ>Ajil!_v_ zQFMwM@neKA7YU>iq-bfU?QXU@;4rP;16o?#RD5%JQK6xrq`^Y9<4`#(f!emarUwxx zt<>6F4bQ*J7D~XJ&%&HPi8)WeoF`z;!wgvWL&VlOA2LV9C>u&z>=2 zNVre?o+I0~Unzh%6xi%;N{+gl$S>!U`Q?Hv>gBhRAeUC)rDv82g?iD*mno^(&WXl^ z&=zY`B2?XO>Zs_|3-A2WJ4TZdR)U2_)?8(Esi~;QP-rN^^c6wSl~$HlVD~S8 zAy`F#gOo^462pD$w8L$*SW>VP3B#-i^TaYDE6Uej>*4OMR4Luv@j?(#>q~GhFDmN4 z9`EG*Bv4@?sPG)9kOV3ufePV-0>(V-Zbwp=ydY^SZ4c}b zUtK{G+n-^ik=}_YOL&cc11sSJ6xrBpR=vATpq4YV)X(qdpMO4FrW%_-CDY57v%+SY zZa`jLeLcxp*}&Al^u~Dji+f=^?xyy@emw2=LiEVDCVO>SS`?4#s86Y%nEHHkBDD)O zN9(XIQdlq$<s8Oj02&B6+aT%p~&Y-{NlhlH&ksBI8*$bb~QFd`h{<5!lNfblhz zu}oQ?Hxh|K3x=5+$-<=P=vMOQ0nm`HKR?@p>uoYsRS`oGG6u?y5WA{^Sdq@Zfzinq zK_#om-${ELFM`*&=Z3T$`^kolQ>A|K4t4j?AjvML15`UfZpJ4e=#+|g^_#rr0kpT%e5 zE8_@8jORhxn|_`aBibSoSOPpFN9s5oewEVC%pHrOq#CN0xEQ~GhZ z!3C~Z3;%&?;|*hp-Da=8b?Umg-_%}Q4=i6JESSo2;&v4pVa=37T~$^iI0?E|*VI&j ztegjr`4RBIM{H#6TC51lNE|@wfRr*^4=i9c*;&!s6Xw234THrSD$>{Ahiua6=Z{U# zK+AtPiIbLbpyfEwawKRO2U^B~mOh5E(y~GW+!&Z(hWxyILjj>KbsCVC9p!~$g$GcD z6v_@0HCAvlH4}Y!mowZq>n1z5Hd_c}Y$pFE%#0`)8PoXbOnp5TqBkpcI8MYZ`*@iS zlJOe`__Lnr$UXY0V)2=4M?9WLH1E7&@SI2Rmamv)ufMIMZc(?G&`_+Rl`Jbvy2GV# zNxZk?dqfC6q;BdNagU@=eM<5R@{3F0zQR&R_@%6*l*})R3s{-J4iVDQ0B{LT@25VV za)n&u>8@~7DAk@`-d>(sGKo~ge|*w8FRL&wyD%?XF)ynyFRL&wW8l*j6Hgh62gWCH zp`aXSjqM4H?H2wE@|cDjz(j4+KthpN6Sl(B;k1H z)9rAfc=U-ACrf=IXCzRZ|3^}~Qywv5WN$o%ju{-`*4#?H_4b4q-;$G@=k#f>ytDZE z1@mSlP3RZXr+2T2{v*Z^EWMB_r|PM#J)W7PKKk3QZF48}syeW9({0KpVeb5S3txYB z@RZ&`p-2menfT^YW8ti4QUz2Qb%I_BcKC#qj2ky@NNjYh7pL&;6W?q2I3&kg`z~@h zI*R2Bdw1rcmSa|A<9WDP?73LG^3_B3+eWkFf%!CvjY%D2| zxXEQgE&|>Mg3_=w0cyd6mq>s}!A=zm5l**K0K$mDs#ie{5?G9~!b)dQ$~6Z5yNWpX zP{$q~@bf)AA-8>e{UNQ@3O6^U#={eP?@HZO?r>9(G^JI0sN_VhqfQ1mfUnTc%YzF4 ze{d(xH837)U=G&6Y^;IqthXM<{DLB&LEvhb)v&z?qZ{&J2V(^qOz??|U=OPa;0RTX zX@J)&gm_;e0oF8a6m2`y>5A8=~$SwU%(i~Q#dm^IzpqD#m;)d(&RsJ;>1*zO63&wkM}U1 zJ9GZzl`@ZBQ%Q~E&#I;ddB0KP7Jjhw?KhuJ7?i6+G3r)93*u(1Z0_87z09V@v|l*v zu8ZeV;ZN7Emv$md%QTwq-Gw$oUEy4=hq0^_J0-)W0r^31BjJLQGMK{|#a3+q6MaufTw?ZUGaLI+h-4FRcd&@DkF2R%${)Pwc7;>&w&A9sZ^VjN^Yg zi_7xz0%iO`nGjIM3(|?yGRpxzW#t&Mjiv&a!pT6+bp3o z=FEr}l$9->J9UUJJ;meq6)RTkuj~}{n7~b;3hCGBNpv4P2he?Cg=avcU8Ob%mrR-U z#N6rBdAb)ppB_es)8ptV_*n)gS`HVw1Hvv3!RvL*vUgq~u_gG;G{o}8*3xQuxRTdv z$}(y~Vjf-mWdDHgf4x~zbfPTcbpp-cDO#%Y%KT;(7gJu>P*;u3WC@SLqca(c%WGY6RsWhjXz{c~%arkpjH(^YNd&ylen@$chx%d8!khjT>+4>}bOSYLrz% z)Y_n3frY@`w{oJ?k@})rVD6>i8-$}hq9)9qIb(pteD~PFBfn?yR%O456DPzf?Rgi@ zpUuZ<6CBknB}#DAB_{$J8OHZY7BPaxJI8() z=TS45b?Vp-T!N#gX~)8F+)GUb(Z5i<6cN}Y*LF<^~UAo^Tg35y6@90TAeCAle8=x9udSs?e6 z2oA9u4aD9J52zj@4AhBU25xmBt-RjX!8b9rNI1 z#9)it;dqXO6ZwHy|HF;SZClo&5=XKjGcTvQqw?1TevbU^xlom<+?;#z*mkV@MMQLR z^elgUxL2=%ckroBk-Ju_RSH(F`sv4?ZZ}soE8`Or6YtJ?>+v zR6+hPOrp}LR+^%6k&m*I`VEY+iCQUoZ&A8F23?3ftgMU$D+2o$cbpvt45}co8Z^6g~C?Q4>D}tRky^UPJQ-tls z?Y0J2gqi3TWP{@Mhi+k!ex1U?n8urqyrC?SMaO;D5k#gC?3Cmn5Sc=HYs6fwZ_T}# zn_rfHH}guKy(9h3)$^&BuU*TzdFAHq^h>FCvk;-albug(X0ZlJ?l(8H!KtaaxviDR z+`E3SxP?IzgYxWRbJ^Xyce9I1Dw}Ow%(=T)_WgEjYfQ}6wJSHKSKPaKbnEt=2T$cz z$Yc8qige=}Z|vT_dG9r7J60g*{qoCP;CJ8MqcHwfvqLGTzW8Fd!4dYvJMX+R!=EqR z{{=T5&g#!#kYDh=o!UxYyEY!*e+5Xo86eU=>RXTZUi;T;FE4(*KXv?zPxas2-TitZXgxM&3C4wpf@b1iS5q_6L$NwvV z+Q`K?veRp=wbjPzmPVeG(X;`;r0n)OD{t*+r3EbwO&#rR;Ef_yWU~eJR4Uov(W9fI z<-S9Q4js}fSmCeVQ6d`ox9>`6`86jH0ey;jbo8pihKvV;S!qY}D5nX0joQ^dwftq?tRmDu*7iB)L4nd(a9DYhZuL*Sz7N1*C= zpz3F!>PMjJN1*B$f+)cffU^aqkPm@D+%57a=sf({`A)($fFlYF`H0-Wb}cO_g3E^g z8jvbS2~)%bD+3?ufy(IpXc3C45w1bh4N)7Q%aolc5k%rysQ}O?%ym1jr)X|)-S9qw z%JV<}xb?=JbC<7IDmsGZKEL4A_uhVbSX8*Wt>m%k!y|imi!JRZ*8RNc$Mgc-mYx`OmnD6}^qHQoXI|B{T|c=Gav zQyF!s=eiTgJnx?jquqzdHj*^H@Mw;oCw>DgJoBMI#;o&pTBK(O)*wS|?e zT>10)`}cF}^gcF8SdmQzPE&=%#?JcSgAZPcDGDk0X(dMmP8uvV=DhRvlIg?y`pOhi zH{bZ7W9Pj*_rfKs->79{fnXnit#X>$%Djf+DIL7`PlT^d9yQ936IpHYR)JVA6gFJH z4ygJA@)WJV+bZv~7=AiQdgvN5`To$Dy+QuMGGHTM5{$moP=b&{W$7ga@fL|vAj4*W zlvoMiC8oSUG8IxSL03>wtMo*K!B-n0@Ba9|f7)qhkQ~XcAuYB;T96#cH?VStz&$m= zlE6e4!F9;Zg?kTt5>`h>#(nr2Bv%aP6Fl2Igm>{pq(+emsmCm`c(LmT3DXnoR_{)9 z6~Sci00`-@vtk+%M_3yy^cVmh_;?0igkN-&I@n8@XFF2vB+(VTum zB0Z%n?LcJ1BW}ph0I5_WbCU;!2S<%!>5O$4GtA_5@Q*Do;GW!3oky2H|e+wP6H zFusV`Kv6{W|Ih0=`STI*=hNWN*TA2TfIlArf5sIODFDc|kbqE#GD43La0MnYK>UIOa@X#xI?5s_FIBlt{POG{|+o-LOP+ae~-m^y8AxV`G$ zhD0J9_(dAz1*(Fvcy}B^ zmq;#v8vBf51urF`9*Y%PYD?xkM{;NI6h0r#P^ZR%)8BN-4%uNk+|BLkRf(rcFGlJ$ zqxDm!#9`=0g6U8K^WEft-jg%ol)05>tfhVkI-MEbBQBEhn=i)+e@Vd@m*a;hCWc3!Me z_;_igmIkI#oCqJ`7R3dXUj^;DH`b|z6=b&@4o-hM<2cXDwl}x ziJ7`2{dkJ(-u-M#!UrTH{|a>n*&=3Wl26^AdSb!T&n43&vlDs<^)*u(dG^&82Q-YgN9|SKY2H+prk_*qSNU@9J_9IKr3ie zNKW|dvm?b^pDGetF#sZf$A}cHT9Qg)B2#E03#zabB_%cHCQC_kRc&2E1N94w97jGy zOJgO=N{9U=P2*upl@--cEbg zmmqmzLGm_p@%_6Q6`dL{#9>DL*V8jINBZcc4hEKPd%LBjD-U4-_+vHr;}`J90`SKI z!XJjrdl~nT$WL--Gw&DV69+jrD-%`7fJO2R#ykRx00<{oY+e?YvqmnmTL=OIYy>Bv z?Z9&MKA@6BrjRQXBy|Gt4v`yTsSN3mqA4&P#=t4=jc0%7Gl_o`*5OCe!&N3rN8Y~m z+*sHbvk~B$%P&*yICZzd{`0{sz#jLGYynQP1UBR%IvFaeFJO!>d}nO_3V0aDS?sv{ z?ahyl^2NM*4taDQv21^KTC=OfinTGgWBppi+#d9ng z?#!njp@-8wp}#Vz_aRN(y~cbvf%*}hJQ|_0B!JhlsSSenpXlNC2*L%cAYE2K@+2ce zkNmtErodXfRwXCUfLQn-pEXk!Em^W;@!~QfNP?$)y|)PWRg$~K(e)@UD@CfkskEZHwiXy$C2UVQxI zK(rR#s=>pW)=p%U16~2}0_%ns0F>1G*nwEIx3spjw2`0>0zi;f){gd0u8(J6P;gLi zNJwZ%aByIN)(fdfS|9(wKp<|uFlg1@1b{(O41tfm{ei*!FE8-lB$v~(P9ySa5P2&j zZ%JLAwH}=LEjV)%IP+srRS(XT-OWf(OH0eoJ)DfPxB;L1b{6dlQ`4`flEzinGw?>YUZUAc$~X-=iudwhY#&LdGO$g(}#C&+k0@=J~XOfhg)0QI_i!0vvad@OlBLI zi{h*2ZeO_s@B8ej)tgdNQ`0k2Q!}x%c*(6RMo7vt7lFb987?g0l&3G{)zs8z`iE6i zFeNA>TI75#57~X#=rM5MK#%%{@{>P@^;MOooj7{*=&1`AHlDb2{*uWw2|(UF;JzP2 z{k(=`DAI`XO?nkL_;Ge5+Pu`2TOD;7*Ww2y4E3x#wQk+IFMi0-4wwVZd1A&GKg5&T zX-|~XN+(R{8xR_O?a=Z2NEj`z7%+SS7ZmPpIDY(iL$8VPlAxq{ufP8KjNzfBzd^bF znoGyL4D(oi?R0@x+?09q=FNX{v<|%dgYezQXAYk`nYx2#4xTqrzj7FoFdx|QG04Jt zmh}<)L=79&J1`)(pjr?=bpf&P9trHo-q>YIDSu7fdkqk1i@B}?v?1vuNFY&e%*~1H z+gt)Dw59E<6hQLa2(5%vWi_{|2Ttv!gy$`UQdCPJ^B8+uaduh$bz{4m8nR#NrnXmA zciPoLJKt*V6v`O8h(&*MHaKWNTzGI$5Rk$!Jo-YCgbxXC=1Gt%`_Lt$-nXO85f<-9 ziA2GXGPO$BBLpHxfn+EVCk+h=>(Qe}VAM?FPy9S!`K)2_@$td%(R{*V2aWFM=NTI0 z6BQMd`0Ui)evUR-wA#C8&pr{OM-S>ZB58syJ|+;ovh~9xn!pfWm8{+7uryS(;RehZ z2e#jA$4x6dqQU@e)}`({a<>+DpqAA&bk=CJm;nK8wm{rDHSxpUFi6#>2y@1?A>@H* z5{pteu>K2d3M#vIxcXz%bghH#OjnX`O8R<{nrcB?z4O2n(p8%j;W8ySc!j{t9Y@on z$tK-ns2X;K>&kCqCB280v>Io4A7^+QXBgUPZ?A_VR1f&88O?1PP|SfJz*9}A&T*i( zro1Ynx~c$@J|`ERbzM7z77_$-Sz___hhV`UNPZhmN*qn}L*ogRJ6fBXNmd$t4{d5v zuUP~_+Sbx&Zn4o^WIN^CXY|OCgQGN6=g(!WF-7onpnHA#jjLxa=I8MP7c6nX zzLJdxN&l9owYyH+JFl%>wer(nE)*zYBEH&i@G8~`a3hgTENE>gUBuGw0}W5{sF^)` z_S8sg-BGa6ZzfbRiBseQg6h)lT;H(f+_~KslfaB0LT-F2m^W26gTzzLA=q+@`c3rK zi~&@BdLr^Rc2eIn3$gx?@kzTCU&|~haTRGSgaxx$q69NslgG^&3-iEL*Y}n1m1)xw z0#bS)SLr_bxul~C@#g)^%JRzc{It78Fdp;MPw(Eb``E!l7eVxAS+~T*(L)FJAD*b2 zVqv{3mWDCIh7Ic(J$PWklqpmC(#6H?5s6%c)!c&WSBget$||J&{oG_?3l?)b?`WTL zRk-y=hX?W`--q^mUH!%WbGg-3m3bF;Zn)GW6e&_CK@D$!a-9LS`U=utpNE0CT=2uM zqz~)s{On}xe=vZdrVbof{N8(}gR57sK3vw`UfBSSZpgWR=i0>!sdq}-91+mquq5*B zBL6{&Nuvfw3OJGfBV(Rxbd2R=F%!q(H*6NZyR?$TFVnN5u`)O9`t`JO0XN84Qc`Xz&N3onokXgPPLq!~tcLTy zoh)u^hF@yL2P%M4R94i~qqe84g5)QuZ7P+Wgb^j7aSTxC{T* zXe0#FWXRG^83csFL!lJ8d-(vgL~^>fr^a38Y_Jm;5#kXV2yj!54`{TWS}%>g!R8+u z>5XggWEm_w|#** zUIK1=9o(iodoiORzcep1^WwRUKYTJ4sT-tJ=3V|n;pb~sfAPf=V1?)SXOaB11d?_Z z(u&>$54?n*m%_^@&yT(<4sEB0SQccqMsds+AM@{C~in)VfLYg2h|_n0kTez|Iu4p&+%|7`nj2R5gC zwd>fC!|TT(Ec-I@1qT3J>&NWbsh^G{-{;Zxw+~|cQHbutiA5kV0IBf7&flKqUrbKm z$FeMK_%VThiH%*g>MJ}pU#k$t4;s|3Z@9l4?BMR}+3&emUw!q-0X~kKm(E{F&o|a} ztPrWq_io;D25WW$A6pHNEsff&c;m^(CeC|%@#3k&V#mBa7XHUpss@Sj9?X-^Et`&T z&Rwbl&Wn}~CeAXEllXGH!qU_1AmS@`NG12P8Z(cZ|Py#lWhDT1;L|ZGCg4P!)>Gz;Iu_2^qrY6JYxgI&l_c2Bw;BCpXA zD6a(@wA$Q+Dw(*`WNf4aU3L@6eMt-z;`Jc+MWu9VK`bH=2oGgRem>lSl2Yflvy^ZJ zzC6?jXF5MX)|&Hkr1JlJVg8-Fqk|IxQGJFC&fME~6VP|?GwLxlGkv=AQ_N$JF+x14 zzd8e0eTERdAvdHYLd8AHPZQGk~sRg0yrqV@Kn-%lMpeDLVmaHp0hw#Z&q0x_6XrkVL3g65KU-+gyT zd%Ikrm&H6bQzPmHkEepADmeM53BFl}e*68vEm2$&ne_E4hg}prXw+Q7Z$zOS6bxg?oCcP*JT^ z$RSUC(3g|o!oJ8F)u2#FLvpGq6xjXaX`EV{AiYn4Ci6g(ZlpH~QVfIuu#N2BiZWwi zem)6&;dTk+OQM&#xh6pBEDCc-ZOfl%lOn4aZg!iE>!(`-E;C|&n`^zb~{Vj z=FL;Jb(*U?GEbd8d-%}d8k@V@Nkd8V}om37U zoiNEi=g7VjhteeRemhCsZSj-rZ+q9Wv0YMrGEeJL zAP@MYDSp#t%zSiGl1?}Dhm9N7Z`{0n*REZMF5a$S3$hT>O8;Q_vSrIj zeE~})OP4KsWm2yIQAdjozpj+O`}o)>N7YMDc1S=pPfuyPxmg%F%H6$hxQ|ajkXGg< z(ktZ}H#cNT&7M6b49N;9(I|RCg=nd21j#AZ4J_G-jvF>mms5klkS8#-kcZ}g zI;T_A+wYv*_`|k?`?qdAkc|`qO9!cDc|x|Us>Ofw*hl9?W#|z7drZ(&51APf9gQ~j z1_^cUY$bG}*lM?5LYdK*>(5brN5)Xm(JAWD`Q}NpUR(CzD@*2#9obhS*64tj@05M` z^zim8=Wkyt4M00EYPF!a5(3k$q-4pGm)?3~zTDET7ur3B44K&#c9Fd@P(FW@s1#6I zN_erM&>5mfm>$i05S2%`tpwr-BDA2eyd28mGV2EU1F%?xs(yx3gO<<{VyFtC2(@55 zpAxQ7s+_TQ#M`ybAUk4co+`N<0#u>$_ExDNE&u5x|3-Hpa*Mb_5o9u_CeS@@5g>>VRZrv=z5hSbz5}qSs_p;Wo1N}9-FrdH z-czBAB?8J8#Q`EH;%2^%+$1e13cd;=3Ni!)L=;evB~Vs@Qo1+YO*h@UP5!^qhIagO4{ z+^6TAI*Wc#S5+-p^5QTn=UKD+>J&D}6rF|@3=TWxzh-F)Q!80nnSqTChb7eU5|TVg5{@tW>5wCpewQiuEx2eFKenHU?y@39O$ zG81ECCdNhx20W=0(9Z$8fg2Y6fkl@lq|)rW1trih3LsR)NExw}B0kPg9z+Ig2K9{- zptuqVkkTeX({#4Cb($em>iU8bx5uBD)n8&fKvAYQJrRlJ%98+BIT}B<^zx29`_ju( zN}WcBVg30@@EenK?zS{^(%Q9apB(_zNsi#88mOY+ekFHOOHxW1=muUhuaNAyU1_gZ z2M!x)m6Ms)85jl%i7(ksS$%`qYuJiNXdDIv2{v)d&qtZp;<|_ie4Bd7pvYl1RYe7t zK}&9T*mIy~Wy)EsScg);1%UcQlIpU`)B_1teLW%vvlZbm#$zPY+^rJyAy@*a8Ibh> z$O|MD{f0&=D#*#mF%}o(LKP{%pt=j|&$bKm5;GHsg&+lHy%R9DNcI7Eu?LW)r-^+dyrf?I}<&aJ@gNtZ>8B=&`Rtv!faU3KRx1qpDD}oIVq~KA4SD0 zj(?u(%{a{-xSqINdKSm0KcvAA($F8$;0Cke;5FS)6t=9q3G~*s9pr zl9If|uvF&J@8;aECbtBd82tK67T=IOGmz8|*483ktwtTUk+1+{)J?@7o|DvLJ!8wlUaw zc-Y(9xP>7D%F(lzFBW@)q^U_R2WF$&V7Y?1xPs-lf;Vvmb8!W8aRs4iX~^b5G6Ujh zNGKKa6Em@@)69=P6;%?VZpZ<$9;Sdyq7?kZ1>Hn=B)UPe0a-!R(6Ycu@n5}k<1Te; zPMCQn$F~4n%I%CHOotxJ`f>zcSHm1#=c0}I?z>axQ(EjaKD`1IgDSI$=Ob8j|vwddWt@`XWf*sxJXI1@Wt<~?A>%2li1dV_!_;jD(0@k*PKVeGKf z)b#SpuT6z-V>|3k@`oHj&cm;O@cfKVK5|_7!YC`I?ANQeJAiFfjjr<`=90w28_yCE zsLa6=EXjW4Ks&-dcLg$W<;rFdoS6DX^(s8Nb06J&L;C4MtE!MFgH(vpQiS5+vIelc z9J&B4BA6RQRTdzdf>OAqrAUeBZWH`OHGn5<9o-=H;50$NTGWP!;)~aZdU_(Piqr^% zxR8wssD(T57I1f9Rl!syhVJ3`uy404dlsqzvWg<(FQcjmclP&V0yHU7d*pEkt zMj%<>Df7RqKfhE`-5jecOz`n&EuI25$WrrgI3gyQ7a+#;EX>3e@{e|$xPFT=+i>i? zh?4Hj;fKOeEqibEYQg43nU$l10(r4$_||s4XsJHb-aE+{)_T+oUuz!A;qgY4Zs90L82IY2q#UhW{UaCHwrv?w4jFeuQ^ z$H(2>6O11J9n~0& z2ed;v>&a;mVNuM;3W3hL#Wty@kMweK)3^tZocL}esH?8xkuMu(>)EgOW7Dt>(I`I$ zNBCCu4EF-}972cR0NnK{XN#!!o5BPUbF&`b1M-GnKGRPpYX^K!Dn~L76lC!^SSFAmX4xBV4~gVu4_s zGFq$I9f2~Eag)h-uucKSXs}1XTPkmdxk5@Z2gIBXByVQJ0AR%(9g#}n;R2{8v=%~Q z!t8c+cZ4IEWnAxE5RV`0#+9D0fXer zbu#EXSl)GZBA8ta>l30xBG&^w%Wc^@5x&p9$U<-fu5TGUpu29|ijRcFPF~XY)!TlF zyLR!)jpSS3eL58~HXC1h#5@6+osYngIS6ye5^xr5`gr3+xH0=6L1`d9dd|FD{%vMK z`6ZC)1By6`U$^d^shZqf2sY=U>P-Xr?;ESSyhe`j;@9!K_sEf+UDZuM%yu`RCQi;# zd~f$%&E&V%=_&Sn2Jz-LPRgk{Th5WSK=MxV@4o%cM53U=GM9ig1v&1LH(!3`&F3It z2}qVcgWb-tEICYzHKB zCn~R!$VoprQ_v9EW1`3w_UhATZXcbZp}3&ZCBzLb;&p=DJ=CS5prlbJ$G`gYnuN@G z3ydtPvEbB3O4=WT1&DxxScyOw1D`x>85lJrs>o&rB!ku|{8!9=KspqB@q|n+f|CJN z674W9&2V@@W-+(J7YbJ94$d=gcqL ze0DJ@*LCKC+0TvPYtxY<^QxoS$*Cpp3JOu4{q<~0jnYX1sF8P8)^A@!ioUohYZytW z(~1C;rye{#Zs5n4Pie?3lWW5I~Md04rG~moNH1 znM9TV3^M7SAWP`cL346aQoykZ0ByGp78BTSFt5Ow!Xsdaq&%|uIGr2_X_CebRCc)J zhbfpfDEHoq^egeMF+VO9zD;W|O@{LE3?e1ZLj!#S^UX`v}VofmnM3y2kp=! zk7GceW_JPEdY1baw-WbPxsj{_EW8K!#lEd`piLeD%s3WHTPeUSzhSL92UX@4?zgey z9?U5Fu#Lq`a+F*0ayVzl*udaMwvRRu6ZScgGOXm4xp>@_iwl%mpIY??@pC%;(Q-xyi1m9=iJ2`?GoE|?QIP$NRt{f!->s<;#LV}#qB2oxw zdL{Q*0fArS20{!0wOAG$bzTLgKcNObN|-HC1ly8FXlvt0daxcU>g`X6xhui@%n z!_|*WONTL?4Ye&R6SY>L-+@%Ss4$*1^u#-NFnn$&#NUcfjlTu9QILuzB_$`J3JbWD zw6-!CCT!6k2fdVf711RD>#_m83g#jSC%{O<5E0}Ax`QJ{A8y)jI>X;Q@#QpAlj)wR z3VN&!L?OZSqv@0>+jQ6556btb(ruC+1)5j)* zX_;x6S%Xi)O;b%Dnl6~)Ot(xKrn9Dtcy$SnQd21sqDKg7a@ICCpFgP8Ld9H*!MB!c zs@L)LDd)p6P@m;i3;-jbY0bL#-+%vQI2wLtKih0gBHg+xCfZ;yY~QL62Qo5@fAQ^c zkFyexm^1=*+lh6%7dMWZ#0>#hGM4R-?Q8S?mO;B?jX09`m%O=43 zpt}w(lutN6jgEI6IYHnGn?PLNrqki%ekl}eeC=_v3gx@+;sT!V)sL*Igu_S-c$8OF zS3!#`E-uK+ZvstnO&yGj`ufHOfFY}!np+xC*chm%d$rXVpULRWPCzKcX%6UB19rFM6YZe(MUm2~)#Ee6}GO|v3|8_Ms`PQw3q-$p{ z-H6-2@0aZC-VI|P+P?q(jKH{p_g}hr?Zj{L$qOT4OYFsJE$~a{fopp+Y_M%m0KKo76# z>ERjZ;ThD!nFaYUFkx_JW&?8%f^`Z86~MH}g1wNRpPPdjypSesh^SzNnUSApRAGzzqR z8n77v#VR}c>!oiufzf5Q^z(!HH(>@C1#b92L8P)(UCwK2sHtF`U7TPqQ*_l-oR5x) zzTDDh$IYNLMbJvVetgT7j#r5v;autueyh&LfnePAaqWbPbCD0w;z= ztHSKFzS#Cf0^7RP&>Oivuo_`SmQ$sN(h?&@lL|n4RR+Dhu&|^Q3r#+)EwIZ4D=canHc`at6}g2tb@eo3fwQng5rf2UYH*i;=~(zkoU4D zhd97_t@kXW>_?rO@-yK?Z*;n>VfnaswvDDL3CJ^LfN1b&mc%LEn73DfZ@-G6y#= zMT5ybz|U7*k@5YGZ`@qWm1S3VZTsw-qd#w1|Ix=gPF^3?I~f&ipa)jv3B7*W2dvJ& z&y>ptj2JzlpTCGF@^l_N`&HnHB0X5r%`4Ya^2=KVwj6(YY2e+sJJ(-^wfsHC*B*ZBC27d@e4l)}(X1v-SMxy`OyY$+b$` zz9UD6Jv!W9jp=+PJ@mtr=6m(^o#rl?ZTPf#^T1~kJ}N9M{PCw3y)tvnJ1;I?w&=NO z)22W1;)_d{Eq!U(!pDqdFRx+YXmXE!At7M{P+9`jJUrd8;06YU3u+Y6GByc1+lm>9d9}STOa8 zr{_+aFmck%d2{E^m@(t&dEqV|v*v^6SKvJQ3B4>o73A**M3Z$#OJf_NEMAlStS}rq z+B!QpPG#@z72qVW8vON0V+^G6D@fzFkj5BDV+^EmJO#aJGRF;Z9YhO=85lkyayi01 z3SW1T5mObkVfq$AcKOAW@QjcpD7E25lH}w8%>fEDa*vv0LSa*(ss<0h4V_At85rR1 zZV`L~{SKZ}7<(v|B)O;wjp-!3W0JF}bphR_wJka6%Z zzuj!^e|qzal64ze)^ZR)~4EDCu0GA4)hQ&!J+vCyp@lc zSMd9wEh~_|aVNX1mD3NGc9rEO-X1-AjD6KbOfwrUmN|{Re7V}U|C8^(#+q9k2ZoOu z(7*qHu!#N+wUuqpuZZmDReAZcUNd9roM)eV=9#Cag-;lP{v0uK{HQTl8q#3!m9Z`U z&&&@#fBYaHKH%P|V~4IV=b_W*&Y3-T%AhGjL;4RGG-}rCK;nK0G0tT9?0bG0Bok~z zia~=0jT{#2Z>P7i4-6hMd?03!icsVcie4YKLA@-v{Myv169QUqe*xw)t-Z{~M$%e= zRimi9nNwQH89tS7xzXZ3Vd_*Qq~mj1Qp7D_C(u4Y#aCWI-b|o45a7wn#!?L@Abkrl z=EA$gCIGf4Ob+lFNlJ@YtN_YFa{O~Kp%Nh0&}V=VVLpSqxSNpNBBThqDq;d~7hze4 z2@XO_n2&^pkRUZ3&IDSCaB~u~$cTg(J|gmbs?{o_pxUafh~f)&6Mfa5>fMtw$q8)V z@FGc^9TC?0Kc0oSu5E;zr$f$fK+ZQp&Nq^rBeft0m=@eDFW@>S2A0D7yKqp5IG3W5 z!ouuqU^C%dNApk%p+)gbzS3NqTr?l)^>A@yA}${zhlYa)FH-nnW80EV4Y+p)aebn& zOf<2$Nbv+tY#V~H0Qzo8N|P9Zk9`0APZyKhJv3D}29FyhjX!(rr|C21h5Ne?sJ*#i zPj-7noyP=EZ}J_!t8gfXRe$%^7BG!--zG-DuZmEM^t0=k#)?RcjtTIh&gHZ*VdmEr z#}e!sPknWGZ#V*QGqotYr)+DeZ$ZvPG74v@P3O5Nc2_kWzAK5jp{?)INj;{{adsVy zF_XIv+l@NC0xCMNxe|m`^0+9|d2^7}2{1jj)U`F;+NRY`#`^T6c?`P_AINsg zF}Rod>StocCaB@Z$ZyyI6?!u~AKMY7iN+@+)?45k#a0efRJGfqFA`|<0H*VkY^VH# z#S`s}3UfGS_=AoPKHgql?oK`$wZ_NY$J5Hu!#fJulh;Le%EzkLCV883DkZ^TFNMQJ zdr2F{_G3nb87y}Q40Om$DAAEFsw(|nuhW0~`GWbc5J(7j{<+{G(fF56R&poPp^vu? zHkQBzKS$vvt@?wf7JR;ID|Y$5R6o8a#f=bbVy+vU|Db@&%P%a3t3^Zw<>lt*8;wZS zMtA^oAx(#vKJiQr*boEmaoez0=m&5cF@7cH?)_#lfSE47yLIZ2 zuw}EMUtR7QnUuu-4PaEC_KeK+(6DIC(#V9-+cPqkL}BWf zTN;fuTPG{L`N#0-^}INIY(H4I;PbEljp6e{IYY{J&sa<>)xpVV8H#&9U+@fQy2b0~ zp6{g}4@JUQX0(hTQRR4W1VI6SdO^@ADF0 zq!tX~9zob+2zrC~go7A|2hW_6Q!>(15}_-jWHv}D)8Q&ULA-rjheTlAIyC-<35bUd zMeunx)Y#eFzHJfEQiIKJDWXdgckBoVfFkDr%&Nd`SSYaVI&Rgf4O0PdC`1c?iTT70 z9ydM4Uny3=A322=0gCqVadGq!JlPxQfeDk1))BUB)dr_~H5HXLbx{3Kdj~|9Koqgt z@Jb7uyc-nQI%KeJaPskX_we?EV&?7T3EoQpi*UUUf5$T06MLl=*Ni=XXRqv^pO0v9 zAq;BrUI18v036T=fBh)}a>V{F(SZP-fvwmi@RC1$z$AvrFl)C#abE(1HyS>nMdnwy zZ;#AEhW-}_1P_L7Ou75hkYI!20Z6;q4hO?hn7!+vwmffsmD_)GGXDD|yx2qWU1AEJ zVSb<23Gfvq3+w>34m*H9177nZ**<}d)Ukc|Gl;JZuV9YckNE9vAPn^I5#_O5T=FK^ zr&~bf+X-_ijTun@f+p~lK$-04WQhMh_}x!`|Lz`qw+N)#zp}5mm9RqJ6+jyaW>Pjt z$A%c#Ac4P3A6&iv0lhZ#FDyi`HXo2Wa;KG)A^{drUJ&~H)u#ydf%gN_#d!P~@S5-b z(+8~E(BHuUu~_2q11}U9@WB{`ATHv4fAuNCet$Z`UmemS6Yh|S@qSV^)K9|PF%H}r zZ*2v$fD=)WaQ|6R*#ZI?WM&nd6;z0;BiMxhGKl&DS%(#Wb5nFA!gLD>CdO_0kW^aW zO(cS!Exxft5NuZT6&+#gz__t=MKJ2D;vP(+;4`c8Vx#+rDGL|QTlykcSqxAu1?k#6 zBU#1vg^k#)twpQT>3%7^v3c_iy%kjspwM`ES;197g`f!Hie{AGvZ8tb6u)+JwGV#q zAC_yR*+B`pS3&M&xJJsI7T86>z>y%zF*HBmQ-S#j2Fso<*fR)w24PPZ z?CF9%ZEoBwC`e1o&ySB!Pe&an2{02~8^aM(`HIVqBR|;$?$+%{({aWL9$@bpamOkT3b+;wW_eNsY&!> z0f5R&H`?^>WpD2ZsA*7;hX)%eSU&w@RKJdMe29}aC)E<6#j+xL0rGRxQ`1sYfFR3& zZbxuI(sQY<0cePLo{=`|iC#CG+FM#W%>Yo7bq@+}V0S@UDv|4n_sI*BUoaxJAsEMl zFv34jZasATT->c&CqJA9>9pn;>pr#haFiHZ5KwA17@0j#w7+#6t}4F->W5a-95 zF`VgsNc!#zudI0c33dwJ-KmHwEkX!)G0+?u_<*fBXM8aQ}TCOuTZVEICSlM zA}g&?=A(kUNMxm*^7iT(zT1Ax0Z5?Kd+~kO;;W;bGq!@2wzS$3fWeF=wTuJIo-AunWdrCfp#Zo(#M#Rddd2!C zn$r#%4=+zIaI#?H24qNtqXSL^sKq}$q_~dFhJ;d$gZm7UA}eirW;R>|WEy9KhE&Y` zfG&=4Tma1iDZ^=*xv9yiSwtVJb?A^aU>pG+ib@>Ntzg>+%Zi9i+&;dUUfsrYi24ecDY2ZxKCbgeenf z^)Z0VfYw3%@!=EaQ8FOu)QLS|=C=h&&CL^gqA+7D67TPD+fTsCHMI%+;p^2Hj+;DR ziaJpe_Q=TL!-qaH?$L4M!@?s2W$EWH$2AOma(N`%&5-VciTfo$HO^o%KQ37rh2maX z^rQK6ZkX2Q-S<|%v0}-h7ZxpAy5g<(1m5?F6;mJ!mso*#zl7yr1r+H6Sa6SXPuUtm zD^Z*Y3Md8rz)35nj6f45QxMuq6$n$xORDcxmz4q#taXrg+F}+Ixfc=AK`+Y%YoKjq z=zO(G0a`{N4fyC#(C^qmi=@z=NV`L?s15>Xoha1j<>{vJ^6_$Y1}n-xJ*L>_|1+Kj zZa)!E1GU?NrvWn)@L&()X%yrRl2LG{2Tnsa^nbR~*Mf;^d>6qk`%;52YC z{25OJbgzh~0VKeJr*Q)D+Mn?>e|;d!to%O4s79gL`-_LuR98H^zLJhO*=rf5{QKN zO*iefLkACjcl5%Y{Q3?Yz-3U>q>14^4JG>l1(bfgZ}0Apqwr0tHU0#GdY_*L+Sq%; z2K~n;jZ&YW;Qo(J3-jaowQHBmo&}Xv~!Pb6#4zmgfT^pBUda z$WKvp{3G24z0=Ho-KBsM>gp1!@9{bsCH)wwUr(d_o?sUlRE0EpiA-lx6F&(noi_oXIiFhw8 zDvP2>cSDecwf(?0H`J9-7+w0@Hy^uD(oA zl_h;cQ#Igy{&39uu-;9S={$?0RN52(2Jm%xDhxD{N=!BQj!|%V2#F|3}dj`d$?%G;`k>= zv-D;Jdh;dp=F{lS2=rzIdNY9P>6zgO==@8P)mVFUBJ_hLQ2s^umKk9@110i4W_{iY1wfrU z5-+4A>_I2K2n$Ld!BV)XtjuXPtrUpS%%AAvu>G21NLa*@_oJx6ao@{s>B9l&I*KtA z1rOF)xVyHo4cwY_lc2qCKonpTR&=T!r?p-&d$g(o>}D+?Q|5mV#tvzrLg4(j6($tw zG@MM$cb8@qzA!GtomU$V!mqXSus(q5?;v^)YXS-#1@_Th@btn1jA2cTo$MM8K}laa zL&FOM7Ov9E8%8;(-|#g z30Ya;S+8lAx%Ba~cXqV3S7QEBb$9jd#R9P#y!JK~@foP=R$5rWP*ujH*R(K&okFUH zv`e~M+FBYLt*x1?=eu0W)$F@>lUjI&q^W|Q2&a!9gUuynK4_c!_LmCuSGd76l_+zt zfF_kDmi!{m2ghbCL+cETymp+@xUOHHHjUK^lJmodpWi4McksYPJI3KUa(vz7&ru@%`fyx ziI2OLoL*)uOpiVG-VZ+vduIOhsL{iF4GDYdtyBJIuEbrw_`|VNzg@cp7l)vWM8wYr zp`Cw8zBzjI>qyL})6L$dE)2dF1TPOtSB&uWolXR}H^ANz0WZ}}wn_SgukYA5BCw`r zvlasyE0Xia^!f$UE8lub$nJ+`CuP z3XVLjEG`ZtXK@BOr%|rGn?dX(9%{-P{K?NVN)LNyoW{x-VyNXy={KQ;b3xY@MF`bJvEkZymE#~JK!w|ghJY6Y%zmM2xMX%XU>F# z3@-ZYv(J7hZZYnTMnPs+->Vw^fM@3DUg3>X)Lp-1k&6_uM(u_Gn1_wIX+F2>B z{cWrr9lhDx0%B`zjV7jX(DBXnT>?s-*Ye?bS*EKr>w`gS=>lM;J!0LYe<2Zw z-ar?_`iy`a-oQS<#P{D0BF^`6^aJ^-<>&_)`ay<%5aB8Yi;Fr8?Twf9^mItD^qnpC zal$^f*vA(8*kT_UWi|n_M6Ny_>+!$LtN2PPUegKX)0LL4N;|1V&0XUvBycy6tDR5s{m{SN0V^$FbM|hZT8!S(oJP{C_i2Gw) zQv;<`zl{=Or^1_o>XdEX_Q+_qUf@r$CpYjvu%|bK0dyuA7QwFJev#97ghPcq#&HVJ zlymK&5qS3<+r>9DvoQLRQs%uud~$||0BRmZh4b@E^MM8|%cGzM5R%*m@d+lQBUCpE z*4o)Q+X3L`YD>Y3KfWQh*)Uw=I9%gsT;niY<1k#KClDND|A97u(z_s+cpygvjylK# zMtQUh0U*$y=@dw|#7-b>kbgXO!S(YEb>Yyv<~Uxf96oMRsNI+AtCG&1Jo)p6&7k#vO026!M%1$inMG=II1*j6-a5mNaHwABP$ z_5`$504ZrSU*Sds>{*uVH+%?_nxfEIu$8q)9YT9yNMlfgHd%`E=VO?ReHOkn##JGDkQ?T)fJ><8s8+QpbhFZYoGS* zmNpNY5H*)_@CC_RE0MO~r=0r4yt(sytq@+cQd?OmZ3$>2_kKdDT>SjdAwvfa(g%<_ z4)6-y5Ira$W;QZEi-0wSdWttNHnr#xv^8l&oItslkPL}bfc;NX7>^>lR9^zEjT_yC=_*S3{!b19C?5|UI590Yb zkG=8w8;hRR%ZH7Zws8Yrp3DrOG5-`G>lhdptKmP`;5=)@*fC?nr_X-{3(v4VA&(9k zFnARBFC32Z{Pj+%a3{BNC@x29pAyej-33SQ)pOUcUr(>q2h+L%iXC9;VE*F|KBuB0 zj0sfMEX4~1LcuF>6$Ls3TLe5IeHgN$G@=qK0`E+!lY_~r8$JSkFfAV_J#C>$hCS~D z5Eaq@G!7_Cgcp?6<-~&j(Xqt7a>J+@>?dUd{Upo{jqWB!b!ZWX!aypURnLkK!OA_S z1umXO3Kl%3)sz(^T*#8xnLBxDr&&HQ_{$v<7!s_Dm=IShv|8bASetXNRk?!yKT1vXW~wvcp(X6p9!{TE(yXh7Dtlc)NCO z3_sd;8LCxDjvQfjq%pl0j@gY}q%!vhP02zP#m!`~JS~{qv>>*ZtoW&%6ADc;0q^w?u2XYOH7{ z=v109Ah9vi;VfrPtN(vS&`%2n0g5V~Z0RYEo?HcsPV(1#+@!CF3QQoqilwKZ|GGyV zeMQuEg6Y+#^wj=e_c%pgxlB(p>B)hftb6wO+m^`vS!{`V1liFNerSmfOB2-nzInJM z9@5X)FZ3eV;3dvbfO%=K>_Z;;g^{r3-u`%jWOJmwnc;bWy{7IpY~L53e!TDW&D%NCLEP{t!itZ> z(Xs|w_A0XhIbcTEQJ&^!%xhKA8+RT!c0S9d5A;B!uN@KOM>ZmW`W3zS3NJoBBJfA9 zR_b`0N9}Z6)m1%r=|}|9Pk`-{Eqiy>qzQvS0s?i+9X4VYCxdBt9RZRJ+)i#AlCoR@ zTu1?sXgAw}l5m5H&ag0UnCLQ_h&h#VrRfeES)li*`1bqzFotU4fjZva-BHH{KC2HZ zET9U~NNg@MBBiRXuCAu0x}~+XxxTixrmDQOqO7=(^j5h5cawPg;{_NGcs<9#GsR2w zK@RrdP@#fut}cH5{(gRb-rioldiM_Y_k}aj&Dq`6$&rot2L}?zFX`rT^dt4XoN;)- z?}Y}r8_OYqwOYbd7hFwJDH6s|4-k1mG9&_uxl{>>aF|gRn8(Z6}E`POQBC-VrAnNLilA+H*`~5nprnNB~QnJeS)7@Xi zY})tp@nhe9y5XxYU~0S~JwS6MC?5o6`t<4POu~gMASu9Ix+sLDqWBsW5qMv5Aj{m% zf(Cl%qzWZ#2iZhLTK@VlwwYfu|Mf7IC$&>~j6+t(21H}sfZbn=w7u~tv2&RnlDzS# zr|T^Enrt_NV%{)p^=B>GkU@DU8ii-Nq*?%jP%=l4ZH zsagmjf80Dx8l7PZToZ|u%j3B3mVGdDWD8iirJ!tAD0GOrN!Jggxa>XXL~yoN*CA7} zv9<=qG)hbFL8x#;J<-Zagm3XyM9?G%uH@li-JR|2NZvw1jzrAMl1V9553o*AW7Hv! z*}Q%r1!#SsD#7*YPW!`YFJ*?QF|n)D`LN*RKZ+ zq>r#MAUh+^n9}a!h13UG@tu>m;56J2!P#Qup5~@ZiC{+&Be&H`#$3BLMjs9r>t4yA zK^+!ykTGqUfn@HOv0J2VUlI#cIRRbsuCe1IFmO$p{m37F)NaXreEpp_K)_!Q|~ zLDt>WP|@Dq)`;Aps-~I-NqMCKM*!P(Mk91S&|IFqpqPvvJ!FVq zXxQWtBOe(&WW=c8kipjWlcPXEL*EAD&+koUks@M?nu|80Doww$MZsBCSqVNsr26H+ zxm{LSQ4N!_)R=d-vMLq!!>t@tD=LK+^$Sb@Fu{|>(A6mw6ug;oqs_cPG@LR)0?Img zMx3QoA38KN1b+0Uio%50*w~p*E*Q}*Pu}z8_D4s5|6|w-r>_(T4e$TaWfw`KoK@ne zrjFV&+)w~OP;?D_4!2=lf~2AXYPMv_n{O@@xHne`Y?WRfcjnxs+JR9~P|puzf%%4g zFV9N6+2Y@44!J1*&A#GGZULY8ks>p8e*ZxK$34REjhBZ_9TNx_hgNm;Trr(};b|m^ z#Q*RSZa>7sNwy4J$?S7M;{EyOo&x(Eo^i~7@90BeIcDdSlvaZw2t8A1gxU_i7|?hE zV$_Jb;a%00WoR)9+-OxG?IM#=D(fPScFY)U9O_rKx3f+`0(v_*#;UrYAnQZH2Ne+D z=)@`AHO^KNnJuU(HI5z}3khwqnh73_kQ|r3j&9a+sPZ_ft5_1r#REr8NJWBeTx!Cf zj+!g5+#bKkV)XA~^sfQ^yBK3)F~-DTq#`#pR)YZ!Hb@<+Bs4d-ApgT?gxBD1aYgR6i=WZm(k*$v!@PBl+tJv9t$B!RBz(s9t zE`sv1Y}uRBoxIC_{qe_P!}jgFc4FdFvu0enI(PM*?DjEJBll$Yb60`65B&nm{%)UW~#LfdZtUT!@#q*H^iPPkko>8U4rE8F2t& zAj)Pfofj#sA_UoCMO)FDX_4WtjtK{N{UhnZl~0Vq0NjEfKYF^5uCMqUNy7Ie3E#&X zq#yD8N9=FqVKD61bR@r4RKbk_(-nD9cR`<7Pyp|8Z2^qgqT=+lD&)yFmE`B-K&=xk z?=BftPr#@lxsZ@tNK7C*0BVgrC>#}{a18EfH`&;Bb=NdWOs3}Qdx(D2b$3eb?L9oK zoyQKTs&sV9YaWXg$>ECGskW)L5h^!?#KVPtLgPw8l1@UBo?uYk?CQGt@J~D-DTF*A zHuMLP2jg?W6(q*^a>>C|1Q!Al!|sX!K;*)}c+bj3P7%B+g$3Dn!EcKfNFM`{Xi7@P zUATZ@Ht1Y|3m|m}@Z)Z(U8cYi#NlFqqv)dI9bE?dwszcL)iT)5lvspG1HMzA#J1`? zXU?6wa{af8yFXRWU-Z&otK#CzKX3c;>mM&&x`(=I?_rW&P3o-1a3`n3&Ps7N<;N_i zp`#-l@6`SHi^x(vN4BvhI2b94I#(;t0MK-~y56`E*E;N}HLF)+8E@>ew$l4n+>Zb4 z@HfoVkbZ7EN;0hfKEKI1JS@zr0%?324XVw$OjJj%c>T3km(LyM+dr)1+Qn1J_5J<( zJ{<{cDAirs;IVw#m^sT|d-J*3&rh{2Dk{6z?LBFzeM^0%$yWIsy1 zZ`@!n2s#@_FZ-@4Nc+fQA2kSY% zz*t;b0D31RuoQtzrlzK%82)SEf_1I{6`(&5UIVHJHTT-F2%EdnXLh68*i(UI za7su~nY4=|w~>H{emsgH14n@HB`9Sy8erqCQC`$ZjBO(A1iH1W#u2Uv2b9-#adNhY zvkEyYj$oki_I7uI;~R+M|NmdNtY9+8?SERJC&|vbO;5w=Ni;9-S5N+1TW`i?(bn@t z5E!;zE+&R}GV=0RHf{2eSsZQo&HL4H9<-aaTp?6rc)GQGE?lTMfEy0*Bfa0nKBG-B ztH$O*d;g*RVd&@3*AOGa224a+CeYlkz z#8%`7Kz&0SQzB(7ou-VH(xwnd{fEz1hAWjZ4iCgjaGf~COnMqXPoi9m5G5D8L&tcyeUkxuLYK0iw9KD& zxCwF(Z0g_>z$S=|$L2w~czEw(WU%6EO@K#Y6T_y{rUN zM5*=^TZaGHa>_=w+p@d-{+Uxxsr@WAtkRETN&vIShP9=O$JPI2xq7UkxOVuoG;|kHg-t7*QrIX#l+d%k4X&ReQfkNxMWmGYbuE8Ijmrj0 z6L^L&BwrV0V3H7{l+CZ-A3lEkw^L_+-Wg)ph7`~>YS;LTj5~MEUi|HH8nEq3)8v&XFrRJE37BNG-3gZz(TtuCJF19E!B5I#6v$qd~>p4nwe& zZDv|SQxp6eFw9LI-R+Hy<*1_F-PGFAj^bi1jm=#gY)pBRwRIDdQgg}!h6e==3y{;n zsJZJ(%G%n>OY1v&<}R|0h#8tN(qtPw4TC~@h;Aeo#3Py#@TBY;i72txFbU_!C+soTQ#AK%e z@v7R{9#o5+2H$##H3qemZ?A^mrlz6jZdz7Zw~dXJ+FEk2=x$*F(T{$se9ED`w6?Am zMpkbqlU*j6EfxRcc)5*@qPw}dS>Yr%$)TjcuCjyGFU8ONX|4L-d-49-2cK`5aRypu z`4TpcW}ITnyKd%!cYi4y!#c^$O^E~3@twrvTUV|$G*%n2k_zsK=f=i%Qigt8Ctgq~&UU;2 z?3NO+ySlogZS$KQ`~yPv?D+SdZ-34+Rz-qQ;(4T+y(2YzwD-vI37|Lag^ZRlz?i-x zIhu0uTS{kKWPX`DfODTM2z+^Jt{~6Z4t|M)Ie2jhFTTzZ_}mnuj>Sd-y7~(4YuROk zY=T>HVtQO8CfJ|h@BCc4=Goyx{O7=vCgl|1Ji7%4=7;cYtm2;Jp68zC#v&e=jIVwU zU1lvRF@41ZZZXsm5B#Y8h`9o{1PM{VGP4-&ttEH|3Q^0ux0eX~+~IaQnJjef+)%l} z7s)6H!NS1_YyzxTklavl*YYx=&O)+mw;)SS$~6*sAR>Ph8Ky*ar3^2K8&n}f*Rj3P58}I(sPR{8ZptBRo2#Oslj8%u?yc5{O`mK0=A}KKl$xO z!v;I2_O$CiMZu*#A6ZK?%r982-+ii7VHIOo|25koNUn_;b1m9Of?AU>mUyY!-hS}H zl`B_1;lV0@2H^K_E;x}o43elv`wtIi``A9<;=%wnUN&Xy*s+t~xDepU&Ow&p4XN6( z*JDd3p;vQ}o@56)sxGXuTO{v4J$_gusH)lsc!%ZnXF=jpS?QveOHmII@o$~1@UX$K zQ5ON;6Wpb{^7&pbO%%bco8gAtKz+~8_mS%R!a4-O5zug@2)gH{r{xycb#ym^I+DCq zB_);js!Gb6JIw9Ch*ZNng=o9ppF#;7kK$@5=SOZJ5Z1MEyh(06#~C)NJRZp$sLqL4O8_CsDXT>EybNLU z%6rDb>{8;m<*x$}X_pd@VV{&(3ChG|uq>qrd~ED>gZIp)A%`xFm^j(jzFT63X3%Xf zFTHu=^ni(%p4tuFKpV!{>A6L(5)6EcymxOqf8hNhVB&uPZ|=*B^!T@dKcgBH#$d{<0uZ;Ot)6unyj3$I^)?MnxZy^XuKyQiyC9_Zx&fT5Q4(%ZRMOQjfnGCZIb*tzIgXwTY48YzvY zSJ9Re(Uud@mcb2G#d#H_Id`&93G&uGV=2I}C3zL7t%gkO($ebsisHNFn7S=(>KYr% zQj?NW*+XrckkHe%e6Kh;Fo-@|LPB=$ZZ0nGPl3SK`}jyVMQSPW9&OT ze(q6Le-z*E2KMm`KDo#(8b=<%pSS=Fp-i;YS23DcvpJSM)Im4X#10C+c_S$;wVNvSs4VwWt}k^%y~BKq259~g?cMkY12#`$zERF z=7e}oChfGbQ={Oblbwx+x0{EnLKf(0Z|SudO^1z52YaaNl=foRiSytb^xSLcxwp`B z_dWF9GHDmk(Cw|=sCkYt3U5s-7!>g{UA&j0NvF(B?)Gp`%Yb#HRhW`6!fKODg344{ z3I(XSwA3WnH#f89*x1_U=GxfU=G(EcxApS!;)3Hx_wAy|WR-cHR^G|+#lOaW83_m5 zi&*Ivb4|73l%vZzds24K)_prO&!I-iW^~e7_Oqg)t=4M%g2&&Cgg%*p`7Vo7 zw-Unc78h&|V_DL7=RGxF5e~5Nr`(7z5fO#nyoz7!<&Fu-&b^nQ>Q#Mw7j9>odkGQ{ zdF~P<4!d?a2;8pg0=ur)hL|~w4-+c(AoHoCO(L6*Yb-@Ya!WjG z1$sUjJ--z_f1fpOcxGnW%?qb5+)7Hgaq~`MR%U7nskrDqRr-UIn~Nzhug7OScBHz10!7_v$yl~)M#AD`Cne)=4NfBakg$UVFtGg>hI9p+9wpA z2d!0I9jfDYBD;y}{6n@n+ZqhYPOOQrbN(z~!;1=y9d;hpjm1^f#**^FJSa8AMWt2M z<<%4fczkMkWf1}pg%{7Hl$12stBk*0mhguAqsd3k?-pQ<}$V|8#3stJ2G3E04 ziGk)7U;ifE@<^vEIL}zM@dP<)q$1 zFc7!vcaqW*E?v3=ySTmy2K-0Lb7$fB|LwOc=k(6G^)~L#j;6dD$^pyYT>V&(hO;03 z&blzB?K`utpRKzckE|NCtD8#U>LP7vZR6}ca<1&$zV*b7+XZL4pG?mKnz z4O!QJ*|7?A51jT&|UqW-wd2CTvP*MGtnnK=REUY(u5 z8ES?Gv?IeuHZ*Ej^PxT4(=<{~&s3PqYt3&uoa`*_bPAfbbjkA(UUqg)P6HMb6R4hG z5&f+Dy&B0buxp4nZInIs*wE3VMuvlHG)cC!W7jwE7zrA?;K7g1T7=sWS;qeT`v(sm z%%>l)6t;gy=b~1%=Vycb2GYTI)KugX_(L)HnF#22xNDVe{%&rbMx&>jetdm7^011F z${SMCk;0vxns7TAr>7^nr@5`I19eo8d8k)WW_UqTTUv+iAQIU~RBl^pvUg3{NL_|Ikq^PZT7&-=t4nu|nS=*y_nw@o5 z*Mx!IO^u@hSMaZ~N6{9I^=2y-g{~+ZA>O*X%+W5bbuFE0m=<6$R)MvX%EBW{h0`nB z5zhktr=`6|XOuyDDH`ywkNOz-dbi*y%;KX)O(lU%7J77}zqZ~diVkLY-^t+RkF#fK z6KDCbo!pM%>U@9S2Ytp`z6(EtZ-+qJh}I$jQOlmE_fGH$VbhKf-GAG=*kbj1L*1vn zU({Nxe{bju*jt5m`eX0^w4%EKyo2$43G?lH^dgf{`UiG2feSoZhx`u9-hbUv+o@%8 z#FpAjFB|CzNI8qY;P1ToIoqjhsy*Pj(HNju*Lxr=m4G; zU%M*`$o|n@gtC123<^a5HGE~%WgHB606VaE37y!3{`LFq^>934C$&!kJ&AFqQhF6B zSmFuyW#d7QJjupRI&higBvY}U_%mYk!VCNUajyS)tyFmyCwch(wf4P#twEM+<^JPZ zAO3Aei(HyTxr7{wQrTsZ%Fh2mDg!N^op8~!gPduAq3O(ju=sXr9`fx_D^olD5n)nV z_O7G7b(THbEPJZ{W6y`r*@PnM_#M%8MPp|uy{m*ag^iL;#D;uAlutrjEff>`-Fq)F zwPR#-_QKPOc*ilAj$3oqBRDi{|Rn8dfS6uvmM-Y zxO&ud=0g+!h+}&s6LQ*f1ryK)Q*i}z(FRe>QZL27s5CRJ7$NyPDK}G7^X`_WXWmIo z%Q6;L6oU3UF+Uf@1~oZlWjQRXAU5_^A^2n43b81cmw^U3Kkmwvxcp{=qQ8%Jb7!@w zHO#!o`KSGd57k#1lgA%8w)eYn#>VE3V`r`vgqas>*tLr|=o>2GW6KXavlyLT^~?Pj|r>8Zb_gs}t4R}p#$4IMM2FcPZWK0pAEDD%96 zLk7&88U9!p+mEZ(333r-GXJ2Um^#T*F87!;Rp5>-5fs0it5>P&&;4TXudWE`3la^r zGmo5M7`}bf8dRf^2WqTkPU_AgN9H3Ss?sh)2DT#DsuB(Zt6=bxv{?*BIP~K!OPlF!a1pnDsCF>I?6Me>m~A>BECJ8)d}*1Awv?_56=pUV4sN#7EJQN zLFVid__G&Kj}dq=e)EO1ma!j!z91~xJoH5b`XUT-m2XaQbxBfg5y11=V9dX9?RsKz zW_B8IoT+Iyf4e}OK<&Z7?Sd<+PUolec&ly@PjgyUo)O+Ez3(Uc`#3f#QHR6BMbT0O zW`gN6M?1@F&wcUlO}nqbW`f=0?fr@1DpknL1j{0y)SzeJlbYe$RR*7wK8$U2Uj5S* zPrewnSc=SEQ&z58yzj`>8r5)eN1cGZ^&7x_FOFa{uvTv7Y;5rS|5$quz$nV?{eNal zb~l^eNl!?j_g+&dB1LRqSBe#}i(<|01Q1YcpopSeMHKKVwu=ZzlPUsILr5hB2!v1q zr0)JdXEp&W_g?S!_g_e|ZD!ti-}9dLyyraUIe42?N;aB2-lkcOW^dDnd9IfCN%hay zwYai^19?^OR*_e8^u*aZ)mVE@s;gCiqO5MHRjM5^{v{=Bw*=l%-{es`caCib^|Dkv zN?KfSsE;?5gGguyj|}nliHNqrKu4KlvQ1jHZ;O(!`NqWf5{ZBZPV*PnkxjNOoWKXy zk!QygPOpd4>n-W^aC$wQUWaqv+n#?kAmC^oi#H@7miTx|&Dy(l`?dIog*z=2VtvQ1xnF)*1NfN6$|R+EsuEG6FIOqWCemNHNBq`j!9q5j0N+M4nTf>Ta& zef^m;Cr)JB0|L6Xk8jnzOL7XPh0|PENV$Xa9%&%RjTg(r>@>n2r}AS^A__`NON#Q# za&t-3hCD4V?_fb$;labK`dOxgP}0R6;^DJiB5dWZQ!M0dRc_w3KPNvr17GxW?kQfN zss8c7CyUeIIoM=5nl9K^6vhJ%^CSCqZQiuip@vwU)>CWi6yvwbpdHy>>9^jLHfa3X z3<>6u5dXy&gE8(u2K4h1e-9fvY`QZY@A6Mfxo-&*-fk*)h4vipP?!JcOjCX`-%z2r z?m!idQ!Zpj9LI7jKY^3vB-@>nBtf4$QGD!7C9&KrQ{l-vgV}*k)M>SmJ!J#|?9Pq| zNN(M)XSjcKbc=w1l+<=9De>WflwoGkG4Ht{y>BZyqS9&L2#3*2scnwQ+(6IWO3zKE z=cM97*n=W7~#;I<%^hV7Ev0P(cV zZufFss-{e)uX4Vg2$NO4bRM-OV58a#ijTEHOEsOZ@oky-_SzxxP-PEqd z=~WBgo|hEiI(^R6cnq4Aibl@putcfB4G@~51IL}F z+8V`b$u=icZQGnvltE-_mGZvni6Pwrf07L~Z13i`a%yb7pqhGDS>m^2>#h<@LffHt z-ge8-j<`x3=A;pK4Ih60V8{wSMKQ|1X@(xkrolL8M(h7>b3b98|NZyd%PD?r_falf z%1)&S2jFlaGCuI@zdrqJ>GIFM-nJ(vH?Qz;&cVGqxBalRZI0wY#rlc7asEn|9WBtl_$>ILt~=CLJ0-f zyuBMQUZ_5IrcSxwa73Osbd&tkQlaPRLl)*I&F;?=;ys-$l<{Xtp8$7-hTW zj7o}X$XV7n_1k21tDc$HZ>acLj;fw(ega1HA=Vf%-8Y$((nn|kRB>Cr5wh^krjWWmh_~5~hN^!FHriWJ&LCK*4 zAheq&$T2^VS5fa3RtIH#mI+{CZLQPRSZ6u?{U@KSJY|^!o!iBgI%>5+ED|f@XUsisy z@S{bauEdyQ2E?b-{pG}OLvqVQyWTnF z`R8Rpo#~0im~Ly~b?(bdPn5>}yq9yv^j8<}It_mdBqe5(mYUeTSCUtq!4T2DOUrIO z+O|sV*g3vK*Mx?WLSx7DZgH)cIuy zU5X20U2RkeY-Xn>wMvc-3XF`1O>UjsxpV8J_*U)WlTuS#wTVh-*SURiOafI1;$yI1 zqZ=A7Q6ur#d4rw!OJo>V5O|7c2TQ|b|@Y8?R?-BJ~Qq&`FDx$1+xN@foP*m}uiNy?p4~h4h_`{#v-|e%-v0BJ-G@s`azSI_ z2Rl}L1Q+EI)(uBe!?}v$1G^c^^y2SftcQ++n zJ2=f=v19SlFH4Wr2jY;3ZehKsoiW(LR!DEjQset;M#qkbL`hU)-PDqkQ!&z z5-FGz7atQB9~T>&km7q10B=rVO;Fdno_h4Ix95L*$Q{xR8lXj}cKNr+GNDuWnSJY5 zznHFMQ!)F=ValzJU@3zm*h-dg!lm}m=ukUmRMp<4Z#tZV5Y*>9B4K{zixNxnL~(}8 zzZsSnr%ZY7p6;P0t^2vtB%z$SKw!>w@!Ww8yH7Ow*&C{k969YB5$Ib}T6D%37HT+? zv*p0q($d=4u02AI9VlRHX%A#|jZ02Wj*AKpbR@;yH09}0QOHL?)Y&iJKWsEG$bS7F zcxqgGL8R2ReR!n#a9MVDJ!IP=GKza{LB2Jv%czzKAy$ijbYfH_6r!P#(a`uq*&p5_ zHZc~^dKgSBiH>06s0!(n7H$)Q`+Aq^7YLlg;ujp4;8!Pc!)lk=8XSUz1^6_OVsq+3 zbByo7CYcfz$_J6D2a&0+MdhbKr{^8~b=RKUzpW+g7cQJCFQPtC5vTuqDkG@O+%_ft z>1R_U-!(%B59Gmbzs}&u?VszR`hvw=3 ztYNzPd7HhgzJWp~8T7Zcjs5J|OR##9#ZiCh?B7#8U85plHj=t0i3w4ETMf0kbctkV zxe&;vzo(KsHi7Waip8IZ{9Ly9OLO(O@4kJxlF)CVuFdw}e#^jBvp~1nzpyo7JR-QXoSlHw%OQB`$Xjyk!))L8XiDU<^+qOGr`$37+ z96XR;v~w4lFpDIz(h`T2FpSib-&m@pyNwic=&%U9q0i?MSlPGkyFnkk0vK7L-tuHa z-Qe}v2B|Htq-lfIhiGtX66uzz3ek_D3S>o?^ovN53n6K%7uNhST)KnIl$JM0jd|rq zE%5F!7psj`5~(Ax?wdGm^YB*Avr05t@;G|$Hhs6AuY;k^^IVfQvu|7 z!-Z^v)SFkJ<`BQjRPOR!?s6%2IhDJd%3b#NWc44$;VfzXMN}TAzA%8CJW37goMGfi z2*0GH|K}VA0?A);jVP>9d=$zM;N`+w0b?O>bSKoo)T~8YL#7ptN0{t-@`bSDBJ>`BX%=yW>f(vI#kDseKU6{A^+wWF}LD4 zuPaC0a(C;HOWPKa<_(ddVOq{gIrdKejkdn|&mWEjhQ*v?yYqy@eyQfvu`}mu$#p-u zBpoQoGwRYZXR8KduvZA4?m>fdccKW&+ui-_GgEF!%=`9W&0~*U->H4UwCU3q?K#u( zF->`FkSqx>?PkG~{=!3d_VKT;scy^|N`zngq9;NGHIGkeif zd0T6JospGg7mk4`IJ9q-O9Z#=e<39t$UY$pyBU)8eYRpRc|2%9kR-gCN<2+fYc;;7o(o?*DvYU4|zi!(^zDhAx1uArCb8!hmiPuCPujO^T5a$t^c{_cb6uZlD2Maw(? z+L~8ma778Y>>l@>wwa&rkBn;H#dP5PLA)M|B#w4ZwfuPK;IC`Rigoz>(4i;wqcZz; zNQn;ia+)Fse6#1g)A!8o?{2x}yN^iX{^vJ6uFv33or=S|mv3bKflZq>ZCtmSgwFYj z!>6}DbwV01HX41~-t^dDa$5e0B3XkPnP<6eU~9|y>Lct`vz6CuPYr49Re!R`-1){q zx-`y#9m8q4rE_OLwI57BJ~w>x*++{$TJ-+w(?9(FsMECb<5sOhDFu<#YSye)VRvP?ta;}_Tw!M! zRX<0I+>b>^1ACm{Lm)yxVC63889HD3ZE5sb&qO4Eu+gTAerp zy@EE;Fn6Gr32G4=8AxTKz%W?bph*N8D!3aYP2>yuR*I}hB8#vpBANJY{|}ea1f{RLnU-CKZL+@iib_FufKMuR93SqR^w{7x0u%$hG1D zQZfrW!088GORvZT!K+aKj7B${+~|8Z-VLXBLtTT>=xRVgLLOo;W~UQ03pBJaeLECd z=AQ0MXtsM(UvVs4Z}YeQx+jw=x>Hb|&s*kv@a?9Jo7R+E1NYldop*mArGah%2lBLW z=38&i5d>r#Fp%OId&X<)iudIwk7L7K=+s9$@ zwJUa;)87;rTUlAYh0nh{9g@(yjh3#?RcES=-oqeLilDlB7d1sq<+lMw?yn8cKYia2 zi^^0zcvhd0$&_PRM``}Q%{COD0G6{n33lc}nr{G8J(?wT>% zx2GVmdND5ZB=K{#mAxfJ|WmSw)T``*M$%33DyldycV;6lDeVB86+&eYgwkn%Pr$c-Aad&b7teF7t3 z0Efsmlsqb`uS4dFTm}RN2SWB435BGvz61yHluyXC;z4f9|K(~Pi%R%JeNNF=) zPwS!rXeW3`cp0sSvl6}@xNdn`0Au*!9HAVBml%6XV;4Zq7(2&liYi25unRZ3*$26q zoed4s)jj`8r>bWMpEWsqjZtCYLGGHv`(Xin6mjeVn#cr2{20@`^?P^JUF#l1HHXeF zx9fzfl7DC1W~zR-!O5vOG$m#TLG@2XoACp(!^_VHlF^p8318!W(;S1L^q=W~a=WXy zZB6Gy_gSXTm_9wb&TZ79qenlLQGKrBq_5dkV+rUpYRZe_2VXNfBiN_isB6-$!FXDL zU-Vs0IvuNUzB0q`!u2e)a~ztjq-LjJLl)tgHB}Xt)EFUq^`gb5SI-P7RC zZe4V=^r-X>EJl(Uc@#gd^d0L3zfoc2=ChQRmMA~dS#IbAcr;AIs{|z*B)>+L?V`#W zVlqu{9S{@}L?4%;jFcd+~<{nvBpI!I%5CChU$qpP8V zG<48F5<6iXDucL>s$6Vy^8qD-N~FMK`r*Qa4!jsq$xV*vwum{itEL%Ne0UHjQS&Tx%pcp)oze$kR;Yxd=r)LP*Cxm5V;_n&_B&b(J&UGT$BSKD5_hqkUR z`gOq|bk;<;sKP4VVM{g5b&Z zk)mc6yN}&!Q0Bb;&oaAD*NK$2nb^&z#VOe6eQ(Rze=f++P!Fd@?fA)Ql$Ln(n${g2 zO;cVCeCp?CMzV?9ASn+gi%PR^*Vp$WeRa%M~EBy!;OUkkF<=H!64)2u+`TI?gF2M z!5Kj^4x>kk3idV&sBUQk2RD0>`iWHf`q<$x4Db)pH$fp05n;h0h$IUo0M;NjMxnBD zN(HjuCVE>;Hw-w4FcjkY|Bp@Z=qo`9UyDSKLZbU4(fyI=7?w*aRAQ4I6*fv?M9+mU z{kOe`a}OMVX2_fDd#q6S@UYQ*MH?%s>O%nq_#QOn5SuJ-=c(#S^87E>pRWMyXq?+{ z_^mkt_m8QXREO=B|#JbW#S(I)T18vt}a zC;z-ZIdtq&r*V_xwr`KqVpvnC*@(RZdm4He0`YR_#j8#qD?fRX%uR20&zRf>N@=>C zz6=$B4|jE~lzmdQ7?ue9Hdruz!a5lQ|2ILB7ICrRVNp>~>HWupbnB3!!J=kBaM zj!hP*%6~ZavYiU-LM;6XC@Uq&+gbw6KFmPqq%^rS$3PE{>ld3%rr^?FL< zzi+e2bL>r?V+%dUHh7K=`O9N%%p(uFrqP38c$|_&T<{a+Vg3lDNYBtQ7%yWtS~3e$ zn1vmfg()1J!YmAIXsoKNJb9X!+$D-vUu{C*fV+|k&E4W_0h~-!Z2)J9?CjD4ce4B{7h$qGRydUhBDFN zP|XI1H6XrY$BrG^W8?TS)q4z8ta%n$${glHoseH7#yu-;!I7dO z9e>AaNgZ7`HURXKn2C)Jcn)F3G)vk9vk*nF8F2Ts+B+NF8k$NpNzpH zBTtiB^pf!6yC3!Xu)ZR%3Lesy5OjFeC>Luu-^FUBhR%BX?frF1(xgd~5|z3GZ+Gv$ zeY=+2vr{N$T0>BW?tN~WJbBWj8~YAS@71Mq+jgyzJNNFJHfUU(*JNTzj^8(~XRl5U zQ_eQ1svYL|US(x5FrjC{s6_tV8cA$P8UMx8n z?rJtWGNV*Pf~VAMVmA-BKr%FjQqK|>0E-5m0o-g*2?Pg-xeMDs5Kw_oCHcRfC`)F> zUytE`$_BSFhJV5XsVfovM`SAG0G#1Z2jvd=vjcP?e|qTiL$2q;(Sk>kY+`Rl(i@q{ zQdV*>)@!l@Iy(O1tqo`fL zcWJ7wJa;hedTWyTbM@W-KK`k5(+S{D$eIvgXz8xvrfa^ccl=d2F70~PmcIZ9)7wmt{SN%6VjCkEYX=T%993 z8}ZmDG6H>~r{9{t|9j1NM!s`aekIGVH{=(*hrg4=-ygAFzM_ktL7r`pU#r5zv7yX`0Ew4QRk$tiBJk0(~Le9I*OK6f4z*O3O>v;Q|wK1%}x)>F=4;R;`uMyd$z7z?poJ)&S1r%b9d( zHE<@8*9RGqM90i5GpC#T9p}D_7Y0t5PC6iU64YzuL(V&Qe$$6uoO|^3%Cqu5XZPmp z&HCjs-jTEO?dxPz<98^Z@|RQ6$VPd`w1efBziKO`ck_&r@oVNSG}A`UcrzM?YBs0Y zh9G6UbgVu+@mKY(+?Ox+CHPLkD>~pNThB?MBe+M+q7Bi!@JUyDs3F|TXO8J4i^>+S>qzyzchX6<9FdjgHalL)G99V^bTcP>l;;v2oeYc45zaY@ z*XHw=)^1Q{$g{wp-hE)z;Te4s%y|Q8q42%}4B>|{xYuq!$>A+i>+MvAKFxaK*RN291|G@7s8%%kpMtyu9- zuBl)B&s{eTkh{iV+B5x`@nX}kiL?oe!P);QD5bp zJ$RXkL3WRFM%hJQ-6fqLn|S;bu1TJqIUCVJNprmNpY(*j`sxfUkH~%L&ssIzS95Rt ze#H{W+yOu?+i({af3svp{Nb2C_h7X$h8_%|*-u;H<#7jluJb8kf1h+|{aL)&lw;=9V zL1&7ek8O9h5rQli{vS>GVURn*ZNaOGAJ+2EW9Q4Na=&^n1K>`qJHfQ>$0e_24dtVo z*@F%xA^Cb7vX7aj&s*{hH)e4}S5$1mr~3Myl)RJ&o_lU;W^8PP54(CxU0#yuqEo$8 zsVOfFR$Hkp)L81e>?XhXvB&QnKbST2OSYsX?1fvXk3KpQ+xlO4O)JPSwyJjCi$?ZN zqnCbG-cX-Z($pCaL#tMq&5qHVl2QtGlQbwB)~UWLaU4}~(ZgCW& zq-uir?eR)b=Lnw|PY5F6ohTMI{6J5CK-A#rAv{RgIFww5X^N_~)XSJl4LlwYz*F1T zUtD24!JB}0AyDY@|Klli&2T;ACTPMbjNA2$o8XXI<-w;awQIqy9;Jx1K631_DO9ez z;*Y>aemM8244&~4fkr2hUzfpT)PnG8=|#s)RODt%smfIEAbmF?u`2&i(b2PRYk=9$ zbo}7YUw!gX|B<1`4^g@BQT*i(6Y_Y;d-c8pCC1n0_i2w4!T4>ETJJR1SyOTie$HU8 zpy=p9&pi9=#Dv_GUtVQMoo}sjC>SqeGL&Nw=`QtouuFoa-gUOxJEhC0=hBJ4ylyVL zsKj?C(D)(Aa7&brmDfygNte#-zzebg_b;>ODzEr2l7?P5ERx_P1I(YDV zAq@e$Z2oX}UP=Nn9o^rpJdSr_vO}4ywJIg-U4~PMhy={+81)RG0NVJA>l^#N_~^0n z0vSW2jd-3@804w`tp(vj((7wC&|Qc+)x69O#jLdof&5ge&=O_w^QB0Y3?~`nRMpnK z+texx3J=o@ApE@(dUQ-8V=H{9Js8_W#x{|$^*@|@D6a?{z>(eia!6hN%Au0AX*S?& z$2YBBsMqA?$Gg+j_3)Ipceha2>>k7l(c8U-tj4=oCm7LBr8I^)TA4dn3C5VZQ~8IM z0TJyR%5L=zYdYma6V#WU9Hc(3rYb*BTGWa%oZ(QOOv4Fc&Q(T?P)blDmCBQv|FQBD zrA353%Fb4TZ6XyyHQOC#7oi%QzpCA=`G<$K2my~C7Mc(f5f=^P#UH=nxz~8^H9g*> z42w4^@!V^?VjvDmA~3!={Wse*zg*n({4b)UCo1Sii1cxCv60l>eg*}dF~s`a*B`z% zh|$ z*KFFfzdpWa&z|x1`!&Dv;-jVfFQSqrI$;aE_js<@3+gUqm*N+O-<$sdRK;Y%r~L7A zPoIR*CoznFJmY`4<`am5q9YQQSnaT=Mxf{QX0v8I=(<1{{6Xg8bE<3!S~@HzLoY1Gk`P2B-t$2q`I|K%&z686rW_mvm!tf<|w>cWN9tIi+YuyfaueLokS zUALm*_=*+f2Uc!bcVNfL14qAITvYV==lMIoSVexTB|7@}asS||O6nz71q0MBFZU0u zHiIO;Z6(_Hn zJ~PT^T4$UlXrH5y6_%7xL4+Wbye=s%Dk?1{=!VYbf05@kQdGho$s+8Ux);HYhXFnG zm#>&AxKlb4-{u(I^#!>XYoZlmzkw`%@ zQjmfaL?Q)|NP+$Eks=C_0V~{(O#{xQ`JqW=mZ@s>DvUV8lGRJoDI1@{T#Rr~C6)vR z{#(piI*{_$)pkz~hQnzu6Yn2Nk=g}m9q+J7mcQqr*^iYKQKk$^2g0LL+X9d!@(2&Jb5h(>)dRQn=m?VZyE@eF`BLH7&EE3?3m0W6@~=WWie^C z&~Ro_U}(6)Ji7d@-kvtJN3hjhXiqcWb+CXkpiTp;A~jribqI#PtN^@85{;7b)VRIi zeAoQ86~b(q^<6v9^d1+>xOrz| zMrJSX4%5`fvdD)81E#)0L&iIl5bY1U#3>y6C8PBtqcw$Nry%dW@yQ-IOkOgk;E^MT zat`L@9ie9G;X`m3><5BQInbPg2Y%bP7brT>8puX|-Mw=c5j?njdZO;cuHYBumlQ}f z0t6#&u3}mM@2~(n1ztp4(d}x$AmgD>EDSSEpbn2Qfy^j;r`zf}XZZf>*CmbOF3=BY zlwE(sFnz(Q2`qY#u{-QasXK{Qd%=o71VimP?0~zeS()Vab)9roxvDu^XP^-G!~q}% z4?Hk>=l2Ugc>lX!OY7?Eii6}k)qaV5Cq)~o4#5sjQ75ZWO0{}}`keawOOuuL@L@Cn zjA*B}#uA6yNO=|uy-1Bw`>S`TudB1v8CV|eRBxCwE-CeDZ}m2Hs=7@5O#PVh3xm~U zU0=YLIXH)a3K+Vy$;Y`SFv-$}3K>11c*! zal8VmL+TbEJ8=p!eNw?FYI?llWL4#9Dc*46#Hp&P^JlBiovu6yk!4TMuzJRqNUhYj z0lUI{!LSB3RSQZU)E_ZOYByj`@=!=H!pQ6$85R*09u^r*1(?v#77>vlA)$dGA>klU zqoX4u&~D)o;F)7$rP4!-7SVBWEfbOwV`HM=&HTS#)ibABA>)GZ>=u6|<`MA2=1t32 zuivz3;#W%fy?cwvu*8e79r&4 zv;3~~nJg9fxikECF8^Zh;d5bahK1*Jdyuu%@5SQ4j;Ya`59O`;bnc>+v-PY*Fb`i+ z5YpvmKi%fGO!jW``M1Dla+}k3e@I&Xx^opnpUkH+*~2Vi7xQMwlcNUptSaBNPwm{b zRqgJbC6(R!51o{$gBs@HB$(8X0(=QxUV(8*b*BLP!#R@iZ^8kd={h{%H`o@)>|KrM|7Z5ahndE+1d5kqKWeU$iH2ni7&Y5ZYs?~!L z=!KkKgPitY1bQH+kw@^run))sHo*hKnuq=dkj*TAL1a=o#q$ID#;M4EEpqFJ^FQAK zs_V?TQ-<1eLH=j=?Af_(>-G&>iOmF@89wBOapT54_{4KlpMCmK|8Uh|u7w)(1YVZg zfqR5cz4x};2X#z|@@wJiQ+}a!mu>@xjT|=eW|rycjx2Sa<66o}JcdI+>K;Ev(Bg6Z z@1urd_vP#BG+gkr0o>q$iOYlSQ)wwa)GU92Gs0l*`C-X&N*yl3gQ0jtgppftLd5*x z7z4a_zS$Ye@?qI|}- zV7B*&KAld&OK0xF%y#(rw1~P;uxHPn!m6lHYjz-|Cjilv6qX7c8Lttu8aNqOR%LS^v*L5}sCc0fy0FQ^k=9$^SlA68xl$XKf`r%dv%suAG#edo-Zn{K2?1DeATJ4V`W8E)g=D= z*Vh}3myEUs>!nL13fqiYV9S8Or0BqimT}RNOdXd52IJ4)x-!cot|;**GovjLf2SVr zSDQFEJ+Y^6)S(WihpgeWaFAR%S|ShTYYb7g9eMr$i_j;0h7d>*JUA=APkDn)>RarZ>T_Ayll@sXHy=3JX^7K?wXX? zokES|Wf5yKRw;ISwuv>PXhRkPg{ZPHBgz8Z&&jFik*2&6=yqS*w(93~%in$H-EX%38au=diP|i$ zjJo`TMb) zYt}1EPEn%GH2U2B4x`8z@34POWVf839EOVrJ7`MC$js3r2lndFyG@7wL%a1Ke#0H3 z@4Ag;VV2`vD@8iRj5)gT6)UAS#Zn1MyefwB*Ri^OEC< z0)mv}rdRqEpoWLx+@xf1g#irO0A@*y`1i|Uku1ekBWN|58mziSr6m;EUh9YggoAac z;ty_utf9zRM#S~cmi$PX`$_xTd7;rU$&D-Du0Kx9LkJ09V#odhdCXHWa;Kju-}lz1 z%z;N8QSZP1{(^OvEJGf;_uhLeD^J^6ST2>EXY>c`*fBtB(Xp$tWyP1vKUsO)Zi|m^ zlX=sPx7<81B{q7?0jc(Tw*c>=NU6>Yv0vD??A>?Yjfsf~_qt%OJq}@NPNz;eT8u<- za&s{r@X-mLP$ZYdxF|0_=MWti#XVqRx^ezDfePhLxF6ZlXfc7*Y4xT~s}Ja1VM!I) zWKjm^3^&S?_QA>I_k{-V7Iie}~`J|NNc89)QMW5j0aDnslYFu^g_OE?T z0bw+5+r(0>P_SySxu(X~w}$BKV8vInx9=Hy_as;RgE#fZ!*4sA*`>?bvt7Dmp4IFO zS4l|$J~pYyYjYzacq|NHB1tEu2CpC9F{!(I^(U9x7*892=iOpgfRP2LNl;B$R)Sni z9>9s3u~n2gI?a8H!QLhoZ~~}?dWS7`Z`C99{%wnFZ)w9Qsq!8^<>UN2iAnnw-t$_P zk{2Denojmzm6e+ndnZbQ7Gntsnn>V8Syp02Seqa^H9Ph_CJ*o+!3SOaiDza@kN&B( zG_>dlGx|uL-dZ>{U7V0gEOel<$J8U@`|~^w!otQnlCrEn87rT3ba6XF*VU ziH+F%nBTO#skN26?42q**_F+jorV^bF}KS~_11EB10A9?YIdPK31zB>hUl%m^2wFf z{%QPCJb-@*Qi*36&!K1h*=bovT&P?A>uKvxkg+}r;7G<_S=&7RWb-ioLCJldzB$Bu zNXDJ|_!oP|znJky^ORv~p_=vaXY6JC**MXqC|e2Ai2D;?;pI;*>zQzLWhAc$fU$F zP&IqwsYPpcpRWzswUt~2#o{oYv{+8^M6>%wkIv}bxpVZ0+m#1k_H-EIE?kHcicdRm zYBX6flb_ulp&C8+kSCfQl(VESZ=QT@l)Az*hc8&Xm>c&=)CWQ5wX7G#5Q>M0q6 zrvl;^JdPTg~PMuC18ENG%UYZF=RowxHdbabC?x4ujRiJw5GEJ4a|nyji#@ z4Qa|k&7fB5u1eK7L(9T1aTc1J6PnRgBhk6@aPaC9B{rh?EB^u~q=({5|FqUMAK`_b zG_w+U?9q$zy%KikbhQyDWvskbn|JG=(3M!OlE5*c923gwN@# z(+0CWd&<4QorojwF5~huVjsg2tNi3tZMJEHNjqAjtXD^Zg zs;Z0B-s<_Gb!l8_Fs3Lsy@5! z!={-OX|Y7=Gb!5MyZfluO1}WKur6|8F<_!viw(08$>o(diI0WT)^(UUT zNe+5*zqMerS|CMs*(~cnqM;q+SWuKiKiMzU7f)|CRG-@Vi{7-zg#7Y06{k;!Xm)gR z*sVbkVYiOdo8~I^Pr8`oZOzl{vi;S6L@$%whyUMmpR(uGe?%=`^+A^4pu{%IM7jqe zMVD=}3zzEgt=7WZ12+5gnd;MLD=RDO8fr*EItQg3Orpn5oIH*9@9de=Cn4ECb?QXL zsgoy9o~}54O7q@d8ITYY*|>krVpzarN<8cT_S*I9VN%`q+i%;pZQr0Qc0`REcYXJk z!57QR%P$7E?0)^Yn3#mbq&98Zq$I}2Xc*(V9P7VCIT?*!Pe1;^gomDcNf0G-lxe2t z9+~_!s0K|r8&Fr*XoNT0+t}!10ZZVe#Psgfs~0~#d-Ukhy+`++z0wEv=-GqnD1CbO z&H#%Wb>>XP(fmWZw{PFR`%wPTiZg^|%JO$Z0e`5lLW_wAYk}js1xdw0zCJ#|E#ks` z>~@PiG(I6A0d|$Bu3g)urluxP#{jPC)OKCFcW>V+5l>Zos}9|-(oAB9{~2Y}NA|gf zMI!SJe~gECF&527c_AZI>QDd=zsdmt(0yjB1t`U(y!x7@A=_|@>MPkE>9TO-AJMoib`#gA zEYhmMp9C+W&N@T~d%bYAzxYe>*%G=*N>llW^h&P*|3Mi z)xvWo+R|d<&fiL*7xqr>*RNlsNvVUQXaCxNP(|@`O^Is#_!E=vzWEl`*FJ+sUFTyD zh)GWB*neDhYTMzYGfD9pEM>SygH@|)4O=H zKO&>bswRkW6K-aIpsIe#$bE zwWuiSuJ!5eG46@_-#c`eGV6MOhuVLfLm8(jL-AuRBGI%Qb8ivprt2~E9&wHn|VIyG4JQdX$JJYS)DA!pOE zjJIy>IonrUCTs)QLy&I%{^mRnwlI7BGq2GNM8d_DwZpBu9_T?w7F%R6xI&CRVNhvOV@Il$C z6dyZUz7>42+ihm65*Qv7mfG}6pZ|f#li-qNCr@ccehiB3z=8by{RfH)(V^KShG*O6 z&Fk6oyRDQbUdY*Zr46BQ~hiX|`8qb!hj_(nyOW zag&>(yrQL}?&hoK&ZR5sDb2p|Ffdw8Qd#-YY!%R`>Q~>v*S&DCX|~A7pdkJFZJOWu zeGBByFSh`u!sa{IbivSUGhb=j_QP+J{JD3{_)BKyFSOu2S|E4-o$`_Khuz%0X3Si@ zI#aWqthjvlnm)XmEh_Tz{Wrb!zhD1@Z+30;f4Kgy`t{S-u1){{;rdrb{__3L z;Qqf(^xXf}fdjYd_wO0~@05kce|fH-xpuAG|CQ0deE(O*|9AJ_tmnh7nwz$QB8Vjb zAmOIA;J?D%A$q&Ib ziqdRHi%QE$>?+4~Yr&m|LqoKC5NX_GX@w~%SH0^Q2f^@&j*$4$Wekr*kL98J*x0tC za|Er|P>;)iB@sS7qQpQwwBW>0NV@iIrx?q=Bh^Zal-501c>}v|`+oJ7ojZ1H*_Jlw znqkBHcId1H?y4!Jq(A`ipOR?zK5DS>jFo zxf97tjur13u22w5dHI+#+$$@#SU18YgW~q47cp(vF1X!JsAGr?Xc%d%tVfE|R zw_oR&V2jgWwAvi%Z@+0?zG1CKJoX~=g--RELFu@!UJvpQ?|$u&q3M144I0p?&$5ik0L$V>2I?y6jL4xTl2>V6gztLa$d}q&yB?s!|fzp;#TtIj34zGmYBY zb&#bCfhoFPeZL;$yuxo0vCK6*q__o5Bt5d=iKr9eR~fVm#$m(>9pxX za7aInGdi1};X3K=vkpgmIX)8ij&A1f3_Z_zo^9i%m@BS!Ft^|+i`7~4qCGAS`S~b1WuRn(CKY#>H zL4w90L1Va7TWCzMq40U+eX9=qvhRHZsqY4ZvaeUKeaij!XSWEh+xoU6YsRca2fX?}{rGpE?mlrYddTp> zBYL(N40%qGGJ#}|Unr^M<&}AcR3J;ym)=V~W1kX19gc z1WS`$T}W7DR8$ZSXu*u|h)9f&0>~8>79STG1yf}d$?1RhtRCGVmRNrz@6PzkyFrVW z;s*)`&kB4GZH=>Mzkx*>l%<@=4i4gNw#yQalt22e+1~b=!5L-S_iQaKE^r8*>0wi_ z|A&jd_3a(jSh=Cpua)^!-im#V*^0Hjz3k*^?=IIoltC)SY-NTzY0yZ*8rzxbOO-E_ ztxA<@hOX#3br?Xs)`3?}dj5qpB`ffm7Y6lh-#V$=z@ax!dVcDow>rQ?-ty|}3*Y~6zBX+*pWjYS z!^_|ppL9QHxsxlnCgfK@~Y@X+JAbkJsDs*-h_$dAG0uzCc5 z>7e-!xcS}*_uMgd)Ue^5`u7{quba5>*n0a-x;@?3YBR)j=z7B)1KOr)Szd4xyQUTj!NI3a)dUSp8|Zh=Rk&f}hTo3ZYg9;_hA~J_fUcd$px!~I+eBW!V0ibPw}$fa z9{x?>XQG6G?XSJ{_PlrBa?E_?jX6#^xfcP4(`TJwfu@oYlgY21t>}f)Y{kW#t{X(LnqMqNK~PoKaSwgDhrZlH6c!*tB=bTChw4fN*&?N-Fi*=u{&9pmF$^#Y z(fqir#pZP|88rX99~ghf9XAgj*eiX=kde1ucW)f870& z>8*M1E}Y(Ux*8{zkIAkkCPX+))UhZ$?ocBWv?vU71VogijxQC{7Iy=Yz~bUk4wZ<5 z%gH_gTb+g9q03971NVUz1v@%Y;RlXGhn(v37hxfvBrxPZAF4}7Dm~kgp6y4^rqZ*i z^lXTPd?k^Rsfq?iKuDn|(q`bmLO>n(X;7jno5pM)SG^Wwb_%m^iHc%qCieAdw#;EfU4%cECtv6#M{QEM^9aInzbY z^`lq(1OD_ik4$B9=MuNLkvnhBu8o6Kt)%EE*-_{zPznMoVH?GD842kR(Nv5Bqah$2 z9U^WJXs^CdVQi>c+<-eCzHj_JcicUqb=y|Yzxcpi!wIhV_(XQ>Gi<`tpoS~6{fn5fsrM0tER7*9S4SOr8e4uKkS6+Q3D@!VEykMF>YxcXdWVN}Mx%Lu8 z5$7)d@|*8Jd2Ln(ef1RZi+S(NnFNVm$pWJ1SWt5@jKy6+C2#D5FKFOKb%ezNRT%jGd6}u@l@* zf5MWq%*Ec%K&Oi`Z6vYz#WW==J~{Q7(PPG3KYG-#8%Cs!x%;kLhYe2cFz(5z4~^{K z9`H;}tY$cFG6}kr}rdd#LLTIat0Tc zq;!t6K)2sxS1vHLvj(rIR4q}=ABKvDtP@8V3syu{cb4>~xnIZvob6fJRjLQ#@{{>b z-U;I&>!nfE68%{gWGxrb4+^j|9R5-P7Cpd3(A9wde?N<7MuadUy2QgJ9d86npz>p| zdy6G1u4p>h8?S<3tYk#8&CF2Q6!-)M+;?A6x9&rF_3GHAT?RYOY|1{oH17ITO))uG zEM1VI8f^(o(ry_U8F8R6-kK}AV7&E>`SVvSn7d&9g2B8LcD;$xp~`Hr2o={fwf?-r ze7-tAFTY&##t9&ZDCwANGqWD+(3`!z;GE<(o8QwVOz_1cxR3j|k0yLEs29&&7`YMJ zqYJib2;pW3ND-1^R!BrjZmw{;s1+izGz0Xpk`xj4@FQ(o_3C>~m&B;J$VZ-f_JMIX zw{J6Gz=*D`ViTjEn)=d%6UKJLIhpO{BGcYzigWn3=*qpuC@zW%84Ym`a|^lm7$Q4N z>X%3W8sUDOzmK>dQ>QuEavC(vIgBmZ4g6eBM?8phWih>hVRK%eJ6CkvaHLu=ubjWCd|zohz!hx5Lj0ZqLWpX2%a=vstaOWVA9?&o|)B%&yl+O$qeFh9BK;{4z|2f3D%QREnM%wgfX}S)OH9;Dbt+;2{X`HSE zplt|fe$&&+7+RUq)MqhBK@5@`q)V;{ZnNalc}=1_JE)=VhMRoNp<#(_Q!HMI$z55q0F4?qp-G-IxH*VRoC1>5b^-bctcFo!~C9{Hb*%i@k zmR-8ENqA>VTRj`EX5poCn?LllyEp9~OuKKQcYD)ry&}DsVLX_2d15U#y_7DOl~ZF6 z!8ZYoi5aiyyC_Xoj%@Q|ow^R4@InUG?LU<-LT(>>`|bBU(kbVNrR{)Q@4q)GAu=kq zgBG)YKaBULy0ZQIcWvIhdH??XHM~?S_wWBrUZR7hm=-L2Z}#*I4BV&PPnec2|KK$Y z{^#5e80IgSDenTEHN7%z$uj&oFY}d$wHUmk7!g)$U7bB7SnA%&OPj5(&QD%yF=EE? zV49bTDc|%`hFI1`?ntySpgK|3n!Zs2)kPXe3^FNmpKc7nkt++Iyc98zrSRV$=;__p z=-s*W?p%8JHTvu|q+%dur5Jd+DTH-IYYHI2mcB)p^~Hgf%8u2uG8dKb9cerEI=k4S zqn4wHlMIF|8q!J!P_StFUmo=UXWe)*c$-y7JQ@5kOgGl?4%-shmpHAGQwnK4)n*6Y zaQ&EJJzLgY7(A?ZQbbL|g&`wCC?tNt=pWPj&Km}H={I=vb=S4?i|aLf)W{LTGrG0& zfdsW}@||~09N4FA?Zujl7ChXk9Rt}Nv~RfD?@?ztCLRxWm3TaU%Wb^u@W4)QbXA|L zafW&6Tjoo$We)OenfD*D*U_fM3l}W-$JaZ5ef`Z(zgxX^+b@fk?D-|X;ArmV9Ur~D zWciO@eEG#!-~6y^^_L6h&6_)S&inuRVaLwxufIKS{yX!QEnBy3+iuI=?W@pN{<*=Zl~?6f08&tIZV41tW8 zZkS`MGf+p5m7yRQ7u~M%ctNeuycK;O%H z!ITo6tvgTnt;vF_%vSW8z#WVt@sk*g+Aj`xI<`*j}4iv-^qHGaKJ#C=Lqr&!` zm++AU`0^dG(folsgPicQ_@NI(hW~$PjBW+ZM9$wv&X*zQGm-O|BIk6LZeWUr5;Qc-`o+&twuudV_zjTJ9NpZE#pvi%hL`LtX3@?_QIX8>>ai_vRWK z&}6z>Pf^WY4q1BPx(r}Tzj5E0tM~Nk3_PqaLc*)P{pDBb_%@#Qf90JImw)lqj90!{ zv1!AmpO$_p`_t#WXU+Lw&YN$%J@13>W)0zhshD(mDObzBw6uokQLV#lRTC4V{T(K@ zb>c3FNzeu&_4?LG)-n-%9BL#%l#1M6CHTCMr9o_4`3j$)pZHZ+XcBZ}iDm{I#@Nv( z8ku&U2U}WpQ+D3IEcK`~jg`eLdk?n%S2-UBBJo7k$UY#dCPI&?5Q4`_WM1Esc-)8* zaKlR>HT&c%eBv($@TN(?n8E+=4e@C9aY+1hBz`s$KMsi>hs1Xl)DucV{^Kgu#a{ZQ z$sz3#ev$sG9n!c6FKcn}NxPeT()tEWH2GCNY16g0*87ajgvb3|me@CpFFu&jKLdav zgh-Z$?i!Fja6o#$K}^KCeEzyJ*LugOK?C~?AA8s2VQiJ&=4&&Q#|W3q)=%)R`ov9| z;Sih0KYOfenmyLW>H;MD_a18~$BeJ9T=Mo{wy5IPoWK@hsycUGCgfe9mfSBOg0H{* z+I)<^$@+(r-H%x2zq9nSWs6^bQ%XhNN2>JfX<38VbRftg)T(oVX1L(x<@mGj`h2tR z+F}j(lkd8-V8iH6`OliEi>T<6CI_}hP(@O)j?gPIDu~xrT5lRYY_-XUt#9*14PWKM z?kq4fQMS6i^@yIn{CgbP9sv{yyxNidzg)m0P1hq$w;)Y-BTdaLUMbL_;}FVY%@+iL zt~5oOkSBRhEOG9)AYYUHvcpS6g{I@wmPw_jXQK)g$;*SS15-dt7~ML(f9p0~+PClC zXV}==?znkKpQP|Mna|(WYDA{Du_2|?@Ru@ibUXPsTbZh*46z(qv#%<`R}4INP&fs! zVtNef>ObZ+v#1ygbWt~GH2qb5Xs$e2o8PL@Ym8O7oBw@>I9Ly1n^!x-Mr7&b?z z+WgHil(GEFaSRnr`TKE{HSd2KOViY_TgOFeZT{wXS`q>-{hcul`O}!bGi}-%e;C*3 ze><)v=j%>tZT{xi${7FUaD@JOd}TG!+WgHima+byPV5=$iH!AqjP=Wm^+cxKM6`b= z5knD9+y}Z&LNn=tipCeEFArucpBw#KvTV#*T|f!UNZ`!DXv@ys8DIdc!?Xo# z;xgzwN)FY~sC63NFQcDtRO__s9)04eC!=EGdk!Bta?IFUZ@X#a$U%dz9ZHB{Nc+@l z?|JB!P6N8JmvO~voq$95AKtZQ&6>OubuL$x!VbN@u?}}8{C+Nr-?7t|>VI0hB`3Z! zAoO6Rrf98Zo&DzA`AZkgc*7y}J)c6QKX3kY>9Q|Ant}Vs^xA8$y^0^{Wy6d)6nvb5 zRpLOWzigQH>3dRcYzpdqvSr%*yk)|LvP%rR?V)F5e3d4*lP+pn0 z@#!q@ubHnI(58beP9lgwu}9CkR7aT6*EgYm|Ned#?R5l5ekWXn0tzK&KGMoeZTgnk3Nv?^-B!?oH?H_d;e=sqeApX1)V*6ipHKx zIS~*aA0Ob2^7{YSd(XhAs;!UvOvy~zr1t_LgoGYK2L+N)r1vJGSG*SN_1aJ$CMTgN z3IZb71yn>pKv0lgRGOhffDl3|Nk}J!l$rOp&IC~5dhheRAKowTnPle7nVECW-h1t} zSN*RBof?5fnTCm0B)&&9A$%Fxd92PN5z4F(lT#zWZ^EKQxpB5D4aAmU%$udN+0_V) z$#7)RWW?==_JaAsH$W5}m^6p+ME>mwv_6-b>7n#_D1CmTf}%yU#YjE?v+-hviopn3 zbPmuO52oH?~}dC~czuXlXC?1!Qgd{pdQ?Qhi) zQf1MME)v8KMP_GlG6Y-Q9E-zNC<+;no<27YYxiX4o?N*nSMJHopqilr17nS2$?$-) zaf(XJc8X!){$CfXsCy{}dP_c{&G?kGGq`XD?+7FR2C^qK9A_hoJhq8qHs)>eH3bd2XU!Xu1YB%TlK3=DFdg z(M*4GH;zLZ1&M?*h5mhktKX1N+S4$c!HCBcc+z&^x^jdOgS5ry3!4)888_&86_zTZ zrY24$yEA3sPW&OGKvQCiQne&m6VV^zB7^#jdj4OpKhrnB^}z@G#Q>2S*|STparX=y zO%X(jeDM?*_^*FG{rpoeKBl_H#+H?pn?t(cqfACT3kYt}BD!6>jpDX>x$)HE_&~Mc0K04y=0F1#oDh|tzNhG zn{U1ugbn%);L-@b;PEO6jQPa&)m$ZxKQq~{zPk8oadBaBadv)s`pM(zl}@Gxo7JhV zt};D6j}RFs0EUKFQ=?>wj1d4wS*TzpV#|i*F-LIfT*}izLXkr?**n=?a-u~hVMEt4 zAf%G*%Gz$LvlAl29Dvc`>lEI>uDI5D5tY>fRcS^Y*q;S`BjA5=Vr{&8GTwq4)Q|D* z$$aX`d9(bWbQw8jaL>-M z6cp?^wD*MjUi{aqkByixWb%XxP!xg&5bTO&&1Ucp_VX;b>>k+ep27DFR|>ACD2ssb za+#Yv>$7jRZ29@XMyS{_f8K(3=FXeHXyvT=%h#+?gHx+3E0F%L7L+;#`1c5RD;K<+ z(@KxduE`;grx=6?Z^CnIXoz`!o=(m^YH?fJ2;-sVKw-_;QNM~tM2}3SJO~<95W7Hu7Hkl~d zmsMx_PuMbsIW@{q2Kw%RU|f zeLhWWR}Y~3I)mmXW}QL6V-JObUpJ)8NkVh(%;PRR^=~MjsWJ6>@6guWI<@ejsFjC{ zOIsYv?V1NvUOt~*V#h7qSg+StQO+<=jrnl#mn#>(w*&;04Zr-lfA)LJ)^6LmXWw^+ ze%rZY=iXJzKmYu*BdM#t`|?9IMpsrMWm2;*)YR705mhV5)}?^Wni?Bi@if)b>n<*? zE>3EU*zROlG`SasA&i0{%E^akz(m+&a$LNmI`bR;kZM)bvIdzD+J+ejN=p{2d!rfy zUa1IX;s|1_ct>0XBj*<=4mbwYAstve|GU#^;~Bwt-ta2kib{g6pnk|+Ryc?s7lv4# zL`^j4l0Eekr{0&?txH@~r%s(ZhXUkjHF{9m4h{3}KCB}xJXFs^OO}4PbnTCOx9tDr zm%TfFIiOlCdh)o4Y)Tis@2W&3G1(wms8!ZLXco7ESSf8N<&E}aQna)+kd~s1GTZ<1 zxntFkx(m%C99diBnu2Ff)|O063CuN1xgCg4BmfnrqgaAlb#LD)YlLil;G=Ai`!GqdI zCr)_s-hKmm`_)-}%pos@`S~Hp`v-+abV~Tg^V45O`>EGa=2*(Sx%#?YDX2i#`TmE| z!+Q1Z9qrm!Te!n!`*|B);a~PwiKXQ97bFeMpZm!dpRf3I;rv+xSz8c@nWnFy;Jm#G z1?OcZ>ObsHJHP+Y%AXD%J8^zny!|Ejk5+#5=@(yoymZl$ufP8=Q43XjN}U<2b9S?s z6)+vkv$Gp4?z)HweW|emoVKiLi+c;gHafMBX&nXSQp5>?~r=1L%qh$X7WXp~{?;brW&Zm8Rm;Nuk`o<&O?(NzpC^)=B>uzyQAgk3*fylVa0bw6(W{PXwvYo__9?T@=A zzqjz`-8<(kcsF^L8cQH~b!7t=bZ-XeWSh2a!s?4CCr+bEb;O&Oms82}-03r?&)BGg zS5i)+=o1($(@2;rv7L0^B~2q1NrYoSZF58f5sl?sQe$Kx6ah^_bNOQv*Ah!6sxhl1 zA0xi9ClPHTov5*deZqB#l60tFh#f@hiaIPp3CFxJz(X&5uCXo8|HCyj%9sNGsDYo^ zA`Hl*I{x(?+RIqVVyQP{w~!rP^o0MJ9S6)D;hm&7bF4k|0rm5QeH3dba|(bmUM0Xr zXp4=7N1oV259TT~QuOGi2SLn^@Cb=9WT3@zDt)CW10>cksox``dq=i>bo!Ku;|8{S z`l+#fqTSs`jvdgyZ-?k7pBUD?g^TO(QGNT8rvLo2WA2La@ESdCQ2#!0ecyWP!I9m< z!ykHNf?@oq_)!T8aFcGYOdB2-_peu;eEgqd#!FhqQn%M09@V4Aj5nWo`mxDl5|ky% zBG;FvjqK9xwHZ%7{_q5~pO25ew5(L`;}cA+$xuH(eQ8Ol-p>#4*Psx0_r}7)!bW#@ z@Z(t#ee^|zMFh%%66;5*qVDR|t2&=ppI~yjLcP3feAVXV1vs-`z|9}Zii)@J4?hf3 zl@ZYHbPV}VYew5QvFZp!E}}J$=i0n$O`11<;j)hwzW@Ha@4UTu{`?O=`fT~~&p!WX z>EcE2q|ADo&z3G+v~d2sd9O5`=NW$c8fSS^WE{dlUPgF%j$n{6#Hx-*xS2{=$TTg4 zUBcMMb9^X5&LEPHM)R$Ec}zm9l!_oDIW~^xk-Wc`P-*$y{rp~jH_1L;)$BczS*FRJtVldzX2z(DmW4HnSBY%WM0+Nf z@FR^U%`^$2a8j{eG(D`sp_1f;OebcO7@#?#w0ErREbrKxu`V+Z;J5KA@lZ0)*hEt0 ziM{{LnYEcD6>i2eliD+rJh_~wqPMs!X5<3q3Q;|JgS$3GsKqCf;nF-MX277XwK+vC zVwyF$yVTSuExTl2zSe!{m7Lm4Xa7qVyR>cABDP=5^@2VURg)AMot}9$B(m$VV_5-F z8R=KZb7^SkXsqtryJvLXiS7Z;naBK8CFanfV`q-#Uc6jZtCSX(+VV+2EIX7+e8;*q zTlehxVdcurKWzPe^R}((ckWsjQ!`IG^YT1*tzW2oCq3XjPivErHU-lrciLp6O-7g( zU8&2+v?}_q1$+NTWsSah^UqXsB%5#nRWTo0x9;mz`*s~t-_d1dT6I2NO_q%q&enr- z#IXv+aJCp^d4JfX@CDXcmK`GTSbyN6z^>xCF=#q23+b!`Wu-F9>y5|n9o)|&v!GoeOXi6W8-7J5U zc_ESjnZFW6CpHHULQwwvHLVW?Vf}`!s0qDDY(+3tB(O{1%0RbHm4gOs-o8O~>XUik zKna1HPHu|H)77H#8XXO)mH|QI@=RH34TuBY+TP8x`!PadiMn}qg+~&_8bzX*rR%Lg`QmD zW^^cD(^h<%;VP>iwkdvp2^)}*2RZDXshRD1_U$|J(FX=8Gx5&OQL@N1JfqAq4(Z_U z=`AD10JVvUifYxWecQGHwZEsP9zAhNRVHX5wr}wKG}_i9Fc=A|9|OagObp~WSkh#K zzK677eMU`YG-Tl3a~0tQ>lTf^Y$d^!d9S*Px`K5J@yYwfF-Mz5(x%6t#1trTB$PN3 zN^A#kmW)2(B3TZYO4Z`i|=r%XuzUSOm9{ZGB}?6Xfi^1$RNBl^WB5Roz8{f)=(>qfPaz}AVQ{z3Ti zET_KRM-S~QnMg=xzq%Ff9HO%~3LoW!BLJW(}}c~W$C!)ysS7tOPw!Ri8A zfs{lpn%xfYF)_7z`C2?2Ae+LdWv+p%1L`c?T8s&-q5((E(?%S&Rdb*okecE;C6z=b; zxMdwXaVn3xJusG8*3>kT|EsyHnN}hMvY{I@pa1{1LNUJA^`K(wua}qAQZ%5hp{}OdF|PX< z&8=G**G*e@GRBPSuO$YZy>RdEXOH}{cJ->C_v}ANZn!KME(#;cxV8!QWQ-Ztj-fE< z7TpKj)jTpfx^;)RxQR|hCxhTo5r;>v!fUE-W96v%2SlJ6Zes!lsdi|4P~~ zbqah8i*?Sy{qT0?^xcssFQ%}*;OI@?$4l0OlA zeSL~X@i%&=*B^W4b}IN}30HZiF zUdfT}`750`y61LUnQ%}1{h)QzJ%6Pat$T!KeE&={{$9EuC0e@YuXIEAI5>jzk4SVq zn);{nohfI>p0758>~tr}*|TTEow(q^LkTQo&p|hqJJOBa?t3Se*gh&=NzpwI>3n%d z(rMMI_D-bJx8Ky^)Jwk;=`=K4yA$bTU8u-Y@~m3!_MdrWy%g#Do8GyddNyoc_ALcJ z??gRX@7zv3$4?}%XunGD_)G8HMmt*X+)g_K?oJ?7O(9>fDCF*;3_2c_NZT&JY8jL6A~ii)aAc2K@cOTfA; zDp9JfZEoj8AAPiJ+0vy;mn~cN0o#%fmMmGq>qj4{p&@}j{^8ww-qkw7-z}i{!o^GF z6~(nqPF0y(H|njH^6S;b1*J+v+l+JBO0Kod?X1eh#ji#C_U)tG-LFxh?&#hvTej>w zapJ0r+B`nKfB(Tl2M-=PBChw~sjs~J(%3g2fB32Ar%xL*+~_gx$v1`|{LUtjOC{{@ z3o0n)gNg7lgOFxzb31qQ_6}}|VBFTX2x%rc(PVMZiwlf%kx+aae_ zmjY2!T2)*^Ew7xM!jowk*_6{Muh8o&3j~wzcQ)h$V2~&asDft{`cc-ug2L|t8;PHqr=z$PSXGVIRBNM z|NC(+EdrCXxRi35IVGhfxtA&`{(hYQI(h$ooVzlg2Syk{zAz|lkRk?0XjLPz8zR1l zCmQ*Fq>`)T@h{J((FRXz&&S!z0f}uLuwgg0{#wu?_#ZR9z4#$6MbT-+5Xrm2y1R7` z=*9!wO$kWlEoRjcL#+G2mlG&EYu7zH#cN=A&bFL^j^}OLzWGMcPQrblwaO+ZaXGqi zYh7I1E&<_`AozgX#qs68PmNb*di|E?6X%oXc-HG9BbAZ1I9n(FCs;Z|EYo5QjenVHn{-o`K&^wklcjRA_bRFL(2s#~6br8H2uz!DA8n za1tA&-O_UTmv`J49Vsya`Y_DMt}d>alr3cWVg%lG0o=C_46+3uvGbI~wIa_Hq_>cR2`wp2nZQ8VZ6B63Mri?WJ$JUz7 zE&8fT`xwv4>bmAVMm_!bqy21ohkid^?GZg7zE8lIK0V_H-rcQLL|Rr%ztL0gn>gvQ zmtIIvwkaF+&nk~wV^dR$6#p(=y7((aAWhxcw^)3Gn>A||>E+dCJP*_+PYHwtR4jtOGZq=$)>-O!#Pun^ zj5}8f3aV9|&YvsK)vlz|fjRE&+9E&+;7Zr8H#S{KX#U^ea@Pr+1FdZZ4vwehO%yMv z(qT}1XDD9k7j)AyKL0@RZdCkB{XO;6nbXIQ9d|o-_Vg+0|7p+oZE<7B!yMA*x0UTi zrfoJ4ZniaB{yi+r4S6tS+&TMp%W$Hsc+H@&zO7hNqY zFDs-(Q%SMgAL&F1G+rwxE^V}gM@L2Y)fED3S84DI4-Irv12=8jahOWpm(LyEvFVTd zP5mJ`xIVNlg4W&0!F5Zeh+{F&p&Po$}C)G|DC>LhTuG)u*>NOSb10Uv_TZy4Nc zOBRS(GTEgG!#lVnPD@l)Ac8z$-nQ*>pFV*N+uX7)ojZU2a?X|F;;Z0S1IUVL&tMQD zSKXYimsei1S%A}x3{u^8?AUei1Xnt7aMzBe`wQUy0-y|u?z@2$>YQD8B{%bYIj4MRKXCulM5RipR_5uO@88E& zTX5aHywD)k`RJh|r%zq_c`rvpyh1_(n>7mzQJpXR^26J6Q@0w7s`GU)r9gqJzJBXo zVi-5^1xUZNiqN|fIqX=!5Z}zM1l<}9nA2T$?c6SRwtdGg#p^9xvFmu>z1Lam4Q zw(Y;1RK4nnhRVx3M$pu@w7k6H8gvWyi9Ot_YX=XRGVPIPUVI@@*{|%yI{uRCb@5`M z-peaAA|j$0CNzD~Mb+hc6=8Mt4TOle1c!u$HK%N0XfVnB!vD2ZQ0H1%THjDvSyo(p zz1Ec|yld6fRn^yPt2Hho>&e};qc`oin|9odeivF?QID^tx+p*E+$m5tujJ%l6unwj zT3u6Hq=ue8ee{<-+c&LSw{Fw+J--}1ee~%5-9p`mgSs!%NnztDk>hu%vEN?A;5<=iqR0;o*Jt=TpJsVQ~a*0ZeaipCE5% zvq|q{5t&yF1-OG+lO@8GbaPtG-h$zP^P)r~+6^^=5Yu)iTlS zWAbc%-Pm@*^G{Eis8W5UCKnVrq;Nywg)G$qL|kRT)xyH7#buRnS4(lRV&+QH;}Q&I z@oMCLORH+Q1zSVCYH4h|Qo!;AT3XGuGHT3eqc6O*6L&7T=fYd1s<3OjcI`WKXy36@ zr%th|fJ8XKEcnx@L+g&#ovu4kcas0q`pEiz^-tG7SO2&=RCiq;rgzbs_3n(^dEF)6 z2kEY>12K)SBRBaAFj!X@ADc90+;{>3QxELfdoXqPo@4o{D=DRm7A;=5V8Md-RaYG5 z^#TKcpImi4b_{3F(cceKTJq1I{9o?7VOPU14SO5*@xQatr!lti{>BFz@BRPaz6BAw z4YN(q3vMDpJBS`XJI-~WTrOvttE;a!P!Q<4;&kHhW%s`05}EyLoYp&x(Rw_}BbuRTrxn}EtZqsA+#&ZABKoq{8}4ZJsjgtS!E zjK?)cvStCcG#ja)WH#d|1(Dg)>*f)yhoiXDwopc2C_}1sN_4PqNeOl5ON#S~OG@y3 zXBQR~p1TS=gk}v&<81vkA{TwL&;62GT5;{d?>RZEHZNPW?4wV%DPG}-EhjfH3D@MDAhrfSI z)PBD}zyzGoqd|*LF?tzI29G9NS9{vrn>M$n&4i=yk0&7`834x5+}+FSQ^SR%Y9VS{EdeX9LhO29Cy}er2o6^JvB$xB)m0!>@DoE z_D5A;SJ$%alB#;8NMJ9j*FSSG(2cXA~;#UV0C} za7^wYUe&7agMBUTuC|P!Pfi#+cIy4ppMQD!(~rw{uT@+=eOB@DHyh{(FQ2N0>L!k0 zq7QE9B!cr~q7QHYDbyV^;d>_O7h+QtzP<-Pu&RKLqS1@ssRe*2Qb8>4-e0G8{6x>YNvEv8Gx=lmMU zm{03Au2J1mi;7a`7ZuHySBiiDW{WEffiEXVbrWA*T`4KL*E^~@XA@dMXJx&rE2~yh*c`vO zk#XS1FK%QUm~T`LAajL4B?4yks(Hq17LVTTEMcm-p|G%FXkp<{)%?M-OBt(vxO7G} z>mwrcnGq40s+rhZf(j4-XPQ{zh$*$HXgCBfN+!mwV~(p!<=+7jMrCxw&C0*NKAO-C zE|X-*-$<(~x15IAGB&R_b$T#5L9-RYO#_H0@|E8hN^^0^vbRaL?A zUtlo0ntY?8qFTpx?b-z-=Pi)Ia!ypwEz{1xE-Vw;?vRV zB}Q8O)x;D?uryUxy?yWWJx9+?r{|=GI@CX%@s88FhLu7(i!e~>uS*hGA))c- z($A%xOOsu~)_|ZrcaCpmsc|(mwNmI+s=X3V%vMvaZB;durNkRwtCV+DEIG*Aby~En z8aJ}0tj;gGP21>ZLC(c_clYbs%qO%(bE@hkju;%Fr0@Oh_v*l&V;+96Pe{#$OIhyi zM?9{E{f+;Ft4tSq&61Q!eM zISGK{$!e=31di*gYs<3J(y}VAt@?Ju&I1RIU(Crq_xlGQpRElZanC^a{NIRff8Ax~ z_6%EGyO@$gUxQtuK0-;deXJ6jl;x})Ut<0K5M^T#0^|4Mm`DEDp{*pT)Nwegr1Q9- z>?R92T-Tc_=e>0iILw#>zbg;Z^Ai)1viqxX+F-I~%J32|DkDtziUiwU z%FNHnzmk7Nc8RV8yf#0dZ)K=)0`~)yuT%sk>w(RMrHjqY#X^`L@}~Ai7GDC!wFqc6 zP9p|ZVj&M=MukF%#0EH45` zDYes}fDYa%CcM>MgGY@Ubx(Y3Gw* z{Iqt>+RZ9Pi6>0oulxGLdGF2YPY9e8M1D;-ckVD2;eG^5Od>$yAH0}lZ0Ms3rveJc}&Dye8x0W7$oIi{qp zojb?1{@1I|jOg06YmeS!`VHxS&xq-woA5Bc_2wIIKBop}2@`-&h?{^y5WOYZO-Lt; zsCS8EOY#!I8aF;YY%o&B$Yc!m(dsS(B?1+YmJ>iY03}L=S^WDSX=_J2MzkB8q9>f9 z9V5~XPJxU>tczd*GID3mrJbP=a~km!fYPw$F~|~)slBSk@K&$CbWeW*R~9HwDhqTI zUZV!AS$|iMZJ4vIqiVR;w(T`CwTi;R+Ia%q<{1{|8KdeB?iyk@QObCX-Iv!sc6WkL z66_{(##gH|KR%PWS~VDt9Wz>;3i3~#DlV`(zD@HOos`FnU-%atxblv60(AUm<2R z5G9~0l*^vWit(wH%ZAAsyS@~rp>IgDi`Vv0Wji92Q zWW9*7dYF9_V(Vc05Y^JYL%UX?)K~Wni;V8lS+!iwfZM_MkZsshi)LtsLy#witzETv zdpNMfkZ;&-`&?^_4{rrQ3 z@OcRegcErE-S4#C6$wvJ;T6u|uah`6-BglC> zs=FAwRR03N(w|LZr+wC|6zCM6|7*{v*AQxa@D?Ium!ZIX9I zdyV4TQcEa`c1JPOJWWpa>Z&S2mTOCkOG>K%3&wQy+wPw?VanRHbMLQ5Pph6S{e5sv z`1+X&c7OlFPiM`lXJSP~Ng)O*Qnt@rHMC7sJxeP~$j`#;LZptf*?vv+bZy=|$QMIY z^XB0p{@(7cs^|53L2JV>rx3N{WN1{$d7$zK>>U23S)NqSib?{u3X03hYie+NSJkMV z7?&{8;+@5FP=uNJFFKj%1roES>160$ZB)gTykFD<1OF&y7}+bDej^C=EhE5ojJR1V zNNP_x1a7m0yz;mtariQg6_3$lM)ye=6&=&5eXFQp{ky0hX8+K#YkK|lvQU4g@OGi9 z;xX%kS?@1%UsRY}tpZ2g{tJI&8nA*#iH9!#BPx&c{o+XV# z5JK1-^f8!wTg$z@&Akn#J%bf}%Sc5+EV&t@IITn_;Jsr{KG3#Em;^V?CP^*Y4hf++ z`nAkZE&Cut@&N((;-}U?O|s%W0kRH(1{G6GS;{Mji>sLht1(V3$Ncb}vYApD8xa|I zV{3g)u_6L~%x5Q*0vwqc6wcVee0hg7#azV4v;1G;vm?rRLSA!~%e*VWth|~tY{$uY znOhH}+>|@lEhQv&oAQm?qqN!;8%AuexKJ`TtIP8*T)0pX(mip=;307?j?db6Yae{` zv&ztJeYMY2r9YW20>?eWtiA_e`3Hi|_lEstr+43-JA0OO078vC01(zE&3RWa1VFk9 znl*pPC!c)0?1T4Htcw@p-g*~RaNav_&;LN}k$n;Q?cAl@A_V`k;_MufL9ZgH5MO;E z%kf$I<-FpXpB3j`PUo|8nQA2LOcYxgWKwcvLj}y095i7K@}DJHhk)X-9$5o5HG=od zJvqRpMT;~>!rnv)7TO@$MK;t%f0hnHLfIFh`ESmn@jc1$_Gec7F~{4Sk$&MELA{x( zl@Yf>JF01Vvrx}YGl;{T!@&y|tU5x>aA+BlX8-nu*rp$`NAp2A5W(5msyQnsH=mdZ zIRQcvCsEBlSdj75dEGjsUH`^vF7c$dc%2fV%HI)HIixW$2i>;wS)atuT(Mlpb{Xn6yRkzO3A^uSV zA1J(>Uf`+vM@D+pknDBr;Nj!Pj~_m8EF&YQ&L>h;z8S!}9!PNTDEn0Xk3S&vK45>| zb$y8`Z|g>Pzb@TsRsYn~Tyxv*-S6reMZ|Jcm%F-mZ*9y;RsFNIGSMvimoB60Ai`c@ z%Tu-VH&%Yp6xi59{6Y5c_w{hYn2s3WiJdJ7c89$>DB$0I-_b9w5f0mt8~xIRwZMcG z7JzuF7|L`WjjCa}#j+e*Qe8e4M_q1iGM7z}(BNA;e>Wa(sxm;8ur1&t*c{jG!FUW} zJce-H9*jp1#v_E}Z1~B=3wRLEoIHK{#EBC)-p--_oWG!o+@9V*Ai)Cg9oQM-Ch~ z`1{@qwl;0rbm-EPEZ6QV!SS{-#@o&APo>~C_OX`ajj2b@ojo4a!rfl51+?&owbYy; z_Pf=PbX$NYioDJe92wct%T#(A_4CrHgZuXFJ*|e2v(0SG&ZQP>VKK#5vD{u|6(c}9 zSA!@>_K`t@6~dD>`a-a>t0AOpGx>b+m4t?bhlb)o4-X5$WkOi?-~7zco3Zq!7$w|mVS1)`21&y z7`{hfmRtJC0KzW1+XK)`CK%T`_36lB|MABke)8?6O`E>`WXZ=LfBjowlD^=Vm8w6^ zt-LGwFgC*1B|G49?iHN(swYw>HICSjKSi)k!tm6n`m=Zk1O>vsh)>s`DEx!K1O9hk zcl3Dr33>#}hD&e`VfWnzyEA+ZqgsHIb1u zHk-F4EX+cUhR!hwwoF@*dDL~&1LgNT;HtX0rKY->O!>gN7FVinWmI%1FD|JlF0UxT zm`b*Sq#X(;QQab%dwI7AZx$KeBC^?EK6Biy#M&2{F8AR4*J(FGwIi$vTCU3N-Gnjm8xEvsHcgl-qFAWX`rf?*6C@Ts&8srynT|Ymxle# zId#-pv=ylUF*~bzOW|)#=WIw&zSB)n_1FbCXKALYudJ+8b@i*`?f0sB%t%e=6#wiP z?sGWzdDB08{Meba%h+Vjvs7T1xpe%*(cj6VJ^Xv>@e}GyGn~oR(9qZ*SeWo-Abc7n z&zlV&xcSVZ!xEv}@bWr0RAZ8bQWMC)=C0|FLbcrISV8g-EE5 zwhG%3+gjTUw+)*rd^F`JFm5q@X}?4!yh2-7Q=z!JQ~`X^#0M%C+L;sgtE@ zXerSZmoS{I3-r&73{1da^`_3dZh)=aR!P@Su)nDPYQ-o0VTFnI2q@3p z+>GvybgiJfuNGL{3rrDod2@3)U0#BJyR5uI0s>HTO7rrIuVR+T%_o#8Hz(_o>TW|0sKk0EA8ICcBu~DOiQvhm2FVgeE}ARbRIP9u5PWvr~-zUQ0MF$)~vao35MUMU5|T44I4Ui z?9@jdoBq&5ib#*1aNh$DP8cwFU}v>6iVjf}y85beR7iqgas?md(z%P7N@E#$P}F_N zyHb1&8~U}%Mt2_wG1y0KzHrg1Z@xKn`1hm7j-NRG`@vri98Nt>EYG&@S1f*e?wr}` z0EG7$+T(vL>YO>VX3tx&aM3$Umwxhz+9@=YiaQbBfyCQ}HftWyEIg)Thjtx0Mx)&N zwQSjpIv`=8L4kqbz__CsGb$-`AS*w0HZXS~05FUq2WTmvm~1scUF^0bCh`T++#MY# zk_?Ct_*g`&5vIX7BiZP!F4_xPnSp)!SKuA#2f?71mEx?!q$ zMMK!WpdGuyRNaOn0}&qvBLlx>|Iq$V)!fH)#**%9!J(C#2T5TSRLurpj|o`9jfD}0 z>@;82l$*!2d$|9&A}S#79^sVBYD)2jdaYfl81p0-S2 zXI30Ps8VZGZ!#6APC1!~WksswzXDF==Y_O|eyzRJuH?+T2#7E$fe{A$e5800u%&?t zWTs^xg6fn6YBJ8zyO%wKiCv{!yx!Aux~e4DTiZM8PGpF9|A4AwSoH-3U{qyZ$rAID z-YdvQUbQR88h<0+uL*8*=H)cQg03(Lb_?+f|-%vo%Z zx9%$eM^SHG+Xv#6w{=7M7v3>$c+fI0EGxMrYCEKPgufOHu3XuQ!IEXc1=FM-#j#1T z|HIK+bI72PA0IGhJ*C!1dA)c?Nm z|D1uguJ>|HP(2$OK}DK9d+wSw->oGavO!e_vxtr%qC#wy9kGtQ1Q7Dfl%%BevmdKi z`p8po+<*UGy&{}cmu%b(x%gkM6GGkpq9MwC$W- zRCFZ|T|Hkw5Olg(ist|f?Xz`*l+pGv`jWE8F@%yRH3Bs5JjZERwPrb6FbTV&HM=w&&_-WkQTlTS@yr;p6-X?KZ&v#5 z-Mco<&8#|h;aQ@N!*ulvMyvH0wH8x& zvxTlT1bnwHUKgwj&_(LH>F&}^(hb#lW4&~Zii>U@>=!+J{8KN!_~Kh6py@kCxw0qvNUk^v)sIVQkrP_YCrLljN#060)-0eM7-!o_S{rJ=V)0u}iE&qhfoG+H2 zPCui1rTnyY`;N5pOP8-(zxLpl%a#Jx^8|DDc?oQ(b_$LRP_C8Q%25F8u4zFqyR*pu zZIO6p^WoIgefxe~yL!tHTjaZL9uaNZC?!{~+eoM^E0G`=t&aqcLG&t#IcdUUP~5cq z?hNDyL_)2Qw1l#tgo0P-hyVqTr3jjHrx^DAjHKs#)*t`*GQd_ZNKr zxjNrX%W)Rl7K_M;GZETuikWkF!(L$|ltIKJog1DR@KG)R z0|8EyD_R!_eh>^RYrOJqVCNvW@{;QwEZbqtE}T7o%_}0Vd)I_vgT*uzTwQ%=_r@Q7 zSo76aU#(lc=U6rVU(hT3f_#ZJjJH4H^Tq0oE0(YP_PbSIEMK{A$NCkYd^CR`CO@cO zA{Oi;)nHfG$S8gxH*Hj^G9~m_wyW8tsvG_rkT!C&G52v}5HYBEn&{-_%}@OO{P`W= zQ-A!#(dMzVc`Q9WmYyC9HI9WEqot2C9X*|aJb@^|5n^jNCdgCT3m>b|h$S|(GGS0! zPAlc95g|lUX2nGy;{%B|Rij5tdg7_)hV{QIzIW%&UHU!v+N+Na@6#tXEG(?!h>4@e zj~~}B*3Ub-_b}CvmX>|h8euF_$2@y;Rv7??IwdGe&p0eAoSLG87QBjzYrbaJY zv|!=;?|u0B=d0GOUOs2>+@$vw&0Y5Cr^{C?{d`Gs^1KD#efQml)nBhrqobmlHH-GJ z^iRC2ZD?IZc}0VlmaE#DB z`ACZpP@^FRib0ATWppw~djTs$SAY_j_DE5D;Do59Mhkff;15JbaFEnsW1m1Fj%hhu z^1{de{TUr&HjFVF#h6WI%!YC0VT@T6(let3`^By&A~qw%8{TT0c7zcF&sjE*4F__vXboEwto%3y_kL=g4pJ&CDLx+zXJ6qR)?2|GG3Vs$# z?Bn>^rrZDJoU&rawjFDiBqMb^Y=6c(nLLeoi$5maZ~0sxDd)aBj}s%Y%56536_wQ5 zmEiDDmx_8M^dxf#!M?bD?P?V5WrX1W*jYs;W_WlbF#vFGz9^= zRCF*ns_)2V#k~?GIC+9j?Cql&t7TkxBOzU~i6G3t{a2@QjLwjCf5D<10GBh)*tV%3Rh+=(VFk#ZP2k(Du z;(x4a7+^G3UAtzs1P9gG3`FxGf!6$Ol|xV`U4A6yXT!!%<}br&`4%HP!=55#g#NeX z4+Ctrs+x+*{EO+xHubd)b;^GzhZrm27D!l(HN-7j#(yY}2*Rp(CeRgSIc7~p0H%20 zzblp)E1?+@mqTp~)Eu_|R6Oy2c>xD)+|x9B1!x0{o@78AcgAa}B-=%Ha4jp>mqs|P_ z)jE}y`uVw)k&IUCVQfhur*`2rE=i;+7(LugI#su9X#(@?E%)RdKW*Ch-P+|J&q}h+ z1Uud{>AiXLXU$#s{wLotD>wbTUd~ogA5I;wN>hu-7A+N90SiHHfyFh+R8wVBygYPj z3mTL~fDQ?3g%Y^{j<>^j0D2(7ZU_?DG*A-d#JDUr13gkSD?LZ3#N%K`lsl&6H1!Ax zYYPpLKnAei9ku~4un(~qAandTr_toSHq43kj6*x-gyf&LVNUpJLI|8v>^Fed>`I&ybLa)^T_B zACi#JcgmD|$Byb37u&7-G>h9bwL?*@J0TK?Tb?dnPF1Y7lCWQOt-->SVuTrN_j;1~ zE3uJNhAhSH>r_mr2fkmjc&c z3N=(NM2wx`ylCzB-+q;{b=z#!E4jGQ4kNZ=4XeW*$k;EYMQs_I&0Wkp3fi@eq3 zL_j6=#ZqDggD6J}))c152I;~xnvTk%uN~Fo0eM2hf19)|I+%{Z=z%us*sM(eb8}(^ zh{#Wi%M@*ZO&Uq+rsJqA{@PJZE|91Hd`ia%Co;lA7~x@za3VL5$Ot!w!xP&8iwEvn zL}_^v;9SUCLd`CUbiuu{!czbnP(5Mp4ZH|ru_9Qiz7O2p5IJn(sKJB#_v{eWp=;M( z{qGvnAMfKk=&G~KlM>pNS60$L|C0k)3&WpbOa^=@w zuUs}yFgE|iI8k+C{>SS#|Mc^h+!MH&a;n-SFdK}{y2uv(R-+WAa#XdVRj_S9-K7w= zh|2OLK)p;F3CYXm3__)>yFdlnXd>*~$cu0jD`q(=6iebV;FwYmTSQ}d62w$ylhniZ z{f}pG^!dQ1J{OPgK>B>3tg2cMJ7jmr1Jkp}#1iAc9@cPfkcp&Ep{J$S9mRS@-xF&Z zS`FP!Qti1jaj}<7))1q~*x&*hq@4ovU=&9<10K&qi>SmFK z0{$i%SYkqysUc_Yp1oOCUE2gTQs;X1q^_(gH@)zxuDay9sj;G}zV_sqj2g4^#fv8o zY}>l#Fg88dtEsL|VznERd|QN{KX$n;?cDK0oue@qd)Vr7^G7dsR&_(HDqHieg= zF7|L^Kuu6TwRnk?k-O3AE)%HHRCgI_f?-5zy)4xr1Zm$3Yp;mB%tQDWM=eqx`ahmT zW0rVBc(}|_?Y-kDBKgc__fDSF zvRRlnCKMM>gVHeWdVPJz9=!*T96o$#|FL5ysiv?n5A6PSYk&qM-At*kue$DJ1W*G* z{J{fP3~KZJA1q$@!J17}tlhV3*H63lzQ2bR_WM=Kmw&Wm$%kJqU8GvW|H$+U_3$2` zhN^38gE8ENcvPv)iH^Ln0B9-&ove8)L?jS;q9NL1AhDV(20}O5d&g0{$2`4VLb?xpS~XVxnjNfFftTO8yHN%H z^uGQ3a#7PBwLha<`ku%jGjwa$CV)Nnnr;rT&dzEqO2AK0~$Y zi;AcRFK5x}!5dpfwG33bThYoy;1F4l7mZi5%iUrvbCP5U4Mvw>bqvNq8fSngMR_Kd zqDetx_s-vU@BzVe>p(wsr=Qxw2in31{2@mA2l-m0Pi^9G@l2+^#+HQNCTsi2t}`Bt zHuW8+gb#fd?{bpMtC1bX21E?L?~#eB5@}UdCMnI5ll z+@EfEk8XO@X~9=-zrA=4?q>V5%2J|)tV*XOFvFF!WZ7?@><>ZFW&~jo_ci9kxN$So z9pO%dO(tvm$^J}Ek!EE+3S^~&WC$a^?VB2fNbSg_w0u7;pG(UfwIc=MSX!NxNx@bj z5(!5VR>2=LlWveeatN7SVIO>kA|mfJ@-|9col7A73)mk_jUhryWSIqNL%1aV0 z-HWi;LBvl@z&tS>1I0Le5Cb^+iv3mRxeJ!8`gYaAxk-Z%VD2N_VLXrf?T_nHW)C17 zWfY#+XYBuSN=ccm&Ut45i3k%wVVq`v$myN8Kl=EyB_pukS@F0};33$(FF+tqVkmtY z;&_3Q>22J7Vd{aN#_M(0jb8rj2Y@6&ovBEz+G>Nbs-75AXJ@y%dW!e?OU+f2Ns9M* znT@_M97oYrN=pg=xTZX2QfFA^mPB>QGRGfNj^UN4e34@{riE#Q1TPs_8T8D{MwV=r zj3Cr5v}JA7Fl@4~(`qK#L?x4>KtJL}YciX7!CCm@Ycg#cvG=L{sFxzeePuy$7yN4x znkW}RW(}-wFG3CNMIpUDacq8l9pCz*pyr&Sf)>0|wafA9HdxyA&;{;>ppT)J9 zr=PrcV9(etJqApg{%rRyVa|=LD{f)!dJdZ~a`529o*i0=2siK!B3vsqq_JEhXd!N+ zT?qZEM%-?HC)V}H0RJ$G`3ytKq7Pt5Zz0S*Z+|Fs<;sme?>Tkq3-`FF5f_$d*gQvIK2iloWJ4IgP#ilB*)RnEJT#;cToXKy zMsQG?LtvyXkrXiflfc+TD93`f>EUTXF@b!3Yk$^Jj4t6EQ1e$5!Sz1PewSr6V z$JyA+9V@g0tF!$}FH9ZYuWR?7gCBU}m7YDKJTa}?+*@?+^}-9`VGlesW^}^n0h1?B zwCU^gt-AjMwk&i~P+@>hToqx0SFeF~?&GGd;@qoa70+N5FHjR%wVxsPG3nipmoB9! zxX{Pbo~ypwvFiZgsr$BV{Wdvi>y~)?tDZ^kFSgE_ot%`sapRUBKL6|;HArJrzU~H? zlgT3xn(+6~m=>$N2yfadZ()@WQ`<9@FrUC(=n6(XJBJVjqD;}wq#y=7CsRZ!VN&dp ziYp{Qm>_4t&&<9<4bo|asT`pCNCQ;(0S8!v)n^$LeH7mb!Hc3WXoWRkA({#E$5))*F-q6DhNsMSnf`{Khl=+N+La{05&ksyS@$f!c7_Bk#o9pVu$g`Rf42@*z@Kdf8_fPj`4^Qd<=S!XLIB z1cB?|77%Lp9$obaD%>0;%GUYew*i#YMvoSQ0V@#Q6ph z4K40pfHfpn(bL;M7}$}2`+Z0M^^Z`9^D**oP$op^lm{Y^x(KH*V0p(S3yF)IFT-o_ zYeC#r%ZL_{k*#peuon@5ooc49U3!MkwQuo#>9Dv15v`k{~pyo($(f_VIgdPS5{g^mBGb>+N%Hs!~yf4LqOE8;BoxAYuC9N&g0@u>~m0? z)zqGO`lv~JV7b#zP= z;!mp>-nNP1U8|YFe2cG(dK4xfw87R~FK2~_>)G9Gc zKWBeQzkGkHWPiS8n15*NE-;mY3kpo-k;aDf+VWGzKvO|Mut~+bs0Epci4Crujw=E8 z*i2L<5o^gLq&uWf?AR9t9v(j9;qLB^kr|9WG`Bl{)zMq6ntGr$J{=V)WV9>S@jneC1@ycvxZ}$TZ`}IRs&&iv?Au(BajCMPHZAS4>YbT>qOLgCRo~EHa?35P zJCmNtF=5C&A^71!!A(plaR`Y9CYz+q?1S)w5a#evVhIzEF=f90`X!AXq=us46gbnu zrZ^@g-aa0a^mEo^-AVSx@cIR>&s&q@l}~3X&&un(naXIZ1MSYl8UBh!p`WSjvL@@@ zmAyRcm4EWwU2*3*K#8y>JNnk?55L37syNZNR%JI|8l*kHN6=KVcEBn>((YEUStN5b zoV#tpW>H*DpSeJgZDuC0m;fv*b+ZS8AQDT;&uN7!*7oK&vq#)%q6OoGaZ&_CB-z`B@_V7AoN$IUDFZAy-PI#!8;3! zxQX}g1dJ3c&v@cOTINPwBaWQ}yN^9a_bD1=UwdCYXN|Xmvvx2c;oxj#psuH`B|_~i zvYdX@ySs89Q>=zQn>Y8FZ7j*Xl9{SAG$t9F`3HG*ls(5)@7U8D{?O9~s0jU^O#)AxNe;?dUUdjTn=+yu3XaC(4ZKN$u4qLyc7O%<)&kf zc1e_nL;y*|;SFRYi%gHp1TLkMy~Q2Oq`0q^f0AtR3uxV=Tj#G<|M24vKPdUunFhP7 zr!ff}Ft=n&byd-o^6Ez(7&md+BizqSL)pbViib7SCtGB?Um`FG)P1?FnZyU`l9_19 z!aPvu5nfn-xVc-OXni9Y1#)+i?Rg`kz#YDUk%ySs3+c)ZD{x9=Fu3H9Vj{ba6m+ej~X{i5f|SHpIf+0{-F< z@~}04l_HKLP4wgD)3bpUVSZXzSdjD@>+L2NRx|>u4Nhh;DX?7GXFFBbRb4)^LP|hM z0bO8OyuZ&gRi+*I{c>(YV^Mn6$$3dm31Vyl+9BRP+G_9*aMxMQEyg{b7xv^6Pb9f! zre1#T)t8@n=9&Kfvx5iqA2@vU@S#Hob?lrtsLmdFE&cMv+G^)cqn{A_!zV`1`|k5p zFgsYC-2%KJxv60pLjZxhzOS4mWAvMs` zcEya;ZWYnn6%RQ-r|;wK(~M`+RxbQ_F#Y>7MvMn}PR3|4)&7k6z5Uh2dgaoW12Kb4 z!9sG1__txYQ;fj~tE<0%0v-P5bC**ylgv*%@nm@3hwOdody<^Ikz5{0EY5B z-HT)=FCYl;pGuO~8y!1#=`dhm*EU_d4U2sK`A9~=YBjsLmC@JruJpU}%mWMo1Mw9j z;_4s9@EDqDW6+#xZ5N50$YHt@?v#_6heHKpnndS7&u|Ji1R#1_r%0x*RaCF)I+2_p zv)RqPr9c}V9vA~5QGQ+l9%81z&PIg0ub0x|Z-4FR`+vUzb=6j&HOdpWtw7F4fsVUe#Z4Z>L8vv#dP)Hulq7;l_OTG z-pz!!u5+hAL-ENIoEdC>NMEUTY986j-=yTQ6f|00NUAt?w4l+G@;=VuRToDzCzjNI zu#y#LDfc2?<%nCyG_ELpF4^RI+Inl%uahcQXT!>KwG81d3E%+RQ-L2dzi;P+^G87QD&ByamgKlJM)hv z;QYDsB=DRgPT>573mF$l{lxVlIX&mnG`sw4?h_v*!=FSaQ~XM#FzPhgC1MsSUys+| zDYC>akbnH6&Xu)(BhNL;3UZhCDaVyd2t3~_dpJ_4rI)e_}RGM5kX*w3fi z$Zg)MY~a_MM8|+0N;rB*S)`lsx@ssaY}qnJZ#0ULpT#yKBT3h*m({tdQQ&QJu2fmg zRhE`)wkq`*+fdtFGJpoy2H9TXF~~OCw%K;pw!!v2k4x-tw!Lp#XnU6c(v!CBwr^}7 z*$#1hxow4QrELe_*=E~k`-T7gw!O9#+Z_Jq+kUWpZTs1F-j=C?j#5;V!h&0Abt=CpqLoYTmR2H-la)$vCcAQBBVk|223@>3NK{xYufD~eL(2H7Dubvvc|0SN{YQ{F z*=$atLZ>yjV-9nCbJXP4IZSVYr9jq#f^X+#K-DW$&B^bxcTk6O=MYVyXm=kI!eRv3 zfuy7ZBH!8}`i9symIN@;7y(=O+1wq*XZ z6d?ibLbcu3j5T#R_x9V#vX8b1AIGsz`MX)Husj}C$jW4-g>SJbWiWk~b4q10;f0XH zAky*8Q4cCjkJ7w4OQWb$6~cq%%{Er&C`~nXtgmtl=DE&-{QOh_@thAUH^*@^x{E@< zZkv-5awZj}!V07?IKw_%&h_h;>?n)uQn$`N>54Z8!~`aRRanKP{t=Hby0PZC4x3=+ zE?-XOoLD?7W!KT~rNSy?DaTZ(#s|9{z+-%nB#Opv6z)lx@4Clshs;TVjbbY@-&<|X zmil@SLI{AW$l6+nsMX*m$Qv43LE@pyylUwvIluSkBd0QR_J5mzXeFd!a@W9E;DlvbL)j++1A~Hs+Hj_m)1_LFLP|A{LCs@ObIPDYTw*ckIp2)a8DMVB zq=+|!8D6+}@xpmz?>m*oz7V|0y#X75cB2W+D1C*?;hN~X@#w+*`}X{{?@UphM?iH$ zlhxMNRCspZCd_!=L43yzI4&;|7e!th?}H2)NE#a%Eb?zWiEY7T&t;EHjZoHHI+>A^ zdF;%!CjZ2RkG{;}1n+vFfFA<#JmDQ}b;2|c+U4;6Y^>l7+BdMO zxAS230YHSonp48IZBzf%T##S5cMtnS0Iv4PAt2!w_sKr zK5VlcKAf!X-+$ZsVTt?qL*t@ByOKM3GM;_S4scofq?D;<_HgHbT8S{N@{0Vt@^WlU z(~w$;h$SCC1K5!<@zbDMDV2R2{{bl}T*YoEuADlZs|TF=84eSPS31^J&aXdbaXdB6 zV3gmG*dMy1-8!{iGikl?urnUvtE({><_W9?|HK3+I3IA2n>j5bwbx85HPkTrEj>** z?nx^tc^AXc62x9>xQ6I9Y(0_eK^e=<1-HE(UQAws$H7b{m#N6=i4Ejl`~A+F5kucx z3iMcRUM`A$R^aOEVP`0}!Asg$PoOB7AP6h*Cou^cl!Y~b?PV)lCx88w$DVn45?0j> zP{sDJU6`I`33_+;P#Hoj+}$Tkm^fkLxKPnhgt)l^X$ypumEvOX=Iha9S`R`?CYl_qM+qk0krksmHw1*5cRdX}URpZ%W>06F zK7T$VGu8D-uN)8*d(V(Qn^Q<{@;~#=UF`XRk)Q9)b?&0jhn~bScjN#f$1*e}=@$(g zUOvpj8221GDpzS#6?NW|A6+z9ci+=6qunwqkgvhBqVpF8?=#ol#QN3Ve$Dx=z1j8Q z_k1tA6@`#YNis;< z$S2v`9zt@wa{?G{c_lLlEJCt{*v|q}h0aU4G3@j$BFBYmY&u z*%xv4`#HG92E?(ixK_sL?<1*q(_jR2%O>TQAKF_sujE1yu4bPH=QV;_E?p$qJaXhv zYRZ8F2U1cG9ohEfm)k@#xD=e!BTkCQoh#OCsw-UKL^%L{RVaRCs?olUEgeHFz#LG!$IFjD(m)WtEi`B`}Q(Nd6F% z1LXn2N3`Xwt?i;A+#7rIeuKs(Cd9>fLEk0$tEi~5LPj}et*VIh@QCbe+b;Qw@07pT z|5^UB&R@LnRUi4In%+_VKK)+aP}ePgQmy|d`O|jF-7#b&D}6}mSa-FL z^ieA?=6^@&!%hn`-dXgd=8^int96q8`_{Q|G4ss-Vsr+K-tou(a&$U%w$AVb3cY;s zzZjpDX_-y88=rS)WCMTJs)}X z4D0ZUU9j-}2OeDX_|iw&0zsLOnwnY{ID_&F&^rM$eB*MZ+S@;TR6>F%E3n%SQ*z-4 z?X%t2pK^ZeMb_?Q3=fVwqGD_}_i%Su4F1m66J$}I`-cHzx-2wY|Da^Qx_i!Ntn ziT!K<9M#w$M+4{}5+rRwcOZbFb ziG3wy|K30L?ROk85*-&hRs%&RDNCk>P<{*eJpP?{ESGmXp9MKfsBPmx`oWUWSy-=S zL8i?!1#fVDYjaJ7 zg%6I*&nX!^cIQc`FUWEdGHl$qaU+aI8E8(fpF45lL{>pTs}EIB^q19?*IBE}va{BF zaK*yA=$zEP;h|6eJ;-v-7zLM*F+KfaaZODPh?QMqVqyjdB6@t_$dRq-5)F>O1x$jq zH2~$8SKzaBwhwgASIKSV{U4O~@Dt>6?WXy~`xA`(c+K z!q+W&q#{Vd=+RphNlQyReDp-t*;J3->d49!D_5?2d)-PYzgLe+LN}1PcBv(suSM;cbjt@LU3vku_|U*TaFTit0M=p_3n7ylB+oso`(DxnwY6 z)5*hc>YYKV_h&wm=amM7k{7)^|6CEOi)VEu$d)V>H<2Zznzgno6|NQa4F#UP^_cFw zGA#H`t4XMzpRdnQdssM{+SFTv>HT?H9Cj=E_rGhJqrFL^A+M(r?X7?WDIpucphLv}O-a3LCD0}&*$p_r^($Z);L!DKVQ9TDNb7L4Lm2C{6pm-5 zr5-(Y^5lu5P{9rzI*fe>CwpwH9kkOtw;c1QLM1x*4B%Y=FiAab#|$PAr=3Hmhhr4wVZTm(HF!bM@?T&=VbIn_(2} z*~8`K)mGiWs&S*LmS76-s}X-x0;Gdt*T$MSFdHGiPb;TErW<8jV|6L^-zpky(Bp0* zDFs$l5Gh4O1o^n@;W^<|t{%pah@c>UZv#wP7kA$vQGLAFHPqzCF8r}M5dhT(0B3zf zXWn@rW#8UC`wksGdhGaN(Kvtpcvov-DuSzz);I+w&IdeINlP^XeU&08G*quWy>#i5 zk3F$$3QI;>5@KtAAB@ExQShN++4Il7`mcW_Vz91-{UnM;w7OEGlwyi3RrPiip&eVm zq|wyM+(b9bdy?`k%_x!y!Wl+Ij~X98ea4I#Q{&?%B~6N*FkaCPqeLzP9Z)MsC;`r7 zoDtz?Aaw*;6l^QKi`vPc)?#jS0|*Fgclz4ehIWO9zIiyy+Zt-o3mk`D0%KjW)Uw18 zLDn{Tth%l3Yo(>RIi+RS*d9UTT3S}9_l+F+_4}iz?}Ot<1S9zh*(2Y74~B)05=*d9EkOt0XMaY7+Q4NzKbG%E3d^B+dYWGcU#aA9z#j^^*Zt3#;qv zKK*!4##O7%b??5TS$o~v0wd}@H8-xO{qgfIG2l-Lt8PhM)Hi$8BX?|LfN)WCnvD7Lh<7gF#k#JtRcc0Ixl|)mm*y4rsPQc|!IA zF9u-s$L1&!OzR=j&RVrvsYI41Pfz5G@$vHX5gm81 zGbmXN77N{>;M|1^=Pq2yDJ(982TBaURy#xiu`KXwW1mm24-JhO9UUFLV8MbpQ^o?f zB=tAx@3gMPnjF@36j?WL^AMiOaipeqg{QOo-PBfMNp*#m+5G~k)5o`3H(YuJ0g)Cc zhTW5&ay>83K8suNi)!(hY(_tc_Bn*{u+Ks6Nx*;5#ZW8gNTtahi5&R=<*fT&HK?I2 zWZtrwu?ZMBcw*O6xK*;| zQGkmDI)nxEV2w@J@_brM4jw1+MYwtmAMWL7nXdW0@2uNQ)9h(FOr(Po9gyNrYD>DJ z3)iDyUHF_0IEB&+;)ulHIDIr7+W+#J^@Mf2?$!@_XAokN+<@9%-+*H^U<#cN`3lJHEUP!{g<; zc+l^@cq_x>Z^!gaw=z7A@iHd%Vt7;yjvP3M{pWh1s3KKgqIZ<~gZQmz*QNLVj(q12 z-)x5?4_GKrc4dnSo5!k+>|4R%Pvrt#;*mV$19Kc))Ur|^pHfi~!W4tqU~+;kkO2pH-Lbabwv-=!_~u)bvaqWZkDw2JWZSz+QE21XChj%dX-j7*X8C|28aAY> zlmb(1DI3_klp(0g1k{Cg7KWhDhoCjNYk7i!w^^=dRMpj6tYw$aZ~SCGBmjY_dabvU zffZND)G(S`9eJtN{m_}0Y92_?#(~y;6yj4J21Fwk-vaG}?T?y)J_6HDU11ecQ`MYI z*}iq_){RnXDupF|$M-wd#Mzg)+&B5LCzd{Z-?Xu#$3$9Kg@CBr{!#ti1+CYh@rlS! z6)WYZ@*Jh+?QP8wVLk?(zQHdvJbdoL8Fc%mCUx5l3(_$d2u)k-*HX zn`I?Ar&E9b{ZMAAu~ln#!qM&T=H{$Y^6l2Tnwq-04vj_urfNk` zD;Z8zImx-Eo4IF?A50Zju3o(=xhwGDC4TdN{+$S7?N`xmOZjck2mLh~OX;=D<44nx z$&twSTB?h(PVVS#HiLh2RaHfkHZ*p@<4-PmI%5!R;k}u_V~Pc^QZb_ ztl~bDd^5nuRo+wu?MCYrG?aR@cf%nA>7&v4^%&3@=BZm9wIJw?`(6GUg0WlMk zl3sY_DX8ZDXjhr5kB5e9m+M_>t`t}e0Y1E{yM=+8@p?EqW^~-Nd4K!c!r5`-$KN&Q z;is3q_`(}+CPBse5Ng+IvW?7a0^O?-%OujVTjhGavcs;_xw@!sZU4B=_Me9Kk3;*% zq5T7@E!ZJavKC}XDu)%+V6)4haH(6$OTO808qM#As4K0Tr>m;Hxy5c*I?1uS)T{Mf zjuhM5TI-6=rT+5mcfxWZ36Tf?0=Vb}K)jyf-~VuRq7E{fYq_JL_mCro$?2;IC z0PYkxVC>8}GvkMc_ZByCR ztPeikhdDrjO(>b%CTK4gwA$e`s=INi%t=(f`_sweC-W{-j+;rC0mdNp$AeJb=OXpT zSbKmy(*Adv0|eL7tA)t^S-OAse?I!?qjy&2mL_}Jb55T<^W%p5#y;}+KNiiKIxc3^ zKsypr3qa``^@2thIC2aaMPs5lTe9b9^3y4Yg@(Gh7#jUULnH3Jf2MR_PkF1|oq78^ z`-1Sx)ibnTvd5A*1c3#Qoc{8{dGj86_^B6PUHaTo2-A1Cb4RPy3MgW*!37ZGenz{} zvCa-hJCKKvthM2gyy0jE?*R8HsP)r_uMshu&>cikA&BBMZmytS=rRtiypamVmzkBiPE&_-E=YQ zY-V~v%8nlnU&+Zkc(uG;9yQDz8j@#IlM8F9;nMaVy4J=Ux<$u(h+jp;%^H>UMw9!X z{-HxchD{hhWW*>aQMkftEx#$dkp9R1JQe(3YByG#Z^H*emP3q}vAZRwAeG9o-uY4G)SZm+&^!_o?ZJbACYhMMh} z#Yio7=6K59eY3G{5bSReOuvO_zjaDave|) z_-7T(rRR_I%6I60ao3k$Z{K|~x2m~0ATV^`@G%o7#zse)Kp4thWd3xu-+&6x7=zx)iIcf8S8s$hvL>Q}KWlM*=*;;nctqEzG}JaK^v+`khPc`lTJ6{o z1{Q{MWbM_g_9o0CUWL(U_~cYwWBtZ$AFo`o_S^5beEb766u~)v-_Oj5WLpYnUDp_# zg4*44i+ph}KTEESxA73kF0Vbk_`Z9iM zqrR=Wy2HiIMcqfz_FQkU)RkX5eQ?(=l(y@B$l7D@6_b7XsC_a2_F8iyf|?#f^z8y} zJ=CmDK|?l{s?sZ&nT5#GeDln$g7)q28HJ6kBQ$K#uuji9Qchkftg3NR*~+hE96OQzeahJeHaHwqn+`#<6yz4yA*fi{R9$6h z?IUO&aMz#1JChZ{oQeS5;u7vLZa9Q0f(zt4I6ddgx!~>1sX@ya&5ecl4dOx|Xc&?t zXq7qVvaZ$1go)R?Ai@a7s*DU)1i*4Gh%2XEc*K)SyYb+p zBFYa&zwqDW;MfaUTBCmTr6u>@A3kv6lDFcqavyTJKRi4@t7>XIeEhoHc-Ot-Dm4o zZxk_O`^)7nzP`RLKn^!U^TB>npRb||Vx(!Cz6c-{`c39H2uoo4gK`a?Bv<_{HRKV+ zOo7Z4#4K&y7xumcZhpbhS~O3DqooIQu^>5m@0U;!^IT=QVGv0OMRat|&` z09iPaFl!%NxM0ronX?&hoP-%-kDynnI%-Qxvq4^;U0Pb(p#sprD=;M3=&v`p%fPXw z&@&G~A00i!iLRpcmN^(RfE3#F2Z&U{}l*qf|iP_VRt@b?vZMEM-fdr=D24P8R zZWW^3hHZVcW*p7nsq9h)HBRksSoZK7e%$v6$U5-zCdkBP2PQfgRZiSN`41 z`U#NeDm_%6 zs_fiMJ0ku$rro`tYrR}tdg(}-tCOg$HyZ2HjK(zjEh-S=T3Aqw7&2hM>?x*0q= zzxBvP5}siWo`K|If{AxCNP`IbfK(=45IF-lbn(i*iZr`OgOrnfDaU{~0&r}$V^7gW zI)b&e4IuytQ1LIYOt1SNxG!l~M8wa#3;QQN_2#S3B?Wa9msNO+?n{>r!Yx5pj-&tG zOPiW5?%$Xou`w-^eYEo|u#5}|+@D6^(^&0>e{bCU>E_Li^_O>kMmjkm)uuTm|CrU$ z+He3p0f=nm?u#CpjCJ8dEK$d#_jib0d*dOX5AxyUE4f)(f*?PD z^lz4z7!c+mSHSZG+69dUN{s>mF65k*f?$xd-535m5PSfX@9pc`&l4Tp4-J8rbQx`r zc3nx=I;4=~r%MV6T_oksiXr&{>H`BY2y*bCnLa@T?-FDgJ|!m;(w%e3gjDNrD6PO=gzfE96lg zWkuzp6qHln8g8PX93XkMX7jp@Yfoff%ee~ofW~+D=-7#)hR4PZ^A&O9f)$I^lo)w zW_ri*VU&JFqZF8#Isx$mJdgb%iXO0G%Lpk9;fOK{Y*MdrNNw1+71>=qjrf=2^H}Nk1Lr;e^!@YMQbC8!n3C-5Yzz}w zc}}5dc4;wq%ajl~mCoJU6VM^uEuLq+8e_1*+HQB!IAH^y1qr$m%~A^JV2# zqU%`Sfsm0qw|2ygnlOGsc!S6c%XjbIv+qRO z@w6j9fAQsJ5%Zd`+*u!F3JMICDAjnS=!70C#JeDmh7S}ipy@Q~?!tLB@!8>R(C5<$ zGCBcJQS2Sk+(%=U#(l5VbBPE=>V7kM=V>HD`RhJgg|pSY&R($sd(YU1UJ+*x=09fP z0yNU)xS~fRz0K-;uVZz7PFkI7`>f9YSJ6yVv)h!sKxBI{^*&{Wp~C(tmE2W3%5U$9 z9_)u6q^w9@CeUt5J8g1=aOUNMf&{ZWQ-wr7e0NavLo-CxS6*IUz0dRSi;Q3YK5XJQ z6`y^+b`A4Epo+=lIeo!`>7pFrqy_mlUc zQ*$>_KTHGB6CO0#(Fa{KHO=l_(O#H|n*=CJ(2R-%HeCdKmSrK`5NS=n(JF{F4+~y* zAKs>2wu@|!7{}f-V~$|2NbS-hp@uV}~4tdd5*evs^-g%&;XvcTo?f4Fg9j*X*L|Aaw>5d?6XdMUzbT4)8eP-VlyJd_3l5L|k(S_DT z8YMtfNG;r@&(is1T}k?ElE@-N7AVp>dWiZ3*!nP$f#?qzCb2zrOYhuL))bK`MSdlf z)!e5n8Z9KV0q7gbapi83lmq~ANMdC?!&p}hnDETP!LiRgh_qT$rd%p};DIucaqAK3 z`6j<&wTR?Z_SelMwEU$@eAyp=l!-EuSy=~|<1y`1gUBU;MW7t8qf#AdwxdY;K-7_9 z1(CUeAr~##PT=%Y5oH2qIILa^Ct(|SJM#bh%(ij!k(ot(ynL{JXnDFg*`5e((+!&w zo^VZ0v`3gRv#A36H&q2%n!`mE*kcIDjodFBkLi-wQq*96A?heS=9iwoI-V#DPc#%y zM0+SokwGEv=)ls&Vdx?@f++}EPFb8Nd%3MyT4|HHc0-(fu41J|nZM_QczC*A zR|wx_S`ZOzM{uNGaMqvOvu@qG-_BKI_qaynl-}`R0C-=nGfcYg*#ux4-es%h4^19S z03+y2lzwUYW3RpT+9T7WVRFWZW**KdKnvs?LTjMN3QQ$H8P&pZOQc@X8k_vzokTM> z`H4rM$L~TtiFIZS>gmzY&`?!VXKio2SqXS+d07#xm=e+BSmv=~r?az9A4^HuzweJd zd-f3MPjuk@XjP&W0(ieSWvL#YY=T2d?qjp~d=d|~s7=luMIN;Jd579QNMT9XH~W-r z=0=eI9#!pAyiA~UOSHg_{2F82_D`a^{gqmuOK7Oo;KSsChXN@g_?8w#s%R(%8a<4l4Us>xb8|el z2cB2#-da()2^&;zsLjhaY}<9{;QpQHo_hq3wx+hW=8}@;wxdTc-E0Rqp#A2hqn%@k z#$oRrTgfg!HeJF@Z6ZAwPns;Cczh(93Wppe;;OljBf?)KqJ0z4zCjdE@K=Z$K(Rz) zSeVJ%%|!y$=-vATg#`v0Jqc8!bMp=qx#FJ?p{7l~+!P!$e281=shtR`dKxYLlK#De zXU?9^NIwlYV)o?=>_Z_DuCO?~|6z=bbT2=b^4aRutGA^WHv2|`#Vzj|!ecIEU&#}> z4keb}fCq;J0MFo}(fRa`9tE{wTzG%XX8wVJLxx60Fq6RUkH-*v$M61Gv+ln8p1Tu) zOB_EoX8hz?3m49qoG^6g(20|#EEJ_G#*H5ZOay>6NfNRq5zrc7YvKW}89$M#r~yzK zJEdfB@PZ2vt~|K)h!Bt_R#A#+|0~1|9)>_S^#I)%pijAL|~=9z1gT;?+W!FU2_*GL9ZRbLQ;XtSeWpASCNd=cpQw@$oq3 zpC?U@Y=w`Tcz;;Jcvk$7&w~6H)IHrOWIu$Jo|k~OfY}npHY4+N+R3yt=gwxF%1HH- zfVKp)1;~~VwtcY4Eb2~BTSdrS@#D2sXLtVc`wg#=_syB<*HMsi8?l`iWSbS!zYlWQU8xYk}RVxF= zkB)Z1YCQB#P~E`5;juFpFJ3$|b~u94dOnMONK#oVVRs321K2M3;_?gfkwF#K!3jSJ zyNkIK8X}flXXn0X=qTiB^+USWe*L_C@xdP$T>y21g97R*tH99R0Q^^j)nxPw4Db)` z9~Kr8&O8Omt%7yPx1Ya%KjQMC&%`<6hbip_pu{z`V65r=xwD@YowH|O#7`>Cm=ey? zg+57uPEWz1_uPuJo``XKN1Qc9g|L)6@Y+m z)!!LsEyd2<5ocXdUS0tV-yLz*QcR8iop9Dz*<9cyxFg8ASF8Ic6HM!$#M?r_JqoU(r$3eT?E;rK(@HPIe zQo-H@(Pv?y2yY2%Tcw6&i-iM}JoyyejlY;lzaT4!qIBI*Dp^4k6D+%$laKg(2_(Z6 zTjR0Yv0s#3nK|={$bI)!JUI=`Oo@Q$f7D=nduC8)c?%M;{BHdvele3$M|-dDoy z7;yidXz7;vI)IlQk&X&i_9e&)1lf?RtRW&-Wy31RzXaka*_DWhE283bM%H=uxgZ0X z54eX;X9R};dFlApr-Z)o*eyYUQ0BTzD6~Ldca)&0vN_pUW6$jokBw-}uG=>>u|u8T z92&xHb#8=ILFWe93i+>OOtoOHDI&=Ep%+9JEXaf7^Yi0HL^(cXe_59D+s}I_v|^bV z;n9e=l&2rkXbwT=E-ua|M&EY$L~%3d+@K|)zi_v3uaV_PwtA*M(BDDw{X z{t~jdBwvP)z>Z6{(~ZvgijW4xPe%;Vgftk57WE?klp|nSB81B!iniSAS4A&KFyeL? z5G?TV!nAE~?T~{CRrH!YDQ?cZ`(`F4#z(o-0ePBttrFYDexjG*#OaF#WktEUSs7<9 z1smlpwJlC^S0kS06>sx@R;^gMcEi6v*qN4EAKL(AqgNGD&}*F|rp{UX z(Cpda5x#-Wt<~4hR*HBsj9(nELq+6yB}MSk6&IF@UZj1&(nBhB(#yzy1;RH_hk$ns zBOC|pP$826G8$p}_i*#Vc)#N{j<)O?qg^9NXM*h{2R$5X6=dx)S-YrETR-@keJ03a z($izWXrOvw3D|CA=Y|eFCt{;RuEf?U&TE&6PzOk98#08)<=x|xGE>ssbtdqMIqu$K z#hmQ8JEykB1Hv*NE|r-*Ma)TH(>jYlbx~z*{{dd!#yU&z^6FE9Cra={2{gw@jAh^z zsV5$XQ^yXU6ghuwoc$$k6|9J5LL5us?`F@5+_ie7x8k0KwkEeM?(Y&wNI+bXQq4Un<($>gwT=k)1nrJVR7?u`4x*(f~)CnX6nT za;5DwgZ^$txhUpQzUFV|!J(!`Ocjb58VxwIS?;GqU_XszamBRiD5WFee35=zsr%vfcj2JP{50V!V71ew~ z^sTL_M)(Z=?#Um}(p9Yw3ya6uBfqK_xxdp%Y^}h~xF31rH4hG=+qQhJ}U= z2tz(Z>1F_u*MZ>@k>n0p^T8M=Yk9Bm`wQ=O>;|g z8+LGP?T7|VME;8pJFIqRIp3cB%i08d{ippk`RB*0z^BwvBk22smer*R%v4Rz{b$XZ z1?LX;5_YlQVIMkMtv(@DB1PAP2wrtkWSL4`vh%+wy6V_h}TJ!o7#GhioF%h}{3EGd!Cp%8>jpx77Z)6ot z_xKZzKOinHpi^V@LyHVWUyny$6Jy>G6AxDh!b!4ML4! zHVXpM@m?*g>KhgDoi1!MI2GLf)EIXA3Wm1OHZ@P1{=nkLo_%RqPY8s-#omqJAhgEr z+q8yWFO!MR)_Uw(wY0ajSk7jZJF5}d0QxsIo)YR7m=>j{l`LJ3*IDcBfDhKTw>4OP zK3T5QQxF5pIK9af8|#n+TE`Zm-Y=rww2mz>@jl*I-mxlh@G(>EdkiygHLO3pkwk;! z90Y<0Wk4KKpm(a{wPF4Gb?erzU%U1lyw`5nz}Uc|^7>}$S6dHXD&~BQY3W(lZWhk!)z*R|Mg=#Wqm48Q1@kjX z4Jf%B@TsRYrP;_>)!1HBSy@qDS5t*}6~K-wt1Wf)b)eX-&oIJ|MVx*VNf9+-)VOiO zhsV#DGI`SU>EogtJ!eACnJ|V4^BIhu3&t3BDXu^=w47_jMMT^HnRiZ2-IJ2CCpGnu zC&z5NJG31@J}0FBzwF;p6(vitV$i%5W9po&#<^X4SNHSsk2 zot;KgofDkP%8Gq`i_6NcOT$u^o|3xjO4ivE>HCUr9>ZGr79e9XPyxbEsgtyQ-8H3Z z@0oYs`R&+;Ckr>@jK>6!LSpw1Bw1?z*jWG0v3F}tBKp)MFaOF_7INpBM9(^W_P?mf zB~}UmpSRTMcdSXu<}Jk7I`P(;ICa+~5p+i3aId{!j9I2{*VY|^VM+DFAF76v; z<=x%a$5Jlko;;P7b}A$5a_&{M)xQL(`6H-A(DD_6Q0xv|Y>$@blAox~Uafzwvk38tqv1Ox=wd5HKJn zK0ba-P>>pN;>FpgPoKVAR>lR0w=^%sK)<@|^*AP|UU`1;gB}5o{A(JP^pAnAq4>ow z2xtYmy=t}tZJjEZy4xDK{uXR_f&1G#BqYSt4>5&7!NIaBG5GYMBfsZfJfD6#9oksh zNoZk55O$iD2JH;UtjyEtY^7l8ZhmY%V*$#2RefZ Z45&OeSD3Au{DRD*>1T?{E}>lZ{{Y=nJdywa literal 0 HcmV?d00001 diff --git a/css/Inter/Inter-VariableFont_opsz,wght.ttf b/css/Inter/Inter-VariableFont_opsz,wght.ttf new file mode 100644 index 0000000000000000000000000000000000000000..e31b51e3e9388ae61767c692885e5d77ff7b5346 GIT binary patch literal 874708 zcmd?ScU)A*`aeEr&hFXW!@9Jk2#AP^*iaF%D>m$1V`A*RYizM3(HL8z(Zs~W7-Niy z#*(Nps3;axR4kxkLji?F5OC?R^qt>(b{9*MdvEUjeD3%2$B%hE@8_A`o|!XqcIH3? zLWm1MK?e2g-XnBm=r8XRq6;I$r*qFi1Bb5P{QDlFa9&C1hzUK14)1z*+lJYM9QPq) z?3caX=rSZawsRXo0*VM}e5^n03FAEsgoyAb`vwke7U=fQ^aH0{@_iv&YZt|LJ%6L~ud=lRlXOf6eO0j)?yr{C}D{e!)C& z2ZWyluAVw;>6C*_KXxTT*?K~2nx-QBJg0**i8SLWk+SoqO&&ifFnZ)~2>%q}gQvm4 zb+z{>*f+vnZ`$mIOR|n+JtSo0FN8D=pEY;lc)RvWg zIpb$fp84}OS2UK95KZ*FxeFHFoH?eH2+KkUSnW0l0|R)>xE?ktEXM8$v>O zG@DHuW4ET}b~o8?itpx@Et~&j@#S$8r0cjbYknJR-=Ua@)^rVv*L}v50ROJst!ftX zOM+D-)Rv1vSzGvwC<#%vRK&n%yOq-zvXOp|ULi^mu9PT}31v>?2L_wvZ^+R8LwSK1 zv(`tQM88BcAPx#LMi~d@gI3_>azK1a^h86nbLTBsMs#rFp&;)XzO-4MSrA>yG3PDzxz6rUNO+F1d*eY|)6ImX;T-%ZjEEDT|eGP1U(dD@u-IC-m==Qz8tIF zU%H^*1fBvAeNGhY0*}KItTA_wB3@+U?SS2V;4h$*yg&2=BZf2~l((mV(aGCTfdBM` z<7WvWNXM51kai@L3?XC446>N4BAdu|a)2Bs7f395MADHK>L;|dw6diIgm!$~S=@=x z-jajWy#s;1+T1%L?`pXdaepCqMk%>FVxaHPYc{z#L9f~57QEL`eqK+?dz+$HYH1+F~2^{*X0%kxlZVEs&5W|IXXOY>|pB|2$> zO%};0NwUcb;-NI!WF-k!9=FL7+;`Yy6&a@NWs}uJtqifr8sek$vB`FjE%ihhw6DlD z1Yhy7EsTx?i~DSH&6j0ECEUf&tnTPfEoy10i=Zd2;vAdoLcGKwHrbUl68&wm8wsHq zHrbujqfs`wCgfc<*#q)>HrbPSP+L0)UXX3|5^51A8f**WO&qA7O|A`hy-oHZO@!*a z>JWF~md)Ll_zH(?a$OQ2thCAXh!RJ(1P7b!51ljH0CrduzTy(C<%dax0{>$|eURuX#4PHPV@4 zliNVQ2iWAc2p?*b+d-}_e|yN)ZPfvCfGtc%$kpxE39`4%y)$H2n;e3&$u_wQB!ba8 zm&_*P$sE#~ECioLrjv=NeH)Z-A$X9DH@EN>Hun~!wap%E9?dMs_wh%50$sb*&Sn4nbxz9q% zO%Z<~;>;&Y5qdmAk4H)`Q|}4acP;j5kQTsi5yFOAYQGS%d0d|JLX=`Y(wj_%!#>Ya z=BAdiE`aQV6kbn<*J(PmGC@MRmtXB;tnP6AOvr7UNbJwMg1qEbzb-3Z8ZmN zu+S2EA@bqzUk$@+GSSw5T$`+Yy#T#19jQ;W^w2*F(;Z=``kMyl0V*yGSY7585J|ZVtO2y021fkxr#O8J2rSu^! zP%=(FuaxZdl5^eXr5TSEgjE~IqsREzdv!#Nhb}KWw2tw(ET4J{uGO zoQ!rJV(HV_&=+2+E~tgIMHax$>r@@SH)N|m`e01=wv@}NEwB1i*NwL)p96VXTx;fI zX7sW2i_eP`1CUB}noUuT$ri2U^XU9PwQCT{+Z?6^X>Zf2rg%nh6NH(1B77#myiCvM z59$5A`0WvEDCrMs|4LeaEZZOLub2E~8Td>)5A!IW*{pNB51u!?4#8*(uDQI=UxxP~ zgAjHu)|Qh|BR*SqL9g(cleEO_7UF%zAa7yP8AmcmxxP&%f;ou_r))PUx^0bQ?VRa zDOO>nNfZ)LtxyAHg$#61)C2k{{DF-XO@IN47Qi4yYhYUi*7AyeihjTWiXp&ZirK)q zin+jd6!U=#706xjuHpmWO2yZ}ZxjcChZMg7k0~w#uPIW1X^LmSOhpbbPmu>KP?(9J z6qG2B(oSgy)G2j92c;v>S?L0FQ+fgGE9(OrDH{QsD4PLWC{a#jure6fM%feF;Wchwsa4eASD2wNjXGR5f%1Vs$r^Oz}c#Gz>ihX9M$Kl&w*d4z5pIjT>@TF zMFVfDZUS$sZUZw^CSZxG1p8@G4Xsd5RUn+0s3mt+L{I$XoaSQ2JNhAr9nGu+H26xnogQdzz|IrU^h)SU{6g? zU|&sNV1EtTTr*HJ5I9&f82FY3ZLXQEnFn01Spob|vj(_MgF0$9YxV<=YM>{YYnmIt zo0>S_1I+_qf+iLCOorad1Lb#!C@++k12@Qzfdz5_uvErd4?AKqT&neIA?9BX1O zA_j7?KMOD)5+;(9!!5!F@*7-OfASOgaEe9P!-dx7Jfw4^D;F@W?rX}Ekv%m>y0dr|g@!mVh$Yj$gj8+CYX^(VszF3X@HS+ZH}pG2|uX0<~o_Uju_j-em zCi@FJvQEs0WGxM~@SLSx!CS)K$-;A&b!H)~2}yc4h;?U;NOD0v7RnlvxpW}<%1IDb z)MI_j$wKG%{aGJ2m~5LJWZ~b~c;w_J7W;V%e~+xHY6tsT^03T}^<#eI)^cAqfCZ6d zlV`Gl%%8+9scqr+mw2+laKE&;1{=z{l4I|Q7M`@Qf(`*p>`I5wGZk!8>arW#j=0P$)sA410tB)t^E!;5Kfpb9`t+jA@v>LoM>55@HYc|yQVTD8yPS<Rug59<$LQ&@=LH)aN#3e~;6 zq%E7vR*~O^k6?@8KBRvywuH4M7w5ZL`12V$@Q&oo_S>*bsbSRr6=svX%#ApcqVp(mr zgdE(e5z#qazAmZ7{Djc2OB~pnB<`DHH|9(dw`X)>?MdQ}0@R$`-BExBA@Ms39N27b z(}F$RQCNdbBB|femPoqYcV-8qneiRkl$(VhR}X zuu(tuw&3>YnTwUVvI)fWDBYU{lcxBHCChrbLyhs;etSvo` z?1CZWDQG2IiBhk`m|sagV(+u}$=7T;rpsfjFY8Orux_jyiDqpugD-R-@E%^QBuo_Ls4_l7Q z4apv0eX?##s?dO}x8RpwUQoM{PqzN-;VyLkZk1Ui*S`;|lu69)<7Fyx8z@4)TEsLX|Iy8gAtX!1$ymZGQR!`P%84fzU~O$u(_ zY@i`q@2nOxNYcFxRUYKt{VkOh9?=bx7YgXYwp;)Nc7)fi1iQMk?GGm?b3!e zs#f~lzbvk5wELH3RernAEw1w2eQvqA?oZ*A*sX?jo}c*nMB9epU%cmWH+)}s8~^arW7I+62~FgNCy#em zcQ~1rTD$kj%v8_br;;0#C{KACLpyL76mhjjWygrvf%W!AJP8)lB8*jQTAhA;)UV0u zba$_^(^)(H8gpoJI@`25=yXB5-0r7MfzSJ$Hnn@+jYId- z@N7yGWz)0iHMLG>wU4{@IrrVj+Ue&a+SE>u3?JpyB=Y>*a+An-Euf+qTA zd;OE>xGqJ_qaU{`bBRu`Q`tP)SX(ki7c?qp7A?tb+FVU&C>LF=s!`G4s$Affebp|c zM0G8xjWX?;(MOSe{Y20F`q$5ouB>(aMpvJh>*;Rhyz3ct#q8_3&Sl=$3!RH=ae(9k zC~?mBzV6Je{&k9K+%P=Jt9K(~AWOMn^vr8|!#J=anFFi?E1usl4u+IBuq5S1k-yO= z=3-0MASS7vu0hP>Mowul$-%jvF@}+4>X=kJ!7(P&v!qT;wr8Q9g9`^A4i>itG1;{s z<@7?}oIZej&s-M{bz({a@|`*8V=4l&^|43C>r!Hq#^=xzK7Q5xwx>+(NzwD+9q+2JONw;rB`_-*~E8a)#a4W-8_PqUV z>$-t)nWA5DTxJWu^ti(QMPgi0Ku(Rg(hfN_?u2&~Gwxhz6I1ify=JbtcOEvTPwymn zX}s?gx6jwzDejoBy;IUYzs8-i_SqKD-6`w9rH*Ox>NT zj+xp!tbJyUJ6iW_`#U;^eEU024u}DOVT8s)pLVushfbM+GLEV>sTtV_C=@~qJF-cvu7>%FvR@JP!9WIlWDdC#PW zZ$-Q3G6%bR7pI(p?{&?PqX(6zfrsJ-X8q)<*&9-j5QxJac)J(BqlYqbIGNyFN;71uFnjYYuJU zk=8oh<jy!kt#PRSEZ+rMV>B<1rq37P}-o+X|29OIoc2qZ?8&)=YEfP&2{k3CMOv(XtyMT&{hgTS8vVl$r_o&MCDM z3M;bfClpq3XqxSkP!gJEpHNYOa21^Zs#XBE;8QLM1IyDEBwiSi=8~A)P_5)3C8jh~ zD-sQjU`=oRRG*mT^wcLY%L`Jr)+i?yNKb2X@JTdD$v%lCp-*dY&?T12PdyS#>%=*6 za89fUP1Yq=Qbf}wX9Okc8p6#f`Dswn#Sy7`4lYSCeIccFOmR$l?vm=3lnAV;PiOFErZh&{?vW)q`gUe zlGZfI15oDjq()kbv% zT@#ZXlHE)+C)qz5Qs~BHhZLmSCdJ_Tq;*PVzhsSJU5HCx!`VSKn;N2a_r0s2;p-f9`DWs((hCWoLCASh(X+rbo{nGjL zm43TfhC_NnEeFnSB1s z)Vy`&N>)na%IxR&nMU{gLCdFw&!0bcsl%bc^E`K@z47uWIm#G4UfJ75q=zYWlz+`S5 zq&FC|@A(&3W+%40UyyAWl+qykx#mt0hvMv9&CTMR*hWr@oV)$9PUbuql2eqE7?Ksv z;bcy7a6(Z|a!C5goHSpxBKO>iE{?g8-LBQn&64jG=04ZlF5pm@n-QfYFXOU~QpW%N=$;gFnX>?I`URXua4pI4RXP&ZGglq&N)jA}*xGq)@=2bPag z<>%+ufMia#ua|E&*!vcQdnCIToc2g`FL>gaEEWhQIbIxU74n5wq0!Jdw=g$R=~{@) za|`o4lvzcWf>?P`iCR%mT%2oXS6o~Skd1ae#d40Fw<$^v^*6>B&9asncH(6 z2m6w&zVLYN_RPMdFi~(SQKj3dOHX$+t4nWpHa9PQ;-hvgEi6>kD6KHcg=P6ltyor3 z0jMfB)h-jt3%of%sst#@b846Km0)>Zres%M>|H69^JBpG8kk3@*tXxR+Q)LO2<#hz1 zveL)mCEyRSm8CjIs#cpzsds#W9}P6zb`sNSrBX$sj{$ZzNl1`SE!_vu{W}%a+BIDm6xh3C55W8OqJ8;dY-DF`}HCf zW12Fx;2oZ?VM0-%Mp?6@Sl(+0^p_70H@b7EDWB+!gjG5k!jzD7Zz?9KHiaBpn192j^t9?$SVqg}$=P{TBP zeU6=-V{x(eYD*zs+rfB6UE{W2p{B-d|3a0H&rLe1{Pzm`XH##)*yoQ1`0FY%9PW0` z%X3H`2jJ@~2Yn6WGsp0br5OM*!_mLobjyhe6; zJ!o@%JaiUX8x;>dKD9#+l{V-xf5hv(F_Y*qiC}W+us2p8XgYPoMP*EhbB!Jw{(~zL z*yZY!=;%~4rjL`ehq^{?l4o0&gaohjCQu=HH^h5;dm9Ws zF5XY>`o{FBDc5zclNjf3FvwB<`ar`Gf9~73UQ(s5sj2R@bAj9w z@E)1JGuUxt(vm$FecI1ytjPH5ZWOkqDPNN@q!Q)FB4wEC{qXd z>c-afVs8evzgg76^{!M^aPK4krVo-DHjV$$zg4`^UmveG#3nw{n-giWKKX{vu;gUf z$B_BdJGHb{t@JNFywa0BDluX9gri0G9^??Fe@K*&rA$51YKw@jJ|*@AG6`rH2dW@F zdPI=HgAbF@qkqI`E)`W(B~lIdY*qb}1+odXf?9Khmi7Jy9X7qry6kK{isF`;>E>2y zSLIf2SK*Oqa;s}^OoFq?-#dP9u&L|gc8xDz?qCQsbue{^@7ORt1=&Op4_m8R-o$K^ z+dny_4>Vq20=YoVB+A<;-$)LEf^_gs)tSrjDAXF`$r0W=M!igodVQ>=wz9WeRVJ1R zRjgQ1R#A>=83Pi%DnoA(%-b7n?2o?GN0C7P*m%5*xAvvwJ@YC?LpC2B)&Khg3!5G> zg-$y9&5~P@aJ#1U^NkHO$hYq|eR?n7PKf_Kv`uB_qU$SJwqD9-MOxbZt-Z9Dh$I+vI8a3*`8}UctjTl`=2I-Fs z8uZ)1fyevw3GdY_s)Hd`Z@3vR-?huadLbTd2hdQjaz;6Mo#7^tu$&Wvd{*jEv_=?DO zjvyFho$ixH@$vrg7-uG9dcJ*@nHCdW1yNRBzEN4FsCpzaq7yRtb3~9B%@Zj#Q_)OS z_z{axXS3+c6e>ZXs;5lBiP%avk{LkadJqiY2vP^R^8>7kAL-uML%%(aGsvlpQ`@DP z(!4WrGpW&3Ov@!R74WKGfr(w)-r4N(+|^tw)hD?w^Eb4*-Xmi0;URlQ|2XZG<$xtm zOI*3~)BHoJkx5B*sYzKnqp={hKvQH_%f41EztkQ*MvunfVV)~S1DYQVOU-y>l<19? zBbSzU&=Dxz1@i@Rp;Npmo)|DfiX<~ZIe1a=N=+(PLCN$+|M(tc@Zk}u_(cUjuHh8# zT+18O&&k(h8`(`xk?SM_zmD<4wmd(&EA({m9q+sPWSt=}$1BGtr=7{QTsAYwBnR0J ze=2_5^Gk#w%~-^!wyuBsjy-w~8a)Z`Y%T9me~frwC}y(GJD^2ikHMpH&dQ?HA8**W zKjPBmxO+*DQ`0jtjoBsTm1aq;vD4Z+>fLJj_|^?*)vibA;1Q#zEnI|>;GJLy%&ULZ zM`x{WR$q?Q?k`<K+sSHq3gBX6&i($)?I6__0`r@+pE2toV4BP`ailN`~roN zE}#qKg&hB(X-XGpLl9P_cGtRTwc6TRA9W422M_P$$bH~n^KTeYtI|qZq7}4MtI(=> z3i;kk-pAAiv{b@Wc`$tA6(6z4{ z$)5JMx%8oL;H=HRXWl>@t9dLnwU}notEQ0zBfH8g!2{(cQsAG=2q{q8MG92c)Yj0} zQF~hbxr_ASpN+zSw*Fr-&8gJ{(Eo|?qug|{pdW2OYt!2DR*tV3Yp4JoO4i!KLUxkX z)JdkR;nE1Eu?4yG@0+_$I!fsvI-F8%2ZbZOt8jGEY1>j?>MIZ7_=*`%$I@V%X+zu4 zUNjgdcc#5?tM&_~^XNRa9ml_6=2NTTakOeBoi9(5r`g=*tF^S}pUjI?52@j+26cF8f?ts>u4)ynL0xR7?hC2fmeLs!sjJo)WKKN?Lhiw&ht(ps@5Esu|Lh{wk|oK;R(LB0V+nLUk?WN|DN@^oI@im_GJ`JSa z)LYb0lc*Kj(pl7Zue+)p)D8~L z4h~umZ7r>z+EwkKttEBVhHC3OX{?-X)OJ-nshykzCl%)~8Y{{o49p@`<}K0#(tLv?g_?&XSte zrxU3cwen816YWSl=wzKnXZPRF!KsGWT5KkYqL)}#tcN46jbR)_cd;Fs*2ltw#WrFq z8cSo<3JHp=Q~mofw5_y_9qhHuwB58pPFig*xrVm4gQK>AgPk@&>+ev*L8}ecHqd%G zNlu~@bx=4E2hl-rplV6$sg~rPw2iEjme9r8j?(829#VVh18Iq@k;l@Q>oRwmX2J98~T)#eZohI6g=i~Ty+ z1kBHvx3Sl*-dodmMK|nExv}j_ZM#6e=ff`VRJ@LKL>-!<>kg&_K z?r`|t#=4)uPDQL;y>oqSe}Yv$o;MtAYkI8HeQi(f7wcckD!HdTo3DNOy4JFSeYw)) zYgRwodhx|7(Yp3~eGMSJitz}tVSyobfvGm#=XC9^2HMMP(_}bb|nnzoSD!M?d zDLdd9$6KhO)PSzj_QdYGk?d)CKD*%A%Ad4q^bn8D*AbRftbN4MNl@#hEfxuL{Y#xz zJIRZWZUQ!;RXJ z8?A=!mK?5uex+TCiTdeG@|X!}~MiW|w!VhyPk#=>wqO`1-Z(RZYo zY8O#s@#i|a7H?-9Xeb?E?N_*?N2Nw$9kGUW^<`ak$=;GTt&kSsH2)IXLh`|SsT4cf zkNFy2ena-eN(*{y(}KyCy_7{0q}sF_Esz7O@4s+Aqz`qq?SRiqb>yLRrCdwyD|Mk) zu@)xKTCOoJTxT5g^c>7S{tn5>PTO8;s%@`zc2a8FNsYAav`#eNGNPN&WVM}xMlI3@ z@(iu)q|mmOMrm7X9WiEVOOvFAa;Vf%_Qe_Ue)3yVU%46mmM-PmCry^U)Ec@}8Y+#1 zX+xLF?s5xxraY70mWN4$@wVikJc*u>dP}|0#-G#ca(}v6?L;4mILJq@;g-=^dRE#< z?>RU*IMJWw2~vNlzdV74NmJ$VbTgg%!t9_wP@}vYc??m%(Cq(xL2Sy6yi|Ti?j?Ij zpGcqJ3951MlGo$;n*P6^-+w1XZ71B>x+k}m8)1Ymk=kPh7{kXlAMaeo+F%T?jog#3 z9PkQVl9o%$iDWroDp|^8Cu?XM?k6_K#)#if(cwn6PTSFP0GNpU3k6uV!tevZcD5O1 zr9(|VgV1#ObLk6}8R^dw3IJ)kF%&}t`V`2*2Y45E{y zk8$T^9Sx$ju=|)NkCuDO?}{xYNN{Obm4QbS&gyEI#*4bo`&ZT`+%{~`C)zlHbtwFs&4G1>yI8bAEXRgCq|RuJ3; zu4?@D{A;&gDh9!oD2^zOm`AZMsf_y?4(89zO2JcTCbTt2n3v#YOtxTej$mzthqzAg zvH1b3BaFwLlF!X)=I3MuuCJt-Gg!DViallH%vtP|@B!||3^12h$)Yb=OV%{zGUG;9^$$~XA+3e zwXmfKBwg7XLI-vi2j9E1S%Mdf!zuY+*kke)8-!~KwJg7XSblZjy=ZRBdI(x_RB0Z_ zI^ZHsabxovxGa%Ga`om(>?0I^kHL`;E1^+$tIt^$J&{*k$_D_igpyU4PMC+q0PZsB5;Gkd=_>K{fXu@8(anXN=Gukt}l4Z$6xagij5EfeOj zA&De}MS?<-39h7Bh5`R^2)`<^2Dmi5=2!$y=yxR_?XCe%$+o;7(WEWOx|BoUOUP>D zZFIP69W2nx9~T*buh=VF62<^GVpFBKkqGbqWPInFTT6^ ziTRk|J7=uffU6dlQ%8b#6KXwaZ7yIsp_6MoQ?j|j%yr|LjB69Ale@8+xQHN*8EfJF z=e5H9sM93e^F8K?dnc)*x`1~h*C%-~FBU9}IU2(1uogm#Pu~XbEwudTZRQ6>dobA1 z!dY((^T+*~;<0Adi0R4m(QQ~0sLu7bJy^=~g*8VyOASWwsU+T^JMQ^}k(3pU zSs>isPR?RMxR3KhYQkDE2U60fDd$4%uJtW^`mhS9kRN%_%ih9S2riUBfs@8OVC||N z(>Ft>unxGNGWbRo3&9}SwrnKs`8bdRLqfrulgz$1I2RndMX?^Z*YaVHh0wbJg3q8) zxDU~dlz5M2{h3VWj~R#iIwILOu`}-JxR8{keK;31?_eMye80}^*xR_YHhXdxHrAX> z_RJs7#^M^-f-w`=WLBH(n*9du&&(ou8SZQvem#tN?1)EDwYz}S?&6z)h&BKtKH*XM|&jt$ey)L*Y7D}=oH04}qohh=V=8s8#RZk54 zx5??@!Qi7v%sqSX=_KjyS+>l4MA#&SvSq9%ZafTN%Tb?w!O(wvjp&Ox-S8{sYxi=@ z4+X!qjY3#0q2AgXPPl6FWNrKnwt^hxjvj*iS>;_8LY{4@EMh}R@fX?Cac$(mR;dpz z)@DH{n@*M<3YDFoWPp=%hM|J zM7sRfJ1ud^i2d5i6L+Xi{n|&3D#rX;$J0E4t~wMKXf_F7fw5tN$Dz0Y^O4} zrX7t9z}=`(oCOH;ISUnZoH>%T-(#AxY2^3cpW36G6T)sZ!L6ztVX@8GBBAH8>owVO zl6fpQmt=xnZ)`p*tUrFO5mO6GIMWL6apoaR;;f0#owF$OyM-6sc_Fc}Qwj_;5pFteF#JayR*m#yC8te!BAvqr+3oOKoSoJ|l`aW;u; z;%pJ=cFM~hm0r(n6Up?5OZ8Et;haq)+ag$f++}JHn>kK+?{ri>q&bDN31sH!@(lJK zvDnsNXnoD^>@V!v#RgQ6uIkwEle{V=c(--xB(x zqH&+poz4c5sy%1gvSDPyBd5!_1ozoyV+ZyY`C(r`V-z_3!cX0CJ#%yMkKND+JEAMk zfo;ES@6=XMd(Q$ph*HC9r_(-~aK0*5u>;-_+VgA`fi2A56jye&g;=b{{-{ zuPOQd;Jte_$>D`QMZ=d^V1#->1%~OIr019n=Qz(-*3k>CFNo3^YX~1u)Q~$kWa#l zH+;ywV;`2~lC8(qTyISFANv^i;Mg~qx;Sdd<>MQ3a>(lAyRJ1Nr;h&wynOtq#U%{# zq43qf!{O_rYhmKq80}AvhJOiM8@@Z*Ulwur0{2G{c zvh~HjBj`+%z~xF_Q4OZCad(|e=pk=3Vv0j}jJtU6r@ zdDm%@D3fo_e1EYnW|RFF>td$(>VhxXdv+V}%d@+IYtJ44eqzDZXAeP6JZr`TGXLxl z$hXdYQr3<{oeL{*B$p%CNBWQ_kw4(W8|xzvMCyQt;IcCEC~#BcsmNr(DKZZ7s>sA- zANFP>!4&e71>-o9k-!0h&f+pB@(uV-h#Uj=Bj+?RV97oEFwrjN{&Xg`kkaH!88V$*>{AmkA!Fo>$7W@_c{ktq;$Pkdszz8&xpnqhFWRJF&WGaaN!8mq(Qj z`}q76=Sv&YyXU5FKDDsw$Yw0>`+mM|$wR--zh0Ev_w$p>9a?TF_b?6k;^cB&i!bAQ z*xlOt^R&7*x1Jo<;KtUQ-BpQSt((&6#@DBY_{DC!AC%F3o9py!mu*2KQhI&!uziD` z+Ye9Dg>8@Y%!&N=yQPlfzP%FKDCygWt<$>faO-=#+jqyhxVHQ5Sa;XfJ5LU+6T9G-(Ua0zVogfi#-nSitO$l%%RP$sD<$%y9^zYLw02~f7*PP4EQ0ehgyaj#yR*zxDz<~Sbu`Pfj$+W@EAKOg5%zs|#&`@)CTirW|6 z&mmyn)3JBk?2jK**m!@T{hihaid^G@57d+5IviZJ_*~P25$|}mI(TJ7OwV6##3cs) zx^rQ{#b0+9918q3rD>Agq1gU$frlQ2#Q7fHxaeHhBdZrnU61Tqe5?79#9lX=A3YFy zx5d%3lYeS()G7X}Ccl68ZcNkPe_s&OH0*~Nss~{|yptUic63TX^RVzYZ#afsZW7ll zEOy3?CSeaI)}mp?AXtl?i?zqXCrRPQ&d#HIkDc#)z45Wz?*i^DKgf>Zbaue8%mvr$ zAIp6oZpFm_!TEaQ%mk1oCzef;zrkgmCXGZ~I-7d?gZa2tsJQWud=W*)M61Zh}0Lm*b`$eo-bg6#C z5AOhe1CKrvoD90?v;GTzngF>G`D@5tk-Iis=G)z=J&FzC!y;^-pgc6;Pt1Kd6WH0%Lr-Tz|Bna&~4JkIoOcCp#nZL==;o()?9I6V!JM=vxw z+bQIN@42I8zID%?d+S2|bLah3P0w9@>wKNa9|k|E6M1|Yrn1vxB7-6$r$M?f2@o{_ z5IqiXZ6x5vQ$Q@)rii>X0McD=z@rer<8~Z6M0!Ui2SLi>vtm{~fN3>{<&o;ha&p-x zvQiBz;l55z0B<~gBAbjoTYUb*chA+na6r0DFWd?|>vSQ}KW@l{g5tBv3*{tkNK}HZ zaAZ_rpVK!kJ)M4zUjAaNx%~3i3p{);hfRvky?lPGSr;7@7(Vvu*?JfJuIBdtz2>!@ z6O4PWotSg3-?jJ|$K$R!_=fq#oSHylV$O^sw_>>zcgEMb@J`(9)LM7qXGJ*NeKhNr zVRw_<%ZA1ucXlNYPfS1e?9mP8@0EDg>#oD&gR`$0pL{bczQz-S*MrLBUuNujnjGF) ze>3^M@2;%mc>m)~lM`LxmKd<>1|A1_BstG3XHFRrZuS)!_r@L&GcN~T zcgp%=siD5{$>>v!bMxwcZp=@5|HmZLFW&yDl2mMpN^gd!OYx^xm;Kh=^>O9FnND>o zWBVx-Rfbt7uT_ocf%^a(gK-Qx*}brerMMSV2q*jL?rW237MAONpZHagi=kU~Ov|sm zV=62sd-*96yw#D))um23?in8G-Ugp!zhwX9CdmQGfyqJ1ZIZ1E>ktxxN6pJWtQcNe z;1>cL$7=VNF7Qx>2)z_Xz}JwV|8x4COi+TPEHXhw!E-<+P;V=}Qa+7E=HXv0_uq`K zDkMR#lClE{`vd8vVWh(gw|KHl`LAi7CCp_KCOt-6@V}%KqL*8cFiwh_|4(Tx&rg|w zwrmJ$fi}k1AXRHX?|_z~Ek9zbx&NQabrR(~$(|`JoUBxqqCNfwNjicPWB;0z1t?bm z*=yrmevPDHlnri@gVkMSAYp1-7?mB$-VZd4Oi+IY+6Vd)v;%IrR+9Fz)jS+e<6nj4 zVW=PK+mCa04Rn(pC9dGbaJS=rkmEpkATuZ*Zusq&d}VX*4Q_ya7U&xqM0SFQiJi#= zxL*@jk_qB^v~d(D5flx&1*&T$YkElc5nFAIV=afcj*R-xh?Tz@Pl>+A7w0BO=z9r$ zkFUT@ka*wUuu^qcMSIpk0bO`u|C~%xaC%_lkd@=mAAdp8aip~uezUFqinVCZKM^y( zaJxVP6+aL?zC{YZW^qDvFwUilQUMssTu3E+z^+NkslOFX9!n_2#^!>{?O6ca_B5#CS z2MQqWN?)?@-=<#>w-NN>pHVXsqHG{UzZRwl@p3?%|2K)}@h4*CuRdQe?*#oLQZ6EG zU(k1uIdQi+c&>(#HWtq1e@1T-iTkTmXeZ7Gp|5`eZG#Rxx4Ln6NnuW}4ufw_F8fT#TX%?=qg*gEJe@ByO7RjKv?nr;a^XmxsIB>|)Eb!H!EYjJ+5k`t2 zdORzpal-S!!mI6n;_s0miVL6!v~L7i57JoZ6zm|?cNX`Z_mjg*wR}8 zdpaoWCGmXN5#;k8JwrNke|p(UkWbi%$4Td9QA~Q3Q2y8K#EaW)Whz1(p3g&W7f0a# z4K!ybsHHxZFuiEFg};w5sWv)eSzAm%-y8zXA$oBLs2=);uQ`Iqdb+JzW~s2(r9rI9 zl71rEnf_zq{qQO&%ekFhv9$Rg{9X;mW$6IfsdPg48_wUU9ooD3Ekkd1G>?j-IXR;>4dfI_fdc@M<*VA$)Z zvDZ;0SV{VUeeo~D@bqdUudzt;0@gg>D2HCz7<&kmMR6L>n@vdX5NWHrPueQ-$@@ya z2N@2+{y|x3UE3?KTplqy?&&VvNhb3EN;uN23#hcK$%L&N`(pe??yfm*uU+^^Ka1q8^Zo^r=?GS z!MxND^8}~0Hr|g+fjmvQ1-#ntuedYf{vF-^pM=4C0_*dLU9s`Q;44A%@H~nJ?R$ma zg?t}=L#>=Yx7gdDd!R7vIf51V|0&pmHhg^!#+p8Zy}TLeR%%7YNi;DHN`lK+cc>i+Ab7vjw3wAh~g~%zZ3S2mNoPg z;;l#}b3||O^B9*qNI#65qCy0BCY*S-jujKE_@3QAF-s5TUtnEU^;Vb^zEyU-J#$<`&9LDhz zj9(M>NbA50!69S)xeTXPKj7uJg!>upAZ2s136ugl1NsWI6m%5yHRw2Kn@z3`_de`A z%vny<`fdpK_eeF(5`72*%>xawa{TWqtNkzCmCc0#RwCzY>2l8Lu8qRLFTO(D&sHu$ zg8U!EBbQNTQccPR7;l)1X@9Z{y0?pStS@#c+{h4h02u;#2;?aW9k`B6q0h-9&^VC6 zM$5sMi_^(xpylZIWw5^*<~2Y3Hw9n3r&>Y{SWoR2B-)Tfa#=h?K2z{_+0Dr~e$bYa zVx>?`L6FDbxv>Sb1G4)c!~e6e)D`vOgmYCGGjZZtvdcna$Pyt3WpO2gU>_tzlPTgS zWQr1VuL^a2)h^_cP2vNxi4MTLHvnyvg*MuOG$FsB8CE}w9kS?5CV>XwnKB79NZFSh z0%e0@ZDauFbP*Ky3UU9RL3=^}Nt|SHFK!g(kogw6gubI>KHeK%R1PDP@J?c1HDO#1 zQN)rV$}d6bn46xFF?i;V5%-W);t!;om`DZ)TgjXFUmoKWabylCOtBH`hW9ZyIFa2- zXT*ypr>K#oD>aw{mLZL4(Enf(DtiWV0BoPr16C$D_IVvIw!g;G_x(WL4z0cjxKKDNNta;wlSP^;LC+ zUenmfb3QL)pX2b?5-O6n*_;#Kb$*!wbzmN}wmQowtBZ``JpKO);Nj!8{@QCe9kDQgDEB~KglpVGrhfgZNPcByXJgh_n2GD~0l_uwh|;p()!4CQ%u z%phn317ZCCku<)PTFwrsA$oI`oGz`O(@h2l!{&R{T~}EBc3ix+U)wpP5UDw=VwsTBZ*)|F5Yzs|J4GDD*A9Uw(qf|MNCV zapJYAn)b;!-QWB2 z_iM~py0yP1*qiy1zg}lq(%Y;TX|vwLx($tdCT)Ixf?uzG7FWjRG_PlOAXQJ$8 z-NEc(-MUY8=l)Z$r}Q7~4+VQj^l{AJL+oJRhJ6(~AhD-n^d9Vg2IpJ%FK&~AH9@l< za4f+c>-T0o-Cqx1e`722xanpt$9n7x)^oST@ny3n?XNqV`c13>Pcv)2!J4gEXZ5+c zzIt;#)2w?Mck+Wh7^WHiX&q6|XP-K;J{Zm`|7oso*4m6aYg4zbX+?tdt6SE+BEkAn zpo2Tkro4JF`N~7t{B>&P_=9e2#a^cyoBa>-VbQ*Z2}v^I0-aQdBu|llCCS`YjXm;QA2O(~_eZ=~o63ZYKR4{hpcE zJGE$+cd_5RM>@L?GpEVGTqHH|rDpEYNE#V;+G%p6sWh=amyR~!6YX!l{jqe5JjI-% z5o;@rrHPqmM6y%&&2cw2q0yUJn=st)&;3|~Xl5VB4jRZ*zpa|^@i_YU=iqx|8Scl2 z4V!f-I~Qfm%9`&e<~jJW@1gHK##pbfKB?lbpPDsPpBsOe{@CxUP5)~8cGHiV^xGzyBtRqDW68>o!Xx4tsx@T-2Ya(VHB?zDBL(TdqX(pTYY1U-@b=O&PRx8GA5u+w?NrnMnQZ#&-G6ys?s58}`>@O@D$v`j{Pl+KF;HtS>V5pJAZR zni>7!KiTPq9^XvV!<$knx=%_a{Z5_C!Ohgw94Q<9kbZEupHKfD?Q(y`9=clWRk9v7 z!093bu&YA$b}8gNE2U#~(PJIo?S9Ucy$Vv<>_6yc- z4od-p^~#Owl(mtuZ}NMDc~2jsD{QwJ?V-63GIqR(dtql?KuY*BX{vUtFFcOlB{G>X zL;UMMNH=Y$>c@YXHnZPA8f|nQ@_Gxz-N%~Yr&q&|^8|aQcXE`uS*077@Yja5uJn{+ zru`-Lds6x^2V*{F_-hE~op2a-!{&rsj%!s=U4exAb-q_Z#{C-8qmUKIhGbX*#PwG< zVEV>CNXWJAFcP@#yRGBgk?&u_WLPTlXGXXWxe;*fua>X~xdcoFp7N9eYBg(B3G_Uv zr#HfJI1kmBFM^qa8n{0VP$ki$nZGjce@ioU=d*Fn)V*u@Vr!U+#x+PX|@v zTy0J|aIJR2tqy({z=6aW`I^ShF|K>axL+fW*WL%xx>fQ0H$PF>40?_x-Zi=-e{5meTs3&Uqv+cMb5-=owEMb1}L}hcY(#a1^GV$$8hUm+8=Q(0i(as zas2D}`6~-!lmfvRg=a_#8OZO31V8G2TxSgg8vEx;w6j~FzuE5VC_nexI3wvF2U;^( zQv#zq)qy^rdn z$!2_YV+~=Os4kMpNk?D(pj0yVRh#`mA&>N9-A5}59B?%ncw7S351IgE=IW6 z7B~ohiX;sJeybye4K3h1;CsrHkQeA@Qa%ZtU?|YHrCblxMk)oFp*Yk8!l&vBjO$aq z0UyFQa8V?66mmgXcpRRGQLqr!!ag`Dl14#hC=U2bLwV9{g#&O#BrQ5mn+-|<&hHo_Nho;MDY*DT!MvXIv-PeE^(0q?>!Uz@XQjTg zQr}tW$FgDrS-*x0BH273KiSAnHu96L4GaYImks@8+YX1}vdBH?@16`$1gb$Z=mq0p z383qHz65lg9lzPLLunx1?CqcrP~Pm6B|BxwPPuc?&E?1n*liB%HV1Z_1G~*ZJ?D56 z&{vLc;a8EINpLSb2v0x<7z%UX9ry%(f~z9A?t;SbFgye3G8g*KH4o5HF7%iiJ?2JF zx$^^cn43DxO&#XOE^_|>e~8>$85+au@E%}y_hNT>u)93iT^{T%PXp)(!(cY7guNnp zNjER)=A{nvQipjdcivCoJD}|Os>8F;8zum0ggMMjAfrL09M zYf;Ks6ulIE3fjU;fUb&_1>!D7+{Jc_6vtkR*MR15Ql!Kr!0t+7cO|jAlGt6zI?x6N z!gN>zdqhgr1oCx%en6M^qs#lzWoZl0W$BWDE=#w9{xAjJfsaHUcoH_i5x62!CJhvT zO3(BVXmnS2^-kZUtbB9BA? zn|lPicm%t61Uq_U6f6Yl_K|&XQly#!{8htWHT+e39$tcJ@Fq~F)xL#aMXD#kz3?F5 zw>o~S4~AK=3W&2haaJdPHPAgjP zg_2MUT0wunUoHIA!e6bQ;17}7=&*KPcnGk$+MNI!syz?h1^m{=Z|$ohkKF}@;bCYD z=lA5?&bK>t&(B5W0TybTP5>97K}!(pHd^$Ax$J5cxamy0}64}KMC z&{yP1?kP`Fzfbmq$*>Hzz(M$h|4|kR3_`W$r|oJ5lCNbAfVnq8y!V$O^@wI&1^(lbwGP>5>HZ!h`TM zbcZo82e7HGMd5(R3w7X2I43N%0BLq>4}*ZdqdVp8L0joD5@y0m*a=5~dh3}63IOHl z*$DaodFxdYYQnS7Q=~U>_a^S%l&3do_NF|&DNpaofWCX9@7~1O`?5%%RFDVCL46=U zeaKIr>98E|+vgzsBGNYoxdFd@@!Pi>jD|(9j;TIn?)NHu2>alqNPh*H0loDnkNsOf zUqJ8uUl$p05Bv+T!2#%Qz}Ij=?`ot_IlCVCr@V z`WZrbhl~U2W(fX32h6z;32*f-gm8knb7fdj|QQLB3~@?-}HK z2KtzZoy^2eX7&K|I`f*ytfqhs&vv0FU^la0gZ1z^oMvgvf~-&yY65wgL%!#X0P-@2 zH0GuU>R~SRGM6~#lEz%>V(w8Ooq5>GywxJ}(eZqAJij751(az%@yve(eiM0xGQC21 zU-F;1Nd_nZ=w!){@TbV@wEx%1-|J)G9FMvb1I4Z2-}?%o>nO)M!mT5o^=SYdug4zNlg@h5Sw9je*ZTKiFPvbxO&~jzhbGVmXhR#4VLx0D z*_aCQKouaajXmHDzr2$dXv3SHfDSMiW&rx%v;(MvP1L~$DdAoy3y;I|fZq>h!77o> z=ymgEkq@!S56R<)lJ&lYUTLW?So(@tc0C#1g?neNCO3+5>VzHU11{NZ^vG^AhI(J z6oY!u6(+(`AfBDX^RYmC9(9U97nlI+;A@dj2=~cwm@l$B2aw0z_}Psee2Tt5%?BM} zHT){FCkfEuo{~@to`&wQM&vW}_SwJS31|hRy|*EB2GZC|8hek!4YqeP0_|oWcD9ea z?z<+kKRxsYYhBBk@&$GNL%g4ckP%ZUaX}zQGp0 z;rl__*+F!1@U+Oc*vhvjL=N2p_XGXdAg2#<9w<9LVqS(ohF}6giO_W{R97-IMRaw<4#eh@3`G zrW&a zLm$~X!7@?zgQsq&Gxd_MvCVa-$J z&kThDUFEL}EucHR43s55I?cZxK85e$qA0e_RDrvJaug^F*iwOKpbHF!`EXKH!Ky(1 z3zB}pv!V)hfW4v$li$KsfIQt-1hB39-WT;R>XNlfRiq|RhebvM8Fisk|GSClx4 zVnaobz%QbTMF3kVRshg>vD(lK2w!XnOa*jYjQES~0_wlmIZ<>Cs(3me-r}X91~h@; zun3Z2JCIHZ4frcj4cYw-Ogbl_Z{$rJy170qnEn1~@GK)>x_BoU1}-QTOAo z^kbqPSPS34AEL@+hO*EY&{rAsQHDH}Sq)o6mHih`KV=D57Co2k1TVrkmXP)C)p$I8XvOHr&#sw%|&a3xqEs;U6~AGr(Wi>ijbReJ!c18t*P zTfoMvjR0)5+UxKh?0~P}IQ%ZEx(k`1Fgyg`|k2Mrk2VK{B1+d#X*F@E&&g)Ktjc{F5 zJ>q|S8SEETpS;yC3h3bp>hOsNVUwr^glRzj8vG^d$wu%Rd?u=4I;al)U=^GY^;8ac z3dX|$QH>~TBkG`W1t1@d--ZtWoisi#stIW}Axx78pf0q9mtY3qx5*Cp9)1_~bZW>C z72zr90wdrRcn>~<<8VV%(~M9Qs>8F;8z#W(@Bw@U=S4l^LJoKU>Oxz131-0Cumiq_ z-$gY`4f&xWJSFPcqEKB_^Ym~ZVC&7BLJt@Vi(vzxhvw+NMFg@zDX0ytVE{~p<**eF z!6i{GQ$QXl4-KFr41>9l47=edToctQJ=_OXp(*r$v9K66zyUbJmUslRK`E#WtziI6 zh2^jn4#6c+&!vDoP#zjUM;HckMYW+$+Z2aaf&RG7=Wtq7Tl(U*S)nA*w%fLX{xAjJ zgpc4`_*GQ9B)AtIgvSBBw?psku+4V#A?@(n4m*B62awkD#P|Gr@EIHz)t)rk4+Q$r z4o?E%JJ8>=7O6UX2!!o$QB=n$YO;1PHRdcruMop-0N=}w(>r_Op%XFaI19@JTnC!hlig;_wHJw68F z>}dgM_j~~A0(H}qy6yQ2P=`IK!=A?94N<+Q-(L3v_1miz^oJ?%27CzLz(rBLQ$v2J z2(-Q4U0?#d4j;f*a8^_w((aQLibHjvJbfrnAIj6`Rag(7!)Z}{@!OYt_Qh{s{Puko zdIRzHCEmU};5eY8es@7(co?3B?l1<3v)@PXE&M8~e-h+^2cRxI2mN6ZP^SKrsXuw` ze_qr87byP#{0_kHfVS`w%z(Fn^aqgsfE%J-q>f&sj$S1F7aId*dXe;BTnNPZB6@le zJq;wzf%gD99rzfug1$id1Ig>awQyS0OBQ5>2LV4XanE~+{(De9psyZ8KR>7)^nr=6 z6uAEnqJJ37J!SC2@C3XIU&94aL%2r{DFpN_Ll|ESp?wbFJ}_iGd=8{L1YHhI3FL8T zQ6Qb6V_`mU4;p$@)UbPj{11B`(A}^%VIzD7*G0YD61oBR#g`YsIw0)Jr$i0ckOfLW z4QK}3Ux$;%@Wp`N;d@1mP>>c1LS=Xw$m@s&Kwd|X*O3pwlh6S+!2waD$kQnFIqEl2 zqtV-F@;Ca7s4<+!Oabyb<|Fu4)L3*k7Tt{{U*jHy^`gd;r|}y^O^5>ZHZdj>)}6O(_7xF_pTXIt$(qHLWnL0P-{)|I@LV>0FzE9nGMOGb_NyqGsg= z^e}6>sM(AaW_O1%fSs`3s^*}FIVVKTCBC`DHLp0J=lRb7Wq743ppyl}vEWrv3-5s* zq823q*I%74YVnJ(4}K8!S|RvG)RJs)AGClkM7{nL5bsjrUAjlq8|3ee5r7_+q5EaD zk!8C9+hQG4y_p)YjW2D}dM0e12h;Z{@! z{H&n7Z`TH{t*iv-dDU!K23z1Dp!3x+KxeFdD%L&~YoBU$Hy90z0NYu;A5MvSM?)6C z?>qGY_vE_J7G47Kll(U90P>RjyQp{D0Qz223#gy>>cACI@4o?SfqcJz9BznOn-TIu zMR*Fjzz87TwQs>Tz{b{6wskI0H|xp*d02-&*S!qH&3dR>KNx6N8<012g^@55mIC?O z_&i(_wP^xS#t&?GN7Uv=0Nehsy{IkdWeYaG1smV;CQ$!dz5&wzCT@r3u(un`hGf_ZUy9nt^?h95$Mt=qfH?Qx1$lw;?tcV`Yrk>-Nz~^a6a~uoIXe2B zcs@TV>Og9s9u7PUBY^T8!2Z6-3H<>1%Lq{3uWXVL+#cX`_eH&-c&3W>H6IA4iDiNL3gHpNe9wQ~f|* zfB0F{k52*p+K;=Hq;raVoazLBiaLEa!W#+Jb0)5fmEZ*+e;20#b#alhU;ItfFU8>_ zI05AGSN#6k7&^cXQI}Febs+qu*{}f)i~8+Bm?i3R9w0xLY0sA@!8`DSsNYiqb@BT^ zSO(`sUC9T1;jE}Xo(A&pXFb6F{v_?IBZ0bQO;TN3C+aWi;jf;eu9tvqqHffNX0TYa znD^IJf!XjqoQE5tRca^=&0rnu7Oib~7S@Zl(nB{m2xs8B=tvsK2_=9yBgDzGvYKBZ z(GlXcd3l%J3UIfV!zS1(+KIqZ@I3T~rSLxN0^(;aP`kwIHipTt6L5Dg32*m>Oi&0O z2jcg7!U*^dNSFU?*U`J6EVPH$;eu%MTJu<2plnHvpdD-h%AcYdj1ip@KPd^9vMX#5 zo$6kw0QG>lQjwoj9|7s4eiW7fc~AX0{46>Rc}r6O&_^2blQtJT1P!4JjDlsb3-F(g za;78Rbi?5d_z(_>z6*WdbszMGRq%!A^u=H-%!TDZ{?b#H^!QJI2Cj*|I|bYi^`SGo zBRWG2n!{^=4l~w-mjRt-LRXoxKuKr{gvm4=Ho|A3GiL*2}`a>}O1>ekLF?d#WlET?2(#rlTiCj{vUZ`TuB z%$pWHiSp*bBuODDC6%O>G?G@*$z76OGD;@NELkM0WRrU&yX26Zl1p;Sy^=@rNeN+#*{WBu z;(80m5_$#4l6p4BQhGSY`*k;trFAoo59miZmeHj-mesji^(tOWr{Y*#{mHQeU3ahI zCDne8rPQX@9iMNj*0k>2rIT9Ordx|vYI4Uetvaf){&Bc}9MrjIr*5iWmu{^)t8QKS zqB?dbZ>mlA=R3Di&AWFmR$MjYSVGn2SW;EtSW1=Qc)u#bv9!v=@d1^EV;PmEd(W2L zRjhl@uH99n2eI)lInQ`T>= zGWk+9q>R4w2vP=LGE0Pa`x5&+D!m_T8vn>%5plEYjO|vG%Ko>s{xQivrtps`Ii|#R z@5XjBW4HHUv$?R>yx3|%VNZ)V*kL1%`=zX(P9vNO;u8Fs5tB2rXJNJ(VvyEcCE z&&Su)t6eW&#r+jmR-9hGO!;i(b=llyV`XDyc9vOMX5<6A9_aW$hEj)0?J2dQ)WA~Z zA$zHTC9lIFcpoN1d#D8IOARcs_kj#0R+Ly!{8Y}RIZNcsl;d*trgz8EH@K^Gx@+l9 zrTa47*avo{E0A_{nwe?Zq)C_hLh3y!cBUwuG%$80);wCmlSsKp3hSV?)aqqb)z|cJ z-A!FryVXopKsmg`ll!%wbH;w+xQq(edoxEaSl`F{k&Ez-@?yHUE}={EuG#x}Yw`oSjCpUheo&X!59tcL z!Mu{Ltnc+^db7OQriRp2&0ed{th2~HI-AZWIlMYvJ<4!1y=?UCGi0XBVzeqFv9a>hyHRIZNEp zj0kR~@+fzOQ*v6)$XPkZc;R;}7{$uCf#Qmt2<{M(Ii`OGT8e z9OWucMZFpx_i*k>+xUp=kgtgUkerlsUPdpYn$#XLAFEgBrlpY-v{qZ| ztk0~EtYqs$>mBPo>l5o;YmfD@^{KVS`oLOmZL~I7Ypq?@25X=7zO~odZEd!;SX-^_ z)(&f@war`Xz2+_PUiX%IZ+OeRH$Cq1)cPcADz!e{S|t&0h&TE+9p{nM`nY~-95ln&#q=Sop zbmwiwi`~iYUOz;6yN*-Y>BbwWUv)>i+iwpMO&7f@nkJeynmU>)nm&3r8qoGm`(qJ% zx4n;VVUE-hIqTWHGuiVZp5;Zon3v?G@KSoIy)?YdJr!^DzRSDrRrhXqcY7JU^jJeC zFSC~gd#K^n^d9wUdGANX%NkW)Ht!xUyO$%Xy_{Yy?_RHfm(R=X<@X9kEw7MQ*t;(p z@rrmwy?=Sdyy9L7ucTMTE9I5;9`Nq>9`wq3<-GFVLtX{1qF33g$QtIrq;drQP8dk4L5yuIE&Z@2e}_nG&p_l38|+wbji7C39Y zW8MkxxOXU;(mU;K^3Hgld*6EBd5694y(8Wa-jCi--cj#o?-%ciciFq@UGjeO{_uYF ze)rCL=e+aY1@EGF&HIabKFe4$HFxB!w3#aMu+-uH+DKYS8+k$c$QYR{Q~ZAEHU1az zb=v(-*-zU)>Gvj=c>pA;(x|(5ll46+yUM5X^G53uijj+Is2ZzhRcqB(4N=2cj2x*Z zt0`)#nyzN3SJVRamU>sMQSYhu)q3@T+N`##UFtKnSM5`us{`tgI;@VUU(|2vhIVw4 zPQkPz6+Lp1u=lN|tLu8YseVSc)tz-;J&1mN0{!_6{ia^7-_omevVK=@*4y-FdY}GA zAEH&C)Ti|sdVtHO^;?z|u~J%Tth81}tFTqXDrJ?m>RQiOEv%MSE332B#p-IkV0E*) zTZ643)=+DB;*K-Xnrtnw7Fvs}SFI)1GV4ui6@A)zdMwj-eL#P;jb7^u>nrP!^__LZ z`oa3yI&Gbc=!hMe8(9;1&wj>kW}mG5_96Q_`+NI{{iFSpebkxnyyqNr zzIA?ZjylJkv(b-Y>0)=sio{;#o#Z28V|Xw5^4NQ^wXt=vqp{<$6S0%A)3GzLv$1or z3rQ+TCuK^?oRlT$9)3kCY{lG1YPieXH{Ip#TkZ3;0)azAl*yPvvy+|S&-?mlj&El}FvDDy!aVpcuoH|}GMZyZ5@qTvjZ?o_MNMcRRVA-7aod_XW3`+uiNq_H=u>z1==;U$>vz-yPt- z=niyWatFDC-68H!cbNOKJKP;XZ##&zFCWv)YhkHc(A+hHa{`-Nb&H(a$4{O&(TJ5_D(|Gy6tKDd+|L&+iKCk$XZ1|t%E&tuo@GX5_BRi;} zDq$UU4{K>mSWk#9!G8|UaAQd_f-JX&UBmE&X%mN`M@ zVx5y^9u_*CUnhFic~xG)R+sSWH%nQWU5L%TDT|!9owvn|wMWWgW5MzoHat#V$BrlR zTe{mBgTG-cT9#qcpU86VqkH5ne|ERRpF_Rv&kk2ILVa4^@n;{&%wR{$8h>uJh53Da z7OYtNV@?1O=LU113^qFpOxf<^7|fHI={gylj0WS6)#}RXs%2JjnB=fN7-Q6@q!gI- z789-#M||POT*-3h6|P++q(!g6niBO;8JRZi&+~#*%?ydY$f@MK;EZw>`F)a^y*y|? z=2UdLIwPHh^fauuhxr^N577hFvFrMMLw#znf&C;k*w}7qx3b&VZSB_fb9OuXdAp3VushpbZW%vywO_Ei+1>3Pc2B#P-P`VC_qF@k{p|tvK>H4*)Q9} z?Gg4!dz3xee$gIdkG03y<0((tzn6y|ucpN8IrcnBvR|=ZkyQ3Vdy%BJ7u#>(f0eyT zGBUE*$osJvxk+X-a+9o#-F}pN>{E`#4+KP=n3QqOI~Sy!bJe-Zy4dX{{-^OvO@AC1 z9|g{_=bAL_d8EJCe$8HLzhN)2Uk^t#XY8|#YR-q_noIU?_GKsPk8l34|747FEga?W z6HJcf#7u6;iK)TIXdi-x8&k?>&_dG9+~vJ4g`K=kJ}19Zz$xeyatb^5IsbBsI7OXe zPI0G%Q_?Br-0zfj9&pMyWu0=)gHCzpA*TYhTiJQo@7HV4&T2WeX=invdd}lcedkH% z38w)zXe`!H8u>m}Br>s{jcxjSAM7z9&0-eHPEW#({o91wG+&xzPhnhVT;nlm<|kIX z-@WO^s-)JVHG5>CFVt}wS`ndHQZyPh*?LMwihpHrXH8o9z$nE%rzDR(qSh z-QK~>#*DoGjb$?S&f(^CbGf-0gXeMcy7}DvZc+C>_g`)iw}4yFE#wwPk1^|e>wE0h zv`gi;%Lr}cRk3Lm$@DR6><#?1%lG#8au=h1eu0x#7?oU%@iS6Krr&!cnuV2$quwrW zCF=xDxzm+kokEzC>|y@1fU$c!w}D%M^iyzeDP(7`BausyA0vAr??)Cy#zzK3+C}QK zewmXM%*)ncR-;yOZyIQ|wrW_ZSX(-X_7=0E*`$&pqs9-_ObSSD}<{#$Nh%Z z6l;)hGb?)RU5aC7V2S&H_U{8WW0Betr7v$tkJ-p>${N2Vu5EZm3pN|c@bH&d`vzZW3|CG+fI&u{Cu3cTgG_7m%=NX)F!=)D?AZl&Ixy|mgv>E(37hraXGv)R%5l=?6&(-=-xvQ z!W|nn=Y(tgzj2BYOEc9>-(tbKKK?QhM>QjbzPTf74e>j($vY)dtoo{R@sBPL`ZiY% zu$q6seE2P~AQt1kffP6JxqO@J64G0uj@;%p8Q(k!66An&G^GaNZxGKCuyjqzo#P+w zJ4$tirM`QFl{7x=3n-3CP4rzYRY_Hnk$kTN{d_mg2#}t@j0060-$e;`1orrpV&982 zl&T6bON+;mfQX9i^40<37!;3rjrjTu)r%o5?%-(B?|^aE~K5dmL4C zYss(M=$9ov)^D%J>2cD5b-C62T6MC1kN2Rg)th80J>@o;&3)!$S-@T9GkKLYy2G-R zS;0A3!@kfp*`}{s4P?93&}yRYv6@=FRBrBu!_`A(R-kH|S%Ip@J@K;YXkCpsYEtA0 zJCmAcXR)*BR&K%fFB0^zS3N2$|D_rtC?qsd;9-N~SXRIVo#dw;Cmf)L8YZYN}qt zih8IE>QA-F`rZ0NeN2YxtKD{1yNu4nn6s%aZBMW#=titn&DM?W74{0<%E{tn(XE|z z&U*cvv(ee8+d3aOn{_*9yR%)lcXm3T=nl>v=L_A%x#8TX1~QG&Fbl0r-#arJG*Vgq_P3zwP=Xxq-;Rl{NmA8h!|6?h)aY zWd92HL_f@PGo^(oC$B=||hrQm_YKN2*FWo;t}uY$@2kEy#HOihn$vaN)ThT+Hz_@kCs9 zo)Qw{b|gy<%a^hXT`reJ?2C?P+F)RNQnQIJSPobULj2XwHZY8%g^PSx6 zMWk{g&Y#TGk2!~&{q*>ooMgtj3s`v?>kMJVwX@UOX~Mj^8e`HD%+a$u=^3Aiec3+4 z3dGl}LvOX$*(;fm&9x`9-Z7Ask@oB~HDp($ie1JoV&~xwnZ|Y^*CM|}PDBnz4n%fE zHnS_WJhCVxy;GI?B@!`{9xk{*x2_S10_hPWWG)@K3Mn`(K^#zdGT6Rl@(Ogn!za z?~hq<;v6q8>r`$|`U&(yh(OvDN*hBdIh0n1QV`Z?D<0P9E0BV)MrVOr5Z34|?q)O> zNI_Vmy}&I9Ycv?R1!0XA<8DTeffR%_x(wWcutuM8H>1%&3c?z#25v!EquIbM2y3(( zcQg78q#&%(ao`q&4fKqr5@C%L4{M}&SR=*58Yv#uNbz*hZ5TGtvvCW;26{GbLD)dg z_zJ@s8#VbeZb8_MVY-2y&D9{?K+mKQhBX!&k0sEvxf-M!=$V{^=^EROhY0j+t_I}| z^laRMbOSw8iZHCPYwbQar4Y;(9hx5H`>= zr47RddNyuB*g((5EeIRv8U2J|13eqJAZ(y#;}(P^4}Kk?(J*YFXX6%x4fJf>g0O*} zxfX^E^laRMuz{Y9TM#zTGqHtX13eqJAZ(y#;}#D~s);-sDXwQD#r15YcvvIF^=zad zY@lay9EJ__Y}|sdfu4<95H`>=WemdxdNyuB*g((5EeIRv8Lfn213eqJAZ(y#;}(P^ z|9%~z$1rT5XX6%x4fJf>g0O*}aS6i)dNyuB*g((5EgqJTiFAz=*Rzr0dNxu#tdZh+ zHc}8a&@-uqVFNuIw;*hwXX6%x4fISt!?1y#jav{l(6ez1!UlS#WMSAq&&DkX8|c}% z1z`g{ql++Xpl9P2gbnm;+=8%_-LE6G7lsY=Y}|sdfu4=qDrRL%)LfqP8p%^TJyd(u zoRzEEJPTGv72$cFEIdcy@Qmp%a)Mdj0of&+S?e7}cC*7nyv+5>15md)@ z8TBRm$JHDyzqIjl@O_oWaYFnSSc8-U(x6EQO4?f^>)31k@`|S zUr%M6K1lb{9T}xJ(sguI#^}X(LOPpH$73_sSrI2C9aXVP)_aeL^4B2bf80)@yjuZ;_s<$LnFtBD%8j*pxX$4W7j=#S9{+ z&cJhnT3ul+?x;G*vxVEq;cA{Fo6nxiXjWc(5qC?~h!wA@tYDPliL*Q^n@Z0LRs=ik zBi-CQ*x(hH{}^JfT}BVoJ(#I9*9~=T<|$=(8Zi&Elr-8=*LYU(1kWrUV8>@O&owSr zi+HAeyc(wZsjjLG`LC~Ps0yS}Naa)+NP+iBvJ!Tb9ihFlT{h7YR2?C%Wp zQ@__wJ;|g_PRv~9ihtgeIB!gxlN0CFiSsJXDf||5+CQ%*y!pP0b6)I#o)8~#$;1_x zOl)z<#21%LjB&}tX(W2WPzu7*6NYX?-rZdZ#?aXoJI`eokZ{00%pfTQO5~Jnc z`C0C#THtoS)^#z@I^X^oW{Iq24F<9GX{|z76@9%*4$Z-*^8FYpU}n!KfvyVVx8 zm=Q=vdfYN9C;Iw@r_#)S7A<0Be-LZQEg4>K)QJ?;$Mqjy(!g+)&UxNx?{siFI-OWG?!wb|-JBl&6SsYN)^ULIBF{Pwat5=i zJk%NP40B$l1clknt->1YlZ>QBusS=Rr-MG@*`y=v#oCMyYgv!;bW&UPVJETk`Xx`~ z9JQ|f+uWO{*W$Xlx!+*+$czO>qGn|E6O!UTPygY%J|0}xN5ZzwYsqPW`TTxCtIS*> zCiL#dhwyzM9-!4Bmd{d zdnyBanyYydXg<5gqv@5*4$K?u^>py1rJ?j%C@l`9S3~LbP+Ag7i$ZB(D7_L&^FwJ~ zD9sHe_AC=Q;QtpB(t=Rp>AZy7%ut#UO8iHB;_9?eni@(|LdpDELA;ccLTO?sO$a5P z&P@0l7fNG8iQgeeTpb-sqe5w9D2)iE;i2?$C=Cmxp`kP+lm>^=pip`#lm>p3 z%YaboA4>f~sc$It2_^osArZD$DD@1b9--7dl)8n|3!&6Cl=%IGM0`9Ch7`6eYYUd; zw>-C#f5lF;2+OnwYbotUc&>r=cxx#A_5bUwf_}j5=(mbvq4ZNI{TNC=gwoHUbTpKX zgc47$CF1)oln#Z`x1n?}l)ee2uS4l@D18-5Uxv~bp>!aWJ`bh+p~Qb$Ci1d3l=v@> zgxj7_`ZSbwhtemZv@4W84yB!;v?G+ZhtjrC+8Rn9h0>N#`Y@C>htdb3v?-J}hSG*m zS|3X5LTPO%y&p>Nh0>Z(dN-7kL+PDRS{+I&L+R~MS`kW&0TP;A9!hT}q=){;{*US{ z=4QA3*V%3VXJ-D#ENqkJiM>RdT;}KNwl>Ll)U-)|9O#!r(Ne>(skcg%R7X;MnQCjQ zWvTk4s+Mw4%3dkUr;Mc-o?<|X`YAFcT~A8(_rfQ#2VOpUEV`Wk4Nd01Q@irtuPOLX zy)ryqo6XhEUPe0cJ>)(1RGyvB#NP0s$n40(NUz9!{@!jw_H?!0tOv3#GnVVMVS;4n=5A{ z5qyn{&NN=5z0x3B(ny{IK2uZ_0l{gQHa@7cFAO&!OukNwDD=3!slu00>?Af zd0Zx0=S-Z|S&k>IlemoIIcaiX{mF5m^Ydw*0kKy3?pP`L!$Qd))=5|_ zlYHf(p$WPDnk{ZJkK7&>qqXSK)W#6QH)+E2HoA43a2 zhUOfvSkK~e%4&u@-g?H5t*IYF+z#scKI`}{x7J-*T+Ui$IG(W{z|SPBH1bL7e%~jy zZk6+;^#8TmlR{Qu@|WEzL|hlGf*j9VriF~P@^L(C<>h$A%0rx|tlY@cty~<(TSYle zvWnn-*2;`L%gV{|l9hww1S>o4r>%_0W2{WbXRLdW$648sPgq$wp5(4xm8Yi4Ft-oe zpmi6%u3PDRjitrcSt|`LW3ALCgp~?;ypJEuS5ctvj^^eO$JA8sqh zNj#H^HuXl_PiyW&&Kb>}CpJcJ;5g3wx|ef8uj6t6W3uDf%brfGL*gs~(V>Cf?u?uW}H-52?+rah8=7mm|) zXO3gFX_XT+?U6jTMINL5R;m3~sh`LF-pE~avZ1Y5aNWcg*;0?#&KF$4>kN4tNSrlGa(XIK`SALw4sD`(mG-jvETnv zUn_+)Z7nISPHAG)v=ej`Bd$|AHS&2)tB9S|rUi`G3ZIW^S`k`H<>wMhI~S&;dqDcc&;3Mso%JAR$cP_{>pK@`i0|I zb<{)}eY+hicz3jNMazv{SG3lo^lCjv z{}=jVSJYbKJEcrp91s8{OKKIziE1Us3F>XGp61OTCSPyiGDa=u zI8M=KDIaZ?JboaSbDArgIG@4weX(L=aDUI7Nrz{2`F=vZ>z_~G*)OoBZSuO%uh|8} zcviiF%UCs!<9Icf<1{rN_v6%ToqwoV`hlb@YN{V{iXU>4A9At@!~X}E@Dn(WQ{!oi=$EzkBC#lA`pXI+D=r5G%H7=<~IZjYDaX+o9AdgWG zBmXbn-UG~vqHEW#)YVfx3`mZ%_Y5%PoHIzyNrL1Y3?LZ;3L*-K0xDUN3?c}Kk`*w5 z!@uXB}Sk|{c-d}?}{(YTK~(P2-kmX7I747-$jc*%1RujJz)eHq#QlHQC1jF z25AuoF`WThKq-m+`0ICq?kU&tZ3W6o?8WqTfjbLQx*OB0u$jnJ_V+szEC^3qu`II9 zQml7FEW!M1Vlk$Z#Ue~!|K~iyyEh#T&sFRjR~6QK3-gy@hqNVKTT}3MPFv#4#iYp6 z>}a+5M~m(M`0g&AisZvUY`b(B%WBE|5?@2j|r zlBNEWS0?C5d>>v1ojb0d*mNjSN=thze+_R{l%l`+aROH#mNk$3ovy*xQT~*gR+yhH znqZ214!<+veT?=<*W3;GbhO9FQwDFQbVfBJ+fn`o_}f(YiKk&Nd&HNhC;b?GL>B*V zJ-5kjGh%!}+B@tP`I`~@y#_y;(wPm@H!#u>r63OTGw_t7JdKT{z+u}wxHr<#HKvo0 z7Ah4V^Yn&B=jCB~6;>a){!yz;7Wf^!QXNdep5saJ_dF53f^&%64=~ap=YEdJ@2m7( zWF+?`Ed81rYnM#n63_1W4a;Oi0PK#6HK=TJ22f9oW%5Wa2nII!C6eN2WaI9ehPlZH1c(A zaF3#31rK78j#wRwmb+LNis@pqc6=R^z_+>KKY-ZFU=bC(pijNI^6CtQ1wS| zpU&v((h@B~bS7R z_bjHu!&?!)iho*I`!ZU?XM6W;8*e0jjYAvvEbpFu#2bM%e)ooB*(+%Ip5y&y+j_&W z>{k!Z0emHo=K#Kz$8!K*^$bFXEd(Su|lrprA% zUGN2eZ|wIaZ<%fF^~2m)Z>eqN^+h^gL_0mc2eU1`UidWzE&OkLi){-JPayAn%t=g> zW4^`oT+9hfPsbd?^i<4or0oUt3YcMO#0d{Aj;qikqp8;x^P|vXV47{>wZZ(e=s$qb zSul2sXP?2>E^pX|UTe&c#ONwhY<-OE;-TGgDf)HP^;%+n82TN&X6tw@Fh2yn5MH$p zd(AOF2raG?Y)#zJ6xuuKxiZjJ$8>MpB>%*AG4J( z|C7Vh2Q8-+G5@2(69Mh06)^2<%VXNd;yYoqua?D_D$4)C#o$*jwDXq1-+J28Sa!?d zY2knG=xNczmcsl^N6(Dzwj|c=W=q8MLB4jyeV}*@J+ZskVwiTeMKSGUi(uN(7RI!L zErjW#wjieMZ2@fC4x@75`Q@iMJeB+#4);_4y2HK1|BlA;_OIcSROMfdd4TDa82Uyp zdbd`u%Gx&D|2TqUga15wV@&QpMsB2W8*TQ#j@-!OHp(~upN`wuBRXp1RT{C8$7}p! zw8noPtC2@))C2!9Qloy6M{4}PGfpE$QTca00~%33gvcG}|N00nF$ZEB#vH^*8-E=W z=iiOF@f1dk#8{F4WgMJIe~cj+^Bs?e^S?6^M`WzY|D%y6$?L+i?Z1yT`3~OK|I;{= zng1MPl17+(_-|uO@(7bt|DTR9`N#N@2P0!k^2m~Zj4S!iqe}LOj4Ao#U&fRi9UW7W z#xVZ>KB8o!|1ydajiA&QBPo%^&@tq&w+t+k{9KMRo;6LPN z^fURH{VaY~KbxODGAic($(WcT@@9&6tmdb2kEFL$^sJ@(C4OsoAM~3>D;Lda)C~Ny zlUXBEy00cM=#MBcB1Qy^T15T!B6z!@`512nG(Y3zh9)s~AG84DO@S6jiZGT&94N|o-$Q9!2H+u9uPDx-Kevc{fm;I0>D)>(-UrZ9pbVDNNbod1czMj% zgjN8RF<%E-g~7MoLeP4^I}feK;5%IbUnPQ<2Bi_^0eYkg_-_%s>Cjq?w+Q-h6nf{a z9fg+B@rXvCZWJ1$qF$74(E3p(LK{SR9g5NFBP@W@_QY49F(U(L6Gmo$Hf7KkNYI$; zk#X0lOqw&2^0EaZDQztoITDI7-wAxOD>L5_i9%=pN-@su7)H;`{bG3I<^+<6+i z1EXvaawimH))R6!^if70f_7l!2`EO1A|#b>C&t4V6{zcs;H`mnVKAbCz`Jz>8v~{M z0@j0eXDscr2ZOQO1nLVSC@P0ujG{d2%_vHHA4c_sMp5H2-;YsmLi;o7UFc&BJR^Y` zjR^341nM{Kl8I=qjz^MDsfsCd!KgH;5&_Rr*G91ikDu*GAru-htXiC>GMn4T5 z&geIwBN&XDB~TL*K_7$CdVofg!nj!x^hGF*q5|~q(B~MBY-|+ct%5$!cnhIq6TrI- zozGY*JK6?Nzi71WM)2}L7cpK4x|s1+LYFXJJaj4Jt%fdRENx5n1njra<%~@SeV@U* zp%5z=?+TRm3#@@sx`B;_(td$G3nhC3HY=3M4|qR8*Mg6*oUYH08B6)Lj?vSh>lvF1 zx`FX7KsPcr6}pMB4?{OIcyktF3*%jde!|#o&`%jlf8Wa3?NIs~@Mb~j_ve6Y=8GtF zj$cO6(66GzK);Ss9l9e5+0{<48*~SIz&=2~_eY_14=|3-@gO6ML&=_w;5w}jJ<6ai zS)evE0^V~3>Ng{JKR{0~vM%%_IE`(o+|Dp)a}oF+DuVH$WS2mXfF?6&8xkUgk-tOF zGiW~&;sS#fC4uirBA|^)2r2`hvOzB~Dh$2MD7qJttz5%8bRB=ksD9AvAPvi@+-@>x z5fvE0EP|o?-z`R4D4i#4N9TtA%%DA5pq?lKTBwD%!+1YJ?=oo77UCE18@8PZrE>%N zP3S!ay*3c|ol%XT4;b|9h`KrWjq{a~L5mxzX&8EvOV~sN>`x-U2&%i19;32AaSjAM zD7n4S=GI3A$$^YD)yP*HSE(2z39Z=O3Y- zg2Gk^x|gFBAc7hM&BLI+I}(QkZ3aT&dY2&vwc(LCCK7g}W?(*nL0!3!iHt%y%DfC} ze}v4(s9Dhb4BZQ55`#8yAqy~cZ<7TX^&YekL)WJ)%&7IyA`D%xvM7VrdjvLT=-Q=g zl|b9RkR=$pj%7(kZHJa(=$b~GO9ZtRT843S9%UJ|7g{a~m3eta(Y6($JOr)CppSx( zl^E9&S~&{keHBJsfmV$|(_8bO;e=$C?E z+)-$|77Ti=ARu=X+D9u!lO45=G6MPtqv`lJ5o1$~S`e=H#>t;A|T=>vKWl+r}30hA7)--Zs1@*(so zMpN1bMIk#M%xFs6kSHHPhcfzI=&&ezpu-vTeiM?)i8uvDGU#(B&?3y7hCUPJ7W7%h zNa%A>et?c*(EASoexv*drMv*>8;HQY3`Ua^lJX8HvTL#lK)wR%n zt`PKP#-Z%xD-6BYqwPBa`4Dmfqnbh|GV~5EUu7ci_meOWTu$g~j3Qg3^Z~@xLl9hs z-YFxt1^aUGm?wJzE;p2H2%zsDg5feAT_aT9K#)yNV@zQv-HU-*44uKy`-Pmzh;N{? zqGX58X3&otL2(&%9{M)pT0zMsfGY|mJ0L0mDtn+~q4T1UUCfVC54wQSxu6RfR{*+* zagRVLtpt_%5=J+JE@fO1=&~r3-uDj3pFfIn68bTtFG1HuxeHy-=zGu&j0-_GGA;qS32a8* z)`4zeyc*C?7_T<;Q^s|KZe?6!=r+bZ3Ed7p$9^dPzhK-D=vNH+g6LBdLG^;}U|c@v z&L}ib_5su}=q|?5x$llr3%ZBVS)h9vksi8_Q5w3RL2EvOyE5uC=)ovWp@$fq2&MH1 z%F81R`lKRQYZS^?N)ymzU&o_R9+T|?^k+rD)+m(U-!hu)?o<@Y`_l~i!Xj`hqh5xd zWl#etVGx9P(Vn zGp;!_#8BM>`GFDMhbAztIFz;_I)c26D*=W55e~MbC@%onh@$fc8fB{rFzA0FRKX~e ze}x#T6HC=prSMZP5YyC0D3o6B^gcoF2&G$qAJa3I!+me z?g6SSqpv~BMZvYI$}^hmq5`9zgI0{P16m1G#<_QelDz_|XH``hsvlIxGmg%wHsiWO>oB4ov@S#Ubw$5{qw}uMP(7Av5T!q~Aw#|> z`OFc%fi`C7eM(VUfu{Uy%Fw%vYR2ex(B=%?|H$tp^krzvC}ekZ{wOO)c}e>sD2-Ge zz_o?azS>856bc;WWd{ITbCge=Kxe$~4TF-M0W%Ux`3}qoXtyXkp_E6&Tc8KyC=Yr@ zc?a5yG3TMZ8AoN?C(06NU(gRM1eABgCP3u`%xvi6QT9Qfh(dYtWRz6sfG8KC0~tp) z{1h03d=t>YQK;O9Fgkz^WwZ~a^9P#l4Z|5t`AO#wG?gdi3vkF^^)%xskI4>zZV!Ey zaWqdh4K(Hb3yem-E3yNiJ3z;P7cozHGnUbm7cVicDs&v<$QE8<9NGSOMn4aoz-T)D zL`Khmz6vHG&6Jm9XFwl@zRu_q(8-Lx4V}X1-=I?&LwWcn<7z>v?18HdoyNF^(CLgL zTc%?J*AzOFah0I67#};O=QQESj^;3i%H-`RbD?BUz){)KZ{i?$H_BP)yeRje^BGMx zuz+z?c4TM3;o4J+qV$9=j?x>tlyRM*%NR#x_a381L6=8)1NuJW$WB&(mDoS+^8?1! zhpqyv0quJYgZ>AC>h1{#yP>)d!j^!3#MlbZj~Pd0x{h&AL)SA7wx~8Rmh5CBW2-|q zfz3jAC!t%I*y7Mn81EbCr%WuRXDbt14!VtrEeqWaJ`*CA()Ky{62Y|#Lcd~y0?@A+ zdm6ffv2>m8WMWH0>3F-b?K$WkCbm3uFB4k^x{rx11>MiY7KR>RVv9l#GO)*rZPbtC>6V!)Z zWrDiUYv4PqPsh0qZeX7BKaKGyzi%=D<>&X{4%VTvzRLtu*1v!USPr|=g0VP`mW;*m zwPMgyQ)sM2SR7Lu#v&b5Pfei5r=a^RfqtTb>aYm(_!RUWK%h6Lp!+g`{-A>D%Lw!$ z74+^)Se&2E%mi;jvogV3(A*6AH41unCoImN-fswtJfL?PdVfJ5S)q$D!AvO1Ls!D` zS`4(4&)>H452=I20rfu}Iv0y+!~$2=Wp1ViRg8HNO6Ly@ogbA4Ft|_a z4;j-2x)!X%@_|tL+Xl>&Eo=mvFi-innQ?oeTNp!W`2>6lTiOTR${0%5wkTDh+rekp z52f*Q@D=9im|rty2lNjvV16g~2I~O5i!oFtyBUM@={-@ZK`BkZP)b>mmDyw_o0gi?IFj$r9O~orv6{R&)Gj0?V zD@AAu^%$xPG#2>S_6jtHk?Elh1fU*>jZzbuj&ZODvSUJ4hSD~`je};0QUUr9BT7Lt zM!~tzwM~cy(9Dc`4w@xOacEXXybR42r35s46xwGFhT2k0PDWON=894pnmbB4XdZ^z zg-kp{ZO$gdP@T641KiKt7%1+WCO@bMP&S04^uSgk(76|k0^2r)Krv7Y&^eR@bdHo( zK)$!3G?fLgZ9{1yDF4ed^qy@f?})2_j!ozH9iZa@w+u?>Nu&WfHUV2R)uP;nR*!NQ zT7#iyg`u(}U`LU(BE7mL<{xIrH#4=P+=td-$i7Y8C=a0Z7=0c}`vo4^Ae}St$nF|~ zMws`ZjX@L4lWmbL0*`F18EAp|hoF>>R>-TdP_kn{@5lz%1K}1!DSf~r8)?hXv(B^w z?XmuD=%Wn1TbT}^Bi1K7?Zl}0(9Voo4(-CY_n=)Fx{ggZ#-r=0JL6t~_F&xm(4LGN z5ADTxanRn3n*i;@xE0X8jGGAU$GDZy{)~GS`WWLrfIiN+Nzf-4w+i|s<6eUfVBBix zK*qfeeTs2wpo17U8A`Se+=tL1QOE{|GMdVn@{~Y%nBj~^*VYKeO@WSN+*;_lovpfEz|iEHvpY8aEqXHp1?~7 zrE>$+rer96KsSX_nuwy{jVPC(Z!+!$=v$0i3Y`{((m$Q?GDBxXDG8k!lpk~q;68?~WE|xU9Rs*^&{d41d|J)8_0TnpqdfbN zaT}m(8F>^+`2su%rDG6P0PPoe*`c&Ag6?OuJ@7o}#wcVDn;6eRH%FoS+!jVrT0V)= z9QrBa<$!Kw+zjY8#?ijFgU@h|Goha|j*jz1l(Nt-quhjk#W*VSucQ0~CHn?~@^xpF z-=NMNRHAHn!`Nxucy8lx%ysq6v0-chvanD2P7{JjrP5T9IK9u$a9NFEkQHnu-i$dr9J0qz)?=hZ$-e=qb z=mW;VX1$UOr6YY!#;YtuJhma+0B9WJo`U9P+#qNj#wJ0L*MuzqZ3r4+9(EhwjIl+b zEf|Yq#-qFm3%iYP%~)Jt@$DE}4%!|(ighYMJ2AEjv?pV`LVGc`AG9~mNy(7L;rpfeQLZ2TcCr*(0C z#UsCiF3{slfa^H^1h|alJ)p2dBG?Os4HEd)QN+Ur3_^~f`ub472;_YzmQgt0 zP&!6D3WYro3g;V2&xj7t42(j#gdSoghw77=iK(WoA@&D4iz|PeQXYRNo!S z#)tvX>JX_Sle-bl!~^O?lOrp?-X! zCJfc*g_<(5EwmYCirm9t`cv$QjUnj2;5*&&Zk3#~3{n`Zz;< zi$YH@dMxxwM$U&0VDwASfeiII3Q_t1)merpO@R6zg(w|>>MuidY#_Hl>3D$Z-a^9| z`3ZD5Lv?VW5sds4I+CF}xe%2TkXxZtHh}8sLeDaC8}vDb>g__K80u#edY;iopf50T z7j!g3^-!TP4E41My~yZqp_CUu?t{|#1APig=M3b2D4i$J$Dw2&KpudSO#rIL3XvTE zc@Rov52!vX^eQ6{Lnkph8TuL{Db24lIt4nJp?)->DUALOI+c+>LEm7g9xC)EBY%dz z#ZbLeXc{ALL&=5!)lr4WZh%@1C0hYh_Z*tVXgcm}Mqh`{VWlu*%-N48P(2a~34Bf;~yK-nVBk0=N!caSNh^{3daE*mN zWvIP5w3QLVpxYQV3c8&U!=axs>Urqrj2Hp^f}!^6(3gxD3H^$pcI(jBjCdNlgHdCk zI~j2d`VB+vg(153fH)4_%~1PcXb&ShK)JhdW50A#i65&E(4|X0qSQQqBH?r7J7oAe#fDcj4lWLmXY0|rx;xxdYX|v zpl29e0eY5^J)!3qYF`Q^GqM*ng`qa5(0NAEwR3^dm7y0IxgDCyP`gCv5+kV$FEi9e z5xT-iD$lD7wO53$G1P}KME4FrZ5JWBo`Iw?y1`HzMktMuR7N)$YR?FL&qyl&TMV^t zgnnSCe_!ZFM$`U&VkDjS&y1#h-)1D`!5v1^aqcqI2Qc&tLv{b5Ul~Vb`Wxf=LVsr* zmFYdk^@HAL9F^$<#`PC;8H%uv`(8UJ?hD}zSdb6;5Xg$;aZ>`^0qp(A!vrYrQ@~)?XmtQXa`23+`_P9LSBJ( zVI;~e+>Mbaw{UlWbE0~SFzxFp%x8uUVyHefJeYB?&F~P$yZ{}_I7%Dt1q9Xig@-fl z40HrT^;_YQj5`aZya3b=8h(ax$xz&P2&%^kKgT$fV|Wxp^&jEq83$Vqj{)PcKAqD9 zFhvM~atTibGqHRUbQYM0<)1+3gJoEL6#5=mjrly#HQ-~+BQL`1!3NAf3Ec=bW80$8 zEsQ`pg=t?vRE2)Z1eC6=jKj4R-o^x@pxc>%@`uh11W0!n*HReQ2i0GOVfW!3n4baN z32=`SFF|*K-I%BHrE&rS+l2Qr!D#3{CU_CLp9yH&1K<$$xfFVsap$1Nz;P^J4m|-* zVV{&{Dl-s_fl@hvfb5gD0l_$EGQc%Nb!=fO1K{vGOxxo=LUoGaOW-o*agB$s0NfKK z2w!I;UJN=cR_M-4_47FK> ze`ly2D146zCPMEs!3#np2oQ_9g&(000c;n9pqUu&BWPyETMNwsa4mbVm4s|e5D(4H z1lVCh4v-V;pbQgo0i?@AnIz-^*tduBO9(UGK`3pLh%{JeUdCGw&Bu5s_XL~+5m364 z81F2!0OKLA5(+W_&LN>BV{buAfzsGNrJ)Qchcq05mIoCv4_ip6#02S~l|dCOM;Rtm zWdb_yYK-?Lv^wL>hSp#L%BPx4KzUP(2`FD5W`YdR+Ds4!t;2X5p>-LH>ms2Z=Snv|or{j!cEbSZDQo<`(j$Gru+i79rO*xc7(pk*xt~$7&`z;c?}$$>vYC;fl_(E7HoIu9LDy6z75{N z^8QdNKMC{-D;X~@^aI8#2weqM!)8iDKV-ZD(6x+L9QqOX80!>)u4AY!Hi7IIc=@55 z7>};w&5V}-{elAM8LuLA8{<`ho?-&p-)Sb;06hcF3gJHiy}|@^obMTo zo34?&T?gv|g&J`#qOBkvN6BK-;qJ0N^YR~bgYUK4TuA#5fn?1r!hp)DD^1KJ9- z#_~PTM?f3QBVQBSf{vK~4B82F#r#faH^w5b61#)Zu;Paa1B_4Qd5H0;ypAwFl_%N9Ijr9odJ&{z zzBiQe0{G3L*BBr7rMxI7!lKOcqMQhu4~p|AsP9Q$lrdpZ{&_PpHVj4i5#B9mF2?4C z(!PL=hr$*J3%kjSYnbp-pbZ)CXDH5}@NPnp*1UZ%e;tav%!_>V--XU(`~}cij6WZ* z;+pYWK@H=>Uh?5Q2){KHl1f_F#P2OMyO&-w29)Bm62t6wJ){rJ&gvzcjQRper-Y9(E{gLbeBAGf z;<_nH>(qkgVf@FT$VbBOFGR6XjKzJR*l0%LIK>|@7y(n1$iaAXgjhO{37!#R8P*}( z2qBi^9!vzWLL>{uuLo5C%fw75&NI0v=C47E0pyd)2Ss@hY9q8MqqaeD4utv|ihBy7 zc0pS(3iqmHTyM#pu^joEjBA6?ZK1ek2#xD58RtyskKidSWf$%!FaHjWICTkSbhge+XEXwmw=_1N1i1w1MguT_l@M`UwC<1KXY|jnBSea2f{Rex>xlr{q#(bD*!s|PAU&32A1N7_*apyy05%+4 zg=PoH*Wfxd5#+^u8Zd=D+cVC+|sQWR8#o#Wh7DgoT9e4JNG6;K`XOQAJDJIsFt zMcJfu#5~S91!a>0`|&Bwy%<{_`UT_dfF1%^*AGFFCq!&RDDvh!{vM;D$mjDXFnz2x82u|0`AO*UP@f5YfFfTBALW)t`vT%? zDDsw&i=lAi_ZFh&0bw%>dADT*;bWBGyR9P#A2Efqu0k2c zC`tq+;{~x2QK!F9ZPas!kUbTn``%P{w9x?~eP_^_byl5Q=hKCBFazGggnhs5WOuM^)WzFBL zyW$VUACJEre>MK+_y?i4Ll44!I72vVI6ho3TrylSTr*rR+$`Ka+#%d4+$G#C+%No8 z`1$bo@Rab(@VxM{@api#;V;97!^gv?!Z*V|h3_ZCCKOC)o6tU?Tf*Rk$qDZytWH>$ zusLB{!r6o?3EwBmM4f08U1H|M?1_033nUgzERk3yv1($o#BPbB6JJT3m^eLgcH;Yq zA11C#+>&@8@lfKC#B+(MiI)?v9eFANxPHwCtXUqi9xSzfocVM z6zEf6e!i~A2Az~b;PLChDcWMlai#&do_y@HX#mm)&@U2Xqn2?byY6?JWl z6WvP>)6eR0u!PBakv^@z*T3lB%@{NPZW+GinLj;&;a%jz1B9CH`9c z?;*6tg;B#DekhzHoDeP)E)}j6t`%+=ZuzGrbPx9r4+@V7PY6#9&kD~EzZYH;ULXEC zd@Ot-d?x%u_;vzn+Y(AfEMZW>u!N}z^AkQy*qHE1!l{TQz*kN5B9@RXF$7B}@~0)V zOzfWc5-eeA;*7-QiE9!+O57Z^gtLhkqn6;o5;FeH5_;#Io=?LPlK!-WeXxY?`Pah| z=ED-!Caq7}oU}dZ%cNa@T7vg4mM{jE@N?7>GW}aiP|4|%izPQnZkgN#mheRKi^&s` z-%OsFyf}Gh^6BIsVF@lJT}sB3>?w&6ODLLBDWyhA`;<-*OE{DQQ#il#{K+(#=F-wd zEFnIv04$+l+7MX6gtVz?bJFIfElWF`b|URu)DrLn`qzK-9*eupSbfSL?Z4nZqj%xg z6Mhdwh-+i<+@*i_aBu(nzXxC6{`LBR>picxy>ecNE4T+-eh>sqIp7r}j_jliD+7V(NgD#TV|KPe^_2{N(empMUNAq|_%b;$7+Dx#SfW ze@n@qTK?j&iyLUGi-+kw;o^5fT*7@DdwJo~Tcm&eclFTKwA3T1_wd)dS1Zw9&RV(uWsr^#>T+D+bWCSr69$dJ6 zVa0_HE=;+ArwEo{Zx>5mIC5eCg*`|?;R}V%Z#X|UWo~jN`s*M6Ql30_Fy+~l1}XJZ zg5=A|7nA2C$0z4HcjMfpt=@O@#pRRlI!qMl>R6IVm*S9@Jc0W_F zUc&T*vx#`}@dYLn zn3(@iev{O+aK8eH1(FIp#A(SgFU!uXPh@>9d(P~+gvj%G&Ov!LLpSDWhUu$9Jd{6{ zqALpwTK>8E)};5~p<}%3VHh`iB}%;-`KJH)N1p8;|NO>&bE+-yd-@an@BJVA`*_6s z>dO7`cT(zI=dMJ##(!~8cOOqtL%#zR8z~?D+kcwIe*5>o#9m5=N171na)2tJVLH4& zK)0sbp6;i(ba6%E%EVQOt47P{UmVuKv=S@L)4w=8U*lSXcG2G*S^qkXYl^>vzy6D> z6;~fi8~(8@uI?Xmq;XXcb+JZV-MISyyxm{dimX}cpO^gW?>L-cTzv4zkw!$=S4FBEU#3;VIy%k<=yU&~A?V?eutwXHr zQy4k_G@@w#yRog$MaIrg!|3@K{Z}aQuxKROh;Cw{cwH*fsxF#{ z?xKh2DSCcDP)!xU4T5;QgtNtnK|E>&o}#3Pd>gKz?g{JwiPse^*B3La3>1 zh{DlKJ*FO4PpE!~AH4<96h1bO;Mtm6=h5*xhtBT(sB`L>dX}E8-_%!BBb|y^5t&6g zc~8WtETXAuE}E$pqPc1*TBuf{o$BOrsCJ^adQ$XJ14LgnQ1nwziT-Mkn5w3WH`EOA zrkW|eCz1WZMs1K;E;-K0t4ym;+ui7hP)gk$iI_X-YB^1F3UufB1`CWvZRiarF42(QOC$eIwTwGuxz3erWMdWNf+~slA^)qsj>mZlN<$9V)a)sqdeM#Qc z_gu&&xH_(`SHr9AI=aU0NjKCDbHg!$QdiZ><&*i;F_$E3X~ZT~S{zoN$)-9{eyr!o zb$Y&BuNTM-dZFB?7pa#;29-`cq%w#aDnU$E)5J72TV_<>%4RyRY_9XUuzOe?#`rs3 zR1Yy-%@M0yF=&h&ZA?7f00>vW3nsTk0e^ zPIr?p>+W)sUMx53CGt%@P`;~QbcqF4o=Im12f zp3sN&3HTbn8R2TV+ODB%=9;=nzHvRwP&L;LL}Zk+>Yi@x8o4HjNz&VOc8}^&>J3-V z^>n>leZ0#qv&GyqwuG(Vo^?;Vk+!O>?w$0$jVtOd(=@cy_Z4Fcgx*-`Ug+D zSKRw3oF-jp%7WLEQo zSb?uQJ}|$UCSDfP%**E9bRWC7+&VYSt#{Mi26@zzriz!{RP#Jj9TAyps5Pd!m&46) z8{JGd%WX1c%}qDk{3PDTHz6y{?{1FUEDM=C>UU9E#fvg3Bx>Pnp9yN5Y2miGx7{c1 z9rvl5>$aMfUQSunZIjj9b~!{hki*<(a=824z3aZvcl3{lF7y+;y`SAY_oaK@EphYR zSEiV`ERT7@EpT7k0=AGXY)jhGww$eGDlj<&> z#5Vy0(EooR`uabmMv6h|X)#Gn6*JV^VhQ@;FGZjGW$0hOP|XwXq2K&2^pW4G*2zrj zl+3J7%Pi`Qj91^ukh(6z>V`~EX);MGSwL%9P#alBKP1cQjIx~0B+Ki}vaK#6+v%dR zy)Gsn)y3u0y19Htw~(*s9&)_yDJSS&a-!}n-_lRXX?l>Ht_RB*dWf8x9Fwv6TM7+s^61a^>VpQzc0VhN8~PjRPNQ^%6(z1NyW)sL#m5 zI$0jkDe|oTPX1(!{MmT&wy|openp?v*HuZLUz<8WhY%qcGjh37hOhn)n#QjT~2n_bxif78{SGB3E( z=2dsbEOBRb9hYoAb1CMCyI`I)1JHM>m?$nvimJi$?qV=1cr_Sr-ZIn7bTh-u6fcQ! zVw2b`z6?eOF9t7}rDmCV&z%d#1Y?77=5zCf`O=*?tIZnomHAqJC$GyJh*0%>@N)1< zFu^376gf~1GPzAglRubc6^ zChx3wCdg=S+5Pr_cgQ>J9r2EN$GsCldhe2VImqI@>CFx@2iby$f(${Nx6zLG-u32t zGrXCIsXo=_rdSw3_8@DJBgh#v4{`;$gFLpTEpF@EhPIJyVw>7#_7U60wzVy6OIrr9 zjE2A~8;%%A&mcb1D8x>B0THBLM#QKwh{n_oae{i=0d6oN2z5uKp`M6C)CYdtV?i^o zgjdR|Ad+$Y!(|><+uf?zNxVt@bnfx&6X^X}_{x+nx3syUXsjn*-lH z7Q_S&F*|Y~T1QUA=g94M_j~!>Y(C>`*t_Mw>(BES`1Ac%{v&>CzpdZN@9cN=yZB}O zihg;&a)dH|1-~3>9EjR}U7H`isE^nmSrG#ygP&b(j7Fq8pxf$Bh=IlNzIs@Ssg~2W zvaJy(=~2W@>V#-YeG%(vAR<7GwJ*W1eHF2yUPm0MHxOfL8sbmQK$NIOh#<8T(WTx; zfem`$Upi&`9J%&{M*5gv2FZ2v9%FX zu6t~c7}N{Lq9zyKQ{>C*!FgB4*Wi#1awva7FA$U`L#7?nGwvw&oBeIQbE8EHT@=@7Ac9flD zXW2z|mEB}_*+ce}y<~6MNA{KdWPkaXd|W;upOgb|#Xcnm$-#06-uj2h;i9k{0l)HT zQA9o?pOw#vyK4V7sOHyq53f%YldHr&xmvD~ABz3t zBO>12NAhF2PW&p@%MFNsw@Gf6TjVG5Q$)ktCb!GaBd+-{kM|p1dy~z`)KciDyA6rQxl5 zsQ2+HB%CM>pAwZZDxHc$G`|e$A(c^OQkfCkFDv}n>?()K3Ewt1#IOD}( zaR8oW2;OdjN>q7OK9yf3AtGQwRY(<9MO0B$OchrpR7q7zl}4<S0w|)lqd-Jyl;dPz_Zh)fn%-WmOZ^6uxwG)dHSyIe5b5RcpL; zwoz?SZw+5Wbx<8uCqyOeA}Xk^cvJXITvXjv4|oN=RBzQs^@RuCUsQ~G<4>vqYM^>b z4N`;E5H%EU6qWEsaZDT$N8wA4P$SjT>KXMcT1Q5y=TR>{T8&XJM&1|4sh3epKK?IP z`s-@4nj$Kr73B?41@-4|scC9Dq9x80Rn;stTg_2#t9L{-HCKG6-W6NaJT+e}i11X$xk|@4&$u}aU|6-bzIDc z*L)HY9#0`c)fxCu=MZNxMV*JQb`jp%C3RU{LEOb_iegvYP-$o#`d-~aRLCFIPwHpH zW4xmf&q~x&zpCHV@9G{RMm`W}qCWX};wf=oE4)kLn>_7dT&)ID!sm;D#~C2dW;z|G z)1!@v;$US$OH&q|RWuYsbv7|r3=%_hcGQaGL|js?AqnZQPSA-uug<6Q>m*%37euR6 zVO>NQMTTqo^%_@*6oC*4_h z5yQnWF#<6&yXo$_2U^8?>E02)wV&<}-}!O@8-o~@WA#gDPk9+{bmI}7b0S*VCZToZb+oTcLHw^b zP%rcrB6m*LGotlGb3{x1HllgX)$i(gdcJ5So<^(OLbS##)=TtKy-dHSmm}im3cXT) zpjYYDs7D!vZv__UHR4(EjCf9esMqR`^v8M~VuNnb8}%l=S#Lq4&`|uRGA4QbWTkt|`jkE`9zomS zS$z)iM^i)_eO_M>ZS_T+iio6_^%Z?pU(??qKIsjerf;I<=oX@t{s^DqXVDH`{vG%i zzlh&ODxRpn>EHD|#4~+hgpp#RQKG%k#u(37ZXRN`4 zdfWZeM;~H_M(S5apoZmXw75Pix|rw8DD%8|f&6!I8D9NZ^OCq>#+jGRD`vczU?!SZ z%_Q@hdEHDlQ_NKJhI#WJZ_4EHn>nZncn7Vw@0xjLzFA-vnnh-@S;BA4X1RIatS~Dh zbpiCo{NbN90_)8N)B|iXo6Q#UiTTuQHQUT~^BKQi)0?%~VRlC940fA6X0O>Nu9^Mj zfH`OmnZxFYIckoXgJ-rRsUe-pm( zE%O7s`k&0t@a*rHyXF`8$G@51%{_A;EzrV~@a{GId(X2GAKyj1{B&L%`Sa}Sdzrl~ zURE!gm)*P8m*6GhEjXW--%G+>gjvYBlbo7HBs*&|*?E}L7d7aP!L#}glmb>b`WC0ZTg(SjDX2{zH@ zwfSs*#Az)cGKh>~g)J!Jsis7P#22=REh@H(ZDP52UxZP&H%&|z31W&!FK&wO#YXX# zSSEVdV(Q>4i&TKhb-c~@ZVg3)Ov&3vMM>5+)% z*)Hlsb&PmXT~M#n&32D^m%VKt+t>CZACu~O;9(7j_*jGNV0fxSBfjbgc&krG{MF~8 z9_#3cw>lPn*tm%AIv)PmM0jMAB0kw5~XycThkKSqq=^@xSM(QdMv(Z=&h#An@RxBty= z-SMaAx(8nEKJw@7LHN9f?Gby_9<#^o342of6p2WATKp_-qyG0S>VlIawZRwcMYJPb zvX>Fb{3_awzP8uwcc|mNVbknQw11rwb?o z?9cYLy<_j%U+k~;Hw;JmyS-=c+XuezrLTPL8{hliiGB%RJExxuo^~ET{@+GzE#w#W zi}*$TVt#SIgkRDx<(Kx$z~e3lZ@U8g?n>~xtN2ynWmorW_%;1n{=@LQ>-cs3dVYPs zf#1+?gcjc>epA00+I?I2E#bYlhX39M{(C#Wz5l4+!S5LH=E`Mv!< zeqX|FRR0bCP5&)_nm^s2;m`DE`Lq2w{@eaL z{@h6Y!hF;)ER6W=OaAcO{rCM9{!0G?f0e)5UjrX~t^bk#vA@n=?{Dxo`kVaC{ucie z|5Jafzs=w7f98Mgf8l@Wf8~Ge@9=l}-}t-y-ToebufNaV?;r3F`iK0({t^GEf6PDb zpYTum-}w`j`5Eo2U)Ag*NjaBlQ8d{X70$ z|Cj&8+MB>xQ62l^x9{4wdlrV(B`$z4E|J`AHZ{V`+}nVnAd9G|ILn0@7-q&<*n&b_ z(8#c8+!OaiNnB!-m}nB@C7Kwsyu`%i8Izd2L{ZF(Pm?@^nLGdQsjAc6GYsnQ^Z9>1 zFx{v6be%eNs_N9KI;U^!K;nk%gp8^{4(6%p1%tW}ms$ z+-7b!cbGStH<@2HZ#Hi+Z#8c-Z#VBS?=-(+e$~9oyqkXg`9AtZ=dYPxHy<<~GIyHa zFu!R&Y(8Q>YCdLu%Y5AYw)uqlq`Axdj`@`NUGsb9)8;eg_st)eKg2H@KWpwW|HJ%| z`D61r^Lg_p=1X7)#wW!m$4|hoxx5pf8ZSp?^`!X8 z@l)cTjh~8K;pgHB#9>p2!Ddu$FJ2X|j@KZwcY6Gc_~%uA51F@G(ZXU9)^WzJU&s!8fKfXAALHt7GQkNhXcZtd`#xILEP&TePo=3cQ zdAudws&aeCwXTY{$2;Pk@yp{~l6PGlUxO^&I^^oU5Z{0t!CakdV5u-~cd7_wr>hs9ry|33ak{15Rz#{U$5GydoJ zU*i8A|0^c7_{$Bj=@&AecBmRE;f8!s-KaBq~{;&AISgqCyYo*m@t+Lv!4y)6;TxH38IkR=vdg}{_@n3;R{6@s^zlbRQ z)z&o>t-sE?9ufK*kT>nKwp!beL*0Q4>P^;{t(z%_dYg5-b%%AQ^%d)@)?L=!);-p} z)_vCf)&thptgl-SS`S$}t#4T0v>vt|u^zP^v%Y0LZhhN&!g|u$Wqrqb%K9#HR8L#a zxEU)iU-g{zy!8|7r`8MB&#V`%pHtTASJr=8zqWp3y=48?`knQ%^@{bX^_ulx*6Y^q ztv9ScSbyYPGUbE+YW)+P@S^KSz ztOM3TtKS;13f7<jy=uB z_IdVvdx5>sUSywdFSakRFSIYRm)IBEm)MtbT-$E6o9t#gZ!fc#+bwphigMd+_A0yG z?yx)U%k3_^+wQS@?bY@gd#zR5)!M$ip{chgZ_{nZ@{acW%2@5HhNiBL_GoQmSAKOq zRx9A%_SQtQJ{zNJHpkaog088AzNWaGs#Moxj^W8%Ej`DTp5vO~=e1I=wpz-ya;=vu zwUf-%$@_Kkex1BuR~e~o>uhPT=xMIlIRLlF-@_x0vS1s*VOZ(N*ezmk;t=q4U z&S_ZH*xaDVNhEWrSZ$7miprU+^gki|Na!;0Gm*-q`Mv5S*H5Oxb@?7aR$XP+H??*( z^{!ghmS1btH+S?jG&SYhdsJPF8Ig{zau<`;j5F;}R}#tU+Gw_^0TLqBxo5eA44H5; z$DB##lBr0xv7w6zNjfPryPn=&8q0~G&nKVArl0_a?T|1Rh z&k3^zIT^l#%qFG@!KrF{mPZ{?oJtDcQ^N0*rb)&zRU4kAyO*i2V$QJrNMtm}*p4KU zPGu<9(_rGiZ#YCkoRkblN`@mP!;umlOv!MhLmQJOmQ`*q66-YX1 z=~P-em6lGWrBhjXKP&HNg(+EKo~UgmQ5Bj!Eoslzt-HdK(*jge<-+PAU9*>WHLT97 z8?t|p&e@(QDKK#guL!RcNCXxgrYQ}gJKI$V-AuhBcVM7HB9o|OEVu(}?6bV?vnFL&0Mb<~ z5E&7OWLnHdj)x|f77LdajZ2fpB{DoYCo(L@5X0)o*$thY4VZzHb3lzG({<4~t_4Zv z7&opOw+J7L2__E`6F3a|Ig3j!BQ#`)1}Jhicup_1kah?KrqufAxxR)o7149uhN0<# zFeBz7Bj$pckj!OjBj>hsbgT^J+LsgF3|nO|k4eg~bxve3rz(q=$q5^HicX}fDud@T zNy+-U$lR6=*!;PohjU4?is!cU!gK2CUDejm+r#|H$^@6=UO<8fp5kc{+Q~YeVk4AD zvGKuG;CgT}tY^85FswotRw08{FVE}cIXfQ7TvncI&w%F)z=dC;kGZV8&mLSd#~vK6 zDSK{rTSIpXvwrTXR^Zd)LALyRSJ2PGVWE>AUIK9q@)JXl_xU1%t&OM`cQpKEAcW}m$onT@C$#XkJY*8h14FS z*Y1@gT@n>YL8ljq^C(x+t5PWpq0y-j{i+bXst|pukO^1#0`6BAP zb&)&wF?ls2t2IJ>4UG^?Q6_v5G%IS0}<-C+=Mxi3nV3 z5fN>hYN?F(nDm?$`dTJASIfRZO8nepmM5@OO7Pcen=0+pvM8n!=~x3xajd};)nu-w z)@tz9v~=r#oJd9*$b%=d&jvS<5yPF)a-IY7yk|x-*@{pDe2NAFsT=KF2uoy!2UFa; zqoTCEU?IZE9LXVo1uj}a^G!X^<@;ZJ#d){hl6W`j6m;YMxJlhrb46(S@U zJmMT%LM%7go}|QcNLE+dO&)iwreVU%3wlf^B)o}CCK+q?o4g7!Pom1HR5IG^&H`wS zwJ9g-{+z7$b4d}gBvT7FjWt|EE?Hk0XN77Z6XE3$3wyb5c$W`(A(2d_qsu*(B^~KPDj8lr zt-HxmnwzRj;Fg7mS%|K%CCRXtgeqitaTRH(sfe_wF52|BjhqFG?v>k+3b!b_T0FX# z;G{4%B@9hzYGj~fA(^Y;)-rWMkwRckYIvdypb>>Od$kqNh7bQUhsbpY$#rqZ5J!`Sl~*p9n&kZ)nKhV8ebrt8KTmPE2JVW$wd}}(BbO1aW%HKI z>4}G9*H937&J#aYNFsCW;G!)3l1N659bCZ0+_IMfhg4XbOWPg3l~aeToaE|XIkDg6 z$e1|121YQB32?-tb%eoAm8==6(%}xRwHc1PV1-l_>BwtI$W*ajWjK6{)qspqg}$=B zWqA3WNMXr;xhPPVUy(A|NEg{j`pZv>x|X$ewbv;eG&^$clOrjXj|NRMJkxJQpcvN9d*@pqF1 z6V2$W2uh_0T9uGjQx)prsZ_P5?<(PRnT$?AsiY1i0i~XMD_y6sQU{Q7RpQpcH-oDe z?Ip_?=_S2Ws+ACjJf*S_qKCbthml@d_Vo_c#v#)TFA9=6_RMhvOh2@2!Qcj-dFIB3Yt;X@QO|Dk!EKW z;pgIBPf|4hs>B&q6vJ)q)r&3nJsph8RcU`1eRZEoNDuv%U>qm}9@i-S4^n5RAuUt({be(@8Ekfc*vpRMp>8%XMYG4dWNU&aKgOhR6k#)MecuJ82 zJZ0x3gQWvpd6nx7uYuh)v5Vh0cDvqWq?+k)tN>bTqi5(#oDc@!Oq2svVI9FC(V9m%OWx~WP-nu@KP%+6Lnfv%IRz-N_`fli?z>icmeEYOOcVlP=*6T2s4YfoUB5axoKS&u`I($ z69j~bp@;(WY@gvU5&~y3#O#YFGMp~LRr}s7Wf_hgAcUjgGKoq8H^c5S-V?Z9K}cLS z!#+Kpvwh2mo1Nk83gEFb@(wdT>C{z)8=(S`0sU*jrCTi?LxDD&NOwNf8Mc_w7n97#DlvY2-y)a!`<;Yh03kp;I?l?ty1i6Q(q z@!!YvlrSsBQ@0~YN=H_LjwCA`SrIz2-@%c{l_NnC{(6dX%9Fx57M?sXyADQ)l7|LU%>}XUr|hI5^@zJCYW1 z#3nnO9zrBUXp@wWBP%FJ0=kZ5ARO_h9mzyEl7VotRalRytqj4nt)VVAYA@o(egNV* ztRLz6RHU2CWluwIte1shtXzlmt#lZ$NXouZl-!9UD*;DxD2}WM9LX*?;w3x0_(jk` ztBzPq$&onX@jIM3Lgs;N46aq7&X!ga|Cg&gYY0?FI+Rtn=&z3=NfS8Yu{x42aKv+U zByHe`2kS`sz!6W@kzltY`)M4GkiwRWGIO{JW#O)@1030B<46S7;pkb?;dMH$QePt0 zj#CvSy%p2WB9h3kh~O$ln#CHXUW~QQC`#rr!(xp3#86zt=(4+jvP=;y)|eUu4`vi031eL78a6$&2UZ(^~Bh7ZVY9iU%N*<+z4CA`<#P9J!zkle<%x`Y#zW5X`clb zWqDux{0v6~;qeIH*a=67i1>_aHk|A2qP!Uj^fIp1(awA~jEFJ|DAK#wO+d8ZS`%(- z?a~iv;8BU>1rLcqD5r{>IuHwN>h0>vw>Pa5RAf0&i=-1*nlo{wZsDq*Yx0O~fYWdR zE6=ex9xjxb9ZA`Im8`3`*5|uArqSx#ghQtSCCOM%3$zff^0JOzEGq#qyIa@NO}xET zSE3htmPOTDxb&>)z{TWOdzQuMU3G26-UB>W81TFsJL1~aO}=$`OAl^hgt2E?vEXA` zY{p7$T0#yg7YSB^+l-R#iv$Z5%%%>^n5Z7_%y+eRG^^%_xwv62PisPeR^jW;BiNw$ zt_ld==G)uy%M@9~ROk{SiyE;=&+A82(2q(~X=NTkOM0(Viu!Sxhg8}e?ZYgfG7r73 z5w*V#f$E|EKV5L1Cw^v!~iZRp8&sXi-E{_@_2HWW$)uFIyr zvD-U(^nI+etydv2SGBg&jlHV3t*5oKZ5@E|=GN7%&BR$%O;;(wW{9qiwHifn0|F5o z&&4}-LpNw??Ous{B1aF7CP$acJX&@4hY&2lVs1uCqg>JjP(<|Wx=f+2>-IzdCDI)! zmU(>=$QXae1UA~&I{}SyyC?v&yA$$+4pTfnQJaukYy+*vEh~SfHlAPGgqa^f7J-Pt zVJnIaeWK=UwNtrdHA#|4A}Tie=3`$dF7B|~MhbeKSn8ZKxWDa*9#0tr3qgDR5H zV+@r>t8%P?uhz2G<-J|`W^9V!4kRnAj?R30V{cnqzDN3$F6rrNZD?QKmLCEZOU4F0 zRZ&}O`-Is>m=;K>u~M%QD%<&0?lYTFay_qQDyz^mSeT~5n4vWIE7bc7NI0N>^eRZA zLhca||I`6(0aT%6s9u#F zqEwcrQN=^lDemf$gp3-5O13Mv>L6I5lO=3O56JL^pW!Q?6 z2%(R}QEad>(REa+bZ=^-t<&0jS9L|(=^_$WXXaOR_N>E<9c`G_4X-!a$d@L*H1h?M zp?bEAFUX#$QVU;@*HNVvd|AmCe9)V%lU#?RRO#Cg1m<+_3{M) zTvb}b7fidVgfNS`tfxzy=WGB<-8=sjjt9jwm*a*fL^U}p1TvMX+ftrY)h*SL$sWZL zp7y;5Y6e&J*bBf-F}VY}g2i(>;pbUh-HNAxyAt*pbakKgQT6=RN0osN$l!f{PkeT-2!IqDB=LHLAF%QN=}#DlTeNaZ#g+ ziyBp26gzQIql$|fRb14l;-W?s7d5K5s8Pj5jVdl`RB=(Gii;XmT-2yGcI#1XbVs!j znACm2^cK6fRL#3pqmV8gTCQT5QWlVTrtW+^Q$^w#!!^%TNqDAig=cOZ=9wxo z&wLLQ&s33krf!I5ZUy0)DhtolE%Qv>G0)uU%rmOZJfmCY8CCaqriy-^sj}u7y~T{7 zYTm7%XX=qE(*hAG4w+~;PLDFL+mgI+M44A!Nsf4+tlwjoEXfN~zG?bNR%`*H)C6RH*{u2 zzNsfd!4r70yjOi1wOx#;F?_?;CeA7Ahi;rDkxs#XPOzO7>o7Fbw((QT8w0RNU}UYWfU!rTN$X5sY8ikTy`D5t@qRwPJQk)K?G z^+}g-YS%xI0M>ZX#>LwKhVHFizVun9v-<|4$ zo!*PsyYDY>3sQ|~w3^$OH!f{z$v3SmqT9xXwziI*wMc%qHY^AK=-S&!16+YaQB>|? zIjP_PQ>uakWT^_?a)vSwX=QD~Rw+I#hOiB0rQEM+Y3;$8WN2)@xv^s{W9HBvRz_?Y z>Nr-Ps!OJenp+!Ib*S7tP(bs&!qT<+8KKnAsIYWxsJoX(zfu&lvN9QI$wT*8SUI+e z-D7NPD#iBHX&c0uXq-!`9n*=X^Bv1t+m+gO_aGq6nA4neNu>n`YzW~_b8;n>=6paZ z%}EnnMRO$pCy^`EV>s*-zt4FC?15yvovzQ7V1t1g5fGp=nb;Q~W}I^|D6{2BN;nRu z9|&ENgX*ayZ}P{L(bjR=B2~fpQZQSVsl2rXJ8n2YC&{x^Mus)RXXj96P0mPnu;Go` zljIhjvsuq@;0|SOH^XP^QW+7K45u|v=Kf?j{);jXO@`yVD6?2e{1#T1Sh{xfU%iKVut}%*yRo^RR?2h^%B^*3)-?+S28E-`Tq}9=w@f)#bhiGb$8!sab zY?f#g7rbZ`mvxwlOBN$lE-VdbN=$XCrn-b5=~Xd3u1#079qjGYZHuLXN!E~3WocoW zb(w3hRpnOMGKyIgkY}Crdt?1pN(1BAlBcFs%w*UqCwJIqCwJH}CwDpb&T;Q%>N^{U zAN75FmU}nby*tajJKMdR4aDf%o7uxXXW@M2kAafyS`Qmu&M-1;4oPI*UG$5JBa&TW*1mLK^Q zL-2TtApln~r0i1+{fes?x{9Y50?~@0-*6Q}0Di^LuSo1oimVt?a1}$!q8Pd=uj-`t zvasXQsAY(plSWG2QCdede{}p+w?q>?R7T1^mC-e{$_%MNDkBiT%4m>JWdt--Wdx|a z0PL?$(W;?D7a;#wc=5AzrN2O1P~}Oc!uO=YS1pyS_C4VZtr{C`drg;7OR@*nS2yFs zg5`(z1pc7Ez+@1;nD3OnYI-qwQUSq3l`m275-%a5t06`T&BwDc9ol4m0y0`d*aQ&* z0=YFxspviCH*{>2(2jGFNC=gnV+h)fqg8EcGyPBkT!>;jw!Km}sq*gC# zsL11!g)zf3Dx|pc`mKcf1V?inI+z_VBcGU(szW{dsy%9LfYSRmS(|^qr2Wa z89%o&{&zA=2J?U=R1n~-u0l{UcL+mdC_Ea}c}`zT^?sh_hP33b#u_CpRrpD*@KD34 z^JA;@^HQCN5XrzcRDU~_er%O~Y`k;5s$|ijJxXRtABCVQ4&-7sAha67ar0L>{j$M8Q&#PJSz20!}%Jk z;Xao6#Fi{g)hJIQ8K5kYIrdFENI;2@92?t)hZs7n673A~_6;yCr1Rc9H{Y0g?Dfb|T2qI5O zy5N5ykO1z!Yx+0|1ua4L+?R(;>H>!Dx zih^ICv3FX(lgwdfv{%5^W?zB%Q{wJY_C2fiQ;v<`-dojv4plS0l68j+8SI0f{fG%E z6s6uyUmr?*6-7N{v0j7$-=+Ym3}JfJ;Q${(b*8{ybjkNd+Ivo8?G=->JKlX8Pqm#R4^54@ zSDD7#EAozc_gxxy|5LT8ou1OHijUv;3*KPk#I8@qGJYA$_&p?i*TIJ%=YJ>XlS}Lh z_f?Zgf+=uV%Bg4c=?z~+|J0YsdIAUPJfXryEKkl+@I)I0|2rak(yS-T3GA_^Mu%+e zB*^eRkpw&v=|lBxA}#9_5qGJ^7e~CQ90iZ3*uv<4;S2NXCe9?Y<)fnD5ptd@x ztpij6i6hml3a4`0?#?MHa!?;TQ*=pHmQtBVra(sfkO_zeCMEJP`f%L!4JjDv^F5q; zpDd@bM@Ds&aHjD^7q8%R{Sb@ugcJo&bnqn+Mo_HjAPm@jAqNzwIL3V_i+;4|Kgmg~ z9(rPz-YeFL5Gs`+wB>HnAjky7V^vJFwVPy1-~?%FqI#p~ z+!TP*4gvIXjUzSXZURSF9pTCFcL=+`J@*~&$?$jF^w1&gcu(Y=qcm-N9LL|#NRRYe zTwA|OZ#A)!95uotVa$<~oEq}d;h#uLBvZ1+r^$*I{668sH-TsvxMJFqszlP~^(2rN z^s31kp9MDQbEpz>w6#i2MY`bG5uo&8ku^SJku|<9Fd1ZL-8H`Ghb$RKSWj2G3wpht z_8@uzUw!BjzZ&T{+E|4Bj!=T?VM}~@!@tC@9&w3JsXBZNkJ1-79Gk>P@wz%1DAxdR zota}(Op#`s_oK))9#BL&k(3?8m=je*6=lYkO4bKwpT(g1s_+~riWF*tb7zMa>XS;z z>Wq0lHrsY1`x?A(zPYNYp;M;`Kn#V=;e}G-5?3ZN#RpH}I8&G%6EJ0Waiuf>t~owA zR)f_|47-`IfvX4m8|kXB#~T|O*U`Rkv#GlWiHW9`I34Uz55q7EtBKf@{^Ub-rIqh) zY3u0dbRWr%XXJgjDj$fz;a}!UiVsGBAHpv=wo@U;4lA-*bGh_e5M)Z0BY)M?&<2at z3I3D~E#gS&FbHVzzcQ=>o&v(SlUaL(I$Z+82E;`xRqD094aY`Xo8o+=0pLiR`Yr^$ zkgUZ~V72d`52-Mcii|oK(1tH#wA*y577UeR+Bk>HBbDUM*621RFd!y0LU>LI2#g>l zC~&}C#Dq_;V&uACr&IO*dPL}+X^4?6d% z&IvcDb8^^b)D*_1m8LLMr%63aC2)X+r}AXDv$ZoHQ-8oNORgfXwN;K|%~w)_4Nmzk zLR*omu;^4Vfh=N;_cS9P;>d8u4M9BSL`DQABj44^@HZN89*bo?!zXr7X3k{fyIUDK z+9xTXgrt6OrIZ=2f?t2AiBH`DPP&0BD{4l*yOiP6FNuskq|M1}Y{_TA&hQAJEbnp3 z4o3#0%qcM(O<)5dsV|f%rKOx);u8E@xypO;U9k+G*Z^NxA~KFV=S(E_*2;Ul^8@u* z(lVl88Ihn2pTWja1l>=ehf`qax4fs%Q1dB7z*$l=eCiNDt8DifUML$&bO1p&UQqRzkR zX7sf(O6ZbdoP-jXeB{KDbK;I1p?2g5wIfHU9r;j!!-xH0<*BQ1HHj4{Q_DDc$D)vy zFKlJ&9J33bVQ7al$2ahxErO757^hp6qm(<$DO?Z97_LW2Aikvx5LZ;G89|tbZBtEP z!!EV}@qN!4Y(mNNTgzIP^{gwRDmZeE4L;4IxK*?zfMNpRYZSfRu&|29IQq;xZ<5Rz zlu4;^r5@ud`EEJql4PX>oP-uvQVv{c@Np%j!j<#JNgg|tiF3FT=Wx|#QQBd13`dBJ z1UBSQ{n{En1d7igaD7-EqC3H68cP-$VO$e(ssZzPk>975>fCICzeS=}Is(^dMkmnk{Ym6FfKrSt?ZpN~t) zcLlJhB77-1RGgAS#wq>%MEx}o{nN3Kc)1Cn#3NaMBh`IKJ5#M{-@+ra!QV@ zrR3aUO7uP@dY=+Ik;-QHqeEE9H?_+xw$*m$7{%% z3U4cS5_?@}hZt?`myI#*ZZr9mQCiN5rDdW@i(yX7gq0QpotB9!ErvQR6Ifadc3LK~ zv>5KROlWC2HlCHQ-DbtgWaUGWS@|qVR*rLK<%5!0`8-NiJ}jA)&!lAK1Cv?#TuN3x zG?|sprex(XWmXPTX5}ztR;*Z7z6Y3<4@6|;kY!dr6p_`2p1<7(LoSA0oX!lNV+NdE zsEmBRI>TFXYAe|lN~FbgO6#w$$XB!+cENBYl9UcdY{jZde;oO^og@8n{w0P8u=}8 zlhX1TjbC!%W?L!oWx9vBbk;Hj%kt8w47v2%SpzxoM24L3C6UXY)s2Z z#k8DMOv_2dw0v7E%_b_5)>15PNLo%VrsY#CY5Cq%T0YyBmJddyWu1_gbwXM`@Rb%f zB`t1CTD}00<~^B-H1F`iRpwUN;e%)jjYd{>8X=&f$|{y3aGGflzvR)%%E9Zb9K6oT zx74z7@H)#olTcs3Pg007&&gT&Olwx8Co58u6=})Jq3Ns~n$F6h>8yNwD68p`uR3Ps zvnW~l8fI3cEGtr$73s>dnM4{%`XL8>vvSZkD+hhE;&WtW>6aC$%!<@!WnqxeW>-Go zm68**DREO%OkX0Es~f9B75=zQ5rthODpZ8oPrhT))-+C69}-G&k$9mOR2?x$QcUp? zY7BuJ_eC$xGa|5e#D;sl_dV)+n?|_=Q_RL!%OkJ1smvops?jt7g`L*h6z32umZn97 z8S|~GR6+&XSfGmhOxT1Cp-nBQr@2O(oUtZV6>(Gz4i*YS9!deH6k3~?`<@mvKpvMi z!LRCSXl!k7?p3(T)Xpyc4FI=>+DQRV0F9^`qwojYg?ZVg00?~!kN&BEp+{^nF-C=7 zRP`b%3Gy=6dM(8>93bo&r673P*?>i-s;>cvEOd1!fUUk-AnzBo!QE8uje1b33&Kyg zER~B{WONB8lieL^#VyIpW7M~h6wNISZOh~pY~^iVPUGVP+6pKYu+^ktISv9!v515K zU#C+oa6wn}oGtYXu`-|Ur6A^7a9rJ#vV^z+Q=4QN{Dpjb_q|O;iVUiY2fhO2bEKHz z2*BQ^(M&J>(uZmrn1(jr)Q*!&;5W5HH8jMNeGMU8QfIs;RsL6fVE_B$yxK|_4#nX| zJqC|g;XAA`Npou=CEs#Q>2D-TI3*4G>@HNja!NH%YE9uPZ^Ak$7*)PRD z5%n!fRlxU2v2|4nII@|df;!!(;(<$*_YZfrq^+R~o0HX{S=D5zob{y*ph&|IRs?6J zsU~do4jdW$uW=1E8OEjUv?&e?5_n2ZZW;l(gsi!*gcO^vgm38jO33>8O8Bn7uY_!z zuY|8`QArtS%+-~UdbU`4sESEub;*=3Y}M+Ux%y_2zG=`md41ENZ`N7zI=)+0L??tr z&r2{OuoOr_Ems0>YI!NVE|UikKb9i(mg6_hD*~qnW(ESr#IlLTxQP{$t12drONYuPmc=Glqzzmrk4smi z$4#y%S5L=HQ6*JnTp6AdjGqWJw0~e;@bQ9a%y{+H!k*V&GiJ=#vu7}1%-DmQ87K_~ z_G~Kb*)(IuCS%5?J;sb3I|_R`3zwO1?hnkEv9WfBu?e+N$53wtHlfa@8E@XZX~w-X zW?W_*f7y&TXP_|SGW-h!1DQY|a$e*m>QEqUq>ahO)T!m^Xf$Sw0g;*DF>eGh+aAN+vEV|`%U7zpTg%|B?k{TU4}%!%9* z)Ah@_ejsKq4&CCt7n`Zyi|z~jm#-bK{q4r*8Lsd}Uefi4;qmHk^V4g({IK2*tnl#r z;BOhf^58KSPxReDqYwUoZr9px9P5Lxap8TMe*^EUfp+1Q#(mu1!jd8IKI2y(Ihuc7 zyVi}ycRl>(s~V4ef-z}C_>{xIrw)NX>EbEZa7B;XZaI%LW?nP2)4VNA3v3ELYN%sTe^s;tC!eFPw~8*Z(;X2n8~Ob%1WeowAex z!iJ@NE~AR){!yK3VN?S# z)T$>r18q_N=&JY6ruSu(@!UVkQ!R|LdOvu#`@Vm)@xFQjX_HY$?|k+l{#WC5JT){= zF+mKQ9uA!vhK>|ZF0oaZTuSj<-*|iCI82f9f3i$3zX&%DWP*1OGzMSo@4jTOna z2f{Cak}-i}An$>6*~HZN@#Dr~nj3d)P+ccanV71q$V{C&d13|b)a~SnyMm*>`tzo2 zN3FBvlI3qaUbxG+^sz6`*>YLoc4OAOtL9vHW8sO&ODmsTb=ehXjV=jaaDQ*-uGXEK z8_vFR&b`yS&uQ2U!l~hyGRL)waG0?^B5+q>i6Y1czm3f9uEG{S9MjJO1Ru*C(Dm(_ z!S@P341@z`K>6N8$87xI_ep_tATXu^2fg5JOp1;hH|EoI^Nc(1ym(3dj6Z&IZ^IXz zb^EOk)y`X2PT~mtjLT$A)PulIAQX%5NOAk zj2@p(of)i3mxMy4&lX;O_+jIeXD|QRy(`D=9(7K~=GmKHyK>`e*UsMDan7jSV^-e# zv$0Pbi{JaZvG|#~jo;}$r)S=Y*B3r|c6Z^U>rR^6JEv>crh3RD+QcuE-vW88fM}P2 zQK|70#$n9CtTN*hs0crE^R|gKrK655UU=@nL!W}8(13BrhN~mtTY|x4rm+0rKR$^< zaA|#Ef#M2=c{Yj5Ey3{MzZ6_?i{zZ((Y?VVNQ{QTFrQ9H!Bx=0II^OG=Ma1*C|0=JC$!CN03SYb@~6pY#(o3nq{H$RCxkeNfW~>E!<#S{kl={^#=<-Dp`N6VOtXQ4Swrz4f?e9aBZ`KM&^(REF{s}7I=;=mFh29Fowf^ zX$z+K01?yhP1=I_;jmzR0v|I3{W}J zEEMfzUW8{8FP5rSqVEL$ta&j6zSp>ZBslu;nh&lV4Z?#(5j)GfV9jUKbjJAi1#b1# zSJN5S-yS&NgDZ7rc%N>!FeltcW5DoRgWEJW{B&B|hS1~U$713*AAYr%U_AQ+zh`*i z(lEH@!f!EpJ^bW^Q+<13;5HAB2e-bb;e}a4@N5nINn~QVACAT0MuDSUm;W$Fs$Gnx z%MZ++e!BXME&Lww)$Jb^Eh4J~SBoy97wxQf>237rwND8A$%W7M!KWMsK6MCOEoPV> zdKRD1cLIwQUeAJ{@{mu zWc3kw04o>;zlh*hjDADvKTNBPr3T(b{VUy7#@h@RztUU93<~@ys_(60_6hucf_tl& z4+Z{Bf|Gwccnvx_00jpjy9X*dCCQ`FhGq=L*L#r9PaJy7gK3#FW&4V{!5$ z#gyy{T{}<{yn5gYH8{bY0~-Rt!M6*G!m47ankhmRV<5WEIsa^d9-*F+3>TZ~m~Qw`bUFA*F1j7>%l$l3%3MVKs4 z62g)GF9d%&a7FNHmV(00;5DeL+=jK7NZ|?}-uZ$PJOR?AX4%lVpB8?%bEk3Qv#Wo3 z|H`q?O_ zHG7sH^|_VbF)lc+uqOEV!w&c5zuCTJ$rCp%9&7Zz^x(jlA!B}Zm(0vLX%1Tg$i2_X0g62SEjl>n}Ps01*aC4ju1 z$eFMSwlY=ld+>iU{_ffxD%rie|AWX&0~>x&*q;yI!|qaHPc+~%HfEm@M&pVY@N<)|E9HIy zVw(mwZ|a}s4%NN)szrj=o|=gU-!m8D{bQ7RPn}u;?VT)L7)!H_OgGVJ$Cv+f1$qHw zduCm~^s=k7yT9=Jv7K++|AT=Qz}XhOx!-Kxp3QD=4_-BJ<=5V5*W*spC&mS3@IAU1 zjg<%P8}2AXZzo6LzTu8S^j@0C?;GwY#NJ@|a7Q8bPk|3}6k=}*e3+vU+beLNqoC-F zolfD-`-}mSV>jFhxC(b_I0e3xV-fj)9E?DP#CMgS~8OfaORZd~e`0)apxsFZAHn*13LY3&m}7j$r88>X`duYXw}Q$W z$-yHmR^MT$*(TH=!V@$+?9q1%Pg1LY0X!R1i#o=$f3SmEZ3F(DUVZDvfJyK>3QLTe zU20&U1IH29?jGD9#1bPI4gQ0znc|`i>ca#`D7sM(}r} z(9aJy7~xS2A8s(hCklL+!3d8P_%MSJ{*1tf8H`W|8I0$L8w|je!O(Cv7>60n!NZL{ z3`}q;8JOn}H83M_PSx^rcw|mO!La@bodV9Q=^eITIqUd4?qVZe_`l$P7WNzF&hS+` zcOJY_E&Ql5sn)wn<*Y4(%W2pNPO8Q5zp_|q_y zSBoXWg2mNT4?77P!}kV$IT9QZlVf~vrCMD7y}@UFaFqaH_`X4oPz-Aqi?r!nzc44% z?$K%8YIKsYa((5U5FQBm$n^up+5Y`9Yf;%|z`Te{5 zy1V*z|KHfXugsfw{r2tGhfh5C+ADj35pRZ{Rby;eXjgf-uhN()_*%y!B@D93y8z^DNFW>g@Jfsi?QccBT`B~oP=mFfJ7YU76=4P0Xt#_kS< zg1;O1+v?zE?O9oyH{`=@!_ z0{B7f0&Fb(8(IYh6KGE38aD@7N^}*6Lilu`E1y?=+;VEZ0 z+7aIX?QHSC2bNzgLov$_hk^V#!Ba38mnjw+M?#mP_R5qJVTf6HIz#=z^1`K&{%AMUKMikIeYE<**J;K?H{-2SDUxLz8=8R3NfkCSO)krD>MaKTPYbC8%eb;#Mv2i; zJaNjT;*(1z;jgj6DE+<>3K^rueeTo~OA1eZz3^Dc)Ke2jjt+)`!N^Pf3nw=9bSI}> z-n(p4=n-rgS}<+(Rhy^hwqM(RLjR+o1(R1^edFxvtFHY*vj0&YwdiIVh!c1qg7<3~ z1{@aKAUM{I0xv94#QWf17sl<0Wj(lkMuY>MkP50FRd{rJs5fF+WO4zR1AZ!5r_^Ln zh3~)-m2wJ#BX0&TJt8}I>d|pCxce^yf1PsnWiuy7Ohn=OXHMz4X~Fq-T{A6o6M8=H z^eY~|vf~GRC-y%k^0`2{j|Sp05QX;wgSw}Y1;hb{?+q4^yYUeR%fD68}<7#t2LN5y=Ss*0HiNx*s6&HO@u~ORw9`lghoMz z)Xsy+G%Z9HJR6CJC(pd>tf>Qk{mbs489%!G+}woVrEiAjpV;?I&;`IK{1vm9f2|ew~keaiqNrr}wu?xrMosGV+S7wOo@00Cl%2qB=ULnJgp?XMUZclVP z$)c)X&Uj$4&rp_Iw@=!G=R?YnEmbaD8oPo_kCHW*o@h+bY4t_bmKu&tehm@mQtVss z#h^KVTJ9!uc_Tn*6tc*VJc11X&l)Eqsr$;a*S@-Oz2f+068ewdyVot``AhK-}Yx$yk*`pfRR z;$!+1S+OB%Mn~^{YSE&iQAG!?()saE?xq@j?#??#mvl#U--GWBvVXvAapBP9R$qN3 zEsSR$oEPQY4zIu6*fJ74`l@l_NN`vv!H-Ro7=4Z3z97in`FaJX$x^jj&TutZj@&-- zmyi3}fj7?dVlyNy)R|tjsZ!u?2fpXk_u%$%#wZU?a}D(yme9v1ufFwGkOBmP|5X?N zyM8?8JsNHwGX(C|w-@^Pq}E_U=l3)mewvR@pocB+@Nu{?81d-$@L+Rh=ZNqrhk;KW z0!KcR>9NZ-T+!pok=mSz@l;xU2wiIJptg_r_($(C{7|_hxHY67G(&QIS1nch(fdIc z*Z10mx_Wepx(;g}@)lM7fgCqtBhX5f|3dkvpwvxz0{)0vjTyCuZ2jY;C*Zrdz5x%| zTa69A7ki)K!8rt%)ffV^rofvCE|EIa|EIv`5nNVdfcFdh0)lhWbCNP!YHOv)#q|ZO zN3|J3{SZ9n$>NcJg5`-6^#_9CkFhtQ>U;0)-}Py%%;C64Qt9I|OR1V$Seu3zU8>hX zqv5^hE0eKC8PY}XsW}6MYcR2^jrW0xy1$5dOp^U#6UQODK_L(o1_@1guJ97JjZJ&b zIKJ@aj@Q?(e|?8>{46h<{(WQaAOCF3eP-vj!oas5D-3Mk>CdcVW(;F5VqgL}$Fg0} z+Z~AQf$)e6kNtVosy81i+-6+1dPi;TjcW=oMqb+Z%C**TSzY%t~@e=V{rgDCkCkP5M?P0EQ8#gd_yS+on}niTzIJv7>Sb5?n2^x zIV%mFt95D%V!FhIaaPna-fa(Q`754CKs8txthU;4>|C-48s{U`7K81Xh< zdi}J*Q?FOw`H2#)2q`Rz)`AeVpQsXq;H1W6>=8wdMQv29qHSgTEW$(_5m+aizcgH(bMTHsn{bWhQPwu$>#g48QL-}nt@0z`!_}EdU*d}mb z+P*JbfBAnJpZVs4#`s@!Ufc7x125iX6hHWYQFPDkg?~TtRN>z{2A;lV!^00}lEL7u z;fTn|UXoE-rRYtqLkDgm*6E(4t~B~qZ?CJ{0Z9$sS-AZb15vUyzrOa`Uvz&Xw{h`> zSI<81a`2vk1{t-zB($4EXm4hr#mX3sLTDNOwsAeXd=PS9eQY}Vt1IL_I412=d~h{s z6CP`S;B6mVDFws(4BhS$f4fkYZ_9E7aB><{eYL^{TsaK^y-?iZe-H2<;yJ=Jg?sVd z5eN$$qYR1xA7e_@;S^0VG4#iQJzc>Q3(uWc=yA2+P<}>l-d}V??AA-$_Jd0!uvRq< z%0X6XNwrqS&KKHU zU0#kfC00V%csqW=WZHQfMVxN@gb9c+U{Msj@xH%rj{n$fykl|xzDv*BJa5;o7p~vC zpKcdnAwT{;<8BgWuK%8{Zy#^0Amt*sd9MpUGVp=W z2|VPnD*Q_Lfk$~{x*Z-1cs}a+A!Ada;yhJn!m80&s5Pk?8Vp4&R)S5~oJCga*5Hkg ze6VrRH8-AdOVu?E^L8{``&98Wk*ogl3#VNZf}bH*u3t&qq?43 z-?IIZ>QN_^RxCZ&Sv7sls>`2jY5C!mH}6iT$Del5thsCJDpoDL;`=QUXQ_s4PEzwM zwq#EnhgB7C&Yz3}aR3{?l)RE(=JysU8Q z?#RhIcOG~pas>+d5q!Kn-VEY<+hb1~(AT_tCxP=O- ztGP|T7P@fT&AVqWu#Yb-{e1s_6GID&uGwImTX-UKYA|?bI1+uC`3t%m2R|2nA^HH! z5HeO1ltq=se;qu~HxNBQum2|2v%fEL&>IvUI7qN!CZZ~Xdm}SI77i6*@0w7h$db(t z;o!l?3m@&+z4Y<*>%VpB?j41LN8IrK9idnIt5-gB*<}x{4872Q(jD*Lpk_~C`ZpHy zma@SHL`~N5I0-NU+VJLGYS}*Utzpq~l8_lpcS=nE!q4A7dv^ zDm@DKqxU^mX!*ln;K*s?PaJ>pF@q`~JNve;%%A_2+h&Ju^4}xe|M>;L}698P5H|^q}ytaGi#F zNc~ML8zNn0O1$1M%2Vf@H0PvqKW9w+{lFW+#!zSBxbwc;*>%e~#-ICl4BU-*0TJy; zUt})?AJZ%51#rnQ&J1!w60>XY{u2heUNgRCoR$3IO=qPlFD)<;!%niPCx;LPAD z^B2Y@-7(>)83XGE?g}=E$eujG1u=K89yc-j>cu6DqxYu17GpSWPrF`6%A%QHyV%l+Es>w+kG$a1Vt&L zpmRQ0&3vM)@KASIJnkt3vsvIL6WrYf5&N#%?qIwYyn^7o-C;P3 zwi_uQiw+r@0=8&Geyz4Upg-8x1-o>8f#vqFJ;SW(ADml_C3p{8_TdI?11I2%A@Trw zIROC)tmDQRNF|kHMKy8URlz$?U6vkq@iP|>{3~2E(3x0SG4A5;UK}jh9QsrJb#-S( zLp!hUPtM$yQ+Hr4p(MF{Uy%m6+)ABcxH2gOhkW?qN!?DHz$!`5V z;|7wm^>BB6aJaj-l3R#?9SV7gyMfNbaX?at4?Ik#DL#%L z9=&!;;0Gjx!g>g$>Nkc0r>^e7P+;4wF=<*@Ely}4j*TcURC&kYBhLBa>Yk0aT-*^l zuK)FLQUB`~-?Tb^@ud}?`P@lkjy$6Q*L_v{Oax3F8+4Wn1!2XG~Y1GK5OZPN7Y`QJ)`BQd3Uy7wl+7( zDlVCHQq*b4q|Pp{S=OC6dfr`E4;*)U`GV?u&b=c0mEv)u?2uW0_KefcJrR;aG*Vpp zph!*?acUSGyRqPE==!lHNCMTjCkC*;Rly5OLT|cwkY!Z$Y4WZiNg;f`_a^E06pwuH zvxdOWblaKihr8|IRAAsbk1ij4s|OF4`S70*;B$4u>81J~L+=#*-Y|Nn4uMa1=`Hue zUAm&tcYoz_F&dQGQESU@Jm|k%=-aP&(2&Lq9hMPN91pko;66@AHI|t5mT}rh9Kyc7 z9V5ZvWFm~LM>_gB7ytgi4IVu5TTNwjUyf=Q&JR!asIe~!!cB9reAUJBuv+_#0N)Y7 z4sQ*Q_38Td(HgGsM>U>ui|Qkb?Y1-U{t#|d8buzyh~`E#`rb(@XMOkJ2jLUKYgsc5 zic!E-LYTyjlT0WHsw^;8%Hz^-7$!w=2>;BG#8-k3iDou!{CuYMobxW9RR+!c)?%%h z*G+HQ!kSsv3C#?z)v8%kI`QOat}&CGQ(lwrA=M22`Sug%Lp3+#ST*BQ=S=shCRiF( zs-{`vijq1)=~Ky}!UWR>g^PX&FkNqS3S0@>7FI|4Ub$j^_-eh z7kzQ|{F^QdR?J^pd-~O{eC6zu7S33FA&i%?6uc9&HhJaBfeSvjVEPHnndpZ~B!lpU zG5zf`p3xkOeyBJm@V&uDJh{-O{!xJG%7ACq|8)dDS_~`ex1STfSo3%vsxx*k**E__h)1n|blEPaeGg(c>GU}8#X-Mvg-T2z0Y0_qN!0L4*EuOb|(=x431%j`=jeCOBSU1u)J?G9CPz57mu>M z%3V@%!!EMCgwOZhB>kT1v=4sP5I8Kav@_WcciT~xcaKMx55Cn0Kg);z1eKpv{TN0s z)&CfJr|9>F(K~er9F|w;E%(D+x?p)@E*F)uP;FRV%Y(v72=whz4+_i6n88sQA=LwA z>s+{x)5`J+{4JwXBBKSMcl){Mdt&m@5g0M)zGwwP6{zc+|im za;TH#2o_VYJZ${V!=fyyz;_sMqjjsw@-o~$TEhq4AJW@Om9f|Lm1U*+u&nTdm1W%< zz7iglR;8)slab2BwmS+(s}DUy$<6Vr($`m9<%}8o*shW>*-fkZUS7BM<%e3g&UCh} zSl%~t*486tUvk`|*dqF0$1a)GhxG0fh5l{#UH)uu&-dF~p5Cxw_e#)WajT%EPRYLA zFnDF~b6Cx2YC?gd0$3D`DTkA-@)(~;)4@wZxx$Y=maa(69d{h~)+Z4sM68ayNh|@Y zDUm`1sF?*;W<;h8KR_A$>Pp3=uYHr4wDpyBYhSs4<&N32wy$d4Hv6pYg^hP^n^;|1 zdR)=MbHT7<9|>Rj#4$@|Z7mEu{_VoSZFhG6XwB-~ovlB-^2+~M73?>fyrZ&rfqlfL z=npmfAn4x+sP9MJ4ME3X`r!AFNx2(tzYh+#|6y|b??%*p*x+N4{V4{&WY@?SI>}>2J56w`0xu6CN!}&Ahm( zbBE*X=&ZhUcB<%+@r&2qbi@uL{K&VA$c~fecF#_pGOOa4-rb$;KkPZSYR;+2vwIhu z0x54nN2fu`LjkxGV^mOKlAbh1ZudLE7jGD-G^T!H>)33Ept*6wry&B}UK=?c@sSEF zU&^PdRGfSRXTpR}A}F*YH+{`*lRj5G?$ol1sYlEzoc@WVg+B9xN4^>f-5d(lWq;ZK z?N6dmnFz%uGEHaElyr+)0V}u~Ffz>q$8xoOBsj7pyu#-C$`CW2zQfetrt7PAUHs~! z6KI#FSYJCd{{x?lAn6>2CVEKV*j~;i)Y;g$Mzu)6I1aFYGAx@@@nDHVRu@-E`9S>?sWedab z{EaJmaL}M@BhwdtY}-vw%vxA)BqfdoB$iA zX*}eMOju zP;TI(k7XKWg9GC=5g*4WY~=qCgLxcwqc|$+fT#e*j1RbErj}2Q;;2M<73I~DyZv+& zhF|E~GJV(lNv-*FzBzu|%V```aqstcE1`sH(s^9#4mI(tU))7 z2U0U#%=)RlMqIljVbO}cMpb&S4l7A+jq*jneyoj!Ubb5+;sLO zH!q&E^M=*NhJhuxISb#n{E75jZ20WKMT5r#j71v^DkkC z&I)#fPccXuZZA=vNFAZoZT?nWDzPLx=+#!s9mc*t@Lk4&S=@&mv-koJJ1s1Uc6(vB zi^YRl-(x7OgdfM&!1u+fEce5a8RiVI@>o1RD$fCJVsYwn0v@p+U!SpDTOHqfF)Zcg zd9>0ZiTV8A`G?XP!^f(=@6~qu5VNNo20nENTrGWxb9T9gD=J++V{^CSKP|C+)RMo0 z_YNJeVN#_Q%S64>hC^#JZ=kZx2&&bES5wGeRy7ntQ#(Vd^j9qWq3>X&(u}{{p;R{1 z+&`dI_Ri&^vfm4@-ccw#@^Im!VLE%oBc!v7HqJu3d!eGg!VENuw;hoi9FtE$?p<=z zc{3J{O_iKFb-|qQhYeTNjf>wIDOz;i!3ya0SiJcP-VD*THj4qxJjvo=W>0 zLvUPr=bZ=dKLpd=z@%nD7gOjEObZKl&ifRYKvU=j>J9GHHmV`?z}%1fA7D8;G3ID% ze93}2g|5(Qq9z=EAQ)8r3O#q{o#-1V3Y~UXJZihn-5Sq?;stXq4rUKUgTbbb#l8a( zxSmBQFxPlO!QqrDrVBh|8s8>S(u{}(H$SC>WjYTI-rtP(qxOpF3MyFfHfkxs6s$BL z)Ec7t1iz138;f1M$53ko51BzcCd2H6&!UxI1^ZIvRWv(RsC3F?%D52!*y+Ts#E|S> z#CJkzT^MZI@SU}jZ#rV;y2j0qmW|1Mv2F7!8&p{Di_Vy`Z(a3=t-hFE!I|Fj`Ad&K zw`$W3ZytX^{l(N1u?_Im&yXObnF8S8_HQ!mj>Z#7|$D5SHQE(zt$tA7`AYw9n zL9aZ&|5s{5w|Wk8;mA)<`7m$!zzd(gBN!~zkt#5GYVamCC^$U*rNOfq9zKh#48dV# zFqR|0w+A=_3_Pf>W)ZI6XNdJ7IFgitf0fZPBA#s{Lqp1D=I$CGzs1fk`?yU3 z6OLCgVQoa^0+(>d$TPXC&PMLcZC1LX=>5guNNR46bO*|aI1-EUCRXK5V<1+K>DqpN zxRRhYbVxr7+3BBA5}q9jEuO-&(=IL_Tl?2PV#rALq|PvAY=|exJA;rN@^xbq(gLZH zK@%2OUES63Le}C}BTubP^3R?`S)Jj5%h;Y1{(6yiq;O_MsXIl4ixwb!>BN3RAW$j#4Um)1 z2zgLO(q>A|`zp4DC9g)XWuL62Ktbw!C*3;LjI_PcXJTzZ6mYbBjvVDip{qW^IuKzF zMOJgdG!U$Bwb4?9{ChnzK@&h%>Kdh0Un3oVAvEd{!KP`P>d`{gr`|fbQ3SyfI246A zD6`&NMv!qq;R8+&3*6kNu!k~*x3j=Lep^8;{XFk*Onz!mj(2r+o8G%}5)VhY&g?gR zYxtS9zgb1!*d?zlcF&9ABk|?**s#~v%a1cx4&067d_y*GmS2%4-+*wI%Y{;BDMH~z zaH0jStKe!|za$?LA)fCt>|#1tJj%`7x}C;yu${~Nz=8Qm{#lbw$q~Ydo!O|Ly%Opo z2tt`g?B$ans@8X9Wm$r z5t09KCyk&Invy%@Er2yGb*A7DE~Q9{QODE1JqIsR;AZK(Jn3VSYMPvtv~T1MHcC5f zFwG^~x9%7(kN7WL`2tx<`&KTy)+@i>?5n@MfG2s6P3pl)KZ)DWI>m%tIsAok%R%K7 zK3|;V+NBVaK388XP>O(9@kDs#)l$T>Jya{~oe6+D2vk}loTLyBV0RpL!UVpe<8aup zJ2?Cl;2zx8R^UI#yE*(*z&mk^ox{~X%YShAe~^i)y8#;q#8#pey#1TgfWMU@Km+|( z3LG<{FuLJa(1?BO<^i|g(V;*-gDPL$uT%V^=nJbez|}Z>Nj*7>x<(qHG>(BBvmXF& z&F4L(mWOq)gi-!8z+W;czrGF=BH(`_SJfs1uEdf7=4Ushx&uQo6Fm7vR!tt!o=WlmXeor-fs7QHoK zOWX)J$a7bYU_lsfZ1Y<>TG&ANlp_W^Pq9g*3wn0<^#o%C&fDU@e_9Yw!YOuPB(Da$ zkbH-zcinD1oHtHz_#x^5r;an~NE2T*#+oSI1goKg5n?Qzpy$k?unx}A8)CA+24HNm zlrzzx78DjnUR*i@>1gfp)?0$h)vMdYzT39de!30KP|uSMD05U1;|r*0hR>B_L3+~{ zOE3L)K6f*=7LdKcf;F3-;~uysfc+}B8#Zp_K$n*fIZTQlg5J8PP6}E!VGkQOsNa~F zfhXM3S6b==`;f-{m@s5#T8>sg! zjZg!+VRnW3A!ofB(f_SMNZiy!x09foJUx#`clPx++GkdBX|>Qs4$c-VS?M?I7ws=v zTPdWksMECZT5)liz0g~;d1@8uanD>~)LYw4xHApsBT+E5T`-S={V(8~B7!TEO= zWKY4x2#1oHLNfuZSDw`n!%W-12D(vLE!&!zQEfk>bC*^<+q*Ygy-fP@S&)y5ByG`{ z13@c%NLjrdg#SX|!DCjGC6}Pv6lerNXXaD3Jc3SFpa|wvx0ru}fx`e5iNknw0j=ou zCnBwJNgTI5Icr#p4e?rrHK~puVFA zc}yJC&!<#d(GY*?xH5eI#c*hh!w2|FeE2xrFnsxtV@m&s{wl-Qkq;kAE5pZ28-@?& zo=IsMzKMq6dx;Mpm8anY+&Fw@b{IY%Ln}0_e*8NvhA+zq$M7lli&Sa05@o zgm298RbDOhmLh6B_bVDh9U~>uwmY^u=>P~V;Phu5Ar{7!i~g_jW3099tX$8uGMn~R z^U9{ptca}6pYFY2RQ2-6@k`vLh^I4BMsH3FPMZ*XC`a2`_+?p~%R&#WgHRWqI3{(P zcATO~c(ZLNlK1(A>x&_xnqXYTKE3(RSk+s^u~%OJNs@z~fn)^H2HKPke_$!4`(+dh zpdG;^c|oEsK#wO;r#yY>R zsp+n^=kqxI)9fdWYZS}6x5-PRr;TL;=M)&au^}03l<1AmVbRbh_1u6EcaR5-c4DM# zJ*P^Yh=Vz|&_NExw^@$!$w1BkzsJTkRvtl_SaIsP&ZM6EW*1H1kOZZEEOYer&oHfD zzqSW{1*UN0S{?gpCiuYji%bkkf7F9%sl)51JcWqFrGM}Qf^zHrR=*kQT0c=yr^3m08X zQi8wGHQo$45=5v4+8S;~gyv8#-W;m_If|e`bstM9Qh~l~i}4IrBrltz3WaaDo7rGc zqkhV<9xxL0fkus^KC6FlNJzgP2ZBSVy*Wh**LtoS;$O**48Lby^3_MT%5ZT3>&-yeG_n=dtA>O8tgeM{luQLt{%e2FsiiY4BjBg zH^D21zwgL-&6KB1;if!j0$2Khg*x>wNA;&ph&=!AVq}S99oXg2a6doaEp&#vSa=cP zdTj6E;$+#egODlu`S+SI(y2p-j;eGq=*5Lj+R~wWhoPOPCwJ`L(V<@iV&0Xb>;J4J#Rs*2 zvD_xKqp)n_M(v^(Qj5>gwh_&**$&)vf*{0-aB^bR5{Jp5AEr24?uw(!dXeD_D?`3C zgg=fp9eLmqE)Iu}uRF|T+Ks23eIaPmSIbiWvylbJ2g6o-PFf8YXImj%t3K%{^u!V8 z^i5BiRACF&1k?^hA2$QgUGRmt=Lh=y^uRS63GY!lq zKas9oeb6ObOjsYhKfyfNU<>ZojG61)uOt{>u+O``g$YWpdtnUvsVvTYF8s^3@zZ5u{ASiTS@ zVxe(cv)R8O^&E9y1ji6n7YKnqNZHgWLy2&V_SZiIDNCnK0;%WzV~}!_fz(X}DRbwC z@QJjL9G38JSSx$4x}bjKOe#Oxvg@uGE9bV&#`W~if&=z&$Q*&45}Hj+nyhtI7I^&o zHaa(AgLUQeHJZ9W`|3R3|+p!!$-3Z8}jha15)vWo%WN+a-Ix^F;#1;C}%!sBBcbN{TZ>bp+5v)iRs`go)OxbX)r9@A^Qo}A$(Zi z|EuP6EZnyy!=-oMBFI?y*-%<-UL?VusQZqF)&SM0AJjp=KtH7cew()T_>im3pYT(4 zaIl~>X(!$ye1b}gZ~>dNNIpdF0w&k6@Ci5GsXX}*0Vm`AGPO+(U=y}d9Kb{;Y&wkN zYy~=DduJ$RH&ZeslCza|FdSJqyYSU#pM51CWZvsDGK*`4Zt~4*Y=ji?){gBpWwBx5 zakVb&sjPm!J_wwWq6{I%O4TCOS5;e_nXz6z#imfbo!HR#<#*tYbcqWOi!H0!zT+(+ zhS$f;=GXlsehmH6PYI*w<*eKj3IDxmt~_06p%pUaDY?;vj!UYRTD{snHzuhdAn4qd zjq_hmS+aX}hrPD(^J8+SM<3i$y6Elp6IVs`2@8#J8|h=&zh_x;)P^9pWqy4}M~AtO z^|tNZvusK9)~V>ze8>&qU_sZYliXN1>w+Gft+Dw>Q~e8~#fpkH9t;Xa5wpkC*ftY_qicc;`4TmiWPo;0uZLS9}oRA0VD zdsb*{SJU4B?_gG-!xQyzfu{N)pi~cE&6)I%n;rC?nZ<(j-viqruFFyvFGPt4^?T3G z{5U<|9&yIm7G-Nhljd;gaO&&xH*Pr>7*x1)?h2bDmWfr#E8kwUe*3}b>A5lS^KHfO zlr2FUqLRyc_P6vI=^is}szc-|x2fA^F5Y6_+tz!md)Uk|Xdc=|N>kk>$ulTm%6w@3 zeD*=^_*S-NEUFh>&CYy#X;ys6$@y1T?c2AsVB5+SGd$hfdu zty^E+mV9nr<=XhgrH8zdCd^r%JMY_NMZe4e$-=dNQ6SlQM-Uf=4 z|3m3Jn|Jz$!%?SG`V8;mJ}1^a`b2VcWq-FG^QOk7y2s6R@BebY?67$WlKr#)NcP`; z=UtxR9qlopzmu=m)U@d_>%0bc+hH{%*n4E?O;`n+= zxlvJ{=#O&DKpP05+L{B^V>~RJZ0tLE*p*=R*tU(CKRf5`)=b)BZoYL5zO|Brg;!Av z?VDoNWvTdyd-Za(wV_IA6^4)tC74N9?*l~VB~G!jIbqXn_t|bQv+xGZ$l(ajn-Uf? ztLP1tTGg;gATIj~OLR>tYRjYnS{kum*Bq+$m~3GWygg-=;3fN3YyH|hqh6BDQeiTA=s$18Nm=#`~RBqF3Y2;<$Z#e)jxTZ*&gxBS_fe` zkGr<2`w2Hh#+i*Zn^4ai*)ZZ%F9dv-#^@18h6≷9WK0CApUWn4bB=meh;stxvU@ zINy8LhJb)gG2<7xIgOpM$=_+^=Ahs;{?&eiT#6@pm6lImvv$$>XVV3*_M4bg`6^Rw zoLVv8ahh}Nxs2=!alL(<`wxl^8xuRmt#F<+>e*XUCx*HEB#c=fA6*tK-saf%k^d6b zLXLSNC|aJ1x-B)Hmak6tb$Zp2Tj$I1sCdamLh$|u2zn%(+VCvY4s%B#K7y(g!*RSzEBxo5%JzGG*`4;T<1HMTdoakSIp zR*f0EGEUfEyUV%X_U!c8i`EBp?;fyzQLU^@Dk_RBBOi|51=kogRc6nu1n2tw5O%Fy zs|{+b4+gT5FOUdG+Yh=;DjGO$R4H)7CRlwqj4CWZ_4V-8gfb8a{MSQ`wP5?+F`%~8?F?7=1RfZ)Hu*ZLqID-CxowmC$4%_%Cu=IqfW)WV-73|w8jKR^yKr8#U8(_#9yJKW8f(jral{V@?D#n~5-j zvbY;jQ8(hWbEODSfVazW|45!47%qQ*mHQ=)pbSvKlDyrny{x%26sXyN};Yfh#E0>%9pbjM&N5K%?-V1V0lvxi*w~3iWUVYk}S-@IbC$l=_JimGZ>`bUTPK9$ZA` zAA(UGvF(J#tHMXlj+7o59y!h$n(^#4CP#i$Fx;E*{)YY{#_zUfZTLNpukCQ5*&K-BVr^S zLfEn;D6M%Tq$DX0F7ECy;^1n;W;D0l;POI(aqLpli3VUz@4VPYLCmRxgY)42{eu_! z_jy-->#6*G7yBVY`}Yrt_wWBM8+A!`gLf%GyKBgR7W=yd1+Y!pgTlK)kC8)K?(aT5 zL{1faYu_aqLBcQ+FIaPy&bBrXEoAIA;aa{TEgWd+Jl*r9Zd~Aoae@2oifr8@GShFc zyBxnm)wF681l;}ke0fZtPR8j zk51jVw#P)#$;sijE9_y9s0{F}>hTU$o7UjlXFH6lU!~IV?JzX(F$FeD&U|zNZM4w?$=2BQ!oTcp z)hqJ&FIcL>$Zjn03)w}m(|*O$h3~atE^de|?F#%2T55x8usf$BZ=^{o95{Ha#re)t zv=#a~*}V1T4@Y^-HPn10do~0o6T0gd>S#{WpAi{sI5_k&DlRBA+%WNL5 z!zNAjUY<2)O~S-sE(1Gu9ni_8Z@2uUWJfRc*6s&CIoDsF9YQb95?xOx(DS>tfX znkk$lR)H6ptuiTZgs;=X;g-k98Oj&3AzaonMYIHwv>Cz0AU2%KKlG_z>Hp`A(=jpjFR?enVG}3OCbU8WY&87RtE)7H? zRSbMJYQ+HximV5r5YZ~RXdP=*AJJVhAE8zBO-1iTx=n^oQQ3EAcEf}o*M+Xno{EyZ zvL#hq33k_Ap9;@6`Lw_Sq&sn`1jQNjbZp{sHv}aYvQhtVv77iZ?k%i(kz3ucc_HDf z5XY2DgAzYOF4ID(cWpm$W$jR@ePyMv6F)_jl}hJTWAO>PMuE=Mebv~S+DGU=PQ&!Y zy+I$LR30kU;xD*Wz}3jsZ4C>Ks_hO(elO(qW7Z6rf+;^&ck4V>*>CFPjvc+H43N_$ zZ|$M5(y?uO_=JsMW92u-#82Gjy0+I-Q%_Dr^y>#+pSXbU6;Jp;zIf&g8~?%T57_uK zXXJ|?tX{qQK~~m--K#4eW@SChKQrbS8+Pp)b2@rde)rln`Npw**$;M>-_OpzU%vA} zw)~Yk`0v9Qez1o+6n*c&)1hNA3`0Pd0~yYKT+2nxGccp7{A}PWf9}oO^XDr;XTxtE z%767{n6}{TSzL0Tb?{kbP*CNwgKUMo_QH>I=KOepB|jYq%3|YKbqoyChbY_K2~oe; z4B>~evFO%7Zo69yZwGb6GZ0u?wX?OzWbrAl1kXJA$A%4moQw|Ho5JF?KkYl4{OS7Q zPm|B=ubk9l@56Ol>(0f-o~zrs?&01ZlPX#N{;c`Cd+%pt+~0d${@(!k7K9iM8(evR z*Q}S+*6b!Uq5kH9e}dQ0)uB>be(k*4JVUuU6raAO)WhJ|Mm@$G2?PZ$1aY|T>QFPY zW*v3#cn&w-RG`+ur*OFL>QIz#t%FB$xbX&qnHn$!@S%m0+I(g7b){dLnRK&(0nTqW zsM~-dy(piU)D~WG+*448)~USQPHoz+Ud?|HNAdb}{lk0H{?Yr1{JpN9a=T`x{d6!# zKg~4Fe~_~D_1Eh!zL(1Q_qu-Ld((c?`<48?uK##%+JAaqz~AfWAh$z*jdZ|!@oWBG zM-TqqNDt$C<;I6JxgBvJ=4+gfCcFh$qQsfFnsp!ezBlaQVeN477R23nV;Zq!FH!*3 zJW;Cz&#HrvhKok83$~HFqL3@5;M26o@<_o}zFu_Tko?IT#g9(T6l!j{mKk2>y z?wYNCmzMs$bMZHA>V%AcV>kBs2&dL&aJx*w2*x52;k$dTK+oZ|YplG1Z`-6LK{<{0AyFBYkoMHpcgoitUh4>8#&c+!tyCAHd=HWG`c6N{Xl!j-l_AT zUJip_8fLHh=-KrAb6JCY7mmzM={zA|&Z=N(|INaLPwLF`OGI49@fy@HtsQP74JPB2c(EWUJYA)ciNHPlT@JkwA zQ~L%yA8z+pHqzo}-j&_hw_nPM+Kuue<+v>FDQYfU)6Nq%Yn^KU7B*_4Os*{1Fo19NP$}A!nq$>U)2lcS`xMKpVF+IMRm=G?CxFdC&I)Z<>p@? z`%8+jMmtbzCcL5zQ|O9#Kf)QpM3Vy1R*$m8J?EzV*V+_1xnm+o5!m-Vt4jzv1Jmykk$>V?nFv(}}_x3>zL z_B7M{P$hS}P|0wRJNmIu@k*im?DXL$x%=67H8t0@Q3}GfH?>iUkg4d1csrVVXd0uT z!@!BHMZ4(U?jeI)`iWa=v>#7sF(|ZPPUVQ5W#UB`1^sG2&wVp<42ah7UgK(9ZDdJj z8CziO1>%ksz+Gqv7L-KEC&41!sUg0yhSgv}}Q1|^mn+9#(l|S3P zoAr4&=LYMI5&8T^TH?;n7cc&NXX1`87BBuHxh8%G>vH7^>#}2qeCNs)`Oc1G@$YZi z^nSeZuUw?g`8yN+r~d9ozx4ebBP~ULMG^{4U(N+BhqUV0+1fv%RhfD1QY$ ziv3=n65xTRh5mG!nuq~q zL1Juy3k}KnZ>WqbbzN;LkVsmy{dQ7KQ0dInFJr>KlWS|_{?cJ_9aZbcjxSg}&d=-I zNFR3TX_d;Shu(Nq|B0$wkQyfHDnHdVf^A^7cVi*>fZiKeU_Zcns81&_{&52;5p$n@R&rgOd7hf zn^u3yrQKN5SwaB*ZCxiEm7`*6q_BTl*^R5H5K2i-o_75A^__s=8D;%TXDG}NkgL0t z1X&hnA*P(N$J&%RN{Kb{_2d%Vf}O+)<@h7G!~MIXOs+j}n=SlQ)z53?v?*C`2bbK) z&HW$+$v$=wuhvyc?-Q>!Pv^4XVRnma2*8_>_&P|K3ZXeXBrI9LcclBtWYUd{evM07 zo#&T_P1Ft>M_4O5;&K#>jMxU0z-{&YE-t=S5zj{`Z&L2(nVFyGPRgSZvcw2!7YJ2W zdQwctI<>@SDMm>C`_e`EDMl!1a$)P9qTLvmegk%uTb*N5e?GxFzCG`l{O&ke4q%#l`YZR`&}`^qgO{ z!b5WM@f_E7|9;9Wkd0xcRU%da96Jo8K5J=XGKh`hJD24DmPiY0Ek{Jw6s{#iR{G`Y z5V@Irk@Y5Qt&TA#YJJngynLpvxyUs6`qoMQPRMuPo(HBRt86(83bl$13gzy@g}??b zTwIpMw*`)Nlcv+8!zYel<``g%T#fQfB~+M??!)j`C{ zY`enB0z9Y?Y&zUEUOfPRb)krNx+k6K#R;>Ots~S9qJ!Htdi+#;Tf1+NFkPTGMJ!Ku%kn|N=a z1HCu&*NDG?9!XxXb=MH0Q^{7el-3s?8)HG1}`S0p+ z7j3TSTKjpXVQdUgeQVhCpGgx_IehT`&=H+Tqf)vmXj5gOtZ_1|)UnH{KiQ`wNYllg zQYu87^2OIyt+={q{u?V+f18)|US3dsgryDZ8n)W2C~aLpK+?Fvl(gar**MzuYueek z*(a7RD)DaSqzQbrXyv%}dW-sP&wbOvjLh12ts_AFBdY zz1>BeZLz@Swy)_vzR!tjD=E=rE5AriWQXKX)I=)Fl~fsRpYSj~>ZdiuQtS0168SRM!cXu4#uk(=py{&B`)*PR^_s3te2DtjjR$lG-w=h4MF4Hkgk9AT*|5bkIsr5<6zh`hUwHrU8B zAPgq5Qyj7lFfLn>lwY!J(HfU37XQU{j_>8H%-7OKSGlZ9Ua&GHWrYFXcpsmrxtCX; zlm9F-pgXzbW=>M|?0WdpSptdh6>yf|XMSy_fw1~C$%G4Si7U)4D*66+D8Hr``I3qG zTMI(_jH>xIC)zmA^3T{bCeb%<$BKP3dZyfr+K~rBE{4RHe^{RVSZ{8{Q~LO%fkmiiMu zP2h!7=G@Dz!z^(Bm-DJz)-Fsd&QDrVrbDZ2MM}!b1kH{UowmlylRNGQpHVFTJ6 zU{;T(zxReum-=p)lQV3FZ&$Hr=!~fj^6O$dd0j;MX#dh^ z$Mr?>cb9thwTL{CHhcg1mHQ_|dUPBqsa3-6HL~5Z?de`q7Yt2VaRmsF{-9HSvF~_m=%XHIBLy0JYor6!OiWC4(|=o!8r*83 zw`!sWAN$6>smB@IHsRl>f1YFb#sF2?fxR+^Rv&4Y$}?3Dpr?3}LvH8OA`qg1H6Xtj z5Xq(6B4k~#ux{Iqk|GGp%kO1m+$&f9tvGzR0zY4gX6>t)*B#e8%&WRuK7&8xGu69s zCAyTyB-TN`C)ei5wemf|dh=$I6H152BL~QUFdPGDEI*(aTl=;lKVaS4_^^*A3Pgjc z43B`f{FA~|R3RrO!Lgm#9$m*VgeSsy3VTvYBhfGxUm#;PFHNkbl+GAT(1Zri%x+Z2 zJN2L%^Q=1T-~oQ10X(}73p!nSXU@hPL-Q3T<-3}|b9CizQ9Yc66kJx`qFO!$w37*R z6~A{lh_^{Bqm<3>r!Q;lJ1$14Tf#Bf@U+vj>as;y!gmyMb*z2293dMQ^7rJQP%l8}0Aas+F{(S*ziR zGip73y<4}Gwrg7BfgO&&8pFdwfxuFz`$tgsjz)FgleWPet#$U2vo-m4bw?h#8OW_B zuJ_%ohxT9all<4Dv<;hGd=HdPPd*(fFAzKB-~D~p<{zFMJAZ8NbW;3p^z38t@!8pH zLkC(|TaH{hGhxMOwaQ{^akExwIrElR6s}0mBi)Z{2kV|lM>$Vh(mq*3zP2YL8`wzw zGMDS}U)s3U7F1&sW(Cqvrf$)U&KQbUw{C@jL&}~E3FXwEg$0-1(^Eewpvx(h{2;Cc@v%i-xOp-{{NOkf3JYZU#Bup)u`WMI!o`#&uP>bQ z+Vb>^v`fXCdNX)rtnyJXgFPqtId#+ ziL-EoDh#Pjc?wNpC%&E$B=9mhOJP^SeG8RBwxo%7*Fwm1TNRo8#)9qgovR&QlYiS7 zuzlg636aC1E);gydSkZpTVfGV-{CXvFts%c&LJiNaSF=R>i6-eufue&{+i}W~d;Ym(L6sRgO@9Lc@*xP9 zkNufl9kMUEI^~tn>g+#{S65pVvNo3mYwg3$S7*+=x|#mEjNZzL8uKTAahd&1-3&4` z#OO(Y&z@)siW(A(@K^>%QwIvxlv2vM_{j-Q*IiP5(-Yzs9*bG9ba_DP= zCij~BBu$0g)RTvo(zJ|u84F}&LU5$WKkRHe>YX?0=kEreU3_7d z<(n6@?)NR_wWZ||N0+4@4v#sxbirY|po>BRx`GC6%{t(2O^P*wt%jr#I?*=-8&SY5 z*%aN&*|pQ3CC7|U@tL-72@V<04vmfu#qX}O6PL`E??@4QS4SsWHTPIHB{bjnl|c8< zsIp*BcR!FDO^Ly1N(FK&@kha088?^guSN1B(t7k>wGc=lj)XzG4dJDPLx*@V)>i(F z7X?Q&mxHM)aLSJ)Nw9!25Y>$+rB38K>n&fDF7ZA=9vhER>K)c&XwQ!3_QHpUs%qa~ zDp<<7?b|xb7uh|#pr!kku+z1kh*26(0JtT7J1J|?f$>=9;Q9)5x8(kO zGOWe*b`gk*=>oEV4;9uVZzs)Crm;~SS%>5+>wRY(S>))wbi#~1bLDu{U2so(|6O*g ze0=IFvl7mxS#7WkO!f@P@kYRj(pIwEUx$7+ABSI+aS(m+TZzlrz|n?#gvUK#=qV`8 zf@0FktJy#{>LYmzTlkUuoqT-mu3hzeAi8R=pcX51{wLJJL&YfQgr;u;39MN0)Ou79 zZV!Xy$*G&`YOR-6UK(v$ zy^amO;5}3@A1(ZW9&qg?*`Qes9)Ks|D%s$AkW+-1c$Rb+(H8@R6$E^*N2pmR(+$@OEbA+X)LVFVlXbS^5O|C7@X*iw}Tu zH-$!qlvfToT4zTJ4F6jy(j;u}GTOGag{RHL{xAOnFiKTL3!zR$XKA5Ss)H~@hsIO5q7k)wSxcX7k`6(qj5skxxptEc62N^y* z1dzP0q@P7C8#8CYf;lh)#-F}*>-4jj>VCrPzSa8!Ci{i#Kf0-`iXROAv9eNzZxv^+ z1>FDp6ZgRwL)e^o2pi4UN(fso3Sr}6^lD2qFANAQO%1g@+;XT_#0X`X=p5!TwB-@o z(3Ii;F8|lH&jeNbenCmzLk3KC?gF<6mW-X9Jo^ptTH@DNuLIbVJ=8;FUJ8W|$PmCr zGZys+T2fiT>e_C&NnC!YH`~H?DGJ9lqxzx}cX#Cg^w)@cM5YjldxZ4O*2@qN0t^Bd zeFzeBmmJv*J?M^EC1wpKwj+;N6F;Pcf|*6kTRun$1=$7dfcM#mYvLP*go1S5o3wyj z!;kyDq|YQ{vg)@pI0z6at(cVTeg7FN-A5<-`!62d(Phn|`6uV4e3-|aU_r_?6@HuN zhp(70VMTc6yD3Yr=M;aJt*S6spYk2`6OK1cDZO;fV1R}z`<8m8uUa~6b?&Tv(eeXk zv;FtAeZw}&BWM~x*sxtieDpe>3by3ajU(2rjZ@9lPYmb>xylA&Ty%V)ABzT#ISQb& zRRZYro`itnlu+9vEr)u9IV;>TBEoBE%fmKPQcLB@U<>_t^z~cfHKd=XQ)eYte;4P; z1BQ4f1@!}4P)-1}ueLz@V`0S1rs~13n+G0iL4er*N$e*ai`fS9#}0!R8)|apr~dXFwlio zBORG45S`Z$chXW1d)AIdfxr}K5`CvjOicmu@HMgQQ)kL&E)dNA?f!c9$EE$z3EHS) z_OJKXXUM72a!foLigS}`rbbwYl#hQ1he;~(RWdmFOWM3Tb?HSNn zejVK_5Gvzm?eeKD0%I+$gBw_V7&M~T5KyKUxv3rb#y3|w_401okrGly-BuSZB7~%l z-u+8n^EaDCZ=EuGcSJy*e|7ac`9BqJxU*6^*Xnd`REhu0+~J!>m#hlOdi13x?@8LF zxYsj+7dTBw_4AxQ60W$KjhPqX;;I%eNO4)M5)Ricvihy6z^IwQY%+UB%;`3yK^G8mXRMX~Dn)T~UKWNhmV5S)^95+`b#3xCDs zO_-23M?O&_w5t3fKK_fN+P1>9N{>u`{|qlW%)9J*TH3oQYY^8hnhgW;|G{Vx8}tI| zn(ofZ56)>D;)O}#N#}9(h!eav(m>bRNlv2s#M*b9q(ra&vlGXqT~0{2oHl-ObRX}@ zHxjC3$~tdTwfMUDW=Vj~^byWJXnrKv2KT<= zFXb*hLWK2#5yBVr-JSJ%)(#VzrN-V_>g-tSKa@Sv{#RnnV%JaJ`HFns@rRQ?E{gf` zq>A0U<*2n8DlI;8+W0U<9VoVC8JdoS7=^={dHyMvBr7X#M># zVQU-hqj(`-j6ZNdm_u>)wSQRF;{5Vg2zr0Bxb_IJ?~kq|>uQB_h}Q{HSd!A^|ItN| zc1$zx$xY4BCpS+rr`A~AvzOa4$4m0uEAsi4^7*&qc<~#ULX{O2;^b$3(h2&(S@Y6o z3sJMwKr>**d2!u9GVeqWMZQf(kukKxQX|}V| zk)F<}Z0EE0BuDzu{`m2qqn&GB9<*|K@}_r_K32I5aP#aKdw8zjw#5_D2MrrAeMmsG z_UT@E;9jju^}z|*y*oP292pcfGd{aA0DT{bzCYFVU4KrWMjdQ~DacNL`4T^Gg8Q*m zPoEA`-FP}|%eO*Adsjb)mzp;dYVK%%u=kGguv0Zti6=paw@OR3I9*ihVZXFAD0Jtl zfnq7>;y)_m)yzfb3wzfdg+fcKb5N!7xfwHrGB@jW3vZV~x^=<+kbhechYq#~Z>RQs zTfTMR0PB4*`Mo{yR#$8%&t2uW^3%+$FV|#Viho&l6GoTIzwoK}t#fP&KYZYxQ#Ic+ zF|Z$t~7>D|!tu=+onP^aT4bg!eih~@|E$yW! zVeBjX-|*)Tba>@`o|*M|u9x9Jhq~~|IgR_1bCSmW?-;$>(;w`}m;b(4Bmcd+kDqd| zLr0)EP9WF@|4bkSBjK&oK`(%ZLYfys;Ad4Vc-M*)GgA&3t2LNz^w)DZEBofUQBurv z;|$S7P&FUZYM}k}OjqIL$(C(HlE(MUptQGnVY| z8;&4xw>Os>oR4voL+1yk>h1Isz zq1hgRX=#DQPj~re+}s?{+QDvH&wf_(ugHH^_MZ^@dL)@W~>5S&__Sr9fZ(X+a` zrA2V6$D9og=Iv(W1w?EOJMB7aP(r%O>}ifFQ+q)>QTRL~-ecu-ZHyFgZgAM>#E?N@ zqo-7@)%F&?4vp@+9;j4fU_QlipcT0{>2#vCf(izi3}|v$&>*Poss-26CoTkp$;&UW zlnZRt^ni;ePD>FdtKQb`kgwc-G!QuMKhjNf%^IOa^a(v!h1*5OeH% z2;O)qm1-}alpk-}%vu~d$y#o{F!xMK!r29L-`yZjRrga1JH+9&*Tvyj5bPh!j<7zj zU0~K_fV$WWB#hHbC=%X%Fz~1 z2A!t1H!(Qt?}bsWAcMUo1g2^q2yJ(~JIi&^c=v+vIk%6?J6Xih4U-F^uqWW@pEG%N ziF}eP#dE5U;;R-!WlU<b;_ts@0sq&knuR=m#6nM>mJdC8Bp<5YkGSs(HLO+9 zdijr<3kaIW3ONcc;dK~1o%*k;_4o1M%>mmPnuv16K?_Gr%JdJ(a6O&&QEu*gNh3>F$>)S)YlIHd^V8jY6UVq^1h0B$ z(W3K70b4H@%KKW9VN5OH#0lKSV~E*X#F_nGSAv#xIR*4PyrxTV9vE+QT-DpO)A`@i zOYh{WE~!HbLLy3n1GX(%wroo7r17Z%lXFAFGt#OrRu){FpPLxD+GEvStC*9?9&?<> zC;J4XyYF*J^qI79jB~74(#e>F)AKTq_h{KBdUHhDrA3hURN#6FRBKLKEtbezgNx45 zb%y&e94%6X!X-X(2Qd*@lTNWg+ofi|*2*m_ja7lz)QU8-u#kSgF$7TZyePWMfd#w1@YED(+ zgt$?`8zZ8&OzGdlMz~Wc`#21lwEFb?_1~@l*0j=`jaEgvNdfpGsnQFyja#s83+-N@ zeI&@JeA@Bfkidyo97ep0O>sxg+ltSE`@wAWUKIUQ00Y+OPF3QQguL)`Tf@B9U(d>T zE3@6X4uwk-_f2itJ7KL)_N}btceYF~4h|@q?wT}pNMNy*YXxg}^C}zi@q(C>Ni%cu zv{hrL4a)qcxZwNMmD9_@XKb446rM7EY8eIjkqv+>A|wdN;YcPUdj~l$B=JpcysB%< z$`en&0pSSyWn2sfnYk$V(>(9&RL`5Iu~vr7h`iz&vNVmYf<6|OmEBu0c~#JY?Btud zD{rQ(m>00hcjdj!vx@zKOT!{cgMCY5{L(yJ(*pw2UEP*jAsQ%R_xv`^I-N^SteOK# zC!g!otj+x05zEfU?;D@w7nJ4Yl@;WdG+wyqlIZQdaLky6-rk8W7{pFtS|U+waDDc<=<|;CExsbfpBqnc*=y(P2tnZ z!a#lA`a(&tNVzx#PqT?HMTrk9_*qi)g=0Xg{2c<^Dkn zx<*7v;DI^ejnmYNI)3W8^phIqfSI*l#UW{K%?-@QMtSar`T;zY)zIm5=LP4~|aE@Nj}Ru*W# z?`pB|-HhzF7n>*=4JGv>MWd)0GE-88fY-Gb#HI3zYOZ8z+v;=-tF5K54;u0T=o`P4 zj}WERA+1UckqsuorSPW^y!O+yuv8ZCp4_^fmCu~f`PUNCKUovH(%XAwXlSmtcdiwr zXWlO-*zejteck#jey?Ed=gap77R`uQ=kHe-7G4}k4MUO!U_^PtOgS~kj~AP8n!!V^ zXm25iH!L4}k3HfIufCp;{%L;5DsS&qA=L0H)m^1!xsC9d(sIFPyyfsBf4_Azl$MLo zlfm^c0WXx_xXm^VWo)tfihv6dGuvX(_f@*fv3$bYOu zC%;1{)3`2eK{F!hev0LAvj0xpDxXj$(25G<6p9J%0zOFuc52laJKDJ23~Y<}vr7kh zUWWIlQKecnj?wc;yieiJ678J@TD@)JALGlG_(i>SP7Fg-Mq*9k z#HvB(7jt5GyHe+2DC3!kJ()kxm$$M~vk$Omd9CioQyC?S>mK7iFNPAgf1@w>v(y*c z*?2~|+jYP3@S9?@NRvX#l>S`g+-ZlJX)+IBkY+&TZ*t^NIYo$%eS*+8D2sejF z7MuLToQ;tKL{D~_)zz^c1|sbory~-RIgxh#NzZk49w1@?CnBwe67bxeKksNu&wLBr` z^P*XQX^|9-Zys0>PBnO5$e$m6NY59|-0(gH?|=Ti{_`*V`4)Rg3KN&${jZPc{VlVB zQWzGBYSn-FbG2EFv1qs+-xI{*0^Z)zO+cCLZ9jI z)L!1l_oam#^04kh4!LOdAZNf5A_GQ*4sx72 zLOA+`_Fexhuj%_1ChVE&IBlf-X_&WrX5h3}awZ>fozl6d*MF=QytX8|C~d-Cs}a7# zJi-PB3?C7@(6#oGuz%*(&C_OW37t|h&2?#Xzpq@A+`JZz^~%4Ly0KS{D_|`#HLW?m~iQt3Zs8cRuN2Fw3otBQ4rggs2 zs;cE2XFpzzTt;i091-~rl#`_r1j7eF*rK4lXyT{^UMi0Eic3Iyysv^bC@Ab~k-s^#FVJ?2CA2e6gvXvB zw0nEbo!Wxk^aVwUvNV-|dIvc5?CI;k!;+hN%$d^vHc z_c5Qvqn#Fo9?M!OMDAs?nRoh5`LKMdG*|vj{_x}%>@vhI3UyfM>KNXR{4mI+rR22H zL`_@hI6A@in0L~I+zk2t88(Eq&Msx%EO=+Sd`jN7=NozKC#d5N>PSExSZP`!*%$f5 z@35jf0Bc;g{{1T*meUuZ01duU7j*@3!t0$-DlB}qWlL>4-3QDp9d$O>)#-%rS!@ZV z-$AYN{jKVDTc6OSxB4Q+I&}rvY55(I3M+N$J`l}{QRlCy^Z#h16#xHbq!7km7!i13 zBjt<#zl;=6Elp0+j8m1Hd6`YaNc8BbxGN|PHLXn*A@8N!L9CDGu|*n4*j0e>Bg=ZOzhQY2nSAbCaQ&MmDdJX`SR9sK#u|3llGz*Sjo z|HEfL&vVWJR16V8Qv?xkM3fnsWRiKFK~zLU98(a)AxE5W!g)3awDg*pO*W^QnOUKk zWw+E@U9;?Vy;+K!hxfbo^Bg$fvF`o+{%@ZIp0(FrYwfkCwbx#IZ+uVZ-@Ewt9DG~y z?>+o`9ln3z-_Nw9f62f1@$c&9Onn&cao212V`#k&-!azK(j7i`sqxItxR{CgIqMmI zi})>k@cWkEn%?U!6?AV3{cX!{Uv-!A5D?tx!FFukEW^u%O~XA~w`z2nDopcPu;1z6 zDOuY?6vq?I>$DOkPmQ(E-c!Te{p>~j)2hFPn{%)3FDTf5b?%%m4~!ah;LABxr)SPQ zT~+yhMaBCFtT|U*eQwr;hg;P@x{f*W-O{DsJvVyvbNK!JkNjRx*kcrp4as|4i+=Kyc&0&Hy zXK-D2{{h_m5ZYl@_4!WT~&BScFVgrALhu))1i z_}Q4p!X%G{=04AZwghUF0xKQ6F({XYbrBbcb!4NEJCdd^W2}Y{mQ$CNRhN^yJ}YZI z>tg&X7f_90l2M6OZ<7`Esn~H?%d{brn>1}&68>>=IqS2t*XMxT;;G35Po4^ayjioD zEF3fzN=!C*eE)Rfl&Jk8d}(BV0dfz_X++zMtDKGAZ8=|MPs^!l75oDq3vhI1yX->QMet* zR6cMB-H_O3XrXkWG5g4IOKH&f^wV{ z3#E?0d4nKmoO{I`>@08L8e0qidv~tP5qAiFh7Zl802_r&R=_2tXavUv%mjxR4dN1Gy6(A_e8zPA zrgXy7A{wbJZ!T`}lInBBybbFsJj<0wSQFe`%i$n-g_CBQHR47#RxJ){zC& zh~YiZM4A}O#rl%d{E{hGVl054rC#;jT!)6y9k1Gn7RQb63sPPDZm2 zRDr9D11|hJT%xZ9oST3|UrRfcdW8Yq)RGQ3r7QW^F=+tyqbMKwb~N0#vsd#mJEoNc zcMktbNs&a#@z@k9!FknFL?s|7j1-kNsC+wa)jjg{Z&tfarv8(K=+)wTYycFO)uizmp4hK|H*h+cUMYdj^)`A8` zASdMl%?;}bD)J9(Ggk$&OHmyDPY(Yl!9n&Iydi~B5uGc9Z`zLe*|{^jW8!hcFf58M z3KsL(*JtD-9eY~IN62`-1N2=AIzu!g{7LvJ?P^L9KGEw9L&f8Qh1~Z&mTA?qBg;G^ z_r``+H zn$z(6EJuUj4Co*TyWOLLmCg33e-TV=4wE( zxS~-*m!BmqCmud5eAbb$Pa;Hj#Ir|L=Cinmn)e3bxsEU1hT@^Hl1tHHl!(+;zA zYz|bX_=LKL#s~+*^koS#RjJQtXFw~&zsgp8WFgt>{np>BZ*jJ+~ub66J|MpWvIBG)F#pdH!vOiWiqinc+kEL=K<*kLaJw%0`4wK18tZ&Lg{}MDaSdB>Ab6S3KSwwr{cObL3oliK+5uz2tBRRrB zL%y6V%-^yF*=K-;F#`q+=hiILCXLyeen-|r(!htGBV)*xE$jni7Bp$N3wEYc{2a8$z2Os>TM14Tj&sK=;;3uz0Vr%6YDXiSD!|o3LtULu5 zTdqx#SNL~mX0O@ZU2!4s)8+T<*lY4gyL-z&4ZN`8t{pB&k~GCEPx=)@ys{!fv;|*T z|H*2|&)v%&2Bm6^Dp40?Uc1-`&sAO8o1ee;Qq}BBd-L-4UYcEfuCnr6_3RHTD?hxq zc+pbXi#^dN$uA()_4J^7UH~LGyy=TrB2MP)fd@)D; zE!?U)Q=YqZY}}ZURk1OZBS*)N-JV-<#!%HR8z*(lBqJwarc`h*lgkjpAC*Rl&_^(c zx?lU>luDacai(&@vy-ilKT(kxJ8R^~S+SW*t&dyJJU*fFOog2FUzG}&x!b8!l`*kZ zRI07Hr-XQrfNt|<90L}vw} zena`py2UYpfzgHq^6LLiWGy=WDQPJ>unzC4(*mv=#1_41p~0g9G3?N9n$pDdpNsY% z7<*g3IGRl#+ca&gY+wh)Q-Xs$n=Lck5*(C&^u8)PlaYbGmL|R|P1j!Go`+#{ln9Gl zeoTrlQ0WYjyBoJ&5vND99WKhoKPZeMiM$jkc#v9;F{L?P!NUXk4GT()7g)}u{Bb#h z1EOH|x%CT5O`m*r^q!xJ_U$jay?c!Ka(3!~etr8mx>{IuN(l@sOiS(GcYu9QC%sjt z^sv};Ns+TvW(=^>;?oqXXsIW4NVLT7}H- z>Zwzgu4R8Q`rtk*HgqMMS|-;qOL^U7Hia|VnRbPWzz)j!^AZx~8P%@vlISr7V^z98 zQkA4k<#Ja(f+66en_f>{j}@5KZNXUI9f1i6fxCAX)i4$^VZ@`gOl&g}LJC3>GUkmf z7%|=-)m@0{eo1>-<=^iS?k%eob?j0?f8mPOwPnvDD{JP@F0Hg0+aY{-TtZUPuyE_q zohCmjgtWY7`SPNCFW-WK{IMS1nd2~#EfRx87wtDn7({!g$|Ihx%cHgx6l@z+ur)t_ zYk^nBuwfZq-kHOOWs1Rh+X_Z)&(GgJs$g5*W&&q;d1ZJ51w0~zWnzC=E-sp3yiIfO z3+8s1UP)Cae*@EsC1BXdRSgxwh_C_m@wgudGlT`#i_gWNj~0_)6*vT_~4++l=l`c zK0SHz>BS3APhOk6I5~Mq3jM!089Z`?Tdb4b9>X_9{-G_I8qP(}B*Tz(>+9nq+!7s{ zZi(}J`?~n}xVU*^Rv#nW628@0AS537$m!4H`-~yTeRR!xw&{9Qq&~HFPu9qhqogZx z7tvusNL=01hJUl)5_9QYq^aZ%OEc?7#lp68; z4Ki2D4N@0a4Ix-f(kRw)8V(1r-CdP!Qa?ZL!9$;Fy^Xq?y;g~a)2o-?*y2@Sv=U@; zQ#Q=(1+N+V#&vt9`^2fEW(O2(88i*;c=!bf`z zpAas320IoVv+kW2)jMdgeA6i-#4*HEZirhoc0m55)MURnZ`%WIC9{HppGuIl0W*@< z9Nf31|JVsxu|DzMcJ=NPW(0;#i_z)(CglvHo33lSbWaQKA06ENfNjt85&cGn^+ffS zYQEFkN}UiY8KOx;ff3I|gEQY;Gdd{wbd7>wU$|wv&lg@)GDd)i6y;c`)|dZA>(_HC zG`+Lt_@^fg>NTXRja#qKmu63YdGh3!rzb6)GSFeLjg6=MfGJCoo_cM{l-H_QNLg6a zEbrb!?X10BJZ8s)m4#@NSV(DTLe-FdLps_Gb@Z+r8Cn`5bz}pg20ZEA-F{HdkZ~aP z)T>hl#|*IdvA6T@)<1Uelvkgc{>o%|3i5{za}4Tb8}2-KL~vTd2 z6MWhGpBk<#n|IcAnS5a(i~C*r8haQvhJXAf+R!T^BDP8{Y8|CD_{*MN60zr@l8@WR zP{L)8CP~v%4Ew4++nt-c`?IQ9pX|xY+w;jRTv=51!lblylY%F{I4nh)p78K!-tNyT zD?izjo4e14a+q=*qOS-iujjdf_ ziwI|*Csul}A@aFSHEfgo)JhC_x3?JnW;nWWqmUt_Z!vT;IDk~NSSh@VMWqwp@#hXa z>M(8Q4j4f#bBB))dX6%6_)Sf20FR+gB%rNwKza`3`4RL zAJW`i$l$S224*tCXyQ^PG}=cH3nzB9diIFE%Y@>TrK!nFS7qkrW~Sxmi`8+1riz2) z4+f=0)J~mPJ95c{)U=Y~jI~h z3uwAL)gxv}Nm@qngw)gtN-;zP9nr>Io0%2Mb$VV0Md9cBY2?FMU-stoec9Q&%0_R` zH*6HTr{-t)rw%I12}-fq+Hj1J))8`-5mQw<)Sc7;v%7| z?psW+bfg>)Pnb+5Hnw1Qu%>WuT-Sb`oVpGfyiXni^i`#0opp01UEf}YNy4Q0bTM3p zuukm2#r04GcK1`R`66qiW!l8h#`kJ<2mIXnNGheCb!Y4sGdhR8BxiBo_2j z{lmghpvmGQ)?fDz^nEjSN5zpWH-CROcYl9qW0GN$@TO;_}(J zu{%h%u3`Y&nz+s@jz-Q-?P2TVo#pPwDQWK*X2oUqUD(G-_@;%L$jhTgc@zU1;U-?L zUJTENVvU4HJ6UH(W^q=2XMp_O5$1SjCX0}UesV`X$K;FRpr*^LQ?)!>Q3`Vc5%Ow6 z4^V;bSkq;3kdh>=fOpA_D&Grl8&dsuvsV_u>TxY}8>9B@Y$|qju(p3X-aETUfX@Rr$GaQo4VJr)Q>L+GJFJg|J&30P1|Cf?`zTk&jVm&bU6= z)}#ejJZsgZL>Z&f3(LCdkEjC`h$L4T2dTKh^558)mXV73%x=-r$OC7|9_=a2gokK( zRBu;Spo}G>%WSQlm7i4y5;VB+n9_a=nsMeJcC_${i?;Vx8wE^D)W#< zx6;r~y*hOo)H$J;s_&`bU`g7n)yDYCdr@0y7)>vtdC&nCnwq(-g`rKojDwFA#(`B` zNb6Yb(+BK#bsQW%AS`~E|NKeC2bay+@Z^?`{rvm-CVB=gDIK4_UaTCI>D;@YtxHdb z-d5dgb0YG`r^oen=xy!Xt6LwdZZ>HnLJDZv<0Q^v7j$#f(TT!5$k4-xr`Ngh8!r|f z@o?(Y%fjhNKj+c2nXf!iH^=ZwZ+phH!W;`L#FJDaS(x+1w>%Mk*dvdzET6gK_WSjQ#wd187yV+SeTH1B% zSV~M9{g&lstO#su^jn#gzG8TzVNl&+08!tB(4Tx&M!}J^>{MM*~M)olD@Lo>~j}pU_s1 zCtSL?bYFi`e%C&F!pMH%!trXMqws{IT_0g8=`JR5b^vsjJ~UxKC$N};qMQc{IE=9Y z=7YJ=4jU_WQ`kItvk-tyzbbLjXtq#WQzoy}(u={P5I~Tkdlebr+(ip+*?%!LG}sp} zesJ!@i4V@cIJ%@{^ympCV%4=XXRm$z{+VlQYM0imS-W(}YK)zxv}i}?G1T;ACbp*- zJc7-NGDCJiFQK423N{YP{SQWH)lhcP`}nc{{LSLpCp1tC+ZJw}3#1|-eGepRdP+3E z3wdE}n4gtRZ0v*Ydt0DlV*%U3`5|==zd-yLxzUT3HBTu;5RtBEGW2iL}Y#kU8kbNsL8>q1pd&5 zYUsHuZd^qI<=t;)JQTz(6z>_0nAg(Sl%grcSwG3C!b17Pr25fg4v19&6G!$R7Ut&_ z_tXh(ZMEFwZXGZqK4DrAS`V5$TF)^oAIJ!3Id-V^A!p%uwzDc85px0ZP3=C(qDF9f zS3YX7qUQPGLO4|4n!C+eB4ggE2jv|?u>23*PM+296LKC4Vm7iCMVkZV_5)w%N0e(k ztf+6bnc2B9o3q$+LSaNo>6DayDLr=k;5x$sCtdf_5n0QU&xlnalOsI?lEcE2dv$kX z&(vOX_6V;?nzLK&9-gSh-gKdevMXUU0F;%q$-LLnW5{XB8lsj=4$FV8a@5)^%fDHr zjffmIMBi&wM#kcVcXub1mL~gTc!=5}v8rFe_NNxVH+jUA$N`=ss?r0Z!e_?K-LoV; zI5@4}hgT2FBwL9=6D5Y8bGOCmtGuYY}3{W^XAh+?+Eu)_+vM z*!qd76DK7nPby7|EF2XTH440fCxm%s1X}8?OTt6O1_TrZhnHCEEdn!!hD`{rj`H@7 zjP&-7Vr5}NhJ*xrdW_&&neW~j|ERb|u>N5u3#iUIWQ0T% zcdF{XNG#eLIC@ z4$WFd^_NgwoPhe%y%sV%cf{;axq(el>MwI1)n8CbN)WHV4Z<(%OWk+8LsH~5f{H;I zldzlHh5-q~cE=_54)@)io+*8v^Y)8Ti{+mxN3O4*I13Sda994qHmX@^Af#7$ky)ot z@(XuNjN9!MKOi$*_+{dubt5a8-QuVh-_EJNUxAFfknuD!QkzlZf_E{uMT>-A#NX!| zbYxu=17)R(qK5evFIvRhgkR?Z-no4ozS-UAn#@4O0(DWHMmr{07*j( zpIxr00jX8e5#c*+05{gL{1U!19MT4?p)1lhOZ|ka+BXym#wdx+Lb%~AAzbTOTPxqM zr7i1d`2*oy76Iq4t48xM1&|8)yf}c#fWXY>kPP75T!tUT{v3j>VU9BeIOjPeM1_n($asoi0bZexK*ChWIOzF5Tw^0t zND(04b4a)fnGOkiaXJwyWCAFAb4a8K#GmJiQX%7!D}_U%AAyWiA#*^dAIFJNAtbXW zrySb?`9RcioH!MdhLYalkoXpyb7C;h1@|4-#whtL=+I_HVhf}})C!h>B&m=MatQlU z@CGDVg(Ly8kwf4<=X7=fa+*U@RY)-)D>($tHjcBR*+clAjRKu?6|$=Nq41q%1t1wJ zWOH+sa21iLdR-<#G)$h3Fo^ElTgtCJl&4FYXPYz>6OCS*)cl9=p6H{Ppe_4R7zo>wr19PP9Qre<~%$Y&4AzA51Er(qL`Q_}mk$VTIN!7&E^W2t zz^pK5hk-#)1`U|;Zii)M@;{!m4xSz!T^fiMlMkhx2NP~~`f;;UgqBbxFgx>E2sEkUi>EKrw z)MsEqa7dy2_%R(^O2VScqYu=EW~PL>jToraXIu76-L-J@y9GfdBV32go9p8rS(QBN z;Nq;%pe)x=_rM}mfLSQ3Ev+1=v=D%so80nk-!F?3pN{hkJG@Vp`*tZBIAigyihziS zfWW9I%Mr8FLzWjYjXcYxUv>Gcx&<-8Bch{2LSxZPFenO7g9myOPWJL;>U@?h&k+3O zL&A;@HMNE@=*~1UyaHA9hAJ@4sIvUYqea$qgiEQbyA1D)+4-!B@VK3Ee(mD z6`i~)um7a+5qTT428H(%wV9D&=}A^W6T{uS{imn<1VznGN?#l&vK~phmTY~mfNdNY z>EbgsJ0zIi4aO)h%o2UoeFGBO;*NEwmcJ+HhB}W(`k2EjKM# z00K*!UtrIR$IThBOWmUe)rBSW3K?1#o6I(3pL;!Gnfz@QdeWn6F8~F5x}xHB7k6(GcnZNdV*mhpbQ`@j^Yr zWe&husX}4_ImjWaRLD4?URcL-)vAzU(0Q9fR+~U@$0+EmQ6c%rb)G}kJ_1>%Ldu05 z!cmU1UWJg%Z*j_XCJ>xO0ObuTWF#m*%^@32a9-iLHnl)@g3h?8=a z8esv;5p%e|&7_Wvtwva2+HBY=mL{f{HXA_DY*1hG*$!5U(=K}59#CEj$QpK>#ffCh z?*(KfAh1Aig<*rnU8zAFnO`Qr;8Rj`rYY`93NUKLD;xtOsXj-04S9JB!S7|_6B>ln z<3$>1fM_)Tq1$9ay_LplmBu*G$mTS7>)@2GOz(``0OW4(BrV~lIQr252qkk=t6$56%D3-9yw9y?i?qmr_Mw79cCaXvg644rL!(ovmp;{EKvc zJg~#)#SxK<3&$)Bk61W5AtWd%enhC$RRt~@(F$zp%$FBpX!^>uq_h?3=_~MSW%}ZY z5lP`?W#LH?6PuhLn}mLZRf(QTVfmp+DN;3~$yom@N&hOH+c1BkiT|ynqO+OeKN$Xh zDE{KVP>dxhrsJ-5ig`-&v<6N7uL}QH>3^x5|EJBwtn$nPf1mLGi$?!K*XWi`&A;g0 zL8o);0P~l3+kf(|<|p25sziE5bECcvJEU$pcO_j^x^;k5+ccl2-_y@DlkO!I@bu{B zf9fY{e?op&9Ehb!!I#r_mnxg4BR{1_nMwCk(+^9(GJ~!gZZNrpe1hgV=~uA`---** zg+giJrw-mwac=j z!YvZbN|`EdYMsS?+IuMD&s@~Y+9*NBgP}UzPyBnF^iVj&@iCPkfBWxh+>~FT!`1y( z%1}E%Iw<(-cH$exQ5$|RsCZEwq$J%Ryr}SDpwE~sDpq8o_?2g5q$mCv=Bt04?DKAk z)`7zauKs(C^bS@#G!}N+Td1AR*nJv;wtn_5&I8$JZJ#i87(o zE$TI~A{r9G@$n6!u8r^&;z15a5wtj3E9e-kH{f~WJ5}E(xS<0hi`88Qdxo@xdkPNF z{ih7>QaT~ZeiUW^XE~xVmqlC$HZ@|hoP&#C=wr(pY}{OZu) z0C5V7@Wo96PCh|s0Ow&#?bABY=D9Nh{pcl4M0YFC4I$dXQM9)gvkc`zZTTgy#f!Xq z4oFY$F9{Fj{^{uhdb*D)9_4;Xs6}dn*Yt_*PL|e|^_HD2oO;Ejr@J*Mkz~t)X|quz3hVyc)qz2#dYK- z%{nQWECqKoFFzdmP%;{>3r=#`L2akT{~VM`mF$L_hU=_XeLXAW0h*wKQvEknEL{J% zc2K{-3>BP2hU%}Xx^edl5&bB*7(N?{gF+~-ySpEg63Q=eCHEwSJd`DY^xu0S-xhH^xRfVkzae@oQ&TE*I)_v+jy>5o!Bsd~7p0-eRPDD_kR-LyYqAx5LoDA2&_>U(nTo2w~pkHwCQ6db1 zzf7BrgS=?%e%MjNyvAdT>=D@ph2y0$JDWyvyL=_)Xnoj#zLDX*ge$WQ1AF;+I&{!? zw9=gsx7S-eVcEfcsJEkG;4I;aV_3Ay0ME(=^L*tiXzhOod3vD}5LSOQM!q6$Kf?z2&RbCFIlv`4%+WAlmhicw_fY!|mQPsG zs`&3{uz}JjB9G%%N<%?-sXMfZ_Xlv%zK@MIu&IO%;NrVPjW8iW;)~mxO1S(Eo<6;V z&u1A1^a_vctH{p=fB}|SP!DH70fZO}+#xyomGT#j3XO{RZ7GoWY(P^9sZNowQ61RyD-S++@}MrP zwS!P9#>h5<9OnQ%P=|XQ2de|dUiZ73DF>@h58nIypk>XUbKrB2_9m7*XwSBc=_vpW zR}=GqOp4*M7WMk`kDXk^LZOyw-h0tvFJ*N6S?bC)uFOg_`QX{In=tc1-xBY_p^t!t za`Qj-f6&hSv^NbHE-`sMRqEa{iYk+rfP}WQUe|{|n^YQRY_w-BUB4lcN8gK#6c{X6 zgo8Gs+z<;0o+9x)uXQvywi%(@G{1T{`b3MS`8TwHuDvaTQ_NJar#$t3g z62{G)MoppmTQ%4!1Wn#%TRIA7I&GWujR#ch7S{dM-lk|3x!NdQ?QM$CBqi*)NjeRM zXSMC(d<#wq(ozDlT=?K?zJrV3=B~@gSeKiSu=3l*i@#mjmH$fZhgH$8(_gcg=Y~+yO!OFS=nfW|s5XY!0z)fgAyZa*@OI zIkcT`G=I~LmR`Vsqp^j3V9&m(VDrRovzwk0c2&tM#V(aiGjSi(8qMFdE2IPHb=d2m zYs7J1816urEiRfpyJ`MxapB))i}PnUE#f~wwxoHV{y2ImtZ-XrR-%t<%EKw3PbmfF z_i-KVmD79VCE@|BjXfTa*;>>bt=pulr4-nex=jy0)_I*0_lXCP_72kit)$u8uy=Gr z@i;{jo!V<%zceE6mmL5MAUfe6m75Q~^U_);qw=8*{X5`|i| zN&!5`A=;>-9DI0+OY`7stze*I5Unx+)858XgjkIgQpCeIxo`&f(}hLQSQt3}#%1Rh zSP^gI)-~2pSU|djiC>9dHMs>OCkNnuN|%>9Uewh-D1S5AFCZla5O#vmD-p(n`lw92 zhS~P6JRI^C&7G^aH+M`MvbWNLv&9^>_GkLdQg451-z>LBPv@gLVTIT6L7d)Vvw>Sk zjh|>uKs%!(S6q#NJa@vV59`bJPsv+o z0~^ooHOE78sB5bJf5)h!*1z%GPIHWeggQC;F^s5NIaug)r%*PHrD&mR@^0!UUKMUN zd5WKkbX=}AUA+y^va|N4oFk{<>ZA+!jc`+3K$<3}u;b#V3S8TfpzK(a2Vt@2fyHEdIp{53dYcVWYyI5y%1>8I zfj-n!zNFXR!&2PhqzLR z3PRv&Dw_hur3ELNGxT5S=EFzQ8}2&eC5L{ve<2j@gFGZ^OoO@=Kv5~QHO?OeqNBrf zcMyNfm=<8P&rC7nFkI*{Kq)$JQ*Q32y!?&1xf}C`r4JpNK8*jgTf+iNSWxi|c1k`i zpDAI1P4XH;Q85cFHmqQ$a^Jb3v%*V`CX8U{u0#i|C%+7lI64T6tx!8sCHvfA1UJhRHS|Xdojd5|AXWL?#a2O z-vYNVU5SD5+Re=uheoiL=Rdo3!~1}UM5438nEClnovxd`Ty!#yn%QEHHtq#4zb27z@Nt;d#q_m`UK- zW6YL0+hT6Dl{^ezePy^R9Qd?B*k}0aDw{8t3pKcKw#HB`&oES@Gq&c#+|inoX+>9O zQFPXd?P(akViu*$T7G@vc=m*1E9{fNmraPfKvz82#m?ZROlHpEk&gBM2;ik|9gVibd zJ(UsNCITNJLsl@vFH9A5uOt5!Dsh96Du9?`h!uJpZYV?zKQoDWuoEPdoMhO@-jc(Y zonWuAlR_86FDStM=5O>JaZA&11i)o$#%pG3D$%ag$R)T&O*=f?x+Z-eH}=qRIvNHz z7P>h8#9|(o6sW_2Enc>59UPny4$(SV*iwWC5?x>vbR!)K-r4IiGuLGwKCJx4g_^8q z*1DXWby@gXU$6YePx(7sv^gYU@ZrI6gYo}ikN6?{FVVT)BYyDUcn|zk|EeDxZ@A2Q z%HQ>2S?jVn#mx0N*=w`1)@Ik&E5D5tv*~6~g09Qj!UiP_MqZD>34_FK^&WA92E}=J zBn%pq;Gz8e@EyV+KEOa2YELwZ^P+=xmi9WCmhi^F+rxrp2w6hb47sQ1fWHjS&0sgt zD9*D3z@rr%+BJ`(;n<%@`yPH|%%B~%cW={-mQ2brbCz(N6Q@NV(@HZ8&k@P*gyZ;& z*wFoLC0LH$M}V~!haRzt)6y7h@5=WfXrTBnPDP6z2LB?VL?|t4y78CiqaB3vV+>n` z(lH(W>vPi7C(3#8J3vveH_(i^PgQzDc-0W|P!>iAyV%!;cMapT4fXYnm+Q5I)UHI? z_~m$52lFh*YIll#gK(5kxvyaydslcB*p!DNe6X*jJH-6!=nIFsl*|A-|d$GJeZx3#oYS^8ZyQe`r zu)?rt2K-@CeSP`I@?FBXP+oq4S#Pgwz@hlv4Zz%!52?8b`o5S-`#^v{<1E7TZuk?P z@p3$S31#~ee)3v&lF?;_ZMZtO+Rk=;lnblW2M5TAwjr+a-H*%5KW5fDsv2Ii-|WD?XFdLs6JYUAh9vWK$`?`V%tH&{%YCfuJ+2m869Yj{I`>nY)nr<%6XPa6I} zmabTkU@vtfvM`b1ds3ETjCh>VP8EJI^k{l>s_@6OX$H%w!u^K}rDZ~x;hl$g&xY;@ zU_L|*k5SW)W{nUoM1ml8QuK7W53TUX_tav2^UehpCw1RFcu%{vu~h0}&`lTa8!V;^ zjd(nmOWLGRdy30Rp>f~pim3q1PvlA&?mK5YPO^qf*a%yw$>CMs{Y^S)*TZp~lSbdOH5*-Sj69ULm6NcA+Rek4D2s^9lnn#d zKUxLUGECEoN+b_CL7Zr46b?1Ti7z`qJJKr;W2Np~$WypGqFCD1A|%O2dK3?m9u0N} zaG_xg^CC4iZ52zKwz8rFw5)HJ4^nE8izJn1vzHock?E}ih%hFN95^8FrmUzwG)T!s zXWd$Tww8jLkGP%!pOXeb*dUMJFOL^CNV)7KIo{A^>sEP_bmL)f;W%y)2bJaq$pte- zOCvyODLb*}jbNG_j}o`i@c>;L5ShSxZJvHffmW2;xo&GLSElwk%XjAHwdZ~zEO!alj+)-8Dy z5Sr)Tg(!k{nRQ7;KNeAK0j^w%QX`iMy&-=ruloqM zx5yvgoICd>>TLR7i+5!14MP0r0uiBXMC8IpMC8U?$R1|cXLZ5C1GbYR$xg0Jb>Xag zbmbh_xS-l%YxRN!)qkW~oON(?boe8%d+*+ceN7gdIkarf!oN8>Se)ZY{tk{XSyko0 zWSz)2UR8$pF5}-JP$LM}%)&>AFdIzh8S!I<#b)6d!avQzM~fIG&Cppv(IIBw zs|AD)n1Mg@82J0rTyyx$N*rVs-dXsMS$KCb*(|&-yK5FcNI)>jqjItl!UD5!oVqj% zpKo5DYysQ_=IE~$WV7%cLZeyuVL=JcXjkue{_%oY{Eygfv+xGtvRU|5;i_5quOh6e zNA;r}F2cw(0}m2kHw(vQZ)V}46rE&-PMC1fEPRA$Z5AFOb~Ou+G~XVyQDQf<_|cD{ zGg4FnBiiX-8zX*XmQJj2-z+>%xMCI_F9w-~Cp-qeUMe+ysm)QtB=|IY(_$UZ}7ZJ1%~UraTB zHqOUt!y4ZaGPZ7Q{QR%xqdy(+&FJCZyu6}Gd7q?p5%)>`k5ieXTO#gUV)*6%P|(_` zQsUJ6+x}k{@-8L=bCvxc;g&hL-a){j&PSD{@BSEg55er@q1Ul5`Q}W!#;5PV_M3%! z3YHpwGkk{%q9zn@5K={Kf{%YZ#S@;g&-)7;51;qB5k>{Ln+ARDdrcAo++5>A4ltI5e z{3;%xF-QMb5mxh~d@+eC#z8we+91(vy=a5Q9%ku;szyV5I$^@+X7P++rxi)4Ez^3cWDw;@%bdmA1H2}ME`W9q4nW& z=7X;07aT3q;Y018tUTHa`)}b}2!-ueb;%>*TC6_CneI=;5e+o4J$S~vG|HUy;r}4X zUEv$E>iiE$_^A5z4uYq-YQ%J89^OMRJ3M1L`jk&Q?KFewXuDasCwewxR@e|$?+ zfGs@9q3E;(?X5zko_HGm`~MW-!o&7ZdX#63RV#@*Rt#7N)J_&J7o>HSKkNLta+ zKitPRA}?jpUMh|J340HbA9_7J8V?;*2!9?&ff%31fiqwUaF64i6bB7_7|L6${GKR& z#M2AG&ve&{XcDu}W`<4-O^*&vit%zCXzLp1zfpY>NdBZ((b_S&#qmLbaUp#qt&}=G z!cqLU=|e$cgIK9GXx#L0`$0Mbdy(sNFJ4)y)ZPRf>q?TE;Lyfiv4Ncdz5_n+s2_!5 zx7;Iu-4)t*Vao>j$GO@LboGi!3XV<>ov1Yw^oksx3if?MVuuICkIxyqrijye1+;GR zD;jA#=FxMky1%z0CB9PN7@?Ai3k-@c&K=xM65(Ra#tWq&F~a&IRhc~ zD=`XSxX`#;t!2TiT%)(R!OlO32Ceq#tJ#xn)3FPas*hIl>n*Np#xsi8@+P>{B{ zO<6141Rs19ILSf%mdnw~|0EoJ=Q#cdYEh7D6~{j*XGnC|*G*%DZ{YIM(STBpAFJRC z_6k1aN#Jnm=|pFyk&ZTj(=o#5aympmmg@zRGxV}rb5$+(G9?{y#d3O-PUY@n#_8w!YdYa=?IjT6EkA@aE1REoLZ3u@yqw?Kh7Jfs}82M^L zI9~-HaN^tiE&QA#Ii2?vI`YeeufQ$7fSd~auKYTu^8$Kqg$e!*4cYz}c z?eJI4-Bvi~qqa}BRe^(^$t(_f+c}?dg16ukAHccXO1>gvKW95sxp#AVMaakdc`ICz z8yhNNoF1$Y(7CQr@J|w)+AmY!cwe2|2HFkJ37C&fm3}ebnYAB z?dabSM5BJSzw>&1Rw zD)mPNA8^hWww$zv)A>~`S1rd!e~{L2d?UO@!EYJ=xD|?iqKtn9IEXlq=xA;$@-$E2 zbU2)kTMGPc^W;{z3H}Yu1qCi=hy#hQN(XSxHxKaZIPU4j9)dpRLycN*%*Sl{Y7JR^ z7>|tU%|D3Od3h+m&f@xKWFzJgRCiCMs7bmaCn@GKZRRq1dri#GF5mIr&9_#iluFY=PPso zk0zZ6CBUb-5VJBb^SBm1(VP#Zqs%guj<~c!sTAWB?nI4B$Ek5Rr_vQO(iQy@z*UWE z_OcU1M~;@_IQ2S}>v*EmjK#Xti)vQD9s{*ZJorjIXmdOG4a{5wCoT2jbQJs>Y&_?S z_Qmo!9PSQCe*@>Fnwv4334>dpM13dc%HjHP;7Ztfi@P%BRZ!o8d9am=8&M$*X&CWn zzK7$ixWYos7Ye)|2?TZTFt)j0^D~r(bEC2-HT4CT z;s#6PWkJbs=QB=a3)eaPBQ{K^fOMN6T`uRLz;ECpKa%Soa)!`}Mt_d)Zo%Pv*6pn5 z3}WIH6r{tFr1A*Mjr?`R(hO6$k-vm}c;h+?S{QG{H(h@boa0lAE2fr3`yjk!y`}JB zQbvDarH_g6D)wHbWYGBMoU3X-YLt}_G;Xr8gfxoAsV0Db3DgPSsBxUwevi0J zZoVa6()U5?1kUAOOmP(|aFR)Z|EgKVxpO$@uE1}wp$Z(Mn8^wjx){pKHksfp_~1@( zUUr_Z`DXK7p+fgH_|OWMTK#Lx3UYOA2fv|lRNyF0>xvkq)GXwDb~oQu`kJ^NbWm@K zt_VtfM}^+iDw;*VlnPZScq)BH!)cgc|E6HKqw#wyl(z_l-+i-`Z?Fr+G zm__ksJW2`|IIn*%jz1sm;|`b62p`Px`OLs)02koXtaww+H`Q$VYRx-s(`m*begfr3 z8LdcaWgbF5R-yMhw?JEPsVvjDY+MFDZx9WFE4=y0OERk?vO4B!Xe?a9=M#)sgs(~B z^U3b!`|23^FAg`3kr-1=JV%5dN@e3? z;260%ocOf*i-?ahi|)qZb)_9B^U!r(Y7XVSK!LXJGs-LkTwV|Dv5kEOluc+D`wXDG z&nT4J_Zfw-^>;&AX)SAG5}{1AupRT|+_TOvB@>TNps z7IMBLRrnN+?+y7caX5M%$9Lm!v?=Ka4j-<<6>kUl>I(GH8aF8taCgo7l4_kYOP-Hg zrLpb$?(yQd-g1CxF{L3pA<;N*95F z=ilW-qS5>z;6wGjTj1X|qqdE-O{n~xs3;swpq#b>B^sQg-r`%$E>7bFUai=P(L`^7 zb(1&?{zgc|b|UChL=Yvex;t_E)-8g{|Kd=@f!MZQgomT6E+=|=c-pSMJ#~!yud&I2 z0bOp%51v{GIXE2&9EmbYL-gr2drpbNFSLcD!u-Kk!FRX#h~^Q>C&3bhmuxTOT8WzB zhxlyqOpP2qB}Vs_w(`AMi{36P8ylAFa``rUx5J7}+5P*?$&McHKeRM`{_#=c zPt2HobhvwLX>JEO^wupZl90f5;%Z`ENFf12sD(tNS{DdGE?+#%;^GINv=_x`da<27 z$4tGw+Sb#f=k43_zZO2l^tZYM1SX6BYmAE4Qaob;&z!+1gms&U zpm2=cpF;06k1It&(<8eBYp|M^6}Vz|Y{nWaubGMjJU&~iBMX#)!ndzU`i3>Op67tL zu+s&Nhx7%zt?$F@mM%L&aJ846$LR<1PutEQslWUMTXyCQ5R0Xsg&%OZm;%eO<>H6> zcM);RG3v=~-I^DFKJwGoXLOq9pSP6nu@08$R3J^UZjA5)Tzz^AxMJY^6j@(XbgiDk zp82_ima3)}+!xg?FUDO;CNta9l$#1JvTE+U^l#1?N zOX1KtNt;yJ;#9*GC!%q0!2-E^8wwLqu=_*-3VThrOLVrF1_*pTB$?(lZGX{O_t2E4 zyHPUS81*D*YF}BiW&x1|(q~$o7=T(*k4BomKB}c?kGn?7&`e}?ZM!K;RKzy&KqG@m z5233K+fBE9Ejy8usSO;y35~u5eiHa5+rEIKRRbSzQ%64FV~ucA2Rq=6M!2a18gRuR zC}f`ht!6x+sb~Y5y^wBZ^H|Jtre*<7j=AjZK475D z)0f6&8@A^}`*=O^*_{$R(YMa;gCFh;ya5LpPPO#W#P!Ot>6TTRvDxd zO$KpdziQ7a6^5x~f&zaMO{Bv_ELH_g75H!0o`gQY-y3%?{zVP|rO1~lM@cUOd2bw4Z_xoq@|rISBgwfM^=(ko)-(uCB? z@c1R!vu97;k(yCADsxqq_8sk#e{`MpPIcktqRt)r9G_qI^7P^p6|p6L!BgWBXM`S% zo|Bq5YlPo~k+V(|Lsas={~J2dNu7S9hDs~0`4F-n35nfmoX>FfBLxie?Sl7JaO4>m z7P*9oIB-G7EhORY((j%dEvF?SpX6sxfp?J{qt!tctmFOgj!VnA zNxV<%gUY}Z)vX(t(wk#_-HP7RPk*6w-b>@=URztves21^)X#A5pkZy+?rSyA$c=j< zinnG~UU+(K!^|D*e#axL3^%Oxd*!>&FZ=FL!NTuQa?Xu*_ADK#l~}D|cIa`q3LH{6 zC{noU=(HT|*>}Me?hwL`gfJfSB>DVcr#vG+e5U5w?yPl&1BQFjfTnBgr>E9_J#XC0 zbIM+rrjOYC{epr+-z|Hdb>4$p$a=3j(owe9F|%Rq(-$f;w-%!TE|L?ZI;gw{UF~A) z+`;iW!0YAWfCDBm7omM2GTTA5!Hca<&qant&ni_I>Hhtt+5< zA6M!7!Q0-Sw&B*kj4wvLu{wW?$Hj{Sw`Z?@bIcdn^}lQ>dwUaG`RT9BNo0mIkz>8b zzQTlOs-^k$KhN_)#I4ZL@G1MxFyE>lkItNVfQc`S_AOd+zgD*|_at3UO3T)BG>a(P ziXBREzj_zc5%pO*rzO&}7q;%WCbVUYa>nr`x{ZQXPR^h0YPy)mHf_H#r~1YY!}YgV z(fVf-Ysboy#Fw+#`-YQ~KUlu{;&dFRyR>(CW_1KLH;f9;KpmJ<_^1GC>LZ5Avll+f z8wPawk+5>h-PKo))vZ68p4e%ib&m+=6|Zglt?Sfl2gUnM7WF%K)(g)TB?;0Y3#(U; z7_tw1jT_={L*_Q^7%Uh?+@ayDSl#YM+4XvKD`kX*?6?pTr;6;Re)Sdm`Ox>3DO<_{ z

    s4ToVzzbdvnbS5gPVjeiI(+NtNy7mq)^rsP1mqy3Sn%D9xu@W#K(+eEpkqZQcH zqy6lHst|2Y4Vr=jU2OdzKGm}FMk7_Sud83TcCZjZ`^m-Ax`K?CHkLh;chN6v$WwDX z4O0tP)~g#Q@6G-c50H(SIo$RO>0zS>yDx0{WkuG9$E5BL|2cEM=Y)OP1js&rR0YIL zosD{%0rxy^goIcqJ3W-j`&Snn2v!DLy$z`sLm3?%lUUr<($lQxsXWiZRrxbl#yl~6 zf5i*Ta>EAsfxJiCaKBbx{$TdsoDToQte5y#JU7iJd3_OEF~om`{Nv}|r{!lcywHRq zkCAqOrwu>3PH)s=)YPHZ@-1mF>G z^wPA#y!Tep7c(hcisU+g`$Xmm-p1KOylquH^c^>+ToiDA=-ec5^Nn2Ih_#9JDCP|kQo z+fu+2Wm2U)+PhxpVb~~-zaw-%D1Cni_mY195Yc}$Dk0AaTO7PR#oMczX}HsE+M%eCOV~y9kH{ z5JW`!(xff0NUuwk-aAMKX#&zkR76EVEWwVbSYnBaMola)y&7X;G|^~!G3A+NqVApj zow;`xvAy@cpU?mE&wIFgXXnnGnK^U%nep=Ky?fTQ`B0TxbaQaR0-xw++IHWp4cC_T zeYBup_uIvDcB1KDZ81M+Si0?l&aRF_>A{_u)tfS5dMbg}Lbfh^n;@9Z^$hzDGNQ$> z5!GUdYPrW>EHsV!0zLmL+hT}cM(;Q*X8%C5^0sAXZ_6v)l9IDE*DpQYFEHI!PPx)U z_dh@x^i@TLoW6eFC~I3@?$(^b?WuWNv(LoG#KV7XVU{pj_5XX>o=orTpdT078cGoPZJSqckS7|!SqW_%H6 z^ryfO2TpCK5&|V*NeE@+Zx+{=!V#*jL=2$FZA(1aRh_#)jZ}2WbtfM?K1nyYv$FS- zmgXz_%GSUwx)mk8Vd1L{!*gaGVseg~WQFx&_UJLJShKVCd_%)a^Gi>+wVkUVDR8bjV z66)H&L^9JZd#+1ZsI$2j6W3|%ZR!|V02iiMnOH4a=Ne|>H@~taDs3g5)ajEoclEhC z_mu;ICG7{0y`e?*i52P9QitqLOMBlP8={w#SWdS+&|VU_c9pF5T<_dWSh^f6r|$v4 zgu0Q7LwR0G;-2|cv{0`W+4>=QtKgi?=GPiVu9lTw+uij_75$oMQDs%LlVdafxh^qp zPw|Gk1L*d^-y6%0*XFFKJ(P*3WgM!3iIL+TMi2TiC7cie*MI{B3CmXX!yS+h*UMYL zp?tJzpL3U)8}NIEUctJd)7qi&X2b6;_;l0e>-80HteB#$ZJ(8yXc6MoR+!)FJzuVC zV^p%lANTB>5)?UO1=8KGH^TOKB zDPU!SAf{hoK_mhaKE#nxD}A9#jTmDDB60*Ow=A%i3Bi+qaL0-LD{5sz*6d09knzMBQfH&tnnuQfhTcm)*)nO# z<^{!7&*#tEky+G!Fx`6Y9OvXk;baY;$M@DFuc+qG=^O1+mdEHYO%rti{C3>Lv=Q6B zFhU@}xPKf8o8WC^{v&IkamiBECOoHL7Em(2aP0-u61bsI?#9 zY}>d6QE^QnK1tR;H~-?2?(OIlU{^`F1I3HJGA{&v8-fW zRzZn1lhcWHx=`eU*X6MmDUlhw%M#Y*1(wA7>Q9w;&Grn~9l0b$UJ~i9r|%gE(*UMc z;bLH{P9PSLY2amQUPza~@la~ArQVKuuw7MR6*akOPx5G18oJQ1Dn9Yd;QGp!7B1LX zG<}yrMP=c#jDr1x1J!RXN#4`Ml8# z8b1AulGHq(;goT(x+Xb)Ptk_I);g{IYr_cpDJ6j+sQaLS4_NZUfHnP(q8%+~TW&n7 zGb9}z{_~mrfg}yymv6w8%lGEy$Z#uZ!l(ojYj@sCK&Ft^bx+JrbieGLkdxqEaemRd zw@W^tqpD@4HFoR)Wabj&;Tc_%2$$qt^^23a2INQOtxqi7lNp&}nMkK@+5{7-8XVXI zz?&!#Yvl{(v)>am2O#bNGVziuI9&)5t09o^S5)&fjcBdZ6c=Z1GWRaJP$hg-qph;K z)!Dg~xlG?0+4%QBc0b{*)mba*4y9#h9jZZp;ECuD(u->OW^$~Yn129G$M96m#`WRQ z9_}oX5EqvxnD|s#&6JCj#mI)g*Sf%>l>iH0UVAXj#d%Rv9Q!#XVLyd_y5UrjD?pkJ zh*%|a=XWM&YTy|K42+S1z^6Z1Amo4j{NOLGg?WJm9v931we-a+r||K&Dry_>V!Zs# z->X9}ZnA6NnVX*HnB-NtJ@v({D}Q?a?Na-5Yp?=elJ0bDJJMXxGmdRqb8HhPSGj=A z0*u-YYJwrq5OONiM8GJrDS(a;F&K_;N*JYE*kAr0LVa4eET5XjI&)rHNA&}AF7`TUmESwn3tC|uTxC~sqAf_Q)`Y!8?4jGup zQ|C-|oDyjp)1GgRDoo^Qo>Ltshs}#6?V;$ZXY^ z=uwxc2#<-<(Pp-Yb>IOVta!aJ@4ZsC3_u#iA+YBLUlO)^~%P6#N+Vlmd%556< zWzg^47ch4_5LynfRDlIx2Ea0dWY{Bpm2^&|An*L$t$i=QP`xF`KJF0GfBki&cPP$2XLI$DvuhA)N{q|0&i#_99BKV% zNy@JB4aZ+T^DD+#n@(mle|@5SWTgDW*UcFxH)SE}*E27_Fj&4TWywdaz#h0A4e$Y% z)K%LVpQ@a~9?W@F6gWS$Y@v%=b6LYejaP*g|5`SDt)l9aEgi4dqs%#(?zIOqJPMB2 z4b-+@s2pSmJz$3ylluk-aXqsPkYFl^K_DIKsyDNWb~sXC?jq6-Y(c(r9ZDA#ZLsC5 zEnB~AZvJv>{h6hqm#iYIBOd^|n`TY-m6zBXi49ifg{F&+$5j|W_Wrh*D^LPjwoZAoR@a1>$Q6szQN!9)n z+)(J+Iz5lW!i4(*(FH~nP#bBy_*$1Hyl)N2)hyD3LFog65zz-IE;+`Yh_)UoQ4?fM z3;Gc$S4@uV212Hm$f-N5$N9@rznOBGct&GL(LyOz`yP7jfN7xF1U(O<#Ch+%_u7G3 zfu zhf0tjNDg$AU_EEZgIinOQTJ#VbqFI9^QJki2QyO7Z4TSeKKCuuk-M;>x+)=8AAg9h z>QKo?e=xA5=(Gb(WzQGc#I+_&j?HS!%&GULypA4)o;^UkvE+#6bQtI}jFA>*uk`-< zS^)}NdNMcnN` z*#cpvc_9+;mt&E@wH~v8@kW7X8o;Dr0o!_i?ZCaQhFic8c5l>h^Gh#nHXJ;Au0MS+ zJ2$K5c=>aG4@lPjv$6a{Rc2n+U^-n>@cO#V?-drlyJ^Gg^S1V{h5y#|ui0x?Ra8-L zv6(42v3TS0n3&@myH4aYn=R@q3ajklBtqDJqO08gHwOyfC4g9a#ET7}I|FMa;{c-| zhl^tHY>=(QoleHZ;&238Uxy}x9J7nN^V+=VYQ^g-ayOI*YELdbxuoZnvdhS$sc_NC zX)~90&1-aHT`!$yP0C*FS#qLOM=PLWLtfA8<(J$W3l}dn)?2x#u*rjR)7!0F8`1kt zm+!$pyH+bN=v5ytKhS=??Lg_Vn+(41NSW##RIcv&^Pq3nJG~KWm8bNoj+Gv0yWaME z`EfEdoF}>p1XKtnUK=<=GYW<_MJ9zayVXR_;2EHktBQamJbMOc*_4C2iroJMy>RG1 z4j(M3vp>9a|1y2y*<+}F_V{5V^nxt*J`9hz-vzp3auHx07tzISXMF=;>%Em1+Rt8( zJ+{u%aY1%YRY^_5rAyS=l^2SxU95SbaPf+oy1~Ad``8zxOCspZ0v%Jtd;`)Y!>K6w zoQR-0I?#&H2ckkrEeLT1<}A!t{6K56236G073m7in9-4u5!zniw1IwmWCYc724W3D zeCS5uF<1*tpnduPvMJDLU^qxuDu&<=FdWeEm3$Lh&=UHorHEdcqq#c4Cs&;LJe8ci z{rRlCK9sydWAy38^snf(71*DO#6I2RAM_e~l6yd}F2DZ?y08rUQsFqDhx~&su&1~O z^y)I0TL70j57r7orUho4+xekcL`LRd$A_;ToE0)lOIK=?lKTa_m)U#toQ7zGVG477 zHgAThl!Il^9Hvgh)ZB!@3m~MKG2y0?k4Aw;h^L4sSB#0m4z-`8Y)%MWhlhvAlyTF> z+k26Sx4Qlh_Ffycz4vINSJ=Rpo5IWYHw6%TZ=Fx%Blh06O6Kf9Mt{5+J?Qb=oA(EU zIu3)qH>+xpu&4rX^j!eD%|N&d;TgKR5~Qo<4`4VjATmFPZeVeQD~hCWISjh}ZRRLm zdBHy|%|8$w<n;B4b&;~ZevyfSbq6dDG140vVEvkPsg zEc2WZAM3GaqJxu1PSN6w$aT#T5iRQ?G8Px*csMys+~W}!H^Z~6vLV#2FoRsi4%dOv zp8$MXL_D_w+pXFy1p+B3v*54+3$I##fYar{7&W`~T;;_%`M&=7G6x4^56kJ!E#3v` z(<{#p9?W@nH?@b&1RvDFgU%fbP-}>{rpAs*Gqcu2<0ZS_&4CV2fk!9o%1ASROG>f60OXtiAw1cX)maIIRKmUDv<3(y85)XQ}7NpdN&};5r7GfL` z)Bf7Z@KyKsIB)%OXQOmPvh2!`bIIYdm`;9^F&2Pms+$O;AeaHg<@6qM@~$;n^#vMu z_2bEs*)y!HeT?2Wq{4`5Mddqp3>U7u63y;FW*asD#uBP!z)2coTZ>R8vbE@5c8mQe z-8@RI1#0v@QOwAn%KEYjz5xva&!{e#gdtu)rb}KT#vrgf5xPVGGNJXSY)3juL(3Tk zS|*y)W>^|3(9{!%qL7|}otxF4+yAz5vo|nAG=%`kH+mIkrO&oc%_;Gw$^qQ)Id%1o zKI#2^>4e0j^{q{p)-^SIv!?(E5N7BR00Ml@TqFlXm7t}#pB}&i^x2&|6~y&g1-fu3 zgcT8M*?6U%3wlu5ZF%91AOpHkccidte`e#3IBGtQNk~i2TbM0RGjEh^MLqKl)tD|h zm0x0$lFB(*K(B_Be<08XDUpyQp&-Ep_@_X438t6RcTvHx@X~!} z;U%aFAx@CpbqS(*7Q6*EhzSb>*NBf@Ir^|1*aw&Oh4hGx0x`+yIp@!%HC7Q)zrtXEp{$b3ZtEEYPrO;mx)q%d_h6T(`H7s(O zI#Gg%LT(Xc=xQ7v>Fiq`swesK^lA?JrfdTJNvdA)1t15kG~*AP*RLgD}(ty1MT zi0l&RU+e+KQd(e3VibsSXa$$N0SW*QnhK&g&?)fwLYxta{ejbEO(=psLLY@#_{Z7F zbE~@}mR;-UxV9{!yL#^Av-%Z7Z`Pj2R}O&N?tJx|L-X&)pQFP^Mo!Vu4Y9V%@x2S@ z@%`mCaShRQ-DpS9X_Wf&FDUg)VEYM_tMc>1%#r+VfQSzpfZj+nh6T7=Il_$y;ihMu zVLzZ|-k>bs!mrJoE3@>7HNx-9DVB0#KO}!dyrJt8{yKNzUbLbY-Gsw}`7@uRxTeJ* zAChfM8h@e#P~r4Z-8rV{G<%UUxla0ceMYP&-v>4oTJW-yB71yS(f7|k>`F-Zg@^=7 zlWt64(@6*mAe)SfIDsxmFLzi$J2WT(2s?Xt4B`|$v(=@}U40fMfj_=}oP7b(7vPNJtMwow1t zy&H%=Y!i)DDnN)y)f%`F@di8_7yyBw;-f%OfALU#Sa}wpGF(2uqI0w9qm?i3Wq-Vh z{$)%+yT;yrI%s6RsFfolY~xdbqy-%~fc{e40Ftn~zz=qUIxOd*o?)g7!6}@|1gJ?| zOK=q-V5A3$_TcOa7P=&c8_>l4G_qg)SzFt!jqCch^>nYsS5~gOi}c$*>y<_ZdPW&1 z2Dita-e#BD60mD?-~7Wz*5lvG@SnoVIR_i+59f8PDz03_{`tngmY*;7cX11IoY-xg zxg@o#GpH%PtA~02;6ZJON1dY06ChCvM z2uXD0Nf*T>ERp_2no!T0fga#}AI6HSza3xTf;X%g5Obm2mw{9pT45Ex<|BYPd`BdIQgH$ae~4EQ>j@4#a%A>tW2%;-F5r^49JZd`Ttm71EF zO8uR$UuCaTf!FZ^J1^H+wkA?<2Yiousk@sEr2|*fCipdZURHpAR=ynhC~o&EPN5d! zL-tP3^vB)MKLnZzoxn_QNkU7lIO(4!tdGRK0$PI{phF44en(wwzPcxOH=22}^1`NK z-E+FNJ2N}pX=r?>J!_~|H!8Aw6wT2%xgzJda_AG}w%4bAq#~omf8lo{d%tb;Z%)e^ zZVg=lV>k_q@>D(mz9o|kcYh~KXZ%K$IES;{5-B)^rPTB-##p&7gzsRzF!1JRp9=R1tt z5!eDSNp!DxTiR%4R(8c`MP^p{Xxf(I`CAf^N8(n!-2GLQf)(b~*5*+U#y**<{ka7P z%F7Or>#>b8U(IL`1vtF@w_9(q%`?sx& zzS*?!LbJ{Z?cn**`Rfzn*Uc+jGwp=Vg0ocT@G!c`DFrJxAZG9LNVq<2sM0mvv-ot! zqI0FCnZ-MaN(`nqQj78~ofg$@n-i^;(!6u*ER1#2u+a$S%_ zlbJ;3;JYEzo>2X#b|>`ax;uS;$H_$W&o*_Rav%0NbaHDszQ*@SlpeHn%NE==uJ^Fe zhrQ=P?~c&>c#cR2IM8~1650wDJz$7rkWVn5yx%@Xp&$J=8q$)0UrJ~VMVZH6#TA!P zvA)j~l=CT`hn=W9!+*a0_MgMU^P63qo2!PYK=wR(mD<9tu(YFU*&_50+%h00Qy`84 zAPhMtrX>hHyp0#Ma$0KT@ia)D&JNpSo0n7=BHv%H9Ot2G3t{JvhS53Z09xf`RHW}S z8Rc99(447%_csj}yX!IZx1Wx|$4%Giks%!TR7fd7287e}gLz0M2mpLk+e60G4b>Nj zsZi-19II4W+<6u$g)|*?UK4twW>l2>xH=4VCC(UQC) z&5`N(McIza_*^5LBkxL2?~*4*&rKXkoI5vRD6%Izr@`GhuroKeH#*qQFL+3NN}Y>D zc0SG#Zh=$&M9kaV&L#-mLw2?h&{-f7-Tr_ug&Y{xAX5(X;U_2s<|g-7)LI5 z9jwEo9A$Flz_f_=3Q~IwDcABvoA}nDJ~?wh#8f}>Y)zh_UcR74L9_%tfGUBs3#oEk zyHqY$AMEEMp6>j}mG`03Tsbf^W<4N?wUFWaqd~HVb{+}rQ9p|)bzZXOs)u>Wq?FG~ z2A1?C_coW8%$1W-i!@h_%S(onF8I7;(NFcHCqPl4prj2#L7^_!0++2!#=)QHL9&&h zE>}wGa;2&qX3bR(bC}6*d=9hbs>|xu5m!pa2dzKMVHSZ+^3fb-=(`_kQT$m8Fe}_8 zv^>icBPn<(4J?37$vBn?(IlLFF0Q5P6ghCi6wbZ8$|Ke;$39H9<|58UCqCTmKG!za zR_-x^?sj)$O=`(Xs_4_Cfsl}aq)*v(%h;6<5>^KWu11FI3>P;4Y*oy^w*)|yQD2^fNT~Ki{;hqQ_7Q`SaL4-_!p1# zqH<4k_8z-b6YHI$X{vL>kL7Hn|%kzl%sI!?s^4D zpynhTkPta$$D4y-e?L9TGKDwLEvg#IRj|N@dk2fm#HL_T>3ey3@0U=|Js5`OxD0$)e)W{VwdpV=svuVo%p{!E zkPCjLW+x0Zr5GI4c8$pKtm&7_`)g!bVXoST3{n>M#}9tn+4-%ZePTnD%j{^GN&VTX znsW^^-NWr&A{NBj_2UOe4q`<=K{zm`0|f$GC|Cl3;1b>htVvHqk0Bn2h`eshX&`Fd zm@4p?JJ$pLq+DNv@%;-^3=V0#hGofkXgqTwT(u8QOKloJ2f2z+8T<4j=HL+|?6-?u z5aD7U?mn~OTus&4dJ|dnY?r8pM0)^9Jqb?%39#fpO^K-rB-{TwHE8b`NP+)#mOw58 z`u!$Q0@yaXW7}psZOg8r8rW5k&?4B0lXrr+0WuYi;UZduPh-o z$l3QrS5*4m;I6!a=aQ11E9lz@xZ4fT_QQ6T0;~qahG#hzz{J_mEynEZgp2bAH&sx* zAH^)6U>+W3{v!TqOZv9rEoZ7LrK9RX6y9`}j-DvII6DMo*DRxh( z-JbO7aaKO?ens8Il}L`yEZfxAN4;E{6z) zXc9A%Dg#AgJ=IFe$CN(;CAaZJZuhN6-wBm+4?C41`Yq8J8}thJmnx)Rw>{(+AK%IPImzi=+i z3~6dW@59)dse|69LEw-dES&|2q!2#^IA)+PBEj!85yPNlEETC___8uzyn4ZjE{XO`{RN+G-uziZ#akx}+Pu8l3JTi1z1y(0 z)XQAQpzZ?FKb1kJ@4UVuT5qy>Km$6GyDe*`^K3)Cy%@s3NAy>H(KV8}AyvQTSZV38 z8vWD_BYzBmE_j-T!!-{!ZtI5mI8 z-r6&=_U@o$@Q*bkCHv<|@()#@&vx(Tr3)eAbh9MF3N-&f-5lwMYz_K5uDn)N4h9bu zOW<|9PE~m?Uk;~{%;4&S>;^K&ABIp{`6iUwit|9uA*EuiDEtFgo0n-oZscSd((5u+ zE94v*cb#-PU=vZJPSCzvlJEWZa z&O~!_1@DO$j{Po%TZ!OzXx|Oq1Dh+Dkigl(Ka2h*^~f`MU&yZk)SCb&>8Q$~eNs+- z7eB9lPb>I|-=QA)oz$D~;TRkwUr7Z&T-3nBF$drnvyKBC@GSbCYmZpV;eBBP_db+Q zm^fAr?UQm2jxFl;xzu|Rn$A6|g&noE9_E2N(^y3EqWSsd&d3!ZOYRwT4OAhb;e?7~ ze-e!Cy-RAf!K|0?rO=zTW-kREDC!uxL;8kn&)>Osp_VZ98MP#QE$)<{FZj3Uuhege z<~;V67|sNTwxMS`5Wak2i;L*XoV*Ut!dOz9Jaesq>5srbf5z1Z-Anu(|2vcejLahP zsoBdc=AT8JmZaG$f*lO)L;VS#{MY*MKB-UI(*P}s^rw+O2K~e*pueSp-W{Ny!a;wF zgC2A*;T5j8TYPKq8`m0>$F&B(i(h}@cLF{9PI?z{226lPC5pmoB@pbTzTn5%Wx4rvFQpCur^cA^RK16|w0`uJWA=irs-bXG6+CT3fIM(&U-~j-`BmSV`Be=>2 z?CXHUy5JQh_h^#fVq^e`b;Q1fEI@bBqr#;-*fq5C2ecbIu*-!?py#knN2!Cf=>Eq- z2gM(3=21A-5($jZZ;T%uNMw9uL7YS&h;te6CKThh+b}3?L1cVGlDGa|)57kN{DQs9 z3(;@pA#*G&y+TZbnsQ|xxy{f@I9`ozie889DlmYXL1l3FV)%_4OVL#P>mN6Sou75# zU+>+IS+>;-g8lm=)YIPU!4^dZzxxzDdn##tp%! z4}VjDe?Y6ka4k*orS)T_y>cj}**7S!mdEO%nmqg^S|5Td7f_z;TlHuie5LIl7J9L7 z@P#<>Vc}R!z=9n1b6T$40c!y=G@>C-QgB-R1o1GD@O@SX`LlY6u&Q}Cn_fRc=>vJ`_A!2M}54! z{QSJUeAJ)l2t1Jzznk5(agX9V+NekI4T^IK4t8-248RReK|xMV!NK$O$o9`_Cw>JlFRF{0W{J)s+wzl3x%UH+Kn17pqft(*4vM zkR1oOp!gG@@q;F%qAAeiELb=XSS+!%rG+^OjPbD41x@0SPF&wWHt+l9l;E>>?%)rK zs-oXQM(;kQW6@+3w0mn%PiZ^y6r-ID64x6#b{*shp3hiIpqu$Y9! z(Xe591P%0c`mHf~AW-jjpk=U$0LiW^4$+A(u=|UtRD63KeH1>6(UpFjK;;#)JE;TY zXT`rusN7;WH%Wv|sSm`*!0L4v_=;F~7&&lQZ9~~oC*g`Hk?`ms!SY9*1(XuwJbEX* z!S6;AsRw`bPDM7oOOcJ0J9Z$nV+Z`2Cb&X}K@bGl>CiI79zyt$uI|H}X0V*8(q_5X zH;F{bXV{9+m67-GtwmjX92enRsMbHjMXaNpv&m)JtjSZS8B8`eoxL?(Yof>{lip70 zaB(HS<6mjh2e$U{b0(SE8eebqjCR%4H#E|nnVq=s%5;0HDRbiOVF~Ji6>k;#k&r_~ zL#7^t!dS_K+@BPlfdwFoaQyknBVXcMawK^R--<%!rw@LEtfOfgJb1A4AijXszy~Vd z*L48Y&GG{sFhgI^>(Oz-VTGS2K+BN$pM#Dn&%qoBge2WVk7 zuE@xt3lh+&EL4_kPLDSuCdL5dTGk<>5hZIb#TSt)(Oa&K;EU%dvpu;x3vCK_=8mlX zd96+VFRRhJ;OYj-N`=wBLQ@rGC5L|z;;M@_$qyaHpZ4LaN1qR;!q7DIDl4Hr#lHA9 zyp%^J(mp~d|B@~&r4XF(LSY_tgNx?0M26@!dQBU$J$?vxBD>az=O2FwSy@56s=Jky zKJ9~VV=w9kYl+@K)95v!hmPZ`IO6E@5v@;vM-%@JxN~T#_YjbL32TOITf?6R_+CT$ z9DF}vA9!n6@H{98GkIhq^FnytFxr8?-h$Tj^?i-%FvtXu!6(FZy_}lm;ri;yJ$zikOO66m{eo}o@S92 zm_Xn;fT~7eU>g)`f*QcS1HBL6TzZ`zpMk#0dGkiuXAzs5_6psjG5~fj4Z}JCw2V1kmysFhN8`R7G78Bqs_YuOEVX zXdYS@99$O~S{o8l8|oSw>gpOQXYcry`1_Uk`j+_lm-yBNIXMOeIXaPq$Gd{NR1{?l zboxg@o`R^Smzg{CN^ll|m&J!n_}H`NpEhJa7UOU@yfKS}ya)}rh$4=V>q|I*I7&{v zu-to|uQbGcj;*c73a=tRyFf30iLKXe?Wk(s8CEm=XN#ih0^kdgN1~Z_bVH<~fFxK% z0*D#?+;gZOeprIr|Lp5~ycRK+TR|-pxCPfdRu5aC8%lzTfN-QW3w3MoLpN^W8yjDV zGzJ7i-Pl+)8Ys^k+ETfVYfB4;4#+~}XfybuHsK=OP1M0H_{ENn@94(*IVEZRYB=Fw3Y?5%HUlwN%M{UlgO2 zg+;iv5Op)li_xk=T#NcjajgI;|3vesa9T{nR9f;d=N$w=Yl#MfY9>x|_0%-iG&2q~ zk4Sf$N=deaCRBTBQrjnLhg1<%EQZKN7vUMsqboq13PnshkT=T32N5B28ju%vDb*-W zi_D6PD6|09k(0ZtxL{`<{{7S`X02arO?LMDO(Av21CUH1PQx5L z8YBmZ6Ij6^06s?xcP^TnGw+bxFR4CNe(BtX=W^Qo8b8M0j4b!%8glfbBLNoc839T3dqH<4jJEsE%p`CZFVwN^5(gzW$B2)3+v`)vhm%s0|FN zqpjG3PN%6p_5qc|9;D{2Yq-?2apec~r(a%Jke{^aUuM_Aqo6-IQ2!8U(C18NZA>!H+ZMx$`-$#DRnDZS5W1I#7F2 zUaMGhxmZi5M(Sf{KRa1xa%yC9QP%>YBbVd8vYe+yfZk*qSBJhg=zKJk(?*VY90XUO~QwOj; zg0E!WARwRpAfry!v8<|z>7=9{ww}huGCL;^QKzt*8ID;Lp;xoJDSnpE>?lV)r@0PJ zrqQ(mT5$ndKQDd@z|`dyhX5)jR$XoeAoGfX86YPMs+I`P>8{^j+}2WBSw2A|&9Rrt zCuzz(y#xG(M$}=&?a4z;+U?uAT0-M%y{mkDYPx`kcq;1S-*v?D{7JV^)em54yGgKKP zDo6n=dJlTP$YBw&bP#bAv};`h_K0=S`q!IY@2@atj=h6&2tMOskO~zTW{80o2;&B+ zkk+-3h(WPqcjn~nJaBk>QSr7z`$eA}IDmh}w+|fxK^_jIV*}XFaE-DXrj>w6#wb&d zk(*2QwWUSxPBMi-dZ4TVo#{#e(A5#DeS$b_i>ks zromvr8>V{XLi5?GI#iq&TqNs6?)DZ^G^Yo(7X;?JS0f*LCmUf^VS!7!4-gbL)fY!t zcO7O}nFxz82`G0MbnqDYk~}0_>J(i--~WNo6xlcs($-O!+tvb9&EC#FT~mKrQ3N4t zs4}sUP@&dFoPn%e{e3;d0-Zvqhs~%Eo-A;1QW3m(M~>D>Z6e$v2(811LYrg8D3mlL zJ5et>VV0%bCRo65!yvdshT$d%`acYWcSb#xB#n(BuazHzUAorT7%CbDf*WJT5v>0V zI}o$vMe--88CigRN&{{e7zvyKZi0xC$5!D#F%cFoTmN-)ih8>9S~av*@SMcnj&PFn zaE_BSqkQ&S-JfH#`}M3~vvKUiGFZh<4g+z0Rs1gTaB>H%aVjstI5j5~dvFAc=Vbt= zH5*``zO5jdvIiU1uVwjt0ndA5eB zEIrUCE;GfoY#<9z{9&Jjfs!=8|K6*WP-69;dq$r4xDAT^uK+lmlbwf+FiKy;=SAc= zKyz*#9=huxWtJp5NL1VvC?I5}JWjpYW8RBw1I!a`!sI$l(JiULmHsOWLv!54V%Hq| z;DAY5;r_uve)Kgb*A!cGPhYgRh5a_M*zr>2qFC*=?SU<+GvZrAD}y|1yaRx)$3TLX zVg#1rf>=Xf3X}4XZ>SCN{3#4FccP8}gF3x%>z4Y3t>RWq4?jN$JVde-H~)#Ph|*od z?CUA-s907JXK&>gX&WDutf!q&`TGoC81!2}1m6)~1S&dqu7{8oOJp)#$OL0!y3zGt zS5C>@WOgj>p7^5T8`}6!{6}9)?NMN+R0iHHKx`dzjJ?Y-A0SRsEZDlGO9_Jk(nmqt z>%<~x8^lMo+=tiU>qgx|)2b@;9msY`07D+o)d^5f1eviqP?M*`7Cd6Jn0P?*6V9?ZklCEb)~4}+triD7C0p!-pH>j=Cx1A;XLM3m=MX97P7 z9@P_xMUX(QLs`gp?<+wbSVQ7}Wj}j)hqbet`I2rlp|mk3OiY1MVuIMsH6q@OnIP`b z5jhE0&3CqvSYjh8DL2t8QIFDr1ECly(KgheJwojB0t#2|pE6yjwPcPis zS86HI$MIil{S4QTD$W2NZh)NuUQh%fzZ@+fVgQecM8f4FJ9)muF(RzvA&+Qn%bwTW zE|N-Z>^8;xzbr|fps zu@!2TIE8I5U)(W2EPb{}?3yMC44pDDA~4wB|LvA(lQkLbsk04z+?jnF8#I>ei)f9S zmC_PWnHoMLGOr{>BRUamTXnN$GMU=Kqy!=}DE866;D3pSA=JYOz#uct`5vstol{|< z%a@L!8+h^o;ebZ7i=*lnuV{2D4RsQwMg;{&xck``b}ySZ+uqYN$ki^cJElpLTAwz* z(JC>-*gD$ZJKP6tEK1AHk975z?ID}Xx>N5a7dr_w?i=7&&?zuJ6Vcrk93`QGL_xj) zT(d|e20{o5h^`Qc!U=^I9AD5?|HeJh>L*oAUVP;7X6!v~U1^o5$+xK%Jo%bzV;!oS zms?lu=q8&@CDzsE7c8hlvt@1$FD*duP*;Z>WUh{H^0lfQ-QAto^L3~;uV6tf{Nm=o zp0BSfD5$GJvpw7#j@O{Nym<@i9O3&&Equ>kP(y0jGk?aHmBrim_~F+Q%gPdMy?xlP z;>t;h4|^$oenq0h-ygq}SW%H^;}^i(jJO1PM#6e8P*OO*r^5?th>;fZtA!a`AL7vDL*+`GDSg^#0sZq2GR2jO;0 zN6WJL)|1^`;0tW!0--+66lH*&00!?a6&qlUk@H{(*MGZh%>wv+7=GWZ`W+aLM2a_$ zB=|5Fgg4&~S+E9Ra2*c=NTa-MkU#-wXF}n(hK@*MV+z6@Q+qaa#a1VT3%7Se*G{s@ z)&=#JP8J{>u7`mV-G*=?Ywq|dc%3`Ymfzz9f7RrSi-=KaeO3^F!=5;9*o}T^SR1r( zu19Hb6XRn$+iQk@u!l3eB*`RLaNdQPIj{t$9xMm(g z^01ln>SmqB_B;j)+}y*D=}mPrPeMbQc?6oxTsw~d!?h!Txsi+0VQ#<< z_9QsEJpPI~!pNDniFx7txW%n`#&tg9fkxXOy)6Xbf27|h;eUF+X!*EBp8Cv=ACB^g zj%lPPG-P5#e}h`gh4`tu=5W@G$13hSx12rBAtBE2N13a>;Gbblg@#-q`9bw}n!6cE z)iNal;SNZYppCgl>%h%Oq$k1|$F)Abr>9@TLJhje0Zv^&-J4g$h8sMP;1&$S=Q4=ygIGdtr++0hqitaCi$p)eIN;0# zp*}qV`3z(n&m0Bq{5ZTq{l~$Az273#M-AYB0(n(;+|Ya(JXVjmzrZl!jto?n{7h3P)j-90 z%mZpKG)RnsoKb>T3b};0W${a5mZpb>rZ1hlgn1CPI5s3TH6*r^Tqb76?4dkGTj90; z=6P}nZwd=*3XfO_ie-d*M1;FrL8OnDWy9u`+IJ--^r4c!melikFf(Nq&4p5iVWfQ{?9wov#U&ufmKs7xoa)}I4JGpa7o*wWa4*X$4o9Rjgb>JZ9(f$`=sHeVc~3{ z;gRBEX72d6mMqE1+|1EZ5CHY5xO%qnG+$`J7_aPr$IG^m3h zc=j;qI9>QXM=eC0^rLUhZ5o`O(%PEh;N|7OWT48_)cPD0;Ara)Q`^m~#XpL6z$@UY zCK32#_IUMYe83p~ad&WJbkOdR-Zi^$s-LHqpO25HpK!Hza*9{e=FONTabB=;fW$`Z zulN@wA2S5++@TMAWU#LzV=dBe6 z&-y{(hq|&d7B9-m=`1&Pw{Z3McXbQ+8Fgjmc6McFc9omBnYji9xw-`5-_l!J($kun z(v02ALuWgNg*m3Tw5O;FL6MWl8V9jDzji4+I_ATxA|Qab)3Lg zj>9Ae;Zo`iiHL_Huhu2K0 zoAi7~HU0yCg@5=vJG`8^Y3x@Sq;F!PKgZvys9n#*M9+&G%r+RzMNZFqEOeH7W4Ccd z|AFl61O4SQciZ-y{Jdk)=O=rZZ8Q8l%*{RgX82SkySgTW0fmiWKc*AMwmWC?1GWKM zTqvf#yg#40K8jlBVe?UBSB$^T&c?r?>0Y;PxdDa^356PiB3%I&tRsYzScNy>8cFUG z0>Y+};2pR|l3eifvCx*nsw)k7w+{&B6x_{skz8d6ck>PM_ZX5UMUhWw5}G1Sl1_1y zC{mfucS`P?B%PA`CP@wWq!kpiRq4Y8lJKGm1joTqM7%x*#B)4JXIH;^4Bs6H@Pfck zPp<$f+RxL+$Ir{tPb7d#9c}+{4Gea6^z$2V^@mWIfB*sg;5%sC92(bB$wA{928#e1 z-#B)Lc1Z+j06zTcKX5BzzB;y4sNozO>}($xC?c0fZg2|ncXbSaqt#gr_G6r>+)H+t zkhi+1&TZq*Z{y5qO=26SEo)C9WZU_MY}mK2GreFz?B^y^g+)@PByHonjd> zbG4C?x0R{+6zvHUG$$Gv8XIX%m>|{`#)w@UXBcW~!2c#1OfxXipFCmW1cOOHH9uxf zD(8u2@Qj)^5K1I6q4@S(+!*s>3__8j8B4xeGG)p7C4l2O%vI$?m=j_wprr=D&pki; zItZ;KcViB~hESY>3I(bP?qG3J|V~Ut|=a8Ebbhf)Ngq0V8{u_k6qu$y-fiHtLzZ z^b+B(M4;Cd>Wjf8;%?fUCI@k-$roQ%jRWDo7a z zX~Ph|ZdhLrh7HBXc)D^g1YD8;FR7SAN1k+n5s8;U#1eL1?>yV^>DGQe=CjA=^-Ex@;Ub%*zekqre2zp#^3y{e>cc7tzCcvO2- z`@X=+WuX$^#c3%a0STc2GoAH}riTg#oQe|W)(7tDt=Ul#IJH7AWL0Z^QH+0hu!d&r zR6W=>@6ngI;Dkw>UqLG7H37*qV$7Gwh`Oa38gC@!6X{6bBwS&0vS^kxEk$H*Wn~fZ z%+Q6yGel-iECWf|kf-|&fh=g{;W&8To$bf8dF%yTu zQfJ0Z@zW3Z?zdHk?)_Rb2_$S;-tM04jAD;g80TIx>FGMs4YJkC4h;5eHd zDx4=`N|4arMKxnPIZu%WC(Xl*sX~dk;f#e6d2n-u{Ci3!Iu+KY6?fZuxXW}Iey))J zV{t`Z!7dRX<$8ik(-j(5ANEU}nvWgw3&;IR*eNCDLB(OR`02@;D4Xc+O{sD7W(Ajo z3)@i6H~3{P@Pzy`7KMUBVETcf5*PhmkQ@xaQ56i!Jx+Fzjr&fH%;6A{7o`bB1Y+^@ zSTt5-4~s}*xc;<(e{8I*ZCoBkr-?MXNlby8x1GI1iaI(?oU4lS0!nIUVk(ia-;N7c zn;BwmB>O$d#bkWA8Uw!mVf>N`cq=XmeH_5!fv5iy{@$^%f_r?dtZaU>v$C+Wv#_#r zcr1MEN#N#v@bzj3sMZyH$jNA!DZ9|<>se~)ms#&{jSG;(mz3s-y6}W7sue+(Ko3Dv)RkDjSh~<__ z@%}QeLQdtE%vs@EJEad0 z5fWM^BaWqUda422tnqQ;_3Me3MY~0p6F>bFzA)e2?d)Ax-`DleKTG=S7xZ`D1r_0A zVJL-#AA#leC(e@lGINKT0v1M*=m$L#$$l4}qqyfE^8}!sB*8x$>0VycB`1aJf>V~- zh-Jwa;ZShuJ4HSG2+{petwmi;9T`Ej;M&wn)Xb3)RL#``(>fTuM?l&6FQ#=!D4?G- z7bk*3#5`%qPEqMcDsvyK1aI!6lwUCSC{58{oI>)5F_~zlYJ>7E_7l=5Z&OB!&Q355 za$s^V{znf08R(N>qGNoEV3{9pa39z5|1bd%xc6ksF>saqvnBvmXgA#T?jg2^Z@h8` z&{>VMppW+`5d9MU{DeIow9VyDnB*zJS8yx54FIj049?*An7@a_xKPi!4<3I`RN2Sx zgO4EVrk-rol$Zi;KLcd$yh&oXe{x=Qc}xD2Q$v-jZTpaZrYD+Dk7!&40|cNxIVn_#TTie&~dz-Jud2;r_cb4DpWH>AZM<=I(9Z%$wES8@=9O7%21Ec5LsxYpC9<(1q;Ayb4BC?J@9ACJTyCcpxtRJ;+B?% zhL$dkTPbpi>q(Bx&W=p(i38HK0P9m%ltHjI|9}AruFq5S;t+gEnF~9TK+wriaw9Wfo$Os^wB2;R)K!%d12kGCCk!^i!+xMK2w@*%djP-B+0*@ zSLPUer*lAM_E%|odva#HF%1Qq=*n}Q#*QFW_gqV-?{XmdQ0bt{p@iW1zrLlOJq=Ajajdmr7=)d0sCK2-o$Z;*{=P z%WiWP;`P#n6|Q1;3i%b@&5&?>%)>KK=1P%Zb4~o?^UsIm{G{t`fyR_xzxwvqYgBa! zp4|0MCW32+tC-y9NyrC6>`Neg&WyEw!OuUx{gz}?VBesgS+3$YSA6@eqkptOm#8Xo z9DU)G>t6JyP#5a`7$y5UBh=A$Vf4I^p9J(&wUM_L49{_$36B;>%bForo^WB2{J1L# zf3v_|uh{whhl27ZDD1^cBLvPu8hFSeypl=6D;Xa5yL#45x}FtXZlq@2B$MX(gNoi= ztu%u|ft!hV{s`Uaeo}qJV{`mZcF^EH*77sym6o3=4`WxV_^V%`Ad_4_b!E9%gSq8G z#CPof$Dqo^A8M$xw* zi4T5#Yt}tx)D5AJOztOOL?gJsc|W3 zsmP~B_=9Vg_%d$d`1utr%NWJ=08A*!L-E|GYn!{;J>GMpIE&{Gy0q>m-K#t|ihnib z+gu*=Y`-0}WQog-k3HkTzx)?$@xAoH|Kp>Cr$&r$RX!k8$)AoAWKa)z{slGDa%H#Uq+Uq8h;7fJ{PZ_{&_C4SCpxdU7*1 z4MSnC*ouEOXWfQ^cHZhLd)KbL@Td0Q>>?A^NoYRnXeLLHNk%)pg(e!A>~FuvkjMt~ z+a(wf^uXpJHluHMrH21Y_KSuHSL)4`%Z3e?_U!gc#}JUSw=7r=ibemZKt86x5bG*q z3cl6uN~OrUu;;Q7#a9@-L9dh&aiEH}->X-W|77yZ=k|W+dNn062|q`YQX;6X?*wCf zdZs=;1Fx^J0N}6UYZ#31S9@ssFKYYo7dDdMzYO3NHX{9N+fo2w0w2?eWa$SXB2x^6U@fzlp2VRFox>Ro=yAxx(_7ssT?FRSn6` z9#U2GMDmqZgZ!=V@sjMj?#eECyij=Nl64>=87&`?oZUWY`GLa$`&Ke7S}Dt#(j-=>?s)SGr?}CZdfI z>ztgWb_k3W;i#GhUZH-x65CaKAJZqUBqU!!_go?%fozo*3Ki(N*+2|`sIqmfl@VOk zOv_yjVi)%wG@eIEx9>KgkFA!53a{eN+F0b<)V|fiPiF+n&VwjzHOlA-!I+JOVd-QK z9{$u*!7~u2W%;2);4o76+F<{_` zcV0Fl!8GYp;n$vV1pn_p&p2Y5exQAFybZ&M%;%pn6=&DqvmS+*gzcPt@JAJFt0TGD z^B|i)|KiK_eBabku{yZ#B)aPP$|bk2YwUa59mTy%XFEykIm~*eEV%-Z#CY& zq+-r*T>aOtr(fEOXN?|}`Pjhvf4(>Bp>^Y5yRg_b19CnCITvy{V@o!d^B5TmFVMR~ z&TLbXAk9~P;u0RNQiZH(@HGKXl;0M2<-qCjL+S3J2#XSzi>oC)S9b7=6061K)zt{7 z6`z;7N;mrgY7y}Kj2Q^15|#^nWKRl+qiPAZidA2zK*(Zj$^8de&fc8i8y@)1(Bk_t zgoqbFO<(Tni3nyuvo`|U!Y(~lC}5n~8eRoxBk4R(m}%8pEUd~^#gh&a_LF5?3V24C z=$X{mpz|2fF8ygzkHDU`pQPj$pI_`6q^kJ8^P;=PEIacWp2CnuVTG)U>v(I0w@lv0 z;@)NR5vV#rDM@;4VAdn0(e}GyZ+BMSmG#KIF{WX{ij;zh27@I%aqxo%LlQU*bblmgsqy?0Oib6I7fqJ& zrDhTD(=ocd-yP!~i=;zEGd`+;-5*{<=)Rub(ehUbn-EBdkT2X#U+4XzMe-lJ+ZHzp z<14zGx(3{$Ge^?bTi&0!R8bZPG^Gf-j}Lq)W8`u8JX>Ful`dZnb-xaN>F(F0^HMz) zKHx8V**n43^&?y^?2SfWq3WUeZ#T|cuXn&pI&0&;HUT3 z44k=i$;^R;xs#u*ot*2xYB2ACJBK_n|I?RWJ@MKvZqLj$w6Z1XU~3W1E97c8*NlIX6A6i;mx)r$EAz>nqTZeknP_p17xS@u;7f=WgA7Nr7B44MkT z`wZwpPcgE`JlZmC+Bu=zg^%u?&k_De!o`PK`^hFC;pcA?AI1ci^B*u$cAEQqAwKn3GW+b9iH)%@X(+B zTz;S;Ecll6{K)w9J{3KB^vo=b&S5(;%;Ey+J_*YeAr)k8mg2+(^X7R268!_HH-LyQ znrgmiLL#CjS0SYFJ>LLkHvaWD!6PYNC?=0%K` zhwcqUuOcv)SBf}6kb#gJGzcTM+^FGOiIB6qcqjaQ{jqH+=Sa?O_$P8uj+~vhABDer zDAwWR4!^4${;Y@2@;-!>pYg^Syghfru7q5B4vRfHSBLzt2{Q`Za3;H!d9LqoM5xG9 zs5#619TD<#NEY};=B~(d<~HV?$=jY=l{*utciwE~a%bn2<=5j{pIe_-lDjBxW!?zp za#wKInA_?Qx51kR`4#DTe)knj%x^9@k-s8;FWl<4Q{l?|P5GM&0t*5`IgGpy z^Q%4u#{8{?-C9#L=QD@0fiKI6=f-luHOfK6$_6H%$DCEm@bWwplIrjek0D!s?y8ER ztru4XQ}$jd*nMEifh&di=HteHa>Fj=R%3*i&6!wXGxO->CjP~NXb1DPdqqj$MD^t$2M{lD| zY4Kq(AarU>18^P~ddAQ%`c#8i_Su3kLUcwt3^WJg|>0RDoow~uf zlN)J&9W9%?s-rQ{@A|dr9Z&UR#Vp4Qo9q3R<9+N8`IU!WZ)*4Q%Tu~*E0?*4lt)}1 z;-R>U%JuFYQ;XWdEB$(zHY!8z`f{_of4Ti)>cv;?56V+7rg*O0&E;%#bj2Uj2lwC0 z3oegPvHD_K`2gQFPvciwpH8`($~XJs@N@6-!EU@;a+ea~`i4=;hq+6qT77Zc4V6qg{IDd3kl#y`X%AdwuyB|2XbP+OFJT<;UBuzz~h0ex@zB=?!} z3LlM~-2=<3{IA=^QRUU`(*Ng)d=)cvJDUwJx8zcilzwY#c( z&ZXl+cFhl4d8@^^j+QUHIGz7h8}{!@E-k0qYccHp-JiFm=_qplxqKOmd(l|p*7B9U z^!{{@jru|^KfAtnRNs~cZ6elHcz@c+3FmE*zn)qmetmaC`!&xKw(miF}-OwRu8 zrmrpSXi44n@@*Z7>*nom`!Ug`?rcBga;p2i^4*tHFQ)kG&cz{@U!mRI$EdKKgO}-e zL-~=*Q(R2Bl0ICCesSu5@71OE)6v{6>&KePPjxiywZwDpoO*&C9Zyq!w&QVscueS} z3+0nBf7AKbr~C7%L5%yg+DdzAz$@5SQumgrc*@=ObJtJp({^=rgq8%9*Q!> zyZqX8|1ncXx%l@T9ZmC}k9TeJM*n9;KVI3dQbuRnIP=FU&-K{Ak5#t+rhMFoD(AG7 z=gM;}7RTK*)!7#JMx)A$#(p=N`;F%J@0G=kEiIc*`v3lavJH3h)O2?5{?Do}lEtB^ z?aY4sV_b~?FUZb+%y9AD{olXxJzMhVL)-I}|HQb!eSGSWb|v0>>LoKkPb~M|c4Plr zi5K?||JJOoiMjiG-)MD&-PDm+i-qy^*sF!#=rH$zsS`TRb81<~<6eJEch%HKuRry_ zKc%~N>hynCF7CZkXa2i*|F`kI(eDEP+wSjr%X4vmKXulne4rz9FQs+#+TAd99?~!J zkAJo2QU38R##5JHeC@lMy2^+C7tYsr?;X9fscSkK^S>pY`(Ks+w)YP=(%vt&k^dIy zwO@`a%>&wxeQi{KIr!c~TpNeWV}IW>WB+lh|G3`2ESGcjym#z~EIJaWJ^sG@{max| zynhD`&m1>)l<&2?C+jHB8y9ctmK(?G+Bl(o*nXuHm{ae%QrQ0--8^;Ae{Y$m?*H#C zTgOXwX}|8jHi2<*Z~cB=?)l#z=f{lUKJdeHbG1x8l=;Rd#8o*=jj`r z&-L&_InEI*4fpTQu7{5=&87XO`zv1v^zC?s(sp&_XM_H6eAqpyqLU9_-?^gemD64S zdlHnZ_m%nH_z?G*iZIky%=K}usJk?syRV1B)=aRqNZ0ziY<APT^rMiGe_ls3a=%wm($ToSckbj#;YwI>6hB#cO3c8FD^}WW!tP2FN=H0T8&G?J9_Ptr*Azuw+rtW z+sDHri%N~}+IMyz^dacXGal^Qk9Muk^Ddny`{eKbds`k?L{XQAxTjTiyPSICDDH;J zmHyn^eyc#;zw-~dcI^JL{WR`YD`-EQQ-j;W+O6%TIX}{il z(59qbn)cFm(LWrvmF0hZHSBJxECBxjSBvxiTbO%i<(M0B&+apogKq>MLMn&d7(U;} zrEVNn)ZdlRXUv#jq zzhOCE10x}geMd*oy#$|lF#D(`D0Q5ofiLRr(A~+tOOqz!2m}RxXRUmxoKU~j-lUta z!#4nx&#`j{-;hehY%C4gr|EEDB%jtB2gdnn_<%E!m80y>aWw}ueDNRi4lop&8nii90O-oJ7OBwTh;8m-8PB%;Uyl%Gc1>J9SYjlme1G-b-K8Ssl&|3nMhG*Alc>9os{xoeQ zj-lQ&+P$=L?=cjveVs^HiHQ2o$k`H z_|q!o1=5X`T6voE22TMW@FTWfIInmAIc+@X_;{oBD5QaZP2*oB3Ltg(3ejTn2YgfM zeeyYZN6-n+ko`ifuv8c(yega!CJ85nbHYaHTj@LD16h>^3Oi^f%@jj4F0Cy7L>sKN zh@sj_?NTvSyG^@8d;tG5*5F={rm(!N_|B4&>t0~3P>S9EvQ`m^CZV2A(tg~}Wj>LV zb?!sfk;>cLZ)84bi*X*HtN@R}%Ioll5n0yaT%uaIpU8bD^NCC|+<#A9M04DyQkT;g zz;%^^cXFv~2vweTr=%`aR=68emm!}9Ybf$zek1c)?Yh^e*3kXj|BU(SS>|j0$o(DM z|B(6QCW;zN9ZB=JKO6p7r2Zha45^v_8S@nl=o1u_6h3M{bprT-4t~tv!Tk@J4^4`c zwX_dSrWv#!&7y^LHEr-{8~r=|LZM1erBHcVS**OOysoTM-c}k_{12w;R8!PcHBC)d zGt^<~S=Hsy%z!-7MW`X!DF=mTnNP#HpTKvCOmx*P}}%RXJ}wO@be5q=6?qcV;DZ@h{^EtnBg-Y|8xvG zK6k4GxW7>X@eW_4yIh~sk6ifg5*7k(3?U{`PHqkT9Qp$qt1U=730gAS-K4w#9(CZM zpd{<*HXNumaG0!~$N2*CQKXmfN6F8-C)3}!3Cf;C&$yTJGO37PkNEYVzM;rGzx7C0 zk97EUEWRkelzzzLF4YL`r5e$_R2zt+Gwxi4)*Xk=9j68EDsW#4?zP~Pqfpm+qQQMU zNV9M>;Mj&dces<7oq~H3xF-=Ueck;AG-xvJXEN?*GD=hrZJ7*RnG8*ttoac4u0`4^ z`T=6>!1+%&@8)+_rMBRF!Cj@g5G#Pm88Ls2n6-%cYs9QY%v$9526Fs0V%H+}uMxYJ z1ffh-h@C^XLt4;uq{yL%arX+OpyCK{H=|yfmCnGc-Ah@0fJ+WIa^3K4yAC7DsIXH50-8r-EQ$r~ua8;S-p(BkL> z+#Tnhko;}r4{8$bAsL6K6mOsu&?a)9^b+jnHX2UjsevYWttf})!jh`!6LdO#l2+51 z^cngreV)EZm%_T%(YI+MeTS~6zo(n&Hu?eGPX9=E&_B`Lt(JF~enF4ZFX;*T6+J~y z({Jc^^c+1;Tc{f!lf^gmV43(oXGNoE6`c~GbW#G9&PsRX7Gh=pfjQcE+F~j4s=-ek(u~j>0t`Hl@Vku9`BP*o`q{-wr zQn@shG)a$1v&jb8C>zNxd7wOyd?bG%e?gjQ2n``0(_3gSd{Zs5&RFs7wXT|pOfS2By|$`Qhi*lA}7>U>MC+lU9GMrU#n}>HRP1~uDX%@ zOFf~UAm6B`)idObCO{KFzS9J1x{!04Zkk}?(!^=Zf}k0s86>EhyEH=ujpkm>eL^Qq zh2{yNvu3eosnA37y5@DEr{=es-wL;A-qE}x^wO-?tQT(8%34|It<`Be2|v|FY9oa{ z+Ei_-&{sQM`=k)2eL?%85TSiV`-%{)eN&77jny`48wI^~y>^2Tr`@RCC>XTcwL1i( zc9*tUuxL+fPYVe;StkpLf;}ntQfDtu3U;oz9%rBKTw!mv*V^msi|i}yYwa8D+w9G) zoNAmGIs5*$Yx_z2Ifv{Bbc8tiIP{JLN4oup{iL(zQmV7YQQ%zDzHcDNiar^eaU*~=N>G~z#Yvz#T)Af|isC8E7|N6Q&8% z$Y;U|;eB#I*emQME@7YW6>$sa#oqYvXJ0WxaEL?1A;J*Wk_dN;zZKUAQ^mF7R-sb- zqqtL;CGHYG66Q$3Qm!ypxZFf^KT3zCGomV;lg^1fVG26&7TG4- z#9negxu1Be{Fb~`>`en`fM}#$Xcy7MS{u=<1S>s6ixRGci#FD(hzUxZ5+^1q7R4gk z6^G&w9ZHIlB0819%3v{xwJ>6`QmT}SDauG?q?oFVRmO^G%6MhGn9f=oF+-J9SQm}dVt;j|x>6j#S|Rav z^-c9n{5!=v>O0~c>Uwp(I7t1S`aAJX^*!}Haj-_xXvDi%TO{7A>7nT<-mmGSi4sR? zESe;7oF-F~B~I4lX>J$GS-T`o*9_N;6st9lXeNu#YNl&y#5tN5G%t$tHFcT=;zG?b z&1&&g&6}Dx#RkpWns>zCYTnhnE3VaS(rgmn(!8hHBCgZCulYc1)O@b_TwJg1qwOPZ z(Au;K;=8QX5`U+CSo^TJNjq2jlK6Y=FSWlE-_tJAE)ut}R!jT?Yqi9!+TUw8i`%q+ z(EdT(u05zdB>qwRPwhX&o!VpCW8#O}o3-C+zZE~$p3|Na zcWYa;E#jZGZmnC~qr+4{{3M`XKtFLWwC*V!82^$Ck_QbOM26wZmiM8@DngI-2Qr_$ z5B*ysbjBZLd?SQF3(qqx6ov{_LXB__<`XXqqghXe`NXTx$w#4)-w6#|BX>a~^Muo^ zZxeemy%fX6`{5$QpTR{k9TlU*r^RQ*Jn=bkwpc9A5toX$vz`y(E5tS8Na*hS;v-CV z#V48WBK98fu=tF4L~If3MVFK-z9AJ#L&Wc-d!&()%KAvD3+p4LQ1p>=r9RN?Lz0te zw)8a9Y-ygcE%Hw({z+YqX4WFd z&_Eg}$1V9$EI-U_L;g8zW4t_3ouE#ZC#h4^O1T17Q7u0XOZbiajQYCzru?k>mimr7 z2R5)p{-t_UJtHrM_Uq&ZO(#v1+^C7s43hs0EgwmV<^jzcG*GifvzFe)^q3B1dQ9(S zdQ3~19@G1n9@G1w$H(YM%oID((b^Dg2%W6$sqIP2pxNPc3hR$)CDU*EDAR8`O`EID zrH^ax*4|C4pzDv&C$vv$pQJOiPidc`PibG&zC^2;-qUBaFKb_>HQM>w`Se-sLhV9Y z3rko+pVO|@uBEe?J$RU~KcNe>pK3p)uV_EhenuC< zPX11R#q5MG)qbJryGOM_UydeaWm9T_qLRW4BQ@IVyfDP;w=5X!b1?|rhPD1a8i>cxW zxI3ZAPl8$LKGo%)#_cCRAFDuad0%$Yz9z(7E(0LuxcsWr1xja#hk|)WNva%)+^ zt>xD=2-dO^e>iZnyoz?G-R0L|Ilbi7+*;m*wHV~J+*%r8ElKh_G=-+g?_!RSD{rFt zG++KbEvCitX4;<)kl&+s(Yxe7(7Wl~@>XtF@58RrMv+^FjS@}TufOc2@ zr2L5nE1xQ#(jFKu%QS>pDZLd|`ZVpWKBLx9hx&s00!>n1R9~dYu+l%!6!m>|8@*fo zQ2mhJqkg1*M29geh1E&0$#JmB59m(Ke$9Txq&ccNrkI)4DK=(xN*1i{7A0HTTN|q6 zYQwZ)N&&MyrI6X4Qp9XeDbc2DGnD??e%gM@?aU69fy@q-J79+sltFmH^SClZ`-Ju> zZ%9rv0@tNxMe-mQn%R+p0`sqh{qfW^c-DW^c+|*xTQg zmza$y3z>~6zhX9~EP{=Ft1M>Lr7UOGrMw2~Qj`_Cv%0g&DrQ~E>jC)z`O4~mI|BwQ zZ-{@f%=@9vaM*CdaMmaqI~jwGp~e^=Kf^u`YA1Y;VEl!aKwS&n_bGq#u%Q;?~r zDa;gavYRql3R973Fi&~Va>JZyfoZsDtZ9;InrWtK4&!TD(8k@g9C+Ceai#`qulAkk zxV4w5$+X3^)3lexFdbqP%jzHIOlQ35@kaqa$TeCT&BOx#dYsj1U3hKIth1c9oVJ!Q zXYS@XU%!7f4v@-bk2Cl7p=&!^hggT0qnLkv&a#6!v)OZQ-!~_N4sEnXbSRH3bB@($ z9$+479%&wLE;mm%*P83hi_9y{Yt0+Y+sw`8{pKU)ljd{gb7(IPT4YP0CB)LlqPHYi z(k%s+L6%a>7|TS;5;nDS0`}WsnuO|8F1+Q&iEw@t# zTu%O4jI@`@+_DEP3gl!t2CXZxX0^qDK8>@utQyd}@^X!XG>y1o@mS0f9+$OgBXE>i ztH8eo$6V_|+K^cBL!Ciq=w|3`h%%TB$%Y)m03SaC@t}6YXC!iG7-|^Fkh70ryt7Z6 zvmdg@Uyn1C+v5$}4ATv@hC0I{!%D+i!$!k4AE(4w%o&>PW7~0)?3cO3`iq@mf7`X; zq~RPM-vt^&jD3uHV}dc=aKvyjvHntOV!h)WbH)PCxqaWh95Gho7-TFpjxkO&K5DEs z&Nj~DDVMPj$2oh^_d8>wakFuUagXt!@tENn-YJJGadG%caR_A z61OvFI^j9T^lHyJi0Q29Y~t?3-DZ*b*W(;@i3jY%&7JJSnKK95hbJCzbi3S{L(QR% zBOYgOus4`vd{=E}=a?(G#G{Ev{lm;w&y_jVKG;6kImo`i-jYHxobA93|k25bZ zFL4li6LaQO=2aeaiND2uMQ2`T-)Y}z-o*S%++O=-F0sYq%-enF+Rm(Q&AZ!b_*#)& zF(0ria6$GU^HJveoMWZO*?Tf)KIJ*L@0(jd54+4|TE^mmsbcAA3A4mo?3PSRk!7%D zxFzT^%2>xxd!}WQquepxGR-p6F%m7)w(D@sj%#yXd%3FAPCBm1S>`b3NQN`Fx|NPa zmIapOj#~Si=+hRAFf2{>v5p+e7S_{Sc3SpY4q1*zyDVp{#HzD)vktKKwnkaa)?{lA z@KEbW>v(Iqb-K0IT4!BkU1?ow-DurrZMN=@(O8dIPg>96p>~2T&=z9rW77kr+X`%h zY^Am_wu!b!ZPm8fwt2Q?w$-*q`?SOgd!}u(ZHH}-?V#gn^mJsj)JF7La1eG%({JUtKVQBXeb*x{nS1TA_@ zuhv$YwXt9wnagwi9iw18)N^w`q^X*@Kvkd2I^z``Eo{rnkKzhWX1CMTw z-i|0}Zn7f>63gK-JLeb*n?x?4Opod1s6{-M8_RE{V{J@V^nP6(8)31{pfQ=91Ya34 z55Wjl?@Vx}I}4nHVtS#?AEW#>=a6BAp_d`TV9eQ)bBH;^+?*!(PL8t-C59nh`Ut}~ zjd7gjdG=T%7FNEI&?w5um49`#aQo@Ws_FEA}PH5ATd&hw-I=e_JXLGIAp zp{6rtV%FsjHFv|&n^8Q@Y&IvGb1s?-Mw!isa|tzfXvti2tz``SIvn1)BIbhWxbn^z zOXk9zEt$(^hivxP-nwq$U{*GQpU0&gD4bKW##&aghPmAF%*QOP2IpODmR7RM=L#2C zS6J7;-(cNp-Bq~2=WL>NpY^Quu=RxXtW9Lp+(oucw$ZtZa2{rxkXvgDwRHjx#t~|Z zu~~UYD)ZZ%EzdTPU)9eZKr@*au*dYPf&0KCG<=PONcLAp1ToO_JqvBhJ>Po!SILU7+cuD96KjWDrzoj zPMC&cCPO&SNm$@Pxkqx3B%Db&p0GTjA(13BC2V07&~|#xKzkDo`LH+ac*2?7BZ&lu zu4G|iw~~c| zXp0h8Caz7~n79q0`xB28?axb3JehdTF53h1(p%kZdr00Gd!M{9d9&?$dxAYZZ=$^b z#~}9qdG=EKGWZCCKM}{H_G-`n>Un(oZ2LT*G4^Hl&GyyyM*HTXDEp3rQ2U;OP@seM zWA@VpF?Lr$j6+itRTSmu>gWYOf@33)(~-rn$2m&!mNDlT;uyis{Y8&+)HudD${baW z8g`wxnYp4U&)G58b8hvIIXgKPIz`8dyd91;&S1xeykm~7j$Mv@dB=Rt8O)I5u;YZ| zET`?rJLVLfot(kNjm}U!dWgZ1ha(mKz@nkdIfpq%J15{=;hf=|<*aws=lAls!luF| z=aT$h&Q;EJ&P_OP_t2Ay`o}nTCuPDvz;IHL^QiNbvn5GM3Sw7DJ(I$c;u*qW$8{!- z!AZlD#=@G*Hz{xmrk_RLY zEhz&U$^15#JU+RcT_N`LuhJ!hcCl+8Zu4cn2j2Xv$>sY>mdax8s$%IOpr zuqJ&$NmZLm?Zt3vR~$g8MjR384W4jzRZ^9|Ce@jmmA@u`S87S>kkk>W<5J5~tMYf{ z?<(n)T9Z1rq*sBMdIHD7)D@{~QV*wY0Nz@FmwBnXaO?wxg`VYWi;sdH%$#qPF~dc5 zxr@@4V7+l_+j`@)8ELc9>it(BdnrrOR;8^=+f-DSwmofk+U}xlX$Oi<<}T`QO*@L? zRM9!+(pu7$^dPvN>0#;dMRn=+^vsNr=|x4&B^~;u4^AKM7jZ}lT>9AbvFz;UMwDEe z8&?9CK8g7kyYy-4Gd+~_IUdI(e{oHD>IBr~zWOPtxi(ikh`IEp^rqtK^ex4u={w=} zrXOOiXgXZE&lS%u=~cX%x%A_3XEI2JE~8uVpp4$d^NNx)q6!-_%sGcLk_)G?u}x8P zMh+Y4WDLz1kTIZW8_pv$re}=LD2HF0QJ1kOV`awLjEx!FGMY2?XB^2mnQ<;t&J2VL z$?TJ<$DbjmXBK1*$}G(slQ}W-(ah@1*_rb)m%**hY|Px8xg&E==E2NknWr;d{WSf$ z_UqLz0?yda*)OYKNxvceM)VukudH8HznXq?`z`FZqTiZ+8{oF~+tqJhzr+1b^gEj+ zW_8L6&I--S%Nm#!lV#0H#hKxOS;MkMXHCeeD4dgXD0flTjI3E%^;t`@R%NZr+LX0D zdvMn7tOHp`vrc8TWGmT0**&wvvg5Pu*_qiz*@F#1*~4*+#W5*+TK3HBIoS)cmuEL* zH)U_h-kH5O`%uoL?Bm&Ia!8IYr&~_%ocJ7jPG(L~&fuKkIb(As8G>`B;h2eIPR;`S zuWSQGX*1c#1mlV2g=iU>Lp#S>lf8u)d15@f6C=>QIF9F>$tAhElC8Pj3P$Jl#(2=2 zo1B}IJ0N#x!Gw~zxg$&FGETYUbIVJ1Q2g+--$8axn(T z-G6bfo_nNZVeZM?b9r)JU|vXGpFBPK;{^1=1qHh?@*0F=Oy0!2N70+I9(Er3ztwq- z=+$=Q?a4ccUg~t7D_@h}6@7~_KcasOe52={3i9%u{e$zf3Y+pv3Kg`%Lp<$m{)qf> zy!}MmnP0_@{F?l^g?sZCUfjCjYEAxz{HKlqU2ve_Xu+w1 zmO`a4sIX^YSYdpjy)d(|h+8OcI|~OF4lf*AIO*#3Rd^ikOc5#46?H4>jnP$9k-50B zD7h%7XaM@>kwxQSZ_|rs7u6Nj;;1WHRJ4*=Q_)(Ck~S7?;~3+klDS1kicS`tE0&7` zi$jY06zhu&u77Z-4gE-fBYJhAxE;_Cil@$BMx#miu~jm4XbcNFg_K3IH=^^V1^ z5=}|hl3pbdu-AwZV~MlBSmG?nLK`;(Z5V5f%D7Z&FfLz+QQ8_dL%|q$U&-N;6D4Oc zhUtV}IkbOFe{28L{&`q)EL35E2ML9*A)Po5thEiapf$e_qX1D}xF;Hx~oLMX6?U_mZ?hE?tdkwjpDmn}%F0}c`n@;Dy;zu__2 zU0{hX{G0ea29hY?Ad!`yl5X&IB!N?AGoNEf1z1bCltj-xcIPo=_<B+Fb>OGLl zCrFIonALt~UP5RArxd~uAToUHOM)LHKF9qR;OqFk;|z>{4u_AwOXvPi;LE&J0_QJs zi3N#apa*a)@_G*v_Hqi#s}sLR!plZTG$^f@e+NB=xIHD zg~ZWc(O(e@T}+n|D_u^P6FXf&R}cq%ovtQM>{9qGNybiv^(2+POW!3~Y_9^z#%_fV zNiOy&>?ZxOSK%``~}Me$MtLkaFx! z=uRrIpXe4+siY|-vDo>yil&ydAf#Z*gPQ=V65lNHK~%3QKisZ)MQ zRw=J2zap!ZrOGn$TV=Jfn!Jg@xca_EjaL{!&m+ z>8PhP(u1a>wtkA*$|8w0n`VQ8-6DvYhdo^uwi*nS0$PALg|rZLP=tCl@p{$rde!oJ z)uKk9#15uvS`971{u1D4QO6Rm;}FzwJqe>P)0asIok!;Z&!_WASGs^MfR?-h4d}`> z;AUvRV&Elo3Fz1jgB@nepc7iI6I!kl*ds(=Cu!Ijwi@@+KpU{HXdQhAc{R}{5`%p# z>w%emNYIaMBoI4VKEQ4u>}r8_?w~tB`4jyUC_AA&I<7qdTzgcmJrvq=h){YMI@Ad| zbcFPw|D^vUed$qp6!>#!RbOb;SGdPh^b~Tzei-2IXbXHdbz={fporMvg&i>@nC*!n zU9c=t2YP@KK%%fa20JXFh3F@+OQtj8FkS2dUF?aawYNYct(NGS%T1bS@ddK?Zt#x60XOerJ1l_^R&@Kj|g zaD`F&bZ0w$umkQkSc1SULBlOU150ocjfx#W>$ zmZ<7?>N!HOGw3`C#O|OL5`Y~-7l2)`nE-AxI@rv1(uwUEBC6(3nw^AdK77i||3t?jDq3VgHnX5enHw9eepD7Q3;TN?TiQqL_-5=V)n zNM~+uR&H+r+}`52y>;gHCVAUGwl7HHHimtiu(9sk#zOsUEQ;G$2=+7fLui|&#j_Qc zpfF1d<(3x3EiJ^)(tgU@OMzRLNN3Qepkd6$D1Dkf4a_V}q&2h#nAsbpv*~QuCu>=| zajWait*#rly58L0x^PSD!fmVzWj3bzS(L;rDwh5hEv^8&Y64GYQF`9)3f!)!pIvq0 zc6F1VU3K!at4`dmG_b2f*o*jg%Jx7qySjRGZPPj0q#m+O%k^avG?N zhrS9lk%xu=jpvB*8_Po%0F7n{2eFOd=vyTiZ#El|??Rv*w$(%uhXW0>Ed%Xy`mt?@ zZ8>O^er-zzZ%MqDHroauXNjBzN=MG1DI!8=fcC8-*?QWlkoGljiMJJbQJ5_m=qb=b zZ37sh=WM~)p^g-Cscooj0cZ<>2HWPa63C_2eYTmPeGaq-cZ?DsG}ShVN|JsD6pwEYNK#jz80!Hq zim>iRS(=blx&_X73grP>0C2SeBrT?1NI9(prSAVcIUgz79KFyiC^7-EL$x*Kzm!7Zhh3U4Y|d;Pg%BDb^&z)+GN=Zw7~t1WwSL0DX>4?veB9h zgnP5#PI2eIAp%xt?(@<%G8+4CfhUyCM2dPV#!%?@h-EHMu>>h*TUK~!wLHa4gyKM1 zx+8Rkr3UhY)GXDEM!p4_&XD*x5N|ESIY1Sh76HUt3*il*iJS{70oNh%mq7D)JmhW} z&XDvV&@fIL3N(Zv;X5D?=j|wSfd!T)VL!hm#{!GTxMHJaoF&gAE6V^&7DJF^j>U=4 zIHb+v{6+y~F+^EO(_2w;E3z}BtV5`kqn$uTjv!Bqo}X5Lx|jX-A^P22%=nxox7 zCmE8#)qI>Il+ApUBjjm@h9VRtG9P3}dIV^HE85GDh&nQZ6A z#*oqlXbVTMY4awIAOrITj?fC2QAU(`0?^u4)WDGRSI|~5B-aA1U`Sad9yKrHXf4np zhSZZl3mC$V$fM?ZPCE@Wm!lS-*$gQU0f7@z%mBZc4AFGqym^MD7I{EU<|-bq2lc%|k48kPCFrJc!2|DMnf5nbD`n z{XqkFhJ+>NJPRyYgf5%27)|(xTxw1?SW(i!VieYek4J9N;>%{cWf9OU*>1L)nRfnO znr@zEWWCTga(A=dY)0Bp(Bh5gg`^(xTjm(E5l8_|Z)Ckt08o^1I7{nlHfI`P3*tBK zQKploGc2vE*>uW?UPwaEU^-zs&C-H)(#TrY0HEVW*0O#iJzzR&Isy5?R!zrPnblti z7Sj>aamHDy@Px`g6StX;8PKw7^oVza$Aev(4lzVe0G(t=6DA%-N`@kVw)1#@7S5Xv zo6tV!%G@SE88({)Z)Fn$hgdrLO#Pts|b7C66Lk9vC0pUw0+^=R-5w&3pY(=_Oq%b5$ zB2?rk4Lz5OBiy_797kw!tY;XKzze;p522R98Eu2{D5ot2I?NGdYdpw9M+5EWG&Vxm z%n=(QY-EU!5Slo$fwqn#HbPj#5gQ?JP2eL0t_e~-Xv77&-N^e!O=E?p@Ih)b6vE?w-GC$K2SSr5de zD?xr1Pm1wCJcSHNSa=E<^0e?2GVa5|Q^@OpjxjEHibyO+TG1iS8TVy5z>pLUw2#wX zLvGa^!8c52NDe_LxC4a(ZDfc-`@ACr$i#wtKpPIUh9hXb1*HJzPk?xBv2m4UIiuk# zA;b{R(c3^V9HD*;5geUIJYFC2%b+1|aQ+gg8%NL=1LOso6{r(KQZclg>zX`(ka#X3 zu?UDuh*@>~nRv7?Y5>q#mO{J-dGOqXJkYp=giN3#JTx1~BcW2cH2y#|T3+%yS7Q9W z_=BkP`_O)$LRbzl4{BYG9-E1(s|3E&JZj9jzo_N35h~hUB{t zx{4vpwn=;#LsBqk{C)-KPdu(5jrYWtJ*Ls zPj~}J6OFN;uv(gK8X1lLkDQaH$DNC2v#*fHuWp9L_m! zj>7mKU$U4Uw=rrHP%uza6nZWJy?ords8v9(OVf>eqFDV71Z_nW8~@J(ZClhVPrh;U zqGo``xYR{KUxYcX=D5%(rY}2zdPSjs5#9g_i5dp9+SP2F6g3bfI0u?8ij9IBE_{Jc zUUXum0!(1=Dt^jsq(tGG?SBC-nl_X=8pbxaS~j7ASz2ZD9$>wGi= z*B92K>t}M>e4rT|!DjR;IeHUZxP7ozO3y6-PxZVlA?rJD8QRK^TMru_8n}*Av}XEr zU#K@ND~rc^fD8k9=q#X;R#Xr-5+!9N$l)}!2l`Bo=0dYuIGRF8>~W5I0v+Y(X@nkT zNKOZBAE(U$+QSjb9NWxMHPB9uCL#qShuqMf#%|(hI%u9dN4%cXIE`^$z)=lQJx92X zMR|}Gp|Q9ZFRJyS_%e>3LJH)AP=>}s(h{DQlGw4FRs}SgA#pQ82Xn+qI)owNW+2{f zh-VR+%0tly#5)<1z6MI*G?X&d%uyUrJV&U7*cgsb=kaK5z^?$bUJQ}buK3uH*xo>B zSz~*~hN1+ezEBaPn>a30DF?JbmO?^_Vs)HfIFP~-T0K}V3zZ0oK^cL9Kx<@3K>HQ9 zmZM27Ys^802;#)-XGo$zdt1?NhQv)s!8MWfgE5>ln+wN4HVDlI+Q^WQ2-L)B_X4eB zh|Qy8)^J)h&}trUD$q)fAcL6YJQR8rvxK9EfEF?&$;daD(?)~F>r$)&;&mxbaOK9# zj6wS%OaiTj#|wi_9^+^z&=H1Mi>^P!X?2LlbzFQKXe18}1scvnPaxhVjpC}p3344LNjvNtd7!?_;ikHVq2_e3qh-i zJ!@rcAwyQHFEqx=+QQB1{MgXAO&m>$Jqn?KruihcQyiNeZr1FH72^^-DCTTzBhdTm z_}I0vO`Ntec3rH8b8gH&JT<}ipEgHxeFx{Vn5|ayAe@GxdJuXwAUTpYM?sf?(tu8L z6bp2cBMZ=Rh9uk<`aB=v@}===ZtMUv=8ix$vFVnB4AJJ;WEM|+Qk@cOk4QOZq}D@G%9Ah zehx#LPog)*ZsTZA^x9aA3&csZ*<+8=UubhAmy7f_ATAerD-f3pwE=nLavRV=R#F-b zOR;7KQKHgQ`Kr{FG=^^mi)1ZAF|9SB-Q#Gbxs0PJF`;_s8mHBTA5T;h1ytw9$vmI*xL&(?Le5&Z;$0(~Qf#Xp)KP)mfz7#3gxFEsGj%4n&IHv^j!n zle7xRqfK3axYXoVP~_TMfjehBs@?5nO6`qt+b3r6zv^w2R4rqa9q5 zy`b$`3<-AwrE_#MlO$%w=zW0j6$%eB9W)%}XjF6zT1^%@B{~BA9*<|*Y=rDt3R9yA zIteu2v^pjVclj_l%RINqKmfSQ-@+=m9Z4SmaXXUU0`lDXyFea&2?FAGE-}RKoaZsf zxRK|PYbr26ro#IO<@YA)f%v^4UYSX6gq;gpAp4DM?!ZwKpF4mS8Da?Lv{6y3qIYmK zC29qF7#=U;oB?{pQbe3IrgAht;s_fpOQUIXIC>?Z@j$Z}!nbXj!+FnuZ`3r0S92Qb zBYZkX>wwTlBc2whq7{`f#Pg^%9Oo!EVtPEY%W}j+e~diBfzZeB(6We;tQMqcxFZ>J z3m{mA+Ck1Q9>`;5FGCu84UFH*>XgXc z(U6)XtD_=!MXlwiCUS?t=|SNqFiT^hHIW-JD`jX(WK;A^FKwNnD@VEUz2h++Mrc`l zH})KlrHI!V6ozPXJPH65q}GJj#-H+{>6l9~T5fnbn;r1b@t7SlG(UVK8!ZD(i5wE$ zlOfF~;R7NEakM8qCvpG}t%)p(hfX5isK~s?GhSL2W_XyZ)8?>XmaiNN)RiGR42ZY) z6zx@*j?;z%DIARf5*d;{q0N0=90jRmVSD38AZ-(9Tu;SX$o@!t2xMQYPKh|g<{e@! z?Ck(gu?%P*Lvm-JJq(G@B1LoD5f6%sj~nYnVR5Vs;EZ!>2JXpM%zDwe| zaWubgeT3dai-^LIl!r!y={@{LMTADMzVeIshBqFd$D-(0K}WikT?TnnH|d{J_9AFjVSfdViv}hdC^=pwiM%);P6R3_A{i8 zh^Y==1Z1J-!bkU5#%T%RBf{%F@nXjGSiosNiz$upgpPw8S+w8$7e?{VVk$1#0+Fg=Z@Ux#<=!8B)(TpAwKV-t@j#>hROuJNJ)#&wnKy(T2O3Aug&hq($!Q5;$HESK;>9csKFVo7iR@IcERU7J&;y{zm?5mk5g$P5OM*9W^mW*r zVAQnmB5akn#o{96-LRUlB?uj*1c%HCW^x_{evbw-xiH$a;9(vVHa?i?K@L)k4VwUT zk6aoW#W~-Dd`AQqF&h0iba!wGL*OzfcmPMip__t(y=Yx<7)Ozzt76!_-9x_)Gj~75 zkovRG`f#QP_oyR6XN9%!c)?*kJ)w#iTG9QShZZ`aJ7gu)D?w3Jak!A77Q z(9Q-&0L=#NSa2^!Q{U=)q&wsy)B_#r4*S5H{ov@K-K)K5K=-*ENqx6>U*M%BcSo(V zczri?XZL$A{Wv zRuFMdDQ-j&sWpgu6mY>55w}zkkswkRE-G~)iik@Qky=o+R;^Xs5fLl3RLSrCd~Q&* zwqNb{|9kyj|JQlVJ5Ofj%$b?znVILDd+uDWae015i*!l8my$Zal!sQyx3TluJX&t~ zIzt5|H=^4v8o(%rcW zly76%g52GyH{~u;>ON)l_N2?Ky<$f05|h$SE}K?SCuu_2mF*_EoT?!cIJ#OIcUbDy?f=XYLU9+ zSPeRpFDGXcQU=$^%q^yrQAp{z38aFw))n!J97!2_?W*9qXsdPmFWXZxQby^b_FR{~ z_x3t5XBN`6L3R5rIWH-$V6Up2F(xtB_0Ku2C9YS_xE85PPPZ1RT~1#~`|MSq6$kyc zzc;5-Dz*KRoHCPQpS53rk+x=UZjm-;(=V3Vep>e1Ez-H! zuS(jd?X#NOt-RB+7ZcZq@)qT^LpqGOW3%adzcTK1oRf{c_yb6-&%v6AE9iJpHaa({ zeeY~EBdKRLnnCj0cgwD)l#!rci0fv?opx`vdyly3X+t``*K*V3O7kKMzU1yq9EVI8phEiMz-o zFNr@?)o}vS&BRT%IP(rW_EzeB#EqB4eRl`l*#uUXF&5`$B8^h&jYz{x@?S=(l*D~S z%G>14p$)~0l#e_AcHNYU@7P8y_qLK+r_?`k4YkNS4V~|uWwXF(^nKf`1xO`9b@`WB z+iV12CB<3T3sOPHMAk(nkuRP#jeN7ow<&84662%z$ShkCvB~1ftg$A=HWv5ILg#*e z=Fa&lnK!vQ+>&(@sa-jDL6%w5Ye@65%$g=6HCW01Fr*u@CKLAwXIzkV3et4quE;tb zsXKAUTgkyOL3PpVS(U_{OzPw;8)cSSmemvK)L7fZy7KXq$4p%$Kg>MZa`mY<1u501 zS0JfRz4Mu~2dIbe3d;L_Bk37{$CHHeUWy}Mc@LA^XJT)acP-~??kBOgicTzNOuIM7 zJ}K|gt{pY-+lcFBwUm|sh|EWAtjy=1Na&AsJ zd72BeipZs_m8Kcp?+3i;rmhG}0a&IjQ%4nPWu-h*yqm1?3y}-S-bVun|NK4&*g_&iWkh*xC zOShJ?HoA|x{Ytl4$?QqJt8`0QHPTw|fYJ|2w_3jt_fgp}>le3Q!K-DraK_`D@p{=D zq~8&@w(JU|&BU!Q!=BxhNNdZ^K^lVeY}rJlxk#&3@+SBB(v_u8x8z%8CA;@g%B->h z)-U9HpsY7>w-9$xSy!ax#4ReTK%$kU3(AU+UZGa=N|z%&gVa!ZH_{78HQORd-Hk!F`#uU&vNuhe>NE>c73BBbk)rj}aUu8M6a9WP5{UMoFO61H1< ztVyvW$v3u?w#6nQ9a(B^yMWXY%EwqJtuiU?d{U1nrEPxqxK~<9x*z9mF_jwUX-}!1 z_V|ay4k#T^%6Q{@4jb|{|MA}ThSGyeza-Ae)A)#uLDKl(9qtYIi&htAxY7zqw0i zB%Mv%Ic-0pysMEWD=r)93`tj#?*d8O+qadrVmBeBv}M;Ft5jyBwnv&2e;rBMik-u` zseTzqTz}=eoH(r?dJdjW9KZWiXRWa-rCN=C+7>AHEaLKZpVv0k)|sRRisOv72&r>( zw@V^5ck4ION0O*p$(xcWujDmJJOe4&Bnc}lN%cF=KuXqIoIQWfU5*8%q`OOAl~kF# zpzVMb>829KuE!Iw+p5;8_uXN8HE%ebt<79_Jk=nql8%tsWR6qVeYdhjX7Vc_VsC% zZF`iorQ~b4?yQm=NkdBNN_ryAA?~7LoBM7eU%O)FG@gM}=FC?r&o**yuu^zZQJFKR zq^w1{qL^9L|BTd&bd43HZY#0gHR-)#<_DfdRF-Ti-qIqiE2h_Yc2HS7shGLXo5AWo zK_$~0#aoMQ?t7YZhZWo0_cSFBP{}+AC?1eDmorT2oi@cJ&h45usYR+tV;=Fk1=Ynx zX={-d#OsUmii=H(6&0tq%|kjhwjnw)&GcJIs@LW#lVU|}zAV{+#65EMinIcgVjHrT zrCAS|v^Xu(q@X%`o?Vwa)ERBo#di{S5;b@vzKt0k4Yql@*w#RkRH}DN0)iirH-vZ*!gi@yPEjjKpkT%EB zKGMcEuf@%B_-vugrubyy3gWM~+0cgGkhG?a^#=88(}(XcAqCZKdc+?-yT>*pcJd7>d>5o`_U$S>;SEJQV%Bn#?0bJmcsp;$k44&- zxHtJ83KDTI#cgcyyra!IvDaFpiLpma;=0u_Yq?2dd5g_@ZAfB$5w*gndp+@V5$lPh z6-5ikcWGK>qN}CaSSapTG@rPw#8nm1SN<#X_!ULg8y6s5UQ|aMqoU}HqNzxXg{&P# zW+iJmqixXyBt~+ZE=5-7)kx!tMv?C{q%lR*+3YWCZBZXmkEMR+#%Q_UA8ASqUApKp z%?YgK*1HpeElAi^)?0zKoW5!^Hh2|jG}4j5dL*8_wHXmSjg(BQEPSo-Eo&Qb;{sZ4 zrDWY5ET-gsu?>Z5f`v$|6NOI)4P2M<3ReZQiTe}MW5Eoh%Cs4Uj|3Nx%JcWa<*Ctw zBwO=VQC^PX0;Eih9xJaw5^GGMMzUE%;hccpb?-r{56sp&BVENix%L!cV&RNJ`jnbZ z%<5H0pL)!(g{g7Qr&5K}RNiAq=cc5zS%qg5GCsV=iJPRjXZ+_2Cl*d7?sQ5%Bj|~A zDbi^Hw&tCWG|`^RdFRIq3ablGLHZSO;{&rbtKW$%0=_$#^0XFt9fRu3L4{*UJszpQ zPmlX~NPT?k@r$TI-@@@GamK-_!F|N_DjY!^Yd~QS#Z4!!YvC~B{)p76up1KoN1?n9 zkCm^mo$@`5WQ&hk*sQ{ig+)kHiOW+e>q06OOPtzzWueW9ucS>a_=f{zNY9BgfB!FvT}In13rNwije5icm%?A>irP+jnr&s<{nErpIZFC`Uh z@X(Ccm3-^H%aOVxt@WmH#y|K~gGt^?NN;fNW8PY>L65iY<*h>MPrl{eIL;WY)O(S5 zDpqidhplBvQ}yZ2dSj?qzKEY<9#i zD>%&ykwzm;aCaenf>iDP%~FHvj7NB5lh1Ey<&o+&^qXyBP^9$(ZCazxanBO>|$R_PJb* zi;;?sGLDzD0%@E{mYS~~_p^{TEA>pIw@mVXL%v=nIiJM+{0**I>;aU#w>uT715$-M zDK&mG*>hc;?FMil%yrsJ?g0~Sqj;S78}VkCuJ6t#oNC1^fCWxl zlN}zDan?BTT$t^Iir-H$%I|EF{DRTDMXAc~tdTrK90s!-SM(H9DfSb)imSzSBDZi{ z-D;lDcit23(I&gKM(-qXGW7NRq=Y+Hsh=zUE~C>O#`LZEgvVsgxvdp{us9p$IRnMM zFxPQJt6`y&Cb$azkG zm~gDjY{z;g+p+#HP&oxossz(}uIg5((sI?ex!S8CSKn95b#lm)=M;<94|z(>b2{r> z>+M{9n=M~GpR0S2ycX?SYJtirZ~|oJbkVGc`CBww%5RB}At$ttN4A*fbW@%-qFH91 za|p~;D{>u;S?4{)`4p1nU1Q_!LvYj48+v+^7ZTd8#k=Qzc`Z1h%% zw<|_7h9|4=hAMusxIomb;tfO2b~;*&^S9brZ7NJf{P4UW;XRm%KmBmPKW26y&$q zbEci;YD{@`k{=cy5*Hbrly%w2%4@L-J1bW@@|^&=K=#kK8)|>65?9j$F{)eTyeT~k!l1)^;ht~#-8n|@m*PO!s$~wbnIYk+ z-(0OzZWA%BHCH6ul(nB_d7SmiSs;0<(M|QkBZ~P-F;7UY6}u?r4)K1)TwwHWlYEVM zB+PIE$EtBv9rCtIv zHD_i>%bBTJ#k*9gk=u z&^%lq%PCOL7f5$`S`G8mKM5z*S0`{*fn)kAaI81-v>#NS*6YHSx@n$q?j$u|eV(iQ z30)=8vQF)#bIm66w8G|TXQyn{A-m=3HpcTb2j)qGc}}1h(|Ddr&TGlpj+lHWP<$aw zNKQCrQ3YE23MA*L6?u+nHn(N|>8AXjh&PKi2PT}WV1~|0IL|5ORHHXa@<{P5qdP|O z46%=RohUsz*P869PrUJx4-!X<(v)|(PdH|V$3QMUTaZx_EZPY=HZ0b2y|7wW8@F;a($NE&dv^&xgtKA`&Vm;n+Hoso5am zoT5~n>!d8D{B$l#bdfzD(>W4in2^Yx!)s(t_V|e`nU7;pD(<$9ghbb8Nz`LQc4g z#k0jTjLz%gU&M{#YvNMax~1Q&J{j_AqLlCCq_lINa{gX?Kpbjxw}_vKTNN)^Uqnwh zzf_Fs;2bS^jB?Hvr77=9$pgi8VydsQO?KWCH;8`{-w^*OzA7#gRf2XqOlY@{1e+aM z3G>ATqU_nZMe^;?myN~Lw+ZKGiXRTMTkb?sGmLCc_EVtVF3?$ddj627JG(@S->VtJ zy?_#OQ={IKjkyP^mfUh zDaO{gjMQxAOjXPw;v#XSD4lvq$#cXr#Z*6ZmYlN5k0hTi{sCsB>hH?Zyona?9cy$g zHBa`R*Rn=@t@u|I-wOIIW2CLcJ0FOeb5i$E&KHUqZS>9&JyG`Koh5mIxLPzzNO)sm zrc)`oMto8{LQKtIDN7!q7_;rxsa3$Uc_wU`!DNl@cI7!5W;#P89|bdXPmz`CRZl(R zEmi7Z@pn1r>zUtsiu^7kNp75@M zv6Kyamnr9qN^MfAddV$NyxCiJ>MG8~is=VqP9XU`qodw%drDT_+$ECDwlh}2lMlL%HKcI*H}!7 z-%tDA|CJF(foReXAL(ljdSLI>*x8%kYU;cTf9YGJ%5Y*h@!z-m-=v)TZ{mMYAK|h+bMxQp zI;<`!>3iJWo*(!3d-m}!w)H=yGb=93ofa-j9~bUSUz1f8?#v5p>?GFP>_DFr8?6?p zF><8^JHnm2+eH~$cFWNYTJ>XUVP&Z0;WX0=C5ZO7CA^-JT8-ga(qm z`CHB~4eoiEZd3E2=E2kqXtucfFukDT-CCwsm!;BdbnYIXj8$@w=R0Rf6FR>|vl^HG z*%~NynU&OH?K>mtEz1g1dgTf&)_l4E*ekSvI z_npn-M~>rcXe}MF%?WHzfsvEEPvAM;RyZvlgO@mga(7bhPSpRTk==C8Y9|P?nu|tu z4GwKij@(-{J2Bda8Xf5b)~*q+2Qg~fH;5apjU(RXI<1=LjqK*-Hs3L_msi|;_Q*b7 zQ}eSU2avMbaYk(OuY~JjA2laOY={TVO(QltF{Nx$$^%MyBz_(#e~V8uIx+W6{(kAc zCH_@>+nBbidD)0v_}|YtgGX$z7LJ%>EgW%^wQ$5d`iOWZuG)69+K!lQwH+~Awe7`K z>PI}yRUn_k?A2T|Vv%zoyxZvwmpKQ+hnz#;V@@Bq+KIdKIDd(Io00yh9xw&pcMeKF) z4G}#IKhg9sd~({C785=%t`q-YbTYX1;NhFOw()UsrMOPCn)k4p58q%lAHL3NK74~} z-j~{NRye`vwD#U?t{c9<|BLvR_*e05WALH)kuk;xu#iV?5ZfAdcX+MI;T&W1tdaZN zS&lPwr`y};{a#!n{z2RzGImLQNqpJpryJQD8HT4CqeG3TkLr<|@M5Y%Zy$I>WVdJeAA-yLh3qnh1C^O*P$z{u8_Jy>RPqZ>N<3V)ph6!tLxAe+(FNBMpw5*%m%7HG8?GcW;RfDyxG7|&un0*$5RPxv}&cz2vxnzQmabC1!}Njo}8PlkeT{3?A~L_kj4I z_^|khxI%0aSrf_mgt$s%eI({-@fq=1@pG7M~HH6`vPhFj5OhEsQ6Nr-<{6PJyl4gYQNc9V%Z$ z7si?5mEtV%e(`bf9dVoZnfST5L;PCYWkeT7Ukr@s!k8hpHlhniU#w}KS2@moPFyQK z-&|jLq`OXhL0m8XL3~mCqcP}g`GbSR(Z;Z^*w0eK{^DWA=vs@9J{Gr#pNLzH)Pi+$ z@Dy7&;fCfFgEzSw#h1jF#ZBTXMq674U&PvKGHdJL%WZ8PJT)9(G2uXQnK7CvHi+}Y zJH&0`XX19FQ|P|h+;qfJ_fO($;-AIW#WzH}4f6a&#OFZ9=YaSe5DyS;7T*=$6F(NW zh@Tomd^vcyc!V)xeuT`AM=Yf-`OP(l&vkDV=ZUw8^Tpf61>zmzLa|ZAKjd8eLwJ`l zz(0Y9iG##3#wk z8L5l$FwuIUe=x>KT_ANa_B2u#<8<+QBfrvec60UNV;C>Snc|h=Eb)HvagkMmJln+2 z#LvYY;@9FXBjd&Bi-CAvbJ3uO@ckfThg8Onv9*!01DztXmqC}9y+G^*VlNPTf!GVg zULf`Yu@{KFKW7l^$Ky2QNMK{L#123=xS1F@PxmzdQ+tY*+9 zPLZvr11H$pJ8-qFy#rU%UrmE|qSfUV`pfv4__?@4{90uFA9M@- zW%R|sNFNz9#MVao2+~LS&D8@oVLQfo;%(x5@pf^6c!#)9Y!vSl7mIfpZQKu(H$89( zc4RS!iG##3Ml8tK(})Ebr;FDcRip6(#!(~VJn=ShzIeO1K)gd-C^m}RJ8>5GPVg=x zwJ;tg4id*0v1UkJj6IFi#W-EOUgYlUux(BovylP4%{m~~0kMt&z0qC6fRE6fai(~s zI7_@=d|Z4-+$Me|elC73?lPh^qb~+Vv}Vi@TZ>rPfWGLC_*A_Pn{V}k)C*Ft!{$@3 zWdC*4%ZUE~@gE@m1H^xT_zw{O0pdSE{0E5t0P!3koMc}Q*8ZpSX}jrt6p){D{i)a*ebJih^<3x9b)ScTZh;>#MU9U4zYEJtwU@b zV(SoFKWr7XGh*uyTR&`-**e75A+`>!b{S^zhplo-%nR@TnvKMMZ~Az)$auC8&lcj@ z!mpZ}`u8`V8gd=jY&4%5x}qm?Wn#E8WL!hWHN-dW|C;&6{a-WR7~&g4X1#u!%{T7< zn)$|%8KQrG^Sk@^H$S=mYfcCAllx6IKdA4Q<|p@?YTgjU8-jR45N`p&&jK#6O34QvIfym)>ux(}5N^{g%-JoNda%o6zOQA> z5L=6dB2T7>Z(~gR!Wf<=>gu7c9_s25{nz*HXn;6SREwe$B~KC6Khf_@b`GYu$_`yY zZyC{apT*vzlAjaTitmbOsL#E2&(!Au_Gh>JpNZSW&yDn-ahk|4iXmShUMOB9UMyZB zUMkj!mx-5))5RI$Oz{dM{byV*zAx(P^dIr`pD{yhEf$JJVjFRe(d?bdI z)=5Qk5Zyy`4>QFqF(GD)IbyDuC+3RNpvav(Wga9REFL2E5xEa1 zwXfJu>@OZBV(|xW@;8b*jG_8JRR4$S|4{uOs{cdve{`1RjII{%H9BJ*uetu7c=8FYlp;#og5sSqVv8`AtmWkzJJF&f3 zA$AZuihGHBi~ES3#Li+DabK~kxS!Zf>@M~Y_ZNGL2Z+7I1I6CrLE^#UAz~l#P_eJb zoi07hoi5}~7Y;Dm>VME!f1~6bMq9fNT5fAMRBzi#ebDhdWv)4BDR&w$O*Ki=b<=dM zP`V4HyHL6drMpnN3#Gg0EGs{{TD;fjoaEiv+|+xuc_O{X`S=vQSNr%B@H%m>$Q>0i zcn1*g0Ny0tEH;S05pNN>|056gfABVOzR3L_F$=^y#D!v`xJXoe{Kb;F3nc#%@wejL z;yvQM;(g*$F)1z+@!>cZuZnizRYANeh*t&ist#OXUKPZvf_PO2t}w3((gP5$3gT5k zyef!hAew=ARlUcVSJiu*c~uav3X;FK?xlLqHUARM6!9;S@h_p(ZLay3@EsBV5*hyz z;$K4iOGphNHS9gt{7Y!%&o%$D_gwQYA^s)AzwA9Xyd16$r;B#=)uCN|b$Eqnbyyu< zC0;FBeO8+f*n6D$fDkPoxWaruh@aYfu6csJ$F&+{Ii2gglIE9tHF&$kW~1+kz8Hux zF)pTwp%{s+#0;^um?>t72{BvD5p%^nakw}_93}otJW@PL94j6zjuVd;PY_QOPZF!e zpNr$gUx*XLlf_fS2aLgFu}-{9yj+|v&JbscSBO`NSBY1Pv&3t}dhyrdZ1Gxgj(D9o zSG-=lLA+7CNxWHHDkjBc;{D?9#D~O(#YeUNRHt{oYhxm8#OYtA#PVt}OSK`;km@F+Oi;BsjVzQ`MmQrOi zFC!}BlTpoQSmPsow**K%l^GRJY?8x^zpDDD?DU{hph0B z6&|v}Lsoc*hXwJiAifpEw}SXq5Z?;oTS0s)h;Ie)tsuS?94?L!M~OcZj}(s*$BIXb zs^s&;Y2x{! zdL~pqglc=Jwuir0>T||OHHlP{NOl;h%xIBP?-cJ6mx#X=?-A9{k;;$MpV3Cezhrdo zveDLcg*_M8Z@oPifDel36B&I%Gzn=dq`r_+A*DjJ2dOpW3J{IK=fxL9Ml)AwHri;0 zz8HuxF)q^gq|#zYi(xA^i{nK`Gcgmylf_d+da~;Z8_lpzWHcjRE>0I`h%?12#4E+C z#H&R{GiO~R){DOuXN%X0bHwY!x#IQW4I-nN5^fT27MF@iahZtr$-`)d4~b|W8SO(x zGsFfUHUO~!$Y_R)W{5pN>;YmAaJ7gk1pqP!?sQ8Ce!(Po4H#Z)2JmS&ofqWZ8_3>8>kmOvBHO zA-djgy*+z^=og-;_(|efqI7Meoc!mBjB;c~Ih?1n)Ni4B$;LYI>KPmB$j=#VtV7k) z#yYY}wy}ofD&NLB`8SF$8F^0e0HLqrIAb^t<=*#1{=0xIM4Dt<2h%YGsXF%v(Y)%+2p+DoaemmyyINtyytx2%y2&9xYCVyE_#)l#m*sD zyOnOO^Mreedx`U@d#ihwv(0_peaZRS-Q<4irn&$0irw~JTd&L==ymYMxkq@%ddIob zyc4|f?gidy-b}a7yVASbUEp2g)w_+}9Pc{!PH(<{xNmyzdY`**dtb1d$!Fd!zD)HGHbRWKU-^-r&PFU5exB#~1%4ZTbF##*^0NG4 z{*hike~drIJHj99AL9-7kMmFQhWgX|^Sz_}3;hedWBiN#>%C+7o%f~Q+5R&B0q<&m zqrcIs_uup1_kQi~@OQAUY#4;zoFF|&_pS>vf-G-tuy4@SyD8`vbYoxM{ezy~Z-QRI z!QQRGp~0cvf}npez`G+D6b$he1;c~k-rd2N;3)5&;8($~y!(QCf_uEB!F|E~UNU$v zSmP}Z-VHwR)&?I3pLpwoPlHdrKL(!%&EAHX7t8Vf63dI_^P4Dzu_A9fzdm2;{Vi4= zEBAK9Dq-Ztf%)6``vkOXKY|>AbY(}j7{{uik%TV!}~fmDK^R5Wxpfu zHQO)9`wsiuUFqvrf6IRN-2ZF%Vfdl{cEsyv{yR}Tiv0JYR#ArkQB)ii`&*)Y zqJ8{NqW)2Te`|Djbh!U%G&&mNZ;OtLj`O!iXGN3!&!Z{P6#t9p!l=&Q5lxS#`(H&f zqbvQdqpPD?ffHRH-57Y$%~3-Ti*AkP1!>V8(H%h)HAYK<^yr@G-XIYzkA4^AL@T1l zg1qR7=$W7}`hE0VP!_!(eHgTlwnW>5j=aabBiOgqpjLx|uKeEm&|p93J$n&s$&S*$OM^<}ZXEY_FB`m$JG7VFDmeOat8i}ic!x(A5|i-(AP#6v|{tS^i8WwE|2 z)|bWlvRHqRavm-oAr2NR#UWyqI8>B{`?6kN)*Hxr16gk%>kYkVYRfvh)>^#-!uK-L?`dIMQ+AnOfey@9MZko5+#-aytH$a;eZb;aL{ zYs8PmW@9WN=8Lk5nDid1MWI?0szsq%6skp`S`?~9q1qB^bc7lmp+-lj(GmX2N)FEz zYsH(yo5cq4H{va#tTcR7^5f!4VIk_loz4OU0zPOyq4$`h>SJA#YaJ%_xp33+1^J}IsipAw%IpAnxG*NV@Je3H&t ze-t-}8^ulHE8?r-pTrNukBn}Xm=LqY95GkS6AQ#bu}Ewq7K>zd&_Y(IO_YpgZoy9KVzG7E#Ke3zGUF;$5FZL7<5POLSiU)}Yi-(AP#6!itVn4CJ zc$hdqJWre^o-bY?$|l{5BwsAniI<6&i_=Bftt-2AWw);E)|K75vRikSa>{z$ddasK zy+Ps;;$X2-93obULq(P5sT_Xg+|E@w-mfKJE6x{h7Z->cDeR|d`886!MUpilyt^ge zBi<`&M0gqzo<@YH5#h62#J`JQivJLIivJY9GWsLMsiHK_H#{s}z3NLxzWUZz-}>rX zUw!M(R!;S>KgZ-CO$@~hv5UB`C|hMuS<5L~4SGwKWd^dqK(-eQRQ%zhY#@;KgDS-g z6NejP8b>je6H_^{K8ly7;w2`>n~iC*^E6#0O?IB9Yo+O0X(uX=`Z-PgoTh$GQ$MGv zhtswx&u5}~IE+jV(?zv6Y%Muc%n}o#W|&YjOsKgf%vXGY*hVZCRp+p+?^DbJqI!(S zH1tE{ik|3;F)=R6HX`*>q&X_mxQ@d1n@*43u z@qO_F@k23HPw6p|?pjq@e5;{G_g3>h*pGRTh}|J$caSXNr@=v&6~b+2R!O zm*P3%uf%i3T5+n#=%+O1dw9NxPLOAaGez`3%$4F*;??3c;vFKELLMvyE)wq)7mIg^ zOT^!bcZ>Ik_lnpT=Q4s}QbZ5P_lwva@`K`X@ps}w;=|%2;-lgUu}OSPd|X^9J|V6W zpA=V%Pl->9&xp^8YsKfqb>bhz4dO;|llY4Gs`w}I0})FsTxQ+}WNv`W4KPQ{74yUb zu}~}$+lb5)Npm>mYuy}~rM?6&QEA|uni-(B=#Ph^y;`!nQqHNN<4_bY(SSMa4UM@}- zWw+*ikW+SR-UqVm*1QkoS)#1hyboe-F`D-Qj}QlomEsVwN*pSxH1j^luX4=$KxSrw z*NXGS+rSerNc>pbB7P!n6+adKCVnCA5dSWIDgHy; zDgIOZ%IG_yD~=SWin0XrO30($Hm?L(eQsU}vijV-66D#UdfvPeVy+YCir0%bh&PHi zi8qT4;%`J*f`6;zdE#xNwCPKmzO?B}o4&N^OPjv5=}VjbVx4uDD9!rPtS`;_(yTAd z`qHc~&HB=;FU|V*+qvdj!BET)yNLUW8j0pxkw+uZd@JNbM2#==t%w;Y9xlql&9@@H zN*pE*H^zK%FHxnLhedoJQS~$rt8kflScS{X!-Be&d05E0qIp=zCyF15AB(ad^RP&j z<(P+s{F%7JXdV_!7u9O>u!zYNv&4j$Eov4u4~sncVu9F3EEZK~^RUQMDwc`mVmqEKyb$&X;_< zD60!)b)l@z{4(k$`w1VA{E^Z8GU$q)D4j+z$#GG3Wquhs)j#H)ArBVSQuEG;xk$WJ ztP?L2RsTpeH}8y`_2REZjjU*m}#Z+x&1LmC-E^9SNtP+PBy)^UDGDn$@ z1`ijpUu0?uv0^wx+I8m$-&k)ZPCy8f?lf|>eDdI21bHrbX=ZdxBRFS!m`mid(^F=g+j5WcTBHBQ{ zQoKsMTD(TQLtH3gHRN0*-YG5??-G}YzZLHm?-B16u{F+R4uDCKxeEDy5gSB)P+Ttl zPJBpwSbRi$R9qo8iI0hoiz~$^#8u*x;%f0J@oDiH@mX=L_`JAI{G+%*+$e4mUlCsw z|0I4OVvm`l%twRFE|A#;=7_mso>(9jibY}@kH5|Nn(xl}9@%f)tLd$B_7Aa)e@ z689GO5j%;U#V+E$VpnlLv76Xk>>=(i_7o2gdx-~%2Z;xZhlqW|L&d&gKe4}fm^eT@ zPn;&6FJ2(ZCe25q)fbC(;$`CH;&f4VYd#t|Ww+*|AcBj%lvM`OgiGvpf-^BQcxI%0a&GHBFPLXkyXx2W6cZ!T=;e&Xm z$Y{1bh;I-XWy$98aV^=fd3?x!7GD?N5dR{+CH_@>TYN{{EWRhcFMc3?C~8D_A4}dM zej;uaKNbHbej)A<|1N$h{zKd;{!{!)+y$$6CkpaTl+hOhF($^vG%*w-v6Yx1wiYwR zEYU`86>m5iZRA$*hNIC&ZWZrD8ApgC;V6HU;_>1M;)&u(Vzu~l zalH5oae{cVc#1d`j`B5v%(q3Bt(tF(EQ>YY7P(&hwK!Y6R-7Z6T~(QHm^sRPLukD* zD(E8aD{2NX-;j9C0OlJaYrNZYH{=20K=E+#2vH-;d_(dK6RqDy1(RV_P$ym{UM@}- zXNWegtAZ;;YhzVlZLA8cja9s%X|y(01=hx@z}i?9SR1PXYhzVlZLA8e6X%N8i#Lcj ziZ_Wji%Z3%xJTq)Xktm2(6l>i61y8W%lKF$1L>|ajW>LxJ~>_{9HNzCVnCAP`r(!s=&rk z74Lc(Z5&kvHjb(S8%I^a*GAs;G6rlB+M74NOg5{i;*Bq3D>x{YX(jXCm&Ig@8W+6x zMNHa9aFjg*g;wWL=AlEY?I`omq1AR2?}l0a&qb@}DD&1~x|kuh7BfY=;;1ko+I2_S z^H6Bl9cA7+ED+m>#bSxrRxB0E#B#Bn*k0@`b`kd#yNYAPqeL60qe2^}qe2^}qj;;0 zKH;r4<0<0lik~Rfh-Zjr!m4nRc$PR>oGRK_t_r7#=Zo{;pzt<0%KUeDySPBSLtH2} ziuZ{Rh_XZ8k+U*oX}lw6^6$mxL}|)={LE42<3rQ_sK^&><{TBpMYDuayi-RByi;eC zcFfZ!ez16?;?-BYYiD^bRQyHarDC0UnP}sBlzIJdmRK+TTD0*vinsNwoIAz4#3kZy z#e2jjM75N+^(_AyI4ZI>j^fQdi+M?WUojtuABw3SkY(C^L*^)Zo{Q(1a5vZ4ub47H zv3G2`JFR$ncbO1D>LCe2($I8K?(0p#oJXfwS#h@+E}39Z*jCkaE?mOQB87GPL7j7{v2IB z$B82cY8~%kB8Te3K=F=#6Dd=_JQFz`N%fSWPk1L<~I*UDYk7{%oyyw4u+BT$~S&ixM0^_NxEOeCb{GJfe#1?U^5BZ_nPIby4}f`D@x0 z`Ppjne|tm?KjtW%(X&r9!;ZwOKPLZw%9Ea%I-=R>nbGWwoqO(i+G_Hi|Fy_wPl?RM ztvlr$iA5%`PmAf*x>M_iBqa()MG2)?njPd$rN9;6NDHHHkJ-rm$+;%n&Y7vR+S>K7 z`xcGlJVbnpyvN@})R5aYx9xvw;r~93p4me$ZPrDdbXaXt()XBN`Ar*kx8*A$s_R%6p&Yc&=+-#rwUnK8ru{<|YR(^4Ylo0LU;%Z}sNBmAPMqU@%qqRX(12{wAe z*IF&Mc_*sKTHosF@HNu}+}%4ie>7xF(EnA@Pd?I*4c`mjiWRnSMrk+kbat z42Z_i>n%0O9~O1`F}-Q|x#mwiTF=!Xf+h_jj-#WC7b_@va%!is4UE7^1S z+UBI?Z*EX3yAa!FiMAJU^RtR4HMpm_7P224yAMC)j3&ou{62hi+b`1bY`0zePC9!G zb49Lw4br8}4M78?T3j9H)*|gk+y~9|oWb|rIjc^46ermyD~a?cQaf8ICV79P?pE>` zO0H#R5w7OF?TpqH{oS=-H_g$hG>f-SFxa=4{qJ0MwxiS<$LH#mwkxt~z;6sYC(+^t z$|a3lb>yleriowreVmqYw?$vPzZrR z{z>?nx&{H?#^2-oq+@yN$Wuq2I`Y(MSL9@HUvq7+y!lXe{HkXMlCquWnA@pc%CLxRdN;AtPM-4e@NUb5YhSVBTYuGvY`z5FPrr!F-3EAhk5ltu2 zIX2zg#LmWz>}%W@t|R<`{t4B~XE)dH)}!@wjQu`hRn7J4@sMwiB>f`FF`H|o?(B0$ z+iSQ&oq8qWDm7fChO2zjS6r!vE7fqN8m?61MC=*S@SR?IhrVEM)p)S$i*NLm{vUJg z*{7{+r&`6-4J^h`J?9UvF-HdZtKhVj=lVt zzG(e}wdwuCHDd3u!KziJ`sO(_@B-mmi#LsY+mqkgcvBjCgs@**6TA9;Ux)17*fInB zi$2j@6Pjy6b4_TjiF)|`6wEUjmE$9nd5)SSS+j`h*?EhRTl zaswqdP;vt$H&Ai|B{xuV10^?5aswqdP;vt$H&Ai|CEIG&K*5UHlG2lu zo}~07r6(ypN$KpBYCEOcKB=}#s_l_#JEUrL{0o2gSRq}!2zN69JGT8$Z8ucg3)TG; zD`UqR+yB&dKefG2ZRb;`@W)sk`vN7|4=5S1-{emk9c{idQh6%+&0N{r8T>EvCA)!p zJkMphJ)F`a4@@V=3@&#(QbD8R^PWD z^zg$r^j}}a{I(x`m15fRzqZspeTjc-wI%=6?C{-r`Uh9A@6Y&}HQKuA%^7Yh0&Ah$ znvltwXe+_@XW{S7c>lHSm2<@;S4?unBv(vw#iaLBXTv`;*QY$5XDhF3Ad) zWQ9wrz2)+=&SqVlL6}KkuPNK@D#rd?w%?WQcJ&=khmmW$s@k4cw&RugownPR?R8~4 zUD-ZY_y|q$rOrvV&z0?RWqVw4_3wB_wnvrI{zvsJ>$&atW&3^EZeNV%%XMu-!B!kTD1X4!sOPQ`ad|G)E+?q_848IQdSU~dB0djRY#GMIfuQfoJ#hj@G% zqBWfL%I@zI;1j|0ZVvmKv&OSe?KYr{i_h)>7 z*9qne`~~|kf8ei#w+SCLn_p0iR>q|LgIRs2Y-0?zF(!NgdA)38Ot_K1FE=OC?cSlg z%N`Z(R|NK_aM=^ZWk(bq(R15JK=61O)Z@9l$5VPQPGIi}k53{zKJ4<+39Sejgw}*i z0()3^2?D!Vct3~!|%>Uf?_223H z8|dd73D492|49eE*v&u5Uj9jUEnx$5@J7N*gqI1M2(J)cWu4(poRzqNmAHYGxPg_p zft9#{eaahHi5s|AZE)Wv;NP7@!6fkKc8>`;X=Yigo_E65H2O~ zQIYL%>t9ZoPMATMNw|VfjqW2XC%jLfOxw*ihIO(Ns}sXAV_01btFs+!V{;sCrkmek z?^VZkv1Q*?$3C#K{cCOa+Gqe_Ab~pBuC?s2>O{21_N#UJutwIh-q+%9)#7i};&0XB zZ`I;&)#7i};&0XBZ`I;&)#A<7;&0WW-zN0ignpaQV=a2DMUS=Uu@*hnqQ_eFSc@KO z(PJ%otVNHt=&=?()}qH+^jM1?YtdsZdaT9wueIkh=&^|vz7~J37Jsf5oi^dq)#A_9 z;?LEhU-Q0d(Q7Swt;N5q#lNdXuTA)Pwdl4M-PWSdCiK~aKAX^IlXEC7{mGuCzk&Oh zmk2KtutFaz^sz!8EA-zd@T9>1knj-!tMsu-{}aMi0@mrX7rM_*=>B#Bd!hS(BYZ*F zN%$w$Yfnlg zcc*AOQQAI~A^w2vL1{ZshWG@w`($`E0pGy(l;nF>>@e=~_J{2*&Ns0JyR2X?Z%w$o zGvV^agu8%n2Z47b+(rWL6S#{ByiveAd86$OyKh$1<^0cYsp;w9zpk#MH7~3r{kyBI zJ@xzUD$9Q4_~oOu+OFfTt+?z(Zab0NKIFCwx$Qx2JCFx`oWa3igh2$}pU|qzZsch^ zGYR)6^d#5{{R?L_Pg6#Jvr@A&xvkdGKJx2FN7MOxJ*)R%4;^{v$a|E)81@*$-n#_+ zB@che+fKk|@{c1-Bb-mTfN&wIjz+(1(vceDvX?4s4;YQlSj4+uQp zXq5HS0$;>qj-~M>Xq5HS0$;>qj-~M>Xq5HS0$; z>qj-~M-A&o4eLjZ%acf#Cz0+81fHe3Jdt!?B>a)EfhQgt2`>>|CTt?SLU^^gk~OE2 zHK&p_r;;_Nk~OE2HK&p_r;;_Nk~OE2HK&p_r;;_Nk~OE2HK&p_r;;_Nk~OE&-Nw`8 z&j{NIpA-H@_=2#5@OJ{gHs}6>u#;!d_N2daa}Cd?oAkuL7k>}pe}4aiwW^x6swNmr z+-Uyt3n0Oz+)MBqAgo!{tXVZ-KSF;3zkY*%QjdR9&pKAgI#$U#Rv8X%t_v#(%k(6t znl-H&f2E#vt%h~2nsu#`b*&Pgr5>N99-pP2HLiv=u7)+PhBdARpQRq3r5>N9-l;`1 zKl#}bnyEuGb!esz%_PxG63ryhOcKo`(M%G}B+)<(8mK`7HE5s)4b-548Z=OY25Qhi z4H~FH12t%%1`X7pff_VWg9d8QKn)tGVSR04eQjcWZSpq}ULpL6!1EcO`(yug!W)D) z3GWa#6PT@7f16l;n^=FFSbv*Xf16l;n^=FFSbv*Xf16l;n^=FFSbv*Xf16l;n^=FF zSbv*Xf16l;o6u?vTCGEqb!hV2=W3(*pJ!{V!A-2eP4NPrh8Gfw2yO5%Tb{Y0**Y|v zM6*dWn?$2EXtV~6*05eTh4qAA6J`@waap^YSi758yX(=gX}1pT)}!4Tv|GbE-o!fI z#5&%@I^Kka>(Fo=8m>dbb__NE=uYTC;F+AmGdbq~LNCIBgx&;vcYen*$*))@`3=jY z!>@Td`0)-s%dX~0_B2mA1F*Uucna}fT6qIj-hh=iFb5==1Cq=EN#=kgb3l?gAjuq% zWDZC&2PByTlFR`~c2ZCJ#ItjH63eN_avHE4dv3a?<=C^+`q(nOw>A7-M_AwN@EMWA zGm7rq?RMvGw>!J6C(~#*JFMIDO1snQ&gXaae12EYXGsp9B{}T7o@Cec|HIz7z}Zyo z5B#_HUi+NAXU3Su`m;$ zXQXbqa)lIEa$Q&U|GU@BF=uAZJP1Ah&V0W6{MK*1erxT0_FliW_g?clYp%b}itDdC z>}OV6k39Nja^85I^M>!0HFMTNir0*LP{KVZ(MQ?M_%i;H_5k0HFgo;)wgt$HRynd} za%9cq$ePL8gBQ`u-Qi;B0hhp~a2fQ35cGn}VXCZ>hxmRN9)W4_C`^aP0AEp_fEn;4 zz2~2cTFEM-BXTFW06GKK8(42#2;Bf13~Vs4!N3Ls8w_kPu))9v0~-u%FtEYs4Oc)P zxDxupRnQNvhW;=BlHeK`2-m_OxDE!x^)Li(fT3_B41=3sIM92>2)G4C!mTh0{sg1p zHW&jF;4V1KxqRa#^1KYMz^lNXF_{2+MzIYYm(p8K_S1}e{#aMI|W6@QNMOQHvUB$UN ziF0)l=jtTR#YvotlQt9w|epyMZk2jkCsyA1)Ql8O%8q>cMGHA5Mn` z5I(P;h1?RT*J%a(QXT4cI72!7Vja#=Uvm*0v0B|gRXVs!7oUm*qPJARo-{4$gHWhTj1*2Q0n50`8* zel{8J;>#uD%OzVTGNU<5feir&f&(rv{cqk7Ea1ta!1*ixYLl3wVE`y$M zIrN4rpbzwetAR1JH2{*}8W;%I!XUT~2E(J+HXS(E;9Dl+TPEXMCgWQs<69=baphKz!xJzAo`eOk5MGBx@CLjMT*VbvafJ_~@L|*vSPIMF zZ}2|+9hSofumV1Wm9Pp{!#A)Q81X1ZJZc9px=}yCPS^$ihQ06~*a!awMmNeI-QX#x zpBce$obbORl3Cq;DQnv=wK+Ce*M6zZalo4POIgu=sofSj@I1o>d|wFNk!e43?6=w2 z%{45JYgioDu(n*o+HwtR%Z%+rW^5-iV>^)<+llyQbL?-u_4sFV>}`B+_qqmbu4}vT z&wkYN>us6Ioybh?M6PjhT;t+cJAbJ&9Bu~o7i;D(b#8~Ta0g7{ddT$)Uq2aNKN(*? z8DBpcUq2aNKN(*?*`ZHaO@AqC=`Up^{iUpFCYMt$ob1v!K7EJjoG z1FP#Vd`rfA3^c`p1 zZcLC@Skw0k677vy^+ma@)o_OZ)4WRSNXkHukw4pUgh@z^D*-=<3ZN?oo6Jo*6(uTF>?hg{w`p} z-yGvz^KN^03K!Sw>=cse_8o=k9b5%#x$$UX|R^7}O^-91CR7l;-8o;-xk*xSTO5J6yRXj6h zZcww-Eb|LBTg^7VRL`h+=0DVYRtWxv6@oXJJ6R$4TkB+22==VHwrK~gbL?2VoOPa^ zU?*A~>`Hbe>jJx~UDfJrSGTKMUHo-|t*-t$!B#iBf!)Bm$X_AY>Tb8S+gcafo$bz6 z54)>

    >Hp*`91&WTj>HKem$WwRVa%(9X0ot-xZ|kqlm8|l+oK=2ruvR%YvC3~MtNcF9s=bf0%I{9+Iac{SpjY|z zoHv~}mEo`Qt4x2DUu8KfoR5^_ukWkkoll(m0wkfRelXs*|l9;RdpTL zRn`2pepL-O-Yu_cxmDaM>SVW?TV2(0>$&w*U4Qjob*jJmFYE8J`mZ|8?dkSX_5Ia< zRYQOEU)9K8{Z}<%_20YH8SdTgJ*ow(|K6w0cJF8P-&XD;?jx#=JDoLv+xpK&s`K?4 zz#a4&z@1nF_+@p0UIVzhyVzZHG_PIs3Y6jLdtk{TS-ET*}-Uat*&gRzj+2tuqy5Mm{Q5ObIz=IuhPK@egEf{=8B z-f#uK5)hWlX(JOKU^3;tR0Id~rC!aR5Z=EIBd z61)trz^kwTUW0{z{~?Qj{}3sQf&UK43Xvh!hYYbgWQa3xh?OBjtP2?u{!gSV1^!2* z@Z2S7<1DI(az)br9 zX4(fZ(>{Qi_5sYa4`8N!05k0am}wutO#1+{6EohOp$l|{3!xia1k}M>_95P}5AlY5 zi1+J5yj>qM=fE@YEO1QmPJM_s>O;IwAJQ}T97nuKAL2dw5O2|kOpYVopbzo>e2BN_ z3(xHP&nx;f`_b>4a}4Qc7z@tq=RVIEdH+28#EpL+o#!0$X7;=I&&owUdBc14A(LZ@ zcj`mDQ6J)c`jGi0`~$v*jqp#{1mD1B*aF|eR@esH;XBv?-@^~E6ZXJQuowOV`{2K@ zAN~gi^qqw+tj-v+#sK@Dx5PudBOc-n@euEahj=?Y#Jk}kRy_)_=26I^zgh1nq=Mi; z95jN)!12fX-yz=q4)N}Hh&R9eyA}g@_dCR$ivfybjCFxR+`$;2F5sK}%&I^k-un*m z)^|u<483@AnCA^_j(yf!46znah?RgsHv5ZJfI@jsDTP=CD8w2-Ayxni1u2)+fBZWk z1A_herayx8N09yq(jP(kf;WmoyiXho(iglx9OCWakb~{4^Alo~pO8}?5`f<@?+b@` zTR6nK!Xe%i4ms7JI#56B`Gi=_C*;(Ilc5fr0u6zFVEvvDtM`PsBQt#mL*~Oc$;@(^S_SP$6K1oQ!0EaX7&n~>8)R%TYsLy>y_Ep8}vSh)i>&{TEqAX)-ej{4U`2^*m`7<xnQSr=L;g8?1hx73!rgbQszVNYF zI9?>@@uIpyngJ)Peq!%^IxQ{B-v}7QtI{7<7u+uKGEAyT|leOI_dRVVsi3rY>-pak0o7Q{k<`# zfo|jajQ)|ia^)Nqsq?mFaXrcVRLp5g8EwKr{arh!Wf5J01o;7PT}}tCO~dmwxspD1 zinO6{s@#3zeUpPoy1n#Y;m-gRv_Gsae0+rUzxnScfAbdpG6TD0 zMZcCGVX6Lk?g+CUm28p6nSXx!we2}7jq(dAqyOZO(aQcwnOgR=M=9l`^yl+SN|}%^ zAE}gb?~&&G<@uuYUze0=zo^_gDGR-?QqrYP%6h*XeXXsN@)g&C8HdWPlQR2|oG0k9 z`RggqOU0Ce^9YnzByD&+RwpHYOi(A~EuHS9lqLGROUiQpJNLS;HF~_DuRR=N__nLD z@he!H8=Jq?Y5X>n^3iQ0jW;t*UvEoEUV3z1pB+j#Aa4w#`|pr8Pe|E}?T6Z5-0y}l zR+MW#MZ}~Wh@5ZGW4qd^br`E08uR7XnDekN>oJ5sR(d@}pCiNV*XbxPKmFlIsivRW zo1f~0bLsP*KOUS2bD0yVofBB4l6q(inx8teLB4z#68%>+ zb!z=cUF7+_fWCzDo_Aep3vW?sd-f&IeP-nL*Da|{BHM8=Gc6{6d``&?Qu}*JX`8&& zIjO9A5%r_Y|gZM#Oulq4~{>~4}w$!2en@zZny}79)!aBbU${ekKmlIRR zX+38@U4O(r@#4_=eg^%S`$?S~mI{CSZ8*3!b$VgFA3kIQxwXG0X$ihw+cDFZ^3&vo zqHn|HgwyLjo1QucUutk<4v+o|hkbv~{}^`aE4dPN<@IZ>-$$77lK=5$r}Dg4t{mPL z4(oIga}MEjT3_gVu74+aZH-JBh(zd*L*5-2z~ zbR<eFi_ESmMr-E(B z9a}}(!<=oQ{_|rC&f|vXyb9*~V~2_x3;dB=9sSr&&5Ufb9_QVdx*<~M%ifZ-9))#< z!`{r?`93eg+wo?mZY-QGGF%5GM^>^gBZep(P#P)D{Ao~3Y=I+}^mg4QrF5)+MJF>e+rYW4KU|gY)Kc5`+ zQ=C0f6PHQpjZWQCri8~ZQNjMp)_za)>Pm`>o+DE0C1+KQ4F6)e%aNCWFSg^zbN@Oy zkEH*?&zKyJ#nG>0hf~7QO?xD5IJy=bUTH_tf3)$3+*&gCglqHmX6_Hmg}=R&EL2&6 z!uqm=y|1#EWr*_8=e_*;qogQNmEVRElO2`FV%u<7r5}F(g~tir($syCRfYF?I6gAY zm%URNwAD%F?H}YFX{vC%P7^U+h#EgJPrHyZ3e_JieQ~1I6gj_FN}m+S7bOtwxf4HG!MP9S#r!$hLh~L`ii@HudEO{WhTpxaaNqrs z21WWZ(aYhPfO%n|MRy@Dx@FjTB>QCfLvUBfBD*P3ux zBqDBgt~T=dyVCZz_j+32@ce_{PepyFyfP2tS=Gq;^X)8~RN^z!OEM2#Z6dE@k#gZ$ zyxn>Bug{zx8RzANy}UBLfoZ+`wukkRX$!~^*C$`^_g{hd-%FrDIwOg^Pn+zGVet?1 zU1&f3(sp_4vziwzBjOpnqLGI*-cQ+!4h0TZD8|o?ZqMNs{>MrmK6ezpCKtHhPLsW* zX-Qt&v?*S%w85qIK;iSPcUyMe86`4jb`w7ou0JxSuwjZ9PSt^x2Cj*kG#e|LeBj5A6ft9_W}1i z=H!Q>Oa46(@5Agv&t-&DmF4*Te(BD!LZ!y@TQPHOFXO#Wd|x_|NVUw?ONP*&NEFder+@k!^@fU6Ne7 zT}QZ|N~$z3?r{1qT%5lyfVU*Q74KEI&PxBE&5zc z_&HZUHvf~}JS-owwMBZpTpjhd4aZWa|9m;`godwA`7uSMlGZkeOzpS9tC!x$*X3Ue z(Z&zNc1F z5xMJDq_g5>#_rsAggMq`rVotF<3AT(c)o!&L;N(w9q&3%;rBIar{5bcw?IBUE~oys z;aJM$*Q(#Uj?m@D7L@!p96XRds-QME#9ADAVxZ`u2bOP{OHKIqNO?o)D_!;g&| z2Oaz>*Ph6IS6qp?Z743bpx#SLpIuOMq@mnCJF?u_2M;S>cwM0`=}Qjf8+u48UI?cy z@V+L=!{tTZr^rvcoVDhGj7)ySa%uPb!flA`v&hP#^`Ebsl)n5> zvT&dHib(lTf;^$5xWZ|pgrmKe6D8RRRWUtqy!x;3*smCKil_@*lZwQiP?V(yBG;@b zDZR(LnMb;xN-FpF5a)fIzSg(XTbr&ORmSD(kEPw-H|boZ!=EEtH%I4Y-JIM~%Jv+1 zN$ci>tttLHL(j{F<0F0Iu&$)<$T3Pu>B<%txo%GJIS%=VQIhlt#m|DTD7Bt@7}*4UutW8e2?^%u$RS zZ9H5|(&MEr+;_jEUhnmc_TI{j&c3f3?niH3M(gmt3V;8iI+zD2{X_i`^A~zug05V% z^WJSMDZk%_lHw!dyyRle0g;KyAm{d384~(QBG#cQxj!=sy+c;U@=qkA2XZ)Cj^vZk z^GI^#=E>;eeP3`59dA>{h{*EzEln>FUY{{2G8~pUo<@h|BmFL1m(Su+dkN(y`HGv-9b zd#f|%MTQUh&$pdjQ0K>&>G`|7eBt}?Wyup+ccknk=f0oduRU4#i;N4)rIis*7y0d9$Gxl!{^JyL$y+lPNv(``eEHC|qE?1}R>r$7V?|l6`K6^P zEk4pVU*=!j7ds~(nYOfR_TknWS$1K$@V$e=@ns1g>faI)N7~HqZB007$@5Mb>%!@d z=l8*gu|-Ms9*w>$sr=tkT-3F6F);ESHodM_8P;2l)-L}ZS5bZbddo%Q{wSKGyQa?J z&cz(-Ui9rLi%lhub&s``$ESR6QKkylRMI^;-tmbEr#qhC2buAVeAd6Xx+Bl!#ijW@ z>v`uYA#8u{n%(~Mg_$+OI^+hv%sb_E^L4*e3O~ye{w_;-hx^=PS<3wXm!{Nr>dMe1 zjMrO+la)4p3#W_x=2&VI87@iwe{rr{l6izadPzO^3*%a3cKC%Q9JlP8!+uf!9r`~$ zzmfkdR8q(N7Jd&$`hSH=YFBA--nPt#y@>x$QCj-r6<^ZWCensuyWcWrMb>rv<;-Vu z*ObeA_L$e*rfm0okGUPcMg^rk!}M#|bfV5EJy3LA+bD@jR7IXI%HmU&SlBU${4Lsa z44Qe|iaZ+Q^5bUg@4TWvx=;Um-aX#v%$44h%+zojyp@@s_~EdQ?*)EbI2=xQINycy z5?-8F=DTGr2fN-cF0J2&;^K~#zHpnu_4xf4mZNhrdfs{hNucpQYao{-yRrUq|iNTi5NEw&z6t-&+r4 zwL4;qgIV_;ao!>sv!)hN6;Wgt`wU}|q=gj6^8ZK`s_I0rOg)g*jWy&iEmP7HiH<`L zWc5A{m7J)eNDpN7E226AMR9Wtkwy7)fJgG|LSz{yCo;R%Xzw_7I2$ytZ-{CTkEO*&*9qu!?ztpDxcfW_GSp&~vYf zk5<`bd3P*YUEaW~QHAy1u&l9#b$^ubFa16JQTpNdv?FW6@u{a^E#9W}rOe3`O#lBr z^c(SW!|#fD{{QpNd*{pBR&?)#b;bQ}+3^3QD4O$^X~H)BvQmEkxxEQl)0o%O>xmsa zkTtuw7I}TLo-Z!$g#Os~fVU)TVS!rOWxZ7(Ts!Nf0^#3aphz2jO&bgM->+%QF|OO& zlC`A7DuP+dON>8)Nbl3E)kl!6Ko;-Q9R1vfRLokNACbkIL3y86VIlmT6>(=hPggPP zt8i>_-yD~li;FuR`oe2r{o3{%k2W9EI(lYpKc?mX(F^~b_h00^%n2Xyw|lTg?x)|b zmA_5PyiM7ek(PNkX6^DuXYGyDl~s;5uNX~N*7ANo>9P*|B3rv;J34=t>=uVrPIx`J zP_|hhU%0#i%Dg~Hb#%#&&(BiII#JKdf`2V0p zs0xbyJ1}K=6S;7E9?bo}a0Ls?3;AV6rVsa@A6?QrHzmaV-mVMbdnI{2@~eLxpZ+WE zU9#-PhU`I+;e7e< zpQrOa%^nevUyp8b!`b8V^;|19WlxL@`!fIkaI#d)PA;rJ((qCJ|Acu)M84x0eeJTM z@3|h0aYS*?b{tmU49K2&Sm~oAF6$VCeK{ve_5v!>9Jl}XS)k(Ib)ew?`}|!uCBk^` zpAms$AbKAbJ@@m!QddN;Dtk?I{cl^V*Saj}7v3&=-EX@EMXf75=TbC9&f`UOzn;dM zpS|JNtNldHg?r$rTsI3BdKAJ(XCBsH-#2^XQ84E>Rm740Kegl4h_bG4biC~Ob?V7c z-j1A@f_3L4NK8)UL%IezHGP?MdUn#=nA4!Fp)4zPf2n4*8h5vx2=@_l~5_ChqkU# zPOqb7CG$glqn1@ECn>7FBxR+Xk%yuy&HyVZOVDLsK3clh47|G%C5 zPt@c7^_*wJ>2gy=-_zl0rGzzf#S|v|4XpDaDsBCBBqs@B5rD3n%fmZQ|dta|-PGEhYu>M3iw+FmW*ThnM3->R+P=ynn8*dc5mB8t=m%diJFH`bI}=e=&KzmFt@yahbo7jDLhS{6@O+ zu*~qjJFMh??8L5^6GalF6jNz+j6?D7-^@|MdMQ_vC+ER(F(q$&<*m&{xOM>xF z9qZd!MaPQxbye1TfsqAroNh&B-7|5zuK2vVd}|YA3EQCVJzF1p98+i&@!u4U`$UW(Io#Y;ytqK__9@7Z{L49Dv<38r6{NE5d4`uMSR&tn}LC$}x~ zo{l|5Z`*6Ve`J=^_KR*yIeo;&<=OTiu`#+WiK39>r6yZ)uuafyNifrW+4!0*$2;3{ zx+mkcoFIQfQ$-)=O7GuzxeQIbbY~lz_shBz*%ie$UV5A*HNw;#+o$O-05t$60>oQWKp+b8C`$mMmP#OSkSj6Rm)CHEJth;_7u z@p*f>9vacHj_#!d;~!{TU0S?!(Xlq$px$4OK7L|zEi|*xCzzRR%jx}FKCeI5YyD$- zo540F?>IRhxx8*;Z0^xz`oEeOeax5Bd4sxFW3+8ST}z_wpBUYrqR&q8ayGf#yrcJV zWXbz|wL%VvVr%NYiqZWOlXoQeX8>2%Zs;C}k!EO4(!Y)b^GW0wsmC_K9K_bqeNtZA zvGELeVt%V^HUm|*p}0lxVgVj^I9}9a<10<``Xa0uyx)9b2OSb-G}kg8CmlB z+27ZuF2~e0np&Tz>k?^9Y$YwnNONS-`&8*;+s-?}{bR4Zt}9j)a)Q~At*K=ZU*??0 z*42Gw>phZSEhj=9FKZifo)I}TSK?5ir%{V z4DaZkPcT;!n_zt9Z_RpaMIXuWx`Y_<&tufdHePbCFa9~u(fv?f{PSMCuD4uXY44!n zY^1mTH6pm-)mlzqTQ2v^VEE@ZS6@|KU2lSsh9*AmI@?6Y`fZ8H>zQ2H zzYf~E1XCX~G5TB`qkkiD`g~<;+0HG~d|Q`LpKZK8t}5w0DEgX`D1N`W+Conpkw{LlVG;c+b6Wfzfy}nuGs&`w(ccc+mm3nLKCZd$&~5r=_g71oUD-< z@`-Gh_l+1MQ@%E`%`1#o%q#h8WM0KzWAkeBIlw$>8WUUjpU zs*6<*YqjdBZnf5^KdC=i+tn;J%lb~uRKQf9`d-afZ&*L7H`OM~Q=9EX)y%GF zS5--Nb-Sjz&OXgPO%1Wnu$!qH>=t$lHOy{hw^BFR=i2SmaQl3_gBodfwl7qp?2GIk zYK%S99;(LK!|Y+|PJ6gLT#dJHv2Rg-wkO&TstNWJ_7iHVJ=1<#J!n5;KcgPCpR?zw z>GphkzM5gbWWS`Iv|qCqs+sm8`wcbAe#c&_X4`+W|E8X|Kd@J*x%L`6MZI8W+L`J# z`wROYYN7qL{k3|--ehl9i|sA;Hubi>!``djv;P~g)XG305T{ZC@qzLxJ5VW5L#+?g z4>VR^2hIpISN{r}6=<)v1v&;gshAdDFviCTPoj2`$&U?;# z_I_uXv)ullv%>i(AkN25N+95*IT?X+&NgRzAi+8491J8f;EW4YaO2%dfjVw=w|by~ zTidN2Xy~5qo*roAwsczt8oTY?_JJnu1?~lbGu$q2mq1gur`s#g%)QFJD$v61@AeOz z<=*U$2()xZx+4Rv+&kTS0_V8*xl;oj+=twU0$trl+(!Zzx{ta~1iHCTx-$cpxKFvW z1DCnaxz7bc?p*ifKri=IcR}DPcd@%TaJBoE`);7WyTn}oJ6JnahG!NVr+bOnF;O^LNu@?mIpdJtatHFA{26kuJivA;e>LSn{tS7D->2gDxf=N+ex+6Tm99npOj424Bm+59 zvXHaobL21ht0J4_Uu?JXXUI1Gs_@I+E(X7CMyN)4qat!8qmq<2DjR1bw=!BuP2(J+ z19C@Wj40!FE=_UoMX;I^Md(;IOcrwMG2TMnTyc8VZJFRnQxiz$f@RE&3BRCGv6b2iMb5(s~Pj zBcHF%NA93HB6m`qkS|c3k-Mlal+#spMZQp7NZxLWU!3Z$E@s<9^*|F+AxiG0dXejL zb+edigc>21x<%b0SE-R|q_k7Fs#_&ajZ&kewttO59<3%I-=*%78`VVhB8NM--3 z!)LG!WtSO%(3 z)IX)1+N3r~jQU1>BbTYoY71$;Ro{|+tJ*4!)i$+VPFCNk@5r@7{hQd`YB#Yzs-KA6 ztA0jvP#r|$DNjsWY(q}AO`Fl7&7U~7vXz8v+YU(34%!@Pc8nb(0Xx=?m6Pl^8=t|B zx8tR;UEVHFY=WIYIVagCAt&025@%PiIVSB&c4hQc>}sU1ZdaEWyM|pudf7GYnsSzZ zeU(c7^;PQmS61Y6?KW)N+HIwc-Og?&UAfAhFBjPz><)6CeSv)e`p$M|8Dw{{yU6AC zh4zJLy4l@Gd69h)DZAU-S9QWdFtf3-T;`7V=Z} zQ_{+wZO@i-Y_4p`T-}kMv!6qL-hLi=u05AFzF@yV?JwFdvVGZpneA)#YjVB4&|XNZ z7uk#CQu_`24eEWzUP8S~?WNMw9|55Gz+Qo7jlD)%*dN;;%Nh13_NUU+UTd#K^O^k_ znss)HH1Nj<$czulZK{S75-wl|Z0i@ilU*x%aUN=JLEy%l+zy$yMXy^~h& zvj0uW-S%$k`qBOod5`@Q`n~pE0_|jI;Jmu;L5-)5*rv9xJ{}CZV!x=_JMJMameEX5Ii7f1b+_xjC?S7P|jckW=aFca+q0gT!(o&C)UBQbK;yt~PC>5gG(g|bX@Y!)(_AVzXF4sUymOY*R$Ql@(@ttR=Q-y~U8jT7 zQ7Sr}oX!&ObaA>$bw+z32|B%;%gNi@=_jUhwR5#Jar!&`kq0=}pdaWAl#`rmok4P{ zbDc97`FiJisp$-HhDgAVL8Ql#F-{1JC8b#p?TbSLTWfO9Ip7zY-cv*Kkdw=%-5XP*e-My($d$R*JE>2&E3joWEEoIZU@7m9gGtlY zZ7Vey4WBQj+rjNfUPi;{d%BDb+>jfR3*26AFEosdiDhKW_Gb5HvE30aeeREs(cI~d zC(T{%-E1ehld$t1_a17w&%IA7x|7|>=%=`hxZDTa2gp0sor;|7GVbz6(o)Nv=1wEm zbay)2$K1!r`?&jr#51;jk~A~jnUwRC`xLRW-Py>Dv)RsbU%-a>?tE-`(R~^DRrgi4 z3)}_N!pK`<7)v+XCe1tUU(qaem!bK)%bs#Sa6d#|<*q_r?XH%S-H+Ul(5!Jk zre4PAq|b0OiT%R;0{yox`@;Rs{Q;YIx{SfxUG8oyVeBr3KX#Wo{@7ir`D1r+7`u0q zO0gHkG8Sb-S%(pPJ4W#5F^&)FaeO7l@gb=#z4%j%=r2d^E&Y)PNRm{RYZ%`L8Qo7| z`v7D5mW=6Vu)52W{8f=xWdSGI*Z35+XEJ+i+@kL;^3vS&XTe_?!INssRHD!3uG;KhOk!0B^@!bI22LwI|R-djc+=KsK>C_yEvO<*83vZzw-ooXS(;J^5#=6qFk`nq_ebM;- zgQ@+880|kKX#b%S{=*&AIL;bJnmesKk;hx(7jZX3-(Ds$EU3(Yx@h-lh{LR+CB-YwuZNbiO@i$_%zfn*78?n{_ z%VTWKk4sKd2EIpS=87Cf%)al@JkR%tRk0Gx^F4g8qY++5BdMest0tuMeUJ0C?@?3x z9_{fxT1ZuO79L2YJm16jI^rU{j`Q(1IK!%rcpB$xPs3213wakw%t(d3i))yd>M8BD zkI@JpgR7(JjgL`Y`xuq9k8y+cF$QTLBSCu?hV~_DXkVg+_9bd)U!sQgC2X}oEs#4S zJPHF};w}7yd@sV#Uc@chi#SPp5wZ9WnX*V_C~sBZ=|R8L;j@wkS=yt{E%+i1L>kYke=EDxlMZ@H~I5| zjM14FWXrrDTi*jo)E-C$?SUj}52TLvKoYeFQbBtliP{6HqkWIA+V{9Z`yO4j?{S6p zJvwRM;{xq_bke@Z1-9>dbke@Z1={y$W6!haQSW?vK3nD*+4{c680~vBw0+;Bx%NF0 z@jc$dXIO#5)J&@Mg136cFAnmjV(pq~UXK4?lwe~=o`7@Mw z9h>k&nrc5JQTrk1YCoix_Cqe$en>Cvhg`1xkP6xlNz{Hw9qorCYCoic_Cpf2A5ua4 zA&J@#siXanMD2%Ezz?zTMtnb{v-U$eYCoio_Cxw=KctQJL;7kzq_g%zI%+?pt@cA2 zYCq&=Jdo4zPZ|Uo$fbdX_#tO&KV*pZLqd2UO&Q-e3pA4(1I_V3F2(m~NgChl2x+fl zr1m;SX|Lntc>>OVx@;a7t7Z7 zL;7exMa7W+{WZx&LroEAB+8e2+{gA=T+H&_J=mxnX z%G_-o82jb~DhQFy#_5?(|iUP?>trPS5_NiC;_Q$wmcHJw_F^}}9DLuu?Z!dI!H zeU*yZS1E_DaxNZ18>fw&>9oafakSskNc$}fwBJ%w`z>+WZ#h-_Ehc^oBMs+D{1!v| zE%mhDa*Fm_OzpS0+HYxs-*P>^P}pO+0nf+xSWG;YG331+k0nNXEC~@Fi={o5SnaW# zu059e+GB~;9!q8Iu~^z;aq(E5mU_-}_$>kLw*{)l_Wc><&+bbl?bVdiUQI=RcAs%Mv-_0k`!fmJpQ*0>8C&}^W`sYJ5aG|5+MkKR zpP3-#@MrGA^1JbB8f&knw)SeuX|JZHd%t@>wfJ65ZM>SP_#50Az!Kk|X~dm@N2R^? zXKHJIrkwU?;&2h!H=43mNU zCkLP_5I%Xm$SLoOoH`Atqux3m8X|9f^ru$@>^OZ2Q10optpWYifPQH}+J>!REW8W* zcxZ`mBf^acHzM4)C5#68xG{a&gg$RF3Kqi-_F*F!39rC*k*0*34gnAmhph?EguGKY1IiP0rj>f{#-|-?F`5k zX;%>_vmJfVE>+|_>N~He$oa!zzDTDLun>L_xu7Xv;|17w0XBBV#?I8)dA&%Ns?Zl6 zhjk)d;{ltxZWQT8zHXG)?In?m*caWG0JdC=T^CdD#T!I=Gza!ck1s_oLB0fgFUb_S zv;j}g>SRvAfa38{bD6h{#!2T;+!5H8;=u5qQhrlwCs}ceED(dYQ1KnXNd?j*q zOCav*`6B)61MTn6KJ34nNhWlHNw7lX9>Vt!zK8HVewecEC44X8dkNo1J@<8hi9mi#kjaE6 z6P~;r_KV!#1~BOUB|vxz;VH?mR^)*=!1f0SKY(phW1t<-H&ee9d9Xgvjt6OHGX0xO z8;Ed!FAU@lQn{u&2_|2h>2zf1Vtu0R{!T>|?>-fIK2<2~B( z-X4)9tzZni1G_|))q(9Ie`^FoME>3rsPFICx}4*0d0)WxPl%Ul7`)B!{J$xj70tuFW*@l57||r7jPV86ZTho z$)RmI*qXCOWW9pkFb$~dbL#%QGho-}%K$q!P~V0T@CvXmzo-iXVK%H6`LZHV=9kl9 zoyb38M7}2N*VMO>`ZrSk<^<>gQ(%e67LMyJ{`X#yZz=!VN$@`G7unhhMgi^Jx?N-& zeYb54EQTLMwl@Oy(RSLj{VS30sN*~8_>MMwM;ms~h8^tx9hA9)GIvns_mufPWqwbY zJJIi4A@Xn7gWY?`x91a)pXiIdv}fNek^SREehvVA@iTq#Gkx(hwjRXRgV=KLCD_2| zYBqey{2TLeHuoKZbHs>w7xsw}+XhAh&m_d|5F@TW3>71Or5NQC#7HDhBF|V<90-(C zc|4G}@*Xj&G>73ZA2x|mwKm|#RGk6oVpJo4wH`19R=@!^Ai&0&8^owpQH@Gz_sqciq)?h2EDI=UnP>ATRE zUDk`ymG*U|EnR6#S8TnI^cRx;LegLOKI|8xTX&!>7j**KaM4;Zx(9%8_X$A0i@O2s z??J!y*eJ#&HK8wH|0VR%rIlf;7(GY8JlHHos2(K2Ofh;56XWs@Fcnsd(c6UfVqDQ$ zj6Tnc(YFnZgC($AjH_Az{dd)Jz{Y+p#kjgIjD+{a=pQFW5^YQB1=E4PxF#O3^BUTI z4Sh9`z8FXw1`eZ#Q(CdwXuK#URe*{yxV_){b357_>v zOfg1N*61EE1y;a8F>b^5+bCnqGBL)|r(-7pw%s9O+=;z+V(Xm?fbz#v_xOQ;ZR4rq z&sAYCJPTinF@f-e-Y^Z;ig6cqa*a0bnh2DC*IqFu(kBz?--)ll7BTL|=DSD1Lf9_G zr21mq*AwXD`>=U(06GCSPNtocDdT?1xSular;Ph4V+v(FKwnLzzNyqVl{O?d1=2iB zcv^Qcrc>5qwEr>M{}^RGL77id-mH3JJk?5!+55zJy0sW{+KcgQoEXm~it#+(bBUXa zeRHX6E_KbLu6ao?3n=r21~6ER`PBI$X45`;3YBMd|r&VY4>0Iit+9^F_vKSQsVyBMvUb{#rWV}F;--Y@nL1?12bTQ7%S=X zm4sJ54(r8OwONdhW&!dV(yqA_9)>kyd~8BH7%#>ra{&84jTd7rWv#0tM#@-t3wDW- z+60EdeApyLT5aeDq)kf~Bb|Kd*qA;UR=`0qGFrnJSOh!7$ZQ1kO(w@j=2v25)db4m zKkym+XDcI{dUB{|J;%p-j*s>G#Q3}=jDS~QqZk{=vtc022HNyRMYt5E!fG-8fertd z0xQM%iuQfg0mj3-ut$uqTfj)brmwMSBYnPc5K!Jf`--uN@;71ACT!ZA1arjLG6d$s z7BRjh{kMd_od@*ow$3mKmILK~M>#ua$M;Re*x6HzU8}|Tw}MVUx_^`I-}KGyb}$y+ z0rLIW04Vpz`9QgQ>Wc9b>3>=w#$NL8rJlWn|3h2%#RFykm*e1nwD-VUK%GBR=g-4| zzWRBK7zgW#;dKJ;wn=rKV_YYu$|`^BunTP9UT z0r!Qg60WvG%<4^HpqMp?tFcSWT3yAg-3EpOvoUo{F;7Vsvo2-S#r{)Qi&+m{y$)iY z)=td&lVGoyr?(Wd!APL&hM$Ppm@=Cz7qcmL;4_;oVt}$+kiI3HGe^ueoy2U{90rMr zuWO!{Eav%?(QyV)XD4jx)E#J7rxkEe%nMq>XjlZ;-x>QmlczIvblxLomsT(qmcRiq zyOO>u>ASuIyT!b)C6N9?(s!>c=EbI%my8kfvbAFNWJa#n5SS@uZ~EuT5n}dx7by4Y z=0Km|N19h}60<*dV*B?KGpUJ~*YJJK48Z1r*gO!M2V(O;Y#xZs*R}!bxRyGvrK~}% z02>D_0c^akJxm1B59Xb!!5v@{ye~YyDdv!=V%|Xd8{UFlOes<4Q0yE!2R4X#V`bWw2jNd_407(!W6Z7fJI{ zU%-}EDvJ5)G*~OFFU&N?yQTQ24j%3nr#%h(5hTPf!9 z_F}H6E#`-H#az`w%+(9UToWhe$Jp~J`)}<6F+bz`v!P%rQiuo;V-Bwl1?TOG`%=LU^6Bq*Isk5OVr!E((9&J7?4lWg|K7Chz zlvoXB1NJwpBUWSjt;v^SHSI4}Gtx9~4aA>G-DgtInbgsedRtW%>)icfwV{kQLxDbQ zlPOkP_CZ^0YCA!!cJ0JE?+3BkQ%3tqut}`*vG089JbxkV5vv3FJJ9Y9gga7C$1P&v zGgzJ0i*>;?pq-t0E3k7b7!I=_Rje-bVHe8nLf!ZZRu^<#1%?6gg`~d_d+`x?nFZ+o zZZlw?SQlZ}MKfW8Slt`Iz3`=27uN>rxR^S7!~=OQA@3zUVLD`sb!kl)1aHA^u`X)? zXoX6>Add@0lppz3+;3ALZN+4+O-TO8E~S5bGi8eK=XHY1r{7_D>%v)??&( zoIZJ+zL-I}CrLAtx@L_Q>#03r&E7879QyYecozA&MPkkEFV;Njm`{JqUnbUzJH&b! z8(&Tr>y;T|y^2k*Vc+Yc#9Gt_kQY<#TeR_=20$6_RTXP#O|h1}CDz}t>HWd5T&%yN z`}=rcUoCG617SL>guP;YK>iQ<0eL>a<`3vA&V|;Boe{_Cqic!124e_u|A4}4nUheng{g98VB0IaF_$B zVlke#K5hv^VK$)qgf@Lj-n9>l^%>=SMq5(In|e^Jbn4BZ{!H>^%@iwVvRI!BOcLvh ztz!LyKL48Xzox#8l(TWPSpTG5o9e<6vA#jKnX>;?9q5C9k^f)IV7FLX8bA`ze_Pf7 z^?i$N-wuOkfikyZ&(=;rpKPW5TQ`ffjk2~;);7x8UK=LD`*1+4@2Km$X+XLit>9ju zobQ_eZTX(I{XqZyFbEclwUc)4>;vH2Pu2Ar~r3C1DC>NSOPzY;{Lq~ZW6`Qy2>GsGY9sH za!KcI5EVn7Sl+bY=|mN~2BZzh?JTIzNeNnY%iK;`{rzDH2 zN4WlSQKwU0!#Gim9v9W5j;J$8-@LD=7KB@3drR7NHuas|3ns!s_)=6W+SdvjTMYry zwxX<7J4BsR9ccGC_X4_e4v1<^y4H(Coy+&R;{kme1*B_({cX{=r9ax$g_W>b)Ok~Z z_MEq0)cM%lfpR-ybI1O$L{ul*e8B`!U8uk7`=Yv0?nQlpdM}zMs(S~ZEf>!h)nl2c zp3Q;sLfwHnF5fPy_h_Jx`qUJ4Wdi&lsxS6jHA7UtVepQqt4V+LYEhhnRsVER1Nw?`JX z)b+h#GQ11eHUyi7TnZCmF>DrfLuH`PZWsp(;44u>v3qDIpe;jb>(KS0Zj6C;Faogi zM(iGj&BI#3P?!bSJq#Of!p57h@g{7%2^(+PD{6QXKsS7bsGGNn8i6ghQ2s5i0CkRx zhxRZMD1T(C@TeTL0s7+BIe>1I2`ym=%!E(ifT%w;g(R2;D`1bP(e=X3}agPvwWFb)Iw5Bi=mH>4=+6IV!^exyUYI;i`Je~B@ zN>>AERxL(SMK8&yUvw;vatth<}3kCn)a;%6o$HW>DUYaX|bG;-4h`N%B9r0Jeyl ziS0AT1N(9&w*Q6p{be$&6dt97&M*P!lUWBuJw@L>H3rzHPwf(gU!-P_0K(kwQcqL< z(<6aCdwP$kIn9B-nzImo5cNz;z@BH8i+Wa|Cp-&Zih8ap^oHq>E{Zum^*sCJ`6;kM z)Z9e46zJQz*f%c#oq&BdkMi)d)C;YE`d*;E7Y_a(b#DS6MVb5$SI>0!%;Zi8ApsI5 z9N|7h+hARXJAs4yt z`#z`t->07$5+DP^?tb3S`^!+%J;PKzS3ULAQ&mqjiR&x?l(#OC#IHR7Xy4ap-`8m* zt`7pt1E9S1$4UIAFW>>dDnJs68}R!LX#WN)U^f8c_bupui~7L-CT;{jY@7yI0yskA zcc|}oXwP?V1MvTFgNU0Z1D*%`2tfP49|Cw3uojR-;^r{GgMg0#sM8PVgC9_*A5e!M zP=_CV0J8vV0Hq{ui2y7Dd=Jr`dov04Q@S@V7ny_y};8#BFH9HuT{({QtH? zByLBYw@(8s0sI80C2_}Cz%zg!0re#QJPz^Q~>BfhZA?(N!$|!cm;sZd#XuH7y)=35D&;E5xSkYcL886iTj=f>?H9Q^wlqS z15!!+bv>Yv!~?*^zE*@zCLY9h2UAEq)En?10JIL3lXw{4A9)P0hQy<}BqojnJOx+< zs3q|j>Us=)aqI}7oy6m4=W*2QIB1_31y}^Y|DQ-E@#GUEo|*_)Okz@hzyl9Qxqg4gktc0qqpfOIZg%z0QXL<^xdHd6bok zaZN>gQ!y?VaJ>)*KpoRiUK-k$J{IsSi5U+8J|-~}Q;a@7i=c6D1gM`N)k&(08oAj$}h<$u^eqE2fd0660063u_lPbI{daCpBoxUY+g@d zE9%sWa#~SNJKAV}oFrNaNFYhKiX=n`OES)s7_&(7cnz?VB(HrW`MeFl)&DmD#DMu9 zCn?ZEQcs+54PHc2?~^3;SpwKVQs_jI!giBnCctEp!a;k$e3AwY0pPb05hM+nM$%A} zF+7Q+QJ<28J+1_sp)|Ibq+39D0)9I&jikxIpPE6^ZL>*=dX}W=sN0NGlHh}qVAGTC zI7HH{6q4@r1H44iUHHvCfh5h@MACh8N&2mkqA_Z#9zr`FE+lEu`v8>vNMFDMfK>pL^(bgQih4b|7y!DDW|QB%=p`U~j)4R}u<0HCj)K|7xT{b#lTno0WmIKbn8cmU`;i>xHi z&I17FS@g|wXcyL9={eNpxnz=_$NxY7AW2IGlC(4kfc{*HIxI~i>BZgv;J%38zKGwx zyn&=wKPJhFHpHSXap;HTH8=&>Z9KhRvlO(;V1EB72z5+-9w3D|_3ckddUqH}?}7IFUL<`GL(+$+--n?8Ad zX%og^Qw$)Mr0*91(1y)B0Hq}TFa+={U>g9Re?%L8d(FmA9 z(y@soA!b23F&6L?U;`kZq?40LN{Rs-05p>XzmSx?lB6?Z0BF-$+@Hh!Ib2hQ0MHjH zNhF;Q0?Y)UFH!>mpn1Uyu%4te3rXqtKK(~PB}p0RgUr5wSU@gGSy2G|?qYAiDIrZq%`r{-utRblpWi;ZxX)#I72_&@v&%T&snn$v*jb!~~k|mtelM_fb z%mus;$RgSJILRh6;8T)4XOrw@C4{;I-XUUO@o-t{3ow zSCQO%D&SGTMi<%dwt!+l9!6hQK$kCXh^Sd#zr8UTIrIOsir`aXejp9HW zg0{z^KVv5X(DvAm0ibE~0-#A0}NPc%Q$?t*w`?Vy0fU;MC&W9UG{%0WILBP9!Qj%9c zPV&Fd&VS+m{*7_?7-RMEev&^yn?FJMpCps~sR)=w^5;uQ{^EH6>acbc0QFtFljJW4 z0)Y2r7Rm9cB!6|5nbEImf`!kjsgw1QSxEp%Rqi?p9qC+bsk zbJHT#Y4(#=i~5Qn%Xqi-s4Z#fXsE5PZ!WGVtD)jzgMq40uiCOURc$G#r50T9tJZ+t zMo(W)PhYR*c6~!ba4>mkzdCpR{2Bb}Tt&q>8*jabnFzVaaKJ&2D{`oz^*TY&37%Vp zX=K%%MRS%}g%RkU^SIHm@^EPd!DZK>pyw3QIEBAlK_T~Q3i6e87|SWx8HHF!1&;9w zTwv(h__=JsO&q(Lmp|HBejO)(qvqFbK?A37fK%{wQW(i8SUM>fz(ZH|q7|oEx3a$S z;pin!^er5{@lp$n7z3jj`54iDqgmrtHU#K>jRLM32`WJs2z?2kH4;LjmoS=-ff3~! zF^aGPry*hj>H+>kF~rbyqz7P@{T|<73WAe`_aSP$&FSW zYXc*9lV?rN;!pSHz)|4f(LJfzmtMLr%>BT z!8w*)C5sg&zqPT{rnD7j;~lg5SB~Dn(6w33B{!cF4hgRDB04dM1jE0CfkZxRK7iE& ztAq>^8oY#<>qw+*UiWrh_esveD|zAVoppD~1(lq_9!|mNq%e+Cxc&b&g`2Kwg%>0O zvZo+3gC1Apbk^ibgB^ot6{ZE*toj3tl@uCiv+8^~ODkl=_ahlwE0lp-3uA4CVy?I1 zoaB95aYf%cW>_yb`u6|-6#P3|(S6^Z>C(5goqY>m6`Lu%Z`IDe4Y<5-C5)GZ@nRBO z!gxs-F9m}Yr=U~g6h;EK&=?fdBd&|o2M$|(y3VxUZpD0xq$>4afcFeV65)M@f zDhX%31cPI)p+pa0vz5mKC3>Jlef4PiO0m*lo8Iq5`UZC;HeJh6v>*xh$87r6_o)PT zfi``>FBFD5KPPX18cxGrqQ|E4!XNFy5DlTQ_OVxe@u6iMt7Nm?oS%sFFD{_xZrqpIba${GFyp?^rhmAc=HM>Crk`V`1-NT+jDEKizhD&S z;a9hHq1a~AJ;x}{#`h_pQ;u3$$-|}}Wv0ovGdTIle6?PtEhBDgB>$~Wr0I6(rK@#2 znXQD8l!N)zOp)qB%&$sJ76bg5 zAXXrVwVq%UndEe?`)oe}icAnIn23U3gF_M{S(Z&7GKq^~1)}&8QPC#qkpI*Nn+&$g z_5(#fQ1ruS`GJcW#YpvTHO3*08QDm6uR~bjN9<>!iv?*QqD!zxXE?-K^63 zt!wydpirBQp0m)0T8(F;hAB32&SECO6$o%~)JP`5l}i%5`&~pM-$fK~RnZKN(2Ns( zndr14bCh!z5sKChL2E~$UZLQ+P_)+fa=FR0BZJef0}Ia;CLJqNoUW3QajpzG7}u)} zIIz!jHQQ(kn89hXiSxZ&<}t+#|HssqK#xoZ6Y2Agx+ZB(c`=>6ayY8am z2Z&mytIyH~N~O;M{}w!kMN<)(=pFj2Vo?_fb9YW3~0q=mZm9* znx)B~wzI)|1SRA0HSJ}WDw^%o=A3f!D*WjjKb_;}aQurLzfUK=Q@Tqizf(6(WV-Nn z_V!oywwJxlR(QYPi+*SMss9Fyx#)M~up%)#yEMN#Hr&Hor}OvK>-7!!IVpuXxS_hj zl$183hbW3w4JmX{v3YxVsU9YKqprElv!`rVOwG;BHC}@DaS-`g)TGV-PnzxenuGiF&V7LVq5$^}Kg5dRRdw6wM=Kd+{; zQR%UK`SPEuJV#8PJo(6xBL$|ObC)h%YU{CaWB)OCJ@n8+3ucWQcQ>vN-7)133&je# z|B6!o!d|-9cV$Jw^cbtlKDa7MeV)C9HQD%$mHMI>-#)zt^)`7Rv59wxuc$=P8us#D zbmuLUmX^oCYQBhueu{5m}H>oGmN=py! zI9Xm^e(_i-riRVCv0i9w7ELlq&7wF%HFC6-yQ=2oI`Ci9q;R?LSEnEZ9$;_s;552HVFq}uaa(pU9F_&8<*#4^P zUdjI=vq}ul6>v;wOJ8P_@Ho$G5*6)&!zSUhXIzOn>7p=?ggZ3M-)fjYo)>WN>3q8| zq00`-jOY2F%u9MsSSfWVR_d^6+U&?&A)EAbnrJb;n{3nPX3@{M)7$hXgDDYrm5%oR z865o@iL?&1o&ZPx6&&q{e$GqH&OV=(pP%nOckA|z>J=*)42Iq#_PNLD-oe4adFyxX z{Pwd|t5)saJy(56U8u&W7VUKpR=K%&%~S7kd~-V(^%Hf3VeRUpN00ve#p+0P1U~IJ znG8;UAM5Q1z!Qp2TrezFdf|l^o_tVrb9(42D=RS_b_&lv8$;jFSM(`+;nRIoO3la$ zKSX;C6auK7+L0fge;y+EBN3s$FpD;b58rajE%#0*8?6&&;fwY3mbhTrv}ubXX}vHD zYe2rDTNj@e8>2p~y`EH;(z+X*RktK3Ctobh%`GbiBj3Ob(^-;BVx^Xrmcp|Z_hy^E zwY7DtMV&py&#x|c?AWn4F!^+o+cRUOva+(u`pIriL``3iz=urJzFZbxWWqO>z4c{p znysQ;;j;L~G3KHSL;%rZ@H&&;_mDt?+beRhamuM^M`*FpI_DYh3}NKa2s&2JYy zJiYx)Mj81PtCT_Q?M5$ZXn{1R)`q&eCcy}4(pE>_a(#V$lU?@oH0jlL+kmPj&)&Uz zdp1=GLS;(QA0AqGNAIeP&F{SP&iCn6z3*7?&>xb{Rm8$hp`bu&ZizJ%pUcK-G3@tI zO#V`=_{dN?`|l|A74|a3pYQ1CFb@lk^Ol;kjvqgs(=PRkyl>u|NcCCu6)FA<`Hy`$ zing$q^vX@X_M|T()mPQ$<+%_3^^X&107_|sMaVN&YAQUsb?fTs>NDzmb&mFWPWLN~ zpN{Mto3+J@npG*Y4-%=cR9LXlPPJqit|`yB@2Y z&|F?#QBhTueKGBPR|G9(T7@e zpyzFpB`NoO3Fm-kFrF`9Jm153K7;Xm3gbD!-{0TI6TU%NzL;5q3|wxH?=5xq_NJz$ zb`n(dxN5I+bG&b!GiOe4;hA3#9^AiU$AP4zBzMQ(S3jn=R^Xrh_{bxV+&xBrgA5U+ zKE+;+DVu`s*Kky7fQenhD>3|NW0!f7-Wm z=gyqGQ`@(1cbkiD@z#kLE0I(IGeOXKyUA4(ii(O-(~I=Awf3H}P*qjsHkbLfWfj%b z)HGAd+0&;_A3oCN7TfXl7vUyY>$wHTWbX9N#X?G0$GP+Wj9c?z-sxqCIud^ki{w%haH=B|Jr+;@KAdUq%JvKtff z{!!T;PyA!jkf8wt|adGnN)ZdpqJruvQt>W~|(jH`TUb zB{%8VTGb(lw!uPcK}t$WK`SJRuDAKUcR%<#MF@<1_~D160)_KmuX^`Ab8mLei7nwR z&15vh8bU^dW6?iQ0**$C~;Hmh7+ zQd(MCR;PrjSR=i>1A6+1iwTU%S$&|cTl;xvC$HBeXnPL4l>!~+p`Z%)yKcfy& z33%xRQGlM|lb)HLlbf5HnUS8Ak&>NYm7AN}R6#N>r#>Zlc?~`mY=LKlnaaxAt1n`x zFXY6>j~mAZ)C2qHFVl5HAH=pfmeI0zV0#o&QquEe47Mi^Xs;JM_coQ4k=LN{Luizq7_q_aq-2Bq){Z`p%lmxY{ z8I=m{($K`GE6z6obA%{t6n5KEAYjgI<-M#d;$5VrC|KBNOFkq^G20XTxIJb;Ab34eKI8 z9~q7BG6l9^!|vVReeuypA4RKAt54$fvf-m6#l^)Z*S&YA`ggqI)FtYl)CaZKL+YQ| zW{jO?x2Q`l+k<16Ek*b2SGw+- zv>}iL!e})&zw6u|^>^%#E{$HFZbq-`j>z>$y3`2j6EuxLS$V zKeJ(Y4ZEHVqkjUkVT{! zjq6?2w=_XOl9t43A)JJhIzb{KV!v>hga?N?qoTSiRX8q2LT%tCQ6AS`p*>wQ7T7j$qA8dp3~6TvUC z!}Kd>ho%Ik1fI3(jRxOlMIN_5{%HKEOsjtB?(xeNVdGr3B;5(S-}GQ8R*Qt$+hEL9 ze+De2L=+xPc$V z@#k~=Xpa9g$8U54zmnreaQx3Weh0@N<_3OWjz5j#f64KEIlf0Hes`R2)lH$ZY@C*I zDCYVXPu`Nv6{oy{Rw7U}1z*ybF}6{?59`Iv4$H8o zLtZf}Gm9W)D~eU$w-TMEuth~bmhtRz$dMU3#}xvT_Q-?UinHZcgR_` zwoK9L@VQopfEKp4c;Dw-TU_>A8}Iiqyx))UelO?!9_U8DhjDx#j<0fj$Bb`v10NmF z;-G6d{u+)ymY4sB|Eu_;uc|+byk;IU}K z13-$$`90u`?&ys3WBz!vS)_0?_xSsfU^L()qV{LVcC1|rU%SYE#IPSXj~lmte|-GW zqw(>lPNA^nhNVk)?;byX`SLh}>_Bw2s(~P1))c<18T=gUn$mS$bFQX6x?4@_x7+vG z_hB_{ZmGe#_)VWlfs=f&nl6t?T&l>64}Emvqcd1Y;VSrP7M4>*4k`5&eVee58kkT|~b< ziQdF-2HW&U!VuQMV)Wqcr-yNe2)*EP^2c4EO$UFxg{-*m8l%^h8j}?I!%6gS{05Hx zePQ$o?hqF8a0)^vwD3F!H4$|*(?iO&;^>~$A{}OGv@GgW7U!UbX!Ix_!igH)o({T4HM%(tx>jAe)*q;ClNsZ4buTd680*bzQ>|fC1(QZqzoeX= z#vN!@nJLOalj*Kb9%8yH8*dA()GcOx&kmE!LJ_|QTC>d}_y|KlE&`t_ZF&!|2>VHt zZjHuq<}MuMqau+69Qyl2^zB5<_&Jch6H(Gca6q5)=Tq9L0!C710?ACz%+40YloT(o zloZ;gSY*vv=Pe*CT)g{4qlfyYN5xv(GjjJil?H!&oLpcI@ub zJ`$qwc)&Mc2UuZz;Tc*KQCr(W&NlZcDJea$?+oe>t5i{5K_0BV?WLs|8KtF^je6(_ zetuFFN!1)Xv2aq=)RdIO_372is}T?I_4A8}h!{K|SoZU~`R0g-=xCf(e@^jLLwbjX zh7K}USN9L`@$oUlN?BRHzIAmDOS4P8q~|tGFPMftfKL56zj&f1_+$iO$4>oPW1jM3 z_J(YXd2`MMMcNy2&yagC=F3O?s0g?J#LO_80T^@oRIR4vidD?VW5_V(M&DkF^v9@s z#@>T5?_cBX81vs^%ok$J{|fG!hcRc-D*Y>~VEo7~s%vdS(kNsT?O|+hDT2E*JuSDg zqQz)5iDIn2sS?#5+}>`06;E?`wl&n$)Z1m4Nn2~lTY}lNrM*>$GjDd2+-@6M(d-3_ zk#}2z&{2|n?4c+Act@}LuP%Mu8wm zqaJh){BMJ)bfd$5YFghgZ20gkTZZ-Xiu3kt$$;Op#IxssNw?pATO=$|FNhx>>F7H- z8fH?>@hQGE+P7kTq`FjHYFbfQIb+6*iG8tJDKNa*4Te~`t*)%7=-@jshH+;9_z~KvUd=?aDC7(IG zW5@R0yZ7zecQlzkR|Xr6ZFm-+=KIup`uchaP25*m*8$!#iL%`;in^wDg!D94H-X(7 zOY)0qn%mnlQ?p8HD%(wPe)@O{9Tizw6}h>Mjg6`f!d4NRn{P=wWX!Ck^4W} zD_a>&>S$;XVRUZiw$Ljze4J|3HdcPwfZyn)P@^mp*aSPu^mgDBy|pqiReiu$_GD-&we4(0+DbbQX$xXnA0?!5EP9yuoy z_U+rVWy{`U$Bx|qFZl{L_0dXw{IN$Lee|B2gd1kADD@BQ)%(-DmH`&a*kEJ-+itt< zKV~>x(L2Xtg_fEa^*Qw)j2qRZx)tvf{U>}lY1_7KKkwdk_?KUPDK1S(NJzK=zMSAI z;Bm)By=vzI+}CYBom5v>mtR!pYqxuZ2YACye*=80YpH0AH5q)F7|u>P-*O{->{6zR z*_uqv<+jWqF8S*@vJXct<;WQf`P#OX|FGqk>TcmJ_u<7S@#2}(LeDjK z*8R$!yM#ZUiSsV6xbD*P^0;2bpGCqIvc%`WfdjuDPR&hDuBw_b z15vx~&U?dihF!bTPMxn%#B{a}#r69o(e)y9A#v;0C`9q#<@x5uUw3Z(dDpJJdl5W7 z{F>9;*T>ttwnZ-)={~kB#f{1;Zti|b@|Ih4b?NDt#p$VO)kQ@)ISmbnEp*Q3-rcNA zzKC&ct48>#_#j)K9Ga5L)hSO~oocn!$xB!w|~PTeRM{%4C-cWl}1qV?ag>VLU%UB$tpuUx+VBaWVOLkroJ z96s&J#q35o-qnic{tR#_UFAH-x>n2N@n#4MTs$y|aT$ThcD*>Ce2(u2=?08V96y8O z+ugtq;P^v0{*N5LlH+G};&;EY1w&+`)OwEW!;y*S#EVbj z#W!}=-DTfT|j2Ev6TY+6q|gQpF)3x`3PEqg_!Hq+Jm=R z<>emdt#0Hc+pk~pi5<+h5?m>mNm%bA;E^JAM})l?gJ5AQFPYi0!^|QLHg{*sPD(nI zd?fkM=L+W!lhJAI&>7c-^<6o3?bwwE^v2IQyL9V?dJG#gwy87t>xtel}2E~MIZRfug! zO-)U|n0b*IeAtKFTv&QAMps#SgpLlIAW3nd)2H7)XYTE@X3auOXYHVgw=!gwu%9K=?^PPF@Z(_^=OIS+MjR>BiE7 zC{mly))p7q(o&gPPy}k&51FLaVrKGTA8Ko1>k010)+2a%AyBi_p}{!ipVM#1v||>J zxrxo55K6-Zrdcal*ymzOW)~y%$A$UKo!FP}3-Zw8_}R;xHP3O@%;c=u!B|tnW&9Nm z{uz>i0497Ja6@ z-jaVmDk*8y@Sbt~TAN{1%qYgDrns10i(ACj8raH=%;?oYt%__k`1?{@n+=So_%xdW zd-duSXll;QrNBUgtzTG}2`tE>xfu6Bj}7cMVDRt}BSuV}I%RM#oGz2w+uPcywA2Rn zQ+xspDgp*oL%^6Z)Y{T+<1@Q^>-Y}lDyn2}KJ4uXdxH;28z<($yV6FlgN@#zGA~Tb zCEpP{ao_1XI)P8zcf|ZE+;`-QJE!l6QSIvO(N6ipcVtojhF@{-5v<1!@6r9Z=`g6AnEwnp9lXpm z5O-a3IxzmZjAGXjcTj8t#T^>OL%>E(hrn{mz#Vcrl$sG4*2$4qdbfN?aCo<3I=x%$ zAY7++t1mur?-pb>yjuc0;XEH-I{jO$-T(W!BZ2DG-rsA@VdgEk(hINOzL8Hi?JWP1 z)G&B!*jaw@4pR!9w6=&dmu^O=KCyX9Dob+eS4>|(K>>U48b++aSMKr6% z&wF$)GuQ1gjdmn`K%Z*UPdmh~li2*bI!iz&^+(hPH0pgE)W;!f0P_p?*QiI*Inqv( zW1qs$zptXtw0WqJA3B0f#4Bp=cUkRURk_@l;GjQIovG1}aL_+WuWOvi^6#)L1Xg-e zn~@s*rdBp3|10`6^n*rU)#wxQ>Dxfxf=}rEDx^;}IO#h_ll8tcN}~&ZOt6WGX>5in ze1_^~vtD0=kA*gI&=@vH70gl5%a=`31(Q@;Il85K8IL%rYO_>_B>xsV1biIb5A4V0 zsdjX~w1Q1kc648(+Pyy1!1Ww<#+m zEqXguS)qq-QHRZ!uyItZq=oS&;(xU;>ha17zK&iQ$EoRb$S`YB7c0=qQn6RVWfY_~ zE{U{?UB=@a?R9pwM(+Kqqf~~FNu$`M8S%-uc+vRl?z7rBzMbQza{LC4Z|Z`NeqcFv zoRJ$^b|oC`XJ>j#glqTR3(Y$99ZjeJ8fd)(_;R zPv`iHIQ~XndbJz)NgUsw<0o+ZYK}j!6TkbJ)Peeg+wS4W!#VOkj-1AjwOQ!0?si`G zP>#Nzqw6@j*jctqd^daECH{=d`tBWG{xM#@;70jX96y5Be-Fn`=lH*M13#1F_u=@Z z9RDK6_vyrUHD@q^VXcVJ=FBc_&Pa*OTH!g~X|0ejcduUk*D{qop1r-w-ZWi)kis?i z?lHEp-=(s*QS41iSTWv#;aYF`htgH&pMg)EIzQ9YxX#ZsOs?~f#GO;;cR6NmNRV_hFsk7J;NioyK>Lq%rMuvXLuRE;(LbGxI_N7huOmP z4(|D`fzgA`>}{rK$85OPu7T|nSPU1BISbS%I(H3a8cn`yV0nB%6}yIs*QCmK4J>wz zQRTY^F9%g^*C4PQbWFczyM_ljRqPsY9zmPA-Pe`x8g6k=<+}zy2UWgnXmU_xyM|eu zDs~McyPw)UDjQmK1W?H*U&{K$I8=?H~kN?UlB7UH!-P z(drcSeZXk3*+|A_Mb}r~HyXJ^{tkFST)vc+A__e#H7got%QgX4L?UEr5waF7V#i(5 za?|R{>rRNqW@<)QuNWlWI^xzy$470Zc|kj2+Gba~ABzZ}0~U2Ov!@HcLeFA(8j>RD@kS zKtbq1Liq{i85k4k^bB;x?|NE&8^a$=tVu6G&!cFDj~~aP22B16x=1K{*J(+avn3VI zmZUpdlE7P1*Vz)6{ZZ~>N?T0B0& z2(P|ugxfo*u)g&$UzH!5vN;j(Nl#8C5|7I?Qk>C;B!mPwDdu64wCOOWU z6gz8D%iA-evnFiDb;q$_d3Cg6#lSjRGuKJuD5p`=NyBALb~8*$?fn zIH+oxl>Ivg)h!y;$quUTgX$kR)mzN;d^dIN)OnuN>bk~3Rnw&GKRBpZcFXYSEFrO{+dJo@zca_FQQ%{DW9$PU?X%`rztU1&WPmEj?y?h|Z; zYH$xd3f4cCKLdJHu$g{`yRLc^8}rLFv#w))i6YaaKGJF!3Vdi%0p&@JE~9E2^?V{K4C}Y$?O=~j!N$fPU-Bk zuDi%fxd$Omu%Tmu8t;&=aSEge-Xax;KhkAS-;iF?3oROoRcj(_X}!=(!_9)Hhk(op z$j`+7c@z|7X67>Y>cwqVgWcZNf>?38LOl$Fy{D|CONn-xV09CFJo=6O-D`{E_c z>#qpIdd7K_ASF+EVPSQ3Rfg4|(@6$N2m2xQ@DBV)2~*s5LaTgQy8E17WSi-cy>3U(C>F+(0#q`#gydGOc6iV$dw)s^pQRfL$tICONm}Fw1z2}5Qitz2uSPLZxb4&5%!^qbY zR9sw?Ur?M=US7uh_(i3qIVIHZ)c(A&?V8O%a!%xC(7?CaDQTw$BHAU6Zcf30CZ@4()i&+;==`#SS8 zuq-P>z+<<7$5@t?LEy1~)YPP;)YOwFPo6(^^w{}RCr=`1>McQ4*45+o6sM#&>zE@%tU|mWHT)#Gczsa43hJnOG?VXNAOgvXG@D-hg@a`*~5qlXEell zvxtT@3ulB@oo;LCFj$bFFXg4<=%dq0Y@~XtdaJ3YTGOch<(X%eo>GLdo^hC4{9sI+ zE>D6czr-dL7&W~wlo)0 zo7b4>B+M(!>=RC}pH<&65CvO>Aa*!EdIn0j(!e;Y@WBEbx-8DB*Kb56tdxOUsSy={ z_TrcwuIe&qKg;9tu(K|=pf2AgGeJ=`7lMM<#T9PFegO+pOREsBwc{<4gaOIIVG{12 ziv$IoxSnR5F~6b@iuYqBA$_h<^Z3JCr5|#VhM>X~HX$RK#Xsuj;@#>h%8KQ5Dd#y| zus*Y#E(>{1mz^x9i$KF2vzqCPOv17VtxHeI((T^|d9Z{fG-&M)>pDzH+1Q!jBc2od zF}u#a4`+b6e|jM(ECYpAps)~fft`OCot=F+Ir&^l%9*oCr;?ISr|;j718nT?-07s` zl$4`sNlB-ZlTK!3Wo98O6YQ3;o~32g_MlTcsMJSh>5gJi@In29~SJsdG&{@ zzx?56Wa|9>yYIe(S0Of9^-xX7r~ezh^7h+t(a5el9N&1@VuVnk7uC^0dQVuoG=@H; z(@+J&g;0h4Pk$^e=jmVC|GYsvsZ0nILYZ{32z@zW!^;6HA<6Kv!l@b?Bcus?g#}3N^p3D( zNer^@*9s2_`-C)Gw7tS!s}55CqTiC3xH%f<9md&Jdxe7URZ=IaOY}b_CT@&|v7E>n)X)Wu>JR<<&LSl~o0HJB}%`zp|?8%F@y%rdL*0*XHKt<)V_5ge z9$5C*BhcT+!=o3@kotIgdISXY2wo#R3KVK(tlNWN7 z3auc{`Pik_I%KUT!@{_OT+m$6$UvRNWQ-3=>Y{OU>rrjy_!BvP9>+Iw{Kigv=RCeL z3b$b`lSj8`QMlQQnWo{H#l%rdm{ZemO2NzH98Z9+QNxFMu@iW)b-Y+TD^{Dm66TMD z`NL#y=L(|v#57OXVzzW-v9~bgTKe+;7Gn}>H4nAAAGMl=?- z{HMpBc`+K#UeaDrOf2-Z*V}Ks^VOjrHm>{f%P(mVo1%BCw_)*qUik3CXc)6^Rm0W6 zwrC*)iKjbgC*nDy>09U(1B71M6r{KwuYdOK*Xuuz|Kx)YK6viA&z@R%-=j-jWlH($ zh-1Xdij7I^9l}$>9l{_;)C+i>6}*J|g{N$H*po27g4BL$FT6rklReLVG7hb*DXD0% zi%pn9RaGr5#g!Mca#Mf_f`-G~FECZtXF_in2iG|1eiPag_lh&N4~w$OXjgpmv}>q240xwQ2B+HSbtnZHb1 z!JEIh*X(f@9ypvCnJUG+2Z*{D4mU?m4&_ql$8Rbt#%H zB_>CO@9*+Mr$M@(` z{mwPmsl#>EL+ZX}^M3Xq0eZ-`a5nF~FR*#v5#pTpt|UgEubI~HgogQmhPm|uQyPjw zoVyEGn0a3_ok2H9fX=WvT#!f@d4Z`6_k=)Y(0Y45X68RIGvhEb=VNBh$IKjgbsmFj z&eW^p-Ca#xY%Q+Lbl~1pzB&oQjm&MMmJ?wwEG>@1@ls?*;cl$C?oD`?0rfW`2u5C= zT;ZC6;p%h?t|Z0CtJ5!BQ*vCLkinHGxr{?z?aCp0_Uz6rJag#Kq20T8@BQ_cU#}y} zZ-7Ide){h(z4G_xpMSovurTc2NB{1wu>bEk#DB$#Rp0IX8hJLJc;bnn1t;VGKV}ZS zUd&3XXl-qU1|)Q-9oJbkZh$!?sWrc*4v9Np?}K($f1PE;edgfugRSf=PxOMWD?7CH zj3`|RX{je}^ zQxAX5S}3B?*Z6>1pRG3NK_a#E}z5c`uwc@h1PvSIR5B>cW_=D;>68 zOl!fyyNjf9TD{T24c3t4>z-9-(!>cUyD{X6WN3+ePF zyN2|JE6Y*S>1gpiXz@Z&nSu2>h6M7cD#^v^^4yZD#*Ww?9gS5r?UE$5*Hl&JBM(hJ z%RSStz1`$Z_0143NZIXeYR3c4k&cKxal@pij;4BBcx8hL)+vvM%EILR`}ZdoRt8MH z>#n<|22>vV0*PlnJIaEPMWY8bwcz0=O=L14Djg5h5I2@k5Hf7XvX?k71W9MubJPPo z?Ah3$XWKo3#^JczxFCUmT$w8ge{T15cW$ZzP`e`8C}n zIuwJmPCUoV3%bKJi9MKlvn~AedGsqvY9xZj{%o1ps2Tiq9f zpB@1}h1S(I$U!}^o#uZ!)!uSF7AsW5auIe%+fcI^OeTsQb>-LPxp6f$wSr#{Q%p<@ z9yyfJRGaY6jn`$vT!rQ>rzwiz=DTnX^1YFB<|XUe&v71PbHnZF2iqA$o#sYm>GG59_H&YKi2?U za(^`E-hP|`*K!72&l&JwC-=IKNXbOS%1K3vzb{=@Z4)`8sO>vH_(U)KLt zUWM(v3h8dNU(fM}@Fr~H_=h?E=*#h$#jLBfO|z8vGW=MMeUxM8aBO#a--nmp$niZm zeksR4dwJ>IS{Va4{&0@Jh2vLp{46){wc4{>OPV)+Gsmyt_&ILi|F`wO7Jj}PC+I#{YldV9u-0r2V%D0P&zZF*C*NtUamlweyyl_2=IOlV z$9T<)-Ke?B@v}Jo29EE~@o)aWitlq({hdCy%kv@2zK|VOVOkJ;ZUu~a6&i@!9~cAk ze8?V*iFrQc5sZ;L^C7z&k*Gq==IPI#J!!@B>tom@3CH$uX=I0duD6nL{C*sNEyrKa z@ejIzU&HZ3Iet3FKgRKk-M|mz_-!1Y1wLtdq*nivo%k+!xlrR}?FpEhV>Dh~qw(@K zCogv|$5%&PlH=`c6^7j2%a&nWN3wMo*ZbK*%;k9Zd#G%-Y0~0zg4o^X><#W~ZS`O~ zqAO$P^y{=toD^fzy@5bs^)bb&MS*>xv}FmkE6mXfrnEu8T_O+ zVmYH(q@aRt5qIdICff87W^KHfPqq74&fjTgZdp7iJHPUPb|Q8hJ_(2v{ES5kB84?S zzmfnOyW{*yA2YI1^VPR|sxvjJ%(=^`&e5oja8TW@QEhNg-P;+<8D&O5+O?=Ys!@$` zQ2mWYb&P}Rw;I(F2i0#u)e+5!W6j6^W2$GgSV@*anw{YJ5>y@WoJKPpa8Si#e6wtT_N=lm>;F?)vPW7z-PiT%$^637R8 z%k1lIc(N7S^*7anrcPbA?z%gg3`Ktgj~BjyeN0e!`TY4euyF}e^br{u5jVJ98N~YG zx_gt)(Gl0(jSNDvRlny@@Vabvgeixp-9&t_D<9NJ!G^D`fbiO&{ z>8D4}ia|`=H`*)K5HsM(Cl4Nc^2td0H-h9JeN?;ZAAFE*LQR_**%Rj+&lVH^Sw=nn z8#(T@hCbPsMvhl^Ao@S^s~=C!vl{4^Afk%9Ol0` zRz;Hmz7DS*Lf^iP62UJ{Vm*TB_M^ats`|Eiqy)m3^-cA)4S3FST!^1vm3I*G{Dhk2 z+ixFr^8h@baKP9RNI4Y&jJk^2mNGmv9S;no4g~fT6%{%33s-Y=^ifS8?8%>wbpo}8 zK9$F#%m#qw_4GlN9N+ARd7Ixfr8;%8G8s zC{)~MUNmAv{+9RI-;a5lHkB7pcz8G!R(||3LaKzc7z?~5!QGEn$onL&+OcEmxY=-S z-7$1zP*DE1m5QRQ++G;T93WPlIidcMhzNQX%C^gKb5_0kv13*5ZtM`p!2Vo}pBY## z$B(Ve5262_LI2G{|J{ZD8`Sk_d@lR(sM7UWeBJZCH>};4p2@d+_iUtCiB|hxdPTuD zHw~|s)aS(a@Z87uqSfbN@5V#ulot7hg<4r zD39jAz^1-B{=wq!2r^3*EaHi$q~(eyln(n}*sh<8E-#AHYTb02kc@E}9KG0elv^ zqvfr*zLq!iv9y6=+GDd9yP@eR!kQzBV6MLbWlwi9$|sWIXM)dEFf?@NXDbmT{j&t~`x# zeH!Dc3`H)uUi!A?>O$1MsHUY&4GQ$|Xw0j|;?Z2*Sd$BdC9_O4;A5<*BqbG*RV@;> z3b9Z|q}XtsPHk(m3vgf9Rh8=NE2=sIdSK<1g$|*;9?$F$6qBGijqD_vY!lb69WrFp z;69ScKM2w!xTmjozlc#oh8XSs!-fs(sgv_JzWL^x-{d4zZ(4 zoKj4ue)#y~kJqebrJJ9Be)Z}#pX~eP5DSGpo^<-yFMHRlS-pB-@wa%&=ns|sXT-o; z_PzFs3yAJ}e)a0rh^v?%dGC`iys%&_g1uwCii%{^z{6W^t*dK^)wd$l5}PL4EIt$3 zei5k#XS0{METw(qs}|af=YmHpQcR2f{EvV9W9dsmORa6!6Qj&=`QSUtbLAmtCG;Zuj=~*4G~T6)^}<3My-;Br~%z83{dcT&F2FHoU2(N%HZQP+(t^ z2?h4-p|4;Mf^E?X^zjMlub3uI9zA;Wh+(S1ho4!S3|?bdK<@yrUCIKQo$cAuh6l3L zm9(d&;mJ@4u#m_%C8ew^Cx=R)n3#NwIvrFj_S6)~=wZUG!cKRDcG0+UbL%Ti(6~%Y z%7mpk+?1(_mQSYknAJv5~bLw#XkPE8r|F_zcl zl$M@2ckw*3h}i;~YMT+7(9(iVrnctBCfq8}C0SfzMytN5smUni_3t0j&j%gp+b<*{ zBG@v$4+Q|Trm(c3>f*(VSgD(vN-ySs0dVxFJTGeT z7RNu$@tfSh_vHBD9Df(b&*AuYx`E%D;}7Kc`#8Rl%UUxYosTmSB zeE*Cwy+@5I+=@V8CTJKK~p^zogt?53f<>u4gD6iKNeI zO+a#D!cRX<95eH=#~zzLXy~+Q*+0G>g?%huy4SblM&l7oZ>X`5>hm5%qF5=oIK!*5 z^4e4OQX8)^aakK(P22%o5x?g2H3Q$(^yT)l81(*J^#1SA`!VSK81#Ps?6kDh^9UME zNjVR%b7l_yva`~VO)lkpGBXjJr5}{o9#~QF4}1N#wsxG860l7+#6li2uo=eH?)mJi zufBQ>52I3lgH*@6c1ewRjMU9yPA&4+wid9pXrX$ikXL6>AA$;w@a`Tx2AX>sdfXHh z1v!~2gx~!%)}>gXSFczhB*bRi7%_bIlTSYR$X_h!{P1MNEKO2w z#L0>0D3!|S3q*E*L7S}>!BdI#>(QgJs;asqH5ItibY;1pMd1h-Ua>;Iehh5=vkME0 zOAyXkT3S+2P?QHzcQHG+5b5zsODdUBpqx&ja#+pq4^K>zBvFT!w?l%)LV_}|_4mWM z2>d~72K!d0uLP5kI1q!QFmotl{FdJ=@pt*@VZ z=+LaF`Ae28xqrfxIdf_c{2M_;|2|kBNgvb4(yDhKe+rMv(;qn!6LW2@ceLf2Jio#v z&xdoKzs7>aHlWw$_{HeqzoCa;Lk~ZJ9)1!%JgzGL-08GahZ9a!H8tU}2p+z5l!SAf z2M;Etrl+Q67vv=+9Y1{d@Y>H)O8X4+MRZ_c0lIChPSEGL1poW~ZQn8~$XN$;e3?4G))SAs?0=8*x9ze(;oTT;>Ez2((Nq(=3f#scm2iqG~8!~di1JHVSd zvT*NJm)zvuZMj##nC3tL8&d*=gj5nzNpG7@3d!#3S`NvkZW3UDG*TcWA@m-KacDLe z_udP*cgsz(^!~XcxqywumiNAAC6;Xb&&-`UbLPycljqL8ZGN!Q&R@>RG*dH68(Xim z_X*lFQ&QkJb8t^u5?m}hbh@^>%SatvT-w&wg)Qal*O0e$!yY2-6_dQ2lbTL%+jguD zGfI+3Dc|*aTPW;F&X3*yaYqKwVY4L8y$1o$BHw>D7#5XPk1Lmw}|beFF+9UIr?s z`8=nRv}NPqdnmq`;{T+0cZz>*2yd35hWm|@TJd}h14kYV!<(AISqgU}@as!aH`x1d ze+!b*lkdHj4n2tW-9`I$7{6~6G9Z}<{3!ekg%41;*U*!XB>_p#6A8%cGz$snEH8ER z9VQZx^CM;nh|C#@KPRJBs7%$HA#jpW$1-25X~!E#$GFD`*!08f%NS6^P~T?h(LC#< z-aY@P6?E*brip*oH1TEp=n;#k9%^HBISllD-YkcCktyKgRE1~BG*CGVJmxSyl1No& zG95#5m>AD>t>uGHdw!Z|W1`77KB4j>9F^b+KYRf*@!oKLSXfTRew_2Uowlhew@zFsj;fcs4X zmk@y)N}PjYVsD-mDXXtM_Z_(9)(Zs%wS5AOqgY#jAct*V?(4{}{X?VC{NAFR5{IMU zpEhTc)4N<|Jo56(FW)o&u}9*~)&}dhqO910eY;+GZDlYf52fN5xxbwZm3=T4TE z9E!)~o@aTYgl*f-wI*1#7pCm|{@3HlS$)2X7ola!EVqJjf_(~ZaReue9xICr_R{eg1q=X+cJ2cE6j}6_zx?d%kUk+; zV6ZB1Rz!eb@PtW`fgxal(YMckczIM*)SO%5d~Mu(r$o-W4b4!V1I=;EMt~TLC)&LMYk(;fKc>%$#thieVlAEfHCs+^dN+>FB`k z3Vg#p60;KWCbCCnfN#J3?(0pPHhqh)_21^@A-jx5BvXo^AYl3vU|B)gn5k1^=700e zV&fv?Vu+EeX{7|bKW1Fa{_#hwak1zJp-}h(J}C0?dg|wQ{RA_|57;|bIum4TTD97) z_O@nif*nZURGaFqyY6b4vu4ejfUHDc#10j^y!qyIkDHzZ-MlcRbNckotj|CHJgbvs zQ!0rAG#^W#CdnJw^H&m3M_T&Jp{D5Bo=i`|Jm#O|C0XX#1QEA9-uSTb3G!k$<|GC} zSZip&dG7BQ2(t`4R22&H6aN+A4=xznLApF3kA0}u1}WFGknkIYb#dvb}wd`}(=*~Hi4$vn;a zF>K1+te9bMQL^U{;aKRbJ8LXr4L`KG`15iI@WPJJM+O;k&xC07A4~8d5pv-#yVG$Hs zP--!6YwT=^F&IZCzJ!G61Snaul=+!;W3rH2N66^$%3>EHsC_N#MeHHme0n28<=QnM zzE%AH&L}PH8SLW|5(u*Q@rgh=5PyHhmE(;a67aq&IkC`SL5s!oX@uZ;Cu5EGV*d>I zBL|a*hl`7o6V!CmsXapiZrai{vz%kg%Q>dpB)6IO38`&QDq;-VCw0X|HNUBaQsNVE z<(ID`rPGjB7D4GAgw7kg^O;5bD&l~o04YGEJK$A}_Ag6tNlh2=~+rBPxroQ|0% z)rQO#L-*n`+^2E9PWpXMUVeUNenv)JU4A}vjE6ZTQAumHEYXkZ`=Jz;afA^(9XWO<-o5I3CsX$#$9~emA1BLL0bwI7*kdJ_4)@VfYjVO&e z57`&+61?@uEAtTeyph=`e&?RNyubY|j{I79;snUtil~BT&wlQeciwpiyg)60HW<7B zIZ|LUN4Ap3H12W#3_kh7bRH;&k#wV*X>2VlwJ<0 zS>2oJaY}$c1%^!Zb#Mq6P^lUkl*;~oY$hd9&USXj%!CLYH4)0$GL9`PLpZx>eVlQc z_^L#V3uez5d`!u(a%?Hk7WtBh88LYmSMW>?mlQBH#wJ+5C3V=`h#VUm9S7$_%)b0w zpPAPsnt2^sB0>XdKl0p;Bq-_SWZoZUJFRDGT`9>og-N;! z!7~t>{(L+`D4v1X^sSMSgya$9)3=DMa!Jad^n$XooSYrJZXD%{QI?x=y*ywK9du%1 z;@0fTy}`HKa?9bvhmT!?3l>Jp8V5L8OZqozwfdX%7EU-|WCa>=ZT|70pjhE5uiAa_ z#+PZljRdv4oqCYqL?n?2t-U!m>Yn+`=gfZo^*MV_6esky7D_b32gi!7T2WQ2TgxXM zE3uMDq>RkQRw2W-aP@RW?kP7MYkuCIZjSuf<>BoE8A;UNuOHBHGChnS7^p$d33UVd zeyVwo!F(*jd_0Qzn1cBr5`u$0SyGlya$}wq`=T83HEeG7CI}FTGJJykpLinZ(MKPx zsj2C89AmB=IOeeg1eeH2aLE|6wt%;a+)Iye1u|OXoEU!|3pi#Aiq~c5l39>|;Lr{{#GpoE=@)j^7lQ zVm&nT=FtydcYft!DVow9$wx?^KbJl~T*u6mV{T#i+VL&a(-wT`{C!06rnssVL-?^I zPHa4ASDG%efecA^g(F!|LZn>V=<%dnJa`DhHuK0@9w{p!byM`w=DI0k^?7**$$XJ} zk%g4|eio5(CzTK>_kn1$lso3rnr6G1K5Z?1S{a=!{m|2niT9?nHJjomQT!o_?;OI9 zHF6LeX3Lj<=1=%%w&Q0@J~&%r_MA_9cA`BWrajlwo|QvAkBJZ6%ky{{F&UE7+I;Dq z7G?wa$%l!&noh2vv@nWy;qm+#!MAXhw$OZS3)fIuK5fB?j^HpIK|RGQEx`9vd&&i!6~?i~oFeo{av)ouf!RC%PCQo|sR#6lh+_Ld09BQtP5+Tij$OhB`) z7$$`zB@piglY}NOD8RiVzY+ETq5~kWzuWwNGK+y!1i!uw+fe~iW_|Cy@DCkTUXw3e z*xw)Itb(@#G|JnNm)P|}L2;{8X&Ro53XMvR#iTUQJdFOt+F zBsU}qRVv4jDpZ{2`tpP~!yvb#h?#`>@@541pJnW-D)LH8N~-gUOG~PAvq{E9X@{0% zLeO@sW!-FJjc>}ES__l1k8b@X@FT=-sFC_IVIC4Nw4mzQ!GrURiwrFW{Ap_ceB2FR zM77h6%%`IJm)-M9bsSU5xu*)!u z%Q1@~6%{!MS1&9?jDJy4d45T8NpUgK-yzj-US4HMQ8E52EY3rCP=Xxc>bQvsd)wQ% zVV^ZprJJ<0)TmGbVw%bP%7G zex0@(F6o`R?w;=6zRoVKR@o*#ko7-t)mEGDF#@KIVZY*k9eEhN~+L-yXhK*}Nc52a&KXScs zRJ(5G&BLjTKZ8xxppIkWL1zlk8SiWoMbp9YZ{tYtrb=Qxv<02Hg-f)B6!JDs3#M9P z^|Xa>a|_+Hg?nzTd+2_;6z)f8ft;;H@S+&CXd)} zq^=Y^07w*9poWoSBKa-X!f&}2Y&kj$3xFq{4&>&!0DEo$_MBfuX=!Z)}9W z$=x;=jvPFGVZT14xNoPKC31%W$fw6P^Ua6eSgDu z`}S@5*ZcU)u_4Df`5TYkd)v&JOYeH*jYOd~DoVM4l_T6?EAu^)*KQKW&Y3kmIwPJ* zX7=G#AuX6Sd&vWjz4S&rw%vLp)#+5nGP!8|wr!kz&D;0I-G2ML``?D%SQQww8jT}S^od&F>nmtPE+H|!Ay&_*s{!tNLz-fD(;@tC_z zMq4;fTbN*O!H%}jG}MCG9yolTl6p|Tk)Fj2jxh3}n~7!ZN#AwQMB>eZZ^xWv_3C`q zYC#8LX8?-={x9T15b(m zahWM+?N~d(?AcgQTTXZ{zBuu{XBwIQ-gx8F#%GN87;j_#ds8z0UhDOaz>EY(&?B?m?<>BG!?F|p!iKG9q zn$XRdlyBngddv+7@899=rV}cIw(#$qP^QzNg0^s-(}D1Ind+K4L2Mu?NC0JYG-)A@ z*#V-%PbcW1AfPARb!WkJy1390T<3Hda~@5ntb(?1ol|y9d62Jr=(v}KJZNh=oZR@s z=?XcVuD^<=nu{{p`&95RUP@@Er(@I5={UMHOZGm}c0gOiQ^7N4yy<+-8p4NR1){Lm zW?`>|VXuXey#`)FzDDwsB(R^W_=#l0hOVyBc*+NyxH~qs`)a;IUDBPLuFqb?y?gie zUd>|oFzn`7d9WVlS93nfF9f@k!*} zr5LmQXhf|TzJ5HdCMQ@&BHkBb4UnPx6_U3tmL+vz5b>_FBd)fjlB6>#6XTOP;(bgy zw0JL(QC%J7E;vVG=KqR*5a)iyhuMfvh;w%&G3g}22C+7hCZ-FY5a%ADA_>`ex*R;ZQR@*y?^$kz=;c&2L~TM9EUnB zFNps6{f=*8)O$m?e?OB0?DkdnFNuF?HGEb%`F-zwx$BS5{;?L)FI!U+YkXC<_EhEN zyraLhv@B(QV35O`6)0>!5#{WD6>5d zfBZT7>|OIAeSbs3i4joj-q+XIp>URqdOEtg?Cm|B2DGfTU!=1$Bf)sObF2ErLgF>U z$rTa4(gFUt65XAgw*T_Yzkd69^ZMT2!-w6xqT?n+goP%N%EW5^{0r(qvt47Q3Q5qSu1xMOK5N)B|+=4%C zfiF8eP76}nLOE@r)!f1y+Jd_UW3ae|vCdu>n)N4##o612IUpB9UkSMwUQ9Kgy<=hL zR1lAhg#=5~96a`%g!p6}Of~yt+yiLU*z6bOi=I5tcxMT)g+Egl4KuA!? z_2^pyXJG{7u$+<=A@m47JEPOP_uRR0OKy!rV!i$Q_c!@Q$5}xClFz8`>n-WK@(u1DWnVSDy$`{y^`!EFV>YNw9x-Ey71CN;WypmtV} zic|Wdn1QX7x$fR0oRHME9C>5xCwWSE1<&LhQh@`xWlejpo6`OHA-Wf%g==Mw3LVN=EM#TcGePs^25SLL2NiCmmXOfmO?;1}>x$%@Z^*!jy>A1yHcoqzoeG1c#| z{f0#OqYFa4Tx@l9moG!uVcXgk!Rv7Y(}oM}5D{rl-TFr=jW!`RDAq z{{C9HbGEkDRWaK*SA{~*)6rauDjPL5Rh5?uGZE^TdHyK$Lhnkqz%^jc*5n?}#b7Xs z6>zW7=`SM=r3n!#&F%fHP&2DpXAQrQrj#6MyB;5$`g4;1+_ULRCsFy?b2Y{LVLY(6 zca=3&UCQakzbZVZ%<#MT`6r)z^4$pt4l02FNmEk}oGL0WMk47+v5&m^>Z_}tSso_W zffF@WisLw!B};)!N!0$N;!s_%|Z_Y&`Y3Y(Z{q$UU?~$aG z^sLORf-cwSd!E+JG8$Tu;p^>&|#m@DeRWX&<6NN+of?X zNHU^DG8S!T%RqlmPoGZ5utt2)(EV?gzlK}d)Pl*NDxB*=A{eR#EG8Fkebi`XgMO*p zJvugmk2h2B8wgbMnE~nMT1&P zzG#4^g)7fSBQ!0X*c}bgv~bP{g~G@A)PsmT#lJU_|9eQM!KWX6&b4&T^Xbf0(wUn= zXDSL{QcLl zGp?S4(feIY=kF#uf4k|rRZwil;QWn_wW8QsiZxKI4aE+OgEdfW2gUYLtZAq9kAr0? z)|+ChDb|(ts~yCerC_4}7;arLZsRTJ2U%Fqvu`2tv%3S4pM`VG^7H6VYZ|c!?X`mT z>Op($9qe^W1UV|RVN1KPE1z{9xa@(jH2U|k11@wudDOOMWbr=WRO0j+v`v=7y zq1e9v3YO?45o>_|hh@nZwvq>4s}}l^`NI1jeuZPQh$B!LArtkyDPn|q*p7^Kc=jWtd zJa_bHrAV9p+tyQ1mgeWDXPo)t^hxs1*(Y}FIC=CCzD}G$pEn&ooOCfGx468bysW$e zwXxbEcXa4vaSQ!Taz}D9vCbbpc{2Iz+0w$C>^!&w!68?3^R~q>1$k*2g!E4;L4^IlAM_{F# zv2<>*UyzrbP3Fn(At!#kX?K2)7&j?rU)@R0<>0}>Uc*Z-E$KV>%Wpe(?>%s$+$i8!rKG=?uAbC(;wYrp1$-I1 zkC@F+l>CeJxqAVi(6r$7Gs9Ue_4bry|MuH&DFc{)_ugKCmBLok-Po#aZRphctUzLu z2WCxQdRJtE(0j6bH4(_{ZF)CtO74~|ik&fe;-raT*7fPXZGu4N6o3dOUu%gy+tGwC zEhF%Vjt&j?X1b-5qU2Z-2+pp3@St~0U)87~1 z?TCwC#FFJ_R#vd{!J;G* zOT-en4Zh(^-;{qS9hR)JOYKuQ7nw|~|U zUCd88k#sgK2jPUdsY%JHh_p^VfA&;zN^bVK?c4U8I$K;pDp>T@RN(?uS#=;=0B8>S#*ClkVbOK0BzDr}IOR?)D#n26lP*2O{R_oUd%%T_G4*350{CwDJ>U90W1 z)lbiHWIN9xzDnwl2wLgzty65zclMO<+vB%@` zlyB%X&I&$=yIMoR6)VKLL|V0HAaHHJXvw6w5qG&uKkk~x%Am?TEqE$WppjQ)XX5gf zk$G7Vk34918y6ZMwcc`}d*-9~x=EX|Yi%4Xmk+ym2RhfC%5SMUvu*F6sppz)i6?sK zhRUw2B*HJUaO9p zE^q{a*WGchh{Ta~Wu1&=)Ds@`0qxp_hT6f?2sb({2GCJaQd&LYB02Gqht8#+zH~*e zk;|MNacP{e;;q%O%t4RIb8lHVZ_ac?y-c}z!GgK7=EmPVJ!0bQdGqdhjc6LzO?-9H zEI;=!tZk295U!4%9*%CvTM`g%x@yk8Z-M9O{aREOmsZT_)f9$+ufg-0f~~nLEH`9tW4X6I|vFaGB7o>_SjgLB^$@x1|=PrIA9j zMQ69}xRiPcd?796QfV2OPH72-WQRgsT{U?mGNZ1$urN89bL#BOtrqF^owZe{Op}>e z(^{2xS+5E0=(R#fnpJm8ahqqbxA(@4US7c-ofSEk#rUG9EMs+X+ej0hj7kB^b%b&J?_uaei zZrQT^$1gWW8cu<=Bn?EOa@Kb!Z%tm`3La+Wa}tH3uT$XVk|4BIw6j) zwFoAE{XFAe$V;@jj0t!?mia~#YHQPjdFWL-EO=mcKt#m!>5-8Eb63VXxcgyy26#FM z+YuqxDMm7I8<9xc#@NciF0)S8*K3??Y^H{atgQt)fy@BHY3~tOE3Gx56Fuy3z}dJ* zExy^!-rL*9$IHthXz@~KH+Q%&API_8YHx3kXz7WZO-Bb@nJ%N8jHq847N!xkx7*4) zIBTI;pzntAw!Tr|>Ug~q%7apj05b|QGhhPnhOjX5W{nQ6M{UO_Rz$HS6sx3Ir*W`{ zDAq`^`4sC+vF?M|(QnJg>Hwzt4G~-$@i7ppj*<*CLiHO;2Q)J-h*zf4;RyMS5c|+?~ zHneVYhSqJK1?y(nSVmvB|2lTw)pIarGV%rDjZ**)I#a^LU9mTp) ztbQD9EycD|Y&XT4_FLaL*anLAqSy+G^{4%I4PwXAa@6rPoFkK1XgM+^ucfrtAlhrsV6UUklqVgrl48>-wuEAf$HAshtS!aLDAq`^hHM0tb4g?fLvDOruH4gR+#pYA& zB#M<%to=CH0*X~p>|u&MO|iZIL#%0)0%*T`Xuqduzd_^l+eF9QM6owhtZ5(B3}Wvz z$vm$QmwD2)y27(?l0=G))Iwcf#-($xeV$Fr_%r#!g$t*WGjcOiQ&Tgtk+M@P5>@4P zc51u|3%lzf_@w65bird+e>tzJ?J^{u{IaUrPDnhNm1$cyZ{JRGo^Jjr{YqBK0hn|S zq%`%pA-$4oZ&T?p2r5U5Fjk~;pYZ(i8n5;1b4ZkScPC?`fW-4liNS9LB%V9m4b{6o zfpg_&sMEa{nt*Sa?aX%e{P~5vGT~e1du9)_lUb*H^Vt{i^^VKst?$48{!cY73+Kbb zoAFB)L#$9}+pVlL!q!%WU4q!fOVHV85`mr%V9LKTi$Lr1nTXej2oxmEKT!l< zW|Ezwz3^C;M5c;~iSbiNoPv>aCd64X0Fk~MG3#z*TEeH<*l4`!>rZDw4wBn4ZCZ#x zN^3!P8zdlu48jmZAZdMrjjxM~Qt9UA>($>R@d<;iDa0#a(j=1BD=NU(2Oa>ven>(f z?&wfB^z~`10|LZ-kb{uzTh2lRam)IkL5)(gbKt9&|!?;YNt+^_{+P-=KR^4q7M zl7tJ!i_H{?PD~yhD?t-0AqT5L6Qlz4{g8tri!0L7ET;?jBS?@m*Am)@tPcI+3(M() zd7i_677>YX%6X&lzd$Lm#?>&auD)S{3A;Ed&pX5=FDKLtwGpa~c(TbdEv9KsV9ozjr??RJ zp#pMODlRC*-6;F_Ur!L{@$HmHbainPlamVycrI}vzowHF355*>RV@`@C7C5v^}WKZ zz8)wudvr*Ihh)EfS;*|$n7<2^8AC&N?D%PCVOvSo@tr$&9?xpixgsZ#OJ7^{Y5bgA z0nKH@;jJ4sZd~`#hZ~@}L>?(ZATPY2iCnji=OR5_DA~!ZTeq{!;Cm<7!ctem<^AjY zBb@mTAA1y)3l>NCgR5+04lu{?;M>7l_JXS{BB_t@({E5e9R~YQi(Y|V@r_s}QSMRu zQ(|J`kC#1eB_?YyA-w(}S%TRRFL`t^{13=4HenOJ1ZC$o@>1ib$rF*awZ^)nq`Mb> z-2%J*7Q%LZVtx|6=-Zo(oS-YnOO$Y-Hx&K+BuM@2LIT^@J|9Vo-b6BAyaXS7ma2&{ zr0vgWEF)D7Nhtxe;r3X|~M>k4bd zboBMvx(J(^p<~cP17qk5ZthNyOYJUKW6Zhv!6th(k+_Iww@d~7ziP0e4=dfmIx*m8 zj)a@x{dU66@NP%=8LOQRk9D4!$CB8Zw-CFOH!rd6;UzXdHzKhar<*0Vv5?MNh+o$% z9(fm!OmQP#UG>w=UR`6Hssp8|NP+raK2=G^@&Cc83VQv&b*diAYe%Zo|K6!8Z)|Zu z&L5HsE67*hB2z_xKEr&1CQn8>t?1|=KNb8R{Q{U6N|){J8tbqy>3FB=a?l}({(2Hz zak)u56q1%rYTp!Rr{$HGW}Z2llAf8FR#Xg)P;p^e#>LYoPMo?}Se(Or!*Tut?N1fqtY4I=X*RS7uA^EF= z{lO}k#>sP_9BjCJz;n)Xe_QjHUv{3UvGRLxK3Lfi{`IZRgMNz2Q@{T7?CLeo%x2;^ zB!7eS!Wt7KEth`!c=I5xbsqU;IX}&9~J8Q1!-2+4I8w)BRw{rgdy*ic%N)Em~`#0y4P|4QX z2AvHlJk#43{Nxg0VRd4Nvlj}yDBZoBtrbcye02M1S$HN3K{tz8{my?wocgFGGW znVFnKukYd(YR{?3<9MYZLPi=f8|c zJk}%F0rHq~f}dqVB;+2LZ+rzIpl^_ueEp7({|z=V!~OB6AA0^4V`y=vd?-TP2DV%NI;ma&oPIIMl&rA678#<>A9b;e+3=KT*4T{Heq6>$fV-9t82EeYSmTpwFkJlXNpJLGFtqDw zLsb;5qg#8vA?6)E z1U8O*;tP(>G+BFY9ATE~M%afkv-vV7SsB~6p3O^5O-e#4*0WoG%ea_kUmMtM1; zfwEHQN?;c;+lT65AJP!(P}MN&(B(F29cu02twTL6MeSZlU$}9jn_IAFXIai=iI(B&tn007{AGHo0pw^+Z-5b9Aaoe}wzWeUiUvK;Vi%k=ae?ke9 zY?!!eKC{JS9r{l22*l_i>yVw|E2 zU2(~9b8{3+U>%Z5p|u!h9qJPY1ZYrc&_OZGI#ky%XdP0hXvU6fG$T1wGZITRBL!45 z@&wh4cnxVrUd8$HD$bXWL8*f`Ja!wZ z-dS!0_ui6n0Yc-2b7#*(!F4`m%Mhtu&nO-PLX#?7CaY8v;-^Jn<}Ejp3yAu}bx3_O zL}}yM%!$62R6PC*C~Z*RdnP+O$#PBxFEmTfx5(0o6E|AUu~--b2Av?T$D;!67c6-9 zhS^aZx!K#iUUGo{h1_MQqplIHRNSEy=NC$1^=yVaH zlj%IPe(cUz1no?Wh>Fmk{t9xc=weQ{YIQtR#^m*(>V}Du3+(f_@6siMPh4D_uVSFO zBroI8rmUN$1tKa*(c8uh@-YjK%()XIL+P10Y2pOJ((BKL)VW?BM2Z7@^Y-$DzRMTRKOQcQ z_Aah&9zK44em>*TzideQyZ(5FnBaAeTe>qyw7dj6gTyZoJ+n+jGLQ>!i>ZU~{$3pX z9UtW4U>DfV;de8MY>0PTA{-9tnTeiSp&F^DE;s3^L#Un_M915_fsN-i)vn`esylg2 zwJWc0F6Z^lGk8t4<29PgPSTj`^zK&%|`%;q^89`jMXh2+{yOJe;Pws1bOAF(UkSmfG#a=JO!rW|~T_$Zn0 z$E5a&n%%<08s((JB<4N5K@lH|>aO_NjUh>s*y*O9NhW~T&;qeieTwyZ2Au3!v>-lm zgyT-0jDw}*P4>qhi971kA>(l`GxNgAaGYyaYHD)w>9*cpW=KG~-gQk}Q6UuO<}N_~ zfJaF!U=oQMYAJJqL?ztu)~l~_!j&I<`e}ki=bZTU*Q=`j`qu?a1yjL(%zlhyLS;9^ zV%SG8i&>nSd2k*o9KUFMkz>})!<_q-=-~71fJMWx1bLC2B=R5BlkBgX@NC3wP zU}xeS3|4UOvy5d-4G37i+}Zh#JM8W6zkfXTVCD`;_aXG}MqIB$jZ7qBEMf-yUM9If zNX&^@BZ0jBjGEa3g%I=*a-w@kS_cYUs}{`n;Ykv|aa#czU8`oM$1?v^6S6gFCN`q2 zuh3Q!==5Kp)AeX;?!n&=oJq|^imjqc>Cl|R!4zM`$k&Ls+^nn%$;sD0bghPI77lY> zN^4S|S?*x#MsmIDJ+mTD7X7 zVej5~&|f}p{QH#kkjoF`qMVDPr=eFU5!1UO_?U&%gfqTW|4oPGPkDDjem;;uRg|zuB6$)K%g51_yb;DoAw zfdJ0368%7LM@P3lL5afmmhG##vlh$_adU8oLLnm1K`2a3_4oJiu(h>8IYJj7KQ{;e zNY!<>o7FH~j2owD>qZp^VQ*7oX+wfdSX(EG;UFQusIR4^Pop&SqF$TXllM9Y^57o4 zHtIL$JPsx`o7bdK3u6m#I&n-RJ{D+kLUBwyK0e`(Q;NF9dPU)+LjCYz_Mf}46Cc7( zybC*#&$V&lOfI%xS@yY-icEMpB_$=NrJXr-?ou`+;;eJ1rgSmobkf<=C(tbyXwcS` zW)4Pu)YdjNpyUFL4q<3?NKA-rVpx}mSn=GXoqBBz{{8S_uLmETF7CX12q}HP_%W^1 z{ie9s!$-1OP=c5Xl-o)WSY-~z$ja(=y3HKAkwQZ^lAA<{u8O`&C=@R<&a)nPbmhvG z1qH@`{cC@V`}_sWT2qCnUm1Mob zCNp&FSv~A`GD z<`)qeHPhgT)`Z1*t z7xK|6-eedu-t76Gh|F*wcrm%@5*seD0Imcto`5H>xB)77&q6Agpnt((n^WM_nKNg` z1UrH;T16b7=85(H1}?uuo${W}CfEP<9$ zgdLd_6IC>fM->{&$zsCHnN?LcKpMpQ_aD=SFO_8jDQ)~lY2zZLjkZDB7+ntbp;#A+ zEuz>iiuD@@TSc)A6nm6nCsOQ+aj;Gldn?60NU_H#Ry!&dv4D`MA*Voi=4E#`iakvJ zI)Ng?Mny^oku4M%MUls7uMQL|9|!vh#okY`ODJ{<#RiRo{ZD3U`iPnO2kp0*_A42u z-*SqbK*xNDVjC%T)p=DXQgNVt!NFs!bDg^#;q6p7@wts%7+ zj8W#=3nMIfDzhc8mJlQ) z{Od&_Zz1;D2J$V6e5)~iy3RWPCtTh#>0vm1-59sM7jg%YJRbxPxEI`(mprSAFA#C_ zV&Ul|dIv1ahBsMZRvLUxE)~{P7vY+V-dg4Sy)83Ci{&=rpMnOvqk~-Z!`|deT$hRCEb2~)N zrIS3hxx}h{-+hzg;7zj0_T71|O}|2*T>E>6Z_KK{qpG?%C%;aCCBv~T6!!P(6TF>* zgX~nQl9H@}B`BPEi?lVoXGcOp!sq+aGAq0Eny}t(J#MTLeWTvk2``J9n#Rt~#w(3| zyfb}mO^+>8ns@$EgApVUf*5~=h!G(Ww!NSLzWZXafwgjQaBz}oZN04F%WJ6Fk=%ru zabY%MC4^2d7pF!@7hG|2BE=xA5tlC)iYT>)VbdG)!6|3b2z?I15T^Dw6te6+tVfgLc+M=8^{sbe0y z`OicLjNwXMMSH)fqNw9?A#wC21F&&IRH~?_yQu+waec-_2jW&2K7Glp%kR5?)#B;6 zr7%;`rmLHEf=noAs7EM!);=0ZITcSzGHuSolTN~uPQjB}ZQZKVp}NHBUHdkGkFS>f zykqmu6NvL*NN!liWa!SAtDl)KJ|gSxCgShjZ3q#6%J`J%!?kNa|9l=%J{ey`Lr-vO z_JtR?YbGL&c@Jt9DWQjnt6^>#}St#Zj5}`ZUFzd=E%iJ#`lRrDItp`g6Xs+A<|GOfGzoL!f@&mZ_Z;|j_M zqhWS&YDY>qCz7)>`OU&W51Xp70n4j8efZFxy(bU-`s0uDP>c{Sw!1sdu-UL#O=>nM z>GtXEdj4g8`!FwX%nKSBG->5*_^efki2oG1H{XOo=AX1!A+-3-OtYF zdXL~cE+im-+5L|^`{G>>Jof^Mk<1p0LqbReV4i9eVkrt>lP$(HaST)#q^jr;cQ;n_)c_S&_`jXTCHIe=!Ix*O`NYuf8eb8|s6b#-xui-wD8=EPVN zWK9Vt;*Q7v_Qo?qgu}+hqGAqHs%D-Xe|j*?f|Eb~eDKIuU;niIFqlqDi&Dw_H0JbB z`nh@G(x{k)JomA;U$KG-8Dn~`C)F951P)jGY1kWhO9i##treNM@n&G2!W)@eVr3vT z6^Kn{Q1)ieVtt>-U$7tH!@Jyj#LR=#@N59a;YhPE&tdK-<{i?;aCuY=Di(u^K^*UJ ziGl`ZGrspSuX2ef7C^q;&Lv(u>Y?iJWz?K97hE6X)PbM#owHR*Y&|Y+4q(QnUd{HA z5w`}U10@~5Cv2F8p5GJprlM>9xrOA-fbK^=w;4^XEB!$K6@S69`0xZl6o7x?%^Lg= zzYgG;9BqMkN|`M?BdlTUX@H+4_OJ(y%dnGQ+@g&1%WUsJXKo?fB1j^ItaL;~@c9*fJez;+9E#1JJbz7cg>SiO3Wd|Gr(~@i z&@Q?-W@jHf0p0nP`jRY?ZlS(2BO^K4+g_o_&MrXsQ69;)&`^+?nyr~psVZf=yjM7(*jmQKA$94y)?a0IIV}hCx!1ERV>4(%H@3pvikcFMhx6YqHCl~)!y zcXk$6Lr7?<>vuC{ZQvA(X1X(L)~xjH&RS=RRupe8>~_6tjb?_ps_!NY712TxD*6g~Z#><&{g(emCby@x0A_A*vcQH$n+nN5pwYHbF8 z)v}?XoTGR^&&0>y6sEEl8jVIRqndR~yh*3K4pjCvsO&dT*_YUzUto961xfswbn4)- zV}GpwVjC#xMyXC}%IWbfH}`6h+%K?VyhVN@o@eqAm`rj+wS1mDX3w-B_~D; z7$n|unT6yn9DDMqR9zm%JsS6P{smUbzN*$-=$VW7P->+I>%r_bUr zPQ3`V!-cGh)*fPdsU~smm%HG!72G+X^z`(!(RUBDvKB_maU!u8R&NJ#&(FjWH;k{z5jk+J2o-c(9;G{yS+yz5KCkz7NzSqkVNk~#%yOv3exVddAkQW~Z>wFdL)^)b}f`D$*@oPTSPm7Kb8BzpaC#J%)Zq zEORs{)v_g?VdoVYA*o0^e)8n8y?c*k?#s4p&GR{1aiCrP@4x@P_n^0@ySqDlAHHV9 zmJot>qtr$&wGF)O`R5;spK|+SOXB}6%6OcU)umMQ<6>s$o{x~LcWFj>!H;EiH%>A$ zqNap6$n0!Ake_eX>o2GpwyaC!H`if}K-crzBG-xjxwC6^H zlTRg|ICk(4_6Da| z<;$15c9k6ZfobQM2Zm3}zypxGauIWVWf9H#t<;5wy4GCJ5SF1RQ((vv}M($m>mJr^by24I2#tkplN} z9d=k#OJjWl$_y~w8d-Zgsgu&**UL$?&DfnyjT-3wluCbpd(Vl}X9Ypr#NZ@0*7k0U zC&#pDWKB(^q?(nj40#y3piOLUxSqwi3)6x_&lS&#Hx7fvn>9Men+Poq-J`Cy4)D1+ z8n^PfIMOM+k;0uRd=`PzdnEDAei-ZY7S@UQW)q8bb(kg6aCKE{MaUIdYR#9As$YnfG4l%)xsddrG??zw-}V=J(dI1&V7tZ#mP9HN5X zQDg9Y>l-(J{q@&-(m5gOgp62sg=yUrKgv5OIxwZ3;4_myTSM@CJ1qP z`O{Bhq7YO5YE$l{WINg&pJkdb&$8D23IhB0D^NAi`|vAQIh?0c-PSwj+Y z_1?C8wMXo8N9=P;S3lzNeK>-BSVm$oSF;k!Np1vN5reIeNT6{^aEQa)W#TY*gTo-P2@{?{v{ZR{ zD3(%)HTxXNx$~*4 zUI&S~zu%CYY|!>3hF4Te5P}c{bLB3mr%B3^O`CRo`^`7-DgE|~tvO;}=-(fu@Gt2*w=JA2278>3a{UdwWS~VSWZt6hlqf(%jzBPJ}##MusaINqR1Y$Y$jTjir;b#!V=+ zvg+$Yfzr$6-NfMF?(S@Brx0Qx7^Oz$V-v!k;~lXwT_=w9pua+kgx zJK;9$gpm4rqfsnWkW-^y+1EcXK=LEkcJyj-hO{*`H`HVC%gc#DD4@NaxPU_5#8x?e z{N$-WPaHy}wZkVXjz(MLg(t3*O$8}G%s2iO`6Bq2;+v!b z1FnA`e~e4I+OM(_b4n)l?v==E{me7ZG9=rw%x~V^ra9id7L?*0F<0g3f($9n4*ovQ za!#n6_2^$0!A4rlbedi#nGK@1;(UF5;}|`oSI07^VehcBn~)&Fyzl;a0ZsW!zq1B? zf>(u8QKECh~{jix&k5#A2tJ(NRHeSQ%R@*w>K% z0_F;PPVImyRdnIHnn>DdtyUmPuxhF4^))S{(&Rt%0@jh(9lpdmzJPUn9_u)@`U(3Nm9wwjp1H_TRNl2vj zjnN0EFvoT4v>lqMj!I9TM?5%qco`j%dS8!b>OYcU zhqZ*5Sc$<(($^=-`W%`H7h79n)wqm_1T4JQjfwBr%2%u} zq4&Z(dN1^$_riOI?uGv|yn*)bPx~*R{R?RS(xLu`TSbRQesyjoS=FYrv+xnDe3SSH z+7z?!($jmIz4XQ$xs3MQM0@U}J@?X{?Z)r<&#ffp%)OmeF>LXhEFu&KirJG4>*LAZ zOl2MOn6HVF+e-4e-9g422SI`#^?_oN(d~+-Iis5xT!r`tiosO|t-v@c6FCeQ-`*tO zipe)m?s}QXO?7Tnq=utfP{zDS)dl=dxZ9bu21I{ODnpWy66>)-#R>%^9co41K_;1F z#6*5V6Yn$Ui9+SY1O{PfaQPv>5tD49Xz6{m*#b`V4zO0^Z6Cy1VfL<#92p=rY&a$aAtI_qG(bHR%Wgs=yoVwT*!*uV;1AeNG{!Mg=w3I| z4Y8!iu^L4C3zq@5?;sn(u#Ksp8}_CbLF2!X=hc-58^3;74O5qfzkv zqFjQote+o^9X^U&`QNf+G%9{H?bp*rO{6p91Q1pvjn2#o`cvKD%#3cAnLx1;`qz4j zy+VI-9|zk`u|@Q+ODNVv-M5c}4WQVYDfVv^yOaLZJPtOUVrBHNizv2({^U0f*0cwg z(Z9Z(Vo9xJ%-qF6tl5`&WZfwQR`KBTu6vHw=z^^nP0?)|>uyA;p%`pY9w7TSBo8^sglpTS14xG&tMhl z+_dk?(FpYn7D@TvmT1(-8jQvU>lh5*cjSzb(}--g{~({alizpcWZ$vURGcx`eWBQW z^G))+FSvjYI9~{Mtq;&X49nOehAOwLOIB5-YpJZnxJ1dx$BKGz(`#!h&C}RU+P?kg zU(dG*1f9K;CP4(^1Upk&5=0+-@WFRGHB8Dp{*t*>@p`BrC#SL`C51eLAe`+=n1{~s zV(*F8*p@e-zRbnLWo2iwa5caK3+4yQ6GXU*1cYmB?L9m^qvk}}+S)7Z?U``$uQS~f zM7Y{by~Q;0tDoD8RDgPQ@^fVeF4$>`p2HB}8bA(6#EN=fGj;UaN6~F%YV~?%+`wC_!+r&08B}@zd zYG&&3ITK14yx1@J*O%;(Sj57-aXvFr}cv#zcY*F56l4Nv*@UIXqoJuuuZkM-2o zy)TopV)*i6HKgSzZeLc>ko3MnF2)*(gkt2QkmD+fN?z7VHxtZUbSw+{yB-i=Sb-8_TUHY~Tv?gORu?4-zyJOVNu2RnmkJq=xOUyo|gIR~P#nj4p2p&!K44i~{XI^&oYK~1wNo4;uJyF@vVAy-0 zy_pEjmof@Up}-NiL`HI~uWusj9h?XW+TYQ|-Ia*awoY#D%mhwgu(jpb0od*b=pKv5 z6WxL*dIC=rizk|cCsIK}b(L-A>L+x+S}SjQVkSA6BMZ5yDrQRj!$KqRX}(GN)X<3Sz8+ z`K&PC%Mw6HV24GZ6TD5u@*{@pB>6VM#2*zPM>ERo+r*3X9lx=2K;0!2T1KEGT_?cA zy$Nv2o5#dYq<9&{8z{bp;$4UEW1aB2--u*mm%!p|&wYbPHsX`y_AH!WzCDkH4ET-Q zVJGm&H+kgmC&>-Ad4l-{8;F?-2faoP-+{Tg^(EycQ0C|6W+M0q6DKHWY2oCRl|%!a zTwa%~5mZ*T!KGE;t`x)>R~lEc6%|OvDmv~Fj)xRvJp3@owmy;l!#h#X+FB6HY#02Q zSQQ6@{!TU{BbNC;)V&8>R9E%~{@(N%%Fu?6QUnnJyJDx=YfLoVB&P0~x+$jYZsyGp zO=7wx(U{_@F-?>KHzKu#{;p*hz z=;5JKjdoEvkQFFE@F}8dq8K4vfj)Sm5y%wzZ7b~mvtotvmMLc4(z9QK5fd*AcK$75 z;yS^uvV}JXAPtMS=_k8BU4HN#gB)71 z60GT8>#6qPNI#7l2DY2R8t7mDL1D%8leiC-}lVk60wkPQ0xOI%p2?G7Kpbzb%dlQBq#5$2}vzAzIgK0o+4%)=-D6Kb z^Yr5R;Y>IAPT=N7AUG+i9Gtm)DXCV&X8#Zdm2!dN?Aago!_T+p+>Npi?HbC8i;Bzm z6q>S%ipq+HvT7y7&ZSE+F_)vyoxOB1`bvz3QFuA|dLbsu&9>`QB;rnQm@FeR3r~!&r}XUnqLTc)-0XBf4;d7oxThvUYGC=;vO}eaJd;@B3Dk7gJ`gs;ofFD1xV|%Zm&0ic3n%s}VfO zdp5B`hpMr*)z{I%mh!oiqY6}VK@g0Cjg^Ist(}82c!PIkVto&J6aMr{_BNtvn(VVu z<9Xlu0&9&#iw%Z(>`GW@zhduWV;Q#V0^0(@L5*Z8huf8p^SipySY#AirBw7d5z#~8 zZ`UQHxVeT(ty=oWbBNR;Mm`8ZL!Zndk=BFe$_UyCG~s|}kSUXbqFc}{v74#F_gCD}y|1Z356a15T7v6G#>q?GY^p8kd1PTSwwhEM10=c~b*A1ss znY>A-H`;f|iOLdBOKcPhJgovxD|G8-3P1#dR*=vJsl1VY&salmLiQ(PuO)xbhuOA$L>YZk7Y?CguOwpcNL^GoSJI64+QmCckgI*#xGC-pper^Em&`xQ2{( zBOFC)vGa5Da7Df%nVZl`!B>tmz}@&QyMzg4>Eu{j2Q5Fff?~$umbOZtot0(Xg*S-5Zf9!Bi_TtaV`31Up^$g9bZA(8?$ouFbiOLeYsdG-#8J`jSAPEpdD<@sxO2er z^W3QA%a=?aWyhCKa~sp#Y&ZA0=bl^aXFzMF<%6-2z{bT!BXSLyzwVuIQz!f4;TCTa}n$W@O2UMu>k+Fl=bJYajc591nV2{s$%poZ&-2`7cQjKcPPe(hK7oN z4M*%UB3qT&=?JB%>yZQn&%0obB?Ai@dw{5h!tB(`fBp3r-nJPK2GZSCAQT=?#~Z{$ zXU1gLHOYH!Fl|6#XO~na)<`6%$L{GQ*9f`{V(pzg90)M{bhOL7J>j^epAbWOTx@f}XK z*X<#)y}F3U?x%||j#K>-fR z=||Nr(WRy7SD~d|FE3BI939QX#>FtP*W#{SX3%Hqr~Ui(9sKL$>6DbL%+#AnnMug5 zffSe`X-_(GZbnW@PfuHEX?+J?bxl2-lnp@PjSVv$IBCOF zyn7%dI0^=>dr<@8KiI3jMo8Np<|k$!`g#Aw?1i#-in)OQFEcSrG`Ym6py565agyu) z8@UiOU0^$TNd%*&J0B4BA*Zp4k1uIQ6j96Xzh`&iwXsRV{1AyIH8|hndX@e5Fvolx zi2@V!i&xlv@lmcWx1mge!hS}KxfE7aH&v8ZS63#d_itd8e|kk8#( zA@XAcoV>fEhw1L@=+ragZEON;echZ~oK<6FJ`UD4WCOLaL9;o9g*C#T>>+Jz?Np>< z2jdB3Sz5|v2)VNl2ynB)tEH7SCq?~GnM|&*G`+L1rX2JZc^G6)kfZ$pu1lPh|H;@mA zy@Zp35#r*m#m1Q4#axb>9UOkos#W(s{K!3E1h7NTU2(~P2wMRnY)8yC7Dz3I?MCLp zf+r# zI}QlB6(HpH?)dY5qeOZY{7EWN5{cyQMweH=SVF`HFOx*rVBJ#l^2_pa@^h}ll8kw( z8fU?HmYth_R2|qPE=JdIZZFm}sms{7;tI<;x2eE?zW$>XeDjF7^&iZtf~~ zsxhA|Z*_-2f>=uFa0mf!@-oTVS|Wfo0r|>ysF^FWQi?dKAsvM%zmKWqq_**R_w%*n zs4&7$_JXQ*U42DKd38!fJ}T12Ugo1a`AA&JbISv|u!O>?9Fo5EK&S$I2o;pbpQ%&b zN{)Fo5)zw!@hSW0)Tr^bwI#(lW#uKsW!;_i zP1Uf1kRLEWWd-aAMWo+cRYiFzSOgrQ5h~ye(ui0~OA!=rRwETlL20QN&x3?Oyb}qc zAcNYJ-D0_1X3^{1YiVz-v?ekLBF9l>3;9E`@RZgNJg8GlWKc^>dk2)s`t z&772U%V1Dqd{CUsU599NIvkr&8z!6EsVqJNuwb`liW-V$!s` zj}p(qq*@kkfLdaUC%%PvqH*?mnlw&Rl=vMvfrDok5wSaHHW8l{IC%EZuvvj4N(mY^ zD{$~+pP{k>2T$f1X;vWdJP~UVshhu_kW)r0(S$N@5_?fL`Kppn-X>8M_dz=$SzVhT z8Aw*w8bJDAi$~QXBwhZ#>pSQO=lDjkgX5i=y%lVq+1nf){L7qHk%0Yc(SQAQ@$%*9 zWMX<8oG#Qr-WO0T8koh`m(?X&YlG4%%W1AA)X}$LMA^re;`$PC){xcEE|$VLJc6_? zYFY1#7w$Bh#77dk+FM{0z`54e*9UsCpv`Rp-Qy1LVnN)dkr@^-YewU{TZS|T(i zWlEsnLE$r#t2cG(wD$Hc92Y@ecV{af^Q4s;pb$AZd3$masBlQm9GtZD6iIrEsSDf7(YIxLV^4lq0ZQ>Rf@!QSw!bp4$Qe!8a!P!Lo%KUanfC!@m_}J!O z8z02ax`)?|ZHW1--QcxsA+sC#1mrq)cLISMHKem*D48G9*{-7k0_it|Ghjw+*EQDb zT3XvlMpCzVrW(pCp*6E%gvh1vwh`%}kjr4(wNzMGNWq&zkyvU$@>a}q+|a`f&VIZd z$|PA{zE!g1=T#IH6-yPQ_acs|ykB(iJ!8P}mCpe;vv>PjdQJ!mnly9P1bRrwHqT81 zPNsKce)sJe7JfFxSFWVi_Lkvru)wI$-{od8;EXv`#y8n4ic7Kz3i3}LzeW!Syn}Xd zg{SMvB%1asp;&dX9HjJ^c`?6yd%TAFa5XQ-*l&C?Cx_4{HV?`is~G& zb`W)=mTDa~eOiY}uats=OffHg`ilM*eT8T*yvEWo3~V(Wx0%8g({cSM?80GS$5Plp z3Y$V?c_KXQv20=|Epr{tPFk{`*h#CJS=dQ)-XwO?pllX)(iN^IJ1Ozw z5bvj{=L{wWeRBUkxG5eX(R9tOOf+5vZ$1F!znpxNoy+VaudZ%PBxj7Xn}NHKnwzUY zQB{1CSg;x@ zQCuP;qag~V*Fp_1T8tXA=#f?TtiAvK`;(JlWmTjnAO7~&pNI*wi_^%s540Lh)JI!6 zt@Pd3H@*G#rkCH1QevG)eYtP%d;HIuHA!J8%l00-X;UOrL*qrPvzoa~B1LGMK z`<)Su4faxIoz5dNGSa=d zuHqC544y1!(n%dW3nZFCap~$5S;*1I_s48e;Chr5e001h`^S&9cJBHD5UoJ(K6CB* z_3PKpbT8Nuj?_V1!W{+f_eVnSKFMFZ1oxvx6e=4d!I(8@#Aw-uw_zxg^r`4k4@m>O zO=x>)cwO{sy~N|uQ07yuS6?q~Ju#C=dUZl6c21c{S4`|G#idLoH(Dk`t2dDWmJb7J zCK-@@>tLmCs6pagO{3o0L83aseeBe!Q^&e{jw6ZC6UTYFkknL_D;O1} zp!Rg8TYxQyxB|WomOz53;gbXI zJ@5oQP|`J1*e(j|IRLh204(vQ5?JD!rKdUrg>3*V{195=Ya*r21a=r2C8F>J+jocO zz(ddA4*$j-NMEvN!9$Z16I1_uf6F2K{PN2$2YxviKP*#y&)LMq$6vn=b8^h7Q_+`5 z?)|WAHo3mOwywUu4hAfw)gMl{Sv@S@<#NI`0=RZ%<1;hef@aO0yL?zalzVtsD3sDh zs5;@KYZ&u|W2B9f(_L}rERH#`84bBD{kpKIPTU2jPHt+dcbFRrz1<19%uQLYau(rd%j_i4fRv`04 zbdskK(LQ`{3m9<)?B*TF%E3;(o(zrGFfukPsAL6*SrVY<;w`y3rsyUTt78SZpcZZ> z2&1_<>j~1Zore*u;Z;D;#%JY(qRaz99{<6-bPO4Q`_Cy3vM3JZ!{DHZ!j7f*J5OP& zDD0!dz}8V%1BK0|u)>kyzLvtaQdl<%Ye`{&ydfoeBZUp1u$L(;qQQt{wj2i5IA`PY zUZk-06jnJ5?0+-opZtkPhd@&@9jEs-43ENk4hJPLc7!cL*EPyGK0cH-@GCOK+% z<|mgDmZW)-rGzbMo@6v(O`0beNZ6BhP3D91z&r7Ib$kJ`}0?qX&Zhh@>s6qA=R z%sjl9mn2@a`sgSMdxXNqQCQtDu!$7box;{q*g6X9*ay24IQSY`(jU;0zJWB{0ckk@ zuC=6u8;Lh=Bqk;$rKF%)^$i6=r*0%9T)TE1a~MQ@nl`kS4#zzyPXj|Ss>S zvxuT4=;*M5kxWEtzQ~2b)^_duyH?zE%}veCO-+p`vZkyt>C8lcdgM` zT3V5UKh{=OO6YVpGCtnK)>dh$oqyM=UQmGF7;n!}Zf-WVuA{yE6#fC@gC=>3*V}ZtDAgUqMiw2?_;;@P1j8IP$M0{sQG(^%62;bqpeO3iA?sTW6 zY6lmE0-fa7RPw8m{EFo0NqX@8pn=NtMO3D5p)%dLV-L6MmQh$4g(ZD!Ajr}%Dqe=$ z{jLlKYeivQhS^IxDeP=IZWSH3jKW$C18Y2U220O;}r*fnsEsQTC zw!io`HMO=5dyuWl)r<^3G#-gHR^j1qRDJrb(>SDSO08^45 zeU%wE+MTy0I=gz6mrv+Cip2LXj&+VBp(!ljRk@^ zA05{vN&62Xs5thMPohNAV`HcPgJb@ob#paEp(4zc&X7!x{>a;NF|+i%j`edR5!K0O z%rl>gAD_E^>WrdLW~XTCQ zQhf@`8g%LDb@er{o>o`YG^D4eqLsoO95{c?~ z-KgosNHl2t4P;Uuh>6kzVrBS5{7HnU>C7R#u(M$DzDYh;(sEFd%nBo|iC>J+`gO<> z;&ERMS%UhnnVB(Zot;THVv!COlaScjn}>jkq)XALT6r-#DeWeVX-O$)tf%&kj?T(z ztAb7ymtKmV?S)ez`T% z*b2M^OoI;$Jz*>bZH;iY>Scm%fu$CeaUC6fJi2NnqsF03X`WMme{1cYFXE#fdv2{xQ7N8r|7(|jDf z67eIY`vnrM-PSfj;s+wsvUsEn;Ay<|$hqJ(Z~>{0{4==VRdB(};DQjyumKeoM#>Sp zx#_o8UKlBNgkbSkpZM{s9Xqyf{qQSzzD6z!?r79ALau>4LiOw=gB5T1UkDB}CtSHx|TA`EHyd6*J6 zMyE?FhKy!gO4Iq)YHe9r(b33FsVL5eU-)KOV{v*Jw!)=)M7AAPPlPn5Ua6JSTVXb?20Vuou>Tw<3CFDF85$Ryx&Wo z`^(2#&wqaZ{r>-avGdq9%qcx5;m@zO?cM+FxBm<_T*7`Etq*=85_0n_fBhhM3<%>8 zfYa0pDc`9RxT7bUQqj@XKO{JKZ1^MLw$>zI zrbB7B@S!zV=c!nc=K{q6^FL4*z{z2 zY=4by0&Im^f4@MCwqE!CUTGTg&`GeUo*))MCK*LUB3W})DPeO z3Gpy;nksLXdm|-s2sZS}<9oM%_0<<2yuW?-?%g9N`WZxP5%mz8oPxwDzF*`1xa@?4 z=;-)((#G+QmW_M3(WE>_Lj&6vr`#`9DtW2GIY8KypxuQ@xRf^V!X@NrIfTfL9Xr0* zyydg)+rRiva|bVF+Uj!}-TXW}cI_jA=1rk05ze1s*bSBbFsi3l4dDfIEg6a9rg zxIVcH^VBqxW z)0aQ~=#1&p=S;G;;w6=h-GYbb`ivhxeR^P^&-{l%?OeRjQOn28PDlinP7FV~2cjMg zN@aifk$%yY4AG@^u(F!!FY1?HXk$s_SBOFiRs(}gnidg+glxxPD#*~YagMXChsUT< z$OH0Svh0qwjlrA;DPz*q)E^h!FWI_?WD`N$$(y-80asN+1RUkGAV@gUEqNh0Z9F90 zcxZO2!ot+l__$ww{%zNnUvB$m=N~w+r8A#%YTIZW!bvpkh4XRO!HikCH=5u6 z`5TwM*t+$T%LNICNkG!qoJQK!)dTw_!-gSaagwz=Au@8++O>~Ac<(GPlE%0U ze`;eR#U9WjwjoFB>Vk43Mo6aoYG8z9ge?a~mIEWcx7WQfgCUv_;NHK`pxEAlG2h!u z_7sH8B|+Fyq!W zap=(5vsYp(tIE(a!qDDP9E$)x+!2w00`!s7D1<_ikJeUI-V&`1L;mhd_6-fK=qHdO zwzQO|oQNw!oqT>-NF<~c7h?z`ZR5onjpnNpH;}}4kU3}%-7medyQQPCQR3k&EJi## zzYYQd~Gjgt7z8rn+YJMkb1Mus>#YKtWF|pVN z6&vkFdpks_X3w59Y04CLsT|Gns&yT<0qBs;uY<}g9hQWlctA;fjnWJ?{XJ=yx!HuLJq1h3b6nL7Lc{QXok{_u9K;1 z#kK1Qc|-owl^Dk1Ff3G~D$Znql`WcL4_>->_%LF8If2KN4UxpUiA%#o$>cIIyk#!|Cvt zB^DzwzVhZ*0b*h=UA}NR8rdB2@dwlv4QTRC)>&I?3*!(8XY8a=c8a>THl3x1&9udl zk<5J@r%cF3m$3$2XK!Z<(ploRlRnV=>CxMWg;~IYuMrCh6sE;!Qj8~*L#pLwW}@O$ zs#+Fv}9;Tq!rYPdKB*;m1_oyLx8K0|i6Y)&n6758A)VwpUWi2??AOb9ki@rbg< zu^_`~!-i*G`R6}hdgjG9B1{RjPR900VP|I6d+uHD0^$m;zPB@lcks(VSU%&pDW#QA2Sc6p9$)W|1Y zh8GgUAqJLB-V#$LqJ~1&QpivOsiAN^eQ<-1nL{BvC}chza~g$xdKlOV6!swsdyv8w zQ&@*#U^6M~Yzmu0VVf!JbHl(^P}q4CwvfWQP*|rvSaZ=$-JcUry4Tk%$q0Cz&x~y# zo^-=9lPBF2v2$CUEh`MGW*kvw#fv;LEyV(#jGAbY#xh%%Fbi6W#UR8?NGTS&-X*z|ZGk1IaR|T4u^Jm8hSW;~fO%V{eqrr5EGj>ZkDYU)w1 z0IkjHYAQHkVRUq%7UYrYf{bTX3IC=|;%Mu7jcK$43~mlW8s6goPx z&cK(d%gUNZlExb8X-X?pnzx0pwy?rDp7>8oIgof+9s!=nnPWZhv>td0N=?6+QkZum zHu*Y=6`r`9N;<$@Kl|6=3(*IDIsfPJlkl8fJ$&TIAAkID^fI%JS#k1&DJKerB`TC`E@!1V}jeSQ^+{H5= zfBf;~3d`{`X3Pzp;bkGxssrmv$|`Fr+6AOwgp;JM2wk?T4T>@2$M}u&542zqt<|P< z^1wds=LjG5c!k2jSs65j9JoE)I=l6T!jh^cg1O4N#B>AzHP+Y4#1bi54LK{cL2eHA zDw}cReZ}af<>c*dX@@~Pe8)dY#E`olEDOq>P5w^ta6BNiR;j96DXQlGHVUqcWOJV%M%K7ZQ_dfje^Upu~>c_ie z8%&@qv8j=_0qxqkgHGTsIEk+e8xG9i5 z_&XSF)7$SbT`f&_!+G-?>!{>Yc(_Y(!s~2pVD63==kn$v8{S-GK!iGu1b4-q^AD6k z#X&}oeUHS|*TciV_2^yl>s+*lE6Iy-fh6$r_4Z;Umdd;5;5Wd*Bq#d^aPTH@@Y{rg zOG^>3(f5OrPo;Os!O)Tg>_?37uGluGq$o2b`^LrO{a;?pZ67^<_1#K4#m-N*ehJ6( z2Vdl8YlNjocckBq{F6Mka{k-}%h5aWGiEmv&-`xj<{ay+OP4eMC;TcMH+jk9FTecq z8=;Jvo$8};4p_6&`>r?^8*kZwpVGVJSfNxZ7JL*azdJVd@Ngl=mj0i>yJS;z1Q!ZE zKx3M~>vzSTbcZDxHi?^*fA2JAxMS$J4#SLFL}6ViY#N32qp;3> zuqI9V_N;~cZA7nkI>|z>Pk)E#_2aCGUhg!?q}LCA0=h^?b)chG(NQII)PTNGZ-?bO z%?Qp>SQ`ra5rviVu%XO*>ON@GiGtLC?XTRMPac?_D4fVc^Ct>nI(cwgu3; zFp@xLsYoPhDJ?6^&B;kmOSy@|M$8}DKt~sG_SIj${PN3vr(=IdNs_bK?Jn-$eRuF$ zPLmB{gzTF;WgZ?LJ=w{>?Z%v)RBlAmb@iS&ZRX5LvYNAS@f=TUZ0pr1TgwX4<748I z^iE;W0-useVh-v{`%T`Yt?$$fWmA1?8Xe6fy63_MMMhoEZTt#*K*mi`eC034$pIB{zl6g@pw~ z5xkiQ7j$x3W>yYtRe5A$1mFXzDOL39>{OOf>4$qIV~BB==JEy(`$=&Z~-`t{de zZwcq?wZCo!tFtC*OH9m`i*VW>2sJ#cjEUinbRmnAEjfSS0BFUj1wo?~8ViY?o3yF9 zr>UZ{rd{KJ=(iK)t}9or1mSY!GoHk0>cNRoj07N8*@^e8V0JRs`Ky?z!Bx)`^4B%! zY&-2YY>4oMjnRMEh7HRCy0Hz}&s-Z0-w-Z@M}CVD>sop=7Q*(TW}&T>gNUhV(1e43 za9Q3zQG`Pfj=&Cim%a}-JV7q@!+lXeNfGq*riLcCd+KXYey+5#vZSPjHG2&gTLkky*+x3tx(WIGFN-xJY%6WOCjJa4;%m?#nojeaMsl2Hr)oRTz1*z%jhfY;gv;u(bY5_hvq%i)o)oBNqUpZz1$IiCGk`)EI3YyA`OLK0- zTur@Msd094-~9y$`*Tu8br;)KR#Ic|o)bpw3R*_`=9?gGvH!EHK-vHB*D)rMDdDf1 zu&95F8bjoI-(#KCxG7WT&l*3v3WQ21>4Nv6ofI*W`>%hKS;#SU99!1{_fK+WJ%fed zI&t>=189x_VY+)!TET9f=mIQk3VVvcIP5YALsUWvk! z4Jd_QS%%_*6%}PDHCR#t9kaNsyebvJKL(Vh)^ns3s>Zfgj}kCaR2C$}*8`c17fKP> zIHC@|`)ChX1XV6hj!sT0m7T4A4HvB^`8eSB%wKazuo+p=J``&LN z44bexR~a^lzxwlLb2EZ}S8{6h`R6rON4`;iwd2cedWnU+KK-ILjD($XO!;DznkCn7 zD0clZVr{-)zLUHb7A!0;M~Lor4(-j}Fub8*7vR*yTRBLDY>AvSfVK(Ns#vzWy8_WO zMTOJQrlB$Xd#tI{LbZD$pNN^HK<$|u(b~{%Wt!mLCu~>yA<3Bfa<-Lnverm-; zm5Nz}Kd!CSSk3iODwWpO?F^K48&7(kB|g-rfT8Dsp%uW;3Sh_=1#+xy3r^iAO~0`F z&%B%pGZ=GXd0#-zQ9HXMS0SFngJDpXot%9!mJUZ-^TFvo7lJR4V23x zZDEmbM+))~^zc=w94gD4=J?xrPG7gs#Z763K6c%`L^60SuC87(72fN1W)OP$&Ye5a z!8?4z{mU1xTm8(viv!Kj&yEcYhRtlviv~kzFt8wJ%+WI#^qdEJk|=>mpyx!w@i$?o ztcNO+j%1OX<0s;}yW`_wjYHUZYbk7ZRgm>vZQWv{pkPNy>lf(f>M`B4N) z8gDe*V|Yriv#l+RtaG?8R-<5aYx-4Av1glh=dN$R5n0+j&oR$yMH6}|Dk>tFEBv*G z{r9^>VyKxf+2X!bSpHFl7X5G-#wiDM@{Y2tX&QNUV^rNp-#(^^iQ3aJE}I zz%~!PO5j>qOVC(>eXFdrw6eCLskWxNw5YzcJHEgc+lXEyFbGgU1ck1oA_Gc@2r>oH zaX8JKENbsZschTHd>8@Z_G~4)V}OBW(?^9GHp##L=9e8Ge)!=ooi2=o;ISrLGkZAP zvh?DGFvCK_Ld(>1yLRn5n=ctZW|R)@vIw^E^Du+b@R}4AA>#AvMLozt;3)ywvPHDk z3e;F&q{d!s(3~}l(l0&#%JXx@x%;hDGoFaV-atE=UuP>S!pNWY(J7VlG#ENu?oUYQ z^+3V>QSzSRoV3K_J8p(d^#MD|+Zzn%yl5~{&DB~#hz1d3Z);~~t&oZp7M6A@nO%2) z&{p4T5cKF-J!wea+uPa0vKq0T(K9HR6fl;MZhSz%grMNy06!luzp?(oll+3kkFRxw zF-!qxogK#V-Toxam$!hgLa9rkZpJ~Dj?;2i|a);v6D*9Q)W=HJJ%Vim- zf84X<*s)_5ZuYUnQH~pjeX|$pt3W8Xv6=EXB-kn!ebGVa+TR|JZ9_bigRHIw7R>g7 z^PKIu(@+~ezI5Rta&)i9rqsi?iM6_%NoG!9N%D#viVPjaXFVD*gs_3^##fgEpU0SF zpnD;^k-r{cXR>Z!mLmd$2rWcl!$~HSIS(!JJhNBw_o*W6^nJYc4S4N6cKx$U5RsqA zbTU1x6C1%_A*|HIV61@D+hGhMu}~u}`ZX$QtHxAh=Z6Bu{vqOt#%1%`42v3Ar%Npz ztnt7WjxL;y=Q=#`>LvHDTefW3nustTk%n4J@d4>=>{d}T6yhn__UMq3xcJcu{Dp%cdf zA^zS9T{JO9FO)tF;YZF^Ac8_6V#N|{tqLLT%uA*DkKDTi6^oWGi<~`Y;gXP<^A|2( zwiM2b@d}~;s>dFCZ2d#){cfDx}~BXUf|Ky zLUxDQSS?{#Ekqv*B)da(b!ll5ye6g9kZN*iZ)I^ z($db(50>u0pt1JSx-RDcUq3%y6_Uv`Ry8FhT|Hgxyz*61q2=tu!zaU`Hd-O+YB6M# zbvaF*%xgjp4jQXVxlpKd&Ag7)SV(sR;vZZB+^hy}rU5t8fEyR0`Iymum<~?XF_^}~ zt2Vs&9+~ba>3Hg?j=|I#s=$m;jS=1W=%X7$t1d1?RWYNkAX!Ez8%#xUAu5R(wM0^i zsE5H+k$y`a=^HyBPfgGT-a>lX30`l;n-6&<7#6y@ASGUAgG$t%M{HE2`U|m9k-j=I zPjb{2uoDUZ9j0HSuZCqRjCBPi!kAvhiXFosdQ990PpnOMjz58pA8i_cwP}1M9lsx5 z;dv)KKY^eE7|Ebhn{J^u-9k=pAtFxIm^3CvGyZ4Oq%)+SqI zvbC9mUQ+;nH>0A^VGy-E^cgGS@orvZ@_3trb}c|%J?0>R_F|YCGPamur&_V7t@qHP zzS?@`o_+>>`s1djcQHM^fj+%u|I?G0)gVxCzER%r=dYNU%1TmtGd>18>rcNNK7F~c zkoj})^OtJhzPdWnoG>mc>EI9V?T9azq8L>aWXvGPfH3HpYC&XVU(dIXEnj@;-Phh- z=VOC9n&_uF%t@=OD05h8gr5T$w)BPAey@>JREs_a4-6V%wDoGV&h|r zQ5T;H#~+*1%vS>!Bfq@eiEtv?+5>GM#-D=ysc2Tqtl&xj6NPz4;^ zJ#_iaGIZH_$*{_0d!x|K3<^hdvQi?c?@Idq_qToj?kk%n^^~0+7RPscoI-7hqI{ca zUKTDBR?W9_keMOE$c|GXA>)-Eb01x^DB_-FPppg>YlcW8I!^V=FcK#;P=;Lxzd}^i zgA?CVqa_w;myFNT!7cntI9^DUksNEt)a6d%i8>Tig28M2C(O_%U|#sQ!4tx(2SIB1 z643Y!U^+IjPn%+th$jS#l!wF+9sJ0G$#h&9f#d=CQHk8T6^_J3@p4Q5Cy`qMZwg#U zfiWvGX+!_GL~iuW8snO(XLJ*AHYkc6!>}oQzg#N`kypa-TjkYYkUoEs_?%@tX+ls=**@{v7e?wtOGNz$7PD;dN;`)GH6QC6+l|fO?8A-%m z34v9QQZIZpaGfqECk^S|A{fb~GIaQ{hwJ<5Hj!r&qZa%P!BfzH7!W_wRIA*Da8PCbwrY&7+hIo@FJ0ed%oY_q->H5Xy zi7BSU3`Mvb^S%ZVPr@5Nz5o6Yl=aEUfyIdam?Lrl_n{do`&3qrGyRw&xIfBcy(g>f zkjaLidm!kr+!Nj|Pa#eUpo1stsNyZ<(83V<3z@>gk}iVJLsgiLu%*;wbYqB(xq%8( zi3nm!Kck4&cj2_$sgLgNL`79{gzx1TUw)Mw%V|}H(7Nn76~)E%9yrM@STJ*pw5{yw z;r;vf?9pmuCocR3Bl@owPH>`Yzaz(6BPqO&5DHtFMkKREzDN{<$T8E(h?3kJ;?aOZ z`a-gTa3N)!(8j8x7A}mYCYgWal4Vc68^NDJNPu%Fk~cN1C$h%hN5DC`wybi!y+Inh|9t}3OKr33+fas?16kXTzgx+(^}2i*zgfD&O?5i77FW`Pp3 zup-=HZ|JtPCk3N`;np@C$~@^RD;pbSGHVq1b*L{duQ#^_mz0Jway1JenocqH_WIiWuAVfkh>ZIF{6+j@mA3 zsD`IF;f%?XXAHsa{F79dRAK%wp0JFNREkHN9Z7wCC7wj4>+aSKF3{Zn`b#W6eJmdC zLyTKfT1r9@i;7|s3sNsIXE>Q=AaR+j98Xec4&e@vIxq^<0j-N)2%mh1d&VEcHk#6q%U$Y{}`md?9BL*IUvaeV7 z;Lu|!bO41;GeKYG%`<}hVW86}v_FL=`wTyJ1%*aEqd{hVs}{=7oYIC3r_dS-ThRw= z!u9R8o32E{my>hAm&9`OHQ`K20pZM^zAod$(n?Mh#Pb}C9h+nuIHlx~dQv!5WM!r# zCh*6KE0?Z;0vemn`snDo7QW?TOI>tyedl{lohoCS?n`x@5_GB~eT%N3Y=87gmZ zvMcBIqAu3nb622HRvJWF@vPA`I8T^5-fx#YGuyTpqeK|vwLcWBMUDluMhD*jj-C57 zX)W+Ge{E-tfOr@GR7da9_XtZUNo$BORWcHmjVO6@qAy(O`mOD*l;SChWy zq!df&iNacDZEua4ArX}_zCTD0S5gFXjlcG@ zuYko)aWbL^4UK;ONtm`Kstcj|mx#C&a3-@+kS8Ik&|q|MBZ^R2PA*l1O7lyrN@3B5 z+FRFP(u3}v9s{)kV}_7XD>UYg(PR%l;BxdROt%JLh%Y zVnC(0$I#My;68VV&e*umwe`(d4DK81ENmcf#jq3?qwHYC#? zC->~L#_%U_ggt8ew>bmJd&0*Y?0UIcmi53kIh39|vhfF>`#_DtICrB)Vd$It;IJbb z*BpIy{7yH8-|1-PnHAmX$m6Qb*JmE6)u&Kw8@2kvVP-sVm$am?#$8g_2fF~gLF^X9 z!aEa^iP$Z?Z?RjL5&toytrUK!*D< z?^@V5vT3CZgifQ-{uJ8M1YJR)5#e*23?|UGYB~Ja5(*tkp*0k?>JHd}d-1@NnT~S2 z@nqI7n@qDF2s=>gFQBkSt$%6XYB9kMeB#^Au~8%|6qGWaW6OqtzDcSh1I{0TsX!_Kl`W*Eg@B^QF;7Rxlow@NO+&d`jt+cw;E*eT;jk`$Ippg9* zhk>V#*>qInsiUC}*2LMj@0QKvq{OnkG@Z@o&EniZj!7OkCh^h~E<(lwQf)kV%L7u) z15!1gUGQ&@?^4<01i zqx6G)88tN-`##vro=gvA{*pa(-|FS7mM)l|ABI4ZR5dy(ahfnhuYM|gX~BY*!VoyJ zbF-x`qpYseAgr!yWh~q|Mu+=IYyaB2_pe&%W~Ef7l-s&`Ro1r~dgYdkE&0c$n>jf* zo28rix%&?`_sfUQT}6EF*^3viXI7?R?#QKOuN@>|ZcpQSi#>P^Uul&G4~C1aEaWLS z6Vi7E2lJ1x>EUrI)wqW@Z4P77nUk{2X|v`>gf3XRI+QsiNI#7QMjinXaxV)%KYbc; zS+5wL-K@5Dx0ZD48s%a}SDB&XAE8r4Fs16;tW?O6$I{j~&f68fFERfJoi-=~IzidF zc_@k2jCo^SbYohOXe;=k@!1l$$4d!rM71*$-EWs@bvubf^XJbn^_Vx2Xf8T(hH0}k z?W3fw$djP_bD;bZP<|OGKPDmb`qk^Cg30BWnBRXtbnw8TKYlxSaQANq4j=vNaH_$PItXDt|CBv#OJ=(Fq+qwvis|;rd_oafU_g-o4?5mHM0dUpey4 z4Z~9Yo8|gM{jb^x_8B$|q3qk(r=E&Hlx;oYjh=~Oe>#CecCxjrB10yRA3wG>lzCsa zY|)bN`IEfeQo@)EOoF1+!p_4#c;WKE^bZhY}yc*l`XZ~cAa z#^>L`JM9w0hptDRqt%AVxK`oaB{`RyeLg3c=}FHJ5eFDoc2 zt*kDuBBv=->ht$1n2R`FBKLVxtt-rp*7}xb{mmfrndmlVR_O6rzaPDN^K@)!oz`Bm&BH;c zt6{sUG7=x;8G&8LPG?TCk0L4gFoMSibP=!>gB0 z4)RhdD|tRJydnN@=a*l8p)m%SiMM?9<(J#v!zb+$1S|aq$^!TA6>305okppNoE5#@t1ln?%O z3m*&)Z)1WVOW~6#d^Cl392Q<_f_I_tEQPP8@Qa3pFEGJRq3|UXUPj?3+yalo(Ez_T zL>_UAGr2$jk%kjVr*}4!?@?)S6S9z*g+*YJhJ!zSf@$;_boA47^k3=dQ*If3E=G^Q zDq4zFGzZc&4ARsiFF!voB?E2Dv$D#va!G_~URG&QQAWn~j1&?hpv*!|h3iGgN$)7S z9vxj-!AV;Su3o-;wV*Xh*o_*Pp@#LUbHD%gQ$uw_{_Oof|NPVZl=7yIKaQVD2sJ#V zM(nVfjhoC#cbzV4X(>Csi(_YHaMIV;cv)L}t$964m>v_G63XnCJ-&9$3V+{V-=YYZ zFHlpNIUvjN4fG$k;+`e{N`6)}2a6b^`zmrr!J64}Xx4Ju|7PNY(e)YqcAxVYNFmt#9RIqCH(8xIeg zs_Ri6vhF5ZPq~umvJ_V}q?a>_%9P$ND^+!!RPW6+V&H@8IjQfcg%3TnXcU0g{=f-N zpZ}DU<$6=_>G{)915SMd^~r7;sbLLFV4sA{!-Ig=12HAMl;Z9Ktr!EL#S}V|LK6!o z5A98%Yx|(Zpe?Zq5bcWSbYjq!18fur$S7fd6)w^0R?=-@JtcTodSc2=snkLodhpx5C$gI*meA&;(rw$f ze7bu#Vlh7WPeNW^!tM_^K}vbo7FmwiZ_*66mTN}kIU;oC^cmC2BAJU!0j@ac&G*dd z!J{C`z?Rf!Bt7Ln5WEyFMu5>y<`dy73l^YG=ckyTAZ-3<4dOxEbxN7Fuv@IlC~4^I z71y=2Ggj`FsJT;&wW|4p(YpFH_BK2AnQ2Cnr${xZZ*?4WeX*(rG)XPOGCjtv&xJ zSftier%lD3=~{gp6d^fFAA%wz3zyVj3BH(-nS)v)Nr{OG`}ZD;y%uw3C>mvOfzd5G z5@FHS)z*TShKjuCp{N!dw)T-H7rEKXMGfdq+t8G8>BLZU6oAKq=gvbSVBpvVILU0I z1yq>#759x|%u!fUembVod(E>E$YLNVe4h(8J+p}vC-@G<8sbMx|E-)_@Zq6y4cpPU z06~DS@MZd5L%-YiJ{&1y1Y6o!G#1wurY2WZx%t^B+`MgvVwDS=+F8_)j&ROYcsB5o zXK-cAF;8%7b!F9#7NUq1V{SY-&iFa9O2}8J54oz#*aF zkoDk@x!{mcaEK>K#T|+#c5>Pw7p~`zKZb=3Mf4mL&fTy&C3M1q9eU$3(65&62=xj1=Foz-nHNEi!8OI652izdK~ zp#8tD>d|ynE9t5x`ry}C#|Q)GPpYMTt9o#IPzi;m)+4sf1U;2PyWRn<@VvEZgo;8h zAY&6%FAPa5s5vHxrr;+C_)rq5fx?fYbMG*X?o7es?-+fc_K-)RgD7;83HlU;P8tT< zlR|q?=n4~bGKD_U2W{?zU&f#CGx-z#14b`pH-Ex!m78?h!Et_rj_gK9_B4&$N=Hub z8+mZ(CJNn2q3<(68>RS}KIp-9rkzAb%HegUugO~-`6V~$OoNYjk&b9fM{F^TD53MM z?Hkbq{lDdrk?vZi(FgL#B0Bdubna5q+=X=RH~Z$kz^H|+yG09Oe}Izzrpu?tlJhPR zn^scFxr^toWmT5ur(Qk%<2Kw`nwXY!@#xjsBX=0MBn1g)Q62WG*1f1#(ovX{@bke? zv4u>Lnv$5lD|mj0H(QmKh)GF1UwHnlSDttr!K9lWo8sd$<*`kiAjnm+C5$=4d?9@0 z*~i}rM=`7$%tcw@-0=JEU;o5o8$y}GlJpZN~G(&!}i=YL<;s0wq#u zcTYn{OFDFnPP_T55HQ}!35r?EE$%fcz0k?FzJu-RY01xWc2HT_%X+1TtdsFX!boj9 z8|yl1u2=)kYLc(-BqiX4o^C`5qU)cmBAKT#*Tl0w*Tm;2JC8xkv^(=Y_g)yhxoG30 z7Mv%VKVB32P5`Eu>szTyBn~SjH&6@W2)r8WE);iw__N?41vT_W#A|iy9mpAiMA`Cp zSWE9*1TjkfwpJlu`weaj&WFZTepFh8xW6o=p z=%`G8Jc}2eB@lao$zp9;Sop^u@$ox%o;&x!2RO6;jZ@2eGH>2QZd-LzQ!VGU4URN_aNehOpYO9+!-QBNTnK-ez8ObW0xRI!#36Ifft5il=+1MaA(;*^4 zp;)$zc^LobhgbzpSYVBEkCYb)ClFqgVg9#tuVmD0B|me*d9yJ=@>)Z>daU173=u>3EcF0ja5H3nh+AEwM128?=b_= zA{oEGjK}616H% zPtV}cesqrUMb9ZTE|cGgi9(y*@6A zBwU1y-Zu=JgztT}3oGIQBwKCYzS6MF5N=qEMD8G5^O65eN=ZCsScg3DxrV>1`?-#N zOrsIGcy#BMY6L(q!f_R!TueEG(lFaTDs74a5BB3ZoZy)`FMeY*tC;cP8e~&evsSqL zkz8HOBr$ss6YvfSfc%AOKK~DG?*SM^)wPe`nVsz=o8FU6NJ1J!AoP+#KtNG!6bmY1 z<+H$R!}ex&gIKVOpx8i)2!hgk4TKsJ0_kPbdvCkRmfv%CHVD25U-|!cU}ux;xifR` zx#ymH%5yq!t9QcX(?mxDocq^E(4bMI&iX(2R)R`@w=gIeoG*Y7^anB&khzazJK4hy zar1v%QBLZwDc*1uSvF}vzT$dMHp)EGR4L@GyqyxnGJ zmxcc|ej~;Wj~$nM&m#*xJ)d7ZZ}iCcaTCW+8Zk^48vEBYVahamP7|Q>km%JIS-0TG zXb-g1)5BsiP|ioh!?oxjs`*bu*GND49I@jceu%{DCFtxYU8%w2)4zt@DG&rCbPq^f z7T|t_laqnrhJf%s-Wd}Uvth%AAHVwQho67`aoY_uLl77O*16`dLK@^c0m9f1B*OxN zLp5*Syoq7)_dNP&g6H+a=$I~|dGjzZ0;ak0uc{c=4^*-+Wzqt}C@c&5hu9Gi%r$?Nw;PXOzKY?Vy6By$s<_x z9X&Qc`!`=4QBm62go&ZFNvN^mpuTFR8B%SJ3}i@AKkOJCa<{Q6jM1@J;TWw8c$LZ? z3RppP1(Ay}w}_VI1mYB;_)3fR@M`2-0Bj1Gzjju{5#k@cdUbSt<(ciM8uaJ+%KG|B zlP?OWc=c4)ll8p86V*a0dwaT7z)1c5_umg2``kw$h222Nfh3Ac!yWJsPFm~)gVHWe z3ikI8Run>$EK~&h!~AFP96e^t2%WMAR_yYpo?7_GC}qW=9Xoayh(9uwjMde(bx2sE zxD3_RRWus10x-AGEDWfM$x;bIp|OxlFX8YJ<~@T8k|k(qOn^VRX|-;i(YzYp1XdKj z#Q+AM{MENKS6Btf?#LA;3@1Xzu%d!o<7OD2MOW9ycao)gG8WcYW~)o0{Yky3w!xS~ z%zBY`gmyG3(>q4V9sNxG-9c!{9TLPHyeT(UT)vu9URX$W^9F<%%47ycHi|*s(MGIxR6haC zwE-B7CezJ+B|I0+g_?`Ic?NfL7w%>n?nVR4Z(D1z6?7jFo*SmbKbWkOV@sf(!PmaZ<7m9P{LO)etV-Y!G<`0;DkwsgyZn5O7%skw|O zq{}reZWCtCoH@a*<@{PC-~4ozmS4{pJ2r#oWD3%YJRWl|kxL29U3`z*T2H(lNX7z# zgPZM38h}yup22I&mPLycHX}?yMw`Mbe#VR$@m`9C?~?^AEbjXTMJ)CWJLw^MI~KKg zMBzn6;i#F~jamnKu~G$7HA$v$^yC9}EIFM5!|^P~v6RgqW0PRo{KF8A*w!o*8wD4k z&f2+?KrTLcF8fkCp!m`+WnV4^ixpo+j*lT&CNp;ul?0$5u(5WUWm>I5L~K3i#%?PS zDYb@hY6|Q$J4C;^v#A=`2GvcSEfr9aDq1>wDS9;9ORD%~TsIn0=sgu^r<-WC8W}Nn zpq+LTk-2CE{ZorE(_+m0F=kqfnU*k83FSq+Bu7sjId$|@n#4?`8tC{ESrgYPYwD6L zCx=ForG#WoA-~!S$oKYK@-ulok^Gd(-n|>_^<=8V)M4&0bu>t%M9uVj8vuhzHweQE5|mU{X@!oQoJvF zYiqCjFgu&o?Be6f&;sshj79^mtaRzD5qI1%;@&s#8Ou<4`7~QGoVjyg%X}YN<21Y4 zR?VC{H_cXKSKIG~Ciy=8GRN*^D;AZPQ{;H%nrJUmO`V^=x~1T0UIKu$*B3wfEOk}v zB}S&vE6E;gpC(QHdHrV}LA6R_E?r83Cib}aqvMw@9sh{#02YcaT^cto86Y#2`SoVX9%orB~7B_WW-yfVs|4~NmV*uvH2 z7oERcr-=dXk*50Mg`)gwS3)MJUBKs-X}P8_|!Y^UYMM``2LZxF`>bp@v*Vw0ove+Vw1+d zszWw784Uteq^!bUBk3+igWB4>{E$H*AtBDa3AIDa1}8iZPM88tm;z3S0s-bA-cbq~ zGY0Qvg~dih9~+Ac%Ze!)dO>k%X;Bd%D!=eVI5U9$7$gH)tpGs;gcvQTZHO8&I0#}W zXz-9-2h-uqNDf0!>Y= z{^px+_S6u+886^9!;jSL`Np-i)zuJGlzRz6=u&Rc>_-31Yma`dGq6J2 z4N>?JGZpVX-afv*zIYe;xKgK=#??opP(yI56@Uk8O!o=X@$$&TqD_o@{l6A0=0<5VwplN z;^iC)#n2z*-16YC@4ox)+wV8+^p6+@w?_PLzx|dY57=?4FDxu}-MUzk3F1pteb}*! z#iG8dOT0Y&3{BbmYWtb=;*qrf+Q zE&qQLpydHR9 zaunu*r?S7bGUx2clP7nt`RJpJ47q&xgAYD9QreRDD>gG*uUsJ|uwpDE)p?lW_Dpvii`he+Ba@I~vNbIRYF>qr$_Qh5MP-#^g499MTIE41B}(*_xA%~^Ko}J*R=phfos2Xv zZiu%xuh9khyUMM=lC>&a{eyIA!W>dz4yiES*R7?5-QI>3?Ky0nQw;3tY4}X5mM{kL zCQPwd0xsgGI)ljDg_Pox-5?t3Le~Z&bvZNC7?xw>>+J-AwHwcu{sj)w>IgR{MF`tjy4s}Hs?Fq9Pen825Zw+E1j+Wh*qB# zT75!CCf~Bv7agreI9pxdXcfT&cyg*V(N^=Et$u`7?-N=bs=F3@Iq>{9-s`Pe{ms#; z-Px)htxgwOjnG}Q)KU(j<3oG(KSz7CLDUt5;Gt40(k(me76`2mj3e#$ z0WA3IIAgSd>*}ZTfX6%#O>w7feup1w08l!P&IBGaft#E}rvi_ua074L(B=)iG0__fu#}W3 z@os&6`}Zf&gi(+j!w*6#=XIP!uRZpMLUY?OnTPRQEl)xmX{iFzdO`zS2u58R1$KB0v|E;l4mEvPe=$8h-8gGxHj<7j~zqjmGw_=K2~`x6f(ri=$iA z4@J{Cx5anE1Mm&dKKA1{2v6^NjM5kI1R!x3*ZGW3or-I}i9veHalFf>ab`W2ztOnr zs-D7BO$9Qf@TZsXwMA*6O(=TR(%g*n*havpba%J6w03tx_3G{I?g2P!PY+$I<>o$` zHr&?-sBq3k@nWIe631^$ede@Koh~FaG$cf)(*^57!=s|YLkC4g!9X7q6BP|xI1oJV z`dcGV_xf!&G~yUHYA+dBtp-+8QFa@7AFB{Fyfmue1>!RUVYHHJcuW|6f`-SqkPG+( z4X@UPxZ%^l`bWKYe@7jp6O=Knq@G~r2wHp*Gn`REetbq}JclR7iPk!HdNJ<7UA+u$ zdIQq!T18hQf~)dSU*c*(Q3>o*(5Op`D{=?6+{M^xmJ}D!O`N2qot9&^>!miU&DwA3 zXyG+Bhc%UA?^^o%`fUbsY??g`yfYt=+n0m&V9ce{kw)E!>+_4i-F z&+_H>U#&_<*xxJ_mQn6l8Gg!5hEp3BxnR| z>|xkf9Iiuz?fgo1qx0Xvg&93Fl>8tX9~a zAD(f~+(nPZ8zSI9tgEN2R$E<(^0L^ERaBKW3~bS9#XoFE=)O_n<|3nlRWF4&MWIlQ z8?S+Q|cf342N%ViOHDQkd zW+5YW@YL!=L{!3vR(%{M`(stM^{nr>apQd1^-xgt@})~re&qb+lBRxMh7mbh#3Uyt8(g*Y zDozw9x`V~_w%(8Botu5voik?RgH&}QH8u5wN(7eH#HNayv1604 zyjY_leNE)XeV2!GZCAmGf(T*y5*Hh(J)iX;fLVIDnA-GHnI%j0iS`$mM5kq+Sdw67 ze^M{5GE)0K4+B3y3aE*neY$%m&G$?eEac-)M5zjm`{}~ zU;e}4tn4#9JPc{l<;#~YJ+tN44F*|mE{q+PK2z^x$mWBz)&rwZ2xjworoKJ}$jlcU z$tih|)hLaV?65X|BYJh3mj!!vYIqdf5^z_u7T5XMbs7AcnjABww6wOipdd37m9k4p zfO%fu0KXboPbvZGwVr{A5oShIYmg{p20xvS8qo3T@4tBQ#EIx=Yk4_foUookofn6M z=sabhffZ325@IwG6t8yfKEg2$wC?06^kO*pKOBAIo{bznV|mBb?4rDb=g#JztH>dn z`7}g=Fjl~a;UpHPQN?tDq@zPVFX7WG(_ptqw|m$u_L=rp$TSmt>52B)py}k9++12I zl9@FtEeX)pW4MJ2yNh#XK+LRw!)_OIjJe2c0*FUAd6!`!V!j8J3(t%h!xL|w`R=8HHFkA3N5sOHL?WJkx4%El-Do^>renYHYQd$PjJExk^Rx2E7JR9f z2N)PMOvjfTD-P8N>H2VE6xtcY^OPxjKc=jq!~`c})QrBbgrpk^Umgd9#UP|zGsjl%#~2G};3 z0CUMs{zKSd((|$g{|vi*$){t>)~`prq#ht^Jc5pxgoHF^>sG3j>zRqbEj)zE zF-MrQLma;Whwa2y@4L_O)jfOCaDj#f1*3LRGji0Q=WqdrNrUMvBBH}Jur<)t91(%u z(=%EvPb548FT3!35xCoE-0cuNUj*(p0(a|Hke^>b?-*s2axRy_rc8G5qW*r4(=!$e zPpl56kSkZOMj4%r8*@>#eUAMh$?DN(H*O?rc~M^8gAeK%BuHU@!aVt8ntb~F`O~*< zednF9qsf>hr^sb)|LAw#!IGT8i=KXZ<3>I6@WVW|O3bQNX>MTuTKY6KC}Go54r4B~ zeduHsiym0wJgVMN7I5=MMN3-4*b7B#f?U1*{2(m;<6DmLn1CmofG3=ZC#1PR6YzwA zd3pJH7>V3G#Ifh)8*A(HjmEsZ;>!G7R9xM~yFzcL!UdKt;7PXBG)p9UIv*H`BrR~& zs%<+qeY<+~_EN^K)J4W|+v4LlZ6bGle-7`bdDrKl_AwfDaQvi6r!z{sMm_hOo@s66 zndhEMgT*IpyB$kupMQ;Mop zjqUlne*Re(Qg>*jBL@e}FcI20QajQD>+9?4>tUCzudb}DuQL|p)zR{-RV8%|_4UX@ zcZHfwJ-l$-T3FW2&%LitPbUb|*OwLugSW3Q?4G`!?(V)mTDcVUe|&vC)INSF%o=w~ zT1T%&qE}OSGp;VGu7}>gAypFH>QVN(-`b_CNe^TG~?T z7XLsz*S9zWOq@DJ@aJIex-w%84sboPiJhbATjoL4kVgFPK;(NDiV(CmHKIqTL`4pW zivf2eLsZN%4=A<%fanZ_Q{HRb_;GQA0)oi|hZGI57i1a4Wo1%@l)I%YFw)6iOy9id z+t2i^$|(zOz9TJ%=zx@Yms(*`*;T}yp32Bsdh#SAaiVf$6ndBlo+pE%|#tv8o?)qUT-o32dGb@iybjyY(O86G7$DpnU-dmYC% z@-q~CiZ!3cJ^tXo{`C+Y@{f3S)?o26zs}BXlf`1{uv<8`rxQg(yF0+Geel*J6HN?t zMJBcIGD(N=3UXbmBs|Np45eyrt88ooRZBDHYm&HorOx(8|A`-!%# zO>6Mfv|WKy;YyoE6Eh7^Rnuai9XFzs^&9jMZ*0_PhP|FlHqb-*`mhZf-gpC6E>~f8X3OKDcL}F?09L9j-*@DTmM9S1LvoXmKgS06rJ8oQf`9EQ|v02`Oen z^i`-UX)VdatE82Dw{`+(+TB=8HvR{4P5EKk*Fm=P2M?v|p;khY@(TI2?~3F> z+qOrpPKeX|x+#?`;mLBI{on&4${qUV;Y1YOS@+vYT3h=~@;(%%hhf$E3G?E8%#Ir_ z(bV14)7aC^kQO~kV_?PM6}lJ33>q6cc1+6T$*?c;B!p+ij>QCll!$T4;bFM@81Qlo zc-gC@l=^=lISb($MuDlaii-SvvX7?@PX_yPT0R9%9TB_StG&+y{knSf>a~Tv?*9JX z=1G&bZG+pFn=oN)xQdr6TaO<Nb<6sJHOlu)wXdTSfOg*o-Hdw8OGAOI&y{%F7==>w2+60 zAy(URub3cINP@N!5gMw~y0}>28}{)5I^2z4U3Ss-w=uub)|XGV(ZN;2LyFT1Au6rT zfysg`|8VDs4`gx3bvMNPG9~*M4PgdXFf76POuZ(d2Bz{1(?J;bNW2T&HqhfEukUdh z2$r7@J3&clG1V-fGZ~Gv$FF-*R(R6(x=e&{AxbAv)_5r+<3hvw_3`nv$J5g@3cGlD zw`)Q|LY&`sJJXko0|!lf;DHBd!4DkLg2Ik5FTecs2!{$Nf_EOhK62z8K?+`8x$Vav ze>~E_jvl$(i^hG`)nQ*zSy4q51gKEB7xemdqYrmG9eu0Z6RRJ754#xjfmrW|2q5dD z*Khdh+rwFBuaCoA8HY7#3Z8u|xQ*h$xw&F~U=BE6pk!HBR5+%+d z+y-1wkgv3pS6GA+4Y6t!VAUronyuLo2P9}rKO9i2+7xzt5 z9Xon^mJBv$^GtRbPs*J8FzTy9r(yOAo(3?T6tU-cF`UGhTY9x5XJ;yF#$K9>U+=Zt zOrwu~NW~Z|*ORSNh~A!0*4n!P{QE{KjakBfy~Nm1(OB7hk=nVXdWkkd7A_4OlFEjV z5Fyg;<}#A*^)OM+kPmU95%0p!&9G(Q*MVfTbH-3zWGwEO#<){mWGu!#AS)|BKd01K z3h=vYr!mm zN%MVth7Ju6uD~mChMs)LJp;SYVQ7Qq7>q^#(@Jy0(EmO~HFY)U@9J8FWI#!- z#Dmn9mlhFVfZVj}eNL)6F6`N}=X^U#zO7%cibSNxm|)dOlEq7J+T}`Lqjb`vk3Kq4 zVs!3A6~=?_zyJPE8Ae|S(|_(~pX$Q=LI|Lu0))FXG*pyeDJm$dsD_$z^Dg%Bjqs)> zLC>+X=_23-V_TmMwWnb0mEN>#9c$lU^vXQ+?29@P8H`?eP{+8t5m&Goot|_9=a5Pw zH{^kq6y+5b7hOG{lV4U=oSOr?9_=HnLb3tgBt(aONzka2sVNH=o;!IWJ+szspEgtm zUXc!&_R1?ppoi_gym-lyARiU?5B*5wc>a?7tekzRunHj& zgM98jKcM(@?~&>N*be=xj(qgd6HgrZb}~%B^mv!saHN+e_e>#7Hi5p&Wmem>=K#y9|lEOAID)- zDE7vE;8lvBo&a9eK@D)c{+VZ&-<)aPQ(Q(2R-E35+FS2yDj|^Ibm_}wzkqz8c_3}0>#1g52Jq7BAHStv6-OT zOEjdT57BNi2EjhveFRo`GXTJ(7S3wr45`s=?VO+Pi_gE_luO)`o_zAjBzJOk(^sE= zq4Q%~TZy?Eyj^WRdp0f(dp2I`9UTn5$=l$JZ{pTmC6Nn~p-$nDJ`#-(FXPoZz1(bq zZ5cxeBiHbP5ZA?h)LzH?G*{;dCm{q|A4o;pk zXU-H907`dlUrDje>FSv$c{h%{B!5gBS+vo z{DP0-IM!4v)`gTwlsazsC2Nd?SBxzE+}u~#yQT) zty|~c9}m$KfB*a!sQ=or*B~FrUzk6sZSUT_ZIk9t#jf=dBc%u2R;j7v1*A0|Idb>1 z|5-qL6zwpo8UHte+G%8>=)V!!9)&FomM^XLzY*Q;FfiOM#QiUcnuzhH)$*uBrCN8O z+}*vfz6ueJm34(KF3t8HJWfk{soaKIiZmX_4VEq%UrHqjKC-M#BTs|X+2H+UXlPd6 zp&x$u;ZXkNs;VLtGE!n=RfS(tWG0;sDFvxC)WQv++)qsim6gJS4*h#{RA`_&4W0-C za@W{D{+K%zG1-5@Yed`UM$}m>FGf&e-l@#0s;a!4nnqh!Qx5`ncsD52G|-f4 z$oOcpnL2vYs4f}EHZ_?$MJUP0b(!1R;#?FcO7G%5itX#Ycp0h#?*LE^r^DB;?jN~Le1vq}9`|exPo^~kCs2JtQ z$a2zVl5#XbX$K_uo7YQl3lFY*_CPNcVuE`|SWmL+M6zGuPgIar)SiY6LtqjwWoAvk;k4^V$EB*uN zCVv#S`AvWHiN~|9HuK_R#|96k5*$}2CbM2zchv}{kAB1{ySW9C8$f`4IGXhY>-3r(^qpSooau zV~6(d+xzFvUAqsRETZX7wM8cn?b*HakG=a296o;GLQak#zmb`Pbq$gI5(KKs?KiA` zsGQl|Yh|4ByUF6%0-GR*n_F9v^8zjcPK#2hRhgU0QKY`P`tq484Gm}u@|){H!*YFa ze{WBJe{UO&w_wzykApaO*FvYN$g96zet!!2y$16+9pnEg=JRsQ=Y-bQ_Ac1u5S7u{ z(bm!gf!om7+}_m=HN#j@-zwx9RWx@YAQ#I{XFI+F?xwG%*0$EBdJZLa;dV^G<3Y2v zAo9Rb4XBUCSNGFr7CEw$u+Fo@(u*c~di(lmyf3U0fV}Dj(l1_J78c(%LA)ymYDR@u%wTh0zk6taTF*h@={3U!2Hk*r$cI!q!<*jo z{c=}%f8dU=L0_=!nC?q~jc1A#@Dkeovo*oa4armR}%)+vgK?)p4F)6`03kBm1h^{_H zt|Hi^i)>-285?0{ZraeO*ulQ8fpM@>j2Pq=5)q)%bKPA?oq|}SX;N>pr^#@td(Z@` zKgieWg+tt`w!pOen^EPP^vYBu@_dSfV0!#W*0Z7>Q(8iH_N5EwE?hvUz{Sk0%U7U6 z#f_g%8YatpGatSPL*v%3+$*?Xcz{*M$M`JDw3E|=H>L9f?82DwO zVG)rS?C7Y$gLE=@jo2XGP>F(c!xB!-KLjDWP{ z-P%z)==z%xbF7}TTcI5B!EJ3FeWahm0tt0V%=ALu_rkaA0w)xtwts-r!n6c)<5SFy z-@va+FgISt+!)REw_!T8bab|Lp^lTXwA9_rP2Am9TZyS&-Ofqe0(E}K){t`j;On;H zQ{CN1tNQy%BOmF=%IoR2;5nO4|4E&B`>&YMAN?IfEykDoSQQf7#dZs?I14B=D^2hy z-)Ycn-hLdB1gH0HUb|`~^1x}_=X|j>G_?QN+NlV!y3an@aXe`Mw|Mgfn{SXuqW$08 z`#=1#t0x)ZSvV9Qtp0MneUPhQRQyVAbldN z+7lJ^orr}<0eg- zgq>HO#V}g!6@p}_LBT;Gfm(3ph!L>js{%rzhf*Cwr&ij#+Iz*ERH5z%Z!zkSfPlfU z4#9*H$#K1%M4?39mBnM+z0_lQXRH(>BnFsGKSoSq>uPWB?q@h&VQ%TcFk!IlE{4%f zop#t|IHV|eY4EF#4oD|6OI+aBba&;t&FwYDY7th+HZksCixF#F@?ok3J3_eFZ!^ z7d-k9cr?1dmsT-Us}T!Y(@kOa@G{p{6=a^zE(0iSZEYLM0W(%?e99?X4m@Lw^BKD= z*FPgVqr=e|hUkFsC|JcGfBfj0H7BeA$%~+tJvY&#?dbZSHcuL(Q_DOw7CM#09thXv z<=U8WCr^$aKaOl2*nn-RC%&_t%qE>b6dsl8;%2?L55XidE=_8;R6+{v+)2MdQ;uJX z=`Z{w86XKb#2;ogtCC+zLR{ESF6U34yjhv zAF~M@@<-orGR-rR&ra3WHg;O{aJ*87PFnQt{h8>dUw%Pw=;edEf8UHSw=H|mUdYKW z%sYGf^qI3~kL}*Ref!S6`w*@No*0<4SDlkq#a%yXF}qQI4YS)2&2;p^vE}LM?(6F! zk@j_BE$ijnT)>W@94l3O`+G|1?1qIF;m!e-2e%?k)qJIex~lqHEGiZw-wAjLz+ba^ zv5Pxv>h4Puv~Mv+`#&Q*I~gsf>`lU_E845vh}e$$$EHWFGCH_!lIVNgfi)U@W8UKzk(I)0?(|e)=#$I z<2YU-X@(?nlAJWSmXuV~Akd<=wyvqI!&cRWqNq&2RnLLVO??Jt+&F`472F5drNVC^f+n1do_bCvL|rGNr;-Vy z3#N8rhv|-Goc=$-n8(1)fcOPAxU8m<=KA4PR#90~TUA?C21_GSn<~nRVdmQ(%!ozU zWHYoOgrc%YZ^y~b^!3whcY8k}F6yE84O9B-pv{C}VZUopB1sK?^MyMn%?4J3H`3Bf zR)zn4z}1}LuIN9B^tdy2n8Aj|-Rsf}TRl`LDeNikmb^UI>2qc=%@r?R#Y^R_QuXB4Gpo;$SRr`?xP$r7); z;*)>uIB_vEy(r_vpKHE5fQiPd%}rHB+0cf%Bc~@Ml=GhDZb5y(@V#1G-P8@8^78LH z%TJsr7jQrR#e91?Elwn&-gkGdzpA*d&w3{J6M1BaytBEdcssPS-^)-;oeheRf0lc$ zr?co>5h9mRJl4+I5e4lT=Goe+XU4sqM1CfpE8u^h6(FkGjy)62H%YO-Ko03SrSig_ z+%%1QAN91Bb}7Bx?2UC{NtlkGNDqS>+d}VqF~tD|Ln9%+m%jW!R3Ad)Pf=XZ_0q7s zQhQ)HS&WPJnOV6M?FEgVGR44Z>(?U^h#uV6h0tf$jbDc3~YeP0mb0+lkELQ3q2tT3H~ z-a4BG^dUQ4yYGh+?d|O*yWJ5OaQAzU8zwt6)zf4n^MK3{CwHX`;1LiqbmnufEr~iF zJEpYSq87vZjxb!8zUnShOIgvF*ux=jzj1$r1U?vEtx(&#x?yyvOJ8j;Oyb?_M}Iv} zrLhH=7?_aJ_e`+!#7?0%maz+myE>)p&G*wS7GMoU*SDQ4sKHV?gU)yHnQ>OtYE0oN#R-peHdtzDq~S-9h~0_{Dwp*lP>*(H1Bl(H1Bp(H1Bt(H1BxIa|0n4U>Qubat5f-ka}pJ>x4gPjS~0@ahX_}<3s>*I z-PH%Umr~QKQ*wHBN>8s&8*pBI5qNJ2WY$;EAYRn5*DS6f|P*=B{Z#c@nsMiu$!qu)yU zqbE+Bh@h%+d2rI}ufIM)yXuP-9~q?hO_3;U+P{)v8^oc0@2R`)n)f0clCQk@`XkRy z#_vAF_Wf(OSV<;!OJ%a;jYK&Hg-Z5mi?<8bU+8?HhcMxS4ZST|my!YG9e z8+QJDUR@tWuL_0ZNg`30T6;LTCMI!Qpr4YM%@R*C%Y1 zZgh!^3^ARh8!tu$KHDDtv6r)*OJ-yEySm%6C-=}!%cReH}c^WWMS@d9IM$! z$Sgdo#ML(#;Sde=#{A;C$~Nq=7dwu(H7kGLb2_iK%^Gk=thcL#B!P4l_hy1`$FIpW zZr}bUx2eN7@l8PW8pc#smeWZ?(}>}BE-mfza+As2Tr3FSyL$3`om6YXr`qbOiZbkY z9nc*xgx1x--oR_RkSA1p>C(uNu|XP{t5P3|U65U5(|W5dy&YGNovy6uCQ@l%W4#jw zuD-EPY8d10fgl8}D_j%yUKT;PgTfeJUnY1kW|%H4#AojFQ|Mgt^l)>5Jq0^GFPbIb zu43p99M*sTC7F>K=+|ln=5X1MPV06?YcC}a;cN&UIEXI-g+@Do|2nk>N%h6(KOu{Ib+T1!SGqb?hgMel~Em zj`Mn*xlI-1@8;$Ie~n15k*HmUKeBl7;_0Eimo8m$ zGYpPUmP3?RSnaisJvQ>r=iYnoz0rQMj_Trnzw_d>T9|mTkW{)yCBH($1@x;#by&W zcvx~phS8A`e&8#g=sO<)a8mFn#IW8yCfZ%*GV_6Z#?OnAY2bMYjf@V7Nw{;6XXLO+ za3HwDtV`zM2^HSHUT$uP{>E*#w27u=BAjua;C z0P@n!^|?ZD2voO`7&e?%3trg4csTszjVlWJ-0R>-s#VY^idUUd@H&oMu_7ZA40-5} zpTGM2^X2bx&sX3}x$yv);M?@Jwua8X*itKk|D-K;!F7xo z=Orq0GG<%b$a}$;<3sJ4d-lBi(!60#zN}FBMNfb7<(K#DIb(^t_b;w&Ya>%|+3UGd z<6z2UNn2YzO!xy#sTGNE*|I;bW6Lp8YYW&i>(JR|_?!q03KjLs0|Udt-9=aT?aqWf zvdzZX#9m=xfdR4}tGT1A$0`aBM?TvqnM~{DVXmqW_|nZMkQyN5u-^7x&z5q(AOKg| zuVc(HV}>H)%E6eAKdz1jUrxItC=4h}PfoeR$(MuT#)d>pe&UJv_@R-};~bp%_~Yai zp7*+*GaDR?DOYK=QXVK;UGx2fyHj}_7`UCt}bKf4n^$-DPwXa9m2*Vh9-{Q2jfe^37^JA3nX zu0wL@4!WXe2M9qm0}N(jZ; zs$#HW=D9T+4nbg^&nZ8D`bwq2qpXSz)P4W+@8@cjBSwty>u-}t4j@tSosds9Y}jzP z;DAKZCJPG-B_-!~ZH@3aP4)f#3`fu=fJb*LoWoOr8rv_}xgx|Gj>6w!yim-7nBn#ll-1B!%rP4Braf=!`^+~ioEz+wK5)*RMGt{3$ z+?#?%hxC>1!d>mSQqW}8OTuOKa5U6_zyKQLo}j~4M3R`93M*yG`*0$V~dR#(=7 z8FMq+JA0^Qy}N;PQ$tz~(me{ia!75>#x85Y12t$Ou|Eb<=D3MU*fOHsP{L={wVes4(ZnfjQ4Z-4oB(VY!#ZRy`)R1X;C0;~D@$=kXu5P;WMZ9!~XTR$iI zi`lq1kEYyQy0!T%_>HN9_x3x)<9So=WCw}{*43`(-@&g<<#>#e-VTtWm0E!L( zYVBoZ-s7jmV-fI@$CZ^aQjOMCAyr#wCTt68!$>>=0|Q3gd$%A6`}?7Axq#&YecS|& ztEe)2cxkWaI6Gieg8byy@!Xg(iHMqKQ$C?h{Wu@!(`{SJ&4b zg=ic(bz1Vo30iIJjC)5X-hKc5_b+&E8U&#}3Vt#bjfrRb8~%yqG`9zbNj$^t0C3C1_qyB3~Y1%2a->EP4odVMEv2JH9xKTWQP4Ac={iJwfugd$S2zGw2y<=|KWd> zjk?>)#zgxayz97e(Sf1T@4o-`P(?*WL1Dy@asNAkIhkBRmCdchz`ng`5zVo0BKMS9;{*Ub`i z-5x>LjS+O+CPCL_uIairr~Tt*v|$2m$Zcq=1llO!x=!qQ1lCa<+w-Qh6g@#ehPFw#Dr)%AD>)JC3xw;2--h;#Ks#EXy(;vw z^J#sq(T;|!dqM}yQk{s5#?#K#F-R+hiA+K*W!GLb#0IIQoXD?~=4n<|Rc7z|efxp^ z=~pW%kvs)xf|7h-)UkR`RxEC*?h^MPmsTo8KCR8x2OLu}1c^JNbnbD(BA`2 zmcUZN697^zS1lQoz3V6VM8EiA(QcO{kB z^q#iM#M!*%3x36|+vSWIRTcMsnPR)ys(r zM|^0MeL?>56Gsm1If&qBfF4}Ad^InZY~T@Yp*V3OJNp!r)H7$#@sf*I5k;1pXYg6K zE;4fSue&!MIdUxh?Af!(+V<`$-n$O?DrA;H79317jBV;jUNZU~n1P?Z-;e~e>_%XL z%z&MD4FH+epnumRDsG#A6e~u3J@L?)GzV^DDtUtf}CM5)Bb(vjp1e+t8l78Ldp972k%|CeS7e*R=@OH43!m+tBt3v?B%D zVu7|wpw-@n_NLc$J})o|ZrN87;ku9AaNWbgb>+9AtrTeO0_{fOp3e)k<8MQIUZCX! z+A4uoD$owP4Q+)$J4>L=yP=;g*Jy9ls#}()BeI)MH$5DLHdUl zD2v>z1~rfDCVQ#`SShfmo1K|G-OB%+ae;IY+Nf1s%OQm2^0(sHnT;)tz#K5w1Q$xH=D4Jcy*h2cdG%PVY4<`)c>!8VEVPsvI>yt3Dkvm#o>a$wmR2z{ z8j;?OS5;LZO)tx>I*>}2Du5iasEZ*2*oReQwqYF(uWhHLP5k8TwWX)vy4iKAv=-r^ z6DGLU9)`c>%fnl?1O@_^VUSGLZ(?n7F}9{+xs5gT!#|Ths3daHqD7GQs7(N;mX!d6(dS5&55-Se=YKdRi~7dg4iccwa>4M~*yi5FQd)oOl%%*U*%(hT_vm z!nUzMC4{$)J|KNiq-B$EM6^3zh`6UXFEW`#Y(a_b9*bBFkBSFiOhKXGQM1g3EOG$K zAQjFPiC9)qFSLwakyS3O@$;(@)=TPhjDmJS<3tCrw_H1&&>h9qS?5o$+g604<}NZ> zOV5cDMb&7cyS>*YGq}1$!9WBzv$5BBWbaQq+sJxeBxyUemnR<~XjIC7{mmDj_7RzZ zdt3=sWICZ<-i^qjq?M`-i28=o#lA!SO@SW#?%|m`m!u$goE|F>!|S4E4X$erV(j5; zI=toR<=mF?oq)y@^=adjygXuzzf7XkNB}JDYNgI3s0(g#Prp!Ae>(}MkQZ)1qzLzi z^G_yk?*#s(!xC`wIai41!}RjR1K?DPTf%7+E{0P$TDz5YU=e5&1lnH&T9H5tgu7eP zN(I^`fwo_uH4C(|8)&H&_xflllt`wudVzA3KzT->Y`!VwUpt@6f2Y(7bm|-E{vMx| z%1VU09U)v(EL<~Rpk;1DYZquE1lnqWwpgI`xee`2uj_nP`E6)5!gcSt;kvtp>$={C zwo#y^Yz;oxF5L4;fp+q3XmbSGZh^L0ptTFMF}I;@5NHv@Nv~_Xp`TH&^OhWSNuX5- zw8a8#yKr6If6(5DRWwMTo#dcRB&n$a_3r|8(|=Ik$d5{aI#IZCkwDum(B5tiIk}@l zpq(JlI_E*#ZLWJjxUN{B^%iKS{{6ZOAmtvw9HQBd4`97{fLfjFu58(N#iTt@*Isnt z*skAx*}wJk6#F7%{rzTt$G*g6+3K~w?>l-SKey=gxii^CaId|?rv`2J?XK#ui>%%7 zAvKuQH8Ot)XpsWq$4lCI_BXmZJdI;1^ZVJf$f3jB+(>mnxmCq>mmb;l)1|D6ei^B{ zQq;tRj!k5c-H}bp@6y5yVceX#i(Y;D>Goc%Nmgz6uN%Jl-xKbG4ek30;uK^q4ePlp<}t5T$Ft<0~rSj;kow~9#Z{*j{wxoX`|Z>+mnY&Ex5Vu?=;Ql%-nWjbr{xkO`k3p_av^ewIWRwPd)BZ--`zJZry;-yx7G}3+9zhoJnafwM z)MWqi%P-Y~620LVBVrg!E*G~O+)ce#kDMy*vC7>>J@CNT(CE7!U9gDveCCBElfy`M zcJ}1SlVw$h1}r8c*REZ=Y10pzcN{u&==en=&=g4V0kpbiXI0gYKgPu=$3Oe*v#~Bi zhV+htt94k%Su-3Rh~Y_P8~Q~COKCdB=#DKzDb<@(djv5;AB|KmaNf1_uZD2fz-iqxd4a(6F#D zgBqW3^;>jWdH%9~pp5n}z!&E0c z^VX|0VB-V!m71C)SY);_y}b@(5zKTzU{IB+y**)33g$IEmXS1xi;5*c%cB)+@ZXSV ze^vSsP{5GlM_pYmAFu!A;O>tpfLL?$(WB&7biX!S+1eU0%ugy-s3i=_7q$|H2Q<2C zVHMm4P&w<$s%6UV#*0l&2%EE4R>IKdVYA5;wY5V65ZB|PZSS&?+FHQzl5RoYq)6v8 zF*~0_Z_Y$--i_W2hn82In{_@N&aP7zGTTkX=W=ti&SW4d?9}P(tAO%6mwg2gda0pR zg%_bpUdXR#rL(Onud@?*osFgC^+wEx+>-LDc8eh#TB(nJ5`zo81jI*427vU2b04e&s&GkQ!;L0Xgm6T!17Bn7hy$b@>8 zy||MBOel#< z2w)n+J>`rEecfv(K0E?dB;YHln(d}#9wSG9<&as+VL`$q?Bhf>PGvxiU9B#{KOh8= zMB#pl)^^uW#4Scgqr_)yENaMNRs`u#c`Ma5Pw9(Rw7%Xxh;ZzXsoHVRDh<=xgsIun z4-{^v4f5uCbDaa$kQJdbp{e-cv2kj`UE(KK znU2CKv8brgN$DB&F~&mhQyslFqxJ!h;-`pf^s~Ti_k*6IY|a9=%>uVYBG&HLUoYkq zq05VMF6Ne%WS=^H_EL8C#lj-YvZ8{En>PFcVM^9gsh<{PZLcZF#4hmYmepU-kXx(^ z)vYzfbyU#z0c|T1;wV7b%s2uFnPzim{?4Dnhm063Z>~K6;e4RBSzBD@fppiS;dkqxyrcU0vKb{TL3qDIwc9Dk+#`Q-Si%hBubJ^+4Mp- z2?-=26o(!0m!K z);zU3_>Pq~Ah=7Uju3bN4B-PvHb|y52ATYQy*OWWttTi1j$YM>lq7K{C!=>0M(+`f z-f$Jlg!B0o;=uj%Q^=JOJ9-c(wu48{pFg^LH%Q$MokC}GG48giODFIjaCV>x9!|nERNVO^=&;72yd9W+uABmAH{xpyaeRa z9LqAEzCm-?Bab}&^zE^pybXx=f0O5lz2oVpAANM`q|hLp+^RA8NG4Xb&-BC-Pn&Mb zgwI?Yt}mF|Ox2$fx?3EYPpeI)#EfO2ky|#MdWb{9I^{)rAjV7>F^t)!2CrcSVd})e z-eK@IY)T{lr}*KA&D)Qht3$eJLJziU;5M6jq|#mDu-WT+7{XPiV0(I4lq!X5z)N*= za1$V>PNOu36PX=w%K{95EF1JPHyi>WakaH0SxR9^h>9a9rO=3w@#7DLATp0fSk!2! z2`MRH>IyZ*B71qHiAv$H+4c3&8b5P5fEI1d-L!uo$wk!C-7RfxWo6ykU_WP{XUOJG z!YC5H#)TNgahQYSFb9prHf-r`cTW$NoyOW`N8^cO7oiKNwA$9P)2GRhRkb3^v)9FF zEs7&HqNw(-TQ^itoB4LoEIMUWYNOwUEioY3(k8(E^{ zy%u9XA+&G`8er7Kv*auH!7#oXfnxIxl-cpvlwSZ0j4080>`7# zgp8dyA!F)-dC&~N(x{7SLn2&T+pLl3Fs*igW`=r6ieCCptky=(%E6n4AJ38><>W^M zC(e#*ugR?FKU^w$@T$=ci?K)a;2F_YT$6{dMr=4K#)QN~&7`GSY{!#!X0a7`e*Wi5 z-X~shiP-Xc@rr(8ThAY~Z4uiV#I`fUwim^=^?%TI$mf4`+aaI7><`-V;&tbX*Hwzw ztroBA`h&KEV%r&F+a|GXt=Lxo2W_u;-J$Obmj3_Q&vNm)_g!^eFffn-_WOgjbzsShKeo2#XG)1yyM@ldN=IP9b69@&%E>g&5R+XV2y$FrQn}#>=eXK95G%ou6Odj^&#~ zqMcrMSA7wX$Ed1pjX}=BlG&r8g0MFP$EAc52YXmlLIP8>9eBKhuJDw(&!nSd6b|K6 z(?bdBFnO2^|{e*Qx_7HrBr;HS2zRXRn0UahxSzb^95o{Pvoaw3Go!pu`ue==9p z$@#gcXvA8Rky%oQS~pmWGQ*1`U;FbS;OI$Lc=A3?V_H8>hbB~wD|?bU)ksc+|#OPp#m{@2C8K- zFq^mXv-#=5F$bu%hxryfpI|>fcI?>aX+UZ|tZZ@o0zG%1qopMc|Mf6%n@eyhrHXaB zNL0h^WK~Me0M%+^8Jl=FbmM35PGrkAy)9hI^A*#F(wX5H!kXyoK3Ll(h9-Pvs@wC(vEv&x9qcQkXEiIyUFcYJnhS6Vw(I+^d zsTlo;s(J^c7_$lGxeJO2u^Y^0D1cQ}n6f`}T%fDwR8G#t29m4c9BjCllXI#D^*TyQ zta@!kf*9@jrAfPPYb885nM3pCasEZwI;|F7jE|rqvjdm5a>}j0kh%i@{PCBq9J}?> z0L!kGy6P{(mQEKk``1ezAL&X%p`wk4)nQ;rW+km;ca z9nVRoN6Oxa2n+B>kH^HsjEo5z8yf?wcDt3{=^&ND@txf5sa1HUR(+;@+ttGt7`0a zTX9KAaY-r4xK)&w@68H!)>Y@9%sE{MKWS@8>tN$~SS}~d*Eu=qU~65wlL}*;GE6g5 zY|@7B)+3hrt8Y9=NHLc3sgKPa@;!V4we_6<#vjhGpfS_ON9e*vX-%Vv#x7@BW0*`y zqujf8M5wG1CpcDm^l0fZ{8A}11}QoJ0k>2Z9)uB$TX@3_H;iH}3cV8eTyIw zS-GUOljc}QMXxqv2r*Bm0w^R&9mdaRXuAsSa3nQ1N)%EvwK#(+q9@anL9KS9b;9`q z!nRV3G=y>_a5h(7JhprH?!$)<|NPy@VCa7LqmR~q`57Q*AAj}z;nPQc*}wD0-7pNE z#)W-e%GK-Sl+M2XK`%@qp8Ssar~rc03l-k&@`3%|ENw1F{i`OcUabRcS%+G0q2ZAb zlAr*Uc?3!fMg$v(6sdIJD<){J&=F%t2dN_hK`k~KR~hK>qRh0}Ad#^+wSw-Ai0Gy9 zi;`8Mu(}k+uD53ZQwepfGWkr}Tzn1B3aEF7(1~8LI?LZNa?tKJpaWND(Z#|S!ru(8 z_@BO&-&*&GttW`BCyTAu54UdngVy=iYF#kgTJs03tqlpmmGb`f0Is z^}qEwd!_a7Q6r@GC8=GU#lVNUm*6v!iM-&ooe0dDEQww0JG`SIB6AJg%kR!Xgyvf3 z5aouPg9y(5J_i{zxPZ0Hou0}q-2}{)1)0Iz=~!cmhj(JmH*Nc zWI-f%KvqO@2V_YkcOb5bct=AP#i2X+A8Sjjc>S5;_0JDqzY~qc6_t3L|GgJ~Cs(9B zxjs1oxxRP?a(!}&q3dTru880AF33a%=2-^O5-KaE(T^gZFm8p+D}X=fc-+d>h);PkWCO4emt! ze13_qlh;G*zFs(9$x21oj*}LA`uR53&78XBr!}9vXSG`2`Q#sGx;W`Qfv3LzSD{t$ zY!XPvt6F_sbK#@>Cnz1^@}|B~y#DNcbLY;z{=TvEp;cd^Y}6oVt`)RG2qIT<9AlU? z4}5%|!6mA`;V+A3rQLYrjnii>e>nrx0SFs0FD*(5utwL`BGig(^2!DWVaUWc2LZXw zZbP{^?B31wM7fjl5*DT*r-oNe@T}p7oZ(od9E#NH=)geG5o)wr4XUp~3pNCj^jvJt zs2T@FC@hFd_@h$@I@`k-jVCc0t1%i6VKg4bXiP!f5NH#&y2iH7&PHM`+w6^JebeLg$OyUk83w4?XnIO|i0?>`yKJA%gq4=;E1z{?M4{u<&%?tu|3>P0MHA@#M45z4XGf&%W^N zqxk&t$-y1FaIx$L?bO#ZM8%F;Ey+OPIZiDxjEyb)YD>AFzur&61^&AJlMg@qU>g!N zk3-)o>G!!jFfHDs-em8OabA8uoqv=Z>}OSyNZ3kmT0pZ(2!m}ECHd%P0wuFmm0Y@b z?lSJZ0VzQqb0A1AyZX>&P9$-YY}}?6%EqmgqGmC)F$<#7k+f{Eva-AnnwT10!J;An z*u&MpwyEVPmMv#3Q&15F|Eu$3y(rbDBEOEB6s6L}jT5WUtxq0r$XB1TOZ zjl{Ot1aJW2GN@|c@|j15s0sUhSah^ef_nj9u?I}Gtarczg%OXwUblo5qn6i0h89AG zZjWN{D#4ScVr|GyKxne0-j2B(7Lm|4hZcsBe!TfCPYC>^)i=Wr)lwa@x?8%I$iWgEMut?#*$dPeq>CjpTHRGhfNTZSYluA@PLbc7` zP6V4u>E{o$jW{l(nm1Xk)tK8uRef2(2xH2MY_&`-T>ebShEc`T3ubCTDurZ*Imjh7 zGxewwOu5WVP>9Y*)r-}Mf}+IA;Ch_P5GBDm5&tAc91;)U-y?~%QrJB>uQVr*oLRcG zgq&N_iL*=ov0&)@bQm_3myl*T)C7=DTr7+3beW|d5RT}K68NXgBDP?7w^I$>ze>=^ zS}LeovdfGgiyBCHZR-XHBZ)<3F~)XMusDl}bd!=ORLzhTK`;ta&=0raJft>~fKech z2F$arZlnV@basM~969*_WKcz%23EGaztvWeljCyHG$+U3xw#cBE=#DvFd{Z(X6o$u zn>T;{d1T}}@BAY<`NF}2U%_iAm#Hdqk*&MTOog5kH`1^$yva`2XI9b1|uc@ry zlt{t~mJ+(HEnT1l-HZwyPo|@a!n@R$u<3UJ2{0Fbtpne|2cQt%Nxg)91Ao(f1SBB* z%9zbBJx^AWP~2@(Lw$WiJ%Za^jg37$_|Xmk37A6VJj%)V8KPnVI+dYJANkwmiZ#jr zv+#OR^W5!rksu6eV#x6WgUC=4d`H;(o!4`qYvThhvdY znPMDY#n1Ph$KGt`$CGy?4hP7Z@5eTcnoFcs@$uurll(e!CV5(^B0}PPi8>~k)!)Ae zAg=;+poa3$w^IYW55MCbNq_&$RKZhn9;Y#KsMjkz$*%_vkf$}+PwHp%SEIg5pav`& zYT#tlYX2ZA8VzVpsKiRj72{V@S=Q^d1}hjvpvlhS{0iCG-)ryyKVOxP!il!NC)iNZInn zBXjC&YJ;pWze19j@k9nZEab3Ao_Xd~g5$%%Jhd3vY;gM}-1F?SucQ;YE$Tzn%JFJ| z(y* zReXPKZD|Skk}hdz7DMPzJLHV5BG9+ zXJ<>h7d6cZKHrkefZ&?(*M+MIlg1$L@6Qh)0L}O#l26jJRB9DA0aOXsVi!=!DJ_@8 zvK>X|&h--A4hEA%g4xP=oY+)ogTWsap524}z0FSE)sHj)W@!I4VQr3pe8fXOLLna^ zQ8cBH!{tU?Y>eF2Tx}~lSGYS%)<)$N7w1r#A5&7MeNWKopO5(wb^Nfzlu#ugEFQy; zS)EF4N9tPHeYv@r7?@4)ovEnpB!P`f^R@fhEopQF9kE)2;-jU|44a#~;NC|G5W2F` zLPtlhmZN$g9yB8hGw4vsJuwSBTLc#+Xq&?;I&#~{mu#r!jb&T=C~gL?H8Emt zM5hgB#@|soDv=wq9wIRpV)0!QafS$d*DJNVOR5Vl71&56Y|tIba&oGwa&l@oM7XU| z@=Y(brXxQ*lfRxF_-?}n%(>@KYWv(d4&G=wGt-JZJ|c&k=w~XxS53f`pLAAH8juO8pz9pTLlgdA$TlcT8K@x!o|4BP&TO!zB@TSlKyG>~X*UI3_ z@8S%ZnD(|VuWvvaJ1zg{;c6;sR%WKqA*44z;hl8zYx7{md?y@Ru@PxAO zy3$wG8R#n&={NL}l{0J`H*TDn;4Z)!>_ zMU$tt4leiR=EkblUUcOdF2aO~8?5yC-1YS=JuryM*lLrBnms!PtmY<@l@18N&KL3A zP2*tVu-*`LbQG)}Qyef|=yt6MEOZF28IbuR@wt7n*1W z;lpe4Z}G`up{^fY0{M89tjT#;lP};Gs>ywm3461K^E&FnvUHggykm^YCyJH6C{n5I_W028b*cHT4hz zBLUxd62391sXCkq6F?vzZg8CRxH?)o++hF5byQI30VetV{Oj-k@WXAZ(if(tMme ze?K#EKRrEeH!9Ym(wCf&0RcPCVDp;&`zabAm3qAdej=6Wje02y-&-6<_2S=u&ebva z)}zQc<6B>;p^in^>%+S2?Ewo-Z%;Q&HK?6g3MlYB&9z_*<5)0?)%C76^!L~E`T(uY zw0ZOR@$27vXX}N_Borz`LCp8R-m=ZYzVJeGWAAECHrQB_hk7Gv)JAF}mZEfQeH*Fo z*vDqiryk?7G6aoq9X7v*qHS2FfsvB*m)5N|;7(v8A|AysgM$)=0nQq_aTfLznI z%b-*MN4-jZoFzXJx!=|e(mPjQuTZ>Rynsg#R*miC`Yq!1!mm_}Cp)zct;2A=kSGZi z9z~OteCVN&Q&RC)xxJ#a$W~b0EYERN3&<=*$44^`Z5Vaxh7YnN4@^E}O+&W%yL22t zaGUsCYByc{A|}$~^wD|N-<_qmuA$l1Qe)4eJrb)D8!j=^@IbWH5FDZl4~%1asTQgD z&1R$biItRrdmFZrA}Wz@XRDDZ`?2!Fjs}H)sZcn%u&}xso3eEs_MKVRfz z7aht&lP4c~OEo#NtDvCPt?oq?@pNi8IoOxt;#MGTO?GIq``|66xERCZ*WK-;UwP^I z=b!)UeVpvR`0jZ}kIs85q87Q3w(9+R&%mt){6T#ks(Ume#~#?t8jPpk0K6CyQg;m+ zw~Y4;075T$Dzbxs?0KLUAv~{3SDsF}9t_g2m1Tf#N3~1;|C4oy;eK z559UnwL_cwxClLVua=7Xj>_YBR00~@La$X^!2=Ue z#8N(upD7$^`2AtNO$0?8IdbGP5fstZaPZ@gKR$p!{xB$FnFxw-37`lN-2oJ#M-iaS z07WdlACIK~MQr|Lt>U^79nh&dMqGC*9ZFB5AEI9uj)(AjD0Lm4%`mdR7moc@Gd^=2 zass{}2?IF3kUVnTb!m7r6nL3sS3nUK8ImmsJ|Z296At#4AX(4P1Wzm+?9V;2pj(h; zuy*tn0I^S@BhFccug9m=1H&8O@9VKw!}npUBYOz!MdaK95y1+G2%Z%X5s0Viw7$+B z_<{JIPM=mMUprc%7N|khAQM%9l}e$~EA(m>JP#RI=>dcNk+d+uDd+(ufsnlnZq92-4sV$Omd!C&K zfBhH8Jbnj?5f0{ovB++*_`UKf5_^_Tj1CG^`Ey6nnh(rJoED7cfymGH2eB^jY5Yq52^F#%o!aCTzSUhV5hODMnu5p8aHOb z7(I21mD+2O#xBShof7sx_q77*1X((QDUy;)@5R0*@S^*H`xS5Jqe_#wq73ihWS9C) zyDg%Ql8=Q0yN}j8(YhMHP|ZZQ4WY*UTibtM%l44A&mxBfFNdgm$%_ptJ6^O@JYEO_ zBpI?a7P3UNM3T8aHj1W#N(nRp)3B6q5+mkOQB#ApdoLG=EE!0Q-G*351(q1Q9crUB zsLL%vX?C@hJ32dy&sI*KzH=uwRwgCIuaooZJP!!_!Q+$1C8exoAH;3|ZHd^aoBZy5 zFd0tNn2G5-canlg|I7mI4|Czdz&PMyE7HmHFtX0SpU2y3t1URzgP z4Y%KEqH-b^+>dy<5pnT)eZI9?4Fx+()WXT$sR<5*p+UHJ0TK^R@Pncyq)|mmWkWv8 z-}n4)+7oegqoV>4*Fn#3L(jiO{%#VLs$i(yV*km0w_?lCe==Mi%0FGPm#(7dQRQ?W zMeFWB{X4quG>ttsFaZu#q31)Q{(HsxA)6ddzM|^Fssh_CHphj7V@GWWdXFWgnm1)h z;AAvYxpSdT)KQ!HMt}w1$pX<9d74qXzNW?AdXA(y%TT*M#6QAd8cUK|!v0t7`U0rw z_?DH_`{Y~K;#&*y*9sRa-ZovyLm_^}WJt{{NKG=nbuw0<;ELLUix8Si1sE;V=5}DD z99VYtaB^(u9Vo=;G=s?pI=9p30SeI)ym{-_pL~kP*IS9bzqJW@JWX4XkHpE5l(+v} zUQw%60mVDJjKUnWFd@m;Bee6iaAYypCl495%47|q6q@7^>N!rnYSpUSQ%42R)*+9R zbjbuAWYrQ}=jiT%1a)@{=4*?M_}E(NFdO7#bsccgj8y^I;JaJl8?pokw6qpLYV%uL zh@Dz@3fv2)>H=VE$_YD>t)as$e7wKrGEk?YuXQ^5gk*v(LZ3`RpAdvvI95^jfUlzk z?%tLT&?9wq4I(qbOAQXdaUvLA*X|ppzkmJug<#3M1721DY=(~O`8()>f^@+yyMz4* zSv()5;UmkTRoqQXGQ|Ve>NySt4dZ8lZ~L90eexY<#Vo?q2+K?{D;#qxd{#Wh+B$5a z{nu_3z7J2q5+mjqTx3$I%nfXvTgIcXqve`?8II3N<+hq!aH{6k*hs_DgFpWGkQv zHq!q7ty@b<77JuhE5Pb`Gr!=mcy7F3z#n$tmC-~MlX&kB9dDgM9R_pH$OoZ|njz8~Jxeyzcqgd+tC_qp!XJ$sxhCsogCGKpZYYSUz;Gymw z1Wu0)BDJ$?&;}5$t>}y&)^{%!yOx%IC&%VctvxU-BhQ~NP|uq;PmNep0Z6tFUoIdk z{-HzT#@%=$3P6-_S-{(_%dYW9LiT|J2a3yVs~J+rel4Z`>t9X%SY!K5QmN^-p^yP` z&xRI%>53Ib7J?w~}ojIM@I+A}ow}gtVlc zs~>KDOdlW6EKF%{Z>gxC-+>;y8>`L}Sat3|58i=QC-KthoD-KSs;f&cpGOiOlpy?h z{@ev5=bt-z_|&DdIRLHaoQ1w#Onu3537Qef6KEel<)Lf3qtj4P=4O zLuuU2jaDe4;$3^b$T_Z+`R7XcmDC+g?6)2G|t-FbvjGp$SJcEW$oSG zu#^cSCrt*Hdh(=^<0nKjpkqi(oi;LQ@{}o4CMTH^5=>MOda1UQS@dAM6NN_QI*q%z z0r51NF8+MiKx6=d`(D~T=mO3jWzFMb;$uyL`iZE!JQ5*8d`W~5V^XHgm}v}8NE#Cp zHUj+9Bf_Pe1eH=alLyrvT3g!&pL&wiz8#sIX(B7>Gm~GvAD0|C}9XqyUA3A*&8{gSn zY8@PVIXM_kyPb;|PErtuSjbT5BpC`MN4LRg0Y}l6EhkQ#JXzVrk{YdDm8W)W*|KG4 zZe^`z(xgdhdu7pPj^lnTMEuag%*f51!9^e$L9n}BBo_fW;mAZ7OEMAeg3Z10ZHwTa zL8Vdcs#Q*BW*kYll*eVR8V=g2UR{+bKMG!?qw>txGEwQWMmUa3UZd`Wm3C<1!b5^_ zjZ|H!IFtefG>g=Oj?L*}>H$j<5AH(Zfn}&yV6-^_R_W|=GFo_SLV`3*U@*3(5ChZG z1)Zt8j~e7GBZ(#bo1M)Ii3dX30le2j)`8aGZy7D1n{_wdm@*{^DrC}>l(AEh?mBgB z%Jfw7n4Us1#pB`Bn71R-q(*reAT3mo2E)-4FTg1A-(RRBf@Dr2i z;$`^EE?;&yDuPM8q~a6!bJ&nzGij$q%1~+~4;ewQc5I}MY@0;p~sCIvrk+q z6#!IqjX<7&^Aq-k#pr4(DK3ZKuN)zm^ZEFtv;euyoCK>RaY~MyIa3lm=l0ugpB*e| z`*q8fAAUG)vm@vRX0bukpCax_)U6qzggfcy;zm9ZgzrhxuC8`Fypa`kja`&dNxksG z8rAaUNhS~Gu`glHy#(z~NG6_qashyJA7Fvv@XkizurN0yEm$D9E*Wf2FLRnR$ZEo} zRJiT4&pz8x!>Rp^a1LpF?e@GgXDGqDzh>moqc-3l0S&8!+s$S}p=7$VjuhmqEG+?K z2ChteS1ZFRlu}ggHAF^6>&=W(qhKw-s;n6q9-d+%RbV|*T>#2Mu2?ZWEGnE7NLGzW z1)T$w#PvZyFZkiApjMg@w4mUpRxK>EyoV)>^212!~ZTnzP^A zm~upW5?+ht>`JnH+Y{5 z9237_rpo1pvrVb=lSwt1kF1(yCVd>y{nR~^#wSm@`Kjg0pStVDn{K*k{se7MkX9Rt zEPvn<0)H5+Z7C@0$1(&(d7!MWB>UaB%{01Iu-D11zG7&NC%gLXWLLkEe|8Ppg-hp7 z96o&PTxog1$&;5a3%mO8;{I>k3nY;rGm)N3B!-r;dh+P`C^YNK1zWNeVWC1H4nqFlcgQX;Q6noBXo>I{rdc1 zWm{p+!M{HH(!;Yvw|#at_VeKcGqaq~F}(BkI`hy*&SrPYLYHTVTRhTaUcsJ!*F+A3oU~V`}?3X{=r2|X)T5AoVO%iP0ir( z@^V9TWa8w>lM|z(4JxIajgJct)K8c&bt?9Acvc4meufc*pTVFUY{N28>F@-N!1f*> zA8c!F?I!UBi5OS0V5Hx~NT0y`v0$WEW2EPl6(K-jE5Cg9Y)KjX=;zNBOVpiMvoD=L zec~7%ySMG#xP8~jBPWj^-M{ZhUTHqqUe2C5oqGz(>$&_g3i>rSlXVH)8^Uhh>4d7R z4Acy|N$nq(+bQ$&wBxtI0i+?~rM=tfCESEvU42fx=e34w{RL_VRAO=L173P-t!mnB zOQ2SdY&iD$T>kDMm;7A*kr5w#`pdx!r^^rQJ#w_kKc@Tm@#7~>96NjnDs8DdklN32 z0Xl=A6CfLLEZe>pHJg5};^L#1J@d>n%c5yd!KP0?gJb67uPG1iq~hY`K0vEr127Dw zGpenAcgK@Vs*dpK)1R3S=4z1=;v4uEKA|?r|8BzCLC=5o7HR|4MC-``jbR(gZGVN& z67T(YSZ6!%-Y3*rSvA?0#fqV5I=VhwIYDENfe4~&e@Oc_}0@HmGUGwko`%3k)`a^C@yn^2jNXCWA68V#4fEqocwI-I6gV zSip-}h4PYegnk--kZcA3bFP)JUgXTNa^E0m zr-tuA227Q(wizMu(NQclE{dT7A@P9}?WY_HZNlDD!g1y4>g;mVmoxC-VLx!W5DOFn zUQnfJMp@q0sIn0ZG{h&`TEtZpQcmq0|ap*m46Ze ztuC-vkZXd+muEG2y=sJRc ze-QI?ynv8&54uZjozNeh?e+n8D|Wl`a;Rsv3VTyKNh-A$m((<%vUyE)DRqF8x3*Td zww5(Ea5CLg5s7)? zhKb3MYSuE|4Wbjb+vk@15pG4l%$9tFuET5VIwt`gI;qBe+z9?ni;q&;;i%zQq@wLk#RP+QA0ojZ_sHkqyb5J zij)E{3xN`PUn%xqu=DayQ$JXBrNwoXHP!aUmWGC2FvkP6VMgqu%V`G7uCvFglPUb= zfR$_g+uDo@Mh{r7nUPBSeP#w-ZySnVJ%exbJiZZOVt5+g=xKbTX%{Yl>aMQva@oaG zhc9*v3;>Gi=iftJ0y6{r;3egN_&aJZT{;K16Q{b$SPnsX#i0B;giqfNBP+=Z-c&%V7i*yeWtxR2cgFeN8sO@5@oYRNhHn& z48cbDl)gGu)1(+X^Sab&aYh-Y7K{BP`IT2*d0?6;Qd)gJXV->rkJU8~N-PRE8xoV$ zC5KPKtzmk7MM?s*ABniz^Dngn5uH=uHqE~OiAOhV=3_E5u)Az02mAcum8q|%VPE@) z49e_^+0+D1PHq4N0vF6Dh*z*kY#iFs)~OaoI`eW4ZT@E4>C!GU+i_~sCJ-cvH3eSN z?yZoIpEKvi%y^xDj1m)IT3_G!^Oq}II_kTelCI0A3EX$j&*zVJB6Na!1C0(WYvq-l zoxK?1$mr&xjUWGvFi2TBiiR?*$HQ{TttU2ZI@pM=mMP?VolK&Z%NdGnhSHeS)Mzww z@l0n!4^OEg0!@ZIHaglPB%=Kc`(ov$n*Ag$1jq zo<_L!OMl(*Q-!U*8OuGdr21(eZ`ZF+Ox&|4 zFpxa4$Ye2%V-Sm=$86#hn|7$xJEC5CDGI)mq%~-_{55m#mIb=44@ke=*q==vKT`R zzzzEuj6p#fkd%v0D)yS675yW3p$~{p*or>53o^bEeUOAwRjtkTnmR{yo3qPS+uv4I zXDcqPvG=JAa2%oTv)vBpMh#-#KmSbq%E1P5c6wO3)Ew4$eA}4@tezS*1LFW3yQl%| zbh$CPec&I*ZCa85RhTv@HYP~Y)l!<9lQ!4bS+nnpueN9BOGb@17<$9NUU%ZeiE@_# zw4}fY#J#t6?UwUWIpHQ{jU#8=N*sh^B)ON*?fd3E3bc?v<-v@o;TVBSUR7IL(%xE; z-x;1T=PzmS@UCNz2Z2G705oo32=J2}62@3k3P1*2Xm$uhO;uTWS!sD`>HaKQ>gA9kJ8ni_uY8oEpryFfah_}oH1h(pI8k2@;rQ&C#g!R5Y#7!z*0j#-jbC{8!$!QetU78 zH4{U-4zmFV^YG%g-{$C=ZnHwFc>A5V-g;|yjn#;{rM0#i#OaX^Q*Ezz)PtF*;ba!= zX22~glQM45II_?`dCG{uOrTDW$R-%*`~!kuAsK^%o^BO{+F8pYIhH$@vkF2@!PYhyn=qWC=HfF zJiWO$m4CvLM*Jv}-Mjfr)SD}UvHU*8Ns>x-1D&RNa)M7T4+Mx#pQ4^|MEn>v-zK#Ieo5)_`u3jeRI#NG`@@2-GkFN0)@_9!vXZhA+fT^x@v(9<)D)$JLRYQC(w7Fc=mGW(giGGY z;2+>u^09m@y<^8h{5BpO*I~jDg5MwFACdrtl*-TM)PaFe0{1j1$KC~Ssa2-*SF)H_rsTK^DjPzcf{C&}<>nifAldYjmHx1#SJgk%y;h+sK|I1-R5 z!kAUsDv@SiUIOn^NjZvoRl~Usx*ZHC;C)yi2%3WDf_cy|QM`<-Y6=-m`P{^(H0Yu& zSXU^NE%jRbtNa4`e(etE>lqbmWWAR^8d?d0KDz8>^wFcbx{0up2SKV}#Qmz%)2k0h zM;~@R`g{hABLSwz-V#W)WfS^!a=gyG`8jBZS(S$B?Ke{{i+08&b#PpOI`}qq{QPB0 zUdY6%Fd!T!j4v!%k{PF7cJs0&Y4HF##-}a0V;MD<13*s}OgS7>q%{-vw3z&VJ!>uU zjeLGCXo|(-YibSM)C)4rexFQcrnHvPWUC~TwboXP#C%61yg7}I+Pc~rc;f47YcZ2+ zYHG2VP{%odo`cZ~=pz@|H3wZhB|+jl=_=mBkoPSVV;LO~0Qjd)rw`NxB2ORMV|aLo zG0+gGhpWZ_*FJdwUZDrb@Xwu5uo*Kjx+F?S&?v(^3Ne*XIY9E^S$@pE7pW1YQkpRa?34GWWv|yraEENY?R)1b9+Da zcTQTHon1?PZDyGE(;p5IL7NqRvJQKD&9+Svjy`qB8dD2pud1e|x(fPU86>v`{}3pv zYd~USL#@543S%P$+Z<1MEXZdkd$U=Nfg+?49y3Ygj0r^Yi&l>rq(aA5?oo za5yRj2L=G$0L172JcT&N{4ojv7==)bfr52jKT^n=fHz)hT%SsDCez5?)u{e_b?RyW!jZMj*Dvl|4evJn3e zU5IYf(4XR!hp~cwSQ&qZW=i9)_Z~^n{#qysEEEN@(vXL?wyLZexulM|QhT|*{1o^Pz-7)$;oE@SO;Mn> z?fePR$ju$^y;tv$!|GMkHwrtFQYPEFl{~3$IbRZ`+5HU zCse8@ECdfCWZ?@+3Q8Dy8jn`J66iUr#UJ z@9p>Y{M&zveM|DYNX#n@Ju?bDGYU$Ip$snj;^LZ`i+QL3S&X$4fcO0zLvo<3tVAxy zK+z0uO~aM~s%`AsfJo3%X*y5y0el$!!3UTu^uhD#e6a5fACIXshGX_7C-1e=i|$CJ zT4;@iqZe0NF$Co64|N8*gOpJUpf2)mq!^~73ocWIl80`1Hl3b9FQ9LyfBmbODof{| z;O~d&N*_>xuOF`Hrl#uZu6B~pjy0^k9Udr#q&-<#pVJ9hq`kD&5Cqc7&`^K>=n-My z?83?fBP z?uV3(|MOPd_U(52_U(V(l9T(SQeSmLL$zi6pSR`)2Mq?;gMZ$llV9K0cRl~(kG$po zX4R>PRTn8d|Npk^o3KKk<(o?HUAb988E^q;o!==^++ zLD`ExZRyGFZnqs}8;CA|l zTx*m+<>N=W+k5o#_|8~=ixvyrxuWW`W_hzMnlS94^Rv)P^XHD~6 z;HNRDnNgO;&-Pug8sKddA5BhgUw3csU}IYg3aobECc9hQFozu7*0uiKu7EIw${#j^ zdeDP#fbhSC#l`vfUvcIdND75NE(b_WAhLbNN?bb*&NtGJedD+ke?OAMgAXVHgoyox)F&e!l}Q)D5Zp4fL*E zY5WaWUSE-ZhR*rnt9g7lERj&*SkAv9*>b5o1N^X4u?+qVy92{1UVQPj=Wd9INC-BM zVnfD-$Rroe_)l5-!pfEP^;+#`pLtLRvA@@C9N}s^kd>8nfKVBDEw@kww1%EXFQ=E# zH^b+*9=m7>43ri+kd7w@l3se*#{gSc4uBofn_hcOSp4vsj*O@RK!sj=<+X*-j}L%8 zb~-&rICS`rjr4tg&7ddHadZN$$x5X^z;X;bM)r>*%^e*fVP?NR1Z6ymSVdjM52^fL zcRR| z1^Ae&sSgl{)GouK01%e zf^*?bz?=kFc@|KtZ!V;tq!&UuC_*F*DOA(9(N9_y`Y!pp`5-=$55^I}D}A+C3VzoM zT}=q1u>+kbe$?S~HiEIlUeqA+Mw4F9tGxEpZQ6+F6pOOkPZN;jm3Vu*t)z_0&_l21 zd97}svmc5n3{{z44GsSO?Hz=stDy(;onj=qpg@Jk*+3$|W%l+qC!sE7&CEbMn7I<8Y_=#6 z>8PkaC#)#S4U|Q%5Bc+IS2v}RLI-kskf;b}s#4~$Bt=FR7PbvS5W_U?KHA%qclO@D zJ-9Hq_uSX0ee|zWZ2@zZFCSk}0Pd{PCaRNT$k~x|p`e^Dv%Iy!LDSN4cc%|o&C5cl z<}K->)vWxXZ%kt1&Yfc-)mA+~5K~o7QlPU*WgIzu{`@&yFS-znvsFQ+er@g zm1O_69|S0ra&3D%&>z!Glz#Y{NGFGRD0BB968P)8nQ)KXEF8~Cer4wn9}!@lRJQWv zP}r7BmW(p8)}+$XpMTD~OxWW}&!7A?=jfrs*_*fSJ8`NMu2V;SRbKw7ef#zvJb3iz z(bIX<*Ic44D$0g(qkWW$R2%g&z!IqG0eJseE!E@2_$pZhvEfJ^@F*-v2M%;~^>=~; zL(<<=*U&ZSaaWd9HFdRnlt310RkXLQs;W&Wy~Il>hGUMVrXGbeazLWfm`zBJhqJA} zM>x5s-9=7LdHecU);Z`I+E-Q%&DUoj4J#oHBzBaHFC#oWZ7@4+HEkY>g_{?J%vG$% zL1>~&P;kUgrO`q37#^G1+Bj5+Qx8dltFNbLfL1^n+&z>=25A`Z$y6#O!+R`==gzr2 zNxf;^}OfH!V0x^&OIcP$L-sK|yqY%lK!TX^Ta_gpG$Gea0w6RHthof?Ee+gNCW zyKM}XRMCTS|7fbK%e^(5zOVpSUE0p>XCi0XHRzVtSJkrRtCChJ4Lz%nRq~GOJ z`U!yM3c?_LZM?tl(znETdro%i-H$&w6*ll2>_8u-R+*&(^{035{&X%lL{x>gTK$fE}V ztZwV*uvM28S2gk)$bpJ#ww0AvR<=3vato+de1m?RSZNM(qiB=cGbjoG<&@w|sADgc z+KZdSDEtbH{wk~m&tmjfVDtq{UQ0`0;GZ7((Iia$CkH=4p*VW<&kp?1D)9ekP@K_yM3h)D-Hv)4eS<^_4l>va=^l$j<&@ z=h4fRMW^%b!cq|0RY{hD?8>gNMR%{dJFl!qChZy&7L36zsjR0AMT|mgR^ADPZq=Qc zUww7)Vp&-gNTkk{QVgANWqG)76jgX-dDv6};_ukxF!Ru2B5BO8v`|LXg}h@YyT&Y7 zv}p6@MT_RmOV%?IO>8<+&hFx$XTL3I341!7+DZ=g>9Cgke-WtoGJbhrl-}Fj)>20< zT3FlC-iK_A4iB^gZS=LhMi-k8=bjE%| zandDI`+EmG!wy;a_$8K@S|ID{tLp3Fh6Mtvy0)dIwgLsO5t0);wKewI!$!vI;SoiX zu@VYu`rx2Xg38a}7Ke6-oDDO&Iz}pyU;=@dR4OC-JfVPrM~DRKa1~q$QLjI78c|-f zke39=%V@|;aFm4762BkC;t~>x%x1gPSX{Hu%;0KB%LUh(@E9>-)27cp8$B9fV1;eo zJX=|6YS~MuRrnG|y<`DGZpib66O0>|GA^YEDKQ_?hnGr`pKSBx@URSWGMz4}S?BUIGgT;gD)?+J;vQWTONV1Uy9k*+i`{IF(}5MazOM8l zh+A`yodQrBd-bIg$9_G1_T0JL6DN+JI9~(}yej|KL#NK>{(3sU4E}Cg)foUr;9^Dj zXa-}t?fVU1ehhN_PyY4w@iW`j|M0`NUwj62^OH|DY(h|d{kNNNw@-0gcps>p0PU5so0Ieg&251^^rcA^k4%nUfs$nlcwgF{Gn`2MY%;JUk?p9kxX9GU#> zpe1|~L8M9iXhey|^DoQReGl~Ocb55(ZsNGGnqzyuK;ZKq?|+Kk z*zcX#7PjO`=;u%W{RJ5phD}?glXg#x3jxi6R;W&%cYG@Xl?OE(Q+o7-9XaTKTzfHS z4hWKpT7su9y7#XSEt@x50;p7bE1dx>g8Y%ly2|r`D_&ZFIP2qh?w~4$j?Ma~9=`LrKwLqCpF28~Rak45R;Ctf2AmpJ_HNse=r9{39N zOF(1vBLAGEsShRjyaVlxae9ea!z#5P;2V)J7KsTH7juLG@E!V&k=*~oa*=aRfjZ4r zR?*M}o=cP&X(uw?(COjb1AXv0fH>OI(+tmmy^~h*fH@4b*CXgkeKiu~H_c>>)kd0%{KXqg`>ci|h2!0kVeSGr%{TUhQW4bTy+rNK*YU=*|yY}w;;@Orth{qO!EKtjjP73D)?7+CZN;u|%l%^daEsTky`;c>PK!?6Mt`jJ)$bG-iR; z2u(}b36t+nrM}Ig0&#Vah8BUf5(HnM_}U?#U}5gQ`_?!!J9gHr>mtk&*+3B|x$wj4 zp-)h`mhh&AItTnGHFb^6w&s?WCX`!h#(@xJ(^)XpP(GK-D?xn`AW<2KS?a?ZJ}Nvk z*kFVsL~jTQ(}jhHhlYlShv7gNGBokOwi5P~85p~z(2?)Oo-&*b6k)rB-Qj#uQ6ZR| zi*4mt%r9JmOCuNT;Ag=x#33n@#wHBn0GSymFU$?_ALE_XRD34FRZ+Z>91AS+h2so;hCWSa({EtWdWQoW zRF%VMbZi^yVza>1E6|Nf7e#nH5hGVe*gCOJAZH5Yz?zzD4n$YFJK?$j58GfLl8-qh z`L3vSjjw42_b$FP=%^vwt0Qz#Bk@2wP(%c>gpdn5A|?vJd;kMR#ZlA#M@w-(H=*yR zVBXBfyfI-vH&L`!gBb($RG21E3N#4)R+kqOPVu5*+ny{Xh&VAn@DbWfg|7=V7-`$Z z{a+xOYcLoo+r_<~u3NWuEg&FYbDY9|%6$NcKQPUoBNRgTN|au|7ICm_SyOY1WX#;# zNVNWI#8|$OAzjXuICps_9KXc+14v6!UyqMkahj&X?`36Wy??ZcQ`FYMGuY8wSG!h& zXl>2^kGl7OkE%@Dho5uiOwD9UGMV&b5>g?A&;b$|4q6-Ceuxs;|1NJ0}ys z!e3X#)wLqhMVf`)0wDwl2_d~sdheOZB$@KRo--LB%Bt+X-}n2znanwpGv_?L-uHE1 zcYQNul5c4STNV694nkcHi4AZtP{V5dU^e9w5EQ5xJ>oZa&c(>NhjmSioCBkTP45TL zW3svAO`Mp;h&+p54Glg$c=%M3@Z94lu~$%HNq8=aF#dY?G7BXFAuq-kVWdIP=_@J# z)+q*7$6QrXTwPUGf>Ecg{(v5R2FoY_z2OK4l3@`eh=BV4~;7EO=m* zU>_~6s1!|FaOXUdhG~8;K5ImGFL?ibs%tc%OxEyBR22QB3gAVf|AYw!Ijc{Eap>eR zQBi6$2EgX9(bN9mrkxh1Y4V0iUt3$>U~U4y$6VdmR8ZjE*2WoPvmS_znGSjbmU zc20*&kgpOwP_cSV0H8e@t$%=8>*tRlM(Gvz@9r1I5(R39@}4O$mJIWr5z*)Xh}4XP zGl}ZQQB!`BY!XeDf#AqUHkd?(C!kw2Lai@4U?yXQZ|)XC_wE$+`xk^oB{x)#Co3so z7-;h9x-idrv#eH)sn=3hUD!~KFQm7juy9DGfXJ@XQV1#}QF&BUxyf6hbSdOcD{C9_ z@{>4uIpk8^N*x|&u~HeZQ6P+ItP)84Nt5~tj)2xB6E3(fX5h>5dcRFUnmKdkj1Sao z+LW~5w&dhwg>4h5;n#X=0=0o07-Kr?YP&nEyziPdYYujLB}|_Vqw2Hua=*J@d+jxo zKXpyDb7y4I9gjZx=|JU3o{&o;`6HQ0KO$s+Q70d@0T@ZE49kaIp=>E$3UH+z~~Umq(ert$mEu#>rWU z9FDi7JS~(!)`|Y%OqGbFG75$PI6{5(M2)3o7agNPha2>Yz9szI1wlNyNw5 ziJ(D)0J0SUXm0ZE>0)}ihD2hu%+fjt2MY*OT3=9rEeT|38l9FJsk-%GD^TZuL7jVe zv8n!4S6pnfnawOrdEMvVUjHBbrmCNol+@6`afc3h&Nt8u$RwT;wTdPYFb{)$IPJGJ zi%|&)qenwgq4J#n%O>){jWv;JO-*Odn$5?KdCoV`Tt4_sEk#sUr`Pl1#6;mdCNc2_ zTFnQ)t--iQiFs%-^Uz|JU_MxYQDPoiOvJFHAEPRq2HN}DI(lK_EhoJl&E;T9C~qd^ zJY02(Wn|ESvBEJ(z?g5f4gq6`+r&U%T1^qfT~wg&)meOUBlehI9WQR7?91|&;xn6J zTel^h#3h+GN)w+#9(`uOMTM zAON{g+|<#>VFMwyc60-)A{y%MXsyP4Q{9S;g+m&`kpV0Et95{gr7%#?DoJlA7C0yd zrB-E%=?P{~of zoV&OBqU-}v_JJ6ohphsMBqJj)uPis0C!1kJWVL$r13vRt*I%WuHk>daN+3mk9b%AA zi8O5vX1Bdk=yk#ONFL&0=yb2^UdE(OgFRtKM{8Ht36R94iU2H)pqF}jxkDjlMd zAI}uep}?yQo`7nJ{6a&66zCHmTjRX?D`E1J7Lk~!w)RuZTwvVfRMKnEtY zP8&z{0iCtE7m7-2;T+Xm2@MO=k0}SRx&do1Dd8vC?S6BXFQ4v3(<<^-N>Lt|VTPP`^ z+9d8bP_1<;TvGU`u+Vb!cC=d@Y1~+~`CJaAH2Bj0^*bl2@<4u;=^O)Y4NW8VSq<88 zDX>lXkbEpTpIIy(Gyk=@$_ce>WgCJm)uU@!}^bf7_SBdyVS; zsk+MJ$B#oi8{BbmPa1HD!uxOB7Y7e!x0}3rD$Sj+b@SHwt3+v2is6;#RCb=ZzNEYw zmha6q*pU|&LE{3hiiWPb-0U-1<*m(q-T1AtprD5CmfFq0(@f&H*5jKto#}(ArOeUW zHvnlLMi#L`#SLIQk;KfH5wApRvpH4F0DO6ne%PwPm64^mStfFbq>ipO3Iph~!-qn@ zr41t0wgHiUuv#gSibWDxKh_j5OBig&xvdvu{d|OEpt7>Qo7Qw!Q#%zdB05&8lA(Qw z)sf@Ih00RBSf3bgi9+V(Y;U#z9!Xg&N1=UDcK0N-uTf}U5e9~6j~oG<4UB3s&}j7c z3y3;q+5}z}6%{vO!Zb2<(i*A3Fsr#)_t8fJ@=xlGPO)|^`G{x6{RmMOq-v5qKmIt5 zsPor|%geVdSz=ESiXZSacp0?Y1+=4){Do%@E(Ndx2ssIS933RB32U@%ZBCs|(bk5} z2h_a3KMCR)wVu)<1?%-{qSdW|{WL(Em^g5hCk?g)z<#cii-fRP32+3%xow>(BK&UT zjINR?BAjZt;vC%-mvTV@)$ZVA=-3)Rz4=(?g)Kcj+YjtJ-!4~BLnQ!VNF`k0YT|FY zDTZSETrxF;2SGu#TmHd&OQl6c4r)!V)2gIQTHM*6A{eOJ2arf8%#GLrvR|ml%i{%|^ykrx4!Y`sM@xk zo_=(In|(`;o|pssg>}jom)vaV0z{%O?v_Q*!IFW=**k|0hoXmGeDR4}up#{xCX!o# z`NKXH?|0Z0AS(x1?c~LaA9{7znX2sRn}Qu@Hf`F3ebVcgo99qVA+L)U+aS;K4~>lS zm!`$#UoI*uH&?=Scu6xrOYM~xDvzEhYLAF$Y46CoP+ZpwcV^Cxa$o{^->%}bXG{B~ zQYmBWZlH94-JFW|9s;7Ng2KleTYCmvX>qOnKH&krTB)t0p&HBC;`{hXaF=%hugSX(Yp!jpmtwfjQ9K8x<*ZO786hse|1; zyVCs+G=(Oom zCqODWVFLEeMp$j6{+35loFLEn?nk;`O;ta+aDu)8+!rljv*tejKN(x+ zKm@^a_dNOJBY&sJ&J7Uo{0r7y^qs~&Kz`2~g}O|=;rdSa{`gUx`Ja3!GUUG z?Epb!18`{4_;}rDr*q+)ITPFW?LA$IHil89k2eNj0>0t;e-`z>3iZDp_5bYfoH)7k z@DEuKeHWa+SOx`oA%Ij3MVE_xGXgmW0`0=HpcTy)~xy9+Y={FoW4{63Wc?6zs#(#8K6{*mYTD^Oj~R6S`g7q z*SZx0VdH-{7Yl<;@O6H!&@q1T9k2|eBp_R`1o|A(*YC(-$Veba!TSxbhv!E3fSIAw zF!R2@&%%Vq(IT zLlQsetNf()p4!r)(#qOSsB$ODWF4jT3SSK+aAJJJ!zI{06KhUR75EN-BawqhT&-3q z3u^XA#;zH2kU~7mEL*#J&m7ANDbq07+&>aJ4j)l8?)E#$(g&jV;4y!FA`}bF0+lO~TCGp4O=#*9P~izj*ktS19$@^C*t=sjXb zlZChlk`;PUn6K9&C99E=uaJ_pNXcrXWOi94g$_1b!AFL(xzgRfPaHUK@*>(f+`9`# zc~s1Yg_{@i>*mr~QeuI4wlxDnA4D6E=22hqB+mJJLEC>y|EmR3F0YMpg;e|(LMzVQO ztzH&I&V2NRq@>{9!b9tgM&tTJg}ot3rrCvgNY{m&{KCScP$XQ;D-<}x+v$LWN~wU$ z30g_{mr*fzUB;Qd;2Sm{x}c&$l+ri+Y&36Zf*^%*aVp}XhH5b9^ybiVS!^Z5GORi= zTtVy%J^_Z2fpwU`acn=v-F~Ob;S!r>yR0^J5WC3F&xX4!R+pCoD@A!r%MkU$#Yx?! zM6gwdqF)Smw7>-mHly%zj**ga!o&z=F<>KN#?H84=nahESZV|*b-@FN1gI?nwnuD@ z1@|#%EFxeK1tm2B|;+m#L z(9TdjXLVJvjkKdU;5rJPbE)IpG8khNVO%R39?gfKi|hjOj0I;lynP zRh$YcXt9_@33&r@ji12mBQ}L(SGvODK6z&^bfBn5wQ`_)NR5^(u1e3?{mxwXAny0! zvBIJz&ez*fmz|OE{U>0R`5L_8JU^a+g=;2coJZnbd+nLKArhTA9yCb~!gSYlsb(&S zCRtgfHpAqVt6&MV;^t^Yb(V>n8Bm*@M}^#<>i#klTiVlS^Dm_z{{9$RMuR5i4zNXf z{84@2{qmBDA>FxF`a66!i8-T3EB^y6JLIl-EdwW+?CpOdS zs32=wr(I^T4A_VIyE;J~)z#5$v06-%z12SO;wo3gg@9H$%sX`U?a85Ng)FBaf;&B_ zFk;iOUaF8vydVbjl3?JT$Z^nJ4>_?7I)}B&8H%uMACe5%T=pSWApir`(Vz5a5B~;gNzk%4{9Peo(0MYHdE{8z+dN{PhDVNm!tBC_Zzo1^|Wh{;B=RwrcKzalD& zwkU_ne}lM8WVgZT%5RaGMWCi4P*VvQ2O>~Yqfk?t>gwj4GWdN`1}WRB^7FweWK_Is z+EOd0^S{lq2F9k6o~jO|QU+swS6UkE|BQ;KUtG52cfZ5_qN;w45>n0c2hRXL1nyF7 zD9}rcWGQc)4-a`ayT-T@T+>|BIOq!MtZOt?RYl$P@WlN6mhbbc+uA@pXH>lKw`HOI z)RUpiYmLTcYqNC1tDD}@-qylkrwp2+z`)V)C`{KFzz`41dpxb~apYwY@^TyU64h04 zxwN(*`)omTH*m7OVC~tlqpogfC_n$$u>!1<8=Bi;x~8|=LvCBXJXwod#KGa?C&tCa zjnZOuA#aTzH##xFpyTYOsG1g0(5QFU{wuwTjfjtrx1Gt5O8@vrfB#*(-hA_|4`9vx z9W5(fE0x*pHQ!>#`9s@4)a`H#(E`0Cb|}2;aS)n0*+C9wauZ12Sx8*ICZB6o-9Ap0 z3yS$;?ClKmQFH-D7Y7)DP@G$>Q-R|V!8?Unx>rCw6yXtF;3o)T*S`m~?ntS@h$)aA?gVCV2p_Gv- zez@gDptjUUj!c?%{J2a8*>tJ-494*@(JY#OEDap}%A9%`<1t zoHJ`ms06zeumCAvTo5NM#Tu_f95oMtziU$DUAl<4+AQ0@`9IjJg}k{Z`+PwO)>ip> zm*Iu`(xv=@%egr?XJ3RlFU0dA=RoK!5Az|LHy@FUm&L6OYMi?Q_dG+_nO+-L1) zy-aa^ja?X&TB_@PCfv3#(NReGBg(4VTk8w4r!TB;?S%hfsl5~Bf>npF?~t|D%!hP# zwpAdi!U}Vr2s#!Q_QkbOSk!g(^tt52yAN8k)N9cnQM31YYIcBC(mvAF*Y5-BKkwc? z*l?QbItTj%7gc?OF3@!AmDJx7Vq=CZUGCUADWI^Tt;5QI&CrS1fHH;enVvqU)D+iQ zo`tGxo}B9MumIIB^$$Zu`;yMGJ-`p{DeDx`^?Cw}9PlDW89bhHtPSxYmoGQM$)H&4 zuf*K*(o6S8E3IcXZ{D0~)5PBYl8Kpm+XCS7+$-?iAm=!%kzcxX;W)O|4F9qh8UtqB z3mLA!*W~ruy)z6pKre0KLai!g?Io3gyz;y6j-5urj;Eii8H8~xI6{=`4j0>EAWx1R zEI+yqG)Ja5(8i$-_wO3C>7#>s@{4oM(FT>5dp;X@%be`o%lSCNy(p>_Ja-~#-GPmK z$b%0qoB)Y-xJD&|cUy}~2_o%?hzNhBLhiDd;+ie(D4t%s?ff?%e6-7q2T{ZJt{%{v z`+)tlr=uEc(Q2Ve?e_LIvBKSb{6?YsjEOXO(_Iq-pXZ_ANDtA#YNOx1Dp`OUrMkJL zs8On$OF>_ausf^})?=EobJ)U)2JMUrhFdDe-cPHxr60gos+*IJ)Xmue@#p~KbdH9z@CT4t=TIl6q@*OO9j)hn_~D0htq29n5XlJb+_p**NwsrJBT419 zH%vbE%ex^j{5H=XL#K?}qU|MEHwwCU?$Z*x;i0)yBM)~|4;g4^bob{1_W(q%EczU9 zVcata2gK zC7nh|`lynwvyUh>DpLRu!PtRt6w2&A%OD#DvYzAV^;86c^#%-AuF=scHKeNt8=;;q zqEQa(==@O!VWcAg$fVCHa7;|W-uCBWqw7TKQ+Zwl?&O0b*9~l25Bcm2AnuW_o z=ACy;upx9S>%6yZTd?587hx}U^k^piRVV|~6`dXYP}2Sm*q%&QjMS}h>+QUwqb(L6 z9~g0Ydpm^7S9N#S(H9XoX|eHXY>#lP7LIXvERC(C$A7D5BCJ}tH&i&D7mo45bF{dB z>A4``-tofmsBm-%&kg>{b9&+4(ZVrHI0gvMNq*(I2;ttz!tu0l^b($LegEwBDAe&c~__GH}rAO zuH%>cdGvArDdykM*FF1=U+nODYxAJ`hB`fvc24fM_xxy!$`@O-U+w*#y~i()03qTK zFSt1WwNb#c_4vi1z};`6QD5<>FUqqQjrxKjmQatg)W>dhJ7`nF62hF_P*u5D);

    Bd zVhl(~kj})&kdO#H{;wJiLwuc_Cle#GN(*g2J}2-wYjR7UX*to-;gItyzwrKO=v z)5tV@TJaH5w;cB_$6de2z02_wMJZybCA{LIvB;tu*ZA0&xQK}8IM4N%Sb8x6GVz!g z)0k3hL+YNa@?$V3*-gTFuoeH}PtMcx;YuYr^n9gIzzB{;NdN6d!Pd>5Y-oE;(hEA8u z){wdT);s6T^Ysml3RCUe@bSm<0h%|B`TX+_VV3spu>trJj|QG}T9P&|By!*wy$HBN zo;HbIj1KwybNCvUwPV$EYi-cPFeqF90>bNe_a#jX6Sec|+KY{$2fy99)7RHomz%fv z&%zEx^zA+Od~1R+q3|lEK1Zvtk}WDJEvDj4&-GHeMJb}LiSzPIGJgyUA_(9aUItca z&;`r2{=6TygkXHK+nbv^A+q40&Z0(|rZLzzp(xw2S+Lo#VFcu(zuVT|*F`Vlx9)xl zG<0&djt;Td1kulxu={P$*TC(t^lI49hhxXffwxfz``_1fL9si@R7wjM&-4W=hRWfr z!Lor~0aHamSy=(3 zu9RNN?HIdq9R3t?<&bbnr)mse2AN5INi-V4*^<#70Go89R22N$&4Y z)m?T-K>2XgxQi}(iAV%rO3wby=4Nad96+Brp|RvR(-@sbfqEB-9M}{<-inv6PA;d~ zbrHzGb@0Q5m)cvdv)gqhK$Mt)o*ugd+)-D zr|~(4&%eLh3%`0klkwf+xm)^gMkAs&U3Wa9v)RwaZK(;jr%}SbpN&bL@e7DBVi`tNt6srpCw~WEPa~Pm{|Hes0ZEozi!L{LTv8%&j{ZQLd>?SF<W8JYz|F7q6 zy#D?l%OCYuaP8WV71|3gRKWW`tp;HPqpjN0pN6rUWVl@ly^1lhv_Ft06T+ga=AZP4 zpFMlEN5IaIrG4V+vp?w-KYR9S>xRM|JAXXe=!y3yJ>$Q5mad@bs`5&H-SJ*+<)r!F z#QMMfTA$%-bFjwhIaHO=F&BoYiC#1QGG;adQF$38_Pn< z)Qa_O#_5tGbM^64u(SgxrLYj{xa!){(sC>nYuek}8o_c}RMaz2R#nfoP*}0v?v!>_ zwu;n&1|36L!hvq^GK8?agdHrRo2ydj{;`1vmpHoEC5nq!uGoBE!W@Tf-^4l039IdDn0(yKmYm9FF#pZv-pjDU?y08ant^jzYp|88LzOy{Hp(2AlMrwJ8Z^bT4(8SQ8mh<~E zJG-#3hXb+3(-#N%=)ogf3+*<#=R{pz=|A z0eBFf2;L+YlpvJGmRfx$_y=m#L!~UlfDnt=H?oaHbMLxdJs-V4$<#&qoP( z3dp$i96O+i^mBAKn%i8|3&0Q&S&<|L8hS1K2x-1gO zrDJBNYjAjdt3@lPoVYr)@?lO~B0Zaxn|n5kd<@F=tgQ5Om~*^MJ`%%*+#gm^6d<}r zbP=v}0S#lHB@a*-C#dDtNc(!*+AMvf8n+A#tdYjW7z}Z-X);dl+jr>L>C>l=9ooNl@8133qj-rYdwJzvSTpo?p&q(=t#*?fVhW0w#!EI~ z)`o92$rnDa0OE?h%)OWFTEXN4Io&gdvd&8$nS$b-0^rn}!$2t*3|<32T1IWP6iH%DDX-S2m&Kh>(w%%y>zZTKOaE3+_L=h=+EcLR-W9) zEANxb!-4|<#|;P$^XR#RnHIUDBMoKwRTvO5!_INN1S4o-#Bh9wUpACoVxO}D}1S-diM)_z@oj!c@#7UIT$rHzbqB%x*UY*HtnORwxr%$D) zr=L2VnKhzxW+r{}r6xK1^9qqYKHd%)iYV6h6yQAU8~`e_C65ZF^XNR7xeC5)$Xi0` zh%78?5!is)PmkB8vbD!QrJ@@xEHmKi865>apP@^(!ri-!3l8S``v<{`FM@4wze#&O z@BF2U#VDoXi>IM2JWW33WnM>*dNnrMEM2LhE{mBg+&qnD-ptz2!Tf#twabb5yChieG(^6S_qPa zaie^?ODk$Snwy4vbYZw285|G{!)-$pwxm&pQ0R?Ap`sp}Gbw1OcS!3G(m}l1vyUIo zZpXe9?CE;hP@hc|l{2FCsj#;VRBC5IN)kC}KJeBYu!+1Lrjpwl%~?O}-o5+4#?LoE z=Ts*F9fqVX9s0&}(0BU|+^ec?>I)3hk9`0xo)-GTrq!CNmg+*M-!~&EX3D)Fq09g7 zy}4kg#9Q>L6KY1Wp#(N9ke%nxi z@MTJNNgkX4mD_{|sbcvX;k_Dlx}R5QD`+kSw)eyl=;Dr?I1PmL zX|kOkTk=(|eW>o(WB_0v=q70qLXE91` z$ECYzYwLiu-EeE8Vf?JUA`GFm9=+OH_Y19cySue^jBKq>APs55t@RPwS^@8Ct8cgr zgimX~Y;^L?QxnEdFvui!T(^ON8W|j$t~e%m`po^_C%RXi%t5 zX)W&b^6EIZ<*m2g+H$VLOK45z@^(MWfV~Z6hc|8Axqshx8D&K^fa$d6r{`b-X>Lc+ zwl`DimpMNkZl%O2p{z@{a6Nr?NY+l7QR9_?to`Z0wz}E*BhdqJrRp(`8E4 zN%}6slW$9BcUy68aYa*YeRES|U#c!Y{frsW_O+s|xAxhA6tvUEnxzlYa9nh3ke?UE z7Ar{f#A>v4zd*Sd^xAfHNLGJ;O?@Nu`Hl58{lij=C<6hTo!e3_(ZG@@0|Q5jlnG0i zMHp7m(E-j>OHXTSPYn;yAY3tbTJ2;a5X^8*55k$7V%fSQ?;6*rWCJaA7pb{>N&3Ys z^jgJ~1BV*m*8L_~!)dgP(?A2M173OQ&iK*OCWMS${N$5QCVTS?MkbLmI1&czyyVHJ zAOF)U0j~=V8pW?;klOc;xvod)Tl9~8bdbU#Meprw!+MIDbk}1P|FDl75D!{a)9-%f zHnN$V6W*ER2-ro7!RdjwLZ2HoedS!Nbmp#{2mSkY;hlIBawErK>WO?>|>16#Or z9w0YDWUgnpf2OBnunJOp_x1Sr#7>y8C@E>%wj_%sXxh@h{q1i{rv+I|8nr~q04mRj zQBuNN4U1ArUQB&Cwe|M4Rj-jTGD;jP8!l7NjS7iG4kA*ZiIhMj$^C+2QE8Z?gz$qx}tqp z!JhyxyAo~!E%lZA(Wxc>{P5F{-e`g@4%{>1RJ%Q4@$+-Z2Kp8~zc|5eBDW6%YImSo z($P1~!uvRRn|uDD@hVvgIfMEE|Ag|{dxFs~2e`&2jHW5*$z7ON)-Vr)25K{XOJ9F@ z7N*Qu=%1*i-0y+;UnSn2UgaDOk53+C`3IsWX3YvT1$Fl+RUI8A*{8qH&0V@Q*UPIR zbDPOz+LqbiWm0r@!X(>4S6J1&%w@y6uFK{kXvh%64}eWpqDW6axO3~LpKjfG&|Utc zG4@jLVTl+!JH+4r(lnel{1{geXfn^C7+ zx2>fc5SJ9Jv^L=LDaSm$9vGcn_k%W#xqlpX9WRl$c}^584whoaA?3O7X=A3{VPx8U z-4>KSh{GRo_#uv;`uh-XJ~AASq_0DWDcSW3Q*|{q$>j4we4_hq2jv5+8y|I7Ft$z~i zl5EM=5^6X%4x295OQ~EN+Xi}bDJTqdk-Ee?7+;Xf3ZoOvoiDQC)y(rw-86fQl z()OSE;)`!K;UN!Y} ztwIZ}wyvhRxt{s}q^}xyG{TGe2|)oC@HeJC9Sav;^yTA+HD4@MT!#fhlHlj#zO42M z45Jrg;awyy4n#fwf&SFfDk!_ll~!>sJBQ*_FIu}r``oz; z^vcCc`R8(T&k3#KMl`5nHy6l4v8@aD8#T&)elPrmZTo8Ts8R3|93CDjE7-hwBfYfu zynXZ0qnm|RaU(iZpcj-Y6jgwlUMepssS15!K7|De?@eU8yriU@{#2z{u|QbEZ1()h zJ`q}lXBoL1N1;9ZNBY&V*4}P-T)>I~fXdDuY=>aN+15rEv3)(=y;iHh>(SHM1}+dQ z+QW@$SjUD1>(p3wD5Y4$`3FIO7#bR;(+QqZ)Vg5MPtz+w{<<)Z3qyPO4|J?YFiIT9 zC{c{|a2TV+5sVVEYpwQ9tQWz_ZR@ZHV5_dPc8UvYwYuOStrCV!0&Y}`Wu-S7CTu5z z#*Q5uBZxsZWag+6W71rJ|$p9lqp3zyANqhEdX)zPxxdeGCfs)4Sq4G6gn4mKAXe3^@Ap`}`4;A31a z->86rnzCk?vUQj2T=UM69>f+k^X|LhZ`Cnm1+W0cb9fFIz5kpfkrZvi7X8$ov?Hpy z%a<>o6<~L=4M)LV`sJaj`LS}>`|lsBYix>n{PD-1V4zx65nPO<%kK zK5Vimu^!%M$h-O{UijT(e}Db;RZDJ}6OWZ%NMiE52OoL#(G<8Kr0;k9pLy_}2OfCf zzWbK{Vfll?TulNJ>Wk<`#;9d z2DHAYp|ejr5d;bM-WE`C_)rNlNl*_|)M_?&%3voN8zY5@wKC$-M<1Qp^L<6rMaUiw zU+UuYH!TUowfpXQw6^s8sb$ON`!*f<4Avoss+$LrmYHVv6=vk3N#$k0a<{hxa)}y= zW*}@3`{L78McF{G*B9j!Rh>JQo{rzlJ+>;KunU@+OD^XX!J49|IJj}B$7*RT&Ocv# zuCf4&E4aHpdlACy^aEhoo9*MnS#2^k=;1AvZb!n(KiwXN-ClJ$M6D2}vYb{S0w*5K zypSUm@E-Q0r>Dv6xCEP1(25cbHkS%+z)2K<$%++$qk~i7*gh~eL~fcLr5u18K#+3D z#eQ5R_VUHuy~X0rLkIRBGuM_EV0bat6<&gUZdOiCPDS3u6Og!^$gQgGWzb5oOb7-D zI#Ac$+SVh>%mqR4*{pLHvM*svnQ;^=_1O{$84_s>K`aWnrrH0lWm6d_D!%*b;K4Gu z?f_e0fDHR15eRsD8vC%b)!hB}DVAmcR4n}2xg)1ObnQXByTp^dO~&JNRLFGkuE6Q~upkQ11WR=l-+U zZu}`K^YQ3eujC9fl#sEL$PS(8k*NZD80*Q|fBqCL52~&7$U}JyXJi27V1SI%gxHSp z#EO46@NnVCqjrg>4m>tz5qLTRbr3QAC8NkOe1z9fyvAu6eN`SPX%!9WQj>^2$p3R}QxY^sn*=+K#{AT?U5sl-83ThR@hX z%%3-J-uzn@En0%nbjhM;Z@c^DyKj5;<>Zi< zTVg_zZz1sAG8{uvQW(5Om?2F!Ll+;G)T!1&l!b#y zrhhf3q51V^sbeJMo|XYU44(=QQUMswVamT&{Px=zJy-Kj7$q5T?g4DWRR_@KEXCt z6ZTL@K07;;u+aly?0bQPt*$!3!W0l>mFI<`$hEPDQwWQJiNzk5 z)pfxU0?=ndKK`2&ut@#SuxTFn+VC7f^=PpMw;qk6*d#E_!MMhpFfzcGAU7i6%|x$E zh5_BmvM;D)=ZoR*peJF!h+a9zwbZ$X$0IyrNM-_tQyzklXCUO8(CUU+dEucL{-^MQ z)(HpYi4*A<8`6(M({|iw^n#}X_@xrD#APSmN(@P~MM6k{7m1=#u5gu5dh}PtC9M~I zlbm32Wum_n1anMzH2&kpx8?W|l{YH}0SMy2n1=Hm;Hh8gdSLi78?KgSw`>hx%8P6_yoZ`T?Zi*lL86O4OaJR$aQZYE@_LIY{2m z)eZ^5b9Ts`>_dkZEI_J@TZvy>0!_Dxkmu%*2GYR1^pbG)es1|K$;rvLEPqL*dWohR zFj-Kv!s8K_oUf@jr>{mir7HY_{DLSL)=v{DT)0|ODx{n0plRQ6*OZvn z2~y5N*Gg*&2RPdPXY-!U)fE<^5o{v_$c~pxoCxCw%P3}1gYu&yiUl%q4n+V_&>WC-lepnX zNgwD72yhcE!}|>ylj>(N?fH1)H8F{GhP3aliH45oL_rx@)1csFXf$w3cTF_sOGHv4B^*o&VlFv=t z+BKhlhD43xfleWpP2#*fNbPpwLK@80mO)!q39^ zf_V^O6oT;^VT7S(u1SXy^`b(Fd6>aj#RY_MGon}Q!8wQ&B8&o|!1WH0-wMEzV9&!?Nix{#NVth@BUX$N z=wN6s5$-TN+6AovE=duF5@B#BE*8_$d7ft$FJ6t{m@-AO>si_=F$AxM*%_S`tG;Krr&TM?Z>1 zlDV-+#3-~703YVK5-8U(4J|Ir}XhmFaiNc=~V=V02x-QdVyOT&GX2w&ZKrM-!kBOQZ= zM|$@%4dn>ef=KYoOesndBTyO`KF6)b!<(3}kYut2A)e;J6uO$0B60Av5OMndW*K_M zCQlg(V^b9z=E5;9;qYe?cN^C32f%^ci$9aN;PfL);1>}46SB`Fu7;adxMKx-J2?5p zT1rhCJN)0(Hci)y*X}G(M<~BnhX)d;+4&|0fzgoXU=T=#Y)2SR7J9}4&ko1acLmEr z83K3_2wNRcHMHk8p!hKw8ceJPOibvxfYa=u|GQIRKyB)giYU}(7;$?{R(kM(kPA|L z;ETK$&{-4#(bU-kY#Q#Waa#U(KNFF_)`9^>S)n3dr&5rGDeI{^5OK*I!L8OM$J1@ zQ3q)ApPEu>9Z>tuRAidkccvnHMvr|b9o?v9=Fh@lf>|L9!X%T=+<`FSH-$hLra=gV zu@w#1AEM!TfYr|ICZ7a$)bf9s4h6yIkB+@9*)Tza&C`b+T zCv(MS)FXo&A#NK>fL4Ndg4iKi{J8z(f-VFaV6jsN`V3fxzYC%+he{2%Ve z=>ijv=M9(2|9*bxUWJaW8uS>t$D(7a^g!`}T^Dy@yfli;kbS46&V^6eKe?VlWBWV& zYzhM;X`2hx%n=L@QUfuJR*WX2al0Wo*^owcLx3`3k^835s%RWqv`h81aTFgYzPRh+ z)i}f)5q3v9f1Lm%O~XQuZuO`?lrxCMC@~PlXvHdQ<(3M!kwA~>V+U`87;#7q1 zUJSXGIl)TFyhxIn80tA@G(9AH#-9J?*&xIlgAyEtc!N-aK`245rluNfPN_%@5>z=i zXy+hHa}Ba4QC!^m^(nA}r|hWDi`K8lYLJa7M$3CgG~i&w@TW(;p7ivrT9pExpAC3_ zB3iWyPP@n!Ok2|Bd;CMy{;%hmyv*<;gE6eBp#c(9xn07+0-j+cP80Y)sn=-gJDNm~ z_(iAt;fg&;E z7(jxJq}b$jXGX@hufN`wk#Q%|DBF_vy1zQqf6sEH)#OzVlh=w0*ud5cNpwnR3hkW4 zIM`uEon%9>xGi`hK5G$kcnIn>_FIT zQtLGu`iJZwLwaQAr*{aY?#WT?wK<}>L6%TJkSw@!G&ID3atZlCu52a=%?+|d{Q^*y zp!FOsAV*?kDA4jf@**k7JD>Gb^A3}m<_1}U1KmMWJ zOU-46Ijg-N{y3wykJ^#VrQdvJD&g{;YhF5sTU@!~9O0@O%=HjYsRFP+0?{hqLm<_s3)%=&Ah>Disa>vHy%n@bBz z&E~8Scg+BgxX=wx;0~NCovz~b3iCGaMd>2K`#`oFCi*9Lt{?ZL! zoo|%^Fez(2|JC3?Ov=h(#(2@nl$ZgNvJv`_+QwRXsLxP3IXhvg+!Sn0MClKWhp)dc zJ@`eB{LNkZRN6qb3chH!R1Tn_K4U!%3S(HQx#X~9h#4f*LrlQ;0pCY%&6Z>@+!g02 zB@zrvkL5lyhk2nnnHWWqsf&pOBT$36F+(DDVP%Kq9mX5I#y{HgzfjI8h?7S-zm9TF zL7XXsi7+r6##>=?ip4BMcpQqGmZ$rL?*?;a1w4P1m6n4yxdfvsRAE)s_`eLgF_49q z9+rY;bZ7t+;MD009)+zvOz!P21OuCTv51gq0g?W!eW)YrM4q`rAOFMGJ1>;h42H*g zTZ(sYbpO_Q_SmK2-_DoTIz|DdU$T26nj9<5TsIeE!z$NbV^(iJ@WT(^@7u7xKnL2# zNLv$7?2XJL;q{a{fzzo}-pYaYvCj? z-m6Ey^2*9Y0J;aeYA#;9SZX<11b&;SsSECg3EJY>v*WB#l%M!v`$r#rwCSkHV6Lf! zBf0v9I)JhpF-2g?uf~MeP!F|QE#{02DVS)*;QVD%!S?S2i!=sE*yhoa1UzHNK@1un zU#&*%r`P%T`TJuIfjSLJP(QsN^lQGDRHA=12zR@SK^vHjHZU7)AO>w9hA=)_)HSVn zdK6TmB9WHPIOZyIZ8c>PMm3inHflhe0geUmAfk?uI@D`uz^DLtFhRXuGIs{V|FCIS zc1b-GnV_~5@1k`VJihx!KtXq0#w=e|e(o5PeD~e=E--*RIK+31iG;Q_mAC8R^3T!J zV=gYZ?{_Qj8k008@#e*MKKLXp`QVeK;oypRK{jL;c@q@I=_@psG9Xvuq?^sLKy%t1oQU;7-s-%z># zujK2O+XdZ~Azz*UK?&YqTLM;t#?U2JYy<6okbhQj7KJ z0i#TWduGpb0z?okq_jp~V{>2Sp2?d*khq0c6E~bd8BL&e)99l_juj(Pi1q7Pqs&1Z zjM2a@A0zZ4}ZLig-hjwon4rMG8nUSCyBUD=N$A%8bsd z@bkt)D3vBYy>sVjm7Mw&TftR+cjR3NE}r9ICV0<1!YxadqpIemOITAiQ9*T)Y%Z^Vy6 zea9oc@ks9sq&FVvjYoQY;UgYvqe^o%+&}?CW3H%xCdnufVSNNjeqQ3h;?2p5(tO87 zHs}}V8zRT$lm#>$XbLFHffmdtnK`n@*_S=@%>-MWGZ@+oGOhWr7<&k$;_Bi5>t9A&9_QON{zXyw$=Q^J|VAsmb4|)VFjkNeG)~Qg?`^ zPiRpnni8jrI#&I0{LExxVpgnxg_U^P_&=^9Px1aawOQMvMpc&FH#Kzp_|U2Ml~j(3 z+MZRLV+yRU?y54|Y}GYgfcsQmuC6(MzP9dibu~G~`&(sdms4byh`MbGg{@m8v5TB8 zwajV?^kMuZK(%V+dd5c?D%Y$0{M3Q6P$dcH{hb;{Y?WD64jGt!yJQX({t+`8=dff7 zV+>hl;K;)U5xLXA;==?75kDyq7s3LBp75e;PoXacl~F#aL#`GInQFGU?LLLG*qzlRe>ts({; z-GB}MiBu43)nRp8XD-Jsq5K4|9He}+VhQAT_8}4+BrGzA^M|K|Z#V>?MEq4Md)cMR zWht?$@7J&YUKM-KlO5(-yNV=Jt&cHP8$56M@~2;WZP}>Gt!vh--?;H`1J4+RUjzJ6 zUP53PABRI_nLrw7WPr1^y8M$3&V?%%I!pKLXzS{CHmwVI(cfBMUWGPu_Ok<_w>|dQ zG`&7JCgw%_LbD(w-klL-!JT+_MhF$1DZLb4@Ii|uxX$pCpM?>Emee0mQg{F7C57GD zh}V^pa$nUWYoR^8y**r5r0(Ngr9HYko_+QXT~Fz;Es;sl9DW%twKU&`!y7k#ux8Dg ztz^_PI8l0P`SN*SQgoMFP#|`BLWoz2O(;DYeBgjwt6#b#hCQ_55F4}N4^iQM>_Af` zExW*&7*Md^^w?v!H5~Ztv&^QZI`q+c>;F>eO-Jd`eM$;S?>>wbD-29XFjDFMbDu^u zy5Op!^75kUoNiziYr1Pkwm{(d{D97f%$Xwpd&r3Pi2aYy2x%G6;Qf$q)r|6U=PDg- zx`=u6B6JzU<*0_==){|EhL}$_?cs-6ck%h@Jat3^pf8q|oR0Hur;=ADwEhgXK$#6U z(L&Q3k5Sawbut!oxx25gx4o(rj!mobMmEbJn9cR| z!!QwSg z&ppR8DJi@tC51ePFlWC*a(R+#3aG7btgET5uZ7#anwol0>0r*M6MhD661{`91z4d1 zeEs|>0gA7$UJrDWp9T{>{gCKuZW8jg0(tur^7af7krl|>3ao8z!nW89xd%N#5K>u( zQwK!}Q`lEac^PIY{DT|HVc1h%Rstaj$^fG1LpX4&rU18brJl43mm+ z_f6t>3*LhGNri>*I0{=w1&jHE75UoXcPBDJY-fs#ira1UKbu9VEGz4CU9to;^{7+; zfRys+#T*~m1$lAjAW>_5wQ3;j@si77B^nbf7j?kgr+biL){t-CVn&DX-2L}2NYbQ= zd?O>JsodCvoA1A$SNkMB^uPo6&z~}B7GxQK6AMI*%;LW!laJ)TPrUo(DN|l~1zUhv zxZ>uxAO!3q8ECGruI{r^QR_f=cMGx)uLyFbMk|p>oomSKw{F5l-3(a=Jwf zaDns z-ioGwh-8zTbBBsI|8qUhMwJHeTvTr$>QW`-ncn3NmZZ$ZN_P!OGqNcpWj?;qcP4DC z_ZgYoBomk0?n{z*Z|ZQ;Tnp!`4GqmrXJPuwrgCsZhCb(w_O93XYUn92Anw|qgw{R* zc}qlVpNZB!0ePE%)*eQ?t@~A3QHu5~s8!$xv$U+DgaVXPQ8UOQyBa$>K*5=}`M`>b zoSfR)oE*KnrBUao_wwrMnvmoq<0ms|sp9|3+k3!ARc8I;cY4pXBs1xgLMVpNK|rJz z6&ot*u4}=9uDb53>$~;$GP&5+eODLzx{8R1ND=7-LMI`RUMIaxZ<*f9|NBgctE~Ea z-{0r|`%U0xlH7aeKIb{lc~1GBb4p5FE`{P{&Ma!8KrrzQtzh1~<;&;KU%q_y(xpfh z{*Zs)#@{^h$RGYtejfr6%s*{hOiBYC*|Y?m2rD>EO^HN9!^@oW)R0s<^adUd@buJc zc*F>{NrhjA^=6}C#6;lH6>tJ!n2DER7ZQZ}8~J3NkG&MiOz=S}9HU5AXJ#l>1V{aE zSEMMj9F$oB%B%oouKG=>$uMjNh@ArhAu|=P9>_b$1-!8KdwUUs!m+1!Oz6Ey`Uoeq zDc7nND;=`n6|UJ=Z4vYuwVK982BWEoNWreIraqDj-PhFH+jQ>Sc~s@8X-bK&y>AL| z%OI02fum8&Nv1rlFLRlb)~bfMQ?#1p%Wbx@G9nspx#f;!kR639?zru?J2q^1^s&bt zf8Y-4aUv(*6rg~OI!2j+{l`6`=;qCMc{0qRAz~$goX7>m%K!7L?Y!G7u0#&g# z&7kzHp!5px;0o~Ilxry6o8m{1wg+1bbTmTf0Yd(+F;Z`lVXnzD0?CXQ4zbPOlDcq= z)OtP2$e{mGdpn;Owl;#wb^54Sh|YghC+M{IubF7Vf?Ft5stIs3XD7^v4vfIry!l&# z7mgA9j_c8Sa>@E9pL=Td!kgDW0p04cJ8r*m_R}vufhuF5_+PT%4`(Ccgc&dx`r#F} zQ3{w6QzuOlGUzGSP+p&6W{}!|O$-_|A^Z@T4$~Ovx5%)@;usV#<9Ok~-aO?R(x<2R z8??^A9tZ7|P+m^vLp4VGAG$}1_LD&SRgi6~G~ZT3P%1!0xHlk!##BS1peV5Or9!r_ z@38}d#rvy>~e2 zMBVYac*ZSjm(MO%@cD}3*~`}&f@Kf==Ap76?t)8GSEr6-VpFr55lc0F<`$vuK@m4_ zrH8AYy;3dBIDERX8|`?#`tHWl8JXRCKl*5|Et67U-x(r`A>07if56Mb=b0d+7W*SA z4v@}$QE&}Kwhy^N3BHwqn-ySKMez@vH?|u*4T?PjZhZ#a`ZT!p8F1_DQB-sZhds_w z2O_9Y6$T9kpSQE>k|DuxAbY*HcWBUMbC}={9}L8;0c70T03OK_=ff@0{BG*gqoIr`OXwMim%tW!LXp{-vKF$&d(G4>mZ@)a~PE@FcGRlmw0JLKcC**MR5V42` z6Y(I2p3UQNf$Jg%Au^ZbIr)WUMe>q)vx}6n9Kf*fXLd{}5)(Pbn)jORKA+oWvRceGyWQpS_^$duR*_W(o>>K+A)Rej zfoBT42M9{Th_<{?BJA;Eu2Ho- z5evNi_U`VmdIcIyEmwuRcfU;+rX^4^+cKI+D=0_*`tpJ_XD!kVE;-Z6*FE~^qwC74 zTx|hRHAvb-=O_FZ3Pr+u?pdG~CA*Q*(v=XY7u-V^`s_xN$z-&l&k|l#)R#c6P?$>k z0H_-swOS~nwg3er87vNIcgcbijw=$2(}b)jscsmFv4m1FU8v5IA`e!IdCtdVBrD*n(o|Ezsa;QE>@-v=`&- z9p#qLl)S%x<;t`KmU%2~0#?BUY6H$08j4Dbmo8m8K^h-|AA9^mc!LH=A%y#|@j`?# zD84KM;s(LHdM!Kko3V@*HJ-E_Pf9XF)Zh>`V7ekEiP92T3Slf7#A<^gMohuivHMnV zpx&{W@M-VetI?3c?uWkKx0k%vBn|a@vE->T8YSX66%4La%bBtW!OTT8m3h-mgM&+# zu3ouh_04EI+uz?usmgAqo*+mvW+W1?V8lkXoZicO_Ux&qMG#|nbXZh+gA&S11qTXk zEJ2B6*yy1}LHW+O_?_RUc9e9@SnWrW*_I8Or>C$P@OY(ES{{n$5=04;+tAY5(t-!& z0yX^{r?|MJxH!5r`K#n-)Jw1O7Y@``*Y*so@=z78Qi^)yD_^Q5*~ph_@xGw}Fgivg zvgKblE6hWAfufmIWDKoH5GWc@GZ)mnio-yhoq;Wi$>I*cJ=HU~w}RqUq_CVULS7a% zYXv$}|0iml;>g@N`?Jl|OBIyj=uwqsGQeb$H7YP6dK3`oUd3PVSll5vy9|T-DyT>W zR}P0zhHNHzKekJ}f+skD72gCZyn^}sJ68PSAzjU-eoNB;X}FzwrR+&`F!eUBh%BK#$bI-au>AC83$C97nHM&7o;;9W zoESOx&O7hC^_6aE^7`a+ncEK?I<({6x8K=a+tJa{W7*d)hZQI|jVxB-e8D9Y9k`si zo@&sdWI}{(*o!MK`>ZzUB0W#a5^B_uQHR+YbzzzNBvTeV`Q($67;#TWWu|X5;_B=5 zaV0P#qi9xYt#3(lM@6}df5|8)#b;TduG8_=Hh%!wolFH*sNR!PMEaXHV6rbCx{%%rnodpB9$cPjCHV&&l&A z(+gMrY~9T(mo22Rl_;k!T=wKL)Lp?B@AvmE%oQ*{LOG|GlK;V|*s=o2e|INuP2QAT zDcH2L4FcD!Yx-gtDV30Xp7-vaGu<}t!!O-L{f&AHRcrnTD`q>jliG_p|2yZ2pD)W| z{th@d?(bF?DVRRfGIT+B9Laop8rlc_e0uSfOILLLZ8awvF^bfyp&s9K1HJv^dqcyC zlhqx~^{I}?V}Hl-^pP{CuaFCBYwHx>{&xHg(igOB)M-uTye>|n5j7MX6mA8qqk>gf zz(nzV#>^1YPw*o$>@1?#3A>t-+3&0ex@dIsP#02%>cl-}+x9q9xO^mJ!uH|(u$ zU5~Dtm|Vm)GtjHk0j}6Th=n!KMbuOv7j=E4@xb5g3L|=j)VWeKp;j&L9NHHVck=Z^5k?DN6TX|;z@Fg z&C*8Qz=*lK_S~ucpX}LtbjK@X^6@p}l^sWSZ~Ne{A6K5cZ19)|yBY_X*tppvAH2Ww zY-3{DikUe|nPz6Tr}4s9A9y_j7mjXO1EyJv2;c9}o%omF#rerQxGx{QFyNtH*Rn3v z9DEb$ChUqg-}xJW*uNG31M*Slq0%NVSg_#UWz?%+$oHAQd*%h=$$5X~pYOu2C#O-K z&KkBN|M>ay$N#qJFUQVbJb|BT>{P}L>nZhOoUwf2 z^b9IAa-j;pzhI#(7x4YWMWdE4o3?Nz{=PU1^_o#G3LCN!xE#IFWVS-VwVF*9ixC#D z(QGgqiP4Ka8_`ijagB!I6yC%{><0rBwvqG^TxKwt&35Q`7Ms;%)Q=#4d@-uQAQA&O z50tfU*#I`)zF|UdDOP;P)130Tl!Cq9vhyjLH4PzXa5!ER- zCpR|>C80CZmB?}bzw)rD^}7-4_eHGVKVtoE#QNQc1v9H3I96gl!ka|=Nh$Lkz9hI! z2$^B&W09s7ZeOan*YHq3)Gjf+eK3k(S8`!lR!qY)2FmbiTMGbq3x>^k3Qi4kSV+j08!<;e|WU*gwK)LU{FE?g+> zsHQ&EavYBH@4D|_4(29bo0)&+i7E=^QkTqa$E5h29=*fmc>M9()dp|vhlr?bwa%LM z#D6?a%SuWm(uAaZ?rqQiVewPX-F7z${f#;4US+><`@HGt6>LUY{@Q0(qNdGQP1CQa z_k>R@&haAn+R-ZqtG}id<>bs+t~`eV=|^-Oi1#ezT!80{Sj)(9 zh93famf}5Ss`OWyR6&*g&p&2NYbpl!%+sLb9Q3wga8EI~N72>Q*N2u!UEST#lHjlE zAc;v46bPiaT;%S-r-l@I;%P`J%lN)Fd1rE3auKwaTT%r-vjM1&;D2tCBTOV-d5LNb zB;FeNwN@-jjh?93cykLwP90}Fz$k<|$Y$I|Z9;e1O^n+ZZ*GEQOTM&;k%z23Ia8}J znNT;%3OmP$4y52lL#5%^icMm~1{<%Q(3fwY(YqA26@~!}VImFTYv_P|l6i7tH;e_iNe1I_uy$ zCVEA~hs@886)|HcV@Tp5*>N}G?lIy8b=S4G zHp7H(YNak~bDT~iDv$?k$u8R?y5se1=d%H|lC!MD^N0pQ6wE~Ytq%UT)u!8 zclkZP{M}2>{qDIH*sHdUeK8+`EtTVTk2-$)+m~K?>6u4%^@7QiX|I#lq2d3vz)j}? zV%+Z<@uYC~2D1$Ytd$)b?)|M|$G`C~N>b6r~Xb~*TRBJRS_a>q4no%K&18jg-*jF`_ ze$}SbfU0vq)j6Q522|C6s?vVLNLPOk#24foqB7^Th;>1ifKDc0v5?cXXBG8o(t^gX zmy+`2yGZZo{N~)brAcv8ya{{VlH~O?H4#TGj?~9!iiabOlpluox>Zyd8*mRQ!fkeWqapmONyGUQd!a6Eq_Z?uZBD z@gOZ#h$JdS2G%MTH43ZbUGw)b&LLymfOU8?*5TDLD&&-qz!Wg4Fk{IJVNs%T;mcq) zAUe>g>+C=CiZH_DqShavJ|SSj%$ZXn#9_cA|Hy0nUHQ|?pgC`$-V#4u&bOa@S6i`Z zhn`YsS;AgVZUxL+im^F$M6 z$XE!jv`5#2V1mVL8OB(+i3l1m0KbHi{bZBHmxwPy=paj-rOqW?wUjalH0 zt2hbVctD4;4qXF1gWY|d9i4q7s=-7bQ!O(>@m^zz<@?SqN&ZrNVz+c;fj#`;2X6?|dcp1$+v zF&=oB@Bqg02*$Dje7*sE&aisu$wBi#HD0a#20tTV@9G+{cA~~vRn^(*5d-;4y}pO& z%695Y?Np0JKQPn_t-A$fW?{UYJh5%x$@6$!s5YDX`bI1^+mNmaO{m-Yd-XPY8cU3@ zb0nx>!yL^H_0;b9Os3$Eb~hh}XHzPbOIRIVtn3V)zY2cMlL$qj45f@8wAx2Uoi27B z3dWUBEX-oS+Z1Lp6pH-(Jn7}jvAh}AO`bGCp0Heo{t&B}l8pFBBMu9v@cO$Rh`Jrf z0J!&OKl>ScRlabU2nwMoyJXeIjT@I!cWI{vgZ>y>4h1ZY&$6Q4p26mEQgD8-l1iV? z<+224%@TOv@HJWj{G3_zG(Ji{Ck0Y|CLT0PAWS+U(Lk8VjSC@EAcrO>BxpgJm8ph* zgvF9z0Zv8v?EpN%p$NqahT!M7A{WC7#m3{qjshhI4o=kmjYlJaP}moxry0z6_TgcN z8JHihZ`AIy2U!BjZz177=%_>8CQ@?E++yp5zerEDXs7yoL9Y#vX?PJ#egvArEJT=* zp_+*H!@nHy`HaQ@s=u>&thmo<9K%bl!Thhm{I9|MufhDU!Tc9h_3L}|dd#Z6-)1;< z`tYt@yAD;=RieSh@j4y02_2-8Mz61HX@TltHWdMn?BW6T+JKLJio;`b|t;a+c z9X$HQ$De+hle2Z}Cp#+-96yBn95`m;VpHLmE^XVktp;4Jkw`L%iWCY1;!sY^sNS5E zr%MhWHc?q#nJ(adHdR&_%ga0D5~O0p-Y*$Wdk^{F9R| zkrfseDkK7ELn7!j8bpR963j~k;eH&2(zrXoOeDN9HrSmV3nb{GuDW(a8tiuDP?;J# zO{B!6+2{H9`$&~Yg3GSjmiG~ zt=G7^-H2<9?bYW_9Y3(=>mz4sI&yMet4?DVE-)I0yHJe`o9xMbpVw5M zJ$33-ZEGLu@fn8)nIXH8E;L5OaD0m+#=*-6zRE4kP8w>qY}ry{NMse}?KyP0A1YP> z01|vg!~@E^WH;P0do~JkG_zF6s>op}OXsb<=PtK9sh)J(3opEI z+oa5x(|O13YvyA06!`s+Izla&xzC&CbqOaoW5PcAmB)Z@u-_;W55XmnF@5LUY#;E% zrNR{7_mVk~NMvC9K#J{IF(23-%?Ln!y(vJM(2^OV39V<71JIJKBLf`}LY-r5A4__A zdO=5TJWkjjx+2)#>lp>x8$B`W82#5m?yQB}xdrrJ3%Rova;M~U6QRGc?ZVjuP&JLl zFZS#?Qr*-KfoE(ubMnC6eaH4A8@|YBG!3@ZSA+3SUuKU%ni6Qcn-7aL&ERTX}Xx9@}bIxp)S( zQd%MYqUq^3yuK|&m=4L)07N*i{XgqrKXm#Div!tHvDRb4aV^wqp z?97q64p2KA6u>FY{h56xD$wZjQ*N%~aH%_>;zq*i1EdIZhn42x)38!D4--qw8{`WiUsRJTV?QBe_aTy zdzBv_!`L6g*dN8%AH&!m!`P>Hbo5!hT*}#b30vRUBgeXQ!$dxG=nn5Wwtqi@Cr2(_ zId%N_@v19*kPf|obC*LU5vg5H5GoYb=^CMKIvu*RL|VWVP*oKjwT)V>hMty-7cc61 z2Cek;p=f%J>YK`r;Rx*O;%tV14YaPDO9@!A9JTtBPd@pkvE2n#!qZM9#}SC^@`3^- zBWN8#7||&zDN#e0Ql~Sl2JoNa*M?t7DzK;Sjhu=y&j5&$x7uG z5h){=t5iIu&oX;-+oB-j~geS}YDP5h8J~*+Nfu z_uCSFSJ2_HfZYQLM!;eO|Bo2`m{f~}kUBhwUBG4-B+{q@HFwIPStVUovNH!+&B`t(KV}i#fvi?$G!zjEIZS9r5jRsV99stjEj1IoFcZ8mj+RmpeaLKX>S#E3Y}cos zehS#<@r$i`_-HeZkxihgij)PdsX1?kGK8J@JgwTgH795L_HEk^?D`UCe^FeFSj(lY z2-#&65Os&e;$X&tOs+yvJO^#MXVI#so(hLIYo z5b!YpVm8<3v+D3^eW=w55uHdVMO=Xp(K>|cWFi5FNh!4%DU>&Z;ls!R0US`77wd;s z4Gr1tW^-4s#UHewYNyTKC%&7z9zZ8@(A3AjG#OZVA&YeDS>=2zweT5Rt;^u7iC~9_Aa-9lU z(-A)!)DV{pu`&@f6o#Lgp%G|iEV*yV2BFFqi%mz&RS}~Eao{D04$Wd}2`ni|OCyOy z1mHgjL??$Yk*A{*Tk1v#p(4S@hJc;XU@@8N1=bKe8#-NQCpBj4kQ@b)i+nximE?6? zk2Ric1$&28CDEuf0e27CYLLf?qX}7U>Z+=4D=8EMrm{H6s89`hY*aESo*mY4>X3?h zO1Eu0(jODB(CaRcarvvSs-#X~enA21)DMheyv=?pI|OUvrd;GOas{E#bPcT6g6WG< z(RDHPuvU#WG9D9FAtLj+!dOgENGLG9P!Xfm1Tx|s&_$y*ARo~99Iq5e5`b6G!sWs2 zg2O#_K%gbU6jfyDbhZ;C3-b9!qXVr9&BRM+Hba51kWE2twF=P398gogpCbpEKqo?} z&hh#yC7{(5tbr+@6`Ai6&`M@Bb{#*~(Mt@2)4jSA)Lv~`PjhosRddP?Ft^m3Y4(?2 zeklJ7tW)m$MDI_(h@b`YHK?m;%a0y)I@Kz0zCtmtUqQ1Uc;G3{AD2*n=cY9K=uLnE zGhd`$)TW(ZzFehpIKlBgUrVu%MvYCu@CX-ZKPJ=T3;Gb8O+(B(kq{Ae1uD8ELd%FQ zj=pC^Dc6s5wZS=)7J$#+>T(gb4*9JidW;K5MjffSLu#-HHL2QEC^G_*>Miw5lwM_S z&8cK63>5RJ@CK$&L1BvV=L90r%prR{GX#f6$_ua__qiMQxeNEX8x*=56q*QuHEb|< zHC$SMulX7zQe)h!|yLNtc@W{SnhxY8;^(Blae0mxu40Y3*JvfShienU^P+nqgVU7;U#j1GG(1n31hPnM$|$$^8=Y3wH8 zlZPPdOR<}W!^DL2Bt)VZQR!SBc$aswMhh_Y(FMdSE zJDoowkH~nH>PKXJB>p2Z9?IYk%lJ{tkH~l{B;F6p_+{YeW#H(g;OJ%G=w;yO zg3FyIYzU^#%T;@K?%cVz>hiID`}Q3;b-800BA=){c}C-)uC2Ki%v#&rrW-Wc?Om-+ z)u2{&b8DxaE-;N*p;1%YIc$vQ+DsfcaEkHc9vGvwRHHUeCX>KZMr=^h??)m#al)KT zD~BaoE=`(F(9B{PqauqTE1t6qzgsq^I03!dC@Nm~5Dd}P)Cz4Lb~x0UC-$ct?#oDE z@nSVEHiWSGP@I>!m@Z()qnP|?j4e(>S^YFITLf<83)l$?dUpZ(*<%Ed@?&3uJAK`B~t6DJ(4bzPBYDe)7qUk3BW+XL<0M zUv7NrscXG0b``R@K<}%57C2u3TKm4YB>_Mi+Z7r2v*c1E-}ScG?TE_iuZ;Uyy6U?0LqpekTd??1P7|Cu=4XNPB@n>x{>|GmJ_Fyg`(fxG z`=;t!>%a2A30VKv+WlrTnkhhbk6ZsECKH*H@7n!ab93`DIp13U?DV|s?5)?@{RIVu zg(}#7px_)!ur3~?uWb%guw0N z)_<6Rf|vhp_m@G<IRnLB^Cp#t<7bo%lI>btjLXKX~}?@ydg>WQRPYRehVkJZq5jx!tJF9=d67!PGR1w=rHlR+e0{j*TjN%8g zotuFF)i6A+b3A}iKY&p`gi$|$Q9p!Hm$h~eIQ{mX`b#HaNq%##h4?v|&Ye2FbNkmP z&R(d)efnySAK0;D$AM~y*(nBtE7;Lbkoy41VBI%1Tp=BN_kL7)EahvjscCF!Y3}Ii z8!^yj?VilconIb2UvGgiN%<0ryu7@4-?0z=$z=x3y{A6=?6afQhdzV+{rnuYQ#(Z_ zi`hCGt8u&Gh@iN1;)Gl|yn9ZjR8u+=;qyFx0BKG={bnRL(`7Smm?m|4l#>gjP&JhW z3O>jpE6r8RdwS#I!YRw(VqZ4@*88C9u3tnwsZ|Aor{9e!CIJ|)5+!_Y7q(|2IPOP5 zSGXn(aN_z{%FLXiN%=~CEJ&A07|^xCRGKOmudFmgGQwfDBsVX&pde2vL?6)@L!6<8 z&o)Pi`jk@`3|?B1T1X584v$1}JbqNHiTXT4j?{bwgK@ZA#C$e_wq=g$%FdtPTgFNt zWfvVNhQP=UhbQ+&^-H0S$LIAU8_w$TVqa2uy*E6Gj&;zjeWOkSAhBA_(9jW+ zx?etI27#P*rz7ZhIRLwZh#%7ni5F!$=6c+VQjWP!S!O^mq$xZ=#Ktm^7Y#28L#w=U z#pwb`tf(U3@%pU&DKAQT0-@FSc5i?8olWmmwGJf`fB!p4_e}m#tAxX|?nqVD$&(Ug zmO71P9`mCxe)F5(+<()`b+!iY52IJm);YWF1t5m77W-h*A!Qv$`zuoDMWTgBk zRIEZCg`?ar2-x+YW+((dibAVQB+TA;0BxF+VXrUY52pMm8GaBW$`y+wp){2Q4Y1j4 ztj%7n(&b8vcmpA?#}x#AfDt2W{UKdn4`idfIX!~K23 zRaL`iLSPzpz<9)KaF|wQ<>X?8Vs)SI55si+;tO?dVUdDZ&oUW5j&8rbm*JGD8*(}o z3XDQVtAG;)t6+`aIXS0g|pyUod!e8=VLOF@2(P}F#$0u zn0!*n9R_rECcp#4dPc58%HbBr9iv_-kmya!K#P#3dRQ1RG0p%0P;_uKO_R<$nYZY1c?|a=H zh4V#!LTP|a)E~4G$hWmCCJTiVm#)QFX{jX67K$ViRjw+#XrhWL(MllILP2YPe@ZEi z&{93J8~aC0R~mqttM5QJT?u5FbHo%1*{x2DWo%~e!YI~b6t`g%w_y~w5zZd)At|w| z7Y1bK$wMD~^ub5(z5Brj^pfQNlDzxjk;6wTkDsWlJXzHR_I+Mki~)LGE)K`y^x~bv z!{H1%J!H-=w^x6?fB(Ky^_R``1WnEdEzONBy6(<4**=?DZ?wBTqmw_|nu`@^Ht9M$ z+S@)u$Xg9!RQ%VGwzgw`ufV%EkGHi|zKwSZ#Zm--U*`P!{`>F0?^jEK^g%e7o?vr7 zK6}>8^78AZO-zhaV`kbBhc&9_vjFGkm!cWr?nH36^S;f;M^n_T@kpwc`OZfvl764w^ zX&F*!j$E3}WOI3Z7TPEYy)c`KF)PDzrB~-_ z3Fc}E=4u({Y6<3Q3Fa!_)!OLsbT?k9Zmz#{{`BWNYOd_qUensqU0q+>>ct+DYu*)w zEvKg>5|+J32Y2Oec(%U&dqu@Z-+&)Cxx;i0DHu6Efw@4f$CdUMZ!@z|MD zwVSE;w7H543!Kg~jXo~d+*EtU>Ad~|CSw{Ojc3Z2!sZkT#M~Edz4^{N%4W@*cfIn@ zzkU9Xin((pm9D#Q<(lX4h%ak%eK(y~spedlfElYPo1;=)yw&IHvuUPJpKQS{ou8SN zr&Oe=^Cx7@nO{D=ybO_y^65pn^3sel zO44}|I={cQ&d{T4sjV66Z@*l9uCk&1fTAWeYW z)4vXdI$J*9vgOSt-S}=!XiE?`%RyU$xVaIu%@ua!i%fN=&(^uE4VM}k&Yh|{`p(}D z9l0%3`>&%3E8B5j#ldv4*=lyM{gS+^qL&tDBni zCgX_wE&8>8obGqC0UpUwSJz&=aHU(H1_^LN=kxU(g9I3jDiCkZp_PfUiJH>#iHdOI z`IRf?%$_%U<{a5FN{ePX|Hb{&9VIzz3J(xobYuC%iABl~^0PBJE(ClgX>%CiKqQLP5?E59!05U23Z945_3H)R4}>?ioqO;{t51PMaYRaM>YOEPZCO zRgX!Rd{DJWGlu5R)i8nTM|B*VZSP*4Zu|D;=J(#C06R8^VMA6`?ML36nNAghB!Dm` ziz5;(&DghZN%C>VfhxYDd-ZB(`SIh_w`KQO;yU52OQ`LPTkho+Tvq9q`OpjoWr1)t zs(1#2a*tXJ(QzV21qjEoHpAG!^1YS0n=_YTZiea`ca@f zo%LVQG`*|dh@==}{XuAEho}#=fOv9aF6f2s7*DZ(e{U>z)v8r_v7Y_39EAm;mBM=p z=jqrNdj+1igkXa4#h$|FPkCrLSU`)eny<@z5tc|HVnx;#8cv)rXx36$X#%LI!Zana zZ_?9ba3aalBqRz2tbzzm*N6{)_Mt|hmX}0LQNQ08c{!^YO_l~bnlEF3SDM>l`L;J- zhSGQi;7aIuz&4QW66Fd^CSNE9_hu~QH;v8iI7;p2@2r`;AEQ^maO@)C`5&X1-fqK+ z`7ydx_6?r^?En$`VcKyRE-%enF@aEP#ba0L#eZxCD8`?%;vuqZ@WpxX=2;8YLnytj z&GDtEry%t6tzBEl0trtTSems-hEPQ3{U{+NYT1tyQ_G6NX+h`x7)h@Tcbx)RbAF7h z9(QzH9%X_TNzJP)(DrK0D@t(s^u>#Ht*ygQ?|D$EP`-n#nAcu|^I3>q)Zb`%@WbA8 z(;Ab3Cfj8NPoa=P8c@H|^5Ax@tURMo&}fG#m8PYqM=3etYzWJ91RPWVJ$n|3Y@MC$ z$j{*sKS_cK4@t@=T13_`o}>s*@~xiJ+S*aujp&Y2)>l_w4@ADi{P{ zW+IbNE93yr1vuH8ZvyWlQLI{X``zo;g9WkD6|_`FrE?q?arywUAcGxapd-n){c(e-q!YN1p7~(>5aFv?B0#$ zYq@57Z)aQAAnFud`d3IWzfMJH|+IdQWadk$JpCFa?|S7*gpdS zo=}25-wE%i5mgvQy+XKiGi8CEKme7&MzMd!-445h#2MnG5hb|K`Fb@Ql431o|vGilOnM0%#=ioxyhPWtqnaFgns&WI4R zCr)?+q@xyJ#0>f9X-Of&jerJ^jqF+x2Qcn9MSSdRh8IbrF%DahhG;N57!+Vjk#Q)o zHVYLzff03+osN`pC5WVGBnWvPt&l}Ay?&3&i>|jANd@mnqv?-)=DXchxo4knN@W=+RxfoX(v)@o_bnfzu68&k78N zusky}jyboElF~S4#4;h=+)j(l6AU;EbYYK!o2h#LofF+j%(A)G=bJuVq4@MuEVoZF zuirFcUJ-sJ^V){AkxrWk``^Iol`9Q~g)Ga6!4>91@D)zZ0@UE~CvPl8#^;b?kwk*E z7=G}Fn49y60n`-6iUy`>3*zxn?5Qs7yfKPo0E@#Av=N3tA{H=0ZhxFE z1PCu280xp1tq_=&LD=c2`QFtf67lSS01VmSDk}&A`KQ;DEEk6r-eY|iQn^AQ#udAH z*b+(t5j^IbP-Epy)Qm-ON&e|X4LYhpf5N|%8YS#)IN#ic+_c)d_k>a;R&&H=smzQ* z-*Wln`uf;MAHDo?EEWvD@(T5;Ryw2EV7TwT=Jxg(iVPbO(^v0BLb~X3mNYL_KKmlj zc)xo>s8Q_Lp|SZ40tF?EW?omwo?BR0s<}}ZgKEJ{M1cCjXT)L#)Z6XpIduyCbvxVI z5Y&;H6QNLMW-t;lhd2~?_v${Qxw_A2fWU>WNbG4ufyw8l#v$m9afYzj)YQKxkebaH zg-m8b*&O+;J)e9+h7bt+yI5cS-I_tFTOr?{QR+R_I!oc z2_R_-QTxT0OokxuY{-ZPoTGIwSpwM8!8TsFjg541O>a~p+4A>2H6viF?&=dq_DRy; zfB(R?Prv$P%Nu|B>tFx68y%?%k@oJl+dF|@83-zoOF8exLW5z&3NANFZtn836m<4& z3$pm4-~Y#kTjwEJf6=;o9^R-FKL7mvk3I3w?^goX{SK15HjyvJQ_z$uw1qxj0&eJ1 zh)9jdi;5U^7p)4#sZ)A=AO)JFD7&0!UEhp0B=p;>gsqH zv+^)z<#Ei)!yMf0?RZxWjcqNY$TQ1CWXau0J19=f7-s0Pz`u*?! z@cc3)w7x<8MfsmEzWCz$#qbshr!2YeH@|!FR}WDyY182zjBw#%#D{S_Hk%F}>XSp- z3<6C;!I7A6$&!454_knZBgt7pm#1Ya+1S095{X!uSEy2m%GQ<5wuKTeEm?gF)MZq!frwmA<_-)Bz-WILP(eNSy-58Zg~Ec zQ*MXu%#r=OkWPK5u8*{9_x4=`sHhsHQ~`y~5c>5O55V8=vX5Y=>m9Pi#ERp`d)?@B zKAG?7Jx&W3%*n>2iiEB;w-S40Gxb{LA2)9N<*$DAt4D5LI1!s~>B4o$ioTQjwYJFT zbNKkm(uoQM9O?)@6&4p}@k73)OG~5?+&&^LSxT2A*-6X=n>X5Bd+0=WOu!T3CVWN; zsV`sw+!c(Y&Qwt><{b@j)v8EjXvjhq69W230{}`9d9_I>7~u$A(%I04$!a*&V-tP7 z`-}Z&A=av^FC+N~3SDzk1JL!2NWv+CTc7Rs4%USb{`c8HID_8kiD1z?UBu(=b2&%H zW+r9dXxOlAVBbL6qoyya?^aftOotBjp<6i5jCBQan`thG01OyRPTE|>;s2`PC{`Gj zD}W)==@N-GYo4W6X?ec0=guhc!%L74FX1^}f_!)h&oRHIzK^VEh+-s+jda!3TsVrwvKt67g8DjeuBHm{ z@>B4C4-K`{oc;Riug}yFRA3|ZrgomeFgyY@BFf3=`%DP=VfVJ1kYnI*x<}0@*$Zs_ z)$Be4N^4l{h(X{C%P2j+Zj4S z$(g6cS`@Ba1nbgg*xuYNO2?iYFsTJ%~{piQ>YAAuoT@q)9yQ z&;^|?AS^&malX*6qvy{rQ-ZvF5sybuM(JhqS3msRbI(2b?6Xfj^ZfJAKe}dq9(L%0 z`S&0f?Ed-1P-cqe--Xb`I)HlSq0S9z1~HlF{K{eT(jZtQs4>StSOzaD?3*{QKoUkU zC@j&;D=6T32eHBiy}SZ?K1z&$HN#L=z1ZRsP-9Ub3#thxfewH?CdubUsVb<#cp+gO z9|1;Q}X3S~}a2_ysU(^UUv;lmUvX2_u z`i2p2Xl-n`bne`x>S3pyV7c^d=jyJU!#;GTuA#Q-Sk;9~%`o&0UD$}GB7+g=_-;t) zW}q=@@wj>zLpFoOX6PRT)C~rcndFpWD}wPvPiq{aGSnY^u)lE_3zI7lrIBiKnViA% z7Y}`eo#@ND4l5^(7aeFmfixYm6-`Y~=R)kYL5x+|q|$V#2pKX_%rc}m4v+ZL^7GT{ z>IP%{62LujIF34c+6_~rN$NMxt)DY74Wganwb@9GU#?V@ef`g|4gF&I+*0J?3G?UO z4%u}Z_MxdB52eh?ng)#Ani3#2h*kFjv$FP4=|)W7x={N2_yr9WGe!*^mOam zdpqGM=;$C;B(P0@Z;s-pL0G1Juv&a^7GNbo??|^EG0HG12w|rJ2S+0z&s826-;Hj+ zx*OdBSyR&0--+$#%rUF(%ArHGgF{2aNJA_N(M3eD3b+K3SwSc1p9^)clPx~BG8D4ZQ%2`10>N2NmjUk zyrWARyIQ~;I?I)l?}Xy)un|3%p+~fK=pbr3I&|2JIXVhV?eMutJrPFS z?`I`E))8|cU^de7oSbwa?;jsw@oe4t51uRs8G=$i(~dGRHef~~q)vpCmb=}71oIaU zKK9sSzqtPw%!nK5HB!DPdqQz>j);jxwTtivjo|tIHpcx@o!JM%+Ahf5->m$gm?s#^o6qum`EA*Xs^MfXf~vj=BiP*ZH*oe$o57rVH3CD36gTa~6@djd8W#kQk% zX)_xi;GxEN;V15`Z_&5!WJle8+#A&z{&ARnVb&n-&Ga)Lx{JNCx0oQ$_-u}xZ3}wH zy-{eaneFy^ue~?P85e`k(65t|O5b2L{oj3N+r0JSYKkyI?E;}3xBjNDkvHq<~l7D9V628RPvyY=-MYuo8(2~9qV<^}_$jS8k zDdww;SD82VQ_E2EMnQedv?0*NVsMxaJH^o(sTe*L9qrt8Tf=eFiXNi z_rR2InS+@meMC6;9ISu*M=%oEr<=8K<>GJ`IpmQ;i+1xSk{SQQSG-S7sU(MWsD_$q3oF#`R$U#XC7k}Ui=C|_!N?xVz2JPm}c#qyN|2}{JIzQDq zrixtj@Ywkn^uM3~ule3fPEAb++QO0!n>d{R$tOyELTbLReIo7;aTGyjyP_yE0%#%2$l&aK2Ht`a=_64|D`WY?wVqjKah7NSo%rSAVB&LaU#oSjDwpI-YZ|0macU;d`1{u^@h zUy*~H9QIxFWZ%EyFgYb9u^%JvxL1GtmSg^1#{ZR^VkC#uS{c3i<4;_%gPig;IXp@Z zVsgOe{=^lSw5wO7o+^#}h}G~DSCDCk3mJ!$Y`l@Y`)O(V-?`%ZGm;YTpW)8Lxr+Pe zlEap3XYu>=#UyKCoZ;{iIb@RqNbwURNzKS9a^A1Wfk_Tu{NxoMkyC2OVHG({A%|=E z@%vBpCOM^&98%&nPTuYR$i_VH6p8;wpYg9NQfunk zryA!IE=9B;CmF(PIG~ygdCMhllYhQuBtLKs;a^mtA=hyJyVnqTa`kiA-@As?IU>Ev zNM_a5I!6>6x$~!ifBL=l)lKZ!8~;VUp?y`$TD_WDt$mffYSq%Eh#S0`dVA-cc>BKg z)iHTHj;|s0cSx?@Y>YFD!Gv80XAdmEB!drcCZ7?7`wArxwP~@KkP5-f3{!%b*PGN5 zGm)W1LxHTm2rS!h-x(OTr~15bmwEeY;F_y}Dg+#O#M%Q7NuM>G+_|&f%0q1mzO{ZQ zomDD@d7oCg7_EaAl`7z@QIsx3?$ms0tyY^Mkt$&zE2Uy;(IO2>!43_@rG<;=EH2tr z2T<-7rksS!gBgh$!?0&K2=uAo>;h6Yif%bfwpc8nNDMYW@x^qO)e0Yj&5AUq0h<-> zVt}aNI<{EQ z1x}n~@CFr82kOvib7CLXzgDvs2w1jP(RxojVu*A>8{2f9wcS6JEQf>XrR zbq)^2^UWsxWne~f5Y@N#gIBCM)_8mvXF)d%T{$K02>R1hY(3wBVk5?wMaC z#h|1`b8o%%);W{X)iWTwwOL3h2Hi5Hd}`UUVks(Y^hTt`%LHOoCcuFlnD-o^3@8bh zw+I&rg#vgP1wjFdsAOt$;PInCz9=e9puQz3!v#y9;_^^mM~sS3PLLCTY>;i#ISLQ_ zs59kHH7CLYRKHe9BIO7TA{3hd8U!vv+**s+eKf>mk9nDu;NDDdFYz)f!Es7(oB);K zN9JgN4>6LZ(cqv zP^qL+F`sXT=3K!7Wo1?A!?FMb6a+|{`x=(j?AdLceNAD>=`PnW_E)u*R8 zoC(*loXAWZMiJqUw;Z%dBo&n96=_+-A_E!Opm7rA#kJ1;pOLkzH8U!+`c{rgM4C) zn-1q=PaqE!{O78jt5#iqJ=WkR_2OJ<5J(iKGs~K>poXe1o*QzJh*Ycx73MGvh?F6< z+<|a{KS`H|qa$_AC3)P09WKjG5*;(a6KU~zGO`-4wCLU83TM*1 zNyGxgX&G#^-VvwE;E0yVl;{dYaE+8yTgGBJ;Rv)qqOnn_6K&(!Jc>*F$>M?n6%W&i zyk15eX{$`6Y0%~J5T_tJGcQ{VFBU&9LoUxCjdas;W=$x}&d)_Dc?HBeTI)cnvn5oN zxlcX!QF#IvN{?&2V*DM&Tn!kHvCVu3I-1Go6t&c?yC%a^ z3P#m=s=UJ73{H$L>M;bl;*Z`wtcxSi;%T@vJUkhI-yJ*t@sF+B5NJHq1Sg#aT{=WN*1==c_^EhpFfg0XW^nUweaO4zD&Z52^AS)7T6{~mydvO zRv=JXnw_1Vg=tpuAl}p2Y^&qBg)Ne(BeILsk4r>J>W|D%#(i3IVnlw?SCOf+b zjU=$GO`4iLb7nM(Nf+OB*G&^3OcW|Q9ci2)o6U|1KxDQb&h8Fm{%QrxfmTp3#9&{g zeKA(t1gy9Tp#69SO8M}xgu_vEcT1%r2vRni3TPQ~%e4?Rg;S=?rDjimMJ|z)4Gjr) z?fUp*4u{D^=Gh)?TG8d!*2Rmf>*|USl!Vw7==CFIWxRRwR;|*)iFwma*bmao1^M~S z%{Z8;2;|OYc`_+N>jpl*tPE`jP+AiRYep!k37|Zz1Q3yUN03S(*hi|=Gua`=pfgs8@K9Z~BNG3N+C6Zsn zzxrxy9N0YeXvJh4S4UwsZ!N+c^*U<_h1hSUf`=mfJ_noM?gB$L-Z z|JA4WdN%`28fNF`>rvroXb3ek$DW>>dUnCpXCIGpti?Qh8*>MOy6(z_3++}9@%P8t zPG73osn_p6ak#b%PEW)Q91dzwt4Jn=g)?S=kpe(ALv%UMZqrKmARwu|Bozxes2W<* zHW*-WKK|%%+bALm=AP=)l|uIJ-Cuw9>HdQUzuJy8xUZ{WixwfhFB}MTAaL965KWjc zA)P(i+B#)QW1|jHn|`wujkV~K>(oJ`ebM~sQ}V&%rPp6qJ||*ay*dl)=9b?+{|IW5 ztzAeh))q0DveFqd%4W=%QIsu*4%B-Ri@{_%bH;$Iw;WBLlFOt^1VTP4cF3fB(&-|D zBMpc6sx3~1%nYmmd+@cJCVxi4>MK3wW7F?9@>sC0B9MW3qd4i zBp9&uwjv;Nsn(E4IO2VMs*DU#%V(dTf-=zF0hbJtq+F<&6~ZPG_aRx=oxTkB3Wc71 zb`d0J-0P3TQAj?;QNI93J&4uu3(U%c;HU?|QB&P*4TGSyq2J~AIxK#4sXz+KAWC(U z;FHblhVR;oOjVNwiWoJd6~|&3Q#5gh7lM$DV!mluvJ|4VSebxjxMO}lI-XRZoSMh( zr>C|w>M5Myu5YQ{{qCC)rlY-b-`<^{e6jPh58pwj$W{Y3v%V{p+y1$$s-F5vdmZLJ zEz(jRn7CG&W*!ENek!9pEGQY zWEJ0Z)5@85Kl0Ems76$z;z1u^DzmP;VacjV)CBBAf{eTz2~iKEQ3UQl)tZ3;J2nE5 zLJdkxRf`w|gB0T^rqWZ>3UUaLUXUddDGMg#D#fzQSdzn%7ix+iAoJ5ykk8o)36~K^ zyfIgM9h>bLv9aVr5F!Z4Q05}A;miqb~V&w>mtKzDHl z7kPGJi^a)lKVN4d%4`bGnTYd6>?BnA5Ft!77-glSVg`c)j^|?gTf1h>4Okv>c9@5y zEFfhx5+a3!FS$C~t1#O)VYXLcw!f`ECq}jhESZ=zL!6QB6HAe0qUOs0;*s&8ba|Lm zh8CM9iw_`xUyH~g`IA`4#V!hvHV7vX@B%IS^ZnO8GifAIZ+YMMzTcgb>^bMmu4}Ko z_TFo)y;ehM`iD3k7L=CO-hPgx*JUlrS6^jRwqYc;m7n@bhUT<^Ixx3>@c4$Nn5!>? zk?+Ydp_olVW5!ONJb5geWy6ub`W8z_FeH~#r;dlndE9U>=jhRK11*4qC4AUuIjR97 z`Nl@Ch=Q!tt-E225rD00X~jr_U>D1`238FV?-Go7n@7#J1spwEguZEOZ-t>ztVe(& zjCN3!F^pb!PcIR_3L$SJ1NEal*MBNrdgr*@+)0E>O>a!N|! z^~3~6CfMTDHo!43A7=r8d;lOH0LTYmoWZFQGs52q+2|^VpE#@NB}w4+r5HU&jIdaC z?tBXjCaZQ-TwGe(h!kVlkt3z0At9H*^`KRI;q2J~1J0(1p;1vb4Q4FNCsq;XpKPo- z#fa0ffPuvy_-7R>&YiQdMs@Y>)@XSGHR#Z2sUxkdM&=U{0p~>Z7}gE67&!yG^A%Wf z)L=3Ou$t<~(m^9X6Wv$`LkDNivl81x*zapsbVi-V(Nxc^>}w#ks%vtPwSA?%6+81* zCuwgePC8X7Y4uo;>9vsjO|EDq5)l-FBlPH3kLHT315ktB_T6{i6Pbpc*+gJ{eKkz}RFqV}G+ZXA0@lv-1l5M>`c`G*# z(l+kdQ_`RVV(A)6_Q;dN0u(f(3ED6WZB+ciVkTd9*=3WaPMwl8fByU_F<}x{xF3Uq zqm#nSph8f-pEhmU2sY7X8?W~?8B91DYl0K)5FmYszrU}^49PK+r}Qi~c3_|$fP_b^ z*pLuEE%)T9^$(FJL!S}v>tP723Xr62EV(8ydU%jKDpUn_eh#2^xmDSsV zu?v<2SVo7cuz^&zTcNb07+}=WhI7A?7MumBKjAGxPf0>vd=71sRMnRWXTs$japVr0ESAkUZX&Jrv)ctiMnW& zb}pNM@kobaG=i$2x{8w-y}aNX0*wOM_fqkk^jU5No|C@+ZNPJ`&%)2jmK2m%HZ<44 z+Jg?qVeqLtS7i>;(703TpoayK?|ESM**eT9Bih@c$=6ObYXCKb=Sw4`^d1@}X(~=+ z3;x~^9>P#i0N$Cf=8;F8Ds1-){_4}6X=P2IwH{hgZvlpBO49O=e}h@&`)qh(@nGk3 zGvH-?{Siznw&5-4lSAMW{nP+MXIMFVY1*nXPo|&RyW@CMSP0v@H?3NU1nzk>7w(lu z`AO$1s%K0ZVw7MXH>c+K?tNK+WJP7IX~=j?ESC&|?qq3EezQ4p{P^)Pm{~@&wV9)a z3>gWY$TbV6Pag%c>~GeaETK`c5niI;w4Q$6M!%?m zVZl1Mcma^wJ^Wgl%$Qw#btb>Cfss+c-a0^rrk4@ubL?M53?Y*#G}IG44nri3ZO*n< zzaaxL9Yc`?=l5y})=dsrnMH3ZKse?Hx0$})jWH8O5Z^|Nv#GEQF9=Nyuxs1kkhD(n z7UIAZgA!VCNW2YwRF?Mr_djE|UG325J+w|oyGE;=tsyI2E!EI1sBLU%cXr?Fbh9#R z@{ho@;vcGGCP9ia>d9n+(@fiBo!K`gCdQ8@8BuI&YH4csumx6D*5Is3ITTZ2`>Cm} z4%;CdKsU(Ib>$hDSPI(T(`(Akn_(LOy{xUvh5B!9odGKY1Ev{0_NV#h*qB>wfv34a z8<;uWshC%;y6egauZ0T+)HXMuGd%|m!*n%lAed3$?aZ?2 z2M!z-1F9Tj3xt_h@coL<)>>dwrCuS8HZQT^oE+U$Pc}>K$2|Sw#|>)*n-A@RZVJW~ zRKL)=CYoo^T{^Mu#1>oNm@xyeF!a)DHJCl&;v00zpoPw1bU}3~=9a?T3kSEp`OYux z7tSNs%8D9n?Q1GZp>|e6GY#x0ofjlsvki?70MH`CLN(5Y2F$`(cVHH7aLw|5k=^wm z{2~RHK7m#P$vzewn6E@d(+C|TrV0Al`(%OU$H09M?4KQ87aOdJsg0{ZYBy50Y89o3^vd_tr zY3Zl@gFqw(2Kb*kc@lgxZ%lp|OYpF4@PZEmCdtsD4vHK+I2r^X6yFiPU>kb3H#WoW zy0fKS<7LK#ZNM2%i`i%hivebZMTQx?ED*1HX|yH}qcPAI*aEUEsq=WxR&lMQ*7IaR z8pj0dDuAd}&D5&)GJBom1MNFzFiAD%pd%FSOXtCY!*^}}3N8C)7q!Y#U)g5#fO){Q za*fg8X_0+O3NwLmg(WeAp`|rsNDQ?~Fq+^%2g;Xl%Yp5%Ap5i#yoe&VwF|_*)wfft zq~L*G?RAZk#9BbO%xJqoN0wQcA&h%`3kpN4yp2#7(|G$Cjb?9yM$*ETK3^euxv(HA z5?s+p>}@&Lg7qRvUw!-O-xHK;HElIe`_X7(GPY!CgiWs$0Bz>RV^tA9_m*d$efAc} z6xP~2nwkPe%!NwT+z|md3L_YG$p#&ayg&yP{5cPp5WzWEKR+*%9*!N%g_T(ARl*=m zW@Zj7ox+HjUf{>35UoqS$fJV=IPL`;hjtFrZrJ+W*O1h*8Tga6p%eF}LzE^p)Ydl0 zn)5z8->FGBg=^sY3X$4`osw| z`q*Iz+P%ogVHH`oq^jWv5j-Mj$@z(H>{zJs!oaI&(;fWi<1coXYi!#0->=L%aY(jl z{l_FFCEZ|4!YK({;x%U8cHNJ_aBWbuP5Y-m4J`lhN7<5_o1KNRkp}bJ85#LSXsZ_; z&I<^j4I}@Ojk|Y$_d1Su+jj3xQtqHjeBpbvaJ0>9@suf3CWPax^v^&0PhY&))iNzw zrbo;4Xc=E+p=CBmxl%}fhmDYJkV3ISd~rE-Mi)8X#i?2-)kZbdY&BINy~^u1)k%s1 z2?v!0kp`8;wSVgRO;t##4ydW%bAnRo=@sh0$FsKAc9l|;CN-6pnyQRmK^)sps&b05 zLroQ=rt+p&t$m~t8NraNSZz^L!Pz0z5YR|*apHTgUn6C^UQIJVO~aKo(6oN)_m8Er zQ(4>8vSzAf`O~Xo{gzcgQNB=9#j2^esz7tUsRAffp_*!znyQ*!L1U_)c4bkNBWfzx zZ6^qFXrRD;Qk77ulWMAoYN~pARori?|H*bKRGk8~hQVqLYw1;A_@rfoPh7PmS4|b7 zros^Eu8DW?m~>JUd$+Q9DyT*D1H&IF)wgP@7PYKsdIhDce0Q)_R;qf;A6TiPNB@Q{2To&t1nr~M13Q98kU}FK zF)1;Nx~J(q_2h1h&zdz<%cWF)u(0Pj6D58CSCT?mXRfq&&OT@m}O~wX2So;a*a6jfX*CnOL~Q`rP} zZvEEp7FYeoSreQzT31*Z>bD9xe(b8>eCRMz{T5QrDiqy0Qm+scp;?VyDNj)DM=Nio z9Csur(~x61&ru9|lqImgs!;TXo)>ce0l2jAk(Y`o;((iw*PA>qaVTN-DvF~|U{));61Q4tfOqYSj&{07NCL%IM~J&G_y~E! z)wfGwOO4=m8(ZS8pF6*9BZS#u4-oa5Liue=P-dZTt9g6bOt^=&Ca^En-t|vlZ{cYx z<$CNcjOHE~W9m6Z;O$KsJAj0hN3)o1D>eqK((gPG)mrA zTVDGo_5kIU{V2P>x0e4au*VD7Z2{$?GaW`?w~pYzTY*7v z_%>ptRfvBKz_cdN!lZDeRdagjBhVU#D&67Ru7iiqRUFH|M!s8+??=e@E#&(J@*Vi~ zH=n_I#Jit<{qxQqtgw2|tKL2_nP9VLr>2_s;G7%0GV_nxsJWo)@Dbtio zUDpg{Jl>2|E^}R@@ia+^aa~imljJ9rCsEfDn;xc`T1{CMj$}_qMV1LC#EOfnOj(ZP z7)O?=s#vS_2#a$VLn9+2LyeBOFb}xg^@wxuXGffeW>x67N8nMZurUAdx38yQrzEd< zyf$VpwPRn0 zkhMwXZQIhB-3uBkUUrtgZPmbvjGbT~ZrGJs(H;Vh&(M&ziZcY{GZhZ-^a7kfdEIk& z;`HJTcU_L9#mnpy_7gkdx^}aV@g`Zi51&h6g6U=UA^X8~eX0!)zx{S}wf?>L*6%)g zG^Z|ESC@12XzG{m*@nYP#x^K%e0TI{36#TphmISUbmLuj-F0KqxN(Dhw0GSFFHQrn zyA2yR?((}<4I4j{1d)RW4I4je9?WOYzAS0{f?4=193M%noB_-3y&bk%@4n-{WeZsf zPW!BIU8`X@aT$97@6_u>k7c)FqkqpW*I&<7HNUj!!@|M?%|1E>4xn{D=0Ng38*BDy zN{(srF~>ros;#lQrj&bZF0H9|@VHSp@|uZEOm{#Dmrz2GW% zDPMfm6<05Sj^&jLuD$^#e6P9Y78scLZ~VXDx?X{Q)mP!@_*I~}dc6YENA%`p*G0d* z$zF3`nui~L>M0CzASe`bAaOG2lA+S$1cwxrsmF7&C;XJuLypq>uNwZVs*?)Op-jLT zF6d9e;3KpNX?^sougmooYRLpNyVi20`Jw!U-7^j12A8`ba?WD zCKJCD%d9Q~bw}R=b>?pxsQnCNP3W(Vka3Z zcymM%)XRAYPdjHj*SM}9x{!k9|2!!e_&-Q1?(Rt|n4SFs-`XF$uFY`UZD;j@w=mFk zSy&z39B12H*Ks`62%hT28Z?~NB#=1aVrnz<_fVVOsFBCNi5%%%Ps~JbJz6= zTaP#U0g?l*>oD7gH^=Z!y-wii!wyHyA81?z4N{rD2-?iwM9}&nc$(ix@tyH7P>aQ*9^aGv{TCmiNr)vSzFxvp|{ zjyVJa=4+$tGGmG95Pa}Xz5I}(gq?L=MXm`;81-+QuwwXx)z6Ifzhy$Zg_Q%^<*dMU z5!}mJ5&qA+t}<54-;^@-qTkN3O84cS5hwq9X2jnn%H4Y5C-8w>rGsrjuP0)1m`i7aEGe|sbw>qA@Kag$#-M-_N6=b ziUejw?yKCnZ%%?5Md=OJ|Nerm|y_E1r=T9i%`feq}bt@s)UBa9Mb_6Ay zcb9N-4)nacl%PWi7oFw?f05)opCL$4kLQo1=ix574_Wn$08}FYRk!_T!UrH@U6UrtaS5>90$DD!|QciW(Gq$u<+OHZ)s`_Y5Kpx6Yq7uC!p z1JA+1^e;n2BeS5QJoNyx*!0!a0(79EeQvBO^$Qs`8$d+tZ4fH_di2^qNs&pmFxfZq}vnNcRKHL-nUK;o}37sf)70pwT&us|1;0? z)it$M71igetIl7@_G}kKXyB>;sB8!vVYVAXhC#qK%+L(Yv}QvH)V)G{1SmHMK2cDP zje-yDL7O-4I(!B?7iSLd+T7Q@Iu;4i7hxLNPEk!V&>+U`UtEDI+i<7peztoH#koKM zmw6wjI~)n;kR}|ss6P{w2N4H!i-&Ztnu1tlOX*%tcPt2`h*TI6Fx*zUg3dHl5=rx* z(t)-S=0w_3#AdSeB6zX{!9a$1kIAfcMYz z3<+&#{eCaJ2SloP8iO_kM~W<@E1rr_z7cT{))P8qf~Y%i`oNLzcE2lRWXu9X`F32l zSZ5(ztHddSHfmn^kC{11$^zv%WxnT};vEt6zvI(ScQE0JMTzWFc1Y`UZq{A#aY@(` z!UntN0pBsD^e;w7i%?zT#YozgiiR3vp^ZIB7`)cnS$tb{V~wG;LRVPmWio-&&3m8l z<)Yq)ER}lU848652u6D-=qkm&cr@>zDEla=>bSelg<~X@aZ9sE;zsk9WY0gHB||kH zr+}o9)3+XS-@p0!jgc}7v#~H)vKPQwF4Sr;7fKeNkU-bHT6cG@ZVOC2*Nq#$ST9RN zOl8SG>ea7hiD;*cW5w`Z%AS-ZqL#AMo_)qCOGG1OsU&uGCEkECs^hNLe82wY^UvM+ zY6iUYt~=im{)u2!eBjP^$h?K|?tEvKYUgy#HwaJwcantxSNs75e?UQ40Z%;7XslmS z_IlSGuuT?rpGs5`l|M+$173Jp*1O0*nsNIBiC7+G3X@UnCib1IcM<<)S#KIO$!o~G ztA@&Y7wNB-^*0V5vox%vxIDQ%agJ z`nYiPec{+JxReov$&-bsfkNC+p`wc8O(@cjkJR8FHu}((;gNzT)bs>slEP#s2(?E} zZca{KK_1RlE#4J~dEngC3i_!}h&coDK161}@T+^eyq$T3kL?CSduId#i6BS}Hq)YI%s#*Z`svNu> z0&6HPXw%7cI!6)=Xm^}khelmKOv_;^z`8~O_OCQTOS(e1iWuy`#8!`41+~WG zhj6?TGklOpCc`dhbg~h`n1Uj^9S2sBUcks$2_we7K5);76SwYH%&GyNsiVA8Q_TwE zDr0G}7$Ls}6_)>jm|bTXS4JOo&9alm^rRQA*mjCNu4@u2Q3rQyD?3ELU2g*%-R}uw ztq_Nz{+Ro^&y?Nk(@beQ0|Px$Kyv5aRz3j8;}j}>#;I@{6ZxFbl*u4aTBJ0uO zeqx@a=sXR;t-@v|Fyp~U4gMvZ(iibB41{$7`f#0WU@5G+(a6M~L|=k9z@C+VCO<0^ z9-+zlA9o6Uh72szS;&lxt)!~wNum!N2p2GTMa5=v;$D()R2NF%apT6Fy!FN#Pn@{q z{V%?N+;Qu+pAOLH=RG?&e~nXL8^77~eG!&1~!dnvFednavG5&!E{%gF$eCi#U#jRgIde`>tQ>Xqo zG(P^?Yj3~(=EaL}yyA*$Zd`gB4AVaF;QhDXJPYUhC(c^>yr7s%(MaQoOQ3~ai1-*qY{x1NU)WpH7iohan?ab`-oOo|?|nYsgc&(OxrahV z_Y6vX*i^b^XLpUeX*!ALVUG!Ac)?K*%BVmYr|EeUc1HMWgkGImV%PN}t?Jw}$rQWb z%>*S!Sipn>fx=!Uq`fcH0AGrQG8Om3&{mp5CPqq_6Gi+B79_z(r_<9H%tULvVO|>! ztMVYE%rCI#=ly{7LJCuonUKte?BBoda9T!A0nDcqWS>eqb^xx9cJI#3h1JlLhttwe z=jI+dbodxd>BIWot31&YyZh`@-Y4kHr^<0erkIKEBbN zKdJ4U9Ld-XOv9EiiA_U0rvcM^eK4ugXck~jlfJMQ@qyjb9Qc`l_|>k_G{YlMs?8f# zAv`q!?NU`+BRn>>fs^>vJZGvi6Qac1l;z6J%2mpAWw;V)o3BLTdpe$P#{0XK$%^Dm zbN*Ouc@VJSy zX3c`{r2}bc4pCAZttaIOw5F)I=-93wFkw;C?4(EV!>$9Ti;9Xt4U>c0+S}?bV0I_x z8gxp^>}#_4cz~kF5#AOrzqCT%@na?>499^5e1<1X8Z$1UH7m^zb5?S&^rsIXVq z7xoK#_`L$K{nLv4dV{?`hke`khX8xOKpbPFeygvl?hbq6dV~V+qk#A1%On_hZ_2V~ zWx)7+dIpL&z(x>AFXkm#axlkd3CaMFsR8UWnSNQgXV28Bn*p6xTTtoEIEcz7Ca}#bT-7!WWm)fqguGXF>#b$7H`TqovK;HU zqH?uJY5+NG6<}dktH|Eeir~;-EbIJ&EfzGz+#|;Sm+f(n+Z(#mOqFE1X(m=gKKA0` zOn96kva)mM@#128suh;Tn(G_es-*S+ji*VYv02%*^CA5h!P?mYP`lOaBlb1>i9u7y zHjhD12=`y(-D>tOD-?zb3GA8`R`wNfW>@8Lfw{xm6o@1jqv=6xyrpzb-GiTkC+k4lPRE> z!6#c`C0lghWDsa2DCWet;ZjSZq8SNNni65p$<3{3k;J?_Y88F*@&iO@8PVOBKwS#2hv*H944 z6qAeWX7S8iWLK9$uE{JGjSgpLBjG_%rN%myGe|UYFlgk6t~8R!RVh%i5ag~8LAF(*utlPpCI#>shoQ*LQgN1#J@@z6Yl%OmC;a|WG z5p5T<7}eOS@1&AMma}LafGlU`QK3YZVJ5GGN}6;Av;hY^4RAi~#i?ZX7#&oy&lV8@ zE-&hue69Jn#|n(T4bgcnfq!HSr}Z5 zKKdj2XcF)S8!{FFGxzwwMO<#?i6cio`wD(t_WX$Z!(gXCAv+9fWZdo9zUjn~eW|HM z<$L$0Zr$|B-#_1iA7q%kYiyk7{M30+iB_IaUQq5+{s8^b@yZC(;i%*_FTM8W-~PH9 z^zB)?tf+ai=kn#tAG!OFH!Z#6zaBp|d^S7C>FFD6t#J3BW`Rr9AdC~PNU_br>z^1j z^)z??;elmvZ}b5>!-|E#EeR|||EERcCf;@DLr;^_BV1N&kVpoUr2LsK-Scm}e&Eo0 zp!A=G%gU3KP^iQ%)vjCp(mQXgOj4GC4o{Qi5J9QRDQbj)FmZ4cXq$&uQ!9f-FCC0; zCc__&!>KjFjfyDdXW^`|!DC=dxVI>z*uwVh`*3?*sDIqp2uZ3eg{3loOZ1>nZy4P{ z>u^{!n6{gFd9~-7G*Wnwv%TneW>qPRTXgdbm6t$$a25LCdh`KV*rWA*#1YuzKX!2U zuI!9shYnTO!eQG_yT18iJ2*T?j=1_@_mi4TseAX97p11|J91*v_C4qq^nuqQ zoNSyqY0U7!aib>8c&{Omy~5UNmfwE;C6^~*l4=%0G=~lyLhWp`ge!z`tbuJ3{&e?j ztlHn;GuuJW;in$|uRE6B^vAm&LFcUS3VZFf*WO#d<-6jtgCD$pr#1-+j{vRi^N%(5-|)gp9SsWX#nRk?aDXad1>{!BxNZ;9V(Px2GHwL)*m|X`Wu+zc&Dt0 zfXT!1igUG1_TtJ57k>E}7DgJHJ$$9})ldrt&}~72=p7z2(8GX@c2++8b~M+SV1>Zp zSP_v5r)C>JdFS14)^FLfJ1b+?t`A>)ZT%;|=ptHMQ_-%F88dEP6vs-dGLILvJA=X{ zP1CvBygb_YyXGku%on?0P9B~kIB<__|KginKOH=Jw6^-tp<@}@yLKNuhQ0vIlLL@k zF?z#ipTLnoGI@tbg?O%%0L(*4LI`Vysr$<>x&HR$#5;e*yWzbV6Gp`i9zJH$O!Nmw z_0_9azxd|+AAj)ntLuiJnhh|{gue4V$`itWubQP?p?E;(v?)ah3*#Wwz54d%A2xja zIROkH_txI|`UeNgioe^k{yp>up*Y)%upHPfwljpM-)##<;($TX zUcN?WdvS#*hL0W^5W+Gxe6-uWmvYe>vVsV21}dJDGtk3uxc1|o!^e+*_1TdlxUalW zz$p-5E(dsqM1^~sU~>W(0EZhj-bOOfxPmphNQlX{`mcX`^R<`OBu5>dr;GrhxeNs7 zJ`kQKlxXKc=cnYijH9}96SPG=lq-Z)|2-Q&-UG_!8out;w?Fv!{Wo7k=d8e9 zfm0Cu(+}Nw*Tiv){sihGYza()3c(rnf#%U=5B%ZoSzsWpB5DK~1IW@|T&vj{?VsR%bYE#FyB38$os04A{EH1TR3ApoK`(aYUd&Os7ac3{IbBxb%f%DyhfLao!zloz9fx}Nbkbx=vcX@k6>x>tzB02oh*=ec|5I`x^2Vxf}Srddn`IT zSR;IU9?+FF(z(n}4(hfN?c|t<^7x^aA@N;z?R@=@btp`4HP?4rqjqpo;<326q=E(c zx%TtKOVOBJ>r+2?pRc|&VVEyG&sEc&!P!uK9=@7}C0ttVq6`1Q%+>$FOmzeL{}^VD z#EdlsG;JG-u>nVZ*Ccf&2)koUL^x*jL-ONY2#dr+{m5cQwoSry~ zZBR~w3ABMtl>OAbO%FCu$9_PtGU8!%VF?Kn56)7Sz_rHlBEaAA@ z{CUDeVIa%GIVmf9n|%tEk0Vg_n$I%vy&cbQWeyy>PnU`wvSFRSn^3rnDH<7E@A2J0MUu(A-#sRxv&l}CgZ zUP@AKQi8xrNCGP%NV!RP>4mwfjFs5}#2n1I`jGIa9nU2gcBY#iCvfysENQPNS z!@@m=OolvujH?Cju+) z&@95z1AX>7TWP#`tdC(}?AR%+4c0POuZD0m+%``MSB607wE%tcICia}&Qs1^nlI4T zU(9pvavp=YcRX~h9>>oM5N(KT8@PCJdHKT+KYaW3H;jsn9zL4g7_TfIGI;o?k%JYb zzML(#K{*%lXmrjsw#C9V!Ys&*53-|BJI-V0p!jXZJ|UT1gRzz^m?7+RLNliTzaB*_ zMVKW=5pl@RhrLvL=EGrJzIaiBpWuwgMVIMO<9`jkOK{AvoO%ot+vfAH?BCJ!GO=pPot z7OJ@jwqQXBA2Td^uqJ(PM6eJNvG?#=U@#G|(<1Dt0d9Nmy&7{P8FIwXGds3A6OJ+{&1^|t6CqejOM8G0idu|yaXKWOmi z#W&mmJbCzz8y1fqJSct)2IJ3sSy ziEIoTJv1s-O4}12CWMdbLzXUXfX8At%EokVRx=xQ!4qJW7Hp>cMqLoupY6FnRpsk_$awt% zlw;p~^UX0Opsp@p=#-f=XHFR!(5KAZ3W=gBc3WXjXJU+hY&uneL`W`E)>dlodSG+D3l_5zUfp2_!>!Ywjy_A-|B!i0N-u) zl;{)g*nm6Fx^QPH;Z82J=uYk1zMTe0ZeCFt#&KB@6i4#%bI3nx9wa;XY~TJ1;^pQ7 zcdW3suCh65+6YrL8c|V_{eeE@Vg&x}+hOEUWr%222<=zShsskN_5gdaAX@>>^?Dr2 zJi&7Pjuy1R@Mv4X(UT{S?s{+S+O_ZPIy#T#;i7MLoc9T8s_~DnlbM474 zr15#<=e&j(ELA$#8W2dXOc_-KtT@}I(-st9ZB>|i26p?-f4<`0bk#OknE41Z9|_j0(hs)(%aAbs(C|o=WDBqn*IEIeT~WCP-KMV z@bFNwz8Mkf=hs|RReR2oaFw^8ACSiC+f9f`h~jAFxjR~&s!eNEs@Wl3(*kE;85vin z4kU5iU{(%FJkZs|tqn^aRqpkl*#8?~BKFdk=)_)o^7q93ci@N~_}5jdPt_iVX{6yC z_gEFapz!X2FPMJ!QC~KwYwSML`l-=K0U2o<)U@Z+mmdE_TG!gX&vN6`BD&YvSAE&2 zmg|f5`eL*Np^tnq@_a#~_4YG4_=9;&Em`o_=w-eCkLA8Q%3*s_J1p(f6GLy5^%)OD zXWko)H^O)Ga1c(Iw>d$1MAqBOit&dtMcQ^rFSWJEdP@*o2>C#{=O{6ZQH2#Jo0kXJTkVadCYzgd8AMtg~e(f zzW%?GN3NYFz?GUGYk!38)s@oi!M7ui=kcb{AgHGkq$p_MZr*nT6jt@C`r}E0soMchgpp!+wB}s`=GT2%e62mfU zS%MP8mdPUEl)&1>+0S!mC9~uNXelNGDT)iL3eTX?$)dx<0hI9f4f8dk@#?c$dn%cH z#t|lVmYkP!s*E}VF&!Q#Ib&O*5(N{o)G^7z>eUIX0_Dzrn96-#?WJVl>8D|r3dI%{ zSDm5$LbY`bDAw0M%t-yFKJ)&kbD;hxmHbA~IZ$Y|GCkm4SeQ=hR-w_9Lg1e;Awkht z0l?%$Wi&gQqCr;_7bgfYR_c%BM0QjeouWstUzWpDK^?B5~>v#T=DI^j#ABJ2)PHSBu3v?M0=h zOQ~Sd#2V33i(jBVkV@vyyrl%+VUZdZ6RE-%R?7`~rkA+(x*i6GFVvo|v0EXJ#PW*y z(~At)eb2aH9655NG4iMHMnDNG**tYxkWvh962(f;v@>T`Fxd-!R!kb!2H9VmePBe1 zoNV^+^o1>M*ry8e_4M#qfpd?|&2SNN!OGR*lXY6P)MU{qDEt++#prVlYN$aUXy}9s z3lQw7HvVskgi{V+B>DC8M++=y0W|ONFL;@mKk+bJf07v+`I6F-VtY<@R(4LR)zbkV zks7T|iwd^FfVn@cYXf#xE3*n#MmHKM0Pm#ob-|BIk0V45Jgu4!a!}K%`7pD^;^|@X z@U-{`1P1s=_V~4zR%kD31=&?Ip%o^yg6#6arlAIF9q?i7rCF&~(E-(LK`RQ&Rtu|_ z@w?=wg{PAxdsZoa)xiV3#S|FH?>JN@BER8GgMSHpY7zg?j_3O?HORB)7J-uipcKKn zGCtXuc~?R;+Z&6#)^>?DVp@&V-s+Io_&8a+G;Z0lWlxTi+K)g;`}`4^N2R+izdX;Q zUb^G;ls4<2%0o*zrVm8`d9944`V$ppZ7GGH7Fn8Y%{ z&lj!{xEeQ|&rts~zPx5G2$gX7K)|lZ9;gO~464MKbbQjJ9W++ z%P3VBV5ndc_U=tou2rt(2`>VdS|=BvvN@PgCqagHL>&hZ`i%K@1&ap>qxZ-772-vG-apOq=#U0}G-sH1=ra*hy*l7-evo+E|Zs zu&V^$nC^6k)>lM#0)b6*(gSNk&=YVpwEO@-Y6|G6BxrH`N7OgMM-2SYRIXz1OMs8X zb>S8h&nvD3P~@{%BGo+%t-wh4IuvzsufU9im*w_MVh^R@L_3Z_QthBoq|_@5M5<;A zjFN)*FP0@LCP}#s1a!A(K_MyJhynXX=DB1dAD&Ay3p%}$3}@gyRUELDvQwDO6d_qp{Diqch<`Uiu!x`cYXb>_0PMXQF~c@ zi9M7ezQ7(y!QK5S;)}S`abSI5Nno;$8s+=3FK)f~+xINbb-uEqm9qE%docyU3l!k) ze)dX=TIXzg39oZ6-@MMy3Q2+c->yLv0mkBCAy=2Zegj}~k2L&I9b@wWzoo#C2Y?~- z0l)c7h#ACu_^gDpVDwEN{?+Y%5Dpa<7J#dh2gVNPpwThI?)RxyBX;7HSb+~Jqo61P z46)(MX6n3W?5~00^Kd$5UN|e>`u5vzZ!I1*>cols*r;6DBPupOmp=`{lN2>5KUaRy z`^~qu6tgg%B#aerc}w=d`=~J|PK=3-%FdTPV)L{46Vl`ZO{hE_)&_JUD>+zUbveh6 zAIqs@p3$R6kB;`NuJ-dY*VUM1kD5BOnV#S>*ISklc%!h9qoU2ME*CBWa_g8mdX(&e z_jQO>U5Ds+GMimbh&~7`MS=&142K4toy1L-djJK@9{@xFU~EPX4-O5Rsq>xwRROY507-2ynd?^oPz} zEdpE@f&L5w)aWmtn8{#a74hfd;*!d9RaivWi|tvbGqcWQk@Qy9(2x+50|liW&T|9? z5tXCGA;EW*L8I3RHc@f5Vc#p*BmrkFlO^Ih)1&XT_x<_drE{b`>)E2oExSH@CFSKW zWOr_QcWxlFhV{E!$1k|=>HDuwV4oPSfAFb$7ENP&*Y}pE@Y!FW9N_utw+E$}i`i#C zI3{0f-&2 z$3-iD+JVJ^dPIVHxOTXD9m83Cl7?W9Tx!SqU)!pP&~IUDlEA@UCQJ}7Wm}V+KP#)D zIQ=|JNpLoaXYRTyLAXr3>#j3+lA#Xy&2ei18<+qhJ$8}&yk^`3J{cL%0+uo|d>)W> z;AUjmOOdlyfE`n&vv_)$EZ9M~p4C1Kh*bCev@Z>ag#W}y0os#PHQ27VXC6yS12g_> zn`B^+FOvzP*~+!f&z;%IG3VzAY(4f=>OVUPak?-b=FS2Hy=vp0?A$NJd2V|-oa2G% ze8MI*GpGz;Rl~3riHe8-uXs%7OAhxyz-l63H3_hC+2Jwage}d2blTcTRmx32lb!js z5mf@dw`mN_1A8foK~t<~o#(D0j{e*M4S==u?5-!irA41DS_FgI)^5odtO&Upzz(r5 zgg;DA#B$K1h(#c4qY0+47(9ajrM(UA z4ts#^k~AAU1h$C~L?xRA%p-UMDO6jVhRsT1R14}EsBHjG zG0*UN(*vHU?{LfUlVVYw={4Pf2|y`Dab|H9{B9S=K)YrLxKfZ_TKalMdNAGQ%sngnFk^_hJgW=2m z2!hBF6Mp{r4);Pjcw%62$SktAPjwqUPIH!-_-oe?k2sjzG?SS{+0SdA{3e)6kbrj z#TQb)OHdvb{8MI%g6Dv^ORrwEVZ-)}2G)?_HV6brsOA#DY2Fswd+I%sq5N3vM~9^nmy zY8p)&n~GVOHitg`{+Nm}7hAk=;bS&?diwkO;gmW=YyQ~jL&WB5_VmKfW}36TJi&PS z$IDg69PK)Z@7ay-pMy5piQB5h3?%9dvVH?|wk>o*dq=YfZ$K?SlJz&U_$AJGC{{qU z4QmPMq=YqJ))xazcH*{bG08X`FmvON23Tm@>wd#2JDs14M9Fm4*nNJ^gPfLilrC$h zP8>gR>U0?f!=28KDzJxb3=63p;{EsDpU8Fy*fuzwH{aZ)IuApF`f%4uc|bx=wNfi4 zYm3S%%8CkP586t0)K1psp=0xMTzlbOA%i(o6$-eUqF>nNWE(hSw%|b9mIUPvS?^pj zlHEL#Aj4L}*GxPqC2isCZ_Mrl88=wL09Pdl?l&AVcLOr_0W$XiGWUWG-ivh7YrR-4Z$kBt%&F9mgn=oR;sNqrdJ`*QSyd>5yd)u~cS(M0{%DlZThNjB1=g*(J zpoxI5=#fLBLIgNBZV?4v-vI+6ybm2}ix@L}*w9!XX9b+WjR}XoPkt6UJQuoFWX+^V z2^=%!%(24-N(A|OJ$?H0$i^i zGlOdYN%5<%zi^?V>5@x~wK?{Z(uzhy*d<(-AD}ZB;TGHIWe6k_Ax4u@4KJ)MqORQp zhpExg(Xkj>_#rUZOu}w62OivLH@37iH-p#O)C?4?%(CZHf+bj6&)~Zk>+9Cm3(X81 z)5hkO`c`>7jarMPglw1;7FL%-Go`Nn{Mq`mO_Gr{R1^`Lv!t{jtE>V)l%K3Rot>Rs z%p%cC^>AWZWhaXnMP=mLv_)z&Fq|B6w&tO9_!}-QZEOI!&`a(0=So$H9jzk6F2t-N8)7w);q_+tU zstKkEnntX;wlS4kVdX^&eZXZXkJ$w0R+}JRy2`3nAhEMDDLURD1`yuGuGh0MNd`SRs=&JK~KQ8}4Ixrd=;e>Iowd-D5PhnYM9`?8Rl$I3cXXWIk>RO!e>B=0KSHK$=+8kgn z2Kjnf{0Z>_1AM(rp61AlBdO3C2m{_l0-F-Z>-uOq zb^EQto*0?e#ntlsto&3o+vI9?OFJBdc4@P>$vd*g&uW`P(B?5{^E9+M1Z@ssuzIEe zCP)zWP#Zk3U(CV22>gYtRGWm&j>9JDof2bfy~=jJy#`&s3tc}4U7rGlqvP@#uVu^b zm>tZ@ckyoDS-ygrTc0RQ3Tn?8$1AWkUn(AT^kwrpA0>oth=^wY`7?CFT)qKK!9 zl4Z5HqO7cGdPhg;rhc^gcif8v& zztX1c{L+F_6oR6PvUnlX6toB=nd#6b{DUP8q_v6BRf7eK0AYm z%rZGef3JB>$r|&$C9~NYWe>da>|qp(nt~QVF2f+nQrlEgkd>bW5CV|7rMx(53R(mK z4CueBO#wb8Z<9}ecc58s4&-g>9*ehWkgK0YqMrt#p9Z0yU^}R=Fsr;gtJI!T2qsP; z4s?>hNyl5nJIQC(E(nVWMF}mBE_z zx?~-EdJP8lB`EQTO}#}ev^YThFGfT7V#d%A11MRu=9hGM|Ta(P~Y{1Pk-3Fwy!KAa7^CJ}SV*D%|)Myd`8sa~o6 zq=5YfY={J_&Zg!?yL^)QleR(kT|VDs7_;r!xdfz~oksBMA@?EyjPV#d*1k^s3tP9& zcm>a|tK^&>RVF7R9T5Qq{ov24SbA<^)3abzv z5gwzHXy^=t;Ez?2ltNpHEY(bTZ09Q&2%N(H1nsQW=`F=Z>rqc-$^iAu#z=cXR#D~r@#mV zM%bcaxVERfI-?s(Htw_wyQyhq(gXez?v8EA){>ekAj zi2$anFly9F$yKLKjOs)6g6Yf`UtF0;RRBa&Y$hQ0jyZF7yocJz`fM($q`GLK#(=E! z9+*b4K0;kwZ+KmZ>I?v8g#(zqQ5}2s>D;XR%-pQ4KyR&}7noDSxO>Ap#b4g2MxO}3 zya6wGMs2$G*hGOHpTvZdm%z82E@WZ#@{omhOE0n)Ye0N7)ok=;X0(@J4^U8ERC*A! zrd4TCnAiq3VL*sqpii)0;IJ-le^dRk?e6+n4e%-;!Y(Uh9-DwHJ?b~0x_Vo5ZOwK} zJC~g7Qa{431;=TGti@p%P)K+EtnmJ-fvgct28DFjughC2+066+{EX;F56}f;BxK<_ zD3NF&VZqcbaUAerEb^h5ac-pGW8wK= z6(tetk5-yCFXZWjsT*WWD|qB)Gow53h;c(ZF{i~lgvJqmqksDH4w!P|BhNf?nHmmfc$iS)s!UZ{_^W!B(r406MvW%$g1}KeXYx?f?@67_g1pNX@7VEXHQ)`hzLN3xZk4z zod`fDg6u8A03n}*&_-&!m}25k(WsIX(FVD!p&e2%R8nbYGJ)Nx&=Ct~P}#(1o_Xfk ziDf|tQ8D?TY-QpK7&P{l<>W&lWfSn@GZV@}4x#w9NapQxK8H93Ip=+1v7w(4+Yuj6 zv3}KA#7D@g_KTe^rb9zf)`6%7Ww-a6>>r3I+Tz1b3uM)_V#+il?KXjRG?TtqA%~VWcQt6ckc0 zk%Bk`eW#c}QQ9dmP~b}erR_U~mZCT*;CB|fE9y4|kD{S?0tJy2)c2c$HdH1-$Hlc6!c$4EJg9Bpn`%h3Q7?4-SSBkWgrDj6kMR-FoM2Q@TT&DexS(F z6l8ZvLD<$EOORp(A~jGN^ET ziUAb4i~^1_$NNpeqwo|Nl%kn}{zo8rzte3`mPlR&M=C+fdc;M z*>4JtdJc+A;|@-15Geho;3V(@#Y?9kih}%pQ^Zr0p%k2^po2p6-42oBas2z4?g-I2 z#qK)=uNCj}{S=uafBz$j_xWUs%*)WzUH{{p!-{tbwO1YNcHH${D_#cg!QGT%I0Zzp zewH#azmnkBSfIj%PsF*>@R@6fcT`zf-`85jDT> z6dY~&Sk0pp{IS-RLW6Y-X{UD4Po)KPB(^I7SX|ez5x6gefoF$0L+K(rYyVp3vP0v)uDqpQ$L`5TvT(sVEQF)pw_I&xU%eNwrw1VdsyZIlT^* zS5UEb>|51!z*$#&h)r@SFj@b_aH(LfspTCTen`YAmZ`dCCt7p?XE{b=PI z+GD-%H)X#=*K^J9r*j-0-*ep`7@)44v5@Yz_LtUQRd2Xx10Z=cT)DB^E+D1nEMszyG zi#mfzzl<31%Nv0l^3U-77CYOq^*Jj`Lr=6xYN;)s<5_m5WMT5En?R$eJQQF_JOY3tS zfvkhMAYi0cV2EH!Mp*S5#t~xDI4Zbk9MRaLm8-7X{syE2E7t_&Pt@6*j?lF0raZql z{>A&JXv&vQ2I+LiS7m2t_Yh`@4 z-R-JVpZzfyu-C%mhmJ*~CooPBR`%?V?k(YK$?4SWl@#zt3f$1W7+MC3XQF`C#OlsN z({BpCjHE>fE2e-iQTm^mc^B}RnPv>NjOKpJ;MqGVGEc_afr+4>c7W_#8C6_ev~s*U z)^7^l3A~^tN`_8SGw2&uJbNCbQDnX*AsW_qiWrK=0oFoy|AO7-#n^~NA)(h%Du5Hv z`u=eQ!`#Z|P>@N%2nyUYpL@3Kj#Yf?M;!(?kq9{1F2P;j?dHoI-tI$`g0Fx2Uxo5& zoTJDWDBwlb^;-t-Kfdbu{ku&JBiB%OuVH#`IbiE*G+*^xLXQN$zFW=_FNES1QNSPj zhwguNoB#jpHh-Z$=cI_1%6(%ACz>45Gdi#&x8D?;l1!q=G?%F(>U_T`I1+OFqnSqK z#PvU-c=nvC?WYu;6wnChyB(aCOs050Q^0o~{g3l86ps&GzT2cG{K`I?m+?>UHd|0o z-|gT{<#g>;st|_(rr&;2aKK6w`CSV5V?e(te*bQhH=5UlLzfsze3tIJ9eg(DG?DlD zNJ>G(x9=4HW4q00s>_EI#8IH@w^kgSG=!L)0{%$Dh1#LPs_y0`BZQxf?;Cj|x%>&+dYr|BoUC1uwX}J3H4rc8ydETZ*dIekB=ap~ zzd+W~F0wBGOH+34oYFgNwX|RC3$A?YxE8WRZ!W58Z$qs}z9MxyxrAw zS6lu^)A7;P=k5e#`9HfwH(~0;%Zu0Nh{G4?z3(nB-jPv!l?#HUW)IreXK!G#Vtwuv z^}zj})`6^|4-8|#(cXF0##h&}X*c3J!OqA<;ckj%#OQUw zNtEEU8n5*5^4vWo`AF`-`ZB;;&x&N`i~HVE5-XEx^)tmKO?rGqswrF8HhKzTxAdE~RR!rOmQ1lnK?xdP zTK1#*HR4y&ZlP#paB+&SiR?i&5r2aAUB7AD;e`f?&a;o`YZQBqz9iU=;$H;5zS~>J zKA~tW@NJ5(scboYnc;mMU+pXuUwx--WiL^*GWI%sO=VBemsa%%DX`1?O`FGFq1d_5 z8^hOV_7r^$V_x(Xf|GikCXKoP3oM`ZF_qnb@WH>aX74P_#NI z_uy-U+D~Q>UOZ`8k}K_4jKi@Q4`YFqV=)Jf1%3qP=48X%U1nx3%$#sju6Z~~aQciK z7#C!c)s>M<4=DGZwV@I!Lojy)x?OIT0x^#VR<WR$4f$KH~~wsSl$;m;|uoMebWhA)I=&O)m&if_yFEPe(mW~u!h#CBUSo;pZsH&|0`=)0m z$xJ4_GLS$*=!hU7m_kudRK&V#!?vz%MP0jQW`bZZpRNlQiinDZj`ZGpOK+3jd!5W= z-v4*rok_t&_WSlf5Ar6HIrp7=&bjBFcJGx1@7j9+`w1V~xq0*6y}Ph*<#KlR`7>DE zf{uhVH8s@MG}P5Ki3&wsbwfjOR{FVf>3PL$*uLf&Pd{|`a9IHk#m#P0hJ={#8W~l+ zR~EBi!Gai>4ZC*QfVuK~y`uK`*Gbsw6_@W9JMwVaM`d1K_PJBGv&)mjH`ofcLi5}$ zBODILh+CdZU@NSN>>RTw=H7JEO*m8Y6BHU}OV<|miRPhOZ+(eguP3sV?40V!o3Nnd zrYC3Mbj}UZ^==B$T6ww5KjAiP-GuTVX3^iDI20Wp|JY-aj~+Q<_|Rd)hhw+hb~_57?d>fUTvD{)tLHp=4fv`t0X>vK zr3kwqEfdU(6UDp55h#EB7AgVSJDdhF?tyHidldioElh#TUkBNm4cYS8qM>{C?js30 zcyQ02{rh)q-MV$h?jzWz^U!{v^{P3f0~-T4I-RKD>H;0B8nIP?(~hlD?bfi>tIi5! z^~oKMimW}KfByNwQtv6rx5lQk%jS^WvBvzbI@uTNUokQ%Uxk;k zt9`KbT-f+))R&K5z$+`O))A7Bkl`<{3?Qa+Q*G!AAR}Z$Nurc3vVZ}Z|G9L zaO()v&W(Q{3AJj=**+7b_do^)z!6vZe31$wI1uIL}ytw?F8ZK>g)i5 z-afuqPJx%Vx1h#RItm{mmcLO?bi&u$+a$4YLQChBS2Z;?VQ@8ea5_;YGZ}QE^&~Nz zPJY-MkH5N=n10XxLlCtiySHxLzkkocgL^?KgGQs0%T(BjA}}mWfdi9)$?EjjTa%}F zmmY+S?a8WeI7sO9!r4_I6?>e{MZ$*n>IcSyJ|k|$5+R9TUT?qf!ap8+_~l2(8bPZE z{IWqH?Vtn@D0xtQem?qxJ(kBj09w(JR!;{lJDT1O3Kbvw8PwWdU43cQyK7dxzr<>2 zZK|i!soJ`_nwnZ08d|CiSZvni>cnOd)n#Q)Xb+&M)6j_sa&Ijtq%#EQX!|J~_UxsV zh}F?pT~h%HRh46rY_1by$=TV~j>QpN=id!#EeEwe0JUa-T2GmT!KN#z|!>fV;F6!B$gq44YY>=kT``zD0$kK~PIcTe7vwKb?<_<*ye* zUDb4n0peKcRIvGUs+x2?j!ox>x;jD!W>5lj+JG$5_OR$b;$kB9)L$=u1sks2$bMvd zSUxImaj`mCd1?GYLa-QFr_*04 zB`VXEQgY?*j9p`j48vl|kimoE;^It0hnq~{^ooeW$#MgU(@68Uj)BB!z(2U<%(LUA z$HO2$0h;1fvUtoGOIU4(d!m`Wj$5@E$q+NdmyPbRb@Z4$u(qw+*73|lC$*bVIst3m4;EAge$n7_Q4jmvjB>mbxv>eJpqb71(>7aswRyOwO?~e+NA#ReaV;Y?Ndr7m~Ja}ib%oAZs|DrkrINizSZ1$=p+}FkMaRqQ1KSd@fKn|EqL=5 z>i>flWV{9H6r2d&A>crr!HqiN#sS(?vO+ZAAvR9Lf5{3F&_ECqbjt)xaDE^$R)`4t zdLfqyP2R!`jz7U_OnfC1!MZINA+P`h00oA!BFSRe&mzo=1!acNN}pC?H9YrVJkO@A zeD7xi9?u?Agl7xeBIwD0#gnU?o+o=K*IRBm?dv7lkK&<(-&iztw}tz1Wvs^2>jB|~ zy%PLf4*v<(VFBL74$@7<;yv&KVeZ;Xn1?hF<{2`=TsTOWd&PS&_ei%6eeRKNLnH>9 zTe>@LV0mUqy5+0BVXPei>7c>q8Z$b zXyOVf`~xi`IXs~Q2toxnLIXEK;R*@5l8XQbEpEsoT6`BmwD7SLEwp!dXfYt>LE9wE zZ(z;ztd;!c#iK+Ta>e_OpFd#kXOrNEb1rV;0i`-71(a+QP`V1dcDn>G@-V|`?zy+| z=!uKya20rg!%yV!8#%m-!?*Ro4 z9-RYV^gBM2`n`8F`VDvQq?zig1&y&+Y`Qfa_mOBe*USsBC`slRaaXG=9(61;jy&-Jt0hrL$)t z7Koz0)D+=gf3>1wIWDGpI zrB@kw07qXYOY&3vG*Lm7FUe1ohYACZy!IcYC*Q`Tz~l}dnAGsVWG4kCdwAgSU3*gz z;CS+FLy3wz2Gf%@S@h)2P|uS+H0-PYRQ?#;Zv9s}5@80e|2A%BL3m60Pg{fG%k|&L ze1vEJ1N|4lwLmMH1KrM|Dcnl97AQj#R~yB_9KM0WpWyJ796skifcNL{)EzvSx`G>Z z12^izRcHxt&~juh(QICmp_|BOPaPZP>G6P153iJ@ZzvCK z@8Nf!qo?2fOP`h&m%2wVc7Z#}N3eYF zLqLF!VfiK?!N;(p42b-34126VFOMd)9hdqNECt&W#5YoSZ}O#ZJ2pK}p@92U^nil5 zF@y)~kMI^8JYXNu13!=)=#CRTOoxdcCnS2bY$tjcAN+HA=oFZ$-1IQf!zGO-KS2m^ zdbAJ{oE}C(q=z2;O9p+JBqNIVO@;?1m zb9f{UJaikV-*&0r(Ne!Zk@~Hb`mO5j_kfU_h+@%3qSz;yL^18Zh+?XdSEX1a$9WpZ zc?{>(?;PjmKjM6)6(RQWdjj2lk5Te_it4X?wr9+%J!9T&eVt(+K+Xj6+m4dJY56=} zMYq0^_Jzh))DQB>aLDI_Zp8g zGd+D2o}Nbd>$%NeL(b#X@aHXvyal7Dg<#%7$$!v-n}#;FiE#KYXc){}_&?JycoWeO zzOO_>e~E_0SEpeBpFxMgp#%PnpujssLlaNIog!nH({LaThVgm-GCjQ+(8R$o6>;#y z$tyYd2hYd9SkiREdCGGW`T1mXNa^r=N{91jdqdA&nR_JrogGdP$50ExAt~689;nfS zCDckMJ43esgoCcu%I}_5&U#vrd0J`btx#nDYOUZ%8jee+rB}ScTX!c?`YT` zyaFaQ0>7XWQ7xiq4SFMK;4?%~??ewpb%3X+Q(wFEk{l^yGbWxCxiXq7IWA@4%^xd>ydCzWWbAOxCHrH zNxN?YKZNufi5me%a09dabh0F@J>$a@U84M$${m@~tP}zm;z$A5S_mQ}yOqt5?mpyL z3#}7{_#XKFe3ZtO^z^~|>FHY}>o_8WWWh-;X20}OLcf!l@(odU1mcjm9VXFBk@=E^ zY++N}l%^gN&RTevy>Yg=agO6SCy2ArPYLTp3G3jiV%;ub9WG&ABw;;X!rIV{b^j7q zR1f2wOK{aV92~$Uc;{|57H>&{M;l<)bI0m#FX>aa1otm-9o!e_L{sQ~1^{q#*UR`x zzYQ#LgbKb*Xy8UD;6{(fjUN5yQNenN0^_QPZ@b5m)EFe+-fOIfZ?v-#X*D(cbFyQ| z7Eciz&>nB1ZD==aGD*+MWPhidmQi_+5DlgBKHdzBcO(0jzJD!uYckj@`psJ^Ac1#- zOsxbvpcSS0-few05?nKTnQk2{hi-ZtXy(>~z2W=Va3Y6~W)gfR!CN@Ax(E8ovE~^q z?cbBMPiiEpmAp(;)8vq}+rvFn>z~gqKC}n(;7vZS*^%+zgsar-s5xvE|F)gOcJkl6 zt^(W1Vcj9877p9aVO8C*{mVM`kR=sO+Ilb{&SgEn2#s)Amr)0}tfR6c4sYGE-h1}Y zKypAPtpC&@Q^GBWG;U4+El42gt&YQt(8})`lT6ix| zFj^2RV(%Z1>Rkm(e;iQ2+KGj~*}w z&C9KI#h*zM{4FF2ng4_&K;P~I9QJLsuu*v9riG-#Fp93UEPeOG9!(S%iqnPJERp?7 zSRiacWM{u%N5m&eX!)bo{bOuz=t+9I?OFbGcWh6LgD7%nkAIHc6{U<4r-c|pwBQl8 zq0>P>$sc951eTH_(LkDMCFozREn>^^TyPu-)V|6X*WInb;HP2pA`=|Kc3E9kcXn%yMn?jN?5!`5@y zY!0jAuwGYzrEcI6)CJrKecT9j+z9Qyob%Wo9(~yOZ{pmDbfVtj2}C{L3ZkC(3=j1N zqI=`F5~LK5J3K{#ERrCndm#JoYYoS@mE#-E@%85T8oTlBAJ)ic%rG(=Qe>6Vw~@ku zp_3w8LjmYMzdfmV)7;U_QF_G3^At1s7NKQGBq2c-`eL27@Dg$^$WhJku zRPnq@e4mQS-cdU>M*bg|YT7persN3ybkJ%fVym~5@m7v2pbQ*__;ZF|f zhV4zo-ea}se~60VJBW%|uanHg_>#;7jPuBhhk||mjv{(2ryw>V+<~_-AZ8Xlo(JG! z2n9~R079diep*xio}Dm>lBy}fHW)$kAU8rNl-}ngMEB0*3*W| z(}tS2fkHA*8$^+Q{4qglf;oT$Xu?{615riqLsU`D_E1F$XlY#q^sZ+%wXpAL_LPhu zp_dUPk`eSLveQg(DKZ_0baL~niyKhRZrFi1sFv&}QL>-(lKo_t>?h$W_S1j%)c!|( zQ-2t{IL2Mw821nB;INK9y;XA9mTuUAdh3#UJ5lOwi_}};3Tx2=d8KvWI!BT|F~`l- zFNv!{0ddt%(|aHO^p1K_9f-uA+a-*p5h)qSTRF}+V$03lfu7vBjGjDFdU9g|`+X!Ywx13lZ2)AWK5Xk!Ew%1>nvm;o%|-9k@}ld2lzfAKuIe~rGo29)O3mtriY zb8da*Uc!{ltuKQQLJ8dZ3J62qyxaQfzdan2>uE2&r@j21_NMo=m(-^{k34w1WIoTe zNP-|Mp3iKmknaJao9)>+jte&Y7QI8VQn1N#9>>yVLg&? zr8QS0WxB^oQIShhRKzEUMt<064>tyabV+gY!ID2xME-~%J4ujT-P(%gpx!!z=!Eif z>f%Ih(w)Ti!1%3}+wm^;IDR84(KGv-`+R?Kdr%>659%?=gDR0csILAzsNS}OAoX?B z8j`%sw+UmH6q9xh_4Kv3_tazeU-qWb*Yz=IMm{e@JkU-|x+xSgGj^!^%EpU6n zRgL6-AA90TDoB+cnhh8Tq-RJ+;snco(yfC>xz)7Lq{o)g& z_CR0BJueflPLqIzaj8U$bn;~d4MoVopCm%v%Ac(SAO6gW5XmNXhf)L<>E>Sl6< z-pQ?Lw-w>)hV?SR`ZDo-W_nrE!0UV96-1tPBi_VV1<`F&=ipcPj6x)s*msufJ3jxj z=l73$UwguhyW5@+yK(Oywy!R>TdT~p_%bYd!qd) z>Dfsa!C_XCz&?^-B~Q)XZ997~?mI7DVNY1e=YVLs?TNB(oXN`dNM@g`=YLx_=gwgL zpKGup3Z-~qm^2T{c2Rbms!2bln+)5_xz7;4DShfVu82tRi8b z3SXNsQUjQhKxN0IY!KuS-A&;0pY!P6;UX!B3i?% zh}wA-QEX2Y(SY#(ZRO4&UInV=^?!xD3bc^>!>xa;0_`n7l5SFOk@|lUhpWK6nE?ZZ zR(QgNAe#y+!EpPdcIzWUUc3UZkt$$QWH!l&m9_HM4BCj$BZC#UBbQ+>71C?1FC;*tAo z9=Z2tCDgg*?#MmCUYS*-#hSn3Yr;CI7kCqoPDq)=7%mq=w}s#TvyD87Ud|5C)9lhZ zRS2_-e(RJR@!pn6-xazVo~&sgi`?P~yi4Uo8a4tA5G2`pp4nQ%x?{f--U5aFDZYh! z2L-rrub~JR?pg|Q^%MK`NCDv{<%>!<3KiV48O$x4FgD5k^#1#-;Ulkw`#^Q}LJECbEM=so*dn8y+XEnz-hjRI`kf5||K+T8;h ziC1z?8VMa=&`1#WqM>K-R6z&kNs36p*IkEDAnDKMEj73AYG8K&cL!g;Yv<(*8c9D8 z50v1+|F4<4jr)`HxCeZaYyVD8$=Gh~-#^`JIkr7o!nITFyN_Xn#kUaA+f1fi{&*T$_vYqRc6Y>^JsX z0!E&!wW|r%pzKoK1{8@&<1`wr3_~j;(aKCo4_OP4_kr6D3BVnm2ViQ^PP42B-M(5UOJ{Vh@t zuvHF~X<^`gV6_k)7|t)?BRmQF^jgqh3U>bHg_ju^5n@wI&ItEqZ1w9 z3i9>pD7T&1x_OE=R-{xRl@B!&s%f80q zxAk0b1)VOD;p`rW&hKWeUtF*s&cLcVzdj{dd_uZjRD6BpNV#b7(W>f;%ZjU;z2f}X z%P*(K;*7!*r+0sjb>7SU_LDg#Yt?zD zqh#-;PG@I(d%JP`v}xmw9eA}jXPhXpyRtWCF_kJc7HcFc9O1OiR_u}mMg{w+bY60m zx)Yy=jU4XV+S*!yC6G{%F_Wsw8oHd>*=5zOST?Q=4G(uTVhq=m;e;C>UuAR2Bw}dy zY)w`>`3s)8mg^#KNf)sMu8Pd0s z+vv4v<3_xf^oT`HtBEa|hI-)Q=Z8bv0uE<%;YirEYwNHFL2>5cwRlg%R4!=R8zix{ zC>qkv96y$bQ-wYdTrRAA{|$>zXIUi0rbxtTn~Rdz7sM)3m|)UxW(5SKC|1%hW@UV& z5=>7NUl3maC#?Z$R|n|S)g{B)QJG5Jh4)Yb7P;P_Qdu+xKM>dHl*z0CI9C-XI^y_M zP`3uhu;M+m1dbgfbz`t-LPkIVm6o=)y0r#*4;qY6kN2SEp0zaM$y>;!Dr0G^({D{= z>bZT2hcJ=ToK1R13tCdU9TumiJuJP-Ko!R~K+_x?7`Tpqi|gQGBOmX-D*!|9q< zNx=#~{e(Bruh?xEe{oucxH#NB`t?uS7(DpOHoUzDXoG0gThkLBL@L6A2{;SSN{Dbu z^~R)!QvWN8x6))xGX@SV#~nLL=o;J>Ui;Km(G{-}ahwl2}hh0oo z@yg%fFA3wvCb4}G^&-tf*$-vQh75saO{ZzXti+L$m9>?%?ajJ$#rFK|`7bBp*p&CN z!(TxGR#{)d8CwNyl_<8+eZF$#=k(uKtMzr`9u3?hfqUctxJ!rw`PNQ#18^tA=|7<^ zA#Y_8;>r6c@_s=9c?a^q{rE0mkK*#*aqR2=tQ;&=mYzQWWe8pJ}egr{en}) z=?Y-Ky$5~ZUX4CvryVbl(0AkS3;enMAv@QT2cT$}60&XvnyFA>chOJyi>YSJsra$i z%rzbYAZ)ft!RDIy9^eUd(hm_*qNH@g38#h$!wj5rrK@8-~&U z>C*0qVz9%tp1hza-wAa?w+STf)=~fnT?AXkijg8MeH!{Q!6y~St?A)}eQ4<;) z7c+F?4O0_YDqAeOktLW1Pyc;<{6nLnq6QlblarH^6QbZrfG zbE&Ano-Vbubxln+f73|*;&NeozMu~m+^{lWL%L4J20pomtSIvD^!Sc9zNv?_q$H6| z=K8Buj$Lo^tx(k4Kex(f<7G{O?JE3LrnSm3hi|~2no82Ga%_R%f=1eqC_V&bBgYA5 z*<2RM;_FS19*fT^lbRNCw_9aW!_qupl}W7%O(od4#_6OsVf9&M)F7wupS6gM6s##y zi}U8qvvMoaD#HdM_>@}2$-dn!;>>Axi!@%yFYLVwuo=}e7#>aZ;}!amO+Uiiqs0S3 z^IFdc+fH9e#SfZ{@Rmj#il*FiiUChC;32)G74KbODgk#OWEGbGpN2y3eqE{5D zNJ_w85|m`@YI8k<|ED5xc_^EyNMagW^Hn4?+w+-<~hQ@F8z|zzIU`JujN06E*g%GtwE0l;5wGg5vda)U9zR;72FMX&?o(mvzd&n&D zXhSc~{BL!i2g}~kDB{0_CHzca2^o-(%;rdjb?43KlscN_hJe2AQ|e&Jh#hvb99CzC znF(fW(t^zuQL*KhAkJ4>eQ{nfEWi0?6oxAFO{-QlS$n(?ZSV}XhJ7h!u^(`+G_y)N zzE;grG&tB1o0Djm2B&wK6=Ha6Yq(YVR1kc8JfAu>8Vnh-o1%c%P~bHJc$qM&hcSVv z>~{F5ZFDT9944Cgk+VmRoXN;Ib7aqs9eb=`sr-oFzy1ZMNfX%CrqeB10`7VT6bA$Z z8w|k#ip`r9RxNVcq#Nmggc#4+pxn)X=#VE zBqE~3EeT|g^_B#9ZP5SI!I$ME2SNgggo56p1^0N1FfmYppR*VR3JeAew+%?%)%NPN zWBD0b`_(jotyr;&`CND1mMw8{Yt{@Jxnd{eOE)|-b2$CSCYd-%xyN*o)jk!BWBTNw zj#>92p{*H`KP2BzhW!O$($Ky@EiHAy5gk}+sx#Kq80kOutrV?ZMMXhDTJ7}d>UrR) zY#2tE?xM+bk^bw)jni|!ku3J^sS;S5F(#$Xa-p|SXKnNYXY}$h8V0B^^p)(DF!VJH zrDmvV^l(Zh(`994T};ohQLp#w-dX(4M@yG}L_KG^ugk{&{^i8r>$b$L$<>KxMVZ4P z%N(9LoO#_gKJe*VLfX;YM3`6>oou8`<>#lO{xTh$GNvu9Lyuc*I-GJD92FH@W2`YU zohc+VaI6qr6r`PQ(#=VjGb=7`7EPF}Mq(F@Rpi*$?)P+2LwuC_=TmdyY`ce_#7p9#ho4p- z?V93v4?l^K#6S---7@=s8UYs!nWg#g%AP8>Z9lwiw`=cA2(!t_=x$c*7nOkUjC8GqsQ`{3CW6qWKEMSU#{YyDPZ*y z{S3L-LMgv6ANwffX3%WC->_x#=1rS6uV23b*N#0q_Q3LOtZ8Lpedgv3ILf302Y$9W zeMgKh%d^d&fA(xr(%!wu@C*tGi5@$3>eTV!;Yw^cn1A8ev16C8Rbhy7%N8a5x3WRr z9Sv=rjTPD1pM7?r!midRltF{SAAKzD=bwiUPfNSw4(p(kC(}ym>+37do;4y=FlrEv zavd>YLdQw{9e3#IztvDvT~kw4UWEu^O=GUjR&vzf?c?L?>*MEVz;(Xud>ivOqL&yz zlAfTG(W#?GZ=JWdR;|{0X?=dpvxzU;As z_7k5w&8exl9i{i(daA#fmX>PRxZNNnzBGIzjdV{kh-?a)(fWtIhGz&|u+(6BX!ugJ z%X}#sj{5Bd>+#S%@thP{2;Z5t6vFj4d{J78xS5)%=Q*bsY3WAqXZOwfMDjJ~AyVE> zI#UAM&RnDhqOYWy)4n@Wch;;=Rf!lcNqk#b%G!-jLBmhj?~(YzF+hNo)>qqP6e zxD?J;D73-I!;>b`3I!_akbW~!qVi8T9LX~tE+*x6M)Qn^w91uwKm;5ZXMb36c#?)o z#}r;U*TQSyL%VC>`>s6a4&R&gG^A20p`F9rCA5fAQfG!;WpCDSL}BnX8SHD^Mshy2@S>N&RP-Xz zYlw-lA<(`trzJ7no$M8>%@t}M~v18W(FWQhWEDrzTe1n1%op7QW z#*L$RB8=JRr%g|q`v?Nejg5~y`N|@zZ(d#{ZHnlVmlqow>s@U1Z3DO=0HNCn6T(g6Bwo_WYBl%lk$kgY9HJZ?V>6BkmsnQjk(yb5nqPfSwdiAlcYf~}9G zo8qjL1ksn`lAj3EDJ~g9=~N~Upf))5JU+Q6uBqzX2JPVHi&Rb@XiGP&51vncAo<&# zy<73Xw~zrr7!6sQ9uyMUM~H1AeE3O90#He$%~0WS6FvfV@#7c?4W5WwG+G{pmdB#y z!Qd%by!wqBH%TF9iZ~zI@5x;uqKkN~96?*dV#9_FhtH*_r(?Fk#;BEtK)Mvg3EH*SYHZV*$Esc%D{{|2!McvZUQragK{&LryR-(uiA$T&HdGI}%h=OYl zvf2_8h0hizW2e!V=&~BF2)~0;iq6hXFRfXPn^s4m6+>e~6R6bK*aC{NhDH#}!Za2& z+zm}QA30$aMRsKmB2MGgdA z7A4$D7ozE56DC}`w07-8V=YdJuMe2GcH+zjo__l2naEyrnk`;0zr1?&cN-{Fh%QPv(biuRGj@A zWj@(C`L_J*qQau=tjz447Om6iP`U(xIqZxX*Zp|B4wBMv>c{!>y}Z`Vpa0{sWO0FX zEs`()k^N}4sGDqACw@Z!G?BJX)(W9li!eymjvyIkJcCdz!g+Nv|po;`cHu&T1M zs@it>^5qO$rKnS~Hix|d_Sgic)1}sX0e2@8n`D$`Y{@usB)b#aE~}l0rGVs!rl`Gr zDN$CvUjn zhLNi3v>&ir?79M{QxH=9t`8|Yefo4-om$<3dqd3hN0ZR|)$Cj4h07horrbM&S*Wf6 zmsxr5^!x9h{?rWob+>eVsl4yn@ahwXldzZV3bs^sAIk8|I=OFffR9QctGDGhSrxTs zcI@~z8C$zPLzi#rhO>3y>eOK>%pi7!uT}|l1e?L$(%hv%0<-1lnl<~X71oJOEe`ypvlf|0+#3oC zONwkIIhm)=m()7J9cNQRWo1KkGk7I-wO80mt4azm{gy+7TPtf_I*nH4Gt4Nn4Aw-B z88bvrIe3k)8ZFVo+G{Hs+MwYYOY`&XDtEkv{2sDAC??;>??L(0I&-SDrOHzBd?Z)M z*KK*we!C9>Y7wCBfn?qZ$&9Y9t{~A=Dl;-Oi)*UeW%T%t^5Tp$r!Eu}6v8&Dvs>h? zRq&DGkc8E!U6hLLblNdhs(e_p1KMBdY(^m1!62RXRtCYe#w*muNa)hIxP8l)Z!A!Z zn2#jOZ86TwweK(abeKug(%@_@##kSR(+Rh_{^Cz#~I<=nYA*nu(&(!$jV9&eLWc(g{}`|Hz^h#6SJ~WJUsza>hxpbp zR1cgYzcEgsaCO*a_~}w9)F4o2GxJe`U=D?kUXRfY!BXoOWXc*Jpi^rhOZD|ZLx)BM z`Dr>Zr8<0rjUmzTL)A*HUszO_2JF{`GKbk5>FVsTD-@I;_rkbtsdV~K{ZC7W!)_Pl za*M*pAZzbPm2}t?6Jz(8P!j@;9uIEEW5JNWG?Oab@f;Z1I9*p}^n;ea;xoUar_QEb zJasPZ{K;b{Po6$|;rv?fu8s~SSBPqO!lKKJm@;GfsQ0YjWXK{Gy!z^^3;ZiXLlfCn zwpAV!WV1>8)%$#N4mAWVOs(|t31k6g$ac|Mi^cNgzTC#hq&ai)64-ab>#v)AAImQ* zEy&7lkWnUwDXfg8Cqn11RlGD(mGM!Ml#qQyvApYs#>PZa0+=1<%_A-IDYJM#f@Dk5 zm{5It_IZ>M%Vy4eI2ox{X-{O$TggMZ4z66eGBQcZjXntb{0SeItJN787o`V(T*e7j z|F*W~#=7#V#@h0lCTyzSP*+`(>nm5N5$^7C3U-FI-!e0XBlC}sj~zNBJZk9RaMQ5i z!-mBS8DyH)?>D(UG+^y-=~{ty&y|BFn8|UtOM&tZmt5A6 zot0l&mj9R;%mT}N^XAM+ifqi?_a%6_7H$~xo;dZI07p~q@!h+3|L|@iHsgL>zI*p| z*Co=Xf2JXsxSmZf&S8%fs za8N*SBnq8OLl8Kc_Q&sV3SSQjKLrXe28FK&g|BBqBy1m~zd3m+Iz=zK$;lFxMHE0S z7Ukt;+HASmnHgCqFUrf#&dDvn%wpRaAvl~O3gj@buyn;GBJ|fKLZmzKDYUk>wzf53 z(k(a7Rpjl&F%mfY;2!g!_OiV5+t#dEv*B=lWajBDDB#QX4XLdi_16b)yY03Su$m>hqzy{53Deej$)F)a0$!j9H*`BW>@hK61{ zTtMDZe+}T?L-Ev)DyKs;_?mm)ped8C@7Oxwxvf+qaBzxS)q20WaB^sj@;Me>52cWViASfg>*hlN*9~2ZA;D*XA2?G3o*-2K>8&sCoiu6Gng$SBNOT(C%>>TD+~OAC}-gW z9QsQ^L4Iy_*7@_Wi}KSE#*w!ROtglGNL@gT9qN<7 z=EdJKv^!0t>EpU1UG|6_5r!!&rX**th~U|VVCO8)ut8hj-@u&dN>!d;++-2F zBP|JRt#Efis_^EUa0@gq-4Yw z_^q<_qSlQB`4$UbAA2^tM58Q@+NA$D^_R;%%`XrP>QrNz1_IGN==x`cn;mCTMX ze{rgKt*`-IkVb|2ljE&S*4imrg*K-p(beHv#=cCQ0H3+Mw79sCs9RZCQC3o1T2xR_ zUJf0CVNnj3k^TZEl$4b>G*na+6;)vy6uFnk1ZxP03(CutptV}zCA7CYF<5B0;9z)B zTqB0rjQV$F$_x? z{ufW+W2v`i-+R4Y*>*w6gyDzjXY&GK1w*kHHX{7W^JK{5t&qnBkjKf8$H|b# zhzwh4&d=MnZ`-tK)7Ia&ZP>7O?Qfe=g8BQ#-+pGFnFWnfiaj`-I0-P-3)PYihZDIr z!75+o8}q?U;x$m63voolFzK2KpZpDB&6*?})es=Qq5SujEn9y4I1%5u1`?HE7GlfH zYE&xkIhtMLOoj9R?(WR0P~lE*uyAUONFjg-M^kNkbVyJLN)bbYL!zQ^ z#zA-hlGu@^XhaZ%4u_e}VNohDwWNB%*j6eP4E(f4=$cwZhsIu2Qc}_$eqB?OM$=J_ zbdRmJ!vP;(=+K#Ain_X-m|;@Az*yg|mdeu7Qg`en8FZcpI{yW9P6nNmLFd6R3m{*a znOV8Hc{J8(?ka@#EyyK}keQ7DF$t5)tas7ycDR%(#46OV@GJ_s67yH8@@!fo%36|8 zSc5C%tFxM0@mer@`i!YlCq?{#!*N}a6o`R4Ir9=^m?PwFMx(wNDzy? zS`%F6XMz_GgBOp27c;?&ndE5^A8ZJn(0OAS>Df3R>Lnpuv+TRIW`ba`vcKAOL^Yq{^ZN8Wz>ZL^|T5bB$)h~_536<;oE zc>1MT(4Qz@m#!V`sP47Nlb)U(4Ezk!pQZzwzLhm(wIrb?`*Zd=21cqGW+!u545G`K zbJ>Ff ziJd=hgma6k=idgmo(H!))$?Nt3NkWar4WY-_!!E}%Fe#bKLe8V%EU)GIj2ut%tz>i zb-fSLnANadFp`>@m!kAW!+)vFEwch^DR!vqtlPA6>uJ&Mgtwe5J}O-==+|cNIbGkd zW#{h~%TaW&_bm`J$eEL++ zUkpv%f1xS&uj2hygmDtVsr9;;A@T7+UV%C??339V>3UE9{6FT7hz^U34buAA8&Tw* zj4&Hrsj6q5nKdLhjKS(Rv)S(pZKY;e={S^kr_d$b5o{JvB`37@{rKKtsbdNYD=T3V z6JrZ${zJ04ysWH-e+GG{7d|R4Ep2J9E5fQTCVwA9GOJ-Iqj$bOOHpW~;lBjOgjvV< z`GtlW@dA^PL?-r!1P290@XwG`rWbt_7#I^9Wn{yz)}S;y%>wVA2Ja~*G7BSh7L8Pl zM;lx!&inkF?5y+**wJ};7`rfw)6>&3kWaK4h0YF#2v?VdUN=CWvz}ecdk362n|={1U^&8^x!P zkJ)RsYDfLHclVyc(Dv|K|1pyjo$U|?3y%s)l~v_z4-bi-o?uNBw74F6D4AvA+J;Ej ztB8xfumB;qQ>=}30U&@?u_9a-Sv7OO&JGtwOV=o2x^ROKfCSG{L}6PK;TEh?zA-ZU z#EB1PV2*v1n>UKZKc6nFwx}!5ojB1jcJ3%^G)AhejE`H4+|si0s!G&378hew!{x84 zs;)$O)oK(~8l~Xs!qHlGj0&31T^2PR`s;$@YK_KFH3ad2i=i7B6dV?cQUZT}3~LyP zFcU*jkax9K_$a1T4>Q3@%Cn3BCu!9~a9(L?S&1z-&juSTFDC~}28g#K`YI;Y-=uPdt%`bv>tLCKE={ zyJoX8SzL!0Y?A2f`cj61RJ7BQI zNeLCm0Y!e$C@v^jOgr?0UhX$Cd0YXsAQWC-PPAXvb-2hGV}R z_#VIk_f--F6~x!ID-Z10`+Ei=Xcs@5MROi5 zoS28}0pUIH=4G?`#EFw<(rmSzR_NNI%(Ew_PoM5pv@IF-pmb@s+SJo`?ATcv8jDIe zOXwRK<}$=4OdpNc^y#tM=2o`or5j^IN_JYA|LmC$KJmZ<4?KL&%-Q1+&5SdRm^pg} zqDuwR^}&$YGe!98;?0u>=l}A{FJ+;jH{AcoBab|{aPrO3A^3Xyjn6+ngMIJ*`cJ|m z5P`*JHA0csj~hES#HtG(GkWaxGof8NMnB%3_7k(ai)8DU?1s(fdE^UwtkTjOH?oRtU?afZrc3n@fV2sf{?vak@oxD6%<#dwQh zIu$`3;~jF~Wrv{CWaa=Vx&&3MW-g^tbaWE`1Pe|@2SZ^hib@ve2Xze1tJjBy`uV{@ z$LV0+dc7a&bo~8{n1l^@>mM2#6#Zu{>d5`~G|9(u93=K`NbH@E*f@;kX(l?O9fFFk zNK%R}e~4jNAK)$yEa`hdT?vdl`&0Lcny{nDkYh$%X0$o|LLOml-XO3 zwM8&I;devmtT$q|9*b+dI8nN;7a#KaXw7a!$xog2tn`V$-gHfnFxR+!@#4k5mN>ei zZ?(>4^}-@yzA!-;DP3}Q2~){2wnFhdV&3y^V?PU_!d=3(xEA1RV+Fl*VYgwpvj-60 zOOY*FgoNlymV*iY6|?G_*5Xp-ARnu-^iuVpxVa1O8Xx}U>SIMYJ4&OU<>B;Cl`l;X zO#A#jYg~CFfS>)gy&n%8bVSu2+`jn~@Edyx&^ zyZPsjFlS+D+!X(0c8INFv)KvXH=l`*9O-=-!I4Y4k;!i)qZpsACHi?IM+_Z0bi~L} zIPiJgDE0!PR4HcVOn>%BVWCA9b=xzuFg5LGHt9OWmKmP8eGq%&%~xJ{0@KM8FTDKv zo7SMR(%S0sTB!NbGP2+*tE)=E;WH@EQbJuSVS(%9YOPi)r%a%VA>I*{P?f;e!`QQ>VHfSZ`K_<=GSrZT=pSM z-x!k%NvBzbd4ZO}#`p>6GOO$kjS*2BA7SYS$cvnAEOQJ=N=mx0;CZCF5)tHnkG-6# z#+09~4G;hOO)L$wj1#u^X0}$ka8zyZ%L|9b-SzMjw+(}_5#U-W1WvGC!r7BxWij*SAURhC!01`auGV-lTO3P^wqbwAwLO?>KFi^5#mPiX% zULqjo7FcdB7izjdTnBNGl!2{ z&Pt>DiVQT2qeY9;nw!m02M=atwN)HHe*9#5NsEA`#ii**^(8oxI3vHPyxD1uJ8L_$ zcKMn$adF>$xBUC_WfxCug^+ALQPUYZX3Uu2_L{ss_!p`ox;@lYy#7P$EEH<}j4;Y7c+@ZlgyY9&!6y73zpcQhzp>8^ zFaP~nyj~9}TZ_7i?@K~%oB;_~DP30CEH=Ze)N0!ru~u4MU$6DGD1#zo&2}x%u00ja z@^L1)R2-rBB7=>10cCO4I8FNQjAm3o*s80Wvv-|WsG?A*JxHVQi^lo(Ljx3!W(4Ap zNDd;x#7*$=u|`x@9=-?@!^@X7v?{#44c@YLoHyU-(#RY#EclHrEcJ?j!G~yaWN=f7 zE69Z9j)MY1hYdp=Pi$OhU;q?DU@-WlY-;lIZEUo94W@GDI;{jURN?JC7QA^4u0qsn^P_d@w!FM)lLz~1T1!fH ztj1C4W!~PS=FFLMO(OKdJ8X&Sg_}nB)!+bG-?vPIf#R?^^G5G9|9ZaEtVvz`-SK@} z57{OVccs&s;X9h%nt{{<|LtSV3lFi~dylU{>AH96`ClzNN8WPNFf?Mava!ckuleQh zz5|!q3{jLD%Q$%Ow{O1v=Iiw*H!fSX?iI_OXyMu+BT;|BzZl5-Ki0W|4C!f9l{`+6L*8U5( zZ!4~k@*6yQ`fax%ie(Cu2~krfM_QQ63X&J0vI>FcqC#7730y%K-~}kiLx{Dws4yS& zr%#Fwr62$nEr(Gh!#Ytq_ZcyKXP3i)W&R480-T0%pu|r$AiK%d@5^q30rF=I3=A~- z8TG2t2v;IauUmv~ni7|ShXFINZHkUz)0#T0QUTBu?C=iKHPIeB&NU|Hr z4bIKZDK4@}5nD`GxonqN1y2YjxtJ=K++1HIv*Z+4z@$VOGi!LZICAn2%YQnR-AP52 zokH{F9m~J`eo|~h`k7CXagMZf$v-~JR+&{xm##=lJ6&W6?hwKY92)T5Cu;ccTj5GP z8CMXVxqPWLY(hxdnJ1t7+ilm63qe#~6Fhp#?1!Iw@=TiF*gI!pfhI4K+r&PV-NT+Y zlRPe#-#%?xs8wlq>RJ_QMD=S=ocK8j!DC#?Wfxj?W1hhAJJx98TxnS;I4IeAbSur3_w4HCfaL)6Y1O*{w;3Is1C1CRW}fS^E} zs4j(NF>M5}p;v8%^CSU0`2u?3JLrW3NJ#2d9EDV8HmDSa= z3}-*bq0n9*mEmPOx%n&0JWeoJa{t5cPApfkuu`0_^3fx>Lt@hgb(#?clhW-1_g`xJHC&L zn)NKW{q)SJ`0>GKj_upG@1kgB2|}>$si$sCy7$SyJpINSZ@hU2dlxyYLo62s4=VSi z5F&&b!kxk_VYYCcFb=LjoDeL#|NaN&Vfo$$d^uZC3&Y%(0V_JYSUF2)n`Li4fc7nB zbCP6mELPk(W>UDf8Oqn9H^fhx6sgsyy>zWCflsh#7&wqZQ8Ur zoi&Qbi1Lx;FDXepf(9)+{ah0oPFK)7eSkQ#f5w zdP?f+eUa%83=E0F@!%BP)nkc!aBzS@KkRDoTw>-xV#q#Nh*3NT5;F((!B9vrW&pT; zP>KkB;4RF|Y*_!8F7S?CIg|xcizps}D0i7vQv6vZw;+;>$Q4A)4Cg!5udt*}*!s)i za>HD_#_0;qd`?L*ocV3gXi($Tnt62N4|r`k182Z&GWhRYhsgY?h)%y79tBU;TlVk2 z_F8EbXJ=zWfXbEXH%iwmj2Ry3)l5h<2aOrCfFeM!`qa-pTUS>fasA^nz&w6gR{SoIuSchUgdcE`(7GMxO z0;plUqgNn{{SR77q1p;1IQ2){Pg<}3Em?dE=|>IySZn6C-+h4i8+iK)czYjsdoOr9 z3SN9>4prslW#R34`1D!u=Rr)A;~<2Rmz{CpC~dOOteQSY2X`Xl0%8-iq6AAkJu(P}?z z?Jy~(Ja^BpTelUI7H`~$rAfNOM=GF@n~Kh(1S%06@fW^m!Ip+$Uv&(L8>1~cuwuoE zgPH%B2%%D||M@D$@l(%OlMsxrW=GU-&Q-(biFlqeZadi#VI;!|V(fo>7e4U#LWC1< zzhMl5IzG3~!G=34*cNsWKHhfx+|Djg1MpnVnq~Kd*)2mGI<%;a9d_;Hi_3OiNY5?q z(hf>YOpJ2jps+Bzxg1WA5H=HKIs_Io4BJ9Q28Kl;P8$ZoVxidsBc;+0_Xz zUl6(w8H8ffs^Q{d(ajw3W?XZ{WO0Ug6aLQ-Z$ZlDRm~=HtTHx#y@YZhbEPjm;=SbS z&FR%&?}TZ&^J_~IR@>kbEYR|!;g5e9eEX_@BZ6bKci}OoR^2j*>f8J$-LjCXuDF@E zM7hw%_lBG9PQoX+@^ixv$TlJwe*@EG>8 z4*o6Hf%jj_-k2FZm#xGfP;Y{3Io^LOd7$63ZxQoZgo4M&ppWPN?Ww1pda|4fPNL?1 z{Bs^q%KQ1_SuhZ<7Y9k#H1RpbvSnX@6gHQZ zoegMp$W@3Gh@u?9NPBy0n;pTDfB#mB8MMG_;Qy=80#maN?>`4gh9_|8QjRU_;u$!6XR}I*k*v+NW#<+a z<(xuc?!nW$&Ry8Qt1@rbf#XN_Y+1j3*Ou)F`k7+_hlM*FOe3=u*CD=MQ`=>`a2he4 z69*4}zGg4V$j@e!)H^Kdnkt2tjAh#_a=W7obvNyGttz-X@$qLr`e;j8yYiZQ@R}Xj zesSZvb>qi{INHjvSKIlE#p{lyT}+RTcyIB?%f5Iw2}SIw;>)s>FCi_^=95C=BiidK z@4UskZp%k#hfU6InDXL_FW!E`T?s)hj0$_JQ?6i7JaI?T>;=^qGh3m2Rq8wLd&24! z-kNqEI^&*u#!;2?GT%3nf@{{jgV(Q3gRXn{ZLAT|UCyY`4!iEw`|i7M?wCY4o~smZ z-#zz{N9I7q9A@W`8Qcqd@dy03u`IR^TEu3KQYb>=hx?D2DYT~x^N`=VS=T(;u>s9~ z@%x3e(hf^7%AuW1@!4npI?%;#yfZ`z9cc{sZS`l?snt%g8IcDBm71HY%c|{Nh@-WY zm$dt`y!7K|tHh2rN_nfC_R``OdrrB%xu&k2DV=u2O*>O_a-}F&cf9h&pv5_2F{1w+r^X%+k6*Uo0MMZ7!;A@4I7yr%q4_nWS~c z>>##>pjCkEQTlP5Z=q1{p3gkhg;e85tMK^RT7?9cnj*UdNR;{&0v0ztpoQtvvGi_4 z&)K4jFcx0NSa=UDEJ6#5@Z3plZ4DKa#Hp_Xb?VA1>zkYEY7qaaZ*6Hu7Dw5Zce)tG zF(_Td1X)nof+%N4TPsX}Ye^|0`i*X*!pMm>yGSXclmlOBN5MI*VG=H#IC1dcal0J; zTvL5Rt5dN0WZ63WqNAgueC!v#|KSkwQpTMuBdak7!OQi#&YjcFnl&rxFs0>`3l|0k z(gMB5#mBYg@3x@aYtzO9*gzvz6uVl>vB(X!dPTwR%rRke5!=?L9sCxtkX41DQ|`qS zXn*_FSIxRd+q(pxz+iuc9ffv2*G-LRJa_I~P2`j@abu?>A_n)pA~a}7(ql=qF&+EB z@Zw+Z8`*@3y)8CL5Kf#ZDVZhJ`cvp-4T@H*DuifWe45HOSD9n5K|x8A!Y>k%GX37W z6bCd4*aEJ=34eL+x#wnw>g%>ITei%H!Yr`(%(@h~xH%V(AKG~^qtZHQ!>^lmBMY>5 z`}Xa>Z`=O=*!%7Pr^@W@_uky}DVg4Tqr=cUQ&B*gh@z+{f(nWi6Bbu8SRA zQPEYFVnd1`Fhd*4z%adMdhdNElYGy)85CCCpZoj%`TqEFA;~1Sc-S-+lkDPe1+iyK?uGl=J7$ox6B3HTCS#Z~pbKe|_`A4=Bio-;U)AK`J~1 zg#iBp0pNM1jF=b|7>*1cIa@nmFWNph$yCk&C}{42&zYFqmHI*PL!>{pcJ?;RL(q+zeY_*j53HwR-bycB)_I3p`4sF;&;?p%6$W|2(gE#c&pOY_KPP>qtauj1IH8tZjhzGLrE?x%V;-PD{z(953&ql{| zGY5eZ6?gBwN+t5dY>Bo+_x<~u8jiywvGe<^7T*yOZ+-C1dCU$`C6`+ivd-p;)grfX zBf?HD^>^QW_Z<;^jfXI9_LU;VLSfQI`_)5+fBJTE^6As@Y$rD0mSoG9dq0$bRq`ik zds}|@1h49Ir-`$07fk1eG3yZF=QL#(GRw>Dh;(9vdrawwF>paUY*_+kojvF1Z^Dxc z7R30O!Y*CBgbSsfIeGHZrL3&l+M=TCXHubx=3Y%X_}NznfBJsk5ylHnEdz)v0Pjh(G#PKn(D#mbSW!>R%#_)mm9vTtont=hOGC0`LQDf<$^_K5$ ztgLCQE$>o7KZa+fzp0@cmT#}8jgqnD02%2UIqNF6-4EqZRDKlQ-X^`g8*@Hiob+(W;gB>cW z$V#hgZf>ZqswOCi&f?_c+SiAnArRgO52GKVOl9?OsOmgmLE~#zS0#g+hJ?4(G(iWe z4N8L!_L+ok#EVYj1ShV}FFeF8#2;^*umcQl=;*QztN*YFFG7;&EL~kN1Z_92WEn8} z3u-WPAbtQivJ49{zV7VktuL?d9ssbzz8V@Z*0x}*nIIt^!B~3)W6j~%$#nc6Xe+0V zf4Gy9tY~`%NZr@pg4Sc43wYquz}jp;L#{%2^(A!kQ?_Si27^IATSDVJaf8;CfBhrO zqIsgoH(oy{Mi!Tnj)53QO=YRMv^e(~rr^(?Bpo~t@spmBm4%kaI@gq>`Fn~SXpGN3 z`|P8SVr*-LEg*S=hv#MawoefM|1c3laV<3du?QTE;y!Bs68bC~`%+a56D+`*uKC{&u8V7(Q;?=rN(bS_}YZZ@1``>(;G{^|D?(bL#B1 zg0l7`xvnK0QHMAP9&Fsk1+mb=c4_apBZB> zckDnWn?po&^rX0`0BzDJ2oLlHRD#wjUWYb`@*HSIAhOTdcl@;R6Q@j`?x%zFPy|Jd z1OR80r-)KXBJb$yFGdPFL8eVy`+HCznT24iMPRJCW30gv zbVTs}riTMSRp#Py`}5CCFfkzs7%bVPt26y$ArpUncPCq95*FbkJPf84G8gB6T#L0m zaMQcz9KIu8HUEuZ!NwebEHjrC*`Je{#l$R04vrokI#);gbEy3@`<=>hr@!@gNUjbL zV2W%uJ%QX$1pU6<-;szT?!<2+)O;eYn}~T7T3iHgEF?!+334{_%t);%E-FOqgS;KZ zgwBFPS_gp)|0t4`$e$yF2g{^Hf6)VOPn%Jz`}}GU+I`7LzJEVK5m!6)AwkuWd+d8W z_J6%!N5A_4I4OYXRzmuum$)u@87^Ywi!Z)7sk>XHF{p>$xx`gF_MwL#t`=FfQ9C`} zEeE)_pG*AW@2AAFFPv1oDr9o}eQVaNS$qGS`026iBs(dKh=5C41u0Yd^=F;MKA=T0w?c^N~?%Hd|m5><>KXLYq18CPY9cH#l|L>Xxs@J8z( z{%jVDO>#kFpa6lO1W1M+oQMR9P**n{p)}KJ(|#n;koS>eQ+4zWeS6|M={yy?gh5`*Ui!l`wk$>iQPS_#ObJweuu2knL7?1nIcw{Txys2cG4!Q9=1E2kp%-0?*Qs@i>wqWJ@5A zZe*8)7L}F#B7lmE5qHg_JXF{Y5K%zal$AjqsW{R~5&N?x>ELT6N!~v|vlgoyXVQIs z`SO9U_mk*3mTRG45@m+2R{zv58%aVcFfEo~w@7k^GzyA}+Q0umk&>M}RflClfxz2Y zq%drM`y`8=F@!}$nRM}H^W^yDYaw@*Cr%&b%Bx~2%&vYUKB$x=R9O&caX@&B1cPTq zO?f%?d+wQswEexv0CAl(3I2DbILp>EL1ZU`RW*PV69O-0;NngoEB~C~<==wQbTJdY~H0f?%GoysD!*jg`DMk$v6NY-r%nk7c<;UU z{`S`HW)8thzVZC|uLlrsAv9%vq*ij2z_%u`sQ);4XF`}AnAr+j8M82 zR)9Q$ol~k0jH!a+vPzg+c_cRJ>vDvj90dp`yA3X2YA3m3@PPOaxx-i!^<{foU$@wq7_+szbG1Yke=k?0Ufni`Xo{4S>o|=CAihNii z081+nKhTnd)cSaMM2x7ono*Eh$&exk`UKf$SE`+j8h}Cs+>kbU0Wj@I0FuLo^BVnl zJTpq$8{AVD65ze;v;VERXZUa*id#|T8Rxxzqy6eLk7DuOo7vpljj{_uV@-YLmucEz zE-}$;9rf;3*#aNc72^;H2TIGrLQzflc1rreI#yX$TmU1en$}JFx(F^QM+-!p-2r8R z@J@1t)EKZ>;owKxQa}-_zi|jc3Y4RTWdvS7KvNKC#>(a4>ZFCq;|1$S`r462{kikm zrQILN$;ykIfI7baK>ZG`ZzUH63E2R$5U%)3JWJys1mhqK<6tDlK>)@<0PfY9;7RG{ zva^eeinCJF*|+cwit6O#%*>{yWR_WO56?U!h97Nj7Fo zak0ovIAw~JOoj^=5;zBL8W^D7WnTfw6~|6+-*60eKn-@@yLZ6?kvo)cawgbMb5C6z zV)&gX{GN_49UR&~@8$1ka`yEFxGOZy*VWI}FDi(I;7dX1>JmAJ0lV(G7QbORegoNX z%kdkQ<2QI-{^(#?P3FIjWcKKe)wC8|KJ~-i&yVi=TLQ-6UADbI)<5T*^!}&&kDR=m zn_Y13;)V1A7-}zzVHG*0t?ZyrXGNP$Hh@|fHFdp?RVqIx-YV0;OY5m@Pu)x0E|B`R zjokiJ(+F?xefvg5ySlQ;$nb!#_V_yf9#ufEv1#k{YzzLsrgTxjKcb3^jfFb#lx*iu zKVh~#XM4;vqE6kFnbXPI2DCD5b9rvH#nP+NcsYQo!EeaKFc%%*+C80(@`2vwa?}hP zs&tU`p=Q8OzR^tbGXzN0ig4SZ8JguJkPS0MCX$u$5a_u#Zh`EFBRryTx69k9; zh|#m=#znch)pvDTTu#2T=Tb&S9TxNMuGTTbJRO~ZuN>N<>d0ZPKjyWzb@aL@p_?i+ zvY0TPySGa-fwua)26}ZPBO;<5d+pNy(hicGsk@O z+V$%>xf$7cxoKA~UAl55?Rs|h*Ln{riw}t*Hh3u@@$73PW3rf|M`Pj;fV%@ks`^jQ zAb9dEafmDhCuw72Rn=Sx(CK6Iu>}Bq9V>13+BOKLcmDMqM%oji2;l!$AHwywx1=}d zE-(TrUwP>zQMF3x=mBvU`S9n`3-h4&Uq_ zcGQ=b7hbs5-c3OqM<1hSt+}>#A$oT01-2;h3+hE7E&N>77Wg0L@4rgq>Y4 zVxB&_5TNUPy#su`yhB4GBErH#1ATp`{?Y!D?4e9+J^JS@^v__XwZETamq=(CfnucS zDS9UDkrVuiOAs-KI{@A$G9OEd2`*DuoDOH;w_wTZw*mkFNVL+@hYU9|w)*-+AgCA< zE@yYkWbts;JdD8TX6?6G&3$kObXb}$@25l?`ut&-pPl*FFT8deTqd+F0eEIV5z)yG zA3kyfJPL55XhqqUq}Sh$1z!Dl(Ej8qOqy=1OCw@EJntQ|{MeD_p+|3ebZDbX%fT&% zSj^JPF!Pu&b@KSAs08>=t^yLD%KoOAKPixW2f-5)fR|2Y1@x6XDr(I1sSnSG2k*$> zW?%{23T_2Fc;8D2{|<FcoFZaO_-q(0rn{S5Uxa zAPZfg5W%9v{9IiSAcwmIetNP(S%8A*qi;Qa=;S0~yP) zDFvi$pdSH%#0`5py98e&d^Kf001$i5q@%bHIjnkT0(c=h4n>fn z{1&SL+7Ip=naB!QaL`fkFgW20tCbS^5aWini=>EGvFU9KZGqSV0FHWDc1!zlV!s zKh_bXehi3{_)VME&-XHa3S3QIRYOA?IhZ@@Eq)Ux`tzH3-hbjmKTCZFQiTWF8XBsy zvD>F+ulegY8Rtp^rIpiiPS~6|IZADP&pzhYu3bBucw?|#LAJzu%p`kZ-IGr~hf#I9kaaLs_Zu;X}n zSK^>qA-^wj*s!~Y8IIy9c)@z~BI7A|bO9mOxpm5?Na| zsDXm$18ENtz!2Glj)iy-92c14Hd{ZN;z61eCDH=(qa!7X=p_iwM^HWwcMsx}z}NU5 zxm~`<2C(OCxFh3M?<`q&M1TW~CoaSrV0sK*3hoZ+S~i0dzL2h`kRkR5l%!pEBKwAX zUiV`(->`4ODKLRdT)R;Tt1BNa71E>&q(>3KO<~};!TlqdmXa2Yt5dp%2{O)R!ha)! zS^1c4z2m3p70pq|0$Vn$x#$dKD7p4Ix^x$xIdqgBmnsHaGyrb}3jp{k3JNj3YTKkRy$iGFi`NqoHFy;ty#nOrgIs!gtS#CWjdg{? z8r{=l;QD$+5ZD`dg+?i=s!x3M+O}=GPSl9X&}ADoy!2?I?^Z_W(aH0G&_BT%xHxX( z;1-8jjXH4r6C?S%pc&G35)=O#eF*gBF$fIW(%4~qa{}L7XQ$Xc?ujR2i9-Y;;4ip? z$)gP1sL3LC;ktn@IDSdwTikE>82Gz9T1Cz&b#NH)2kM=-`oOE++}hfV6pk`;O#xc)d&*_Hok!wTpM%qvj@66 z8w#%#7hS(zUT9{^fSTHZLle=EC>bYh?+Sl;!3)!Y^%@<_HHhulLCNL|Okt2A@{3$E zoi}*o?I>g2kY7HwkJnlFb>HIG9l)=fgb{R1DNNLtQDMX zE-1*)FUSUN2tk9)tjz1z;LAfy>GBmoc``E(s^G1qwXx4;klEnWG?J$Ws0kzA2QO}{ zb+2_Vf9%-D^R4~Z`k?AW+B~h*)>Lb{wZvM1h3!7uX8yyEKVD&7hE)GnS+`hU zwQjU-us&~n*1F#MB)+3>>G@vkcIz4IIqPL>uJxKV%UWp7!HGZSs3EUoXXlB% zdOdQ^HbGC>#C5hA_=dC-u{2+}bwf>x4V!@CSql^V2lfN^PB!?AEL)OHCewTG7+EY~ z%=q=|XWYR`v6-O~*P$|)D>UF}Y#xX>?Up^2IAg{N-k)2`J%Mc(_Y-%GtKcfR7Oss~ zARwaQ{pnkJ=KZ)j?lO0j+k@?WZYO8t{>p8}k!$ra{2WoQ*s+5-J10J#h#)F$27sZr zb3q;k-hF(c=vdy4so2+JG68P%W4Xu$xfoPR)ypsA2P_OS2G^M@k*QW$S6e}W7t|L3 zprNgew9fVx#Ge|{RnYQj*jML)#5E6JA9sq;`1|<>28RH$ zh(}m(aG<|`^nX0uKCTvETv3MUMkZtPFs>Ft--yqj1fK?yB*RcDwH!Cp)z&9Nn6|wFc}mIGd%F=Y@9wws!Itamy>4V}gUP}C z&Ru_xhiH2i9=OMZ55B+H)+V`Y6kT07;s?WAgiIfT?!6K{y%gHiLzsBjAo}is;CW{o zRENO7{e9f zOI4Lb@FL0)mJWM-^Bl;$bJz~ULf!$$*csYj4S$SF4Y{At5VV1fTIoQXrRp(Ex>vBO zy?M#N^|mG;!i&*K8^7(%i^d2w9sqNB1W$=}>f_bwu(1AqeB|V!^FuF9pEVr11q&5L z$FPV{2aQ_G_jh)?x;Y!2ipe8a3 z{rocI$0o>+QRwGU_MB1-Pg?G_f< z1L;5z25=B_h&+IhiaU375j5Vhwn@ki9u3BWMR*qiwK)LFvp=-r`)&8y*7G0y_+tY8 z^n{{(zipEYV1~B{Klm5hZNko-7ca6HYb9Jm!^K!~$*EI{I@(&BI=Sv1gHDk5HP)WQ zEIQFWZ1rQ0EkY0oyO>xEk_8@mejy1#)((hq59@$C$&Cx-BBafgHDZDLTDIxoWy=yc z73a*V@b_*WVT^d#P^Vc5>k*!R{=WNgrdLI-xlz=<^G;(Gz|#ew2JbZra%VrUSaP3W zQ_plMfGhV3d*>aKXN?(|0LVl{+G#0RU!cm=K&q!gIb{73<_vZQW+@^CwtmsmTL<;h z%hwl4r!Js$fy&|yVL$Zu_qNZ>Y3Q-}n3;<(Gbf?PCShj!R@Gch`Q*#@cEU2>`QDeG zfBwM-yT5|c@YU{5&<%i-)iot20~poZ(bdsw&?F~gcJ^8=m?K8n#Zh;jnFAws6&OGk z{<7U;Tg8=>#G!lcwXG6%!;sodA=HO4>(+=|_i0fHWY@9F*RvZ;DkzZOxOkR6rh5%Su79TjM_(c&vlCFA%~*|IdyTld zp0#bk{NF6-|Ni&C@3;*293B}6o*x2=H{MCryo#ST3KCD zUS4F!b)5KQ>%of>t|KQG*}@>uY8GV#B{K>q%aB+)m;vds=U-o3K&M_le1=dC**|^( z)a}#I{?`a48DD>4h?ivMO51uaX-5J)a#S&Mh3yIGlU-e(+AuC&g-oj}79+p;{4G3= zRx3tLoHB8Q9o6CF?Ge3NLUpk8i)YTK6;^eLAdF;B7=;8IdSHztCC|Zo_zx}?p8M0% zmX38bAG)GFIV*)VabC#Nd6V1Ay#v=p0jq~QY617UwB5%|;a*QG&aBHR$1g}wkJ@AJoH#5&J)DM5ni4a4Y>*l_h8O6+(J#^; zvJ$YM;9(D}dFGjCHa&6QB4Cvsd2YkY>(<3XG9-bgT-0uOan-6-YhL-whF4z&de|hq z{CI!=E5Nhs*=qEP--sbYVn%~?TpJYR=i?s`5EeRQ3?NqV4{uyNapcI6qfv%3JaWje z$S@aCHfiR}NyhQ9lOUP#)wDz(;7o>09O)l`myr{v&KMRsEq3(WxT#20m=!;3@`MQ! z65>b57W);Kyj^x1(R4kK?52JuK=pS2J5L3DvZ;qD@9#6LUo zh$1k%AhRi7Y%l^76$ue78~n{HMGtWRlY zW6H5BSVYeq=mx>fI$NByZ45l!I0?>&I5zfyM<0E3PIPoYEwq6ux%J>z7XY|FaXaP+ z`OxlyD}}bjwwHO-z8016saz|l>gJ3SxbB+5YgeXBnX0Qf30br4WR>Hzl$3^`5p(~t zmUZ>Gjfk6t#vc_EH^Qm0zUQHJ@goB2Q&Nn~Ju!CPop;@J*V6g1iIYcz^Jesv*u+=k z7%@lXhQ68@>x8U!=gC8=f#A*a@E9`>`z~KJdH%%V*mvThb?XwKEl4}FPx%$|A+AiK za?F@9lg5V!I~%pmA(7)JjfNnpihL~I?#p1?G>?T>o|`vsc34*`U{@RTI)#%1>mrH#maymumjk(~z`6^^G>e2+jlz6nFSvVqJ4rk>AgaNqL|%hb6>!FR zxP|`N^VwIN%@}WKkcj6X5jR63ZiYmhMNSDj_yDi;773#OtdtOm82oU^rbNChLVq_s z@MF;qb~3pl2k$?6B5L2|di z;@`i2|97Xa_k={j9pXD&xP1JZ{fieb8lx4w)4%@uvZqx(at*(oghTI(6lj^rWFkWtHIj?``4T4_YPS8tLkE)-D@iQPv zwbm0{0xMF_CoRXS>KXC)>_N@`AUkiHYFjLaW%5C0i&3fSAxAhrKb{qF3Dy%ft)n3W zC+X?$QDuIBon~W%Pi5yJIQ6t~kB>0kN%$;mJAlgl2vtx13g_BR{6&B7z`%N2`s_6Y z+OBu_od9Nj&QKAaeXMwCLA0%OU?7$_Gu%)*4D=PQfj%z2zV4oWURqb*fOvL6+P+f1 zbZ@k#sfi!G6zrb+*?DNRUqApMJ6hoy9`2HxR$-t7I8_yq(C*%O-@S46+v6XK9Gir1 z42p`hT$k{0SL|jLxyXkGzgOLR-_B2Q$iJ$Lvj`7k2XNq(-Xf_MXf25{4b-q@@X8V) z>`em#G6=E=!<-27${os)&JJ8I;59XH`n-k_ZnbT3B3FE$wO^ot0cYBF83Ysenj>Tp z{ucTr2d>RHi#$DcWC&jAEs}9jhE1YcgI~`Tg-D1jmfX)b9?n3MAjMCj(sOeF-rUK_ z8A*O#-Ynw3yt;j!h?pnWF;6aHo`{$yKxVO7=rl}-n||~yv*4yl@COsh@#YzYiGqCZ z-=7qNGsyAQ8Dx|N1$l0dg*N?);(s#{{>)6nTtTsKQLSPB<>1#mKH|TbV}dltjF5qW zaR>%r(Er≠1AU|K%)`OS25Z{|o%e%0D+5{>04U9*Q*1!q^8U;HJTG%Y?XTKK#Ml za=dwRVa|a2;rHjq;AC>Vbut-eVfX_?aAQK?^_Dqt)0Fsw>E(Fy{KB;PZ*M^I&_!V0 z#0Cp&A@rGF@osj1t7L{swk7XkCMw_jyi$UanDdeCepDcU1SfI>1EL_YoQcfC!MQ*0 z{C-47x9`id+49<89@nMrAuD~YJSr(E3dQDpw2l;YlY6_kKu)BjBsYgHgY$d7HYFvA z0S;pE9=S9Q9vf`j`ln-#nDlGvPaQ zp2!D}`t44386!tq{T5%lp8&O`^fZ#5is(rznxxhx?z13BrEo~4LGClHMl$Y6pWt~- zMoz}H-2BYKZ1eYec^}fDn8G+4*80Q9mb~4z8oaNsd;)^K zJ^kH%bzqGe{Luc(h`Ujb`$XnEuwN2)BO6&#LaGERf|52=SO7F|0U{*7|0z2z(elbb z-GP55Q1tRL;_1EV112+Z#+|ca<=zX0(}Y0#v$nO|r(eyrEw{Z3`m)imo@UtQBL8(0 zthvAZH~Yd*d7YBD*%Sl&68}ti&rf1|L^jF5_TZmMdT&}wuwnS@^59r9A@Y_7asv?# zrH`@9Hz4BnIikFU@K_u{!2N|kzc2R!JQXiXMZHrEY>a`8H6g2%PXWakJA2~;dT$yD zV-t&U1=2F0V^&lWw-O>SzyB!;aRDdJ1`J@IX1odS`GJ4?fDM>Ny138^>yG>jsG05_ z03n061Pmp=_bChiGpDqViD4KMaTpU5F(w8pyD$$drW7zDofFc49%eHZMiGXINdu(- z5ix{n(vUbuO?wE0Fl`7*rvb0 z0kmda1l0w3i$xe?*iJ26IlhzTu6X#V=U&-Fd{vZL1>KB_jxQBOmr^Qr)YGLGp#>c( zO?~stRHFwZI)(8D)gjD(d;sYzLXk3Qpv@qDgs8@!E;J9aUbG`fI=#UeFgJ*3(bFgDzaeFWIxgYE$N0MsoF83ANY@8 zkzKCxii_r~l1tfpVNsZ7Ti>>jS43*=XEW9~C4 zy3F;ZopsqTFHMR_ou5mD??ifUzt&=0i+snMajjq23G+cwf(Lld3heBJMY&m78R^%v zD1cu;469fM!A^Mi7j{BFQvIR$0}?1l3<74(Aags>Mk5e^A9x3cB7BZ87`BPFd(~S% zK6LsF9!K7v2VZiSwbR;f)7ZRip@Z8<+cMR*oj(F_b2JG~#m-yV34<0dTROwHxAB1` zb8nv=JJ0p#mMvQjRQ6jV78>thE!^|)Tnv@aG65hOPLpC{`&p6X&qCmi}GPiXJ= z(dc)oF!N%tjE%+^8~qDEphJDBbhD3~$-YtVy{@}n*c%cO+*6qR;fEillyv&S8u#faPW$G=9dGnC zFa5_~cYbyf3=#S`YXcW9}hoTgMy!f$WeM ztUaP^Z%XZe5t`l`+~T{C!RS>nA}TB_Z0LxQdm%atHKk*ifAr&5TX082T6 zMcM-$f8HbhV0U}}%trtGf7HSMi>QMQ=o!je{SXp!17^+!$n_amLf~<( zp!&S9o#}xTRk&K<7zd7tU=s9@_K++Hs+dlTBK9eQo10K=ShtZnfn)>#vJjO)sw7YN zMJ`E0FdMw553z$p6vY{M2w0;_PCgrnwDf5q2tS1n7tWpgFdj~Ain*-|)W7@n-hb^s zbUvr53-U$gF?`0%S(99=s=gAp!5I#Qp6Bh;N)QiZpgy1fwr%jsmAt`^SGf2^jh``R>0OU*+{p7GaSLWd1qZ7uzTK+dW^_sPZ!5*G zuIq9hN$~K|7f#Y1l1)dvo{}#FL}e`We|=raj<_9TbT!84Ll~pC zV~pO8F&c@{NWYLUc$A(;@Hz-|fO+t5kl!>X59A{NRX{yM3L4xsv$=ozgb5>bxQ|}* zWwSioYeMv}aZe^t!o0MZ0EBr4l35SnVsoWPWgzf}K9+o9{BzNj#A~|=A^sijeZtfR zxkLZ8%H}>(B>KM$jX};VO5h{XU`X?Ftk0Opc6QR*YQ)C@;UH(5j_^Q8CkDa=Q$tH6umZ- zY9ro=5&$5FUV%KoFmp4b!W-a}P^u;GEm+fYkBNdDcWOf&NvIJ6Scc>ROm$*|=CZS6 zQPJT6;p+>fnV-yqbxrxqAiR*amuvXmIeBzU+-jtuqnWXN#x5BV&jlU{%0nWS+f<3QU?(&28#is9 zjX5lFU*P!-_E(3^PbZFKG!tc0?_U#(4}X9)`i5u{-YW%E!QjEGG`zL3IJflSnUUVS zrs|&xTuLrV(Eit)6NP?)e~KFT+Tm62Cq73v9V)W{E(jl^Cm;iL&FZ@l+y)r-bhm%(>UP?>L&B@KmuBbS4hgoE_$$oHQv8 z#&{vvv1IcrD&`xxo>Q@?i&~<{U3+(8{A^^j>bdog#ehv|(IO({+lq$Cv7aN>`jc&l zun?hQKn>Q@c3U=9eKWu&b<= z2=lE*_b#S%)+jwa)G9|+cQ0tAVa3P?_vNo91Am*Ktv zaP9JyjDoWAq71Z$RJ#adXL{%4%(ua$6N^6e^4cwSlS_h8i+Nz4`=AMeq zPWYYaQC`(ZpH^3Z$hpQ|5s4Z=n=rF~^)gU1KQD^IE5j8E{s8p>;MO>j2Z&GyQ`y9i3-kgjiakV>){*{8WIF74vVx%}>q_|?FXfaaUf;nUm zLOAh)1?<61d}?y4$a7uWD5SKRe;$0~Tm?K6UtxTLV*3Tmn-@?@zF#I-c|fd1h)EnP z!y4x3=&F}h!VHw8F`;el3>MfO!M~twJW-2462tIWpt+%*VTQp}@}R76`~%r9J;CRJ zf2Ak%u%M7>i7+e`-K0B)S@ig{+>cVF26$hRWFRB|5igOU zY?GKIePn<`#?DkjS0@SRFnDz*$R@!m9kE52hCJDe5TEWHMr2D`*;9DG^LwMfBTMt5 zxJA$e22wqVSfDGs22m@aINvCoZxYTIieD3o^LgZ#mcY9L^$d``0=&=5D=S3`CXW%c z&HnVE-`O}R26{W!(L#aI=LIj1l-TMR+cWYv-$Z1FY}?p?ku1`4!1I!=lx zp5oknW1(*yf#b) zJ@V-MZPJs6r?U$@4^Vkf%+Ukyy}X>9S=3E?NG(m=1T*m)r{FgZauYDWVuX8OTb*?B3$%ifE)=@gq&&g{US;p@%$YVZVEX)Jk!|b}N4&T4`o5&*Cu> zJUt*%=(Oa6u|Lrta!gE2M3OB4uIJ?;c?L#J-cbR(XvkvMK=orb3+u-a40Ux?H5C@B ze42|3T4WBsC$2W@LdH*yj*i~5=Zj+*SAIGH+EHcbJG;Oa|96x{o@CqTQV|{=9@f!~ zj9y!1%8z}0U;i}$2~)J8D2&m$q5_Rj(AC-5)l!_EUHLTWCP|tXURV~;-QA5vb0NwU z2%FYE4L57BYTiN+1^8=SoM=tcs;FL}i!x}DO7+M?kT#>o7^RBeK)~Gk1KrH1+eDe{ zSa+(CF9JKP`n-TFPytECXb8p$3}TRlf*Xj~-tmy+4Z!8#<3%!v3pF_TWicG}DmoB{ za@-aLH6xA|Va7ma@ciIBjKmmS6>R5jh{QacA=UGd8Xm%hZcZuV5yp0v1}faw6db0g zMsd+kGJ(m^{iJ$wN|EA2(6y0Mq(X1>)@Gjl5{#9busz4=lj0~H)Ap+KJMZi}`1Y2h zq$CWGswRcHgfIl(5N>Mf54VrR^l9y@>8W9wgdVdBIIiPNV{nU0{B%Xd6~ zzE6j?Kv}W!8W=mrwqe|#{TVAzPK(jCoCY@{n-tun@IqPnS($|mU=or!VMC?@B6jX> z?B;fn180hcag&BQQSMSsvnv7#u5J;x9BiP2?XuSytvwp8Jt|nnoYC6Oj05ryj!zjJ zj!J^ilopm8mti;~Ge&glt;}w8b{-fIgkFmxBI3gjZR~xKp)mN)q+y9j2B}shp-QSS-SZ|PlfRQ7EzWX&(8e~~}#<#PL1~vr*{M9M3 zN%mG#;!bX?ST~N>^V!f0u%FC+@F(_z>+A|8)$9`@0q>%1IixUtrXsNQ4}Sra`u z(bENb@*Z5}X?6bX5vjC`#77iP?~>^$@;@E{HcOGU&{G&ay+u!;ZI;^P_pc${)fw9F zE_%|@lVb80Ab3jK!?sssJCJ>mXzid4vE=)08`WE1 z+xhAK6ZtOV;R_fdB9%t5Z#xX}59rMYc=OkNBERoUoq>0l;fOq`Gf1b|w=a$vXp`T$ zdIkLbcVeJp146r!HQBv^+%{=@nDfGCSxBqd#omywUmabvkHv7~(HMGhHXI7C54r@m z5B_b(^3$`P_Te4~Gw{wnC`=Rf08~~=N~&sU0V5rV*BTq@Q4GqQty3^04=U|X3OTe_ z11D#w!t^_>h`H?vZZIz&&*0#oApgk7$N+yoRDkl(P5kX?j5v+Gmo(_58}-zQFIB@h zCJW$LiKu4-Mp>#OxeVxCs1+ki&!!q>zO5`5?_@?j} zcM`(x8K-T#KG^cc9)wi(ys>51$Ji zru;G&nRQ_!eGGz9#d>-TeD(FBtrD#Yw`NTYE0+0qjafbyG1@&W6Qt6srEMg55B9L_ zvZX_uoR`3`zRFG^$74S}`v)Suc2DwKc)thvAK!s4_!^Bkejf%9J18oK?cP1idSXuk ztiJ~&YV11%Q;szpP&~UgdbeWp;z?b-Sm1=7E~`RefLQ0Zj;1Vtrb;BIB7>(5FmEu_ z68(53u;h)XWF%R|4H#JsNRUTMUl4tHiv_uY{nWrceK0hr$(3*xW1tz4K}s1uP!6_^ zri_6=q_6rQxhE9hN0El*q?b)ScW$aN7GA&4zy;EG zidZxh0Z)G(#N+rbLHix$QerO++ZJ-xKuB?|WYr$)_MRP!NY%2 zG~Q?$Kj@3UG>vS}p7G=DRTyP5n-fi&msM{6*mdyYk@U*a7~7-r>S{3UZx>&uW^!dm zey5o*Iw2W}gKn2>$dL9Uv5+6x^30WqV`qY&$iV$&MFLV>lCbHSULV7*iLX<0&6MT+ zTXXr*&=X*ZTRLPYSBpS#TX$!_#%u;>T(9`LLtACvO-<$(8L87vnKE{4@X(msCr*u- zI%USx$YIe=PP1pTsnXt=H#M8II^K;|yBn?cG+OO$$e_C+gN8ystg0x;zjieXj-Qf} z+=BdKNFAvXwK=O6m8J{fp+l?-qf4(ABtHlI1A?gJZ6t(jJjz2T;bN2_*$7Z7xgavZ zI253Jf1$gf^vdberx!0?jEtw0_uqei`t$<_Mm%)nyQ{j05wCs2^8jwt;;hy_P!?2I zH}_dZrDfa-JjPo@S#2%lFt2|0*#|`K*%w7%ofH7KpKl(WkN_v^J`}gYrU2$dGhrq` z?Vt~WQCv|Wi63QM^0FZl#@+54gI!*pH*e}x_($~W1INor9F!jisQvQ(?Rf0JERqa- zT+eojvVfgC{Y17CRUeVVu--^blaKl|Ev-oH0MHG|b!89~QjKYIeGY0!SJcB>U0KO4 z;oKIuduwx5WdZU6OC{G|6^IOB{nFY~R77+=b@k@Xs;Zo7;?V$^%T$KR)#uX}T%r2v4O2v>r5?**=(ITuOZ~wtaFbUfMYKEcr2jd|c@!ZVp5~w4vU7=r< z!A8eBA~uCP9cjGa{LHVsR#MVbOwlJWW1jhgabW^F*V+S`?SCH}(eR@x9V`P)t>|zx62fmtZP_yf2SyzYqI44|*90*J*RAds^ha<#81?eO8lCzFp?RvzyG zH-dlP)ygwRKltFo11TvjCyXNI-$oDpyXi$%phCNA&eR1>pD{yuuJXXKOR-3fyGOpQ zJ}e9p$>62+4f3PJz0}kTJ!6-@^wLYcy*;f~C#^B4;qsMJ`~JbKEg45Y0tR*KSA|`k z@Gxm>G3nghQPz%mCTil>qDIA&_e>58o7k3;l2W5d3>$_qMVK?_BEC<|pL+LWk3F^^ z=Fa)LQgnZvFv_zBGfF!mj70mY#5jq^BP2pU63DjUiaY0K>09otG+5 z#N&M>JST8cAvFc1H0#PsQuWS&WZ7ggOE2jn3?6QT1|f-+yoIa>5v?Brra2I1xI25e zdHVRkbpnhE9JsFTNJv3}OMk!+eYdYVX2QYoFWqTLevL%V8riey&> zBO(a9stEJ(G8t4wET3LVSm6x|Q;ji7w(-q2s-82$T-iu_5_Jqt;E5CLr@wbYg5#wE z*;%*R$G05MH?Ta~tyela#`ct+t)ajDp(02x)q_W?gPFuyNj_KeFCyXN-$_IVDpcwDFlvH>sO!fwEeBWzN<=wOxF z?_G}H{u?hkT-a#k@9-3Zk_~t(8QOTVUr&K!h-Z`T7$Bt^cg$7(2+tZ9`W*+S?Km9C zon%~d6W7hSf+nuv)_c8yX0%s^aAEayTO;AVMN2P3OP?n1c@=$V{>{DK>IsnUHJ9#k z6g^2}1EWM5O*A&t|JJkAxZEYlYr>_WkME{SJ_n=mBf6Rc6b!tSAPfjKimEwL>(V)> zmcrI!oh-DFU8QtODCKx|I_j6BhvA$tc;E@gGvrB-LR(cqC}iz@OVPN zQp81rz8j49KsP=Ep}$EaMh|_A?P+Nc+$L_(^Y>XW@V4Qd%u0qCUcDXI0&$D|x6H@6 z?!&pB#JTRksGrX`76FRbVg&0-GICQd=j9d_);1KCLEXtJEvl)=2QKUC<=jlH9*SPM z9F<@MQQe1#roX?Vi>^=PWEN4a=<4pRsX;(fVNy0gHmmFFdtz*NIbJ&YOG_*`-dbdxW5(zbIW_FvFW9F5rhXugnK^CZ*lWZoMx0@0 zW$v^YbC<7t`o(y-j^6=gR6E9amZ+%m_pf3Lz7*A)HZF@@xG?VCjhjU7$9W>ZW!n}4 zfGx-N3~$)7#U#2mHFfv(^$0d64b_=$s_*OML?^zhzomm}uPD?5G8h>OWq&`E`98Ej zOACk-JDOV*@DR(yh)G3Ykpunzp_{*+rHKF`2%IUL?ze*dK@a z^EFuX%%!O6Lnv~sQqJ{O5NbKMyq8z19L5BA_=h+v<#JXjDtmeW5V5p(8~iQZZSC!S zv=sKXSyT+QWV-C<$<&pgD5siU&sz z9X={z#XVQAvQE(l#E#6`)6xcQzA~pAKK7W68#gWw=}AKwJ}H@8^DT3< zUCqh84p@`fA`ebncHd$$vrQ)lT*kf@U%deu&#GE2*5ptJ3!62|G;Gh$ zxlqE)B`809`qYWzKOQ@A>U<#~lxvF4pEz>l(8vEe_|M(n9RrOnh$K}i&N3i4s0{wR z0_Zfs0@Y06B=mqI#aWGkp>cNx+gQITs8!}js4c#iYOmL~lon?f)G`qZV+TF(-oG5H3Pv0u z_*Y4OLr2i_zyp!6<-MyF`(%%_WzyXuw|YGw=d^L_SO42z{N0mW406TLyYC(%H+h5l3J;iAc2B<) zij-6UFCaGS>Z&T5n(FIozzEXO-p8@)`R41Q7K6-=P=tnOltNiz6u_tOB|w%zn@3a{ zSQ}8}Aw~+%V^C;lh_~M`!QI;l`j-J=I7CMY{-==t`UX)xlRW@CD24AqztURz%Agcp zN>UgI=OXf~T?f5&RaIF*W)}D`5Sm8DNH+R8=Xyp}HY#CWxdiX50RNjgnW8uWJq=Yw zBn!`;Jay_aN?~SXgKHTzrFtQCG=j+&wapI^1M&B%@GTW&ojdf|c0-bbrsmWSA15Fc zY>BizrP=jEmZQVgq^)qSiL!H3r=Cj+kSPR|(gYqzqtlIBfofWde5^G;00Md$7^FUv zEG8xz<>}C&>I)ZMeQeIOY10znHxL2zF9Fk3kq{U$bp*_p>$ZKXc0T5w%>0Xm*hlxZtHsOUDlupkN$U z0fZ^J@oEaQ3H@DC)!5Z<>uhdps79^Jii(=L>Y~=3fu7z@sMaz>U^!#33ItzP4wY}Z zdb=sefkzK7FgYY9^_Nm*9A>e&xsmr6wNQ~s?cwF3Q0d$`nWH!HNr8wh&|j$*fIoJF zj%7b9h=5UjfN=&-wP#?UqKzuEb@N)POwk5?sK;XW7!iX_EP8wfdi(~1jel)@MM)`^ z4>&ihDTHImb!lZN`}ZMC!aU@-+6xQ*QX~= zoG9z=9vDC!_wMexqeqX9ntgiYZnB$VIA@S--ZS}UWaOieDxfuJ`OEaEsb~F0aN<*%xl1=;3!L4ZjI7_} zuU2)rQDNi(tXWz|9h%r5hOwkIFCOA%vxCE)ELLw4^jLgfP>E^;niN>9IZ_UxrgR0M+eeH&75y&hAJT6YV) zTU=BWp7p=q&QIIhnaz%eu6djbgWS1rL2~tTGiKZgk$1>03_)r@@svYA5`70h>o?M0 z``9VkxRo>G;D0^Mg<_&mwClqyZpk)VGP|Wu3Zy;>zj!e{E7NRlXlQM%twp+59>z;< z`XvmT3s@r3uB7MZ5Vn)`Z*g-3ML4^qgIuK!3RfvR33l!8KQM6lGMrKt2L~mnG;|IQ zzUU~WJj_F9@2D+qeU6TO?3UhA|Jqx3p|_qyZ#|FRx(mHU{%?aTKVX5*N~O>F&0Mc%drYKxs~IU?A!l^oV+;^%_miYgSQhtpJ#^v$MS1D(X5rS!Z%`b7yCB za&qUD`Nn0{;!EcpkH?ubX8qf}UoLuC%&9G>Q!{D2V=-rSycwkQ_N z6$^zud$w;!cdo>>Dzf2@O3AJe)fK`p$gpWc`3g~1TFRnCSx8irAFPQNMD~J_hUDZ2 zkO#Gc$oAL|3alGGi`F7JT93ZVqm9#OVv!hii5qW=VR^#FM;~9Rj0Jf06D}r3@|L3; z&*74h-Viy?fKsa?>%aX3k0b4oq73wTpNKM-Pd;%GxlhiE?7Y$7>L3#a06v7K2a7@m zSh89z;)$^Y|6{4)G11=7(No5;De{rl$RD*Di9+uuVQHGiaGT7EyGuMl!@v(K@J+GcDcMx zl(n}5^dRePZwIKQy;5YAb{Y2sdfk9t--%vdgI-^QULTJI>B8|Jj-5|CfUv7nwHJKh zsd$hIQ&NIbff*RN?X_vB_4xf>@R97_4=oD6qNWBt-P%-t>mcyPH15RMV80y<0%-Yu z-a-(!m-$2A?TmxMj}2H&ffavu;;ne#^&g{6O$rY%J2}0*vB-;EuXGDCzWruKYH2Jq zox6muzIyXb42{P`y{0Qa`8bLho=tAOWptQSpyHXpemg{j88c?gTzngrz+!F~axgw- zrL+l97!qEdjSpVKY_8zQ_bZd60VzXyd27~8m=Fu-a~HQ__-%-3eF=}~PMl;MQlc+% zPd*uos4cP6x8s&N^M3sF{deARaT+I$p6>=nk4g23v-1b2Df90`Xgo|0pCMz0cq3ia z*B96>rQAgq5CCT{dDClay}TS8JUpB_teEz?@R4ra-NS~VR+V1U(%RnL+uv{FD{mQ6 zCRINTv)_)Z@c{UBJNoOoDjhK?OA+`et*k(W>4pZ7Gp?HE5&c{D9F5&EIrO&P0vm{efl&k-PVrYe(QGC@t=+#KlAg?XO0_Z6|^`( zA9QTa_n8G%X%?a{-ZR&)Uq92^(Gj{Krkm2sTdnr-R$)GI&OZ4m(7MkSl$87=U!0x& z3~7UTt~j4}An9`HhnmDikn+*4!Hq3#_HVS*CV$r%Y-zX?0aYTffn{OMd@SGrKg-rn96vsWU6DOu-L*i}I!DY{vU26hv0p@I ze~FdR;HpvgmgHCB(%>g7fqjy#yg!D$uJne2Tvg@AX0_I!z01mL+cHUI%g#(o%gBNf zk&|)xaz=3pz01rx3}7~NDIM(Fn#nG z0BF>+yQQmNgB3C;C`b(yNH^^Ou79HnupwIKx_a#rQjSskTW|BX_H)Yd%r4tg64k?5VB6dXtQ7oY1qln$iPC{3zh#e3S6%YX%NbfCxKziBq5C~}` zr0#ydGqa%x`uM#6Zw4~6vpX~Qo_p@Oryo~RGAB#UlhWOtkvIaEuVXp+r@Kp$7Zg7W zlmuX85=dt8^x2Pj(m@52_|3cTlzr=q%Sh>WaV8`#SX8x5G+- zEDLDndbIaBIysK^uBL|eCOtE$nvr3~8iLN>R9nBhE~h(NL(gx= zT{vV*nh1qEE1^&IH}n#Bhm}(9>-Qaf?b6ZLaJjGC9eur3vrnCr@UmROOJK2tl^!x) zwEP#W443sW36ef0j-^roS2TH(tdY5Ln?{D9OmMns&Ctqs*U;S-`OYT!PVF6c-rnYR z85`VLI=!xm^`^rW@J={5v0*sS=ydqO-V%4gl9lZ(xEWe@_iMNt$EE_@j^hUrE~T9X&__?HdEgKLp5= z1UCv#3MnapZv_0n&u5%Eb0#aRG%FYU;`zKRIGJ%C<@r2v%FAhWgsBD~lvU$&zAHJ=wtj-^}VdKVchn_F1ZTj(- zou?6mnatJ+>ge?JXm%n-(0sNFoNno^&jfBjhM<`}5pmxk6KAJzRaNoqEB2K#{m9Yt zacqNj%7YWfb&BW_QJB~vnQ3z(x^?P2?!Hm?$FX(jwM9@<^aFcjqo5fx!cVRC9Wh4W zetJgG{QO6)PFMTK&nY35m7O~~X|+z~7QKhBvzuGE6JlIFbkY6>-c6^hHlqqF>?6%8 zI7DkA^yDa_k3b5DWX&yY`|w(Hr(Rpa$P1OVbG$3-aoE1Y?na zv53c51Z8Fx7M@1r3}Trto;j0t`gD5w`7@_alDiV31mW$hY>tR%7Q9H|bk3o0Lfs)K zoN}$@b$mJ6oZ= zHw5{fiHsWz`0LOdDP+yjV)2INsDOn(6Hl5WF{r30E32vq%5^m0Aw6SX$f*Hyta z5&3~@s<4U#7>kQhQHm5q?G6Qzq#|m<@}M>v#UPVOqcM9Sy|l^8*Ino2?!~K}-Mx^Q zpF9%1KqZ_E0RaYC8KkfnNg1T6CuLBhLXRR1^7?N8+zo)c8E`iM?gqxGRp8RFsjPfD zpMD7DyWl86)%*1N)9Z7R6>~14=GUg7RCsDA{=2}UF_pQ(dZ)6sRkl^uCo5WOE(t1A zY-DWYV=&vg)C=-z))6;EetR8pL*AJv|C5thaN)v5626es`fOG~#@X|vvf)F_YFn!S z7Zi*3dK(-{T9}4QQRkexm|s-pp8yT;9Q$lFc^ZAYem&LDJ*!DiMJzOEz;lA?rSoZr z4;~db53NOeWp@ixZfC zC%U7`qQD)3QpcJd){==BA<{Fw2)Iqe2vH9BsQQ{Km&(ctib{YxPF~<6J)P<=Bk2JW z*;QPwu2)h@MS}u*w}`ejyHbZrpSJ7B-mEsT`d09fTJAawcP$Kp1=Nww+BPu?7D=UV z@Uo>Vzs=x05~fa_8t=iSf4gGovS43Dbv4&giWyMaf;lZ{yt@P=nkxepbtP9|{<90L zGnjJ%p+)m}7q|~up>OA#o0wLUq6)mS=jX%c5@5Xgh9&d&M|#w5NRSIQs5aEOg-v)r z9$)lR+`%o?tzM4di+AtO7xFh4iv;^|+$Lt@$Qe>r^-c{-5&G4tZN)4R9*^wUq< z_rR(hq1E#C>kSPpN@q-J?WJ z#Z*{ug?BN#8Jv;M1F1UeFT=6!QVk5`g72k_B9%(5gD%UYXl5=v9$4Q5<7ngcRtvAg z^BQGZ*TEA~)WEIY8l$Q?1R&~auxfGp!nmW1JFtwq#yHg#u0ozee|KkJ=}s5v&UMnA z1=5{^Zrpiy-c{oBQTGp|3kCfIFdO(2(!smZJqlpn?a>W?2#6=Z4H74pOE+}HQ}n{` zDf*H8mB}*~q960RfI!1CMD~{!WOLf<}BpGw&;Qq7JQR&`Ethk1#`74 z&ga!?A=f=YS)Dd%d-v!P**QGO=ht}XE4Gk>54=0c?juFvTG$cj<&ZmXiSp&7B zsUruLt^Q#zmb|pAvfB2PsAIbi=5ZcN=Zhb1K9<&sh`-jfW1By;%wK9&8ie3*X|n~6V|&Aih}m( z9o8!*M2U)oUE}(Nb?*ag!+|uf@+;ST2RvQ=7GtiI{+vo{xfM;X+#*G`(s}S*v%)(q% zbA9IaH4)uoqjXi}JHgKEEU(f=#dMEYyFIh62?O10{P#-yO#Z^#Spz4hXE zTMn1#Lr{9-O9#*MrD0NtQ{n#a*FQh|t>*?~_}$=?C`H3#sYaPaXoEQ;C&ARHg9JEmcL_|1SxK>#ex+zY`B|B8Y#p=n;F3)Bl-{#JU-0^H~ z;8t+!6!&x%5|qluRLFAV?Hw*CtLh4vOz|S|=ANK3QeUP+2%xsb?j{g7&}xR+688;6RBOZ#UW}a($?%ux=w6HjW+J+84!!-%W<)ll?*` zT++T2MO<2%z~aOtt~bmOo=hq%3=_l%v~mNktYT}RId1I{upmolIt;aff_=dI8QL%9u2cPb*OVUriN9Ay`wQ4<0{-Dvc%;Ts!=P(B=` zJo{J@wW4CpCvY@CWWN1%URb-&+G=s(Z1rU&C%0|=@ypfUE)yU)Pqj~SUa`Lwu{CC{ z__t4?>9EgHEIzt!7}#n!$YXzByY1kaCghgPg|zazpugU9^-?VmudS_YT|AO<&f(S< zS~}aD0-e2}T~0{KK8?v<82QMYIj=t&nZNzdrdhM@@71+w%iOth*PN*hoh7o@hanPu ziR|RUzcXXH4DM^5T+$0q1+g0fNh{G7#~shw2jNWVbp zb5sNk;5g)nkdngT8-=~82{~k?RPR{q+M2{6$V&x7mZyiDyVHoIkf@O-dFdc!7&bF9 zs<^vTUVQx0+uZ}<^>EFXT;z-<7WsdFvpj2O0+%V$bUtu-CUAKsaJdh>U{SwPs<{fe zD4U8SBG6REY(ixT#VMyVJhn1Xbp*K@^5CiN+{P1@VngJk!h*B_fQf}tnxeoSA#AHn zp`uV!B~Q5xX%qd{e!2a8`K6kASfcB#M|OUlif!fQJ?YI5Ivo)ax;1+%J9UbTxVjJd zisVAQ6SbDZh1QH?KSA;C{9BK{LlYAd2lng|;Gwbb0SVhKG*}E(C$>HE$hLI|PX1xr zfcp|;Vbf`-xsxgrw(2Lw`9d!{lVO?ekr!_$Kk8w zVNttAbXvHu6KXO@5zESFI%UJ6OISF`N!Y#asQZW!BNF0DqK^3D}OJ};x9 ziU>EAagmTuLP}(nqzbAPs4IfxAH{HPhCilcC2vKdvv!+>w}Fsx7D~U{(xRX;3o0ar zgkajz-UgpoX%qW+d3&J3Cgr9>5<;^lRbxZ|s*HVvZc++aN(4;=9yLeYOh$4P`oFwO z#;{j`VV?lQeguZS3JiM{7}h^64bgI7U9nw&4pUA&l{k$L8HI+BHFBfs#Uf5N1h zL3PXMEIor9zjAZn5txkS-eQD-5UFN?Q*9`8(}pU*yeTxoOC|8g#)RN*qpht?sdOM6 z!kz76JhEOj32e&<)I98lOn7qk;nDV2_<0*opFS;@0iJ1pUAtiQ=imRfd;d=BNQ6KS zmTG$UL&m)Okcj#vrLbrblA&~w_6hb`n$*ol4j+ct6=!Qi!(MyMV(i*4*yGA(2tG&3 zs;UCr)fV2|IU>wm;9;fFsx3yfdmxI1AHfecU-1a+(^-t?mSb0|N5h-n%ymKJbBMHS z;El1Htx}*$t6TxNiPfS^1(ZQwx~l! zKXm-~jdH)LnK9}%tx4=(SBDr-xG7+}LTFlA3MRYqQblcjWoaRDg;it5R3Yk~%Bj&= z%bng9Z+Ex%aNfRs`(=>CmX`X9XTBc}2i9J=OnZ0xM1IwA zKytKw6!y_}S6eeJELdQeU1N_z_IjHjvVh?j`t?{vX%|@viWgz}hd)jgW-jm_DikG1 zmHd>w5BGX$ynUGc84V)b=BA*8f#~ly>Fq)Af@ux#ba7I<^cy$({prXa7VoAG!%8f# zyPvXoidSBHPs+Zkc$ETtBA))22gnK1tfr$l_h`2l{I~a6S`c*#jAzB!vzXgY9XXlT!HyVBs{s%7&& z_MQ4jn{`}=i+CDj1;My0PtM%lsh~c1Go7k0ee`U*s z$g0;=tCCTCawV3QFoQmFOLL1w-B!zEj)0-Sq4 zjdB`K_jV4d{&9(@ot&H7?q}9fvXI%cSqW-m{uI&UCWlYi&NniryG``;>J@*ueZ3}H{D>{DQ@X4H7Lr~%RLYJP= zQQg9u(=&KJ?|d77JvS{azrmu`2X*sv)AI3PO%|wT^!MvlyUyVvr`p)m&2PXAq%RYP zLWHnlGU6|+BBf;*D~uSZO(|HmB8is+u3=4;Hd;7=j)aI%Fe2S+DFfyS?nvN^lEuV$ zx7m5*N=1GjJIc_)BNhy|%`Rg0!qr4@#fl1l6gw?Da4JHWiCP@$K^c8rk-!%?iIlzW zMrQ*M6}B;QvnGcmBk&Uot+df<3KXU*o>*OxCzb(Q|DPHlVRASyd3Z2tO9m@hI50UJ zP5~xLN(TQJ*=vSm36q4sNY;;Zv6)$U*%$ZFcZ@^G4T3;Hl#~d!HB}}jSJu{oh*#Iw z(fQFNxS)-9?C-0OFJF!n39ALCMT>ysMH$yrA-)#F>#xsweWdEz=F~+;2M<0fG9&7@ zO&r-1bUCo+2s$5w*|kH;Ku49WHX&0s60xIfy)K!}6r5gsk@)GVMkrD(jzM->pIP1* z`p}C($BzezCTS?GMWx92T2WqHcKHApZcso32SKL}3enCni~^vI?Dtd~-pk$7FZ9me z$Q;StZOK=qhz7-YwwA2NE0InB2b~lkitq}uiQnqsaJvOPn_3VSS`l_M#8L{V(1IZ7>MqrlD*-9lmDo(?Kb@r{Hj_SI3=^ zBiZu3X8GQ7`QCfvdn5kKz12VfylI5&F)&Z|7%)NV#zmuK!T$V!Hr zoaFwtA>+wE!I_uueL%jqK)$z4zSn%$dv%x%a(rsL>?`n!><+RH$r)3VxeQ3VuTle$ zW!drOFwCV$tO{MRDo`f4NURF3*PDQNpdpuQt~NF`L0!6A5RgJzYr?_uKPEalin-y1>91B-U-b-q^d}08ty`39}<7*fc@lR0=_B3PvU19^Dnn6xZzYn5I}}@(W8Y{(1adI?@dq-Q0Ap6%|nK6VLDM z;U>Crl+6w9weHLW7u$9X3}MP{-TEX7x?w}$34c%DE}gqM^cE`gOog72--rr5Q=w-9 zdXgfbt@tyI{v4z~VS;RTB3L{Cx)Kk%;?U{1xk6E)L=IAr3{V#iaUlv*g<7#8GGgF=8lN<+b9Y53lMT9 zA!H`pMsVoCCeh&p@=)M#4~V5aKPW1eveHphY)3>~GU>`@Cdb)B$!{fds}u*T=rIY4 z?m}EqoFMQ2$#izlOk7(hIJ{b@C3kpzNiCUyw{e-ci-;i{&*C^0HRA+ENNP<7h3V-1 z?RGyYs~I`FQQm4`m6VI^Mpmkm0Zg??qKjUT6TgfK}Z^as>h-VpiZ?d9Ver+|w zYh_B_5gU9)${ax>P&fDns}prv!wXG5s;H z{}Y7a_3$X*(My6b9s_95xw82n`{1t-?!XfTAbgVoNoo9A1TqpsNW`CsLW&Hxj^E!- zu_QeDpK|(1DFdZD6!d)Qg*#k<%7mG8mQQ~I9e3QFe`uF-2RC`+x2a0! zN-28)JVN&oKsnqi&s)7?uIT{*TAPsGhP>#0*^a@O3C!VxPgG*A1Do<+8VeN4rEC0& zH0c`8b?OYSp?MO;p%9C$JQQlH|XHMz+%x z{sfj19Jx-cf5(g>*ix2>+v&nS0Sa|Q@Tb6A@+kr=E(k367U)vH4Coip1@hLT>Sg=z zP=wV_;gZn?e|&I`=kDP=(3cIuTtR{~-v0d$6!kA-4`mr8zdwS|4I0UPvML^R%p6h|< zx}b%W13DScO{C}ARdxl4#V%oCvXY)n*$F5XqdzWC2%z)=ZUzS$!e!x_nqD$*5q@zW zeWDjleF`vLARxVA$4GREY zSV%xcMR&U|zw_sBmWm^gocKZeILH2weI_p~|CZ>7M)0q!JOr)fk~5LHk?m8o&0=xfJVDW^(=dl}QuRP+Wd7MT4x!NSTV`f%4PI)i&W-foMP?Gi z5HJaqI0^RT+y)6zRy0AN#8y^{R7mhUcZUT{hKJ3eH726KNA4cyHjNTlK-i<1Hc^A_qI|;NDR@6YugGtZ|RfVKA z%7oDn<&@!sMoHe4*PB{xq#(k&Y;lKdi=ufLNF~zfNj5(e7eZ*NcR<$Z5_|{lzD}2VXh84So|(%e#K@{`Bl}<=nYZQOM7$+-_2~TU0hfWcD?s#oYzTChtW- zohSXXBQsZFSd*9jE->&fs7aECk`)`AUm6&yBC$mk+-KZd6odOQoXVWLC%#Dr5Zc!w z%lD-5z05)JV3Lv*Sy`C;9|-YmFI&&O+S#H~nqWQD85Mor*74N_Gu`4=OMg!{}y(=BBu}B>l;2q$H;ETWhLc(J5y(RDJQRr!h@2yuh z&=dk5v~77vgATo?WY5OxVMd%5sp_F8b7(#xKOuT4&K{@S^6d&7-&Zd9^pjL#rKhU? zEGRy^aX}*TD#TG4$mBS?n_%wTxnKOy{{8#MB@XD^xx42zP_=q=MICCPQ;?Vk!?SK&mD)80?V5siQSPHDXx9|9s}~g7NPwQ7S6G;rhk^{y zaFe#)@woufd_e(Z`#d zt3-W8wL5&D%gT(dRrpj^<%&;^@09eN7&$H`W?VP(#OKZC88gh?o}V}=F>%u1h$(ZT zqTYNnV(^?PGw!=@#+cr(z*&Gp>Q*d03%TB7=+i5pR-Xu(^z=@hE)@p^iTdz>;!El2 zk&%~*gXmKbeR6!Kr0>LD&?X}>eNk~iL2)r^LO{JtW+2DsLTC^Q3!zUaF0o=*h}ns0B28L4AozaBl8Kqq{pQNI-k+3dOSHvn$jOuB41`hx@*V z2hlQUY5tez%M{EPSP%g$hy)h+0tg11)WBv(rEBcqY~rUE@U6;F((NM-V3mce~s|q-oVzU1RaIF%bTb# zFSVjZ7xD~U0lNdELdPcRz04*Qo+&%lP6m2f9rh8&V<$a`*}Y27nib7uB6}&0E!Wf>EA@EzeKti<>4TLS)4yg}` zy0Quwd!SYZ89)k370=l;yr}c^Hn|v~NXAsZc~ypAC%~^eS`&lTbOQW30e4jpWcR7r(_)gf75bVo5Xm`8z5v^1JMy?uP3 z{Jisb67(tRi*&~%HFRicutL2iFrz3yI%SEX5C^L~JA{vQDAj|*R)Lkvu_9{ROLVR0K6Wj$aXc-3vZ_E@}aYcS7t1)4y^{x}~6;hj9s?{+=h9tIm}h3&j;vuz9Z z^EkKL!@)v4CG_o`mKM`HCN0gS0o6tuT*9FL4DT1B(`Z8ajYpc@@uzlt{`u!SP8C6O zT;%voIQ>RsBM?2%!iAzxawv}DaeVLn7vZy;&8~6Lv?HuP8!lV#Bj3$s{4SAuiu**` zpTc=4%Gv$RbtFQ%pIgLj;!Xi-*tk;%k%LQ};DyD$4NxN~03%0sFIe=z0EGK`di!~M`udrS9!QMj zZUjB@3kX1P6b?S#zJ9@>fqpUn?pe~zAxaSnZ0rMU?64!4Htg85VdtJhhYs!8`RmrL zTYm*6PZqfGh(s?g&dW+i{)qIfykb!Yr3X+86m~!x7RE$GlU)R=` znX_I-WI|@_F}66iiSKJN^_`Ol_oOc&eSEIi9UP4Gr>ZY=IWG50oF%(nDJW`bZh&pC zp&6bRd^;i>jVDfMwPj^hwGB|DH`G>@iLQIsor3cse9Y4`v%p|vAtj{I$tgG(8UC&} zL*8hb4?azF{Qs5H1y0y799)+k8| zd**We`QJ7lKm~$l?9&vVgA}4H&-GT%Ok~VMop<~5s^yuU4RtC}H@L6fntiz@cJRbR z1ko*p^zoH)_6SfyBkI)UWik&ShZoDtt85TmOG_a1VI?jtECmfLr~2APX?lVLf^mV! zfaTT|6||}9y}Q}f3<~H)MYHexl{Av7yHPQ@42ex5Vp2?tq*M&;kl{&Ecod`mFh<`Y zDVU55wi9XCqywigT4hjmKx$vI3!x*yNd^;RT-wL+2wZ#GYwo^{Kc-zNJYov#t~zS( z66$iWp#12jwZA05PySK+!-}OK!UFGZs_FKsBfYuv%iU_64Dcbo8$cO10Y3E$*)N#b z8`w8)Qzs0wwNzC3jv6<@ucoq?b&GrE(90U79ZhsnxDQ;VmUFN|m&P z2bk1%z}{$;Vc!9I^oZyp0bYhXGn!<$cYrIcdE}@6@phSU~Pm5qC*WU|0tf zO*W}j$yf=buPjiZ$XZrEK>r>6 zmW<3<KXol-4iXv82h>XG8=1aEF{K|hGA2GCdR7`l-m8LGSeTI&@ckH-{Xy{xt`lHdX zbRn4!_F33pk@i{kms#O_6ytC%g-Zg26OkU{E&9Fz#djL5@OMAbg9=D;$dhlo!|;m&^%4~@Zemw5+v z{J8IYb;TLq9(|oL6$ReSrK;=WRK(#EN|(>i>|Eu9tXqSJ&rSqlti>!`rk^!(aIdP% z*?D{n=AvM5^7LycOGoCWylOuWZHikdB=}1tz?Y(m7l=YJ=AvK#!6K}ng_MJx(9I07 z+=rxcvllpA@V=Ofcl_#RKPd-U(ybKdwAFh0yfynYQhMcNThE_H48Ku@`fwJ?= z*QDdGw~KMMAGn=6$J%~?Th31VbGGkn?Kpb@L-Cw_C(ggaIyy~ICMKRg58JHrV(Os9 zi7b_++84w9X;A7#F;1c3UVnWo{(b<}+*KTxam&~hh^1HAYupD44c4`{1?5*?jTkZ7 zK9D7{U+v$sP4@ow{`fl(=f8qka;>*o>#M9~^>6|#EwfhDBME}BwN|K5(LqvsN5G-;+fU8J=<*=a9-NbE?WvXMINo*B2A2){!DaiH&Uj7B*&^nl{oA3R<-AOhO zs^CaA5Qm)&oHua@d;xG|Z!(C5z!y{$R2)~lrF%u-W5Z*^A5G?}^#WfAG~SOox*QxT zLA^tyQ;A|s!c*VCQ=h?8CjkzV0EZrV1z3%8ka8ZfEj;$34wcG%(Un@hvTpku5lhz@fFZ z&U*HIQSfxTeYDe=y?eK;TD5A^-u>5y+wHIQtb(t5eXH0bN%P_ZUQ@##Ft+sXV?r2F zBO>>wjT|!~K0dyG|9+^n)IG-aiHT3lMB<)D2gCXu2hT038Us!u5azGfYu&=5`i`F9 z=f~VS_3nEQ_^;@`G2U%{eh($}3=enH>ct+#$b5uML?u{8A>=}`foxMMIq#9NT-xvp z$*-!km>ot<>lCK8gi6aB7!o)Zfy79>HB>?hC1@{l`OwPGi9OtD?KLB_JcL|0^BJj7 zsO-x}R(5H_FC>>Kc=JX4FVB`Rf$VcVgXI_)hs>hlRXi2&gPUglo)l*OZOM*`{a^V8 z*wV_qS`HSn`SUjwAsaY9zV1S|(SpXK+l40&$ipF$!Pk4-vpD=c4UnJJ)F?iHH z1IJ8!Y{F#Os{Sz$vZ&}l3XsvwEi~ok%(%PNYY>c6Dd2Oa+&Z14Hh0*bsj_m1O-YrL zY0U>xl7ylVRo^W%H9@DK4EFU67TL2%R-tU^2J5x*Y z(8-^qHxNHM3Q(an(IL6IyU~EqMv@FR43)yIqPz&gW$&SMSTeSFsI=Zy&%J83e)Sd5e3sxxv82bY zYx+-l=9y=zs;)ZuXs(xC$6yaRabk$bd;){(_iX%O)0VRe|G+^1P6-b@@W6wKU4jGC z(&8!E8M%o0!MkI0fWG0_+K)f}7@okMe1pEF0=SqN9i1t}Wb@Rp1CpYP?z~HAW*@Kmq8RG0os-`A`p$T@&BxVrxJv)0mJd^W$ za`+JFy>+#TF)_8ZF)@j?q8_9P6&y&bgdx`1F%DAH50q>W(@~Xy^g&Sg-1?0)s?^3{ zwDC554r%M4uftzxLXx#M_$qQr5He928tUW}XMarL6dEcWAOPqB|Ezwg)!5alTP}_Q zZ@rwdj3f7HjdYG9$iGF!b!(-i;+DzWJ)6NdC(Ztcx)ZnpN0@rR)) z7iPHpeA0zNS0;}?$eMQa#pX>Xujuxm;`__V^7yme9peu-e2l+G=bkek_T+{Pe^MI% zLV5fPG5#>@mB5Gv<+L>Zi1?!MCnYPYn|i=71tqjJ{x`po#{X6i7)M3+jvd`qEDEd}0MVa-80)0h+zeqaD&-w=#=HzEtf9LsD zg5fuAKR9paBlqKx{Ntu~Yo6!7pN(horA6h|qkNl+U}$#pGDXSf^0UbYhvXRIuR5&v zum-&*aJ{@CwLpQUh+|)f7Sf?{%6Yg{Cvf~F9JeAbC9PC(W!Ov_THXh&M`m#9gIYHYN zi7=gNliGJRJ<5qAy`IzXa zMJ?kv=%+%$aqsriXVR|LHymB>z;R!V##h4eKtun*69PN;dVIVC$Fmz6vLzh1g7GZE z{4OqX;5ewIO~P@L5h{MnZxn`>alGT3e`*^_Rgm>9kLXV_te}`KXq!#c*VP?9eEe`- z-SNZV8FZ$yya3NG9#^ka5LJHg!Eq0oOyeG8&w?t48w+iXxmu@iqU2aYVGb)MvXcz{ z40wUINt7JxGt6G@M0RfdMsAzWEpYPzuJ`~~^i&&J`cG{W5a4KYJnZmQG_J1h>S4KQ zo&kAfT`I0z^$_(Bg8rgq9uLY*vlSY{8^fKnxuPDevP%D{X$W(KUPSIE6&jZQQ_F7o zLYgVTXjek8taTe4tYAZfIim|kPX*O1m{MYM_+$gGZfiyYZRG7ma7mjBw1~CUl@(Vm zQ%%Rq%-s#YqK#Z!!5K+6Y03;?VPS@{r3ojyVA+v#I-v!DqN1& z(yK^4A*rTuRJg^&#Ky+N^r9^~I;L0e-Z7KhYAW+G&Ky6Jfr4y#&>5!ZRl?+`*BLNi z2A$pj3!Oq^&}$*3X!XtrJTM^Hh+k}M*T|?&p@=OF?GzQ+HMUQmE>WQf{tJ!j(&uJo z$GdyD3@6!7@gZS$+h4Wrq~q_ni#*!Yzy!FhF2)Te6kSBQ`%86NHwe!@2Tolm ztFNzjiWoF#kRQ*QiV>;!$6C|?+$1u;(C1&6{`iFZ$Hv{$FLBf;g;MM0=M&nqZ||-l zY=+&~q@#DwGK*NKKOGU*pmwMP4TnJ!=5Dg`-fQknN z;@#Y3&NSlo@8llSIz@hGBqRKFt%oKHbPyG18QYhLcXnCY+T0wQ{1d% zs(gY}xZ=ng;XJVX2E$J}e zumceRn%zIt_EuOH+L^lQ*zw+$>K*F(uj_N_bC}xC<6)Pl1OB5nK5y)6?90^P{h>EK zq=qvwC%iPo18B z(fSJ%5(2-f-_Ma?b6*#rZMWC4AGkvgL7nruz*iPsww{0@N#K3m{7hZwOv20m_EZbk z!sMq~pnJRdR0}lk$?{Vz0H6F+I+LE-p-ZE-{zaE2w^gCUB$pejNIZGVbCFXqkvUqc zXa_dP&85@-{mFxm-uC3(@&CU)dFWrB{EuxW{3Txj&s*?Umv7C@24Tr2Ujan;LAY$Q zi@J3GvUQgOvo`>vH6b}$Yl7-FWr@1rx`bJSuLjOU)P<%tXi7|V%w#`NS7a?I1Ia8S zF9C#fLVV;oQRimz@C5PnWbhKWaaD$0lw%%sK|7*A9il+z-DRm<3MEB+FolkyMTifD z_d*IXVIj&@qvAw&-@vpJIptY}1%;9v$1KfA9XIhlU!jrl%c7)^pdwv9C@HgG)fWZQFOym3(uuTEweKV5hLi zvND9bAPf>+;ZdrD=TcBX3O3>9R=WtMillcbcoA{56s(B8-SIOCJ3`RIF~PDnDFi(X zp@>7`UnEVE#J_+GLEIjA)jX*FN%4wIc>9yFLM!YRNO)MRAb|1xdNvm*!VPaisRXv@0IBx!IQ{!*q|WV6?wcX+Z$1PT4GByOcL)pD+?1+B z;szUn;*l@|wbG$NPT@Yx6ct(P#9^om@rbm)?ELA4#>UH%NU(j~r-^Vfkaq5~4=L%( zR1o8L1R;p4nG%c#F#iDoPA)z^n>YV*-f2J*+&>?gQg?3U%9T*^r>OcP`C!fX5#Uib zvu3WFv_~O(|C<7vMD?maQs9QsOhlfYpV<+|zMZ9@;KvWiaS&StE^-1qTG)3Mm;8+j z*PXt~aUmfguDRbXa-^qGetclU`w7UMfjuR>wT}8+4Ju7XX=}O$NVK%J04#83133q0V`q_HRfmXvT+8=38zReC47McTMFd~a9j$ZwY}u#j7yfef~8 z-ma_`ziF4BA&poEPeN4mud7LRe&%n<$~KH6@3hc+^}1x`I`W*l`gCACTg~PsE61{I z##_sZDvrVNo)$FafrI?COO_E*&fl9u5Zwuu=G%3^9g{`TqMcES@T%r1-!VfxIVWD6UKxPRae$jo>jsqW4Gw-R5=(77t9SBLD8mc7JOh%ME(*_Kl1j4tmEw&>>K16hDfinF^Wvr8!kXG%dvn>3ngzu_%u zJ4+FI_9E|L`fBZVP5-L>E}e3>wx=+sL6Hh1a#KVOy8m@@@x{))e*e9!m6fvnG?lRL z$9XH~#`WRGNP8cgzwt&qxTd|}nfeMr-+i~YxY-noJPom?=Hk8Ity%L+c8#V>TwGii zO-=SMYv!rzp6JpgBI4d;=9Qv$u7|4&(kK4*WQyFkX!Pv~^i4qDqJfXmjPvq@VH`pe z+7H2rq)_?`uX9hRLO{1+3MfS20oeoc;ap)Yzz?A}GN7g~L@Kh9(SIm4Sl4&oeP{Z8 zgF=E{dG+ZD10gbpg$K`mJ?XaQQ z_=Z+^wrI8btJej$_5gR3F4U+I&~J9;-3@4(pwM7}@RVQybq*zdk}M5HRMNUoDfQlf zO4gE#39MZN9maV=!Gl!-#s-QjAd7>Wk-YBy?)UQejKKJe#riRYDcJ~&&j_p^A*6Ga zHW~}cc0kK1or05~{09YAF&XkC8E7~QEmoWW=9cJ*pmmtq;Q(mtKk|{ulP5p<(7pE$ zfn?jSpNxYaxIA-zA5c_HQ18KyKRu{xgqW;Gc1#;R1jeLee-0PlQ!QViktjLo96~TMoiQ>F zodYvUqk*xY!NA`_QB-3#&^F)b>pN@K^k*JP>WvD8y+=Iyj40}xtI@COW-7&`61g>N z5@AB7{S~cc-jcQJ)-G8fB4!moG5QNMV&%F z3B!PKw|pmKRA00kEIGu$o0Z%A2y-PQLZfLTaN~ep$vy z#)qc?fhIrm@JNuKks}{|B@qc4)KriD3$epY)o@T2j$3PFXLgblGqU%onlN0WRF9pN@hQKOIGRUxgjxIe4i^K|RV< zt3Fzm0Iz~M_UR(irZOdj##%vDs|RWI6}48VnGyX|RG0z+U{`3N>;~!sC*;p^>pP?M z326O9w7xT_OJ~NxmmFF%nGHdqN#GO~0nVL334AC3h{;(3$Wl-M8UrsWq~^i$fC&c= z8W5+zu5nOrjURt+Ulfj?@!XK!;qYX+-lPtW9y0YAceM}Hi`=y<*BW&e4}VwJSfE0z zd*ijd;~*txs;WvsN)!tgi4+Mz`@8&t1&dcLNdTMjg8g+}O6sRS{`lk8&*n;54rke) zRV?{<89lMJP2IvaLgp4&2fNU9MNXNItQtOq=o5O1YcCi zumzuRnod@c*-{uc*Jx8fkBw;r1G>a2NK-20?YG0iW{;cz2#%XLvhRZrKm1VA;HUx5 zzWnl(d%F68Rc-PT6%{;>I38Q04F)wg|H5l_+n&9-*3eEvhYk&Aww%2p`A|N?eeww< z&3g}Jf-TraLR0-7Xv<5$m^pAZ!#c^WSV@3>k!F}?$%@Z+VEar|1cPX*)Ec!~!CTz3 z^MgA3X*xyqxhE|b9_2L(wSSZtLKMQW20@essmq6>v}6o~%Q;Oq^iq++>~oC@8&O^u z6BNJ?Ac-{PLTI+ptdllpE{R2Wp#tT|r&6{w$ z!UxM`bV5QxSCMh65F(5Tr(L>w^^c#f1QU@YQLE{|$JxjRz67HQGf67fiiul)4W{;T z(h(Gu$Y_TARitQ!e!$H`#wW)Wxd)Nx!8r6_I(p#n3pRTIDX0;cT;wG~NBr|ZS@L{o z;(?lw!T?BIoS!N^USUzwTcGG?;T|3{XLgP3*Kc5tZcz5moH^s^`vyhQ=dNAx`K6a$ ze&*h|D3Pf}joepth+t^%ioSMI^T@%h(1Zyp>2nV_gG+k7vG7z{g0q+Gv$(l1SH7(P zGL4dQ0No4XJOfU1@7a@qSg&&Puv)#Vup}oZA&&F`(C$azxBDuyh^%P=r!ojC0|BTk zl~bc=0hJTYtrR_q;OHPtN)Ri|PIRm=uP9X-Fq=h9<7Trpd$_^rL7I^uVWf=!<79CA z>#wABnPQ8>0Ml^5)QB-RGES=j2tmtug@(WgTV8>M9I6H8+Sc~K14i(aMlWBrIyBTz zrvh+NRO3d*$H%(~%xx}n{qxT?YYO)5J9u(0)mb}qan~Q{kEpgC`~#Uw$^@pEtD!!? zrdMc@a)oH9u`uMh${(dxd7)KZ(W=g970URsD8$A(0Z-8`TDLi5JOcr>A>9Z%OqO&I zC8!!0w|lqXqDM}+9b3NLw(ZAVCu;P1QCSfYQ6V@V08f16C{Lk|)+0bwyfM*g)?s(_25Sb>Nsa)LOMnRDa4M)A_xQ7M>`V24 z*q5Jk8aN7^)o)@CmGR3}_NDzJyuGI<<=fhKUAlV3~vVwBr zX^K=(6P@G%Ao&7xT#{!I*{WgNg(uTZTeThw7N{-OEG&c69+2F5T+*9y>^nu=pafW< zxH(A)NJsV)+axMm4z#qOB8i2yh-2*WaMSE=PXKO@M0Utn1ZsrZ&Gtx<*M=f zoU|gmbcoPlUBL=Z?E$r$2L~OZURXn*6Y97qjX&`TH2$F&|6BM3@{Yr-(@JQGi7nx! z!ou8awuuu_#aHEG_Tl;PP(QsED?Mng-9{>#dU_|?1L<2x- zK+`cJ=~zw&B0B_F@&Oc6C0-juU47%MT~y8L)_v%s@H_vQA*+|iBacja^7)sd(=W2{ zz7I^B78()|9MnS$xNw1_va+zQa5gtKFqSeAO^Iw&|eE~M1 z2XxG$1J0#U1b0BaAz=xBT5g;v#Q=~Su+bnlbkxy-NDqxNxt!kpCK)Ri0-`Gc(UpMc zLO^sOAUYJU!$zy83`v{=wty!zz_=o9#Adrf2R zH1NKJ*of#MqeqYKZt&_B57vIv(Aa1fNWo!Yk3T-CPme(SyiTKW_v{*joNHotMH@vJ z-H_YsYY`=HlqB}_?DkekVsEakWypSbQ;4jqYd3@U^JlgKaDEqa^sQ(UTm!%O`sXue zmaSO3^>+}rwcngc18_199{F_HhK;{(`Uzn&+fVM^wg$TFkC%M*%g-pmx?;sgAFcXq z+t2vB)yBaa_Mc>x%FH-Yq?IdkSbk<`VYf8xmr!+OIt^p#h71p_dV{qKMF?bn{X z_n{{sagFUYWzKADb7oCR0^^e~e9Fs-(B9JivF7Ec#^bht@nWZc!fF*oOrf9|VRefD zpN_kWw*<;?fd$}cs0akPk~2j3=*4tLT*QKA~ryG8O3F!mj(@kMeM!297tEBMFD4pg$jK6k2nPv z3`BQ;-4O34f8U1Kz}cIKjWY*c(~3&s0sePfL3BGIIlfWg%m0V>OY0xiv!8?Yc_G&4 zfxza0j0=EPM?z>6-P;O?(xF=@CA-%?% z|9RpCXo)SfbI&Bf_yZc|T6sI^h0gm9m8nfaRY_$EUqj`RUpK5slFErA&IerA``pJL zCqjuj37q^}Y5}!imaSer3?E3#FvI?$=HpcxHf&h42$ch%c~{J{h)Gy`-vXA>{x17i zR5Ym7f@|pQ1Xo|xc!%zk7#3M~>TfNEFfKu_utR+cib|J&P{fqda)wYrgatwPkqIT{ zJvd^2lRLz7Ns|f%s;o(sRH-CSy8h#>GTkEI_s22PPhzBR_`cIyApDTBLkR~QNMM{B z!okS9PF93Lct@@lTqd75injz04FXMr8{uyQT&6=6<>6%P+tD^ur~{ zp8mr?h5-wN!Ye*(5^XHSr)e+6)lI{?0V!Ms(s z0^R>_fnHqSa#=Lr1@Kr79WW(!4#J=edz>OV<__?KC;gjWOVn&C#$YJ|!Cn67kowHNK6O`)Em)QkU^rf08SPx|c zC>RQgb!^bOw~NsehmRcU*LCp3DQ{y72@4-^??Vqye){RBr%j&n80wCWn>cRl&;gIX z_~N7ehQ^|zgL7{&x&`*ytW$sd@kf4Di_KPThj7^1aP`WS8YsG%&ZYcHL+hU1M>B&v zAsdYTN@d1wQ8RzZich}#)G~h<@SikSFCwgY?b@~LKb?!rGQ7AL8!0K~E}c6J8ZQ_j z@b@i+_}MbbEJ{CrRkPx=EqnIt*}YYaMl4jlJ^~uk9zl>C%z>U-HOx&~5_r`uwM_=I ztEOAGnBHkQP<>rfYlFLq(NO6EZ15zcjlf0(qGLtuxF&rot)e7@BQ(Gc{R63TEcnj} zWT{Ans)@m@W~b070F=NB`Xff0Txhj3NcpE^r8G|ERSzp3Hd-U4ZzcW@rl|kz1rmk~ z0)~)XiE{c30)`9%h6FphLBmhzP<5G9O`y-zM=bXRFaaY=Ma zOn}9CczjY!%m7q5b$(|;d|y{o9KG5jao9`oY#y7ZN=WF@!!<-qhLf2~QIo~X$zTW@ zI(Sg5XJdOg^eShqJ$k@<9L64}PnIs6KY!uk6)V2qngI3Ef9&t*R_r+V#~-_P?fV4L z;O`K7r26o~jT=ARN-dgea;vSqf<|8Rh`gi@4h}Xti3$Y-ieLwN(?HN*oxeybCiD#- zNdA*_3-)=Ka0LFEs|q{p%nFXASuwc7Py`N}V6KFb#@!u-Wbi@Kqr)@+iT|$mN)RTE zj~^iH2MCkK#|sda{O4#m6|!us`>^Qt!dzTuPYRTuD8mA280V!BN!lA|2aZy+F)Pk0#d#q%XkIU?7<5h$TL zE=z6F2H(uVs0~Iw#bD4{7_pTk+Kclz{xObc+_9CgRNxu}WN{u-lFTd4lkTSohK}s8 zI`jz{D`Dma%#;WMu|mrtDBW8c$0x%@Sh@fE% ziE=I%4?N2Q@Nt1;08En&3>YPmwJ*~So7vgpwa5IrbcqK&OIAMlqzXE^)5{p#S{|{_ z&@5lL_3+M}^XJFYzyV%4r=@K(@jkiCWs0v z^I6O`!Wl?&7=B>ewr$!(F?#Qaid^QmSZ~dz$wIe7#{ja#E!LY*y;M?rfa8`gNpJ6m zw@(4oW&vvbFn;~OoB0umgKm)ywBXZ$R4NbvIE5drEk&T7#33X17QI8FmvkLo$FPQo z6oh!=UMw?YiDEs5dwLGMAL*WFj8Us6Pae|`*zMk}+klZtV-f}pN78*Fh?Je{)mQPG z7oHNGdiJcTsnz-QiHnO%@icpQc!h*Ubne=%dyk%NN;hMh-VUZ17L;mzTLARjiup@F z9FE1|P07agD)_89%0q{iFNdp{xA^tfUw^v%qlMz4MJrY=p1*eM=FOY8{P5j(-^Js{ z@A85mSS%F80)u%|RW`bm;@Q$vgSf^km#eOnmz3w`UOb;$uTZwaq|(w>pPN}G`qAQ$ zkHr{t2r>kXmAw4Yb@**_2+niEii>~vV}tPoqu6G*AsYz4j(aAgp1#W7mUSBm5%8t*VD&}dM8FVZi0%PTtu4og0G0B~Y7o~+%L>K!(A zd{W=&ZWx8=gx-(bH{->ZAMev|;3JPTqW!)DJ%RVChGzzn?qzlTDjMpR~sMI-HUW)-1{JBkq@)A=B7|`$?!dev z%!ITc|8Sp77ZQQtRC{$IV3r6BPsD_bq{W{$NpXipNW2pGSb$ENb(lB6>canzy!Q@| zs@ndC_c_xiGnve!O%jq!LJGZv4gx83=}ov|!7HfOUa#d|&Ey0@KtNFt1qGxiND%}< z=^#xAJ&+K>r1#$2%7*~TL;m0J?dI$LXUr0)I_(Tg4wRX?& zzCArDe~V6|i5Nb7Sfrn4b4i-5ngA?Yns~krO_VJ}e6{+=Ez7V1SiASk*|SSN-@NxY z?zrRp*`r5~p4_(SyY=fYU);L=+tosRcxiD#VL{d%7y=qADvE%LnO|maY*IAab@-|Y zEeHL-mKP#iGL5JT098mDf$VG=!eaeJc6b-nq;_{8V+(TG&_=S20|*uiBV1C8sHQ?f zX;E5%EK6fm6lX|x1qX${zJ*A+ zkBo{}9*aUiL^MNTaoC!nG8E{d-2e7UsqdoEch92lXoieN-$gSnI7nnsG%?EYAh83f_h6f( zH;xNJDbZz6v|*$XFQG%~Nt)B~RS=2`h76Vm%?OULV*_&*5Ana}H2>hR@bIqv2ThnB z6Etv4^tAUs_+U=YY0**0ztXSEfB~`7(RVXreZfKCyEFa~L4oPRSEEaas7pO``oQ_~ z=Zlpl`e^}5ZQH}5-g@W-|5c5ht=FM^LG z02^Jru;Gi%P}3jyWy6Ng<6s#kY@64#pMJIO*s)_PR;>8=V?hCj&4!w$ga};_EOt`TES^-90=e zPmUWtq~{~iEME7~ys3`_2S4-7w5j8VP9c!qRc;^5o7|_*+wVL-@98PCVhFa!?V}fG z3>x&oe_na-YA*pPM&#vM}6aqCEC8lgSUx0s&rLJmA52FE4)p zA_sVQ$mo-Y2j!^r_41-@h%zrPF`=!O7yTyl>gg4L9|Ju+>`0htN1`@_5cnXUy>uad z^6@j7D}MQGV$N|^d^3phMawq zVQHQXJtx$dqP}wuKFxre!g$9h?Bj3-{Z8M_#J5x6;qWM~n+Ug3`u-{Bbx+`0U~D<2 z35y`TLv~~>3FHF8ZDw-uxJCSrRSwL!u!8eK5Y{1Cz%M4@1q!!pa(5sGL<5}u7SfFQ zDbSaKFy(jxmzbm#sC*Ct2p<9nFc^fS5iqu%D9#Mz{X+Rzx`7^@-Ysxz5XhfQ=*gv%I<;$5qVd-i4V7txDJ><2+1`mwLz1rVLZM*Cxu*mc0FDGBl zyOU8~YlrM>F97;p`S}aU$rpC~bnwK{y&E^~-FtAy-ro-j+YM4%z%vSCnySAXu%=X5 zz%XN!QazW8^^>E1KB%7#^^=3wVPT5V!gJ|1s{)^iTW*SqX1{P9!+{F{M#)s?ckkZv z!^xxPg(Y03&B~b&hfb*ZR$z)>c5d6Y^B1Uj-1K;u;5-)(l9A9CPkku%3ZdK#z}=tq}+k5nVOC!$1=~V1NX<7kh&WncmS(>>K!B4y3Lg2i$RWd!+jI;* zkH=?nKYbHCc?!<&@8}G)2TEQ$3*`MFA(;y@zO0NF-~`~QWeqJ5E^2|dNFIZ!*}PkK zNMv0t7Oe`-yK2EV4?-`vAlMe#U;)@UmYB$%6cVEy&kBjOwQ?B*QX<;CmCG<@ zfzk|c8+`>fUx0g%__-Y4iC1*NohdFh92C$6_w9l^YwDYt>X8f#FRQA)3MVBBL{!^`RAYEDz^9F$%BH%4RHmYXbw*|tI91CAfK7+k`==oPO=8S&XAn}(91MF zM!`)hiS(clj@P<-`gCv)g+9X)jYJDoMWLZOzOkGE0AHa-AQN6zmD-2{BcjHPm6e!( zODgNo4Iqa4%KS<}Q(se~hT5>6$C>qwIB2FK(%sSK#7(gQ){K+l|Ns9{OR@!aMqiSz zpwkuz9it{ep(;}k z&YY1t(no`9TF{Z5{S4?->PV?^AtK7;EI2_K7!@7XB`R85-pK1hLyL;4$a$hJK~cxl zYD~F6=$8Q?hI%o-vL5}1UaYGuL625qN^Yj#>+8#F6BM=OI8W?D>Nj!2c%gTx4+*z_ z5NbC7wIkeq#P+nKR5yuIFI>+)3m9(i2dVYvpN}7JN3jZtVks}nAHcDCKAy78ct1$I zo}N$%7~Jm>Fcpf|v>{@kgpcn}QtD^!_sF@vK0Le~NxMlT?ar!=%-!n!AX!&eHv{Xt z4Qabcq+L>QyAlZzJn6aZ$-Ak^$;o@6bq1n!`lEGvp><%VY}fZ1>ie^oZKr_KhPJuC z|HZZ;8ql?)fd`uAesXX&jaV=YJ9weukOx^PoMgd1=OzrGaKg{s?-7XDGHr=OLquYd z*LySMOL&k_#CB;*EE*yfnWF+TEVau01k>6oL^D7J5PhnNW}+TWGgodLJck4&52Kmm z$A4}|Gfpmd`h1LIu;XP=%;1MojKksiAjP=dqnOdrv1}s}$Zl*)FX7?!52Kg9y(h-9 zi4Uch=H}{$(Mz_iwgApC64*;ScGyHIRwT|(6hk`HzD|mv@zj=Eh!pe3xgCdrIr=b) zp_X}&Yb0HbfwSq`XqZ@PmL#!uw8;b1a<5HhS)};p=b4Pnu<(d=<`@)IG@ppFkXUZ|BpC(30R!q9 z>uT!o=cp?Me=Kc6$T@(?ny}T@28c5zCQRpi`SqGLUwyR(9)8$aSD@bd1oi3DC&XuLw8+32X#)gq zLI8q=TU*B2KCKPo50`5<4gNTdNux0VrcxneGKyQ)YW>9t9rHP+1I!9Ky&eiFS}Vcj z#UFveg;|8dT$q`e3JBxW^vgHWGV=rntJK%o;`ccUb=eq?|b{abE|cTcxQUGyQ!XoC!@A zVoFj|uinDJt*bdXf8W9PkJrEKV2Q5_fmM?GyWtwr z%|36Dyp*1`aEQE-sw_^)t;5pGCn0H@j+ew_{KOS`KOGuobs}>->)yPs?RSkE66Rj2ZSbHynHNp zOxD35dJP*kcIMoFUAyKdb#UwyL<+eW0U@M^(|m0pLB965UQ$cYoju~Vr~jP^S* zeX($m136i6W^!^Suh;XpE?&IVAeT4PRM*s0m6umhQkdGtMq6-Zk;-6D6=k~KO zzMhuu-8***kL(uAq~0dIffuONi?!+vn4+`_i6=^}vI7SW9TH$U;4U(N&01+$T7es3HVXo@78avyl@$KbKqzo}e=LRKgzK@9XAh%ACKL#bh|>U?w0JbI~0 zi_0sD3Q9^ixpP{0zSwPRYnfIsA+3U21zOtEWW{ zE+LgQkjhGWIl3QgV7a-rlxx>iDJjXsdkk_uB zJ9=1}nOq1z6(>LW+?A3C%jREonvgJvB}Ntvz|)Ud}Y6fns~ z39`qJ9qSki==w@l$<3ZEAoWF-*_@S|dk4OJM~>`Ey?pu189@!*x0hFd*&HgUZ_DKe zlNK-Dds0A9Yhy!0V_jYO-<74HATtwr!pjBqq2CW3I(6j8sY57>dU{IA!C8X_32H|Y z%%)9EJdy%EK6`d76@;zeJ03fRr-tCUxw!%GU_5{3%;n3e`;HtD)czj+{=P`IhydT) zC-*L1oCH8pLEXR_8fxq6Y8#}o&S1#CApWjZoQOS!rUHz#J#3wZ~gA1Ac2@@rrIbggSqPHCkW7K%H`Xq>$F!V9s`o_)!_ zU2A;W}XtXBZU@ud!S!>n-I13KIV6U*xac(gK`;C||E@o8Bu%W{ML+k7B3mgc* zB6+g_gsVbNZ*VtyBTJV7sY59D5;WLQ(3ymQl9K$q;_@;q+RMs|^YTlos!ECr;crt| zTv8?Vk_EN4VBA|ky%z8Z3xj_chujkyKu;)vla8L%A2*v6qKD``iEfi@_jIUKZsWMp}n$cvJh zClXU?4iYlvVh|A3I+_-;+J&B(nMpU{`*ZV5dOkwXKSv#E5U(@)-{xy!jWEl@kwquj@YlQbF) zmShX$9 ^q}(zjMGi)7yP~$;P+J(yndgmEC|BiOQ%=LqSzMlZ^UA8#zX)(9(d*0s zotdpzuGZ`I%@wu^lc4-`H%K)nWjsub6CESPozlU~g8_0PRP4Q=tM=4t%d<*rnpn}9 zqg&zVylO>mxmJibrXl!k&u5Pf!>V=WqtQ`426j!vsX<*j2f8s+m!YwvM~{vk<|fw) zoMUwe_xt**?tag%xbs;L<~Of}855*L%B1xU{>M+-QubPzCmKv^D%mjbWj zc=dF0j)!1tF&U^g9gu2R6A3bP@V{mXvKo!12AY!Zp!d=Br9w#94GrMZD*Ry!dq-I zf;N@mA$A!7n~Fzbn_6(c5f(X{Vv**ju1@r1Orw~cME1kEr=^LQx|BDCOk85f2KX_Y zQ}AS%qr4_gojf-Bk**$^XJ4E*b>u|gSg(s$?PQCi}?F;m#0Bvo08fGW=?CeZ9O=nWh z5;8!E@fIj9w6G{T&+B!V{H01#HFaJh5HGMt$^4`vCqwaEF^inoQp5$`MoXnt;PAP~x9&D-bSFg? z3=^4($k9T;uLo}`kOeKWH{vD+^^fw-heh=CZ&0c1+4EJby@wI>PA9!Y~OczEF8 znLzZLiToQsRaBPerl#FEcl5})n>Utz`q|pen|2&HdSu_u7#Ol&R&PF;gK(vslbZ-z zgud8#GF$v&10c?Xd4Lo?$xZ?FeS^5KgNA>lxbI@SVbbiSJ^F=Cho^LH5bAS z68B(E4KJX};$FthdLtIfHtfpRXT?Cghyhwej?{Z_Bu1*vjP%Uxyu1uA4RRZ*)EbRe zt2Y<`0H{)dM7@wpO6)tM$sGwijqWBdUtii=Y0Pm!F$9dfMz6Q=bMhL7SgCxB2O*_k^COPoy`YiYBylkiS_|1oh=b z#EJd6W9OM{bRzwREUZ=6j~zR<|EHgR`bEI<_i5$Ut>3R%wgOW>In#W=ee%hpV$MlI zN+TDR=ddb!0|*^+??3f@oI36yr#|nPD@69_7OsLaq6|UcKq{97Lt)T2Dk`ee+wZ4=s+5Kh+^aLh{J!fmFrZt&AanA+wyk{GWq4pa62!^yMDc- zz^ceAxPHB$MDQ#qu-VWSqBm$^UVdR-VPQVXWrrmYR&YBS$N4UJ`uT})_oAbyzpvTf z7i!vvpDMLO0NR0KP&%L;pp?f8$o!($DfnMj76K!z8gOYiAAn&a(ME?EGde^>4DxyO z(HVpAlb2UeW{F#7kS8)OpIEr?#Ia+Wzy0KsA2%c0vpp=8X}7JTx+3uPZg+kYnij!)Fj#S+4kU+i=Qu)`3|@*7ykTneXOWzeM-H4)=;Q& zhZ0--M%)1#rYc&#Y{lY*iGWn;0oanc{NlZZv8Y@h$G?eYj(&pyJ{2t}L>93M#ls(O(7u28wQ;@UADU_lH!#=a01#MqBGl%hE1l*c?! zf)1b)KhOzjVZtpyg@?Npc$ZA6ku_;#)s4y!S^ z%Ay?7U(iJ$E(j#01zDv~g($U9R6t$f`~i2Xt#7Jr;2lcjsr7<}f*Go-kyEgws=B(q zwz9Iawq7u}6c8?b#NSEpogt16$4iD6hgYRjdbglPp{Rc-#;QwW;V$;BEuGxV$z@AG zB`D~D9GmFKee#rX!$wXHi|iX=35^@nU2s=;29(z#a-lT9TM^LNPjFwfdeMq?pZ~OZ z+nyixeD~cqf;)=NLm^zn;~*n1LvXi=JrA#6(P1t_phyEm9EfhdI6}N6Q31SqMW;FA z{l81&HVmbI1^w|l`Xda~8RkT(=9QRuDDQ$y!7F7R)#8pi z!R05aPp{71LPJA4c8=;(SFaB0JOr-PLplf4Kk?+_Go}GKW#Wv-pA797HFm~xu?(?~ zY@zJA86!F(uRA!lzqdDq$mk7TatXqfR(}4C&^zlk z*15N{N-$fOWMvm&jh&a30UyQ;@rU%bKjdbmt=;xq&fEAD7~ zxc0)2n#_+FdgNLSQ{)5Zh~tn}eB{OHk5+Gu-XN_b^6#ak+_6#I7np+b#h?z}z5fdg z;Y5UG+`et)!ox|1c~q8=aC`$L0DjTY4Zj2iTDhz&L7tJFmxXJ=OdWEgm#?RXbliSU zA)l@`a|$B8ECSmju=C=emd+)cKIyV3E{eQB$foa|Mu2^kH!U6VIr6!u-MUG#f;SNg zWI8keZ8(CUhirV_Jtn>W{_FEAl^y%`?Gh5yef<2&ytHhM;2j$qEC~7zi@z8 zhxHlKzkj4WJ4x`)&HzyognK4LUnF_Y&MzQ&!JEoO<&!oXL99aZKV(;=z(MG0Ob~yO z?A^;uw%xxxUuqSt<(@@N3*M!*kkAVsIzZLL;|YnJL&xbia*kpyhtW{Sw+?u*w;r|f zV3%&-UOyzV5#T8ZSH&hhjdVt%K_lZpBcbSnP{zTN8TSH5BqJC8WSk+5Wb2fhH?B}F zcJzsXW+klHKuqWaa8JfQ90^{0NYoCkXP+6~n%P%dj=Zn?ez}|HId=YgF<45h<6e4I z@Q>}*p$6mP=9NFso;{m%;I8{PX1g6KMcVDCoQOOY2q4b-nNM_0!beZtN&r*2%k z5+3Qn6&+qN0=7-bPke;npJw-Q$DGaS{lE@P%CcmHRM~EvKY8+GvS7$20XMIpsEqQE zqpJ+!^pAD~38BAQ>;^9qXMOzwf+@Q=I_H5iq!uEtQ_8J97*y)?I%VFskjnQ~3euV0 zg|Go}&X$&Z?K+;SP$(60+VHM`RFZ(PrVzX<=a28=AJ}EkBVE;%^I!e<%M*I&3T*j; z0lNJ;v!_m-I(yFakt2sjcqH(~PQwKyzis!OmV7Mw-f%p@$8Q%&b}UHN?3gLvDI4C; zT`+v}&FZy30y6T)wX47RX3Lp^1is+R7Qs6|J0~wc4+z~E5GNoM+Oi9w`xcPdwy_BY zo#tk|H#OUVC}3}H61>gG^lSD-0#GC53kZrHUjBg{{Qv57sXfS}+k*BW-MP!7+wIB~ zo9)uItCy}_x^xZtB+!4WK+h=@F>DLlqT1)@vBPiQ4o|_YCM>MRZZD{U4Bu1WdiRd8 z7upL|lN#LTHcXo9Cb+pJCAle;c_on5s|4U=*=!Xhr4`_#B?4h$5Tdfk@dUTvU=NRw zz@U)8kdVOpev?W|h=e+n)`ch!+Gg>wj<@Xj_BqgNVH4u{v;$uVG6GH&;$zuPZVs;% zWT2vx_^C2G8EXz~@X^^F1O11GAVZkWd1(&!v`AQdzGw@Qx+urb>VEB7cO%am1r@0d z20im!9ALhzPUCv8>V4Ftryt^f-PT;6>5g2rRV^gYO?>>y__L?a`yh z)OTZGIyxo<_%w*E+ykxLqhC@|zv9bi>5F1ZD+O7HfXQ!-KnT=1xTBn7U+`~D4j?o2 zN1UOB_QhyQnh`U2i2 zMusK;GLR0jSs`q(-lHBJ(yueLP~N~a;j})`G5V0RS7gtABN00~b1nk#=g!2@WH`Ld znfOTT@a{ryi18&QwJr5k<(LQyv9v@~M5YaScvy2mZeC7CW?o@wReeK4RYi-t2?Xa4 zg~dTo!3Toksd<4P=RE@+E3)~$0%r)S@0&XuHy=OQLEwJ) zb0ny8G;DVNaeV1`K~N1SM+!EzuY7@5vV;rE)c`HeAg5tvubw>t2FCe)fZ!kPB z#87AncQsptU=~=ql`kkr7ZAM!V&dJKYEt^D)PBUCB|io0Ohu3cYzCDtuHy+DxJGV*dMJlG-g zF#7t4M`E1_IfSyL(u&xL9Ktk|+v!g7P2`|P^Ce=9BIAS+teo=zB-3JcPJZA^8h216f%Faqqo)-eK1CWbqGI0ha2F8i9E&XiVJLBY8K0XNOg&MGWGGH$dfAl#8m zP5>M!OtAib-ah7gU$&Mj_`!0e-iE6vv~>5%6~mgjAi4I@e;v(Az!A-WR4ovZ`5ob!GP>Q`1w{!%62Hc z7cE>UFxlDjC)SR`T6wCY1NJw-&1vO~ROG;^J1ZFAw+F3q$e=+nkV0a}{|0txT1w^Q zqX=NfkeOv_voDf)!NtRT?>n(g32lWk$&>cciME1xHLwQg4G^ye5ghXw)O}VZ(#|wP zTwvVOFOm-VTQb(jV`j{M@?|k zM&RjC=Y;`#m<+;zj^r+X1IHb&fI%qypb&)`vNfg`XzHxMW`_ELV`|9Hm^GlDYyCO1&6a|DU=sLP4XvQs4GLe4^9!lZyW#n9 zcz!gVPit(^N?w$MrPrY^^BN(mT)%v`u&hgGH^H;Cw5mEk=OzpGYOaR}CpTCOrNT6% ztULqiq zP1Zn2U?N-?w{YutdxpKyo+9p9_7Z!8L+Q|h-t>-vf@gB_$^E|~8|cOT-+#Mi&6@Qa zZ=@v)9>Tuge*5i8@~ZE4?%a9y+f}OpNANPt^skb6zEcc8aTFT(7-B?ZLdCwy@ z=tTU7T;GE^>(q&TySDH9c|V=&=5FcInH3k-H^cY6yp-(oVjs~OfmSrMMlhg_XpP`T zVWJ2am-MI0xlRLP;LfpBb;&kzz{H7Ae8tPc!oU>e>0#;V@Q~ZN<>O^1e_8YWiWT2| zCoFf%%!K?wiVRYjE_b6~cf=D*+S27x&%mb=-d1uJ1DX!TQ-c{Y(=O5*9At>D#P20`f=k@EaBd=+)1B2ZZ@Ln$^LGz3kDeDEn@uM@ZCw{;^}mMYClB zvQeEs0!8*un>TOXwe5IPwR6h!@vI^Z&p$oaZQ1_qx9hk3u;sgro6elvzjfpKFBcHF zTLQ*yf@7-Sr$y3tTtr$X1*=KF35^tOg;`~SyBI%+5u8K%kW3Vg29SgR00Gxsj29$- zEmD6V9=QF_V#`ON4W^+DrlAd-5ie$6+ypCUFeOsyJ9lafUr{f37djZy4`QAbB?WYm zFQlXvGzTY#Bn@~CJRBxJ$WAO`#+;X4dTH$N;Uk73bYaM2uf6%~ql5Z&g_~HvDN`m+ zo;-Qvpbol-fs=%YTdCP4EzEJ_+V%9JO1N1xm1^LO(NtA<^}54RSCtQGtsy@Hvr5Fu z`4oD-wVO6?`gHY@gk_&DT)h@vGa$@GAAkJuhVRyI-n9N(0THpm!5!W8aj`?X z1>2#7YGFvaVf65|bna19*yI}(7Z(@X$r2`@Ueu(}d(k@~n36aM;z4XH7&8WUGp0)! zMmhK$F+AQ^RQ(wZqy+?u3&;xEKnhs?%PYjz9*x$X3_5=dbUqpsHyVT#hNi>=(Ub(l zayu0&4II)2Ah2}GIYMuO73n0vlWElr&5sp^SZpGA3;~}x1mm;`)37p{HD%IZi^Vc{ z`kY6fra9v?4_LOJ7}pak#=eg}`pl8c4mmRhFN63N!tL|Q3 z3bV_Vn~lv7V-}3U+U@1mbPw;U7km%<>W-E1R$7F=XtV;);?q^%BFE8p3l@B`c=6Ju z0;GA!6@sBLFT1c70%BmGwyFivZ-NT_9RL%YLkL3yi?u~&5D;!I6eVbB+5lGub*luv zi`sG@V-lpv03!|gYG_NKN@IBh$t;Zekv1b@VZFS4pg_SPZE!V$Mpgoc{vU56@qs?* zm3bHgA7BjhL9g_2S-uJXR$Mz^FhCoJf1n35?0q8rV+>I5w9g15igFL6g$bD0IlL=` z^2i}Wr@*7{>4|Z(=3|@p+@t~BdII0p_|dF!gDoI;L*Rf>v)}t~=b-+Ak6d0;U9B;g z&5dx&fU1?bGROEBjlH6snPbkv4t@>m3dkkVj`uZt_kq{!{r<}}pNzoFPkSOE%=}-b zl<~3KYw9X%3ewUrlQ-5iH?qG^F=K{NRUA4PJ`d)Z5#`kTlg*fc_9KoRjGqTH&OG!I zaRd=6)dY;72^c|f=&Lx4pb)ZJh!_ue!*IcXazPA0Dxsf%)CBQZ0h~?;uSjrVXD5&w zAu_XI0qXtalXAIsP}~scA-awo)C0V#Pmh1hkE8et?2A?ZoE{epxDTXp&_YfE&Hz`N z(ZQZzQZh)dM~>dMvk(lRuu$90qk{M6pPxGQ+pc|we%rtQ@c!*TBS3cRp0Ae2L5>l3 z)zUAP?KyE`*Bac+ZnrR{DzIyC`mF>Q_LWK`n+aC(MQ{?hlGMLfq=3rNxR3@3MhC_Z zO>I<1NI>)x(6+FGKw3L!=9&^*7}E*Bq@v_H7>^-jt`PAh@P?6ru|v}#s14XYG+tmp z{2y-7%Dt%VN4Mc#AtLj-gH|TYw?aNuBKFAx)hCW^C?z6I@%)u!F5BJ~m2D&YSx zwu!3>x2V}8Vi!D~pFXBr$MKIpH7TYydZtU%5CI@gzZ;IQ9oGJ@#Ouhons^P%d2gI>ZHPQ@_ zIuxvm=dFaT4*`MW6R5&q8DiEo@F`5-SPwXv7EK=FY$38Npj7@ZZX@=9D;Hx|a4M%P zOZm1a{WqSj#&R73p^Yiv~VgSAwxXy#Jnd00TMcdba3|$-3kP36@@UkP!9C> z?O^HB?~#}xLk7KWPv9p zKM@0xkRPOjj4bLjq9rVEVND{C2Ty{EyLVtj#)h`1NbOA8#<3Wck7881w2i@{9hYbp z;tUv|$OCESl?EkrELQXb3|foi;1QxY0eZnG1uOJ?GS=Q9ZW7G6V<*S;fp1fvzHxKn zq?l0liRy`w0Me+e3>TEOSIhh;0RU4aFpbsF$elQz+cv$OCSmb`oh#!ZkS&OIe5hWz zboX@vKk7rW0X{?Q zBsqcD9_m8k8jhjuArS#CH;d$8=xD<>3GQH@Ab|j5(3LU*{0<#Nf*W`rv_gM|i&JkBtelq~AJz z{CFDU!()VyjvdR&0<=cZp3$pMI)HMJUlL7BQW86qx7lm-$w(C8={%{v(x3!prQtO&--vEbS%lP3jx8P$&3%G&0#+S)15OGozZ zA3I{oj0t0BPniM*cn5b?f|+BF41MI0M?#T#xrR~&$mQowpF4LhS0=FFQ>zho{nOrK zdrzG`eF|89t52Oeb^7OR-)~&~#phpsyLvSY^5BeSCxT)jkCBcTZs4H!QG#9f8o@wf zBAOB8K=a4ly9=QGhVYJQ&jI=)C~J&xv=5PHo8Kg=>w#A21F9Q|aqUEsXMWjl&Y^iNr_$+5xVtk`_R` z!C#_yoap`foKVx?Db3zvUY`8cJC@Fqy*x+Fer>wILal650t-rD{S#SVE8Cr@7FK+I zVCR|xJAm=`fnymOW4>a==EY0bEhf+R=NXiw=dG+)0=tmFZY9!jw?uC|8Px^@AlO9| z27iQa1p$-_fudj((Q%?Tm=dZC$&6qYB{v>IyL#|Usn$b55t}eBPGDRN#kd&CxS&9g zJRGE;u~2Yu;KUfgc$8rSJD1D&K=4KyC|MLEBeEF|Lz)d* zUy5r=D2SJ_WvUNfePZg=sZYEbgRtUdZqNR6=7`?m;o-eU%>1X6!Xwf6^6U|jk^g@G z#aXdkyLRo;t@D3B0tmYHqmM=o2HrsQq^IFPD#$;0^NE-KEhQaHO59~ zA*=3Bgj`pTtR7{Q2wC}%M934M+Zq4_y9sb2p+u_l@jd|>y8$%Is246>3B}9Gr3*)4 z!jUqn&W6YDYjE+EGA2I-@4z<{2qQ>bgdECGz%O_P+zsa{KKbNpF=O(79aizW_i+6i zxH)-!)BXy46g#)gORM(+EA`PU;ODHkKa+BOGqQnrQxYbnm`PdNHj{E7OdiB2?ofb{ z)JX*Oh=D`wID^)_;A$AukcZGM8tNeQ6?GTm0^K2OfP3gE7D;5kw3;HGD`r>z`->&2 z8UU(#161`NP!$xOcsVgvkuQ)aiewV}#~Ub%xP1U|vy2Rsn#w1xw}_rldelS|UZ9V# za-1X44TXfd0vBVdV3DuA_RlH9`}glZe9Avx8$D)FCtqlJd^-&q(<*FW(bf0i!iN7} z5sOOzapu=;CxEaxq>M#+Y{$Rl@vx4;>40Do=LT%pu;;)L7)OsE*t22XLS#%>vT)rl z$|$sJ9nvI#@nL^o_1U^DPiEijsXgRPnf`AY}?n+`BQ1AwlS7>z%;e|w2<6mDU^(0B$|3FX9 zLr*#-ZcmdEeTek|iFa6Zk^Pf1L4$XPU^xMIj?#f?X4-(A=fL){g1XF(NRGGPojk~`bG%J_vyW41V+(%QXf*s}Gv!>4a0-#UHxx2=hZ2M$C#-ZLb8_PO=r#qg8b zvuE$lHER+DbK5cI>=bdl-IxD2NH9T{mj&-%1Wgf4YB19!3DJ&;0-G+NjESP;;_{~>Po-9Wx?5>* zqf4Ox3DHw2)0TGY2Km!TTpAri{lZIb2(SWrr%vqet8p`R=%6c2o%qs_E0!V`kX8w+Np0 zHL$(wPc%20J&dF56%a7iz!d+}&tF|FNJ`p50lC=uElEi+IQUSu^1{Bu0y1P56eue~ z<;`jJ71!my%7OwvrQivvj^rrPkHaPeB>^}pN(A8$Lm4p~gv;eS}5v?ml>>43fqXi`4nsAF@GsYT>Ru_p%Y6ct&h~ncv z8uCsw+sl0zL!!|_wb6Hme^$_JU!!0&zcfA{I41{q{=k(%#@z8QnFZs%qep-EdCSq` z`wH&dsV%C%d#6A!+S0BzmgH#p=4Pc^PFZ7enoTf5Tt;mrp*fqFSeAfv*R*g*R8p03 zERd8;76(yQCj3~~njbhxs-Y4!d;lK?@F0NUh;~dC|I;xO|KYQAl>%3};sWl&@i;+e zKTH2aSPwvXh`f)kc;Gp73O`ExzYJ~-_g_O^3I?>k2U_3hrJxX1S-d>?_AT4~c-ELO z0M~Y z`zr3ehm*D7-kls90PtDr@Zn<4;oc))Vm;3++#GJ-&;)rYZk@!n$P(m1xOERYl3J&e z1yLXCX@Gf3wqO5w!gAvL+VfK1YFrW|u~f#<+4k?CB3Cc)jl`z=l9LN!4@ z0}p|c%eP|P@Vta?q9S|~zot>M5%;Kb!Pt<`BGb9BPD*_{)Z26}z!Oj`%<+CEpq$k5 zCg7ZO4@I^PK%J<8brR;Mng|aYK7v`S0l=PV00@mi)c_cpHphnEz2O=nwyFv}rb5ku zq=$DY*4PRhFa$j#U$m7)3f5@0$BNizEEcTnxb)_VfF`U9iV6usyM<1ypv(lIiefJi@Zh@uT_P@T zg;n8RkBerz)JKm7#Kh!yPGIC+AUy<=j@AXXl96#n>%v=UX#(1BTgVmCkh4gup{S^& zjsyo*S5nl?%d4BS9jGRhzo(P4CQj52>ThhQNo$YWwrE`)j+$s){e9f8QtRq%)Ivn_ z5c?VzJ9}D5J^fhcceSow7Hi_7E2@dB!z(IMwXW_)O+0~GhX!rC^E~x`JCqxfhrh`M zHspHug13U$&-YXK@V}vOS{k%c_+48Y3g7VEclRhfXyEn(|9c7#?DhT=e@)@V=aTB~ zr|`hPqHyh|tS&^}Sy{!6YG|g^jm2Q9_l`yK7Hdry85Z;^VaE`CO1(;ginM9=AK3QO z4_kNa+P*E?@rw5N(IdYdJVY;sq-(NkXDdI2NuN)VeM`QS6SciAhaLU0kzP{&I!Cii znY4b_vRxcfw$0-{zD0Ms*4^Igp8@nDT5WxvOpEyCz5@sK>D{mY8(Pkhk&~N6FX~|u z=-cn6E}it4@vuM%fZ8@H2HR9qO^|yBu^$7Zx-zGljRqRX z468@pfcY;iJ%gs+w41h++s6#Q?A;52eAlks*olJ${uM@D_rW{P0`v*kK^=Py88Rd) zAV2|E-1IAF&YZcHkx_3Z9w!9o0Os0Qk)N9S`R7;i8&qn!!Y4T3*?FDr9MuL20q4)( z$cEY;0o*1m9J>c20I+M{zV+utA8n%t3G_5xsI;^?;ysFsii;6LO5hU}*?LpyX^*yO zfg&q3iLk%U%fuBe$GeAh3=0blCfO}4yt8I9pt87579(;~2EjEo#L_X;5<+C+Y`aj< z3B`g90-c0{3y!srhSEtSBBMcx3@#RcaOJDXw{6D^EQH`X%^5DZbV5@9Hz0)I7LEM`A%?_ONTju+C8%STvd zYytjZwxGi^^e^)M(zURQ`Z$;@}@mK5#fwe)M*TYEujWLCFsv zJbK~}x0y5Zk8eM2R@xD*uKwut=U;f?h38-YXcRPitDGP*tDxii=%Y8@c;iF(z`Ub@Jw_a;utr5Y=kD-je%T}K~a*2w^N*V1N24+I4Vq90oE)VA;5yJNw^h4J9;y{xJO^gzPc zlWo>UAVJ!z+>O*|$5=f1IV0^~Rq0xtnB*S9hX z@{gTi9##MepfH5Krzy(y^y?h?*pqC86%cSJ)IAhFlHVT~S<#N!M5;fCq`QF*d(gkG z&N0H{8j{6b2r3HDR3hQOBbqhZF+Uz7keRUz6vv}D^O0_uZ2J>Yd(tfr#}&a?0?|s4 zG(_S%>9esa60JeBMiMP&=x3{>XB6Q5g-swnwICLuXmFg+{O8|u=gjF8+5fTEDHilc zcG!SVA#SR=x>KiaHW<2&e_RL#Py{YP$|CbX;P&)JhSf&tvqx`y9^i{Izb{+8YQ<;K zj{m9?6E|-7{FBcXqt)Lx;&{UnbU>q*iJiSIa|^_FP1qtzGch3AWAM6p8T42NDKgY<$B(RL3h0vaP)Y0?3$ z!^oCMZyqeu`Zt$Ld@lm69csBJ2S=c_-HHorx1cM+%mtet$_~0QY}oJvkni?EIZ1Cn zY!HCuh9QR?f0|#vsZQJdRYl(QrJ7TmCNNOLT|Xts!{R9lZq8d+pUUSxGh>DVFraFc zD);93^EY!z8o^{q8{V{fpCDhguBFzl^3wL}5n}G%-5k=RA2XghbNSY-3m2faJ9G#+ zjgGTLR)d$9p4HXKp$wE`V$pj^+@G)sT&n`YCQzY1R+K-QEdl#qYGp6evfzNUEbzj* z^L>21$O=`2)E=%iD$y$fz6sj^?(zZ_6>Ok@xK)sQJi3z5v@T8^(Y;vM{uWsd;;yM*i4Ey6iT&fExv_u z;oi3gT5OqrXcYlf0p|j$2mmP)07&pM0{0^TBokH?(kHQ;6oWkyPbFVVioqUi;Yfs` zKA?`^XyuE{vg8AP=l1QJDXF(b2!&J%z=C5;3r1HnaB>=&noW>@=t7ct=z5wffLo@S zHG5|F$e(Zc#Jm#2K4u@wpNTft`sliF%xva@Ccao%v2O*Xl*P^^RWQoh@*Kq&xFrgU zmVUZ?c|rb_t;VCP<>rV$3b_T*%tf3h85sR+L05AQb$;Y8!$ce;mke0((qgX9U(8syvBCiAso% z%5mL2J#I8xfp}ctBG4xG60HHqU!0;V3EvOaXQm}19)%1+3Kp~ubOiLG6cHh4-$#P6 z|NOV38tg>SslZa7cKCp(*w|ivNOwv+ z6;@K{*#uxY(qs=i0Q|LdQ2oP2h_tL;5SZHiCba>{7!T@N{vPv}8O0!h`OXn935>Gj|< zW^Zb0YH4Q8R{hPJ$u|(J2%YWCTPZgM#b$?y3aLuJdibZ!Kd|jqJ;xQAow$%1!iAi~ ziMWueh)J)$^7P~EMXMfSOpX&7vKYva3YA0e300zsSGbwHj8J6gNzsNuhoJ=&BMRvO z%4l7qRi)9pspU!*VnsINOPAr_j*h>RdgFQe98y0MWT2SD=`wnV4#abwA3iC$8{`i_S zfH6w{;g~Ei)0h`ue5seG%&oDJ(|T)=a1^NZF>EH?GTse-OB61J$<3?Yt<7mnqCLW$r5chCno~{!k}9z- zkyC02&<=UG;4zO#I;8eN?a-FGV+NRujvPAl>knUlnJBpzEc;ae$i+9P9x6E+fuE02 zOo6z+=6IRUD+J`GM_*rrs@v_3jxqgXMoyRz9V^HJIz)66YpCtkcM9tE(hGrp9Ry7_ ztUXAt1#ftSp8;AERpU4ff#o5^7M$SQd@I$2VzR0Vkd7{kvh!Erdk>B z1=>CFz8Z}Z*lz-I%9*U}9r1ak&iR^jxfESZ$jhY5^|V0WNA!lw+jng;yQ8^9)@M$i zIWnKk{7R<=nZe=qeaCXg`^+r2c+Y}+90_}1v2vM9$4mlTK7>2DjNL02we)?qK4e6S zc8rH#1`3rZ%@Pmirgc^}6gADVvPba-zx=Cs3$(I*cmoLyyvI$cz1z)TP>}-m36mBO zNCEuA$^`bO!1fC4uE5~+N}1e9YNU7-@{5WF|i?<(xfB}=4FjK=06|IiRq-lmoYIIWC6aujaRQ; zZNkzFo7va5S^9({UtfW_-`%?9!rf9IpS!Sg-%Z~d1Ge;vZq?Rb@8X9oTQ_fk%z#^j zAfz{xt`KiQA=LOJ-NKlYg>b$cpb*j=m_ycDO!*`gP-Tdf(3>X^U+=FUZ}Rp;&>x~& zylIm3CM#~H>A`EAed>}eiPl1R6)GdyJ|3Y25SSG!AtFC+qV}pJ!xa zk{{1=QSkDFxvc0CQ6B*D|qzT_uqei{-b@dOzbPf^Oyy! zYG_{IPzkdG`j>{X&LN`mK}An=`M*3%s$~z2p!j)>|Oaa<^&<^=fa2mBw#Rj~KVylc(S zM1R0-``XPb%L`FSzDK@j41?Mb3K+4V&j0ceKulZ1)^JhmGWSmw!`87+xYu4A0k__D z>?}LWMH}>CdSh*^F(9DVz>y;d_F|ND+hpocXVohphcxpF3dhngxz$Ti0imIxfrmG< zPA@vv>UCOyKe-)vu)tdM*uSB;c-z+iVfuR8A;c4ra(uGYtGTwhxjrYSzWLOt6%zazavHWZ1oE(e5^I>QLK~ zK?ocQ55a;FT{3A>47Dm=HBheC%fH)SZ9FCgY!H7a6N=VULVsIX8m1# zP&=S?A8!UJ23_g;lh`5eEdsC`+zEc&(;MLH9)MmqS-kUryiUODW#t6BUINhd!jdW` zwODTpH2!F@e=yJ?wHR6%bwQJna4M-9W$tzDl#Mob_jc+=1_cKykLd>2NIFLuR}U@N zlZvnJ=^hoL?lIou*X7H9ZL5G}9Krr2LlyDA)kJb1v1sTLf=|`fo&r>4^3{tc;XiTm z;?-nZK}ku<)${xJ?>~Pvr9_}+MWI|ONnY2#tx^)=j~+>u?kOf&2>DC@opuuIli+l0 zM~|Mma4jtxn$PUCYZp!(1-{O;>rm+<$HOIQUZx1&XVI>6L5YNbJ6Fu6f;Mv67|=}4 z&8w|7#bq8i=GE4u(rQt%1d9z#qqzOnY8?s|EiVtIe+nF(d7#Ivs&kFOlIi6E%S)?O zN=(Qt-M#&Dh`8Z1CJ(x2o@r;DsY|XKo3taz64Yv-X>Xqi9zJ(Cy}a7mV={ool8#96 z(_U?DGOaeP;2Hzr4 zfIu?@0zW@5v&~jsR#uMXne*K_3TW+2yap30-v8$eCE}>T`D7>SjZuRPB^WhQQAtRF zeuL-T+jeZa?U;MpF5R^8uZ?6B1Mqi7vb_q$-oG}InazcTe{Ce2;ORL5S>@Um74Gxy zZM&qP;Iw<&&Vcc_oedD)Sv+rOXB-Q)SPbp#Vy!k~vWfS`IG~jwt$oc5YhRI75>JM? z8X&+8a9?SLepd^`lX^@mnBW=wG+%ZsvzI#*RzCeVBn^Ixc7Ek=HfyxK+-m65yLTt| zYWQ!}x_6#5X;LSFaj(7~1O3x-t2(Gl9|RHi>0-fb4)tWW;NWf-u42V_{lXuI4<9}y zXpt%m5#>}3cK`?|2<{M_#fQeEtI#O&6u77t4&sfRK=o0HaY>lrEY)X{nA1P!t3X z+?>fqM1Kzm@tBQanBjRV>_<3_Mh_{nd7UfuNL|+v%{CZfOcak{3W|;nlIk1otnVPy zw=3$~1NBAM$;B*7G$6I!{yso!^9l?$TeM<@j~`Enk7wRFB~aCG-+m>hAvpNlxy$K4 z`}E!@*nh@6zknM#E1G>D@2qxFn`#%Q z1ALaJRL1BiQS=h)Q@0LtFY*gtysQ% z*|OzJ5Q)BY3F~ajtEj8ny!nEyIw0WOxl5@zwu0)atP5K=Ro68a1CF$?oDYsjy>cB* zgZ%RDN~^r^#EBDyR)3|tDLA+bC7l|C?0K+HIV#N1hvVqzK^22OhVl~-SU>Axe`$5`M0iR=%Q;pvz)GCRA{&EG#9 zzKr4i{%)1orKQ!4kem663V&F*Ltv9PckI}aD-}EpGNqh|q_w6n%9Yn5KTUN7fjQMS z*$m~C#C~>-MyjFZnyoP z&aMN#sdDYV=bWr&v`O1^PdX{P?A_9`EeeV{eQu&U7z2LYpSv|2b*W;QD>v&HGDEa&pf5p7(jz^FHJMz|ljG30#B2 zh>>F_j!B(*%Pq5JX3m&9rEk=s86b6q^#$6&gRts}XjEQ)wFFxTZC>7<0|)lx<-xQ_ zG@E@rv5k^O{rfQ1Y%P>Y2uSE~43kEg;1%UaSiJaogqQw=?U9J2wOT7fLxG^~8d|=d zphx?v>W(p!sKpMz=>*5EBeMasuP6M`r@vhPg8nDpJRo%IeIln$jqKxO-E|m@I*<%A ziJ>of>d@3&Qv)xPKdi&3d56yYc^wd`DTC#)3tS|2N@=V}{8cSD^hamgV8K{jr9%`_sHkCD2Yc1f_s>{o(D$z8?{-!ZQ z*`Jdmiek=svpLw)M*D(v$pbrrgFC8D>^+Ez0=LuMSRRqB5)xbQRzr<%VIQ&=(ZQd` zWbqH6lD~=1ok&~zCXK-UmNY2atE#Fn5nON13HH$9^wA4NNAr&z%RhDbmn&#G*j~Gv z)+6R6DQ~c+jdy?Y!PfV-e)7qe|JkwqbGoCyu_F=JDIx;i-XmiH@g&hH1$3GnBV%{O zxcre|4-yEJ9GZXnMo1s05Kv3;xoT_c8(>$~(g^e5`f5vnt4ANgg!1aEuWj782?>&L z57m5Rx{zRH%gOwZ@ECmvp!@?9Gv+utY#?Bahk#L01IMIeD-j#3X=-Y!yaLH|k=5#m zf>y+92m~!&{P3dN?p!i|)`SUdZ6jtaUj5*z`=5RBuhZBO=y5hk<08gGO@JuhfVhZj zX$i$_5fcV6&VT^}7*N|!AKFuL0S>Xp^Yf4WbPQNF`KMnBaM4tE(_Fy@cu0Ee=#isG zj~sznU*1pnMz3SXkO!z6dni?iza-i9PriV;S5Ob%DaB{u-=~+lFo;=4p~iQ$#(2ux80WGL&X7-p^A!>+3kSA%kiO7p~W*&$$jTseV6oIZrpNrJS!dP zyGkvPA(j+NPk^2b^*5eD7Zj8KKnb4Mv0=jo_LB&WoUQwMPhzwj z8aeob`$fiw%wGjW8g{R!j!d1DF=cG(l+59gX37&*02Kg%ucZ6HIYEX19fC_AM@s$x zK)?sGxSRiT(U7bsZoyN!tta4_2kQw?PQIWRNGm9^QM&VNc?==Zpi@8*lGt(T9Vuc@o6J$ibdnBiQec}M2EY?&{^{#aCMBGA#I2N|KBS8WBmjbHhYxI6;y3OJ9{X=q0X;|a8*^g`|6 zdW!1cbq#2@l|Ti!M%Cg3AZzekHx^+)SUP*VZ%D9@9G44U>^}csh>Z}%mo{=R)Ccx$ zAP@w6F`@m*%~F4u1WGRir5B<<^n1I5uH@2G0#P>(uIGd18jlJ<5IhplYC%_EvnUmO zHsB{XIi`QOW;=5sxR6~u^LKIMZ0MdStp=oEE))l(i+e$_n|$4iKc5c3#e1E%J1@b| zXbyKtqPW^<%mjh|_ITOZvTWs=HETnvA6k+ua(qy(JhP@is~ag$V_fc6%du< z;ObnEzoVa(mzPp4p{pd4^I*5K0W;w2Av2n6Msx?2$=``&hKU*)A{pe$<~}$UBKT8zA>y)g?Y#R^I3fgLsqaKqO1^&4^{~h z9+4h8Zt17gcYlA)=_1)?1${W_g958U-0jjbdDs0n6qG?1NE93eK9eZ8EdmZd-cZ>V!Q`a)mm-YkvrtpxMFWcdEt0&; zbMaj8{|DW&(XT|Mvx9?j9*>P+1UP{vHLqnmSP9Y&PK|g+-{7-{ZNmUK#LnUDdiFIt z$gZmiuZ zh;kn+iPx@WbK}R0%8s`By4qT62fq1&Pl+S>a^xb-a?Ef%jWh$_zjIsxi104Q z*N#%h5BT+l<1M5MNc$ZBcD(C2gYzFEZNn8`Ilgxsa2%G>ddGT44zBzjXAj|;QZqCI z_4Vs{fX9gPR>U>3ShBhLdQoZADQvWfDpgxQIez0g4m7v0!lKAy~>K%d-GuFr3!# zhKf5F)-!=Y!I(^Ap~J_EMmzl*{C~~X)nAC4rT9U^(O*)~UsBLt0(j0^+J-G*2kq$^ zdV8@o7fjFR~KR zx1{8gAL<;At^=QxoWFSSycxV^WE?Ct2*SwU5Tpg#1xpk$bbal1Ym-c+>S?yxopL$g zOfH{ksXA9zS5#74ckzO?6zeZoD&{M~gw5LKwAq?#t-c{iNr)swI#$BD^hMptBZ+V> zA?;?c(M-plD-Yc9m!i<6L3lp7Ws>YOfDa0>Wu?((*|Y(7JdJ04TNEtqQ7SVxd}z8e z)ERCfb%mgA@6?)QpJ%_m2kXWtV5AL|T{id6ZKT&GyEpIgZQ_3Ug;#VlQ#zcVGWiQb z@D0@3z;|eviGv?Uw*trNYF^#R`g3w?pETshb+<1ZL{l*s!Iz@ zM;H4}%;vhgbA@>lsbaQdzLrDlw#J8@#1#sK z`_~}XFR3J&+sS{-2>Q8kBWcJ8?H#P$RTLK+D+=%Yi@oD2hBF#r9n;M!(3susa%eUa zAUV^zviANtetiE-tdEhQZ+KdYi9$yw>kxAVOvD1qOLvZU+%R#4a0_SYtcqPj5mjQY zC_s{?=|8$?L}QHn4o4tA&lu~n3#_fJm3WReL<)w`2HIiBn*{p}3Q}wE7iI13c8A)m zJA1XAb=RCd27A%s{rjg$2*~|fC-=@&06+Tdt2m%Sc zm|9dr0IJpj$3WB8B+9f}nVdHVaG1j#N+*!e;5aTgd;-*1^VBMpk19GE(n~w8RH-ht zK}#;R8FBO@LCJA3uBcoiCC?%#lxVw*gTQOgoPmDl@}*0cul#)a6D<_V*WdvP#ELQt zV546@a^UQVW5=)5_eLhP*sKjr4aEg}He&@f#`1QsQI}wAceT{6AS5@yPb8RG24#MVsZlZ1O~8Di8vo zSq{LrGIS|Gr(p?z4g0W>cznSwliB-dt3m!WpU~_@b~2E@yNv-Xm&-2i<_~ut7G=pF zX6&mZUrLsr!9I2FWn0(^^wpQLWtO_qx&jkZWXtp+x}Zc@&boXl70B1k{gDj?^s{I! zo~zppHjYylaMyvu@F)kZb8rEzR58h1YyXU*!4qllL^sV#-~j=`4>W+4@V;DV;%=~M z>;l`uJ!C>K0pbKrL~w9|badsFh^H8Wya@HhZ7pCZDf*?Tg3Re8Rt~%IA4ET_k%WFn zR+@OhU6%Dw*vBWGPiWsSu7Ozr?CqL*dN5iZ6A@nZwI9*^t<9nd@+#A7=IJnfXk<-z zf5m=?FVgNW8{^lxaln}+;^SL+0TM9r~_A$mehv{~3 z8vBgvK?t@5M0{km7y%|Vc2 z##GK2zjW!+OqlgrrS!hA)Ohx+s0~e9v}DO$cinZ*y-V3$qGDQJUS4xV7Ad$XO%u6} z&TNIs7?A)$w%MSU1ohGfRiEAt9aifJMe<3>)OKD2f^Hl1AUHVIzY#5J{M z%e$Lv3(qsmcD!5!gDwb8L8fqa0o{~v=KkosRO1BHm@Mj3(Kiy%Lld!X8-6;ncYod? z@bZVqlXU+9CW918}V!H*VjHQJfcRet|{EN;~R6tZai8hi1&4I`{6y zi|>B!xz%gT;lELdc}sgGr&mzqrbqYh(~z|{bhC!$#^&n)V)O4)l({T?QA0?6BRX(l zi`|Zips5L3%C4?z$PUV&t!rzpt2al$^aTzc(3(MQMtDt#0pTSCc8xhuGYM@p0MrZu zHAAI!txxe!goN;cX`z_oco9xct&TiJUfz~>-`&FaAIl+TStO5{Wj)Y?z@v|GAAX~} zmt7f8?+oC%PeEE>;%*0zyq!;@G&xJPX>*yqj#cQp2!q(sQ^K~bLZDz z{FEmT3YxND0i37%id8hWm6wG&Vz4(2N==0wuk_r%zbm!THuNlgNVYdB`i66!JKU>5 zzwGoBE=hAfOrK!(ak(#jH2`Cj+DNtgT6hw!KOkk4M4a{_&`59#uG^wIgJ_k1PY zGfleZN4k6teV|gF`jADy&3=B!QF0hTG3V*4nm+crt&ijwk^jv-Zf>9Grequzl_oKV zvthFN4;ch<5XEteaS{fj5<%KT7Sp#gLH{DZ1mddQQes-%(V-~P(eAHhhYrHu+8ybr zuq_?ZtS8O=k&cuU>1gY&Udyk;9E~%Ed51YdES}_K-M_1cn3X`U?COtkME26ey?FcL z$-dsXeLEJ?#iD?q8L-Ch4$OTEerYZqg^N?C!|ZG`=E-fMP#qFdZI;I|`A`@Qg580< zKt8ovJ{`MA0k|FjilrUc@0PN2=k_nWO9_m9>+Akf)T%VLnJk4orI=&`hpPKasZN|; ze{(5AQJ2Z63-RA>bUovRLQK0YwS~t4ty9-zx%~56|JVh70$M(u)+jfyibn6~^_zB_ zI(4cH(d+L(MBxl6J%Zq9DuAy}YU5qmMqyPIKPt zd*{?;%a+|Sd&-C*V+J~=62 zQXu=fb_AR4Qf7hi6B9Z-5JoDXwb(Y!CxW~Z&* zYJohe;<~8swY9g@mYzSpecQHeZ@*5F1)?KtR;%^8JbXmPt+%dR`M{dTo__wXfBox| ztMAU5bH~gGwvuv)N=Y4?niQ%*=qf>N2#Xs$&Wn)I%O?*XK72?@Qd|rz^D zkqa~D%%1zq3s1wISKZku?`dspXSx8LsPDG*IOQsZpn@NRLK~(Bhz>Ic-8&Ai2@@v) zj%n)D>GPJ{|LD_CKfUh7>0k}sgk$Y1><#vc2^(4D)!tTH)nJq9)Y*Q#0)`{%-%vgc zl&5%6vq1SYP(BTmk1DMMpW4;dPz@0_MB3JdHqyp-HTLd^w!3Wvq0e$V8HG{(5NL+^uEQ2Y&sq4g9m|-HV!g|=S|SU zMX1#+6rHH5{(3j~LPw|d%5kqg%Z)ZMR;_(@2DVhMB7_vBP0XZx=7E)u ztQgH!Q9e;A6SEd%rKiS4Mh-|!P8l-Bs|BLQtI$X~|BxscgeN7%MTYo87SkRTAD3|J zUGpFSjp*#u^mKK>GuTh3RKf>G7vSsFxufO}!%)x#CqR9Zk(rr+7tAN-WGz~;YRU2? zYzp1q-d26Ju(+z(*3;9{VuR|>i|EoUL*suUXij>1ve_F7nj0_ug7M!_QIt>c0WFq` z=ic7@Ev9@q9FloOuc#@l=)n$GXsx+yRhbp9?!@@7!uVf^@jqQk6s%+=0$8pz>J9=2-f!*%`@N83VadnQ#Pz;9en>X&7x_tTaJLgUrHgwz|k=2W$5Ev8b zFRBA$r9k4NM|0gd#>raLR+*X-0|!&vfRvPhw=Z2F=GxHb$jM}sjwEGO{3 zG;k=}|&T6Cw+k+Z|z(JrckVhq^41cZk>6|S{!4CpWubQl9V zi~$|UHaXO&PEJlviVM?^oR&5ey6(u~1A+~gue7s~Vz9rzTB-H*vA2%PNFQwQ>TxQG z9+xb2we!%`7J(+Sy{K|OW=RiKHjBa3)vO)z0OR4(#!~NI7FxC+L18|9@6ApA5ktnO zrKOD>9IN*W)s|so;;yKa?6er%*$KWmcE*EGOo;%>t5;779T5#4y}7+trRpR(-RN^mld8pw*QA8X1O(g z{YP%Y%hO@9@d$KOlR;eac)!OrXp0Mh4bb6~$yD;Ls&j|E25r>EogaMg!S+2TuQhgc zMMlOX3>prU#gym(8LxF-q`^is9W!OZ&?rBsnp8AhdJ{IP;T$Xpe!fytQPMHP^SC^rD-=5q77ncNI+GPc?v88t&wb#;B44L-0SGBkO=ZVQqN#G@_9 zi-7!U<8c`u6Q&WhCFPY4Kdz-|*P%AHOY~F0V9%fF5kJAeI%y0W8roYrweM_P|NYf7 zKYq7q^G|G(C~9xCwzB+P7_6_*VCCil5O&>62{2(B!Wq3Si2^q9f18Vjr_1BB?qmZ} zAqN;k=L#Ac>ka4b2fgq z{e#UruuT)Smc9RB;VO(}tyW`?%uWn zlh;E8Zc2&Y^3T^Y5QOGAa5Ia!SFt?zbvA25OG`uLl{06`Z#Zw%ocVuuwsAAFZpoZ5 zEG1>+q{$Pd=LDmXZK*q^q;FRbI2y+Tfd@ z=xlCmv)|O&f;btZNR%pAFBid0;_mcxwjo?17mk2Qeo_kJ)LacY(;IyKEjZCHZzJiC9%Yd9M;GIgw_UY>X=b%) zERV}7l60vIu2;r{Yj#Sxg8V13O@8W#&O&_wEvcbtL$LJU=QYnJz)T^liWi41L%Un4N`ana7RK`s; zr2MFUa#sx>G_!a!Y&lFUQDny?-!Mzx4wYs&1{y4qNW^_* ze~+b|GWqOzON~uXvr77=bgY7p@C9R_d0M-cA9fB$qmlZmiJLqbV)wq5qn`i!`CPx= zmrs^6%azaExvqTn?|0>c{5<)P?Nao=C@lEP98m+Z-z1p-u_sSpNF_$ zA3u*hic>JkeaQs5I>Z&@`8~4rh>rwMFNG*{_OctEAJ=cVc?v3pv{^To0+@|1Q8;(< z*eTosN?brlvvT(7+&92CcUvL!P$a#EJa;6EM_60)}8oVOc}{)gvFg+tA!z*-&3q zUeg^Mb^gRDNJxT1L*qxl3wmTk1W-G%YlBVT#q#nS@RLjjVB==Zx~JAGoi$>__j|(T zu9=M;(!1GHJ$>3Rdi3bQQ4q7Ij2zi@!mRc7 zTgZb02rT;%%LBr61FW^ZDgo0iU2;-#a$;N}*x2NOLt=CY{n66Wj3^us^SQmDh!%u4 zaFb;j22e%Cm144Lf=Qv;QrgqK=zeM8xqV=ZcV zdxBYu9a~$M&mF7b>CPU29?mau!sLk)7Y$N7%vv%BsA_E>WHMMIsyn-?ip#2QI2t=U e%~}iuoh~$xLRo=ZP+Q8*9y@uqqUH)pW&aByuM@uj literal 0 HcmV?d00001 diff --git a/css/chat_style-cai-chat-square.css b/css/chat_style-cai-chat-square.css index d626dbb1c8..854fff607c 100644 --- a/css/chat_style-cai-chat-square.css +++ b/css/chat_style-cai-chat-square.css @@ -16,6 +16,6 @@ } .message { - padding-bottom: 30px; + padding-bottom: 2em; grid-template-columns: 70px minmax(0, 1fr); } diff --git a/css/chat_style-cai-chat.css b/css/chat_style-cai-chat.css index 618184cfab..d7b1ba88e4 100644 --- a/css/chat_style-cai-chat.css +++ b/css/chat_style-cai-chat.css @@ -1,7 +1,7 @@ .message { display: grid; grid-template-columns: 60px minmax(0, 1fr); - padding-bottom: 15px; + padding-bottom: 2em; font-size: 15px; font-family: 'Noto Sans', Helvetica, Arial, sans-serif; line-height: 22.5px !important; diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css index 50b9402f4d..f1eb2531c1 100644 --- a/css/html_instruct_style.css +++ b/css/html_instruct_style.css @@ -1,74 +1,99 @@ .chat { background: transparent; - padding: 24px 19px; - padding-right: 19px !important; + padding: 0; padding-top: 0; } -.chat > .messages { - padding-top: 18px !important; +.chat > .messages:first-child { + padding-top: 0 !important; } -.message { - display: grid; - grid-template-columns: 60px 1fr; - padding-bottom: 25px; - font-size: 15px; - font-family: 'Noto Sans', Helvetica, Arial, sans-serif; - line-height: 24px; +.chat > .messages > :last-child { + margin-bottom: 1.7rem !important; } -.message:first-child { - padding-top: 0; +.chat .message-body p, .chat .message-body li { + font-size: 1rem !important; + line-height: 28px !important; } -.username { - display: none; +.dark .chat .message-body p, .dark .chat .message-body li { + color: #d1d5db !important; } -.message-body p, .message-body li { - font-size: 15px !important; - line-height: 24px !important; +.chat .message-body p, +.chat .message-body ul, +.chat .message-body ol { + margin-top: 1.25em !important; + margin-bottom: 1.25em !important; } -.message-body p, .chat .message-body ul, .chat .message-body ol { - margin-bottom: 16px !important; +.chat .message-body p:first-child, +.chat .message-body ul:first-child, +.chat .message-body ol:first-child { + margin-top: 0 !important; } -.message-body p:last-child, .chat .message-body ul:last-child, .chat .message-body ol:last-child { +.chat .message-body p:last-child, +.chat .message-body ul:last-child, +.chat .message-body ol:last-child { margin-bottom: 0 !important; } -.gradio-container .chat .assistant-message { - padding: 20px; +.chat .message-body li { + margin-top: 1.25em !important; + margin-bottom: 1.25em !important; +} + +.user-message, .assistant-message { + font-family: Inter, Helvetica, Arial, sans-serif; +} + +.message:first-child { + padding-top: 0; +} + +.username { + display: none; +} + +.chat .user-message { + padding: 1.5rem 1rem; + border-radius: 0; + border-bottom-right-radius: 0; +} + +.chat .assistant-message { background: #f4f4f4; - margin-top: 9px !important; - margin-bottom: 12px !important; - border-radius: 7px; - border: 1px solid var(--border-color-primary); + padding: 1.5rem 1rem; + border-radius: 0; + border: 0; +} + +.dark .chat .user-message { + background: transparent; } .dark .chat .assistant-message { - background: var(--color-grey-800); + background: var(--light-gray); } -.gradio-container .chat .user-message { - padding: 20px; - padding-left: 0; - padding-right: 0; - background-color: transparent; - border-radius: 8px; - border-bottom-right-radius: 0; +.chat .user-message .text, +.chat .assistant-message .text { + max-width: 40.25rem; + margin-left: auto; + margin-right: auto; } -.gradio-container .chat .assistant-message:last-child, .gradio-container .chat .user-message:last-child { - margin-bottom: 0 !important; +/* Create space between two assistant messages in a row */ +.assistant-message + .assistant-message { + margin-top: 1.5rem; } -code { +pre > code { background-color: #f3f4f6 !important; } -.dark code { +.dark pre > code { background-color: #1f2937 !important; } diff --git a/css/main.css b/css/main.css index 26e58b5cd1..a886bcd945 100644 --- a/css/main.css +++ b/css/main.css @@ -1,7 +1,41 @@ +:root { + --darker-gray: #202123; + --dark-gray: #343541; + --light-gray: #444654; + --light-theme-gray: #f4f4f4; + --border-color-dark: #525252; + --header-width: 112px; + --selected-item-color-dark: #32333e; +} + +@font-face { + font-family: Inter; + src: url('file/css/Inter/Inter-VariableFont_opsz,wght.ttf') format('truetype'); + font-weight: 100 900; + font-style: normal; +} + +@font-face { + font-family: Inter; + src: url('file/css/Inter/Inter-Italic-VariableFont_opsz,wght.ttf') format('truetype'); + font-weight: 100 900; + font-style: italic; +} + .tabs.svelte-710i53 { margin-top: 0 } +.padded.svelte-12cmxck { + padding: 3px 0; +} + +div.svelte-sfqy0y, +div.svelte-iyf88w { + background: transparent; + border: 0; +} + .py-6 { padding-top: 2.5rem } @@ -19,7 +53,7 @@ height: 39.594px; align-self: end; line-height: 1em; - border-radius: 0.5em; + border-radius: 0.375rem; flex: none; } @@ -46,10 +80,6 @@ min-height: 0 } -.dark svg { - fill: white; -} - .dark a { color: white !important; } @@ -62,14 +92,20 @@ ol li p, ul li p { border: 0; } +#default-tab, #notebook-tab, #parameters, #chat-settings, #lora, #training-tab, #model-tab, #session-tab { + padding: 1rem; +} + .gradio-container { max-width: 100% !important; padding-top: 0 !important; } #extensions { - margin-top: 5px; - margin-bottom: 35px; + margin: 5px auto 35px; + max-width: 880px; + padding: 1em; + padding-left: calc(var(--header-width) + 1em); } .extension-tab { @@ -86,20 +122,29 @@ div.svelte-15lo0d8 > *, div.svelte-15lo0d8 > .form > * { } gradio-app > :first-child { - padding-left: var(--size-4) !important; - padding-right: var(--size-4) !important; + padding: 0 !important; } .header_bar { - background-color: #f4f4f4; box-shadow: 0 0 3px rgba(22 22 22 / 35%); margin-bottom: 0; overflow-x: scroll; - margin-left: calc(-1 * var(--size-4)); - margin-right: calc(-1 * var(--size-4)); - display: block !important; text-wrap: nowrap; z-index: 90; + position: fixed; + display: flex !important; + flex-direction: column; + height: 100dvh; + width: var(--header-width); +} + +.header_bar button { + margin: 0; + padding: 0.75rem; +} + +.header_bar button.selected { + border: 0; } .dark .header_bar { @@ -113,23 +158,23 @@ gradio-app > :first-child { } .textbox_default textarea { - height: calc(100dvh - 271px); + height: calc(100dvh - 201px); } .textbox_default_output textarea { - height: calc(100dvh - 185px); + height: calc(100dvh - 117px); } .textbox textarea { - height: calc(100dvh - 241px); + height: calc(100dvh - 172px); } .textbox_logits textarea { - height: calc(100dvh - 236px); + height: calc(100dvh - 205px); } .textbox_logits_notebook textarea { - height: calc(100dvh - 292px); + height: calc(100dvh - 221px); } .monospace textarea { @@ -149,24 +194,6 @@ gradio-app > :first-child { color: #efefef !important; } -@media screen and (width <= 711px) { - .textbox_default textarea { - height: calc(100dvh - 259px); - } - - div .default-token-counter { - top: calc( 0.5 * (100dvh - 236px) ) !important; - } - - .transparent-substring { - display: none; - } - - .hover-menu { - min-width: 250px !important; - } -} - /* Hide the gradio footer */ footer { display: none !important; @@ -227,11 +254,13 @@ button { .pretty_scrollbar::-webkit-scrollbar-thumb, .pretty_scrollbar::-webkit-scrollbar-thumb:hover { background: var(--neutral-300); + border-radius: 30px; } .dark .pretty_scrollbar::-webkit-scrollbar-thumb, .dark .pretty_scrollbar::-webkit-scrollbar-thumb:hover { - background: var(--neutral-700); + background: #ccc; + border-radius: 10px; } .pretty_scrollbar::-webkit-resizer { @@ -239,7 +268,8 @@ button { } .dark .pretty_scrollbar::-webkit-resizer { - background: #374151; + background: #ccc; + border-radius: 10px; } .pretty_scrollbar::-webkit-scrollbar-corner { @@ -251,20 +281,26 @@ audio { } /* Copied from https://github.com/AUTOMATIC1111/stable-diffusion-webui */ -.token-counter { +#default-token-counter, #notebook-token-counter { position: absolute !important; - top: calc( 0.5 * (100dvh - 218px) ) !important; - right: 2px; z-index: 100; background: var(--input-background-fill) !important; min-height: 0 !important; + width: 0; + text-align: left; + direction: rtl; + right: 5px; +} + +#default-token-counter { + top: calc(100dvh - 200px) !important; } -.default-token-counter { - top: calc( 0.5 * (100dvh - 248px) ) !important; +#notebook-token-counter { + top: calc(100dvh - 171px) !important; } -.token-counter span { +#default-token-counter span, #notebook-token-counter span { padding: 1px; box-shadow: 0 0 0 0.3em rgb(192 192 192 / 15%), inset 0 0 0.6em rgb(192 192 192 / 7.5%); border: 2px solid rgb(192 192 192 / 40%) !important; @@ -272,15 +308,15 @@ audio { } .no-background { - background: var(--background-fill-primary) !important; + background: transparent; padding: 0 !important; } /* ---------------------------------------------- Chat tab ---------------------------------------------- */ -.h-\[40vh\], .wrap.svelte-byatnx.svelte-byatnx.svelte-byatnx { - height: 66.67vh +.h-\[40dvh\], .wrap.svelte-byatnx.svelte-byatnx.svelte-byatnx { + height: 66.67dvh } .gradio-container { @@ -310,7 +346,13 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { } #chat-tab { - padding-top: 0; + padding: 0; +} + +#chat-tab > :nth-child(1) { + display: flex; + flex-direction: row; + gap: 0; } #chat-tab button#Generate, #chat-tab button#stop { @@ -322,7 +364,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { } #chat-tab > :first-child, #extensions { - max-width: 880px; margin-left: auto; margin-right: auto; } @@ -342,61 +383,49 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { .chat { margin-left: auto; margin-right: auto; - max-width: 880px; min-height: var(--chat-height); overflow-y: auto; - padding-right: 15px; display: flex; flex-direction: column; word-break: break-word; overflow-wrap: anywhere; border-top: none; - border-radius: 0 0 0 8px; + border-radius: 0; visibility: visible; } .chat-parent { - height: calc(100dvh - 98px - var(--header-height) - var(--input-delta)); + height: calc(100dvh - 98px - var(--input-delta)); overflow: auto !important; border-radius: 0 !important; margin-bottom: var(--input-delta) !important; } -/* On desktop, automatically hide the chat scroll bar - * when not hovered. */ -@media (hover: hover) and (pointer: fine) { - .chat-parent { - visibility: hidden; - } - - .chat-parent:focus, .chat-parent:hover { - visibility: visible; - } -} - .chat-parent .prose { visibility: visible; } -.old-ui .chat-parent { - height: calc(100dvh - 192px - var(--header-height) - var(--input-delta)); - margin-bottom: var(--input-delta) !important; +.chat .message { + width: min(100%, 48rem); + margin-left: auto; + margin-right: auto; + text-align: start; + padding-left: 1rem; + padding-right: 1rem; } .chat-parent.bigchat { - height: calc(100dvh - 98px - var(--header-height) - var(--input-delta)) !important; + height: calc(100dvh - 98px - var(--input-delta)) !important; margin-bottom: var(--input-delta) !important; } .chat > .messages { display: flex; flex-direction: column; - padding-top: 25px; } -.chat .message:last-child { - margin-bottom: 0 !important; - padding-bottom: 15px !important; +.chat > .messages > :first-child { + padding-top: 20px; } .message-body h1, @@ -423,12 +452,12 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { padding-inline-start: 2em; } -.message-body li:not(:last-child) { - margin-top: 0 !important; - margin-bottom: 2px !important; +.chat .message-body li:not(:last-child) { + margin-top: 0; + margin-bottom: 2px; } -.message-body li:last-child { +.chat .message-body li:last-child { margin-bottom: 0 !important; } @@ -460,7 +489,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { margin: 0; } -.message-body code { +.message-body pre > code { white-space: pre-wrap !important; word-wrap: break-word !important; border: 1px solid #666; @@ -471,7 +500,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { color: #1f2328; } -.dark .message-body code { +.dark .message-body pre > code { background: #0d1117 !important; color: rgb(201 209 217); } @@ -481,8 +510,18 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { padding: 15px; } +.message-body :not(pre) > code::before { + content: "`"; +} + +.message-body :not(pre) > code::after { + content: "`"; +} + .message-body :not(pre) > code { white-space: normal !important; + font-weight: bold; + font-family: unset; } #chat-input { @@ -492,6 +531,15 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { border: none; } +#chat-input textarea { + padding: 0.65rem 2.5rem; +} + +#chat-input textarea::placeholder { + white-space: nowrap; + overflow: hidden; +} + #chat-input textarea:focus { box-shadow: none !important; } @@ -504,6 +552,14 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { display: none; } +.chat-input-positioned { + position: absolute; + bottom: 0; + max-width: 54rem; + left: 50%; + transform: translateX(-50%); +} + @media print { body { visibility: hidden; @@ -539,7 +595,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { #show-controls { position: absolute; - height: 100%; background-color: transparent; border: 0 !important; border-radius: 0; @@ -548,7 +603,8 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { #show-controls label { z-index: 1000; position: absolute; - right: 0; + right: 30px; + top: 10px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; @@ -630,7 +686,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { position: absolute; bottom: 80%; left: 0; - background-color: var(--background-fill-primary); box-shadow: 0 0 5px rgb(0 0 0 / 25%); z-index: 10000; min-width: 330px; @@ -641,7 +696,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { width: 100%; background: transparent !important; border-radius: 0 !important; - border-color: var(--border-color-primary); justify-content: space-between; margin: 0 !important; height: 36px; @@ -663,7 +717,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { opacity: 0.333; } -#chat-tab:not(.old-ui) #chat-buttons { +#chat-tab #chat-buttons { display: none !important; } @@ -694,23 +748,37 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { } #chat-input-row { - padding-bottom: 20px; + padding-bottom: 1.5em; + padding-left: 1rem; + padding-right: 1rem; } -.old-ui #chat-input-row, #chat-input-row.bigchat { - padding-bottom: 0 !important; +#chat-input-row.bigchat { + padding-bottom: 1px !important; } #chat-col { padding-bottom: 100px; } -.old-ui #chat-col, #chat-col.bigchat { - padding-bottom: 80px !important; +@media screen and (width <= 924px) { + #chat-col { + padding-bottom: 100px; + margin-top: 32px; + position: relative; /* Ensure positioning for the pseudo-element */ + } + + .chat-parent { + height: calc(100dvh - 98px - var(--input-delta) - 32px); + } + + .chat-parent.bigchat { + height: calc(100dvh - 98px - var(--input-delta) - 32px) !important; + } } -.old-ui #chat-buttons #clear-history-confirm { - order: -1; +#chat-col.bigchat { + padding-bottom: 80px !important; } .chat ol, .chat ul { @@ -725,26 +793,36 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { } /* ---------------------------------------------- - Past chat histories in a side bar on desktop + Create the sidebars ---------------------------------------------- */ -@media screen and (width >= 1327px) { - #past-chats-row { - position: absolute; - top: 36px; - left: 0; - width: calc(0.5*(var(--document-width) - 880px - 120px - 16px*2)); - max-width: 300px; - margin-left: calc(-0.5*(var(--document-width) - 880px - 14px - 16px * 2)); - } +#chat-controls, +#past-chats-row { + width: 260px; + max-width: 80vw; + padding: 0.5rem; + height: 100dvh; + flex-shrink: 0; + box-sizing: content-box; + z-index: 10; +} - #chat-controls { - position: absolute; - top: 16px; - right: 0; - width: calc(0.5*(var(--document-width) - 880px - 120px - 16px*2)); - max-width: 400px; - margin-right: calc(-0.5*(var(--document-width) - 880px - 14px - 16px * 2)); - } +#past-chats-row:not(.negative-header) { + max-width: calc(85vw - var(--header-width)); +} + +#chat-controls { + padding: 1rem; + padding-bottom: 0; +} + +#chat-controls > :nth-child(1) { + padding: 0.5rem; +} + +#past-chats-row + * { + width: unset; + flex-grow: 1; + flex-shrink: 1; } /* ---------------------------------------------- @@ -761,12 +839,12 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { position: fixed; bottom: 0; left: 0; - width: calc((100vw - 880px - 120px) /2); + width: calc(100vw / 2 - 600px); + z-index: 10000; } .pfp_character { position: relative; - z-index: 100; } .pfp_character:hover { @@ -780,10 +858,14 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { } #past-chats { - max-height: calc(100vh - 195px); + max-height: calc(100dvh - 90px); overflow-y: scroll !important; border-radius: 0; - scrollbar-width: none; /* Hide scrollbar in Firefox by default */ + scrollbar-width: auto; +} + +#past-chats::-webkit-scrollbar { + display: block; } #past-chats label { @@ -794,6 +876,24 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { border-radius: 0; padding-top: 8px; padding-bottom: 8px; + position: relative; + min-height: 42px !important; +} + +#past-chats label::before { + content: url('data:image/svg+xml;utf8,'); + position: absolute; + top: 12px; + left: 12px; + margin-right: 8px; +} + +.dark #past-chats label::before { + content: url('data:image/svg+xml;utf8,'); +} + +#past-chats label span { + margin-left: 29px; } #past-chats > :nth-child(2) { @@ -801,23 +901,264 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { } #past-chats > :nth-child(3) { - gap: 0; + gap: 0.25rem; } -#past-chats::-webkit-scrollbar { +#past-chats input { display: none; } -#past-chats:hover { - scrollbar-width: auto; +#past-chats label { + padding: 0.75rem; + font-size: 12.5px; + font-weight: 400; } -#past-chats:hover::-webkit-scrollbar { - display: block; +#past-chats .selected, +#past-chats label:hover { + border-radius: 0.5rem; +} + +#past-chats label:hover { + cursor: pointer; +} + +#past-chats-buttons, +#delete-chat-row, +#rename-row { + width: 100%; + justify-content: center; +} + + +#past-chats-row, +#chat-controls { + width: 260px; + padding: 0.5rem; + height: calc(100dvh - 16px); + flex-shrink: 0; + box-sizing: content-box; +} + +.sidebar-hidden { + width: 0 !important; + padding: 0 !important; + overflow: hidden; +} + +#past-chats-toggle, +#chat-controls-toggle, +#navigation-toggle { + display: flex; + align-items: center; + justify-content: center; + cursor: pointer; + user-select: none; + border-radius: 3px; + z-index: 1000; + position: fixed; + width: 2rem; + height: 2rem; + top: 0; +} + +#past-chats-toggle svg, +#chat-controls-toggle svg, +#navigation-toggle svg { + pointer-events: none; +} + +#past-chats-row:not(.negative-header, .sidebar-hidden) { + margin-left: 112px; +} + +@media screen and (width <= 408px) { + #past-chats-toggle.past-chats-open { + top: 28px; + } + + #chat-controls-toggle.chat-controls-open { + top: 28px; + right: calc(16px + min(260px, 80vw)) !important; + } +} + +#past-chats-toggle.past-chats-open.negative-header { + left: calc(min(260px, 85vw) + 16px); +} + +#past-chats-toggle.past-chats-open:not(.negative-header) { + left: calc(112px + min(260px, calc(85vw - var(--header-width))) + 16px); +} + +#past-chats-toggle.past-chats-closed:not(.negative-header) { + left: 112px; +} + +#past-chats-toggle.past-chats-closed.negative-header { + left: 0; + top: 28px; +} + +@media screen and (width <= 924px) { + #past-chats-toggle.past-chats-closed.negative-header { + left: 28px; + top: 0; + } +} + +.header_bar ~ *:not(#chat-tab) { + margin-left: var(--header-width); +} + +/* Positions for chat-controls-toggle */ +#chat-controls-toggle.chat-controls-open { + right: calc(min(260px, 80vw) + 23px); +} + +#chat-controls-toggle.chat-controls-closed { + right: 7px; +} + +@media screen and (width <= 924px) { + #chat-controls.sidebar-shown { + position: fixed; + right: 0; + } + + #past-chats-row.sidebar-shown { + position: fixed; + } +} + +/* ---------------------------------------------- + Dark theme +---------------------------------------------- */ +.dark .header_bar { + background-color: var(--darker-gray) !important; +} + +.dark .header_bar button.selected { + background: var(--selected-item-color-dark); +} + +.dark #chat-input textarea { + background: var(--light-gray); + color: white !important; + border-color: #292c3b; +} + +.dark #chat-input textarea::placeholder { + color: #9ca3af; +} + +.dark .hover-menu { + background-color: var(--darker-gray); +} + +.dark .hover-menu button { + border-color: var(--border-color-primary); +} + +.dark #chat-controls, +.dark #past-chats-row { + background-color: var(--darker-gray); + border: 0 !important; +} + +.dark #past-chats .selected, +.dark #past-chats label:hover { + background-color: var(--selected-item-color-dark) !important; +} + +.dark #past-chats-row, +.dark #chat-controls { + background-color: var(--darker-gray); +} + +.dark #past-chats-toggle, +.dark #chat-controls-toggle, +.dark #navigation-toggle { + color: white; +} + +.dark svg { + fill: white; + color: white; +} + +@media screen and (width <= 408px) { + .dark #past-chats-toggle.past-chats-open { + background: var(--darker-gray); + } + + .dark #chat-controls-toggle.chat-controls-open { + background: var(--darker-gray); + } +} + +/* ---------------------------------------------- + Light theme +---------------------------------------------- */ +.header_bar { + background-color: var(--light-theme-gray) !important; +} + +.header_bar button.selected { + background: white; +} + +#chat-controls, +#past-chats-row { + background-color: var(--light-theme-gray); +} + +#chat-controls { + border-left: 1px solid #d9d9d0; +} + +#past-chats-row { + border-right: 1px solid #d9d9d0; +} + +#past-chats-toggle, +#chat-controls-toggle, +#navigation-toggle { + color: gray !important; } -@media screen and (width < 1327px) { - #past-chats { - max-height: 300px; +.mobile-top-bar { + position: fixed; + top: 0; + left: 0; + width: 100%; + height: 32px; + z-index: 2; + opacity: 0; + pointer-events: none; +} + +@media screen and (width <= 924px) { + .mobile-top-bar { + opacity: 1; + pointer-events: auto; + } + + .dark .mobile-top-bar { + background-color: var(--darker-gray); + } + + .mobile-top-bar { + background-color: var(--light-theme-gray); + } +} + +@media screen and (width <= 408px) { + #past-chats-toggle.past-chats-open { + background: var(--light-theme-gray); + } + + #chat-controls-toggle.chat-controls-open { + background: var(--light-theme-gray); } } diff --git a/js/main.js b/js/main.js index 3028afac1f..997701f594 100644 --- a/js/main.js +++ b/js/main.js @@ -18,16 +18,18 @@ document.querySelector(".header_bar").addEventListener("click", function(event) if (extensionsVisible) { if (extensions) { extensions.style.display = "flex"; - extensions.style.maxWidth = chatVisible ? "880px" : "none"; - extensions.style.padding = chatVisible ? "0px" : "15px"; } + this.style.marginBottom = chatVisible ? "0px" : "19px"; if (chatVisible && !showControlsChecked) { - document.querySelectorAll("#chat-tab > div > :nth-child(n+2), #extensions").forEach(element => { + document.querySelectorAll( + "#chat-tab > div > :nth-child(1), #chat-tab > div > :nth-child(3), #chat-tab > div > :nth-child(4), #extensions" + ).forEach(element => { element.style.display = "none"; }); } + } else { this.style.marginBottom = "19px"; if (extensions) extensions.style.display = "none"; @@ -132,8 +134,7 @@ targetElement.addEventListener("scroll", function() { const observer = new MutationObserver(function(mutations) { updateCssProperties(); - const firstChild = targetElement.children[0]; - if (firstChild.classList.contains("generating")) { + if (targetElement.classList.contains("_generating")) { typing.parentNode.classList.add("visible-dots"); document.getElementById("stop").style.display = "flex"; document.getElementById("Generate").style.display = "none"; @@ -255,7 +256,7 @@ for (i = 0; i < slimDropdownElements.length; i++) { // The show/hide events were adapted from: // https://github.com/SillyTavern/SillyTavern/blob/6c8bd06308c69d51e2eb174541792a870a83d2d6/public/script.js //------------------------------------------------ -var buttonsInChat = document.querySelectorAll("#chat-tab:not(.old-ui) #chat-buttons button"); +var buttonsInChat = document.querySelectorAll("#chat-tab #chat-buttons button"); var button = document.getElementById("hover-element-button"); var menu = document.getElementById("hover-menu"); var istouchscreen = (navigator.maxTouchPoints > 0) || "ontouchstart" in document.documentElement; @@ -290,12 +291,6 @@ if (buttonsInChat.length > 0) { thisButton.innerHTML = newText; } } -} else { - buttonsInChat = document.querySelectorAll("#chat-tab.old-ui #chat-buttons button"); - for (let i = 0; i < buttonsInChat.length; i++) { - buttonsInChat[i].textContent = buttonsInChat[i].textContent.replace(/ \(.*?\)/, ""); - } - document.getElementById("gr-hover-container").style.display = "none"; } function isMouseOverButtonOrMenu() { @@ -339,6 +334,8 @@ menu.addEventListener("mouseleave", function () { // Add event listener for click anywhere in the document document.addEventListener("click", function (event) { + const target = event.target; + // Check if the click is outside the button/menu and the menu is visible if (!isMouseOverButtonOrMenu() && menu.style.display === "flex") { hideMenu(); @@ -361,10 +358,9 @@ for (var i = 0; i < 2; i++) { parent.insertBefore(elementToMove, parent.firstChild); //------------------------------------------------ -// Make the chat input grow upwards instead of downwards +// Position the chat input //------------------------------------------------ -document.getElementById("show-controls").parentNode.style.position = "absolute"; -document.getElementById("show-controls").parentNode.style.bottom = "0px"; +document.getElementById("show-controls").parentNode.classList.add("chat-input-positioned"); //------------------------------------------------ // Focus on the chat input @@ -444,20 +440,10 @@ function updateCssProperties() { // Check if the chat container is visible if (chatContainer.clientHeight > 0) { - var numericHeight = chatContainer.parentNode.clientHeight - chatInputHeight + 40 - 100; - if (document.getElementById("chat-tab").style.paddingBottom != "") { - numericHeight += 20; - } - - const newChatHeight = `${numericHeight}px`; + const newChatHeight = `${chatContainer.parentNode.clientHeight - chatInputHeight + 40 - 100 - 20}px`; document.documentElement.style.setProperty("--chat-height", newChatHeight); document.documentElement.style.setProperty("--input-delta", `${chatInputHeight - 40}px`); - // Get and set header height - const header = document.querySelector(".header_bar"); - const headerHeight = `${header.clientHeight}px`; - document.documentElement.style.setProperty("--header-height", headerHeight); - // Adjust scrollTop based on input height change if (chatInputHeight !== currentChatInputHeight) { if (!isScrolled && chatInputHeight < currentChatInputHeight) { @@ -568,6 +554,8 @@ function moveToChatTab() { grandParent.style.display = "none"; } + grandParent.children[0].style.minWidth = "100%"; + const chatControlsFirstChild = document.querySelector("#chat-controls").firstElementChild; const newParent = chatControlsFirstChild; let newPosition = newParent.children.length - 2; @@ -586,6 +574,7 @@ function restoreOriginalPosition() { document.getElementById("save-character").style.display = ""; movedElement.style.display = ""; + movedElement.children[0].style.minWidth = ""; } } @@ -612,3 +601,206 @@ window.addEventListener("beforeunload", function (event) { }); moveToChatTab(); + +//------------------------------------------------ +// Buttons to toggle the sidebars +//------------------------------------------------ + +const leftArrowSVG = ` + + + + + +`; + +const rightArrowSVG = ` + + + + + +`; + +const hamburgerMenuSVG = ` + + + + +`; + +const closeMenuSVG = ` + + + +`; + +const chatTab = document.getElementById("chat-tab"); +const pastChatsRow = document.getElementById("past-chats-row"); +const chatControlsRow = document.getElementById("chat-controls"); + +if (chatTab) { + // Create past-chats-toggle div + const pastChatsToggle = document.createElement("div"); + pastChatsToggle.id = "past-chats-toggle"; + pastChatsToggle.innerHTML = leftArrowSVG; // Set initial icon to left arrow + pastChatsToggle.classList.add("past-chats-open"); // Set initial position + + // Create chat-controls-toggle div + const chatControlsToggle = document.createElement("div"); + chatControlsToggle.id = "chat-controls-toggle"; + chatControlsToggle.innerHTML = rightArrowSVG; // Set initial icon to right arrow + chatControlsToggle.classList.add("chat-controls-open"); // Set initial position + + // Append both elements to the chat-tab + chatTab.appendChild(pastChatsToggle); + chatTab.appendChild(chatControlsToggle); +} + +// Create navigation toggle div +const navigationToggle = document.createElement("div"); +navigationToggle.id = "navigation-toggle"; +navigationToggle.innerHTML = leftArrowSVG; // Set initial icon to right arrow +navigationToggle.classList.add("navigation-left"); // Set initial position +headerBar.appendChild(navigationToggle); + +// Retrieve the dynamically created toggle buttons +const pastChatsToggle = document.getElementById("past-chats-toggle"); +const chatControlsToggle = document.getElementById("chat-controls-toggle"); + +function toggleSidebar(sidebar, toggle) { + const isCurrentlyHidden = sidebar.classList.contains("sidebar-hidden"); + const shouldClose = !isCurrentlyHidden; + + // Apply visibility classes + sidebar.classList.toggle("sidebar-hidden", shouldClose); + sidebar.classList.toggle("sidebar-shown", !shouldClose); + + if (sidebar === headerBar) { + // Special handling for header bar + document.documentElement.style.setProperty("--header-width", shouldClose ? "0px" : "112px"); + pastChatsRow.classList.toggle("negative-header", shouldClose); + pastChatsToggle.classList.toggle("negative-header", shouldClose); + toggle.innerHTML = shouldClose ? hamburgerMenuSVG : closeMenuSVG; + } else if (sidebar === pastChatsRow) { + // Past chats sidebar + toggle.classList.toggle("past-chats-closed", shouldClose); + toggle.classList.toggle("past-chats-open", !shouldClose); + toggle.innerHTML = shouldClose ? rightArrowSVG : leftArrowSVG; + } else if (sidebar === chatControlsRow) { + // Chat controls sidebar + toggle.classList.toggle("chat-controls-closed", shouldClose); + toggle.classList.toggle("chat-controls-open", !shouldClose); + toggle.innerHTML = shouldClose ? leftArrowSVG : rightArrowSVG; + } + + // Mobile handling + if (isMobile()) { + sidebar.classList.toggle("sidebar-shown", !shouldClose); + } +} + +// Function to check if the device is mobile +function isMobile() { + return window.innerWidth <= 924; +} + +// Function to initialize sidebars +function initializeSidebars() { + const isOnMobile = isMobile(); + + if (isOnMobile) { + // Mobile state: Hide sidebars and set closed states + [pastChatsRow, chatControlsRow, headerBar].forEach(el => { + el.classList.add("sidebar-hidden"); + el.classList.remove("sidebar-shown"); + }); + + document.documentElement.style.setProperty("--header-width", "0px"); + pastChatsRow.classList.add("negative-header"); + pastChatsToggle.classList.add("negative-header", "past-chats-closed"); + pastChatsToggle.classList.remove("past-chats-open"); + + [chatControlsToggle, navigationToggle].forEach(el => { + el.classList.add("chat-controls-closed"); + el.classList.remove("chat-controls-open"); + }); + + pastChatsToggle.innerHTML = rightArrowSVG; + chatControlsToggle.innerHTML = leftArrowSVG; + navigationToggle.innerHTML = hamburgerMenuSVG; + } else { + // Desktop state: Show sidebars and set open states + [pastChatsRow, chatControlsRow].forEach(el => { + el.classList.remove("sidebar-hidden", "sidebar-shown"); + }); + + pastChatsToggle.classList.add("past-chats-open"); + pastChatsToggle.classList.remove("past-chats-closed"); + + [chatControlsToggle, navigationToggle].forEach(el => { + el.classList.add("chat-controls-open"); + el.classList.remove("chat-controls-closed"); + }); + + pastChatsToggle.innerHTML = leftArrowSVG; + chatControlsToggle.innerHTML = rightArrowSVG; + navigationToggle.innerHTML = closeMenuSVG; + } +} + +// Run the initializer when the page loads +initializeSidebars(); + +// Add an event listener to handle screen resizing +window.addEventListener("resize", initializeSidebars); + +// Add click event listeners to toggle buttons +pastChatsToggle.addEventListener("click", () => { + toggleSidebar(pastChatsRow, pastChatsToggle); +}); + +chatControlsToggle.addEventListener("click", () => { + toggleSidebar(chatControlsRow, chatControlsToggle); +}); + +navigationToggle.addEventListener("click", () => { + toggleSidebar(headerBar, navigationToggle); +}); + +//------------------------------------------------ +// Fixes #chat-input textarea height issue +// for devices with width <= 924px +//------------------------------------------------ + +if (isMobile()) { + // Target the textarea + const textarea = document.querySelector("#chat-input textarea"); + + if (textarea) { + // Simulate adding and removing a newline + textarea.value += "\n"; + textarea.dispatchEvent(new Event("input", { bubbles: true })); + textarea.value = textarea.value.slice(0, -1); + textarea.dispatchEvent(new Event("input", { bubbles: true })); + } +} + +//------------------------------------------------ +// Create a top navigation bar on mobile +//------------------------------------------------ + +function createMobileTopBar() { + const chatTab = document.getElementById("chat-tab"); + + // Only create the top bar if it doesn't already exist + if (chatTab && !chatTab.querySelector(".mobile-top-bar")) { + const topBar = document.createElement("div"); + topBar.classList.add("mobile-top-bar"); + + // Insert the top bar as the first child of chat-tab + chatTab.appendChild(topBar); + } +} + +createMobileTopBar(); diff --git a/js/show_controls.js b/js/show_controls.js index 1ff88e52aa..1a87b52d96 100644 --- a/js/show_controls.js +++ b/js/show_controls.js @@ -1,4 +1,6 @@ -const belowChatInput = document.querySelectorAll("#chat-tab > div > :nth-child(n+2), #extensions"); +const belowChatInput = document.querySelectorAll( + "#chat-tab > div > :nth-child(1), #chat-tab > div > :nth-child(3), #chat-tab > div > :nth-child(4), #extensions" +); const chatParent = document.querySelector(".chat-parent"); function toggle_controls(value) { diff --git a/modules/chat.py b/modules/chat.py index 787277b997..92808fb7de 100644 --- a/modules/chat.py +++ b/modules/chat.py @@ -352,13 +352,17 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess for j, reply in enumerate(generate_reply(prompt, state, stopping_strings=stopping_strings, is_chat=True, for_ui=for_ui)): # Extract the reply - visible_reply = reply if state['mode'] in ['chat', 'chat-instruct']: - visible_reply = re.sub("(||{{user}})", state['name1'], reply) + visible_reply = re.sub("(||{{user}})", state['name1'], reply + '❚') + else: + visible_reply = reply + '❚' visible_reply = html.escape(visible_reply) if shared.stop_everything: + if output['visible'][-1][1].endswith('❚'): + output['visible'][-1][1] = output['visible'][-1][1][:-1] + output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True) yield output return @@ -374,6 +378,9 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess if is_stream: yield output + if output['visible'][-1][1].endswith('❚'): + output['visible'][-1][1] = output['visible'][-1][1][:-1] + output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True) yield output @@ -606,9 +613,9 @@ def find_all_histories_with_first_prompts(state): first_prompt = first_prompt.strip() - # Truncate the first prompt if it's longer than 32 characters - if len(first_prompt) > 32: - first_prompt = first_prompt[:29] + '...' + # Truncate the first prompt if it's longer than 30 characters + if len(first_prompt) > 30: + first_prompt = first_prompt[:30-3] + '...' result.append((first_prompt, filename)) @@ -1087,9 +1094,8 @@ def handle_delete_chat_confirm_click(state): def handle_rename_chat_click(): return [ - gr.update(visible=True, value="My New Chat"), + gr.update(value="My New Chat"), gr.update(visible=True), - gr.update(visible=True) ] @@ -1100,8 +1106,6 @@ def handle_rename_chat_confirm(rename_to, state): return [ gr.update(choices=histories, value=rename_to), gr.update(visible=False), - gr.update(visible=False), - gr.update(visible=False) ] @@ -1209,7 +1213,7 @@ def handle_delete_template_click(template): return [ f"{template}.yaml", "instruction-templates/", - gr.update(visible=True) + gr.update(visible=False) ] diff --git a/modules/shared.py b/modules/shared.py index 894ed6fe56..599d64926b 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -81,7 +81,6 @@ group.add_argument('--settings', type=str, help='Load the default interface settings from this yaml file. See settings-template.yaml for an example. If you create a file called settings.yaml, this file will be loaded by default without the need to use the --settings flag.') group.add_argument('--extensions', type=str, nargs='+', help='The list of extensions to load. If you want to load more than one extension, write the names separated by spaces.') group.add_argument('--verbose', action='store_true', help='Print the prompts to the terminal.') -group.add_argument('--chat-buttons', action='store_true', help='Show buttons on the chat tab instead of a hover menu.') group.add_argument('--idle-timeout', type=int, default=0, help='Unload model after this many minutes of inactivity. It will be automatically reloaded when you try to use it again.') # Model loader @@ -191,6 +190,7 @@ group.add_argument('--ssl-keyfile', type=str, help='The path to the SSL certificate key file.', default=None) group.add_argument('--ssl-certfile', type=str, help='The path to the SSL certificate cert file.', default=None) group.add_argument('--subpath', type=str, help='Customize the subpath for gradio, use with reverse proxy') +group.add_argument('--old-colors', action='store_true', help='Use the legacy Gradio colors, before the December/2024 update.') # API group = parser.add_argument_group('API') @@ -213,6 +213,7 @@ group.add_argument('--checkpoint', type=str, help='DEPRECATED') group.add_argument('--monkey-patch', action='store_true', help='DEPRECATED') group.add_argument('--no_inject_fused_attention', action='store_true', help='DEPRECATED') +group.add_argument('--chat-buttons', action='store_true', help='DEPRECATED') args = parser.parse_args() args_defaults = parser.parse_args([]) diff --git a/modules/ui.py b/modules/ui.py index c07beeb466..f8a2d8aba9 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -50,6 +50,49 @@ button_secondary_border_color="var(--border-color-primary)" ) +if not shared.args.old_colors: + theme = theme.set( + # General Colors + border_color_primary='#c5c5d2', + body_text_color_subdued='#484848', + background_fill_secondary='#eaeaea', + background_fill_secondary_dark='var(--dark-gray)', + background_fill_primary='var(--neutral-50)', + background_fill_primary_dark='var(--darker-gray)', + body_background_fill="white", + block_background_fill="transparent", + body_text_color="#333", + button_secondary_background_fill="#f4f4f4", + button_secondary_border_color="var(--border-color-primary)", + + # Dark Mode Colors + input_background_fill_dark='var(--darker-gray)', + checkbox_background_color_dark='var(--darker-gray)', + block_background_fill_dark='transparent', + block_border_color_dark='transparent', + input_border_color_dark='var(--border-color-dark)', + checkbox_border_color_dark='var(--border-color-dark)', + border_color_primary_dark='var(--border-color-dark)', + button_secondary_border_color_dark='var(--border-color-dark)', + body_background_fill_dark='var(--dark-gray)', + button_primary_background_fill_dark='transparent', + button_secondary_background_fill_dark='transparent', + checkbox_label_background_fill_dark='transparent', + button_cancel_background_fill_dark='transparent', + button_secondary_background_fill_hover_dark='var(--selected-item-color-dark)', + checkbox_label_background_fill_hover_dark='var(--selected-item-color-dark)', + table_even_background_fill_dark='var(--darker-gray)', + table_odd_background_fill_dark='var(--dark-gray)', + + # Shadows and Radius + checkbox_label_shadow='none', + block_shadow='none', + block_shadow_dark='none', + button_large_radius='0.375rem', + button_large_padding='6px 12px', + input_radius='0.375rem', + ) + if Path("notification.mp3").exists(): audio_notification_js = "document.querySelector('#audio_notification audio')?.play();" else: @@ -232,10 +275,10 @@ def gather_interface_values(*args): def apply_interface_values(state, use_persistent=False): if use_persistent: state = shared.persistent_interface_state - if 'textbox-default' in state: + if 'textbox-default' in state and 'prompt_menu-default' in state: state.pop('prompt_menu-default') - if 'textbox-notebook' in state: + if 'textbox-notebook' and 'prompt_menu-notebook' in state: state.pop('prompt_menu-notebook') elements = list_interface_input_elements() diff --git a/modules/ui_chat.py b/modules/ui_chat.py index 57143cd8c0..e372f5c223 100644 --- a/modules/ui_chat.py +++ b/modules/ui_chat.py @@ -12,7 +12,6 @@ inputs = ('Chat input', 'interface_state') reload_arr = ('history', 'name1', 'name2', 'mode', 'chat_style', 'character_menu') -clear_arr = ('delete_chat-confirm', 'delete_chat', 'delete_chat-cancel') def create_ui(): @@ -21,7 +20,27 @@ def create_ui(): shared.gradio['Chat input'] = gr.State() shared.gradio['history'] = gr.JSON({'internal': [], 'visible': []}, visible=False) - with gr.Tab('Chat', elem_id='chat-tab', elem_classes=("old-ui" if shared.args.chat_buttons else None)): + with gr.Tab('Chat', elem_id='chat-tab'): + with gr.Row(elem_id='past-chats-row', elem_classes=['pretty_scrollbar']): + with gr.Column(): + with gr.Row(elem_id='past-chats-buttons'): + shared.gradio['rename_chat'] = gr.Button('Rename', elem_classes='refresh-button', interactive=not mu) + shared.gradio['delete_chat'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu) + shared.gradio['Start new chat'] = gr.Button('New chat', elem_classes=['refresh-button', 'focus-on-chat-input']) + + with gr.Row(elem_id='delete-chat-row', visible=False) as shared.gradio['delete-chat-row']: + shared.gradio['delete_chat-cancel'] = gr.Button('Cancel', elem_classes=['refresh-button', 'focus-on-chat-input']) + shared.gradio['delete_chat-confirm'] = gr.Button('Confirm', variant='stop', elem_classes=['refresh-button', 'focus-on-chat-input']) + + with gr.Row(elem_id='rename-row', visible=False) as shared.gradio['rename-row']: + shared.gradio['rename_to'] = gr.Textbox(label='Rename to:', placeholder='New name', elem_classes=['no-background']) + with gr.Row(): + shared.gradio['rename_to-cancel'] = gr.Button('Cancel', elem_classes=['refresh-button', 'focus-on-chat-input']) + shared.gradio['rename_to-confirm'] = gr.Button('Confirm', elem_classes=['refresh-button', 'focus-on-chat-input'], variant='primary') + + with gr.Row(): + shared.gradio['unique_id'] = gr.Radio(label="", elem_classes=['slim-dropdown', 'pretty_scrollbar'], interactive=not mu, elem_id='past-chats') + with gr.Row(): with gr.Column(elem_id='chat-col'): shared.gradio['display'] = gr.HTML(value=chat_html_wrapper({'internal': [], 'visible': []}, '', '', 'chat', 'cai-chat', '')) @@ -60,25 +79,6 @@ def create_ui(): shared.gradio['send-chat-to-default'] = gr.Button('Send to default') shared.gradio['send-chat-to-notebook'] = gr.Button('Send to notebook') - with gr.Row(elem_id='past-chats-row', elem_classes=['pretty_scrollbar']): - with gr.Column(): - with gr.Row(): - shared.gradio['rename_chat'] = gr.Button('Rename', elem_classes='refresh-button', interactive=not mu) - shared.gradio['delete_chat'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu) - shared.gradio['delete_chat-confirm'] = gr.Button('Confirm', variant='stop', visible=False, elem_classes=['refresh-button', 'focus-on-chat-input']) - shared.gradio['delete_chat-cancel'] = gr.Button('Cancel', visible=False, elem_classes=['refresh-button', 'focus-on-chat-input']) - shared.gradio['Start new chat'] = gr.Button('New chat', elem_classes=['refresh-button', 'focus-on-chat-input']) - - with gr.Row(elem_id='rename-row'): - shared.gradio['rename_to'] = gr.Textbox(label='Rename to:', placeholder='New name', visible=False, elem_classes=['no-background']) - with gr.Row(): - shared.gradio['rename_to-confirm'] = gr.Button('Confirm', visible=False, elem_classes=['refresh-button', 'focus-on-chat-input']) - shared.gradio['rename_to-cancel'] = gr.Button('Cancel', visible=False, elem_classes=['refresh-button', 'focus-on-chat-input']) - - gr.Markdown("Past chats") - with gr.Row(): - shared.gradio['unique_id'] = gr.Radio(label="", elem_classes=['slim-dropdown', 'pretty_scrollbar'], interactive=not mu, elem_id='past-chats') - with gr.Row(elem_id='chat-controls', elem_classes=['pretty_scrollbar']): with gr.Column(): with gr.Row(): @@ -180,29 +180,39 @@ def create_event_handlers(): shared.gradio['Generate'].click( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( lambda x: (x, ''), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False).then( + lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then( chat.generate_chat_reply_wrapper, gradio(inputs), gradio('display', 'history'), show_progress=False).then( + None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['textbox'].submit( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( lambda x: (x, ''), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False).then( + lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then( chat.generate_chat_reply_wrapper, gradio(inputs), gradio('display', 'history'), show_progress=False).then( + None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['Regenerate'].click( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( + lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then( partial(chat.generate_chat_reply_wrapper, regenerate=True), gradio(inputs), gradio('display', 'history'), show_progress=False).then( + None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['Continue'].click( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( + lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then( partial(chat.generate_chat_reply_wrapper, _continue=True), gradio(inputs), gradio('display', 'history'), show_progress=False).then( + None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['Impersonate'].click( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( lambda x: x, gradio('textbox'), gradio('Chat input'), show_progress=False).then( + lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then( chat.impersonate_wrapper, gradio(inputs), gradio('textbox', 'display'), show_progress=False).then( + None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['Replace last reply'].click( @@ -234,21 +244,21 @@ def create_event_handlers(): ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( chat.handle_start_new_chat_click, gradio('interface_state'), gradio('history', 'display', 'unique_id'), show_progress=False) - shared.gradio['delete_chat'].click(lambda: [gr.update(visible=True), gr.update(visible=False), gr.update(visible=True)], None, gradio(clear_arr)) - shared.gradio['delete_chat-cancel'].click(lambda: [gr.update(visible=False), gr.update(visible=True), gr.update(visible=False)], None, gradio(clear_arr)) + shared.gradio['delete_chat'].click(lambda: gr.update(visible=True), None, gradio('delete-chat-row')) + shared.gradio['delete_chat-cancel'].click(lambda: gr.update(visible=False), None, gradio('delete-chat-row')) shared.gradio['delete_chat-confirm'].click( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( - chat.handle_delete_chat_confirm_click, gradio('interface_state'), gradio('history', 'display', 'unique_id') + gradio(clear_arr), show_progress=False) + chat.handle_delete_chat_confirm_click, gradio('interface_state'), gradio('history', 'display', 'unique_id', 'delete-chat-row'), show_progress=False) - shared.gradio['rename_chat'].click(chat.handle_rename_chat_click, None, gradio('rename_to', 'rename_to-confirm', 'rename_to-cancel'), show_progress=False) - shared.gradio['rename_to-cancel'].click(lambda: [gr.update(visible=False)] * 3, None, gradio('rename_to', 'rename_to-confirm', 'rename_to-cancel'), show_progress=False) + shared.gradio['rename_chat'].click(chat.handle_rename_chat_click, None, gradio('rename_to', 'rename-row'), show_progress=False) + shared.gradio['rename_to-cancel'].click(lambda: gr.update(visible=False), None, gradio('rename-row'), show_progress=False) shared.gradio['rename_to-confirm'].click( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( - chat.handle_rename_chat_confirm, gradio('rename_to', 'interface_state'), gradio('unique_id', 'rename_to', 'rename_to-confirm', 'rename_to-cancel'), show_progress=False) + chat.handle_rename_chat_confirm, gradio('rename_to', 'interface_state'), gradio('unique_id', 'rename-row')) shared.gradio['rename_to'].submit( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( - chat.handle_rename_chat_confirm, gradio('rename_to', 'interface_state'), gradio('unique_id', 'rename_to', 'rename_to-confirm', 'rename_to-cancel'), show_progress=False) + chat.handle_rename_chat_confirm, gradio('rename_to', 'interface_state'), gradio('unique_id', 'rename-row'), show_progress=False) shared.gradio['load_chat_history'].upload( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( diff --git a/modules/ui_default.py b/modules/ui_default.py index 112acd2358..af418ed7dc 100644 --- a/modules/ui_default.py +++ b/modules/ui_default.py @@ -20,12 +20,12 @@ def create_ui(): with gr.Column(): with gr.Row(): shared.gradio['textbox-default'] = gr.Textbox(value='', lines=27, label='Input', elem_classes=['textbox_default', 'add_scrollbar']) - shared.gradio['token-counter-default'] = gr.HTML(value="0", elem_classes=["token-counter", "default-token-counter"]) + shared.gradio['token-counter-default'] = gr.HTML(value="0", elem_id="default-token-counter") with gr.Row(): - shared.gradio['Generate-default'] = gr.Button('Generate', variant='primary') - shared.gradio['Stop-default'] = gr.Button('Stop', elem_id='stop') shared.gradio['Continue-default'] = gr.Button('Continue') + shared.gradio['Stop-default'] = gr.Button('Stop', elem_id='stop', visible=False) + shared.gradio['Generate-default'] = gr.Button('Generate', variant='primary') with gr.Row(): shared.gradio['prompt_menu-default'] = gr.Dropdown(choices=utils.get_available_prompts(), value='None', label='Prompt', elem_classes='slim-dropdown') @@ -63,20 +63,26 @@ def create_ui(): def create_event_handlers(): shared.gradio['Generate-default'].click( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( + lambda : [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-default', 'Generate-default')).then( generate_reply_wrapper, gradio(inputs), gradio(outputs), show_progress=False).then( lambda state, left, right: state.update({'textbox-default': left, 'output_textbox': right}), gradio('interface_state', 'textbox-default', 'output_textbox'), None).then( + lambda : [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-default', 'Generate-default')).then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['textbox-default'].submit( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( + lambda : [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-default', 'Generate-default')).then( generate_reply_wrapper, gradio(inputs), gradio(outputs), show_progress=False).then( lambda state, left, right: state.update({'textbox-default': left, 'output_textbox': right}), gradio('interface_state', 'textbox-default', 'output_textbox'), None).then( + lambda : [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-default', 'Generate-default')).then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['Continue-default'].click( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( + lambda : [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-default', 'Generate-default')).then( generate_reply_wrapper, [shared.gradio['output_textbox']] + gradio(inputs)[1:], gradio(outputs), show_progress=False).then( lambda state, left, right: state.update({'textbox-default': left, 'output_textbox': right}), gradio('interface_state', 'textbox-default', 'output_textbox'), None).then( + lambda : [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-default', 'Generate-default')).then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['Stop-default'].click(stop_everything_event, None, None, queue=False) diff --git a/modules/ui_notebook.py b/modules/ui_notebook.py index 799328447c..fccb1c9874 100644 --- a/modules/ui_notebook.py +++ b/modules/ui_notebook.py @@ -23,7 +23,7 @@ def create_ui(): with gr.Tab('Raw'): with gr.Row(): shared.gradio['textbox-notebook'] = gr.Textbox(value='', lines=27, elem_id='textbox-notebook', elem_classes=['textbox', 'add_scrollbar']) - shared.gradio['token-counter-notebook'] = gr.HTML(value="0", elem_classes=["token-counter"]) + shared.gradio['token-counter-notebook'] = gr.HTML(value="0", elem_id="notebook-token-counter") with gr.Tab('Markdown'): shared.gradio['markdown_render-notebook'] = gr.Button('Render') @@ -48,10 +48,10 @@ def create_ui(): shared.gradio['tokens-notebook'] = gr.Textbox(lines=23, label='Tokens', elem_classes=['textbox_logits_notebook', 'add_scrollbar', 'monospace']) with gr.Row(): - shared.gradio['Generate-notebook'] = gr.Button('Generate', variant='primary', elem_classes='small-button') - shared.gradio['Stop-notebook'] = gr.Button('Stop', elem_classes='small-button', elem_id='stop') shared.gradio['Undo'] = gr.Button('Undo', elem_classes='small-button') shared.gradio['Regenerate-notebook'] = gr.Button('Regenerate', elem_classes='small-button') + shared.gradio['Stop-notebook'] = gr.Button('Stop', visible=False, elem_classes='small-button', elem_id='stop') + shared.gradio['Generate-notebook'] = gr.Button('Generate', variant='primary', elem_classes='small-button') with gr.Column(scale=1): gr.HTML('

    ') @@ -66,22 +66,28 @@ def create_event_handlers(): shared.gradio['Generate-notebook'].click( lambda x: x, gradio('textbox-notebook'), gradio('last_input-notebook')).then( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( + lambda : [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-notebook', 'Generate-notebook')).then( generate_reply_wrapper, gradio(inputs), gradio(outputs), show_progress=False).then( lambda state, text: state.update({'textbox-notebook': text}), gradio('interface_state', 'textbox-notebook'), None).then( + lambda : [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-notebook', 'Generate-notebook')).then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['textbox-notebook'].submit( lambda x: x, gradio('textbox-notebook'), gradio('last_input-notebook')).then( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( + lambda : [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-notebook', 'Generate-notebook')).then( generate_reply_wrapper, gradio(inputs), gradio(outputs), show_progress=False).then( lambda state, text: state.update({'textbox-notebook': text}), gradio('interface_state', 'textbox-notebook'), None).then( + lambda : [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-notebook', 'Generate-notebook')).then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['Regenerate-notebook'].click( lambda x: x, gradio('last_input-notebook'), gradio('textbox-notebook'), show_progress=False).then( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( + lambda : [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-notebook', 'Generate-notebook')).then( generate_reply_wrapper, gradio(inputs), gradio(outputs), show_progress=False).then( lambda state, text: state.update({'textbox-notebook': text}), gradio('interface_state', 'textbox-notebook'), None).then( + lambda : [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-notebook', 'Generate-notebook')).then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['Undo'].click( diff --git a/requirements.txt b/requirements.txt index 4e976d8eb3..24c92391ca 100644 --- a/requirements.txt +++ b/requirements.txt @@ -4,7 +4,7 @@ colorama datasets einops fastapi==0.112.4 -gradio==4.26.* +gradio==4.37.* jinja2==3.1.4 markdown numba==0.59.* diff --git a/requirements_amd.txt b/requirements_amd.txt index dd537a66ca..b7093d5099 100644 --- a/requirements_amd.txt +++ b/requirements_amd.txt @@ -3,7 +3,7 @@ colorama datasets einops fastapi==0.112.4 -gradio==4.26.* +gradio==4.37.* jinja2==3.1.4 markdown numba==0.59.* diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt index 275bbb22f1..88682aea04 100644 --- a/requirements_amd_noavx2.txt +++ b/requirements_amd_noavx2.txt @@ -3,7 +3,7 @@ colorama datasets einops fastapi==0.112.4 -gradio==4.26.* +gradio==4.37.* jinja2==3.1.4 markdown numba==0.59.* diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt index daeab92a2c..6588278d03 100644 --- a/requirements_apple_intel.txt +++ b/requirements_apple_intel.txt @@ -3,7 +3,7 @@ colorama datasets einops fastapi==0.112.4 -gradio==4.26.* +gradio==4.37.* jinja2==3.1.4 markdown numba==0.59.* diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt index 99b5f3c244..1fc9795bb3 100644 --- a/requirements_apple_silicon.txt +++ b/requirements_apple_silicon.txt @@ -3,7 +3,7 @@ colorama datasets einops fastapi==0.112.4 -gradio==4.26.* +gradio==4.37.* jinja2==3.1.4 markdown numba==0.59.* diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt index 446a47cefd..53fedd7ec5 100644 --- a/requirements_cpu_only.txt +++ b/requirements_cpu_only.txt @@ -3,7 +3,7 @@ colorama datasets einops fastapi==0.112.4 -gradio==4.26.* +gradio==4.37.* jinja2==3.1.4 markdown numba==0.59.* diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt index 73b9f764c1..9f52b17283 100644 --- a/requirements_cpu_only_noavx2.txt +++ b/requirements_cpu_only_noavx2.txt @@ -3,7 +3,7 @@ colorama datasets einops fastapi==0.112.4 -gradio==4.26.* +gradio==4.37.* jinja2==3.1.4 markdown numba==0.59.* diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt index 5ea4f65416..9ad138d8d4 100644 --- a/requirements_noavx2.txt +++ b/requirements_noavx2.txt @@ -4,7 +4,7 @@ colorama datasets einops fastapi==0.112.4 -gradio==4.26.* +gradio==4.37.* jinja2==3.1.4 markdown numba==0.59.* diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt index f6ba93bdfd..e2daebd9ec 100644 --- a/requirements_nowheels.txt +++ b/requirements_nowheels.txt @@ -3,7 +3,7 @@ colorama datasets einops fastapi==0.112.4 -gradio==4.26.* +gradio==4.37.* jinja2==3.1.4 markdown numba==0.59.* From 517fcc1f2324c02b952030c99b2e34cbda9e4283 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 16 Dec 2024 20:12:16 -0800 Subject: [PATCH 0095/1701] Better centralize the chat tab --- css/main.css | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) diff --git a/css/main.css b/css/main.css index a886bcd945..7d090e5742 100644 --- a/css/main.css +++ b/css/main.css @@ -968,10 +968,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { pointer-events: none; } -#past-chats-row:not(.negative-header, .sidebar-hidden) { - margin-left: 112px; -} - @media screen and (width <= 408px) { #past-chats-toggle.past-chats-open { top: 28px; @@ -1007,7 +1003,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { } } -.header_bar ~ *:not(#chat-tab) { +.header_bar ~ * { margin-left: var(--header-width); } From c43ee5db1152c82a97b1fff13d32e66b0e452c46 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 17 Dec 2024 07:59:55 -0800 Subject: [PATCH 0096/1701] UI: very minor color change --- modules/ui.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/ui.py b/modules/ui.py index f8a2d8aba9..9ef958abc3 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -56,7 +56,7 @@ border_color_primary='#c5c5d2', body_text_color_subdued='#484848', background_fill_secondary='#eaeaea', - background_fill_secondary_dark='var(--dark-gray)', + background_fill_secondary_dark='var(--selected-item-color-dark)', background_fill_primary='var(--neutral-50)', background_fill_primary_dark='var(--darker-gray)', body_background_fill="white", @@ -82,7 +82,7 @@ button_secondary_background_fill_hover_dark='var(--selected-item-color-dark)', checkbox_label_background_fill_hover_dark='var(--selected-item-color-dark)', table_even_background_fill_dark='var(--darker-gray)', - table_odd_background_fill_dark='var(--dark-gray)', + table_odd_background_fill_dark='var(--selected-item-color-dark)', # Shadows and Radius checkbox_label_shadow='none', From addad3c63eff903f853055a95eb9487df6143a3e Mon Sep 17 00:00:00 2001 From: Diner Burger Date: Tue, 17 Dec 2024 15:43:48 -0500 Subject: [PATCH 0097/1701] Allow more granular KV cache settings (#6561) --- modules/exllamav2.py | 20 +++++++++--- modules/exllamav2_hf.py | 26 +++++++++++----- modules/llamacpp_hf.py | 10 +++--- modules/llamacpp_model.py | 38 +++++++++++++++++++---- modules/loaders.py | 12 +++----- modules/shared.py | 65 +++++++++++++++++++++++++++++++++++++-- modules/ui.py | 3 +- modules/ui_model_menu.py | 3 +- 8 files changed, 140 insertions(+), 37 deletions(-) diff --git a/modules/exllamav2.py b/modules/exllamav2.py index 0498c4882e..be5ef8d3b6 100644 --- a/modules/exllamav2.py +++ b/modules/exllamav2.py @@ -2,17 +2,19 @@ from pathlib import Path import torch + from exllamav2 import ( ExLlamaV2, ExLlamaV2Cache, ExLlamaV2Cache_8bit, ExLlamaV2Cache_Q4, + ExLlamaV2Cache_Q6, + ExLlamaV2Cache_Q8, ExLlamaV2Cache_TP, ExLlamaV2Config, ExLlamaV2Tokenizer ) from exllamav2.generator import ExLlamaV2Sampler, ExLlamaV2StreamingGenerator - from modules import shared from modules.logging_colors import logger from modules.text_generation import get_max_prompt_length @@ -57,12 +59,22 @@ def from_pretrained(self, path_to_model): model.load(split) # Determine the correct cache type - if shared.args.cache_8bit: + kv_cache_type = 'fp16' + if shared.args.cache_type: + kv_cache_type = shared.args.cache_type.lower() + + if kv_cache_type == 'fp16': + cache_type = ExLlamaV2Cache + elif kv_cache_type == 'fp8': cache_type = ExLlamaV2Cache_8bit - elif shared.args.cache_4bit: + elif kv_cache_type == 'q8': + cache_type = ExLlamaV2Cache_Q8 + elif kv_cache_type == 'q6': + cache_type = ExLlamaV2Cache_Q6 + elif kv_cache_type == 'q4': cache_type = ExLlamaV2Cache_Q4 else: - cache_type = ExLlamaV2Cache + raise ValueError(f"Invalid cache type for ExLlamaV2: {cache_type}. Valid options are: fp16, fp8, q8, q6, q4.") # Use TP if specified if shared.args.enable_tp: diff --git a/modules/exllamav2_hf.py b/modules/exllamav2_hf.py index 320a8d2467..f6b943c8b7 100644 --- a/modules/exllamav2_hf.py +++ b/modules/exllamav2_hf.py @@ -4,18 +4,20 @@ from typing import Any, Dict, Optional, Union import torch +from torch.nn import CrossEntropyLoss +from transformers import GenerationConfig, PretrainedConfig, PreTrainedModel +from transformers.modeling_outputs import CausalLMOutputWithPast + from exllamav2 import ( ExLlamaV2, ExLlamaV2Cache, ExLlamaV2Cache_8bit, ExLlamaV2Cache_Q4, + ExLlamaV2Cache_Q6, + ExLlamaV2Cache_Q8, ExLlamaV2Cache_TP, ExLlamaV2Config ) -from torch.nn import CrossEntropyLoss -from transformers import GenerationConfig, PretrainedConfig, PreTrainedModel -from transformers.modeling_outputs import CausalLMOutputWithPast - from modules import shared from modules.logging_colors import logger @@ -45,12 +47,22 @@ def __init__(self, config: ExLlamaV2Config): self.ex_model.load(split) # Determine the correct cache type - if shared.args.cache_8bit: + kv_cache_type = 'fp16' + if shared.args.cache_type: + kv_cache_type = shared.args.cache_type.lower() + + if kv_cache_type == 'fp16': + cache_type = ExLlamaV2Cache + elif kv_cache_type == 'fp8': cache_type = ExLlamaV2Cache_8bit - elif shared.args.cache_4bit: + elif kv_cache_type == 'q8': + cache_type = ExLlamaV2Cache_Q8 + elif kv_cache_type == 'q6': + cache_type = ExLlamaV2Cache_Q6 + elif kv_cache_type == 'q4': cache_type = ExLlamaV2Cache_Q4 else: - cache_type = ExLlamaV2Cache + raise ValueError(f"Invalid cache type for ExLlamaV2: {cache_type}. Valid options are: fp16, fp8, q8, q6, q4.") # Use TP if specified if shared.args.enable_tp: diff --git a/modules/llamacpp_hf.py b/modules/llamacpp_hf.py index 6611a7c1a8..7909825049 100644 --- a/modules/llamacpp_hf.py +++ b/modules/llamacpp_hf.py @@ -9,6 +9,7 @@ from modules import shared from modules.llama_cpp_python_hijack import llama_cpp_lib +from modules.llamacpp_model import get_llamacpp_cache_type_for_string from modules.logging_colors import logger @@ -196,12 +197,9 @@ def from_pretrained(cls, pretrained_model_name_or_path: Optional[Union[str, os.P 'flash_attn': shared.args.flash_attn } - if shared.args.cache_4bit: - params["type_k"] = 2 - params["type_v"] = 2 - elif shared.args.cache_8bit: - params["type_k"] = 8 - params["type_v"] = 8 + if shared.args.cache_type: + params["type_k"] = get_llamacpp_cache_type_for_string(shared.args.cache_type) + params["type_v"] = get_llamacpp_cache_type_for_string(shared.args.cache_type) Llama = llama_cpp_lib().Llama model = Llama(**params) diff --git a/modules/llamacpp_model.py b/modules/llamacpp_model.py index 96f7ed56b5..c8b3456e64 100644 --- a/modules/llamacpp_model.py +++ b/modules/llamacpp_model.py @@ -10,6 +10,35 @@ from modules.logging_colors import logger from modules.text_generation import get_max_prompt_length +llamacpp_quant_mapping = { + 'f32': 0, + 'fp16': 1, + 'q4_0': 2, + 'q4_1': 3, + 'q5_0': 6, + 'q5_1': 7, + 'q8_0': 8, + 'q8_1': 9, + 'q2_k': 10, + 'q3_k': 11, + 'q4_k': 12, + 'q5_k': 13, + 'q6_k': 14, + 'q8_k': 15, + 'iq4_nl': 20, + 'bf16': 30, +} + +llamacpp_valid_cache_types = {'fp16', 'q8_0', 'q4_0'} + + +def get_llamacpp_cache_type_for_string(quant_type: str): + quant_type = quant_type.lower() + if quant_type in llamacpp_valid_cache_types: + return llamacpp_quant_mapping[quant_type] + else: + raise ValueError(f"Invalid cache type for llama.cpp: {quant_type}. Valid options are: fp16, q8_0, q4_0.") + def ban_eos_logits_processor(eos_token, input_ids, logits): logits[eos_token] = -float('inf') @@ -75,12 +104,9 @@ def from_pretrained(self, path): 'flash_attn': shared.args.flash_attn } - if shared.args.cache_4bit: - params["type_k"] = 2 - params["type_v"] = 2 - elif shared.args.cache_8bit: - params["type_k"] = 8 - params["type_v"] = 8 + if shared.args.cache_type: + params["type_k"] = get_llamacpp_cache_type_for_string(shared.args.cache_type) + params["type_v"] = get_llamacpp_cache_type_for_string(shared.args.cache_type) result.model = Llama(**params) if cache_capacity > 0: diff --git a/modules/loaders.py b/modules/loaders.py index deee00a7f9..4cb7e349d6 100644 --- a/modules/loaders.py +++ b/modules/loaders.py @@ -31,8 +31,7 @@ 'llama.cpp': [ 'n_ctx', 'n_gpu_layers', - 'cache_8bit', - 'cache_4bit', + 'cache_type', 'tensor_split', 'n_batch', 'threads', @@ -54,8 +53,7 @@ 'llamacpp_HF': [ 'n_ctx', 'n_gpu_layers', - 'cache_8bit', - 'cache_4bit', + 'cache_type', 'tensor_split', 'n_batch', 'threads', @@ -87,8 +85,7 @@ 'no_xformers', 'no_sdpa', 'num_experts_per_token', - 'cache_8bit', - 'cache_4bit', + 'cache_type', 'autosplit', 'enable_tp', 'alpha_value', @@ -103,8 +100,7 @@ 'no_xformers', 'no_sdpa', 'num_experts_per_token', - 'cache_8bit', - 'cache_4bit', + 'cache_type', 'autosplit', 'enable_tp', 'alpha_value', diff --git a/modules/shared.py b/modules/shared.py index 599d64926b..41865e2279 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -142,8 +142,6 @@ group.add_argument('--no_flash_attn', action='store_true', help='Force flash-attention to not be used.') group.add_argument('--no_xformers', action='store_true', help='Force xformers to not be used.') group.add_argument('--no_sdpa', action='store_true', help='Force Torch SDPA to not be used.') -group.add_argument('--cache_8bit', action='store_true', help='Use 8-bit cache to save VRAM.') -group.add_argument('--cache_4bit', action='store_true', help='Use Q4 cache to save VRAM.') group.add_argument('--num_experts_per_token', type=int, default=2, help='Number of experts to use for generation. Applies to MoE models like Mixtral.') group.add_argument('--enable_tp', action='store_true', help='Enable Tensor Parallelism (TP) in ExLlamaV2.') @@ -166,6 +164,10 @@ group = parser.add_argument_group('TensorRT-LLM') group.add_argument('--cpp-runner', action='store_true', help='Use the ModelRunnerCpp runner, which is faster than the default ModelRunner but doesn\'t support streaming yet.') +# Cache +group = parser.add_argument_group('Cache') +group.add_argument('--cache_type', type=str, default=None, help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.') + # DeepSpeed group = parser.add_argument_group('DeepSpeed') group.add_argument('--deepspeed', action='store_true', help='Enable the use of DeepSpeed ZeRO-3 for inference via the Transformers integration.') @@ -213,6 +215,8 @@ group.add_argument('--checkpoint', type=str, help='DEPRECATED') group.add_argument('--monkey-patch', action='store_true', help='DEPRECATED') group.add_argument('--no_inject_fused_attention', action='store_true', help='DEPRECATED') +group.add_argument('--cache_4bit', action='store_true', help='DEPRECATED') +group.add_argument('--cache_8bit', action='store_true', help='DEPRECATED') group.add_argument('--chat-buttons', action='store_true', help='DEPRECATED') args = parser.parse_args() @@ -270,6 +274,59 @@ def fix_loader_name(name): return 'TensorRT-LLM' +def transform_legacy_kv_cache_options(opts): + # Handle both argparse.Namespace and dict here + def get(key): + return opts.get(key) if isinstance(opts, dict) else getattr(opts, key, None) + + def set(key, value): + if isinstance(opts, dict): + opts[key] = value + else: + setattr(opts, key, value) + + def del_key(key, fallback_set): + # only remove from user dict, can't delete from argparse.Namespace + if type(opts) is dict: + if key in opts: + del opts[key] + else: + setattr(opts, key, fallback_set) + + # Retrieve values + loader = get('loader') + cache_type = get('cache_type') + cache_8bit = get('cache_8bit') + cache_4bit = get('cache_4bit') + + # Determine cache type based on loader or legacy flags + if not cache_type: + if not loader: + # Legacy behavior: prefer 8-bit over 4-bit to minimize breakage + if cache_8bit: + set('cache_type', 'fp8') + elif cache_4bit: + set('cache_type', 'q4') + elif loader.lower() in ['exllamav2', 'exllamav2_hf']: + # ExLlamaV2 loader-specific cache type + if cache_8bit: + set('cache_type', 'fp8') + elif cache_4bit: + set('cache_type', 'q4') + elif loader.lower() in ['llama.cpp', 'llamacpp_hf']: + # Llama.cpp loader-specific cache type + if cache_4bit: + set('cache_type', 'q4_0') + elif cache_8bit: + set('cache_type', 'q8_0') + + # Clean up legacy keys + del_key('cache_4bit', False) + del_key('cache_8bit', False) + + return opts + + def add_extension(name, last=False): if args.extensions is None: args.extensions = [name] @@ -298,10 +355,14 @@ def load_user_config(): else: user_config = {} + for model_name in user_config: + user_config[model_name] = transform_legacy_kv_cache_options(user_config[model_name]) + return user_config args.loader = fix_loader_name(args.loader) +args = transform_legacy_kv_cache_options(args) # Activate the multimodal extension if args.multimodal_pipeline is not None: diff --git a/modules/ui.py b/modules/ui.py index 9ef958abc3..ff5bc16547 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -130,8 +130,7 @@ def list_model_elements(): 'no_xformers', 'no_sdpa', 'num_experts_per_token', - 'cache_8bit', - 'cache_4bit', + 'cache_type', 'autosplit', 'enable_tp', 'threads', diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py index 34d581776b..eb61cd820b 100644 --- a/modules/ui_model_menu.py +++ b/modules/ui_model_menu.py @@ -118,8 +118,7 @@ def create_ui(): shared.gradio['flash_attn'] = gr.Checkbox(label="flash_attn", value=shared.args.flash_attn, info='Use flash-attention.') shared.gradio['auto_devices'] = gr.Checkbox(label="auto-devices", value=shared.args.auto_devices) shared.gradio['tensorcores'] = gr.Checkbox(label="tensorcores", value=shared.args.tensorcores, info='NVIDIA only: use llama-cpp-python compiled with tensor cores support. This may increase performance on newer cards.') - shared.gradio['cache_8bit'] = gr.Checkbox(label="cache_8bit", value=shared.args.cache_8bit, info='Use 8-bit cache to save VRAM.') - shared.gradio['cache_4bit'] = gr.Checkbox(label="cache_4bit", value=shared.args.cache_4bit, info='Use Q4 cache to save VRAM.') + shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q6', 'q4'], value=shared.args.cache_type, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.') shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming_llm", value=shared.args.streaming_llm, info='(experimental) Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.') shared.gradio['attention_sink_size'] = gr.Number(label="attention_sink_size", value=shared.args.attention_sink_size, precision=0, info='StreamingLLM: number of sink tokens. Only used if the trimmed prompt doesn\'t share a prefix with the old prompt.') shared.gradio['cpu'] = gr.Checkbox(label="cpu", value=shared.args.cpu, info='llama.cpp: Use llama-cpp-python compiled without GPU acceleration. Transformers: use PyTorch in CPU mode.') From 3030c79e8c6ff24c1b51047faa648f7e36b64eef Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 17 Dec 2024 19:37:19 -0800 Subject: [PATCH 0098/1701] UI: show progress while loading a model --- modules/ui_model_menu.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py index eb61cd820b..024ba1893a 100644 --- a/modules/ui_model_menu.py +++ b/modules/ui_model_menu.py @@ -194,13 +194,13 @@ def create_event_handlers(): shared.gradio['model_menu'].change( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( handle_load_model_event_initial, gradio('model_menu', 'interface_state'), gradio(ui.list_interface_input_elements()) + gradio('interface_state'), show_progress=False).then( - load_model_wrapper, gradio('model_menu', 'loader', 'autoload_model'), gradio('model_status'), show_progress=False).success( + load_model_wrapper, gradio('model_menu', 'loader', 'autoload_model'), gradio('model_status'), show_progress=True).success( handle_load_model_event_final, gradio('truncation_length', 'loader', 'interface_state'), gradio('truncation_length', 'filter_by_loader'), show_progress=False) shared.gradio['load_model'].click( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( update_model_parameters, gradio('interface_state'), None).then( - partial(load_model_wrapper, autoload=True), gradio('model_menu', 'loader'), gradio('model_status'), show_progress=False).success( + partial(load_model_wrapper, autoload=True), gradio('model_menu', 'loader'), gradio('model_status'), show_progress=True).success( handle_load_model_event_final, gradio('truncation_length', 'loader', 'interface_state'), gradio('truncation_length', 'filter_by_loader'), show_progress=False) shared.gradio['unload_model'].click(handle_unload_model_click, None, gradio('model_status'), show_progress=False) From ddccc0d657bde02e294232fc3ec3126315de3bc2 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 17 Dec 2024 19:39:00 -0800 Subject: [PATCH 0099/1701] UI: minor change to log messages --- modules/ui_model_menu.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py index 024ba1893a..189bedfdf1 100644 --- a/modules/ui_model_menu.py +++ b/modules/ui_model_menu.py @@ -298,7 +298,7 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur downloader.check_model_files(model, branch, links, sha256, output_folder) progress(1.0) else: - yield (f"Downloading file{'s' if len(links) > 1 else ''} to `{output_folder}`") + yield (f"Downloading file{'s' if len(links) > 1 else ''} to `{output_folder}/`") downloader.download_model_files(model, branch, links, sha256, output_folder, progress_bar=progress, threads=4, is_llamacpp=is_llamacpp) yield (f"Model successfully saved to `{output_folder}/`.") @@ -318,7 +318,7 @@ def create_llamacpp_hf(gguf_name, unquantized_url, progress=gr.Progress()): links, sha256, is_lora, is_llamacpp = downloader.get_download_links_from_huggingface(model, branch, text_only=True) output_folder = Path(shared.args.model_dir) / (re.sub(r'(?i)\.gguf$', '', gguf_name) + "-HF") - yield (f"Downloading tokenizer to `{output_folder}`") + yield (f"Downloading tokenizer to `{output_folder}/`") downloader.download_model_files(model, branch, links, sha256, output_folder, progress_bar=progress, threads=4, is_llamacpp=False) # Move the GGUF From 60c93e0c66210ec7e3f99684858697abec8002a6 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 17 Dec 2024 19:44:20 -0800 Subject: [PATCH 0100/1701] UI: Set cache_type to fp16 by default --- modules/exllamav2.py | 4 +--- modules/exllamav2_hf.py | 4 +--- modules/llamacpp_hf.py | 2 +- modules/llamacpp_model.py | 2 +- modules/shared.py | 5 ++--- 5 files changed, 6 insertions(+), 11 deletions(-) diff --git a/modules/exllamav2.py b/modules/exllamav2.py index be5ef8d3b6..9b6da83c87 100644 --- a/modules/exllamav2.py +++ b/modules/exllamav2.py @@ -59,9 +59,7 @@ def from_pretrained(self, path_to_model): model.load(split) # Determine the correct cache type - kv_cache_type = 'fp16' - if shared.args.cache_type: - kv_cache_type = shared.args.cache_type.lower() + kv_cache_type = shared.args.cache_type.lower() if kv_cache_type == 'fp16': cache_type = ExLlamaV2Cache diff --git a/modules/exllamav2_hf.py b/modules/exllamav2_hf.py index f6b943c8b7..62d1e0547c 100644 --- a/modules/exllamav2_hf.py +++ b/modules/exllamav2_hf.py @@ -47,9 +47,7 @@ def __init__(self, config: ExLlamaV2Config): self.ex_model.load(split) # Determine the correct cache type - kv_cache_type = 'fp16' - if shared.args.cache_type: - kv_cache_type = shared.args.cache_type.lower() + kv_cache_type = shared.args.cache_type.lower() if kv_cache_type == 'fp16': cache_type = ExLlamaV2Cache diff --git a/modules/llamacpp_hf.py b/modules/llamacpp_hf.py index 7909825049..f9964fe8b0 100644 --- a/modules/llamacpp_hf.py +++ b/modules/llamacpp_hf.py @@ -197,7 +197,7 @@ def from_pretrained(cls, pretrained_model_name_or_path: Optional[Union[str, os.P 'flash_attn': shared.args.flash_attn } - if shared.args.cache_type: + if shared.args.cache_type != 'fp16': params["type_k"] = get_llamacpp_cache_type_for_string(shared.args.cache_type) params["type_v"] = get_llamacpp_cache_type_for_string(shared.args.cache_type) diff --git a/modules/llamacpp_model.py b/modules/llamacpp_model.py index c8b3456e64..6a76ee4e95 100644 --- a/modules/llamacpp_model.py +++ b/modules/llamacpp_model.py @@ -104,7 +104,7 @@ def from_pretrained(self, path): 'flash_attn': shared.args.flash_attn } - if shared.args.cache_type: + if shared.args.cache_type != 'fp16': params["type_k"] = get_llamacpp_cache_type_for_string(shared.args.cache_type) params["type_v"] = get_llamacpp_cache_type_for_string(shared.args.cache_type) diff --git a/modules/shared.py b/modules/shared.py index 41865e2279..cab612268a 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -166,7 +166,7 @@ # Cache group = parser.add_argument_group('Cache') -group.add_argument('--cache_type', type=str, default=None, help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.') +group.add_argument('--cache_type', type=str, default='fp16', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.') # DeepSpeed group = parser.add_argument_group('DeepSpeed') @@ -295,12 +295,11 @@ def del_key(key, fallback_set): # Retrieve values loader = get('loader') - cache_type = get('cache_type') cache_8bit = get('cache_8bit') cache_4bit = get('cache_4bit') # Determine cache type based on loader or legacy flags - if not cache_type: + if cache_8bit or cache_4bit: if not loader: # Legacy behavior: prefer 8-bit over 4-bit to minimize breakage if cache_8bit: From b051e2c1614d5641c9b4841ff5b5b7377f254a0d Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 17 Dec 2024 19:58:09 -0800 Subject: [PATCH 0101/1701] UI: improve a margin for readability --- css/main.css | 5 +++++ modules/ui.py | 1 + 2 files changed, 6 insertions(+) diff --git a/css/main.css b/css/main.css index 7d090e5742..4affff0d8c 100644 --- a/css/main.css +++ b/css/main.css @@ -36,6 +36,11 @@ div.svelte-iyf88w { border: 0; } +/* "info" messages without a title above */ +.block > .svelte-e8n7p6:not(:only-of-type) { + margin-bottom: 2px; +} + .py-6 { padding-top: 2.5rem } diff --git a/modules/ui.py b/modules/ui.py index ff5bc16547..4bfea9fade 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -83,6 +83,7 @@ checkbox_label_background_fill_hover_dark='var(--selected-item-color-dark)', table_even_background_fill_dark='var(--darker-gray)', table_odd_background_fill_dark='var(--selected-item-color-dark)', + code_background_fill_dark='var(--darker-gray)', # Shadows and Radius checkbox_label_shadow='none', From ac0f60eb1a09e59d98214239bb21fc63fe195c4c Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 17 Dec 2024 20:02:04 -0800 Subject: [PATCH 0102/1701] UI: make dropdown menus more readable --- css/main.css | 2 ++ 1 file changed, 2 insertions(+) diff --git a/css/main.css b/css/main.css index 4affff0d8c..0fba06dfe5 100644 --- a/css/main.css +++ b/css/main.css @@ -835,6 +835,8 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { ---------------------------------------------- */ .options { z-index: 100 !important; + border: 1px solid var(--input-border-color); + border-radius: 0px; } /* ---------------------------------------------- From e83235a0cc4d11eee8da2f5192842369b7894149 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 17 Dec 2024 20:11:51 -0800 Subject: [PATCH 0103/1701] UI: fix a font color --- css/html_instruct_style.css | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css index f1eb2531c1..f6ceb93245 100644 --- a/css/html_instruct_style.css +++ b/css/html_instruct_style.css @@ -17,7 +17,9 @@ line-height: 28px !important; } -.dark .chat .message-body p, .dark .chat .message-body li { +.dark .chat .message-body p, +.dark .chat .message-body li, +.dark .chat .message-body q { color: #d1d5db !important; } From b27f6f8915295da37ee6466d2ce05e63d8bf1273 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 17 Dec 2024 20:13:32 -0800 Subject: [PATCH 0104/1701] Lint --- css/main.css | 2 +- modules/ui_default.py | 12 ++++++------ modules/ui_notebook.py | 12 ++++++------ 3 files changed, 13 insertions(+), 13 deletions(-) diff --git a/css/main.css b/css/main.css index 0fba06dfe5..f33b5a5967 100644 --- a/css/main.css +++ b/css/main.css @@ -836,7 +836,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { .options { z-index: 100 !important; border: 1px solid var(--input-border-color); - border-radius: 0px; + border-radius: 0; } /* ---------------------------------------------- diff --git a/modules/ui_default.py b/modules/ui_default.py index af418ed7dc..ccae9a5ec3 100644 --- a/modules/ui_default.py +++ b/modules/ui_default.py @@ -63,26 +63,26 @@ def create_ui(): def create_event_handlers(): shared.gradio['Generate-default'].click( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( - lambda : [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-default', 'Generate-default')).then( + lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-default', 'Generate-default')).then( generate_reply_wrapper, gradio(inputs), gradio(outputs), show_progress=False).then( lambda state, left, right: state.update({'textbox-default': left, 'output_textbox': right}), gradio('interface_state', 'textbox-default', 'output_textbox'), None).then( - lambda : [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-default', 'Generate-default')).then( + lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-default', 'Generate-default')).then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['textbox-default'].submit( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( - lambda : [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-default', 'Generate-default')).then( + lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-default', 'Generate-default')).then( generate_reply_wrapper, gradio(inputs), gradio(outputs), show_progress=False).then( lambda state, left, right: state.update({'textbox-default': left, 'output_textbox': right}), gradio('interface_state', 'textbox-default', 'output_textbox'), None).then( - lambda : [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-default', 'Generate-default')).then( + lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-default', 'Generate-default')).then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['Continue-default'].click( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( - lambda : [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-default', 'Generate-default')).then( + lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-default', 'Generate-default')).then( generate_reply_wrapper, [shared.gradio['output_textbox']] + gradio(inputs)[1:], gradio(outputs), show_progress=False).then( lambda state, left, right: state.update({'textbox-default': left, 'output_textbox': right}), gradio('interface_state', 'textbox-default', 'output_textbox'), None).then( - lambda : [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-default', 'Generate-default')).then( + lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-default', 'Generate-default')).then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['Stop-default'].click(stop_everything_event, None, None, queue=False) diff --git a/modules/ui_notebook.py b/modules/ui_notebook.py index fccb1c9874..b234ac5753 100644 --- a/modules/ui_notebook.py +++ b/modules/ui_notebook.py @@ -66,28 +66,28 @@ def create_event_handlers(): shared.gradio['Generate-notebook'].click( lambda x: x, gradio('textbox-notebook'), gradio('last_input-notebook')).then( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( - lambda : [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-notebook', 'Generate-notebook')).then( + lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-notebook', 'Generate-notebook')).then( generate_reply_wrapper, gradio(inputs), gradio(outputs), show_progress=False).then( lambda state, text: state.update({'textbox-notebook': text}), gradio('interface_state', 'textbox-notebook'), None).then( - lambda : [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-notebook', 'Generate-notebook')).then( + lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-notebook', 'Generate-notebook')).then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['textbox-notebook'].submit( lambda x: x, gradio('textbox-notebook'), gradio('last_input-notebook')).then( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( - lambda : [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-notebook', 'Generate-notebook')).then( + lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-notebook', 'Generate-notebook')).then( generate_reply_wrapper, gradio(inputs), gradio(outputs), show_progress=False).then( lambda state, text: state.update({'textbox-notebook': text}), gradio('interface_state', 'textbox-notebook'), None).then( - lambda : [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-notebook', 'Generate-notebook')).then( + lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-notebook', 'Generate-notebook')).then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['Regenerate-notebook'].click( lambda x: x, gradio('last_input-notebook'), gradio('textbox-notebook'), show_progress=False).then( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( - lambda : [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-notebook', 'Generate-notebook')).then( + lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-notebook', 'Generate-notebook')).then( generate_reply_wrapper, gradio(inputs), gradio(outputs), show_progress=False).then( lambda state, text: state.update({'textbox-notebook': text}), gradio('interface_state', 'textbox-notebook'), None).then( - lambda : [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-notebook', 'Generate-notebook')).then( + lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-notebook', 'Generate-notebook')).then( None, None, None, js=f'() => {{{ui.audio_notification_js}}}') shared.gradio['Undo'].click( From c48e4622e863230eab574fe55baf807b27caddda Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 18 Dec 2024 06:28:14 -0800 Subject: [PATCH 0105/1701] UI: update a link --- modules/block_requests.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/block_requests.py b/modules/block_requests.py index 886930f0c0..6adc385a75 100644 --- a/modules/block_requests.py +++ b/modules/block_requests.py @@ -47,7 +47,7 @@ def my_open(*args, **kwargs): if len(args) > 1 and args[1] == 'rb': file_contents = file_contents.decode('utf-8') - file_contents = file_contents.replace('\t\t', '') + file_contents = file_contents.replace('\t\t', '') file_contents = file_contents.replace('cdnjs.cloudflare.com', '127.0.0.1') file_contents = file_contents.replace( '', From 0c069e5b3f7be5a4f6f0d11308458a1a417ee795 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 18 Dec 2024 12:16:26 -0800 Subject: [PATCH 0106/1701] UI: remove obsolete js event --- js/main.js | 12 ------------ 1 file changed, 12 deletions(-) diff --git a/js/main.js b/js/main.js index 997701f594..fd2fe758b0 100644 --- a/js/main.js +++ b/js/main.js @@ -463,18 +463,6 @@ new ResizeObserver(updateCssProperties).observe(document.querySelector("#chat-in // Handle changes in window size window.addEventListener("resize", updateCssProperties); -//------------------------------------------------ -// Keep track of the display width to position the past -// chats dropdown on desktop -//------------------------------------------------ -function updateDocumentWidth() { - var updatedWidth = window.innerWidth || document.documentElement.clientWidth || document.body.clientWidth; - document.documentElement.style.setProperty("--document-width", updatedWidth + "px"); -} - -updateDocumentWidth(); -window.addEventListener("resize", updateDocumentWidth); - //------------------------------------------------ // Focus on the rename text area when it becomes visible //------------------------------------------------ From 636a6621cc85b4ba089a22984c78e716eadf2320 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 18 Dec 2024 12:16:59 -0800 Subject: [PATCH 0107/1701] UI: fix sidebars closing when typing on mobile --- js/main.js | 3 --- 1 file changed, 3 deletions(-) diff --git a/js/main.js b/js/main.js index fd2fe758b0..06a44c79cb 100644 --- a/js/main.js +++ b/js/main.js @@ -740,9 +740,6 @@ function initializeSidebars() { // Run the initializer when the page loads initializeSidebars(); -// Add an event listener to handle screen resizing -window.addEventListener("resize", initializeSidebars); - // Add click event listeners to toggle buttons pastChatsToggle.addEventListener("click", () => { toggleSidebar(pastChatsRow, pastChatsToggle); From 0a15cff6a08db744cfb89e0fa63d8ee28c8d3525 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 18 Dec 2024 12:27:06 -0800 Subject: [PATCH 0108/1701] UI: close sidebars by clicking outside their areas on mobile --- js/main.js | 36 +++++++++++++++++++++++++++++++++++- 1 file changed, 35 insertions(+), 1 deletion(-) diff --git a/js/main.js b/js/main.js index 06a44c79cb..a8018175db 100644 --- a/js/main.js +++ b/js/main.js @@ -344,6 +344,21 @@ document.addEventListener("click", function (event) { if (event.target.classList.contains("pfp_character")) { toggleBigPicture(); } + + // Handle sidebar clicks on mobile + if (isMobile()) { + // Check if the click did NOT originate from any of the specified toggle buttons or elements + if ( + target.closest("#navigation-toggle") !== navigationToggle && + target.closest("#past-chats-toggle") !== pastChatsToggle && + target.closest("#chat-controls-toggle") !== chatControlsToggle && + target.closest(".header_bar") !== headerBar && + target.closest("#past-chats-row") !== pastChatsRow && + target.closest("#chat-controls") !== chatControlsRow + ) { + handleIndividualSidebarClose(event); + } + } }); //------------------------------------------------ @@ -656,7 +671,26 @@ headerBar.appendChild(navigationToggle); const pastChatsToggle = document.getElementById("past-chats-toggle"); const chatControlsToggle = document.getElementById("chat-controls-toggle"); -function toggleSidebar(sidebar, toggle) { +function handleIndividualSidebarClose(event) { + const target = event.target; + + // Close navigation bar if click is outside and it is open + if (!headerBar.contains(target) && !headerBar.classList.contains("sidebar-hidden")) { + toggleSidebar(headerBar, navigationToggle, true); + } + + // Close past chats row if click is outside and it is open + if (!pastChatsRow.contains(target) && !pastChatsRow.classList.contains("sidebar-hidden")) { + toggleSidebar(pastChatsRow, pastChatsToggle, true); + } + + // Close chat controls row if click is outside and it is open + if (!chatControlsRow.contains(target) && !chatControlsRow.classList.contains("sidebar-hidden")) { + toggleSidebar(chatControlsRow, chatControlsToggle, true); + } +} + +function toggleSidebar(sidebar, toggle, forceClose = false) { const isCurrentlyHidden = sidebar.classList.contains("sidebar-hidden"); const shouldClose = !isCurrentlyHidden; From 2bea4dfa966ace5b5c879fc5dcecb979caf5df77 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Alu=C3=ADsio=20Pires?= <88967089+Aluisio-Pires@users.noreply.github.com> Date: Wed, 18 Dec 2024 18:16:56 -0300 Subject: [PATCH 0109/1701] Fix an issue caused during the installation of tts (#6496) --- extensions/coqui_tts/requirements.txt | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/extensions/coqui_tts/requirements.txt b/extensions/coqui_tts/requirements.txt index 747f99a068..b0b691e8fe 100644 --- a/extensions/coqui_tts/requirements.txt +++ b/extensions/coqui_tts/requirements.txt @@ -1 +1 @@ -TTS==0.21.* \ No newline at end of file +coqui-tts==0.25.1 From d01dd2e1c87121a3d3b70a54e68df8bda15458cc Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 18 Dec 2024 13:35:09 -0800 Subject: [PATCH 0110/1701] UI: fix a margin --- css/main.css | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/css/main.css b/css/main.css index f33b5a5967..65c963167f 100644 --- a/css/main.css +++ b/css/main.css @@ -37,7 +37,7 @@ div.svelte-iyf88w { } /* "info" messages without a title above */ -.block > .svelte-e8n7p6:not(:only-of-type) { +.block > .svelte-e8n7p6:not(:only-of-type, #chat-mode *) { margin-bottom: 2px; } From 228caf0f3c0375b561516a762fec586ca45cc546 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 18 Dec 2024 15:33:05 -0800 Subject: [PATCH 0111/1701] UI: add a scrollbar to the right sidebar --- css/main.css | 1 + 1 file changed, 1 insertion(+) diff --git a/css/main.css b/css/main.css index 65c963167f..fef3d3f1bf 100644 --- a/css/main.css +++ b/css/main.css @@ -818,6 +818,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { #chat-controls { padding: 1rem; padding-bottom: 0; + overflow-y: scroll; } #chat-controls > :nth-child(1) { From 9fd12605aca9f530b6c5ee0d7b34d5d5824d96f3 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 18 Dec 2024 17:58:53 -0800 Subject: [PATCH 0112/1701] Update README.md --- README.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index fb22d09945..1429e6919d 100644 --- a/README.md +++ b/README.md @@ -4,9 +4,9 @@ A Gradio web UI for Large Language Models. Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.com/AUTOMATIC1111/stable-diffusion-webui) of text generation. -|![Image1](https://github.com/oobabooga/screenshots/raw/main/print_instruct.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/print_chat.png) | +|![Image1](https://github.com/oobabooga/screenshots/raw/main/AFTER-INSTRUCT.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/AFTER-CHAT.png) | |:---:|:---:| -|![Image1](https://github.com/oobabooga/screenshots/raw/main/print_default.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/print_parameters.png) | +|![Image1](https://github.com/oobabooga/screenshots/raw/main/AFTER-DEFAULT.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/AFTER-PARAMETERS.png) | ## Features From fee23df1a55fa8a8ab7b0b9d468857fdef9ad51e Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 18 Dec 2024 18:13:01 -0800 Subject: [PATCH 0113/1701] Update README.md --- README.md | 27 +++++++++++++++------------ 1 file changed, 15 insertions(+), 12 deletions(-) diff --git a/README.md b/README.md index 1429e6919d..0e16aa30a5 100644 --- a/README.md +++ b/README.md @@ -202,18 +202,19 @@ List of command-line flags ```txt usage: server.py [-h] [--multi-user] [--character CHARACTER] [--model MODEL] [--lora LORA [LORA ...]] [--model-dir MODEL_DIR] [--lora-dir LORA_DIR] [--model-menu] [--settings SETTINGS] - [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--chat-buttons] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--cpu] [--auto-devices] - [--gpu-memory GPU_MEMORY [GPU_MEMORY ...]] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] - [--force-safetensors] [--no_use_fast] [--use_flash_attention_2] [--use_eager_attention] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] - [--flash-attn] [--tensorcores] [--n_ctx N_CTX] [--threads THREADS] [--threads-batch THREADS_BATCH] [--no_mul_mat_q] [--n_batch N_BATCH] [--no-mmap] [--mlock] - [--n-gpu-layers N_GPU_LAYERS] [--tensor_split TENSOR_SPLIT] [--numa] [--logits_all] [--no_offload_kqv] [--cache-capacity CACHE_CAPACITY] [--row_split] [--streaming-llm] - [--attention-sink-size ATTENTION_SINK_SIZE] [--tokenizer-dir TOKENIZER_DIR] [--gpu-split GPU_SPLIT] [--autosplit] [--max_seq_len MAX_SEQ_LEN] [--cfg-cache] [--no_flash_attn] - [--no_xformers] [--no_sdpa] [--cache_8bit] [--cache_4bit] [--num_experts_per_token NUM_EXPERTS_PER_TOKEN] [--triton] [--no_inject_fused_mlp] [--no_use_cuda_fp16] [--desc_act] - [--disable_exllama] [--disable_exllamav2] [--wbits WBITS] [--groupsize GROUPSIZE] [--hqq-backend HQQ_BACKEND] [--cpp-runner] [--deepspeed] [--nvme-offload-dir NVME_OFFLOAD_DIR] + [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--cpu] [--auto-devices] [--gpu-memory GPU_MEMORY [GPU_MEMORY ...]] + [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] + [--use_flash_attention_2] [--use_eager_attention] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn] [--tensorcores] + [--n_ctx N_CTX] [--threads THREADS] [--threads-batch THREADS_BATCH] [--no_mul_mat_q] [--n_batch N_BATCH] [--no-mmap] [--mlock] [--n-gpu-layers N_GPU_LAYERS] + [--tensor_split TENSOR_SPLIT] [--numa] [--logits_all] [--no_offload_kqv] [--cache-capacity CACHE_CAPACITY] [--row_split] [--streaming-llm] [--attention-sink-size ATTENTION_SINK_SIZE] + [--tokenizer-dir TOKENIZER_DIR] [--gpu-split GPU_SPLIT] [--autosplit] [--max_seq_len MAX_SEQ_LEN] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa] + [--num_experts_per_token NUM_EXPERTS_PER_TOKEN] [--enable_tp] [--triton] [--no_inject_fused_mlp] [--no_use_cuda_fp16] [--desc_act] [--disable_exllama] [--disable_exllamav2] + [--wbits WBITS] [--groupsize GROUPSIZE] [--hqq-backend HQQ_BACKEND] [--cpp-runner] [--cache_type CACHE_TYPE] [--deepspeed] [--nvme-offload-dir NVME_OFFLOAD_DIR] [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] - [--subpath SUBPATH] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--nowebui] + [--subpath SUBPATH] [--old-colors] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--nowebui] [--multimodal-pipeline MULTIMODAL_PIPELINE] [--model_type MODEL_TYPE] [--pre_layer PRE_LAYER [PRE_LAYER ...]] [--checkpoint CHECKPOINT] [--monkey-patch] [--no_inject_fused_attention] + [--cache_4bit] [--cache_8bit] [--chat-buttons] Text generation web UI @@ -232,7 +233,6 @@ Basic settings: file will be loaded by default without the need to use the --settings flag. --extensions EXTENSIONS [EXTENSIONS ...] The list of extensions to load. If you want to load more than one extension, write the names separated by spaces. --verbose Print the prompts to the terminal. - --chat-buttons Show buttons on the chat tab instead of a hover menu. --idle-timeout IDLE_TIMEOUT Unload model after this many minutes of inactivity. It will be automatically reloaded when you try to use it again. Model loader: @@ -291,9 +291,8 @@ ExLlamaV2: --no_flash_attn Force flash-attention to not be used. --no_xformers Force xformers to not be used. --no_sdpa Force Torch SDPA to not be used. - --cache_8bit Use 8-bit cache to save VRAM. - --cache_4bit Use Q4 cache to save VRAM. --num_experts_per_token NUM_EXPERTS_PER_TOKEN Number of experts to use for generation. Applies to MoE models like Mixtral. + --enable_tp Enable Tensor Parallelism (TP) in ExLlamaV2. AutoGPTQ: --triton Use triton. @@ -311,6 +310,9 @@ HQQ: TensorRT-LLM: --cpp-runner Use the ModelRunnerCpp runner, which is faster than the default ModelRunner but doesn't support streaming yet. +Cache: + --cache_type CACHE_TYPE KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4. + DeepSpeed: --deepspeed Enable the use of DeepSpeed ZeRO-3 for inference via the Transformers integration. --nvme-offload-dir NVME_OFFLOAD_DIR DeepSpeed: Directory to use for ZeRO-3 NVME offloading. @@ -332,6 +334,7 @@ Gradio: --ssl-keyfile SSL_KEYFILE The path to the SSL certificate key file. --ssl-certfile SSL_CERTFILE The path to the SSL certificate cert file. --subpath SUBPATH Customize the subpath for gradio, use with reverse proxy + --old-colors Use the legacy Gradio colors, before the December/2024 update. API: --api Enable the API extension. From 836a868abcca409eb9c7f8539dad1b8942b204c2 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Thu, 19 Dec 2024 12:21:28 -0800 Subject: [PATCH 0114/1701] UI: improve the heading fonts --- css/html_instruct_style.css | 19 ++++-------- css/main.css | 58 +++++++++++++++++++++++++++++++++---- 2 files changed, 58 insertions(+), 19 deletions(-) diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css index f6ceb93245..5591f7d199 100644 --- a/css/html_instruct_style.css +++ b/css/html_instruct_style.css @@ -17,28 +17,19 @@ line-height: 28px !important; } -.dark .chat .message-body p, -.dark .chat .message-body li, -.dark .chat .message-body q { +.dark .chat .message-body :is(p, li, q, h1, h2, h3, h4, h5, h6) { color: #d1d5db !important; } -.chat .message-body p, -.chat .message-body ul, -.chat .message-body ol { - margin-top: 1.25em !important; - margin-bottom: 1.25em !important; +.chat .message-body :is(p, ul, ol) { + margin: 1.25em 0 !important; } -.chat .message-body p:first-child, -.chat .message-body ul:first-child, -.chat .message-body ol:first-child { +.chat .message-body :is(p, ul, ol):first-child { margin-top: 0 !important; } -.chat .message-body p:last-child, -.chat .message-body ul:last-child, -.chat .message-body ol:last-child { +.chat .message-body :is(p, ul, ol):last-child { margin-bottom: 0 !important; } diff --git a/css/main.css b/css/main.css index fef3d3f1bf..7b3910f95b 100644 --- a/css/main.css +++ b/css/main.css @@ -433,12 +433,60 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { padding-top: 20px; } -.message-body h1, -.message-body h2, -.message-body h3, +.message-body { + font-size: 16px; +} + +.message-body :is(h1, h2, h3, h4, h5, h6) { + color: black !important; +} + +.dark .message-body :is(h1, h2, h3, h4, h5, h6) { + color: white !important; +} + +.message-body h1 { + font-weight: 800; + font-size: 2.25em; + margin-top: 0; + margin-bottom: 0.8888889em; + line-height: 1.1111111; +} + +.message-body h2 { + font-weight: 700; + font-size: 1.5em; + margin-top: 2em; + margin-bottom: 1em; + line-height: 1.3333333; +} + +.message-body h3 { + font-weight: 600; + font-size: 1.25em; + margin-top: 0; + margin-bottom: 0.6em; + line-height: 1.6; +} + .message-body h4 { - color: var(--body-text-color); - margin: 20px 0 10px; + font-weight: 600; + font-size: 1em; + margin-top: 0; + margin-bottom: 0.5em; + line-height: 1.5; +} + +.message-body h5 { + font-weight: normal; + font-size: 1em; + margin: 0; +} + +.message-body h6 { + font-weight: normal; + font-size: 1em; + margin: 0; } .dark .message q { From 24a4c98d42a60283df1bc7e583604be222de4626 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Thu, 19 Dec 2024 12:23:03 -0800 Subject: [PATCH 0115/1701] UI: improve the style of links in messages --- css/main.css | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/css/main.css b/css/main.css index 7b3910f95b..637dbae5b9 100644 --- a/css/main.css +++ b/css/main.css @@ -501,6 +501,10 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { list-style-position: outside; } +.message-body a { + font-weight: 500; +} + .chat .message-body ul, .chat .message-body ol { padding-inline-start: 2em; } From c8ddb86c22e8c8a2b56d161e4acf311c79fb1c17 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Thu, 19 Dec 2024 12:23:21 -0800 Subject: [PATCH 0116/1701] UI: improve some light mode colors --- css/html_instruct_style.css | 2 +- css/main.css | 4 ---- 2 files changed, 1 insertion(+), 5 deletions(-) diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css index 5591f7d199..14fd919d00 100644 --- a/css/html_instruct_style.css +++ b/css/html_instruct_style.css @@ -51,13 +51,13 @@ } .chat .user-message { + background: #f4f4f4; padding: 1.5rem 1rem; border-radius: 0; border-bottom-right-radius: 0; } .chat .assistant-message { - background: #f4f4f4; padding: 1.5rem 1rem; border-radius: 0; border: 0; diff --git a/css/main.css b/css/main.css index 637dbae5b9..ec11891102 100644 --- a/css/main.css +++ b/css/main.css @@ -437,10 +437,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { font-size: 16px; } -.message-body :is(h1, h2, h3, h4, h5, h6) { - color: black !important; -} - .dark .message-body :is(h1, h2, h3, h4, h5, h6) { color: white !important; } From e2fb86e5df464b671984762e2445108b15c9145c Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Thu, 19 Dec 2024 13:42:17 -0800 Subject: [PATCH 0117/1701] UI: further improve the style of lists and headings --- css/html_instruct_style.css | 5 ---- css/main.css | 30 ++++++++++------------- modules/html_generator.py | 49 +++++++++++++++++++++++++++++++++++++ 3 files changed, 62 insertions(+), 22 deletions(-) diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css index 14fd919d00..7f74bf8801 100644 --- a/css/html_instruct_style.css +++ b/css/html_instruct_style.css @@ -33,11 +33,6 @@ margin-bottom: 0 !important; } -.chat .message-body li { - margin-top: 1.25em !important; - margin-bottom: 1.25em !important; -} - .user-message, .assistant-message { font-family: Inter, Helvetica, Arial, sans-serif; } diff --git a/css/main.css b/css/main.css index ec11891102..2ea7b96007 100644 --- a/css/main.css +++ b/css/main.css @@ -460,7 +460,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { .message-body h3 { font-weight: 600; font-size: 1.25em; - margin-top: 0; + margin-top: 1.6em; margin-bottom: 0.6em; line-height: 1.6; } @@ -468,7 +468,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { .message-body h4 { font-weight: 600; font-size: 1em; - margin-top: 0; + margin-top: 1.5em; margin-bottom: 0.5em; line-height: 1.5; } @@ -495,6 +495,14 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { .message-body li { list-style-position: outside; + margin-top: 0.5em !important; + margin-bottom: 0.5em !important; +} + +.message-body ul.long-list li, +.message-body ol.long-list li { + margin-top: 1.25em !important; + margin-bottom: 1.25em !important; } .message-body a { @@ -505,23 +513,10 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { padding-inline-start: 2em; } -.chat .message-body li:not(:last-child) { - margin-top: 0; - margin-bottom: 2px; -} - -.chat .message-body li:last-child { - margin-bottom: 0 !important; -} - .message-body li > p { display: inline !important; } -.message-body ul, .message-body ol { - font-size: 15px !important; -} - .message-body ul { list-style-type: disc !important; } @@ -834,8 +829,9 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* { padding-bottom: 80px !important; } -.chat ol, .chat ul { - margin-top: 6px !important; +.message-body ol, .message-body ul { + margin-top: 0 !important; + margin-bottom: 1.25em !important; } /* ---------------------------------------------- diff --git a/modules/html_generator.py b/modules/html_generator.py index 01b2086610..57eac1b11b 100644 --- a/modules/html_generator.py +++ b/modules/html_generator.py @@ -69,6 +69,52 @@ def replace_blockquote(m): return m.group().replace('\n', '\n> ').replace('\\begin{blockquote}', '').replace('\\end{blockquote}', '') +def add_long_list_class(html): + ''' + Adds a long-list class to
      or
        containing long
      1. items. + These will receive a smaller margin/padding in the CSS. + ''' + + # Helper function to check if a tag is within 
         or 
+    def is_within_block(start_idx, end_idx, block_matches):
+        return any(start < start_idx < end or start < end_idx < end for start, end in block_matches)
+
+    # Find all 
        ...
         and ... blocks
+    pre_blocks = [(m.start(), m.end()) for m in re.finditer(r'.*?
        ', html, re.DOTALL)] + code_blocks = [(m.start(), m.end()) for m in re.finditer(r'.*?', html, re.DOTALL)] + all_blocks = pre_blocks + code_blocks + + # Pattern to find
          ...
        and
          ...
        blocks and their contents + list_pattern = re.compile(r'(<[uo]l.*?>)(.*?)()', re.DOTALL) + li_pattern = re.compile(r'(.*?)
        2. ', re.DOTALL) + + def process_list(match): + start_idx, end_idx = match.span() + if is_within_block(start_idx, end_idx, all_blocks): + return match.group(0) # Leave the block unchanged if within 
         or 
+
+        opening_tag = match.group(1)
+        list_content = match.group(2)
+        closing_tag = match.group(3)
+
+        # Find all list items within this list
+        li_matches = li_pattern.finditer(list_content)
+        has_long_item = any(len(li_match.group(1).strip()) > 128 for li_match in li_matches)
+
+        if has_long_item:
+            # Add class="long-list" to the opening tag if it doesn't already have a class
+            if 'class=' not in opening_tag:
+                opening_tag = opening_tag[:-1] + ' class="long-list">'
+            else:
+                # If there's already a class, append long-list to it
+                opening_tag = re.sub(r'class="([^"]*)"', r'class="\1 long-list"', opening_tag)
+
+        return opening_tag + list_content + closing_tag
+
+    # Process HTML and replace list blocks
+    return list_pattern.sub(process_list, html)
+
+
 @functools.lru_cache(maxsize=None)
 def convert_to_markdown(string):
 
@@ -168,6 +214,9 @@ def convert_to_markdown(string):
     pattern = re.compile(r']*>(.*?)', re.DOTALL)
     html_output = pattern.sub(lambda x: html.unescape(x.group()), html_output)
 
+    # Add "long-list" class to 
           or 
             containing a long 
          1.  item
+    html_output = add_long_list_class(html_output)
+
     return html_output
 
 

From 2acec386fc6647782e30cbddbc298652a2224b3c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Dec 2024 14:08:56 -0800
Subject: [PATCH 0118/1701] UI: improve the streaming cursor

---
 modules/chat.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 92808fb7de..2638c79431 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -353,14 +353,14 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
 
         # Extract the reply
         if state['mode'] in ['chat', 'chat-instruct']:
-            visible_reply = re.sub("(||{{user}})", state['name1'], reply + '❚')
+            visible_reply = re.sub("(||{{user}})", state['name1'], reply + '▍')
         else:
-            visible_reply = reply + '❚'
+            visible_reply = reply + '▍'
 
         visible_reply = html.escape(visible_reply)
 
         if shared.stop_everything:
-            if output['visible'][-1][1].endswith('❚'):
+            if output['visible'][-1][1].endswith('▍'):
                 output['visible'][-1][1] = output['visible'][-1][1][:-1]
 
             output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
@@ -378,7 +378,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             if is_stream:
                 yield output
 
-    if output['visible'][-1][1].endswith('❚'):
+    if output['visible'][-1][1].endswith('▍'):
         output['visible'][-1][1] = output['visible'][-1][1][:-1]
 
     output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)

From 89888bef56018099b95e04317c9393f1952c5aa7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Dec 2024 14:38:36 -0800
Subject: [PATCH 0119/1701] UI: increase the threshold for a 
          2.  to be
 considered long

---
 modules/html_generator.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 57eac1b11b..0d831af25d 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -99,7 +99,7 @@ def process_list(match):
 
         # Find all list items within this list
         li_matches = li_pattern.finditer(list_content)
-        has_long_item = any(len(li_match.group(1).strip()) > 128 for li_match in li_matches)
+        has_long_item = any(len(li_match.group(1).strip()) > 160 for li_match in li_matches)
 
         if has_long_item:
             # Add class="long-list" to the opening tag if it doesn't already have a class

From ee3a533e5cc43612d887bdf275a786edd581eec0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Dec 2024 16:11:29 -0800
Subject: [PATCH 0120/1701] UI: improve the message width in instruct mode

---
 css/html_instruct_style.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index 7f74bf8801..dcc19c2910 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -68,7 +68,7 @@
 
 .chat .user-message .text,
 .chat .assistant-message .text {
-    max-width: 40.25rem;
+    max-width: 645px;
     margin-left: auto;
     margin-right: auto;
 }

From 0490ee620a1d479d304cd24b3d6d6c0d44032f7b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Dec 2024 16:51:34 -0800
Subject: [PATCH 0121/1701] UI: increase the threshold for a 
          3.  to be
 considered long (some more)

---
 modules/html_generator.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 0d831af25d..f07b6e75ec 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -99,7 +99,7 @@ def process_list(match):
 
         # Find all list items within this list
         li_matches = li_pattern.finditer(list_content)
-        has_long_item = any(len(li_match.group(1).strip()) > 160 for li_match in li_matches)
+        has_long_item = any(len(li_match.group(1).strip()) > 224 for li_match in li_matches)
 
         if has_long_item:
             # Add class="long-list" to the opening tag if it doesn't already have a class

From 39a5c9a49c35875d14f35cca54411c6864fe66ae Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Sun, 29 Dec 2024 11:16:17 -0300
Subject: [PATCH 0122/1701] UI organization (#6618)

---
 css/main.css             |  2 +-
 modules/loaders.py       |  1 -
 modules/ui_model_menu.py | 71 +++++++++++++----------------
 modules/ui_parameters.py | 96 ++++++++++++++++++++--------------------
 4 files changed, 80 insertions(+), 90 deletions(-)

diff --git a/css/main.css b/css/main.css
index 2ea7b96007..314b36e0eb 100644
--- a/css/main.css
+++ b/css/main.css
@@ -38,7 +38,7 @@ div.svelte-iyf88w {
 
 /* "info" messages without a title above */
 .block > .svelte-e8n7p6:not(:only-of-type, #chat-mode *) {
-    margin-bottom: 2px;
+    margin-bottom: 0;
 }
 
 .py-6 {
diff --git a/modules/loaders.py b/modules/loaders.py
index 4cb7e349d6..1cfdb31b25 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -26,7 +26,6 @@
         'compress_pos_emb',
         'disable_exllama',
         'disable_exllamav2',
-        'transformers_info',
     ],
     'llama.cpp': [
         'n_ctx',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 189bedfdf1..c4bb8f0133 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -80,59 +80,52 @@ def create_ui():
                 with gr.Blocks():
                     with gr.Row():
                         with gr.Column():
-                            with gr.Blocks():
-                                for i in range(len(total_mem)):
-                                    shared.gradio[f'gpu_memory_{i}'] = gr.Slider(label=f"gpu-memory in MiB for device :{i}", maximum=total_mem[i], value=default_gpu_mem[i])
+                            for i in range(len(total_mem)):
+                                shared.gradio[f'gpu_memory_{i}'] = gr.Slider(label=f"gpu-memory in MiB for device :{i}", maximum=total_mem[i], value=default_gpu_mem[i])
 
-                                shared.gradio['cpu_memory'] = gr.Slider(label="cpu-memory in MiB", maximum=total_cpu_mem, value=default_cpu_mem)
-
-                            with gr.Blocks():
-                                shared.gradio['transformers_info'] = gr.Markdown('load-in-4bit params:')
-                                shared.gradio['compute_dtype'] = gr.Dropdown(label="compute_dtype", choices=["bfloat16", "float16", "float32"], value=shared.args.compute_dtype)
-                                shared.gradio['quant_type'] = gr.Dropdown(label="quant_type", choices=["nf4", "fp4"], value=shared.args.quant_type)
-
-                            shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
-                            shared.gradio['n_gpu_layers'] = gr.Slider(label="n-gpu-layers", minimum=0, maximum=256, value=shared.args.n_gpu_layers, info='Must be set to more than 0 for your GPU to be used.')
-                            shared.gradio['n_ctx'] = gr.Number(label="n_ctx", precision=0, step=256, value=shared.args.n_ctx, info='Context length. Try lowering this if you run out of memory while loading the model.')
-                            shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
-                            shared.gradio['n_batch'] = gr.Slider(label="n_batch", minimum=1, maximum=2048, step=1, value=shared.args.n_batch)
+                            shared.gradio['cpu_memory'] = gr.Slider(label="cpu-memory in MiB", maximum=total_cpu_mem, value=default_cpu_mem)
+                            shared.gradio['n_gpu_layers'] = gr.Slider(label="n-gpu-layers", minimum=0, maximum=256, value=shared.args.n_gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['threads'] = gr.Slider(label="threads", minimum=0, step=1, maximum=256, value=shared.args.threads)
                             shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
+                            shared.gradio['n_batch'] = gr.Slider(label="n_batch", minimum=1, maximum=2048, step=1, value=shared.args.n_batch)
+                            shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
                             shared.gradio['wbits'] = gr.Dropdown(label="wbits", choices=["None", 1, 2, 3, 4, 8], value=shared.args.wbits if shared.args.wbits > 0 else "None")
                             shared.gradio['groupsize'] = gr.Dropdown(label="groupsize", choices=["None", 32, 64, 128, 1024], value=shared.args.groupsize if shared.args.groupsize > 0 else "None")
+                            shared.gradio['n_ctx'] = gr.Number(label="n_ctx", precision=0, step=256, value=shared.args.n_ctx, info='Context length. ⚠️ Lower this value if you can\'t load the model.')
+                            shared.gradio['max_seq_len'] = gr.Number(label='max_seq_len', precision=0, step=256, value=shared.args.max_seq_len, info='Context length. ⚠️ Lower this value if you can\'t load the model.')
+                            shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q6', 'q4'], value=shared.args.cache_type, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.')
+                            shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
-                            shared.gradio['max_seq_len'] = gr.Number(label='max_seq_len', precision=0, step=256, value=shared.args.max_seq_len, info='Context length. Try lowering this if you run out of memory while loading the model.')
-                            with gr.Blocks():
-                                shared.gradio['alpha_value'] = gr.Number(label='alpha_value', value=shared.args.alpha_value, precision=2, info='Positional embeddings alpha factor for NTK RoPE scaling. Recommended values (NTKv1): 1.75 for 1.5x context, 2.5 for 2x context. Use either this or compress_pos_emb, not both.')
-                                shared.gradio['rope_freq_base'] = gr.Number(label='rope_freq_base', value=shared.args.rope_freq_base, precision=0, info='Positional embeddings frequency base for NTK RoPE scaling. Related to alpha_value by rope_freq_base = 10000 * alpha_value ^ (64 / 63). 0 = from model.')
-                                shared.gradio['compress_pos_emb'] = gr.Number(label='compress_pos_emb', value=shared.args.compress_pos_emb, precision=2, info='Positional embeddings compression factor. Should be set to (context length) / (model\'s original context length). Equal to 1/rope_freq_scale.')
-
-                            shared.gradio['autogptq_info'] = gr.Markdown('ExLlamav2_HF is recommended over AutoGPTQ for models derived from Llama.')
+                            shared.gradio['alpha_value'] = gr.Number(label='alpha_value', value=shared.args.alpha_value, precision=2, info='Positional embeddings alpha factor for NTK RoPE scaling. Recommended values (NTKv1): 1.75 for 1.5x context, 2.5 for 2x context. Use either this or compress_pos_emb, not both.')
+                            shared.gradio['rope_freq_base'] = gr.Number(label='rope_freq_base', value=shared.args.rope_freq_base, precision=0, info='Positional embeddings frequency base for NTK RoPE scaling. Related to alpha_value by rope_freq_base = 10000 * alpha_value ^ (64 / 63). 0 = from model.')
+                            shared.gradio['compress_pos_emb'] = gr.Number(label='compress_pos_emb', value=shared.args.compress_pos_emb, precision=2, info='Positional embeddings compression factor. Should be set to (context length) / (model\'s original context length). Equal to 1/rope_freq_scale.')
+                            shared.gradio['compute_dtype'] = gr.Dropdown(label="compute_dtype", choices=["bfloat16", "float16", "float32"], value=shared.args.compute_dtype, info='Used by load-in-4bit.')
+                            shared.gradio['quant_type'] = gr.Dropdown(label="quant_type", choices=["nf4", "fp4"], value=shared.args.quant_type, info='Used by load-in-4bit.')
+                            shared.gradio['attention_sink_size'] = gr.Number(label="attention_sink_size", value=shared.args.attention_sink_size, precision=0, info='StreamingLLM: number of sink tokens. Only used if the trimmed prompt doesn\'t share a prefix with the old prompt.')
+                            shared.gradio['num_experts_per_token'] = gr.Number(label="Number of experts per token", value=shared.args.num_experts_per_token, info='Only applies to MoE models like Mixtral.')
 
                         with gr.Column():
+                            shared.gradio['tensorcores'] = gr.Checkbox(label="tensorcores", value=shared.args.tensorcores, info='NVIDIA only: use llama-cpp-python compiled with tensor cores support. This may increase performance on newer cards.')
                             shared.gradio['load_in_8bit'] = gr.Checkbox(label="load-in-8bit", value=shared.args.load_in_8bit)
                             shared.gradio['load_in_4bit'] = gr.Checkbox(label="load-in-4bit", value=shared.args.load_in_4bit)
-                            shared.gradio['use_double_quant'] = gr.Checkbox(label="use_double_quant", value=shared.args.use_double_quant)
-                            shared.gradio['use_flash_attention_2'] = gr.Checkbox(label="use_flash_attention_2", value=shared.args.use_flash_attention_2, info='Set use_flash_attention_2=True while loading the model.')
-                            shared.gradio['use_eager_attention'] = gr.Checkbox(label="use_eager_attention", value=shared.args.use_eager_attention, info='Set attn_implementation= eager while loading the model.')
                             shared.gradio['flash_attn'] = gr.Checkbox(label="flash_attn", value=shared.args.flash_attn, info='Use flash-attention.')
-                            shared.gradio['auto_devices'] = gr.Checkbox(label="auto-devices", value=shared.args.auto_devices)
-                            shared.gradio['tensorcores'] = gr.Checkbox(label="tensorcores", value=shared.args.tensorcores, info='NVIDIA only: use llama-cpp-python compiled with tensor cores support. This may increase performance on newer cards.')
-                            shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q6', 'q4'], value=shared.args.cache_type, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.')
+                            shared.gradio['use_flash_attention_2'] = gr.Checkbox(label="use_flash_attention_2", value=shared.args.use_flash_attention_2, info='Set use_flash_attention_2=True while loading the model.')
                             shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming_llm", value=shared.args.streaming_llm, info='(experimental) Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
-                            shared.gradio['attention_sink_size'] = gr.Number(label="attention_sink_size", value=shared.args.attention_sink_size, precision=0, info='StreamingLLM: number of sink tokens. Only used if the trimmed prompt doesn\'t share a prefix with the old prompt.')
+                            shared.gradio['auto_devices'] = gr.Checkbox(label="auto-devices", value=shared.args.auto_devices)
                             shared.gradio['cpu'] = gr.Checkbox(label="cpu", value=shared.args.cpu, info='llama.cpp: Use llama-cpp-python compiled without GPU acceleration. Transformers: use PyTorch in CPU mode.')
+                            shared.gradio['disk'] = gr.Checkbox(label="disk", value=shared.args.disk)
                             shared.gradio['row_split'] = gr.Checkbox(label="row_split", value=shared.args.row_split, info='Split the model by rows across GPUs. This may improve multi-gpu performance.')
                             shared.gradio['no_offload_kqv'] = gr.Checkbox(label="no_offload_kqv", value=shared.args.no_offload_kqv, info='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
                             shared.gradio['no_mul_mat_q'] = gr.Checkbox(label="no_mul_mat_q", value=shared.args.no_mul_mat_q, info='Disable the mulmat kernels.')
+                            shared.gradio['no_mmap'] = gr.Checkbox(label="no-mmap", value=shared.args.no_mmap)
+                            shared.gradio['mlock'] = gr.Checkbox(label="mlock", value=shared.args.mlock)
+                            shared.gradio['numa'] = gr.Checkbox(label="numa", value=shared.args.numa, info='NUMA support can help on some systems with non-uniform memory access.')
                             shared.gradio['triton'] = gr.Checkbox(label="triton", value=shared.args.triton)
                             shared.gradio['no_inject_fused_mlp'] = gr.Checkbox(label="no_inject_fused_mlp", value=shared.args.no_inject_fused_mlp, info='Affects Triton only. Disable fused MLP. Fused MLP improves performance but uses more VRAM. Disable if running low on VRAM.')
                             shared.gradio['no_use_cuda_fp16'] = gr.Checkbox(label="no_use_cuda_fp16", value=shared.args.no_use_cuda_fp16, info='This can make models faster on some systems.')
                             shared.gradio['desc_act'] = gr.Checkbox(label="desc_act", value=shared.args.desc_act, info='\'desc_act\', \'wbits\', and \'groupsize\' are used for old models without a quantize_config.json.')
-                            shared.gradio['no_mmap'] = gr.Checkbox(label="no-mmap", value=shared.args.no_mmap)
-                            shared.gradio['mlock'] = gr.Checkbox(label="mlock", value=shared.args.mlock)
-                            shared.gradio['numa'] = gr.Checkbox(label="numa", value=shared.args.numa, info='NUMA support can help on some systems with non-uniform memory access.')
-                            shared.gradio['disk'] = gr.Checkbox(label="disk", value=shared.args.disk)
+                            shared.gradio['use_double_quant'] = gr.Checkbox(label="use_double_quant", value=shared.args.use_double_quant, info='Used by load-in-4bit.')
+                            shared.gradio['use_eager_attention'] = gr.Checkbox(label="use_eager_attention", value=shared.args.use_eager_attention, info='Set attn_implementation= eager while loading the model.')
                             shared.gradio['bf16'] = gr.Checkbox(label="bf16", value=shared.args.bf16)
                             shared.gradio['autosplit'] = gr.Checkbox(label="autosplit", value=shared.args.autosplit, info='Automatically split the model tensors across the available GPUs.')
                             shared.gradio['enable_tp'] = gr.Checkbox(label="enable_tp", value=shared.args.enable_tp, info='Enable Tensor Parallelism (TP).')
@@ -141,16 +134,14 @@ def create_ui():
                             shared.gradio['no_sdpa'] = gr.Checkbox(label="no_sdpa", value=shared.args.no_sdpa)
                             shared.gradio['cfg_cache'] = gr.Checkbox(label="cfg-cache", value=shared.args.cfg_cache, info='Necessary to use CFG with this loader.')
                             shared.gradio['cpp_runner'] = gr.Checkbox(label="cpp-runner", value=shared.args.cpp_runner, info='Enable inference with ModelRunnerCpp, which is faster than the default ModelRunner.')
-                            shared.gradio['num_experts_per_token'] = gr.Number(label="Number of experts per token", value=shared.args.num_experts_per_token, info='Only applies to MoE models like Mixtral.')
-                            with gr.Blocks():
-                                shared.gradio['trust_remote_code'] = gr.Checkbox(label="trust-remote-code", value=shared.args.trust_remote_code, info='Set trust_remote_code=True while loading the tokenizer/model. To enable this option, start the web UI with the --trust-remote-code flag.', interactive=shared.args.trust_remote_code)
-                                shared.gradio['no_use_fast'] = gr.Checkbox(label="no_use_fast", value=shared.args.no_use_fast, info='Set use_fast=False while loading the tokenizer.')
-                                shared.gradio['logits_all'] = gr.Checkbox(label="logits_all", value=shared.args.logits_all, info='Needs to be set for perplexity evaluation to work with this loader. Otherwise, ignore it, as it makes prompt processing slower.')
-
+                            shared.gradio['logits_all'] = gr.Checkbox(label="logits_all", value=shared.args.logits_all, info='Needs to be set for perplexity evaluation to work with this loader. Otherwise, ignore it, as it makes prompt processing slower.')
                             shared.gradio['disable_exllama'] = gr.Checkbox(label="disable_exllama", value=shared.args.disable_exllama, info='Disable ExLlama kernel for GPTQ models.')
                             shared.gradio['disable_exllamav2'] = gr.Checkbox(label="disable_exllamav2", value=shared.args.disable_exllamav2, info='Disable ExLlamav2 kernel for GPTQ models.')
-                            shared.gradio['exllamav2_info'] = gr.Markdown("ExLlamav2_HF is recommended over ExLlamav2 for better integration with extensions and more consistent sampling behavior across loaders.")
+                            shared.gradio['trust_remote_code'] = gr.Checkbox(label="trust-remote-code", value=shared.args.trust_remote_code, info='Set trust_remote_code=True while loading the tokenizer/model. To enable this option, start the web UI with the --trust-remote-code flag.', interactive=shared.args.trust_remote_code)
+                            shared.gradio['no_use_fast'] = gr.Checkbox(label="no_use_fast", value=shared.args.no_use_fast, info='Set use_fast=False while loading the tokenizer.')
                             shared.gradio['llamacpp_HF_info'] = gr.Markdown("llamacpp_HF loads llama.cpp as a Transformers model. To use it, you need to place your GGUF in a subfolder of models/ with the necessary tokenizer files.\n\nYou can use the \"llamacpp_HF creator\" menu to do that automatically.")
+                            shared.gradio['exllamav2_info'] = gr.Markdown("ExLlamav2_HF is recommended over ExLlamav2 for better integration with extensions and more consistent sampling behavior across loaders.")
+                            shared.gradio['autogptq_info'] = gr.Markdown('ExLlamav2_HF is recommended over AutoGPTQ for models derived from Llama.')
                             shared.gradio['tensorrt_llm_info'] = gr.Markdown('* TensorRT-LLM has to be installed manually in a separate Python 3.10 environment at the moment. For a guide, consult the description of [this PR](https://github.com/oobabooga/text-generation-webui/pull/5715). \n\n* `max_seq_len` is only used when `cpp-runner` is checked.\n\n* `cpp_runner` does not support streaming at the moment.')
 
             with gr.Column():
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index a2665e0d21..727a152860 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -27,79 +27,79 @@ def create_ui(default_preset):
                 with gr.Column():
                     with gr.Row():
                         with gr.Column():
-                            shared.gradio['max_new_tokens'] = gr.Slider(minimum=shared.settings['max_new_tokens_min'], maximum=shared.settings['max_new_tokens_max'], step=1, label='max_new_tokens', value=shared.settings['max_new_tokens'])
+                            gr.Markdown('## Curve shape')
                             shared.gradio['temperature'] = gr.Slider(0.01, 5, value=generate_params['temperature'], step=0.01, label='temperature')
+                            shared.gradio['dynatemp_low'] = gr.Slider(0.01, 5, value=generate_params['dynatemp_low'], step=0.01, label='dynatemp_low', visible=generate_params['dynamic_temperature'])
+                            shared.gradio['dynatemp_high'] = gr.Slider(0.01, 5, value=generate_params['dynatemp_high'], step=0.01, label='dynatemp_high', visible=generate_params['dynamic_temperature'])
+                            shared.gradio['dynatemp_exponent'] = gr.Slider(0.01, 5, value=generate_params['dynatemp_exponent'], step=0.01, label='dynatemp_exponent', visible=generate_params['dynamic_temperature'])
+                            shared.gradio['smoothing_factor'] = gr.Slider(0.0, 10.0, value=generate_params['smoothing_factor'], step=0.01, label='smoothing_factor', info='Activates Quadratic Sampling.')
+                            shared.gradio['smoothing_curve'] = gr.Slider(1.0, 10.0, value=generate_params['smoothing_curve'], step=0.01, label='smoothing_curve', info='Adjusts the dropoff curve of Quadratic Sampling.')
+
+                            gr.Markdown('## Curve cutoff')
+                            shared.gradio['min_p'] = gr.Slider(0.0, 1.0, value=generate_params['min_p'], step=0.01, label='min_p')
                             shared.gradio['top_p'] = gr.Slider(0.0, 1.0, value=generate_params['top_p'], step=0.01, label='top_p')
                             shared.gradio['top_k'] = gr.Slider(0, 200, value=generate_params['top_k'], step=1, label='top_k')
                             shared.gradio['typical_p'] = gr.Slider(0.0, 1.0, value=generate_params['typical_p'], step=0.01, label='typical_p')
-                            shared.gradio['min_p'] = gr.Slider(0.0, 1.0, value=generate_params['min_p'], step=0.01, label='min_p')
+                            shared.gradio['xtc_threshold'] = gr.Slider(0, 0.5, value=generate_params['xtc_threshold'], step=0.01, label='xtc_threshold', info='If 2 or more tokens have probability above this threshold, consider removing all but the last one.')
+                            shared.gradio['xtc_probability'] = gr.Slider(0, 1, value=generate_params['xtc_probability'], step=0.01, label='xtc_probability', info='Probability that the removal will actually happen. 0 disables the sampler. 1 makes it always happen.')
+                            shared.gradio['epsilon_cutoff'] = gr.Slider(0, 9, value=generate_params['epsilon_cutoff'], step=0.01, label='epsilon_cutoff')
+                            shared.gradio['eta_cutoff'] = gr.Slider(0, 20, value=generate_params['eta_cutoff'], step=0.01, label='eta_cutoff')
+                            shared.gradio['tfs'] = gr.Slider(0.0, 1.0, value=generate_params['tfs'], step=0.01, label='tfs')
+                            shared.gradio['top_a'] = gr.Slider(0.0, 1.0, value=generate_params['top_a'], step=0.01, label='top_a')
+
+                            gr.Markdown('## Repetition suppression')
+                            shared.gradio['dry_multiplier'] = gr.Slider(0, 5, value=generate_params['dry_multiplier'], step=0.01, label='dry_multiplier', info='Set to greater than 0 to enable DRY. Recommended value: 0.8.')
+                            shared.gradio['dry_allowed_length'] = gr.Slider(1, 20, value=generate_params['dry_allowed_length'], step=1, label='dry_allowed_length', info='Longest sequence that can be repeated without being penalized.')
+                            shared.gradio['dry_base'] = gr.Slider(1, 4, value=generate_params['dry_base'], step=0.01, label='dry_base', info='Controls how fast the penalty grows with increasing sequence length.')
                             shared.gradio['repetition_penalty'] = gr.Slider(1.0, 1.5, value=generate_params['repetition_penalty'], step=0.01, label='repetition_penalty')
                             shared.gradio['frequency_penalty'] = gr.Slider(0, 2, value=generate_params['frequency_penalty'], step=0.05, label='frequency_penalty')
                             shared.gradio['presence_penalty'] = gr.Slider(0, 2, value=generate_params['presence_penalty'], step=0.05, label='presence_penalty')
+                            shared.gradio['encoder_repetition_penalty'] = gr.Slider(0.8, 1.5, value=generate_params['encoder_repetition_penalty'], step=0.01, label='encoder_repetition_penalty')
+                            shared.gradio['no_repeat_ngram_size'] = gr.Slider(0, 20, step=1, value=generate_params['no_repeat_ngram_size'], label='no_repeat_ngram_size')
                             shared.gradio['repetition_penalty_range'] = gr.Slider(0, 4096, step=64, value=generate_params['repetition_penalty_range'], label='repetition_penalty_range')
-                            shared.gradio['do_sample'] = gr.Checkbox(value=generate_params['do_sample'], label='do_sample')
-
-                            with gr.Blocks():
-                                shared.gradio['dry_multiplier'] = gr.Slider(0, 5, value=generate_params['dry_multiplier'], step=0.01, label='dry_multiplier', info='Set to greater than 0 to enable DRY. Recommended value: 0.8.')
-                                shared.gradio['dry_allowed_length'] = gr.Slider(1, 20, value=generate_params['dry_allowed_length'], step=1, label='dry_allowed_length', info='Longest sequence that can be repeated without being penalized.')
-                                shared.gradio['dry_base'] = gr.Slider(1, 4, value=generate_params['dry_base'], step=0.01, label='dry_base', info='Controls how fast the penalty grows with increasing sequence length.')
-                                shared.gradio['dry_sequence_breakers'] = gr.Textbox(value=generate_params['dry_sequence_breakers'], label='dry_sequence_breakers', info='Tokens across which sequence matching is not continued. Specified as a comma-separated list of quoted strings.')
-
-                            with gr.Blocks():
-                                shared.gradio['xtc_threshold'] = gr.Slider(0, 0.5, value=generate_params['xtc_threshold'], step=0.01, label='xtc_threshold', info='If 2 or more tokens have probability above this threshold, consider removing all but the last one.')
-                                shared.gradio['xtc_probability'] = gr.Slider(0, 1, value=generate_params['xtc_probability'], step=0.01, label='xtc_probability', info='Probability that the removal will actually happen. 0 disables the sampler. 1 makes it always happen.')
-
-                            gr.Markdown("[Learn more](https://github.com/oobabooga/text-generation-webui/wiki/03-%E2%80%90-Parameters-Tab)")
 
                         with gr.Column():
-                            with gr.Group():
-                                shared.gradio['auto_max_new_tokens'] = gr.Checkbox(value=shared.settings['auto_max_new_tokens'], label='auto_max_new_tokens', info='Expand max_new_tokens to the available context length.')
-                                shared.gradio['ban_eos_token'] = gr.Checkbox(value=shared.settings['ban_eos_token'], label='Ban the eos_token', info='Forces the model to never end the generation prematurely.')
-                                shared.gradio['add_bos_token'] = gr.Checkbox(value=shared.settings['add_bos_token'], label='Add the bos_token to the beginning of prompts', info='Disabling this can make the replies more creative.')
-                                shared.gradio['custom_stopping_strings'] = gr.Textbox(lines=2, value=shared.settings["custom_stopping_strings"] or None, label='Custom stopping strings', info='Written between "" and separated by commas.', placeholder='"\\n", "\\nYou:"')
-                                shared.gradio['custom_token_bans'] = gr.Textbox(value=shared.settings['custom_token_bans'] or None, label='Token bans', info='Token IDs to ban, separated by commas. The IDs can be found in the Default or Notebook tab.')
-
+                            gr.Markdown('## Alternative sampling methods')
                             shared.gradio['penalty_alpha'] = gr.Slider(0, 5, value=generate_params['penalty_alpha'], label='penalty_alpha', info='For Contrastive Search. do_sample must be unchecked.')
                             shared.gradio['guidance_scale'] = gr.Slider(-0.5, 2.5, step=0.05, value=generate_params['guidance_scale'], label='guidance_scale', info='For CFG. 1.5 is a good value.')
-                            shared.gradio['negative_prompt'] = gr.Textbox(value=shared.settings['negative_prompt'], label='Negative prompt', lines=3, elem_classes=['add_scrollbar'])
                             shared.gradio['mirostat_mode'] = gr.Slider(0, 2, step=1, value=generate_params['mirostat_mode'], label='mirostat_mode', info='mode=1 is for llama.cpp only.')
                             shared.gradio['mirostat_tau'] = gr.Slider(0, 10, step=0.01, value=generate_params['mirostat_tau'], label='mirostat_tau')
                             shared.gradio['mirostat_eta'] = gr.Slider(0, 1, step=0.01, value=generate_params['mirostat_eta'], label='mirostat_eta')
-                            shared.gradio['epsilon_cutoff'] = gr.Slider(0, 9, value=generate_params['epsilon_cutoff'], step=0.01, label='epsilon_cutoff')
-                            shared.gradio['eta_cutoff'] = gr.Slider(0, 20, value=generate_params['eta_cutoff'], step=0.01, label='eta_cutoff')
-                            shared.gradio['encoder_repetition_penalty'] = gr.Slider(0.8, 1.5, value=generate_params['encoder_repetition_penalty'], step=0.01, label='encoder_repetition_penalty')
-                            shared.gradio['no_repeat_ngram_size'] = gr.Slider(0, 20, step=1, value=generate_params['no_repeat_ngram_size'], label='no_repeat_ngram_size')
-
-                with gr.Column():
-                    with gr.Row() as shared.gradio['grammar_file_row']:
-                        shared.gradio['grammar_file'] = gr.Dropdown(value='None', choices=utils.get_available_grammars(), label='Load grammar from file (.gbnf)', elem_classes='slim-dropdown')
-                        ui.create_refresh_button(shared.gradio['grammar_file'], lambda: None, lambda: {'choices': utils.get_available_grammars()}, 'refresh-button', interactive=not mu)
-                        shared.gradio['save_grammar'] = gr.Button('💾', elem_classes='refresh-button', interactive=not mu)
-                        shared.gradio['delete_grammar'] = gr.Button('🗑️ ', elem_classes='refresh-button', interactive=not mu)
 
-                    shared.gradio['grammar_string'] = gr.Textbox(value='', label='Grammar', lines=16, elem_classes=['add_scrollbar', 'monospace'])
+                            gr.Markdown('## Other options')
+                            shared.gradio['max_new_tokens'] = gr.Slider(minimum=shared.settings['max_new_tokens_min'], maximum=shared.settings['max_new_tokens_max'], value=shared.settings['max_new_tokens'], step=1, label='max_new_tokens', info='⚠️ Setting this too high can cause prompt truncation.')
+                            shared.gradio['prompt_lookup_num_tokens'] = gr.Slider(value=shared.settings['prompt_lookup_num_tokens'], minimum=0, maximum=10, step=1, label='prompt_lookup_num_tokens', info='Activates Prompt Lookup Decoding.')
+                            shared.gradio['max_tokens_second'] = gr.Slider(value=shared.settings['max_tokens_second'], minimum=0, maximum=20, step=1, label='Maximum tokens/second', info='To make text readable in real time.')
+                            shared.gradio['max_updates_second'] = gr.Slider(value=shared.settings['max_updates_second'], minimum=0, maximum=24, step=1, label='Maximum UI updates/second', info='Set this if you experience lag in the UI during streaming.')
 
+                with gr.Column():
                     with gr.Row():
                         with gr.Column():
-                            shared.gradio['tfs'] = gr.Slider(0.0, 1.0, value=generate_params['tfs'], step=0.01, label='tfs')
-                            shared.gradio['top_a'] = gr.Slider(0.0, 1.0, value=generate_params['top_a'], step=0.01, label='top_a')
-                            shared.gradio['smoothing_factor'] = gr.Slider(0.0, 10.0, value=generate_params['smoothing_factor'], step=0.01, label='smoothing_factor', info='Activates Quadratic Sampling.')
-                            shared.gradio['smoothing_curve'] = gr.Slider(1.0, 10.0, value=generate_params['smoothing_curve'], step=0.01, label='smoothing_curve', info='Adjusts the dropoff curve of Quadratic Sampling.')
+                            shared.gradio['do_sample'] = gr.Checkbox(value=generate_params['do_sample'], label='do_sample')
                             shared.gradio['dynamic_temperature'] = gr.Checkbox(value=generate_params['dynamic_temperature'], label='dynamic_temperature')
-                            shared.gradio['dynatemp_low'] = gr.Slider(0.01, 5, value=generate_params['dynatemp_low'], step=0.01, label='dynatemp_low', visible=generate_params['dynamic_temperature'])
-                            shared.gradio['dynatemp_high'] = gr.Slider(0.01, 5, value=generate_params['dynatemp_high'], step=0.01, label='dynatemp_high', visible=generate_params['dynamic_temperature'])
-                            shared.gradio['dynatemp_exponent'] = gr.Slider(0.01, 5, value=generate_params['dynatemp_exponent'], step=0.01, label='dynatemp_exponent', visible=generate_params['dynamic_temperature'])
                             shared.gradio['temperature_last'] = gr.Checkbox(value=generate_params['temperature_last'], label='temperature_last', info='Moves temperature/dynamic temperature/quadratic sampling to the end of the sampler stack, ignoring their positions in "Sampler priority".')
-                            shared.gradio['sampler_priority'] = gr.Textbox(value=generate_params['sampler_priority'], lines=12, label='Sampler priority', info='Parameter names separated by new lines or commas.')
+                            shared.gradio['auto_max_new_tokens'] = gr.Checkbox(value=shared.settings['auto_max_new_tokens'], label='auto_max_new_tokens', info='Expand max_new_tokens to the available context length.')
+                            shared.gradio['ban_eos_token'] = gr.Checkbox(value=shared.settings['ban_eos_token'], label='Ban the eos_token', info='Forces the model to never end the generation prematurely.')
+                            shared.gradio['add_bos_token'] = gr.Checkbox(value=shared.settings['add_bos_token'], label='Add the bos_token to the beginning of prompts', info='Disabling this can make the replies more creative.')
+                            shared.gradio['skip_special_tokens'] = gr.Checkbox(value=shared.settings['skip_special_tokens'], label='Skip special tokens', info='Some specific models need this unset.')
+                            shared.gradio['stream'] = gr.Checkbox(value=shared.settings['stream'], label='Activate text streaming')
 
                         with gr.Column():
                             shared.gradio['truncation_length'] = gr.Number(precision=0, step=256, value=get_truncation_length(), label='Truncate the prompt up to this length', info='The leftmost tokens are removed if the prompt exceeds this length. Most models require this to be at most 2048.')
-                            shared.gradio['prompt_lookup_num_tokens'] = gr.Slider(value=shared.settings['prompt_lookup_num_tokens'], minimum=0, maximum=10, step=1, label='prompt_lookup_num_tokens', info='Activates Prompt Lookup Decoding.')
-                            shared.gradio['max_tokens_second'] = gr.Slider(value=shared.settings['max_tokens_second'], minimum=0, maximum=20, step=1, label='Maximum tokens/second', info='To make text readable in real time.')
-                            shared.gradio['max_updates_second'] = gr.Slider(value=shared.settings['max_updates_second'], minimum=0, maximum=24, step=1, label='Maximum UI updates/second', info='Set this if you experience lag in the UI during streaming.')
                             shared.gradio['seed'] = gr.Number(value=shared.settings['seed'], label='Seed (-1 for random)')
-                            shared.gradio['skip_special_tokens'] = gr.Checkbox(value=shared.settings['skip_special_tokens'], label='Skip special tokens', info='Some specific models need this unset.')
-                            shared.gradio['stream'] = gr.Checkbox(value=shared.settings['stream'], label='Activate text streaming')
+
+                            shared.gradio['sampler_priority'] = gr.Textbox(value=generate_params['sampler_priority'], lines=12, label='Sampler priority', info='Parameter names separated by new lines or commas.', elem_classes=['add_scrollbar'])
+                            shared.gradio['custom_stopping_strings'] = gr.Textbox(lines=2, value=shared.settings["custom_stopping_strings"] or None, label='Custom stopping strings', info='Written between "" and separated by commas.', placeholder='"\\n", "\\nYou:"')
+                            shared.gradio['custom_token_bans'] = gr.Textbox(value=shared.settings['custom_token_bans'] or None, label='Token bans', info='Token IDs to ban, separated by commas. The IDs can be found in the Default or Notebook tab.')
+                            shared.gradio['negative_prompt'] = gr.Textbox(value=shared.settings['negative_prompt'], label='Negative prompt', info='For CFG. Only used when guidance_scale is different than 1.', lines=3, elem_classes=['add_scrollbar'])
+                            shared.gradio['dry_sequence_breakers'] = gr.Textbox(value=generate_params['dry_sequence_breakers'], label='dry_sequence_breakers', info='Tokens across which sequence matching is not continued. Specified as a comma-separated list of quoted strings.')
+                            with gr.Row() as shared.gradio['grammar_file_row']:
+                                shared.gradio['grammar_file'] = gr.Dropdown(value='None', choices=utils.get_available_grammars(), label='Load grammar from file (.gbnf)', elem_classes='slim-dropdown')
+                                ui.create_refresh_button(shared.gradio['grammar_file'], lambda: None, lambda: {'choices': utils.get_available_grammars()}, 'refresh-button', interactive=not mu)
+                                shared.gradio['save_grammar'] = gr.Button('💾', elem_classes='refresh-button', interactive=not mu)
+                                shared.gradio['delete_grammar'] = gr.Button('🗑️ ', elem_classes='refresh-button', interactive=not mu)
+
+                            shared.gradio['grammar_string'] = gr.Textbox(value='', label='Grammar', lines=16, elem_classes=['add_scrollbar', 'monospace'])
 
         ui_chat.create_chat_settings_ui()
 

From 4ce9d13dbec04230b4a56b905e8e96ae9f4f4d8e Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Sun, 29 Dec 2024 12:25:26 -0300
Subject: [PATCH 0123/1701] Preset cleanup (#6619)

---
 presets/Big O.yaml                                       | 6 ------
 presets/Creative.yaml                                    | 2 ++
 presets/{Debug-deterministic.yaml => Deterministic.yaml} | 0
 presets/Divine Intellect.yaml                            | 4 ----
 presets/Instruct.yaml                                    | 1 +
 presets/LLaMA-Precise.yaml                               | 4 ----
 presets/Midnight Enigma.yaml                             | 4 ----
 presets/Shortwave.yaml                                   | 4 ----
 presets/Yara.yaml                                        | 4 ----
 presets/simple-1.yaml                                    | 4 ----
 10 files changed, 3 insertions(+), 30 deletions(-)
 delete mode 100644 presets/Big O.yaml
 create mode 100644 presets/Creative.yaml
 rename presets/{Debug-deterministic.yaml => Deterministic.yaml} (100%)
 delete mode 100644 presets/Divine Intellect.yaml
 create mode 100644 presets/Instruct.yaml
 delete mode 100644 presets/LLaMA-Precise.yaml
 delete mode 100644 presets/Midnight Enigma.yaml
 delete mode 100644 presets/Shortwave.yaml
 delete mode 100644 presets/Yara.yaml
 delete mode 100644 presets/simple-1.yaml

diff --git a/presets/Big O.yaml b/presets/Big O.yaml
deleted file mode 100644
index 2ab1826876..0000000000
--- a/presets/Big O.yaml	
+++ /dev/null
@@ -1,6 +0,0 @@
-temperature: 0.87
-top_p: 0.99
-typical_p: 0.68
-tfs: 0.68
-repetition_penalty: 1.01
-top_k: 85
diff --git a/presets/Creative.yaml b/presets/Creative.yaml
new file mode 100644
index 0000000000..3ed04190f8
--- /dev/null
+++ b/presets/Creative.yaml
@@ -0,0 +1,2 @@
+min_p: 0.02
+xtc_probability: 0.5
diff --git a/presets/Debug-deterministic.yaml b/presets/Deterministic.yaml
similarity index 100%
rename from presets/Debug-deterministic.yaml
rename to presets/Deterministic.yaml
diff --git a/presets/Divine Intellect.yaml b/presets/Divine Intellect.yaml
deleted file mode 100644
index ac750e40dc..0000000000
--- a/presets/Divine Intellect.yaml	
+++ /dev/null
@@ -1,4 +0,0 @@
-temperature: 1.31
-top_p: 0.14
-repetition_penalty: 1.17
-top_k: 49
diff --git a/presets/Instruct.yaml b/presets/Instruct.yaml
new file mode 100644
index 0000000000..142fcd82e6
--- /dev/null
+++ b/presets/Instruct.yaml
@@ -0,0 +1 @@
+min_p: 0.2
diff --git a/presets/LLaMA-Precise.yaml b/presets/LLaMA-Precise.yaml
deleted file mode 100644
index c5f9cae256..0000000000
--- a/presets/LLaMA-Precise.yaml
+++ /dev/null
@@ -1,4 +0,0 @@
-temperature: 0.7
-top_p: 0.1
-repetition_penalty: 1.18
-top_k: 40
diff --git a/presets/Midnight Enigma.yaml b/presets/Midnight Enigma.yaml
deleted file mode 100644
index 0bd1763c6d..0000000000
--- a/presets/Midnight Enigma.yaml	
+++ /dev/null
@@ -1,4 +0,0 @@
-temperature: 0.98
-top_p: 0.37
-repetition_penalty: 1.18
-top_k: 100
diff --git a/presets/Shortwave.yaml b/presets/Shortwave.yaml
deleted file mode 100644
index a2528abdb4..0000000000
--- a/presets/Shortwave.yaml
+++ /dev/null
@@ -1,4 +0,0 @@
-temperature: 1.53
-top_p: 0.64
-repetition_penalty: 1.07
-top_k: 33
diff --git a/presets/Yara.yaml b/presets/Yara.yaml
deleted file mode 100644
index 87bb019ec6..0000000000
--- a/presets/Yara.yaml
+++ /dev/null
@@ -1,4 +0,0 @@
-temperature: 0.82
-top_p: 0.21
-repetition_penalty: 1.19
-top_k: 72
diff --git a/presets/simple-1.yaml b/presets/simple-1.yaml
deleted file mode 100644
index 30a106590b..0000000000
--- a/presets/simple-1.yaml
+++ /dev/null
@@ -1,4 +0,0 @@
-temperature: 0.7
-top_p: 0.9
-repetition_penalty: 1.15
-top_k: 20

From 292cd489e9916104b3f965c8da3ccb9f36253b90 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 30 Dec 2024 04:31:10 -0800
Subject: [PATCH 0124/1701] Bump ExLlamaV2 to 0.2.7

---
 requirements.txt               | 10 +++++-----
 requirements_amd.txt           |  6 +++---
 requirements_amd_noavx2.txt    |  6 +++---
 requirements_apple_intel.txt   |  2 +-
 requirements_apple_silicon.txt |  2 +-
 requirements_noavx2.txt        | 10 +++++-----
 6 files changed, 18 insertions(+), 18 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 24c92391ca..a04cb04f8c 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -50,11 +50,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # CUDA wheels
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements_amd.txt b/requirements_amd.txt
index b7093d5099..e5bec4bf45 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -39,6 +39,6 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.5+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.5+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 88682aea04..d408b4cf18 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -37,6 +37,6 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # AMD wheels
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 6588278d03..da0423a21e 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -33,4 +33,4 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 1fc9795bb3..dd1933bd18 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -35,4 +35,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 9ad138d8d4..db21c0fc98 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -50,11 +50,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # CUDA wheels
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.6/exllamav2-0.2.6-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

From cca4ac56fa2e093af99552a4a9ce6aece21bde41 Mon Sep 17 00:00:00 2001
From: mamei16 
Date: Mon, 30 Dec 2024 13:34:19 +0100
Subject: [PATCH 0125/1701] Fix interface loading with dark theme even when
 'dark_theme' is set to false (#6614)

---
 server.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/server.py b/server.py
index d6069d5e34..31e1c4c60a 100644
--- a/server.py
+++ b/server.py
@@ -154,6 +154,9 @@ def create_interface():
                 if ({str(shared.settings['dark_theme']).lower()}) {{
                     document.getElementsByTagName('body')[0].classList.add('dark');
                 }}
+                else {{
+                    document.getElementsByTagName('body')[0].classList.remove('dark');
+                }}
                 {js}
                 {ui.show_controls_js}
                 toggle_controls(x);

From d24b83132b207b5b95f7ba3754e6b3f9b99e1f84 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Mon, 30 Dec 2024 09:35:20 -0300
Subject: [PATCH 0126/1701] Bump jinja2 from 3.1.4 to 3.1.5 (#6601)

---
 requirements.txt                 | 2 +-
 requirements_amd.txt             | 2 +-
 requirements_amd_noavx2.txt      | 2 +-
 requirements_apple_intel.txt     | 2 +-
 requirements_apple_silicon.txt   | 2 +-
 requirements_cpu_only.txt        | 2 +-
 requirements_cpu_only_noavx2.txt | 2 +-
 requirements_noavx2.txt          | 2 +-
 requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index a04cb04f8c..8b36f54f1d 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -5,7 +5,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.4
+jinja2==3.1.5
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_amd.txt b/requirements_amd.txt
index e5bec4bf45..1b85b20fdf 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -4,7 +4,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.4
+jinja2==3.1.5
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index d408b4cf18..1bafc1419f 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -4,7 +4,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.4
+jinja2==3.1.5
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index da0423a21e..345368b79d 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -4,7 +4,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.4
+jinja2==3.1.5
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index dd1933bd18..a3c3055bee 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -4,7 +4,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.4
+jinja2==3.1.5
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index 53fedd7ec5..a4b7882cc4 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -4,7 +4,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.4
+jinja2==3.1.5
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 9f52b17283..878aea0699 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -4,7 +4,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.4
+jinja2==3.1.5
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index db21c0fc98..1a3456113a 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -5,7 +5,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.4
+jinja2==3.1.5
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index e2daebd9ec..95fe3add4f 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -4,7 +4,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.4
+jinja2==3.1.5
 markdown
 numba==0.59.*
 numpy==1.26.*

From e953af85cd45b3c24d4fe953a9ed68181d175ed8 Mon Sep 17 00:00:00 2001
From: mamei16 
Date: Tue, 31 Dec 2024 05:04:02 +0100
Subject: [PATCH 0127/1701] Fix newlines in the markdown renderer (#6599)

---------

Co-authored-by: oobabooga 
---
 modules/html_generator.py | 27 +++++++++++----------------
 1 file changed, 11 insertions(+), 16 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index f07b6e75ec..08147bb72a 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -150,7 +150,6 @@ def convert_to_markdown(string):
     result = ''
     is_code = False
     is_latex = False
-    previous_line_empty = True
 
     for line in string.split('\n'):
         stripped_line = line.strip()
@@ -168,20 +167,16 @@ def convert_to_markdown(string):
         elif stripped_line.endswith('\\\\]'):
             is_latex = False
 
-        # Preserve indentation for lists and code blocks
-        if stripped_line.startswith('-') or stripped_line.startswith('*') or stripped_line.startswith('+') or stripped_line.startswith('>') or re.match(r'\d+\.', stripped_line):
-            result += line + '\n'
-            previous_line_empty = False
-        elif is_code or is_latex or line.startswith('|'):
-            result += line + '\n'
-            previous_line_empty = False
-        else:
-            if previous_line_empty:
-                result += line.strip() + '\n'
-            else:
-                result += line.strip() + '\n\n'
+        result += line
 
-            previous_line_empty = stripped_line == ''
+        # Don't add an extra \n for code, LaTeX, or tables
+        if is_code or is_latex or line.startswith('|'):
+            result += '\n'
+        # Also don't add an extra \n for lists
+        elif stripped_line.startswith('-') or stripped_line.startswith('*') or stripped_line.startswith('+') or stripped_line.startswith('>') or re.match(r'\d+\.', stripped_line):
+            result += '\n'
+        else:
+            result += '\n\n'
 
     result = result.strip()
     if is_code:
@@ -200,7 +195,7 @@ def convert_to_markdown(string):
         result = re.sub(list_item_pattern, r'\g<1> ' + delete_str, result)
 
         # Convert to HTML using markdown
-        html_output = markdown.markdown(result, extensions=['fenced_code', 'tables'], tab_length=2)
+        html_output = markdown.markdown(result, extensions=['fenced_code', 'tables'])
 
         # Remove the delete string from the HTML output
         pos = html_output.rfind(delete_str)
@@ -208,7 +203,7 @@ def convert_to_markdown(string):
             html_output = html_output[:pos] + html_output[pos + len(delete_str):]
     else:
         # Convert to HTML using markdown
-        html_output = markdown.markdown(result, extensions=['fenced_code', 'tables'], tab_length=2)
+        html_output = markdown.markdown(result, extensions=['fenced_code', 'tables'])
 
     # Unescape code blocks
     pattern = re.compile(r']*>(.*?)', re.DOTALL)

From 64853f85095f8671a28298b9563896e619d531da Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 31 Dec 2024 14:43:22 -0800
Subject: [PATCH 0128/1701] Reapply a necessary change that I removed from
 #6599 (thanks @mamei16!)

---
 modules/html_generator.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 08147bb72a..40a56731e7 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -176,7 +176,7 @@ def convert_to_markdown(string):
         elif stripped_line.startswith('-') or stripped_line.startswith('*') or stripped_line.startswith('+') or stripped_line.startswith('>') or re.match(r'\d+\.', stripped_line):
             result += '\n'
         else:
-            result += '\n\n'
+            result += '  \n'
 
     result = result.strip()
     if is_code:

From 7b887247115b40825854c31f094703af399bd6d2 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Wed, 1 Jan 2025 18:33:38 -0300
Subject: [PATCH 0129/1701] Make responses start faster by removing unnecessary
 cleanup calls (#6625)

---
 modules/callbacks.py       | 15 +--------------
 modules/models.py          |  1 +
 modules/text_generation.py |  5 +++--
 3 files changed, 5 insertions(+), 16 deletions(-)

diff --git a/modules/callbacks.py b/modules/callbacks.py
index 2b039ef13f..2c04cc53ac 100644
--- a/modules/callbacks.py
+++ b/modules/callbacks.py
@@ -65,7 +65,6 @@ def gentask():
                 traceback.print_exc()
                 pass
 
-            clear_torch_cache()
             self.q.put(self.sentinel)
             if self.c_callback:
                 self.c_callback(ret)
@@ -84,22 +83,10 @@ def __next__(self):
             return obj
 
     def __del__(self):
-        clear_torch_cache()
+        pass
 
     def __enter__(self):
         return self
 
     def __exit__(self, exc_type, exc_val, exc_tb):
         self.stop_now = True
-        clear_torch_cache()
-
-
-def clear_torch_cache():
-    gc.collect()
-    if not shared.args.cpu:
-        if is_torch_xpu_available():
-            torch.xpu.empty_cache()
-        elif is_torch_npu_available():
-            torch.npu.empty_cache()
-        else:
-            torch.cuda.empty_cache()
diff --git a/modules/models.py b/modules/models.py
index 3e8ae3f227..7a52c07c70 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -90,6 +90,7 @@ def load_model(model_name, loader=None):
                 raise ValueError
 
     shared.args.loader = loader
+    clear_torch_cache()
     output = load_func_map[loader](model_name)
     if type(output) is tuple:
         model, tokenizer = output
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 86245098ab..c999fa8124 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -79,7 +79,6 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
             all_stop_strings += st
 
     shared.stop_everything = False
-    clear_torch_cache()
     seed = set_manual_seed(state['seed'])
     last_update = -1
     reply = ''
@@ -288,6 +287,9 @@ def get_reply_from_output_ids(output_ids, state=None, starting_from=0):
 
 
 def generate_reply_HF(question, original_question, seed, state, stopping_strings=None, is_chat=False):
+    if shared.args.loader == 'Transformers':
+        clear_torch_cache()
+
     generate_params = {}
     for k in ['max_new_tokens', 'temperature', 'temperature_last', 'dynamic_temperature', 'dynatemp_low', 'dynatemp_high', 'dynatemp_exponent', 'smoothing_factor', 'smoothing_curve', 'top_p', 'min_p', 'top_k', 'repetition_penalty', 'presence_penalty', 'frequency_penalty', 'repetition_penalty_range', 'typical_p', 'tfs', 'top_a', 'guidance_scale', 'penalty_alpha', 'mirostat_mode', 'mirostat_tau', 'mirostat_eta', 'do_sample', 'encoder_repetition_penalty', 'no_repeat_ngram_size', 'dry_multiplier', 'dry_base', 'dry_allowed_length', 'dry_sequence_breakers', 'xtc_threshold', 'xtc_probability']:
         if k in state:
@@ -393,7 +395,6 @@ def generate_reply_HF(question, original_question, seed, state, stopping_strings
 
             def generate_with_callback(callback=None, *args, **kwargs):
                 kwargs['stopping_criteria'].append(Stream(callback_func=callback))
-                clear_torch_cache()
                 with torch.no_grad():
                     shared.model.generate(**kwargs)
 

From 725639118a84fc78a1bdb9c49538a941899eee58 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 1 Jan 2025 13:53:50 -0800
Subject: [PATCH 0130/1701] UI: Use a tab length of 2 for lists (rather than 4)

---
 modules/html_generator.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 40a56731e7..8160f8b69e 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -195,7 +195,7 @@ def convert_to_markdown(string):
         result = re.sub(list_item_pattern, r'\g<1> ' + delete_str, result)
 
         # Convert to HTML using markdown
-        html_output = markdown.markdown(result, extensions=['fenced_code', 'tables'])
+        html_output = markdown.markdown(result, extensions=['fenced_code', 'tables'], tab_length=2)
 
         # Remove the delete string from the HTML output
         pos = html_output.rfind(delete_str)
@@ -203,7 +203,7 @@ def convert_to_markdown(string):
             html_output = html_output[:pos] + html_output[pos + len(delete_str):]
     else:
         # Convert to HTML using markdown
-        html_output = markdown.markdown(result, extensions=['fenced_code', 'tables'])
+        html_output = markdown.markdown(result, extensions=['fenced_code', 'tables'], tab_length=2)
 
     # Unescape code blocks
     pattern = re.compile(r']*>(.*?)', re.DOTALL)

From 9163951f3aa65956ffcc7518d4a9c6ccca353b0b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 1 Jan 2025 17:49:57 -0800
Subject: [PATCH 0131/1701] UI: reduce the CPU usage during text streaming

---
 js/main.js | 20 ++++++++++++++------
 1 file changed, 14 insertions(+), 6 deletions(-)

diff --git a/js/main.js b/js/main.js
index a8018175db..76a0184b8e 100644
--- a/js/main.js
+++ b/js/main.js
@@ -446,25 +446,33 @@ function toggleBigPicture() {
 //------------------------------------------------
 // Handle the chat input box growth
 //------------------------------------------------
-let currentChatInputHeight = 0;
+
+// Variables to store current dimensions
+let currentChatInputHeight = chatInput.clientHeight;
+
+// Cache DOM elements
+const chatContainer = document.getElementById("chat").parentNode.parentNode.parentNode;
+const chatInput = document.querySelector("#chat-input textarea");
 
 // Update chat layout based on chat and input dimensions
 function updateCssProperties() {
-  const chatContainer = document.getElementById("chat").parentNode.parentNode.parentNode;
-  const chatInputHeight = document.querySelector("#chat-input textarea").clientHeight;
+  const chatInputHeight = chatInput.clientHeight;
 
   // Check if the chat container is visible
   if (chatContainer.clientHeight > 0) {
-    const newChatHeight = `${chatContainer.parentNode.clientHeight - chatInputHeight + 40 - 100 - 20}px`;
+    const chatContainerParentHeight = chatContainer.parentNode.clientHeight;
+    const newChatHeight = `${chatContainerParentHeight - chatInputHeight - 80}px`;
+
     document.documentElement.style.setProperty("--chat-height", newChatHeight);
     document.documentElement.style.setProperty("--input-delta", `${chatInputHeight - 40}px`);
 
     // Adjust scrollTop based on input height change
     if (chatInputHeight !== currentChatInputHeight) {
-      if (!isScrolled && chatInputHeight < currentChatInputHeight) {
+      const deltaHeight = chatInputHeight - currentChatInputHeight;
+      if (!isScrolled && deltaHeight < 0) {
         chatContainer.scrollTop = chatContainer.scrollHeight;
       } else {
-        chatContainer.scrollTop += chatInputHeight - currentChatInputHeight;
+        chatContainer.scrollTop += deltaHeight;
       }
 
       currentChatInputHeight = chatInputHeight;

From f011787a835b8eb2d3cf354cec4b76a8109defbc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 1 Jan 2025 17:55:18 -0800
Subject: [PATCH 0132/1701] UI: make codeblocks scroll horizontally on overflow

---
 css/main.css | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/css/main.css b/css/main.css
index 314b36e0eb..a08350ad14 100644
--- a/css/main.css
+++ b/css/main.css
@@ -538,8 +538,8 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 .message-body pre > code {
-    white-space: pre-wrap !important;
-    word-wrap: break-word !important;
+    white-space: pre !important;
+    overflow-x: auto !important;
     border: 1px solid #666;
     border-radius: 5px;
     font-size: 82%;

From 979e1f1bd68ed9e44b9af0633017f9ecc456bbcd Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 1 Jan 2025 17:57:09 -0800
Subject: [PATCH 0133/1701] Fix a bug after
 9163951f3aa65956ffcc7518d4a9c6ccca353b0b

---
 js/main.js | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/js/main.js b/js/main.js
index 76a0184b8e..efb8423807 100644
--- a/js/main.js
+++ b/js/main.js
@@ -447,13 +447,13 @@ function toggleBigPicture() {
 // Handle the chat input box growth
 //------------------------------------------------
 
-// Variables to store current dimensions
-let currentChatInputHeight = chatInput.clientHeight;
-
 // Cache DOM elements
 const chatContainer = document.getElementById("chat").parentNode.parentNode.parentNode;
 const chatInput = document.querySelector("#chat-input textarea");
 
+// Variables to store current dimensions
+let currentChatInputHeight = chatInput.clientHeight;
+
 // Update chat layout based on chat and input dimensions
 function updateCssProperties() {
   const chatInputHeight = chatInput.clientHeight;

From 13c033c745b54ce3b35c805ca0ff12f8f93bfb5d Mon Sep 17 00:00:00 2001
From: Petr Korolev 
Date: Thu, 2 Jan 2025 06:06:11 +0300
Subject: [PATCH 0134/1701] Fix CUDA error on MPS backend during API request
 (#6572)

---------

Co-authored-by: oobabooga 
---
 modules/LoRA.py            | 14 +++---------
 modules/logits.py          | 25 +++++++++------------
 modules/models.py          | 46 ++++++++++++++++++++++++--------------
 modules/sampler_hijack.py  | 15 +++++++------
 modules/text_generation.py | 28 ++++++++++-------------
 5 files changed, 63 insertions(+), 65 deletions(-)

diff --git a/modules/LoRA.py b/modules/LoRA.py
index 4fd144ba00..e1ad01d769 100644
--- a/modules/LoRA.py
+++ b/modules/LoRA.py
@@ -1,11 +1,8 @@
 from pathlib import Path
 
-import torch
-from transformers import is_torch_xpu_available
-
 import modules.shared as shared
 from modules.logging_colors import logger
-from modules.models import reload_model
+from modules.models import get_device, reload_model
 
 
 def add_lora_to_model(lora_names):
@@ -132,14 +129,9 @@ def add_lora_transformers(lora_names):
         if not shared.args.load_in_8bit and not shared.args.cpu:
             shared.model.half()
             if not hasattr(shared.model, "hf_device_map"):
-                if torch.backends.mps.is_available():
-                    device = torch.device('mps')
-                    shared.model = shared.model.to(device)
-                elif is_torch_xpu_available():
-                    device = torch.device("xpu:0")
+                device = get_device()
+                if device:
                     shared.model = shared.model.to(device)
-                else:
-                    shared.model = shared.model.cuda()
 
     shared.lora_names = lora_names
 
diff --git a/modules/logits.py b/modules/logits.py
index 73cabb41b8..f8a1e80c28 100644
--- a/modules/logits.py
+++ b/modules/logits.py
@@ -2,11 +2,10 @@
 import traceback
 
 import torch
-from transformers import is_torch_npu_available, is_torch_xpu_available
 
 from modules import models, sampler_hijack, shared
 from modules.logging_colors import logger
-from modules.models import load_model
+from modules.models import get_device, load_model
 from modules.text_generation import generate_reply
 
 global_scores = None
@@ -57,23 +56,21 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
         scores = sampler_hijack.global_scores[-1]
     else:
         if is_non_hf_exllamav2:
-            if is_torch_xpu_available():
-                tokens = shared.tokenizer.encode(prompt).to("xpu:0")
-            elif is_torch_npu_available():
-                tokens = shared.tokenizer.encode(prompt).to("npu:0")
-            else:
-                tokens = shared.tokenizer.encode(prompt).cuda()
+            device = get_device()
+            tokens = shared.tokenizer.encode(prompt)
+            if device:
+                tokens = tokens.to(device)
+
             scores = shared.model.get_logits(tokens)[-1][-1]
         elif is_non_hf_llamacpp:
             tokens = shared.tokenizer.encode(prompt)
             scores = shared.model.get_logits(tokens)[-1][-1]
         else:
-            if is_torch_xpu_available():
-                tokens = shared.tokenizer.encode(prompt, return_tensors='pt').to("xpu:0")
-            elif is_torch_npu_available():
-                tokens = shared.tokenizer.encode(prompt, return_tensors='pt').to("npu:0")
-            else:
-                tokens = shared.tokenizer.encode(prompt, return_tensors='pt').cuda()
+            device = get_device()
+            tokens = shared.tokenizer.encode(prompt, return_tensors='pt')
+            if device:
+                tokens = tokens.to(device)
+
             output = shared.model(input_ids=tokens)
             scores = output['logits'][-1][-1]
 
diff --git a/modules/models.py b/modules/models.py
index 7a52c07c70..d906535b5c 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -21,11 +21,12 @@
     AutoModelForSeq2SeqLM,
     AutoTokenizer,
     BitsAndBytesConfig,
-    GPTQConfig
+    GPTQConfig,
+    is_torch_npu_available,
+    is_torch_xpu_available
 )
 
 import modules.shared as shared
-from modules import sampler_hijack
 from modules.logging_colors import logger
 from modules.models_settings import get_model_metadata
 
@@ -56,8 +57,6 @@
     ds_config = generate_ds_config(shared.args.bf16, 1 * world_size, shared.args.nvme_offload_dir)
     dschf = HfDeepSpeedConfig(ds_config)  # Keep this object alive for the Transformers integration
 
-sampler_hijack.hijack_samplers()
-
 
 last_generation_time = time.time()
 
@@ -172,17 +171,9 @@ def huggingface_loader(model_name):
 
         model = LoaderClass.from_pretrained(path_to_model, **params)
         if not (hasattr(model, 'is_loaded_in_4bit') and model.is_loaded_in_4bit):
-            if torch.backends.mps.is_available():
-                device = torch.device('mps')
-                model = model.to(device)
-            elif is_xpu_available():
-                device = torch.device("xpu")
-                model = model.to(device)
-            elif is_npu_available():
-                device = torch.device("npu")
+            device = get_device()
+            if device:
                 model = model.to(device)
-            else:
-                model = model.cuda()
 
     # DeepSpeed ZeRO-3
     elif shared.args.deepspeed:
@@ -380,13 +371,34 @@ def get_max_memory_dict():
     return max_memory if len(max_memory) > 0 else None
 
 
+def get_device():
+    if torch.cuda.is_available():
+        return torch.device('cuda')
+    elif shared.args.deepspeed:
+        import deepspeed
+        return deepspeed.get_accelerator().current_device_name()
+    elif torch.backends.mps.is_available():
+        return torch.device('mps')
+    elif is_torch_xpu_available():
+        return torch.device('xpu:0')
+    elif is_torch_npu_available():
+        return torch.device('npu:0')
+    else:
+        return None
+
+
 def clear_torch_cache():
     gc.collect()
     if not shared.args.cpu:
-        if is_xpu_available():
-            torch.xpu.empty_cache()
-        else:
+        if torch.cuda.is_available():
             torch.cuda.empty_cache()
+        elif is_xpu_available():
+            torch.xpu.empty_cache()
+        elif is_npu_available():
+            torch.npu.empty_cache()
+        elif torch.backends.mps.is_available():
+            if hasattr(torch.backends.mps, 'empty_cache'):
+                torch.backends.mps.empty_cache()
 
 
 def unload_model(keep_model_name=False):
diff --git a/modules/sampler_hijack.py b/modules/sampler_hijack.py
index 24dbcf2ee1..62ceca8d15 100644
--- a/modules/sampler_hijack.py
+++ b/modules/sampler_hijack.py
@@ -5,7 +5,7 @@
 
 import torch
 import transformers
-from transformers import LogitsWarper, is_torch_xpu_available
+from transformers import LogitsWarper
 from transformers.generation.logits_process import (
     LogitNormalization,
     LogitsProcessor,
@@ -14,6 +14,7 @@
 
 from modules import shared
 from modules.logging_colors import logger
+from modules.models import get_device
 
 global_scores = None
 
@@ -339,12 +340,12 @@ def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> to
                 break
 
         # Normalize the probabilities of the remaining words
-        if is_torch_xpu_available():
-            prob_topk = torch.softmax(sorted_logits, dim=0).to("xpu")
-            prev_i = torch.multinomial(prob_topk, num_samples=1, replacement=True).to("xpu")
-        else:
-            prob_topk = torch.softmax(sorted_logits, dim=0).to('cuda')
-            prev_i = torch.multinomial(prob_topk, num_samples=1, replacement=True).to('cuda')
+        prob_topk = torch.softmax(sorted_logits, dim=0)
+        prev_i = torch.multinomial(prob_topk, num_samples=1, replacement=True)
+        device = get_device()
+        if device:
+            prob_topk = prob_topk.to(device)
+            prev_i = prev_i.to(device)
 
         observed_surprise = -math.log2(prob_topk[prev_i])
         self.e = observed_surprise - self.mirostat_tau
diff --git a/modules/text_generation.py b/modules/text_generation.py
index c999fa8124..db415dce7c 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -16,7 +16,7 @@
 )
 
 import modules.shared as shared
-from modules import models
+from modules import models, sampler_hijack
 from modules.cache_utils import process_llamacpp_cache
 from modules.callbacks import (
     Iteratorize,
@@ -28,7 +28,9 @@
 from modules.grammar.logits_process import GrammarConstrainedLogitsProcessor
 from modules.html_generator import generate_basic_html
 from modules.logging_colors import logger
-from modules.models import clear_torch_cache, load_model
+from modules.models import clear_torch_cache, get_device, load_model
+
+sampler_hijack.hijack_samplers()
 
 
 def generate_reply(*args, **kwargs):
@@ -159,18 +161,12 @@ def encode(prompt, add_special_tokens=True, add_bos_token=True, truncation_lengt
 
     if shared.model.__class__.__name__ in ['LlamaCppModel', 'Exllamav2Model', 'TensorRTLLMModel'] or shared.args.cpu:
         return input_ids
-    elif shared.args.deepspeed:
-        import deepspeed
-        return input_ids.to(deepspeed.get_accelerator().current_device_name())
-    elif torch.backends.mps.is_available():
-        device = torch.device('mps')
-        return input_ids.to(device)
-    elif is_torch_xpu_available():
-        return input_ids.to("xpu:0")
-    elif is_torch_npu_available():
-        return input_ids.to("npu:0")
     else:
-        return input_ids.cuda()
+        device = get_device()
+        if device:
+            return input_ids.to(device)
+
+        return input_ids
 
 
 def decode(output_ids, skip_special_tokens=True):
@@ -328,7 +324,6 @@ def generate_reply_HF(question, original_question, seed, state, stopping_strings
     # Encode the input
     input_ids = encode(question, add_bos_token=state['add_bos_token'], truncation_length=get_max_prompt_length(state))
     output = input_ids[0]
-    cuda = not any((shared.args.cpu, shared.args.deepspeed))
     if state['auto_max_new_tokens']:
         generate_params['max_new_tokens'] = state['truncation_length'] - input_ids.shape[-1]
 
@@ -383,8 +378,9 @@ def generate_reply_HF(question, original_question, seed, state, stopping_strings
         if not state['stream']:
             with torch.no_grad():
                 output = shared.model.generate(**generate_params)[0]
-                if cuda:
-                    output = output.cuda()
+                device = get_device()
+                if device:
+                    output = output.to(device)
 
             starting_from = 0 if shared.is_seq2seq else len(input_ids[0])
             yield get_reply_from_output_ids(output, state, starting_from=starting_from)

From 75f1b5ccde221783b4fedef92da6afc3a44824f0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Jan 2025 16:24:18 -0800
Subject: [PATCH 0135/1701] UI: add a "Branch chat" button

---
 css/main.css       |  1 +
 modules/chat.py    | 15 +++++++++++++++
 modules/ui_chat.py |  5 +++++
 3 files changed, 21 insertions(+)

diff --git a/css/main.css b/css/main.css
index a08350ad14..870f4192de 100644
--- a/css/main.css
+++ b/css/main.css
@@ -980,6 +980,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 #rename-row {
     width: 100%;
     justify-content: center;
+    gap: 10px;
 }
 
 
diff --git a/modules/chat.py b/modules/chat.py
index 2638c79431..81328385d5 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1092,6 +1092,21 @@ def handle_delete_chat_confirm_click(state):
     ]
 
 
+def handle_branch_chat_click(state):
+    history = state['history']
+    new_unique_id = datetime.now().strftime('%Y%m%d-%H-%M-%S')
+    save_history(history, new_unique_id, state['character_menu'], state['mode'])
+
+    histories = find_all_histories_with_first_prompts(state)
+    html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
+
+    convert_to_markdown.cache_clear()
+
+    past_chats_update = gr.update(choices=histories, value=new_unique_id)
+
+    return [history, html, past_chats_update]
+
+
 def handle_rename_chat_click():
     return [
         gr.update(value="My New Chat"),
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index e372f5c223..b09d008164 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -24,6 +24,7 @@ def create_ui():
         with gr.Row(elem_id='past-chats-row', elem_classes=['pretty_scrollbar']):
             with gr.Column():
                 with gr.Row(elem_id='past-chats-buttons'):
+                    shared.gradio['branch_chat'] = gr.Button('Branch', elem_classes='refresh-button', interactive=not mu)
                     shared.gradio['rename_chat'] = gr.Button('Rename', elem_classes='refresh-button', interactive=not mu)
                     shared.gradio['delete_chat'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu)
                     shared.gradio['Start new chat'] = gr.Button('New chat', elem_classes=['refresh-button', 'focus-on-chat-input'])
@@ -250,6 +251,10 @@ def create_event_handlers():
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_delete_chat_confirm_click, gradio('interface_state'), gradio('history', 'display', 'unique_id', 'delete-chat-row'), show_progress=False)
 
+    shared.gradio['branch_chat'].click(
+        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
+        chat.handle_branch_chat_click, gradio('interface_state'), gradio('history', 'display', 'unique_id'), show_progress=False)
+
     shared.gradio['rename_chat'].click(chat.handle_rename_chat_click, None, gradio('rename_to', 'rename-row'), show_progress=False)
     shared.gradio['rename_to-cancel'].click(lambda: gr.update(visible=False), None, gradio('rename-row'), show_progress=False)
     shared.gradio['rename_to-confirm'].click(

From 973255cb0bc61c0cf50e6f562bfc632a3d6fab1c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Jan 2025 16:48:24 -0800
Subject: [PATCH 0136/1701] UI: fix codeblocks overflowing on mobile

---
 css/main.css | 1 +
 1 file changed, 1 insertion(+)

diff --git a/css/main.css b/css/main.css
index 870f4192de..564a39a137 100644
--- a/css/main.css
+++ b/css/main.css
@@ -540,6 +540,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 .message-body pre > code {
     white-space: pre !important;
     overflow-x: auto !important;
+    max-width: calc(100dvw - 39px);
     border: 1px solid #666;
     border-radius: 5px;
     font-size: 82%;

From b8fc9010fa4667c78a01adb6f1449e515e94709f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Jan 2025 16:57:04 -0800
Subject: [PATCH 0137/1701] UI: fix `orjson.JSONDecodeError` error on page
 reload

---
 modules/ui_chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index b09d008164..ecebc28ed6 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -18,7 +18,7 @@ def create_ui():
     mu = shared.args.multi_user
 
     shared.gradio['Chat input'] = gr.State()
-    shared.gradio['history'] = gr.JSON({'internal': [], 'visible': []}, visible=False)
+    shared.gradio['history'] = gr.JSON(visible=False)
 
     with gr.Tab('Chat', elem_id='chat-tab'):
         with gr.Row(elem_id='past-chats-row', elem_classes=['pretty_scrollbar']):

From 4b3e1b37573057fe878ff48037569253b9df5576 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Jan 2025 18:46:40 -0800
Subject: [PATCH 0138/1701] UI: add a "Search chats" input field

---
 css/main.css       |  6 +++++-
 modules/chat.py    | 40 +++++++++++++++++++++++++---------------
 modules/ui.py      |  1 +
 modules/ui_chat.py |  6 ++++++
 4 files changed, 37 insertions(+), 16 deletions(-)

diff --git a/css/main.css b/css/main.css
index 564a39a137..64da16021b 100644
--- a/css/main.css
+++ b/css/main.css
@@ -876,6 +876,10 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     flex-shrink: 1;
 }
 
+#search_chat > :nth-child(2) > :first-child {
+    display: none;
+}
+
 /* ----------------------------------------------
   Keep dropdown menus above errored components
 ---------------------------------------------- */
@@ -911,7 +915,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 #past-chats {
-    max-height: calc(100dvh - 90px);
+    max-height: calc(100dvh - 135px);
     overflow-y: scroll !important;
     border-radius: 0;
     scrollbar-width: auto;
diff --git a/modules/chat.py b/modules/chat.py
index 81328385d5..694c137bd9 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -593,21 +593,26 @@ def find_all_histories_with_first_prompts(state):
     result = []
     for i, path in enumerate(histories):
         filename = path.stem
+        file_content = ""
+        with open(path, 'r', encoding='utf-8') as f:
+            file_content = f.read()
+
+        if state['search_chat'] and state['search_chat'] not in file_content:
+            continue
+
+        data = json.loads(file_content)
         if re.match(r'^[0-9]{8}-[0-9]{2}-[0-9]{2}-[0-9]{2}$', filename):
-            with open(path, 'r', encoding='utf-8') as f:
-                data = json.load(f)
-
-                first_prompt = ""
-                if data and 'visible' in data and len(data['visible']) > 0:
-                    if data['internal'][0][0] == '<|BEGIN-VISIBLE-CHAT|>':
-                        if len(data['visible']) > 1:
-                            first_prompt = html.unescape(data['visible'][1][0])
-                        elif i == 0:
-                            first_prompt = "New chat"
-                    else:
-                        first_prompt = html.unescape(data['visible'][0][0])
-                elif i == 0:
-                    first_prompt = "New chat"
+            first_prompt = ""
+            if data and 'visible' in data and len(data['visible']) > 0:
+                if data['internal'][0][0] == '<|BEGIN-VISIBLE-CHAT|>':
+                    if len(data['visible']) > 1:
+                        first_prompt = html.unescape(data['visible'][1][0])
+                    elif i == 0:
+                        first_prompt = "New chat"
+                else:
+                    first_prompt = html.unescape(data['visible'][0][0])
+            elif i == 0:
+                first_prompt = "New chat"
         else:
             first_prompt = filename
 
@@ -615,7 +620,7 @@ def find_all_histories_with_first_prompts(state):
 
         # Truncate the first prompt if it's longer than 30 characters
         if len(first_prompt) > 30:
-            first_prompt = first_prompt[:30-3] + '...'
+            first_prompt = first_prompt[:30 - 3] + '...'
 
         result.append((first_prompt, filename))
 
@@ -1124,6 +1129,11 @@ def handle_rename_chat_confirm(rename_to, state):
     ]
 
 
+def handle_search_chat_change(state):
+    histories = find_all_histories_with_first_prompts(state)
+    return gr.update(choices=histories)
+
+
 def handle_upload_chat_history(load_chat_history, state):
     history = start_new_chat(state)
     history = load_history_json(load_chat_history, history)
diff --git a/modules/ui.py b/modules/ui.py
index 4bfea9fade..a3bf520f3e 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -230,6 +230,7 @@ def list_interface_input_elements():
         'start_with',
         'character_menu',
         'history',
+        'search_chat',
         'unique_id',
         'name1',
         'user_bio',
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index ecebc28ed6..b92dd9ae4d 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -29,6 +29,8 @@ def create_ui():
                     shared.gradio['delete_chat'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu)
                     shared.gradio['Start new chat'] = gr.Button('New chat', elem_classes=['refresh-button', 'focus-on-chat-input'])
 
+                shared.gradio['search_chat'] = gr.Textbox(placeholder='Search chats...', max_lines=1, elem_id='search_chat')
+
                 with gr.Row(elem_id='delete-chat-row', visible=False) as shared.gradio['delete-chat-row']:
                     shared.gradio['delete_chat-cancel'] = gr.Button('Cancel', elem_classes=['refresh-button', 'focus-on-chat-input'])
                     shared.gradio['delete_chat-confirm'] = gr.Button('Confirm', variant='stop', elem_classes=['refresh-button', 'focus-on-chat-input'])
@@ -265,6 +267,10 @@ def create_event_handlers():
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_rename_chat_confirm, gradio('rename_to', 'interface_state'), gradio('unique_id', 'rename-row'), show_progress=False)
 
+    shared.gradio['search_chat'].change(
+        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
+        chat.handle_search_chat_change, gradio('interface_state'), gradio('unique_id'), show_progress=False)
+
     shared.gradio['load_chat_history'].upload(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_upload_chat_history, gradio('load_chat_history', 'interface_state'), gradio('history', 'display', 'unique_id'), show_progress=False).then(

From e2702200e131944e58e19a9894b8ebadb24c6f75 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Jan 2025 19:26:50 -0800
Subject: [PATCH 0139/1701] UI: fix the font size of lists in chat mode

---
 css/chat_style-cai-chat.css | 1 +
 1 file changed, 1 insertion(+)

diff --git a/css/chat_style-cai-chat.css b/css/chat_style-cai-chat.css
index d7b1ba88e4..93276bd347 100644
--- a/css/chat_style-cai-chat.css
+++ b/css/chat_style-cai-chat.css
@@ -9,6 +9,7 @@
 
 .message-body {
     margin-top: 3px;
+    font-size: 15px !important;
 }
 
 .circle-you {

From 3815f46838e4bb2542d32fbdc5151bc4f2a5abb7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 3 Jan 2025 04:35:29 -0800
Subject: [PATCH 0140/1701] UI: minor style improvements to chat tab

---
 css/main.css | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 64da16021b..0ccaf1062e 100644
--- a/css/main.css
+++ b/css/main.css
@@ -839,7 +839,11 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
   Past chats menus
 ---------------------------------------------- */
 #rename-row label {
-    margin-top: var(--layout-gap);
+    margin-top: 0;
+}
+
+#rename-row > :nth-child(2) {
+    justify-content: center;
 }
 
 /* ----------------------------------------------

From 9f24885bd2f70f7d07742f766107983b301434af Mon Sep 17 00:00:00 2001
From: mamei16 
Date: Sat, 4 Jan 2025 19:41:31 +0100
Subject: [PATCH 0141/1701] Sane handling of markdown lists (#6626)

---
 modules/html_generator.py      |   7 +-
 modules/sane_markdown_lists.py | 336 +++++++++++++++++++++++++++++++++
 2 files changed, 340 insertions(+), 3 deletions(-)
 create mode 100644 modules/sane_markdown_lists.py

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 8160f8b69e..e61fc55824 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -9,6 +9,7 @@
 from PIL import Image, ImageOps
 
 from modules import shared
+from modules.sane_markdown_lists import SaneListExtension
 from modules.utils import get_available_chat_styles
 
 # This is to store the paths to the thumbnails of the profile pictures
@@ -174,7 +175,7 @@ def convert_to_markdown(string):
             result += '\n'
         # Also don't add an extra \n for lists
         elif stripped_line.startswith('-') or stripped_line.startswith('*') or stripped_line.startswith('+') or stripped_line.startswith('>') or re.match(r'\d+\.', stripped_line):
-            result += '\n'
+            result += '  \n'
         else:
             result += '  \n'
 
@@ -195,7 +196,7 @@ def convert_to_markdown(string):
         result = re.sub(list_item_pattern, r'\g<1> ' + delete_str, result)
 
         # Convert to HTML using markdown
-        html_output = markdown.markdown(result, extensions=['fenced_code', 'tables'], tab_length=2)
+        html_output = markdown.markdown(result, extensions=['fenced_code', 'tables', SaneListExtension()])
 
         # Remove the delete string from the HTML output
         pos = html_output.rfind(delete_str)
@@ -203,7 +204,7 @@ def convert_to_markdown(string):
             html_output = html_output[:pos] + html_output[pos + len(delete_str):]
     else:
         # Convert to HTML using markdown
-        html_output = markdown.markdown(result, extensions=['fenced_code', 'tables'], tab_length=2)
+        html_output = markdown.markdown(result, extensions=['fenced_code', 'tables', SaneListExtension()])
 
     # Unescape code blocks
     pattern = re.compile(r']*>(.*?)', re.DOTALL)
diff --git a/modules/sane_markdown_lists.py b/modules/sane_markdown_lists.py
new file mode 100644
index 0000000000..1e1d76fd8e
--- /dev/null
+++ b/modules/sane_markdown_lists.py
@@ -0,0 +1,336 @@
+# Code based on the Sane List Extension for Python-Markdown
+# =======================================
+
+# Modify the behavior of Lists in Python-Markdown to act in a sane manner.
+
+# See https://Python-Markdown.github.io/extensions/sane_lists
+# for documentation.
+
+# Original code Copyright 2011 [Waylan Limberg](http://achinghead.com)
+
+# All changes Copyright 2011-2014 The Python Markdown Project
+
+# License: [BSD](https://opensource.org/licenses/bsd-license.php)
+
+"""
+Modify the behavior of Lists in Python-Markdown to act in a sane manner.
+"""
+
+from __future__ import annotations
+
+import re
+import xml.etree.ElementTree as etree
+from typing import TYPE_CHECKING
+
+from markdown import Extension
+from markdown.blockparser import BlockParser
+from markdown.blockprocessors import (
+    ListIndentProcessor,
+    OListProcessor,
+    ParagraphProcessor
+)
+
+if TYPE_CHECKING:  # pragma: no cover
+    from markdown import blockparser
+
+
+# The min. number of added leading spaces needed to start a nested list
+MIN_NESTED_LIST_INDENT = 2
+assert MIN_NESTED_LIST_INDENT > 1, "'MIN_NESTED_LIST_INDENT' must be > 1"
+
+
+class SaneListIndentProcessor(ListIndentProcessor):
+    """ Process children of list items.
+
+    Example
+
+        * a list item
+            process this part
+
+            or this part
+
+    """
+
+    def __init__(self, *args):
+        super().__init__(*args)
+        self.INDENT_RE = re.compile(r'^(([ ])+)')
+
+    def test(self, parent: etree.Element, block: str) -> bool:
+        return block.startswith(' ' * MIN_NESTED_LIST_INDENT) and \
+            not self.parser.state.isstate('detabbed') and \
+            (parent.tag in self.ITEM_TYPES or
+                (len(parent) and parent[-1] is not None and
+                    (parent[-1].tag in self.LIST_TYPES)))
+
+    def get_level(self, parent: etree.Element, block: str) -> tuple[int, etree.Element]:
+        """ Get level of indentation based on list level. """
+        # Get indent level
+        m = self.INDENT_RE.match(block)
+        if m:
+            indent_level = len(m.group(1)) / MIN_NESTED_LIST_INDENT
+        else:
+            indent_level = 0
+        if self.parser.state.isstate('list'):
+            # We're in a tight-list - so we already are at correct parent.
+            level = 1
+        else:
+            # We're in a loose-list - so we need to find parent.
+            level = 0
+        # Step through children of tree to find matching indent level.
+        while indent_level > level:
+            child = self.lastChild(parent)
+            if (child is not None and
+                    (child.tag in self.LIST_TYPES or child.tag in self.ITEM_TYPES)):
+                if child.tag in self.LIST_TYPES:
+                    level += 1
+                parent = child
+            else:
+                # No more child levels. If we're short of `indent_level`,
+                # we have a code block. So we stop here.
+                break
+        return level, parent
+
+    def detab(self, text: str, length: int | None = None) -> tuple[str, str]:
+        """ Remove a tab from the front of each line of the given text. """
+        if length is None:
+            length = MIN_NESTED_LIST_INDENT
+        newtext = []
+        lines = text.split('\n')
+        for line in lines:
+            if line.startswith(' ' * length):
+                newtext.append(line[length:])
+            elif not line.strip():
+                newtext.append('')
+            else:
+                break
+        return '\n'.join(newtext), '\n'.join(lines[len(newtext):])
+
+    def looseDetab(self, text: str, level: int = 1) -> str:
+        """ Remove indentation from front of lines but allowing dedented lines. """
+        lines = text.split('\n')
+        for i in range(len(lines)):
+            if lines[i].startswith(' ' * MIN_NESTED_LIST_INDENT * level):
+                lines[i] = lines[i][MIN_NESTED_LIST_INDENT * level:]
+        return '\n'.join(lines)
+
+
+class SaneOListProcessor(OListProcessor):
+    """ Override `SIBLING_TAGS` to not include `ul` and set `LAZY_OL` to `False`. """
+
+    SIBLING_TAGS = ['ol']
+    """ Exclude `ul` from list of siblings. """
+    LAZY_OL = False
+    """ Disable lazy list behavior. """
+
+    def __init__(self, parser: blockparser.BlockParser):
+        super().__init__(parser)
+        # This restriction stems from the 'CodeBlockProcessor' class,
+        # which automatically matches blocks with an indent = self.tab_length
+        max_list_start_indent = self.tab_length - 1
+        # Detect an item (e.g., `1. item`)
+        self.RE = re.compile(r'^[ ]{0,%d}[\*_]{0,2}\d+\.[ ]+(.*)' % max_list_start_indent)
+        # Detect items on secondary lines. they can be of either list type.
+        self.CHILD_RE = re.compile(r'^[ ]{0,%d}([\*_]{0,2})((\d+\.))[ ]+(.*)' % (MIN_NESTED_LIST_INDENT - 1))
+        # Detect indented (nested) items of either type
+        self.INDENT_RE = re.compile(r'^[ ]{%d,%d}[\*_]{0,2}((\d+\.)|[*+-])[ ]+.*' %
+                                    (MIN_NESTED_LIST_INDENT, self.tab_length * 2 - 1))
+
+    def run(self, parent: etree.Element, blocks: list[str]) -> None:
+        # Check for multiple items in one block.
+        items = self.get_items(blocks.pop(0))
+        sibling = self.lastChild(parent)
+
+        if sibling is not None and sibling.tag in self.SIBLING_TAGS:
+            # Previous block was a list item, so set that as parent
+            lst = sibling
+            # make sure previous item is in a `p` - if the item has text,
+            # then it isn't in a `p`
+            if lst[-1].text:
+                # since it's possible there are other children for this
+                # sibling, we can't just `SubElement` the `p`, we need to
+                # insert it as the first item.
+                p = etree.Element('p')
+                p.text = lst[-1].text
+                lst[-1].text = ''
+                lst[-1].insert(0, p)
+            # if the last item has a tail, then the tail needs to be put in a `p`
+            # likely only when a header is not followed by a blank line
+            lch = self.lastChild(lst[-1])
+            if lch is not None and lch.tail:
+                p = etree.SubElement(lst[-1], 'p')
+                p.text = lch.tail.lstrip()
+                lch.tail = ''
+
+            # parse first block differently as it gets wrapped in a `p`.
+            li = etree.SubElement(lst, 'li')
+            self.parser.state.set('looselist')
+            firstitem = items.pop(0)
+            self.parser.parseBlocks(li, [firstitem])
+            self.parser.state.reset()
+        elif parent.tag in ['ol', 'ul']:
+            # this catches the edge case of a multi-item indented list whose
+            # first item is in a blank parent-list item:
+            #     * * subitem1
+            #         * subitem2
+            # see also `ListIndentProcessor`
+            lst = parent
+        else:
+            # This is a new list so create parent with appropriate tag.
+            lst = etree.SubElement(parent, self.TAG)
+            # Check if a custom start integer is set
+            if not self.LAZY_OL and self.STARTSWITH != '1':
+                lst.attrib['start'] = self.STARTSWITH
+
+        self.parser.state.set('list')
+        # Loop through items in block, recursively parsing each with the
+        # appropriate parent.
+        for item in items:
+            if item.startswith(" " * MIN_NESTED_LIST_INDENT):
+                # Item is indented. Parse with last item as parent
+                self.parser.parseBlocks(lst[-1], [item])
+            else:
+                # New item. Create `li` and parse with it as parent
+                li = etree.SubElement(lst, 'li')
+                self.parser.parseBlocks(li, [item])
+        self.parser.state.reset()
+
+    def looseDetab(self, text: str, indent_length: int, level: int = 1) -> str:
+        """ Remove indentation from front of lines but allowing dedented lines. """
+        lines = text.split('\n')
+        for i in range(len(lines)):
+            if lines[i].startswith(' ' * indent_length * level):
+                lines[i] = lines[i][indent_length * level:]
+        return '\n'.join(lines)
+
+    def get_items(self, block: str) -> list[str]:
+        """ Break a block into list items. """
+        # If first level of list is indented, remove that indentation
+        if (indent_len := len(block) - len(block.lstrip())) > 0:
+            block = self.looseDetab(block, indent_len)
+        items = []
+        for line in block.split('\n'):
+            m = self.CHILD_RE.match(line)
+            if m:
+                # This is a new list item
+                # Check first item for the start index
+                if not items:
+                    # Detect the integer value of first list item
+                    INTEGER_RE = re.compile(r'(\d+)')
+                    self.STARTSWITH = INTEGER_RE.match(m.group(2)).group()
+                # Append to the list
+                items.append(m.group(1) + m.group(4))
+            elif self.INDENT_RE.match(line):
+                # This is an indented (possibly nested) item.
+                if items[-1].startswith(' ' * MIN_NESTED_LIST_INDENT):
+                    # Previous item was indented. Append to that item.
+                    items[-1] = '{}\n{}'.format(items[-1], line)
+                else:
+                    items.append(line)
+            else:
+                # This is another line of previous item. Append to that item.
+                items[-1] = '{}\n{}'.format(items[-1], line)
+        return items
+
+
+class SaneUListProcessor(SaneOListProcessor):
+    """ Override `SIBLING_TAGS` to not include `ol`. """
+
+    TAG: str = 'ul'
+    SIBLING_TAGS = ['ul']
+    """ Exclude `ol` from list of siblings. """
+
+    def __init__(self, parser: blockparser.BlockParser):
+        super().__init__(parser)
+        # Detect an item (e.g., `- item` or `+ item` or `* item`).
+        max_list_start_indent = self.tab_length - 1
+        self.RE = re.compile(r'^[ ]{0,%d}[*+-][ ]+(.*)' % max_list_start_indent)
+        self.CHILD_RE = re.compile(r'^[ ]{0,%d}(([*+-]))[ ]+(.*)' % (MIN_NESTED_LIST_INDENT - 1))
+
+    def get_items(self, block: str) -> list[str]:
+        """ Break a block into list items. """
+        # If first level of list is indented, remove that indentation
+        if (indent_len := len(block) - len(block.lstrip())) > 0:
+            block = self.looseDetab(block, indent_len)
+        items = []
+        for line in block.split('\n'):
+            m = self.CHILD_RE.match(line)
+            if m:
+                # Append to the list
+                items.append(m.group(3))
+            elif self.INDENT_RE.match(line):
+                # This is an indented (possibly nested) item.
+                if items[-1].startswith(' ' * MIN_NESTED_LIST_INDENT):
+                    # Previous item was indented. Append to that item.
+                    items[-1] = '{}\n{}'.format(items[-1], line)
+                else:
+                    items.append(line)
+            else:
+                # This is another line of previous item. Append to that item.
+                items[-1] = '{}\n{}'.format(items[-1], line)
+        return items
+
+
+class SaneParagraphProcessor(ParagraphProcessor):
+    """ Process Paragraph blocks. """
+
+    def __init__(self, parser: BlockParser):
+        super().__init__(parser)
+        max_list_start_indent = self.tab_length - 1
+        self.LIST_RE = re.compile(r"\s{2}\n(\s{0,%d}[\d+*-])" % max_list_start_indent)
+
+    def run(self, parent: etree.Element, blocks: list[str]) -> None:
+        block = blocks.pop(0)
+        if block.strip():
+            # Not a blank block. Add to parent, otherwise throw it away.
+            if self.parser.state.isstate('list'):
+                # The parent is a tight-list.
+                #
+                # Check for any children. This will likely only happen in a
+                # tight-list when a header isn't followed by a blank line.
+                # For example:
+                #
+                #     * # Header
+                #     Line 2 of list item - not part of header.
+                sibling = self.lastChild(parent)
+                if sibling is not None:
+                    # Insert after sibling.
+                    if sibling.tail:
+                        sibling.tail = '{}\n{}'.format(sibling.tail, block)
+                    else:
+                        sibling.tail = '\n%s' % block
+                else:
+                    # Append to parent.text
+                    if parent.text:
+                        parent.text = '{}\n{}'.format(parent.text, block)
+                    else:
+                        parent.text = block.lstrip()
+            else:
+                # Check if paragraph contains a list
+                next_list_block = None
+                if list_match := self.LIST_RE.search(block):
+                    list_start = list_match.end() - len(list_match.group(1))
+                    next_list_block = block[list_start:]
+                    block = block[:list_start]
+
+                # Create a regular paragraph
+                p = etree.SubElement(parent, 'p')
+                p.text = block.lstrip()
+
+                # If a list was found, parse its block separately with the paragraph as the parent
+                if next_list_block:
+                    self.parser.parseBlocks(p, [next_list_block])
+
+
+class SaneListExtension(Extension):
+    """ Add sane lists to Markdown. """
+
+    def extendMarkdown(self, md):
+        """ Override existing Processors. """
+        md.parser.blockprocessors.register(SaneListIndentProcessor(md.parser), 'indent', 90)
+        md.parser.blockprocessors.register(SaneOListProcessor(md.parser), 'olist', 40)
+        md.parser.blockprocessors.register(SaneUListProcessor(md.parser), 'ulist', 30)
+        md.parser.blockprocessors.register(SaneParagraphProcessor(md.parser), 'paragraph', 10)
+
+
+def makeExtension(**kwargs):  # pragma: no cover
+    return SaneListExtension(**kwargs)

From 0e673a7a4278deaedb3513b4e406cdcc45b19b4e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Jan 2025 11:40:24 -0800
Subject: [PATCH 0142/1701] UI: reduce the size of HTML sent to the UI during
 streaming

---
 modules/html_generator.py | 136 +++++++++++++++++---------------------
 1 file changed, 61 insertions(+), 75 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index e61fc55824..1f12d77a5c 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -268,29 +268,24 @@ def generate_instruct_html(history):
     for i, _row in enumerate(history):
         row = [convert_to_markdown_wrapped(entry, use_cache=i != len(history) - 1) for entry in _row]
 
-        if row[0]:  # don't display empty user messages
-            output += f"""
-                  
            
-                    
            
-                      
            
-                        {row[0]}
-                      
            
-                    
            
-                  
            
-                """
-
-        output += f"""
-              
            
-                
            
-                  
            
-                    {row[1]}
-                  
            
-                
            
-              
            
-            """
+        if row[0]:  # Don't display empty user messages
+            output += (
+                f'
            '
+                f'
            '
+                f'
            {row[0]}
            '
+                f'
            '
+                f'
            '
+            )
+
+        output += (
+            f'
            '
+            f'
            '
+            f'
            {row[1]}
            '
+            f'
            '
+            f'
            '
+        )
 
     output += ""
-
     return output
 
 
@@ -298,44 +293,39 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
     output = f'
            '
 
     # We use ?character and ?time.time() to force the browser to reset caches
-    img_bot = f'' if Path("cache/pfp_character_thumb.png").exists() else ''
-    img_me = f'' if Path("cache/pfp_me.png").exists() else ''
+    img_bot = (
+        f''
+        if Path("cache/pfp_character_thumb.png").exists() else ''
+    )
+
+    img_me = (
+        f''
+        if Path("cache/pfp_me.png").exists() else ''
+    )
 
     for i, _row in enumerate(history):
         row = [convert_to_markdown_wrapped(entry, use_cache=i != len(history) - 1) for entry in _row]
 
-        if row[0]:  # don't display empty user messages
-            output += f"""
-                  
            
-                    
            
-                      {img_me}
-                    
            
-                    
            
-                      
            
-                        {name1}
-                      
            
-                      
            
-                        {row[0]}
-                      
            
-                    
            
-                  
            
-                """
-
-        output += f"""
-              
            
-                
            
-                  {img_bot}
-                
            
-                
            
-                  
            
-                    {name2}
-                  
            
-                  
            
-                    {row[1]}
-                  
            
-                
            
-              
            
-            """
+        if row[0]:  # Don't display empty user messages
+            output += (
+                f'
            '
+                f'
            {img_me}
            '
+                f'
            '
+                f'
            {name1}
            '
+                f'
            {row[0]}
            '
+                f'
            '
+                f'
            '
+            )
+
+        output += (
+            f'
            '
+            f'
            {img_bot}
            '
+            f'
            '
+            f'
            {name2}
            '
+            f'
            {row[1]}
            '
+            f'
            '
+            f'
            '
+        )
 
     output += "
            "
     return output
@@ -347,26 +337,22 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
     for i, _row in enumerate(history):
         row = [convert_to_markdown_wrapped(entry, use_cache=i != len(history) - 1) for entry in _row]
 
-        if row[0]:  # don't display empty user messages
-            output += f"""
-              
            
-                
            
-                  
            
-                    {row[0]}
-                  
            
-                
            
-              
            
-            """
-
-        output += f"""
-          
            
-            
            
-              
            
-                {row[1]}
-              
            
-            
            
-          
            
-        """
+        if row[0]:  # Don't display empty user messages
+            output += (
+                f'
            '
+                f'
            '
+                f'
            {row[0]}
            '
+                f'
            '
+                f'
            '
+            )
+
+        output += (
+            f'
            '
+            f'
            '
+            f'
            {row[1]}
            '
+            f'
            '
+            f'
            '
+        )
 
     output += ""
     return output

From 049297fa660125467d8d2aed00f91901871dbf52 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Jan 2025 14:09:36 -0800
Subject: [PATCH 0143/1701] UI: reduce the size of CSS sent to the UI during
 streaming

---
 modules/html_generator.py | 29 +++++++++++++++++++++++++++++
 1 file changed, 29 insertions(+)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 1f12d77a5c..e3550ed5a4 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -15,6 +15,29 @@
 # This is to store the paths to the thumbnails of the profile pictures
 image_cache = {}
 
+
+def minify_css(css: str) -> str:
+    # Step 1: Remove comments
+    css = re.sub(r'/\*.*?\*/', '', css, flags=re.DOTALL)
+
+    # Step 2: Remove leading and trailing whitespace
+    css = re.sub(r'^[ \t]*|[ \t]*$', '', css, flags=re.MULTILINE)
+
+    # Step 3: Remove spaces after specific characters ({ : ; ,})
+    css = re.sub(r'([:{;,])\s+', r'\1', css)
+
+    # Step 4: Remove spaces before `{`
+    css = re.sub(r'\s+{', '{', css)
+
+    # Step 5: Remove empty lines
+    css = re.sub(r'^\s*$', '', css, flags=re.MULTILINE)
+
+    # Step 6: Collapse all lines into one
+    css = re.sub(r'\n', '', css)
+
+    return css
+
+
 with open(Path(__file__).resolve().parent / '../css/html_readable_style.css', 'r') as f:
     readable_css = f.read()
 with open(Path(__file__).resolve().parent / '../css/html_instruct_style.css', 'r') as f:
@@ -35,6 +58,12 @@
         style = match.group(1)
         chat_styles[k] = chat_styles.get(style, '') + '\n\n' + '\n'.join(lines[1:])
 
+# Reduce the size of the CSS sources above
+readable_css = minify_css(readable_css)
+instruct_css = minify_css(instruct_css)
+for k in chat_styles:
+    chat_styles[k] = minify_css(chat_styles[k])
+
 
 def fix_newlines(string):
     string = string.replace('\n', '\n\n')

From d56b5005689d17472626872c6fd40fc4208dff30 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Jan 2025 16:22:40 -0800
Subject: [PATCH 0144/1701] UI: add padding to file saving dialog

---
 css/main.css | 1 +
 1 file changed, 1 insertion(+)

diff --git a/css/main.css b/css/main.css
index 0ccaf1062e..dc89fbef49 100644
--- a/css/main.css
+++ b/css/main.css
@@ -226,6 +226,7 @@ button {
     max-width: 500px;
     background-color: var(--input-background-fill);
     border: var(--input-border-width) solid var(--input-border-color) !important;
+    padding: 10px;
 }
 
 .file-saver > :first-child > :last-child {

From 3967520e71ba0ab386893d7c7e946fd621e25b06 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Jan 2025 16:22:59 -0800
Subject: [PATCH 0145/1701] Connect XTC, DRY, smoothing_factor, and dynatemp to
 ExLlamaV2 loader (non-HF)

---
 modules/exllamav2.py      | 30 +++++++++++++++++++++++++++++-
 modules/loaders.py        | 12 +++++++++++-
 modules/sampler_hijack.py |  4 +++-
 3 files changed, 43 insertions(+), 3 deletions(-)

diff --git a/modules/exllamav2.py b/modules/exllamav2.py
index 9b6da83c87..0289bb2109 100644
--- a/modules/exllamav2.py
+++ b/modules/exllamav2.py
@@ -1,8 +1,8 @@
+import json
 import traceback
 from pathlib import Path
 
 import torch
-
 from exllamav2 import (
     ExLlamaV2,
     ExLlamaV2Cache,
@@ -15,6 +15,7 @@
     ExLlamaV2Tokenizer
 )
 from exllamav2.generator import ExLlamaV2Sampler, ExLlamaV2StreamingGenerator
+
 from modules import shared
 from modules.logging_colors import logger
 from modules.text_generation import get_max_prompt_length
@@ -122,6 +123,10 @@ def generate_with_streaming(self, prompt, state):
         settings.token_presence_penalty = state['presence_penalty']
 
         settings.temperature = state['temperature']
+        settings.smoothing_factor = state['smoothing_factor']
+        settings.min_temp = state['dynatemp_low'] if state['dynamic_temperature'] else 0
+        settings.max_temp = state['dynatemp_high'] if state['dynamic_temperature'] else 0
+        settings.temp_exponent = state['dynatemp_exponent']
         settings.top_k = state['top_k']
         settings.top_p = state['top_p']
         settings.top_a = state['top_a']
@@ -143,6 +148,29 @@ def generate_with_streaming(self, prompt, state):
             if len(to_ban) > 0:
                 settings.disallow_tokens(self.tokenizer, to_ban)
 
+        settings.dry_allowed_length = state['dry_allowed_length']
+        settings.dry_base = state['dry_base']
+        settings.dry_multiplier = state['dry_multiplier']
+
+        # Dry sequence breakers processing
+        if state['dry_multiplier'] > 0 and state['dry_sequence_breakers']:
+            dry_sequence_breakers = state['dry_sequence_breakers']
+
+            # Support both JSON array notation and comma-separated strings.
+            if not dry_sequence_breakers.startswith("["):
+                dry_sequence_breakers = "[" + dry_sequence_breakers + "]"
+
+            sequence_breaker_strings = json.loads(dry_sequence_breakers)
+            # Prefix with 'a' to get the correct encoding of the token at the end of a text.
+            sequence_breakers = {
+                self.encode(f"a{s}")[0, -1].item() for s in sequence_breaker_strings
+            }
+
+            settings.dry_sequence_breakers = sequence_breakers
+
+        settings.xtc_probability = state['xtc_probability']
+        settings.xtc_threshold = state['xtc_threshold']
+
         ids = self.tokenizer.encode(prompt, add_bos=state['add_bos_token'], encode_special_tokens=True)
         ids = ids[:, -get_max_prompt_length(state):]
 
diff --git a/modules/loaders.py b/modules/loaders.py
index 1cfdb31b25..a4edf8225f 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -194,6 +194,10 @@ def transformers_samplers():
     'ExLlamav2': {
         'temperature',
         'temperature_last',
+        'smoothing_factor',
+        'dynatemp_low',
+        'dynatemp_high',
+        'dynatemp_exponent',
         'top_p',
         'min_p',
         'top_k',
@@ -204,10 +208,16 @@ def transformers_samplers():
         'presence_penalty',
         'frequency_penalty',
         'repetition_penalty_range',
-        'seed',
         'mirostat_mode',
         'mirostat_tau',
         'mirostat_eta',
+        'dry_multiplier',
+        'dry_base',
+        'dry_allowed_length',
+        'dry_sequence_breakers',
+        'xtc_threshold',
+        'xtc_probability',
+        'seed',
         'ban_eos_token',
         'add_bos_token',
         'custom_token_bans',
diff --git a/modules/sampler_hijack.py b/modules/sampler_hijack.py
index 62ceca8d15..d202af1f2c 100644
--- a/modules/sampler_hijack.py
+++ b/modules/sampler_hijack.py
@@ -495,7 +495,9 @@ def get_logits_processor_patch(self, **kwargs):
 
         sequence_breaker_strings = json.loads(dry_sequence_breakers)
         # Prefix with 'a' to get the correct encoding of the token at the end of a text.
-        sequence_breakers = {shared.tokenizer.encode(f'a{s}')[-1] for s in sequence_breaker_strings}
+        sequence_breakers = {
+            shared.tokenizer.encode(f'a{s}')[-1] for s in sequence_breaker_strings
+        }
 
         warpers.append(
             DRYLogitsProcessor(

From 11af199aff41d0863a06d3407a22c0a13fb2709b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Jan 2025 17:52:57 -0800
Subject: [PATCH 0146/1701] Add a "Static KV cache" option for transformers

---
 extensions/openai/typing.py | 1 +
 modules/loaders.py          | 3 ++-
 modules/shared.py           | 1 +
 modules/text_generation.py  | 3 +++
 modules/ui.py               | 1 +
 modules/ui_parameters.py    | 1 +
 settings-template.yaml      | 1 +
 7 files changed, 10 insertions(+), 1 deletion(-)

diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index f63c1f3911..dfac8e03bc 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -42,6 +42,7 @@ class GenerationOptions(BaseModel):
     truncation_length: int = 0
     max_tokens_second: int = 0
     prompt_lookup_num_tokens: int = 0
+    static_cache: bool = False
     custom_token_bans: str = ""
     sampler_priority: List[str] | str | None = Field(default=None, description="List of samplers where the first items will appear first in the stack. Example: [\"top_k\", \"temperature\", \"top_p\"].")
     auto_max_new_tokens: bool = False
diff --git a/modules/loaders.py b/modules/loaders.py
index a4edf8225f..e1a41bb1fe 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -183,7 +183,8 @@ def transformers_samplers():
         'add_bos_token',
         'skip_special_tokens',
         'auto_max_new_tokens',
-        'prompt_lookup_num_tokens'
+        'prompt_lookup_num_tokens',
+        'static_cache',
     }
 
 
diff --git a/modules/shared.py b/modules/shared.py
index cab612268a..f2ae05a641 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -46,6 +46,7 @@
     'max_tokens_second': 0,
     'max_updates_second': 0,
     'prompt_lookup_num_tokens': 0,
+    'static_cache': False,
     'custom_stopping_strings': '',
     'custom_token_bans': '',
     'auto_max_new_tokens': False,
diff --git a/modules/text_generation.py b/modules/text_generation.py
index db415dce7c..3e9788b81a 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -302,6 +302,9 @@ def generate_reply_HF(question, original_question, seed, state, stopping_strings
     if state['prompt_lookup_num_tokens'] > 0:
         generate_params['prompt_lookup_num_tokens'] = state['prompt_lookup_num_tokens']
 
+    if state['static_cache']:
+        generate_params['cache_implementation'] = 'static'
+
     for k in ['epsilon_cutoff', 'eta_cutoff']:
         if state[k] > 0:
             generate_params[k] = state[k] * 1e-4
diff --git a/modules/ui.py b/modules/ui.py
index a3bf520f3e..3c75f6cac9 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -220,6 +220,7 @@ def list_interface_input_elements():
         'custom_stopping_strings',
         'skip_special_tokens',
         'stream',
+        'static_cache',
         'tfs',
         'top_a',
     ]
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 727a152860..f22f6233d1 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -83,6 +83,7 @@ def create_ui(default_preset):
                             shared.gradio['add_bos_token'] = gr.Checkbox(value=shared.settings['add_bos_token'], label='Add the bos_token to the beginning of prompts', info='Disabling this can make the replies more creative.')
                             shared.gradio['skip_special_tokens'] = gr.Checkbox(value=shared.settings['skip_special_tokens'], label='Skip special tokens', info='Some specific models need this unset.')
                             shared.gradio['stream'] = gr.Checkbox(value=shared.settings['stream'], label='Activate text streaming')
+                            shared.gradio['static_cache'] = gr.Checkbox(value=shared.settings['static_cache'], label='Static KV cache')
 
                         with gr.Column():
                             shared.gradio['truncation_length'] = gr.Number(precision=0, step=256, value=get_truncation_length(), label='Truncate the prompt up to this length', info='The leftmost tokens are removed if the prompt exceeds this length. Most models require this to be at most 2048.')
diff --git a/settings-template.yaml b/settings-template.yaml
index 59c76c350b..d5ed47c392 100644
--- a/settings-template.yaml
+++ b/settings-template.yaml
@@ -22,6 +22,7 @@ ban_eos_token: false
 add_bos_token: true
 skip_special_tokens: true
 stream: true
+static_cache: false
 character: Assistant
 name1: You
 custom_system_message: ''

From c0f600c8879fb28c5b84dd74bc6384619bfcf4fa Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 5 Jan 2025 05:45:12 -0800
Subject: [PATCH 0147/1701] Add a --torch-compile flag for transformers

---
 modules/loaders.py       | 3 ++-
 modules/models.py        | 3 +++
 modules/shared.py        | 1 +
 modules/ui.py            | 3 ++-
 modules/ui_model_menu.py | 1 +
 5 files changed, 9 insertions(+), 2 deletions(-)

diff --git a/modules/loaders.py b/modules/loaders.py
index e1a41bb1fe..191126b34f 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -9,12 +9,13 @@
     'Transformers': [
         'cpu_memory',
         'gpu_memory',
+        'load_in_4bit',
         'load_in_8bit',
+        'torch_compile',
         'bf16',
         'cpu',
         'disk',
         'auto_devices',
-        'load_in_4bit',
         'use_double_quant',
         'quant_type',
         'compute_dtype',
diff --git a/modules/models.py b/modules/models.py
index d906535b5c..cb1ba218f7 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -254,6 +254,9 @@ def huggingface_loader(model_name):
         print()
         model = LoaderClass.from_pretrained(path_to_model, **params)
 
+    if shared.args.torch_compile:
+        model = torch.compile(model)
+
     return model
 
 
diff --git a/modules/shared.py b/modules/shared.py
index f2ae05a641..891c055683 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -104,6 +104,7 @@
 group.add_argument('--no_use_fast', action='store_true', help='Set use_fast=False while loading the tokenizer (it\'s True by default). Use this if you have any problems related to use_fast.')
 group.add_argument('--use_flash_attention_2', action='store_true', help='Set use_flash_attention_2=True while loading the model.')
 group.add_argument('--use_eager_attention', action='store_true', help='Set attn_implementation= eager while loading the model.')
+group.add_argument('--torch-compile', action='store_true', help='Compile the model with torch.compile for improved performance.')
 
 # bitsandbytes 4-bit
 group = parser.add_argument_group('bitsandbytes 4-bit')
diff --git a/modules/ui.py b/modules/ui.py
index 3c75f6cac9..30d4163c3e 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -109,12 +109,13 @@ def list_model_elements():
         'disk',
         'cpu',
         'bf16',
+        'load_in_4bit',
         'load_in_8bit',
+        'torch_compile',
         'trust_remote_code',
         'no_use_fast',
         'use_flash_attention_2',
         'use_eager_attention',
-        'load_in_4bit',
         'compute_dtype',
         'quant_type',
         'use_double_quant',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index c4bb8f0133..f281440183 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -108,6 +108,7 @@ def create_ui():
                             shared.gradio['tensorcores'] = gr.Checkbox(label="tensorcores", value=shared.args.tensorcores, info='NVIDIA only: use llama-cpp-python compiled with tensor cores support. This may increase performance on newer cards.')
                             shared.gradio['load_in_8bit'] = gr.Checkbox(label="load-in-8bit", value=shared.args.load_in_8bit)
                             shared.gradio['load_in_4bit'] = gr.Checkbox(label="load-in-4bit", value=shared.args.load_in_4bit)
+                            shared.gradio['torch_compile'] = gr.Checkbox(label="torch-compile", value=shared.args.torch_compile, info='Compile the model with torch.compile for improved performance.')
                             shared.gradio['flash_attn'] = gr.Checkbox(label="flash_attn", value=shared.args.flash_attn, info='Use flash-attention.')
                             shared.gradio['use_flash_attention_2'] = gr.Checkbox(label="use_flash_attention_2", value=shared.args.use_flash_attention_2, info='Set use_flash_attention_2=True while loading the model.')
                             shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming_llm", value=shared.args.streaming_llm, info='(experimental) Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')

From 03a0f236a483d2b05094cc6e5546cfa28bd1668b Mon Sep 17 00:00:00 2001
From: FP HAM 
Date: Wed, 8 Jan 2025 14:54:09 -0500
Subject: [PATCH 0148/1701] Training_PRO fix: add `if 'quantization_config' in
 shared.model.config.to_dict()`

---
 extensions/Training_PRO/script.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/extensions/Training_PRO/script.py b/extensions/Training_PRO/script.py
index 5365154c35..01bcf67d4c 100644
--- a/extensions/Training_PRO/script.py
+++ b/extensions/Training_PRO/script.py
@@ -789,7 +789,11 @@ def generate_and_tokenize_prompt(data_point):
     if not hasattr(shared.model, 'lm_head') or hasattr(shared.model.lm_head, 'weight'):
         logger.info("Getting model ready...")
         # here we can disable gradient checkpoint, by default = true,  use_gradient_checkpointing=True
-        prepare_model_for_kbit_training(shared.model)
+        if 'quantization_config' in shared.model.config.to_dict():
+            print(f"Method: {RED}QLORA{RESET}")
+            prepare_model_for_kbit_training(shared.model)
+        else:
+            print(f"Method: {RED}LoRA{RESET}")    
 
     # base model is now frozen and should not be reused for any other LoRA training than this one
     shared.model_dirty_from_training = True

From 1f867229770b902939cd319a9950802b42c4a104 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Wed, 8 Jan 2025 16:56:55 -0300
Subject: [PATCH 0149/1701] Update safetensors requirement from ==0.4.* to
 ==0.5.* (#6634)

---
 requirements.txt                 | 2 +-
 requirements_amd.txt             | 2 +-
 requirements_amd_noavx2.txt      | 2 +-
 requirements_apple_intel.txt     | 2 +-
 requirements_apple_silicon.txt   | 2 +-
 requirements_cpu_only.txt        | 2 +-
 requirements_cpu_only_noavx2.txt | 2 +-
 requirements_noavx2.txt          | 2 +-
 requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 8b36f54f1d..fb8ca6b3e3 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -17,7 +17,7 @@ pydantic==2.8.2
 pyyaml
 requests
 rich
-safetensors==0.4.*
+safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 1b85b20fdf..88b4b887f8 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -16,7 +16,7 @@ pydantic==2.8.2
 pyyaml
 requests
 rich
-safetensors==0.4.*
+safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 1bafc1419f..ef43da993f 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -16,7 +16,7 @@ pydantic==2.8.2
 pyyaml
 requests
 rich
-safetensors==0.4.*
+safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 345368b79d..369d34d881 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -16,7 +16,7 @@ pydantic==2.8.2
 pyyaml
 requests
 rich
-safetensors==0.4.*
+safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index a3c3055bee..9eefd8b890 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -16,7 +16,7 @@ pydantic==2.8.2
 pyyaml
 requests
 rich
-safetensors==0.4.*
+safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index a4b7882cc4..f5313ecffd 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -16,7 +16,7 @@ pydantic==2.8.2
 pyyaml
 requests
 rich
-safetensors==0.4.*
+safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 878aea0699..aefe31d022 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -16,7 +16,7 @@ pydantic==2.8.2
 pyyaml
 requests
 rich
-safetensors==0.4.*
+safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 1a3456113a..9e1633ca85 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -17,7 +17,7 @@ pydantic==2.8.2
 pyyaml
 requests
 rich
-safetensors==0.4.*
+safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index 95fe3add4f..45003f0db5 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -16,7 +16,7 @@ pydantic==2.8.2
 pyyaml
 requests
 rich
-safetensors==0.4.*
+safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard

From d3adcbf64b782ee04262af1fca8eb1a837a1ea67 Mon Sep 17 00:00:00 2001
From: Jack Cloudman 
Date: Wed, 8 Jan 2025 14:30:21 -0600
Subject: [PATCH 0150/1701] Add `--exclude-pattern` flag to download-model.py
 script (#6542)

---
 download-model.py | 22 +++++++++++++++-------
 1 file changed, 15 insertions(+), 7 deletions(-)

diff --git a/download-model.py b/download-model.py
index 306784a355..8fe94371f2 100644
--- a/download-model.py
+++ b/download-model.py
@@ -72,7 +72,7 @@ def sanitize_model_and_branch_names(self, model, branch):
 
         return model, branch
 
-    def get_download_links_from_huggingface(self, model, branch, text_only=False, specific_file=None):
+    def get_download_links_from_huggingface(self, model, branch, text_only=False, specific_file=None, exclude_pattern=None):
         session = self.session
         page = f"/api/models/{model}/tree/{branch}"
         cursor = b""
@@ -100,13 +100,17 @@ def get_download_links_from_huggingface(self, model, branch, text_only=False, sp
                 if specific_file not in [None, ''] and fname != specific_file:
                     continue
 
+                # Exclude files matching the exclude pattern
+                if exclude_pattern is not None and re.match(exclude_pattern, fname):
+                    continue
+
                 if not is_lora and fname.endswith(('adapter_config.json', 'adapter_model.bin')):
                     is_lora = True
 
                 is_pytorch = re.match(r"(pytorch|adapter|gptq)_model.*\.bin", fname)
                 is_safetensors = re.match(r".*\.safetensors", fname)
                 is_pt = re.match(r".*\.pt", fname)
-                is_gguf = re.match(r'.*\.gguf', fname)
+                is_gguf = re.match(r".*\.gguf", fname)
                 is_tiktoken = re.match(r".*\.tiktoken", fname)
                 is_tokenizer = re.match(r"(tokenizer|ice|spiece).*\.model", fname) or is_tiktoken
                 is_text = re.match(r".*\.(txt|json|py|md)", fname) or is_tokenizer
@@ -140,7 +144,6 @@ def get_download_links_from_huggingface(self, model, branch, text_only=False, sp
 
         # If both pytorch and safetensors are available, download safetensors only
         # Also if GGUF and safetensors are available, download only safetensors
-        # (why do people do this?)
         if (has_pytorch or has_pt or has_gguf) and has_safetensors:
             has_gguf = False
             for i in range(len(classifications) - 1, -1, -1):
@@ -148,8 +151,6 @@ def get_download_links_from_huggingface(self, model, branch, text_only=False, sp
                     links.pop(i)
 
         # For GGUF, try to download only the Q4_K_M if no specific file is specified.
-        # If not present, exclude all GGUFs, as that's likely a repository with both
-        # GGUF and fp16 files.
         if has_gguf and specific_file is None:
             has_q4km = False
             for i in range(len(classifications) - 1, -1, -1):
@@ -312,6 +313,7 @@ def check_model_files(self, model, branch, links, sha256, output_folder):
     parser.add_argument('--threads', type=int, default=4, help='Number of files to download simultaneously.')
     parser.add_argument('--text-only', action='store_true', help='Only download text files (txt/json).')
     parser.add_argument('--specific-file', type=str, default=None, help='Name of the specific file to download (if not provided, downloads all).')
+    parser.add_argument('--exclude-pattern', type=str, default=None, help='Regex pattern to exclude files from download.')
     parser.add_argument('--output', type=str, default=None, help='Save the model files to this folder.')
     parser.add_argument('--model-dir', type=str, default=None, help='Save the model files to a subfolder of this folder instead of the default one (text-generation-webui/models).')
     parser.add_argument('--clean', action='store_true', help='Does not resume the previous download.')
@@ -322,6 +324,7 @@ def check_model_files(self, model, branch, links, sha256, output_folder):
     branch = args.branch
     model = args.MODEL
     specific_file = args.specific_file
+    exclude_pattern = args.exclude_pattern
 
     if model is None:
         print("Error: Please specify the model you'd like to download (e.g. 'python download-model.py facebook/opt-1.3b').")
@@ -336,7 +339,9 @@ def check_model_files(self, model, branch, links, sha256, output_folder):
         sys.exit()
 
     # Get the download links from Hugging Face
-    links, sha256, is_lora, is_llamacpp = downloader.get_download_links_from_huggingface(model, branch, text_only=args.text_only, specific_file=specific_file)
+    links, sha256, is_lora, is_llamacpp = downloader.get_download_links_from_huggingface(
+        model, branch, text_only=args.text_only, specific_file=specific_file, exclude_pattern=exclude_pattern
+    )
 
     # Get the output folder
     if args.output:
@@ -349,4 +354,7 @@ def check_model_files(self, model, branch, links, sha256, output_folder):
         downloader.check_model_files(model, branch, links, sha256, output_folder)
     else:
         # Download files
-        downloader.download_model_files(model, branch, links, sha256, output_folder, specific_file=specific_file, threads=args.threads, is_llamacpp=is_llamacpp)
+        downloader.download_model_files(
+            model, branch, links, sha256, output_folder,
+            specific_file=specific_file, threads=args.threads, is_llamacpp=is_llamacpp
+        )

From 7157257c3f4691b9e4b56ceddb8c428802ac4c54 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Wed, 8 Jan 2025 19:28:56 -0300
Subject: [PATCH 0151/1701] Remove the AutoGPTQ loader (#6641)

---
 README.md                  |  2 +-
 modules/AutoGPTQ_loader.py | 74 --------------------------------------
 modules/LoRA.py            | 38 ++------------------
 modules/loaders.py         | 21 -----------
 modules/models.py          | 29 +--------------
 modules/models_settings.py | 41 +++------------------
 modules/shared.py          | 23 +++++-------
 modules/ui.py              |  8 -----
 modules/ui_model_menu.py   |  9 -----
 one_click.py               |  2 +-
 10 files changed, 19 insertions(+), 228 deletions(-)
 delete mode 100644 modules/AutoGPTQ_loader.py

diff --git a/README.md b/README.md
index 0e16aa30a5..bec686ad46 100644
--- a/README.md
+++ b/README.md
@@ -10,7 +10,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 
 ## Features
 
-- Supports multiple text generation backends in one UI/API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp), and [ExLlamaV2](https://github.com/turboderp/exllamav2). [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM) are also supported but you need to install them manually.
+- Supports multiple text generation backends in a single UI/API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp), and [ExLlamaV2](https://github.com/turboderp-org/exllamav2). [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) is supported via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile), and the Transformers loader is compatible with libraries like [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM), but they must be installed manually.
 - OpenAI-compatible API with Chat and Completions endpoints – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples).
 - Automatic prompt formatting using Jinja2 templates.
 - Three chat modes: `instruct`, `chat-instruct`, and `chat`, with automatic prompt templates in `chat-instruct`.
diff --git a/modules/AutoGPTQ_loader.py b/modules/AutoGPTQ_loader.py
deleted file mode 100644
index 69e8f299cb..0000000000
--- a/modules/AutoGPTQ_loader.py
+++ /dev/null
@@ -1,74 +0,0 @@
-from pathlib import Path
-
-from accelerate.utils import is_xpu_available
-from auto_gptq import AutoGPTQForCausalLM, BaseQuantizeConfig
-
-import modules.shared as shared
-from modules.logging_colors import logger
-from modules.models import get_max_memory_dict
-
-
-def load_quantized(model_name):
-    path_to_model = Path(f'{shared.args.model_dir}/{model_name}')
-    pt_path = None
-
-    # Find the model checkpoint
-    if shared.args.checkpoint:
-        pt_path = Path(shared.args.checkpoint)
-    else:
-        for ext in ['.safetensors', '.pt', '.bin']:
-            found = list(path_to_model.glob(f"*{ext}"))
-            if len(found) > 0:
-                if len(found) > 1:
-                    logger.warning(f'More than one {ext} model has been found. The last one will be selected. It could be wrong.')
-
-                pt_path = found[-1]
-                break
-
-    if pt_path is None:
-        logger.error("The model could not be loaded because its checkpoint file in .bin/.pt/.safetensors format could not be located.")
-        return
-
-    use_safetensors = pt_path.suffix == '.safetensors'
-    if not (path_to_model / "quantize_config.json").exists():
-        quantize_config = BaseQuantizeConfig(
-            bits=bits if (bits := shared.args.wbits) > 0 else 4,
-            group_size=gs if (gs := shared.args.groupsize) > 0 else -1,
-            desc_act=shared.args.desc_act
-        )
-    else:
-        quantize_config = None
-
-    # Define the params for AutoGPTQForCausalLM.from_quantized
-    params = {
-        'model_basename': pt_path.stem,
-        'device': "xpu:0" if is_xpu_available() else "cuda:0" if not shared.args.cpu else "cpu",
-        'use_triton': shared.args.triton,
-        'inject_fused_attention': False,
-        'inject_fused_mlp': not shared.args.no_inject_fused_mlp,
-        'use_safetensors': use_safetensors,
-        'trust_remote_code': shared.args.trust_remote_code,
-        'max_memory': get_max_memory_dict(),
-        'quantize_config': quantize_config,
-        'use_cuda_fp16': not shared.args.no_use_cuda_fp16,
-        'disable_exllama': shared.args.disable_exllama,
-        'disable_exllamav2': shared.args.disable_exllamav2,
-    }
-
-    logger.info(f"The AutoGPTQ params are: {params}")
-    model = AutoGPTQForCausalLM.from_quantized(path_to_model, **params)
-
-    # These lines fix the multimodal extension when used with AutoGPTQ
-    if hasattr(model, 'model'):
-        if not hasattr(model, 'dtype'):
-            if hasattr(model.model, 'dtype'):
-                model.dtype = model.model.dtype
-
-        if hasattr(model.model, 'model') and hasattr(model.model.model, 'embed_tokens'):
-            if not hasattr(model, 'embed_tokens'):
-                model.embed_tokens = model.model.model.embed_tokens
-
-            if not hasattr(model.model, 'embed_tokens'):
-                model.model.embed_tokens = model.model.model.embed_tokens
-
-    return model
diff --git a/modules/LoRA.py b/modules/LoRA.py
index e1ad01d769..1f4883e25b 100644
--- a/modules/LoRA.py
+++ b/modules/LoRA.py
@@ -2,13 +2,11 @@
 
 import modules.shared as shared
 from modules.logging_colors import logger
-from modules.models import get_device, reload_model
+from modules.models import get_device
 
 
 def add_lora_to_model(lora_names):
-    if 'GPTQForCausalLM' in shared.model.__class__.__name__ or shared.args.loader == 'AutoGPTQ':
-        add_lora_autogptq(lora_names)
-    elif shared.model.__class__.__name__ in ['Exllamav2Model', 'Exllamav2HF'] or shared.args.loader in ['ExLlamav2', 'ExLlamav2_HF']:
+    if shared.model.__class__.__name__ in ['Exllamav2Model', 'Exllamav2HF'] or shared.args.loader in ['ExLlamav2', 'ExLlamav2_HF']:
         add_lora_exllamav2(lora_names)
     else:
         add_lora_transformers(lora_names)
@@ -48,38 +46,6 @@ def add_lora_exllamav2(lora_names):
         shared.model.loras = None
 
 
-def add_lora_autogptq(lora_names):
-    '''
-    Adapted from https://github.com/Ph0rk0z/text-generation-webui-testing
-    '''
-
-    try:
-        from auto_gptq import get_gptq_peft_model
-        from auto_gptq.utils.peft_utils import GPTQLoraConfig
-    except:
-        logger.error("This version of AutoGPTQ does not support LoRA. You need to install from source or wait for a new release.")
-        return
-
-    if len(lora_names) == 0:
-        reload_model()
-
-        shared.lora_names = []
-        return
-    else:
-        if len(lora_names) > 1:
-            logger.warning('AutoGPTQ can only work with 1 LoRA at the moment. Only the first one in the list will be loaded.')
-
-        peft_config = GPTQLoraConfig(
-            inference_mode=True,
-        )
-
-        lora_path = get_lora_path(lora_names[0])
-        logger.info("Applying the following LoRAs to {}: {}".format(shared.model_name, ', '.join([lora_names[0]])))
-        shared.model = get_gptq_peft_model(shared.model, peft_config, lora_path)
-        shared.lora_names = [lora_names[0]]
-        return
-
-
 def add_lora_transformers(lora_names):
 
     from peft import PeftModel
diff --git a/modules/loaders.py b/modules/loaders.py
index 191126b34f..4e331dbbfd 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -25,8 +25,6 @@
         'use_eager_attention',
         'alpha_value',
         'compress_pos_emb',
-        'disable_exllama',
-        'disable_exllamav2',
     ],
     'llama.cpp': [
         'n_ctx',
@@ -107,24 +105,6 @@
         'compress_pos_emb',
         'exllamav2_info',
     ],
-    'AutoGPTQ': [
-        'triton',
-        'no_inject_fused_mlp',
-        'no_use_cuda_fp16',
-        'wbits',
-        'groupsize',
-        'desc_act',
-        'disable_exllama',
-        'disable_exllamav2',
-        'gpu_memory',
-        'cpu_memory',
-        'cpu',
-        'disk',
-        'auto_devices',
-        'trust_remote_code',
-        'no_use_fast',
-        'autogptq_info',
-    ],
     'HQQ': [
         'hqq_backend',
         'trust_remote_code',
@@ -191,7 +171,6 @@ def transformers_samplers():
 
 loaders_samplers = {
     'Transformers': transformers_samplers(),
-    'AutoGPTQ': transformers_samplers(),
     'HQQ': transformers_samplers(),
     'ExLlamav2': {
         'temperature',
diff --git a/modules/models.py b/modules/models.py
index cb1ba218f7..9c58b2799c 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -3,7 +3,6 @@
 import pprint
 import re
 import time
-import traceback
 from pathlib import Path
 
 import torch
@@ -21,7 +20,6 @@
     AutoModelForSeq2SeqLM,
     AutoTokenizer,
     BitsAndBytesConfig,
-    GPTQConfig,
     is_torch_npu_available,
     is_torch_xpu_available
 )
@@ -73,7 +71,6 @@ def load_model(model_name, loader=None):
         'llamacpp_HF': llamacpp_HF_loader,
         'ExLlamav2': ExLlamav2_loader,
         'ExLlamav2_HF': ExLlamav2_HF_loader,
-        'AutoGPTQ': AutoGPTQ_loader,
         'HQQ': HQQ_loader,
         'TensorRT-LLM': TensorRT_LLM_loader,
     }
@@ -164,7 +161,7 @@ def huggingface_loader(model_name):
             LoaderClass = AutoModelForCausalLM
 
     # Load the model without any special settings
-    if not any([shared.args.cpu, shared.args.load_in_8bit, shared.args.load_in_4bit, shared.args.auto_devices, shared.args.disk, shared.args.deepspeed, shared.args.gpu_memory is not None, shared.args.cpu_memory is not None, shared.args.compress_pos_emb > 1, shared.args.alpha_value > 1, shared.args.disable_exllama, shared.args.disable_exllamav2]):
+    if not any([shared.args.cpu, shared.args.load_in_8bit, shared.args.load_in_4bit, shared.args.auto_devices, shared.args.disk, shared.args.deepspeed, shared.args.gpu_memory is not None, shared.args.cpu_memory is not None, shared.args.compress_pos_emb > 1, shared.args.alpha_value > 1]):
         logger.info("TRANSFORMERS_PARAMS=")
         pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(params)
         print()
@@ -229,21 +226,6 @@ def huggingface_loader(model_name):
             if shared.args.disk:
                 params['offload_folder'] = shared.args.disk_cache_dir
 
-        if shared.args.disable_exllama or shared.args.disable_exllamav2:
-            try:
-                gptq_config = GPTQConfig(
-                    bits=config.quantization_config.get('bits', 4),
-                    disable_exllama=shared.args.disable_exllama,
-                    disable_exllamav2=shared.args.disable_exllamav2,
-                )
-
-                params['quantization_config'] = gptq_config
-                logger.info(f'Loading with disable_exllama={shared.args.disable_exllama} and disable_exllamav2={shared.args.disable_exllamav2}.')
-            except:
-                exc = traceback.format_exc()
-                logger.error('Failed to disable exllama. Does the config.json for this model contain the necessary quantization info?')
-                print(exc)
-
         if shared.args.compress_pos_emb > 1:
             params['rope_scaling'] = {'type': 'linear', 'factor': shared.args.compress_pos_emb}
         elif shared.args.alpha_value > 1:
@@ -310,15 +292,6 @@ def ExLlamav2_HF_loader(model_name):
     return Exllamav2HF.from_pretrained(model_name)
 
 
-def AutoGPTQ_loader(model_name):
-    try:
-        import modules.AutoGPTQ_loader
-    except ModuleNotFoundError:
-        raise ModuleNotFoundError("Failed to import 'autogptq'. Please install it manually following the instructions in the AutoGPTQ GitHub repository.")
-
-    return modules.AutoGPTQ_loader.load_quantized(model_name)
-
-
 def HQQ_loader(model_name):
     try:
         from hqq.core.quantize import HQQBackend, HQQLinear
diff --git a/modules/models_settings.py b/modules/models_settings.py
index 1bb00ceb6a..8d65852345 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -11,9 +11,6 @@ def get_fallback_settings():
     return {
         'bf16': False,
         'use_eager_attention': False,
-        'wbits': 'None',
-        'groupsize': 'None',
-        'desc_act': False,
         'max_seq_len': 2048,
         'n_ctx': 2048,
         'rope_freq_base': 0,
@@ -111,26 +108,6 @@ def get_model_metadata(model):
             if 'architectures' in metadata and isinstance(metadata['architectures'], list) and 'Gemma2ForCausalLM' in metadata['architectures']:
                 model_settings['use_eager_attention'] = True
 
-            # Read GPTQ metadata for old GPTQ loaders
-            if 'quantization_config' in metadata and metadata['quantization_config'].get('quant_method', '') != 'exl2':
-                if 'bits' in metadata['quantization_config']:
-                    model_settings['wbits'] = metadata['quantization_config']['bits']
-                if 'group_size' in metadata['quantization_config']:
-                    model_settings['groupsize'] = metadata['quantization_config']['group_size']
-                if 'desc_act' in metadata['quantization_config']:
-                    model_settings['desc_act'] = metadata['quantization_config']['desc_act']
-
-        # Read AutoGPTQ metadata
-        path = Path(f'{shared.args.model_dir}/{model}/quantize_config.json')
-        if path.exists():
-            metadata = json.loads(open(path, 'r', encoding='utf-8').read())
-            if 'bits' in metadata:
-                model_settings['wbits'] = metadata['bits']
-            if 'group_size' in metadata:
-                model_settings['groupsize'] = metadata['group_size']
-            if 'desc_act' in metadata:
-                model_settings['desc_act'] = metadata['desc_act']
-
     # Try to find the Jinja instruct template
     path = Path(f'{shared.args.model_dir}/{model}') / 'tokenizer_config.json'
     if path.exists():
@@ -178,7 +155,7 @@ def infer_loader(model_name, model_settings):
     path_to_model = Path(f'{shared.args.model_dir}/{model_name}')
     if not path_to_model.exists():
         loader = None
-    elif (path_to_model / 'quantize_config.json').exists() or ('wbits' in model_settings and isinstance(model_settings['wbits'], int) and model_settings['wbits'] > 0):
+    elif (path_to_model / 'quantize_config.json').exists():  # Old GPTQ metadata file
         loader = 'ExLlamav2_HF'
     elif len(list(path_to_model.glob('*.gguf'))) > 0 and path_to_model.is_dir() and (path_to_model / 'tokenizer_config.json').exists():
         loader = 'llamacpp_HF'
@@ -215,16 +192,11 @@ def update_model_parameters(state, initial=False):
         if initial and element in shared.provided_arguments:
             continue
 
-        # Setting null defaults
-        if element in ['wbits', 'groupsize'] and value == 'None':
-            value = vars(shared.args_defaults)[element]
-        elif element in ['cpu_memory'] and value == 0:
+        if element in ['cpu_memory'] and value == 0:
             value = vars(shared.args_defaults)[element]
 
         # Making some simple conversions
-        if element in ['wbits', 'groupsize']:
-            value = int(value)
-        elif element == 'cpu_memory' and value is not None:
+        if element == 'cpu_memory' and value is not None:
             value = f"{value}MiB"
 
         setattr(shared.args, element, value)
@@ -251,15 +223,12 @@ def apply_model_settings_to_state(model, state):
         loader = model_settings.pop('loader')
 
         # If the user is using an alternative loader for the same model type, let them keep using it
-        if not (loader == 'ExLlamav2_HF' and state['loader'] in ['ExLlamav2', 'AutoGPTQ']):
+        if not (loader == 'ExLlamav2_HF' and state['loader'] in ['ExLlamav2']):
             state['loader'] = loader
 
     for k in model_settings:
         if k in state:
-            if k in ['wbits', 'groupsize']:
-                state[k] = str(model_settings[k])
-            else:
-                state[k] = model_settings[k]
+            state[k] = model_settings[k]
 
     return state
 
diff --git a/modules/shared.py b/modules/shared.py
index 891c055683..6a83baae92 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -86,7 +86,7 @@
 
 # Model loader
 group = parser.add_argument_group('Model loader')
-group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav2_HF, ExLlamav2, AutoGPTQ.')
+group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav2_HF, ExLlamav2.')
 
 # Transformers/Accelerate
 group = parser.add_argument_group('Transformers/Accelerate')
@@ -147,17 +147,6 @@
 group.add_argument('--num_experts_per_token', type=int, default=2, help='Number of experts to use for generation. Applies to MoE models like Mixtral.')
 group.add_argument('--enable_tp', action='store_true', help='Enable Tensor Parallelism (TP) in ExLlamaV2.')
 
-# AutoGPTQ
-group = parser.add_argument_group('AutoGPTQ')
-group.add_argument('--triton', action='store_true', help='Use triton.')
-group.add_argument('--no_inject_fused_mlp', action='store_true', help='Triton mode only: disable the use of fused MLP, which will use less VRAM at the cost of slower inference.')
-group.add_argument('--no_use_cuda_fp16', action='store_true', help='This can make models faster on some systems.')
-group.add_argument('--desc_act', action='store_true', help='For models that do not have a quantize_config.json, this parameter is used to define whether to set desc_act or not in BaseQuantizeConfig.')
-group.add_argument('--disable_exllama', action='store_true', help='Disable ExLlama kernel, which can improve inference speed on some systems.')
-group.add_argument('--disable_exllamav2', action='store_true', help='Disable ExLlamav2 kernel.')
-group.add_argument('--wbits', type=int, default=0, help='Load a pre-quantized model with specified precision in bits. 2, 3, 4 and 8 are supported.')
-group.add_argument('--groupsize', type=int, default=-1, help='Group size.')
-
 # HQQ
 group = parser.add_argument_group('HQQ')
 group.add_argument('--hqq-backend', type=str, default='PYTORCH_COMPILE', help='Backend for the HQQ loader. Valid options: PYTORCH, PYTORCH_COMPILE, ATEN.')
@@ -220,6 +209,14 @@
 group.add_argument('--cache_4bit', action='store_true', help='DEPRECATED')
 group.add_argument('--cache_8bit', action='store_true', help='DEPRECATED')
 group.add_argument('--chat-buttons', action='store_true', help='DEPRECATED')
+group.add_argument('--triton', action='store_true', help='DEPRECATED')
+group.add_argument('--no_inject_fused_mlp', action='store_true', help='DEPRECATED')
+group.add_argument('--no_use_cuda_fp16', action='store_true', help='DEPRECATED')
+group.add_argument('--desc_act', action='store_true', help='DEPRECATED')
+group.add_argument('--disable_exllama', action='store_true', help='DEPRECATED')
+group.add_argument('--disable_exllamav2', action='store_true', help='DEPRECATED')
+group.add_argument('--wbits', type=int, default=0, help='DEPRECATED')
+group.add_argument('--groupsize', type=int, default=-1, help='DEPRECATED')
 
 args = parser.parse_args()
 args_defaults = parser.parse_args([])
@@ -262,8 +259,6 @@ def fix_loader_name(name):
         return 'llamacpp_HF'
     elif name in ['transformers', 'huggingface', 'hf', 'hugging_face', 'hugging face']:
         return 'Transformers'
-    elif name in ['autogptq', 'auto-gptq', 'auto_gptq', 'auto gptq']:
-        return 'AutoGPTQ'
     elif name in ['exllama', 'ex-llama', 'ex_llama', 'exlama']:
         return 'ExLlama'
     elif name in ['exllamav2', 'exllama-v2', 'ex_llama-v2', 'exlamav2', 'exlama-v2', 'exllama2', 'exllama-2']:
diff --git a/modules/ui.py b/modules/ui.py
index 30d4163c3e..e66de434ad 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -119,14 +119,6 @@ def list_model_elements():
         'compute_dtype',
         'quant_type',
         'use_double_quant',
-        'wbits',
-        'groupsize',
-        'triton',
-        'desc_act',
-        'no_inject_fused_mlp',
-        'no_use_cuda_fp16',
-        'disable_exllama',
-        'disable_exllamav2',
         'cfg_cache',
         'no_flash_attn',
         'no_xformers',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index f281440183..eac0cba66a 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -89,8 +89,6 @@ def create_ui():
                             shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
                             shared.gradio['n_batch'] = gr.Slider(label="n_batch", minimum=1, maximum=2048, step=1, value=shared.args.n_batch)
                             shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
-                            shared.gradio['wbits'] = gr.Dropdown(label="wbits", choices=["None", 1, 2, 3, 4, 8], value=shared.args.wbits if shared.args.wbits > 0 else "None")
-                            shared.gradio['groupsize'] = gr.Dropdown(label="groupsize", choices=["None", 32, 64, 128, 1024], value=shared.args.groupsize if shared.args.groupsize > 0 else "None")
                             shared.gradio['n_ctx'] = gr.Number(label="n_ctx", precision=0, step=256, value=shared.args.n_ctx, info='Context length. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['max_seq_len'] = gr.Number(label='max_seq_len', precision=0, step=256, value=shared.args.max_seq_len, info='Context length. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q6', 'q4'], value=shared.args.cache_type, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.')
@@ -121,10 +119,6 @@ def create_ui():
                             shared.gradio['no_mmap'] = gr.Checkbox(label="no-mmap", value=shared.args.no_mmap)
                             shared.gradio['mlock'] = gr.Checkbox(label="mlock", value=shared.args.mlock)
                             shared.gradio['numa'] = gr.Checkbox(label="numa", value=shared.args.numa, info='NUMA support can help on some systems with non-uniform memory access.')
-                            shared.gradio['triton'] = gr.Checkbox(label="triton", value=shared.args.triton)
-                            shared.gradio['no_inject_fused_mlp'] = gr.Checkbox(label="no_inject_fused_mlp", value=shared.args.no_inject_fused_mlp, info='Affects Triton only. Disable fused MLP. Fused MLP improves performance but uses more VRAM. Disable if running low on VRAM.')
-                            shared.gradio['no_use_cuda_fp16'] = gr.Checkbox(label="no_use_cuda_fp16", value=shared.args.no_use_cuda_fp16, info='This can make models faster on some systems.')
-                            shared.gradio['desc_act'] = gr.Checkbox(label="desc_act", value=shared.args.desc_act, info='\'desc_act\', \'wbits\', and \'groupsize\' are used for old models without a quantize_config.json.')
                             shared.gradio['use_double_quant'] = gr.Checkbox(label="use_double_quant", value=shared.args.use_double_quant, info='Used by load-in-4bit.')
                             shared.gradio['use_eager_attention'] = gr.Checkbox(label="use_eager_attention", value=shared.args.use_eager_attention, info='Set attn_implementation= eager while loading the model.')
                             shared.gradio['bf16'] = gr.Checkbox(label="bf16", value=shared.args.bf16)
@@ -136,13 +130,10 @@ def create_ui():
                             shared.gradio['cfg_cache'] = gr.Checkbox(label="cfg-cache", value=shared.args.cfg_cache, info='Necessary to use CFG with this loader.')
                             shared.gradio['cpp_runner'] = gr.Checkbox(label="cpp-runner", value=shared.args.cpp_runner, info='Enable inference with ModelRunnerCpp, which is faster than the default ModelRunner.')
                             shared.gradio['logits_all'] = gr.Checkbox(label="logits_all", value=shared.args.logits_all, info='Needs to be set for perplexity evaluation to work with this loader. Otherwise, ignore it, as it makes prompt processing slower.')
-                            shared.gradio['disable_exllama'] = gr.Checkbox(label="disable_exllama", value=shared.args.disable_exllama, info='Disable ExLlama kernel for GPTQ models.')
-                            shared.gradio['disable_exllamav2'] = gr.Checkbox(label="disable_exllamav2", value=shared.args.disable_exllamav2, info='Disable ExLlamav2 kernel for GPTQ models.')
                             shared.gradio['trust_remote_code'] = gr.Checkbox(label="trust-remote-code", value=shared.args.trust_remote_code, info='Set trust_remote_code=True while loading the tokenizer/model. To enable this option, start the web UI with the --trust-remote-code flag.', interactive=shared.args.trust_remote_code)
                             shared.gradio['no_use_fast'] = gr.Checkbox(label="no_use_fast", value=shared.args.no_use_fast, info='Set use_fast=False while loading the tokenizer.')
                             shared.gradio['llamacpp_HF_info'] = gr.Markdown("llamacpp_HF loads llama.cpp as a Transformers model. To use it, you need to place your GGUF in a subfolder of models/ with the necessary tokenizer files.\n\nYou can use the \"llamacpp_HF creator\" menu to do that automatically.")
                             shared.gradio['exllamav2_info'] = gr.Markdown("ExLlamav2_HF is recommended over ExLlamav2 for better integration with extensions and more consistent sampling behavior across loaders.")
-                            shared.gradio['autogptq_info'] = gr.Markdown('ExLlamav2_HF is recommended over AutoGPTQ for models derived from Llama.')
                             shared.gradio['tensorrt_llm_info'] = gr.Markdown('* TensorRT-LLM has to be installed manually in a separate Python 3.10 environment at the moment. For a guide, consult the description of [this PR](https://github.com/oobabooga/text-generation-webui/pull/5715). \n\n* `max_seq_len` is only used when `cpp-runner` is checked.\n\n* `cpp_runner` does not support streaming at the moment.')
 
             with gr.Column():
diff --git a/one_click.py b/one_click.py
index 8fc1edf0cf..ca11efacd4 100644
--- a/one_click.py
+++ b/one_click.py
@@ -394,7 +394,7 @@ def update_requirements(initial_installation=False, pull=True):
         textgen_requirements = [
             req.replace('+cu121', '+cu118').replace('+cu122', '+cu118')
             for req in textgen_requirements
-            if "auto-gptq" not in req.lower() and "autoawq" not in req.lower()
+            if "autoawq" not in req.lower()
         ]
 
     if is_windows() and is_cuda118:  # No flash-attention on Windows for CUDA 11

From ad118056b87ff03f4892baed34c605851db3a923 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 8 Jan 2025 14:29:46 -0800
Subject: [PATCH 0152/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index bec686ad46..071387726a 100644
--- a/README.md
+++ b/README.md
@@ -10,7 +10,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 
 ## Features
 
-- Supports multiple text generation backends in a single UI/API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp), and [ExLlamaV2](https://github.com/turboderp-org/exllamav2). [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) is supported via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile), and the Transformers loader is compatible with libraries like [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM), but they must be installed manually.
+- Supports multiple text generation backends in one UI/API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp), and [ExLlamaV2](https://github.com/turboderp-org/exllamav2). [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) is supported via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile), and the Transformers loader is compatible with libraries like [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM), but they must be installed manually.
 - OpenAI-compatible API with Chat and Completions endpoints – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples).
 - Automatic prompt formatting using Jinja2 templates.
 - Three chat modes: `instruct`, `chat-instruct`, and `chat`, with automatic prompt templates in `chat-instruct`.

From 91a8a878878e57f221ca3992ec8c425cf3805d88 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 8 Jan 2025 15:07:21 -0800
Subject: [PATCH 0153/1701] Remove obsolete code

---
 modules/shared.py | 2 --
 1 file changed, 2 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 6a83baae92..7829e46200 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -259,8 +259,6 @@ def fix_loader_name(name):
         return 'llamacpp_HF'
     elif name in ['transformers', 'huggingface', 'hf', 'hugging_face', 'hugging face']:
         return 'Transformers'
-    elif name in ['exllama', 'ex-llama', 'ex_llama', 'exlama']:
-        return 'ExLlama'
     elif name in ['exllamav2', 'exllama-v2', 'ex_llama-v2', 'exlamav2', 'exlama-v2', 'exllama2', 'exllama-2']:
         return 'ExLlamav2'
     elif name in ['exllamav2-hf', 'exllamav2_hf', 'exllama-v2-hf', 'exllama_v2_hf', 'exllama-v2_hf', 'exllama2-hf', 'exllama2_hf', 'exllama-2-hf', 'exllama_2_hf', 'exllama-2_hf']:

From b9e2ded6d4ea32ceccea87d1f0a37ed003f0316e Mon Sep 17 00:00:00 2001
From: nclok1405 <155463060+nclok1405@users.noreply.github.com>
Date: Thu, 9 Jan 2025 09:17:31 +0900
Subject: [PATCH 0154/1701] Added UnicodeDecodeError workaround for
 modules/llamacpp_model.py (#6040)

---------

Co-authored-by: oobabooga <112222186+oobabooga@users.noreply.github.com>
---
 modules/llamacpp_model.py | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/modules/llamacpp_model.py b/modules/llamacpp_model.py
index 6a76ee4e95..c79755e460 100644
--- a/modules/llamacpp_model.py
+++ b/modules/llamacpp_model.py
@@ -122,7 +122,14 @@ def encode(self, string):
         return self.model.tokenize(string)
 
     def decode(self, ids, **kwargs):
-        return self.model.detokenize(ids).decode('utf-8')
+        detokenized = self.model.detokenize(ids)
+        try:
+            # Attempt strict UTF-8 decoding first
+            return detokenized.decode('utf-8', 'strict')
+        except UnicodeDecodeError as e:
+            # Log the error and fall back to UTF-8 with replacement
+            logger.warning(f"Invalid UTF-8 in detokenized output. Using replacement characters.\n{e}")
+            return detokenized.decode('utf-8', 'replace')
 
     def get_logits(self, tokens):
         self.model.reset()

From e6796c3859350a082174436922709818d2aa74ef Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 8 Jan 2025 17:24:21 -0800
Subject: [PATCH 0155/1701] Bump llama-cpp-python to 0.3.6, add macOS 14 and 15
 wheels

---
 requirements.txt                 | 24 ++++++++++++------------
 requirements_amd.txt             | 12 ++++++------
 requirements_amd_noavx2.txt      |  8 ++++----
 requirements_apple_intel.txt     |  6 ++++--
 requirements_apple_silicon.txt   | 10 ++++++----
 requirements_cpu_only.txt        |  8 ++++----
 requirements_cpu_only_noavx2.txt |  8 ++++----
 requirements_noavx2.txt          | 24 ++++++++++++------------
 8 files changed, 52 insertions(+), 48 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index fb8ca6b3e3..6539161c09 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -32,22 +32,22 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, no tensor cores)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, tensor cores)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # CUDA wheels
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 88b4b887f8..2e5f2da74e 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -31,14 +31,14 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.5+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.5+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.6+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.6+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index ef43da993f..b1eb7d31d6 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -31,10 +31,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # AMD wheels
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 369d34d881..6a9bf7f72a 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -31,6 +31,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp310-cp310-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 9eefd8b890..d8928d58bb 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -31,8 +31,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.5-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp310-cp310-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index f5313ecffd..84658a11dd 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -31,7 +31,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index aefe31d022..5944d5a75e 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -31,7 +31,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 9e1633ca85..fda4292dee 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -32,22 +32,22 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.5+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, no tensor cores)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.5+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, tensor cores)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.5+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # CUDA wheels
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"

From 4ffc9ffc7a84c39f2442d060eff531cf8f70a084 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 8 Jan 2025 17:24:38 -0800
Subject: [PATCH 0156/1701] UI: fix a list style

---
 css/main.css | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/css/main.css b/css/main.css
index dc89fbef49..6febb4714f 100644
--- a/css/main.css
+++ b/css/main.css
@@ -500,8 +500,8 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     margin-bottom: 0.5em !important;
 }
 
-.message-body ul.long-list li,
-.message-body ol.long-list li {
+.message-body ul.long-list > li,
+.message-body ol.long-list > li {
     margin-top: 1.25em !important;
     margin-bottom: 1.25em !important;
 }

From 5c89068168e6a5e65d55832e7a725f4688bc9434 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 8 Jan 2025 17:36:30 -0800
Subject: [PATCH 0157/1701] UI: add an info message for the new Static KV cache
 option

---
 modules/ui_parameters.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index f22f6233d1..c8fd6bc752 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -83,7 +83,7 @@ def create_ui(default_preset):
                             shared.gradio['add_bos_token'] = gr.Checkbox(value=shared.settings['add_bos_token'], label='Add the bos_token to the beginning of prompts', info='Disabling this can make the replies more creative.')
                             shared.gradio['skip_special_tokens'] = gr.Checkbox(value=shared.settings['skip_special_tokens'], label='Skip special tokens', info='Some specific models need this unset.')
                             shared.gradio['stream'] = gr.Checkbox(value=shared.settings['stream'], label='Activate text streaming')
-                            shared.gradio['static_cache'] = gr.Checkbox(value=shared.settings['static_cache'], label='Static KV cache')
+                            shared.gradio['static_cache'] = gr.Checkbox(value=shared.settings['static_cache'], label='Static KV cache', info='Use a static cache for improved performance.')
 
                         with gr.Column():
                             shared.gradio['truncation_length'] = gr.Number(precision=0, step=256, value=get_truncation_length(), label='Truncate the prompt up to this length', info='The leftmost tokens are removed if the prompt exceeds this length. Most models require this to be at most 2048.')

From 619265b32c9c436cd2687849ba2c56fa65a18033 Mon Sep 17 00:00:00 2001
From: BPplays <58504799+BPplays@users.noreply.github.com>
Date: Thu, 9 Jan 2025 05:23:44 -0800
Subject: [PATCH 0158/1701] add ipv6 support to the API (#6559)

---
 extensions/openai/script.py | 36 ++++++++++++++++++++++++++----------
 modules/shared.py           |  2 ++
 2 files changed, 28 insertions(+), 10 deletions(-)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 03d99e8ded..f23caf9b15 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -353,23 +353,38 @@ async def handle_unload_loras():
 
 
 def run_server():
-    server_addr = '0.0.0.0' if shared.args.listen else '127.0.0.1'
+    # Parse configuration
     port = int(os.environ.get('OPENEDAI_PORT', shared.args.api_port))
-
     ssl_certfile = os.environ.get('OPENEDAI_CERT_PATH', shared.args.ssl_certfile)
     ssl_keyfile = os.environ.get('OPENEDAI_KEY_PATH', shared.args.ssl_keyfile)
 
-    if shared.args.public_api:
-        def on_start(public_url: str):
-            logger.info(f'OpenAI-compatible API URL:\n\n{public_url}\n')
+    # In the server configuration:
+    server_addrs = []
+    if os.environ.get('OPENEDAI_ENABLE_IPV6', shared.args.api_enable_ipv6):
+        server_addrs.append('[::]' if shared.args.listen else '[::1]')
+    if not os.environ.get('OPENEDAI_DISABLE_IPV4', shared.args.api_disable_ipv4):
+        server_addrs.append('0.0.0.0' if shared.args.listen else '127.0.0.1')
 
-        _start_cloudflared(port, shared.args.public_api_id, max_attempts=3, on_start=on_start)
+    if not server_addrs:
+        raise Exception('you MUST enable IPv6 or IPv4 for the API to work')
+
+    # Log server information
+    if shared.args.public_api:
+        _start_cloudflared(
+            port,
+            shared.args.public_api_id,
+            max_attempts=3,
+            on_start=lambda url: logger.info(f'OpenAI-compatible API URL:\n\n{url}\n')
+        )
     else:
-        if ssl_keyfile and ssl_certfile:
-            logger.info(f'OpenAI-compatible API URL:\n\nhttps://{server_addr}:{port}\n')
+        url_proto = 'https://' if (ssl_certfile and ssl_keyfile) else 'http://'
+        urls = [f'{url_proto}{addr}:{port}' for addr in server_addrs]
+        if len(urls) > 1:
+            logger.info('OpenAI-compatible API URLs:\n\n' + '\n'.join(urls) + '\n')
         else:
-            logger.info(f'OpenAI-compatible API URL:\n\nhttp://{server_addr}:{port}\n')
+            logger.info('OpenAI-compatible API URL:\n\n' + '\n'.join(urls) + '\n')
 
+    # Log API keys
     if shared.args.api_key:
         if not shared.args.admin_key:
             shared.args.admin_key = shared.args.api_key
@@ -379,8 +394,9 @@ def on_start(public_url: str):
     if shared.args.admin_key and shared.args.admin_key != shared.args.api_key:
         logger.info(f'OpenAI API admin key (for loading/unloading models):\n\n{shared.args.admin_key}\n')
 
+    # Start server
     logging.getLogger("uvicorn.error").propagate = False
-    uvicorn.run(app, host=server_addr, port=port, ssl_certfile=ssl_certfile, ssl_keyfile=ssl_keyfile)
+    uvicorn.run(app, host=server_addrs, port=port, ssl_certfile=ssl_certfile, ssl_keyfile=ssl_keyfile)
 
 
 def setup():
diff --git a/modules/shared.py b/modules/shared.py
index 7829e46200..a0070b1ff7 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -193,6 +193,8 @@
 group.add_argument('--api-port', type=int, default=5000, help='The listening port for the API.')
 group.add_argument('--api-key', type=str, default='', help='API authentication key.')
 group.add_argument('--admin-key', type=str, default='', help='API authentication key for admin tasks like loading and unloading models. If not set, will be the same as --api-key.')
+group.add_argument('--api-enable-ipv6', action='store_true', help='Enable IPv6 for the API')
+group.add_argument('--api-disable-ipv4', action='store_true', help='Disable IPv4 for the API')
 group.add_argument('--nowebui', action='store_true', help='Do not launch the Gradio UI. Useful for launching the API in standalone mode.')
 
 # Multimodal

From 03b4067f3197de0ef753d2fe415785e0a5dffebb Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Jan 2025 11:58:33 -0800
Subject: [PATCH 0159/1701] Installer: ask 1 question for NVIDIA users instead
 of 2

---
 docker/amd/Dockerfile    |  2 +-
 docker/cpu/Dockerfile    |  2 +-
 docker/intel/Dockerfile  |  2 +-
 docker/nvidia/Dockerfile |  2 +-
 one_click.py             | 45 +++++++++++++++++++++-------------------
 5 files changed, 28 insertions(+), 25 deletions(-)

diff --git a/docker/amd/Dockerfile b/docker/amd/Dockerfile
index 365e88e3f8..cfbcf7e451 100644
--- a/docker/amd/Dockerfile
+++ b/docker/amd/Dockerfile
@@ -13,7 +13,7 @@ RUN --mount=type=cache,target=/var/cache/apt,sharing=locked,rw \
 WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
-RUN GPU_CHOICE=B USE_CUDA118=FALSE LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
+RUN GPU_CHOICE=C LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
 COPY CMD_FLAGS.txt /home/app/text-generation-webui/
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
 WORKDIR /home/app/text-generation-webui
diff --git a/docker/cpu/Dockerfile b/docker/cpu/Dockerfile
index 04ccf94a95..8f643d2f1a 100644
--- a/docker/cpu/Dockerfile
+++ b/docker/cpu/Dockerfile
@@ -17,7 +17,7 @@ RUN --mount=type=cache,target=/var/cache/apt,sharing=locked,rw \
 WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
-RUN GPU_CHOICE=N USE_CUDA118=FALSE LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
+RUN GPU_CHOICE=N LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
 COPY CMD_FLAGS.txt /home/app/text-generation-webui/
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
 # set umask to ensure group read / write at runtime
diff --git a/docker/intel/Dockerfile b/docker/intel/Dockerfile
index bc67a1855c..d2ed671edb 100644
--- a/docker/intel/Dockerfile
+++ b/docker/intel/Dockerfile
@@ -13,7 +13,7 @@ RUN --mount=type=cache,target=/var/cache/apt,sharing=locked,rw \
 WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
-RUN GPU_CHOICE=D USE_CUDA118=FALSE LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
+RUN GPU_CHOICE=E LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
 COPY CMD_FLAGS.txt /home/app/text-generation-webui/
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
 # set umask to ensure group read / write at runtime
diff --git a/docker/nvidia/Dockerfile b/docker/nvidia/Dockerfile
index 66a717a7a4..900a432967 100644
--- a/docker/nvidia/Dockerfile
+++ b/docker/nvidia/Dockerfile
@@ -13,7 +13,7 @@ RUN --mount=type=cache,target=/var/cache/apt,sharing=locked,rw \
 WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
-RUN GPU_CHOICE=A USE_CUDA118=FALSE LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
+RUN GPU_CHOICE=A LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
 COPY CMD_FLAGS.txt /home/app/text-generation-webui/
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
 WORKDIR /home/app/text-generation-webui
diff --git a/one_click.py b/one_click.py
index ca11efacd4..e78a2450b3 100644
--- a/one_click.py
+++ b/one_click.py
@@ -232,33 +232,45 @@ def get_user_choice(question, options_dict):
 
 
 def install_webui():
-
     # Ask the user for the GPU vendor
     if "GPU_CHOICE" in os.environ:
         choice = os.environ["GPU_CHOICE"].upper()
         print_big_message(f"Selected GPU choice \"{choice}\" based on the GPU_CHOICE environment variable.")
+
+        # Warn about changed meanings and handle old NVIDIA choice
+        if choice == "B":
+            print_big_message("Warning: GPU_CHOICE='B' now means 'NVIDIA (CUDA 11.8)' in the new version.")
+        elif choice == "C":
+            print_big_message("Warning: GPU_CHOICE='C' now means 'AMD' in the new version.")
+        elif choice == "D":
+            print_big_message("Warning: GPU_CHOICE='D' now means 'Apple M Series' in the new version.")
+        elif choice == "A" and "USE_CUDA118" in os.environ:
+            choice = "B" if os.environ.get("USE_CUDA118", "").lower() in ("yes", "y", "true", "1", "t", "on") else "A"
     else:
         choice = get_user_choice(
             "What is your GPU?",
             {
-                'A': 'NVIDIA',
-                'B': 'AMD (Linux/MacOS only. Requires ROCm SDK 6.1 on Linux)',
-                'C': 'Apple M Series',
-                'D': 'Intel Arc (IPEX)',
-                'N': 'None (I want to run models in CPU mode)'
+                'A': 'NVIDIA - CUDA 12.1 (recommended)',
+                'B': 'NVIDIA - CUDA 11.8 (legacy GPUs)',
+                'C': 'AMD - Linux/macOS only, requires ROCm 6.1',
+                'D': 'Apple M Series',
+                'E': 'Intel Arc (beta)',
+                'N': 'CPU mode'
             },
         )
 
+    # Convert choices to GPU names for compatibility
     gpu_choice_to_name = {
         "A": "NVIDIA",
-        "B": "AMD",
-        "C": "APPLE",
-        "D": "INTEL",
+        "B": "NVIDIA",
+        "C": "AMD",
+        "D": "APPLE",
+        "E": "INTEL",
         "N": "NONE"
     }
 
     selected_gpu = gpu_choice_to_name[choice]
-    use_cuda118 = "N"
+    use_cuda118 = (choice == "B")  # CUDA version is now determined by menu choice
 
     # Write a flag to CMD_FLAGS.txt for CPU mode
     if selected_gpu == "NONE":
@@ -267,18 +279,9 @@ def install_webui():
                 print_big_message("Adding the --cpu flag to CMD_FLAGS.txt.")
                 cmd_flags_file.write("\n--cpu\n")
 
-    # Check if the user wants CUDA 11.8
+    # Handle CUDA version display
     elif any((is_windows(), is_linux())) and selected_gpu == "NVIDIA":
-        if "USE_CUDA118" in os.environ:
-            use_cuda118 = "Y" if os.environ.get("USE_CUDA118", "").lower() in ("yes", "y", "true", "1", "t", "on") else "N"
-        else:
-            print("\nDo you want to use CUDA 11.8 instead of 12.1?\nOnly choose this option if your GPU is very old (Kepler or older).\n\nFor RTX and GTX series GPUs, say \"N\".\nIf unsure, say \"N\".\n")
-            use_cuda118 = input("Input (Y/N)> ").upper().strip('"\'').strip()
-            while use_cuda118 not in 'YN':
-                print("Invalid choice. Please try again.")
-                use_cuda118 = input("Input> ").upper().strip('"\'').strip()
-
-        if use_cuda118 == 'Y':
+        if use_cuda118:
             print("CUDA: 11.8")
         else:
             print("CUDA: 12.1")

From c08d87b78db0acf13cb3d7ced94f8c9b5b78ac91 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Jan 2025 12:23:38 -0800
Subject: [PATCH 0160/1701] Make the huggingface loader more readable

---
 modules/models.py | 37 +++++++++++++++++++++++++++++++------
 1 file changed, 31 insertions(+), 6 deletions(-)

diff --git a/modules/models.py b/modules/models.py
index 9c58b2799c..f551b82838 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -160,8 +160,22 @@ def huggingface_loader(model_name):
         else:
             LoaderClass = AutoModelForCausalLM
 
+    # Determine if we should use default loading
+    should_use_default_loading = not any([
+        shared.args.cpu,
+        shared.args.load_in_8bit,
+        shared.args.load_in_4bit,
+        shared.args.auto_devices,
+        shared.args.disk,
+        shared.args.deepspeed,
+        shared.args.gpu_memory is not None,
+        shared.args.cpu_memory is not None,
+        shared.args.compress_pos_emb > 1,
+        shared.args.alpha_value > 1,
+    ])
+
     # Load the model without any special settings
-    if not any([shared.args.cpu, shared.args.load_in_8bit, shared.args.load_in_4bit, shared.args.auto_devices, shared.args.disk, shared.args.deepspeed, shared.args.gpu_memory is not None, shared.args.cpu_memory is not None, shared.args.compress_pos_emb > 1, shared.args.alpha_value > 1]):
+    if should_use_default_loading:
         logger.info("TRANSFORMERS_PARAMS=")
         pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(params)
         print()
@@ -174,8 +188,20 @@ def huggingface_loader(model_name):
 
     # DeepSpeed ZeRO-3
     elif shared.args.deepspeed:
-        model = LoaderClass.from_pretrained(path_to_model, torch_dtype=params['torch_dtype'], trust_remote_code=params.get('trust_remote_code'))
-        model = deepspeed.initialize(model=model, config_params=ds_config, model_parameters=None, optimizer=None, lr_scheduler=None)[0]
+        model = LoaderClass.from_pretrained(
+            path_to_model,
+            torch_dtype=params['torch_dtype'],
+            trust_remote_code=params.get('trust_remote_code')
+        )
+
+        model = deepspeed.initialize(
+            model=model,
+            config_params=ds_config,
+            model_parameters=None,
+            optimizer=None,
+            lr_scheduler=None
+        )[0]
+
         model.module.eval()  # Inference
         logger.info(f'DeepSpeed ZeRO-3 is enabled: {is_deepspeed_zero3_enabled()}')
 
@@ -197,16 +223,15 @@ def huggingface_loader(model_name):
                 # and https://huggingface.co/blog/4bit-transformers-bitsandbytes
                 quantization_config_params = {
                     'load_in_4bit': True,
-                    'bnb_4bit_compute_dtype': eval("torch.{}".format(shared.args.compute_dtype)) if shared.args.compute_dtype in ["bfloat16", "float16", "float32"] else None,
+                    'bnb_4bit_compute_dtype': eval(f"torch.{shared.args.compute_dtype}") if shared.args.compute_dtype in ["bfloat16", "float16", "float32"] else None,
                     'bnb_4bit_quant_type': shared.args.quant_type,
                     'bnb_4bit_use_double_quant': shared.args.use_double_quant,
                     'llm_int8_enable_fp32_cpu_offload': True
                 }
-
                 params['quantization_config'] = BitsAndBytesConfig(**quantization_config_params)
 
             elif shared.args.load_in_8bit:
-                if any((shared.args.auto_devices, shared.args.gpu_memory)):
+                if shared.args.auto_devices or shared.args.gpu_memory:
                     params['quantization_config'] = BitsAndBytesConfig(load_in_8bit=True, llm_int8_enable_fp32_cpu_offload=True)
                 else:
                     params['quantization_config'] = BitsAndBytesConfig(load_in_8bit=True)

From 3020f2e5ec823eef91475bcbcc929b1b2d9ca614 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Jan 2025 12:44:03 -0800
Subject: [PATCH 0161/1701] UI: improve the info message about --tensorcores

---
 modules/ui_model_menu.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index eac0cba66a..d5116938df 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -103,7 +103,7 @@ def create_ui():
                             shared.gradio['num_experts_per_token'] = gr.Number(label="Number of experts per token", value=shared.args.num_experts_per_token, info='Only applies to MoE models like Mixtral.')
 
                         with gr.Column():
-                            shared.gradio['tensorcores'] = gr.Checkbox(label="tensorcores", value=shared.args.tensorcores, info='NVIDIA only: use llama-cpp-python compiled with tensor cores support. This may increase performance on newer cards.')
+                            shared.gradio['tensorcores'] = gr.Checkbox(label="tensorcores", value=shared.args.tensorcores, info='NVIDIA only: use llama-cpp-python compiled without GGML_CUDA_FORCE_MMQ. This may improve performance on newer cards.')
                             shared.gradio['load_in_8bit'] = gr.Checkbox(label="load-in-8bit", value=shared.args.load_in_8bit)
                             shared.gradio['load_in_4bit'] = gr.Checkbox(label="load-in-4bit", value=shared.args.load_in_4bit)
                             shared.gradio['torch_compile'] = gr.Checkbox(label="torch-compile", value=shared.args.torch_compile, info='Compile the model with torch.compile for improved performance.')

From 0e94d7075e5404537320b1a17c0a37bf2164a187 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Jan 2025 13:12:30 -0800
Subject: [PATCH 0162/1701] UI: minor style fix on Windows

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 6febb4714f..9d99a8761f 100644
--- a/css/main.css
+++ b/css/main.css
@@ -990,7 +990,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 #rename-row {
     width: 100%;
     justify-content: center;
-    gap: 10px;
+    gap: 9px;
 }
 
 

From f3c0f964a2ef51c0828495b88c654a0c0ab32b42 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Jan 2025 13:18:23 -0800
Subject: [PATCH 0163/1701] Lint

---
 modules/callbacks.py | 2 --
 1 file changed, 2 deletions(-)

diff --git a/modules/callbacks.py b/modules/callbacks.py
index 2c04cc53ac..0f918f3d54 100644
--- a/modules/callbacks.py
+++ b/modules/callbacks.py
@@ -1,11 +1,9 @@
-import gc
 import traceback
 from queue import Queue
 from threading import Thread
 
 import torch
 import transformers
-from transformers import is_torch_npu_available, is_torch_xpu_available
 
 import modules.shared as shared
 

From 15bfe36619d5c8b9b2be32e14f71bd47802a2223 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Jan 2025 15:58:14 -0800
Subject: [PATCH 0164/1701] Installer: update miniconda to 24.11.1
 (experimental)

---
 start_linux.sh    | 2 +-
 start_macos.sh    | 2 +-
 start_windows.bat | 4 ++--
 wsl.sh            | 2 +-
 4 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/start_linux.sh b/start_linux.sh
index 792daca840..256604cb7e 100755
--- a/start_linux.sh
+++ b/start_linux.sh
@@ -19,7 +19,7 @@ esac
 INSTALL_DIR="$(pwd)/installer_files"
 CONDA_ROOT_PREFIX="$(pwd)/installer_files/conda"
 INSTALL_ENV_DIR="$(pwd)/installer_files/env"
-MINICONDA_DOWNLOAD_URL="https://repo.anaconda.com/miniconda/Miniconda3-py310_23.3.1-0-Linux-${OS_ARCH}.sh"
+MINICONDA_DOWNLOAD_URL="https://repo.anaconda.com/miniconda/Miniconda3-py311_24.11.1-0-Linux-${OS_ARCH}.sh"
 conda_exists="F"
 
 # figure out whether git and conda needs to be installed
diff --git a/start_macos.sh b/start_macos.sh
index 6761f53169..02f1011afa 100755
--- a/start_macos.sh
+++ b/start_macos.sh
@@ -19,7 +19,7 @@ esac
 INSTALL_DIR="$(pwd)/installer_files"
 CONDA_ROOT_PREFIX="$(pwd)/installer_files/conda"
 INSTALL_ENV_DIR="$(pwd)/installer_files/env"
-MINICONDA_DOWNLOAD_URL="https://repo.anaconda.com/miniconda/Miniconda3-py310_23.3.1-0-MacOSX-${OS_ARCH}.sh"
+MINICONDA_DOWNLOAD_URL="https://repo.anaconda.com/miniconda/Miniconda3-py311_24.11.1-0-MacOSX-${OS_ARCH}.sh"
 conda_exists="F"
 
 # figure out whether git and conda needs to be installed
diff --git a/start_windows.bat b/start_windows.bat
index ebcc199706..c167cdc501 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -25,8 +25,8 @@ set TEMP=%cd%\installer_files
 set INSTALL_DIR=%cd%\installer_files
 set CONDA_ROOT_PREFIX=%cd%\installer_files\conda
 set INSTALL_ENV_DIR=%cd%\installer_files\env
-set MINICONDA_DOWNLOAD_URL=https://repo.anaconda.com/miniconda/Miniconda3-py310_23.3.1-0-Windows-x86_64.exe
-set MINICONDA_CHECKSUM=307194e1f12bbeb52b083634e89cc67db4f7980bd542254b43d3309eaf7cb358
+set MINICONDA_DOWNLOAD_URL=https://repo.anaconda.com/miniconda/Miniconda3-py311_24.11.1-0-Windows-x86_64.exe
+set MINICONDA_CHECKSUM=43dcbcc315ff91edf959e002cd2f1ede38c64b999fefcc951bccf2ed69c9e8bb
 set conda_exists=F
 
 @rem figure out whether git and conda needs to be installed
diff --git a/wsl.sh b/wsl.sh
index 7b17132f09..c5d28b1694 100755
--- a/wsl.sh
+++ b/wsl.sh
@@ -26,7 +26,7 @@ fi
 INSTALL_DIR="$INSTALL_DIR_PREFIX/text-generation-webui"
 CONDA_ROOT_PREFIX="$INSTALL_DIR/installer_files/conda"
 INSTALL_ENV_DIR="$INSTALL_DIR/installer_files/env"
-MINICONDA_DOWNLOAD_URL="https://repo.anaconda.com/miniconda/Miniconda3-py310_23.3.1-0-Linux-x86_64.sh"
+MINICONDA_DOWNLOAD_URL="https://repo.anaconda.com/miniconda/Miniconda3-py311_24.11.1-0-Linux-x86_64.sh"
 conda_exists="F"
 
 # environment isolation

From da6d868f58e519ec90796eef137cef26e67d4fd8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Jan 2025 16:11:46 -0800
Subject: [PATCH 0165/1701] Remove old deprecated flags (~6 months or more)

---
 modules/shared.py | 21 ++++++++++++++-------
 1 file changed, 14 insertions(+), 7 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index a0070b1ff7..f478df0551 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -203,11 +203,6 @@
 
 # Deprecated parameters
 group = parser.add_argument_group('Deprecated')
-group.add_argument('--model_type', type=str, help='DEPRECATED')
-group.add_argument('--pre_layer', type=int, nargs='+', help='DEPRECATED')
-group.add_argument('--checkpoint', type=str, help='DEPRECATED')
-group.add_argument('--monkey-patch', action='store_true', help='DEPRECATED')
-group.add_argument('--no_inject_fused_attention', action='store_true', help='DEPRECATED')
 group.add_argument('--cache_4bit', action='store_true', help='DEPRECATED')
 group.add_argument('--cache_8bit', action='store_true', help='DEPRECATED')
 group.add_argument('--chat-buttons', action='store_true', help='DEPRECATED')
@@ -228,14 +223,26 @@
     if hasattr(args, arg):
         provided_arguments.append(arg)
 
-deprecated_args = []
+deprecated_args = [
+    'cache_4bit',
+    'cache_8bit',
+    'chat_buttons',
+    'triton',
+    'no_inject_fused_mlp',
+    'no_use_cuda_fp16',
+    'desc_act',
+    'disable_exllama',
+    'disable_exllamav2',
+    'wbits',
+    'groupsize'
+]
 
 
 def do_cmd_flags_warnings():
 
     # Deprecation warnings
     for k in deprecated_args:
-        if getattr(args, k):
+        if k in provided_arguments:
             logger.warning(f'The --{k} flag has been deprecated and will be removed soon. Please remove that flag.')
 
     # Security warnings

From 7fe46764fb2d675c4e281592a1328293c0c56b07 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 10 Jan 2025 07:07:41 -0800
Subject: [PATCH 0166/1701] Improve the --help message about --tensorcores as
 well

---
 modules/shared.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/shared.py b/modules/shared.py
index f478df0551..8926320551 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -116,7 +116,7 @@
 # llama.cpp
 group = parser.add_argument_group('llama.cpp')
 group.add_argument('--flash-attn', action='store_true', help='Use flash-attention.')
-group.add_argument('--tensorcores', action='store_true', help='NVIDIA only: use llama-cpp-python compiled with tensor cores support. This may increase performance on newer cards.')
+group.add_argument('--tensorcores', action='store_true', help='NVIDIA only: use llama-cpp-python compiled without GGML_CUDA_FORCE_MMQ. This may improve performance on newer cards.')
 group.add_argument('--n_ctx', type=int, default=2048, help='Size of the prompt context.')
 group.add_argument('--threads', type=int, default=0, help='Number of threads to use.')
 group.add_argument('--threads-batch', type=int, default=0, help='Number of threads to use for batches/prompt processing.')

From 17aa97248fa3aa60d46a1a355ec8d2f5705bad38 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 10 Jan 2025 07:22:25 -0800
Subject: [PATCH 0167/1701] Installer: make the hashsum verification more
 robust on Windows

---
 start_windows.bat | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/start_windows.bat b/start_windows.bat
index c167cdc501..960cfdb787 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -41,10 +41,18 @@ if "%conda_exists%" == "F" (
 	mkdir "%INSTALL_DIR%"
 	call curl -Lk "%MINICONDA_DOWNLOAD_URL%" > "%INSTALL_DIR%\miniconda_installer.exe" || ( echo. && echo Miniconda failed to download. && goto end )
 
+	:: Try CertUtil first
 	for /f %%a in ('CertUtil -hashfile "%INSTALL_DIR%\miniconda_installer.exe" SHA256 ^| find /i /v " " ^| find /i "%MINICONDA_CHECKSUM%"') do (
 		set "output=%%a"
 	)
 
+	:: If CertUtil fails, try PowerShell
+	if not defined output (
+		for /f %%a in ('powershell -Command "if((Get-FileHash \"%INSTALL_DIR%\miniconda_installer.exe\" -Algorithm SHA256).Hash -eq ''%MINICONDA_CHECKSUM%''){echo true}"') do (
+			set "output=%%a"
+		)
+	)
+
 	if not defined output (
 		echo The checksum verification for miniconda_installer.exe has failed.
 		del "%INSTALL_DIR%\miniconda_installer.exe"

From 83c426e96b6c2cd4349d38c4cd212cbb0afd2044 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Fri, 10 Jan 2025 18:04:32 -0300
Subject: [PATCH 0168/1701] Organize internals (#6646)

---
 extensions/openai/typing.py |  48 +++----
 modules/loaders.py          | 266 ++++++++++++++++++------------------
 modules/presets.py          |  36 ++---
 modules/shared.py           |  30 ++--
 modules/text_generation.py  |  62 +++++++--
 modules/ui.py               | 150 ++++++++++----------
 6 files changed, 314 insertions(+), 278 deletions(-)

diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index dfac8e03bc..5f0e01281e 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -7,48 +7,48 @@
 
 class GenerationOptions(BaseModel):
     preset: str | None = Field(default=None, description="The name of a file under text-generation-webui/presets (without the .yaml extension). The sampling parameters that get overwritten by this option are the keys in the default_preset() function in modules/presets.py.")
-    min_p: float = 0
-    dynamic_temperature: bool = False
     dynatemp_low: float = 1
     dynatemp_high: float = 1
     dynatemp_exponent: float = 1
     smoothing_factor: float = 0
     smoothing_curve: float = 1
+    min_p: float = 0
     top_k: int = 0
-    repetition_penalty: float = 1
-    repetition_penalty_range: int = 1024
     typical_p: float = 1
-    tfs: float = 1
-    top_a: float = 0
+    xtc_threshold: float = 0.1
+    xtc_probability: float = 0
     epsilon_cutoff: float = 0
     eta_cutoff: float = 0
-    guidance_scale: float = 1
-    negative_prompt: str = ''
+    tfs: float = 1
+    top_a: float = 0
+    dry_multiplier: float = 0
+    dry_allowed_length: int = 2
+    dry_base: float = 1.75
+    repetition_penalty: float = 1
+    encoder_repetition_penalty: float = 1
+    no_repeat_ngram_size: int = 0
+    repetition_penalty_range: int = 1024
     penalty_alpha: float = 0
+    guidance_scale: float = 1
     mirostat_mode: int = 0
     mirostat_tau: float = 5
     mirostat_eta: float = 0.1
-    temperature_last: bool = False
-    do_sample: bool = True
-    seed: int = -1
-    encoder_repetition_penalty: float = 1
-    no_repeat_ngram_size: int = 0
-    dry_multiplier: float = 0
-    dry_base: float = 1.75
-    dry_allowed_length: int = 2
-    dry_sequence_breakers: str = '"\\n", ":", "\\"", "*"'
-    xtc_threshold: float = 0.1
-    xtc_probability: float = 0
-    truncation_length: int = 0
-    max_tokens_second: int = 0
     prompt_lookup_num_tokens: int = 0
-    static_cache: bool = False
-    custom_token_bans: str = ""
-    sampler_priority: List[str] | str | None = Field(default=None, description="List of samplers where the first items will appear first in the stack. Example: [\"top_k\", \"temperature\", \"top_p\"].")
+    max_tokens_second: int = 0
+    do_sample: bool = True
+    dynamic_temperature: bool = False
+    temperature_last: bool = False
     auto_max_new_tokens: bool = False
     ban_eos_token: bool = False
     add_bos_token: bool = True
     skip_special_tokens: bool = True
+    static_cache: bool = False
+    truncation_length: int = 0
+    seed: int = -1
+    sampler_priority: List[str] | str | None = Field(default=None, description="List of samplers where the first items will appear first in the stack. Example: [\"top_k\", \"temperature\", \"top_p\"].")
+    custom_token_bans: str = ""
+    negative_prompt: str = ''
+    dry_sequence_breakers: str = '"\\n", ":", "\\"", "*"'
     grammar_string: str = ""
 
 
diff --git a/modules/loaders.py b/modules/loaders.py
index 4e331dbbfd..cd864e406d 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -7,102 +7,103 @@
 
 loaders_and_params = OrderedDict({
     'Transformers': [
-        'cpu_memory',
         'gpu_memory',
-        'load_in_4bit',
+        'cpu_memory',
+        'alpha_value',
+        'compress_pos_emb',
+        'compute_dtype',
+        'quant_type',
         'load_in_8bit',
+        'load_in_4bit',
         'torch_compile',
-        'bf16',
+        'use_flash_attention_2',
+        'auto_devices',
         'cpu',
         'disk',
-        'auto_devices',
         'use_double_quant',
-        'quant_type',
-        'compute_dtype',
+        'use_eager_attention',
+        'bf16',
+
         'trust_remote_code',
         'no_use_fast',
-        'use_flash_attention_2',
-        'use_eager_attention',
-        'alpha_value',
-        'compress_pos_emb',
     ],
     'llama.cpp': [
-        'n_ctx',
         'n_gpu_layers',
-        'cache_type',
-        'tensor_split',
-        'n_batch',
         'threads',
         'threads_batch',
-        'no_mmap',
-        'mlock',
-        'no_mul_mat_q',
+        'n_batch',
+        'n_ctx',
+        'cache_type',
+        'tensor_split',
         'rope_freq_base',
         'compress_pos_emb',
-        'cpu',
-        'numa',
-        'no_offload_kqv',
-        'row_split',
+        'attention_sink_size',
         'tensorcores',
         'flash_attn',
         'streaming_llm',
-        'attention_sink_size',
+        'cpu',
+        'row_split',
+        'no_offload_kqv',
+        'no_mul_mat_q',
+        'no_mmap',
+        'mlock',
+        'numa',
     ],
     'llamacpp_HF': [
-        'n_ctx',
         'n_gpu_layers',
-        'cache_type',
-        'tensor_split',
-        'n_batch',
         'threads',
         'threads_batch',
-        'no_mmap',
-        'mlock',
-        'no_mul_mat_q',
+        'n_batch',
+        'n_ctx',
+        'cache_type',
+        'tensor_split',
         'rope_freq_base',
         'compress_pos_emb',
+        'attention_sink_size',
+        'tensorcores',
+        'flash_attn',
+        'streaming_llm',
         'cpu',
+        'row_split',
+        'no_offload_kqv',
+        'no_mul_mat_q',
+        'no_mmap',
+        'mlock',
         'numa',
         'cfg_cache',
+        'logits_all',
         'trust_remote_code',
         'no_use_fast',
-        'logits_all',
-        'no_offload_kqv',
-        'row_split',
-        'tensorcores',
-        'flash_attn',
-        'streaming_llm',
-        'attention_sink_size',
         'llamacpp_HF_info',
     ],
     'ExLlamav2_HF': [
-        'gpu_split',
         'max_seq_len',
-        'cfg_cache',
-        'no_flash_attn',
-        'no_xformers',
-        'no_sdpa',
-        'num_experts_per_token',
         'cache_type',
-        'autosplit',
-        'enable_tp',
+        'gpu_split',
         'alpha_value',
         'compress_pos_emb',
+        'num_experts_per_token',
+        'autosplit',
+        'enable_tp',
+        'no_flash_attn',
+        'no_xformers',
+        'no_sdpa',
+        'cfg_cache',
         'trust_remote_code',
         'no_use_fast',
     ],
     'ExLlamav2': [
-        'gpu_split',
         'max_seq_len',
-        'no_flash_attn',
-        'no_xformers',
-        'no_sdpa',
-        'num_experts_per_token',
         'cache_type',
-        'autosplit',
-        'enable_tp',
+        'gpu_split',
         'alpha_value',
         'compress_pos_emb',
+        'num_experts_per_token',
+        'autosplit',
+        'enable_tp',
+        'no_flash_attn',
+        'no_xformers',
+        'no_sdpa',
         'exllamav2_info',
     ],
     'HQQ': [
@@ -121,51 +122,51 @@
 def transformers_samplers():
     return {
         'temperature',
-        'temperature_last',
-        'dynamic_temperature',
         'dynatemp_low',
         'dynatemp_high',
         'dynatemp_exponent',
         'smoothing_factor',
         'smoothing_curve',
-        'top_p',
         'min_p',
+        'top_p',
         'top_k',
         'typical_p',
+        'xtc_threshold',
+        'xtc_probability',
         'epsilon_cutoff',
         'eta_cutoff',
         'tfs',
         'top_a',
+        'dry_multiplier',
+        'dry_allowed_length',
+        'dry_base',
         'repetition_penalty',
-        'presence_penalty',
         'frequency_penalty',
-        'repetition_penalty_range',
+        'presence_penalty',
         'encoder_repetition_penalty',
         'no_repeat_ngram_size',
-        'dry_multiplier',
-        'dry_base',
-        'dry_allowed_length',
-        'dry_sequence_breakers',
-        'xtc_threshold',
-        'xtc_probability',
-        'seed',
-        'do_sample',
+        'repetition_penalty_range',
         'penalty_alpha',
+        'guidance_scale',
         'mirostat_mode',
         'mirostat_tau',
         'mirostat_eta',
-        'grammar_file_row',
-        'grammar_string',
-        'guidance_scale',
-        'negative_prompt',
+        'prompt_lookup_num_tokens',
+        'do_sample',
+        'dynamic_temperature',
+        'temperature_last',
+        'auto_max_new_tokens',
         'ban_eos_token',
-        'custom_token_bans',
-        'sampler_priority',
         'add_bos_token',
         'skip_special_tokens',
-        'auto_max_new_tokens',
-        'prompt_lookup_num_tokens',
         'static_cache',
+        'seed',
+        'sampler_priority',
+        'custom_token_bans',
+        'negative_prompt',
+        'dry_sequence_breakers',
+        'grammar_string',
+        'grammar_file_row',
     }
 
 
@@ -174,155 +175,156 @@ def transformers_samplers():
     'HQQ': transformers_samplers(),
     'ExLlamav2': {
         'temperature',
-        'temperature_last',
-        'smoothing_factor',
         'dynatemp_low',
         'dynatemp_high',
         'dynatemp_exponent',
-        'top_p',
+        'smoothing_factor',
         'min_p',
+        'top_p',
         'top_k',
         'typical_p',
+        'xtc_threshold',
+        'xtc_probability',
         'tfs',
         'top_a',
+        'dry_multiplier',
+        'dry_allowed_length',
+        'dry_base',
         'repetition_penalty',
-        'presence_penalty',
         'frequency_penalty',
+        'presence_penalty',
         'repetition_penalty_range',
         'mirostat_mode',
         'mirostat_tau',
         'mirostat_eta',
-        'dry_multiplier',
-        'dry_base',
-        'dry_allowed_length',
-        'dry_sequence_breakers',
-        'xtc_threshold',
-        'xtc_probability',
-        'seed',
+        'dynamic_temperature',
+        'temperature_last',
+        'auto_max_new_tokens',
         'ban_eos_token',
         'add_bos_token',
-        'custom_token_bans',
         'skip_special_tokens',
-        'auto_max_new_tokens',
+        'seed',
+        'custom_token_bans',
+        'dry_sequence_breakers',
     },
     'ExLlamav2_HF': {
         'temperature',
-        'temperature_last',
-        'dynamic_temperature',
         'dynatemp_low',
         'dynatemp_high',
         'dynatemp_exponent',
         'smoothing_factor',
         'smoothing_curve',
-        'top_p',
         'min_p',
+        'top_p',
         'top_k',
         'typical_p',
+        'xtc_threshold',
+        'xtc_probability',
         'epsilon_cutoff',
         'eta_cutoff',
         'tfs',
         'top_a',
+        'dry_multiplier',
+        'dry_allowed_length',
+        'dry_base',
         'repetition_penalty',
-        'presence_penalty',
         'frequency_penalty',
-        'repetition_penalty_range',
+        'presence_penalty',
         'encoder_repetition_penalty',
         'no_repeat_ngram_size',
-        'dry_multiplier',
-        'dry_base',
-        'dry_allowed_length',
-        'dry_sequence_breakers',
-        'xtc_threshold',
-        'xtc_probability',
-        'seed',
-        'do_sample',
+        'repetition_penalty_range',
+        'guidance_scale',
         'mirostat_mode',
         'mirostat_tau',
         'mirostat_eta',
-        'grammar_file_row',
-        'grammar_string',
-        'guidance_scale',
-        'negative_prompt',
+        'do_sample',
+        'dynamic_temperature',
+        'temperature_last',
+        'auto_max_new_tokens',
         'ban_eos_token',
-        'custom_token_bans',
-        'sampler_priority',
         'add_bos_token',
         'skip_special_tokens',
-        'auto_max_new_tokens',
+        'seed',
+        'sampler_priority',
+        'custom_token_bans',
+        'negative_prompt',
+        'dry_sequence_breakers',
+        'grammar_string',
+        'grammar_file_row',
     },
     'llama.cpp': {
         'temperature',
-        'top_p',
         'min_p',
+        'top_p',
         'top_k',
         'typical_p',
         'tfs',
         'repetition_penalty',
-        'presence_penalty',
         'frequency_penalty',
-        'seed',
+        'presence_penalty',
         'mirostat_mode',
         'mirostat_tau',
         'mirostat_eta',
-        'grammar_file_row',
-        'grammar_string',
         'ban_eos_token',
+        'seed',
         'custom_token_bans',
+        'grammar_string',
+        'grammar_file_row',
     },
     'llamacpp_HF': {
         'temperature',
-        'temperature_last',
-        'dynamic_temperature',
         'dynatemp_low',
         'dynatemp_high',
         'dynatemp_exponent',
         'smoothing_factor',
         'smoothing_curve',
-        'top_p',
         'min_p',
+        'top_p',
         'top_k',
         'typical_p',
+        'xtc_threshold',
+        'xtc_probability',
         'epsilon_cutoff',
         'eta_cutoff',
         'tfs',
         'top_a',
+        'dry_multiplier',
+        'dry_allowed_length',
+        'dry_base',
         'repetition_penalty',
-        'presence_penalty',
         'frequency_penalty',
-        'repetition_penalty_range',
+        'presence_penalty',
         'encoder_repetition_penalty',
         'no_repeat_ngram_size',
-        'dry_multiplier',
-        'dry_base',
-        'dry_allowed_length',
-        'dry_sequence_breakers',
-        'xtc_threshold',
-        'xtc_probability',
-        'seed',
-        'do_sample',
+        'repetition_penalty_range',
+        'guidance_scale',
         'mirostat_mode',
         'mirostat_tau',
         'mirostat_eta',
-        'grammar_file_row',
-        'grammar_string',
-        'guidance_scale',
-        'negative_prompt',
+        'do_sample',
+        'dynamic_temperature',
+        'temperature_last',
+        'auto_max_new_tokens',
         'ban_eos_token',
-        'custom_token_bans',
-        'sampler_priority',
         'add_bos_token',
         'skip_special_tokens',
-        'auto_max_new_tokens',
+        'seed',
+        'sampler_priority',
+        'custom_token_bans',
+        'negative_prompt',
+        'dry_sequence_breakers',
+        'grammar_string',
+        'grammar_file_row',
     },
     'TensorRT-LLM': {
         'temperature',
         'top_p',
         'top_k',
         'repetition_penalty',
-        'presence_penalty',
         'frequency_penalty',
-        'ban_eos_token',
+        'presence_penalty',
         'auto_max_new_tokens',
+        'ban_eos_token',
     }
 }
 
diff --git a/modules/presets.py b/modules/presets.py
index c8118fb3b7..b841af53d6 100644
--- a/modules/presets.py
+++ b/modules/presets.py
@@ -13,40 +13,40 @@
 def default_preset():
     return {
         'temperature': 1,
-        'temperature_last': False,
-        'dynamic_temperature': False,
         'dynatemp_low': 1,
         'dynatemp_high': 1,
         'dynatemp_exponent': 1,
         'smoothing_factor': 0,
         'smoothing_curve': 1,
-        'top_p': 1,
         'min_p': 0,
+        'top_p': 1,
         'top_k': 0,
-        'repetition_penalty': 1,
-        'presence_penalty': 0,
-        'frequency_penalty': 0,
-        'repetition_penalty_range': 1024,
         'typical_p': 1,
-        'tfs': 1,
-        'top_a': 0,
+        'xtc_threshold': 0.1,
+        'xtc_probability': 0,
         'epsilon_cutoff': 0,
         'eta_cutoff': 0,
-        'guidance_scale': 1,
+        'tfs': 1,
+        'top_a': 0,
+        'dry_multiplier': 0,
+        'dry_allowed_length': 2,
+        'dry_base': 1.75,
+        'repetition_penalty': 1,
+        'frequency_penalty': 0,
+        'presence_penalty': 0,
+        'encoder_repetition_penalty': 1,
+        'no_repeat_ngram_size': 0,
+        'repetition_penalty_range': 1024,
         'penalty_alpha': 0,
+        'guidance_scale': 1,
         'mirostat_mode': 0,
         'mirostat_tau': 5,
         'mirostat_eta': 0.1,
         'do_sample': True,
-        'encoder_repetition_penalty': 1,
-        'no_repeat_ngram_size': 0,
-        'dry_multiplier': 0,
-        'dry_base': 1.75,
-        'dry_allowed_length': 2,
+        'dynamic_temperature': False,
+        'temperature_last': False,
+        'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ndry\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nmirostat\nxtc\nencoder_repetition_penalty\nno_repeat_ngram',
         'dry_sequence_breakers': '"\\n", ":", "\\"", "*"',
-        'xtc_threshold': 0.1,
-        'xtc_probability': 0,
-        'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ndry\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nmirostat\nxtc\nencoder_repetition_penalty\nno_repeat_ngram'
     }
 
 
diff --git a/modules/shared.py b/modules/shared.py
index 8926320551..928747f708 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -29,39 +29,39 @@
 
 # UI defaults
 settings = {
-    'dark_theme': True,
     'show_controls': True,
     'start_with': '',
     'mode': 'chat-instruct',
     'chat_style': 'cai-chat',
+    'chat-instruct_command': 'Continue the chat dialogue below. Write a single reply for the character "<|character|>".\n\n<|prompt|>',
     'prompt-default': 'QA',
     'prompt-notebook': 'QA',
+    'character': 'Assistant',
+    'name1': 'You',
+    'user_bio': '',
+    'custom_system_message': '',
+    'instruction_template_str': "{%- set ns = namespace(found=false) -%}\n{%- for message in messages -%}\n    {%- if message['role'] == 'system' -%}\n        {%- set ns.found = true -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if not ns.found -%}\n    {{- '' + 'Below is an instruction that describes a task. Write a response that appropriately completes the request.' + '\\n\\n' -}}\n{%- endif %}\n{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {{- '' + message['content'] + '\\n\\n' -}}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{-'### Instruction:\\n' + message['content'] + '\\n\\n'-}}\n        {%- else -%}\n            {{-'### Response:\\n' + message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{-'### Response:\\n'-}}\n{%- endif -%}",
+    'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{- name1 + ': ' + message['content'] + '\\n'-}}\n        {%- else -%}\n            {{- name2 + ': ' + message['content'] + '\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}",
     'preset': 'min_p',
     'max_new_tokens': 512,
     'max_new_tokens_min': 1,
     'max_new_tokens_max': 4096,
-    'negative_prompt': '',
-    'seed': -1,
-    'truncation_length': 2048,
+    'prompt_lookup_num_tokens': 0,
     'max_tokens_second': 0,
     'max_updates_second': 0,
-    'prompt_lookup_num_tokens': 0,
-    'static_cache': False,
-    'custom_stopping_strings': '',
-    'custom_token_bans': '',
     'auto_max_new_tokens': False,
     'ban_eos_token': False,
     'add_bos_token': True,
     'skip_special_tokens': True,
     'stream': True,
-    'character': 'Assistant',
-    'name1': 'You',
-    'user_bio': '',
-    'custom_system_message': '',
-    'instruction_template_str': "{%- set ns = namespace(found=false) -%}\n{%- for message in messages -%}\n    {%- if message['role'] == 'system' -%}\n        {%- set ns.found = true -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if not ns.found -%}\n    {{- '' + 'Below is an instruction that describes a task. Write a response that appropriately completes the request.' + '\\n\\n' -}}\n{%- endif %}\n{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {{- '' + message['content'] + '\\n\\n' -}}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{-'### Instruction:\\n' + message['content'] + '\\n\\n'-}}\n        {%- else -%}\n            {{-'### Response:\\n' + message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{-'### Response:\\n'-}}\n{%- endif -%}",
-    'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{- name1 + ': ' + message['content'] + '\\n'-}}\n        {%- else -%}\n            {{- name2 + ': ' + message['content'] + '\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}",
-    'chat-instruct_command': 'Continue the chat dialogue below. Write a single reply for the character "<|character|>".\n\n<|prompt|>',
+    'static_cache': False,
+    'truncation_length': 2048,
+    'seed': -1,
+    'custom_stopping_strings': '',
+    'custom_token_bans': '',
+    'negative_prompt': '',
     'autoload_model': False,
+    'dark_theme': True,
     'default_extensions': [],
 }
 
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 3e9788b81a..152b2b8df0 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -287,30 +287,61 @@ def generate_reply_HF(question, original_question, seed, state, stopping_strings
         clear_torch_cache()
 
     generate_params = {}
-    for k in ['max_new_tokens', 'temperature', 'temperature_last', 'dynamic_temperature', 'dynatemp_low', 'dynatemp_high', 'dynatemp_exponent', 'smoothing_factor', 'smoothing_curve', 'top_p', 'min_p', 'top_k', 'repetition_penalty', 'presence_penalty', 'frequency_penalty', 'repetition_penalty_range', 'typical_p', 'tfs', 'top_a', 'guidance_scale', 'penalty_alpha', 'mirostat_mode', 'mirostat_tau', 'mirostat_eta', 'do_sample', 'encoder_repetition_penalty', 'no_repeat_ngram_size', 'dry_multiplier', 'dry_base', 'dry_allowed_length', 'dry_sequence_breakers', 'xtc_threshold', 'xtc_probability']:
+    for k in [
+        'temperature',
+        'dynatemp_low',
+        'dynatemp_high',
+        'dynatemp_exponent',
+        'smoothing_factor',
+        'smoothing_curve',
+        'min_p',
+        'top_p',
+        'top_k',
+        'typical_p',
+        'xtc_threshold',
+        'xtc_probability',
+        'tfs',
+        'top_a',
+        'dry_multiplier',
+        'dry_allowed_length',
+        'dry_base',
+        'repetition_penalty',
+        'frequency_penalty',
+        'presence_penalty',
+        'encoder_repetition_penalty',
+        'no_repeat_ngram_size',
+        'repetition_penalty_range',
+        'penalty_alpha',
+        'guidance_scale',
+        'mirostat_mode',
+        'mirostat_tau',
+        'mirostat_eta',
+        'max_new_tokens',
+        'do_sample',
+        'dynamic_temperature',
+        'temperature_last',
+        'dry_sequence_breakers',
+    ]:
         if k in state:
             generate_params[k] = state[k]
 
-    if isinstance(state['sampler_priority'], list) and len(state['sampler_priority']) > 0:
-        generate_params['sampler_priority'] = state['sampler_priority']
-    elif isinstance(state['sampler_priority'], str) and state['sampler_priority'].strip() != '':
-        generate_params['sampler_priority'] = [x.strip() for x in state['sampler_priority'].replace('\n', ',').split(',') if x.strip()]
-
-    if state['negative_prompt'] != '':
-        generate_params['negative_prompt_ids'] = encode(state['negative_prompt'])
+    for k in ['epsilon_cutoff', 'eta_cutoff']:
+        if state[k] > 0:
+            generate_params[k] = state[k] * 1e-4
 
     if state['prompt_lookup_num_tokens'] > 0:
         generate_params['prompt_lookup_num_tokens'] = state['prompt_lookup_num_tokens']
 
+    if state['ban_eos_token']:
+        generate_params['suppress_tokens'] = [shared.tokenizer.eos_token_id]
+
     if state['static_cache']:
         generate_params['cache_implementation'] = 'static'
 
-    for k in ['epsilon_cutoff', 'eta_cutoff']:
-        if state[k] > 0:
-            generate_params[k] = state[k] * 1e-4
-
-    if state['ban_eos_token']:
-        generate_params['suppress_tokens'] = [shared.tokenizer.eos_token_id]
+    if isinstance(state['sampler_priority'], list) and len(state['sampler_priority']) > 0:
+        generate_params['sampler_priority'] = state['sampler_priority']
+    elif isinstance(state['sampler_priority'], str) and state['sampler_priority'].strip() != '':
+        generate_params['sampler_priority'] = [x.strip() for x in state['sampler_priority'].replace('\n', ',').split(',') if x.strip()]
 
     if state['custom_token_bans']:
         to_ban = [int(x) for x in state['custom_token_bans'].split(',')]
@@ -320,6 +351,9 @@ def generate_reply_HF(question, original_question, seed, state, stopping_strings
             else:
                 generate_params['suppress_tokens'] = to_ban
 
+    if state['negative_prompt'] != '':
+        generate_params['negative_prompt_ids'] = encode(state['negative_prompt'])
+
     generate_params.update({'use_cache': not shared.args.no_cache})
     if shared.args.deepspeed:
         generate_params.update({'synced_gpus': True})
diff --git a/modules/ui.py b/modules/ui.py
index e66de434ad..4f7ee785ac 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -102,55 +102,55 @@
 
 def list_model_elements():
     elements = [
-        'loader',
         'filter_by_loader',
+        'loader',
         'cpu_memory',
-        'auto_devices',
-        'disk',
-        'cpu',
-        'bf16',
-        'load_in_4bit',
-        'load_in_8bit',
-        'torch_compile',
-        'trust_remote_code',
-        'no_use_fast',
-        'use_flash_attention_2',
-        'use_eager_attention',
-        'compute_dtype',
-        'quant_type',
-        'use_double_quant',
-        'cfg_cache',
-        'no_flash_attn',
-        'no_xformers',
-        'no_sdpa',
-        'num_experts_per_token',
-        'cache_type',
-        'autosplit',
-        'enable_tp',
+        'n_gpu_layers',
         'threads',
         'threads_batch',
         'n_batch',
-        'no_mmap',
-        'mlock',
-        'no_mul_mat_q',
-        'n_gpu_layers',
-        'tensor_split',
+        'hqq_backend',
         'n_ctx',
-        'gpu_split',
         'max_seq_len',
-        'compress_pos_emb',
+        'cache_type',
+        'tensor_split',
+        'gpu_split',
         'alpha_value',
         'rope_freq_base',
-        'numa',
-        'logits_all',
-        'no_offload_kqv',
-        'row_split',
+        'compress_pos_emb',
+        'compute_dtype',
+        'quant_type',
+        'attention_sink_size',
+        'num_experts_per_token',
         'tensorcores',
+        'load_in_8bit',
+        'load_in_4bit',
+        'torch_compile',
         'flash_attn',
+        'use_flash_attention_2',
         'streaming_llm',
-        'attention_sink_size',
-        'hqq_backend',
+        'auto_devices',
+        'cpu',
+        'disk',
+        'row_split',
+        'no_offload_kqv',
+        'no_mul_mat_q',
+        'no_mmap',
+        'mlock',
+        'numa',
+        'use_double_quant',
+        'use_eager_attention',
+        'bf16',
+        'autosplit',
+        'enable_tp',
+        'no_flash_attn',
+        'no_xformers',
+        'no_sdpa',
+        'cfg_cache',
         'cpp_runner',
+        'logits_all',
+        'trust_remote_code',
+        'no_use_fast',
     ]
 
     if is_torch_xpu_available():
@@ -165,87 +165,87 @@ def list_model_elements():
 
 def list_interface_input_elements():
     elements = [
-        'max_new_tokens',
-        'auto_max_new_tokens',
-        'max_tokens_second',
-        'max_updates_second',
-        'prompt_lookup_num_tokens',
-        'seed',
         'temperature',
-        'temperature_last',
-        'dynamic_temperature',
         'dynatemp_low',
         'dynatemp_high',
         'dynatemp_exponent',
         'smoothing_factor',
         'smoothing_curve',
-        'top_p',
         'min_p',
+        'top_p',
         'top_k',
         'typical_p',
+        'xtc_threshold',
+        'xtc_probability',
         'epsilon_cutoff',
         'eta_cutoff',
+        'tfs',
+        'top_a',
+        'dry_multiplier',
+        'dry_allowed_length',
+        'dry_base',
         'repetition_penalty',
-        'presence_penalty',
         'frequency_penalty',
-        'repetition_penalty_range',
+        'presence_penalty',
         'encoder_repetition_penalty',
         'no_repeat_ngram_size',
-        'dry_multiplier',
-        'dry_base',
-        'dry_allowed_length',
-        'dry_sequence_breakers',
-        'xtc_threshold',
-        'xtc_probability',
-        'do_sample',
+        'repetition_penalty_range',
         'penalty_alpha',
+        'guidance_scale',
         'mirostat_mode',
         'mirostat_tau',
         'mirostat_eta',
-        'grammar_string',
-        'negative_prompt',
-        'guidance_scale',
-        'add_bos_token',
+        'max_new_tokens',
+        'prompt_lookup_num_tokens',
+        'max_tokens_second',
+        'max_updates_second',
+        'do_sample',
+        'dynamic_temperature',
+        'temperature_last',
+        'auto_max_new_tokens',
         'ban_eos_token',
-        'custom_token_bans',
-        'sampler_priority',
-        'truncation_length',
-        'custom_stopping_strings',
+        'add_bos_token',
         'skip_special_tokens',
         'stream',
         'static_cache',
-        'tfs',
-        'top_a',
+        'truncation_length',
+        'seed',
+        'sampler_priority',
+        'custom_stopping_strings',
+        'custom_token_bans',
+        'negative_prompt',
+        'dry_sequence_breakers',
+        'grammar_string',
     ]
 
     # Chat elements
     elements += [
-        'textbox',
-        'start_with',
-        'character_menu',
         'history',
         'search_chat',
         'unique_id',
-        'name1',
-        'user_bio',
+        'textbox',
+        'start_with',
+        'mode',
+        'chat_style',
+        'chat-instruct_command',
+        'character_menu',
         'name2',
-        'greeting',
         'context',
-        'mode',
+        'greeting',
+        'name1',
+        'user_bio',
         'custom_system_message',
         'instruction_template_str',
         'chat_template_str',
-        'chat_style',
-        'chat-instruct_command',
     ]
 
     # Notebook/default elements
     elements += [
-        'textbox-notebook',
         'textbox-default',
-        'output_textbox',
+        'textbox-notebook',
         'prompt_menu-default',
         'prompt_menu-notebook',
+        'output_textbox',
     ]
 
     # Model elements

From c393f7650d558d8ee1311adb5f66cc505e73fb78 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 10 Jan 2025 13:22:18 -0800
Subject: [PATCH 0169/1701] Update settings-template.yaml, organize
 modules/shared.py

---
 modules/shared.py      |  4 ++--
 settings-template.yaml | 34 +++++++++++++++++-----------------
 2 files changed, 19 insertions(+), 19 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 928747f708..4d873cb9f7 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -40,8 +40,6 @@
     'name1': 'You',
     'user_bio': '',
     'custom_system_message': '',
-    'instruction_template_str': "{%- set ns = namespace(found=false) -%}\n{%- for message in messages -%}\n    {%- if message['role'] == 'system' -%}\n        {%- set ns.found = true -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if not ns.found -%}\n    {{- '' + 'Below is an instruction that describes a task. Write a response that appropriately completes the request.' + '\\n\\n' -}}\n{%- endif %}\n{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {{- '' + message['content'] + '\\n\\n' -}}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{-'### Instruction:\\n' + message['content'] + '\\n\\n'-}}\n        {%- else -%}\n            {{-'### Response:\\n' + message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{-'### Response:\\n'-}}\n{%- endif -%}",
-    'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{- name1 + ': ' + message['content'] + '\\n'-}}\n        {%- else -%}\n            {{- name2 + ': ' + message['content'] + '\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}",
     'preset': 'min_p',
     'max_new_tokens': 512,
     'max_new_tokens_min': 1,
@@ -63,6 +61,8 @@
     'autoload_model': False,
     'dark_theme': True,
     'default_extensions': [],
+    'instruction_template_str': "{%- set ns = namespace(found=false) -%}\n{%- for message in messages -%}\n    {%- if message['role'] == 'system' -%}\n        {%- set ns.found = true -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if not ns.found -%}\n    {{- '' + 'Below is an instruction that describes a task. Write a response that appropriately completes the request.' + '\\n\\n' -}}\n{%- endif %}\n{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {{- '' + message['content'] + '\\n\\n' -}}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{-'### Instruction:\\n' + message['content'] + '\\n\\n'-}}\n        {%- else -%}\n            {{-'### Response:\\n' + message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{-'### Response:\\n'-}}\n{%- endif -%}",
+    'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{- name1 + ': ' + message['content'] + '\\n'-}}\n        {%- else -%}\n            {{- name2 + ': ' + message['content'] + '\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}",
 }
 
 default_settings = copy.deepcopy(settings)
diff --git a/settings-template.yaml b/settings-template.yaml
index d5ed47c392..b61dc4e01a 100644
--- a/settings-template.yaml
+++ b/settings-template.yaml
@@ -1,31 +1,38 @@
-dark_theme: true
 show_controls: true
 start_with: ''
 mode: chat-instruct
 chat_style: cai-chat
+chat-instruct_command: |-
+  Continue the chat dialogue below. Write a single reply for the character "<|character|>".
+
+  <|prompt|>
 prompt-default: QA
 prompt-notebook: QA
+character: Assistant
+name1: You
+user_bio: ''
+custom_system_message: ''
 preset: min_p
 max_new_tokens: 512
 max_new_tokens_min: 1
 max_new_tokens_max: 4096
-negative_prompt: ''
-seed: -1
-truncation_length: 2048
+prompt_lookup_num_tokens: 0
 max_tokens_second: 0
 max_updates_second: 0
-prompt_lookup_num_tokens: 0
-custom_stopping_strings: ''
-custom_token_bans: ''
 auto_max_new_tokens: false
 ban_eos_token: false
 add_bos_token: true
 skip_special_tokens: true
 stream: true
 static_cache: false
-character: Assistant
-name1: You
-custom_system_message: ''
+truncation_length: 2048
+seed: -1
+custom_stopping_strings: ''
+custom_token_bans: ''
+negative_prompt: ''
+autoload_model: false
+dark_theme: true
+default_extensions: []
 instruction_template_str: |-
   {%- set ns = namespace(found=false) -%}
   {%- for message in messages -%}
@@ -67,11 +74,4 @@ chat_template_str: |-
           {%- endif -%}
       {%- endif -%}
   {%- endfor -%}
-chat-instruct_command: |-
-  Continue the chat dialogue below. Write a single reply for the character "<|character|>".
 
-  <|prompt|>
-autoload_model: false
-gallery-items_per_page: 50
-gallery-open: false
-default_extensions: []

From d2f6c0f65ff72c96999a51655b096f42d037fe32 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 10 Jan 2025 13:25:40 -0800
Subject: [PATCH 0170/1701] Update README

---
 README.md         | 30 +++++++++++-------------------
 modules/shared.py |  2 +-
 2 files changed, 12 insertions(+), 20 deletions(-)

diff --git a/README.md b/README.md
index 071387726a..7e2dec95da 100644
--- a/README.md
+++ b/README.md
@@ -204,17 +204,16 @@ List of command-line flags
 usage: server.py [-h] [--multi-user] [--character CHARACTER] [--model MODEL] [--lora LORA [LORA ...]] [--model-dir MODEL_DIR] [--lora-dir LORA_DIR] [--model-menu] [--settings SETTINGS]
                  [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--cpu] [--auto-devices] [--gpu-memory GPU_MEMORY [GPU_MEMORY ...]]
                  [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast]
-                 [--use_flash_attention_2] [--use_eager_attention] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn] [--tensorcores]
-                 [--n_ctx N_CTX] [--threads THREADS] [--threads-batch THREADS_BATCH] [--no_mul_mat_q] [--n_batch N_BATCH] [--no-mmap] [--mlock] [--n-gpu-layers N_GPU_LAYERS]
+                 [--use_flash_attention_2] [--use_eager_attention] [--torch-compile] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn]
+                 [--tensorcores] [--n_ctx N_CTX] [--threads THREADS] [--threads-batch THREADS_BATCH] [--no_mul_mat_q] [--n_batch N_BATCH] [--no-mmap] [--mlock] [--n-gpu-layers N_GPU_LAYERS]
                  [--tensor_split TENSOR_SPLIT] [--numa] [--logits_all] [--no_offload_kqv] [--cache-capacity CACHE_CAPACITY] [--row_split] [--streaming-llm] [--attention-sink-size ATTENTION_SINK_SIZE]
                  [--tokenizer-dir TOKENIZER_DIR] [--gpu-split GPU_SPLIT] [--autosplit] [--max_seq_len MAX_SEQ_LEN] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa]
-                 [--num_experts_per_token NUM_EXPERTS_PER_TOKEN] [--enable_tp] [--triton] [--no_inject_fused_mlp] [--no_use_cuda_fp16] [--desc_act] [--disable_exllama] [--disable_exllamav2]
-                 [--wbits WBITS] [--groupsize GROUPSIZE] [--hqq-backend HQQ_BACKEND] [--cpp-runner] [--cache_type CACHE_TYPE] [--deepspeed] [--nvme-offload-dir NVME_OFFLOAD_DIR]
+                 [--num_experts_per_token NUM_EXPERTS_PER_TOKEN] [--enable_tp] [--hqq-backend HQQ_BACKEND] [--cpp-runner] [--cache_type CACHE_TYPE] [--deepspeed] [--nvme-offload-dir NVME_OFFLOAD_DIR]
                  [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT]
                  [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE]
-                 [--subpath SUBPATH] [--old-colors] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--nowebui]
-                 [--multimodal-pipeline MULTIMODAL_PIPELINE] [--model_type MODEL_TYPE] [--pre_layer PRE_LAYER [PRE_LAYER ...]] [--checkpoint CHECKPOINT] [--monkey-patch] [--no_inject_fused_attention]
-                 [--cache_4bit] [--cache_8bit] [--chat-buttons]
+                 [--subpath SUBPATH] [--old-colors] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6]
+                 [--api-disable-ipv4] [--nowebui] [--multimodal-pipeline MULTIMODAL_PIPELINE] [--cache_4bit] [--cache_8bit] [--chat-buttons] [--triton] [--no_inject_fused_mlp] [--no_use_cuda_fp16]
+                 [--desc_act] [--disable_exllama] [--disable_exllamav2] [--wbits WBITS] [--groupsize GROUPSIZE]
 
 Text generation web UI
 
@@ -237,7 +236,7 @@ Basic settings:
 
 Model loader:
   --loader LOADER                                Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav2_HF, ExLlamav2,
-                                                 AutoGPTQ.
+                                                 HQQ, TensorRT-LLM.
 
 Transformers/Accelerate:
   --cpu                                          Use the CPU to generate text. Warning: Training on CPU is extremely slow.
@@ -255,6 +254,7 @@ Transformers/Accelerate:
   --no_use_fast                                  Set use_fast=False while loading the tokenizer (it's True by default). Use this if you have any problems related to use_fast.
   --use_flash_attention_2                        Set use_flash_attention_2=True while loading the model.
   --use_eager_attention                          Set attn_implementation= eager while loading the model.
+  --torch-compile                                Compile the model with torch.compile for improved performance.
 
 bitsandbytes 4-bit:
   --load-in-4bit                                 Load the model with 4-bit precision (using bitsandbytes).
@@ -264,7 +264,7 @@ bitsandbytes 4-bit:
 
 llama.cpp:
   --flash-attn                                   Use flash-attention.
-  --tensorcores                                  NVIDIA only: use llama-cpp-python compiled with tensor cores support. This may increase performance on newer cards.
+  --tensorcores                                  NVIDIA only: use llama-cpp-python compiled without GGML_CUDA_FORCE_MMQ. This may improve performance on newer cards.
   --n_ctx N_CTX                                  Size of the prompt context.
   --threads THREADS                              Number of threads to use.
   --threads-batch THREADS_BATCH                  Number of threads to use for batches/prompt processing.
@@ -294,16 +294,6 @@ ExLlamaV2:
   --num_experts_per_token NUM_EXPERTS_PER_TOKEN  Number of experts to use for generation. Applies to MoE models like Mixtral.
   --enable_tp                                    Enable Tensor Parallelism (TP) in ExLlamaV2.
 
-AutoGPTQ:
-  --triton                                       Use triton.
-  --no_inject_fused_mlp                          Triton mode only: disable the use of fused MLP, which will use less VRAM at the cost of slower inference.
-  --no_use_cuda_fp16                             This can make models faster on some systems.
-  --desc_act                                     For models that do not have a quantize_config.json, this parameter is used to define whether to set desc_act or not in BaseQuantizeConfig.
-  --disable_exllama                              Disable ExLlama kernel, which can improve inference speed on some systems.
-  --disable_exllamav2                            Disable ExLlamav2 kernel.
-  --wbits WBITS                                  Load a pre-quantized model with specified precision in bits. 2, 3, 4 and 8 are supported.
-  --groupsize GROUPSIZE                          Group size.
-
 HQQ:
   --hqq-backend HQQ_BACKEND                      Backend for the HQQ loader. Valid options: PYTORCH, PYTORCH_COMPILE, ATEN.
 
@@ -343,6 +333,8 @@ API:
   --api-port API_PORT                            The listening port for the API.
   --api-key API_KEY                              API authentication key.
   --admin-key ADMIN_KEY                          API authentication key for admin tasks like loading and unloading models. If not set, will be the same as --api-key.
+  --api-enable-ipv6                              Enable IPv6 for the API
+  --api-disable-ipv4                             Disable IPv4 for the API
   --nowebui                                      Do not launch the Gradio UI. Useful for launching the API in standalone mode.
 
 Multimodal:
diff --git a/modules/shared.py b/modules/shared.py
index 4d873cb9f7..93cd227232 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -86,7 +86,7 @@
 
 # Model loader
 group = parser.add_argument_group('Model loader')
-group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav2_HF, ExLlamav2.')
+group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav2_HF, ExLlamav2, HQQ, TensorRT-LLM.')
 
 # Transformers/Accelerate
 group = parser.add_argument_group('Transformers/Accelerate')

From 02db4b0d06e9573de9e399b49006f882b996571b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 10 Jan 2025 15:05:08 -0800
Subject: [PATCH 0171/1701] Bump transformers to 4.48

---
 requirements.txt                 | 2 +-
 requirements_amd.txt             | 2 +-
 requirements_amd_noavx2.txt      | 2 +-
 requirements_apple_intel.txt     | 2 +-
 requirements_apple_silicon.txt   | 2 +-
 requirements_cpu_only.txt        | 2 +-
 requirements_cpu_only_noavx2.txt | 2 +-
 requirements_noavx2.txt          | 2 +-
 requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 6539161c09..c7ced3df54 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.47.*
+transformers==4.48.*
 tqdm
 wandb
 
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 2e5f2da74e..87ee93d1d9 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.47.*
+transformers==4.48.*
 tqdm
 wandb
 
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index b1eb7d31d6..fa2f5ca745 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.47.*
+transformers==4.48.*
 tqdm
 wandb
 
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 6a9bf7f72a..e983829542 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.47.*
+transformers==4.48.*
 tqdm
 wandb
 
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index d8928d58bb..bef02feb7b 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.47.*
+transformers==4.48.*
 tqdm
 wandb
 
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index 84658a11dd..32f1a50a91 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.47.*
+transformers==4.48.*
 tqdm
 wandb
 
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 5944d5a75e..938848bfaf 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.47.*
+transformers==4.48.*
 tqdm
 wandb
 
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index fda4292dee..e18cbe6437 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.47.*
+transformers==4.48.*
 tqdm
 wandb
 
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index 45003f0db5..a034ee6180 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.47.*
+transformers==4.48.*
 tqdm
 wandb
 

From 3a722a36c85f31f7d5d4529b8dfea3faec7b9c37 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Sat, 11 Jan 2025 12:55:19 -0300
Subject: [PATCH 0172/1701] Use morphdom to make chat streaming 1902381098231%
 faster (#6653)

---
 js/main.js                      |  3 +--
 js/morphdom/morphdom-umd.min.js |  1 +
 modules/block_requests.py       |  1 +
 modules/ui_chat.py              | 26 +++++++++++++++++++++++---
 4 files changed, 26 insertions(+), 5 deletions(-)
 create mode 100644 js/morphdom/morphdom-umd.min.js

diff --git a/js/main.js b/js/main.js
index efb8423807..ab2499d4b4 100644
--- a/js/main.js
+++ b/js/main.js
@@ -147,10 +147,9 @@ const observer = new MutationObserver(function(mutations) {
 
   doSyntaxHighlighting();
 
-  if(!isScrolled) {
+  if (!isScrolled && targetElement.scrollTop !== targetElement.scrollHeight) {
     targetElement.scrollTop = targetElement.scrollHeight;
   }
-
 });
 
 // Configure the observer to watch for changes in the subtree and attributes
diff --git a/js/morphdom/morphdom-umd.min.js b/js/morphdom/morphdom-umd.min.js
new file mode 100644
index 0000000000..6746f0e805
--- /dev/null
+++ b/js/morphdom/morphdom-umd.min.js
@@ -0,0 +1 @@
+(function(global,factory){typeof exports==="object"&&typeof module!=="undefined"?module.exports=factory():typeof define==="function"&&define.amd?define(factory):(global=global||self,global.morphdom=factory())})(this,function(){"use strict";var DOCUMENT_FRAGMENT_NODE=11;function morphAttrs(fromNode,toNode){var toNodeAttrs=toNode.attributes;var attr;var attrName;var attrNamespaceURI;var attrValue;var fromValue;if(toNode.nodeType===DOCUMENT_FRAGMENT_NODE||fromNode.nodeType===DOCUMENT_FRAGMENT_NODE){return}for(var i=toNodeAttrs.length-1;i>=0;i--){attr=toNodeAttrs[i];attrName=attr.name;attrNamespaceURI=attr.namespaceURI;attrValue=attr.value;if(attrNamespaceURI){attrName=attr.localName||attrName;fromValue=fromNode.getAttributeNS(attrNamespaceURI,attrName);if(fromValue!==attrValue){if(attr.prefix==="xmlns"){attrName=attr.name}fromNode.setAttributeNS(attrNamespaceURI,attrName,attrValue)}}else{fromValue=fromNode.getAttribute(attrName);if(fromValue!==attrValue){fromNode.setAttribute(attrName,attrValue)}}}var fromNodeAttrs=fromNode.attributes;for(var d=fromNodeAttrs.length-1;d>=0;d--){attr=fromNodeAttrs[d];attrName=attr.name;attrNamespaceURI=attr.namespaceURI;if(attrNamespaceURI){attrName=attr.localName||attrName;if(!toNode.hasAttributeNS(attrNamespaceURI,attrName)){fromNode.removeAttributeNS(attrNamespaceURI,attrName)}}else{if(!toNode.hasAttribute(attrName)){fromNode.removeAttribute(attrName)}}}}var range;var NS_XHTML="http://www.w3.org/1999/xhtml";var doc=typeof document==="undefined"?undefined:document;var HAS_TEMPLATE_SUPPORT=!!doc&&"content"in doc.createElement("template");var HAS_RANGE_SUPPORT=!!doc&&doc.createRange&&"createContextualFragment"in doc.createRange();function createFragmentFromTemplate(str){var template=doc.createElement("template");template.innerHTML=str;return template.content.childNodes[0]}function createFragmentFromRange(str){if(!range){range=doc.createRange();range.selectNode(doc.body)}var fragment=range.createContextualFragment(str);return fragment.childNodes[0]}function createFragmentFromWrap(str){var fragment=doc.createElement("body");fragment.innerHTML=str;return fragment.childNodes[0]}function toElement(str){str=str.trim();if(HAS_TEMPLATE_SUPPORT){return createFragmentFromTemplate(str)}else if(HAS_RANGE_SUPPORT){return createFragmentFromRange(str)}return createFragmentFromWrap(str)}function compareNodeNames(fromEl,toEl){var fromNodeName=fromEl.nodeName;var toNodeName=toEl.nodeName;var fromCodeStart,toCodeStart;if(fromNodeName===toNodeName){return true}fromCodeStart=fromNodeName.charCodeAt(0);toCodeStart=toNodeName.charCodeAt(0);if(fromCodeStart<=90&&toCodeStart>=97){return fromNodeName===toNodeName.toUpperCase()}else if(toCodeStart<=90&&fromCodeStart>=97){return toNodeName===fromNodeName.toUpperCase()}else{return false}}function createElementNS(name,namespaceURI){return!namespaceURI||namespaceURI===NS_XHTML?doc.createElement(name):doc.createElementNS(namespaceURI,name)}function moveChildren(fromEl,toEl){var curChild=fromEl.firstChild;while(curChild){var nextChild=curChild.nextSibling;toEl.appendChild(curChild);curChild=nextChild}return toEl}function syncBooleanAttrProp(fromEl,toEl,name){if(fromEl[name]!==toEl[name]){fromEl[name]=toEl[name];if(fromEl[name]){fromEl.setAttribute(name,"")}else{fromEl.removeAttribute(name)}}}var specialElHandlers={OPTION:function(fromEl,toEl){var parentNode=fromEl.parentNode;if(parentNode){var parentName=parentNode.nodeName.toUpperCase();if(parentName==="OPTGROUP"){parentNode=parentNode.parentNode;parentName=parentNode&&parentNode.nodeName.toUpperCase()}if(parentName==="SELECT"&&!parentNode.hasAttribute("multiple")){if(fromEl.hasAttribute("selected")&&!toEl.selected){fromEl.setAttribute("selected","selected");fromEl.removeAttribute("selected")}parentNode.selectedIndex=-1}}syncBooleanAttrProp(fromEl,toEl,"selected")},INPUT:function(fromEl,toEl){syncBooleanAttrProp(fromEl,toEl,"checked");syncBooleanAttrProp(fromEl,toEl,"disabled");if(fromEl.value!==toEl.value){fromEl.value=toEl.value}if(!toEl.hasAttribute("value")){fromEl.removeAttribute("value")}},TEXTAREA:function(fromEl,toEl){var newValue=toEl.value;if(fromEl.value!==newValue){fromEl.value=newValue}var firstChild=fromEl.firstChild;if(firstChild){var oldValue=firstChild.nodeValue;if(oldValue==newValue||!newValue&&oldValue==fromEl.placeholder){return}firstChild.nodeValue=newValue}},SELECT:function(fromEl,toEl){if(!toEl.hasAttribute("multiple")){var selectedIndex=-1;var i=0;var curChild=fromEl.firstChild;var optgroup;var nodeName;while(curChild){nodeName=curChild.nodeName&&curChild.nodeName.toUpperCase();if(nodeName==="OPTGROUP"){optgroup=curChild;curChild=optgroup.firstChild}else{if(nodeName==="OPTION"){if(curChild.hasAttribute("selected")){selectedIndex=i;break}i++}curChild=curChild.nextSibling;if(!curChild&&optgroup){curChild=optgroup.nextSibling;optgroup=null}}}fromEl.selectedIndex=selectedIndex}}};var ELEMENT_NODE=1;var DOCUMENT_FRAGMENT_NODE$1=11;var TEXT_NODE=3;var COMMENT_NODE=8;function noop(){}function defaultGetNodeKey(node){if(node){return node.getAttribute&&node.getAttribute("id")||node.id}}function morphdomFactory(morphAttrs){return function morphdom(fromNode,toNode,options){if(!options){options={}}if(typeof toNode==="string"){if(fromNode.nodeName==="#document"||fromNode.nodeName==="HTML"||fromNode.nodeName==="BODY"){var toNodeHtml=toNode;toNode=doc.createElement("html");toNode.innerHTML=toNodeHtml}else{toNode=toElement(toNode)}}else if(toNode.nodeType===DOCUMENT_FRAGMENT_NODE$1){toNode=toNode.firstElementChild}var getNodeKey=options.getNodeKey||defaultGetNodeKey;var onBeforeNodeAdded=options.onBeforeNodeAdded||noop;var onNodeAdded=options.onNodeAdded||noop;var onBeforeElUpdated=options.onBeforeElUpdated||noop;var onElUpdated=options.onElUpdated||noop;var onBeforeNodeDiscarded=options.onBeforeNodeDiscarded||noop;var onNodeDiscarded=options.onNodeDiscarded||noop;var onBeforeElChildrenUpdated=options.onBeforeElChildrenUpdated||noop;var skipFromChildren=options.skipFromChildren||noop;var addChild=options.addChild||function(parent,child){return parent.appendChild(child)};var childrenOnly=options.childrenOnly===true;var fromNodesLookup=Object.create(null);var keyedRemovalList=[];function addKeyedRemoval(key){keyedRemovalList.push(key)}function walkDiscardedChildNodes(node,skipKeyedNodes){if(node.nodeType===ELEMENT_NODE){var curChild=node.firstChild;while(curChild){var key=undefined;if(skipKeyedNodes&&(key=getNodeKey(curChild))){addKeyedRemoval(key)}else{onNodeDiscarded(curChild);if(curChild.firstChild){walkDiscardedChildNodes(curChild,skipKeyedNodes)}}curChild=curChild.nextSibling}}}function removeNode(node,parentNode,skipKeyedNodes){if(onBeforeNodeDiscarded(node)===false){return}if(parentNode){parentNode.removeChild(node)}onNodeDiscarded(node);walkDiscardedChildNodes(node,skipKeyedNodes)}function indexTree(node){if(node.nodeType===ELEMENT_NODE||node.nodeType===DOCUMENT_FRAGMENT_NODE$1){var curChild=node.firstChild;while(curChild){var key=getNodeKey(curChild);if(key){fromNodesLookup[key]=curChild}indexTree(curChild);curChild=curChild.nextSibling}}}indexTree(fromNode);function handleNodeAdded(el){onNodeAdded(el);var curChild=el.firstChild;while(curChild){var nextSibling=curChild.nextSibling;var key=getNodeKey(curChild);if(key){var unmatchedFromEl=fromNodesLookup[key];if(unmatchedFromEl&&compareNodeNames(curChild,unmatchedFromEl)){curChild.parentNode.replaceChild(unmatchedFromEl,curChild);morphEl(unmatchedFromEl,curChild)}else{handleNodeAdded(curChild)}}else{handleNodeAdded(curChild)}curChild=nextSibling}}function cleanupFromEl(fromEl,curFromNodeChild,curFromNodeKey){while(curFromNodeChild){var fromNextSibling=curFromNodeChild.nextSibling;if(curFromNodeKey=getNodeKey(curFromNodeChild)){addKeyedRemoval(curFromNodeKey)}else{removeNode(curFromNodeChild,fromEl,true)}curFromNodeChild=fromNextSibling}}function morphEl(fromEl,toEl,childrenOnly){var toElKey=getNodeKey(toEl);if(toElKey){delete fromNodesLookup[toElKey]}if(!childrenOnly){var beforeUpdateResult=onBeforeElUpdated(fromEl,toEl);if(beforeUpdateResult===false){return}else if(beforeUpdateResult instanceof HTMLElement){fromEl=beforeUpdateResult;indexTree(fromEl)}morphAttrs(fromEl,toEl);onElUpdated(fromEl);if(onBeforeElChildrenUpdated(fromEl,toEl)===false){return}}if(fromEl.nodeName!=="TEXTAREA"){morphChildren(fromEl,toEl)}else{specialElHandlers.TEXTAREA(fromEl,toEl)}}function morphChildren(fromEl,toEl){var skipFrom=skipFromChildren(fromEl,toEl);var curToNodeChild=toEl.firstChild;var curFromNodeChild=fromEl.firstChild;var curToNodeKey;var curFromNodeKey;var fromNextSibling;var toNextSibling;var matchingFromEl;outer:while(curToNodeChild){toNextSibling=curToNodeChild.nextSibling;curToNodeKey=getNodeKey(curToNodeChild);while(!skipFrom&&curFromNodeChild){fromNextSibling=curFromNodeChild.nextSibling;if(curToNodeChild.isSameNode&&curToNodeChild.isSameNode(curFromNodeChild)){curToNodeChild=toNextSibling;curFromNodeChild=fromNextSibling;continue outer}curFromNodeKey=getNodeKey(curFromNodeChild);var curFromNodeType=curFromNodeChild.nodeType;var isCompatible=undefined;if(curFromNodeType===curToNodeChild.nodeType){if(curFromNodeType===ELEMENT_NODE){if(curToNodeKey){if(curToNodeKey!==curFromNodeKey){if(matchingFromEl=fromNodesLookup[curToNodeKey]){if(fromNextSibling===matchingFromEl){isCompatible=false}else{fromEl.insertBefore(matchingFromEl,curFromNodeChild);if(curFromNodeKey){addKeyedRemoval(curFromNodeKey)}else{removeNode(curFromNodeChild,fromEl,true)}curFromNodeChild=matchingFromEl;curFromNodeKey=getNodeKey(curFromNodeChild)}}else{isCompatible=false}}}else if(curFromNodeKey){isCompatible=false}isCompatible=isCompatible!==false&&compareNodeNames(curFromNodeChild,curToNodeChild);if(isCompatible){morphEl(curFromNodeChild,curToNodeChild)}}else if(curFromNodeType===TEXT_NODE||curFromNodeType==COMMENT_NODE){isCompatible=true;if(curFromNodeChild.nodeValue!==curToNodeChild.nodeValue){curFromNodeChild.nodeValue=curToNodeChild.nodeValue}}}if(isCompatible){curToNodeChild=toNextSibling;curFromNodeChild=fromNextSibling;continue outer}if(curFromNodeKey){addKeyedRemoval(curFromNodeKey)}else{removeNode(curFromNodeChild,fromEl,true)}curFromNodeChild=fromNextSibling}if(curToNodeKey&&(matchingFromEl=fromNodesLookup[curToNodeKey])&&compareNodeNames(matchingFromEl,curToNodeChild)){if(!skipFrom){addChild(fromEl,matchingFromEl)}morphEl(matchingFromEl,curToNodeChild)}else{var onBeforeNodeAddedResult=onBeforeNodeAdded(curToNodeChild);if(onBeforeNodeAddedResult!==false){if(onBeforeNodeAddedResult){curToNodeChild=onBeforeNodeAddedResult}if(curToNodeChild.actualize){curToNodeChild=curToNodeChild.actualize(fromEl.ownerDocument||doc)}addChild(fromEl,curToNodeChild);handleNodeAdded(curToNodeChild)}}curToNodeChild=toNextSibling;curFromNodeChild=fromNextSibling}cleanupFromEl(fromEl,curFromNodeChild,curFromNodeKey);var specialElHandler=specialElHandlers[fromEl.nodeName];if(specialElHandler){specialElHandler(fromEl,toEl)}}var morphedNode=fromNode;var morphedNodeType=morphedNode.nodeType;var toNodeType=toNode.nodeType;if(!childrenOnly){if(morphedNodeType===ELEMENT_NODE){if(toNodeType===ELEMENT_NODE){if(!compareNodeNames(fromNode,toNode)){onNodeDiscarded(fromNode);morphedNode=moveChildren(fromNode,createElementNS(toNode.nodeName,toNode.namespaceURI))}}else{morphedNode=toNode}}else if(morphedNodeType===TEXT_NODE||morphedNodeType===COMMENT_NODE){if(toNodeType===morphedNodeType){if(morphedNode.nodeValue!==toNode.nodeValue){morphedNode.nodeValue=toNode.nodeValue}return morphedNode}else{morphedNode=toNode}}}if(morphedNode===toNode){onNodeDiscarded(fromNode)}else{if(toNode.isSameNode&&toNode.isSameNode(morphedNode)){return}morphEl(morphedNode,toNode,childrenOnly);if(keyedRemovalList){for(var i=0,len=keyedRemovalList.length;i'
             '\n    '
             '\n    '
+            '\n    '
             f'\n    '
             '\n    '
             '\n  '
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index b92dd9ae4d..61be17e3c9 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -20,7 +20,7 @@ def create_ui():
     shared.gradio['Chat input'] = gr.State()
     shared.gradio['history'] = gr.JSON(visible=False)
 
-    with gr.Tab('Chat', elem_id='chat-tab'):
+    with gr.Tab('Chat', id='Chat', elem_id='chat-tab'):
         with gr.Row(elem_id='past-chats-row', elem_classes=['pretty_scrollbar']):
             with gr.Column():
                 with gr.Row(elem_id='past-chats-buttons'):
@@ -46,8 +46,8 @@ def create_ui():
 
         with gr.Row():
             with gr.Column(elem_id='chat-col'):
-                shared.gradio['display'] = gr.HTML(value=chat_html_wrapper({'internal': [], 'visible': []}, '', '', 'chat', 'cai-chat', ''))
-
+                shared.gradio['html_display'] = gr.HTML(value=chat_html_wrapper({'internal': [], 'visible': []}, '', '', 'chat', 'cai-chat', ''), visible=True)
+                shared.gradio['display'] = gr.Textbox(value="", visible=False)  # Hidden buffer
                 with gr.Row(elem_id="chat-input-row"):
                     with gr.Column(scale=1, elem_id='gr-hover-container'):
                         gr.HTML(value='
            ☰
            ', elem_id='gr-hover')
@@ -180,6 +180,26 @@ def create_event_handlers():
     shared.input_params = gradio(inputs)
     shared.reload_inputs = gradio(reload_arr)
 
+    # Morph HTML updates instead of updating everything
+    shared.gradio['display'].change(None, gradio('display'), None,
+        js="""
+            (text) => {
+                morphdom(
+                    document.getElementById('chat').parentNode,
+                    '
            ' + text + '
            ',
+                    {
+                        onBeforeElUpdated: function(fromEl, toEl) {
+                            if (fromEl.isEqualNode(toEl)) {
+                                return false; // Skip identical nodes
+                            }
+                            return true; // Update only if nodes differ
+                        }
+                    }
+                );
+            }
+        """
+    )
+
     shared.gradio['Generate'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         lambda x: (x, ''), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False).then(

From 58342740a5b061c2836f46d93dd70832f894c6e9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 11 Jan 2025 07:59:49 -0800
Subject: [PATCH 0173/1701] Bump flash-attn to 2.7.3

---
 requirements.txt        | 8 ++++----
 requirements_noavx2.txt | 8 ++++----
 2 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index c7ced3df54..9bc5956ab2 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -55,7 +55,7 @@ https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/oobabooga/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu12torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu12torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index e18cbe6437..1755ac24cb 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -55,7 +55,7 @@ https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/oobabooga/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.2.post1/flash_attn-2.7.2.post1+cu12torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu12torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"

From a5d64b586da0ce39c36d01a59d991fbc76e16362 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Sat, 11 Jan 2025 16:59:21 -0300
Subject: [PATCH 0174/1701] Add a "copy" button below each message (#6654)

---
 css/html_instruct_style.css |  2 ++
 css/main.css                | 53 +++++++++++++++++++++++++++-
 modules/block_requests.py   |  3 +-
 modules/html_generator.py   | 70 ++++++++++++++++++++++++-------------
 modules/ui.py               |  2 ++
 5 files changed, 104 insertions(+), 26 deletions(-)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index dcc19c2910..fcd0558f0d 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -48,12 +48,14 @@
 .chat .user-message {
     background: #f4f4f4;
     padding: 1.5rem 1rem;
+    padding-bottom: 2rem;
     border-radius: 0;
     border-bottom-right-radius: 0;
 }
 
 .chat .assistant-message {
     padding: 1.5rem 1rem;
+    padding-bottom: 2rem;
     border-radius: 0;
     border: 0;
 }
diff --git a/css/main.css b/css/main.css
index 9d99a8761f..48c6727a1b 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1142,7 +1142,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 .dark svg {
-    fill: white;
     color: white;
 }
 
@@ -1221,3 +1220,55 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
         background: var(--light-theme-gray);
     }
 }
+
+/* ----------------------------------------------
+  Copy button for chat messages
+---------------------------------------------- */
+.message .text,
+.message .text-you,
+.message .text-bot,
+.user-message .text,
+.assistant-message .text {
+    position: relative;
+}
+
+.message, .user-message, .assistant-message {
+    position: relative;
+}
+
+.copy-button {
+    position: absolute;
+    bottom: -23px;
+    left: 0;
+    padding: 0;
+    border: none;
+    border-radius: 3px;
+    cursor: pointer;
+    opacity: 0;
+    display: flex;
+    align-items: center;
+    transition: opacity 0.2s;
+}
+
+.message:hover .copy-button,
+.user-message:hover .copy-button,
+.assistant-message:hover .copy-button {
+    opacity: 1;
+}
+
+.copy-button svg {
+    stroke: rgb(156 163 175);
+    transition: stroke 0.2s;
+}
+
+.copy-button:hover svg {
+    stroke: rgb(107 114 128);
+}
+
+.dark .copy-button svg {
+    stroke: rgb(156 163 175);
+}
+
+.dark .copy-button:hover svg {
+    stroke: rgb(209 213 219);
+}
diff --git a/modules/block_requests.py b/modules/block_requests.py
index 35f983cf55..29fc66336d 100644
--- a/modules/block_requests.py
+++ b/modules/block_requests.py
@@ -3,7 +3,7 @@
 
 import requests
 
-from modules import shared
+from modules import shared, ui
 from modules.logging_colors import logger
 
 original_open = open
@@ -58,6 +58,7 @@ def my_open(*args, **kwargs):
             '\n    '
             f'\n    '
             '\n    '
+            f'\n    '
             '\n  '
         )
 
diff --git a/modules/html_generator.py b/modules/html_generator.py
index e3550ed5a4..b565c63af5 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -292,24 +292,34 @@ def get_image_cache(path):
     return image_cache[path][1]
 
 
+copy_svg = ''''''
+copy_button = f''
+
 def generate_instruct_html(history):
     output = f'
            '
-    for i, _row in enumerate(history):
-        row = [convert_to_markdown_wrapped(entry, use_cache=i != len(history) - 1) for entry in _row]
 
-        if row[0]:  # Don't display empty user messages
+    for i in range(len(history['visible'])):
+        row_visible = history['visible'][i]
+        row_internal = history['internal'][i]
+        converted_visible = [convert_to_markdown_wrapped(entry, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
+
+        if converted_visible[0]:  # Don't display empty user messages
             output += (
-                f'
            '
+                f'
            '
                 f'
            '
-                f'
            {row[0]}
            '
+                f'
            {converted_visible[0]}
            '
+                f'{copy_button}'
                 f'
            '
                 f'
            '
             )
 
         output += (
-            f'
            '
+            f'
            '
             f'
            '
-            f'
            {row[1]}
            '
+            f'
            {converted_visible[1]}
            '
+            f'{copy_button}'
             f'
            '
             f'
            '
         )
@@ -332,26 +342,32 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
         if Path("cache/pfp_me.png").exists() else ''
     )
 
-    for i, _row in enumerate(history):
-        row = [convert_to_markdown_wrapped(entry, use_cache=i != len(history) - 1) for entry in _row]
+    for i in range(len(history['visible'])):
+        row_visible = history['visible'][i]
+        row_internal = history['internal'][i]
+        converted_visible = [convert_to_markdown_wrapped(entry, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
 
-        if row[0]:  # Don't display empty user messages
+        if converted_visible[0]:  # Don't display empty user messages
             output += (
-                f'
            '
+                f'
            '
                 f'
            {img_me}
            '
                 f'
            '
                 f'
            {name1}
            '
-                f'
            {row[0]}
            '
+                f'
            {converted_visible[0]}
            '
+                f'{copy_button}'
                 f'
            '
                 f'
            '
             )
 
         output += (
-            f'
            '
+            f'
            '
             f'
            {img_bot}
            '
             f'
            '
             f'
            {name2}
            '
-            f'
            {row[1]}
            '
+            f'
            {converted_visible[1]}
            '
+            f'{copy_button}'
             f'
            '
             f'
            '
         )
@@ -363,22 +379,28 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
 def generate_chat_html(history, name1, name2, reset_cache=False):
     output = f'
            '
 
-    for i, _row in enumerate(history):
-        row = [convert_to_markdown_wrapped(entry, use_cache=i != len(history) - 1) for entry in _row]
+    for i in range(len(history['visible'])):
+        row_visible = history['visible'][i]
+        row_internal = history['internal'][i]
+        converted_visible = [convert_to_markdown_wrapped(entry, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
 
-        if row[0]:  # Don't display empty user messages
+        if converted_visible[0]:  # Don't display empty user messages
             output += (
-                f'
            '
+                f'
            '
                 f'
            '
-                f'
            {row[0]}
            '
+                f'
            {converted_visible[0]}
            '
+                f'{copy_button}'
                 f'
            '
                 f'
            '
             )
 
         output += (
-            f'
            '
+            f'
            '
             f'
            '
-            f'
            {row[1]}
            '
+            f'
            {converted_visible[1]}
            '
+            f'{copy_button}'
             f'
            '
             f'
            '
         )
@@ -389,8 +411,8 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
 
 def chat_html_wrapper(history, name1, name2, mode, style, character, reset_cache=False):
     if mode == 'instruct':
-        return generate_instruct_html(history['visible'])
+        return generate_instruct_html(history)
     elif style == 'wpp':
-        return generate_chat_html(history['visible'], name1, name2)
+        return generate_chat_html(history, name1, name2)
     else:
-        return generate_cai_chat_html(history['visible'], name1, name2, style, character, reset_cache)
+        return generate_cai_chat_html(history, name1, name2, style, character, reset_cache)
diff --git a/modules/ui.py b/modules/ui.py
index 4f7ee785ac..df948a1428 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -19,6 +19,8 @@
     css += f.read()
 with open(Path(__file__).resolve().parent / '../js/main.js', 'r') as f:
     js = f.read()
+with open(Path(__file__).resolve().parent / '../js/global_scope_js.js', 'r') as f:
+    global_scope_js = f.read()
 with open(Path(__file__).resolve().parent / '../js/save_files.js', 'r') as f:
     save_files_js = f.read()
 with open(Path(__file__).resolve().parent / '../js/switch_tabs.js', 'r') as f:

From 1b9121e5b87625edbbc13d2dc0e42624173553ca Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 11 Jan 2025 12:41:41 -0800
Subject: [PATCH 0175/1701] Add a "refresh" button below the last message, add
 a missing file

---
 css/main.css              | 29 +++++++++++++++++++----------
 js/global_scope_js.js     | 23 +++++++++++++++++++++++
 modules/html_generator.py |  7 ++++++-
 3 files changed, 48 insertions(+), 11 deletions(-)
 create mode 100644 js/global_scope_js.js

diff --git a/css/main.css b/css/main.css
index 48c6727a1b..6368197919 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1236,11 +1236,10 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     position: relative;
 }
 
-.copy-button {
+.footer-button {
     position: absolute;
-    bottom: -23px;
-    left: 0;
     padding: 0;
+    margin: 0;
     border: none;
     border-radius: 3px;
     cursor: pointer;
@@ -1250,25 +1249,35 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     transition: opacity 0.2s;
 }
 
-.message:hover .copy-button,
-.user-message:hover .copy-button,
-.assistant-message:hover .copy-button {
+.footer-button#copy-button {
+    bottom: -23px;
+    left: 0;
+}
+
+.footer-button#refresh-button {
+    bottom: -23px;
+    left: 25px;
+}
+
+.message:hover .footer-button,
+.user-message:hover .footer-button,
+.assistant-message:hover .footer-button {
     opacity: 1;
 }
 
-.copy-button svg {
+.footer-button svg {
     stroke: rgb(156 163 175);
     transition: stroke 0.2s;
 }
 
-.copy-button:hover svg {
+.footer-button:hover svg {
     stroke: rgb(107 114 128);
 }
 
-.dark .copy-button svg {
+.dark .footer-button svg {
     stroke: rgb(156 163 175);
 }
 
-.dark .copy-button:hover svg {
+.dark .footer-button:hover svg {
     stroke: rgb(209 213 219);
 }
diff --git a/js/global_scope_js.js b/js/global_scope_js.js
new file mode 100644
index 0000000000..79b673d794
--- /dev/null
+++ b/js/global_scope_js.js
@@ -0,0 +1,23 @@
+function copyToClipboard(element) {
+  if (!element) return;
+
+  const messageElement = element.closest(".message, .user-message, .assistant-message");
+  if (!messageElement) return;
+
+  const rawText = messageElement.getAttribute("data-raw");
+  if (!rawText) return;
+
+  navigator.clipboard.writeText(rawText).then(function() {
+    const originalSvg = element.innerHTML;
+    element.innerHTML = "";
+    setTimeout(() => {
+      element.innerHTML = originalSvg;
+    }, 1000);
+  }).catch(function(err) {
+    console.error("Failed to copy text: ", err);
+  });
+}
+
+function regenerateClick() {
+    document.getElementById("Regenerate").click();
+}
diff --git a/modules/html_generator.py b/modules/html_generator.py
index b565c63af5..79a8dc6487 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -293,7 +293,9 @@ def get_image_cache(path):
 
 
 copy_svg = ''''''
-copy_button = f''
+refresh_svg = ''''''
+copy_button = f''
+refresh_button = f''
 
 def generate_instruct_html(history):
     output = f'
            '
@@ -320,6 +322,7 @@ def generate_instruct_html(history):
             f'
            '
             f'
            {converted_visible[1]}
            '
             f'{copy_button}'
+            f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
             f'
            '
             f'
            '
         )
@@ -368,6 +371,7 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
             f'
            {name2}
            '
             f'
            {converted_visible[1]}
            '
             f'{copy_button}'
+            f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
             f'
            '
             f'
            '
         )
@@ -401,6 +405,7 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
             f'
            '
             f'
            {converted_visible[1]}
            '
             f'{copy_button}'
+            f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
             f'
            '
             f'
            '
         )

From f1797f4323b6eba98521d14c1cc011ce45f4db42 Mon Sep 17 00:00:00 2001
From: mamei16 
Date: Sat, 11 Jan 2025 22:39:44 +0100
Subject: [PATCH 0176/1701] Unescape backslashes in html_output (#6648)

---
 modules/html_generator.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 79a8dc6487..3ddad51d33 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -239,6 +239,9 @@ def convert_to_markdown(string):
     pattern = re.compile(r']*>(.*?)', re.DOTALL)
     html_output = pattern.sub(lambda x: html.unescape(x.group()), html_output)
 
+    # Unescape backslashes
+    html_output = html_output.replace('\\\\', '\\')
+
     # Add "long-list" class to 
               or 
                 containing a long 
              1.  item
     html_output = add_long_list_class(html_output)
 

From a0492ce325b951a9c000fa3cad45806adc8d8926 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Sat, 11 Jan 2025 21:14:10 -0300
Subject: [PATCH 0177/1701] Optimize syntax highlighting during chat streaming
 (#6655)

---
 js/global_scope_js.js |  2 +-
 js/main.js            | 57 +++++++++++++++----------------------------
 modules/ui_chat.py    | 40 ++++++++++++++++++------------
 3 files changed, 45 insertions(+), 54 deletions(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 79b673d794..983d60f14b 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -19,5 +19,5 @@ function copyToClipboard(element) {
 }
 
 function regenerateClick() {
-    document.getElementById("Regenerate").click();
+  document.getElementById("Regenerate").click();
 }
diff --git a/js/main.js b/js/main.js
index ab2499d4b4..c5c47d0472 100644
--- a/js/main.js
+++ b/js/main.js
@@ -177,47 +177,30 @@ function isElementVisibleOnScreen(element) {
   );
 }
 
-function getVisibleMessagesIndexes() {
-  const elements = document.querySelectorAll(".message-body");
-  const visibleIndexes = [];
-
-  elements.forEach((element, index) => {
-    if (isElementVisibleOnScreen(element) && !element.hasAttribute("data-highlighted")) {
-      visibleIndexes.push(index);
-    }
-  });
-
-  return visibleIndexes;
-}
-
 function doSyntaxHighlighting() {
-  const indexes = getVisibleMessagesIndexes();
-  const elements = document.querySelectorAll(".message-body");
+  const messageBodies = document.querySelectorAll(".message-body");
 
-  if (indexes.length > 0) {
+  if (messageBodies.length > 0) {
     observer.disconnect();
 
-    indexes.forEach((index) => {
-      const element = elements[index];
-
-      // Tag this element to prevent it from being highlighted twice
-      element.setAttribute("data-highlighted", "true");
-
-      // Perform syntax highlighting
-      const codeBlocks = element.querySelectorAll("pre code");
-
-      codeBlocks.forEach((codeBlock) => {
-        hljs.highlightElement(codeBlock);
-      });
-
-      renderMathInElement(element, {
-        delimiters: [
-          { left: "$$", right: "$$", display: true },
-          { left: "$", right: "$", display: false },
-          { left: "\\(", right: "\\)", display: false },
-          { left: "\\[", right: "\\]", display: true },
-        ],
-      });
+    messageBodies.forEach((messageBody) => {
+      if (isElementVisibleOnScreen(messageBody)) {
+        // Handle both code and math in a single pass through each message
+        const codeBlocks = messageBody.querySelectorAll("pre code:not([data-highlighted])");
+        codeBlocks.forEach((codeBlock) => {
+          hljs.highlightElement(codeBlock);
+          codeBlock.setAttribute("data-highlighted", "true");
+        });
+
+        renderMathInElement(messageBody, {
+          delimiters: [
+            { left: "$$", right: "$$", display: true },
+            { left: "$", right: "$", display: false },
+            { left: "\\(", right: "\\)", display: false },
+            { left: "\\[", right: "\\]", display: true },
+          ],
+        });
+      }
     });
 
     observer.observe(targetElement, config);
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 61be17e3c9..8497f7df57 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -182,23 +182,31 @@ def create_event_handlers():
 
     # Morph HTML updates instead of updating everything
     shared.gradio['display'].change(None, gradio('display'), None,
-        js="""
-            (text) => {
-                morphdom(
-                    document.getElementById('chat').parentNode,
-                    '
                ' + text + '
                ',
-                    {
-                        onBeforeElUpdated: function(fromEl, toEl) {
-                            if (fromEl.isEqualNode(toEl)) {
-                                return false; // Skip identical nodes
-                            }
-                            return true; // Update only if nodes differ
-                        }
-                    }
-                );
+      js="""
+      (text) => {
+        morphdom(
+          document.getElementById('chat').parentNode,
+          '
                ' + text + '
                ',
+          {
+            onBeforeElUpdated: function(fromEl, toEl) {
+              if (fromEl.tagName === 'PRE' && fromEl.querySelector('code[data-highlighted]')) {
+                const fromCode = fromEl.querySelector('code');
+                const toCode = toEl.querySelector('code');
+
+                if (fromCode && toCode && fromCode.textContent === toCode.textContent) {
+                  // If the  content is the same, preserve the entire 
                 element
+                  toEl.className = fromEl.className;
+                  toEl.innerHTML = fromEl.innerHTML;
+                  return false; // Skip updating the 
                 element
+                }
+              }
+              return !fromEl.isEqualNode(toEl); // Update only if nodes differ
             }
-        """
-    )
+          }
+        );
+      }
+      """
+    );
 
     shared.gradio['Generate'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(

From ed16374ecee6f547c49fe4af0630294add317399 Mon Sep 17 00:00:00 2001
From: Lounger <4087076+TheLounger@users.noreply.github.com>
Date: Sun, 12 Jan 2025 03:35:22 +0100
Subject: [PATCH 0178/1701] Fix the gallery extension (#6656)

---
 extensions/gallery/script.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/extensions/gallery/script.py b/extensions/gallery/script.py
index ff0242c8fb..54f9c74593 100644
--- a/extensions/gallery/script.py
+++ b/extensions/gallery/script.py
@@ -93,10 +93,11 @@ def generate_html():
 
 def filter_cards(filter_str=''):
     if filter_str == '':
-        return cards
+        return gr.Dataset(samples=cards)
 
     filter_upper = filter_str.upper()
-    return [k for k in cards if filter_upper in k[1].upper()]
+    filtered = [k for k in cards if filter_upper in k[1].upper()]
+    return gr.Dataset(samples=filtered)
 
 
 def select_character(evt: gr.SelectData):

From facb4155d4a0d343b6f0cbae93f112456f20875b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 11 Jan 2025 20:57:28 -0800
Subject: [PATCH 0179/1701] Fix morphdom leaving ghost elements behind

---
 css/main.css              | 4 ++--
 modules/html_generator.py | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/css/main.css b/css/main.css
index 6368197919..1a7efe70ef 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1249,12 +1249,12 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     transition: opacity 0.2s;
 }
 
-.footer-button#copy-button {
+.footer-button.footer-copy-button {
     bottom: -23px;
     left: 0;
 }
 
-.footer-button#refresh-button {
+.footer-button.footer-refresh-button {
     bottom: -23px;
     left: 25px;
 }
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 3ddad51d33..245c833ce8 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -297,8 +297,8 @@ def get_image_cache(path):
 
 copy_svg = ''''''
 refresh_svg = ''''''
-copy_button = f''
-refresh_button = f''
+copy_button = f''
+refresh_button = f''
 
 def generate_instruct_html(history):
     output = f'
                '

From c85e5e58d08a18e86e94106740d482250b4c0594 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 13 Jan 2025 06:20:42 -0800
Subject: [PATCH 0180/1701] UI: move the new morphdom code to a .js file

---
 js/global_scope_js.js | 24 ++++++++++++++++++++++++
 modules/ui_chat.py    | 27 +--------------------------
 2 files changed, 25 insertions(+), 26 deletions(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 983d60f14b..f4d9c67376 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -21,3 +21,27 @@ function copyToClipboard(element) {
 function regenerateClick() {
   document.getElementById("Regenerate").click();
 }
+
+function handleMorphdomUpdate(text) {
+  console.log("Morphing!");
+  morphdom(
+    document.getElementById("chat").parentNode,
+    "
                " + text + "
                ",
+    {
+      onBeforeElUpdated: function(fromEl, toEl) {
+        if (fromEl.tagName === "PRE" && fromEl.querySelector("code[data-highlighted]")) {
+          const fromCode = fromEl.querySelector("code");
+          const toCode = toEl.querySelector("code");
+
+          if (fromCode && toCode && fromCode.textContent === toCode.textContent) {
+            // If the  content is the same, preserve the entire 
                 element
+            toEl.className = fromEl.className;
+            toEl.innerHTML = fromEl.innerHTML;
+            return false; // Skip updating the 
                 element
+          }
+        }
+        return !fromEl.isEqualNode(toEl); // Update only if nodes differ
+      }
+    }
+  );
+}
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 8497f7df57..e80fa33bc5 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -181,32 +181,7 @@ def create_event_handlers():
     shared.reload_inputs = gradio(reload_arr)
 
     # Morph HTML updates instead of updating everything
-    shared.gradio['display'].change(None, gradio('display'), None,
-      js="""
-      (text) => {
-        morphdom(
-          document.getElementById('chat').parentNode,
-          '
                ' + text + '
                ',
-          {
-            onBeforeElUpdated: function(fromEl, toEl) {
-              if (fromEl.tagName === 'PRE' && fromEl.querySelector('code[data-highlighted]')) {
-                const fromCode = fromEl.querySelector('code');
-                const toCode = toEl.querySelector('code');
-
-                if (fromCode && toCode && fromCode.textContent === toCode.textContent) {
-                  // If the  content is the same, preserve the entire 
                 element
-                  toEl.className = fromEl.className;
-                  toEl.innerHTML = fromEl.innerHTML;
-                  return false; // Skip updating the 
                 element
-                }
-              }
-              return !fromEl.isEqualNode(toEl); // Update only if nodes differ
-            }
-          }
-        );
-      }
-      """
-    );
+    shared.gradio['display'].change(None, gradio('display'), None, js="(text) => handleMorphdomUpdate(text)")
 
     shared.gradio['Generate'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(

From 53b838d6c5637406533371cef3b440549f43f4de Mon Sep 17 00:00:00 2001
From: Underscore <47636331+Th-Underscore@users.noreply.github.com>
Date: Mon, 13 Jan 2025 16:01:50 -0500
Subject: [PATCH 0181/1701] HTML: Fix quote pair RegEx matching for all quote
 types (#6661)

---
 modules/html_generator.py | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 245c833ce8..c14a28b42f 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -73,7 +73,6 @@ def fix_newlines(string):
 
 
 def replace_quotes(text):
-
     # Define a list of quote pairs (opening and closing), using HTML entities
     quote_pairs = [
         ('"', '"'),  # Double quotes
@@ -84,14 +83,22 @@ def replace_quotes(text):
         ('‘', '’'),  # Alternative single quotes
         ('“', '”'),  # Unicode quotes (numeric entities)
         ('“', '”'),  # Unicode quotes (hex entities)
+        ('\u201C', '\u201D'),  # Unicode quotes (literal chars)
     ]
 
     # Create a regex pattern that matches any of the quote pairs, including newlines
     pattern = '|'.join(f'({re.escape(open_q)})(.*?)({re.escape(close_q)})' for open_q, close_q in quote_pairs)
 
     # Replace matched patterns with  tags, keeping original quotes
-    replaced_text = re.sub(pattern, lambda m: f'{m.group(1)}{m.group(2)}{m.group(3)}', text, flags=re.DOTALL)
+    def replacer(m):
+        # Find the first non-None group set
+        for i in range(1, len(m.groups()), 3):  # Step through each sub-pattern's groups
+            if m.group(i):  # If this sub-pattern matched
+                return f'{m.group(i)}{m.group(i + 1)}{m.group(i + 2)}'
+
+        return m.group(0)  # Fallback (shouldn't happen)
 
+    replaced_text = re.sub(pattern, replacer, text, flags=re.DOTALL)
     return replaced_text
 
 

From c832953ff723aa5dafcb3aac4f19acd0df56bb5b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Jan 2025 05:59:55 -0800
Subject: [PATCH 0182/1701] UI: Activate auto_max_new_tokens by default

---
 modules/shared.py      | 2 +-
 settings-template.yaml | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 93cd227232..f1e126736f 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -47,7 +47,7 @@
     'prompt_lookup_num_tokens': 0,
     'max_tokens_second': 0,
     'max_updates_second': 0,
-    'auto_max_new_tokens': False,
+    'auto_max_new_tokens': True,
     'ban_eos_token': False,
     'add_bos_token': True,
     'skip_special_tokens': True,
diff --git a/settings-template.yaml b/settings-template.yaml
index b61dc4e01a..93a64abbeb 100644
--- a/settings-template.yaml
+++ b/settings-template.yaml
@@ -19,7 +19,7 @@ max_new_tokens_max: 4096
 prompt_lookup_num_tokens: 0
 max_tokens_second: 0
 max_updates_second: 0
-auto_max_new_tokens: false
+auto_max_new_tokens: true
 ban_eos_token: false
 add_bos_token: true
 skip_special_tokens: true

From f843cb475bd3b880838a6a3a6ff200a2e290b115 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Jan 2025 08:12:51 -0800
Subject: [PATCH 0183/1701] UI: update a help message

---
 modules/ui_chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index e80fa33bc5..395300d0be 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -164,7 +164,7 @@ def create_chat_settings_ui():
         with gr.Row():
             with gr.Column():
                 shared.gradio['custom_system_message'] = gr.Textbox(value=shared.settings['custom_system_message'], lines=2, label='Custom system message', info='If not empty, will be used instead of the default one.', elem_classes=['add_scrollbar'])
-                shared.gradio['instruction_template_str'] = gr.Textbox(value='', label='Instruction template', lines=24, info='Change this according to the model/LoRA that you are using. Used in instruct and chat-instruct modes.', elem_classes=['add_scrollbar', 'monospace'])
+                shared.gradio['instruction_template_str'] = gr.Textbox(value='', label='Instruction template', lines=24, info='This gets autodetected; you usually don\'t need to change it. Used in instruct and chat-instruct modes.', elem_classes=['add_scrollbar', 'monospace'])
                 with gr.Row():
                     shared.gradio['send_instruction_to_default'] = gr.Button('Send to default', elem_classes=['small-button'])
                     shared.gradio['send_instruction_to_notebook'] = gr.Button('Send to notebook', elem_classes=['small-button'])

From 1ef748fb203730aae92b8f28f44abb68699accb4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Jan 2025 16:44:15 -0800
Subject: [PATCH 0184/1701] Lint

---
 extensions/gallery/script.py | 1 -
 modules/html_generator.py    | 1 +
 2 files changed, 1 insertion(+), 1 deletion(-)

diff --git a/extensions/gallery/script.py b/extensions/gallery/script.py
index 54f9c74593..76be4a5838 100644
--- a/extensions/gallery/script.py
+++ b/extensions/gallery/script.py
@@ -5,7 +5,6 @@
 from modules.html_generator import get_image_cache
 from modules.shared import gradio
 
-
 params = {
     'items_per_page': 50,
     'open': False,
diff --git a/modules/html_generator.py b/modules/html_generator.py
index c14a28b42f..299734121f 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -307,6 +307,7 @@ def get_image_cache(path):
 copy_button = f''
 refresh_button = f''
 
+
 def generate_instruct_html(history):
     output = f'
                '
 

From 5d257397678e03694ded7eca2a9639d04368039b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Jan 2025 16:59:36 -0800
Subject: [PATCH 0185/1701] Make the update wizards nice

---
 update_wizard_linux.sh    | 2 +-
 update_wizard_macos.sh    | 2 +-
 update_wizard_windows.bat | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/update_wizard_linux.sh b/update_wizard_linux.sh
index 3ada9a1e47..c81d9d9b20 100755
--- a/update_wizard_linux.sh
+++ b/update_wizard_linux.sh
@@ -23,4 +23,4 @@ source "$CONDA_ROOT_PREFIX/etc/profile.d/conda.sh" # otherwise conda complains a
 conda activate "$INSTALL_ENV_DIR"
 
 # update installer env
-python one_click.py --update-wizard && echo -e "\nDone!"
+python one_click.py --update-wizard && echo -e "\nHave a great day!"
diff --git a/update_wizard_macos.sh b/update_wizard_macos.sh
index c5add61ecc..f58bb9e94b 100755
--- a/update_wizard_macos.sh
+++ b/update_wizard_macos.sh
@@ -23,4 +23,4 @@ source "$CONDA_ROOT_PREFIX/etc/profile.d/conda.sh" # otherwise conda complains a
 conda activate "$INSTALL_ENV_DIR"
 
 # update installer env
-python one_click.py --update-wizard && echo -e "\nDone!"
+python one_click.py --update-wizard && echo -e "\nHave a great day!"
diff --git a/update_wizard_windows.bat b/update_wizard_windows.bat
index 2b23f322f1..fac251a764 100755
--- a/update_wizard_windows.bat
+++ b/update_wizard_windows.bat
@@ -30,7 +30,7 @@ call "%CONDA_ROOT_PREFIX%\condabin\conda.bat" activate "%INSTALL_ENV_DIR%" || (
 @rem update installer env
 call python one_click.py --update-wizard && (
     echo.
-    echo Done!
+    echo Have a great day!
 )
 
 :end

From 2344366c9b4d3869bfc24d4b2d79efa5c183450d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Jan 2025 17:23:44 -0800
Subject: [PATCH 0186/1701] Remove a debug message

---
 js/global_scope_js.js | 1 -
 1 file changed, 1 deletion(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index f4d9c67376..6bf0f0e3b3 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -23,7 +23,6 @@ function regenerateClick() {
 }
 
 function handleMorphdomUpdate(text) {
-  console.log("Morphing!");
   morphdom(
     document.getElementById("chat").parentNode,
     "
                " + text + "
                ",

From fe96678692a834783e4df523875a1ac6b998a18d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Jan 2025 19:28:48 -0800
Subject: [PATCH 0187/1701] Update some comments in the requirements

---
 requirements.txt        | 4 ++--
 requirements_noavx2.txt | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 9bc5956ab2..188724312d 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -37,13 +37,13 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
-# llama-cpp-python (CUDA, no tensor cores)
+# llama-cpp-python (CUDA, with GGML_CUDA_FORCE_MMQ)
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
-# llama-cpp-python (CUDA, tensor cores)
+# llama-cpp-python (CUDA, without GGML_CUDA_FORCE_MMQ)
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 1755ac24cb..69e497e0a8 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -37,13 +37,13 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
-# llama-cpp-python (CUDA, no tensor cores)
+# llama-cpp-python (CUDA, with GGML_CUDA_FORCE_MMQ)
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
-# llama-cpp-python (CUDA, tensor cores)
+# llama-cpp-python (CUDA, without GGML_CUDA_FORCE_MMQ)
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

From 0258a6f87793da90006045a3ed8e601c5b3c3897 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 16 Jan 2025 05:21:18 -0800
Subject: [PATCH 0188/1701] Fix the Google Colab notebook

---
 modules/block_requests.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/block_requests.py b/modules/block_requests.py
index 29fc66336d..5a4b533f74 100644
--- a/modules/block_requests.py
+++ b/modules/block_requests.py
@@ -40,7 +40,7 @@ def my_get(url, **kwargs):
 # Kindly provided by our friend WizardLM-30B
 def my_open(*args, **kwargs):
     filename = str(args[0])
-    if filename.endswith('index.html'):
+    if filename.endswith(('index.html', 'share.html')):
         with original_open(*args, **kwargs) as f:
             file_contents = f.read()
 

From c32f06d62f52dc59f57203a8af088201b5891029 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 17 Jan 2025 07:03:22 -0800
Subject: [PATCH 0189/1701] Update README

---
 .github/FUNDING.yml | 3 +--
 README.md           | 6 +++++-
 2 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/.github/FUNDING.yml b/.github/FUNDING.yml
index 7fd111381d..e2e1621278 100644
--- a/.github/FUNDING.yml
+++ b/.github/FUNDING.yml
@@ -1,2 +1 @@
-github: oobabooga
-ko_fi: oobabooga
+patreon: oobabooga
diff --git a/README.md b/README.md
index 7e2dec95da..4b22f8d533 100644
--- a/README.md
+++ b/README.md
@@ -359,7 +359,7 @@ text-generation-webui
     └── llama-2-13b-chat.Q4_K_M.gguf
 ```
 
-* The remaining model types (like 16-bit transformers models and GPTQ models) are made of several files and must be placed in a subfolder. Example:
+* The remaining model types (like 16-bit Transformers models and EXL2 models) are made of several files and must be placed in a subfolder. Example:
 
 ```
 text-generation-webui
@@ -400,3 +400,7 @@ https://colab.research.google.com/github/oobabooga/text-generation-webui/blob/ma
 ## Acknowledgment
 
 In August 2023, [Andreessen Horowitz](https://a16z.com/) (a16z) provided a generous grant to encourage and support my independent work on this project. I am **extremely** grateful for their trust and recognition.
+
+## ⭐ Featured Patreon Supporters
+
+* [Become the first one!](https://www.patreon.com/oobabooga)

From 096272f49e55357a364ed9016357b97829dae0fd Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 17 Jan 2025 07:03:22 -0800
Subject: [PATCH 0190/1701] Update README

---
 .github/FUNDING.yml | 3 +--
 README.md           | 6 +++++-
 2 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/.github/FUNDING.yml b/.github/FUNDING.yml
index 7fd111381d..e2e1621278 100644
--- a/.github/FUNDING.yml
+++ b/.github/FUNDING.yml
@@ -1,2 +1 @@
-github: oobabooga
-ko_fi: oobabooga
+patreon: oobabooga
diff --git a/README.md b/README.md
index 7e2dec95da..4b22f8d533 100644
--- a/README.md
+++ b/README.md
@@ -359,7 +359,7 @@ text-generation-webui
     └── llama-2-13b-chat.Q4_K_M.gguf
 ```
 
-* The remaining model types (like 16-bit transformers models and GPTQ models) are made of several files and must be placed in a subfolder. Example:
+* The remaining model types (like 16-bit Transformers models and EXL2 models) are made of several files and must be placed in a subfolder. Example:
 
 ```
 text-generation-webui
@@ -400,3 +400,7 @@ https://colab.research.google.com/github/oobabooga/text-generation-webui/blob/ma
 ## Acknowledgment
 
 In August 2023, [Andreessen Horowitz](https://a16z.com/) (a16z) provided a generous grant to encourage and support my independent work on this project. I am **extremely** grateful for their trust and recognition.
+
+## ⭐ Featured Patreon Supporters
+
+* [Become the first one!](https://www.patreon.com/oobabooga)

From f8a5b0bc43ee7dcdb9c3a43f78ddf462daaac59e Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Mon, 20 Jan 2025 17:41:03 -0300
Subject: [PATCH 0191/1701] Update accelerate requirement from ==1.2.* to
 ==1.3.* (#6683)

---
 requirements.txt                 | 2 +-
 requirements_amd.txt             | 2 +-
 requirements_amd_noavx2.txt      | 2 +-
 requirements_apple_intel.txt     | 2 +-
 requirements_apple_silicon.txt   | 2 +-
 requirements_cpu_only.txt        | 2 +-
 requirements_cpu_only_noavx2.txt | 2 +-
 requirements_noavx2.txt          | 2 +-
 requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 188724312d..cb4a93eea2 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,4 +1,4 @@
-accelerate==1.2.*
+accelerate==1.3.*
 bitsandbytes==0.45.*
 colorama
 datasets
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 87ee93d1d9..13616a9260 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -1,4 +1,4 @@
-accelerate==1.2.*
+accelerate==1.3.*
 colorama
 datasets
 einops
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index fa2f5ca745..b1fa3957e2 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.2.*
+accelerate==1.3.*
 colorama
 datasets
 einops
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index e983829542..5c62e0b751 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -1,4 +1,4 @@
-accelerate==1.2.*
+accelerate==1.3.*
 colorama
 datasets
 einops
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index bef02feb7b..93ead21563 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -1,4 +1,4 @@
-accelerate==1.2.*
+accelerate==1.3.*
 colorama
 datasets
 einops
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index 32f1a50a91..f0db20167e 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -1,4 +1,4 @@
-accelerate==1.2.*
+accelerate==1.3.*
 colorama
 datasets
 einops
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 938848bfaf..80d0f03965 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.2.*
+accelerate==1.3.*
 colorama
 datasets
 einops
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 69e497e0a8..b3a1423bf8 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.2.*
+accelerate==1.3.*
 bitsandbytes==0.45.*
 colorama
 datasets
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index a034ee6180..3d6c922fb8 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -1,4 +1,4 @@
-accelerate==1.2.*
+accelerate==1.3.*
 colorama
 datasets
 einops

From ecb5d3c48545a9d3ad41cd34bd77767e93f6ed3b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 21 Jan 2025 08:35:35 -0800
Subject: [PATCH 0192/1701] Installer: do not redownload wheels for each update

---
 one_click.py | 67 +++++++++++++++++++++++++++++-----------------------
 1 file changed, 38 insertions(+), 29 deletions(-)

diff --git a/one_click.py b/one_click.py
index e78a2450b3..4910f8c70b 100644
--- a/one_click.py
+++ b/one_click.py
@@ -101,7 +101,7 @@ def torch_version():
 
 
 def update_pytorch():
-    print_big_message("Checking for PyTorch updates")
+    print_big_message("Checking for PyTorch updates.")
 
     torver = torch_version()
     is_cuda = '+cu' in torver
@@ -343,6 +343,31 @@ def update_requirements(initial_installation=False, pull=True):
         git_creation_cmd = 'git init -b main && git remote add origin https://github.com/oobabooga/text-generation-webui && git fetch && git symbolic-ref refs/remotes/origin/HEAD refs/remotes/origin/main && git reset --hard origin/main && git branch --set-upstream-to=origin/main'
         run_cmd(git_creation_cmd, environment=True, assert_success=True)
 
+    # Detect the requirements file from the PyTorch version
+    torver = torch_version()
+    is_cuda = '+cu' in torver
+    is_cuda118 = '+cu118' in torver  # 2.1.0+cu118
+    is_rocm = '+rocm' in torver  # 2.0.1+rocm5.4.2
+    is_intel = '+cxx11' in torver  # 2.0.1a0+cxx11.abi
+    is_cpu = '+cpu' in torver  # 2.0.1+cpu
+
+    if is_rocm:
+        base_requirements = "requirements_amd" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
+    elif is_cpu or is_intel:
+        base_requirements = "requirements_cpu_only" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
+    elif is_macos():
+        base_requirements = "requirements_apple_" + ("intel" if is_x86_64() else "silicon") + ".txt"
+    else:
+        base_requirements = "requirements" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
+
+    requirements_file = base_requirements
+
+    # Call git pull
+    before_pull_whl_lines = []
+    if os.path.exists(requirements_file):
+        with open(requirements_file, 'r') as f:
+            before_pull_whl_lines = [line for line in f if '.whl' in line]
+
     if pull:
         print_big_message("Updating the local copy of the repository with \"git pull\"")
 
@@ -362,6 +387,11 @@ def update_requirements(initial_installation=False, pull=True):
                 print_big_message(f"File '{file_name}' was updated during 'git pull'. Please run the script again.")
                 exit(1)
 
+    after_pull_whl_lines = []
+    if os.path.exists(requirements_file):
+        with open(requirements_file, 'r') as f:
+            after_pull_whl_lines = [line for line in f if '.whl' in line]
+
     if os.environ.get("INSTALL_EXTENSIONS", "").lower() in ("yes", "y", "true", "1", "t", "on"):
         install_extensions_requirements()
 
@@ -369,30 +399,16 @@ def update_requirements(initial_installation=False, pull=True):
     if not initial_installation:
         update_pytorch()
 
-    # Detect the PyTorch version
-    torver = torch_version()
-    is_cuda = '+cu' in torver
-    is_cuda118 = '+cu118' in torver  # 2.1.0+cu118
-    is_rocm = '+rocm' in torver  # 2.0.1+rocm5.4.2
-    is_intel = '+cxx11' in torver  # 2.0.1a0+cxx11.abi
-    is_cpu = '+cpu' in torver  # 2.0.1+cpu
-
-    if is_rocm:
-        base_requirements = "requirements_amd" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
-    elif is_cpu or is_intel:
-        base_requirements = "requirements_cpu_only" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
-    elif is_macos():
-        base_requirements = "requirements_apple_" + ("intel" if is_x86_64() else "silicon") + ".txt"
-    else:
-        base_requirements = "requirements" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
-
-    requirements_file = base_requirements
-
     print_big_message(f"Installing webui requirements from file: {requirements_file}")
     print(f"TORCH: {torver}\n")
 
     # Prepare the requirements file
     textgen_requirements = open(requirements_file).read().splitlines()
+
+    whl_changed = before_pull_whl_lines != after_pull_whl_lines
+    if not initial_installation and not whl_changed:
+        textgen_requirements = [line for line in textgen_requirements if not '.whl' in line]
+
     if is_cuda118:
         textgen_requirements = [
             req.replace('+cu121', '+cu118').replace('+cu122', '+cu118')
@@ -416,16 +432,9 @@ def update_requirements(initial_installation=False, pull=True):
 
     # Install/update the project requirements
     run_cmd("python -m pip install -r temp_requirements.txt --upgrade", assert_success=True, environment=True)
-    os.remove('temp_requirements.txt')
-
-    # Check for '+cu' or '+rocm' in version string to determine if torch uses CUDA or ROCm. Check for pytorch-cuda as well for backwards compatibility
-    if not any((is_cuda, is_rocm)) and run_cmd("conda list -f pytorch-cuda | grep pytorch-cuda", environment=True, capture_output=True).returncode == 1:
-        clear_cache()
-        return
-
-    if not os.path.exists("repositories/"):
-        os.mkdir("repositories")
 
+    # Clean up
+    os.remove('temp_requirements.txt')
     clear_cache()
 
 

From 5e99dded4e9bae4501127c0bb905a6d27711722d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 21 Jan 2025 09:05:44 -0800
Subject: [PATCH 0193/1701] UI: add "Continue" and "Remove" buttons below the
 last chat message

---
 css/main.css              | 10 ++++++++++
 js/global_scope_js.js     |  8 ++++++++
 modules/html_generator.py | 15 +++++++++++++--
 3 files changed, 31 insertions(+), 2 deletions(-)

diff --git a/css/main.css b/css/main.css
index 1a7efe70ef..b10d19805d 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1259,6 +1259,16 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     left: 25px;
 }
 
+.footer-button.footer-continue-button {
+    bottom: -23px;
+    left: 50px;
+}
+
+.footer-button.footer-remove-button {
+    bottom: -23px;
+    left: 75px;
+}
+
 .message:hover .footer-button,
 .user-message:hover .footer-button,
 .assistant-message:hover .footer-button {
diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 6bf0f0e3b3..f308edb961 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -22,6 +22,14 @@ function regenerateClick() {
   document.getElementById("Regenerate").click();
 }
 
+function continueClick() {
+  document.getElementById("Continue").click();
+}
+
+function removeLastClick() {
+  document.getElementById("Remove-last").click();
+}
+
 function handleMorphdomUpdate(text) {
   morphdom(
     document.getElementById("chat").parentNode,
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 299734121f..c836f6634e 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -304,8 +304,13 @@ def get_image_cache(path):
 
 copy_svg = ''''''
 refresh_svg = ''''''
-copy_button = f''
-refresh_button = f''
+continue_svg = ''''''
+remove_svg = ''''''
+
+copy_button = f''
+refresh_button = f''
+continue_button = f''
+remove_button = f''
 
 
 def generate_instruct_html(history):
@@ -334,6 +339,8 @@ def generate_instruct_html(history):
             f'
                {converted_visible[1]}
                '
             f'{copy_button}'
             f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
+            f'{continue_button if i == len(history["visible"]) - 1 else ""}'
+            f'{remove_button if i == len(history["visible"]) - 1 else ""}'
             f'
                '
             f'
                '
         )
@@ -383,6 +390,8 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
             f'
                {converted_visible[1]}
                '
             f'{copy_button}'
             f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
+            f'{continue_button if i == len(history["visible"]) - 1 else ""}'
+            f'{remove_button if i == len(history["visible"]) - 1 else ""}'
             f'
                '
             f'
                '
         )
@@ -417,6 +426,8 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
             f'
                {converted_visible[1]}
                '
             f'{copy_button}'
             f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
+            f'{continue_button if i == len(history["visible"]) - 1 else ""}'
+            f'{remove_button if i == len(history["visible"]) - 1 else ""}'
             f'
            '
             f'
            '
         )

From 2bf8788c3036f4e83677f41c225af4fa868d9b7a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 21 Jan 2025 09:31:06 -0800
Subject: [PATCH 0194/1701] Installer: Fix a bug after
 ecb5d3c48545a9d3ad41cd34bd77767e93f6ed3b

---
 .gitignore   |  1 +
 one_click.py | 34 +++++++++++++++++++++++-----------
 2 files changed, 24 insertions(+), 11 deletions(-)

diff --git a/.gitignore b/.gitignore
index ca307c4a95..7d1099b669 100644
--- a/.gitignore
+++ b/.gitignore
@@ -32,6 +32,7 @@ venv
 .direnv
 .vs
 .vscode
+.wheels_changed_flag
 *.bak
 *.ipynb
 *.log
diff --git a/one_click.py b/one_click.py
index 4910f8c70b..04a488a07b 100644
--- a/one_click.py
+++ b/one_click.py
@@ -362,13 +362,17 @@ def update_requirements(initial_installation=False, pull=True):
 
     requirements_file = base_requirements
 
-    # Call git pull
-    before_pull_whl_lines = []
-    if os.path.exists(requirements_file):
-        with open(requirements_file, 'r') as f:
-            before_pull_whl_lines = [line for line in f if '.whl' in line]
+    # Call git pull, while checking if .whl requirements have changed
+    wheels_changed_from_flag = False
+    if os.path.exists('.wheels_changed_flag'):
+        os.remove('.wheels_changed_flag')
+        wheels_changed_from_flag = True
 
     if pull:
+        if os.path.exists(requirements_file):
+            with open(requirements_file, 'r') as f:
+                before_pull_whl_lines = [line for line in f if '.whl' in line]
+
         print_big_message("Updating the local copy of the repository with \"git pull\"")
 
         files_to_check = [
@@ -381,16 +385,25 @@ def update_requirements(initial_installation=False, pull=True):
         run_cmd("git pull --autostash", assert_success=True, environment=True)
         after_pull_hashes = {file_name: calculate_file_hash(file_name) for file_name in files_to_check}
 
+        if os.path.exists(requirements_file):
+            with open(requirements_file, 'r') as f:
+                after_pull_whl_lines = [line for line in f if '.whl' in line]
+
         # Check for differences in installation file hashes
         for file_name in files_to_check:
             if before_pull_hashes[file_name] != after_pull_hashes[file_name]:
                 print_big_message(f"File '{file_name}' was updated during 'git pull'. Please run the script again.")
+
+                # Check if wheels changed during this pull
+                wheels_changed = before_pull_whl_lines != after_pull_whl_lines
+                if wheels_changed:
+                    open('.wheels_changed_flag', 'w').close()
+
                 exit(1)
 
-    after_pull_whl_lines = []
-    if os.path.exists(requirements_file):
-        with open(requirements_file, 'r') as f:
-            after_pull_whl_lines = [line for line in f if '.whl' in line]
+    wheels_changed = wheels_changed_from_flag
+    if pull:
+        wheels_changed = wheels_changed or (before_pull_whl_lines != after_pull_whl_lines)
 
     if os.environ.get("INSTALL_EXTENSIONS", "").lower() in ("yes", "y", "true", "1", "t", "on"):
         install_extensions_requirements()
@@ -405,8 +418,7 @@ def update_requirements(initial_installation=False, pull=True):
     # Prepare the requirements file
     textgen_requirements = open(requirements_file).read().splitlines()
 
-    whl_changed = before_pull_whl_lines != after_pull_whl_lines
-    if not initial_installation and not whl_changed:
+    if not initial_installation and not wheels_changed:
         textgen_requirements = [line for line in textgen_requirements if not '.whl' in line]
 
     if is_cuda118:

From ff250dd800e122e3d40a6f0182abafea9bf7da83 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 21 Jan 2025 09:58:13 -0800
Subject: [PATCH 0195/1701] Installer: simplify the script

---
 one_click.py | 109 +++++++++++++++++++++++----------------------------
 1 file changed, 48 insertions(+), 61 deletions(-)

diff --git a/one_click.py b/one_click.py
index 04a488a07b..b0a77b7283 100644
--- a/one_click.py
+++ b/one_click.py
@@ -102,31 +102,24 @@ def torch_version():
 
 def update_pytorch():
     print_big_message("Checking for PyTorch updates.")
-
     torver = torch_version()
-    is_cuda = '+cu' in torver
-    is_cuda118 = '+cu118' in torver  # 2.1.0+cu118
-    is_rocm = '+rocm' in torver  # 2.0.1+rocm5.4.2
-    is_intel = '+cxx11' in torver  # 2.0.1a0+cxx11.abi
-    is_cpu = '+cpu' in torver  # 2.0.1+cpu
-
-    install_pytorch = f"python -m pip install --upgrade torch=={TORCH_VERSION} torchvision=={TORCHVISION_VERSION} torchaudio=={TORCHAUDIO_VERSION} "
-
-    if is_cuda118:
-        install_pytorch += "--index-url https://download.pytorch.org/whl/cu118"
-    elif is_cuda:
-        install_pytorch += "--index-url https://download.pytorch.org/whl/cu121"
-    elif is_rocm:
-        install_pytorch += "--index-url https://download.pytorch.org/whl/rocm6.1"
-    elif is_cpu:
-        install_pytorch += "--index-url https://download.pytorch.org/whl/cpu"
-    elif is_intel:
-        if is_linux():
-            install_pytorch = "python -m pip install --upgrade torch==2.1.0a0 torchvision==0.16.0a0 torchaudio==2.1.0a0 intel-extension-for-pytorch==2.1.10+xpu --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
-        else:
-            install_pytorch = "python -m pip install --upgrade torch==2.1.0a0 torchvision==0.16.0a0 torchaudio==2.1.0a0 intel-extension-for-pytorch==2.1.10 --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
+    base_cmd = f"python -m pip install --upgrade torch=={TORCH_VERSION} torchvision=={TORCHVISION_VERSION} torchaudio=={TORCHAUDIO_VERSION}"
+
+    if "+cu118" in torver:
+        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/cu118"
+    elif "+cu" in torver:
+        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/cu121"
+    elif "+rocm" in torver:
+        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.1"
+    elif "+cpu" in torver:
+        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/cpu"
+    elif "+cxx11" in torver:
+        intel_extension = "intel-extension-for-pytorch==2.1.10+xpu" if is_linux() else "intel-extension-for-pytorch==2.1.10"
+        install_cmd = f"{base_cmd} {intel_extension} --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
+    else:
+        install_cmd = base_cmd
 
-    run_cmd(f"{install_pytorch}", assert_success=True, environment=True)
+    run_cmd(install_cmd, assert_success=True, environment=True)
 
 
 def is_installed():
@@ -340,69 +333,63 @@ def install_extensions_requirements():
 def update_requirements(initial_installation=False, pull=True):
     # Create .git directory if missing
     if not os.path.exists(os.path.join(script_dir, ".git")):
-        git_creation_cmd = 'git init -b main && git remote add origin https://github.com/oobabooga/text-generation-webui && git fetch && git symbolic-ref refs/remotes/origin/HEAD refs/remotes/origin/main && git reset --hard origin/main && git branch --set-upstream-to=origin/main'
-        run_cmd(git_creation_cmd, environment=True, assert_success=True)
+        run_cmd(
+            "git init -b main && git remote add origin https://github.com/oobabooga/text-generation-webui && "
+            "git fetch && git symbolic-ref refs/remotes/origin/HEAD refs/remotes/origin/main && "
+            "git reset --hard origin/main && git branch --set-upstream-to=origin/main",
+            environment=True,
+            assert_success=True
+        )
 
-    # Detect the requirements file from the PyTorch version
     torver = torch_version()
-    is_cuda = '+cu' in torver
-    is_cuda118 = '+cu118' in torver  # 2.1.0+cu118
-    is_rocm = '+rocm' in torver  # 2.0.1+rocm5.4.2
-    is_intel = '+cxx11' in torver  # 2.0.1a0+cxx11.abi
-    is_cpu = '+cpu' in torver  # 2.0.1+cpu
-
-    if is_rocm:
-        base_requirements = "requirements_amd" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
-    elif is_cpu or is_intel:
-        base_requirements = "requirements_cpu_only" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
+    if "+rocm" in torver:
+        requirements_file = "requirements_amd" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
+    elif "+cpu" in torver or "+cxx11" in torver:
+        requirements_file = "requirements_cpu_only" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
     elif is_macos():
-        base_requirements = "requirements_apple_" + ("intel" if is_x86_64() else "silicon") + ".txt"
+        requirements_file = "requirements_apple_" + ("intel" if is_x86_64() else "silicon") + ".txt"
     else:
-        base_requirements = "requirements" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
-
-    requirements_file = base_requirements
+        requirements_file = "requirements" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
 
-    # Call git pull, while checking if .whl requirements have changed
-    wheels_changed_from_flag = False
-    if os.path.exists('.wheels_changed_flag'):
+    # Check and clear the wheels changed flag
+    wheels_changed = os.path.exists('.wheels_changed_flag')
+    if wheels_changed:
         os.remove('.wheels_changed_flag')
-        wheels_changed_from_flag = True
 
     if pull:
+        # Read .whl lines before pulling
+        before_pull_whl_lines = []
         if os.path.exists(requirements_file):
             with open(requirements_file, 'r') as f:
                 before_pull_whl_lines = [line for line in f if '.whl' in line]
 
-        print_big_message("Updating the local copy of the repository with \"git pull\"")
+        print_big_message('Updating the local copy of the repository with "git pull"')
 
+        # Hash files before pulling
         files_to_check = [
             'start_linux.sh', 'start_macos.sh', 'start_windows.bat', 'start_wsl.bat',
             'update_wizard_linux.sh', 'update_wizard_macos.sh', 'update_wizard_windows.bat', 'update_wizard_wsl.bat',
             'one_click.py'
         ]
+        before_hashes = {file: calculate_file_hash(file) for file in files_to_check}
 
-        before_pull_hashes = {file_name: calculate_file_hash(file_name) for file_name in files_to_check}
+        # Perform the git pull
         run_cmd("git pull --autostash", assert_success=True, environment=True)
-        after_pull_hashes = {file_name: calculate_file_hash(file_name) for file_name in files_to_check}
 
+        # Check hashes after pulling
+        after_hashes = {file: calculate_file_hash(file) for file in files_to_check}
         if os.path.exists(requirements_file):
             with open(requirements_file, 'r') as f:
                 after_pull_whl_lines = [line for line in f if '.whl' in line]
 
-        # Check for differences in installation file hashes
-        for file_name in files_to_check:
-            if before_pull_hashes[file_name] != after_pull_hashes[file_name]:
-                print_big_message(f"File '{file_name}' was updated during 'git pull'. Please run the script again.")
-
-                # Check if wheels changed during this pull
-                wheels_changed = before_pull_whl_lines != after_pull_whl_lines
-                if wheels_changed:
+        # Check for changes
+        for file in files_to_check:
+            if before_hashes[file] != after_hashes[file]:
+                print_big_message(f"File '{file}' was updated during 'git pull'. Please run the script again.")
+                if before_pull_whl_lines != after_pull_whl_lines:
                     open('.wheels_changed_flag', 'w').close()
-
                 exit(1)
 
-    wheels_changed = wheels_changed_from_flag
-    if pull:
         wheels_changed = wheels_changed or (before_pull_whl_lines != after_pull_whl_lines)
 
     if os.environ.get("INSTALL_EXTENSIONS", "").lower() in ("yes", "y", "true", "1", "t", "on"):
@@ -419,16 +406,16 @@ def update_requirements(initial_installation=False, pull=True):
     textgen_requirements = open(requirements_file).read().splitlines()
 
     if not initial_installation and not wheels_changed:
-        textgen_requirements = [line for line in textgen_requirements if not '.whl' in line]
+        textgen_requirements = [line for line in textgen_requirements if '.whl' not in line]
 
-    if is_cuda118:
+    if "+cu118" in torver:
         textgen_requirements = [
             req.replace('+cu121', '+cu118').replace('+cu122', '+cu118')
             for req in textgen_requirements
             if "autoawq" not in req.lower()
         ]
 
-    if is_windows() and is_cuda118:  # No flash-attention on Windows for CUDA 11
+    if is_windows() and "+cu118" in torver:  # No flash-attention on Windows for CUDA 11
         textgen_requirements = [req for req in textgen_requirements if 'oobabooga/flash-attention' not in req]
 
     with open('temp_requirements.txt', 'w') as file:

From 41f4fee085a08ab67d2e9d6afe6b570a43fb8e32 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 21 Jan 2025 10:01:52 -0800
Subject: [PATCH 0196/1701] Lint

---
 one_click.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/one_click.py b/one_click.py
index b0a77b7283..54e295011f 100644
--- a/one_click.py
+++ b/one_click.py
@@ -388,6 +388,7 @@ def update_requirements(initial_installation=False, pull=True):
                 print_big_message(f"File '{file}' was updated during 'git pull'. Please run the script again.")
                 if before_pull_whl_lines != after_pull_whl_lines:
                     open('.wheels_changed_flag', 'w').close()
+
                 exit(1)
 
         wheels_changed = wheels_changed or (before_pull_whl_lines != after_pull_whl_lines)

From 079ace63ec1462bd8402a23ce29c67323c9a5e9f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 21 Jan 2025 10:14:05 -0800
Subject: [PATCH 0197/1701] Installer: minor change

---
 start_windows.bat | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/start_windows.bat b/start_windows.bat
index 960cfdb787..2e42d6fa93 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -41,12 +41,12 @@ if "%conda_exists%" == "F" (
 	mkdir "%INSTALL_DIR%"
 	call curl -Lk "%MINICONDA_DOWNLOAD_URL%" > "%INSTALL_DIR%\miniconda_installer.exe" || ( echo. && echo Miniconda failed to download. && goto end )
 
-	:: Try CertUtil first
+	@rem Try CertUtil first
 	for /f %%a in ('CertUtil -hashfile "%INSTALL_DIR%\miniconda_installer.exe" SHA256 ^| find /i /v " " ^| find /i "%MINICONDA_CHECKSUM%"') do (
 		set "output=%%a"
 	)
 
-	:: If CertUtil fails, try PowerShell
+	@rem If CertUtil fails, try PowerShell
 	if not defined output (
 		for /f %%a in ('powershell -Command "if((Get-FileHash \"%INSTALL_DIR%\miniconda_installer.exe\" -Algorithm SHA256).Hash -eq ''%MINICONDA_CHECKSUM%''){echo true}"') do (
 			set "output=%%a"

From 39799adc4739c769e057ce253d31dbd08b0695c6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 21 Jan 2025 11:49:44 -0800
Subject: [PATCH 0198/1701] Add a helpful error message when llama.cpp fails to
 load the model

---
 modules/llamacpp_hf.py    | 14 +++++++++++++-
 modules/llamacpp_model.py | 14 +++++++++++++-
 2 files changed, 26 insertions(+), 2 deletions(-)

diff --git a/modules/llamacpp_hf.py b/modules/llamacpp_hf.py
index f9964fe8b0..b3761e0f53 100644
--- a/modules/llamacpp_hf.py
+++ b/modules/llamacpp_hf.py
@@ -202,7 +202,19 @@ def from_pretrained(cls, pretrained_model_name_or_path: Optional[Union[str, os.P
             params["type_v"] = get_llamacpp_cache_type_for_string(shared.args.cache_type)
 
         Llama = llama_cpp_lib().Llama
-        model = Llama(**params)
+        try:
+            model = Llama(**params)
+        except Exception as e:
+            error_message = (
+                f"Failed loading the model. **This usually happens due to lack of memory**. Try these steps:\n"
+                f"1. Reduce the context length `n_ctx` (currently {shared.args.n_ctx})."
+                f"{' Try a lower value like 4096.' if shared.args.n_ctx > 4096 else '.'}"
+                "\n"
+                f"2. Lower the `n-gpu-layers` value (currently {shared.args.n_gpu_layers})."
+            )
+
+            raise type(e)(error_message) from e
+
         model.last_updated_index = -1
 
         return LlamacppHF(model, model_file)
diff --git a/modules/llamacpp_model.py b/modules/llamacpp_model.py
index c79755e460..db25c66c51 100644
--- a/modules/llamacpp_model.py
+++ b/modules/llamacpp_model.py
@@ -108,7 +108,19 @@ def from_pretrained(self, path):
             params["type_k"] = get_llamacpp_cache_type_for_string(shared.args.cache_type)
             params["type_v"] = get_llamacpp_cache_type_for_string(shared.args.cache_type)
 
-        result.model = Llama(**params)
+        try:
+            result.model = Llama(**params)
+        except Exception as e:
+            error_message = (
+                f"Failed loading the model. **This usually happens due to lack of memory**. Try these steps:\n"
+                f"1. Reduce the context length `n_ctx` (currently {shared.args.n_ctx})."
+                f"{' Try a lower value like 4096.' if shared.args.n_ctx > 4096 else '.'}"
+                "\n"
+                f"2. Lower the `n-gpu-layers` value (currently {shared.args.n_gpu_layers})."
+            )
+
+            raise type(e)(error_message) from e
+
         if cache_capacity > 0:
             result.model.set_cache(LlamaCache(capacity_bytes=cache_capacity))
 

From 4bd260c60d0c3b947b74021a9608c49bd18e5c09 Mon Sep 17 00:00:00 2001
From: FP HAM 
Date: Wed, 22 Jan 2025 10:01:44 -0500
Subject: [PATCH 0199/1701] Give SillyTavern a bit of leaway the way the do
 OpenAI (#6685)

---
 extensions/openai/completions.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 2cefc22bde..0f1f26a80e 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -205,7 +205,7 @@ def convert_history(history):
             else:
                 chat_dialogue.append(['', current_reply])
         elif role == "system":
-            system_message = content
+            system_message += f"\n{content}" if system_message else content
 
     if not user_input_last:
         user_input = ""

From b76b7f6bf5dd2c5d14bf9efaaefb1585c2d48a58 Mon Sep 17 00:00:00 2001
From: Shay Molcho <152275799+shaymolcho@users.noreply.github.com>
Date: Wed, 22 Jan 2025 17:02:43 +0200
Subject: [PATCH 0200/1701] Minor README change (#6687)

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 4b22f8d533..407fdff0d0 100644
--- a/README.md
+++ b/README.md
@@ -380,7 +380,7 @@ text-generation-webui
 │   │   └── tokenizer.model
 ```
 
-In both cases, you can use the "Model" tab of the UI to download the model from Hugging Face automatically. It is also possible to download it via the command-line with 
+In both cases, you can use the "Model" tab of the UI to download the model from Hugging Face automatically. It is also possible to download it via the command-line with:
 
 ```
 python download-model.py organization/model

From 7f8c1c1f073f5460633d01e2d201a4ad78fa329d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 22 Jan 2025 08:45:17 -0800
Subject: [PATCH 0201/1701] Docs: update the API examples

---
 docs/12 - OpenAI API.md | 30 +++++++++++++++++++++++-------
 1 file changed, 23 insertions(+), 7 deletions(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index 9b4f89bf17..daabb43eca 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -14,7 +14,7 @@ Add `--api` to your command-line flags.
 * To create a public Cloudflare URL, add the `--public-api` flag.
 * To listen on your local network, add the `--listen` flag.
 * To change the port, which is 5000 by default, use `--api-port 1234` (change 1234 to your desired port number).
-* To use SSL, add `--ssl-keyfile key.pem --ssl-certfile cert.pem`. Note that it doesn't work with `--public-api`.
+* To use SSL, add `--ssl-keyfile key.pem --ssl-certfile cert.pem`. ⚠️ **Note**: this doesn't work with `--public-api` since Cloudflare already uses HTTPS by default.
 * To use an API key for authentication, add `--api-key yourkey`.
 
 ### Examples
@@ -51,8 +51,7 @@ curl http://127.0.0.1:5000/v1/chat/completions \
         "content": "Hello!"
       }
     ],
-    "mode": "instruct",
-    "instruction_template": "Alpaca"
+    "mode": "instruct"
   }'
 ```
 
@@ -86,7 +85,6 @@ curl http://127.0.0.1:5000/v1/chat/completions \
       }
     ],
     "mode": "instruct",
-    "instruction_template": "Alpaca",
     "stream": true
   }'
 ```
@@ -131,9 +129,6 @@ curl -k http://127.0.0.1:5000/v1/internal/model/load \
     "args": {
       "load_in_4bit": true,
       "n_gpu_layers": 12
-    },
-    "settings": {
-      "instruction_template": "Alpaca"
     }
   }'
 ```
@@ -241,6 +236,27 @@ for event in client.events():
 print()
 ```
 
+#### Python example with API key
+
+Replace
+
+```python
+headers = {
+    "Content-Type": "application/json"
+}
+```
+
+with
+
+```python
+headers = {
+    "Content-Type": "application/json",
+    "Authorization": "Bearer yourPassword123"
+}
+```
+
+in any of the examples above.
+
 ### Environment variables
 
 The following environment variables can be used (they take precedence over everything else):

From 0485ff20e8b7abbd4fba24a5993eed3c11dfa946 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 23 Jan 2025 06:21:40 -0800
Subject: [PATCH 0202/1701] Workaround for convert_to_markdown bug

---
 modules/html_generator.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index c836f6634e..3edbef5e79 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -154,6 +154,8 @@ def process_list(match):
 
 @functools.lru_cache(maxsize=None)
 def convert_to_markdown(string):
+    if not string:
+        return ""
 
     # Make \[ \]  LaTeX equations inline
     pattern = r'^\s*\\\[\s*\n([\s\S]*?)\n\s*\\\]\s*$'

From 5d6f3e6f923b59c8314c72c4d2b8c4c34eccdc5d Mon Sep 17 00:00:00 2001
From: FP HAM 
Date: Fri, 24 Jan 2025 09:23:44 -0500
Subject: [PATCH 0203/1701] Training pro- removed monkeypatch references
 (#6695)

---
 extensions/Training_PRO/script.py | 19 -------------------
 1 file changed, 19 deletions(-)

diff --git a/extensions/Training_PRO/script.py b/extensions/Training_PRO/script.py
index 01bcf67d4c..f553e482d7 100644
--- a/extensions/Training_PRO/script.py
+++ b/extensions/Training_PRO/script.py
@@ -557,12 +557,6 @@ def calc_trainable_parameters(model):
 
 def do_train(lora_name: str, always_override: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, raw_text_file: str, higher_rank_limit: bool, warmup_steps: int, optimizer: str, hard_cut_string: str, train_only_after: str, stop_at_loss: float, add_eos_token: bool, min_chars: int, report_to: str, precize_slicing_overlap: bool, add_eos_token_type: str, save_steps_under_loss: float, add_bos_token: bool, training_projection: str,sliding_window:bool,warmup_ratio:float, grad_accumulation: int,neft_noise_alpha:float):
 
-    if shared.args.monkey_patch:
-        from alpaca_lora_4bit.monkeypatch.peft_tuners_lora_monkey_patch import (
-            replace_peft_model_with_int4_lora_model
-        )
-        replace_peft_model_with_int4_lora_model()
-    
     global train_log_graph
     global WANT_INTERRUPT
     WANT_INTERRUPT = False
@@ -600,10 +594,6 @@ def do_train(lora_name: str, always_override: bool, save_steps: int, micro_batch
 
         time.sleep(5)
 
-    if shared.args.loader == 'GPTQ-for-LLaMa' and not shared.args.monkey_patch:
-        yield "LoRA training with GPTQ-for-LLaMa requires loading with `--monkey-patch`", zero_pd
-        return
-
     if cutoff_len <= 0 or micro_batch_size <= 0 or actual_lr <= 0 or lora_rank <= 0 or lora_alpha <= 0:
         yield "Cannot input zeroes.", zero_pd
         return
@@ -865,15 +855,6 @@ def generate_and_tokenize_prompt(data_point):
         yield traceback.format_exc().replace('\n', '\n\n'), zero_pd
         return
 
-    if shared.args.monkey_patch:
-        from alpaca_lora_4bit.autograd_4bit import Autograd4bitQuantLinear
-        from alpaca_lora_4bit.models import Linear4bitLt
-        for _, m in lora_model.named_modules():
-            if isinstance(m, Autograd4bitQuantLinear) or isinstance(m, Linear4bitLt):
-                if m.is_v1_model:
-                    m.zeros = m.zeros.half()
-                m.scales = m.scales.half()
-
     class Tracked():
         def __init__(self):
             self.current_steps = 0

From 71a551a62247228729b86c912891fe80d8bf4e84 Mon Sep 17 00:00:00 2001
From: FP HAM 
Date: Fri, 24 Jan 2025 09:37:20 -0500
Subject: [PATCH 0204/1701] Add strftime_now to JINJA to sattisfy LLAMA 3.1 and
 3.2 (and granite) (#6692)

---
 modules/chat.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 694c137bd9..60ded0b0cd 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -30,9 +30,13 @@
 )
 from modules.utils import delete_file, get_available_characters, save_file
 
-# Copied from the Transformers library
+def strftime_now(format):
+    return datetime.now().strftime(format)
+
 jinja_env = ImmutableSandboxedEnvironment(trim_blocks=True, lstrip_blocks=True)
 
+jinja_env.globals["strftime_now"] = strftime_now
+
 
 def str_presenter(dumper, data):
     """

From 3d4f3e423c28694d35fdc431e37028c4a201de38 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 25 Jan 2025 07:28:31 -0800
Subject: [PATCH 0205/1701] Downloader: Make progress bars not jump around

Adapted from: https://gist.github.com/NiklasBeierl/13096bfdd8b2084da8c1163dd06f91d3
---
 download-model.py | 165 +++++++++++++++++++++++++++++-----------------
 1 file changed, 103 insertions(+), 62 deletions(-)

diff --git a/download-model.py b/download-model.py
index 8fe94371f2..8ff1d69c77 100644
--- a/download-model.py
+++ b/download-model.py
@@ -14,6 +14,7 @@
 import os
 import re
 import sys
+from multiprocessing import Array
 from pathlib import Path
 from time import sleep
 
@@ -27,9 +28,10 @@
 
 
 class ModelDownloader:
-    def __init__(self, max_retries=5):
+    def __init__(self, max_retries=7):
         self.max_retries = max_retries
         self.session = self.get_session()
+        self._progress_bar_slots = None
 
     def get_session(self):
         session = requests.Session()
@@ -186,73 +188,112 @@ def get_output_folder(self, model, branch, is_lora, is_llamacpp=False, model_dir
         output_folder = Path(base_folder) / output_folder
         return output_folder
 
+    @property
+    def progress_bar_slots(self):
+        if self._progress_bar_slots is None:
+            raise RuntimeError("Progress bar slots not initialized. Start download threads first.")
+
+        return self._progress_bar_slots
+
+    def initialize_progress_bar_slots(self, num_threads):
+        self._progress_bar_slots = Array("B", [0] * num_threads)
+
+    def get_progress_bar_position(self):
+        with self.progress_bar_slots.get_lock():
+            for i in range(len(self.progress_bar_slots)):
+                if self.progress_bar_slots[i] == 0:
+                    self.progress_bar_slots[i] = 1
+                    return i
+
+        return 0  # fallback
+
+    def release_progress_bar_position(self, slot):
+        with self.progress_bar_slots.get_lock():
+            self.progress_bar_slots[slot] = 0
+
     def get_single_file(self, url, output_folder, start_from_scratch=False):
         filename = Path(url.rsplit('/', 1)[1])
         output_path = output_folder / filename
+        progress_bar_position = self.get_progress_bar_position()
 
-        max_retries = 7
+        max_retries = self.max_retries
         attempt = 0
-        while attempt < max_retries:
-            attempt += 1
-            session = self.session
-            headers = {}
-            mode = 'wb'
-
-            try:
-                if output_path.exists() and not start_from_scratch:
-                    # Resume download
-                    r = session.get(url, stream=True, timeout=20)
-                    total_size = int(r.headers.get('content-length', 0))
-                    if output_path.stat().st_size >= total_size:
-                        return
-
-                    headers = {'Range': f'bytes={output_path.stat().st_size}-'}
-                    mode = 'ab'
-
-                with session.get(url, stream=True, headers=headers, timeout=30) as r:
-                    r.raise_for_status()  # If status is not 2xx, raise an error
-                    total_size = int(r.headers.get('content-length', 0))
-                    block_size = 1024 * 1024  # 1MB
-
-                    filename_str = str(filename)  # Convert PosixPath to string if necessary
-
-                    tqdm_kwargs = {
-                        'total': total_size,
-                        'unit': 'B',
-                        'unit_scale': True,
-                        'unit_divisor': 1024,
-                        'bar_format': '{desc}{percentage:3.0f}%|{bar:50}| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, {rate_fmt}]',
-                        'desc': f"{filename_str}: "
-                    }
-
-                    if 'COLAB_GPU' in os.environ:
-                        tqdm_kwargs.update({
-                            'position': 0,
-                            'leave': True
-                        })
-
-                    with open(output_path, mode) as f:
-                        with tqdm.tqdm(**tqdm_kwargs) as t:
-                            count = 0
-                            for data in r.iter_content(block_size):
-                                f.write(data)
-                                t.update(len(data))
-                                if total_size != 0 and self.progress_bar is not None:
-                                    count += len(data)
-                                    self.progress_bar(float(count) / float(total_size), f"{filename_str}")
-
-                    break  # Exit loop if successful
-            except (RequestException, ConnectionError, Timeout) as e:
-                print(f"Error downloading {filename}: {e}.")
-                print(f"That was attempt {attempt}/{max_retries}.", end=' ')
-                if attempt < max_retries:
-                    print(f"Retry begins in {2 ** attempt} seconds.")
-                    sleep(2 ** attempt)
-                else:
-                    print("Failed to download after the maximum number of attempts.")
+        try:
+            while attempt < max_retries:
+                attempt += 1
+                session = self.session
+                headers = {}
+                mode = 'wb'
+
+                try:
+                    if output_path.exists() and not start_from_scratch:
+                        # Resume download
+                        r = session.get(url, stream=True, timeout=20)
+                        total_size = int(r.headers.get('content-length', 0))
+                        if output_path.stat().st_size >= total_size:
+                            return
+
+                        headers = {'Range': f'bytes={output_path.stat().st_size}-'}
+                        mode = 'ab'
+
+                    with session.get(url, stream=True, headers=headers, timeout=30) as r:
+                        r.raise_for_status()  # If status is not 2xx, raise an error
+                        total_size = int(r.headers.get('content-length', 0))
+                        block_size = 1024 * 1024  # 1MB
+
+                        filename_str = str(filename)  # Convert PosixPath to string if necessary
+
+                        tqdm_kwargs = {
+                            'total': total_size,
+                            'unit': 'B',
+                            'unit_scale': True,
+                            'unit_divisor': 1024,
+                            'bar_format': '{desc}{percentage:3.0f}%|{bar:50}| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, {rate_fmt}]',
+                            'desc': f"{filename_str}: ",
+                            'position': progress_bar_position,
+                            'leave': False
+                        }
+
+                        if 'COLAB_GPU' in os.environ:
+                            tqdm_kwargs.update({
+                                'position': 0,
+                                'leave': True
+                            })
+
+                        with open(output_path, mode) as f:
+                            with tqdm.tqdm(**tqdm_kwargs) as t:
+                                count = 0
+                                for data in r.iter_content(block_size):
+                                    f.write(data)
+                                    t.update(len(data))
+                                    if total_size != 0 and self.progress_bar is not None:
+                                        count += len(data)
+                                        self.progress_bar(float(count) / float(total_size), f"{filename_str}")
+
+                        break  # Exit loop if successful
+                except (RequestException, ConnectionError, Timeout) as e:
+                    print(f"Error downloading {filename}: {e}.")
+                    print(f"That was attempt {attempt}/{max_retries}.", end=' ')
+                    if attempt < max_retries:
+                        print(f"Retry begins in {2 ** attempt} seconds.")
+                        sleep(2 ** attempt)
+                    else:
+                        print("Failed to download after the maximum number of attempts.")
+        finally:
+            self.release_progress_bar_position(progress_bar_position)
 
     def start_download_threads(self, file_list, output_folder, start_from_scratch=False, threads=4):
-        thread_map(lambda url: self.get_single_file(url, output_folder, start_from_scratch=start_from_scratch), file_list, max_workers=threads, disable=True)
+        self.initialize_progress_bar_slots(threads)
+        tqdm.tqdm.set_lock(tqdm.tqdm.get_lock())
+        try:
+            thread_map(
+                lambda url: self.get_single_file(url, output_folder, start_from_scratch=start_from_scratch),
+                file_list,
+                max_workers=threads,
+                disable=True
+            )
+        finally:
+            print(f"\nDownload of {len(file_list)} files to {output_folder} completed.")
 
     def download_model_files(self, model, branch, links, sha256, output_folder, progress_bar=None, start_from_scratch=False, threads=4, specific_file=None, is_llamacpp=False):
         self.progress_bar = progress_bar
@@ -318,7 +359,7 @@ def check_model_files(self, model, branch, links, sha256, output_folder):
     parser.add_argument('--model-dir', type=str, default=None, help='Save the model files to a subfolder of this folder instead of the default one (text-generation-webui/models).')
     parser.add_argument('--clean', action='store_true', help='Does not resume the previous download.')
     parser.add_argument('--check', action='store_true', help='Validates the checksums of model files.')
-    parser.add_argument('--max-retries', type=int, default=5, help='Max retries count when get error in download time.')
+    parser.add_argument('--max-retries', type=int, default=7, help='Max retries count when get error in download time.')
     args = parser.parse_args()
 
     branch = args.branch

From 75ff3f381556f80050850c908fd3479cfb6da70d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 25 Jan 2025 08:22:23 -0800
Subject: [PATCH 0206/1701] UI: Mention common context length values

---
 modules/ui_model_menu.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index d5116938df..1264a9fd67 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -89,8 +89,8 @@ def create_ui():
                             shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
                             shared.gradio['n_batch'] = gr.Slider(label="n_batch", minimum=1, maximum=2048, step=1, value=shared.args.n_batch)
                             shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
-                            shared.gradio['n_ctx'] = gr.Number(label="n_ctx", precision=0, step=256, value=shared.args.n_ctx, info='Context length. ⚠️ Lower this value if you can\'t load the model.')
-                            shared.gradio['max_seq_len'] = gr.Number(label='max_seq_len', precision=0, step=256, value=shared.args.max_seq_len, info='Context length. ⚠️ Lower this value if you can\'t load the model.')
+                            shared.gradio['n_ctx'] = gr.Number(label="n_ctx", precision=0, step=256, value=shared.args.n_ctx, info='Context length. ⚠️ Lower this value if you can\'t load the model. Common values: 2048, 4096, 8192, 16384, 32768.')
+                            shared.gradio['max_seq_len'] = gr.Number(label='max_seq_len', precision=0, step=256, value=shared.args.max_seq_len, info='Context length. ⚠️ Lower this value if you can\'t load the model. Common values: 2048, 4096, 8192, 16384, 32768.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q6', 'q4'], value=shared.args.cache_type, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.')
                             shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')

From c49251e95d27c210408080065dc16f6b0b0cd7a7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 25 Jan 2025 15:03:09 -0800
Subject: [PATCH 0207/1701] Installer: change a message

---
 one_click.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/one_click.py b/one_click.py
index 54e295011f..9124e8333a 100644
--- a/one_click.py
+++ b/one_click.py
@@ -400,7 +400,7 @@ def update_requirements(initial_installation=False, pull=True):
     if not initial_installation:
         update_pytorch()
 
-    print_big_message(f"Installing webui requirements from file: {requirements_file}")
+    print_big_message(f"Using requirements file: {requirements_file}")
     print(f"TORCH: {torver}\n")
 
     # Prepare the requirements file

From 87de91dd65e7f720b852d9ba1b75df3d457fa4f0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 25 Jan 2025 18:29:11 -0800
Subject: [PATCH 0208/1701] Docs: fix an API example

---
 docs/12 - OpenAI API.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index daabb43eca..364c6b0926 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -193,7 +193,7 @@ while True:
     assistant_message = ''
     for event in client.events():
         payload = json.loads(event.data)
-        chunk = payload['choices'][0]['message']['content']
+        chunk = payload['choices'][0]['delta']['content']
         assistant_message += chunk
         print(chunk, end='')
 

From 1c9dfa871bd151bccd1ee50b6674f3b878ef4d25 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 26 Jan 2025 18:17:31 -0800
Subject: [PATCH 0209/1701] Revert "Installer: change a message"

This reverts commit c49251e95d27c210408080065dc16f6b0b0cd7a7.
---
 one_click.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/one_click.py b/one_click.py
index 9124e8333a..54e295011f 100644
--- a/one_click.py
+++ b/one_click.py
@@ -400,7 +400,7 @@ def update_requirements(initial_installation=False, pull=True):
     if not initial_installation:
         update_pytorch()
 
-    print_big_message(f"Using requirements file: {requirements_file}")
+    print_big_message(f"Installing webui requirements from file: {requirements_file}")
     print(f"TORCH: {torver}\n")
 
     # Prepare the requirements file

From 053911b6294dae5547e385ef22e69fbd4ad3b57b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 27 Jan 2025 09:07:39 -0800
Subject: [PATCH 0210/1701] Installer: don't ignore .whl requirements if the
 commit has changed

By the user manually switching branches or calling git pull.
---
 one_click.py | 45 ++++++++++++++++++++++++++++++++++++++++-----
 1 file changed, 40 insertions(+), 5 deletions(-)

diff --git a/one_click.py b/one_click.py
index 54e295011f..e1b2be5390 100644
--- a/one_click.py
+++ b/one_click.py
@@ -1,6 +1,7 @@
 import argparse
 import glob
 import hashlib
+import json
 import os
 import platform
 import re
@@ -148,6 +149,11 @@ def check_env():
         sys.exit(1)
 
 
+def get_current_commit():
+    result = run_cmd("git rev-parse HEAD", capture_output=True, environment=True)
+    return result.stdout.decode('utf-8').strip()
+
+
 def clear_cache():
     run_cmd("conda clean -a -y", environment=True)
     run_cmd("python -m pip cache purge", environment=True)
@@ -351,10 +357,21 @@ def update_requirements(initial_installation=False, pull=True):
     else:
         requirements_file = "requirements" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
 
-    # Check and clear the wheels changed flag
-    wheels_changed = os.path.exists('.wheels_changed_flag')
-    if wheels_changed:
-        os.remove('.wheels_changed_flag')
+    # Load state from JSON file
+    state_file = '.installer_state.json'
+    wheels_changed = False
+    if os.path.exists(state_file):
+        with open(state_file, 'r') as f:
+            last_state = json.load(f)
+
+        wheels_changed = last_state.get('wheels_changed', False)
+    else:
+        last_state = {}
+
+    # Check wheels changed from state file and commit differences
+    current_commit = get_current_commit()
+    if last_state.get('last_commit') != current_commit:
+        wheels_changed = True
 
     if pull:
         # Read .whl lines before pulling
@@ -387,12 +404,30 @@ def update_requirements(initial_installation=False, pull=True):
             if before_hashes[file] != after_hashes[file]:
                 print_big_message(f"File '{file}' was updated during 'git pull'. Please run the script again.")
                 if before_pull_whl_lines != after_pull_whl_lines:
-                    open('.wheels_changed_flag', 'w').close()
+                    wheels_changed = True
+
+                # Save state before exiting
+                current_state = {
+                    'last_commit': current_commit,
+                    'wheels_changed': wheels_changed
+                }
+
+                with open(state_file, 'w') as f:
+                    json.dump(current_state, f)
 
                 exit(1)
 
         wheels_changed = wheels_changed or (before_pull_whl_lines != after_pull_whl_lines)
 
+    # Save current state
+    current_state = {
+        'last_commit': current_commit,
+        'wheels_changed': wheels_changed
+    }
+
+    with open(state_file, 'w') as f:
+        json.dump(current_state, f)
+
     if os.environ.get("INSTALL_EXTENSIONS", "").lower() in ("yes", "y", "true", "1", "t", "on"):
         install_extensions_requirements()
 

From 340022d4b09a01ecd62799dcc3b0c65900896cbe Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 27 Jan 2025 10:02:21 -0800
Subject: [PATCH 0211/1701] Fix after previous commit

---
 one_click.py | 23 +++++++++--------------
 1 file changed, 9 insertions(+), 14 deletions(-)

diff --git a/one_click.py b/one_click.py
index e1b2be5390..4d2a457102 100644
--- a/one_click.py
+++ b/one_click.py
@@ -359,20 +359,17 @@ def update_requirements(initial_installation=False, pull=True):
 
     # Load state from JSON file
     state_file = '.installer_state.json'
+    current_commit = get_current_commit()
     wheels_changed = False
     if os.path.exists(state_file):
         with open(state_file, 'r') as f:
             last_state = json.load(f)
 
-        wheels_changed = last_state.get('wheels_changed', False)
+        if 'wheels_changed' in last_state or last_state.get('last_commit') != current_commit:
+            wheels_changed = True
     else:
         last_state = {}
 
-    # Check wheels changed from state file and commit differences
-    current_commit = get_current_commit()
-    if last_state.get('last_commit') != current_commit:
-        wheels_changed = True
-
     if pull:
         # Read .whl lines before pulling
         before_pull_whl_lines = []
@@ -407,10 +404,9 @@ def update_requirements(initial_installation=False, pull=True):
                     wheels_changed = True
 
                 # Save state before exiting
-                current_state = {
-                    'last_commit': current_commit,
-                    'wheels_changed': wheels_changed
-                }
+                current_state = {'last_commit': current_commit}
+                if wheels_changed:
+                    current_state['wheels_changed'] = True
 
                 with open(state_file, 'w') as f:
                     json.dump(current_state, f)
@@ -420,10 +416,9 @@ def update_requirements(initial_installation=False, pull=True):
         wheels_changed = wheels_changed or (before_pull_whl_lines != after_pull_whl_lines)
 
     # Save current state
-    current_state = {
-        'last_commit': current_commit,
-        'wheels_changed': wheels_changed
-    }
+    current_state = {'last_commit': current_commit}
+    if wheels_changed:
+        current_state['wheels_changed'] = True
 
     with open(state_file, 'w') as f:
         json.dump(current_state, f)

From bac652bb1d145b0151ecf8a4c0366fcfd38a1e91 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 27 Jan 2025 10:22:36 -0800
Subject: [PATCH 0212/1701] Another fix

---
 one_click.py | 17 ++++++-----------
 1 file changed, 6 insertions(+), 11 deletions(-)

diff --git a/one_click.py b/one_click.py
index 4d2a457102..d0f5458260 100644
--- a/one_click.py
+++ b/one_click.py
@@ -365,7 +365,7 @@ def update_requirements(initial_installation=False, pull=True):
         with open(state_file, 'r') as f:
             last_state = json.load(f)
 
-        if 'wheels_changed' in last_state or last_state.get('last_commit') != current_commit:
+        if 'wheels_changed' in last_state or last_state.get('last_installed_commit') != current_commit:
             wheels_changed = True
     else:
         last_state = {}
@@ -396,15 +396,15 @@ def update_requirements(initial_installation=False, pull=True):
             with open(requirements_file, 'r') as f:
                 after_pull_whl_lines = [line for line in f if '.whl' in line]
 
-        # Check for changes
+        wheels_changed = wheels_changed or (before_pull_whl_lines != after_pull_whl_lines)
+
+        # Check for changes to installer files
         for file in files_to_check:
             if before_hashes[file] != after_hashes[file]:
                 print_big_message(f"File '{file}' was updated during 'git pull'. Please run the script again.")
-                if before_pull_whl_lines != after_pull_whl_lines:
-                    wheels_changed = True
 
                 # Save state before exiting
-                current_state = {'last_commit': current_commit}
+                current_state = {}
                 if wheels_changed:
                     current_state['wheels_changed'] = True
 
@@ -413,13 +413,8 @@ def update_requirements(initial_installation=False, pull=True):
 
                 exit(1)
 
-        wheels_changed = wheels_changed or (before_pull_whl_lines != after_pull_whl_lines)
-
     # Save current state
-    current_state = {'last_commit': current_commit}
-    if wheels_changed:
-        current_state['wheels_changed'] = True
-
+    current_state = {'last_installed_commit': current_commit}
     with open(state_file, 'w') as f:
         json.dump(current_state, f)
 

From 0b9ab1438dd079ea3ee9175d4c32d2d6f9c9073d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 27 Jan 2025 10:28:59 -0800
Subject: [PATCH 0213/1701] Clean up

---
 one_click.py | 2 --
 1 file changed, 2 deletions(-)

diff --git a/one_click.py b/one_click.py
index d0f5458260..9105e057d1 100644
--- a/one_click.py
+++ b/one_click.py
@@ -367,8 +367,6 @@ def update_requirements(initial_installation=False, pull=True):
 
         if 'wheels_changed' in last_state or last_state.get('last_installed_commit') != current_commit:
             wheels_changed = True
-    else:
-        last_state = {}
 
     if pull:
         # Read .whl lines before pulling

From 39365897550683b3c1ff02e36f1942016a880cca Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 28 Jan 2025 12:53:55 -0800
Subject: [PATCH 0214/1701] Update README

---
 .github/FUNDING.yml | 1 -
 README.md           | 4 ----
 2 files changed, 5 deletions(-)
 delete mode 100644 .github/FUNDING.yml

diff --git a/.github/FUNDING.yml b/.github/FUNDING.yml
deleted file mode 100644
index e2e1621278..0000000000
--- a/.github/FUNDING.yml
+++ /dev/null
@@ -1 +0,0 @@
-patreon: oobabooga
diff --git a/README.md b/README.md
index 407fdff0d0..3642fc58c5 100644
--- a/README.md
+++ b/README.md
@@ -400,7 +400,3 @@ https://colab.research.google.com/github/oobabooga/text-generation-webui/blob/ma
 ## Acknowledgment
 
 In August 2023, [Andreessen Horowitz](https://a16z.com/) (a16z) provided a generous grant to encourage and support my independent work on this project. I am **extremely** grateful for their trust and recognition.
-
-## ⭐ Featured Patreon Supporters
-
-* [Become the first one!](https://www.patreon.com/oobabooga)

From a1c353a4b36d906fd1c88062b6c5e8c67cb74d8f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 28 Jan 2025 12:53:55 -0800
Subject: [PATCH 0215/1701] Update README

---
 .github/FUNDING.yml | 1 -
 README.md           | 4 ----
 2 files changed, 5 deletions(-)
 delete mode 100644 .github/FUNDING.yml

diff --git a/.github/FUNDING.yml b/.github/FUNDING.yml
deleted file mode 100644
index e2e1621278..0000000000
--- a/.github/FUNDING.yml
+++ /dev/null
@@ -1 +0,0 @@
-patreon: oobabooga
diff --git a/README.md b/README.md
index 4b22f8d533..f163333405 100644
--- a/README.md
+++ b/README.md
@@ -400,7 +400,3 @@ https://colab.research.google.com/github/oobabooga/text-generation-webui/blob/ma
 ## Acknowledgment
 
 In August 2023, [Andreessen Horowitz](https://a16z.com/) (a16z) provided a generous grant to encourage and support my independent work on this project. I am **extremely** grateful for their trust and recognition.
-
-## ⭐ Featured Patreon Supporters
-
-* [Become the first one!](https://www.patreon.com/oobabooga)

From 9ddcc91a9166ae5b61f7f390a2483d461de04ab8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 29 Jan 2025 11:20:28 -0800
Subject: [PATCH 0216/1701] Bump llama-cpp-python to 0.3.7

---
 requirements.txt                 | 24 ++++++++++++------------
 requirements_amd.txt             | 12 ++++++------
 requirements_amd_noavx2.txt      |  8 ++++----
 requirements_apple_intel.txt     |  8 ++++----
 requirements_apple_silicon.txt   | 12 ++++++------
 requirements_cpu_only.txt        |  8 ++++----
 requirements_cpu_only_noavx2.txt |  8 ++++----
 requirements_noavx2.txt          | 24 ++++++++++++------------
 8 files changed, 52 insertions(+), 52 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index cb4a93eea2..4ff7a6dfa2 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -32,22 +32,22 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, with GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, without GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # CUDA wheels
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 13616a9260..e30f30eec1 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -31,14 +31,14 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.6+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.6+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.7+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.7+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index b1fa3957e2..15d25caa4e 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -31,10 +31,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # AMD wheels
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 5c62e0b751..b614acf4d7 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -31,8 +31,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp310-cp310-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp310-cp310-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 93ead21563..ca9cc3ac16 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -31,10 +31,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp310-cp310-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.6-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp310-cp310-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index f0db20167e..e9a9790517 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -31,7 +31,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 80d0f03965..c435767655 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -31,7 +31,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index b3a1423bf8..40cbc7b0e5 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -32,22 +32,22 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.6+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, with GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.6+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, without GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.6+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # CUDA wheels
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"

From b7c17727b00e3500ddf5b9263973ec57467557ef Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 29 Jan 2025 13:57:56 -0800
Subject: [PATCH 0217/1701] Update .gitignore

---
 .gitignore | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.gitignore b/.gitignore
index 7d1099b669..318e147d1b 100644
--- a/.gitignore
+++ b/.gitignore
@@ -26,13 +26,13 @@
 .DS_Store
 .eslintrc.js
 .idea
+.installer_state.json
 .venv
 venv
 .envrc
 .direnv
 .vs
 .vscode
-.wheels_changed_flag
 *.bak
 *.ipynb
 *.log

From f01cc079b98e422948000b92ad2e0dd8b3031014 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 29 Jan 2025 14:00:59 -0800
Subject: [PATCH 0218/1701] Lint

---
 extensions/openai/completions.py | 2 +-
 modules/chat.py                  | 3 ++-
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 0f1f26a80e..f1a6064597 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -146,7 +146,7 @@ def convert_history(history):
                 for item in entry['content']:
                     if not isinstance(item, dict):
                         continue
-                    
+
                     image_url = None
                     content = None
                     if item['type'] == 'image_url' and isinstance(item['image_url'], dict):
diff --git a/modules/chat.py b/modules/chat.py
index 60ded0b0cd..0e47da2947 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -30,11 +30,12 @@
 )
 from modules.utils import delete_file, get_available_characters, save_file
 
+
 def strftime_now(format):
     return datetime.now().strftime(format)
 
-jinja_env = ImmutableSandboxedEnvironment(trim_blocks=True, lstrip_blocks=True)
 
+jinja_env = ImmutableSandboxedEnvironment(trim_blocks=True, lstrip_blocks=True)
 jinja_env.globals["strftime_now"] = strftime_now
 
 

From b614ea659673e6ac2f4cbf46dbffa1a747ec2d68 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 29 Jan 2025 14:05:39 -0800
Subject: [PATCH 0219/1701] Installer: small fixes

---
 one_click.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/one_click.py b/one_click.py
index 9105e057d1..effc7d4371 100644
--- a/one_click.py
+++ b/one_click.py
@@ -367,6 +367,8 @@ def update_requirements(initial_installation=False, pull=True):
 
         if 'wheels_changed' in last_state or last_state.get('last_installed_commit') != current_commit:
             wheels_changed = True
+    else:
+        wheels_changed = True
 
     if pull:
         # Read .whl lines before pulling
@@ -409,7 +411,7 @@ def update_requirements(initial_installation=False, pull=True):
                 with open(state_file, 'w') as f:
                     json.dump(current_state, f)
 
-                exit(1)
+                sys.exit(1)
 
     # Save current state
     current_state = {'last_installed_commit': current_commit}

From fea98f82c541d7bee1ba3c2f0cd77e6f17d52740 Mon Sep 17 00:00:00 2001
From: SpyTech Labs <80442847+teufortressIndustries@users.noreply.github.com>
Date: Thu, 30 Jan 2025 22:34:23 +0500
Subject: [PATCH 0220/1701] DOCS FIX: WSL Port Forwarding Loop. (#6519)

---
 docs/10 - WSL.md | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/docs/10 - WSL.md b/docs/10 - WSL.md
index 3e9865c168..e0d66393cb 100644
--- a/docs/10 - WSL.md	
+++ b/docs/10 - WSL.md	
@@ -135,9 +135,12 @@ When you git clone a repository, put it inside WSL and not outside. To understan
 
 ### Bonus: Port Forwarding
 
-By default, you won't be able to access the webui from another device on your local network. You will need to setup the appropriate port forwarding using the following command (using PowerShell or Terminal with administrator privileges). 
+By default, you won't be able to access the webui from another device on your local network. You will need to setup the appropriate port forwarding using the following steps:
+
+1. First, get the IP address of the WSL by typing `wsl hostname -I`. This will output the IP address, for example `172.20.134.111`.
+2. Then, use the following command (using PowerShell or Terminal with administrator privileges) to set up port forwarding, replacing `172.20.134.111` with the IP address you obtained in step 1:
 
 ```
-netsh interface portproxy add v4tov4 listenaddress=0.0.0.0 listenport=7860 connectaddress=localhost connectport=7860
+netsh interface portproxy add v4tov4 listenaddress=0.0.0.0 listenport=7860 connectaddress=172.20.134.111 connectport=7860
 ```
 

From 461d1fdb761978436a0dd10550053881f997d182 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 30 Jan 2025 09:47:59 -0800
Subject: [PATCH 0221/1701] Update README

---
 README.md | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/README.md b/README.md
index 3642fc58c5..396e8e1e07 100644
--- a/README.md
+++ b/README.md
@@ -392,11 +392,6 @@ Run `python download-model.py --help` to see all the options.
 
 https://colab.research.google.com/github/oobabooga/text-generation-webui/blob/main/Colab-TextGen-GPU.ipynb
 
-## Community
-
-* Subreddit: https://www.reddit.com/r/Oobabooga/
-* Discord: https://discord.gg/jwZCF2dPQN
-
 ## Acknowledgment
 
 In August 2023, [Andreessen Horowitz](https://a16z.com/) (a16z) provided a generous grant to encourage and support my independent work on this project. I am **extremely** grateful for their trust and recognition.

From 32cdaa540f26d0241cd33a4e75860c2cd4f2c035 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 30 Jan 2025 09:47:59 -0800
Subject: [PATCH 0222/1701] Update README

---
 README.md | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/README.md b/README.md
index 3642fc58c5..396e8e1e07 100644
--- a/README.md
+++ b/README.md
@@ -392,11 +392,6 @@ Run `python download-model.py --help` to see all the options.
 
 https://colab.research.google.com/github/oobabooga/text-generation-webui/blob/main/Colab-TextGen-GPU.ipynb
 
-## Community
-
-* Subreddit: https://www.reddit.com/r/Oobabooga/
-* Discord: https://discord.gg/jwZCF2dPQN
-
 ## Acknowledgment
 
 In August 2023, [Andreessen Horowitz](https://a16z.com/) (a16z) provided a generous grant to encourage and support my independent work on this project. I am **extremely** grateful for their trust and recognition.

From 0360f54ae8db04e7491a44107fae0bff04473db3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 2 Feb 2025 15:23:16 -0800
Subject: [PATCH 0223/1701] UI: add a "Show after" parameter (to use with
 DeepSeek )

---
 extensions/openai/typing.py |  1 +
 modules/chat.py             | 10 +++++++++-
 modules/shared.py           |  1 +
 modules/ui.py               |  1 +
 modules/ui_parameters.py    |  1 +
 5 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 5f0e01281e..288205a5d7 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -47,6 +47,7 @@ class GenerationOptions(BaseModel):
     seed: int = -1
     sampler_priority: List[str] | str | None = Field(default=None, description="List of samplers where the first items will appear first in the stack. Example: [\"top_k\", \"temperature\", \"top_p\"].")
     custom_token_bans: str = ""
+    show_after: str = ""
     negative_prompt: str = ''
     dry_sequence_breakers: str = '"\\n", ":", "\\"", "*"'
     grammar_string: str = ""
diff --git a/modules/chat.py b/modules/chat.py
index 0e47da2947..2852aaf333 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -412,8 +412,16 @@ def generate_chat_reply(text, state, regenerate=False, _continue=False, loading_
             yield history
             return
 
+    show_after = html.escape(state["show_after"]) if state["show_after"] else None
     for history in chatbot_wrapper(text, state, regenerate=regenerate, _continue=_continue, loading_message=loading_message, for_ui=for_ui):
-        yield history
+        if show_after:
+            after = history["visible"][-1][1].partition(show_after)[2] or "*Is thinking...*"
+            yield {
+                'internal': history['internal'],
+                'visible': history['visible'][:-1] + [[history['visible'][-1][0], after]]
+            }
+        else:
+            yield history
 
 
 def character_is_loaded(state, raise_exception=False):
diff --git a/modules/shared.py b/modules/shared.py
index f1e126736f..2e91f4d5a3 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -57,6 +57,7 @@
     'seed': -1,
     'custom_stopping_strings': '',
     'custom_token_bans': '',
+    'show_after': '',
     'negative_prompt': '',
     'autoload_model': False,
     'dark_theme': True,
diff --git a/modules/ui.py b/modules/ui.py
index df948a1428..b776e19c37 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -215,6 +215,7 @@ def list_interface_input_elements():
         'sampler_priority',
         'custom_stopping_strings',
         'custom_token_bans',
+        'show_after',
         'negative_prompt',
         'dry_sequence_breakers',
         'grammar_string',
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index c8fd6bc752..265840ed4c 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -92,6 +92,7 @@ def create_ui(default_preset):
                             shared.gradio['sampler_priority'] = gr.Textbox(value=generate_params['sampler_priority'], lines=12, label='Sampler priority', info='Parameter names separated by new lines or commas.', elem_classes=['add_scrollbar'])
                             shared.gradio['custom_stopping_strings'] = gr.Textbox(lines=2, value=shared.settings["custom_stopping_strings"] or None, label='Custom stopping strings', info='Written between "" and separated by commas.', placeholder='"\\n", "\\nYou:"')
                             shared.gradio['custom_token_bans'] = gr.Textbox(value=shared.settings['custom_token_bans'] or None, label='Token bans', info='Token IDs to ban, separated by commas. The IDs can be found in the Default or Notebook tab.')
+                            shared.gradio['show_after'] = gr.Textbox(value=shared.settings['show_after'] or None, label='Show after', info='Hide the reply before this text.', placeholder="")
                             shared.gradio['negative_prompt'] = gr.Textbox(value=shared.settings['negative_prompt'], label='Negative prompt', info='For CFG. Only used when guidance_scale is different than 1.', lines=3, elem_classes=['add_scrollbar'])
                             shared.gradio['dry_sequence_breakers'] = gr.Textbox(value=generate_params['dry_sequence_breakers'], label='dry_sequence_breakers', info='Tokens across which sequence matching is not continued. Specified as a comma-separated list of quoted strings.')
                             with gr.Row() as shared.gradio['grammar_file_row']:

From c6f2c2fd7e430ebb078a1a6ae06b9c282bfe32a8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 2 Feb 2025 15:34:03 -0800
Subject: [PATCH 0224/1701] UI: style improvements

---
 modules/html_generator.py | 49 ---------------------------------------
 1 file changed, 49 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 3edbef5e79..6bad0f892a 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -106,52 +106,6 @@ def replace_blockquote(m):
     return m.group().replace('\n', '\n> ').replace('\\begin{blockquote}', '').replace('\\end{blockquote}', '')
 
 
-def add_long_list_class(html):
-    '''
-    Adds a long-list class to 
               or 
                 containing long 
              1.  items.
-    These will receive a smaller margin/padding in the CSS.
-    '''
-
-    # Helper function to check if a tag is within 
                 or 
-    def is_within_block(start_idx, end_idx, block_matches):
-        return any(start < start_idx < end or start < end_idx < end for start, end in block_matches)
-
-    # Find all 
                ...
                 and ... blocks
-    pre_blocks = [(m.start(), m.end()) for m in re.finditer(r'.*?
                ', html, re.DOTALL)]
-    code_blocks = [(m.start(), m.end()) for m in re.finditer(r'.*?', html, re.DOTALL)]
-    all_blocks = pre_blocks + code_blocks
-
-    # Pattern to find 
                  ...
                 and 
                  ...
                 blocks and their contents
-    list_pattern = re.compile(r'(<[uo]l.*?>)(.*?)()', re.DOTALL)
-    li_pattern = re.compile(r'(.*?)
                2. ', re.DOTALL)
-
-    def process_list(match):
-        start_idx, end_idx = match.span()
-        if is_within_block(start_idx, end_idx, all_blocks):
-            return match.group(0)  # Leave the block unchanged if within 
                 or 
-
-        opening_tag = match.group(1)
-        list_content = match.group(2)
-        closing_tag = match.group(3)
-
-        # Find all list items within this list
-        li_matches = li_pattern.finditer(list_content)
-        has_long_item = any(len(li_match.group(1).strip()) > 224 for li_match in li_matches)
-
-        if has_long_item:
-            # Add class="long-list" to the opening tag if it doesn't already have a class
-            if 'class=' not in opening_tag:
-                opening_tag = opening_tag[:-1] + ' class="long-list">'
-            else:
-                # If there's already a class, append long-list to it
-                opening_tag = re.sub(r'class="([^"]*)"', r'class="\1 long-list"', opening_tag)
-
-        return opening_tag + list_content + closing_tag
-
-    # Process HTML and replace list blocks
-    return list_pattern.sub(process_list, html)
-
-
 @functools.lru_cache(maxsize=None)
 def convert_to_markdown(string):
     if not string:
@@ -251,9 +205,6 @@ def convert_to_markdown(string):
     # Unescape backslashes
     html_output = html_output.replace('\\\\', '\\')
 
-    # Add "long-list" class to 
                   or 
                     containing a long 
                  1.  item
-    html_output = add_long_list_class(html_output)
-
     return html_output
 
 

From f074ffc31b90f4583aa74cfb4db3825a756d5df3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 2 Feb 2025 15:39:36 -0800
Subject: [PATCH 0225/1701] UI: minor light theme improvement

---
 css/html_instruct_style.css | 2 +-
 css/main.css                | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index fcd0558f0d..4613b38055 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -46,7 +46,7 @@
 }
 
 .chat .user-message {
-    background: #f4f4f4;
+    background: #f5f5f5;
     padding: 1.5rem 1rem;
     padding-bottom: 2rem;
     border-radius: 0;
diff --git a/css/main.css b/css/main.css
index b10d19805d..234923384a 100644
--- a/css/main.css
+++ b/css/main.css
@@ -2,7 +2,7 @@
     --darker-gray: #202123;
     --dark-gray: #343541;
     --light-gray: #444654;
-    --light-theme-gray: #f4f4f4;
+    --light-theme-gray: #f5f5f5;
     --border-color-dark: #525252;
     --header-width: 112px;
     --selected-item-color-dark: #32333e;

From f28f39792d52b100e9829f5f3210b17d68b12635 Mon Sep 17 00:00:00 2001
From: SamAcctX <87765660+SamAcctX@users.noreply.github.com>
Date: Sun, 2 Feb 2025 17:41:36 -0600
Subject: [PATCH 0226/1701] update deprecated deepspeed import for transformers
 4.46+ (#6725)

---
 modules/models.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/models.py b/modules/models.py
index f551b82838..3951fe822b 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -33,7 +33,7 @@
 local_rank = None
 if shared.args.deepspeed:
     import deepspeed
-    from transformers.deepspeed import (
+    from transformers.integrations.deepspeed import (
         HfDeepSpeedConfig,
         is_deepspeed_zero3_enabled
     )

From 44e569c3a260d08c455b20103d38ad7aa0f87587 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 2 Feb 2025 16:15:33 -0800
Subject: [PATCH 0227/1701] Remove obsolete convert-to-safetensors.py from the
 repository

---
 convert-to-safetensors.py | 38 --------------------------------------
 1 file changed, 38 deletions(-)
 delete mode 100644 convert-to-safetensors.py

diff --git a/convert-to-safetensors.py b/convert-to-safetensors.py
deleted file mode 100644
index 3b721e7cd4..0000000000
--- a/convert-to-safetensors.py
+++ /dev/null
@@ -1,38 +0,0 @@
-'''
-
-Converts a transformers model to safetensors format and shards it.
-
-This makes it faster to load (because of safetensors) and lowers its RAM usage
-while loading (because of sharding).
-
-Based on the original script by 81300:
-
-https://gist.github.com/81300/fe5b08bff1cba45296a829b9d6b0f303
-
-'''
-
-import argparse
-from pathlib import Path
-
-import torch
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-parser = argparse.ArgumentParser(formatter_class=lambda prog: argparse.HelpFormatter(prog, max_help_position=54))
-parser.add_argument('MODEL', type=str, default=None, nargs='?', help="Path to the input model.")
-parser.add_argument('--output', type=str, default=None, help='Path to the output folder (default: models/{model_name}_safetensors).')
-parser.add_argument("--max-shard-size", type=str, default="2GB", help="Maximum size of a shard in GB or MB (default: %(default)s).")
-parser.add_argument('--bf16', action='store_true', help='Load the model with bfloat16 precision. Requires NVIDIA Ampere GPU.')
-args = parser.parse_args()
-
-if __name__ == '__main__':
-    path = Path(args.MODEL)
-    model_name = path.name
-
-    print(f"Loading {model_name}...")
-    model = AutoModelForCausalLM.from_pretrained(path, low_cpu_mem_usage=True, torch_dtype=torch.bfloat16 if args.bf16 else torch.float16)
-    tokenizer = AutoTokenizer.from_pretrained(path)
-
-    out_folder = args.output or Path(f"models/{model_name}_safetensors")
-    print(f"Saving the converted model to {out_folder} with a maximum shard size of {args.max_shard_size}...")
-    model.save_pretrained(out_folder, max_shard_size=args.max_shard_size, safe_serialization=True)
-    tokenizer.save_pretrained(out_folder)

From edbe0af64797b271ac866d3084b006522ff9c5e1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 2 Feb 2025 17:04:56 -0800
Subject: [PATCH 0228/1701] Minor fixes after
 0360f54ae8db04e7491a44107fae0bff04473db3

---
 extensions/openai/typing.py | 1 -
 settings-template.yaml      | 1 +
 2 files changed, 1 insertion(+), 1 deletion(-)

diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 288205a5d7..5f0e01281e 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -47,7 +47,6 @@ class GenerationOptions(BaseModel):
     seed: int = -1
     sampler_priority: List[str] | str | None = Field(default=None, description="List of samplers where the first items will appear first in the stack. Example: [\"top_k\", \"temperature\", \"top_p\"].")
     custom_token_bans: str = ""
-    show_after: str = ""
     negative_prompt: str = ''
     dry_sequence_breakers: str = '"\\n", ":", "\\"", "*"'
     grammar_string: str = ""
diff --git a/settings-template.yaml b/settings-template.yaml
index 93a64abbeb..74935a609d 100644
--- a/settings-template.yaml
+++ b/settings-template.yaml
@@ -29,6 +29,7 @@ truncation_length: 2048
 seed: -1
 custom_stopping_strings: ''
 custom_token_bans: ''
+show_after: ''
 negative_prompt: ''
 autoload_model: false
 dark_theme: true

From b54bf359bf6145ff7a1a2f676079ebac339c862b Mon Sep 17 00:00:00 2001
From: Manuel Schmid 
Date: Mon, 3 Feb 2025 04:11:49 +0100
Subject: [PATCH 0229/1701] sd_api_pictures model reload fix (#6720)

---
 extensions/sd_api_pictures/script.py | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/extensions/sd_api_pictures/script.py b/extensions/sd_api_pictures/script.py
index 3a31771af6..f216da3849 100644
--- a/extensions/sd_api_pictures/script.py
+++ b/extensions/sd_api_pictures/script.py
@@ -11,7 +11,7 @@
 from PIL import Image
 
 from modules import shared
-from modules.models import reload_model, unload_model
+from modules.models import load_model, unload_model
 from modules.ui import create_refresh_button
 
 torch._C._jit_set_profiling_mode(False)
@@ -38,7 +38,8 @@
     'cfg_scale': 7,
     'textgen_prefix': 'Please provide a detailed and vivid description of [subject]',
     'sd_checkpoint': ' ',
-    'checkpoint_list': [" "]
+    'checkpoint_list': [" "],
+    'last_model': ""
 }
 
 
@@ -46,6 +47,7 @@ def give_VRAM_priority(actor):
     global shared, params
 
     if actor == 'SD':
+        params["last_model"] = shared.model_name
         unload_model()
         print("Requesting Auto1111 to re-load last checkpoint used...")
         response = requests.post(url=f'{params["address"]}/sdapi/v1/reload-checkpoint', json='')
@@ -55,7 +57,8 @@ def give_VRAM_priority(actor):
         print("Requesting Auto1111 to vacate VRAM...")
         response = requests.post(url=f'{params["address"]}/sdapi/v1/unload-checkpoint', json='')
         response.raise_for_status()
-        reload_model()
+        if params["last_model"]:
+            shared.model, shared.tokenizer = load_model(params["last_model"])
 
     elif actor == 'set':
         print("VRAM mangement activated -- requesting Auto1111 to vacate VRAM...")

From cf9676c4d5c32cb3f1346a0cef2ef29b151c3965 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 13 Feb 2025 18:09:21 -0800
Subject: [PATCH 0230/1701] Update README

---
 README.md | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/README.md b/README.md
index 396e8e1e07..40c242c87d 100644
--- a/README.md
+++ b/README.md
@@ -4,6 +4,8 @@ A Gradio web UI for Large Language Models.
 
 Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.com/AUTOMATIC1111/stable-diffusion-webui) of text generation.
 
+[Try the Deep Reason extension](https://oobabooga.gumroad.com/l/deep_reason)
+
 |![Image1](https://github.com/oobabooga/screenshots/raw/main/AFTER-INSTRUCT.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/AFTER-CHAT.png) |
 |:---:|:---:|
 |![Image1](https://github.com/oobabooga/screenshots/raw/main/AFTER-DEFAULT.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/AFTER-PARAMETERS.png) |

From 16f4f1a1c33a57001ceb09afa10c27f5ae401fa8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 17 Feb 2025 17:20:10 -0800
Subject: [PATCH 0231/1701] Bump transformers to 4.49

---
 requirements.txt                 | 2 +-
 requirements_amd.txt             | 2 +-
 requirements_amd_noavx2.txt      | 2 +-
 requirements_apple_intel.txt     | 2 +-
 requirements_apple_silicon.txt   | 2 +-
 requirements_cpu_only.txt        | 2 +-
 requirements_cpu_only_noavx2.txt | 2 +-
 requirements_noavx2.txt          | 2 +-
 requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 4ff7a6dfa2..b4e358fb06 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.48.*
+transformers==4.49.*
 tqdm
 wandb
 
diff --git a/requirements_amd.txt b/requirements_amd.txt
index e30f30eec1..0ceb9d9e21 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.48.*
+transformers==4.49.*
 tqdm
 wandb
 
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 15d25caa4e..330c73d174 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.48.*
+transformers==4.49.*
 tqdm
 wandb
 
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index b614acf4d7..185e6cadbc 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.48.*
+transformers==4.49.*
 tqdm
 wandb
 
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index ca9cc3ac16..f70d1c438c 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.48.*
+transformers==4.49.*
 tqdm
 wandb
 
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index e9a9790517..6467f9967b 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.48.*
+transformers==4.49.*
 tqdm
 wandb
 
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index c435767655..cbebdbbc3b 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.48.*
+transformers==4.49.*
 tqdm
 wandb
 
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 40cbc7b0e5..23d1c20d5e 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.48.*
+transformers==4.49.*
 tqdm
 wandb
 
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index 3d6c922fb8..f20a73329d 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.48.*
+transformers==4.49.*
 tqdm
 wandb
 

From dba17c40fc67fd4e64a26214c47d745bf5a42d18 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 17 Feb 2025 17:31:11 -0800
Subject: [PATCH 0232/1701] Make transformers 4.49 functional

---
 modules/sampler_hijack.py | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

diff --git a/modules/sampler_hijack.py b/modules/sampler_hijack.py
index d202af1f2c..e0df49c3ce 100644
--- a/modules/sampler_hijack.py
+++ b/modules/sampler_hijack.py
@@ -5,7 +5,7 @@
 
 import torch
 import transformers
-from transformers import LogitsWarper
+from transformers import LogitsProcessor
 from transformers.generation.logits_process import (
     LogitNormalization,
     LogitsProcessor,
@@ -19,7 +19,7 @@
 global_scores = None
 
 
-class TemperatureLogitsWarperCustom(LogitsWarper):
+class TemperatureLogitsWarperCustom(LogitsProcessor):
     '''
     A copy of the original Transformers temperature logits warper.
     '''
@@ -42,7 +42,7 @@ def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> to
         return scores
 
 
-class DynamicTemperatureLogitsWarper(LogitsWarper):
+class DynamicTemperatureLogitsWarper(LogitsProcessor):
     '''
     Dynamic temperature.
     '''
@@ -100,7 +100,7 @@ def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> to
         return scores
 
 
-class QuadraticSamplingLogitsWarper(LogitsWarper):
+class QuadraticSamplingLogitsWarper(LogitsProcessor):
     '''
     Quadratic sampling with smoothing factor and smoothing curve parameters.
     '''
@@ -127,7 +127,7 @@ def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> to
         return transformed_logits
 
 
-class TailFreeLogitsWarper(LogitsWarper):
+class TailFreeLogitsWarper(LogitsProcessor):
     def __init__(self, tfs: float, filter_value: float = -float("Inf"), min_tokens_to_keep: int = 1):
         tfs = float(tfs)
         if tfs < 0 or tfs > 1.0:
@@ -167,7 +167,7 @@ def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> to
         return scores
 
 
-class TopALogitsWarper(LogitsWarper):
+class TopALogitsWarper(LogitsProcessor):
     def __init__(self, top_a: float, filter_value: float = -float("Inf"), min_tokens_to_keep: int = 1):
         top_a = float(top_a)
         if top_a < 0 or top_a > 1.0:
@@ -194,7 +194,7 @@ def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> to
 
 
 # Exclude Top Choices (XTC)
-class XTCLogitsWarper(LogitsWarper):
+class XTCLogitsWarper(LogitsProcessor):
     def __init__(self, threshold: float, probability: float, filter_value: float = -float("Inf")):
         self.threshold = threshold
         self.probability = probability
@@ -312,7 +312,7 @@ def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> to
         return scores
 
 
-class MirostatLogitsWarper(LogitsWarper):
+class MirostatLogitsWarper(LogitsProcessor):
     def __init__(self, mirostat_mode: int, mirostat_tau: float, mirostat_eta: float, filter_value: float = -float("Inf"), min_tokens_to_keep: int = 1):
         if mirostat_mode not in [2]:
             raise ValueError(f"`mirostat` has to be a an integer 2, but is {mirostat_mode}")
@@ -361,7 +361,7 @@ def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> to
         return scores
 
 
-class SpyLogitsWarper(LogitsWarper):
+class SpyLogitsWarper(LogitsProcessor):
     def __init__(self):
         pass
 

From 12f6f7ba9ff327a67ee334e84da4e7f292819a25 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Mon, 17 Feb 2025 22:35:38 -0300
Subject: [PATCH 0233/1701] Update accelerate requirement from ==1.3.* to
 ==1.4.* (#6753)

---
 requirements.txt                 | 2 +-
 requirements_amd.txt             | 2 +-
 requirements_amd_noavx2.txt      | 2 +-
 requirements_apple_intel.txt     | 2 +-
 requirements_apple_silicon.txt   | 2 +-
 requirements_cpu_only.txt        | 2 +-
 requirements_cpu_only_noavx2.txt | 2 +-
 requirements_noavx2.txt          | 2 +-
 requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index b4e358fb06..d09f6bf582 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,4 +1,4 @@
-accelerate==1.3.*
+accelerate==1.4.*
 bitsandbytes==0.45.*
 colorama
 datasets
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 0ceb9d9e21..124ad6b64a 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -1,4 +1,4 @@
-accelerate==1.3.*
+accelerate==1.4.*
 colorama
 datasets
 einops
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 330c73d174..9e3063c32a 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.3.*
+accelerate==1.4.*
 colorama
 datasets
 einops
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 185e6cadbc..0ef8db345e 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -1,4 +1,4 @@
-accelerate==1.3.*
+accelerate==1.4.*
 colorama
 datasets
 einops
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index f70d1c438c..9c4bcc1166 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -1,4 +1,4 @@
-accelerate==1.3.*
+accelerate==1.4.*
 colorama
 datasets
 einops
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index 6467f9967b..9f19238a1a 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -1,4 +1,4 @@
-accelerate==1.3.*
+accelerate==1.4.*
 colorama
 datasets
 einops
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index cbebdbbc3b..042dbbd82f 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.3.*
+accelerate==1.4.*
 colorama
 datasets
 einops
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 23d1c20d5e..f488cafcd5 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.3.*
+accelerate==1.4.*
 bitsandbytes==0.45.*
 colorama
 datasets
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index f20a73329d..1bc2e385c6 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -1,4 +1,4 @@
-accelerate==1.3.*
+accelerate==1.4.*
 colorama
 datasets
 einops

From 01f20d2d9f4e3a96556c6dce560d0068bc2eb492 Mon Sep 17 00:00:00 2001
From: Alireza Ghasemi 
Date: Tue, 18 Feb 2025 02:38:15 +0100
Subject: [PATCH 0234/1701] Improve SuperboogaV2 with Date/Time Embeddings, GPU
 Support, and Multiple File Formats (#6748)

---
 extensions/superboogav2/README.md         |  42 ++++++-
 extensions/superboogav2/chromadb.py       |  36 +++---
 extensions/superboogav2/config.json       |   3 +
 extensions/superboogav2/data_processor.py |   8 ++
 extensions/superboogav2/optimize.py       |   6 +-
 extensions/superboogav2/parameters.py     |   8 ++
 extensions/superboogav2/requirements.txt  |  16 ++-
 extensions/superboogav2/script.py         | 146 ++++++++++++++++++++--
 8 files changed, 227 insertions(+), 38 deletions(-)

diff --git a/extensions/superboogav2/README.md b/extensions/superboogav2/README.md
index d25b3a5eb9..0460c40154 100644
--- a/extensions/superboogav2/README.md
+++ b/extensions/superboogav2/README.md
@@ -1,5 +1,41 @@
-# superboogav2
+# SuperboogaV2
 
-For a description, please see the comments in this Pull Request:
+Enhance your LLM with additional information from text, URLs, and files for more accurate and context-aware responses.
 
-https://github.com/oobabooga/text-generation-webui/pull/3272
+---
+
+
+
+## Installation and Activation
+
+1. Start the conda environment by running `cmd_windows.bat` or the equivalent for your system in the root directory of `text-generation-webui`.
+2. Install the necessary packages:
+   ```
+   pip install -r extensions/superboogav2/requirements.txt
+   ```
+3. Activate the extension in the `Session` tab of the web UI.
+4. Click on `Apply flags/extensions and restart`. Optionally save the configuration by clicking on `Save UI defaults to settings.yaml`.
+
+## Usage and Features
+
+After activation, you can scroll further down in the chat UI to reveal the SuperboogaV2 interface. Here, you can add extra information to your chats through text input, multiple URLs, or by providing multiple files subject to the context window limit of your model.
+
+The extra information and the current date and time are provided to the model as embeddings that persist across conversations. To clear them, click the `Clear Data` button and start a new chat. You can adjust the text extraction parameters and other options in the `Settings`.
+
+## Supported File Formats
+
+SuperboogaV2 utilizes MuPDF, pandas, python-docx, and python-pptx to extract text from various file formats, including:
+
+- TXT
+- PDF
+- EPUB
+- HTML
+- CSV
+- ODT/ODS/ODP
+- DOCX/PPTX/XLSX
+
+## Additional Information
+
+SuperboogaV2 processes your data into context-aware chunks, applies cleaning techniques, and stores them as embeddings to minimize redundant computations. Relevance is determined using distance calculations and prioritization of recent information.
+
+For a detailed description and more information, refer to the comments in this pull request: [https://github.com/oobabooga/text-generation-webui/pull/3272](https://github.com/oobabooga/text-generation-webui/pull/3272)
diff --git a/extensions/superboogav2/chromadb.py b/extensions/superboogav2/chromadb.py
index 3381fb1436..c9e450e4b5 100644
--- a/extensions/superboogav2/chromadb.py
+++ b/extensions/superboogav2/chromadb.py
@@ -1,7 +1,7 @@
 import math
 import random
 import threading
-
+import torch
 import chromadb
 import numpy as np
 import posthog
@@ -16,9 +16,6 @@
 posthog.capture = lambda *args, **kwargs: None
 
 
-embedder = embedding_functions.SentenceTransformerEmbeddingFunction("sentence-transformers/all-mpnet-base-v2")
-
-
 class Info:
     def __init__(self, start_index, text_with_context, distance, id):
         self.text_with_context = text_with_context
@@ -77,11 +74,23 @@ def should_merge(s1, s2, s1_start, s2_start):
 
 class ChromaCollector():
     def __init__(self):
-        name = ''.join(random.choice('ab') for _ in range(10))
+        name = "".join(random.choice("ab") for _ in range(10))
 
         self.name = name
-        self.chroma_client = chromadb.Client(Settings(anonymized_telemetry=False))
-        self.collection = self.chroma_client.create_collection(name=name, embedding_function=embedder)
+        self.embedder = embedding_functions.SentenceTransformerEmbeddingFunction(
+            "sentence-transformers/all-mpnet-base-v2",
+            device=("cuda" if torch.cuda.is_available() else "cpu"),
+        )
+        chroma_client = chromadb.Client(Settings(anonymized_telemetry=False))
+        self.collection = chroma_client.create_collection(
+            name=self.name,
+            embedding_function=self.embedder,
+            metadata={
+                "hnsw:search_ef": 200,
+                "hnsw:construction_ef": 200,
+                "hnsw:M": 64,
+            },
+        )
 
         self.ids = []
         self.id_to_info = {}
@@ -110,7 +119,7 @@ def add(self, texts: list[str], texts_with_context: list[str], starting_indices:
 
             # If there are any non-existing texts, compute their embeddings all at once. Each call to embed has significant overhead.
             if non_existing_texts:
-                non_existing_embeddings = embedder(non_existing_texts)
+                non_existing_embeddings = self.embedder(non_existing_texts)
                 for text, embedding in zip(non_existing_texts, non_existing_embeddings):
                     self.embeddings_cache[text] = embedding
 
@@ -139,7 +148,7 @@ def _split_texts_by_cache_hit(self, texts: list[str], new_ids: list[str], metada
             id_ = new_ids[i]
             metadata = metadatas[i] if metadatas is not None else None
             embedding = self.embeddings_cache.get(text)
-            if embedding:
+            if embedding is not None and embedding.any():
                 existing_texts.append(text)
                 existing_embeddings.append(embedding)
                 existing_ids.append(id_)
@@ -323,6 +332,8 @@ def get_sorted_by_dist(self, search_strings: list[str], n_results: int, max_toke
     def delete(self, ids_to_delete: list[str], where: dict):
         with self.lock:
             ids_to_delete = self.collection.get(ids=ids_to_delete, where=where)['ids']
+            if not ids_to_delete:
+                return
             self.collection.delete(ids=ids_to_delete, where=where)
 
             # Remove the deleted ids from self.ids and self.id_to_info
@@ -335,12 +346,7 @@ def delete(self, ids_to_delete: list[str], where: dict):
 
     def clear(self):
         with self.lock:
-            self.chroma_client.reset()
-
-            self.ids = []
-            self.chroma_client.delete_collection(name=self.name)
-            self.collection = self.chroma_client.create_collection(name=self.name, embedding_function=embedder)
-
+            self.__init__()  # reinitialize the collector
             logger.info('Successfully cleared all records and reset chromaDB.')
 
 
diff --git a/extensions/superboogav2/config.json b/extensions/superboogav2/config.json
index 0f1034f521..5de3d8706a 100644
--- a/extensions/superboogav2/config.json
+++ b/extensions/superboogav2/config.json
@@ -127,6 +127,9 @@
       "default": "\n\n<>\n\n"
     },
     "manual": {
+      "default": false
+    },
+    "add_date_time": {
       "default": true
     },
     "add_chat_to_data": {
diff --git a/extensions/superboogav2/data_processor.py b/extensions/superboogav2/data_processor.py
index 0a96d4a43b..3c5e5c9fea 100644
--- a/extensions/superboogav2/data_processor.py
+++ b/extensions/superboogav2/data_processor.py
@@ -6,6 +6,7 @@
 
 import bisect
 import re
+from datetime import datetime
 
 import extensions.superboogav2.parameters as parameters
 
@@ -154,6 +155,13 @@ def process_and_add_to_collector(corpus: str, collector: ChromaCollector, clear_
     data_chunks_with_context = []
     data_chunk_starting_indices = []
 
+    if parameters.get_add_date_time():
+        now = datetime.now()
+        date_time_chunk = f"Current time is {now.strftime('%H:%M:%S')}. Today is {now.strftime('%A')}. The current date is {now.strftime('%Y-%m-%d')}."
+        data_chunks.append(date_time_chunk)
+        data_chunks_with_context.append(date_time_chunk)
+        data_chunk_starting_indices.append(0)
+
     # Handling chunk_regex
     if parameters.get_chunk_regex():
         if parameters.get_chunk_separator():
diff --git a/extensions/superboogav2/optimize.py b/extensions/superboogav2/optimize.py
index ebdd03c6e2..3597fdf100 100644
--- a/extensions/superboogav2/optimize.py
+++ b/extensions/superboogav2/optimize.py
@@ -39,11 +39,11 @@ def _markdown_hyperparams():
 # Convert numpy types to python types.
 def _convert_np_types(params):
     for key in params:
-        if type(params[key]) == np.bool_:
+        if isinstance(params[key], np.bool_):
             params[key] = bool(params[key])
-        elif type(params[key]) == np.int64:
+        elif isinstance(params[key], np.int64):
             params[key] = int(params[key])
-        elif type(params[key]) == np.float64:
+        elif isinstance(params[key], np.float64):
             params[key] = float(params[key])
     return params
 
diff --git a/extensions/superboogav2/parameters.py b/extensions/superboogav2/parameters.py
index 8bb2d1a6fb..e691dae18d 100644
--- a/extensions/superboogav2/parameters.py
+++ b/extensions/superboogav2/parameters.py
@@ -251,6 +251,10 @@ def get_is_manual() -> bool:
     return bool(Parameters.getInstance().hyperparameters['manual']['default'])
 
 
+def get_add_date_time() -> bool:
+    return bool(Parameters.getInstance().hyperparameters['add_date_time']['default'])
+
+
 def get_add_chat_to_data() -> bool:
     return bool(Parameters.getInstance().hyperparameters['add_chat_to_data']['default'])
 
@@ -331,6 +335,10 @@ def set_manual(value: bool):
     Parameters.getInstance().hyperparameters['manual']['default'] = value
 
 
+def set_add_date_time(value: bool):
+    Parameters.getInstance().hyperparameters['add_date_time']['default'] = value
+
+
 def set_add_chat_to_data(value: bool):
     Parameters.getInstance().hyperparameters['add_chat_to_data']['default'] = value
 
diff --git a/extensions/superboogav2/requirements.txt b/extensions/superboogav2/requirements.txt
index d9031167de..6de51e6304 100644
--- a/extensions/superboogav2/requirements.txt
+++ b/extensions/superboogav2/requirements.txt
@@ -1,10 +1,16 @@
-beautifulsoup4==4.12.2
-chromadb==0.4.24
+beautifulsoup4==4.13.3
+chromadb==0.6.3
 lxml
+nltk
 optuna
-pandas==2.0.3
-posthog==2.4.2
-sentence_transformers==2.2.2
+pandas
+posthog==3.13.0
+sentence_transformers==3.3.1
 spacy
 pytextrank
 num2words
+PyMuPDF
+python-docx
+python-pptx
+openpyxl
+odfpy
\ No newline at end of file
diff --git a/extensions/superboogav2/script.py b/extensions/superboogav2/script.py
index 77c5cced78..13c58df9e3 100644
--- a/extensions/superboogav2/script.py
+++ b/extensions/superboogav2/script.py
@@ -9,6 +9,13 @@
 
 import codecs
 import textwrap
+import docx
+import pptx
+import fitz
+fitz.TOOLS.mupdf_display_errors(False)
+import pandas as pd
+from odf.opendocument import load
+from odf.draw import Page
 
 import gradio as gr
 
@@ -46,11 +53,123 @@ def _feed_data_into_collector(corpus):
     yield '### Done.'
 
 
-def _feed_file_into_collector(file):
-    yield '### Reading and processing the input dataset...'
-    text = file.decode('utf-8')
-    process_and_add_to_collector(text, collector, False, create_metadata_source('file'))
-    yield '### Done.'
+def _feed_file_into_collector(files):
+    if not files:
+        logger.warning("No files selected.")
+        return
+
+    def read_binary_file(file_path):
+        try:
+            with open(file_path, 'rb') as f:
+                return f.read()
+        except Exception:
+            logger.error(f"Failed to read {file_path}.")
+            return None
+
+    def extract_with_utf8(text):
+        try:
+            return text.decode('utf-8')
+        except Exception:
+            return ""
+
+    def extract_with_fitz(file_content):
+        try:
+            with fitz.open(stream=file_content, filetype=None) as doc:
+                num_pages = doc.page_count
+                text = "\n".join(block[4] for page in doc for block in page.get_text("blocks") if block[6] == 0)
+                logger.info(f"Extracted text from {num_pages} pages with fitz.")
+                return text
+        except Exception:
+            return ""
+
+    def extract_with_docx(file_path):
+        try:
+            paragraphs = docx.Document(file_path).paragraphs
+            text = "\n".join(para.text for para in paragraphs)
+            logger.info(f"Extracted text from {len(paragraphs)} paragraphs with docx.")
+            return text
+        except Exception:
+            return ""
+
+    def extract_with_pptx(file_path):
+        try:
+            slides = pptx.Presentation(file_path).slides
+            text = "\n".join(
+                shape.text for slide in slides for shape in slide.shapes if hasattr(shape, "text")
+            )
+            logger.info(f"Extracted text from {len(slides)} slides with pptx.")
+            return text
+        except Exception:
+            return ""
+
+    def extract_with_odf(file_path):
+        if not file_path.endswith(".odp"):
+            return ""
+        try:
+            doc = load(file_path)
+            text_content = []
+
+            def extract_text(element):
+                parts = []
+                if hasattr(element, "childNodes"):
+                    for node in element.childNodes:
+                        if node.nodeType == node.TEXT_NODE:
+                            parts.append(node.data)
+                        else:
+                            parts.append(extract_text(node))
+                return "".join(parts)
+
+            for slide in doc.getElementsByType(Page):
+                slide_text = extract_text(slide)
+                if slide_text.strip():
+                    text_content.append(slide_text.strip())
+
+            text = "\n".join(text_content)
+            logger.info(f"Extracted text from {len(doc.getElementsByType(Page))} slides with odf.")
+            return text
+        except Exception as e:
+            logger.error(f"Failed to extract text from {file_path}: {str(e)}")
+            return ""
+
+    def extract_with_pandas(file_path):
+        try:
+            df = pd.read_excel(file_path)
+            text = "\n".join(str(cell) for col in df.columns for cell in df[col])
+            logger.info(f"Extracted text from {df.shape[0]}x{df.shape[1]} cells with pandas.")
+            return text
+        except Exception:
+            return ""
+
+    for index, file in enumerate(files, start=1):
+        file_name = os.path.basename(file)
+        logger.info(f"Processing {file_name}...")
+
+        file_content = read_binary_file(file)
+        if not file_content:
+            continue
+
+        text_extractors = [
+            lambda: extract_with_utf8(file_content),
+            lambda: extract_with_fitz(file_content),
+            lambda: extract_with_docx(file),
+            lambda: extract_with_pptx(file),
+            lambda: extract_with_odf(file),
+            lambda: extract_with_pandas(file),
+        ]
+
+        for extractor in text_extractors:
+            text = extractor()
+            if text:
+                break
+
+        if not text:
+            logger.error(f"Failed to extract text from {file_name}, unsupported format.")
+            continue
+
+        process_and_add_to_collector(text, collector, False, create_metadata_source(f"file-{index}"))
+
+    logger.info("Done.")
+    yield "### Done."
 
 
 def _feed_url_into_collector(urls):
@@ -107,7 +226,7 @@ def _get_optimizable_settings() -> list:
 
 
 def _apply_settings(optimization_steps, time_power, time_steepness, significant_level, min_sentences, new_dist_strat, delta_start, min_number_length, num_conversion,
-                    preprocess_pipeline, api_port, api_on, injection_strategy, add_chat_to_data, manual, postfix, data_separator, prefix, max_token_count,
+                    preprocess_pipeline, api_port, api_on, injection_strategy, add_chat_to_data, manual, add_date_time, postfix, data_separator, prefix, max_token_count,
                     chunk_count, chunk_sep, context_len, chunk_regex, chunk_len, threads, strong_cleanup):
     logger.debug('Applying settings.')
 
@@ -124,6 +243,7 @@ def _apply_settings(optimization_steps, time_power, time_steepness, significant_
         parameters.set_injection_strategy(injection_strategy)
         parameters.set_add_chat_to_data(add_chat_to_data)
         parameters.set_manual(manual)
+        parameters.set_add_date_time(add_date_time)
         parameters.set_postfix(codecs.decode(postfix, 'unicode_escape'))
         parameters.set_data_separator(codecs.decode(data_separator, 'unicode_escape'))
         parameters.set_prefix(codecs.decode(prefix, 'unicode_escape'))
@@ -237,11 +357,11 @@ def ui():
                 url_input = gr.Textbox(lines=10, label='Input URLs', info='Enter one or more URLs separated by newline characters.')
                 strong_cleanup = gr.Checkbox(value=parameters.get_is_strong_cleanup(), label='Strong cleanup', info='Only keeps html elements that look like long-form text.')
                 threads = gr.Number(value=parameters.get_num_threads(), label='Threads', info='The number of threads to use while downloading the URLs.', precision=0)
-                update_url = gr.Button('Load data')
+                update_urls = gr.Button('Load data')
 
             with gr.Tab("File input"):
-                file_input = gr.File(label='Input file', type='binary')
-                update_file = gr.Button('Load data')
+                file_input = gr.File(label="Input file", type="filepath", file_count="multiple")
+                update_files = gr.Button('Load data')
 
             with gr.Tab("Settings"):
                 with gr.Accordion("Processing settings", open=True):
@@ -258,6 +378,7 @@ def ui():
                     postfix = gr.Textbox(value=codecs.encode(parameters.get_postfix(), 'unicode_escape').decode(), label='Postfix', info='What to put after the injection point.')
                     with gr.Row():
                         manual = gr.Checkbox(value=parameters.get_is_manual(), label="Is Manual", info="Manually specify when to use ChromaDB. Insert `!c` at the start or end of the message to trigger a query.", visible=shared.is_chat())
+                        add_date_time = gr.Checkbox(value=parameters.get_add_date_time(), label="Add date and time to Data", info="Make the current date and time available to the model.", visible=shared.is_chat())
                         add_chat_to_data = gr.Checkbox(value=parameters.get_add_chat_to_data(), label="Add Chat to Data", info="Automatically feed the chat history as you chat.", visible=shared.is_chat())
                     injection_strategy = gr.Radio(choices=[parameters.PREPEND_TO_LAST, parameters.APPEND_TO_LAST, parameters.HIJACK_LAST_IN_CONTEXT], value=parameters.get_injection_strategy(), label='Injection Strategy', info='Where to inject the messages in chat or instruct mode.', visible=shared.is_chat())
                     with gr.Row():
@@ -313,14 +434,14 @@ def ui():
             last_updated = gr.Markdown()
 
     all_params = [optimization_steps, time_power, time_steepness, significant_level, min_sentences, new_dist_strat, delta_start, min_number_length, num_conversion,
-                  preprocess_pipeline, api_port, api_on, injection_strategy, add_chat_to_data, manual, postfix, data_separator, prefix, max_token_count,
+                  preprocess_pipeline, api_port, api_on, injection_strategy, add_chat_to_data, manual, add_date_time, postfix, data_separator, prefix, max_token_count,
                   chunk_count, chunk_sep, context_len, chunk_regex, chunk_len, threads, strong_cleanup]
     optimizable_params = [time_power, time_steepness, significant_level, min_sentences, new_dist_strat, delta_start, min_number_length, num_conversion,
                           preprocess_pipeline, chunk_count, context_len, chunk_len]
 
     update_data.click(_feed_data_into_collector, [data_input], last_updated, show_progress=False)
-    update_url.click(_feed_url_into_collector, [url_input], last_updated, show_progress=False)
-    update_file.click(_feed_file_into_collector, [file_input], last_updated, show_progress=False)
+    update_urls.click(_feed_url_into_collector, [url_input], last_updated, show_progress=False)
+    update_files.click(_feed_file_into_collector, [file_input], last_updated, show_progress=False)
     benchmark_button.click(_begin_benchmark, [], last_updated, show_progress=True)
     optimize_button.click(_begin_optimization, [], [last_updated] + optimizable_params, show_progress=True)
     clear_button.click(_clear_data, [], last_updated, show_progress=False)
@@ -339,6 +460,7 @@ def ui():
     api_on.input(fn=_apply_settings, inputs=all_params, show_progress=False)
     injection_strategy.input(fn=_apply_settings, inputs=all_params, show_progress=False)
     add_chat_to_data.input(fn=_apply_settings, inputs=all_params, show_progress=False)
+    add_date_time.input(fn=_apply_settings, inputs=all_params, show_progress=False)
     manual.input(fn=_apply_settings, inputs=all_params, show_progress=False)
     postfix.input(fn=_apply_settings, inputs=all_params, show_progress=False)
     data_separator.input(fn=_apply_settings, inputs=all_params, show_progress=False)

From b131f865840aff5ccb7516535efc2c683f763cf1 Mon Sep 17 00:00:00 2001
From: SeanScripts <64337075+SeanScripts@users.noreply.github.com>
Date: Tue, 18 Feb 2025 06:56:28 -0800
Subject: [PATCH 0235/1701] Perplexity colors extension v2 (#6756)

---
 extensions/perplexity_colors/script.py | 275 ++++++++++++++++++-------
 1 file changed, 201 insertions(+), 74 deletions(-)

diff --git a/extensions/perplexity_colors/script.py b/extensions/perplexity_colors/script.py
index 2a986ac40b..849e4e630c 100644
--- a/extensions/perplexity_colors/script.py
+++ b/extensions/perplexity_colors/script.py
@@ -1,9 +1,14 @@
 import time
 
+import html
+import functools
+import re
+
 import gradio
 import numpy as np
 import torch
 from transformers import LogitsProcessor
+import colorsys
 
 from modules import html_generator, shared
 
@@ -28,7 +33,7 @@ def __init__(self, verbose=False):
         self.verbose = verbose
 
     def __call__(self, input_ids, scores):
-        # t0 = time.time()
+        #t0 = time.time()
         probs = torch.softmax(scores, dim=-1, dtype=torch.float)
         log_probs = torch.nan_to_num(torch.log(probs))  # Note: This is to convert log(0) nan to 0, but probs*log_probs makes this 0 not affect the perplexity.
         entropy = -torch.sum(probs * log_probs)
@@ -42,9 +47,8 @@ def __call__(self, input_ids, scores):
         if len(self.selected_probs) > 0:
             # Is the selected token in the top tokens?
             if self.verbose:
-                print('Probs: Token after', shared.tokenizer.decode(last_token_id))
-                print('Probs:', [shared.tokenizer.decode(token_id) for token_id in self.top_token_ids_list[-1][0]])
-                print('Probs:', [round(float(prob), 4) for prob in self.top_probs_list[-1][0]])
+                print(shared.tokenizer.decode(last_token_id), [shared.tokenizer.decode(token_id) for token_id in self.top_token_ids_list[-1][0]],
+                    [round(float(prob), 4) for prob in self.top_probs_list[-1][0]])
             if last_token_id in self.top_token_ids_list[-1][0]:
                 idx = self.top_token_ids_list[-1][0].index(last_token_id)
                 self.selected_probs.append(self.top_probs_list[-1][0][idx])
@@ -60,7 +64,7 @@ def __call__(self, input_ids, scores):
             pplbar = "-"
             if not np.isnan(perplexity):
                 pplbar = "*" * round(perplexity)
-            print(f"PPL: Token after {shared.tokenizer.decode(last_token_id)}\t{perplexity:.2f}\t{pplbar}")
+            print(f"PPL for token after {shared.tokenizer.decode(last_token_id)}: {perplexity:.2f} {pplbar}")
 
         # Get top 5 probabilities
         top_tokens_and_probs = torch.topk(probs, 5)
@@ -73,14 +77,15 @@ def __call__(self, input_ids, scores):
         probs = probs.cpu().numpy().flatten()
         self.last_probs = probs  # Need to keep this as a reference for top probs
 
-        # t1 = time.time()
-        # print(f"PPL Processor: {(t1-t0):.3f} s")
+        #t1 = time.time()
+        #print(f"PPL Processor: {(t1-t0):.3f} s")
         # About 1 ms, though occasionally up to around 100 ms, not sure why...
         # Doesn't actually modify the logits!
         return scores
 
 
 # Stores the perplexity and top probabilities
+# global ppl_logits_processor
 ppl_logits_processor = None
 
 
@@ -93,9 +98,9 @@ def logits_processor_modifier(logits_processor_list, input_ids):
 
 def output_modifier(text):
     global ppl_logits_processor
-    # t0 = time.time()
+    #t0 = time.time()
 
-    if not params['active']:
+    if not params['active'] or ppl_logits_processor is None:
         return text
 
     # TODO: It's probably more efficient to do this above rather than modifying all these lists
@@ -111,110 +116,147 @@ def output_modifier(text):
 
     end_part = '
            ' if params['probability_dropdown'] else ''  # Helps with finding the index after replacing part of the text.
 
-    i = 0
-    for token, prob, ppl, top_tokens, top_probs in zip(gen_tokens, sel_probs, perplexities, top_tokens_list, top_probs_list):
+    # Initial space added to deal with some tokenizers...
+    # Used to find where the message started generating, for working with "continue" generations
+    # Doesn't work for longer messages... Not sure how I should handle this
+    full_msg = shared.tokenizer.decode([token_id for token_id in gen_token_ids[:-1]]).strip()
+    # Space at the beginning to account for tokenization spaces...
+    text = ' ' + html.unescape(text)
+    # There was an issue with tab lengths being off by one...
+    # Seems like it might be model-dependent...
+    #text = re.sub(r'( {3,})', r'\1 ', text)
+    # Subtracting 2 to hopefully help with the tokenization spaces and continue issues,
+    # Though it's possible it could overwrite the previous token if it's the same in the last 2 chars
+    i = text.find(full_msg) - 2
+    if i < 0:
+        # Backup, try removing the extra whitespace (needed for continue)
+        i = text.find(full_msg.strip()) - 2
+        if i < 0:
+            i = 0
+
+    #i = 0
+    # Add token index for ability to regenerate from there
+    nonwhitespace_token_found = False
+    for index, token, prob, ppl, top_tokens, top_probs in zip(range(len(gen_tokens)), gen_tokens, sel_probs, perplexities, top_tokens_list, top_probs_list):
+        # Somehow this works without issues, but not sure how...
+        if not nonwhitespace_token_found and token.strip() == '':
+            #print('Ignoring initial whitespace token...')
+            continue
+        nonwhitespace_token_found = True
+        max_prob = top_probs[0][0]
         color = 'ffffff'
         if params['color_by_probability'] and params['color_by_perplexity']:
-            color = probability_perplexity_color_scale(prob, ppl)
+            color = probability_perplexity_color_scale(prob, max_prob, ppl)
         elif params['color_by_perplexity']:
             color = perplexity_color_scale(ppl)
         elif params['color_by_probability']:
             color = probability_color_scale(prob)
-        if token in text[i:]:
+        if token.strip() in text[i:]:
             if params['probability_dropdown']:
-                text = text[:i] + text[i:].replace(token, add_dropdown_html(token, color, top_tokens, top_probs[0], ppl), 1)
+                text = text[:i] + text[i:].replace(token.replace('\n', ''), add_dropdown_html(token, index, color, top_tokens, top_probs[0], ppl), 1)
             else:
-                text = text[:i] + text[i:].replace(token, add_color_html(token, color), 1)
+                text = text[:i] + text[i:].replace(token.replace('\n', ''), add_color_html(token, color), 1)
+            
+            # This might be slightly inefficient
             i += text[i:].find(end_part) + len(end_part)
+        else:
+            print('Missing token:', token, '...', text[i:i+20])
 
     # Use full perplexity list for calculating the average here.
-    print('Average perplexity:', round(np.mean(ppl_logits_processor.perplexities_list[:-1]), 4))
-    # t1 = time.time()
-    # print(f"Modifier: {(t1-t0):.3f} s")
+    # Fix issue with mean of empty slice
+    if len(ppl_logits_processor.perplexities_list) > 1:
+        print('Average perplexity:', round(np.mean(ppl_logits_processor.perplexities_list[:-1]), 4))
+    #t1 = time.time()
+    #print(f"Output modifier: {(t1-t0):.3f} s")
     # About 50 ms
-    return text
+    return text.strip() # Remove extra beginning whitespace that some tokenizers add
 
 
 def probability_color_scale(prob):
     '''
     Green-yellow-red color scale
     '''
+    # hue (0.0 = red, 0.33 = green)
+    # saturation (0.0 = gray / white, 1.0 = normal, just leave at 1.0)
+    # brightness (0.0 = black, 1.0 = brightest, use something in between for better readability if you want...)
+    hue = prob * 0.33
+    rv, gv, bv = colorsys.hsv_to_rgb(hue, 1.0, 1.0)
+    # to hex
+    hex_col = f"{int(rv*255):02x}{int(gv*255):02x}{int(bv*255):02x}"
 
-    rv = 0
-    gv = 0
-    if prob <= 0.5:
-        rv = 'ff'
-        gv = hex(int(255 * prob * 2))[2:]
-        if len(gv) < 2:
-            gv = '0' * (2 - len(gv)) + gv
-    else:
-        rv = hex(int(255 - 255 * (prob - 0.5) * 2))[2:]
-        gv = 'ff'
-        if len(rv) < 2:
-            rv = '0' * (2 - len(rv)) + rv
-
-    return rv + gv + '00'
+    return hex_col
 
 
 def perplexity_color_scale(ppl):
     '''
     Red component only, white for 0 perplexity (sorry if you're not in dark mode)
     '''
-    value = hex(max(int(255.0 - params['ppl_scale'] * (float(ppl) - 1.0)), 0))[2:]
-    if len(value) < 2:
-        value = '0' * (2 - len(value)) + value
+    # hue (0.0 = red)
+    # saturation (1.0 = red)
+    # brightness (0.0 = black, 1.0 = red)
+    # scale saturation from white to red the higher the perplexity
 
-    return 'ff' + value + value
+    ppl = min(ppl, params['ppl_scale'])  # clip ppl to 0-params['ppl_scale'] for color scaling. 15 should be fine for clipping and scaling
+    sat = ppl / params['ppl_scale']
+    rv, gv, bv = colorsys.hsv_to_rgb(0.0, sat, 1.0)
 
+    # to hex
+    hex_col = f"{int(rv*255):02x}{int(gv*255):02x}{int(bv*255):02x}"
+    
+    return hex_col
 
-def probability_perplexity_color_scale(prob, ppl):
+
+def probability_perplexity_color_scale(prob, max_prob, ppl):
     '''
-    Green-yellow-red for probability and blue component for perplexity
+    Green-yellow-red for relative probability compared to maximum for the current token, and blue component for perplexity
     '''
-
-    rv = 0
-    gv = 0
-    bv = hex(min(max(int(params['ppl_scale'] * (float(ppl) - 1.0)), 0), 255))[2:]
-    if len(bv) < 2:
-        bv = '0' * (2 - len(bv)) + bv
-
-    if prob <= 0.5:
-        rv = 'ff'
-        gv = hex(int(255 * prob * 2))[2:]
-        if len(gv) < 2:
-            gv = '0' * (2 - len(gv)) + gv
-    else:
-        rv = hex(int(255 - 255 * (prob - 0.5) * 2))[2:]
-        gv = 'ff'
-        if len(rv) < 2:
-            rv = '0' * (2 - len(rv)) + rv
-
-    return rv + gv + bv
+    hue = prob/max_prob * 0.33
+    rv, gv, _ = colorsys.hsv_to_rgb(hue, 1.0, 1.0)
+    
+    ppl = min(ppl, params['ppl_scale'])  # clip ppl to 0-params['ppl_scale'] for color scaling. 15 should be fine for clipping and scaling
+    bv = ppl / params['ppl_scale']
+    
+    # to hex
+    hex_col = f"{int(rv*255):02x}{int(gv*255):02x}{int(bv*255):02x}"
+    
+    return hex_col
 
 
 def add_color_html(token, color):
-    return f'{token}'
+    output = ''
+    output += f'{html.escape(repr(token)[1:-1])}'
+    #if '\n' in token or '\r' in token: #token.isspace():
+    #    output += '
            '
+    return output
+
 
+# TODO: Might also need message index for the click-to-regenerate feature to work... For now it only works in the last message, which I think is fine.
 
-# TODO: Major issue: Applying this to too many tokens will cause a permanent slowdown in generation speed until the messages are removed from the history.
+# TODO: Major issue: Applying this to too many tokens will cause a permanent slowdown in generation speed until the messages are removed from the history. The slowdown seems to be mostly resolved in the current version though
 # I think the issue is from HTML elements taking up space in the visible history, and things like history deepcopy add latency proportional to the size of the history.
 # Potential solution is maybe to modify the main generation code to send just the internal text and not the visible history, to avoid moving too much around.
 # I wonder if we can also avoid using deepcopy here.
-def add_dropdown_html(token, color, top_tokens, top_probs, perplexity=0):
-    html = f'
            {token}
            '
-    for token_option, prob in zip(top_tokens, top_probs):
+def add_dropdown_html(token, index, color, top_tokens, top_probs, perplexity=0):
+    #print("Token:", token, token.isspace(), '\n' in token or '\r' in token)
+    output = ''
+    # Use the repr to get characters like \n visible. Exclude the quotes around it
+    output += f'
            {html.escape(repr(token)[1:-1])}
            '
+    for i, token_option, prob in zip(range(len(top_tokens)), top_tokens, top_probs):
         # TODO: Bold for selected token?
         # Using divs prevented the problem of divs inside spans causing issues.
         # Now the problem is that divs show the same whitespace of one space between every token.
         # There is probably some way to fix this in CSS that I don't know about.
         row_color = probability_color_scale(prob)
         row_class = ' class="selected"' if token_option == token else ''
-        html += f''
+        # This time we want to include the quotes around it so that we can see where the spaces are.
+        output += f''
     if perplexity != 0:
         ppl_color = perplexity_color_scale(perplexity)
-        html += f''
-    html += '
            {token_option}{prob:.4f}
            {html.escape(repr(token_option))}{prob:.4f}
            Perplexity:{perplexity:.4f}
            '
-    return html  # About 750 characters per token...
+        output += f'Perplexity:{perplexity:.4f}'
+    output += '
            '
+    #if '\n' in token or '\r' in token: #token.isspace():
+    #    output += '
            ' # I imagine this will cause problems sometimes
+    return output  # About 750 characters per token...
 
 
 def custom_css():
@@ -223,8 +265,8 @@ def custom_css():
             display: none;
             position: absolute;
             z-index: 50;
-            background-color: var(--block-background-fill);
-            box-shadow: 0px 8px 16px 0px rgba(0,0,0,0.2);
+            background-color: var(--background-fill-secondary);
+            box-shadow: 0px 8px 16px 0px rgba(0,0,0,1.0);
             width: max-content;
             overflow: visible;
             padding: 5px;
@@ -238,7 +280,7 @@ def custom_css():
         }
 
         .dropdown-content tr.selected {
-            background-color: var(--block-label-background-fill);
+            background-color: var(--background-fill-primary);
         }
 
         .dropdown-content td {
@@ -267,21 +309,106 @@ def custom_css():
         # TODO: This makes the hover menus extend outside the bounds of the chat area, which is good.
         # However, it also makes the scrollbar disappear, which is bad.
         # The scroll bar needs to still be present. So for now, we can't see dropdowns that extend past the edge of the chat area.
-        #.chat {
-        #    overflow-y: auto;
-        #}
+        .chat {
+            overflow-y: auto;
+        }
     """
 
+def custom_js():
+    return """
+
+function sleep(ms) {
+    return new Promise(resolve => setTimeout(resolve, ms));
+}    
+
+// Note that this will only work as intended on the last agent message
+document.addEventListener("click", async function(event) {
+    //console.log(event.target);
+    const id = event.target.id;
+    if (id.includes("opt_")) {
+        const id_parts = id.split("_");
+        const token_index = id_parts[1];
+        const option_index = id_parts[2];
+        // Exclude the quotes and convert newlines... Not sure about the newlines though
+        // TODO: Seems like continuing generation from a newline causes problems whether you add it or not!
+        const token_string = event.target.innerHTML.substring(1, event.target.innerHTML.length-1).replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"r", "g"), '').replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"n", "g"), '');
+        //console.log(token_index + ", " + option_index + ", " + token_string);
+        // Get all the previous text (I'm sure there is a more efficient way to do this)
+        var msg_text = ""
+        const msg_html = event.target.parentElement.parentElement.parentElement.parentElement.parentElement.parentElement;
+        var msg_parts = msg_html.childNodes;
+        for (var i = 0; i < msg_parts.length; i++) {
+            var msg_part = msg_parts[i];
+            if (msg_part.nodeType === Node.ELEMENT_NODE) {
+                if (msg_part.nodeName == "DIV") {
+                    var current_token_index = msg_part.id.split("_")[1];
+                    if (current_token_index == token_index) {
+                        // Use the replacement token
+                        // TODO: Don't have access to the tokenizer here, and sometimes there needs to be a space added before this token
+                        msg_text += token_string //.replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"r", "g"), '').replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"n", "g"), '');
+                        break;
+                    }
+                    else {
+                        // Replace here or at the end?
+                        var text = msg_part.firstChild.innerHTML.replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"r", "g"), '').replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"n", "g"), '')
+                        msg_text += text;
+                    }
+                }
+                else {
+                    // Break tag (hacky workaround because the newline literal can't be parsed here)
+                    //msg_text += String.fromCharCode(10);
+                    // Do nothing???
+                }
+            }
+            else if (msg_part.nodeType === Node.TEXT_NODE) {
+                msg_text +=  msg_part.textContent;
+            }
+        }
+        var textbox = document.querySelector("#chat-input textarea");
+        textbox.focus();
+        textbox.value = msg_text.trimStart() // Fix initial tokenization spaces
+        //console.log(textbox.value);
+        
+        // Add some delays to make sure it's processed correctly. Without these, there's a chance the events don't go through correctly and it doesn't work
+        // It's unknown how long this will take, and probably depends on the size of the message...
+        // It would be better to somehow wait for gradio to update instead of waiting a fixed amount of time.
+        // Hopefully 1 second of delay before starting generation isn't unacceptable.
+        var inputEvent = new Event('input', {
+            bubbles: true,
+            cancelable: true,
+        });
+        textbox.dispatchEvent(inputEvent);
+        var changeEvent = new Event('change', {
+            bubbles: true,
+            cancelable: true,
+        });
+        textbox.dispatchEvent(changeEvent);
+        await sleep(250);
+        document.getElementById("Replace-last").click();
+        // This can take a while to execute
+        await sleep(750);
+        document.getElementById("Continue").click();
+    }
+});
+
+console.log("Custom JS for perplexity_colors loaded");
+"""
 
 # Monkeypatch applied to html_generator.py
 # We simply don't render markdown into HTML. We wrap everything in 
             tags to preserve whitespace
 # formatting. If you're coloring tokens by perplexity or probability, or especially if you're using
 # the probability dropdown, you probably care more about seeing the tokens the model actually outputted
 # rather than rendering ```code blocks``` or *italics*.
+@functools.lru_cache(maxsize=4096)
 def convert_to_markdown(string):
     return '
            ' + string + '
            '
 
+def convert_to_markdown_wrapped(string, use_cache=True):
+    if use_cache:
+        return convert_to_markdown(string)
+    return convert_to_markdown.__wrapped__(string)
 
+# This is still necessary for formatting to work correctly
 html_generator.convert_to_markdown = convert_to_markdown
 
 
@@ -298,7 +425,7 @@ def update_color_by_prob_check(x):
     def update_prob_dropdown_check(x):
         params.update({'probability_dropdown': x})
 
-    active_check = gradio.Checkbox(value=True, label="Compute probabilities and perplexity scores", info="Activate this extension. Note that this extension currently does not work with exllama or llama.cpp.")
+    active_check = gradio.Checkbox(value=True, label="Compute probabilities and perplexity scores", info="Activate this extension. Note that this extension currently does not work with llama.cpp, but it does work with ExLlamav2_HF and llamacpp_HF when set up correctly")
     color_by_ppl_check = gradio.Checkbox(value=False, label="Color by perplexity", info="Higher perplexity is more red. If also showing probability, higher perplexity has more blue component.")
     color_by_prob_check = gradio.Checkbox(value=False, label="Color by probability", info="Green-yellow-red linear scale, with 100% green, 50% yellow, 0% red.")
     prob_dropdown_check = gradio.Checkbox(value=False, label="Probability dropdown", info="Hover over a token to show a dropdown of top token probabilities. Currently slightly buggy with whitespace between tokens.")

From 16fa9215c4cfb0d280ca470db30bda1743afc068 Mon Sep 17 00:00:00 2001
From: Kelvie Wong 
Date: Tue, 18 Feb 2025 07:01:30 -0800
Subject: [PATCH 0236/1701] Fix OpenAI API with new param (show_after), closes
 #6747 (#6749)

---------

Co-authored-by: oobabooga 
---
 modules/chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 2852aaf333..66ab8c743a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -412,7 +412,7 @@ def generate_chat_reply(text, state, regenerate=False, _continue=False, loading_
             yield history
             return
 
-    show_after = html.escape(state["show_after"]) if state["show_after"] else None
+    show_after = html.escape(state.get("show_after")) if state.get("show_after") else None
     for history in chatbot_wrapper(text, state, regenerate=regenerate, _continue=_continue, loading_message=loading_message, for_ui=for_ui):
         if show_after:
             after = history["visible"][-1][1].partition(show_after)[2] or "*Is thinking...*"

From 769eee1ff30d922cb15b89c9840d8147c077231f Mon Sep 17 00:00:00 2001
From: Kelvie Wong 
Date: Tue, 18 Feb 2025 07:01:30 -0800
Subject: [PATCH 0237/1701] Fix OpenAI API with new param (show_after), closes
 #6747 (#6749)

---------

Co-authored-by: oobabooga 
---
 modules/chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 2852aaf333..66ab8c743a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -412,7 +412,7 @@ def generate_chat_reply(text, state, regenerate=False, _continue=False, loading_
             yield history
             return
 
-    show_after = html.escape(state["show_after"]) if state["show_after"] else None
+    show_after = html.escape(state.get("show_after")) if state.get("show_after") else None
     for history in chatbot_wrapper(text, state, regenerate=regenerate, _continue=_continue, loading_message=loading_message, for_ui=for_ui):
         if show_after:
             after = history["visible"][-1][1].partition(show_after)[2] or "*Is thinking...*"

From a12e05d9c0818d5f4a709ddcbf4cd973e4aa7f05 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Wed, 12 Mar 2025 16:11:03 -0300
Subject: [PATCH 0238/1701] Bump jinja2 from 3.1.5 to 3.1.6 (#6786)

---
 requirements.txt                 | 2 +-
 requirements_amd.txt             | 2 +-
 requirements_amd_noavx2.txt      | 2 +-
 requirements_apple_intel.txt     | 2 +-
 requirements_apple_silicon.txt   | 2 +-
 requirements_cpu_only.txt        | 2 +-
 requirements_cpu_only_noavx2.txt | 2 +-
 requirements_noavx2.txt          | 2 +-
 requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index d09f6bf582..a4812ac5f8 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -5,7 +5,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.5
+jinja2==3.1.6
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 124ad6b64a..972979bc2b 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -4,7 +4,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.5
+jinja2==3.1.6
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 9e3063c32a..d40db5139f 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -4,7 +4,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.5
+jinja2==3.1.6
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 0ef8db345e..e82c198ad5 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -4,7 +4,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.5
+jinja2==3.1.6
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 9c4bcc1166..e23f465ee3 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -4,7 +4,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.5
+jinja2==3.1.6
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index 9f19238a1a..fc5c52eb85 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -4,7 +4,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.5
+jinja2==3.1.6
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 042dbbd82f..1201ab05b2 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -4,7 +4,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.5
+jinja2==3.1.6
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index f488cafcd5..38df6e694c 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -5,7 +5,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.5
+jinja2==3.1.6
 markdown
 numba==0.59.*
 numpy==1.26.*
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index 1bc2e385c6..858ffff59a 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -4,7 +4,7 @@ datasets
 einops
 fastapi==0.112.4
 gradio==4.37.*
-jinja2==3.1.5
+jinja2==3.1.6
 markdown
 numba==0.59.*
 numpy==1.26.*

From 39fded487af6c80bcbf353efcd7562494e7c9cb2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 12 Mar 2025 17:54:30 -0700
Subject: [PATCH 0239/1701] Bump ExllamaV2 to 0.2.8

---
 requirements.txt               | 10 +++++-----
 requirements_amd.txt           |  6 +++---
 requirements_amd_noavx2.txt    |  6 +++---
 requirements_apple_intel.txt   |  2 +-
 requirements_apple_silicon.txt |  2 +-
 requirements_noavx2.txt        | 10 +++++-----
 6 files changed, 18 insertions(+), 18 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index a4812ac5f8..77de5853e4 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -50,11 +50,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # CUDA wheels
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 972979bc2b..761805e16b 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -39,6 +39,6 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.7+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.7+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index d40db5139f..799044b36c 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -37,6 +37,6 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # AMD wheels
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index e82c198ad5..435201ff2a 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -35,4 +35,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp310-cp310-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index e23f465ee3..34e8a3b6a3 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -37,4 +37,4 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/me
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 38df6e694c..cd41d69f82 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -50,11 +50,11 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # CUDA wheels
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.7/exllamav2-0.2.7-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

From 0261338910ba4a0f0f7d117d5152ae1fa938b70c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 12 Mar 2025 17:55:25 -0700
Subject: [PATCH 0240/1701] Bump llama-cpp-python to 0.3.8

---
 requirements.txt                 | 24 ++++++++++++------------
 requirements_amd.txt             | 12 ++++++------
 requirements_amd_noavx2.txt      |  8 ++++----
 requirements_apple_intel.txt     |  8 ++++----
 requirements_apple_silicon.txt   | 12 ++++++------
 requirements_cpu_only.txt        |  8 ++++----
 requirements_cpu_only_noavx2.txt |  8 ++++----
 requirements_noavx2.txt          | 24 ++++++++++++------------
 8 files changed, 52 insertions(+), 52 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 77de5853e4..83bd3a5334 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -32,22 +32,22 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, with GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, without GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # CUDA wheels
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 761805e16b..1e757ffe54 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -31,14 +31,14 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.7+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.7+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.8+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.8+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 799044b36c..f74ebf6954 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -31,10 +31,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # AMD wheels
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 435201ff2a..dcdeae3faf 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -31,8 +31,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp310-cp310-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp310-cp310-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 34e8a3b6a3..b823e40e0d 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -31,10 +31,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp310-cp310-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.7-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp310-cp310-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index fc5c52eb85..fe3f522a2c 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -31,7 +31,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 1201ab05b2..014e2e5d93 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -31,7 +31,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index cd41d69f82..6139c46e51 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -32,22 +32,22 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama-cpp-python (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.7+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, with GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.7+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, without GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.7+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # CUDA wheels
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"

From f04a37adc29304cd8216ea38d0af2ea6722947ab Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 14 Mar 2025 05:20:15 -0700
Subject: [PATCH 0241/1701] UI: improved scrollbar styles

---
 css/main.css | 50 ++++++++++++++++++++++++++++++++++----------------
 1 file changed, 34 insertions(+), 16 deletions(-)

diff --git a/css/main.css b/css/main.css
index 234923384a..b4544693ef 100644
--- a/css/main.css
+++ b/css/main.css
@@ -248,40 +248,58 @@ button {
     font-size: 100% !important;
 }
 
+.pretty_scrollbar {
+    scrollbar-width: thin;
+    scrollbar-color: rgb(140 140 140 / 35%) transparent;
+}
+
+.dark .pretty_scrollbar {
+    scrollbar-width: thin;
+    scrollbar-color: rgb(204 204 204 / 35%) transparent;
+}
+
 .pretty_scrollbar::-webkit-scrollbar {
-    width: 7px;
-    height: 7px;
+    width: 8px;
+    height: 8px;
 }
 
 .pretty_scrollbar::-webkit-scrollbar-track {
     background: transparent;
 }
 
-.pretty_scrollbar::-webkit-scrollbar-thumb,
-.pretty_scrollbar::-webkit-scrollbar-thumb:hover {
-    background: var(--neutral-300);
-    border-radius: 30px;
+.pretty_scrollbar::-webkit-scrollbar-thumb {
+    background: rgb(140 140 140 / 35%);
+    border-radius: 4px;
+    transition: background 0.2s ease;
 }
 
-.dark .pretty_scrollbar::-webkit-scrollbar-thumb,
-.dark .pretty_scrollbar::-webkit-scrollbar-thumb:hover {
-    background: #ccc;
-    border-radius: 10px;
+.pretty_scrollbar::-webkit-scrollbar-thumb:hover {
+    background: rgb(140 140 140 / 55%);
 }
 
-.pretty_scrollbar::-webkit-resizer {
-    background: #c5c5d2;
+.dark .pretty_scrollbar::-webkit-scrollbar-thumb {
+    background: rgb(204 204 204 / 35%);
+    border-radius: 4px;
 }
 
-.dark .pretty_scrollbar::-webkit-resizer {
-    background: #ccc;
-    border-radius: 10px;
+.dark .pretty_scrollbar::-webkit-scrollbar-thumb:hover {
+    background: rgb(204 204 204 / 55%);
 }
 
 .pretty_scrollbar::-webkit-scrollbar-corner {
     background: transparent;
 }
 
+.pretty_scrollbar::-webkit-resizer {
+    background: rgb(140 140 140 / 35%);
+    border-radius: 4px;
+}
+
+.dark .pretty_scrollbar::-webkit-resizer {
+    background: rgb(204 204 204 / 35%);
+    border-radius: 4px;
+}
+
 audio {
     max-width: 100%;
 }
@@ -295,7 +313,7 @@ audio {
     width: 0;
     text-align: left;
     direction: rtl;
-    right: 5px;
+    right: 13px;
 }
 
 #default-token-counter {

From 26317a4c7e371c59e03be4ac39c62505c43f7aad Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 14 Mar 2025 10:59:05 -0700
Subject: [PATCH 0242/1701] Fix jinja2 error while loading
 c4ai-command-a-03-2025

---
 modules/chat.py                    |  7 +++++-
 modules/llama_cpp_python_hijack.py | 40 ++++++++++++++++++++++++++++++
 2 files changed, 46 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 66ab8c743a..fd94990750 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -11,6 +11,7 @@
 
 import gradio as gr
 import yaml
+from jinja2.ext import loopcontrols
 from jinja2.sandbox import ImmutableSandboxedEnvironment
 from PIL import Image
 
@@ -35,7 +36,11 @@ def strftime_now(format):
     return datetime.now().strftime(format)
 
 
-jinja_env = ImmutableSandboxedEnvironment(trim_blocks=True, lstrip_blocks=True)
+jinja_env = ImmutableSandboxedEnvironment(
+    trim_blocks=True,
+    lstrip_blocks=True,
+    extensions=[loopcontrols]
+)
 jinja_env.globals["strftime_now"] = strftime_now
 
 
diff --git a/modules/llama_cpp_python_hijack.py b/modules/llama_cpp_python_hijack.py
index f3872a7446..c03c28a772 100644
--- a/modules/llama_cpp_python_hijack.py
+++ b/modules/llama_cpp_python_hijack.py
@@ -121,5 +121,45 @@ def my_generate(self, *args, **kwargs):
     lib.Llama.original_generate = lib.Llama.generate
     lib.Llama.generate = my_generate
 
+    # Also patch Jinja2ChatFormatter to handle loop controls
+    if hasattr(lib, 'llama_chat_format') and hasattr(lib.llama_chat_format, 'Jinja2ChatFormatter'):
+        Formatter = lib.llama_chat_format.Jinja2ChatFormatter
+
+        if not getattr(Formatter, '_is_patched', False):
+            def patched_init(self, *args, **kwargs):
+                # Extract parameters from args or kwargs
+                if args:
+                    self.template = args[0]
+                    self.eos_token = args[1] if len(args) > 1 else kwargs.get('eos_token')
+                    self.bos_token = args[2] if len(args) > 2 else kwargs.get('bos_token')
+                    self.add_generation_prompt = args[3] if len(args) > 3 else kwargs.get('add_generation_prompt', True)
+                    self.stop_token_ids = args[4] if len(args) > 4 else kwargs.get('stop_token_ids')
+                else:
+                    self.template = kwargs.get('template')
+                    self.eos_token = kwargs.get('eos_token')
+                    self.bos_token = kwargs.get('bos_token')
+                    self.add_generation_prompt = kwargs.get('add_generation_prompt', True)
+                    self.stop_token_ids = kwargs.get('stop_token_ids')
+
+                # Process stop tokens as in the original
+                self.stop_token_ids = (
+                    set(self.stop_token_ids) if self.stop_token_ids is not None else None
+                )
+
+                # Create environment with loopcontrols extension
+                from jinja2.ext import loopcontrols
+                import jinja2
+
+                self._environment = jinja2.sandbox.ImmutableSandboxedEnvironment(
+                    loader=jinja2.BaseLoader(),
+                    trim_blocks=True,
+                    lstrip_blocks=True,
+                    extensions=[loopcontrols]
+                ).from_string(self.template)
+
+            # Replace the original __init__ with our patched version
+            Formatter.__init__ = patched_init
+            Formatter._is_patched = True
+
     # Set the flag to indicate that the patch has been applied
     lib.Llama._is_patched = True

From 6ab04698f6f74ac127156807698087bcb68dac3f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 14 Mar 2025 12:03:49 -0700
Subject: [PATCH 0243/1701] UI: improve the light mode left sidebar color

---
 css/main.css | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/css/main.css b/css/main.css
index b4544693ef..8082e3cc43 100644
--- a/css/main.css
+++ b/css/main.css
@@ -249,7 +249,7 @@ button {
 }
 
 .pretty_scrollbar {
-    scrollbar-width: thin;
+    scrollbar-width: thin !important;
     scrollbar-color: rgb(140 140 140 / 35%) transparent;
 }
 
@@ -1181,7 +1181,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 .header_bar button.selected {
-    background: white;
+    background: #E0E0E0;
 }
 
 #chat-controls,

From 677d74a6a0e6bec0417caa6dde002af00b194328 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 14 Mar 2025 12:09:56 -0700
Subject: [PATCH 0244/1701] Revert "UI: improved scrollbar styles", add just a
 small change instead

---
 css/main.css | 44 +++++++++++++-------------------------------
 1 file changed, 13 insertions(+), 31 deletions(-)

diff --git a/css/main.css b/css/main.css
index 8082e3cc43..a3fa9753f0 100644
--- a/css/main.css
+++ b/css/main.css
@@ -248,16 +248,6 @@ button {
     font-size: 100% !important;
 }
 
-.pretty_scrollbar {
-    scrollbar-width: thin !important;
-    scrollbar-color: rgb(140 140 140 / 35%) transparent;
-}
-
-.dark .pretty_scrollbar {
-    scrollbar-width: thin;
-    scrollbar-color: rgb(204 204 204 / 35%) transparent;
-}
-
 .pretty_scrollbar::-webkit-scrollbar {
     width: 8px;
     height: 8px;
@@ -267,37 +257,29 @@ button {
     background: transparent;
 }
 
-.pretty_scrollbar::-webkit-scrollbar-thumb {
-    background: rgb(140 140 140 / 35%);
-    border-radius: 4px;
-    transition: background 0.2s ease;
-}
-
+.pretty_scrollbar::-webkit-scrollbar-thumb,
 .pretty_scrollbar::-webkit-scrollbar-thumb:hover {
-    background: rgb(140 140 140 / 55%);
-}
-
-.dark .pretty_scrollbar::-webkit-scrollbar-thumb {
-    background: rgb(204 204 204 / 35%);
-    border-radius: 4px;
+    background: var(--neutral-300);
+    border-radius: 30px;
 }
 
+.dark .pretty_scrollbar::-webkit-scrollbar-thumb,
 .dark .pretty_scrollbar::-webkit-scrollbar-thumb:hover {
-    background: rgb(204 204 204 / 55%);
-}
-
-.pretty_scrollbar::-webkit-scrollbar-corner {
-    background: transparent;
+    background: #ccc;
+    border-radius: 10px;
 }
 
 .pretty_scrollbar::-webkit-resizer {
-    background: rgb(140 140 140 / 35%);
-    border-radius: 4px;
+    background: #c5c5d2;
 }
 
 .dark .pretty_scrollbar::-webkit-resizer {
-    background: rgb(204 204 204 / 35%);
-    border-radius: 4px;
+    background: #ccc;
+    border-radius: 10px;
+}
+
+.pretty_scrollbar::-webkit-scrollbar-corner {
+    background: transparent;
 }
 
 audio {

From 5bcd2d7ad01a211198df1a133a29cfb682336e0b Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Fri, 14 Mar 2025 16:45:11 -0300
Subject: [PATCH 0245/1701] Add the top N-sigma sampler (#6796)

---
 extensions/openai/typing.py |  1 +
 modules/loaders.py          |  3 +++
 modules/presets.py          |  3 ++-
 modules/sampler_hijack.py   | 53 +++++++++++++++++++++++++++++++++++--
 modules/text_generation.py  |  1 +
 modules/ui.py               |  1 +
 modules/ui_parameters.py    |  1 +
 7 files changed, 60 insertions(+), 3 deletions(-)

diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 5f0e01281e..ea68889704 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -21,6 +21,7 @@ class GenerationOptions(BaseModel):
     eta_cutoff: float = 0
     tfs: float = 1
     top_a: float = 0
+    top_n_sigma: float = 0
     dry_multiplier: float = 0
     dry_allowed_length: int = 2
     dry_base: float = 1.75
diff --git a/modules/loaders.py b/modules/loaders.py
index cd864e406d..88ded1d10b 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -137,6 +137,7 @@ def transformers_samplers():
         'eta_cutoff',
         'tfs',
         'top_a',
+        'top_n_sigma',
         'dry_multiplier',
         'dry_allowed_length',
         'dry_base',
@@ -224,6 +225,7 @@ def transformers_samplers():
         'eta_cutoff',
         'tfs',
         'top_a',
+        'top_n_sigma',
         'dry_multiplier',
         'dry_allowed_length',
         'dry_base',
@@ -288,6 +290,7 @@ def transformers_samplers():
         'eta_cutoff',
         'tfs',
         'top_a',
+        'top_n_sigma',
         'dry_multiplier',
         'dry_allowed_length',
         'dry_base',
diff --git a/modules/presets.py b/modules/presets.py
index b841af53d6..7cab2af00b 100644
--- a/modules/presets.py
+++ b/modules/presets.py
@@ -28,6 +28,7 @@ def default_preset():
         'eta_cutoff': 0,
         'tfs': 1,
         'top_a': 0,
+        'top_n_sigma': 0,
         'dry_multiplier': 0,
         'dry_allowed_length': 2,
         'dry_base': 1.75,
@@ -45,7 +46,7 @@ def default_preset():
         'do_sample': True,
         'dynamic_temperature': False,
         'temperature_last': False,
-        'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ndry\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nmirostat\nxtc\nencoder_repetition_penalty\nno_repeat_ngram',
+        'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ndry\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_n_sigma\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nmirostat\nxtc\nencoder_repetition_penalty\nno_repeat_ngram',
         'dry_sequence_breakers': '"\\n", ":", "\\"", "*"',
     }
 
diff --git a/modules/sampler_hijack.py b/modules/sampler_hijack.py
index e0df49c3ce..e6883289e4 100644
--- a/modules/sampler_hijack.py
+++ b/modules/sampler_hijack.py
@@ -5,7 +5,6 @@
 
 import torch
 import transformers
-from transformers import LogitsProcessor
 from transformers.generation.logits_process import (
     LogitNormalization,
     LogitsProcessor,
@@ -193,6 +192,46 @@ def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> to
         return scores
 
 
+class TopNSigmaLogitsWarper(LogitsProcessor):
+    def __init__(self, n_sigma: float = 2.0, filter_value: float = -float("Inf"), min_tokens_to_keep: int = 1):
+        """
+        Initialize Top-nσ Sampling logits warper.
+
+        Args:
+            n_sigma: The threshold multiplier for standard deviation
+            filter_value: Value to assign to filtered logits
+            min_tokens_to_keep: Minimum number of tokens to keep
+        """
+        if n_sigma < 0:
+            raise ValueError(f"`n_sigma` must be a non-negative float, but is {n_sigma}")
+        self.n_sigma = n_sigma
+        self.filter_value = filter_value
+        self.min_tokens_to_keep = min_tokens_to_keep
+
+    def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> torch.FloatTensor:
+        # Calculate max of logits
+        max_logit = torch.max(scores, dim=-1, keepdim=True)[0]
+
+        # Calculate standard deviation only on finite values
+        finite_mask = torch.isfinite(scores)
+        finite_scores = scores.masked_fill(~finite_mask, 0.0)
+        std_logit = torch.std(finite_scores, dim=-1, keepdim=True)
+
+        # Create mask where tokens with logits >= max_logit - n_sigma * std_logit are kept
+        threshold = max_logit - self.n_sigma * std_logit
+        indices_to_remove = scores < threshold
+
+        if self.min_tokens_to_keep > 1:
+            # Keep at least min_tokens_to_keep tokens
+            top_k_indices = torch.topk(scores, self.min_tokens_to_keep, dim=-1)[1]
+            indices_to_remove.scatter_(-1, top_k_indices, False)
+
+        # Apply mask by setting filtered tokens to filter_value
+        scores = scores.masked_fill(indices_to_remove, self.filter_value)
+
+        return scores
+
+
 # Exclude Top Choices (XTC)
 class XTCLogitsWarper(LogitsProcessor):
     def __init__(self, threshold: float, probability: float, filter_value: float = -float("Inf")):
@@ -525,6 +564,14 @@ def get_logits_processor_patch(self, **kwargs):
             )
         )
 
+    if generation_config.top_n_sigma is not None and generation_config.top_n_sigma > 0.0:
+        warpers_to_add.append(
+            TopNSigmaLogitsWarper(
+                n_sigma=generation_config.top_n_sigma,
+                min_tokens_to_keep=min_tokens_to_keep
+            )
+        )
+
     if generation_config.xtc_probability is not None and generation_config.xtc_probability > 0:
         warpers_to_add.append(
             XTCLogitsWarper(
@@ -589,6 +636,7 @@ def get_logits_processor_patch(self, **kwargs):
         'TailFreeLogitsWarper': 'tfs',
         'TemperatureLogitsWarperCustom': 'temperature',
         'TopALogitsWarper': 'top_a',
+        'TopNSigmaLogitsWarper': 'top_n_sigma',
         'TopKLogitsWarper': 'top_k',
         'TopPLogitsWarper': 'top_p',
         'TypicalLogitsWarper': 'typical_p',
@@ -636,6 +684,7 @@ def generation_config_init_patch(self, **kwargs):
     self.smoothing_curve = kwargs.pop("smoothing_curve", 1.0)
     self.tfs = kwargs.pop("tfs", 1.0)
     self.top_a = kwargs.pop("top_a", 0.0)
+    self.top_n_sigma = kwargs.pop("top_n_sigma", 0.0)
     self.mirostat_mode = kwargs.pop("mirostat_mode", 0)
     self.mirostat_eta = kwargs.pop("mirostat_eta", 0.1)
     self.mirostat_tau = kwargs.pop("mirostat_tau", 5)
@@ -649,7 +698,7 @@ def generation_config_init_patch(self, **kwargs):
     self.xtc_threshold = kwargs.pop("xtc_threshold", 0.1)
     self.xtc_probability = kwargs.pop("xtc_probability", 0)
     self.temperature_last = kwargs.pop("temperature_last", False)
-    self.sampler_priority = kwargs.pop("sampler_priority", ['repetition_penalty', 'presence_penalty', 'frequency_penalty', 'dry', 'temperature', 'dynamic_temperature', 'quadratic_sampling', 'top_k', 'top_p', 'typical_p', 'epsilon_cutoff', 'eta_cutoff', 'tfs', 'top_a', 'min_p', 'mirostat', 'xtc', 'encoder_repetition_penalty', 'no_repeat_ngram'])
+    self.sampler_priority = kwargs.pop("sampler_priority", ['repetition_penalty', 'presence_penalty', 'frequency_penalty', 'dry', 'temperature', 'dynamic_temperature', 'quadratic_sampling', 'top_n_sigma', 'top_k', 'top_p', 'typical_p', 'epsilon_cutoff', 'eta_cutoff', 'tfs', 'top_a', 'min_p', 'mirostat', 'xtc', 'encoder_repetition_penalty', 'no_repeat_ngram'])
 
 
 def hijack_samplers():
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 152b2b8df0..eff6495ecf 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -302,6 +302,7 @@ def generate_reply_HF(question, original_question, seed, state, stopping_strings
         'xtc_probability',
         'tfs',
         'top_a',
+        'top_n_sigma',
         'dry_multiplier',
         'dry_allowed_length',
         'dry_base',
diff --git a/modules/ui.py b/modules/ui.py
index b776e19c37..adbb67b086 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -183,6 +183,7 @@ def list_interface_input_elements():
         'eta_cutoff',
         'tfs',
         'top_a',
+        'top_n_sigma',
         'dry_multiplier',
         'dry_allowed_length',
         'dry_base',
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 265840ed4c..846fcfe77d 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -37,6 +37,7 @@ def create_ui(default_preset):
 
                             gr.Markdown('## Curve cutoff')
                             shared.gradio['min_p'] = gr.Slider(0.0, 1.0, value=generate_params['min_p'], step=0.01, label='min_p')
+                            shared.gradio['top_n_sigma'] = gr.Slider(0.0, 5.0, value=generate_params['top_n_sigma'], step=0.01, label='top_n_sigma')
                             shared.gradio['top_p'] = gr.Slider(0.0, 1.0, value=generate_params['top_p'], step=0.01, label='top_p')
                             shared.gradio['top_k'] = gr.Slider(0, 200, value=generate_params['top_k'], step=1, label='top_k')
                             shared.gradio['typical_p'] = gr.Slider(0.0, 1.0, value=generate_params['typical_p'], step=0.01, label='typical_p')

From 60d67994d9703c44ff1f56fd3aa8a51d0e7658ad Mon Sep 17 00:00:00 2001
From: SeanScripts <64337075+SeanScripts@users.noreply.github.com>
Date: Fri, 14 Mar 2025 12:45:53 -0700
Subject: [PATCH 0246/1701] Perplexity colors extension updates (#6764)

---
 extensions/perplexity_colors/script.py | 80 ++++++++++++++++++--------
 1 file changed, 55 insertions(+), 25 deletions(-)

diff --git a/extensions/perplexity_colors/script.py b/extensions/perplexity_colors/script.py
index 849e4e630c..d032cebd94 100644
--- a/extensions/perplexity_colors/script.py
+++ b/extensions/perplexity_colors/script.py
@@ -96,23 +96,42 @@ def logits_processor_modifier(logits_processor_list, input_ids):
         logits_processor_list.append(ppl_logits_processor)
 
 
+def get_last_token(text, tokens_list, token_ids_list, token_probs_list):
+    for token, token_id, prob in zip(tokens_list, token_ids_list, token_probs_list):
+        if text.strip().endswith(token.strip()): # Whitespace could be a problem
+            return token, token_id, prob
+    # Unknown?
+    print("Last token not found in list:", tokens_list)
+    return '', -1, 0.0
+
+
 def output_modifier(text):
     global ppl_logits_processor
     #t0 = time.time()
+    original_text = text
 
     if not params['active'] or ppl_logits_processor is None:
         return text
 
+    # Space at the beginning to account for tokenization spaces...
+    text = ' ' + html.unescape(text)
+
     # TODO: It's probably more efficient to do this above rather than modifying all these lists
     # Remove last element of perplexities_list, top_token_ids_list, top_tokens_list, top_probs_list since everything is off by one because this extension runs before generation
-    perplexities = ppl_logits_processor.perplexities_list[:-1]
-    top_token_ids_list = ppl_logits_processor.top_token_ids_list[:-1]
+    perplexities = ppl_logits_processor.perplexities_list
+    top_token_ids_list = ppl_logits_processor.top_token_ids_list
     top_tokens_list = [[shared.tokenizer.decode(token_id) for token_id in top_token_ids[0]] for top_token_ids in top_token_ids_list]
-    top_probs_list = ppl_logits_processor.top_probs_list[:-1]
+    top_probs_list = ppl_logits_processor.top_probs_list
     # Remove first element of generated_token_ids, generated_tokens, selected_probs because they are for the last token of the prompt
     gen_token_ids = ppl_logits_processor.generated_token_ids[1:]
+    # Add last sampled token, if possible (it could be past the end of the top 5 list)
+    last_token, last_token_id, last_prob = get_last_token(text, top_tokens_list[-1], top_token_ids_list[-1][0], top_probs_list[-1][0])
+    if last_token_id != -1:
+        gen_token_ids.append(last_token_id)
     gen_tokens = [shared.tokenizer.decode(token_id) for token_id in gen_token_ids]
     sel_probs = ppl_logits_processor.selected_probs[1:]
+    if last_token_id != -1:
+        sel_probs.append(last_prob)
 
     end_part = '
            ' if params['probability_dropdown'] else ''  # Helps with finding the index after replacing part of the text.
 
@@ -120,8 +139,7 @@ def output_modifier(text):
     # Used to find where the message started generating, for working with "continue" generations
     # Doesn't work for longer messages... Not sure how I should handle this
     full_msg = shared.tokenizer.decode([token_id for token_id in gen_token_ids[:-1]]).strip()
-    # Space at the beginning to account for tokenization spaces...
-    text = ' ' + html.unescape(text)
+    
     # There was an issue with tab lengths being off by one...
     # Seems like it might be model-dependent...
     #text = re.sub(r'( {3,})', r'\1 ', text)
@@ -137,6 +155,7 @@ def output_modifier(text):
     #i = 0
     # Add token index for ability to regenerate from there
     nonwhitespace_token_found = False
+    missing_token_count = 0
     for index, token, prob, ppl, top_tokens, top_probs in zip(range(len(gen_tokens)), gen_tokens, sel_probs, perplexities, top_tokens_list, top_probs_list):
         # Somehow this works without issues, but not sure how...
         if not nonwhitespace_token_found and token.strip() == '':
@@ -153,14 +172,20 @@ def output_modifier(text):
             color = probability_color_scale(prob)
         if token.strip() in text[i:]:
             if params['probability_dropdown']:
-                text = text[:i] + text[i:].replace(token.replace('\n', ''), add_dropdown_html(token, index, color, top_tokens, top_probs[0], ppl), 1)
+                text = text[:i] + text[i:].replace(token.replace('\n', ''), add_dropdown_html(token, index, i, color, top_tokens, top_probs[0], ppl), 1)
             else:
                 text = text[:i] + text[i:].replace(token.replace('\n', ''), add_color_html(token, color), 1)
             
             # This might be slightly inefficient
             i += text[i:].find(end_part) + len(end_part)
         else:
+            missing_token_count += 1
             print('Missing token:', token, '...', text[i:i+20])
+            # If there are any missing tokens, then either the tokenization was off, or this is the start of a conversation, or something else went wrong
+        if missing_token_count > 5:
+            print("Canceling token coloring...")
+            return original_text
+
 
     # Use full perplexity list for calculating the average here.
     # Fix issue with mean of empty slice
@@ -236,11 +261,11 @@ def add_color_html(token, color):
 # I think the issue is from HTML elements taking up space in the visible history, and things like history deepcopy add latency proportional to the size of the history.
 # Potential solution is maybe to modify the main generation code to send just the internal text and not the visible history, to avoid moving too much around.
 # I wonder if we can also avoid using deepcopy here.
-def add_dropdown_html(token, index, color, top_tokens, top_probs, perplexity=0):
+def add_dropdown_html(token, index, msg_position, color, top_tokens, top_probs, perplexity=0):
     #print("Token:", token, token.isspace(), '\n' in token or '\r' in token)
     output = ''
     # Use the repr to get characters like \n visible. Exclude the quotes around it
-    output += f'
            {html.escape(repr(token)[1:-1])}
            '
+    output += f'
            {html.escape(repr(token)[1:-1])}
            '
     for i, token_option, prob in zip(range(len(top_tokens)), top_tokens, top_probs):
         # TODO: Bold for selected token?
         # Using divs prevented the problem of divs inside spans causing issues.
@@ -249,7 +274,7 @@ def add_dropdown_html(token, index, color, top_tokens, top_probs, perplexity=0):
         row_color = probability_color_scale(prob)
         row_class = ' class="selected"' if token_option == token else ''
         # This time we want to include the quotes around it so that we can see where the spaces are.
-        output += f''
+        output += f''
     if perplexity != 0:
         ppl_color = perplexity_color_scale(perplexity)
         output += f''
@@ -324,11 +349,12 @@ def custom_js():
 // Note that this will only work as intended on the last agent message
 document.addEventListener("click", async function(event) {
     //console.log(event.target);
-    const id = event.target.id;
-    if (id.includes("opt_")) {
-        const id_parts = id.split("_");
-        const token_index = id_parts[1];
-        const option_index = id_parts[2];
+    const name = event.target.getAttribute("name");
+    if (name != null && name.includes("opt_")) {
+        const name_parts = name.split("_");
+        const token_index = name_parts[1];
+        const option_index = name_parts[2];
+        const msg_pos = name_parts[3];
         // Exclude the quotes and convert newlines... Not sure about the newlines though
         // TODO: Seems like continuing generation from a newline causes problems whether you add it or not!
         const token_string = event.target.innerHTML.substring(1, event.target.innerHTML.length-1).replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"r", "g"), '').replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"n", "g"), '');
@@ -341,17 +367,21 @@ def custom_js():
             var msg_part = msg_parts[i];
             if (msg_part.nodeType === Node.ELEMENT_NODE) {
                 if (msg_part.nodeName == "DIV") {
-                    var current_token_index = msg_part.id.split("_")[1];
-                    if (current_token_index == token_index) {
-                        // Use the replacement token
-                        // TODO: Don't have access to the tokenizer here, and sometimes there needs to be a space added before this token
-                        msg_text += token_string //.replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"r", "g"), '').replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"n", "g"), '');
-                        break;
-                    }
-                    else {
-                        // Replace here or at the end?
-                        var text = msg_part.firstChild.innerHTML.replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"r", "g"), '').replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"n", "g"), '')
-                        msg_text += text;
+                    msg_part_name = msg_part.getAttribute("name")
+                    if (msg_part_name != null) {
+                        var current_token_index = msg_part_name.split("_")[1];
+                        var current_message_pos = msg_part_name.split("_")[2];
+                        if (current_token_index == token_index && current_message_pos == msg_pos) {
+                            // Use the replacement token
+                            // TODO: Don't have access to the tokenizer here, and sometimes there needs to be a space added before this token
+                            msg_text += token_string //.replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"r", "g"), '').replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"n", "g"), '');
+                            break;
+                        }
+                        else {
+                            // Replace here or at the end?
+                            var text = msg_part.firstChild.innerHTML.replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"r", "g"), '').replace(new RegExp(String.fromCharCode(92)+String.fromCharCode(92)+"n", "g"), '')
+                            msg_text += text;
+                        }
                     }
                 }
                 else {

From 758c3f15a5a9c24a69ee3104fe06d5f5a22b3cd3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 14 Mar 2025 20:04:43 -0700
Subject: [PATCH 0247/1701] Lint

---
 modules/llama_cpp_python_hijack.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/llama_cpp_python_hijack.py b/modules/llama_cpp_python_hijack.py
index c03c28a772..8572cd8115 100644
--- a/modules/llama_cpp_python_hijack.py
+++ b/modules/llama_cpp_python_hijack.py
@@ -147,8 +147,8 @@ def patched_init(self, *args, **kwargs):
                 )
 
                 # Create environment with loopcontrols extension
-                from jinja2.ext import loopcontrols
                 import jinja2
+                from jinja2.ext import loopcontrols
 
                 self._environment = jinja2.sandbox.ImmutableSandboxedEnvironment(
                     loader=jinja2.BaseLoader(),

From 01e42a00ffcb82747316a6ba40429919e8efd36d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 26 Mar 2025 06:01:57 -0700
Subject: [PATCH 0248/1701] Bump transformers to 4.50

---
 requirements.txt                 | 2 +-
 requirements_amd.txt             | 2 +-
 requirements_amd_noavx2.txt      | 2 +-
 requirements_apple_intel.txt     | 2 +-
 requirements_apple_silicon.txt   | 2 +-
 requirements_cpu_only.txt        | 2 +-
 requirements_cpu_only_noavx2.txt | 2 +-
 requirements_noavx2.txt          | 2 +-
 requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 83bd3a5334..63b539cf43 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.49.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 1e757ffe54..71679c7e70 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.49.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index f74ebf6954..cfe9bf00bd 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.49.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index dcdeae3faf..4e3c0c116a 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.49.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index b823e40e0d..61c66e821c 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.49.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index fe3f522a2c..0ab64e791b 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.49.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 014e2e5d93..17187b47c9 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.49.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 6139c46e51..63e051ca68 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.49.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index 858ffff59a..533d399b68 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.49.*
+transformers==4.50.*
 tqdm
 wandb
 

From 2bfaf44df0904d8db3cd04e38b33b6be582babb6 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Wed, 26 Mar 2025 10:03:21 -0300
Subject: [PATCH 0249/1701] Update accelerate requirement from ==1.4.* to
 ==1.5.* (#6802)

---
 requirements.txt                 | 2 +-
 requirements_amd.txt             | 2 +-
 requirements_amd_noavx2.txt      | 2 +-
 requirements_apple_intel.txt     | 2 +-
 requirements_apple_silicon.txt   | 2 +-
 requirements_cpu_only.txt        | 2 +-
 requirements_cpu_only_noavx2.txt | 2 +-
 requirements_noavx2.txt          | 2 +-
 requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 63b539cf43..8c76ff1754 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,4 +1,4 @@
-accelerate==1.4.*
+accelerate==1.5.*
 bitsandbytes==0.45.*
 colorama
 datasets
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 71679c7e70..7fcd5c0c00 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -1,4 +1,4 @@
-accelerate==1.4.*
+accelerate==1.5.*
 colorama
 datasets
 einops
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index cfe9bf00bd..028d37ccf9 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.4.*
+accelerate==1.5.*
 colorama
 datasets
 einops
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 4e3c0c116a..284c2dd618 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -1,4 +1,4 @@
-accelerate==1.4.*
+accelerate==1.5.*
 colorama
 datasets
 einops
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 61c66e821c..43615aeb59 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -1,4 +1,4 @@
-accelerate==1.4.*
+accelerate==1.5.*
 colorama
 datasets
 einops
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index 0ab64e791b..9ae8e2a777 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -1,4 +1,4 @@
-accelerate==1.4.*
+accelerate==1.5.*
 colorama
 datasets
 einops
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 17187b47c9..77fe85fe6c 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.4.*
+accelerate==1.5.*
 colorama
 datasets
 einops
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 63e051ca68..4bebd865c9 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.4.*
+accelerate==1.5.*
 bitsandbytes==0.45.*
 colorama
 datasets
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index 533d399b68..a7255c0afa 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -1,4 +1,4 @@
-accelerate==1.4.*
+accelerate==1.5.*
 colorama
 datasets
 einops

From 525b1e020776c31e9dc3fd592855983cecfafe07 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 29 Mar 2025 13:43:16 -0700
Subject: [PATCH 0250/1701] Remove the stalebot

---
 .github/workflows/stale.yml | 22 ----------------------
 1 file changed, 22 deletions(-)
 delete mode 100644 .github/workflows/stale.yml

diff --git a/.github/workflows/stale.yml b/.github/workflows/stale.yml
deleted file mode 100644
index 8eb03299eb..0000000000
--- a/.github/workflows/stale.yml
+++ /dev/null
@@ -1,22 +0,0 @@
-name: Close inactive issues
-on:
-  schedule:
-    - cron: "10 23 * * *"
-
-jobs:
-  close-issues:
-    runs-on: ubuntu-latest
-    permissions:
-      issues: write
-      pull-requests: write
-    steps:
-      - uses: actions/stale@v5
-        with:
-          stale-issue-message: ""
-          close-issue-message: "This issue has been closed due to inactivity for 6 months. If you believe it is still relevant, please leave a comment below. You can tag a developer in your comment."
-          days-before-issue-stale: 180
-          days-before-issue-close: 0
-          stale-issue-label: "stale"
-          days-before-pr-stale: -1
-          days-before-pr-close: -1
-          repo-token: ${{ secrets.GITHUB_TOKEN }}

From 9b80d1d6c2b5331864d314b4d43aa23471c53a7f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 29 Mar 2025 13:43:16 -0700
Subject: [PATCH 0251/1701] Remove the stalebot

---
 .github/workflows/stale.yml | 22 ----------------------
 1 file changed, 22 deletions(-)
 delete mode 100644 .github/workflows/stale.yml

diff --git a/.github/workflows/stale.yml b/.github/workflows/stale.yml
deleted file mode 100644
index 8eb03299eb..0000000000
--- a/.github/workflows/stale.yml
+++ /dev/null
@@ -1,22 +0,0 @@
-name: Close inactive issues
-on:
-  schedule:
-    - cron: "10 23 * * *"
-
-jobs:
-  close-issues:
-    runs-on: ubuntu-latest
-    permissions:
-      issues: write
-      pull-requests: write
-    steps:
-      - uses: actions/stale@v5
-        with:
-          stale-issue-message: ""
-          close-issue-message: "This issue has been closed due to inactivity for 6 months. If you believe it is still relevant, please leave a comment below. You can tag a developer in your comment."
-          days-before-issue-stale: 180
-          days-before-issue-close: 0
-          stale-issue-label: "stale"
-          days-before-pr-stale: -1
-          days-before-pr-close: -1
-          repo-token: ${{ secrets.GITHUB_TOKEN }}

From 1bd208c219a16ec1d333f07e8a2bb2b6dd55d22d Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Sat, 29 Mar 2025 22:47:10 -0300
Subject: [PATCH 0252/1701] Add a new chat style: Dark (#6817)

---
 css/chat_style-Dark.css | 128 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 128 insertions(+)
 create mode 100644 css/chat_style-Dark.css

diff --git a/css/chat_style-Dark.css b/css/chat_style-Dark.css
new file mode 100644
index 0000000000..7f7f0dbfc3
--- /dev/null
+++ b/css/chat_style-Dark.css
@@ -0,0 +1,128 @@
+.message {
+    display: grid;
+    grid-template-columns: 60px minmax(0, 1fr);
+    padding-bottom: 28px;
+    font-size: 18px;
+    font-family: 'Roboto', Arial, sans-serif; /* Modern font */
+    line-height: 1.5;
+}
+
+.circle-you,
+.circle-bot {
+    background-color: #2b2b2b; /* Darker background for circles */
+    border-radius: 50%; /* Perfect circle */
+    border: 1px solid #4a90e2; /* Soft blue border */
+    box-shadow: 0 4px 8px rgba(0, 0, 0, 0.5); /* Soft shadow for depth */
+}
+
+.circle-bot img,
+.circle-you img {
+    border-radius: 50%; /* Make images circular */
+    width: 100%;
+    height: 100%;
+    object-fit: cover;
+}
+
+.circle-you, .circle-bot {
+    width: 64px; /* Smaller size for modern look */
+    height: 64px;
+}
+
+.text {
+    padding-left: 12px; /* Reduced padding for a cleaner layout */
+    color: #f0f0f0; /* Light text color for readability */
+}
+
+.text p {
+    margin-top: 2px;
+}
+
+.username {
+    padding-left: 10px;
+    font-size: 20px;
+    font-weight: bold;
+    color: #e0e0e0; /* Light gray text */
+    transition: color 0.3s ease; /* Smooth color transition */
+}
+
+.username:hover {
+    color: #4a90e2; /* Blue color on hover */
+}
+
+.message-body {
+    position: relative;
+    border: 1px solid rgba(255, 255, 255, 0.1); /* Soft white border */
+    border-radius: 8px; /* Slightly rounded corners */
+    padding: 15px;
+    background: #1e1e1e; /* Dark background */
+    box-shadow: 0 4px 10px rgba(0, 0, 0, 0.3); /* Subtle shadow for depth */
+    transition: background 0.3s ease; /* Smooth transition for background */
+}
+
+.message-body:hover {
+    background: #252525; /* Slightly lighter on hover */
+}
+
+/* Adds 2 extra lines at the top and bottom of the message */
+.message-body::before,
+.message-body::after {
+    content: "";
+    position: absolute;
+    left: 10px;
+    right: 10px;
+    height: 1px;
+    background-color: rgba(255, 255, 255, 0.05); /* Faded lines for subtle separation */
+}
+
+.message-body::before {
+    top: 4px;
+}
+
+.message-body::after {
+    bottom: 4px;
+}
+
+.message-body img {
+    max-width: 300px;
+    max-height: 300px;
+    border-radius: 10px; /* Rounded corners for images */
+}
+
+.message-body p {
+    margin-bottom: 0 !important;
+    font-size: 16px !important;
+    line-height: 1.5 !important;
+    color: #e0e0e0 !important; /* Light color for text */
+}
+
+.message-body p em {
+    color: #a6a6a6 !important; /* Softer gray for emphasized text */
+}
+
+@media screen and (max-width: 688px) {
+    .message {
+        display: grid;
+        grid-template-columns: 60px minmax(0, 1fr);
+        padding-bottom: 25px;
+        font-size: 15px;
+        font-family: 'Roboto', Arial, sans-serif; /* Modern font */
+        line-height: 1.5;
+    }
+
+    .circle-you, .circle-bot {
+        width: 40px; /* Smaller size for mobile */
+        height: 40px;
+    }
+
+    .text {
+        padding-left: 10px; /* Reduced padding for mobile */
+    }
+
+    .message-body p {
+        font-size: 14px !important; /* Smaller text for mobile */
+    }
+
+    .username {
+        font-size: 18px; /* Smaller username for mobile */
+    }
+}

From 79a26d7a5cd24e952b670668e97d50f1369c3e49 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 29 Mar 2025 18:49:48 -0700
Subject: [PATCH 0253/1701] Lint

---
 css/chat_style-Dark.css | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/css/chat_style-Dark.css b/css/chat_style-Dark.css
index 7f7f0dbfc3..368a2a16b1 100644
--- a/css/chat_style-Dark.css
+++ b/css/chat_style-Dark.css
@@ -3,7 +3,7 @@
     grid-template-columns: 60px minmax(0, 1fr);
     padding-bottom: 28px;
     font-size: 18px;
-    font-family: 'Roboto', Arial, sans-serif; /* Modern font */
+    font-family: Roboto, Arial, sans-serif; /* Modern font */
     line-height: 1.5;
 }
 
@@ -12,7 +12,7 @@
     background-color: #2b2b2b; /* Darker background for circles */
     border-radius: 50%; /* Perfect circle */
     border: 1px solid #4a90e2; /* Soft blue border */
-    box-shadow: 0 4px 8px rgba(0, 0, 0, 0.5); /* Soft shadow for depth */
+    box-shadow: 0 4px 8px rgb(0 0 0 / 50%); /* Soft shadow for depth */
 }
 
 .circle-bot img,
@@ -51,11 +51,11 @@
 
 .message-body {
     position: relative;
-    border: 1px solid rgba(255, 255, 255, 0.1); /* Soft white border */
+    border: 1px solid rgb(255 255 255 / 10%); /* Soft white border */
     border-radius: 8px; /* Slightly rounded corners */
     padding: 15px;
     background: #1e1e1e; /* Dark background */
-    box-shadow: 0 4px 10px rgba(0, 0, 0, 0.3); /* Subtle shadow for depth */
+    box-shadow: 0 4px 10px rgb(0 0 0 / 30%); /* Subtle shadow for depth */
     transition: background 0.3s ease; /* Smooth transition for background */
 }
 
@@ -71,7 +71,7 @@
     left: 10px;
     right: 10px;
     height: 1px;
-    background-color: rgba(255, 255, 255, 0.05); /* Faded lines for subtle separation */
+    background-color: rgb(255 255 255 / 5%); /* Faded lines for subtle separation */
 }
 
 .message-body::before {
@@ -99,13 +99,13 @@
     color: #a6a6a6 !important; /* Softer gray for emphasized text */
 }
 
-@media screen and (max-width: 688px) {
+@media screen and (width <= 688px) {
     .message {
         display: grid;
         grid-template-columns: 60px minmax(0, 1fr);
         padding-bottom: 25px;
         font-size: 15px;
-        font-family: 'Roboto', Arial, sans-serif; /* Modern font */
+        font-family: Roboto, Arial, sans-serif; /* Modern font */
         line-height: 1.5;
     }
 

From 1981327285c4411cc23472c6e874a895c8cf4424 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 29 Mar 2025 19:17:14 -0700
Subject: [PATCH 0254/1701] Fix the colab notebook

---
 Colab-TextGen-GPU.ipynb | 1 +
 1 file changed, 1 insertion(+)

diff --git a/Colab-TextGen-GPU.ipynb b/Colab-TextGen-GPU.ipynb
index 8e305e1dc3..ebeafc0296 100644
--- a/Colab-TextGen-GPU.ipynb
+++ b/Colab-TextGen-GPU.ipynb
@@ -57,6 +57,7 @@
         "from pathlib import Path\n",
         "\n",
         "os.environ.pop('PYTHONPATH', None)\n",
+        "os.environ.pop('MPLBACKEND', None)\n",
         "\n",
         "if Path.cwd().name != 'text-generation-webui':\n",
         "  print(\"\\033[1;32;1m\\n --> Installing the web UI. This will take a while, but after the initial setup, you can download and test as many models as you like.\\033[0;37;0m\\n\")\n",

From bb1905ebc5229ada4008ad74b23a4c61e965d424 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 29 Mar 2025 19:17:14 -0700
Subject: [PATCH 0255/1701] Fix the colab notebook

---
 Colab-TextGen-GPU.ipynb | 1 +
 1 file changed, 1 insertion(+)

diff --git a/Colab-TextGen-GPU.ipynb b/Colab-TextGen-GPU.ipynb
index 8e305e1dc3..ebeafc0296 100644
--- a/Colab-TextGen-GPU.ipynb
+++ b/Colab-TextGen-GPU.ipynb
@@ -57,6 +57,7 @@
         "from pathlib import Path\n",
         "\n",
         "os.environ.pop('PYTHONPATH', None)\n",
+        "os.environ.pop('MPLBACKEND', None)\n",
         "\n",
         "if Path.cwd().name != 'text-generation-webui':\n",
         "  print(\"\\033[1;32;1m\\n --> Installing the web UI. This will take a while, but after the initial setup, you can download and test as many models as you like.\\033[0;37;0m\\n\")\n",

From 109de34e3b3187eb3f463bf463086a48444013a0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 31 Mar 2025 09:23:28 -0700
Subject: [PATCH 0256/1701] Remove the old --model-menu flag

---
 modules/shared.py |  2 +-
 server.py         | 18 ------------------
 2 files changed, 1 insertion(+), 19 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 2e91f4d5a3..ea6c581a61 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -79,7 +79,6 @@
 group.add_argument('--lora', type=str, nargs='+', help='The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.')
 group.add_argument('--model-dir', type=str, default='models/', help='Path to directory with all the models.')
 group.add_argument('--lora-dir', type=str, default='loras/', help='Path to directory with all the loras.')
-group.add_argument('--model-menu', action='store_true', help='Show a model menu in the terminal when the web UI is first launched.')
 group.add_argument('--settings', type=str, help='Load the default interface settings from this yaml file. See settings-template.yaml for an example. If you create a file called settings.yaml, this file will be loaded by default without the need to use the --settings flag.')
 group.add_argument('--extensions', type=str, nargs='+', help='The list of extensions to load. If you want to load more than one extension, write the names separated by spaces.')
 group.add_argument('--verbose', action='store_true', help='Print the prompts to the terminal.')
@@ -215,6 +214,7 @@
 group.add_argument('--disable_exllamav2', action='store_true', help='DEPRECATED')
 group.add_argument('--wbits', type=int, default=0, help='DEPRECATED')
 group.add_argument('--groupsize', type=int, default=-1, help='DEPRECATED')
+group.add_argument('--model-menu', action='store_true', help='DEPRECATED')
 
 args = parser.parse_args()
 args_defaults = parser.parse_args([])
diff --git a/server.py b/server.py
index 31e1c4c60a..1f227350fd 100644
--- a/server.py
+++ b/server.py
@@ -218,28 +218,10 @@ def create_interface():
         if extension not in shared.args.extensions:
             shared.args.extensions.append(extension)
 
-    available_models = utils.get_available_models()
-
     # Model defined through --model
     if shared.args.model is not None:
         shared.model_name = shared.args.model
 
-    # Select the model from a command-line menu
-    elif shared.args.model_menu:
-        if len(available_models) == 0:
-            logger.error('No models are available! Please download at least one.')
-            sys.exit(0)
-        else:
-            print('The following models are available:\n')
-            for i, model in enumerate(available_models):
-                print(f'{i+1}. {model}')
-
-            print(f'\nWhich one do you want to load? 1-{len(available_models)}\n')
-            i = int(input()) - 1
-            print()
-
-        shared.model_name = available_models[i]
-
     # If any model has been selected, load it
     if shared.model_name != 'None':
         p = Path(shared.model_name)

From 77a73cc56122dbabf8fb446f1c1c278923fbbfe9 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Mon, 31 Mar 2025 21:01:27 -0300
Subject: [PATCH 0257/1701] Update peft requirement from ==0.12.* to ==0.15.*
 (#6820)

---
 requirements.txt                 | 2 +-
 requirements_amd.txt             | 2 +-
 requirements_amd_noavx2.txt      | 2 +-
 requirements_apple_intel.txt     | 2 +-
 requirements_apple_silicon.txt   | 2 +-
 requirements_cpu_only.txt        | 2 +-
 requirements_cpu_only_noavx2.txt | 2 +-
 requirements_noavx2.txt          | 2 +-
 requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 8c76ff1754..5ab40c4203 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -10,7 +10,7 @@ markdown
 numba==0.59.*
 numpy==1.26.*
 pandas
-peft==0.12.*
+peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 7fcd5c0c00..5f278db688 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -9,7 +9,7 @@ markdown
 numba==0.59.*
 numpy==1.26.*
 pandas
-peft==0.12.*
+peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 028d37ccf9..78d145243b 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -9,7 +9,7 @@ markdown
 numba==0.59.*
 numpy==1.26.*
 pandas
-peft==0.12.*
+peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 284c2dd618..7ab6dfffd5 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -9,7 +9,7 @@ markdown
 numba==0.59.*
 numpy==1.26.*
 pandas
-peft==0.12.*
+peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 43615aeb59..0c84a84c8f 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -9,7 +9,7 @@ markdown
 numba==0.59.*
 numpy==1.26.*
 pandas
-peft==0.12.*
+peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index 9ae8e2a777..b9eb99b1ca 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -9,7 +9,7 @@ markdown
 numba==0.59.*
 numpy==1.26.*
 pandas
-peft==0.12.*
+peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 77fe85fe6c..68ae5c17fe 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -9,7 +9,7 @@ markdown
 numba==0.59.*
 numpy==1.26.*
 pandas
-peft==0.12.*
+peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 4bebd865c9..1a2a670faf 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -10,7 +10,7 @@ markdown
 numba==0.59.*
 numpy==1.26.*
 pandas
-peft==0.12.*
+peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index a7255c0afa..3b61ca39f6 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -9,7 +9,7 @@ markdown
 numba==0.59.*
 numpy==1.26.*
 pandas
-peft==0.12.*
+peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2

From cbffcf67ef12938e6b26b7eddffa4327d83e71b0 Mon Sep 17 00:00:00 2001
From: Shixian Sheng 
Date: Wed, 2 Apr 2025 13:28:29 -0400
Subject: [PATCH 0258/1701] Fix links in the ngrok extension README (#6826)

---
 extensions/ngrok/README.md | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/extensions/ngrok/README.md b/extensions/ngrok/README.md
index 0324bf9852..2e9eb82d61 100644
--- a/extensions/ngrok/README.md
+++ b/extensions/ngrok/README.md
@@ -9,9 +9,9 @@ the `settings.json` file, see the Examples below. Retrieve your authtoken on the
 
 # Documentation
 
-For a list of all available options, see [the configuration documentation](https://ngrok.com/docs/ngrok-agent/config/) or [the connect example](https://github.com/ngrok/ngrok-py/blob/main/examples/ngrok-connect-full.py).
+For a list of all available options, see [the configuration documentation](https://ngrok.com/docs/ngrok-agent/config/) or [the forward example](https://github.com/ngrok/ngrok-python/blob/main/examples/ngrok-forward-full.py).
 
-The ngrok Python SDK is [on github here](https://github.com/ngrok/ngrok-py). A quickstart guide and a full API reference are included in the [ngrok-py Python API documentation](https://ngrok.github.io/ngrok-py/).
+The ngrok Python SDK is [on github here](https://github.com/ngrok/ngrok-py). A quickstart guide and a full API reference are included in the [ngrok-py Python API documentation](https://ngrok.github.io/ngrok-python/).
 
 # Running
 
@@ -66,4 +66,4 @@ To add an authtoken instead of using the NGROK_AUTHTOKEN environment variable:
         "authtoken_from_env":false
     }
 }
-```
\ No newline at end of file
+```

From c010cea7be0ba17623c16d8a2951d55c952d6ba0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 6 Apr 2025 17:17:25 -0700
Subject: [PATCH 0259/1701] Remove CUDA 11.8 support

---
 one_click.py | 51 ++++++++++++++-------------------------------------
 1 file changed, 14 insertions(+), 37 deletions(-)

diff --git a/one_click.py b/one_click.py
index effc7d4371..2a161e1113 100644
--- a/one_click.py
+++ b/one_click.py
@@ -106,9 +106,7 @@ def update_pytorch():
     torver = torch_version()
     base_cmd = f"python -m pip install --upgrade torch=={TORCH_VERSION} torchvision=={TORCHVISION_VERSION} torchaudio=={TORCHAUDIO_VERSION}"
 
-    if "+cu118" in torver:
-        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/cu118"
-    elif "+cu" in torver:
+    if "+cu" in torver:
         install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/cu121"
     elif "+rocm" in torver:
         install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.1"
@@ -236,24 +234,21 @@ def install_webui():
         choice = os.environ["GPU_CHOICE"].upper()
         print_big_message(f"Selected GPU choice \"{choice}\" based on the GPU_CHOICE environment variable.")
 
-        # Warn about changed meanings and handle old NVIDIA choice
+        # Warn about changed meanings and handle old choices
         if choice == "B":
-            print_big_message("Warning: GPU_CHOICE='B' now means 'NVIDIA (CUDA 11.8)' in the new version.")
+            print_big_message("Warning: GPU_CHOICE='B' now means 'AMD' in the new version.")
         elif choice == "C":
-            print_big_message("Warning: GPU_CHOICE='C' now means 'AMD' in the new version.")
+            print_big_message("Warning: GPU_CHOICE='C' now means 'Apple M Series' in the new version.")
         elif choice == "D":
-            print_big_message("Warning: GPU_CHOICE='D' now means 'Apple M Series' in the new version.")
-        elif choice == "A" and "USE_CUDA118" in os.environ:
-            choice = "B" if os.environ.get("USE_CUDA118", "").lower() in ("yes", "y", "true", "1", "t", "on") else "A"
+            print_big_message("Warning: GPU_CHOICE='D' now means 'Intel Arc' in the new version.")
     else:
         choice = get_user_choice(
             "What is your GPU?",
             {
-                'A': 'NVIDIA - CUDA 12.1 (recommended)',
-                'B': 'NVIDIA - CUDA 11.8 (legacy GPUs)',
-                'C': 'AMD - Linux/macOS only, requires ROCm 6.1',
-                'D': 'Apple M Series',
-                'E': 'Intel Arc (beta)',
+                'A': 'NVIDIA - CUDA 12.1',
+                'B': 'AMD - Linux/macOS only, requires ROCm 6.1',
+                'C': 'Apple M Series',
+                'D': 'Intel Arc (beta)',
                 'N': 'CPU mode'
             },
         )
@@ -261,15 +256,13 @@ def install_webui():
     # Convert choices to GPU names for compatibility
     gpu_choice_to_name = {
         "A": "NVIDIA",
-        "B": "NVIDIA",
-        "C": "AMD",
-        "D": "APPLE",
-        "E": "INTEL",
+        "B": "AMD",
+        "C": "APPLE",
+        "D": "INTEL",
         "N": "NONE"
     }
 
     selected_gpu = gpu_choice_to_name[choice]
-    use_cuda118 = (choice == "B")  # CUDA version is now determined by menu choice
 
     # Write a flag to CMD_FLAGS.txt for CPU mode
     if selected_gpu == "NONE":
@@ -280,10 +273,7 @@ def install_webui():
 
     # Handle CUDA version display
     elif any((is_windows(), is_linux())) and selected_gpu == "NVIDIA":
-        if use_cuda118:
-            print("CUDA: 11.8")
-        else:
-            print("CUDA: 12.1")
+        print("CUDA: 12.1")
 
     # No PyTorch for AMD on Windows (?)
     elif is_windows() and selected_gpu == "AMD":
@@ -294,10 +284,7 @@ def install_webui():
     install_pytorch = f"python -m pip install torch=={TORCH_VERSION} torchvision=={TORCHVISION_VERSION} torchaudio=={TORCHAUDIO_VERSION} "
 
     if selected_gpu == "NVIDIA":
-        if use_cuda118 == 'Y':
-            install_pytorch += "--index-url https://download.pytorch.org/whl/cu118"
-        else:
-            install_pytorch += "--index-url https://download.pytorch.org/whl/cu121"
+        install_pytorch += "--index-url https://download.pytorch.org/whl/cu121"
     elif selected_gpu == "AMD":
         install_pytorch += "--index-url https://download.pytorch.org/whl/rocm6.1"
     elif selected_gpu in ["APPLE", "NONE"]:
@@ -434,16 +421,6 @@ def update_requirements(initial_installation=False, pull=True):
     if not initial_installation and not wheels_changed:
         textgen_requirements = [line for line in textgen_requirements if '.whl' not in line]
 
-    if "+cu118" in torver:
-        textgen_requirements = [
-            req.replace('+cu121', '+cu118').replace('+cu122', '+cu118')
-            for req in textgen_requirements
-            if "autoawq" not in req.lower()
-        ]
-
-    if is_windows() and "+cu118" in torver:  # No flash-attention on Windows for CUDA 11
-        textgen_requirements = [req for req in textgen_requirements if 'oobabooga/flash-attention' not in req]
-
     with open('temp_requirements.txt', 'w') as file:
         file.write('\n'.join(textgen_requirements))
 

From a8a64b6c1cf0f609348fa0182308209acd11f563 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 6 Apr 2025 17:40:18 -0700
Subject: [PATCH 0260/1701] Update the README

---
 README.md | 13 -------------
 1 file changed, 13 deletions(-)

diff --git a/README.md b/README.md
index 40c242c87d..542e1ae101 100644
--- a/README.md
+++ b/README.md
@@ -143,19 +143,6 @@ Then browse to
 3) Manually install AutoGPTQ: [Installation](https://github.com/PanQiWei/AutoGPTQ#install-from-source).
     * Perform the from-source installation - there are no prebuilt ROCm packages for Windows.
 
-##### Older NVIDIA GPUs
-
-1) For Kepler GPUs and older, you will need to install CUDA 11.8 instead of 12:
-
-```
-pip3 install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1 --index-url https://download.pytorch.org/whl/cu118
-conda install -y -c "nvidia/label/cuda-11.8.0" cuda-runtime
-```
-
-2) bitsandbytes >= 0.39 may not work. In that case, to use `--load-in-8bit`, you may have to downgrade like this:
-    * Linux: `pip install bitsandbytes==0.38.1`
-    * Windows: `pip install https://github.com/jllllll/bitsandbytes-windows-webui/raw/main/bitsandbytes-0.38.1-py3-none-any.whl`
-
 ##### Manual install
 
 The `requirements*.txt` above contain various wheels precompiled through GitHub Actions. If you wish to compile things manually, or if you need to because no suitable wheels are available for your hardware, you can use `requirements_nowheels.txt` and then install your desired loaders manually.

From eef90a4964d00a94525d7c8ec9dd9ed90c193546 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 6 Apr 2025 17:44:07 -0700
Subject: [PATCH 0261/1701] Update some intel arc installation commands

---
 one_click.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/one_click.py b/one_click.py
index 2a161e1113..72626010ec 100644
--- a/one_click.py
+++ b/one_click.py
@@ -302,9 +302,9 @@ def install_webui():
     if selected_gpu == "INTEL":
         # Install oneAPI dependencies via conda
         print_big_message("Installing Intel oneAPI runtime libraries.")
-        run_cmd("conda install -y -c https://software.repos.intel.com/python/conda/ -c conda-forge dpcpp-cpp-rt=2024.0 mkl-dpcpp=2024.0")
+        run_cmd("conda install -y -c https://software.repos.intel.com/python/conda/ -c conda-forge dpcpp-cpp-rt=2024.0 mkl-dpcpp=2024.0", environment=True)
         # Install libuv required by Intel-patched torch
-        run_cmd("conda install -y libuv")
+        run_cmd("conda install -y libuv", environment=True)
 
     # Install the webui requirements
     update_requirements(initial_installation=True, pull=False)

From 204db283623a277d2831e0952814b7f0890ef1c6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 6 Apr 2025 18:48:31 -0700
Subject: [PATCH 0262/1701] Update the dockerfiles

---
 docker/amd/Dockerfile   | 2 +-
 docker/intel/Dockerfile | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/docker/amd/Dockerfile b/docker/amd/Dockerfile
index cfbcf7e451..66e5863c90 100644
--- a/docker/amd/Dockerfile
+++ b/docker/amd/Dockerfile
@@ -13,7 +13,7 @@ RUN --mount=type=cache,target=/var/cache/apt,sharing=locked,rw \
 WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
-RUN GPU_CHOICE=C LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
+RUN GPU_CHOICE=B LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
 COPY CMD_FLAGS.txt /home/app/text-generation-webui/
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
 WORKDIR /home/app/text-generation-webui
diff --git a/docker/intel/Dockerfile b/docker/intel/Dockerfile
index d2ed671edb..cab62442d7 100644
--- a/docker/intel/Dockerfile
+++ b/docker/intel/Dockerfile
@@ -13,7 +13,7 @@ RUN --mount=type=cache,target=/var/cache/apt,sharing=locked,rw \
 WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
-RUN GPU_CHOICE=E LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
+RUN GPU_CHOICE=D LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
 COPY CMD_FLAGS.txt /home/app/text-generation-webui/
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
 # set umask to ensure group read / write at runtime

From f1f32386b4338ef71cf2c23f93d6aa00b53b545c Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Mon, 7 Apr 2025 19:29:39 -0300
Subject: [PATCH 0263/1701] Update transformers requirement from ==4.50.* to
 ==4.51.* (#6834)

---
 requirements.txt                 | 2 +-
 requirements_amd.txt             | 2 +-
 requirements_amd_noavx2.txt      | 2 +-
 requirements_apple_intel.txt     | 2 +-
 requirements_apple_silicon.txt   | 2 +-
 requirements_cpu_only.txt        | 2 +-
 requirements_cpu_only_noavx2.txt | 2 +-
 requirements_noavx2.txt          | 2 +-
 requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 5ab40c4203..19f2124c83 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 5f278db688..99142de3d1 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 78d145243b..cdb6cff4bf 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 7ab6dfffd5..9dc36546ce 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 0c84a84c8f..e1fe8eaa87 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index b9eb99b1ca..ade65fbe27 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 68ae5c17fe..2bcfe71564 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 1a2a670faf..1db48c2277 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index 3b61ca39f6..bb9ea97c01 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 

From a5855c345cc3e361bc8a436daf995fe6a2a5dd33 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Mon, 7 Apr 2025 21:42:33 -0300
Subject: [PATCH 0264/1701] Set context lengths to at most 8192 by default (to
 prevent out of memory errors) (#6835)

---
 modules/models_settings.py |  7 +++++--
 modules/shared.py          |  6 +++---
 modules/ui_model_menu.py   | 10 +++++-----
 modules/ui_parameters.py   |  2 +-
 settings-template.yaml     |  2 +-
 5 files changed, 15 insertions(+), 12 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 8d65852345..b67d28a0c6 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -17,6 +17,7 @@ def get_fallback_settings():
         'compress_pos_emb': 1,
         'alpha_value': 1,
         'truncation_length': shared.settings['truncation_length'],
+        'truncation_length_info': shared.settings['truncation_length'],
         'skip_special_tokens': shared.settings['skip_special_tokens'],
         'custom_stopping_strings': shared.settings['custom_stopping_strings'],
     }
@@ -53,7 +54,8 @@ def get_model_metadata(model):
 
         for k in metadata:
             if k.endswith('context_length'):
-                model_settings['n_ctx'] = metadata[k]
+                model_settings['n_ctx'] = min(metadata[k], 8192)
+                model_settings['truncation_length_info'] = metadata[k]
             elif k.endswith('rope.freq_base'):
                 model_settings['rope_freq_base'] = metadata[k]
             elif k.endswith('rope.scale_linear'):
@@ -89,7 +91,8 @@ def get_model_metadata(model):
             for k in ['max_position_embeddings', 'model_max_length', 'max_seq_len']:
                 if k in metadata:
                     model_settings['truncation_length'] = metadata[k]
-                    model_settings['max_seq_len'] = metadata[k]
+                    model_settings['truncation_length_info'] = metadata[k]
+                    model_settings['max_seq_len'] = min(metadata[k], 8192)
 
             if 'rope_theta' in metadata:
                 model_settings['rope_freq_base'] = metadata['rope_theta']
diff --git a/modules/shared.py b/modules/shared.py
index ea6c581a61..77bd7639d1 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -53,7 +53,7 @@
     'skip_special_tokens': True,
     'stream': True,
     'static_cache': False,
-    'truncation_length': 2048,
+    'truncation_length': 8192,
     'seed': -1,
     'custom_stopping_strings': '',
     'custom_token_bans': '',
@@ -117,7 +117,7 @@
 group = parser.add_argument_group('llama.cpp')
 group.add_argument('--flash-attn', action='store_true', help='Use flash-attention.')
 group.add_argument('--tensorcores', action='store_true', help='NVIDIA only: use llama-cpp-python compiled without GGML_CUDA_FORCE_MMQ. This may improve performance on newer cards.')
-group.add_argument('--n_ctx', type=int, default=2048, help='Size of the prompt context.')
+group.add_argument('--n_ctx', type=int, default=8192, help='Size of the prompt context.')
 group.add_argument('--threads', type=int, default=0, help='Number of threads to use.')
 group.add_argument('--threads-batch', type=int, default=0, help='Number of threads to use for batches/prompt processing.')
 group.add_argument('--no_mul_mat_q', action='store_true', help='Disable the mulmat kernels.')
@@ -139,7 +139,7 @@
 group = parser.add_argument_group('ExLlamaV2')
 group.add_argument('--gpu-split', type=str, help='Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.')
 group.add_argument('--autosplit', action='store_true', help='Autosplit the model tensors across the available GPUs. This causes --gpu-split to be ignored.')
-group.add_argument('--max_seq_len', type=int, default=2048, help='Maximum sequence length.')
+group.add_argument('--max_seq_len', type=int, default=8192, help='Maximum sequence length.')
 group.add_argument('--cfg-cache', action='store_true', help='ExLlamav2_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.')
 group.add_argument('--no_flash_attn', action='store_true', help='Force flash-attention to not be used.')
 group.add_argument('--no_xformers', action='store_true', help='Force xformers to not be used.')
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 1264a9fd67..c23b991a16 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -200,8 +200,10 @@ def create_event_handlers():
 
 
 def load_model_wrapper(selected_model, loader, autoload=False):
+    settings = get_model_metadata(selected_model)
+
     if not autoload:
-        yield f"The settings for `{selected_model}` have been updated.\n\nClick on \"Load\" to load it."
+        yield "### {}\n\n- Settings updated: Click \"Load\" to load the model\n- Max sequence length: {}".format(selected_model, settings['truncation_length_info'])
         return
 
     if selected_model == 'None':
@@ -214,11 +216,9 @@ def load_model_wrapper(selected_model, loader, autoload=False):
                 shared.model, shared.tokenizer = load_model(selected_model, loader)
 
             if shared.model is not None:
-                output = f"Successfully loaded `{selected_model}`."
-
-                settings = get_model_metadata(selected_model)
+                output = f"Successfully loaded `{selected_model}`.\n\n"
                 if 'instruction_template' in settings:
-                    output += '\n\nIt seems to be an instruction-following model with template "{}". In the chat tab, instruct or chat-instruct modes should be used.'.format(settings['instruction_template'])
+                    output += '- It seems to be an instruction-following model with template "{}". In the chat tab, instruct or chat-instruct modes should be used.\n'.format(settings['instruction_template'])
 
                 yield output
             else:
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 846fcfe77d..c3245a9dc2 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -87,7 +87,7 @@ def create_ui(default_preset):
                             shared.gradio['static_cache'] = gr.Checkbox(value=shared.settings['static_cache'], label='Static KV cache', info='Use a static cache for improved performance.')
 
                         with gr.Column():
-                            shared.gradio['truncation_length'] = gr.Number(precision=0, step=256, value=get_truncation_length(), label='Truncate the prompt up to this length', info='The leftmost tokens are removed if the prompt exceeds this length. Most models require this to be at most 2048.')
+                            shared.gradio['truncation_length'] = gr.Number(precision=0, step=256, value=get_truncation_length(), label='Truncate the prompt up to this length', info='The leftmost tokens are removed if the prompt exceeds this length.')
                             shared.gradio['seed'] = gr.Number(value=shared.settings['seed'], label='Seed (-1 for random)')
 
                             shared.gradio['sampler_priority'] = gr.Textbox(value=generate_params['sampler_priority'], lines=12, label='Sampler priority', info='Parameter names separated by new lines or commas.', elem_classes=['add_scrollbar'])
diff --git a/settings-template.yaml b/settings-template.yaml
index 74935a609d..0343df0abd 100644
--- a/settings-template.yaml
+++ b/settings-template.yaml
@@ -25,7 +25,7 @@ add_bos_token: true
 skip_special_tokens: true
 stream: true
 static_cache: false
-truncation_length: 2048
+truncation_length: 8192
 seed: -1
 custom_stopping_strings: ''
 custom_token_bans: ''

From bf48ec8c449cbf58172c6f3ab83dd6a844b7994f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 7 Apr 2025 17:43:41 -0700
Subject: [PATCH 0265/1701] Remove an unnecessary UI message

---
 modules/ui_model_menu.py | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index c23b991a16..4fc1de0819 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -216,11 +216,7 @@ def load_model_wrapper(selected_model, loader, autoload=False):
                 shared.model, shared.tokenizer = load_model(selected_model, loader)
 
             if shared.model is not None:
-                output = f"Successfully loaded `{selected_model}`.\n\n"
-                if 'instruction_template' in settings:
-                    output += '- It seems to be an instruction-following model with template "{}". In the chat tab, instruct or chat-instruct modes should be used.\n'.format(settings['instruction_template'])
-
-                yield output
+                yield f"Successfully loaded `{selected_model}`."
             else:
                 yield f"Failed to load `{selected_model}`."
         except:

From 649ee729c126c3396b9f97c3cbfc8db8e2e6f7f0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 8 Apr 2025 09:22:06 -0700
Subject: [PATCH 0266/1701] Remove Python 3.10 support

---
 requirements.txt                 | 10 ----------
 requirements_amd.txt             |  4 ----
 requirements_amd_noavx2.txt      |  3 ---
 requirements_apple_intel.txt     |  2 --
 requirements_apple_silicon.txt   |  3 ---
 requirements_cpu_only.txt        |  2 --
 requirements_cpu_only_noavx2.txt |  2 --
 requirements_noavx2.txt          | 10 ----------
 8 files changed, 36 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 19f2124c83..e13cf98408 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -33,29 +33,19 @@ tiktoken
 
 # llama-cpp-python (CPU only, AVX2)
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, with GGML_CUDA_FORCE_MMQ)
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, without GGML_CUDA_FORCE_MMQ)
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # CUDA wheels
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu12torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 99142de3d1..adc77d32d6 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -32,13 +32,9 @@ tiktoken
 
 # llama-cpp-python (CPU only, AVX2)
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.8+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.8+rocm6.1.2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index cdb6cff4bf..22ee57b42e 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -32,11 +32,8 @@ tiktoken
 
 # llama-cpp-python (CPU only, no AVX2)
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # AMD wheels
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 9dc36546ce..553db45a92 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -32,7 +32,5 @@ tiktoken
 
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp310-cp310-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index e1fe8eaa87..e30ce8169a 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -32,9 +32,6 @@ tiktoken
 
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp310-cp310-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index ade65fbe27..e849a45149 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -32,6 +32,4 @@ tiktoken
 
 # llama-cpp-python (CPU only, AVX2)
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 2bcfe71564..e10782c9cc 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -32,6 +32,4 @@ tiktoken
 
 # llama-cpp-python (CPU only, no AVX2)
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 1db48c2277..ab7e59fc74 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -33,29 +33,19 @@ tiktoken
 
 # llama-cpp-python (CPU only, no AVX2)
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, with GGML_CUDA_FORCE_MMQ)
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, without GGML_CUDA_FORCE_MMQ)
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121avx-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # CUDA wheels
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu122torch2.4.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu12torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"

From 0b3503c91fcd3eaf6e0b93de4384794648406ba7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 8 Apr 2025 12:26:03 -0700
Subject: [PATCH 0267/1701] Revert "Update transformers requirement from
 ==4.50.* to ==4.51.* (#6834)"

This reverts commit f1f32386b4338ef71cf2c23f93d6aa00b53b545c.
---
 requirements.txt                 | 2 +-
 requirements_amd.txt             | 2 +-
 requirements_amd_noavx2.txt      | 2 +-
 requirements_apple_intel.txt     | 2 +-
 requirements_apple_silicon.txt   | 2 +-
 requirements_cpu_only.txt        | 2 +-
 requirements_cpu_only_noavx2.txt | 2 +-
 requirements_noavx2.txt          | 2 +-
 requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index e13cf98408..4cf99b6907 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_amd.txt b/requirements_amd.txt
index adc77d32d6..0d2057250c 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 22ee57b42e..93a46a64a2 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 553db45a92..00353bfd27 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index e30ce8169a..7076b386eb 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index e849a45149..c7e2687c94 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index e10782c9cc..2003c5448b 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index ab7e59fc74..d5f456f8b2 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index bb9ea97c01..3b61ca39f6 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 

From 8b8d39ec4e66affac03c22176ac368785095f584 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Wed, 9 Apr 2025 00:07:08 -0300
Subject: [PATCH 0268/1701] Add ExLlamaV3 support (#6832)

---
 README.md                      |  24 ++---
 modules/exllamav3_hf.py        | 179 +++++++++++++++++++++++++++++++++
 modules/loaders.py             |  56 ++++++++++-
 modules/models.py              |  17 +++-
 modules/models_settings.py     |   4 +-
 modules/shared.py              |   4 +-
 one_click.py                   |  50 +++++++--
 requirements.txt               |  18 ++--
 requirements_amd.txt           |   2 +-
 requirements_amd_noavx2.txt    |   2 +-
 requirements_apple_intel.txt   |   1 +
 requirements_apple_silicon.txt |   1 +
 requirements_noavx2.txt        |  18 ++--
 13 files changed, 322 insertions(+), 54 deletions(-)
 create mode 100644 modules/exllamav3_hf.py

diff --git a/README.md b/README.md
index 542e1ae101..63b8931a9f 100644
--- a/README.md
+++ b/README.md
@@ -12,7 +12,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 
 ## Features
 
-- Supports multiple text generation backends in one UI/API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp), and [ExLlamaV2](https://github.com/turboderp-org/exllamav2). [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) is supported via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile), and the Transformers loader is compatible with libraries like [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM), but they must be installed manually.
+- Supports multiple text generation backends in one UI/API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [ExLlamaV2](https://github.com/turboderp-org/exllamav2). [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) is supported via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile), and the Transformers loader is compatible with libraries like [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM), but they must be installed manually.
 - OpenAI-compatible API with Chat and Completions endpoints – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples).
 - Automatic prompt formatting using Jinja2 templates.
 - Three chat modes: `instruct`, `chat-instruct`, and `chat`, with automatic prompt templates in `chat-instruct`.
@@ -78,25 +78,19 @@ conda activate textgen
 
 | System | GPU | Command |
 |--------|---------|---------|
-| Linux/WSL | NVIDIA | `pip3 install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1 --index-url https://download.pytorch.org/whl/cu121` |
-| Linux/WSL | CPU only | `pip3 install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1 --index-url https://download.pytorch.org/whl/cpu` |
-| Linux | AMD | `pip3 install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1 --index-url https://download.pytorch.org/whl/rocm6.1` |
-| MacOS + MPS | Any | `pip3 install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1` |
-| Windows | NVIDIA | `pip3 install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1 --index-url https://download.pytorch.org/whl/cu121` |
-| Windows | CPU only | `pip3 install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1` |
+| Linux/WSL | NVIDIA | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/cu124` |
+| Linux/WSL | CPU only | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/cpu` |
+| Linux | AMD | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/rocm6.1` |
+| MacOS + MPS | Any | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0` |
+| Windows | NVIDIA | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/cu124` |
+| Windows | CPU only | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0` |
 
 The up-to-date commands can be found here: https://pytorch.org/get-started/locally/.
 
-For NVIDIA, you also need to install the CUDA runtime libraries:
+If you need `nvcc` to compile some library manually, you will additionally need to install this:
 
 ```
-conda install -y -c "nvidia/label/cuda-12.1.1" cuda-runtime
-```
-
-If you need `nvcc` to compile some library manually, replace the command above with
-
-```
-conda install -y -c "nvidia/label/cuda-12.1.1" cuda
+conda install -y -c "nvidia/label/cuda-12.4.1" cuda
 ```
 
 #### 3. Install the web UI
diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
new file mode 100644
index 0000000000..3bf44c9be1
--- /dev/null
+++ b/modules/exllamav3_hf.py
@@ -0,0 +1,179 @@
+import os
+import traceback
+from pathlib import Path
+from typing import Any, Dict, Optional, Union
+
+import torch
+from exllamav3 import Cache, Config, Model
+from torch.nn import CrossEntropyLoss
+from transformers import GenerationConfig, PretrainedConfig, PreTrainedModel
+from transformers.modeling_outputs import CausalLMOutputWithPast
+
+from modules import shared
+from modules.logging_colors import logger
+
+try:
+    import flash_attn
+except Exception:
+    logger.warning('Failed to load flash-attention due to the following error:\n')
+    traceback.print_exc()
+
+
+class Exllamav3HF(PreTrainedModel):
+    def __init__(self, model_dir):
+        super().__init__(PretrainedConfig())
+        self.generation_config = GenerationConfig()
+
+        config = Config.from_directory(model_dir)
+        self.ex_model = Model.from_config(config)
+
+        # Calculate the closest multiple of 256 at or above the chosen value
+        max_tokens = shared.args.max_seq_len
+        if max_tokens % 256 != 0:
+            adjusted_tokens = ((max_tokens // 256) + 1) * 256
+            logger.warning(f"max_num_tokens must be a multiple of 256. Adjusting from {max_tokens} to {adjusted_tokens}")
+            max_tokens = adjusted_tokens
+
+        self.ex_cache = Cache(self.ex_model, max_num_tokens=max_tokens)
+
+        # Create load parameters dictionary
+        load_params = {'progressbar': True}
+        if shared.args.gpu_split:
+            split = [float(alloc) for alloc in shared.args.gpu_split.split(",")]
+            load_params['use_per_device'] = split
+
+        self.ex_model.load(**load_params)
+        self.past_seq = None
+        self.max_tokens = max_tokens
+
+    def _validate_model_class(self):
+        pass
+
+    def _validate_model_kwargs(self, model_kwargs: Dict[str, Any]):
+        pass
+
+    def prepare_inputs_for_generation(self, input_ids, **kwargs):
+        return {'input_ids': input_ids, **kwargs}
+
+    @property
+    def device(self) -> torch.device:
+        return torch.device(0)
+
+    def __call__(self, *args, **kwargs):
+        use_cache = kwargs.get('use_cache', True)
+        labels = kwargs.get('labels', None)
+        past_key_values = kwargs.get('past_key_values', None)
+
+        if len(args) > 0:
+            if not shared.args.cfg_cache:
+                logger.error("Please enable the cfg-cache option to use CFG with ExLlamav3_HF.")
+                return
+
+            input_ids = args[0]
+            is_negative = True
+            past_seq = self.past_seq_negative
+            ex_cache = self.ex_cache_negative
+        else:
+            input_ids = kwargs['input_ids']
+            is_negative = False
+            past_seq = self.past_seq
+            ex_cache = self.ex_cache
+
+        seq = input_ids[0].tolist()
+        if is_negative and past_key_values is not None:
+            seq = past_key_values + seq
+
+        seq_tensor = torch.tensor(seq)
+        reset = True
+
+        # Make the forward call
+        if labels is None:
+            if past_seq is not None:
+                min_length = min(past_seq.shape[0], seq_tensor.shape[0])
+                indices = torch.nonzero(~torch.eq(past_seq[:min_length], seq_tensor[:min_length]))
+                if len(indices) > 0:
+                    longest_prefix = indices[0].item()
+                else:
+                    longest_prefix = min_length
+
+                if longest_prefix > 0:
+                    reset = False
+                    current_len = longest_prefix
+                    if len(seq_tensor) - longest_prefix > 1:
+                        self.ex_model.forward(
+                            input_ids=seq_tensor[longest_prefix:-1].view(1, -1),
+                            params={
+                                "attn_mode": "flash_attn",
+                                "cache": ex_cache,
+                                "past_len": longest_prefix,
+                                "batch_shape": (1, self.max_tokens)
+                            }
+                        )
+
+                        current_len = longest_prefix + len(seq_tensor) - longest_prefix - 1
+
+            if reset:
+                if len(seq_tensor) > 1:
+                    self.ex_model.forward(
+                        input_ids=seq_tensor[:-1].view(1, -1),
+                        params={
+                            "attn_mode": "flash_attn",
+                            "cache": ex_cache,
+                            "past_len": 0,
+                            "batch_shape": (1, self.max_tokens)
+                        }
+                    )
+
+                    current_len = len(seq_tensor) - 1
+                else:
+                    current_len = 0
+
+            logits = self.ex_model.forward(
+                input_ids=seq_tensor[-1:].view(1, -1),
+                params={
+                    "attn_mode": "flash_attn",
+                    "cache": ex_cache,
+                    "past_len": current_len,
+                    "batch_shape": (1, self.max_tokens)
+                }
+            ).to(input_ids.device).float()
+        else:
+            logits = self.ex_model.forward(
+                input_ids=seq_tensor.view(1, -1),
+                params={
+                    "attn_mode": "flash_attn",
+                    "cache": ex_cache,
+                    "past_len": 0,
+                    "batch_shape": (1, self.max_tokens)
+                }
+            ).float()
+
+        if is_negative:
+            self.past_seq_negative = seq_tensor
+        else:
+            self.past_seq = seq_tensor
+
+        loss = None
+        if labels is not None:
+            # Shift so that tokens < n predict n
+            shift_logits = logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            # Flatten the tokens
+            loss_fct = CrossEntropyLoss()
+            shift_logits = shift_logits.view(-1, logits.shape[-1])
+            shift_labels = shift_labels.view(-1)
+            # Enable model parallelism
+            shift_labels = shift_labels.to(shift_logits.device)
+            loss = loss_fct(shift_logits, shift_labels)
+
+        return CausalLMOutputWithPast(logits=logits, past_key_values=seq if use_cache else None, loss=loss)
+
+    @classmethod
+    def from_pretrained(cls, pretrained_model_name_or_path: Optional[Union[str, os.PathLike]], *model_args, **kwargs):
+        assert len(model_args) == 0 and len(kwargs) == 0, "extra args is currently not supported"
+        if isinstance(pretrained_model_name_or_path, str):
+            pretrained_model_name_or_path = Path(pretrained_model_name_or_path)
+
+        pretrained_model_name_or_path = Path(f'{shared.args.model_dir}') / Path(pretrained_model_name_or_path)
+
+        return Exllamav3HF(pretrained_model_name_or_path)
diff --git a/modules/loaders.py b/modules/loaders.py
index 88ded1d10b..980a13e617 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -23,7 +23,6 @@
         'use_double_quant',
         'use_eager_attention',
         'bf16',
-
         'trust_remote_code',
         'no_use_fast',
     ],
@@ -76,6 +75,13 @@
         'no_use_fast',
         'llamacpp_HF_info',
     ],
+    'ExLlamav3_HF': [
+        'max_seq_len',
+        'gpu_split',
+        'cfg_cache',
+        'trust_remote_code',
+        'no_use_fast',
+    ],
     'ExLlamav2_HF': [
         'max_seq_len',
         'cache_type',
@@ -174,30 +180,38 @@ def transformers_samplers():
 loaders_samplers = {
     'Transformers': transformers_samplers(),
     'HQQ': transformers_samplers(),
-    'ExLlamav2': {
+    'ExLlamav3_HF': {
         'temperature',
         'dynatemp_low',
         'dynatemp_high',
         'dynatemp_exponent',
         'smoothing_factor',
+        'smoothing_curve',
         'min_p',
         'top_p',
         'top_k',
         'typical_p',
         'xtc_threshold',
         'xtc_probability',
+        'epsilon_cutoff',
+        'eta_cutoff',
         'tfs',
         'top_a',
+        'top_n_sigma',
         'dry_multiplier',
         'dry_allowed_length',
         'dry_base',
         'repetition_penalty',
         'frequency_penalty',
         'presence_penalty',
+        'encoder_repetition_penalty',
+        'no_repeat_ngram_size',
         'repetition_penalty_range',
+        'guidance_scale',
         'mirostat_mode',
         'mirostat_tau',
         'mirostat_eta',
+        'do_sample',
         'dynamic_temperature',
         'temperature_last',
         'auto_max_new_tokens',
@@ -205,8 +219,12 @@ def transformers_samplers():
         'add_bos_token',
         'skip_special_tokens',
         'seed',
+        'sampler_priority',
         'custom_token_bans',
+        'negative_prompt',
         'dry_sequence_breakers',
+        'grammar_string',
+        'grammar_file_row',
     },
     'ExLlamav2_HF': {
         'temperature',
@@ -254,6 +272,40 @@ def transformers_samplers():
         'grammar_string',
         'grammar_file_row',
     },
+    'ExLlamav2': {
+        'temperature',
+        'dynatemp_low',
+        'dynatemp_high',
+        'dynatemp_exponent',
+        'smoothing_factor',
+        'min_p',
+        'top_p',
+        'top_k',
+        'typical_p',
+        'xtc_threshold',
+        'xtc_probability',
+        'tfs',
+        'top_a',
+        'dry_multiplier',
+        'dry_allowed_length',
+        'dry_base',
+        'repetition_penalty',
+        'frequency_penalty',
+        'presence_penalty',
+        'repetition_penalty_range',
+        'mirostat_mode',
+        'mirostat_tau',
+        'mirostat_eta',
+        'dynamic_temperature',
+        'temperature_last',
+        'auto_max_new_tokens',
+        'ban_eos_token',
+        'add_bos_token',
+        'skip_special_tokens',
+        'seed',
+        'custom_token_bans',
+        'dry_sequence_breakers',
+    },
     'llama.cpp': {
         'temperature',
         'min_p',
diff --git a/modules/models.py b/modules/models.py
index 3951fe822b..288bc1b66b 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -69,8 +69,9 @@ def load_model(model_name, loader=None):
         'Transformers': huggingface_loader,
         'llama.cpp': llamacpp_loader,
         'llamacpp_HF': llamacpp_HF_loader,
-        'ExLlamav2': ExLlamav2_loader,
+        'ExLlamav3_HF': ExLlamav3_HF_loader,
         'ExLlamav2_HF': ExLlamav2_HF_loader,
+        'ExLlamav2': ExLlamav2_loader,
         'HQQ': HQQ_loader,
         'TensorRT-LLM': TensorRT_LLM_loader,
     }
@@ -304,11 +305,10 @@ def llamacpp_HF_loader(model_name):
         return model
 
 
-def ExLlamav2_loader(model_name):
-    from modules.exllamav2 import Exllamav2Model
+def ExLlamav3_HF_loader(model_name):
+    from modules.exllamav3_hf import Exllamav3HF
 
-    model, tokenizer = Exllamav2Model.from_pretrained(model_name)
-    return model, tokenizer
+    return Exllamav3HF.from_pretrained(model_name)
 
 
 def ExLlamav2_HF_loader(model_name):
@@ -317,6 +317,13 @@ def ExLlamav2_HF_loader(model_name):
     return Exllamav2HF.from_pretrained(model_name)
 
 
+def ExLlamav2_loader(model_name):
+    from modules.exllamav2 import Exllamav2Model
+
+    model, tokenizer = Exllamav2Model.from_pretrained(model_name)
+    return model, tokenizer
+
+
 def HQQ_loader(model_name):
     try:
         from hqq.core.quantize import HQQBackend, HQQLinear
diff --git a/modules/models_settings.py b/modules/models_settings.py
index b67d28a0c6..51994e2375 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -158,14 +158,14 @@ def infer_loader(model_name, model_settings):
     path_to_model = Path(f'{shared.args.model_dir}/{model_name}')
     if not path_to_model.exists():
         loader = None
-    elif (path_to_model / 'quantize_config.json').exists():  # Old GPTQ metadata file
-        loader = 'ExLlamav2_HF'
     elif len(list(path_to_model.glob('*.gguf'))) > 0 and path_to_model.is_dir() and (path_to_model / 'tokenizer_config.json').exists():
         loader = 'llamacpp_HF'
     elif len(list(path_to_model.glob('*.gguf'))) > 0:
         loader = 'llama.cpp'
     elif re.match(r'.*\.gguf', model_name.lower()):
         loader = 'llama.cpp'
+    elif re.match(r'.*exl3', model_name.lower()):
+        loader = 'ExLlamav3_HF'
     elif re.match(r'.*exl2', model_name.lower()):
         loader = 'ExLlamav2_HF'
     elif re.match(r'.*-hqq', model_name.lower()):
diff --git a/modules/shared.py b/modules/shared.py
index 77bd7639d1..0981f6fb92 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -86,7 +86,7 @@
 
 # Model loader
 group = parser.add_argument_group('Model loader')
-group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav2_HF, ExLlamav2, HQQ, TensorRT-LLM.')
+group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav3_HF, ExLlamav2_HF, ExLlamav2, HQQ, TensorRT-LLM.')
 
 # Transformers/Accelerate
 group = parser.add_argument_group('Transformers/Accelerate')
@@ -273,6 +273,8 @@ def fix_loader_name(name):
         return 'ExLlamav2'
     elif name in ['exllamav2-hf', 'exllamav2_hf', 'exllama-v2-hf', 'exllama_v2_hf', 'exllama-v2_hf', 'exllama2-hf', 'exllama2_hf', 'exllama-2-hf', 'exllama_2_hf', 'exllama-2_hf']:
         return 'ExLlamav2_HF'
+    elif name in ['exllamav3-hf', 'exllamav3_hf', 'exllama-v3-hf', 'exllama_v3_hf', 'exllama-v3_hf', 'exllama3-hf', 'exllama3_hf', 'exllama-3-hf', 'exllama_3_hf', 'exllama-3_hf']:
+        return 'ExLlamav3_HF'
     elif name in ['hqq']:
         return 'HQQ'
     elif name in ['tensorrt', 'tensorrtllm', 'tensorrt_llm', 'tensorrt-llm', 'tensort', 'tensortllm']:
diff --git a/one_click.py b/one_click.py
index 72626010ec..fcca4ff524 100644
--- a/one_click.py
+++ b/one_click.py
@@ -16,10 +16,11 @@
 # os.environ["HCC_AMDGPU_TARGET"] = 'gfx1030'
 
 
-# Define the required PyTorch version
-TORCH_VERSION = "2.4.1"
-TORCHVISION_VERSION = "0.19.1"
-TORCHAUDIO_VERSION = "2.4.1"
+# Define the required versions
+TORCH_VERSION = "2.6.0"
+TORCHVISION_VERSION = "0.21.0"
+TORCHAUDIO_VERSION = "2.6.0"
+PYTHON_VERSION = "3.11"
 
 # Environment
 script_dir = os.getcwd()
@@ -101,13 +102,20 @@ def torch_version():
     return torver
 
 
-def update_pytorch():
+def update_pytorch_and_python():
     print_big_message("Checking for PyTorch updates.")
+
+    # Update the Python version. Left here for future reference in case this becomes necessary.
+    # print_big_message("Checking for PyTorch and Python updates.")
+    # current_python_version = f"{sys.version_info.major}.{sys.version_info.minor}"
+    # if current_python_version != PYTHON_VERSION:
+    #     run_cmd(f"conda install -y python={PYTHON_VERSION}", assert_success=True, environment=True)
+
     torver = torch_version()
     base_cmd = f"python -m pip install --upgrade torch=={TORCH_VERSION} torchvision=={TORCHVISION_VERSION} torchaudio=={TORCHAUDIO_VERSION}"
 
     if "+cu" in torver:
-        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/cu121"
+        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/cu124"
     elif "+rocm" in torver:
         install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.1"
     elif "+cpu" in torver:
@@ -245,7 +253,7 @@ def install_webui():
         choice = get_user_choice(
             "What is your GPU?",
             {
-                'A': 'NVIDIA - CUDA 12.1',
+                'A': 'NVIDIA - CUDA 12.4',
                 'B': 'AMD - Linux/macOS only, requires ROCm 6.1',
                 'C': 'Apple M Series',
                 'D': 'Intel Arc (beta)',
@@ -273,7 +281,7 @@ def install_webui():
 
     # Handle CUDA version display
     elif any((is_windows(), is_linux())) and selected_gpu == "NVIDIA":
-        print("CUDA: 12.1")
+        print("CUDA: 12.4")
 
     # No PyTorch for AMD on Windows (?)
     elif is_windows() and selected_gpu == "AMD":
@@ -284,7 +292,7 @@ def install_webui():
     install_pytorch = f"python -m pip install torch=={TORCH_VERSION} torchvision=={TORCHVISION_VERSION} torchaudio=={TORCHAUDIO_VERSION} "
 
     if selected_gpu == "NVIDIA":
-        install_pytorch += "--index-url https://download.pytorch.org/whl/cu121"
+        install_pytorch += "--index-url https://download.pytorch.org/whl/cu124"
     elif selected_gpu == "AMD":
         install_pytorch += "--index-url https://download.pytorch.org/whl/rocm6.1"
     elif selected_gpu in ["APPLE", "NONE"]:
@@ -297,7 +305,7 @@ def install_webui():
 
     # Install Git and then Pytorch
     print_big_message("Installing PyTorch.")
-    run_cmd(f"conda install -y -k ninja git && {install_pytorch} && python -m pip install py-cpuinfo==9.0.0", assert_success=True, environment=True)
+    run_cmd(f"conda install -y ninja git && {install_pytorch} && python -m pip install py-cpuinfo==9.0.0", assert_success=True, environment=True)
 
     if selected_gpu == "INTEL":
         # Install oneAPI dependencies via conda
@@ -323,6 +331,24 @@ def install_extensions_requirements():
         run_cmd(f"python -m pip install -r {extension_req_path} --upgrade", assert_success=False, environment=True)
 
 
+def clean_outdated_pytorch_cuda_dependencies():
+    patterns = ["cu121", "cu122", "torch2.4"]
+    result = run_cmd("python -m pip list --format=freeze", capture_output=True, environment=True)
+    matching_packages = []
+
+    for line in result.stdout.decode('utf-8').splitlines():
+        if "==" in line:
+            pkg_name, version = line.split('==', 1)
+            if any(pattern in version for pattern in patterns):
+                matching_packages.append(pkg_name)
+
+    if matching_packages:
+        print(f"Uninstalling: {', '.join(matching_packages)}")
+        run_cmd(f"python -m pip uninstall -y {' '.join(matching_packages)}", assert_success=True, environment=True)
+
+    return matching_packages
+
+
 def update_requirements(initial_installation=False, pull=True):
     # Create .git directory if missing
     if not os.path.exists(os.path.join(script_dir, ".git")):
@@ -410,7 +436,9 @@ def update_requirements(initial_installation=False, pull=True):
 
     # Update PyTorch
     if not initial_installation:
-        update_pytorch()
+        clean_outdated_pytorch_cuda_dependencies()
+        update_pytorch_and_python()
+        torver = torch_version()
 
     print_big_message(f"Installing webui requirements from file: {requirements_file}")
     print(f"TORCH: {torver}\n")
diff --git a/requirements.txt b/requirements.txt
index 4cf99b6907..b9b4ea7aff 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -36,16 +36,18 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 
 # llama-cpp-python (CUDA, with GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu124-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu124-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 
 # llama-cpp-python (CUDA, without GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu124-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu124-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 
 # CUDA wheels
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1/exllamav3-0.0.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1/exllamav3-0.0.1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/oobabooga/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 0d2057250c..3d24891f6a 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -36,5 +36,5 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp
 
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.8+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 93a46a64a2..057b631dfa 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -35,5 +35,5 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 
 # AMD wheels
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 00353bfd27..eba21ec2ce 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -33,4 +33,5 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1/exllamav3-0.0.1-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 7076b386eb..2048c99bde 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -34,4 +34,5 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1/exllamav3-0.0.1-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index d5f456f8b2..60b71ac1d9 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -36,16 +36,18 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 
 # llama-cpp-python (CUDA, with GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu124avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu124avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 
 # llama-cpp-python (CUDA, without GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu121avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu124avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu124avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 
 # CUDA wheels
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu121.torch2.4.1-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1/exllamav3-0.0.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1/exllamav3-0.0.1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/oobabooga/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu122torch2.4.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu12torch2.4cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

From d8aad6da948262e8679da2063d27600da0d8ccb4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 8 Apr 2025 20:20:24 -0700
Subject: [PATCH 0269/1701] Fix an update bug

---
 one_click.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/one_click.py b/one_click.py
index fcca4ff524..8e85dc3ac9 100644
--- a/one_click.py
+++ b/one_click.py
@@ -436,9 +436,9 @@ def update_requirements(initial_installation=False, pull=True):
 
     # Update PyTorch
     if not initial_installation:
-        clean_outdated_pytorch_cuda_dependencies()
         update_pytorch_and_python()
         torver = torch_version()
+        clean_outdated_pytorch_cuda_dependencies()
 
     print_big_message(f"Installing webui requirements from file: {requirements_file}")
     print(f"TORCH: {torver}\n")

From ad1ada657421cc830e8cd421e50477c3c05c3332 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Apr 2025 05:17:10 -0700
Subject: [PATCH 0270/1701] Change one message in the installer

---
 one_click.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/one_click.py b/one_click.py
index 8e85dc3ac9..f35366386a 100644
--- a/one_click.py
+++ b/one_click.py
@@ -343,7 +343,7 @@ def clean_outdated_pytorch_cuda_dependencies():
                 matching_packages.append(pkg_name)
 
     if matching_packages:
-        print(f"Uninstalling: {', '.join(matching_packages)}")
+        print(f"\nUninstalling: {', '.join(matching_packages)}\n")
         run_cmd(f"python -m pip uninstall -y {' '.join(matching_packages)}", assert_success=True, environment=True)
 
     return matching_packages

From 89f40cdcf7f0a5eb67a41865d1c67e2ae921dfa1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Apr 2025 07:20:51 -0700
Subject: [PATCH 0271/1701] Update libstdcxx-ng for GLIBCXX_3.4.30 support on
 Linux

---
 one_click.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/one_click.py b/one_click.py
index f35366386a..8dd11b6b61 100644
--- a/one_click.py
+++ b/one_click.py
@@ -434,6 +434,9 @@ def update_requirements(initial_installation=False, pull=True):
     if os.environ.get("INSTALL_EXTENSIONS", "").lower() in ("yes", "y", "true", "1", "t", "on"):
         install_extensions_requirements()
 
+    if is_linux():
+        run_cmd("conda install -y -c conda-forge libstdcxx-ng==12.1.0", assert_success=True, environment=True)
+
     # Update PyTorch
     if not initial_installation:
         update_pytorch_and_python()

From 8229736ec4b359b2a773f36abac1bf96d4eabf56 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Apr 2025 08:38:06 -0700
Subject: [PATCH 0272/1701] Reapply "Update transformers requirement from
 ==4.50.* to ==4.51.* (#6834)"

This reverts commit 0b3503c91fcd3eaf6e0b93de4384794648406ba7.
---
 requirements.txt                 | 2 +-
 requirements_amd.txt             | 2 +-
 requirements_amd_noavx2.txt      | 2 +-
 requirements_apple_intel.txt     | 2 +-
 requirements_apple_silicon.txt   | 2 +-
 requirements_cpu_only.txt        | 2 +-
 requirements_cpu_only_noavx2.txt | 2 +-
 requirements_noavx2.txt          | 2 +-
 requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index b9b4ea7aff..de338696e1 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 3d24891f6a..00281d2210 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 057b631dfa..80fcb71cbd 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index eba21ec2ce..4253a9400b 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 2048c99bde..6962b6fcc9 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index c7e2687c94..e849a45149 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 2003c5448b..e10782c9cc 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 60b71ac1d9..00c31c40b8 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index 3b61ca39f6..bb9ea97c01 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.51.*
 tqdm
 wandb
 

From d337ea31fa05d3d2f60df8c28fff10c07c10156f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Apr 2025 10:16:47 -0700
Subject: [PATCH 0273/1701] Revert "Reapply "Update transformers requirement
 from ==4.50.* to ==4.51.* (#6834)""

This reverts commit 8229736ec4b359b2a773f36abac1bf96d4eabf56.
---
 requirements.txt                 | 2 +-
 requirements_amd.txt             | 2 +-
 requirements_amd_noavx2.txt      | 2 +-
 requirements_apple_intel.txt     | 2 +-
 requirements_apple_silicon.txt   | 2 +-
 requirements_cpu_only.txt        | 2 +-
 requirements_cpu_only_noavx2.txt | 2 +-
 requirements_noavx2.txt          | 2 +-
 requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index de338696e1..b9b4ea7aff 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 00281d2210..3d24891f6a 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 80fcb71cbd..057b631dfa 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 4253a9400b..eba21ec2ce 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 6962b6fcc9..2048c99bde 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index e849a45149..c7e2687c94 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index e10782c9cc..2003c5448b 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 00c31c40b8..60b71ac1d9 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index bb9ea97c01..3b61ca39f6 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.51.*
+transformers==4.50.*
 tqdm
 wandb
 

From 9025848df56e7095febf88fc1cd473b180e19fc5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Apr 2025 10:25:47 -0700
Subject: [PATCH 0274/1701] Small change to installer

---
 one_click.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/one_click.py b/one_click.py
index 8dd11b6b61..9f46a2df83 100644
--- a/one_click.py
+++ b/one_click.py
@@ -21,6 +21,7 @@
 TORCHVISION_VERSION = "0.21.0"
 TORCHAUDIO_VERSION = "2.6.0"
 PYTHON_VERSION = "3.11"
+LIBSTDCXX_VERSION_LINUX = "12.1.0"
 
 # Environment
 script_dir = os.getcwd()
@@ -435,7 +436,7 @@ def update_requirements(initial_installation=False, pull=True):
         install_extensions_requirements()
 
     if is_linux():
-        run_cmd("conda install -y -c conda-forge libstdcxx-ng==12.1.0", assert_success=True, environment=True)
+        run_cmd(f"conda install -y -c conda-forge libstdcxx-ng=={LIBSTDCXX_VERSION_LINUX}", assert_success=True, environment=True)
 
     # Update PyTorch
     if not initial_installation:

From 6ea02062071544a157c3ef409b5930c8b5e8635e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Apr 2025 14:59:34 -0700
Subject: [PATCH 0275/1701] UI: remove the streaming cursor

---
 modules/chat.py | 11 ++---------
 1 file changed, 2 insertions(+), 9 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index fd94990750..7326c033b5 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -362,17 +362,13 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
     for j, reply in enumerate(generate_reply(prompt, state, stopping_strings=stopping_strings, is_chat=True, for_ui=for_ui)):
 
         # Extract the reply
+        visible_reply = reply
         if state['mode'] in ['chat', 'chat-instruct']:
-            visible_reply = re.sub("(||{{user}})", state['name1'], reply + '▍')
-        else:
-            visible_reply = reply + '▍'
+            visible_reply = re.sub("(||{{user}})", state['name1'], visible_reply)
 
         visible_reply = html.escape(visible_reply)
 
         if shared.stop_everything:
-            if output['visible'][-1][1].endswith('▍'):
-                output['visible'][-1][1] = output['visible'][-1][1][:-1]
-
             output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
             yield output
             return
@@ -388,9 +384,6 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             if is_stream:
                 yield output
 
-    if output['visible'][-1][1].endswith('▍'):
-        output['visible'][-1][1] = output['visible'][-1][1][:-1]
-
     output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
     yield output
 

From 297a406e0548557b4f85b0f1082677f6f5c984f7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Apr 2025 16:02:37 -0700
Subject: [PATCH 0276/1701] UI: smoother chat streaming

This removes the throttling associated to gr.Textbox that made words appears in chunks rather than one at a time
---
 modules/html_generator.py | 8 +++++---
 modules/ui_chat.py        | 6 +++---
 2 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 6bad0f892a..144f259316 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -391,8 +391,10 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
 
 def chat_html_wrapper(history, name1, name2, mode, style, character, reset_cache=False):
     if mode == 'instruct':
-        return generate_instruct_html(history)
+        result = generate_instruct_html(history)
     elif style == 'wpp':
-        return generate_chat_html(history, name1, name2)
+        result = generate_chat_html(history, name1, name2)
     else:
-        return generate_cai_chat_html(history, name1, name2, style, character, reset_cache)
+        result = generate_cai_chat_html(history, name1, name2, style, character, reset_cache)
+
+    return {'html': result}
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 395300d0be..a830abfb10 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -46,8 +46,8 @@ def create_ui():
 
         with gr.Row():
             with gr.Column(elem_id='chat-col'):
-                shared.gradio['html_display'] = gr.HTML(value=chat_html_wrapper({'internal': [], 'visible': []}, '', '', 'chat', 'cai-chat', ''), visible=True)
-                shared.gradio['display'] = gr.Textbox(value="", visible=False)  # Hidden buffer
+                shared.gradio['html_display'] = gr.HTML(value=chat_html_wrapper({'internal': [], 'visible': []}, '', '', 'chat', 'cai-chat', '')['html'], visible=True)
+                shared.gradio['display'] = gr.JSON(value={}, visible=False)  # Hidden buffer
                 with gr.Row(elem_id="chat-input-row"):
                     with gr.Column(scale=1, elem_id='gr-hover-container'):
                         gr.HTML(value='
            ☰
            ', elem_id='gr-hover')
@@ -181,7 +181,7 @@ def create_event_handlers():
     shared.reload_inputs = gradio(reload_arr)
 
     # Morph HTML updates instead of updating everything
-    shared.gradio['display'].change(None, gradio('display'), None, js="(text) => handleMorphdomUpdate(text)")
+    shared.gradio['display'].change(None, gradio('display'), None, js="(data) => handleMorphdomUpdate(data.html)")
 
     shared.gradio['Generate'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(

From 598568b1eda5bde68a5729dedbcb40d54b8bf01c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Apr 2025 16:03:14 -0700
Subject: [PATCH 0277/1701] Revert "UI: remove the streaming cursor"

This reverts commit 6ea02062071544a157c3ef409b5930c8b5e8635e.
---
 modules/chat.py | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 7326c033b5..fd94990750 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -362,13 +362,17 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
     for j, reply in enumerate(generate_reply(prompt, state, stopping_strings=stopping_strings, is_chat=True, for_ui=for_ui)):
 
         # Extract the reply
-        visible_reply = reply
         if state['mode'] in ['chat', 'chat-instruct']:
-            visible_reply = re.sub("(||{{user}})", state['name1'], visible_reply)
+            visible_reply = re.sub("(||{{user}})", state['name1'], reply + '▍')
+        else:
+            visible_reply = reply + '▍'
 
         visible_reply = html.escape(visible_reply)
 
         if shared.stop_everything:
+            if output['visible'][-1][1].endswith('▍'):
+                output['visible'][-1][1] = output['visible'][-1][1][:-1]
+
             output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
             yield output
             return
@@ -384,6 +388,9 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             if is_stream:
                 yield output
 
+    if output['visible'][-1][1].endswith('▍'):
+        output['visible'][-1][1] = output['visible'][-1][1][:-1]
+
     output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
     yield output
 

From 4ed0da74a8d9b7f4e363225452333004d197a550 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Apr 2025 20:09:48 -0700
Subject: [PATCH 0278/1701] Remove the obsolete 'multimodal' extension

---
 extensions/multimodal/DOCS.md                 |  85 ------
 extensions/multimodal/README.md               | 139 ----------
 extensions/multimodal/abstract_pipeline.py    |  63 -----
 extensions/multimodal/multimodal_embedder.py  | 178 ------------
 extensions/multimodal/pipeline_loader.py      |  52 ----
 .../multimodal/pipelines/llava/README.md      |   9 -
 .../multimodal/pipelines/llava/llava.py       | 262 ------------------
 .../multimodal/pipelines/llava/pipelines.py   |  48 ----
 .../place-additional-pipelines-here.txt       |   0
 extensions/multimodal/script.py               | 113 --------
 modules/shared.py                             |   9 +-
 11 files changed, 1 insertion(+), 957 deletions(-)
 delete mode 100644 extensions/multimodal/DOCS.md
 delete mode 100644 extensions/multimodal/README.md
 delete mode 100644 extensions/multimodal/abstract_pipeline.py
 delete mode 100644 extensions/multimodal/multimodal_embedder.py
 delete mode 100644 extensions/multimodal/pipeline_loader.py
 delete mode 100644 extensions/multimodal/pipelines/llava/README.md
 delete mode 100644 extensions/multimodal/pipelines/llava/llava.py
 delete mode 100644 extensions/multimodal/pipelines/llava/pipelines.py
 delete mode 100644 extensions/multimodal/pipelines/place-additional-pipelines-here.txt
 delete mode 100644 extensions/multimodal/script.py

diff --git a/extensions/multimodal/DOCS.md b/extensions/multimodal/DOCS.md
deleted file mode 100644
index eaa4365e9a..0000000000
--- a/extensions/multimodal/DOCS.md
+++ /dev/null
@@ -1,85 +0,0 @@
-# Technical description of multimodal extension
-
-## Working principle
-Multimodality extension does most of the stuff which is required for any image input:
-
-- adds the UI
-- saves the images as base64 JPEGs to history
-- provides the hooks to the UI
-- if there are images in the prompt, it:
-    - splits the prompt to text and image parts
-    - adds image start/end markers to text parts, then encodes and embeds the text parts
-    - calls the vision pipeline to embed the images
-    - stitches the embeddings together, and returns them to text generation
-- loads the appropriate vision pipeline, selected either from model name, or by specifying --multimodal-pipeline parameter
-
-Now, for the pipelines, they:
-
-- load the required vision models
-- return some consts, for example the number of tokens taken up by image
-- and most importantly: return the embeddings for LLM, given a list of images
-
-## Prompts/history
-
-To save images in prompt/history, this extension is using a base64 JPEG, wrapped in a HTML tag, like so:
-```
-
-```
-where `{img_str}` is the actual image data. This format makes displaying them in the UI for free. Do note, that this format is required to be exactly the same, the regex used to find the images is: ``.
-
-## LLM input
-To describe the input, let's see it on an example prompt:
-```
-text1text2text3
-```
-where `textN` is N-th text, `` is N-th image, in HTML format specified above.
-
-**The first step is to split the prompt into image/text parts**, so we get:
-```
-['text1', '', 'text2', '', 'text3']
-```
-this is done in `MultimodalEmbedder._split_prompt(...)` function, which returns a list of `PromptPart`s - dataclasses wrapping the separate parts.
-
-This function also appends the image start/end markers to text, which are provided by `AbstractMultimodalPipeline.image_start()` / `AbstractMultimodalPipeline.image_end()` functions. If image start is ``, and end is ``, this function will return:
-```
-['text1', '', 'text2', '', 'text3']
-```
-
-**The returned prompt parts are then turned into token embeddings.**
-
-First, they are modified to token IDs, for the text it is done using standard `modules.text_generation.encode()` function, and for the images the returned token IDs are changed to placeholders. The placeholder is a list of `N` times `placeholder token id`, where `N` is specified using `AbstractMultimodalPipeline.num_image_embeds()`, and placeholder token IDs using  `AbstractMultimodalPipeline.placeholder_token_id()`.
-
-Now, based on the token IDs, the prompt might get truncated, especially if `max_new_tokens` are unreasonably high. Unfortunately, it can't be done simply, just by trimming the prompt to be short enough. This way will lead to sometimes splitting the prompt in the middle of an image embedding, which usually breaks the generation. Therefore, in this case, the entire image needs to be removed from input. This is done inside `MultimodalEmbedder._encode_text(...)` function.
-
-**After the tokenization, the tokens need to get embedded**, the text and images are once again treated separately.
-
-The text parts are turned to embeddings, using `AbstractMultimodalPipeline.embed_tokens(...)` function. It uses standard embedding function from the model, but to support many LLMs, the actual function is returned by the pipeline (as it might be different for different LLMs), for LLaMA it is `shared.model.model.embed_tokens(...)`.
-
-The image parts are turned to embeddings, using `AbstractMultimodalPipeline.embed_images(...)` function. This function is specific for a given pipeline, it takes the images as input, forwards them through vision model/projector, and returns the embeddings.
-
-**Now, the returned embeddings are stitched together**, using `torch.cat()`, this is creating the final input to the LLM.
-
-## Pipelines
-
-All of the pipelines should subclass `AbstractMultimodalPipeline` class. The idea is to allow for new pipelines to be added in the same way as user extensions - git clone into `extensions/multimodal/pipelines`.
-
-The pipelines are the description of the vision part, containing vision model/multimodal projector. All of the pipelines should have an unique `name()`, which is then selected by user, in `--multimodal-pipeline` CLI argument. For an example, see `pipelines/llava/llava.py`.
-
-## Pipeline modules
-
-Pipelines are organized into "pipeline modules" - subdirectories in `pipelines` directory. The pipeline modules should contain a file called `pipelines.py`, that should contain the following fields:
-- `available_pipelines: List[str]` - list of pipelines provided by this module, shown as the list of available pipelines to the user
-- `def get_pipeline(name: str, params: dict) -> Optional[AbstractMultimodalPipeline]`: - a function to get a concrete pipeline by `name`, if `name` doesn't match any, should return `None`. `params` is the user settings for multimodal extension
-- `def get_pipeline_from_model_name(model_name: str, params: dict) -> Optional[AbstractMultimodalPipeline]`: - a function to get a pipeline from `model_name`, should be eager to return `None`, unless the determination can be done clearly (for example: minigpt-4 bases on vicuna - it should never return the pipeline, but llava can, as it has its own specific LLM finetune)
-
-**NOTE**: A pipeline module should lazy-import the pipelines only when necessary, and it should keep its imports to minimum
-
-## Pipeline params
-
-The pipelines will get the extension `params` in the constructor. They should honor the following fields:
-- `vision_device` - string, specifying `torch.device` to run the vision model (CLIP/ViT) on
-- `vision_bits` - int, number of fp bits to load the vision model(s) in
-- `projector_device` - string, specifying `torch.device` to run the projector models (Linear layers, QFormer, etc.) on
-- `projector_bits` - int, number of fp bits to load the projector models in
-
-As a helper, `AbstractMultimodalPipeline` has `_get_device(self, setting_name: str, params: dict)` and `_get_dtype(self, setting_name: str, params: dict)` helper functions, which parse string/int and return `torch.device` / `torch.dtype`.
diff --git a/extensions/multimodal/README.md b/extensions/multimodal/README.md
deleted file mode 100644
index b176eca3d6..0000000000
--- a/extensions/multimodal/README.md
+++ /dev/null
@@ -1,139 +0,0 @@
-# Multimodal
-
-## Description
-
-Adds support for multimodality (text+images) to text-generation-webui.
-
-Note: multimodal currently only works for transformers, AutoGPTQ, and GPTQ-for-LLaMa loaders. ExLlama (v1 and v2) and llama.cpp support are planned.
-
-https://user-images.githubusercontent.com/3718215/233817203-69b57e77-0c55-4fd6-b742-3204bb13b8fc.mp4
-
-## Usage
-
-To run this extension, download a LLM that supports multimodality, and then start server.py with the appropriate `--multimodal-pipeline` argument. Examples:
-
-```
-# LLaVA 1.5 13B has the best performance
-python server.py --model liuhaotian_llava-v1.5-13b --multimodal-pipeline llava-v1.5-13b --load-in-4bit
-# LLaVA 1.5 7B is relatively weaker, but requires less memory
-python server.py --model liuhaotian_llava-v1.5-7b --multimodal-pipeline llava-v1.5-7b --load-in-4bit
-python server.py --model TheBloke_llava-v1.5-13B-GPTQ_gptq-4bit-32g-actorder_True --multimodal-pipeline llava-v1.5-13b --disable_exllama --loader autogptq
-python server.py --model wojtab_llava-7b-v0-4bit-128g --multimodal-pipeline llava-7b
-python server.py --model wojtab_llava-13b-v0-4bit-128g --multimodal-pipeline llava-13b
-python server.py --model anon8231489123_vicuna-13b-GPTQ-4bit-128g --multimodal-pipeline minigpt4-13b
-python server.py --model llama-7b-4bit --multimodal-pipeline minigpt4-7b
-```
-
-There is built-in support for LLaVA-v0-13B, LLaVA-v0-7b, and LLaVA-v1.5-13B. To install `minigpt4`:
-
-- clone https://github.com/Wojtab/minigpt-4-pipeline into `extensions/multimodal/pipelines`
-- install the requirements.txt
-
-The same procedure should be used to install other pipelines, which can then be used with `--multimodal-pipeline [pipeline name]`. For additional multimodal pipelines refer to the compatibility section below.
-
-Do note, that each image takes up a considerable amount of tokens, so adjust `max_new_tokens` to be at most 1700 (recommended value is between 200 to 500), so the images don't get truncated.
-
-To send an image, just upload it to the extension field below chat, and send a prompt as always. The image will be added to the end of your message. If you wish to modify the placement, include a string `` in your prompt.
-
-Additionally, there is *Embed all images, not only the last one* checkbox. It modifies the image embeddings, by default (if it's unchecked), all but the most recent images have their embeddings empty, so they are not fed to the network. It seems as if some multimodal networks consider the features in all images at the same time as if they were a single image. Due to this behavior, by default, the extension skips previous images. However, it can lead to sub-par generation on other pipelines. If you want to include all images, just tick this checkbox.
-
-## Compatibility
-
-As of now, the following multimodal pipelines are supported:
-|Pipeline|`--multimodal-pipeline`|Default LLM|LLM info(for the linked model)|Pipeline repository|
-|-|-|-|-|-|
-|[LLaVA 13B](https://github.com/haotian-liu/LLaVA)|`llava-13b`|[LLaVA 13B](https://huggingface.co/wojtab/llava-13b-v0-4bit-128g)|GPTQ 4-bit quant, old CUDA|built-in|
-|[LLaVA 7B](https://github.com/haotian-liu/LLaVA)|`llava-7b`|[LLaVA 7B](https://huggingface.co/wojtab/llava-7b-v0-4bit-128g)|GPTQ 4-bit quant, old CUDA|built-in|
-|[MiniGPT-4 7B](https://github.com/Vision-CAIR/MiniGPT-4)|`minigpt4-7b`|[Vicuna v0 7B](https://huggingface.co/TheBloke/vicuna-7B-GPTQ-4bit-128g)|GPTQ 4-bit quant, new format|[Wojtab/minigpt-4-pipeline](https://github.com/Wojtab/minigpt-4-pipeline)|
-|[MiniGPT-4 13B](https://github.com/Vision-CAIR/MiniGPT-4)|`minigpt4-13b`|[Vicuna v0 13B](https://huggingface.co/anon8231489123/vicuna-13b-GPTQ-4bit-128g)|GPTQ 4-bit quant, old CUDA|[Wojtab/minigpt-4-pipeline](https://github.com/Wojtab/minigpt-4-pipeline)|
-|[InstructBLIP 7B](https://github.com/salesforce/LAVIS/tree/main/projects/instructblip)|`instructblip-7b`|[Vicuna v1.1 7B](https://huggingface.co/TheBloke/vicuna-7B-1.1-GPTQ-4bit-128g)|GPTQ 4-bit quant|[kjerk/instructblip-pipeline](https://github.com/kjerk/instructblip-pipeline)|
-|[InstructBLIP 13B](https://github.com/salesforce/LAVIS/tree/main/projects/instructblip)|`instructblip-13b`|[Vicuna v1.1 13B](https://huggingface.co/TheBloke/vicuna-13B-1.1-GPTQ-4bit-128g)|GPTQ 4-bit quant|[kjerk/instructblip-pipeline](https://github.com/kjerk/instructblip-pipeline)|
-
-Some pipelines could support different LLMs but do note that while it might work, it isn't a supported configuration.
-
-DO NOT report bugs if you are using a different LLM.
-
-DO NOT report bugs with pipelines in this repository (unless they are built-in)
-
-## Extension config
-This extension uses the following parameters (from `settings.json`):
-|Parameter|Description|
-|---------|-----------|
-|`multimodal-vision_bits`|Number of bits to load vision models (CLIP/ViT) feature extractor in (most pipelines should support either 32 or 16, default=32)|
-|`multimodal-vision_device`|Torch device to run the feature extractor on, for example, `cpu` or `cuda:0`, by default `cuda:0` if available|
-|`multimodal-projector_bits`|Number of bits to load feature projector model(s) in (most pipelines should support either 32 or 16, default=32)|
-|`multimodal-projector_device`|Torch device to run the feature projector model(s) on, for example `cpu` or `cuda:0`, by default `cuda:0` if available|
-|`multimodal-add_all_images_to_prompt`|Default value of "Embed all images, not only the last one" checkbox|
-
-## Usage through API
-
-### Chat completions endpoint
-
-#### With an image URL
-
-```shell
-curl http://127.0.0.1:5000/v1/chat/completions \
-  -H "Content-Type: application/json" \
-  -d '{
-    "messages": [
-      {
-        "role": "user",
-        "image_url": "https://avatars.githubusercontent.com/u/112222186?v=4"
-      },
-      {
-        "role": "user",
-        "content": "What is unusual about this image?"
-      }
-    ]
-  }'
-```
-
-#### With a Base64 image
-
-```python
-import base64
-import json
-import requests
-
-img = open('image.jpg', 'rb')
-img_bytes = img.read()
-img_base64 = base64.b64encode(img_bytes).decode('utf-8')
-data = { "messages": [
-        {
-            "role": "user",
-            "image_url": f"data:image/jpeg;base64,{img_base64}"
-        },
-        {
-            "role": "user",
-            "content": "what is unusual about this image?"
-        }
-    ]
-}
-response = requests.post('http://127.0.0.1:5000/v1/chat/completions', json=data)
-print(response.text)
-```
-
-You can run the multimodal inference through API, by inputting the images to prompt. Images are embedded like so: `f''`, where `img_str` is base-64 jpeg data. Note that you will need to launch `server.py` with the arguments `--api --extensions multimodal`. 
-
-### Completions endpoint
-
-Python example:
-
-```Python
-import base64
-import requests
-
-CONTEXT = "You are LLaVA, a large language and vision assistant trained by UW Madison WAIV Lab. You are able to understand the visual content that the user provides, and assist the user with a variety of tasks using natural language. Follow the instructions carefully and explain your answers in detail.### Human: Hi!### Assistant: Hi there! How can I help you today?\n"
-
-with open('extreme_ironing.jpg', 'rb') as f:
-    img_str = base64.b64encode(f.read()).decode('utf-8')
-    prompt = CONTEXT + f'### Human: What is unusual about this image: \n### Assistant: '
-    print(requests.post('http://127.0.0.1:5000/v1/completions', json={'prompt': prompt, 'max_tokens': 200, 'stop': ['\n###']}).json())
-```
-script output:
-```Python
-{'results': [{'text': "The unusual aspect of this image is that a man is standing on top of a yellow minivan while doing his laundry. He has set up a makeshift clothes line using the car's rooftop as an outdoor drying area. This scene is uncommon because people typically do their laundry indoors, in a dedicated space like a laundromat or a room in their home, rather than on top of a moving vehicle. Additionally, hanging clothes on the car could be potentially hazardous or illegal in some jurisdictions due to the risk of damaging the vehicle or causing accidents on the road.\n##"}]}
-```
-
-## For pipeline developers/technical description
-see [DOCS.md](https://github.com/oobabooga/text-generation-webui/blob/main/extensions/multimodal/DOCS.md)
diff --git a/extensions/multimodal/abstract_pipeline.py b/extensions/multimodal/abstract_pipeline.py
deleted file mode 100644
index 9c49935a2d..0000000000
--- a/extensions/multimodal/abstract_pipeline.py
+++ /dev/null
@@ -1,63 +0,0 @@
-from abc import ABC, abstractmethod
-from typing import List, Optional
-
-import torch
-from PIL import Image
-from transformers import is_torch_xpu_available
-
-
-class AbstractMultimodalPipeline(ABC):
-    @staticmethod
-    @abstractmethod
-    def name() -> str:
-        'name of the pipeline, should be same as in --multimodal-pipeline'
-        pass
-
-    @staticmethod
-    @abstractmethod
-    def image_start() -> Optional[str]:
-        'return image start string, string representation of image start token, or None if not applicable'
-        pass
-
-    @staticmethod
-    @abstractmethod
-    def image_end() -> Optional[str]:
-        'return image end string, string representation of image end token, or None if not applicable'
-        pass
-
-    @staticmethod
-    @abstractmethod
-    def placeholder_token_id() -> int:
-        'return placeholder token id'
-        pass
-
-    @staticmethod
-    @abstractmethod
-    def num_image_embeds() -> int:
-        'return the number of embeds used by a single image (for example: 256 for LLaVA)'
-        pass
-
-    @abstractmethod
-    def embed_images(self, images: List[Image.Image]) -> torch.Tensor:
-        'forward the images through vision pipeline, and return their embeddings'
-        pass
-
-    @staticmethod
-    @abstractmethod
-    def embed_tokens(input_ids: torch.Tensor) -> torch.Tensor:
-        'embed tokens, the exact function varies by LLM, for LLaMA it is `shared.model.model.embed_tokens`'
-        pass
-
-    @staticmethod
-    @abstractmethod
-    def placeholder_embeddings() -> torch.Tensor:
-        'get placeholder embeddings if there are multiple images, and `add_all_images_to_prompt` is False'
-        pass
-
-    def _get_device(self, setting_name: str, params: dict):
-        if params[setting_name] is None:
-            return torch.device("cuda:0" if torch.cuda.is_available() else "xpu:0" if is_torch_xpu_available() else "cpu")
-        return torch.device(params[setting_name])
-
-    def _get_dtype(self, setting_name: str, params: dict):
-        return torch.float32 if int(params[setting_name]) == 32 else torch.float16
diff --git a/extensions/multimodal/multimodal_embedder.py b/extensions/multimodal/multimodal_embedder.py
deleted file mode 100644
index 626077cb80..0000000000
--- a/extensions/multimodal/multimodal_embedder.py
+++ /dev/null
@@ -1,178 +0,0 @@
-import base64
-import re
-from dataclasses import dataclass
-from io import BytesIO
-from typing import Any, List, Optional
-
-import torch
-from PIL import Image
-
-from extensions.multimodal.pipeline_loader import load_pipeline
-from modules import shared
-from modules.logging_colors import logger
-from modules.text_generation import encode, get_max_prompt_length
-
-
-@dataclass
-class PromptPart:
-    text: str
-    image: Optional[Image.Image] = None
-    is_image: bool = False
-    input_ids: Optional[torch.Tensor] = None
-    embedding: Optional[torch.Tensor] = None
-
-
-class MultimodalEmbedder:
-    def __init__(self, params: dict):
-        pipeline, source = load_pipeline(params)
-        self.pipeline = pipeline
-        logger.info(f'Multimodal: loaded pipeline {self.pipeline.name()} from pipelines/{source} ({self.pipeline.__class__.__name__})')
-
-    def _split_prompt(self, prompt: str, load_images: bool = False) -> List[PromptPart]:
-        """Splits a prompt into a list of `PromptParts` to separate image data from text.
-        It will also append `image_start` and `image_end` before and after the image, and optionally parse and load the images,
-        if `load_images` is `True`.
-        """
-        parts: List[PromptPart] = []
-        curr = 0
-        while True:
-            match = re.search(r'', prompt[curr:])
-            if match is None:
-                # no more image tokens, append the rest of the prompt
-                if curr > 0:
-                    # add image end token after last image
-                    parts.append(PromptPart(text=self.pipeline.image_end() + prompt[curr:]))
-                else:
-                    parts.append(PromptPart(text=prompt))
-                break
-            # found an image, append image start token to the text
-            if match.start() > 0:
-                parts.append(PromptPart(text=prompt[curr:curr + match.start()] + self.pipeline.image_start()))
-            else:
-                parts.append(PromptPart(text=self.pipeline.image_start()))
-            # append the image
-            parts.append(PromptPart(
-                text=match.group(0),
-                image=Image.open(BytesIO(base64.b64decode(match.group(1)))) if load_images else None,
-                is_image=True
-            ))
-            curr += match.end()
-        return parts
-
-    def _len_in_tokens_prompt_parts(self, parts: List[PromptPart]) -> int:
-        """Total length in tokens of all `parts`"""
-        tokens = 0
-        for part in parts:
-            if part.is_image:
-                tokens += self.pipeline.num_image_embeds()
-            elif part.input_ids is not None:
-                tokens += len(part.input_ids)
-            else:
-                tokens += len(encode(part.text)[0])
-        return tokens
-
-    def len_in_tokens(self, prompt: str) -> int:
-        """Total length in tokens for a given text `prompt`"""
-        parts = self._split_prompt(prompt, False)
-        return self._len_in_tokens_prompt_parts(parts)
-
-    def _encode_single_text(self, part: PromptPart, add_bos_token: bool) -> PromptPart:
-        """Encode a single prompt `part` to `input_ids`. Returns a `PromptPart`"""
-        if part.is_image:
-            placeholders = torch.ones((self.pipeline.num_image_embeds())) * self.pipeline.placeholder_token_id()
-            part.input_ids = placeholders.to(shared.model.device, dtype=torch.int64)
-        else:
-            part.input_ids = encode(part.text, add_bos_token=add_bos_token)[0].to(shared.model.device, dtype=torch.int64)
-        return part
-
-    @staticmethod
-    def _num_images(parts: List[PromptPart]) -> int:
-        count = 0
-        for part in parts:
-            if part.is_image:
-                count += 1
-        return count
-
-    def _encode_text(self, state, parts: List[PromptPart]) -> List[PromptPart]:
-        """Encode text to token_ids, also truncate the prompt, if necessary.
-
-        The chat/instruct mode should make prompts that fit in get_max_prompt_length, but if max_new_tokens are set
-        such that the context + min_rows don't fit, we can get a prompt which is too long.
-        We can't truncate image embeddings, as it leads to broken generation, so remove the images instead and warn the user
-        """
-        encoded: List[PromptPart] = []
-        for i, part in enumerate(parts):
-            encoded.append(self._encode_single_text(part, i == 0 and state['add_bos_token']))
-
-        # truncation:
-        max_len = get_max_prompt_length(state)
-        removed_images = 0
-
-        # 1. remove entire text/image blocks
-        while self._len_in_tokens_prompt_parts(encoded[1:]) > max_len:
-            if encoded[0].is_image:
-                removed_images += 1
-            encoded = encoded[1:]
-
-        # 2. check if the last prompt part doesn't need to get truncated
-        if self._len_in_tokens_prompt_parts(encoded) > max_len:
-            if encoded[0].is_image:
-                # don't truncate image embeddings, just remove the image, otherwise generation will be broken
-                removed_images += 1
-                encoded = encoded[1:]
-            elif len(encoded) > 1 and encoded[0].text.endswith(self.pipeline.image_start()):
-                # see if we can keep image_start token
-                len_image_start = len(encode(self.pipeline.image_start(), add_bos_token=state['add_bos_token'])[0])
-                if self._len_in_tokens_prompt_parts(encoded[1:]) + len_image_start > max_len:
-                    # we can't -> remove this text, and the image
-                    encoded = encoded[2:]
-                    removed_images += 1
-                else:
-                    # we can -> just truncate the text
-                    trunc_len = self._len_in_tokens_prompt_parts(encoded) - max_len
-                    encoded[0].input_ids = encoded[0].input_ids[trunc_len:]
-            elif len(encoded) > 0:
-                # only one text left, truncate it normally
-                trunc_len = self._len_in_tokens_prompt_parts(encoded) - max_len
-                encoded[0].input_ids = encoded[0].input_ids[trunc_len:]
-
-        # notify user if we truncated an image
-        if removed_images > 0:
-            logger.warning(f"Multimodal: removed {removed_images} image(s) from prompt. Try decreasing max_new_tokens if generation is broken")
-
-        return encoded
-
-    def _embed(self, parts: List[PromptPart]) -> List[PromptPart]:
-        # batch images
-        image_indicies = [i for i, part in enumerate(parts) if part.is_image]
-        embedded = self.pipeline.embed_images([parts[i].image for i in image_indicies])
-        for i, embeds in zip(image_indicies, embedded):
-            parts[i].embedding = embeds
-        # embed text
-        for (i, part) in enumerate(parts):
-            if not part.is_image:
-                parts[i].embedding = self.pipeline.embed_tokens(part.input_ids)
-        return parts
-
-    def _remove_old_images(self, parts: List[PromptPart], params: dict) -> List[PromptPart]:
-        if params['add_all_images_to_prompt']:
-            return parts
-        already_added = False
-        for i, part in reversed(list(enumerate(parts))):
-            if part.is_image:
-                if already_added:
-                    parts[i].embedding = self.pipeline.placeholder_embeddings()
-                else:
-                    already_added = True
-        return parts
-
-    def forward(self, prompt: str, state: Any, params: dict):
-        prompt_parts = self._split_prompt(prompt, True)
-        prompt_parts = self._encode_text(state, prompt_parts)
-        prompt_parts = self._embed(prompt_parts)
-        prompt_parts = self._remove_old_images(prompt_parts, params)
-        embeds = tuple(part.embedding for part in prompt_parts)
-        ids = tuple(part.input_ids for part in prompt_parts)
-        input_embeds = torch.cat(embeds, dim=0)
-        input_ids = torch.cat(ids, dim=0)
-        return prompt, input_ids, input_embeds, self._num_images(prompt_parts)
diff --git a/extensions/multimodal/pipeline_loader.py b/extensions/multimodal/pipeline_loader.py
deleted file mode 100644
index 8fcd0a9b41..0000000000
--- a/extensions/multimodal/pipeline_loader.py
+++ /dev/null
@@ -1,52 +0,0 @@
-import traceback
-from importlib import import_module
-from pathlib import Path
-from typing import Tuple
-
-from extensions.multimodal.abstract_pipeline import AbstractMultimodalPipeline
-from modules import shared
-from modules.logging_colors import logger
-
-
-def _get_available_pipeline_modules():
-    pipeline_path = Path(__file__).parent / 'pipelines'
-    modules = [p for p in pipeline_path.iterdir() if p.is_dir()]
-    return [m.name for m in modules if (m / 'pipelines.py').exists()]
-
-
-def load_pipeline(params: dict) -> Tuple[AbstractMultimodalPipeline, str]:
-    pipeline_modules = {}
-    available_pipeline_modules = _get_available_pipeline_modules()
-    for name in available_pipeline_modules:
-        try:
-            pipeline_modules[name] = import_module(f'extensions.multimodal.pipelines.{name}.pipelines')
-        except:
-            logger.warning(f'Failed to get multimodal pipelines from {name}')
-            logger.warning(traceback.format_exc())
-
-    if shared.args.multimodal_pipeline is not None:
-        for k in pipeline_modules:
-            if hasattr(pipeline_modules[k], 'get_pipeline'):
-                pipeline = getattr(pipeline_modules[k], 'get_pipeline')(shared.args.multimodal_pipeline, params)
-                if pipeline is not None:
-                    return (pipeline, k)
-    else:
-        model_name = shared.args.model.lower()
-        for k in pipeline_modules:
-            if hasattr(pipeline_modules[k], 'get_pipeline_from_model_name'):
-                pipeline = getattr(pipeline_modules[k], 'get_pipeline_from_model_name')(model_name, params)
-                if pipeline is not None:
-                    return (pipeline, k)
-
-    available = []
-    for k in pipeline_modules:
-        if hasattr(pipeline_modules[k], 'available_pipelines'):
-            pipelines = getattr(pipeline_modules[k], 'available_pipelines')
-            available += pipelines
-
-    if shared.args.multimodal_pipeline is not None:
-        log = f'Multimodal - ERROR: Failed to load multimodal pipeline "{shared.args.multimodal_pipeline}", available pipelines are: {available}.'
-    else:
-        log = f'Multimodal - ERROR: Failed to determine multimodal pipeline for model {shared.args.model}, please select one manually using --multimodal-pipeline [PIPELINE]. Available pipelines are: {available}.'
-    logger.critical(f'{log} Please specify a correct pipeline, or disable the extension')
-    raise RuntimeError(f'{log} Please specify a correct pipeline, or disable the extension')
diff --git a/extensions/multimodal/pipelines/llava/README.md b/extensions/multimodal/pipelines/llava/README.md
deleted file mode 100644
index aff64faaae..0000000000
--- a/extensions/multimodal/pipelines/llava/README.md
+++ /dev/null
@@ -1,9 +0,0 @@
-## LLaVA pipeline
-
-This module provides 2 pipelines:
-- `llava-7b` - for use with LLaVA v0 7B model (finetuned LLaMa 7B)
-- `llava-13b` - for use with LLaVA v0 13B model (finetuned LLaMa 13B)
-
-[LLaVA](https://github.com/haotian-liu/LLaVA) uses CLIP `openai/clip-vit-large-patch14` as the vision model, and then a single linear layer. For 13B the projector weights are in `liuhaotian/LLaVA-13b-delta-v0`, and for 7B they are in `liuhaotian/LLaVA-7b-delta-v0`.
-
-The supported parameter combinations for both the vision model, and the projector are: CUDA/32bit, CUDA/16bit, CPU/32bit
diff --git a/extensions/multimodal/pipelines/llava/llava.py b/extensions/multimodal/pipelines/llava/llava.py
deleted file mode 100644
index 09b5aff7b8..0000000000
--- a/extensions/multimodal/pipelines/llava/llava.py
+++ /dev/null
@@ -1,262 +0,0 @@
-import time
-from abc import abstractmethod
-from typing import List, Tuple
-
-import torch
-from huggingface_hub import hf_hub_download
-from PIL import Image
-from transformers import CLIPImageProcessor, CLIPVisionModel
-
-from extensions.multimodal.abstract_pipeline import AbstractMultimodalPipeline
-from modules import shared
-from modules.logging_colors import logger
-from modules.text_generation import encode
-
-
-def expand2square(pil_img: Image.Image, background_color: Tuple[int]) -> Image.Image:
-    width, height = pil_img.size
-    if width == height:
-        return pil_img
-    elif width > height:
-        result = Image.new(pil_img.mode, (width, width), background_color)
-        result.paste(pil_img, (0, (width - height) // 2))
-        return result
-    else:
-        result = Image.new(pil_img.mode, (height, height), background_color)
-        result.paste(pil_img, ((height - width) // 2, 0))
-        return result
-
-
-class LLaVA_v0_Pipeline(AbstractMultimodalPipeline):
-    CLIP_REPO = "openai/clip-vit-large-patch14"
-
-    def __init__(self, params: dict) -> None:
-        super().__init__()
-        self.clip_device = self._get_device("vision_device", params)
-        self.clip_dtype = self._get_dtype("vision_bits", params)
-        self.projector_device = self._get_device("projector_device", params)
-        self.projector_dtype = self._get_dtype("projector_bits", params)
-        self.image_processor, self.vision_tower, self.mm_projector = self._load_models()
-
-    def _load_models(self):
-        start_ts = time.time()
-
-        logger.info(f"LLaVA - Loading CLIP from {self.CLIP_REPO} as {self.clip_dtype} on {self.clip_device}...")
-        image_processor = CLIPImageProcessor.from_pretrained(self.CLIP_REPO, torch_dtype=self.clip_dtype)
-        vision_tower = CLIPVisionModel.from_pretrained(self.CLIP_REPO, torch_dtype=self.clip_dtype).to(self.clip_device)
-
-        logger.info(f"LLaVA - Loading projector from {self.llava_projector_repo()} as {self.projector_dtype} on {self.projector_device}...")
-        projector_path = hf_hub_download(self.llava_projector_repo(), self.llava_projector_filename())
-        mm_projector = self.build_mm_projector()
-        projector_data = torch.load(projector_path)
-        projector_data = {k[19:]: v for k, v in projector_data.items() if k.startswith('model.mm_projector.')}
-        mm_projector.load_state_dict(projector_data)
-        mm_projector = mm_projector.to(self.projector_device)
-
-        logger.info(f"LLaVA supporting models loaded, took {time.time() - start_ts:.2f} seconds")
-        return image_processor, vision_tower, mm_projector
-
-    def build_mm_projector(self) -> torch.nn.Module:
-        projector_shape = self.llava_projector_shape()
-        if len(projector_shape) == 2:
-            return torch.nn.Linear(*projector_shape)
-        else:
-            modules = []
-            modules.append(torch.nn.Linear(projector_shape[0], projector_shape[1]))
-            for i in range(2, len(projector_shape)):
-                modules.append(torch.nn.GELU())
-                modules.append(torch.nn.Linear(projector_shape[i-1], projector_shape[i]))
-            return torch.nn.Sequential(*modules)
-
-    @staticmethod
-    def image_start() -> str:
-        return ""
-
-    @staticmethod
-    def image_end() -> str:
-        return ""
-
-    @staticmethod
-    def num_image_embeds() -> int:
-        return 256
-
-    @staticmethod
-    def embed_tokens(input_ids: torch.Tensor) -> torch.Tensor:
-        for attr in ['', 'model', 'model.model', 'model.model.model']:
-            tmp = getattr(shared.model, attr, None) if attr != '' else shared.model
-            if tmp is not None and hasattr(tmp, 'embed_tokens'):
-                func = tmp.embed_tokens
-                break
-        else:
-            raise ValueError('The embed_tokens method has not been found for this loader.')
-
-        return func(input_ids).to(shared.model.device, dtype=shared.model.dtype)
-
-    @staticmethod
-    def placeholder_embeddings() -> torch.Tensor:
-        return LLaVA_v0_Pipeline.embed_tokens(encode(""*256, add_bos_token=False)[0])
-
-    def embed_images(self, images: List[Image.Image]) -> torch.Tensor:
-        images = self.image_processor(images, return_tensors='pt')['pixel_values']
-        images = images.to(self.clip_device, dtype=self.clip_dtype)
-
-        with torch.no_grad():
-            image_forward_outs = self.vision_tower(images, output_hidden_states=True)
-            select_hidden_state_layer = -2
-            select_hidden_state = image_forward_outs.hidden_states[select_hidden_state_layer]
-            image_features = select_hidden_state[:, 1:].to(self.projector_device, dtype=self.projector_dtype)
-            image_features = self.mm_projector(image_features)
-        return image_features.to(shared.model.device, dtype=shared.model.dtype)
-
-    @staticmethod
-    @abstractmethod
-    def llava_projector_repo() -> str:
-        pass
-
-    @staticmethod
-    @abstractmethod
-    def llava_projector_filename() -> str:
-        pass
-
-    @staticmethod
-    @abstractmethod
-    def llava_projector_shape() -> Tuple[int, int]:
-        pass
-
-
-class LLaVA_v0_13B_Pipeline(LLaVA_v0_Pipeline):
-    def __init__(self, params: dict) -> None:
-        super().__init__(params)
-
-    @staticmethod
-    def name() -> str:
-        return "llava-13b"
-
-    @staticmethod
-    def placeholder_token_id() -> int:
-        return 32000
-
-    @staticmethod
-    def llava_projector_shape() -> Tuple[int, int]:
-        return (1024, 5120)
-
-    @staticmethod
-    def llava_projector_filename() -> str:
-        return "mm_projector.bin"
-
-    @staticmethod
-    def llava_projector_repo() -> str:
-        return "liuhaotian/LLaVA-13b-delta-v0"
-
-
-class LLaVA_v0_7B_Pipeline(LLaVA_v0_Pipeline):
-    def __init__(self, params: dict) -> None:
-        super().__init__(params)
-
-    @staticmethod
-    def name() -> str:
-        return "llava-7b"
-
-    @staticmethod
-    def placeholder_token_id() -> int:
-        return 32001
-
-    @staticmethod
-    def llava_projector_shape() -> Tuple[int, int]:
-        return (1024, 4096)
-
-    @staticmethod
-    def llava_projector_filename() -> str:
-        return "mm_projector.bin"
-
-    @staticmethod
-    def llava_projector_repo() -> str:
-        return "liuhaotian/LLaVA-7b-delta-v0"
-
-
-class LLaVA_LLaMA_2_13B_Pipeline(LLaVA_v0_13B_Pipeline):
-    def __init__(self, params: dict) -> None:
-        super().__init__(params)
-
-    @staticmethod
-    def name() -> str:
-        return "llava-llama-2-13b"
-
-    @staticmethod
-    def placeholder_token_id() -> int:
-        return 0
-
-    @staticmethod
-    def llava_projector_repo() -> str:
-        return "liuhaotian/llava-llama-2-13b-chat-lightning-preview"
-
-    @staticmethod
-    def image_start() -> str:
-        return ""
-
-    @staticmethod
-    def image_end() -> str:
-        return ""
-
-    @staticmethod
-    def placeholder_embeddings() -> torch.Tensor:
-        return LLaVA_v0_Pipeline.embed_tokens(encode(""*256, add_bos_token=False)[0])
-
-
-class LLaVA_v1_5_13B_Pipeline(LLaVA_v0_13B_Pipeline):
-    CLIP_REPO = "openai/clip-vit-large-patch14-336"
-
-    def __init__(self, params: dict) -> None:
-        super().__init__(params)
-
-    @staticmethod
-    def name() -> str:
-        return "llava-v1.5-13b"
-
-    @staticmethod
-    def llava_projector_shape() -> Tuple[int, int]:
-        return (1024, 5120, 5120)
-
-    @staticmethod
-    def placeholder_token_id() -> int:
-        return 0
-
-    @staticmethod
-    def llava_projector_repo() -> str:
-        return "liuhaotian/llava-v1.5-13b"
-
-    @staticmethod
-    def image_start() -> str:
-        return ""
-
-    @staticmethod
-    def image_end() -> str:
-        return ""
-
-    @staticmethod
-    def num_image_embeds() -> int:
-        return 576
-
-    def embed_images(self, images: List[Image.Image]) -> torch.Tensor:
-        # pad it to square first
-        images = [
-            expand2square(image, tuple(int(x*255) for x in self.image_processor.image_mean))
-            for image in images
-        ]
-        return super().embed_images(images)
-
-    @staticmethod
-    def placeholder_embeddings() -> torch.Tensor:
-        return LLaVA_v0_Pipeline.embed_tokens(encode(""*576, add_bos_token=False)[0])
-
-class LLaVA_v1_5_7B_Pipeline(LLaVA_v1_5_13B_Pipeline):
-    @staticmethod
-    def name() -> str:
-        return "llava-v1.5-7b"
-
-    @staticmethod
-    def llava_projector_shape() -> Tuple[int, int]:
-        return (1024, 4096, 4096)
-    @staticmethod
-    def llava_projector_repo() -> str:
-        return "liuhaotian/llava-v1.5-7b"
\ No newline at end of file
diff --git a/extensions/multimodal/pipelines/llava/pipelines.py b/extensions/multimodal/pipelines/llava/pipelines.py
deleted file mode 100644
index e6833ed6ff..0000000000
--- a/extensions/multimodal/pipelines/llava/pipelines.py
+++ /dev/null
@@ -1,48 +0,0 @@
-from typing import Optional
-
-from extensions.multimodal.abstract_pipeline import AbstractMultimodalPipeline
-
-available_pipelines = ['llava-7b', 'llava-13b', 'llava-llama-2-13b', 'llava-v1.5-13b', 'llava-v1.5-7b']
-
-
-def get_pipeline(name: str, params: dict) -> Optional[AbstractMultimodalPipeline]:
-    if name == 'llava-7b':
-        from .llava import LLaVA_v0_7B_Pipeline
-        return LLaVA_v0_7B_Pipeline(params)
-    if name == 'llava-13b':
-        from .llava import LLaVA_v0_13B_Pipeline
-        return LLaVA_v0_13B_Pipeline(params)
-    if name == 'llava-llama-2-13b':
-        from .llava import LLaVA_LLaMA_2_13B_Pipeline
-        return LLaVA_LLaMA_2_13B_Pipeline(params)
-    if name == 'llava-v1.5-7b':
-        from .llava import LLaVA_v1_5_7B_Pipeline
-        return LLaVA_v1_5_7B_Pipeline(params)
-    if name == 'llava-v1.5-13b':
-        from .llava import LLaVA_v1_5_13B_Pipeline
-        return LLaVA_v1_5_13B_Pipeline(params)
-    return None
-
-
-def get_pipeline_from_model_name(model_name: str, params: dict) -> Optional[AbstractMultimodalPipeline]:
-    if 'llava' not in model_name.lower():
-        return None
-    if 'llama-2' in model_name.lower():
-        if '13b' in model_name.lower():
-            from .llava import LLaVA_LLaMA_2_13B_Pipeline
-            return LLaVA_LLaMA_2_13B_Pipeline(params)
-    elif 'llava-v1.5' in model_name.lower():
-        if '13b' in model_name.lower():
-            from .llava import LLaVA_v1_5_13B_Pipeline
-            return LLaVA_v1_5_13B_Pipeline(params)
-        if '7b' in model_name.lower():
-            from .llava import LLaVA_v1_5_7B_Pipeline
-            return LLaVA_v1_5_7B_Pipeline(params)
-    else:
-        if '7b' in model_name.lower():
-            from .llava import LLaVA_v0_7B_Pipeline
-            return LLaVA_v0_7B_Pipeline(params)
-        if '13b' in model_name.lower():
-            from .llava import LLaVA_v0_13B_Pipeline
-            return LLaVA_v0_13B_Pipeline(params)
-    return None
diff --git a/extensions/multimodal/pipelines/place-additional-pipelines-here.txt b/extensions/multimodal/pipelines/place-additional-pipelines-here.txt
deleted file mode 100644
index e69de29bb2..0000000000
diff --git a/extensions/multimodal/script.py b/extensions/multimodal/script.py
deleted file mode 100644
index 6607ce5a32..0000000000
--- a/extensions/multimodal/script.py
+++ /dev/null
@@ -1,113 +0,0 @@
-import base64
-import re
-import time
-from functools import partial
-from io import BytesIO
-
-import gradio as gr
-import torch
-
-from extensions.multimodal.multimodal_embedder import MultimodalEmbedder
-from modules import shared
-from modules.logging_colors import logger
-
-params = {
-    "add_all_images_to_prompt": False,
-    # device to run vision encoder on
-    "vision_device": None,
-    # bits to load vision encoder in, either 16 or 32
-    "vision_bits": 32,
-    # device to run multimodal projector on
-    "projector_device": None,
-    # multimodal projector bits, either 32 or 16
-    "projector_bits": 32
-}
-
-
-# If 'state' is True, will hijack the next chat generation
-input_hijack = {
-    'state': False,
-    'value': ["", ""]
-}
-
-
-# initialized in ui, so that params are loaded from settings
-multimodal_embedder: MultimodalEmbedder = None
-
-
-def chat_input_modifier(text, visible_text, state):
-    global input_hijack
-    if input_hijack['state']:
-        input_hijack['state'] = False
-        return input_hijack['value'](text, visible_text)
-    else:
-        return text, visible_text
-
-
-def add_chat_picture(picture, text, visible_text):
-    # resize the image, so that shortest edge is at least 224 (size for CLIP), and at most 300 (to keep history manageable)
-    # Adjusted to 336 for the values here, due to the increased resolution in llava-v1.5
-    max_hw, min_hw = max(picture.size), min(picture.size)
-    aspect_ratio = max_hw / min_hw
-    shortest_edge = int(max(336 / aspect_ratio, 336))
-    longest_edge = int(shortest_edge * aspect_ratio)
-    w = shortest_edge if picture.width < picture.height else longest_edge
-    h = shortest_edge if picture.width >= picture.height else longest_edge
-    picture = picture.resize((w, h))
-
-    buffer = BytesIO()
-    picture.save(buffer, format="PNG")
-    img_str = base64.b64encode(buffer.getvalue()).decode('utf-8')
-    image = f''
-
-    if '' in text:
-        text = text.replace('', image)
-    else:
-        text = image + '\n' + text
-
-    if visible_text == '' or visible_text is None:
-        visible_text = text
-    elif '' in visible_text:
-        visible_text = visible_text.replace('', image)
-    else:
-        visible_text = visible_text + '\n' + image
-
-    return text, visible_text
-
-
-def custom_tokenized_length(prompt):
-    return multimodal_embedder.len_in_tokens(prompt)
-
-
-def tokenizer_modifier(state, prompt, input_ids, input_embeds):
-    global params
-    start_ts = time.time()
-    image_match = re.search(r'', prompt)
-
-    if image_match is None:
-        return prompt, input_ids, input_embeds
-
-    prompt, input_ids, input_embeds, total_embedded = multimodal_embedder.forward(prompt, state, params)
-    logger.info(f'Embedded {total_embedded} image(s) in {time.time()-start_ts:.2f}s')
-    return (prompt,
-            input_ids.unsqueeze(0).to(shared.model.device, dtype=torch.int64),
-            input_embeds.unsqueeze(0).to(shared.model.device, dtype=shared.model.dtype))
-
-
-def ui():
-    global multimodal_embedder
-    multimodal_embedder = MultimodalEmbedder(params)
-    with gr.Column():
-        picture_select = gr.Image(label='Send a picture', type='pil')
-        # The models don't seem to deal well with multiple images
-        single_image_checkbox = gr.Checkbox(False, label='Embed all images, not only the last one')
-    # Prepare the input hijack
-    picture_select.upload(
-        lambda picture: input_hijack.update({"state": True, "value": partial(add_chat_picture, picture)}),
-        [picture_select],
-        None
-    )
-    picture_select.clear(lambda: input_hijack.update({"state": False, "value": ["", ""]}), None, None)
-    single_image_checkbox.change(lambda x: params.update({"add_all_images_to_prompt": x}), single_image_checkbox, None)
-    shared.gradio['Generate'].click(lambda: None, None, picture_select)
-    shared.gradio['textbox'].submit(lambda: None, None, picture_select)
diff --git a/modules/shared.py b/modules/shared.py
index 0981f6fb92..91a2d361e2 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -197,10 +197,6 @@
 group.add_argument('--api-disable-ipv4', action='store_true', help='Disable IPv4 for the API')
 group.add_argument('--nowebui', action='store_true', help='Do not launch the Gradio UI. Useful for launching the API in standalone mode.')
 
-# Multimodal
-group = parser.add_argument_group('Multimodal')
-group.add_argument('--multimodal-pipeline', type=str, default=None, help='The multimodal pipeline to use. Examples: llava-7b, llava-13b.')
-
 # Deprecated parameters
 group = parser.add_argument_group('Deprecated')
 group.add_argument('--cache_4bit', action='store_true', help='DEPRECATED')
@@ -215,6 +211,7 @@
 group.add_argument('--wbits', type=int, default=0, help='DEPRECATED')
 group.add_argument('--groupsize', type=int, default=-1, help='DEPRECATED')
 group.add_argument('--model-menu', action='store_true', help='DEPRECATED')
+group.add_argument('--multimodal-pipeline', type=str, default=None, help='DEPRECATED')
 
 args = parser.parse_args()
 args_defaults = parser.parse_args([])
@@ -370,10 +367,6 @@ def load_user_config():
 args.loader = fix_loader_name(args.loader)
 args = transform_legacy_kv_cache_options(args)
 
-# Activate the multimodal extension
-if args.multimodal_pipeline is not None:
-    add_extension('multimodal')
-
 # Activate the API extension
 if args.api or args.public_api:
     add_extension('openai', last=True)

From d7b336d37e823541e3a0dd33a060147c3cea5290 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Apr 2025 20:12:14 -0700
Subject: [PATCH 0279/1701] Update the README

---
 README.md | 14 +++++---------
 1 file changed, 5 insertions(+), 9 deletions(-)

diff --git a/README.md b/README.md
index 63b8931a9f..7f440222ee 100644
--- a/README.md
+++ b/README.md
@@ -184,7 +184,7 @@ List of command-line flags
 
 
 ```txt
-usage: server.py [-h] [--multi-user] [--character CHARACTER] [--model MODEL] [--lora LORA [LORA ...]] [--model-dir MODEL_DIR] [--lora-dir LORA_DIR] [--model-menu] [--settings SETTINGS]
+usage: server.py [-h] [--multi-user] [--character CHARACTER] [--model MODEL] [--lora LORA [LORA ...]] [--model-dir MODEL_DIR] [--lora-dir LORA_DIR] [--settings SETTINGS]
                  [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--cpu] [--auto-devices] [--gpu-memory GPU_MEMORY [GPU_MEMORY ...]]
                  [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast]
                  [--use_flash_attention_2] [--use_eager_attention] [--torch-compile] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn]
@@ -195,8 +195,8 @@ usage: server.py [-h] [--multi-user] [--character CHARACTER] [--model MODEL] [--
                  [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT]
                  [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE]
                  [--subpath SUBPATH] [--old-colors] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6]
-                 [--api-disable-ipv4] [--nowebui] [--multimodal-pipeline MULTIMODAL_PIPELINE] [--cache_4bit] [--cache_8bit] [--chat-buttons] [--triton] [--no_inject_fused_mlp] [--no_use_cuda_fp16]
-                 [--desc_act] [--disable_exllama] [--disable_exllamav2] [--wbits WBITS] [--groupsize GROUPSIZE]
+                 [--api-disable-ipv4] [--nowebui] [--cache_4bit] [--cache_8bit] [--chat-buttons] [--triton] [--no_inject_fused_mlp] [--no_use_cuda_fp16] [--desc_act] [--disable_exllama]
+                 [--disable_exllamav2] [--wbits WBITS] [--groupsize GROUPSIZE] [--model-menu] [--multimodal-pipeline MULTIMODAL_PIPELINE]
 
 Text generation web UI
 
@@ -210,7 +210,6 @@ Basic settings:
   --lora LORA [LORA ...]                         The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.
   --model-dir MODEL_DIR                          Path to directory with all the models.
   --lora-dir LORA_DIR                            Path to directory with all the loras.
-  --model-menu                                   Show a model menu in the terminal when the web UI is first launched.
   --settings SETTINGS                            Load the default interface settings from this yaml file. See settings-template.yaml for an example. If you create a file called settings.yaml, this
                                                  file will be loaded by default without the need to use the --settings flag.
   --extensions EXTENSIONS [EXTENSIONS ...]       The list of extensions to load. If you want to load more than one extension, write the names separated by spaces.
@@ -218,8 +217,8 @@ Basic settings:
   --idle-timeout IDLE_TIMEOUT                    Unload model after this many minutes of inactivity. It will be automatically reloaded when you try to use it again.
 
 Model loader:
-  --loader LOADER                                Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav2_HF, ExLlamav2,
-                                                 HQQ, TensorRT-LLM.
+  --loader LOADER                                Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav3_HF, ExLlamav2_HF,
+                                                 ExLlamav2, HQQ, TensorRT-LLM.
 
 Transformers/Accelerate:
   --cpu                                          Use the CPU to generate text. Warning: Training on CPU is extremely slow.
@@ -319,9 +318,6 @@ API:
   --api-enable-ipv6                              Enable IPv6 for the API
   --api-disable-ipv4                             Disable IPv4 for the API
   --nowebui                                      Do not launch the Gradio UI. Useful for launching the API in standalone mode.
-
-Multimodal:
-  --multimodal-pipeline MULTIMODAL_PIPELINE      The multimodal pipeline to use. Examples: llava-7b, llava-13b.
 ```
 
 

From 454366f93ed4767a77fbc5c592646800415f23ed Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 10 Apr 2025 18:33:29 -0700
Subject: [PATCH 0280/1701] Change the ExLlamaV3 wheel version to 0.0.1a1

---
 requirements.txt               | 4 ++--
 requirements_apple_intel.txt   | 2 +-
 requirements_apple_silicon.txt | 2 +-
 requirements_noavx2.txt        | 4 ++--
 4 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index b9b4ea7aff..84d90880f8 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -44,8 +44,8 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu124-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 
 # CUDA wheels
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1/exllamav3-0.0.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1/exllamav3-0.0.1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a1/exllamav3-0.0.1a1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a1/exllamav3-0.0.1a1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index eba21ec2ce..fce0134107 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -33,5 +33,5 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1/exllamav3-0.0.1-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a1/exllamav3-0.0.1a1-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 2048c99bde..0985bb5fa3 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -34,5 +34,5 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1/exllamav3-0.0.1-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a1/exllamav3-0.0.1a1-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 60b71ac1d9..7f6ddb52cf 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -44,8 +44,8 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu124avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 
 # CUDA wheels
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1/exllamav3-0.0.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1/exllamav3-0.0.1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a1/exllamav3-0.0.1a1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a1/exllamav3-0.0.1a1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 682c78ea429721b65b2493a13c19ea7e2fdfa7e8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 11 Apr 2025 20:58:00 -0700
Subject: [PATCH 0281/1701] Add back detection of GPTQ models (closes #6841)

---
 modules/models_settings.py | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 51994e2375..b83544d45b 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -40,7 +40,12 @@ def get_model_metadata(model):
         hf_metadata = None
 
     if 'loader' not in model_settings:
-        model_settings['loader'] = infer_loader(model, model_settings)
+        quant_method = None if hf_metadata is None else hf_metadata.get("quantization_config", {}).get("quant_method", None)
+        model_settings['loader'] = infer_loader(
+            model,
+            model_settings,
+            hf_quant_method=quant_method
+        )
 
     # GGUF metadata
     if model_settings['loader'] in ['llama.cpp', 'llamacpp_HF']:
@@ -154,7 +159,7 @@ def get_model_metadata(model):
     return model_settings
 
 
-def infer_loader(model_name, model_settings):
+def infer_loader(model_name, model_settings, hf_quant_method=None):
     path_to_model = Path(f'{shared.args.model_dir}/{model_name}')
     if not path_to_model.exists():
         loader = None
@@ -164,6 +169,10 @@ def infer_loader(model_name, model_settings):
         loader = 'llama.cpp'
     elif re.match(r'.*\.gguf', model_name.lower()):
         loader = 'llama.cpp'
+    elif hf_quant_method == 'exl3':
+        loader = 'ExLlamav3_HF'
+    elif hf_quant_method in ['exl2', 'gptq']:
+        loader = 'ExLlamav2_HF'
     elif re.match(r'.*exl3', model_name.lower()):
         loader = 'ExLlamav3_HF'
     elif re.match(r'.*exl2', model_name.lower()):

From 038a01258136ea305b9ed56bc5967ebf2ea2265c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 11 Apr 2025 21:12:32 -0700
Subject: [PATCH 0282/1701] Installer: Remove .installer_state.json on
 reinstalling

---
 one_click.py | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/one_click.py b/one_click.py
index 9f46a2df83..99bdc41e80 100644
--- a/one_click.py
+++ b/one_click.py
@@ -26,6 +26,7 @@
 # Environment
 script_dir = os.getcwd()
 conda_env_path = os.path.join(script_dir, "installer_files", "env")
+state_file = '.installer_state.json'
 
 # Command-line flags
 cmd_flags_path = os.path.join(script_dir, "CMD_FLAGS.txt")
@@ -238,6 +239,9 @@ def get_user_choice(question, options_dict):
 
 
 def install_webui():
+    if os.path.isfile(state_file):
+        os.remove(state_file)
+
     # Ask the user for the GPU vendor
     if "GPU_CHOICE" in os.environ:
         choice = os.environ["GPU_CHOICE"].upper()
@@ -372,7 +376,6 @@ def update_requirements(initial_installation=False, pull=True):
         requirements_file = "requirements" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
 
     # Load state from JSON file
-    state_file = '.installer_state.json'
     current_commit = get_current_commit()
     wheels_changed = False
     if os.path.exists(state_file):

From 38dc09dca56830650d80d6f312283eb4fe2b9799 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 15 Apr 2025 09:50:36 -0700
Subject: [PATCH 0283/1701] Bump exllamav3 to the latest commit

---
 requirements.txt               | 4 ++--
 requirements_apple_intel.txt   | 2 +-
 requirements_apple_silicon.txt | 2 +-
 requirements_noavx2.txt        | 4 ++--
 4 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 84d90880f8..5485352871 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -44,8 +44,8 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu124-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 
 # CUDA wheels
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a1/exllamav3-0.0.1a1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a1/exllamav3-0.0.1a1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index fce0134107..a9e88a258d 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -33,5 +33,5 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a1/exllamav3-0.0.1a1-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 0985bb5fa3..4c64c03017 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -34,5 +34,5 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a1/exllamav3-0.0.1a1-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 7f6ddb52cf..0096c9ab2b 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -44,8 +44,8 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/te
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu124avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 
 # CUDA wheels
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a1/exllamav3-0.0.1a1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a1/exllamav3-0.0.1a1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 0ef1b8f8b4ac5d7b97fe08b3205ddef187fc341e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 17 Apr 2025 05:47:40 -0700
Subject: [PATCH 0284/1701] Use ExLlamaV2 (instead of the HF one) for EXL2
 models for now

It doesn't seem to have the "OverflowError" bug
---
 modules/loaders.py         | 1 -
 modules/models_settings.py | 4 ++--
 modules/ui_model_menu.py   | 1 -
 3 files changed, 2 insertions(+), 4 deletions(-)

diff --git a/modules/loaders.py b/modules/loaders.py
index 980a13e617..2fbba58850 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -110,7 +110,6 @@
         'no_flash_attn',
         'no_xformers',
         'no_sdpa',
-        'exllamav2_info',
     ],
     'HQQ': [
         'hqq_backend',
diff --git a/modules/models_settings.py b/modules/models_settings.py
index b83544d45b..b50840d714 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -172,11 +172,11 @@ def infer_loader(model_name, model_settings, hf_quant_method=None):
     elif hf_quant_method == 'exl3':
         loader = 'ExLlamav3_HF'
     elif hf_quant_method in ['exl2', 'gptq']:
-        loader = 'ExLlamav2_HF'
+        loader = 'ExLlamav2'
     elif re.match(r'.*exl3', model_name.lower()):
         loader = 'ExLlamav3_HF'
     elif re.match(r'.*exl2', model_name.lower()):
-        loader = 'ExLlamav2_HF'
+        loader = 'ExLlamav2'
     elif re.match(r'.*-hqq', model_name.lower()):
         return 'HQQ'
     else:
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 4fc1de0819..141869fabb 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -133,7 +133,6 @@ def create_ui():
                             shared.gradio['trust_remote_code'] = gr.Checkbox(label="trust-remote-code", value=shared.args.trust_remote_code, info='Set trust_remote_code=True while loading the tokenizer/model. To enable this option, start the web UI with the --trust-remote-code flag.', interactive=shared.args.trust_remote_code)
                             shared.gradio['no_use_fast'] = gr.Checkbox(label="no_use_fast", value=shared.args.no_use_fast, info='Set use_fast=False while loading the tokenizer.')
                             shared.gradio['llamacpp_HF_info'] = gr.Markdown("llamacpp_HF loads llama.cpp as a Transformers model. To use it, you need to place your GGUF in a subfolder of models/ with the necessary tokenizer files.\n\nYou can use the \"llamacpp_HF creator\" menu to do that automatically.")
-                            shared.gradio['exllamav2_info'] = gr.Markdown("ExLlamav2_HF is recommended over ExLlamav2 for better integration with extensions and more consistent sampling behavior across loaders.")
                             shared.gradio['tensorrt_llm_info'] = gr.Markdown('* TensorRT-LLM has to be installed manually in a separate Python 3.10 environment at the moment. For a guide, consult the description of [this PR](https://github.com/oobabooga/text-generation-webui/pull/5715). \n\n* `max_seq_len` is only used when `cpp-runner` is checked.\n\n* `cpp_runner` does not support streaming at the moment.')
 
             with gr.Column():

From 2c2d453c8ca6a7e9b1cba9d004fc944c1f2e4b67 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 17 Apr 2025 21:31:32 -0700
Subject: [PATCH 0285/1701] Revert "Use ExLlamaV2 (instead of the HF one) for
 EXL2 models for now"

This reverts commit 0ef1b8f8b4ac5d7b97fe08b3205ddef187fc341e.
---
 modules/loaders.py         | 1 +
 modules/models_settings.py | 4 ++--
 modules/ui_model_menu.py   | 1 +
 3 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/modules/loaders.py b/modules/loaders.py
index 2fbba58850..980a13e617 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -110,6 +110,7 @@
         'no_flash_attn',
         'no_xformers',
         'no_sdpa',
+        'exllamav2_info',
     ],
     'HQQ': [
         'hqq_backend',
diff --git a/modules/models_settings.py b/modules/models_settings.py
index b50840d714..b83544d45b 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -172,11 +172,11 @@ def infer_loader(model_name, model_settings, hf_quant_method=None):
     elif hf_quant_method == 'exl3':
         loader = 'ExLlamav3_HF'
     elif hf_quant_method in ['exl2', 'gptq']:
-        loader = 'ExLlamav2'
+        loader = 'ExLlamav2_HF'
     elif re.match(r'.*exl3', model_name.lower()):
         loader = 'ExLlamav3_HF'
     elif re.match(r'.*exl2', model_name.lower()):
-        loader = 'ExLlamav2'
+        loader = 'ExLlamav2_HF'
     elif re.match(r'.*-hqq', model_name.lower()):
         return 'HQQ'
     else:
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 141869fabb..4fc1de0819 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -133,6 +133,7 @@ def create_ui():
                             shared.gradio['trust_remote_code'] = gr.Checkbox(label="trust-remote-code", value=shared.args.trust_remote_code, info='Set trust_remote_code=True while loading the tokenizer/model. To enable this option, start the web UI with the --trust-remote-code flag.', interactive=shared.args.trust_remote_code)
                             shared.gradio['no_use_fast'] = gr.Checkbox(label="no_use_fast", value=shared.args.no_use_fast, info='Set use_fast=False while loading the tokenizer.')
                             shared.gradio['llamacpp_HF_info'] = gr.Markdown("llamacpp_HF loads llama.cpp as a Transformers model. To use it, you need to place your GGUF in a subfolder of models/ with the necessary tokenizer files.\n\nYou can use the \"llamacpp_HF creator\" menu to do that automatically.")
+                            shared.gradio['exllamav2_info'] = gr.Markdown("ExLlamav2_HF is recommended over ExLlamav2 for better integration with extensions and more consistent sampling behavior across loaders.")
                             shared.gradio['tensorrt_llm_info'] = gr.Markdown('* TensorRT-LLM has to be installed manually in a separate Python 3.10 environment at the moment. For a guide, consult the description of [this PR](https://github.com/oobabooga/text-generation-webui/pull/5715). \n\n* `max_seq_len` is only used when `cpp-runner` is checked.\n\n* `cpp_runner` does not support streaming at the moment.')
 
             with gr.Column():

From c9b3c9dfbfdbc3a9877728b98e9670b066ea89eb Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 17 Apr 2025 21:45:15 -0700
Subject: [PATCH 0286/1701] Attempt at solving the ExLlamaV2 issue

---
 modules/text_generation.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index eff6495ecf..cbe5b61b95 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -444,7 +444,7 @@ def generate_with_streaming(**kwargs):
 
                     new_content = get_reply_from_output_ids(output, state, starting_from=starting_from)
                     # check the partial unicode character
-                    if chr(0xfffd) in new_content:
+                    if new_content.endswith(chr(0xfffd)):
                         continue
 
                     cumulative_reply += new_content

From 5dfab7d363eddc6bcd0a00546754fbd5ac041113 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 17 Apr 2025 22:03:11 -0700
Subject: [PATCH 0287/1701] New attempt at solving the exl2 issue

---
 modules/text_generation.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index cbe5b61b95..d62441dfd1 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -263,6 +263,9 @@ def apply_stopping_strings(reply, all_stop_strings):
 
 
 def get_reply_from_output_ids(output_ids, state=None, starting_from=0):
+    if torch.cuda.is_available():
+        torch.cuda.synchronize()
+
     reply = decode(output_ids[starting_from:], state['skip_special_tokens'] if state else True)
 
     # Handle tokenizers that do not add the leading space for the first token

From ce9e2d94b169f657e3ba52fbb68e74d693ac590b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 17 Apr 2025 22:03:21 -0700
Subject: [PATCH 0288/1701] Revert "Attempt at solving the ExLlamaV2 issue"

This reverts commit c9b3c9dfbfdbc3a9877728b98e9670b066ea89eb.
---
 modules/text_generation.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index d62441dfd1..2869e03c21 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -447,7 +447,7 @@ def generate_with_streaming(**kwargs):
 
                     new_content = get_reply_from_output_ids(output, state, starting_from=starting_from)
                     # check the partial unicode character
-                    if new_content.endswith(chr(0xfffd)):
+                    if chr(0xfffd) in new_content:
                         continue
 
                     cumulative_reply += new_content

From d78abe480b06a977388798099c068ed3c1fed812 Mon Sep 17 00:00:00 2001
From: Googolplexed <65880807+Googolplexed0@users.noreply.github.com>
Date: Fri, 18 Apr 2025 01:53:59 -0400
Subject: [PATCH 0289/1701] Allow for model subfolder organization for GGUF
 files (#6686)

---------

Co-authored-by: oobabooga <112222186+oobabooga@users.noreply.github.com>
---
 modules/models_settings.py |  8 +++---
 modules/utils.py           | 58 ++++++++++++++++++++++++++++++++------
 2 files changed, 53 insertions(+), 13 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index b83544d45b..693c4dde6a 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -29,7 +29,7 @@ def get_model_metadata(model):
     # Get settings from models/config.yaml and models/config-user.yaml
     settings = shared.model_config
     for pat in settings:
-        if re.match(pat.lower(), model.lower()):
+        if re.match(pat.lower(), Path(model).name.lower()):
             for k in settings[pat]:
                 model_settings[k] = settings[pat][k]
 
@@ -148,7 +148,7 @@ def get_model_metadata(model):
     # Apply user settings from models/config-user.yaml
     settings = shared.user_config
     for pat in settings:
-        if re.match(pat.lower(), model.lower()):
+        if re.match(pat.lower(), Path(model).name.lower()):
             for k in settings[pat]:
                 model_settings[k] = settings[pat][k]
 
@@ -254,7 +254,7 @@ def save_model_settings(model, state):
         return
 
     user_config = shared.load_user_config()
-    model_regex = model + '$'  # For exact matches
+    model_regex = Path(model).name + '$'  # For exact matches
     if model_regex not in user_config:
         user_config[model_regex] = {}
 
@@ -281,7 +281,7 @@ def save_instruction_template(model, template):
         return
 
     user_config = shared.load_user_config()
-    model_regex = model + '$'  # For exact matches
+    model_regex = Path(model).name + '$'  # For exact matches
     if model_regex not in user_config:
         user_config[model_regex] = {}
 
diff --git a/modules/utils.py b/modules/utils.py
index f4333031fe..81cd085a7d 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -73,21 +73,61 @@ def natural_keys(text):
 
 
 def get_available_models():
-    model_list = []
-    for item in list(Path(f'{shared.args.model_dir}/').glob('*')):
-        if not item.name.endswith(('.txt', '-np', '.pt', '.json', '.yaml', '.py')) and 'llama-tokenizer' not in item.name:
-            model_list.append(item.name)
+    # Get all GGUF files
+    gguf_files = get_available_ggufs()
+
+    model_dir = Path(shared.args.model_dir)
+
+    # Find top-level directories containing GGUF files
+    dirs_with_gguf = set()
+    for gguf_path in gguf_files:
+        path = Path(gguf_path)
+        if path.parts:  # If in a subdirectory
+            dirs_with_gguf.add(path.parts[0])  # Add top-level directory
+
+    # Find directories with safetensors files directly under them
+    dirs_with_safetensors = set()
+    for item in os.listdir(model_dir):
+        item_path = model_dir / item
+        if item_path.is_dir():
+            # Check if there are safetensors files directly under this directory
+            if any(file.lower().endswith('.safetensors') for file in os.listdir(item_path) if (item_path / file).is_file()):
+                dirs_with_safetensors.add(item)
+
+    # Find valid model directories
+    model_dirs = []
 
-    return ['None'] + sorted(model_list, key=natural_keys)
+    for item in os.listdir(model_dir):
+        item_path = model_dir / item
+
+        # Skip if not a directory
+        if not item_path.is_dir():
+            continue
+
+        # Include directory if it either:
+        # 1. Doesn't contain GGUF files, OR
+        # 2. Contains both GGUF and safetensors files
+        if item not in dirs_with_gguf or item in dirs_with_safetensors:
+            model_dirs.append(item)
+
+    model_dirs = sorted(model_dirs, key=natural_keys)
+
+    # Combine all models
+    return ['None'] + gguf_files + model_dirs
 
 
 def get_available_ggufs():
     model_list = []
-    for item in Path(f'{shared.args.model_dir}/').glob('*'):
-        if item.is_file() and item.name.lower().endswith(".gguf"):
-            model_list.append(item.name)
+    model_dir = Path(shared.args.model_dir)
+
+    for dirpath, _, files in os.walk(model_dir, followlinks=True):
+        for file in files:
+            if file.lower().endswith(".gguf"):
+                model_path = Path(dirpath) / file
+                rel_path = model_path.relative_to(model_dir)
+                model_list.append(str(rel_path))
 
-    return ['None'] + sorted(model_list, key=natural_keys)
+    return sorted(model_list, key=natural_keys)
 
 
 def get_available_presets():

From 2fc58ad93580ec739bd3d9db9a1f11010af8686e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 17 Apr 2025 23:10:43 -0700
Subject: [PATCH 0290/1701] Consider files with .pt extension in the new model
 menu function

---
 modules/utils.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/utils.py b/modules/utils.py
index 81cd085a7d..f6be7541aa 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -91,7 +91,7 @@ def get_available_models():
         item_path = model_dir / item
         if item_path.is_dir():
             # Check if there are safetensors files directly under this directory
-            if any(file.lower().endswith('.safetensors') for file in os.listdir(item_path) if (item_path / file).is_file()):
+            if any(file.lower().endswith(('.safetensors', '.pt')) for file in os.listdir(item_path) if (item_path / file).is_file()):
                 dirs_with_safetensors.add(item)
 
     # Find valid model directories

From 5c2f8d828e97c4fd6e16979faa611586077e1072 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 05:42:38 -0700
Subject: [PATCH 0291/1701] Fix exllamav2 generating eos randomly after
 previous fix

---
 modules/exllamav2_hf.py    | 3 +++
 modules/text_generation.py | 3 ---
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/modules/exllamav2_hf.py b/modules/exllamav2_hf.py
index 62d1e0547c..6486e438fa 100644
--- a/modules/exllamav2_hf.py
+++ b/modules/exllamav2_hf.py
@@ -157,6 +157,9 @@ def __call__(self, *args, **kwargs):
         else:
             self.past_seq = seq_tensor
 
+        if torch.cuda.is_available():
+            torch.cuda.synchronize()
+
         loss = None
         if labels is not None:
             # Shift so that tokens < n predict n
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 2869e03c21..eff6495ecf 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -263,9 +263,6 @@ def apply_stopping_strings(reply, all_stop_strings):
 
 
 def get_reply_from_output_ids(output_ids, state=None, starting_from=0):
-    if torch.cuda.is_available():
-        torch.cuda.synchronize()
-
     reply = decode(output_ids[starting_from:], state['skip_special_tokens'] if state else True)
 
     # Handle tokenizers that do not add the leading space for the first token

From ae54d8faaa556bddf3e0805f4f3d04bb56c9be4b Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Fri, 18 Apr 2025 09:59:37 -0300
Subject: [PATCH 0292/1701] New llama.cpp loader (#6846)

---
 extensions/openai/logits.py        |   4 +-
 modules/cache_utils.py             | 115 ----------
 modules/evaluate.py                |   6 +-
 modules/llama_cpp_python_hijack.py | 165 --------------
 modules/llama_cpp_server.py        | 338 +++++++++++++++++++++++++++++
 modules/llamacpp_hf.py             | 220 -------------------
 modules/llamacpp_model.py          | 218 -------------------
 modules/loaders.py                 |  67 +-----
 modules/logits.py                  | 143 ++++++------
 modules/models.py                  |  39 +---
 modules/models_settings.py         |   4 +-
 modules/shared.py                  |  28 ++-
 modules/text_generation.py         |  21 +-
 modules/ui.py                      |   8 +-
 modules/ui_model_menu.py           |  43 +---
 requirements.txt                   |  14 +-
 requirements_amd.txt               |   6 +-
 requirements_amd_noavx2.txt        |   5 +-
 requirements_apple_intel.txt       |   4 +-
 requirements_apple_silicon.txt     |   6 +-
 requirements_cpu_only.txt          |   6 +-
 requirements_cpu_only_noavx2.txt   |   6 +-
 requirements_noavx2.txt            |  14 +-
 23 files changed, 476 insertions(+), 1004 deletions(-)
 delete mode 100644 modules/cache_utils.py
 delete mode 100644 modules/llama_cpp_python_hijack.py
 create mode 100644 modules/llama_cpp_server.py
 delete mode 100644 modules/llamacpp_hf.py
 delete mode 100644 modules/llamacpp_model.py

diff --git a/extensions/openai/logits.py b/extensions/openai/logits.py
index 357e70fa60..280612db5f 100644
--- a/extensions/openai/logits.py
+++ b/extensions/openai/logits.py
@@ -5,7 +5,5 @@
 def _get_next_logits(body):
     # Pre-process the input payload to simulate a real generation
     use_samplers = body['use_samplers']
-    state = process_parameters(body) if use_samplers else {}
-    state['stream'] = True
-
+    state = process_parameters(body)
     return get_next_logits(body['prompt'], state, use_samplers, "", top_logits=body['top_logits'], return_dict=True)
diff --git a/modules/cache_utils.py b/modules/cache_utils.py
deleted file mode 100644
index 0d1368a217..0000000000
--- a/modules/cache_utils.py
+++ /dev/null
@@ -1,115 +0,0 @@
-import torch
-from numba import njit
-
-from modules import shared
-
-
-def process_llamacpp_cache(model, new_sequence, past_sequence):
-    if len(past_sequence) == 0 or len(new_sequence) == 0:
-        return past_sequence
-
-    i1, i2, j1, j2 = find_longest_common_substring_indices(past_sequence, new_sequence)
-    overlap_length = i2 - i1 + 1
-
-    # Do StreamingLLM if i1 > 0 (ie the longest common subsequence is not a prefix)
-    # and the overlap length is sufficiently long.
-    if i1 > 0 and overlap_length > 0.2 * len(new_sequence):
-
-        new_sequence = torch.tensor(new_sequence)
-        past_sequence = torch.tensor(past_sequence)
-
-        prefix_length = find_prefix_length(past_sequence[:i1], new_sequence[:j1])
-        sink_length = max(prefix_length, shared.args.attention_sink_size)
-        removed_length = i1 - sink_length
-
-        if removed_length <= 0:
-            return past_sequence.tolist()
-
-        matching_prefix = past_sequence[:prefix_length]
-        removed_chunk = past_sequence[sink_length:i1]
-        overlapping_sequence = new_sequence[j1:j2 + 1]
-        added_chunk = new_sequence[j2 + 1:]
-
-        # print(past_sequence.tolist())
-        # print(new_sequence.tolist())
-
-        print()
-        print('MATCHING PREFIX=', repr(shared.tokenizer.decode(matching_prefix)))
-        print('ADDED CHUNK=', repr(shared.tokenizer.decode(added_chunk)))
-        print('REMOVED CHUNK=', repr(shared.tokenizer.decode(removed_chunk)))
-        print('REMOVED LENGTH=', removed_length)
-        print()
-
-        # Remove interval [sink_length, sink_length + removed_length) from the context
-        # Update model.n_tokens
-        model._ctx.kv_cache_seq_rm(0, sink_length, sink_length + removed_length)
-        model._ctx.kv_cache_seq_shift(0, sink_length + removed_length, -1, -removed_length)
-
-        new_sequence = new_sequence.tolist()
-        model.input_ids[:j2 + 1] = new_sequence[:j2 + 1]
-        model.n_tokens = j2 + 1
-
-        return new_sequence[:j2 + 1]
-    else:
-        return past_sequence
-
-
-def find_prefix_length(past_seq, seq_tensor):
-    '''
-    Given two torch tensors, finds the length of the longest
-    common prefix between the two.
-    '''
-    min_length = min(past_seq.shape[0], seq_tensor.shape[0])
-    indices = torch.nonzero(~torch.eq(past_seq[:min_length], seq_tensor[:min_length]))
-    if len(indices) > 0:
-        prefix_length = indices[0].item()
-    else:
-        prefix_length = min_length
-
-    return prefix_length
-
-
-@njit
-def find_longest_common_substring_indices(list1, list2):
-    '''
-    Given two lists, solves the Longest Common Substring problem.
-
-    It returns the indices where the substring starts and ends in
-    s1 and s2.
-
-    Example:
-
-    ir, jr, ir2, jr2 = find_longest_common_substring_indices(s1, s2)
-    print(s1[ir:jr + 1])
-    print(s2[ir2:jr2 + 1])
-
-    Adapted from
-    https://rosettacode.org/wiki/Longest_common_substring#Python
-    '''
-
-    len_list1, len_list2 = len(list1), len(list2)
-    start_index_list1, end_index_list1 = 0, -1
-    start_index_list2, end_index_list2 = 0, -1
-
-    # for index1 in tqdm(range(0, len_list1), desc="StreamingLLM prompt comparison", leave=False):
-    for index1 in range(0, len_list1):
-        try:
-            index2 = list2.index(list1[index1])
-        except:
-            continue
-
-        while index2 >= 0:
-            temp_index1, temp_index2 = index1, index2
-            while temp_index1 < len_list1 and temp_index2 < len_list2 and list2[temp_index2] == list1[temp_index1]:
-                if temp_index1 - index1 >= end_index_list1 - start_index_list1:
-                    start_index_list1, end_index_list1 = index1, temp_index1
-                    start_index_list2, end_index_list2 = index2, temp_index2
-
-                temp_index1 += 1
-                temp_index2 += 1
-            try:
-                index2 = list2.index(list1[index1], index2 + 1)
-            except:
-                break
-
-    return start_index_list1, end_index_list1, start_index_list2, end_index_list2
diff --git a/modules/evaluate.py b/modules/evaluate.py
index 35c72689af..dd7ef9a70b 100644
--- a/modules/evaluate.py
+++ b/modules/evaluate.py
@@ -40,17 +40,13 @@ def calculate_perplexity(models, input_dataset, stride, _max_length):
     '''
 
     if shared.args.loader == "llama.cpp":
-        logger.error("llamacpp_HF is required for perplexity evaluation with GGUF models. Please reload the model with llamacpp_HF instead of llama.cpp.")
+        logger.error("Perplexity evaluation is not implemented for the llama.cpp loader.")
         raise ValueError
 
     if shared.args.loader == "ExLlamav2":
         logger.error("ExLlamav2_HF is required for perplexity evaluation with EXL2 models. Please reload the model with ExLlamav2_HF instead of ExLlamav2.")
         raise ValueError
 
-    if shared.args.loader == "llamacpp_HF" and not shared.args.logits_all:
-        logger.error("--logits_all is required for perplexity evaluation with GGUF models. Please reload the model with that option set/checked.")
-        raise ValueError
-
     if not shared.args.no_use_fast:
         logger.warning("--no_use_fast is not set. If tokenizing the input dataset takes a long time, try reloading the model with that option set/checked.")
 
diff --git a/modules/llama_cpp_python_hijack.py b/modules/llama_cpp_python_hijack.py
deleted file mode 100644
index 8572cd8115..0000000000
--- a/modules/llama_cpp_python_hijack.py
+++ /dev/null
@@ -1,165 +0,0 @@
-import importlib
-import platform
-from typing import Sequence
-
-import numpy as np
-from tqdm import tqdm
-
-from modules import shared
-from modules.cache_utils import process_llamacpp_cache
-
-imported_module = None
-not_available_modules = set()
-
-
-def llama_cpp_lib():
-    global imported_module, not_available_modules
-
-    # Determine the platform
-    is_macos = platform.system() == 'Darwin'
-
-    # Define the library names based on the platform
-    if is_macos:
-        lib_names = [
-            (None, 'llama_cpp')
-        ]
-    else:
-        lib_names = [
-            ('cpu', 'llama_cpp'),
-            ('tensorcores', 'llama_cpp_cuda_tensorcores'),
-            (None, 'llama_cpp_cuda'),
-            (None, 'llama_cpp')
-        ]
-
-    for arg, lib_name in lib_names:
-        if lib_name in not_available_modules:
-            continue
-
-        should_import = (arg is None or getattr(shared.args, arg))
-
-        if should_import:
-            if imported_module and imported_module != lib_name:
-                # Conflict detected, raise an exception
-                raise Exception(f"Cannot import `{lib_name}` because `{imported_module}` is already imported. Switching to a different version of llama-cpp-python currently requires a server restart.")
-
-            try:
-                return_lib = importlib.import_module(lib_name)
-                imported_module = lib_name
-                monkey_patch_llama_cpp_python(return_lib)
-                return return_lib
-            except ImportError:
-                not_available_modules.add(lib_name)
-                continue
-
-    return None
-
-
-def eval_with_progress(self, tokens: Sequence[int]):
-    """
-    A copy of
-
-    https://github.com/abetlen/llama-cpp-python/blob/main/llama_cpp/llama.py
-
-    with tqdm to show prompt processing progress.
-    """
-    self._ctx.kv_cache_seq_rm(-1, self.n_tokens, -1)
-
-    if len(tokens) > self.n_batch:
-        progress_bar = tqdm(range(0, len(tokens), self.n_batch), desc="Prompt evaluation", leave=False)
-    else:
-        progress_bar = range(0, len(tokens), self.n_batch)
-
-    for i in progress_bar:
-        batch = tokens[i : min(len(tokens), i + self.n_batch)]
-        n_past = self.n_tokens
-        n_tokens = len(batch)
-        self._batch.set_batch(
-            batch=batch, n_past=n_past, logits_all=self.context_params.logits_all
-        )
-        self._ctx.decode(self._batch)
-        # Save tokens
-        self.input_ids[n_past : n_past + n_tokens] = batch
-        # Save logits
-        if self.context_params.logits_all:
-            rows = n_tokens
-            cols = self._n_vocab
-            logits = np.ctypeslib.as_array(
-                self._ctx.get_logits(), shape=(rows * cols,)
-            )
-            self.scores[n_past : n_past + n_tokens, :].reshape(-1)[::] = logits
-            self.last_updated_index = n_past + n_tokens - 1
-        else:
-            rows = 1
-            cols = self._n_vocab
-            logits = np.ctypeslib.as_array(
-                self._ctx.get_logits(), shape=(rows * cols,)
-            )
-            last_token_index = min(n_past + n_tokens - 1, self.scores.shape[0] - 1)
-            self.scores[last_token_index, :] = logits.reshape(-1)
-            self.last_updated_index = last_token_index
-        # Update n_tokens
-        self.n_tokens += n_tokens
-
-
-def monkey_patch_llama_cpp_python(lib):
-    if getattr(lib.Llama, '_is_patched', False):
-        # If the patch is already applied, do nothing
-        return
-
-    def my_generate(self, *args, **kwargs):
-        if shared.args.streaming_llm:
-            new_sequence = args[0]
-            past_sequence = self._input_ids
-
-            # Do the cache trimming for StreamingLLM
-            process_llamacpp_cache(self, new_sequence, past_sequence)
-
-        for output in self.original_generate(*args, **kwargs):
-            yield output
-
-    lib.Llama.eval = eval_with_progress
-    lib.Llama.original_generate = lib.Llama.generate
-    lib.Llama.generate = my_generate
-
-    # Also patch Jinja2ChatFormatter to handle loop controls
-    if hasattr(lib, 'llama_chat_format') and hasattr(lib.llama_chat_format, 'Jinja2ChatFormatter'):
-        Formatter = lib.llama_chat_format.Jinja2ChatFormatter
-
-        if not getattr(Formatter, '_is_patched', False):
-            def patched_init(self, *args, **kwargs):
-                # Extract parameters from args or kwargs
-                if args:
-                    self.template = args[0]
-                    self.eos_token = args[1] if len(args) > 1 else kwargs.get('eos_token')
-                    self.bos_token = args[2] if len(args) > 2 else kwargs.get('bos_token')
-                    self.add_generation_prompt = args[3] if len(args) > 3 else kwargs.get('add_generation_prompt', True)
-                    self.stop_token_ids = args[4] if len(args) > 4 else kwargs.get('stop_token_ids')
-                else:
-                    self.template = kwargs.get('template')
-                    self.eos_token = kwargs.get('eos_token')
-                    self.bos_token = kwargs.get('bos_token')
-                    self.add_generation_prompt = kwargs.get('add_generation_prompt', True)
-                    self.stop_token_ids = kwargs.get('stop_token_ids')
-
-                # Process stop tokens as in the original
-                self.stop_token_ids = (
-                    set(self.stop_token_ids) if self.stop_token_ids is not None else None
-                )
-
-                # Create environment with loopcontrols extension
-                import jinja2
-                from jinja2.ext import loopcontrols
-
-                self._environment = jinja2.sandbox.ImmutableSandboxedEnvironment(
-                    loader=jinja2.BaseLoader(),
-                    trim_blocks=True,
-                    lstrip_blocks=True,
-                    extensions=[loopcontrols]
-                ).from_string(self.template)
-
-            # Replace the original __init__ with our patched version
-            Formatter.__init__ = patched_init
-            Formatter._is_patched = True
-
-    # Set the flag to indicate that the patch has been applied
-    lib.Llama._is_patched = True
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
new file mode 100644
index 0000000000..983b506f52
--- /dev/null
+++ b/modules/llama_cpp_server.py
@@ -0,0 +1,338 @@
+import json
+import pprint
+import socket
+import subprocess
+import sys
+import threading
+import time
+
+import llama_cpp_binaries
+import requests
+
+from modules import shared
+from modules.logging_colors import logger
+
+llamacpp_valid_cache_types = {"fp16", "q8_0", "q4_0"}
+
+
+class LlamaServer:
+    def __init__(self, model_path, server_path=None):
+        """
+        Initialize and start a server for llama.cpp models.
+        """
+        self.model_path = model_path
+        self.server_path = server_path
+        self.port = self._find_available_port()
+        self.process = None
+        self.max_context_length = None
+        self.bos_token = ""
+
+        # Start the server
+        self._start_server()
+
+    def encode(self, text, add_bos_token=False, **kwargs):
+        if self.bos_token and text.startswith(self.bos_token):
+            add_bos_token = False
+
+        url = f"http://localhost:{self.port}/tokenize"
+        payload = {
+            "content": text,
+            "add_special": add_bos_token,
+        }
+
+        response = requests.post(url, json=payload)
+        result = response.json()
+        return result.get("tokens", [])
+
+    def decode(self, token_ids, **kwargs):
+        url = f"http://localhost:{self.port}/detokenize"
+        payload = {
+            "tokens": token_ids,
+        }
+
+        response = requests.post(url, json=payload)
+        result = response.json()
+        return result.get("content", "")
+
+    def prepare_payload(self, state):
+        # Prepare DRY
+        dry_sequence_breakers = state['dry_sequence_breakers']
+        if not dry_sequence_breakers.startswith("["):
+            dry_sequence_breakers = "[" + dry_sequence_breakers + "]"
+        dry_sequence_breakers = json.loads(dry_sequence_breakers)
+
+        # Prepare the sampler order
+        samplers = state["sampler_priority"]
+        samplers = samplers.split("\n") if isinstance(samplers, str) else samplers
+        penalty_found = False
+        filtered_samplers = []
+        for s in samplers:
+            if s.strip() in ["dry", "top_k", "typ_p", "top_p", "min_p", "xtc", "temperature"]:
+                filtered_samplers.append(s.strip())
+            elif not penalty_found and s.strip() == "repetition_penalty":
+                filtered_samplers.append("penalties")
+                penalty_found = True
+
+        samplers = filtered_samplers
+
+        # Move temperature to the end if temperature_last is true and temperature exists in the list
+        if state["temperature_last"] and "temperature" in samplers:
+            samplers.remove("temperature")
+            samplers.append("temperature")
+
+        payload = {
+            "temperature": state["temperature"] if not state["dynamic_temperature"] else (state["dynatemp_low"] + state["dynatemp_high"]) / 2,
+            "dynatemp_range": 0 if not state["dynamic_temperature"] else (state["dynatemp_high"] - state["dynatemp_low"]) / 2,
+            "dynatemp_exponent": state["dynatemp_exponent"],
+            "top_k": state["top_k"],
+            "top_p": state["top_p"],
+            "min_p": state["min_p"],
+            "tfs_z": state["tfs"],
+            "typical_p": state["typical_p"],
+            "repeat_penalty": state["repetition_penalty"],
+            "repeat_last_n": state["repetition_penalty_range"],
+            "presence_penalty": state["presence_penalty"],
+            "frequency_penalty": state["frequency_penalty"],
+            "dry_multiplier": state["dry_multiplier"],
+            "dry_base": state["dry_base"],
+            "dry_allowed_length": state["dry_allowed_length"],
+            "dry_penalty_last_n": state["repetition_penalty_range"],
+            "dry_sequence_breakers": dry_sequence_breakers,
+            "xtc_probability": state["xtc_probability"],
+            "xtc_threshold": state["xtc_threshold"],
+            "mirostat": state["mirostat_mode"],
+            "mirostat_tau": state["mirostat_tau"],
+            "mirostat_eta": state["mirostat_eta"],
+            "grammar": state["grammar_string"],
+            "seed": state["seed"],
+            "ignore_eos": state["ban_eos_token"],
+            "samplers": samplers,
+        }
+
+        if state['custom_token_bans']:
+            to_ban = [[int(token_id), False] for token_id in state['custom_token_bans'].split(',')]
+            payload["logit_bias"] = to_ban
+
+        return payload
+
+    def generate_with_streaming(
+        self,
+        prompt,
+        state,
+    ):
+        url = f"http://localhost:{self.port}/completion"
+        payload = self.prepare_payload(state)
+
+        token_ids = self.encode(prompt, add_bos_token=state["add_bos_token"])
+        if state['auto_max_new_tokens']:
+            max_new_tokens = state['truncation_length'] - len(token_ids)
+        else:
+            max_new_tokens = state['max_new_tokens']
+
+        payload.update({
+            "prompt": token_ids,
+            "n_predict": max_new_tokens,
+            "stream": True,
+        })
+
+        if shared.args.verbose:
+            logger.info("GENERATE_PARAMS=")
+            printable_payload = {k: v for k, v in payload.items() if k != "prompt"}
+            pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
+            print()
+
+        # Make a direct request with streaming enabled
+        response = requests.post(url, json=payload, stream=True)
+        response.raise_for_status()  # Raise an exception for HTTP errors
+
+        full_text = ""
+
+        # Process the streaming response
+        for line in response.iter_lines():
+            if shared.stop_everything:
+                break
+
+            if line:
+                try:
+                    # Check if the line starts with "data: " and remove it
+                    line_str = line.decode('utf-8')
+                    if line_str.startswith('data: '):
+                        line_str = line_str[6:]  # Remove the "data: " prefix
+
+                    # Parse the JSON data
+                    data = json.loads(line_str)
+
+                    # Extract the token content
+                    if 'content' in data:
+                        token_text = data['content']
+                        full_text += token_text
+                        yield full_text
+
+                    # Check if generation is complete
+                    if data.get('stop', False):
+                        break
+
+                except json.JSONDecodeError as e:
+                    # Log the error and the problematic line
+                    print(f"JSON decode error: {e}")
+                    print(f"Problematic line: {line}")
+                    continue
+
+    def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
+        """Get the logits/probabilities for the next token after a prompt"""
+        url = f"http://localhost:{self.port}/completion"
+
+        payload = self.prepare_payload(state)
+        payload.update({
+            "prompt": self.encode(prompt, add_bos_token=state["add_bos_token"]),
+            "n_predict": 0,
+            "logprobs": True,
+            "n_probs": n_probs,
+            "stream": False,
+            "post_sampling_probs": use_samplers,
+        })
+
+        if shared.args.verbose:
+            logger.info("GENERATE_PARAMS=")
+            printable_payload = {k: v for k, v in payload.items() if k != "prompt"}
+            pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
+            print()
+
+        response = requests.post(url, json=payload)
+        result = response.json()
+
+        if "completion_probabilities" in result:
+            if use_samplers:
+                return result["completion_probabilities"][0]["top_probs"]
+            else:
+                return result["completion_probabilities"][0]["top_logprobs"]
+        else:
+            raise Exception(f"Unexpected response format: 'completion_probabilities' not found in {result}")
+
+    def _get_max_context_length(self):
+        """Get and store the model's maximum context length."""
+        url = f"http://localhost:{self.port}/v1/models"
+        response = requests.get(url).json()
+
+        if "data" in response and len(response["data"]) > 0:
+            model_info = response["data"][0]
+            if "meta" in model_info and "n_vocab" in model_info["meta"]:
+                self.max_context_length = model_info["meta"]["n_vocab"]
+
+    def _get_bos_token(self):
+        """Get and store the model's BOS token."""
+        url = f"http://localhost:{self.port}/props"
+        response = requests.get(url).json()
+        if "bos_token" in response:
+            self.bos_token = response["bos_token"]
+
+    def _find_available_port(self):
+        """Find an available port by letting the OS assign one."""
+        with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
+            s.bind(('', 0))  # Bind to port 0 to get an available port
+            return s.getsockname()[1]
+
+    def _start_server(self):
+        """Start the llama.cpp server and wait until it's ready."""
+        # Determine the server path
+        if self.server_path is None:
+            self.server_path = llama_cpp_binaries.get_binary_path()
+
+        # Build the command
+        cmd = [
+            self.server_path,
+            "--model", self.model_path,
+            "--ctx-size", str(shared.args.n_ctx),
+            "--n-gpu-layers", str(shared.args.n_gpu_layers),
+            "--batch-size", str(shared.args.batch_size),
+            "--port", str(self.port),
+        ]
+
+        if shared.args.flash_attn:
+            cmd.append("--flash-attn")
+        if shared.args.threads > 0:
+            cmd += ["--threads", str(shared.args.threads)]
+        if shared.args.threads_batch > 0:
+            cmd += ["--threads-batch", str(shared.args.threads_batch)]
+        if shared.args.no_mmap:
+            cmd.append("--no-mmap")
+        if shared.args.mlock:
+            cmd.append("--mlock")
+        if shared.args.tensor_split:
+            cmd += ["--tensor-split", shared.args.tensor_split]
+        if shared.args.numa:
+            cmd += ["--numa", "distribute"]
+        if shared.args.no_kv_offload:
+            cmd.append("--no-kv-offload")
+        if shared.args.row_split:
+            cmd += ["--split-mode", "row"]
+        if shared.args.cache_type != "fp16" and shared.args.cache_type in llamacpp_valid_cache_types:
+            cmd += ["--cache-type-k", shared.args.cache_type, "--cache-type-v", shared.args.cache_type]
+        if shared.args.compress_pos_emb != 1:
+            cmd += ["--rope-freq-scale", str(1.0 / shared.args.compress_pos_emb)]
+
+        # Start the server with pipes for output
+        self.process = subprocess.Popen(
+            cmd,
+            stderr=subprocess.PIPE,
+            text=True,
+            bufsize=1
+        )
+
+        def filter_stderr():
+            for line in iter(self.process.stderr.readline, ''):
+                if not line.startswith(('srv ', 'slot ')) and not 'log_server_r: request: GET /health' in line:
+                    sys.stderr.write(line)
+                    sys.stderr.flush()
+
+        threading.Thread(target=filter_stderr, daemon=True).start()
+
+        # Wait for server to be healthy
+        health_url = f"http://localhost:{self.port}/health"
+        start_time = time.time()
+        timeout = 3600 * 8  # 8 hours
+        while time.time() - start_time < timeout:
+            # Check if process is still alive
+            if self.process.poll() is not None:
+                # Process has terminated
+                exit_code = self.process.poll()
+                raise RuntimeError(f"Server process terminated unexpectedly with exit code: {exit_code}")
+
+            try:
+                response = requests.get(health_url)
+                if response.status_code == 200:
+                    break
+            except:
+                pass
+
+            time.sleep(1)
+        else:
+            raise TimeoutError(f"Server health check timed out after {timeout} seconds")
+
+        # Server is now healthy, get model info
+        self._get_max_context_length()
+        self._get_bos_token()
+        return self.port
+
+    def __enter__(self):
+        """Support for context manager."""
+        return self
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        """Support for context manager."""
+        self.stop()
+
+    def __del__(self):
+        """Cleanup when the object is deleted."""
+        self.stop()
+
+    def stop(self):
+        """Stop the server process."""
+        if self.process:
+            self.process.terminate()
+            try:
+                self.process.wait(timeout=5)
+            except subprocess.TimeoutExpired:
+                self.process.kill()
+
+            self.process = None
diff --git a/modules/llamacpp_hf.py b/modules/llamacpp_hf.py
deleted file mode 100644
index b3761e0f53..0000000000
--- a/modules/llamacpp_hf.py
+++ /dev/null
@@ -1,220 +0,0 @@
-import os
-from pathlib import Path
-from typing import Any, Dict, Optional, Union
-
-import torch
-from torch.nn import CrossEntropyLoss
-from transformers import GenerationConfig, PretrainedConfig, PreTrainedModel
-from transformers.modeling_outputs import CausalLMOutputWithPast
-
-from modules import shared
-from modules.llama_cpp_python_hijack import llama_cpp_lib
-from modules.llamacpp_model import get_llamacpp_cache_type_for_string
-from modules.logging_colors import logger
-
-
-class LlamacppHF(PreTrainedModel):
-    def __init__(self, model, path):
-        super().__init__(PretrainedConfig())
-        self.model = model
-        self.generation_config = GenerationConfig()
-
-        self.past_seq = None
-        self.llamacpp_cache = {
-            'n_tokens': self.model.n_tokens,
-            'input_ids': self.model.input_ids,
-            'scores': self.model.scores,
-            'ctx': self.model._ctx.ctx
-        }
-
-        if shared.args.cfg_cache:
-            self.past_seq_negative = None
-            self.llamacpp_cache_negative = {
-                'n_tokens': self.model.n_tokens,
-                'input_ids': self.model.input_ids.copy(),
-                'scores': self.model.scores.copy(),
-                'ctx': llama_cpp_lib().llama_new_context_with_model(model.model, model.context_params)
-            }
-
-    def _validate_model_class(self):
-        pass
-
-    def _validate_model_kwargs(self, model_kwargs: Dict[str, Any]):
-        pass
-
-    def prepare_inputs_for_generation(self, input_ids, **kwargs):
-        return {'input_ids': input_ids, **kwargs}
-
-    def save_cache(self):
-        self.llamacpp_cache.update({
-            'n_tokens': self.model.n_tokens,
-            'input_ids': self.model.input_ids,
-            'scores': self.model.scores,
-            'ctx': self.model._ctx.ctx
-        })
-
-    def save_negative_cache(self):
-        self.llamacpp_cache_negative.update({
-            'n_tokens': self.model.n_tokens,
-            'input_ids': self.model.input_ids,
-            'scores': self.model.scores,
-            'ctx': self.model._ctx.ctx
-        })
-
-    def load_cache(self):
-        self.model.n_tokens = self.llamacpp_cache['n_tokens']
-        self.model.input_ids = self.llamacpp_cache['input_ids']
-        self.model.scores = self.llamacpp_cache['scores']
-        self.model._ctx.ctx = self.llamacpp_cache['ctx']
-
-    def load_negative_cache(self):
-        self.model.n_tokens = self.llamacpp_cache_negative['n_tokens']
-        self.model.input_ids = self.llamacpp_cache_negative['input_ids']
-        self.model.scores = self.llamacpp_cache_negative['scores']
-        self.model._ctx.ctx = self.llamacpp_cache_negative['ctx']
-
-    @property
-    def device(self) -> torch.device:
-        return torch.device(0)
-
-    def __call__(self, *args, **kwargs):
-        use_cache = kwargs.get('use_cache', True)
-        labels = kwargs.get('labels', None)
-        past_key_values = kwargs.get('past_key_values', None)
-
-        if len(args) > 0:
-            if not shared.args.cfg_cache:
-                logger.error("Please enable the cfg-cache option to use CFG with llamacpp_HF.")
-                return
-
-            input_ids = args[0]
-            is_negative = True
-            past_seq = self.past_seq_negative
-            self.load_negative_cache()
-        else:
-            input_ids = kwargs['input_ids']
-            is_negative = False
-            past_seq = self.past_seq
-            self.load_cache()
-
-        seq = input_ids[0].tolist()
-        if is_negative and past_key_values is not None:
-            seq = past_key_values + seq
-
-        seq_tensor = torch.tensor(seq)
-        reset = True
-
-        # Make the forward call. The prefix-match code has been adapted from
-        # https://github.com/abetlen/llama-cpp-python/commit/f4090a0bb2a2a25acfe28d31c82cc1aa273bedee
-        if labels is None:
-            if past_seq is not None:
-                min_length = min(past_seq.shape[0], seq_tensor.shape[0])
-                indices = torch.nonzero(~torch.eq(past_seq[:min_length], seq_tensor[:min_length]))
-                if len(indices) > 0:
-                    longest_prefix = indices[0].item()
-                else:
-                    longest_prefix = min_length
-
-                if longest_prefix > 0:
-                    reset = False
-                    self.model.n_tokens = longest_prefix
-                    if len(seq_tensor) - longest_prefix > 0:
-                        self.model.eval(seq[longest_prefix:])
-                    else:
-                        self.model.n_tokens -= 1
-                        self.model.eval([seq[-1]])
-
-            if reset:
-                self.model.reset()
-                self.model.eval(seq)
-
-            logits = torch.tensor(self.model.scores[self.model.last_updated_index, :]).view(1, 1, -1).to(input_ids.device)
-        else:
-            self.model.reset()
-            self.model.eval(seq)
-            logits = torch.tensor(self.model.eval_logits)
-            logits = logits.view(1, logits.shape[0], logits.shape[1]).to(input_ids.device)
-
-        if is_negative:
-            self.save_negative_cache()
-            self.past_seq_negative = seq_tensor
-        else:
-            self.save_cache()
-            self.past_seq = seq_tensor
-
-        loss = None
-        if labels is not None:
-            # Shift so that tokens < n predict n
-            shift_logits = logits[..., :-1, :].contiguous()
-            shift_labels = labels[..., 1:].contiguous()
-            # Flatten the tokens
-            loss_fct = CrossEntropyLoss()
-            shift_logits = shift_logits.view(-1, logits.shape[-1])
-            shift_labels = shift_labels.view(-1)
-            # Enable model parallelism
-            shift_labels = shift_labels.to(shift_logits.device)
-            loss = loss_fct(shift_logits, shift_labels)
-
-        return CausalLMOutputWithPast(logits=logits, past_key_values=seq if use_cache else None, loss=loss)
-
-    @classmethod
-    def from_pretrained(cls, pretrained_model_name_or_path: Optional[Union[str, os.PathLike]], *model_args, **kwargs):
-        assert len(model_args) == 0 and len(kwargs) == 0, "extra args is currently not supported"
-
-        if isinstance(pretrained_model_name_or_path, str):
-            pretrained_model_name_or_path = Path(pretrained_model_name_or_path)
-
-        path = Path(f'{shared.args.model_dir}') / Path(pretrained_model_name_or_path)
-        if path.is_file():
-            model_file = path
-        else:
-            model_file = sorted(path.glob('*.gguf'))[0]
-
-        logger.info(f"llama.cpp weights detected: {model_file}\n")
-
-        if shared.args.tensor_split is None or shared.args.tensor_split.strip() == '':
-            tensor_split_list = None
-        else:
-            tensor_split_list = [float(x) for x in shared.args.tensor_split.strip().split(",")]
-
-        params = {
-            'model_path': str(model_file),
-            'n_ctx': shared.args.n_ctx,
-            'n_threads': shared.args.threads or None,
-            'n_threads_batch': shared.args.threads_batch or None,
-            'n_batch': shared.args.n_batch,
-            'use_mmap': not shared.args.no_mmap,
-            'use_mlock': shared.args.mlock,
-            'mul_mat_q': not shared.args.no_mul_mat_q,
-            'numa': shared.args.numa,
-            'n_gpu_layers': shared.args.n_gpu_layers,
-            'rope_freq_base': shared.args.rope_freq_base,
-            'tensor_split': tensor_split_list,
-            'rope_freq_scale': 1.0 / shared.args.compress_pos_emb,
-            'logits_all': shared.args.logits_all,
-            'offload_kqv': not shared.args.no_offload_kqv,
-            'split_mode': 1 if not shared.args.row_split else 2,
-            'flash_attn': shared.args.flash_attn
-        }
-
-        if shared.args.cache_type != 'fp16':
-            params["type_k"] = get_llamacpp_cache_type_for_string(shared.args.cache_type)
-            params["type_v"] = get_llamacpp_cache_type_for_string(shared.args.cache_type)
-
-        Llama = llama_cpp_lib().Llama
-        try:
-            model = Llama(**params)
-        except Exception as e:
-            error_message = (
-                f"Failed loading the model. **This usually happens due to lack of memory**. Try these steps:\n"
-                f"1. Reduce the context length `n_ctx` (currently {shared.args.n_ctx})."
-                f"{' Try a lower value like 4096.' if shared.args.n_ctx > 4096 else '.'}"
-                "\n"
-                f"2. Lower the `n-gpu-layers` value (currently {shared.args.n_gpu_layers})."
-            )
-
-            raise type(e)(error_message) from e
-
-        model.last_updated_index = -1
-
-        return LlamacppHF(model, model_file)
diff --git a/modules/llamacpp_model.py b/modules/llamacpp_model.py
deleted file mode 100644
index db25c66c51..0000000000
--- a/modules/llamacpp_model.py
+++ /dev/null
@@ -1,218 +0,0 @@
-import re
-from functools import partial
-
-import numpy as np
-import torch
-
-from modules import shared
-from modules.callbacks import Iteratorize
-from modules.llama_cpp_python_hijack import llama_cpp_lib
-from modules.logging_colors import logger
-from modules.text_generation import get_max_prompt_length
-
-llamacpp_quant_mapping = {
-    'f32': 0,
-    'fp16': 1,
-    'q4_0': 2,
-    'q4_1': 3,
-    'q5_0': 6,
-    'q5_1': 7,
-    'q8_0': 8,
-    'q8_1': 9,
-    'q2_k': 10,
-    'q3_k': 11,
-    'q4_k': 12,
-    'q5_k': 13,
-    'q6_k': 14,
-    'q8_k': 15,
-    'iq4_nl': 20,
-    'bf16': 30,
-}
-
-llamacpp_valid_cache_types = {'fp16', 'q8_0', 'q4_0'}
-
-
-def get_llamacpp_cache_type_for_string(quant_type: str):
-    quant_type = quant_type.lower()
-    if quant_type in llamacpp_valid_cache_types:
-        return llamacpp_quant_mapping[quant_type]
-    else:
-        raise ValueError(f"Invalid cache type for llama.cpp: {quant_type}. Valid options are: fp16, q8_0, q4_0.")
-
-
-def ban_eos_logits_processor(eos_token, input_ids, logits):
-    logits[eos_token] = -float('inf')
-    return logits
-
-
-def custom_token_ban_logits_processor(token_ids, input_ids, logits):
-    for token_id in token_ids:
-        logits[token_id] = -float('inf')
-
-    return logits
-
-
-class LlamaCppModel:
-    def __init__(self):
-        self.initialized = False
-        self.grammar_string = ''
-        self.grammar = None
-
-    def __del__(self):
-        del self.model
-
-    @classmethod
-    def from_pretrained(self, path):
-
-        Llama = llama_cpp_lib().Llama
-        LlamaCache = llama_cpp_lib().LlamaCache
-
-        result = self()
-        cache_capacity = 0
-        if shared.args.cache_capacity is not None:
-            if 'GiB' in shared.args.cache_capacity:
-                cache_capacity = int(re.sub('[a-zA-Z]', '', shared.args.cache_capacity)) * 1000 * 1000 * 1000
-            elif 'MiB' in shared.args.cache_capacity:
-                cache_capacity = int(re.sub('[a-zA-Z]', '', shared.args.cache_capacity)) * 1000 * 1000
-            else:
-                cache_capacity = int(shared.args.cache_capacity)
-
-        if cache_capacity > 0:
-            logger.info("Cache capacity is " + str(cache_capacity) + " bytes")
-
-        if shared.args.tensor_split is None or shared.args.tensor_split.strip() == '':
-            tensor_split_list = None
-        else:
-            tensor_split_list = [float(x) for x in shared.args.tensor_split.strip().split(",")]
-
-        params = {
-            'model_path': str(path),
-            'n_ctx': shared.args.n_ctx,
-            'n_threads': shared.args.threads or None,
-            'n_threads_batch': shared.args.threads_batch or None,
-            'n_batch': shared.args.n_batch,
-            'use_mmap': not shared.args.no_mmap,
-            'use_mlock': shared.args.mlock,
-            'mul_mat_q': not shared.args.no_mul_mat_q,
-            'numa': shared.args.numa,
-            'n_gpu_layers': shared.args.n_gpu_layers,
-            'rope_freq_base': shared.args.rope_freq_base,
-            'tensor_split': tensor_split_list,
-            'rope_freq_scale': 1.0 / shared.args.compress_pos_emb,
-            'offload_kqv': not shared.args.no_offload_kqv,
-            'split_mode': 1 if not shared.args.row_split else 2,
-            'flash_attn': shared.args.flash_attn
-        }
-
-        if shared.args.cache_type != 'fp16':
-            params["type_k"] = get_llamacpp_cache_type_for_string(shared.args.cache_type)
-            params["type_v"] = get_llamacpp_cache_type_for_string(shared.args.cache_type)
-
-        try:
-            result.model = Llama(**params)
-        except Exception as e:
-            error_message = (
-                f"Failed loading the model. **This usually happens due to lack of memory**. Try these steps:\n"
-                f"1. Reduce the context length `n_ctx` (currently {shared.args.n_ctx})."
-                f"{' Try a lower value like 4096.' if shared.args.n_ctx > 4096 else '.'}"
-                "\n"
-                f"2. Lower the `n-gpu-layers` value (currently {shared.args.n_gpu_layers})."
-            )
-
-            raise type(e)(error_message) from e
-
-        if cache_capacity > 0:
-            result.model.set_cache(LlamaCache(capacity_bytes=cache_capacity))
-
-        # This is ugly, but the model and the tokenizer are the same object in this library.
-        return result, result
-
-    def encode(self, string):
-        if type(string) is str:
-            string = string.encode()
-
-        return self.model.tokenize(string)
-
-    def decode(self, ids, **kwargs):
-        detokenized = self.model.detokenize(ids)
-        try:
-            # Attempt strict UTF-8 decoding first
-            return detokenized.decode('utf-8', 'strict')
-        except UnicodeDecodeError as e:
-            # Log the error and fall back to UTF-8 with replacement
-            logger.warning(f"Invalid UTF-8 in detokenized output. Using replacement characters.\n{e}")
-            return detokenized.decode('utf-8', 'replace')
-
-    def get_logits(self, tokens):
-        self.model.reset()
-        self.model.eval(tokens)
-        logits = self.model._scores
-        logits = np.expand_dims(logits, 0)  # batch dim is expected
-        return torch.tensor(logits, dtype=torch.float32)
-
-    def load_grammar(self, string):
-        if string != self.grammar_string:
-            self.grammar_string = string
-            if string.strip() != '':
-                self.grammar = llama_cpp_lib().LlamaGrammar.from_string(string)
-            else:
-                self.grammar = None
-
-    def generate(self, prompt, state, callback=None):
-        LogitsProcessorList = llama_cpp_lib().LogitsProcessorList
-        prompt = prompt if type(prompt) is str else prompt.decode()
-
-        # Handle truncation
-        prompt = self.encode(prompt)
-        prompt = prompt[-get_max_prompt_length(state):]
-        prompt = self.decode(prompt)
-
-        self.load_grammar(state['grammar_string'])
-        logit_processors = LogitsProcessorList()
-        if state['ban_eos_token']:
-            logit_processors.append(partial(ban_eos_logits_processor, self.model.token_eos()))
-
-        if state['custom_token_bans']:
-            to_ban = [int(x) for x in state['custom_token_bans'].split(',')]
-            if len(to_ban) > 0:
-                logit_processors.append(partial(custom_token_ban_logits_processor, to_ban))
-
-        completion_chunks = self.model.create_completion(
-            prompt=prompt,
-            max_tokens=state['max_new_tokens'],
-            temperature=state['temperature'],
-            top_p=state['top_p'] if state['top_p'] < 1 else 0.999,
-            min_p=state['min_p'],
-            typical_p=state['typical_p'],
-            frequency_penalty=state['frequency_penalty'],
-            presence_penalty=state['presence_penalty'],
-            repeat_penalty=state['repetition_penalty'],
-            top_k=state['top_k'],
-            stream=True,
-            seed=int(state['seed']) if state['seed'] != -1 else None,
-            tfs_z=state['tfs'],
-            mirostat_mode=int(state['mirostat_mode']),
-            mirostat_tau=state['mirostat_tau'],
-            mirostat_eta=state['mirostat_eta'],
-            logits_processor=logit_processors,
-            grammar=self.grammar
-        )
-
-        output = ""
-        for completion_chunk in completion_chunks:
-            if shared.stop_everything:
-                break
-
-            text = completion_chunk['choices'][0]['text']
-            output += text
-            if callback:
-                callback(text)
-
-        return output
-
-    def generate_with_streaming(self, *args, **kwargs):
-        with Iteratorize(self.generate, args, kwargs, callback=None) as generator:
-            reply = ''
-            for token in generator:
-                reply += token
-                yield reply
diff --git a/modules/loaders.py b/modules/loaders.py
index 980a13e617..3060406d8a 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -30,50 +30,19 @@
         'n_gpu_layers',
         'threads',
         'threads_batch',
-        'n_batch',
+        'batch_size',
         'n_ctx',
         'cache_type',
         'tensor_split',
         'rope_freq_base',
         'compress_pos_emb',
-        'attention_sink_size',
-        'tensorcores',
         'flash_attn',
-        'streaming_llm',
-        'cpu',
-        'row_split',
-        'no_offload_kqv',
-        'no_mul_mat_q',
-        'no_mmap',
-        'mlock',
-        'numa',
-    ],
-    'llamacpp_HF': [
-        'n_gpu_layers',
-        'threads',
-        'threads_batch',
-        'n_batch',
-        'n_ctx',
-        'cache_type',
-        'tensor_split',
-        'rope_freq_base',
-        'compress_pos_emb',
-        'attention_sink_size',
-        'tensorcores',
-        'flash_attn',
-        'streaming_llm',
-        'cpu',
         'row_split',
-        'no_offload_kqv',
+        'no_kv_offload',
         'no_mul_mat_q',
         'no_mmap',
         'mlock',
         'numa',
-        'cfg_cache',
-        'logits_all',
-        'trust_remote_code',
-        'no_use_fast',
-        'llamacpp_HF_info',
     ],
     'ExLlamav3_HF': [
         'max_seq_len',
@@ -307,66 +276,34 @@ def transformers_samplers():
         'dry_sequence_breakers',
     },
     'llama.cpp': {
-        'temperature',
-        'min_p',
-        'top_p',
-        'top_k',
-        'typical_p',
-        'tfs',
-        'repetition_penalty',
-        'frequency_penalty',
-        'presence_penalty',
-        'mirostat_mode',
-        'mirostat_tau',
-        'mirostat_eta',
-        'ban_eos_token',
-        'seed',
-        'custom_token_bans',
-        'grammar_string',
-        'grammar_file_row',
-    },
-    'llamacpp_HF': {
         'temperature',
         'dynatemp_low',
         'dynatemp_high',
         'dynatemp_exponent',
-        'smoothing_factor',
-        'smoothing_curve',
         'min_p',
         'top_p',
         'top_k',
         'typical_p',
         'xtc_threshold',
         'xtc_probability',
-        'epsilon_cutoff',
-        'eta_cutoff',
         'tfs',
-        'top_a',
-        'top_n_sigma',
         'dry_multiplier',
         'dry_allowed_length',
         'dry_base',
         'repetition_penalty',
         'frequency_penalty',
         'presence_penalty',
-        'encoder_repetition_penalty',
-        'no_repeat_ngram_size',
         'repetition_penalty_range',
-        'guidance_scale',
         'mirostat_mode',
         'mirostat_tau',
         'mirostat_eta',
-        'do_sample',
         'dynamic_temperature',
         'temperature_last',
         'auto_max_new_tokens',
         'ban_eos_token',
         'add_bos_token',
-        'skip_special_tokens',
         'seed',
         'sampler_priority',
-        'custom_token_bans',
-        'negative_prompt',
         'dry_sequence_breakers',
         'grammar_string',
         'grammar_file_row',
diff --git a/modules/logits.py b/modules/logits.py
index f8a1e80c28..c5d9112d00 100644
--- a/modules/logits.py
+++ b/modules/logits.py
@@ -1,6 +1,7 @@
 import time
 import traceback
 
+import numpy as np
 import torch
 
 from modules import models, sampler_hijack, shared
@@ -38,70 +39,86 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
         return 'Error: No model is loaded1 Select one in the Model tab.', previous
 
     is_non_hf_exllamav2 = shared.model.__class__.__name__ == 'Exllamav2Model'
-    is_non_hf_llamacpp = shared.model.__class__.__name__ == 'LlamaCppModel'
-
-    if use_samplers:
-        if any([is_non_hf_exllamav2, is_non_hf_llamacpp]):
-            logger.error("Sampler hijacking is not supported non-Huggingface loaders.")
-            # sampling is all done in c for exllama, so it is really hard to hijack
-            # it should be possible to hijack llamacpp sampler by hijacking all their sampling methods,
-            # but it is not implemented yet
-            return 'Error: Sampler hijacking is not supported non-Huggingface loaders. Please disable the "Use samplers" option.', previous
-
-        state['max_new_tokens'] = 1
-        state['auto_max_new_tokens'] = False
-        for _ in generate_reply(prompt, state):
-            pass
-
-        scores = sampler_hijack.global_scores[-1]
-    else:
-        if is_non_hf_exllamav2:
-            device = get_device()
-            tokens = shared.tokenizer.encode(prompt)
-            if device:
-                tokens = tokens.to(device)
-
-            scores = shared.model.get_logits(tokens)[-1][-1]
-        elif is_non_hf_llamacpp:
-            tokens = shared.tokenizer.encode(prompt)
-            scores = shared.model.get_logits(tokens)[-1][-1]
+    is_llamacpp = shared.model.__class__.__name__ == 'LlamaServer'
+
+    if is_llamacpp:
+        logprobs = shared.model.get_logits(prompt, state, n_probs=top_logits, use_samplers=use_samplers)
+        if return_dict:
+            output = {}
+            for entry in logprobs:
+                token = repr(entry['token'])
+                prob = entry['prob'] if use_samplers else np.exp(entry['logprob'])
+                output[token] = prob
+
+            return output
         else:
-            device = get_device()
-            tokens = shared.tokenizer.encode(prompt, return_tensors='pt')
-            if device:
-                tokens = tokens.to(device)
-
-            output = shared.model(input_ids=tokens)
-            scores = output['logits'][-1][-1]
+            output = ''
+            for entry in logprobs:
+                token = repr(entry['token'])
+                prob = entry['prob'] if use_samplers else np.exp(entry['logprob'])
+                output += f"{prob:.5f}  -  {token}\n"
 
-    probs = torch.softmax(scores, dim=-1, dtype=torch.float)
-    topk_values, topk_indices = torch.topk(probs, k=top_logits, largest=True, sorted=True)
-    if is_non_hf_llamacpp:
-        topk_indices = [i.expand((1, 1)) for i in topk_indices]
-
-    if hasattr(shared.tokenizer, 'convert_ids_to_tokens'):
-        tokens = [shared.tokenizer.convert_ids_to_tokens(int(i)) for i in topk_indices]
-    else:
-        tokens = [shared.tokenizer.decode(i) for i in topk_indices]
-
-    if return_dict:
-        topk_values = [float(i) for i in topk_values]
-        output = {}
-        for row in list(zip(topk_values, tokens)):
-            key = row[1]
-            if isinstance(key, bytes):
-                try:
-                    key = key.decode()
-                except:
-                    key = key.decode('latin')
-
-            output[key] = row[0]
-
-        return output
+            return output, previous
     else:
-        topk_values = [f"{float(i):.5f}" for i in topk_values]
-        output = ''
-        for row in list(zip(topk_values, tokens)):
-            output += f"{row[0]}  -  {repr(row[1])}\n"
+        if not use_samplers:
+            state = {'stream': True}
+
+        if use_samplers:
+            if is_non_hf_exllamav2:
+                logger.error("Sampler hijacking is not supported non-Huggingface loaders.")
+                # sampling is all done in c for exllama, so it is really hard to hijack
+                # it should be possible to hijack llamacpp sampler by hijacking all their sampling methods,
+                # but it is not implemented yet
+                return 'Error: Sampler hijacking is not supported non-Huggingface loaders. Please disable the "Use samplers" option.', previous
+
+            state['max_new_tokens'] = 1
+            state['auto_max_new_tokens'] = False
+            for _ in generate_reply(prompt, state):
+                pass
+
+            scores = sampler_hijack.global_scores[-1]
+        else:
+            if is_non_hf_exllamav2:
+                device = get_device()
+                tokens = shared.tokenizer.encode(prompt)
+                if device:
+                    tokens = tokens.to(device)
+
+                scores = shared.model.get_logits(tokens)[-1][-1]
+            else:
+                device = get_device()
+                tokens = shared.tokenizer.encode(prompt, return_tensors='pt')
+                if device:
+                    tokens = tokens.to(device)
+
+                output = shared.model(input_ids=tokens)
+                scores = output['logits'][-1][-1]
+
+        probs = torch.softmax(scores, dim=-1, dtype=torch.float)
+        topk_values, topk_indices = torch.topk(probs, k=top_logits, largest=True, sorted=True)
+        if hasattr(shared.tokenizer, 'convert_ids_to_tokens'):
+            tokens = [shared.tokenizer.convert_ids_to_tokens(int(i)) for i in topk_indices]
+        else:
+            tokens = [shared.tokenizer.decode(i) for i in topk_indices]
+
+        if return_dict:
+            topk_values = [float(i) for i in topk_values]
+            output = {}
+            for row in list(zip(topk_values, tokens)):
+                key = row[1]
+                if isinstance(key, bytes):
+                    try:
+                        key = key.decode()
+                    except:
+                        key = key.decode('latin')
+
+                output[key] = row[0]
+
+            return output
+        else:
+            topk_values = [f"{float(i):.5f}" for i in topk_values]
+            output = ''
+            for row in list(zip(topk_values, tokens)):
+                output += f"{row[0]}  -  {repr(row[1])}\n"
 
-        return output, previous
+            return output, previous
diff --git a/modules/models.py b/modules/models.py
index 288bc1b66b..48d92bd539 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -67,8 +67,7 @@ def load_model(model_name, loader=None):
     shared.model_name = model_name
     load_func_map = {
         'Transformers': huggingface_loader,
-        'llama.cpp': llamacpp_loader,
-        'llamacpp_HF': llamacpp_HF_loader,
+        'llama.cpp': llama_cpp_server_loader,
         'ExLlamav3_HF': ExLlamav3_HF_loader,
         'ExLlamav2_HF': ExLlamav2_HF_loader,
         'ExLlamav2': ExLlamav2_loader,
@@ -101,7 +100,7 @@ def load_model(model_name, loader=None):
     shared.settings.update({k: v for k, v in metadata.items() if k in shared.settings})
     if loader.lower().startswith('exllama') or loader.lower().startswith('tensorrt'):
         shared.settings['truncation_length'] = shared.args.max_seq_len
-    elif loader in ['llama.cpp', 'llamacpp_HF']:
+    elif loader == 'llama.cpp':
         shared.settings['truncation_length'] = shared.args.n_ctx
 
     logger.info(f"Loaded \"{model_name}\" in {(time.time()-t0):.2f} seconds.")
@@ -268,8 +267,8 @@ def huggingface_loader(model_name):
     return model
 
 
-def llamacpp_loader(model_name):
-    from modules.llamacpp_model import LlamaCppModel
+def llama_cpp_server_loader(model_name):
+    from modules.llama_cpp_server import LlamaServer
 
     path = Path(f'{shared.args.model_dir}/{model_name}')
     if path.is_file():
@@ -278,31 +277,11 @@ def llamacpp_loader(model_name):
         model_file = sorted(Path(f'{shared.args.model_dir}/{model_name}').glob('*.gguf'))[0]
 
     logger.info(f"llama.cpp weights detected: \"{model_file}\"")
-    model, tokenizer = LlamaCppModel.from_pretrained(model_file)
-    return model, tokenizer
-
-
-def llamacpp_HF_loader(model_name):
-    from modules.llamacpp_hf import LlamacppHF
-
-    if shared.args.tokenizer_dir:
-        logger.info(f'Using tokenizer from: \"{shared.args.tokenizer_dir}\"')
-    else:
-        path = Path(f'{shared.args.model_dir}/{model_name}')
-        # Check if a HF tokenizer is available for the model
-        if all((path / file).exists() for file in ['tokenizer_config.json']):
-            logger.info(f'Using tokenizer from: \"{path}\"')
-        else:
-            logger.error("Could not load the model because a tokenizer in Transformers format was not found.")
-            return None, None
-
-    model = LlamacppHF.from_pretrained(model_name)
-
-    if shared.args.tokenizer_dir:
-        tokenizer = load_tokenizer(model_name, tokenizer_dir=shared.args.tokenizer_dir)
-        return model, tokenizer
-    else:
-        return model
+    try:
+        model = LlamaServer(model_file)
+        return model, model
+    except Exception as e:
+        logger.error(f"Error loading the model with llama.cpp: {str(e)}")
 
 
 def ExLlamav3_HF_loader(model_name):
diff --git a/modules/models_settings.py b/modules/models_settings.py
index 693c4dde6a..0af89d2cc1 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -48,7 +48,7 @@ def get_model_metadata(model):
         )
 
     # GGUF metadata
-    if model_settings['loader'] in ['llama.cpp', 'llamacpp_HF']:
+    if model_settings['loader'] == 'llama.cpp':
         path = Path(f'{shared.args.model_dir}/{model}')
         if path.is_file():
             model_file = path
@@ -163,8 +163,6 @@ def infer_loader(model_name, model_settings, hf_quant_method=None):
     path_to_model = Path(f'{shared.args.model_dir}/{model_name}')
     if not path_to_model.exists():
         loader = None
-    elif len(list(path_to_model.glob('*.gguf'))) > 0 and path_to_model.is_dir() and (path_to_model / 'tokenizer_config.json').exists():
-        loader = 'llamacpp_HF'
     elif len(list(path_to_model.glob('*.gguf'))) > 0:
         loader = 'llama.cpp'
     elif re.match(r'.*\.gguf', model_name.lower()):
diff --git a/modules/shared.py b/modules/shared.py
index 91a2d361e2..83761b7548 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -86,7 +86,7 @@
 
 # Model loader
 group = parser.add_argument_group('Model loader')
-group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav3_HF, ExLlamav2_HF, ExLlamav2, HQQ, TensorRT-LLM.')
+group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF, ExLlamav2, HQQ, TensorRT-LLM.')
 
 # Transformers/Accelerate
 group = parser.add_argument_group('Transformers/Accelerate')
@@ -116,24 +116,17 @@
 # llama.cpp
 group = parser.add_argument_group('llama.cpp')
 group.add_argument('--flash-attn', action='store_true', help='Use flash-attention.')
-group.add_argument('--tensorcores', action='store_true', help='NVIDIA only: use llama-cpp-python compiled without GGML_CUDA_FORCE_MMQ. This may improve performance on newer cards.')
 group.add_argument('--n_ctx', type=int, default=8192, help='Size of the prompt context.')
 group.add_argument('--threads', type=int, default=0, help='Number of threads to use.')
 group.add_argument('--threads-batch', type=int, default=0, help='Number of threads to use for batches/prompt processing.')
-group.add_argument('--no_mul_mat_q', action='store_true', help='Disable the mulmat kernels.')
-group.add_argument('--n_batch', type=int, default=512, help='Maximum number of prompt tokens to batch together when calling llama_eval.')
+group.add_argument('--batch-size', type=int, default=2048, help='Maximum number of prompt tokens to batch together when calling llama_eval.')
 group.add_argument('--no-mmap', action='store_true', help='Prevent mmap from being used.')
 group.add_argument('--mlock', action='store_true', help='Force the system to keep the model in RAM.')
 group.add_argument('--n-gpu-layers', type=int, default=0, help='Number of layers to offload to the GPU.')
-group.add_argument('--tensor_split', type=str, default=None, help='Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.')
+group.add_argument('--tensor-split', type=str, default=None, help='Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.')
 group.add_argument('--numa', action='store_true', help='Activate NUMA task allocation for llama.cpp.')
-group.add_argument('--logits_all', action='store_true', help='Needs to be set for perplexity evaluation to work. Otherwise, ignore it, as it makes prompt processing slower.')
-group.add_argument('--no_offload_kqv', action='store_true', help='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
-group.add_argument('--cache-capacity', type=str, help='Maximum cache capacity (llama-cpp-python). Examples: 2000MiB, 2GiB. When provided without units, bytes will be assumed.')
-group.add_argument('--row_split', action='store_true', help='Split the model by rows across GPUs. This may improve multi-gpu performance.')
-group.add_argument('--streaming-llm', action='store_true', help='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
-group.add_argument('--attention-sink-size', type=int, default=5, help='StreamingLLM: number of sink tokens. Only used if the trimmed prompt does not share a prefix with the old prompt.')
-group.add_argument('--tokenizer-dir', type=str, help='Load the tokenizer from this folder. Meant to be used with llamacpp_HF through the command-line.')
+group.add_argument('--no-kv-offload', action='store_true', help='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
+group.add_argument('--row-split', action='store_true', help='Split the model by rows across GPUs. This may improve multi-gpu performance.')
 
 # ExLlamaV2
 group = parser.add_argument_group('ExLlamaV2')
@@ -212,6 +205,13 @@
 group.add_argument('--groupsize', type=int, default=-1, help='DEPRECATED')
 group.add_argument('--model-menu', action='store_true', help='DEPRECATED')
 group.add_argument('--multimodal-pipeline', type=str, default=None, help='DEPRECATED')
+group.add_argument('--streaming-llm', action='store_true', help='DEPRECATED')
+group.add_argument('--attention-sink-size', type=int, default=5, help='DEPRECATED')
+group.add_argument('--tokenizer-dir', type=str, help='DEPRECATED')
+group.add_argument('--logits_all', action='store_true', help='DEPRECATED')
+group.add_argument('--no_mul_mat_q', action='store_true', help='DEPRECATED')
+group.add_argument('--cache-capacity', type=str, help='DEPRECATED')
+group.add_argument('--tensorcores', action='store_true', help='DEPRECATED')
 
 args = parser.parse_args()
 args_defaults = parser.parse_args([])
@@ -262,8 +262,6 @@ def fix_loader_name(name):
     name = name.lower()
     if name in ['llamacpp', 'llama.cpp', 'llama-cpp', 'llama cpp']:
         return 'llama.cpp'
-    if name in ['llamacpp_hf', 'llama.cpp_hf', 'llama-cpp-hf', 'llamacpp-hf', 'llama.cpp-hf']:
-        return 'llamacpp_HF'
     elif name in ['transformers', 'huggingface', 'hf', 'hugging_face', 'hugging face']:
         return 'Transformers'
     elif name in ['exllamav2', 'exllama-v2', 'ex_llama-v2', 'exlamav2', 'exlama-v2', 'exllama2', 'exllama-2']:
@@ -316,7 +314,7 @@ def del_key(key, fallback_set):
                 set('cache_type', 'fp8')
             elif cache_4bit:
                 set('cache_type', 'q4')
-        elif loader.lower() in ['llama.cpp', 'llamacpp_hf']:
+        elif loader.lower() == 'llama.cpp':
             # Llama.cpp loader-specific cache type
             if cache_4bit:
                 set('cache_type', 'q4_0')
diff --git a/modules/text_generation.py b/modules/text_generation.py
index eff6495ecf..16aba3cbb5 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -17,7 +17,6 @@
 
 import modules.shared as shared
 from modules import models, sampler_hijack
-from modules.cache_utils import process_llamacpp_cache
 from modules.callbacks import (
     Iteratorize,
     Stream,
@@ -56,7 +55,7 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
             yield ''
             return
 
-        if shared.model.__class__.__name__ in ['LlamaCppModel', 'Exllamav2Model', 'TensorRTLLMModel']:
+        if shared.model.__class__.__name__ in ['LlamaServer', 'Exllamav2Model', 'TensorRTLLMModel']:
             generate_func = generate_reply_custom
         else:
             generate_func = generate_reply_HF
@@ -133,8 +132,12 @@ def encode(prompt, add_special_tokens=True, add_bos_token=True, truncation_lengt
     if shared.tokenizer is None:
         raise ValueError('No tokenizer is loaded')
 
-    if shared.model.__class__.__name__ in ['LlamaCppModel', 'Exllamav2Model', 'TensorRTLLMModel']:
-        input_ids = shared.tokenizer.encode(str(prompt))
+    if shared.model.__class__.__name__ in ['LlamaServer', 'Exllamav2Model', 'TensorRTLLMModel']:
+        if shared.model.__class__.__name__ == 'LlamaServer':
+            input_ids = shared.tokenizer.encode(str(prompt), add_bos_token=add_bos_token)
+        else:
+            input_ids = shared.tokenizer.encode(str(prompt))
+
         if shared.model.__class__.__name__ not in ['Exllamav2Model']:
             input_ids = np.array(input_ids).reshape(1, len(input_ids))
     else:
@@ -159,7 +162,7 @@ def encode(prompt, add_special_tokens=True, add_bos_token=True, truncation_lengt
     if truncation_length is not None:
         input_ids = input_ids[:, -truncation_length:]
 
-    if shared.model.__class__.__name__ in ['LlamaCppModel', 'Exllamav2Model', 'TensorRTLLMModel'] or shared.args.cpu:
+    if shared.model.__class__.__name__ in ['LlamaServer', 'Exllamav2Model', 'TensorRTLLMModel'] or shared.args.cpu:
         return input_ids
     else:
         device = get_device()
@@ -186,7 +189,7 @@ def get_encoded_length(prompt):
 
 def get_token_ids(prompt):
     tokens = encode(prompt)[0]
-    decoded_tokens = [shared.tokenizer.decode([i]) for i in tokens]
+    decoded_tokens = [shared.tokenizer.decode([int(i)]) for i in tokens]
 
     output = ''
     for row in list(zip(tokens, decoded_tokens)):
@@ -401,12 +404,6 @@ def generate_reply_HF(question, original_question, seed, state, stopping_strings
         logger.info("PROMPT=")
         print_prompt(decode(input_ids[0], skip_special_tokens=False))
 
-    # Handle StreamingLLM for llamacpp_HF
-    if shared.model.__class__.__name__ == 'LlamacppHF' and shared.args.streaming_llm:
-        tmp = process_llamacpp_cache(shared.model.model, input_ids[-1].tolist(), shared.model.model._input_ids.tolist())
-        shared.model.past_seq = torch.tensor(tmp)
-        shared.model.save_cache()
-
     t0 = time.time()
     try:
         if not is_chat and not shared.is_seq2seq:
diff --git a/modules/ui.py b/modules/ui.py
index adbb67b086..919ab8da15 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -110,7 +110,7 @@ def list_model_elements():
         'n_gpu_layers',
         'threads',
         'threads_batch',
-        'n_batch',
+        'batch_size',
         'hqq_backend',
         'n_ctx',
         'max_seq_len',
@@ -122,20 +122,17 @@ def list_model_elements():
         'compress_pos_emb',
         'compute_dtype',
         'quant_type',
-        'attention_sink_size',
         'num_experts_per_token',
-        'tensorcores',
         'load_in_8bit',
         'load_in_4bit',
         'torch_compile',
         'flash_attn',
         'use_flash_attention_2',
-        'streaming_llm',
         'auto_devices',
         'cpu',
         'disk',
         'row_split',
-        'no_offload_kqv',
+        'no_kv_offload',
         'no_mul_mat_q',
         'no_mmap',
         'mlock',
@@ -150,7 +147,6 @@ def list_model_elements():
         'no_sdpa',
         'cfg_cache',
         'cpp_runner',
-        'logits_all',
         'trust_remote_code',
         'no_use_fast',
     ]
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 4fc1de0819..fca883ebb9 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -87,7 +87,7 @@ def create_ui():
                             shared.gradio['n_gpu_layers'] = gr.Slider(label="n-gpu-layers", minimum=0, maximum=256, value=shared.args.n_gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['threads'] = gr.Slider(label="threads", minimum=0, step=1, maximum=256, value=shared.args.threads)
                             shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
-                            shared.gradio['n_batch'] = gr.Slider(label="n_batch", minimum=1, maximum=2048, step=1, value=shared.args.n_batch)
+                            shared.gradio['batch_size'] = gr.Slider(label="batch_size", minimum=1, maximum=4096, step=1, value=shared.args.batch_size)
                             shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
                             shared.gradio['n_ctx'] = gr.Number(label="n_ctx", precision=0, step=256, value=shared.args.n_ctx, info='Context length. ⚠️ Lower this value if you can\'t load the model. Common values: 2048, 4096, 8192, 16384, 32768.')
                             shared.gradio['max_seq_len'] = gr.Number(label='max_seq_len', precision=0, step=256, value=shared.args.max_seq_len, info='Context length. ⚠️ Lower this value if you can\'t load the model. Common values: 2048, 4096, 8192, 16384, 32768.')
@@ -99,22 +99,19 @@ def create_ui():
                             shared.gradio['compress_pos_emb'] = gr.Number(label='compress_pos_emb', value=shared.args.compress_pos_emb, precision=2, info='Positional embeddings compression factor. Should be set to (context length) / (model\'s original context length). Equal to 1/rope_freq_scale.')
                             shared.gradio['compute_dtype'] = gr.Dropdown(label="compute_dtype", choices=["bfloat16", "float16", "float32"], value=shared.args.compute_dtype, info='Used by load-in-4bit.')
                             shared.gradio['quant_type'] = gr.Dropdown(label="quant_type", choices=["nf4", "fp4"], value=shared.args.quant_type, info='Used by load-in-4bit.')
-                            shared.gradio['attention_sink_size'] = gr.Number(label="attention_sink_size", value=shared.args.attention_sink_size, precision=0, info='StreamingLLM: number of sink tokens. Only used if the trimmed prompt doesn\'t share a prefix with the old prompt.')
                             shared.gradio['num_experts_per_token'] = gr.Number(label="Number of experts per token", value=shared.args.num_experts_per_token, info='Only applies to MoE models like Mixtral.')
 
                         with gr.Column():
-                            shared.gradio['tensorcores'] = gr.Checkbox(label="tensorcores", value=shared.args.tensorcores, info='NVIDIA only: use llama-cpp-python compiled without GGML_CUDA_FORCE_MMQ. This may improve performance on newer cards.')
                             shared.gradio['load_in_8bit'] = gr.Checkbox(label="load-in-8bit", value=shared.args.load_in_8bit)
                             shared.gradio['load_in_4bit'] = gr.Checkbox(label="load-in-4bit", value=shared.args.load_in_4bit)
                             shared.gradio['torch_compile'] = gr.Checkbox(label="torch-compile", value=shared.args.torch_compile, info='Compile the model with torch.compile for improved performance.')
                             shared.gradio['flash_attn'] = gr.Checkbox(label="flash_attn", value=shared.args.flash_attn, info='Use flash-attention.')
                             shared.gradio['use_flash_attention_2'] = gr.Checkbox(label="use_flash_attention_2", value=shared.args.use_flash_attention_2, info='Set use_flash_attention_2=True while loading the model.')
-                            shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming_llm", value=shared.args.streaming_llm, info='(experimental) Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
                             shared.gradio['auto_devices'] = gr.Checkbox(label="auto-devices", value=shared.args.auto_devices)
                             shared.gradio['cpu'] = gr.Checkbox(label="cpu", value=shared.args.cpu, info='llama.cpp: Use llama-cpp-python compiled without GPU acceleration. Transformers: use PyTorch in CPU mode.')
                             shared.gradio['disk'] = gr.Checkbox(label="disk", value=shared.args.disk)
                             shared.gradio['row_split'] = gr.Checkbox(label="row_split", value=shared.args.row_split, info='Split the model by rows across GPUs. This may improve multi-gpu performance.')
-                            shared.gradio['no_offload_kqv'] = gr.Checkbox(label="no_offload_kqv", value=shared.args.no_offload_kqv, info='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
+                            shared.gradio['no_kv_offload'] = gr.Checkbox(label="no_kv_offload", value=shared.args.no_kv_offload, info='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
                             shared.gradio['no_mul_mat_q'] = gr.Checkbox(label="no_mul_mat_q", value=shared.args.no_mul_mat_q, info='Disable the mulmat kernels.')
                             shared.gradio['no_mmap'] = gr.Checkbox(label="no-mmap", value=shared.args.no_mmap)
                             shared.gradio['mlock'] = gr.Checkbox(label="mlock", value=shared.args.mlock)
@@ -129,7 +126,6 @@ def create_ui():
                             shared.gradio['no_sdpa'] = gr.Checkbox(label="no_sdpa", value=shared.args.no_sdpa)
                             shared.gradio['cfg_cache'] = gr.Checkbox(label="cfg-cache", value=shared.args.cfg_cache, info='Necessary to use CFG with this loader.')
                             shared.gradio['cpp_runner'] = gr.Checkbox(label="cpp-runner", value=shared.args.cpp_runner, info='Enable inference with ModelRunnerCpp, which is faster than the default ModelRunner.')
-                            shared.gradio['logits_all'] = gr.Checkbox(label="logits_all", value=shared.args.logits_all, info='Needs to be set for perplexity evaluation to work with this loader. Otherwise, ignore it, as it makes prompt processing slower.')
                             shared.gradio['trust_remote_code'] = gr.Checkbox(label="trust-remote-code", value=shared.args.trust_remote_code, info='Set trust_remote_code=True while loading the tokenizer/model. To enable this option, start the web UI with the --trust-remote-code flag.', interactive=shared.args.trust_remote_code)
                             shared.gradio['no_use_fast'] = gr.Checkbox(label="no_use_fast", value=shared.args.no_use_fast, info='Set use_fast=False while loading the tokenizer.')
                             shared.gradio['llamacpp_HF_info'] = gr.Markdown("llamacpp_HF loads llama.cpp as a Transformers model. To use it, you need to place your GGUF in a subfolder of models/ with the necessary tokenizer files.\n\nYou can use the \"llamacpp_HF creator\" menu to do that automatically.")
@@ -147,15 +143,6 @@ def create_ui():
                         shared.gradio['download_model_button'] = gr.Button("Download", variant='primary', interactive=not mu)
                         shared.gradio['get_file_list'] = gr.Button("Get file list", interactive=not mu)
 
-                with gr.Tab("llamacpp_HF creator"):
-                    with gr.Row():
-                        shared.gradio['gguf_menu'] = gr.Dropdown(choices=utils.get_available_ggufs(), value=lambda: shared.model_name, label='Choose your GGUF', elem_classes='slim-dropdown', interactive=not mu)
-                        ui.create_refresh_button(shared.gradio['gguf_menu'], lambda: None, lambda: {'choices': utils.get_available_ggufs()}, 'refresh-button', interactive=not mu)
-
-                    shared.gradio['unquantized_url'] = gr.Textbox(label="Enter the URL for the original (unquantized) model", info="Example: https://huggingface.co/lmsys/vicuna-13b-v1.5", max_lines=1)
-                    shared.gradio['create_llamacpp_hf_button'] = gr.Button("Submit", variant="primary", interactive=not mu)
-                    gr.Markdown("This will move your gguf file into a subfolder of `models` along with the necessary tokenizer files.")
-
                 with gr.Tab("Customize instruction template"):
                     with gr.Row():
                         shared.gradio['customized_template'] = gr.Dropdown(choices=utils.get_available_instruction_templates(), value='None', label='Select the desired instruction template', elem_classes='slim-dropdown')
@@ -195,7 +182,6 @@ def create_event_handlers():
     shared.gradio['download_model_button'].click(download_model_wrapper, gradio('custom_model_menu', 'download_specific_file'), gradio('model_status'), show_progress=True)
     shared.gradio['get_file_list'].click(partial(download_model_wrapper, return_links=True), gradio('custom_model_menu', 'download_specific_file'), gradio('model_status'), show_progress=True)
     shared.gradio['autoload_model'].change(lambda x: gr.update(visible=not x), gradio('autoload_model'), gradio('load_model'))
-    shared.gradio['create_llamacpp_hf_button'].click(create_llamacpp_hf, gradio('gguf_menu', 'unquantized_url'), gradio('model_status'), show_progress=True)
     shared.gradio['customized_template_submit'].click(save_instruction_template, gradio('model_menu', 'customized_template'), gradio('model_status'), show_progress=True)
 
 
@@ -286,34 +272,11 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
         yield traceback.format_exc().replace('\n', '\n\n')
 
 
-def create_llamacpp_hf(gguf_name, unquantized_url, progress=gr.Progress()):
-    try:
-        downloader = importlib.import_module("download-model").ModelDownloader()
-
-        progress(0.0)
-        model, branch = downloader.sanitize_model_and_branch_names(unquantized_url, None)
-
-        yield ("Getting the tokenizer files links from Hugging Face")
-        links, sha256, is_lora, is_llamacpp = downloader.get_download_links_from_huggingface(model, branch, text_only=True)
-        output_folder = Path(shared.args.model_dir) / (re.sub(r'(?i)\.gguf$', '', gguf_name) + "-HF")
-
-        yield (f"Downloading tokenizer to `{output_folder}/`")
-        downloader.download_model_files(model, branch, links, sha256, output_folder, progress_bar=progress, threads=4, is_llamacpp=False)
-
-        # Move the GGUF
-        (Path(shared.args.model_dir) / gguf_name).rename(output_folder / gguf_name)
-
-        yield (f"Model saved to `{output_folder}/`.\n\nYou can now load it using llamacpp_HF.")
-    except:
-        progress(1.0)
-        yield traceback.format_exc().replace('\n', '\n\n')
-
-
 def update_truncation_length(current_length, state):
     if 'loader' in state:
         if state['loader'].lower().startswith('exllama'):
             return state['max_seq_len']
-        elif state['loader'] in ['llama.cpp', 'llamacpp_HF']:
+        elif state['loader'] == 'llama.cpp':
             return state['n_ctx']
 
     return current_length
diff --git a/requirements.txt b/requirements.txt
index 5485352871..84bdbd6289 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -31,19 +31,9 @@ flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
-# llama-cpp-python (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-
-# llama-cpp-python (CUDA, with GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu124-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu124-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-
-# llama-cpp-python (CUDA, without GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu124-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu124-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-
 # CUDA wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.1.0+cu124-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.1.0+cu124-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 3d24891f6a..c94522cfed 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -30,11 +30,7 @@ flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
-# llama-cpp-python (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.3.8+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/rocm/llama_cpp_binaries-0.1.0+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 057b631dfa..a62d4c1ee1 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -30,10 +30,7 @@ flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
-# llama-cpp-python (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-
 # AMD wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/rocm/llama_cpp_binaries-0.1.0+rocm6.1.2avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index a9e88a258d..766a3dd6db 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -31,7 +31,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.1.0-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.1.0-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 4c64c03017..6a4cd80c27 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -31,8 +31,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.3.8-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.1.0-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.1.0-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.1.0-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index c7e2687c94..67c8fe31e7 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -30,6 +30,6 @@ flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
-# llama-cpp-python (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+# llama.cpp (CPU only, AVX2)
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.1.0+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.1.0+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 2003c5448b..3e988c46e5 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -30,6 +30,6 @@ flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
-# llama-cpp-python (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+# llama.cpp (CPU only, no AVX2)
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.1.0+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.1.0+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 0096c9ab2b..27cd2150df 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -31,19 +31,9 @@ flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
-# llama-cpp-python (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.3.8+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-
-# llama-cpp-python (CUDA, with GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu124avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.3.8+cu124avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-
-# llama-cpp-python (CUDA, without GGML_CUDA_FORCE_MMQ)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu124avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.3.8+cu124avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-
 # CUDA wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.1.0+cu124avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.1.0+cu124avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"

From 8144e1031e0f311a69092e75f6517654c7965617 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 06:02:28 -0700
Subject: [PATCH 0293/1701] Remove deprecated command-line flags

---
 modules/loaders.py       |  1 -
 modules/shared.py        | 76 ----------------------------------------
 modules/ui.py            |  1 -
 modules/ui_model_menu.py |  1 -
 4 files changed, 79 deletions(-)

diff --git a/modules/loaders.py b/modules/loaders.py
index 3060406d8a..aaf4c8ed78 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -39,7 +39,6 @@
         'flash_attn',
         'row_split',
         'no_kv_offload',
-        'no_mul_mat_q',
         'no_mmap',
         'mlock',
         'numa',
diff --git a/modules/shared.py b/modules/shared.py
index 83761b7548..9e525bd9ec 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -192,26 +192,6 @@
 
 # Deprecated parameters
 group = parser.add_argument_group('Deprecated')
-group.add_argument('--cache_4bit', action='store_true', help='DEPRECATED')
-group.add_argument('--cache_8bit', action='store_true', help='DEPRECATED')
-group.add_argument('--chat-buttons', action='store_true', help='DEPRECATED')
-group.add_argument('--triton', action='store_true', help='DEPRECATED')
-group.add_argument('--no_inject_fused_mlp', action='store_true', help='DEPRECATED')
-group.add_argument('--no_use_cuda_fp16', action='store_true', help='DEPRECATED')
-group.add_argument('--desc_act', action='store_true', help='DEPRECATED')
-group.add_argument('--disable_exllama', action='store_true', help='DEPRECATED')
-group.add_argument('--disable_exllamav2', action='store_true', help='DEPRECATED')
-group.add_argument('--wbits', type=int, default=0, help='DEPRECATED')
-group.add_argument('--groupsize', type=int, default=-1, help='DEPRECATED')
-group.add_argument('--model-menu', action='store_true', help='DEPRECATED')
-group.add_argument('--multimodal-pipeline', type=str, default=None, help='DEPRECATED')
-group.add_argument('--streaming-llm', action='store_true', help='DEPRECATED')
-group.add_argument('--attention-sink-size', type=int, default=5, help='DEPRECATED')
-group.add_argument('--tokenizer-dir', type=str, help='DEPRECATED')
-group.add_argument('--logits_all', action='store_true', help='DEPRECATED')
-group.add_argument('--no_mul_mat_q', action='store_true', help='DEPRECATED')
-group.add_argument('--cache-capacity', type=str, help='DEPRECATED')
-group.add_argument('--tensorcores', action='store_true', help='DEPRECATED')
 
 args = parser.parse_args()
 args_defaults = parser.parse_args([])
@@ -276,58 +256,6 @@ def fix_loader_name(name):
         return 'TensorRT-LLM'
 
 
-def transform_legacy_kv_cache_options(opts):
-    # Handle both argparse.Namespace and dict here
-    def get(key):
-        return opts.get(key) if isinstance(opts, dict) else getattr(opts, key, None)
-
-    def set(key, value):
-        if isinstance(opts, dict):
-            opts[key] = value
-        else:
-            setattr(opts, key, value)
-
-    def del_key(key, fallback_set):
-        # only remove from user dict, can't delete from argparse.Namespace
-        if type(opts) is dict:
-            if key in opts:
-                del opts[key]
-        else:
-            setattr(opts, key, fallback_set)
-
-    # Retrieve values
-    loader = get('loader')
-    cache_8bit = get('cache_8bit')
-    cache_4bit = get('cache_4bit')
-
-    # Determine cache type based on loader or legacy flags
-    if cache_8bit or cache_4bit:
-        if not loader:
-            # Legacy behavior: prefer 8-bit over 4-bit to minimize breakage
-            if cache_8bit:
-                set('cache_type', 'fp8')
-            elif cache_4bit:
-                set('cache_type', 'q4')
-        elif loader.lower() in ['exllamav2', 'exllamav2_hf']:
-            # ExLlamaV2 loader-specific cache type
-            if cache_8bit:
-                set('cache_type', 'fp8')
-            elif cache_4bit:
-                set('cache_type', 'q4')
-        elif loader.lower() == 'llama.cpp':
-            # Llama.cpp loader-specific cache type
-            if cache_4bit:
-                set('cache_type', 'q4_0')
-            elif cache_8bit:
-                set('cache_type', 'q8_0')
-
-    # Clean up legacy keys
-    del_key('cache_4bit', False)
-    del_key('cache_8bit', False)
-
-    return opts
-
-
 def add_extension(name, last=False):
     if args.extensions is None:
         args.extensions = [name]
@@ -356,14 +284,10 @@ def load_user_config():
     else:
         user_config = {}
 
-    for model_name in user_config:
-        user_config[model_name] = transform_legacy_kv_cache_options(user_config[model_name])
-
     return user_config
 
 
 args.loader = fix_loader_name(args.loader)
-args = transform_legacy_kv_cache_options(args)
 
 # Activate the API extension
 if args.api or args.public_api:
diff --git a/modules/ui.py b/modules/ui.py
index 919ab8da15..c36fe55351 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -133,7 +133,6 @@ def list_model_elements():
         'disk',
         'row_split',
         'no_kv_offload',
-        'no_mul_mat_q',
         'no_mmap',
         'mlock',
         'numa',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index fca883ebb9..cdc0e77ca6 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -112,7 +112,6 @@ def create_ui():
                             shared.gradio['disk'] = gr.Checkbox(label="disk", value=shared.args.disk)
                             shared.gradio['row_split'] = gr.Checkbox(label="row_split", value=shared.args.row_split, info='Split the model by rows across GPUs. This may improve multi-gpu performance.')
                             shared.gradio['no_kv_offload'] = gr.Checkbox(label="no_kv_offload", value=shared.args.no_kv_offload, info='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
-                            shared.gradio['no_mul_mat_q'] = gr.Checkbox(label="no_mul_mat_q", value=shared.args.no_mul_mat_q, info='Disable the mulmat kernels.')
                             shared.gradio['no_mmap'] = gr.Checkbox(label="no-mmap", value=shared.args.no_mmap)
                             shared.gradio['mlock'] = gr.Checkbox(label="mlock", value=shared.args.mlock)
                             shared.gradio['numa'] = gr.Checkbox(label="numa", value=shared.args.numa, info='NUMA support can help on some systems with non-uniform memory access.')

From 170ad3d3ec46bb88db940f379fff7c3dd218f262 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 06:03:35 -0700
Subject: [PATCH 0294/1701] Update the README

---
 README.md | 27 +++++++++------------------
 1 file changed, 9 insertions(+), 18 deletions(-)

diff --git a/README.md b/README.md
index 7f440222ee..777c19b621 100644
--- a/README.md
+++ b/README.md
@@ -188,15 +188,13 @@ usage: server.py [-h] [--multi-user] [--character CHARACTER] [--model MODEL] [--
                  [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--cpu] [--auto-devices] [--gpu-memory GPU_MEMORY [GPU_MEMORY ...]]
                  [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast]
                  [--use_flash_attention_2] [--use_eager_attention] [--torch-compile] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn]
-                 [--tensorcores] [--n_ctx N_CTX] [--threads THREADS] [--threads-batch THREADS_BATCH] [--no_mul_mat_q] [--n_batch N_BATCH] [--no-mmap] [--mlock] [--n-gpu-layers N_GPU_LAYERS]
-                 [--tensor_split TENSOR_SPLIT] [--numa] [--logits_all] [--no_offload_kqv] [--cache-capacity CACHE_CAPACITY] [--row_split] [--streaming-llm] [--attention-sink-size ATTENTION_SINK_SIZE]
-                 [--tokenizer-dir TOKENIZER_DIR] [--gpu-split GPU_SPLIT] [--autosplit] [--max_seq_len MAX_SEQ_LEN] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa]
+                 [--n_ctx N_CTX] [--threads THREADS] [--threads-batch THREADS_BATCH] [--batch-size BATCH_SIZE] [--no-mmap] [--mlock] [--n-gpu-layers N_GPU_LAYERS] [--tensor-split TENSOR_SPLIT]
+                 [--numa] [--no-kv-offload] [--row-split] [--gpu-split GPU_SPLIT] [--autosplit] [--max_seq_len MAX_SEQ_LEN] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa]
                  [--num_experts_per_token NUM_EXPERTS_PER_TOKEN] [--enable_tp] [--hqq-backend HQQ_BACKEND] [--cpp-runner] [--cache_type CACHE_TYPE] [--deepspeed] [--nvme-offload-dir NVME_OFFLOAD_DIR]
                  [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT]
                  [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE]
                  [--subpath SUBPATH] [--old-colors] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6]
-                 [--api-disable-ipv4] [--nowebui] [--cache_4bit] [--cache_8bit] [--chat-buttons] [--triton] [--no_inject_fused_mlp] [--no_use_cuda_fp16] [--desc_act] [--disable_exllama]
-                 [--disable_exllamav2] [--wbits WBITS] [--groupsize GROUPSIZE] [--model-menu] [--multimodal-pipeline MULTIMODAL_PIPELINE]
+                 [--api-disable-ipv4] [--nowebui]
 
 Text generation web UI
 
@@ -217,8 +215,8 @@ Basic settings:
   --idle-timeout IDLE_TIMEOUT                    Unload model after this many minutes of inactivity. It will be automatically reloaded when you try to use it again.
 
 Model loader:
-  --loader LOADER                                Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav3_HF, ExLlamav2_HF,
-                                                 ExLlamav2, HQQ, TensorRT-LLM.
+  --loader LOADER                                Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF, ExLlamav2,
+                                                 HQQ, TensorRT-LLM.
 
 Transformers/Accelerate:
   --cpu                                          Use the CPU to generate text. Warning: Training on CPU is extremely slow.
@@ -246,24 +244,17 @@ bitsandbytes 4-bit:
 
 llama.cpp:
   --flash-attn                                   Use flash-attention.
-  --tensorcores                                  NVIDIA only: use llama-cpp-python compiled without GGML_CUDA_FORCE_MMQ. This may improve performance on newer cards.
   --n_ctx N_CTX                                  Size of the prompt context.
   --threads THREADS                              Number of threads to use.
   --threads-batch THREADS_BATCH                  Number of threads to use for batches/prompt processing.
-  --no_mul_mat_q                                 Disable the mulmat kernels.
-  --n_batch N_BATCH                              Maximum number of prompt tokens to batch together when calling llama_eval.
+  --batch-size BATCH_SIZE                        Maximum number of prompt tokens to batch together when calling llama_eval.
   --no-mmap                                      Prevent mmap from being used.
   --mlock                                        Force the system to keep the model in RAM.
   --n-gpu-layers N_GPU_LAYERS                    Number of layers to offload to the GPU.
-  --tensor_split TENSOR_SPLIT                    Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.
+  --tensor-split TENSOR_SPLIT                    Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.
   --numa                                         Activate NUMA task allocation for llama.cpp.
-  --logits_all                                   Needs to be set for perplexity evaluation to work. Otherwise, ignore it, as it makes prompt processing slower.
-  --no_offload_kqv                               Do not offload the K, Q, V to the GPU. This saves VRAM but reduces the performance.
-  --cache-capacity CACHE_CAPACITY                Maximum cache capacity (llama-cpp-python). Examples: 2000MiB, 2GiB. When provided without units, bytes will be assumed.
-  --row_split                                    Split the model by rows across GPUs. This may improve multi-gpu performance.
-  --streaming-llm                                Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.
-  --attention-sink-size ATTENTION_SINK_SIZE      StreamingLLM: number of sink tokens. Only used if the trimmed prompt does not share a prefix with the old prompt.
-  --tokenizer-dir TOKENIZER_DIR                  Load the tokenizer from this folder. Meant to be used with llamacpp_HF through the command-line.
+  --no-kv-offload                                Do not offload the K, Q, V to the GPU. This saves VRAM but reduces the performance.
+  --row-split                                    Split the model by rows across GPUs. This may improve multi-gpu performance.
 
 ExLlamaV2:
   --gpu-split GPU_SPLIT                          Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.

From c6901aba9fa09c77f9b3af64ba26d3b63c377d32 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 06:05:47 -0700
Subject: [PATCH 0295/1701] Remove deprecation warning code

---
 modules/shared.py | 20 --------------------
 1 file changed, 20 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 9e525bd9ec..60760423fc 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -201,28 +201,8 @@
     if hasattr(args, arg):
         provided_arguments.append(arg)
 
-deprecated_args = [
-    'cache_4bit',
-    'cache_8bit',
-    'chat_buttons',
-    'triton',
-    'no_inject_fused_mlp',
-    'no_use_cuda_fp16',
-    'desc_act',
-    'disable_exllama',
-    'disable_exllamav2',
-    'wbits',
-    'groupsize'
-]
-
 
 def do_cmd_flags_warnings():
-
-    # Deprecation warnings
-    for k in deprecated_args:
-        if k in provided_arguments:
-            logger.warning(f'The --{k} flag has been deprecated and will be removed soon. Please remove that flag.')
-
     # Security warnings
     if args.trust_remote_code:
         logger.warning('trust_remote_code is enabled. This is dangerous.')

From 03544d4fb6ffcde9b92d28ad1858934de6550f45 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 06:14:13 -0700
Subject: [PATCH 0296/1701] Bump llama.cpp and exllamav3 to the latest commits

---
 requirements.txt                 | 8 ++++----
 requirements_amd.txt             | 2 +-
 requirements_amd_noavx2.txt      | 2 +-
 requirements_apple_intel.txt     | 6 +++---
 requirements_apple_silicon.txt   | 8 ++++----
 requirements_cpu_only.txt        | 4 ++--
 requirements_cpu_only_noavx2.txt | 4 ++--
 requirements_noavx2.txt          | 8 ++++----
 8 files changed, 21 insertions(+), 21 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 84bdbd6289..607efda0ec 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -32,10 +32,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.1.0+cu124-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.1.0+cu124-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.2.0+cu124-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.2.0+cu124-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements_amd.txt b/requirements_amd.txt
index c94522cfed..b242d4ad3c 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -31,6 +31,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/rocm/llama_cpp_binaries-0.1.0+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/rocm/llama_cpp_binaries-0.2.0+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index a62d4c1ee1..b6105209f3 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -31,6 +31,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/rocm/llama_cpp_binaries-0.1.0+rocm6.1.2avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/rocm/llama_cpp_binaries-0.2.0+rocm6.1.2avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 766a3dd6db..ce730f635c 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -31,7 +31,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.1.0-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.1.0-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2-py3-none-any.whl
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.2.0-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.2.0-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 6a4cd80c27..a7be282d0d 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -31,8 +31,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.1.0-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.1.0-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.1.0-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2-py3-none-any.whl
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.2.0-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.2.0-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.2.0-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index 67c8fe31e7..2437c2ae4b 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -31,5 +31,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.1.0+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.1.0+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.2.0+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.2.0+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 3e988c46e5..cbaa8e96c7 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -31,5 +31,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.1.0+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.1.0+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.2.0+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.2.0+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 27cd2150df..cce27aa22e 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -32,10 +32,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.1.0+cu124avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.1.0+cu124avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a2/exllamav3-0.0.1a2+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.2.0+cu124avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.2.0+cu124avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 1c4a2c9a71b528de396c51c1375975c9d88b72f2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 06:17:58 -0700
Subject: [PATCH 0297/1701] Make exllamav3 safer as well

---
 modules/exllamav3_hf.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index 3bf44c9be1..0f742fa23f 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -153,6 +153,9 @@ def __call__(self, *args, **kwargs):
         else:
             self.past_seq = seq_tensor
 
+        if torch.cuda.is_available():
+            torch.cuda.synchronize()
+
         loss = None
         if labels is not None:
             # Shift so that tokens < n predict n

From ef9910c767615e7b58cd942402bf28b60b3e8b69 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 06:51:28 -0700
Subject: [PATCH 0298/1701] Fix a bug after
 c6901aba9fa09c77f9b3af64ba26d3b63c377d32

---
 modules/ui_session.py | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/modules/ui_session.py b/modules/ui_session.py
index dfb95b83d5..66386d1203 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -69,10 +69,8 @@ def set_interface_arguments(extensions, bool_active):
 
 
 def get_boolean_arguments(active=False):
-    exclude = shared.deprecated_args
-
     cmd_list = vars(shared.args)
-    bool_list = sorted([k for k in cmd_list if type(cmd_list[k]) is bool and k not in exclude + ui.list_model_elements()])
+    bool_list = sorted([k for k in cmd_list if type(cmd_list[k]) is bool and k not in ui.list_model_elements()])
     bool_active = [k for k in bool_list if vars(shared.args)[k]]
 
     if active:

From a0abf934251e8f376c5809eabaf203f29a79ae5d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 06:53:51 -0700
Subject: [PATCH 0299/1701] Connect --rope-freq-base to the new llama.cpp
 loader

---
 modules/llama_cpp_server.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 983b506f52..39aaf42191 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -245,6 +245,7 @@ def _start_server(self):
             "--ctx-size", str(shared.args.n_ctx),
             "--n-gpu-layers", str(shared.args.n_gpu_layers),
             "--batch-size", str(shared.args.batch_size),
+            "--rope-freq-base", str(shared.args.rope_freq_base),
             "--port", str(self.port),
         ]
 

From d68f0fbdf7311608227a103631023fd8e7114dd1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 07:46:04 -0700
Subject: [PATCH 0300/1701] Remove obsolete references to llamacpp_HF

---
 modules/logits.py        | 4 +---
 modules/shared.py        | 2 +-
 modules/ui_model_menu.py | 1 -
 3 files changed, 2 insertions(+), 5 deletions(-)

diff --git a/modules/logits.py b/modules/logits.py
index c5d9112d00..020e742458 100644
--- a/modules/logits.py
+++ b/modules/logits.py
@@ -65,10 +65,8 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
 
         if use_samplers:
             if is_non_hf_exllamav2:
+                # sampling is all done in C++ for exllama, so it is really hard to hijack
                 logger.error("Sampler hijacking is not supported non-Huggingface loaders.")
-                # sampling is all done in c for exllama, so it is really hard to hijack
-                # it should be possible to hijack llamacpp sampler by hijacking all their sampling methods,
-                # but it is not implemented yet
                 return 'Error: Sampler hijacking is not supported non-Huggingface loaders. Please disable the "Use samplers" option.', previous
 
             state['max_new_tokens'] = 1
diff --git a/modules/shared.py b/modules/shared.py
index 60760423fc..641b3f4843 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -220,7 +220,7 @@ def fix_loader_name(name):
         return name
 
     name = name.lower()
-    if name in ['llamacpp', 'llama.cpp', 'llama-cpp', 'llama cpp']:
+    if name in ['llama.cpp', 'llamacpp', 'llama-cpp', 'llama cpp']:
         return 'llama.cpp'
     elif name in ['transformers', 'huggingface', 'hf', 'hugging_face', 'hugging face']:
         return 'Transformers'
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index cdc0e77ca6..8db486219c 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -127,7 +127,6 @@ def create_ui():
                             shared.gradio['cpp_runner'] = gr.Checkbox(label="cpp-runner", value=shared.args.cpp_runner, info='Enable inference with ModelRunnerCpp, which is faster than the default ModelRunner.')
                             shared.gradio['trust_remote_code'] = gr.Checkbox(label="trust-remote-code", value=shared.args.trust_remote_code, info='Set trust_remote_code=True while loading the tokenizer/model. To enable this option, start the web UI with the --trust-remote-code flag.', interactive=shared.args.trust_remote_code)
                             shared.gradio['no_use_fast'] = gr.Checkbox(label="no_use_fast", value=shared.args.no_use_fast, info='Set use_fast=False while loading the tokenizer.')
-                            shared.gradio['llamacpp_HF_info'] = gr.Markdown("llamacpp_HF loads llama.cpp as a Transformers model. To use it, you need to place your GGUF in a subfolder of models/ with the necessary tokenizer files.\n\nYou can use the \"llamacpp_HF creator\" menu to do that automatically.")
                             shared.gradio['exllamav2_info'] = gr.Markdown("ExLlamav2_HF is recommended over ExLlamav2 for better integration with extensions and more consistent sampling behavior across loaders.")
                             shared.gradio['tensorrt_llm_info'] = gr.Markdown('* TensorRT-LLM has to be installed manually in a separate Python 3.10 environment at the moment. For a guide, consult the description of [this PR](https://github.com/oobabooga/text-generation-webui/pull/5715). \n\n* `max_seq_len` is only used when `cpp-runner` is checked.\n\n* `cpp_runner` does not support streaming at the moment.')
 

From c1cc65e82e226eaaeba1f13a977dc52efac7f556 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 08:06:51 -0700
Subject: [PATCH 0301/1701] Lint

---
 modules/llama_cpp_server.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 39aaf42191..1965a1705c 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -282,7 +282,7 @@ def _start_server(self):
 
         def filter_stderr():
             for line in iter(self.process.stderr.readline, ''):
-                if not line.startswith(('srv ', 'slot ')) and not 'log_server_r: request: GET /health' in line:
+                if not line.startswith(('srv ', 'slot ')) and 'log_server_r: request: GET /health' not in line:
                     sys.stderr.write(line)
                     sys.stderr.flush()
 

From d00d713ace9a1e39bb291175c468eea71a42dfa2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 08:14:15 -0700
Subject: [PATCH 0302/1701] Rename get_max_context_length to
 get_vocabulary_size in the new llama.cpp loader

---
 modules/llama_cpp_server.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 1965a1705c..db500c9c42 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -24,7 +24,7 @@ def __init__(self, model_path, server_path=None):
         self.server_path = server_path
         self.port = self._find_available_port()
         self.process = None
-        self.max_context_length = None
+        self.vocabulary_size = None
         self.bos_token = ""
 
         # Start the server
@@ -209,7 +209,7 @@ def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
         else:
             raise Exception(f"Unexpected response format: 'completion_probabilities' not found in {result}")
 
-    def _get_max_context_length(self):
+    def _get_vocabulary_size(self):
         """Get and store the model's maximum context length."""
         url = f"http://localhost:{self.port}/v1/models"
         response = requests.get(url).json()
@@ -217,7 +217,7 @@ def _get_max_context_length(self):
         if "data" in response and len(response["data"]) > 0:
             model_info = response["data"][0]
             if "meta" in model_info and "n_vocab" in model_info["meta"]:
-                self.max_context_length = model_info["meta"]["n_vocab"]
+                self.vocabulary_size = model_info["meta"]["n_vocab"]
 
     def _get_bos_token(self):
         """Get and store the model's BOS token."""
@@ -311,7 +311,7 @@ def filter_stderr():
             raise TimeoutError(f"Server health check timed out after {timeout} seconds")
 
         # Server is now healthy, get model info
-        self._get_max_context_length()
+        self._get_vocabulary_size()
         self._get_bos_token()
         return self.port
 

From 85c4486d4a5913548e048b0238c2a8165d5f2438 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 08:53:44 -0700
Subject: [PATCH 0303/1701] Update the colab notebook

---
 Colab-TextGen-GPU.ipynb | 1 -
 1 file changed, 1 deletion(-)

diff --git a/Colab-TextGen-GPU.ipynb b/Colab-TextGen-GPU.ipynb
index ebeafc0296..971099f628 100644
--- a/Colab-TextGen-GPU.ipynb
+++ b/Colab-TextGen-GPU.ipynb
@@ -67,7 +67,6 @@
         "\n",
         "  # Install the project in an isolated environment\n",
         "  !GPU_CHOICE=A \\\n",
-        "  USE_CUDA118=FALSE \\\n",
         "  LAUNCH_AFTER_INSTALL=FALSE \\\n",
         "  INSTALL_EXTENSIONS=FALSE \\\n",
         "  ./start_linux.sh\n",

From e52f62d3ff0258e301661fb437ddd1d40b1e5100 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 09:29:57 -0700
Subject: [PATCH 0304/1701] Update README

---
 README.md | 11 -----------
 1 file changed, 11 deletions(-)

diff --git a/README.md b/README.md
index 777c19b621..363a2113e2 100644
--- a/README.md
+++ b/README.md
@@ -126,17 +126,6 @@ Then browse to
 
 `http://localhost:7860/?__theme=dark`
 
-##### AMD GPU on Windows
-
-1) Use `requirements_cpu_only.txt` or `requirements_cpu_only_noavx2.txt` in the command above.
-
-2) Manually install llama-cpp-python using the appropriate command for your hardware: [Installation from PyPI](https://github.com/abetlen/llama-cpp-python#installation-with-hardware-acceleration).
-    * Use the `LLAMA_HIPBLAS=on` toggle.
-    * Note the [Windows remarks](https://github.com/abetlen/llama-cpp-python#windows-remarks).
-
-3) Manually install AutoGPTQ: [Installation](https://github.com/PanQiWei/AutoGPTQ#install-from-source).
-    * Perform the from-source installation - there are no prebuilt ROCm packages for Windows.
-
 ##### Manual install
 
 The `requirements*.txt` above contain various wheels precompiled through GitHub Actions. If you wish to compile things manually, or if you need to because no suitable wheels are available for your hardware, you can use `requirements_nowheels.txt` and then install your desired loaders manually.

From caa6afc88b2b2af7764e7db2d066477b6de1414d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 09:57:57 -0700
Subject: [PATCH 0305/1701] Only show 'GENERATE_PARAMS=...' in the logits
 endpoint if use_logits is True

---
 modules/llama_cpp_server.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index db500c9c42..a6eb9b2035 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -192,7 +192,7 @@ def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
             "post_sampling_probs": use_samplers,
         })
 
-        if shared.args.verbose:
+        if shared.args.verbose and use_samplers:
             logger.info("GENERATE_PARAMS=")
             printable_payload = {k: v for k, v in payload.items() if k != "prompt"}
             pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)

From 8d481ef9d5acdc7209d358ac0ac04539e63340c7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 11:31:22 -0700
Subject: [PATCH 0306/1701] Update README

---
 README.md | 1 +
 1 file changed, 1 insertion(+)

diff --git a/README.md b/README.md
index 363a2113e2..a15e974f58 100644
--- a/README.md
+++ b/README.md
@@ -23,6 +23,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 - Simple LoRA fine-tuning tool.
 - Requirements installed in a self-contained `installer_files` directory that doesn't interfere with the system environment.
 - Extension support, with numerous built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details.
+- Completely private: no telemetry, no tracking, no remote connections.
 
 ## How to install
 

From 51355234290ac3adb0ee0df597aa6a3bb9189cb4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 17:10:26 -0700
Subject: [PATCH 0307/1701] Fix the new llama.cpp loader failing to unload
 models

---
 modules/llama_cpp_server.py | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index a6eb9b2035..8180f97481 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -280,13 +280,17 @@ def _start_server(self):
             bufsize=1
         )
 
-        def filter_stderr():
-            for line in iter(self.process.stderr.readline, ''):
-                if not line.startswith(('srv ', 'slot ')) and 'log_server_r: request: GET /health' not in line:
-                    sys.stderr.write(line)
-                    sys.stderr.flush()
+        def filter_stderr(process_stderr):
+            try:
+                for line in iter(process_stderr.readline, ''):
+                    if not line.startswith(('srv ', 'slot ')) and 'log_server_r: request: GET /health' not in line:
+                        sys.stderr.write(line)
+                        sys.stderr.flush()
+            except (ValueError, IOError):
+                # Handle pipe closed exceptions
+                pass
 
-        threading.Thread(target=filter_stderr, daemon=True).start()
+        threading.Thread(target=filter_stderr, args=(self.process.stderr,), daemon=True).start()
 
         # Wait for server to be healthy
         health_url = f"http://localhost:{self.port}/health"

From 4fabd729c9c05b2a5ee423bd22e06792962fa703 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 17:25:22 -0700
Subject: [PATCH 0308/1701] Fix the API without streaming or without
 'sampler_priority' (closes #6851)

---
 modules/llama_cpp_server.py | 44 ++++++++++++++++++++-----------------
 1 file changed, 24 insertions(+), 20 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 8180f97481..123f94713e 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -61,20 +61,6 @@ def prepare_payload(self, state):
             dry_sequence_breakers = "[" + dry_sequence_breakers + "]"
         dry_sequence_breakers = json.loads(dry_sequence_breakers)
 
-        # Prepare the sampler order
-        samplers = state["sampler_priority"]
-        samplers = samplers.split("\n") if isinstance(samplers, str) else samplers
-        penalty_found = False
-        filtered_samplers = []
-        for s in samplers:
-            if s.strip() in ["dry", "top_k", "typ_p", "top_p", "min_p", "xtc", "temperature"]:
-                filtered_samplers.append(s.strip())
-            elif not penalty_found and s.strip() == "repetition_penalty":
-                filtered_samplers.append("penalties")
-                penalty_found = True
-
-        samplers = filtered_samplers
-
         # Move temperature to the end if temperature_last is true and temperature exists in the list
         if state["temperature_last"] and "temperature" in samplers:
             samplers.remove("temperature")
@@ -106,20 +92,31 @@ def prepare_payload(self, state):
             "grammar": state["grammar_string"],
             "seed": state["seed"],
             "ignore_eos": state["ban_eos_token"],
-            "samplers": samplers,
         }
 
+        # Sampler order
+        if state["sampler_priority"]:
+            samplers = state["sampler_priority"]
+            samplers = samplers.split("\n") if isinstance(samplers, str) else samplers
+            filtered_samplers = []
+
+            penalty_found = False
+            for s in samplers:
+                if s.strip() in ["dry", "top_k", "typ_p", "top_p", "min_p", "xtc", "temperature"]:
+                    filtered_samplers.append(s.strip())
+                elif not penalty_found and s.strip() == "repetition_penalty":
+                    filtered_samplers.append("penalties")
+                    penalty_found = True
+
+            payload["samplers"] = filtered_samplers
+
         if state['custom_token_bans']:
             to_ban = [[int(token_id), False] for token_id in state['custom_token_bans'].split(',')]
             payload["logit_bias"] = to_ban
 
         return payload
 
-    def generate_with_streaming(
-        self,
-        prompt,
-        state,
-    ):
+    def generate_with_streaming(self, prompt, state):
         url = f"http://localhost:{self.port}/completion"
         payload = self.prepare_payload(state)
 
@@ -178,6 +175,13 @@ def generate_with_streaming(
                     print(f"Problematic line: {line}")
                     continue
 
+    def generate(self, prompt, state):
+        output = ""
+        for output in self.generate_with_streaming(prompt, state):
+            pass
+
+        return output
+
     def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
         """Get the logits/probabilities for the next token after a prompt"""
         url = f"http://localhost:{self.port}/completion"

From 5ad080ff25a827c9491e085a45bf63c4b373a75c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 18:04:49 -0700
Subject: [PATCH 0309/1701] Attempt at making the llama-server streaming more
 efficient.

---
 modules/llama_cpp_server.py      | 62 ++++++++++++++++----------------
 requirements.txt                 |  1 +
 requirements_amd.txt             |  1 +
 requirements_amd_noavx2.txt      |  1 +
 requirements_apple_intel.txt     |  1 +
 requirements_apple_silicon.txt   |  1 +
 requirements_cpu_only.txt        |  1 +
 requirements_cpu_only_noavx2.txt |  1 +
 requirements_noavx2.txt          |  1 +
 requirements_nowheels.txt        |  1 +
 10 files changed, 41 insertions(+), 30 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 123f94713e..5319e7af5f 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -8,6 +8,7 @@
 
 import llama_cpp_binaries
 import requests
+import sseclient
 
 from modules import shared
 from modules.logging_colors import logger
@@ -138,42 +139,43 @@ def generate_with_streaming(self, prompt, state):
             pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
             print()
 
-        # Make a direct request with streaming enabled
-        response = requests.post(url, json=payload, stream=True)
-        response.raise_for_status()  # Raise an exception for HTTP errors
+        # Configure headers for Server-Sent Events
+        headers = {
+            'Content-Type': 'application/json',
+            'Accept': 'text/event-stream'
+        }
+
+        response = requests.post(url, json=payload, stream=True, headers=headers)
+        response.raise_for_status()
+
+        # Initialize SSE client for proper event stream parsing
+        client = sseclient.SSEClient(response)
 
         full_text = ""
 
-        # Process the streaming response
-        for line in response.iter_lines():
+        for event in client.events():
             if shared.stop_everything:
                 break
 
-            if line:
-                try:
-                    # Check if the line starts with "data: " and remove it
-                    line_str = line.decode('utf-8')
-                    if line_str.startswith('data: '):
-                        line_str = line_str[6:]  # Remove the "data: " prefix
-
-                    # Parse the JSON data
-                    data = json.loads(line_str)
-
-                    # Extract the token content
-                    if 'content' in data:
-                        token_text = data['content']
-                        full_text += token_text
-                        yield full_text
-
-                    # Check if generation is complete
-                    if data.get('stop', False):
-                        break
-
-                except json.JSONDecodeError as e:
-                    # Log the error and the problematic line
-                    print(f"JSON decode error: {e}")
-                    print(f"Problematic line: {line}")
-                    continue
+            try:
+                # Handle stream termination marker
+                if event.data == '[DONE]':
+                    break
+
+                data = json.loads(event.data)
+
+                if 'content' in data:
+                    token_text = data['content']
+                    full_text += token_text
+                    yield full_text
+
+                if data.get('stop', False):
+                    break
+
+            except json.JSONDecodeError as e:
+                print(f"JSON decode error: {e}")
+                print(f"Problematic data: {event.data}")
+                continue
 
     def generate(self, prompt, state):
         output = ""
diff --git a/requirements.txt b/requirements.txt
index 607efda0ec..b6759806c2 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -19,6 +19,7 @@ requests
 rich
 safetensors==0.5.*
 scipy
+sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_amd.txt b/requirements_amd.txt
index b242d4ad3c..e156bc55c5 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -18,6 +18,7 @@ requests
 rich
 safetensors==0.5.*
 scipy
+sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index b6105209f3..6becd514de 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -18,6 +18,7 @@ requests
 rich
 safetensors==0.5.*
 scipy
+sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index ce730f635c..1223b4d319 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -18,6 +18,7 @@ requests
 rich
 safetensors==0.5.*
 scipy
+sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index a7be282d0d..f47f999199 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -18,6 +18,7 @@ requests
 rich
 safetensors==0.5.*
 scipy
+sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index 2437c2ae4b..007f9ef18c 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -18,6 +18,7 @@ requests
 rich
 safetensors==0.5.*
 scipy
+sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index cbaa8e96c7..4219273d12 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -18,6 +18,7 @@ requests
 rich
 safetensors==0.5.*
 scipy
+sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index cce27aa22e..58e0e5a118 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -19,6 +19,7 @@ requests
 rich
 safetensors==0.5.*
 scipy
+sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index 3b61ca39f6..c210f6a010 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -18,6 +18,7 @@ requests
 rich
 safetensors==0.5.*
 scipy
+sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*

From 71ae05e0a4f8e481731e3803131cd46efa3faa05 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 18:06:36 -0700
Subject: [PATCH 0310/1701] llama.cpp: Fix the sampler priority handling

---
 modules/llama_cpp_server.py | 25 +++++++++++++------------
 1 file changed, 13 insertions(+), 12 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 5319e7af5f..d9b35baad1 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -56,17 +56,6 @@ def decode(self, token_ids, **kwargs):
         return result.get("content", "")
 
     def prepare_payload(self, state):
-        # Prepare DRY
-        dry_sequence_breakers = state['dry_sequence_breakers']
-        if not dry_sequence_breakers.startswith("["):
-            dry_sequence_breakers = "[" + dry_sequence_breakers + "]"
-        dry_sequence_breakers = json.loads(dry_sequence_breakers)
-
-        # Move temperature to the end if temperature_last is true and temperature exists in the list
-        if state["temperature_last"] and "temperature" in samplers:
-            samplers.remove("temperature")
-            samplers.append("temperature")
-
         payload = {
             "temperature": state["temperature"] if not state["dynamic_temperature"] else (state["dynatemp_low"] + state["dynatemp_high"]) / 2,
             "dynatemp_range": 0 if not state["dynamic_temperature"] else (state["dynatemp_high"] - state["dynatemp_low"]) / 2,
@@ -84,7 +73,6 @@ def prepare_payload(self, state):
             "dry_base": state["dry_base"],
             "dry_allowed_length": state["dry_allowed_length"],
             "dry_penalty_last_n": state["repetition_penalty_range"],
-            "dry_sequence_breakers": dry_sequence_breakers,
             "xtc_probability": state["xtc_probability"],
             "xtc_threshold": state["xtc_threshold"],
             "mirostat": state["mirostat_mode"],
@@ -95,6 +83,14 @@ def prepare_payload(self, state):
             "ignore_eos": state["ban_eos_token"],
         }
 
+        # DRY
+        dry_sequence_breakers = state['dry_sequence_breakers']
+        if not dry_sequence_breakers.startswith("["):
+            dry_sequence_breakers = "[" + dry_sequence_breakers + "]"
+
+        dry_sequence_breakers = json.loads(dry_sequence_breakers)
+        payload["dry_sequence_breakers"] = dry_sequence_breakers
+
         # Sampler order
         if state["sampler_priority"]:
             samplers = state["sampler_priority"]
@@ -109,6 +105,11 @@ def prepare_payload(self, state):
                     filtered_samplers.append("penalties")
                     penalty_found = True
 
+            # Move temperature to the end if temperature_last is true and temperature exists in the list
+            if state["temperature_last"] and "temperature" in samplers:
+                samplers.remove("temperature")
+                samplers.append("temperature")
+
             payload["samplers"] = filtered_samplers
 
         if state['custom_token_bans']:

From 2002590536ad0825aa677366240a649ab871b70f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 18:13:54 -0700
Subject: [PATCH 0311/1701] Revert "Attempt at making the llama-server
 streaming more efficient."

This reverts commit 5ad080ff25a827c9491e085a45bf63c4b373a75c.
---
 modules/llama_cpp_server.py      | 62 ++++++++++++++++----------------
 requirements.txt                 |  1 -
 requirements_amd.txt             |  1 -
 requirements_amd_noavx2.txt      |  1 -
 requirements_apple_intel.txt     |  1 -
 requirements_apple_silicon.txt   |  1 -
 requirements_cpu_only.txt        |  1 -
 requirements_cpu_only_noavx2.txt |  1 -
 requirements_noavx2.txt          |  1 -
 requirements_nowheels.txt        |  1 -
 10 files changed, 30 insertions(+), 41 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index d9b35baad1..26ab8f108a 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -8,7 +8,6 @@
 
 import llama_cpp_binaries
 import requests
-import sseclient
 
 from modules import shared
 from modules.logging_colors import logger
@@ -140,43 +139,42 @@ def generate_with_streaming(self, prompt, state):
             pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
             print()
 
-        # Configure headers for Server-Sent Events
-        headers = {
-            'Content-Type': 'application/json',
-            'Accept': 'text/event-stream'
-        }
-
-        response = requests.post(url, json=payload, stream=True, headers=headers)
-        response.raise_for_status()
-
-        # Initialize SSE client for proper event stream parsing
-        client = sseclient.SSEClient(response)
+        # Make a direct request with streaming enabled
+        response = requests.post(url, json=payload, stream=True)
+        response.raise_for_status()  # Raise an exception for HTTP errors
 
         full_text = ""
 
-        for event in client.events():
+        # Process the streaming response
+        for line in response.iter_lines():
             if shared.stop_everything:
                 break
 
-            try:
-                # Handle stream termination marker
-                if event.data == '[DONE]':
-                    break
-
-                data = json.loads(event.data)
-
-                if 'content' in data:
-                    token_text = data['content']
-                    full_text += token_text
-                    yield full_text
-
-                if data.get('stop', False):
-                    break
-
-            except json.JSONDecodeError as e:
-                print(f"JSON decode error: {e}")
-                print(f"Problematic data: {event.data}")
-                continue
+            if line:
+                try:
+                    # Check if the line starts with "data: " and remove it
+                    line_str = line.decode('utf-8')
+                    if line_str.startswith('data: '):
+                        line_str = line_str[6:]  # Remove the "data: " prefix
+
+                    # Parse the JSON data
+                    data = json.loads(line_str)
+
+                    # Extract the token content
+                    if 'content' in data:
+                        token_text = data['content']
+                        full_text += token_text
+                        yield full_text
+
+                    # Check if generation is complete
+                    if data.get('stop', False):
+                        break
+
+                except json.JSONDecodeError as e:
+                    # Log the error and the problematic line
+                    print(f"JSON decode error: {e}")
+                    print(f"Problematic line: {line}")
+                    continue
 
     def generate(self, prompt, state):
         output = ""
diff --git a/requirements.txt b/requirements.txt
index b6759806c2..607efda0ec 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -19,7 +19,6 @@ requests
 rich
 safetensors==0.5.*
 scipy
-sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_amd.txt b/requirements_amd.txt
index e156bc55c5..b242d4ad3c 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -18,7 +18,6 @@ requests
 rich
 safetensors==0.5.*
 scipy
-sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 6becd514de..b6105209f3 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -18,7 +18,6 @@ requests
 rich
 safetensors==0.5.*
 scipy
-sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 1223b4d319..ce730f635c 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -18,7 +18,6 @@ requests
 rich
 safetensors==0.5.*
 scipy
-sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index f47f999199..a7be282d0d 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -18,7 +18,6 @@ requests
 rich
 safetensors==0.5.*
 scipy
-sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index 007f9ef18c..2437c2ae4b 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -18,7 +18,6 @@ requests
 rich
 safetensors==0.5.*
 scipy
-sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 4219273d12..cbaa8e96c7 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -18,7 +18,6 @@ requests
 rich
 safetensors==0.5.*
 scipy
-sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 58e0e5a118..cce27aa22e 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -19,7 +19,6 @@ requests
 rich
 safetensors==0.5.*
 scipy
-sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index c210f6a010..3b61ca39f6 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -18,7 +18,6 @@ requests
 rich
 safetensors==0.5.*
 scipy
-sseclient-py==1.8.0
 sentencepiece
 tensorboard
 transformers==4.50.*

From b3342b8dd8eb3f3c191ef0706ba2ab1493fa8937 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 18:46:36 -0700
Subject: [PATCH 0312/1701] llama.cpp: optimize the llama-server connection

---
 modules/llama_cpp_server.py | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 26ab8f108a..822800b9b2 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -24,6 +24,7 @@ def __init__(self, model_path, server_path=None):
         self.server_path = server_path
         self.port = self._find_available_port()
         self.process = None
+        self.session = requests.Session()
         self.vocabulary_size = None
         self.bos_token = ""
 
@@ -40,7 +41,7 @@ def encode(self, text, add_bos_token=False, **kwargs):
             "add_special": add_bos_token,
         }
 
-        response = requests.post(url, json=payload)
+        response = self.session.post(url, json=payload)
         result = response.json()
         return result.get("tokens", [])
 
@@ -50,7 +51,7 @@ def decode(self, token_ids, **kwargs):
             "tokens": token_ids,
         }
 
-        response = requests.post(url, json=payload)
+        response = self.session.post(url, json=payload)
         result = response.json()
         return result.get("content", "")
 
@@ -140,7 +141,7 @@ def generate_with_streaming(self, prompt, state):
             print()
 
         # Make a direct request with streaming enabled
-        response = requests.post(url, json=payload, stream=True)
+        response = self.session.post(url, json=payload, stream=True)
         response.raise_for_status()  # Raise an exception for HTTP errors
 
         full_text = ""
@@ -203,7 +204,7 @@ def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
             pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
             print()
 
-        response = requests.post(url, json=payload)
+        response = self.session.post(url, json=payload)
         result = response.json()
 
         if "completion_probabilities" in result:
@@ -217,7 +218,7 @@ def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
     def _get_vocabulary_size(self):
         """Get and store the model's maximum context length."""
         url = f"http://localhost:{self.port}/v1/models"
-        response = requests.get(url).json()
+        response = self.session.get(url).json()
 
         if "data" in response and len(response["data"]) > 0:
             model_info = response["data"][0]
@@ -227,7 +228,7 @@ def _get_vocabulary_size(self):
     def _get_bos_token(self):
         """Get and store the model's BOS token."""
         url = f"http://localhost:{self.port}/props"
-        response = requests.get(url).json()
+        response = self.session.get(url).json()
         if "bos_token" in response:
             self.bos_token = response["bos_token"]
 
@@ -309,7 +310,7 @@ def filter_stderr(process_stderr):
                 raise RuntimeError(f"Server process terminated unexpectedly with exit code: {exit_code}")
 
             try:
-                response = requests.get(health_url)
+                response = self.session.get(health_url)
                 if response.status_code == 200:
                     break
             except:

From f727b4a2cc674465558599e5a75e56d7cb261a88 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 19:01:39 -0700
Subject: [PATCH 0313/1701] llama.cpp: close the connection properly when
 generation is cancelled

---
 modules/llama_cpp_server.py | 72 ++++++++++++++++++-------------------
 1 file changed, 36 insertions(+), 36 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 822800b9b2..34aab61393 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -140,42 +140,42 @@ def generate_with_streaming(self, prompt, state):
             pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
             print()
 
-        # Make a direct request with streaming enabled
-        response = self.session.post(url, json=payload, stream=True)
-        response.raise_for_status()  # Raise an exception for HTTP errors
-
-        full_text = ""
-
-        # Process the streaming response
-        for line in response.iter_lines():
-            if shared.stop_everything:
-                break
-
-            if line:
-                try:
-                    # Check if the line starts with "data: " and remove it
-                    line_str = line.decode('utf-8')
-                    if line_str.startswith('data: '):
-                        line_str = line_str[6:]  # Remove the "data: " prefix
-
-                    # Parse the JSON data
-                    data = json.loads(line_str)
-
-                    # Extract the token content
-                    if 'content' in data:
-                        token_text = data['content']
-                        full_text += token_text
-                        yield full_text
-
-                    # Check if generation is complete
-                    if data.get('stop', False):
-                        break
-
-                except json.JSONDecodeError as e:
-                    # Log the error and the problematic line
-                    print(f"JSON decode error: {e}")
-                    print(f"Problematic line: {line}")
-                    continue
+        # Make a direct request with streaming enabled using a context manager
+        with self.session.post(url, json=payload, stream=True) as response:
+            response.raise_for_status()  # Raise an exception for HTTP errors
+
+            full_text = ""
+
+            # Process the streaming response
+            for line in response.iter_lines():
+                if shared.stop_everything:
+                    break
+
+                if line:
+                    try:
+                        # Check if the line starts with "data: " and remove it
+                        line_str = line.decode('utf-8')
+                        if line_str.startswith('data: '):
+                            line_str = line_str[6:]  # Remove the "data: " prefix
+
+                        # Parse the JSON data
+                        data = json.loads(line_str)
+
+                        # Extract the token content
+                        if 'content' in data:
+                            token_text = data['content']
+                            full_text += token_text
+                            yield full_text
+
+                        # Check if generation is complete
+                        if data.get('stop', False):
+                            break
+
+                    except json.JSONDecodeError as e:
+                        # Log the error and the problematic line
+                        print(f"JSON decode error: {e}")
+                        print(f"Problematic line: {line}")
+                        continue
 
     def generate(self, prompt, state):
         output = ""

From 9f07a1f5d7bda07655e924fbbf3d9375104336ad Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 19:30:53 -0700
Subject: [PATCH 0314/1701] llama.cpp: new attempt at optimizing the
 llama-server connection

---
 modules/llama_cpp_server.py | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 34aab61393..5986ac36fb 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -147,19 +147,18 @@ def generate_with_streaming(self, prompt, state):
             full_text = ""
 
             # Process the streaming response
-            for line in response.iter_lines():
+            for line in response.iter_lines(decode_unicode=True):
                 if shared.stop_everything:
                     break
 
                 if line:
                     try:
                         # Check if the line starts with "data: " and remove it
-                        line_str = line.decode('utf-8')
-                        if line_str.startswith('data: '):
-                            line_str = line_str[6:]  # Remove the "data: " prefix
+                        if line.startswith('data: '):
+                            line = line[6:]  # Remove the "data: " prefix
 
                         # Parse the JSON data
-                        data = json.loads(line_str)
+                        data = json.loads(line)
 
                         # Extract the token content
                         if 'content' in data:

From e2e90af6cd5e839830cf1a75218777ea9c01bc53 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 20:51:18 -0700
Subject: [PATCH 0315/1701] llama.cpp: don't include --rope-freq-base in the
 launch command if null

---
 modules/llama_cpp_server.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 5986ac36fb..5071c40c39 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -250,7 +250,6 @@ def _start_server(self):
             "--ctx-size", str(shared.args.n_ctx),
             "--n-gpu-layers", str(shared.args.n_gpu_layers),
             "--batch-size", str(shared.args.batch_size),
-            "--rope-freq-base", str(shared.args.rope_freq_base),
             "--port", str(self.port),
         ]
 
@@ -276,6 +275,8 @@ def _start_server(self):
             cmd += ["--cache-type-k", shared.args.cache_type, "--cache-type-v", shared.args.cache_type]
         if shared.args.compress_pos_emb != 1:
             cmd += ["--rope-freq-scale", str(1.0 / shared.args.compress_pos_emb)]
+        if shared.args.rope_freq_base > 0:
+            cmd += ["--rope-freq-base", str(shared.args.rope_freq_base)]
 
         # Start the server with pipes for output
         self.process = subprocess.Popen(

From e2e73ed22f348092bb8915c29d2e2d68c6c8ec94 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 21:04:56 -0700
Subject: [PATCH 0316/1701] llama.cpp: new optimization attempt

---
 modules/llama_cpp_server.py | 22 ++++++++++++----------
 1 file changed, 12 insertions(+), 10 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 5071c40c39..b0fe1154f2 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -10,6 +10,7 @@
 import requests
 
 from modules import shared
+from modules.callbacks import Iteratorize
 from modules.logging_colors import logger
 
 llamacpp_valid_cache_types = {"fp16", "q8_0", "q4_0"}
@@ -118,7 +119,7 @@ def prepare_payload(self, state):
 
         return payload
 
-    def generate_with_streaming(self, prompt, state):
+    def generate(self, prompt, state, callback=None):
         url = f"http://localhost:{self.port}/completion"
         payload = self.prepare_payload(state)
 
@@ -144,7 +145,7 @@ def generate_with_streaming(self, prompt, state):
         with self.session.post(url, json=payload, stream=True) as response:
             response.raise_for_status()  # Raise an exception for HTTP errors
 
-            full_text = ""
+            output = ""
 
             # Process the streaming response
             for line in response.iter_lines(decode_unicode=True):
@@ -162,9 +163,10 @@ def generate_with_streaming(self, prompt, state):
 
                         # Extract the token content
                         if 'content' in data:
-                            token_text = data['content']
-                            full_text += token_text
-                            yield full_text
+                            text = data['content']
+                            output += text
+                            if callback:
+                                callback(output)
 
                         # Check if generation is complete
                         if data.get('stop', False):
@@ -176,13 +178,13 @@ def generate_with_streaming(self, prompt, state):
                         print(f"Problematic line: {line}")
                         continue
 
-    def generate(self, prompt, state):
-        output = ""
-        for output in self.generate_with_streaming(prompt, state):
-            pass
-
         return output
 
+    def generate_with_streaming(self, *args, **kwargs):
+        with Iteratorize(self.generate, args, kwargs, callback=None) as generator:
+            for output in generator:
+                yield output
+
     def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
         """Get the logits/probabilities for the next token after a prompt"""
         url = f"http://localhost:{self.port}/completion"

From 6589ebeca8012a991ef6a39c90fbd5e588f0c6dd Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 18 Apr 2025 21:16:21 -0700
Subject: [PATCH 0317/1701] Revert "llama.cpp: new optimization attempt"

This reverts commit e2e73ed22f348092bb8915c29d2e2d68c6c8ec94.
---
 modules/llama_cpp_server.py | 22 ++++++++++------------
 1 file changed, 10 insertions(+), 12 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index b0fe1154f2..5071c40c39 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -10,7 +10,6 @@
 import requests
 
 from modules import shared
-from modules.callbacks import Iteratorize
 from modules.logging_colors import logger
 
 llamacpp_valid_cache_types = {"fp16", "q8_0", "q4_0"}
@@ -119,7 +118,7 @@ def prepare_payload(self, state):
 
         return payload
 
-    def generate(self, prompt, state, callback=None):
+    def generate_with_streaming(self, prompt, state):
         url = f"http://localhost:{self.port}/completion"
         payload = self.prepare_payload(state)
 
@@ -145,7 +144,7 @@ def generate(self, prompt, state, callback=None):
         with self.session.post(url, json=payload, stream=True) as response:
             response.raise_for_status()  # Raise an exception for HTTP errors
 
-            output = ""
+            full_text = ""
 
             # Process the streaming response
             for line in response.iter_lines(decode_unicode=True):
@@ -163,10 +162,9 @@ def generate(self, prompt, state, callback=None):
 
                         # Extract the token content
                         if 'content' in data:
-                            text = data['content']
-                            output += text
-                            if callback:
-                                callback(output)
+                            token_text = data['content']
+                            full_text += token_text
+                            yield full_text
 
                         # Check if generation is complete
                         if data.get('stop', False):
@@ -178,12 +176,12 @@ def generate(self, prompt, state, callback=None):
                         print(f"Problematic line: {line}")
                         continue
 
-        return output
+    def generate(self, prompt, state):
+        output = ""
+        for output in self.generate_with_streaming(prompt, state):
+            pass
 
-    def generate_with_streaming(self, *args, **kwargs):
-        with Iteratorize(self.generate, args, kwargs, callback=None) as generator:
-            for output in generator:
-                yield output
+        return output
 
     def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
         """Get the logits/probabilities for the next token after a prompt"""

From 5fdebc554b7ca46afb9695babf89397635e9f91d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 19 Apr 2025 04:59:24 -0700
Subject: [PATCH 0318/1701] llama.cpp: close the connection immediately on
 'Stop'

---
 modules/llama_cpp_server.py | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 5071c40c39..3025aa7d72 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -141,16 +141,24 @@ def generate_with_streaming(self, prompt, state):
             print()
 
         # Make a direct request with streaming enabled using a context manager
-        with self.session.post(url, json=payload, stream=True) as response:
+        with self.session.post(url, json=payload, stream=True, timeout=(5, 0.1)) as response:
             response.raise_for_status()  # Raise an exception for HTTP errors
 
             full_text = ""
+            iterator = response.iter_lines(decode_unicode=True)
 
-            # Process the streaming response
-            for line in response.iter_lines(decode_unicode=True):
+            while True:
                 if shared.stop_everything:
                     break
 
+                try:
+                    line = next(iterator)
+                except requests.exceptions.Timeout:
+                    # Check stop flag again on timeout
+                    continue
+                except StopIteration:
+                    break
+
                 if line:
                     try:
                         # Check if the line starts with "data: " and remove it

From ed42154c78d7d7a63c092684deaa47d22750d796 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 19 Apr 2025 05:32:36 -0700
Subject: [PATCH 0319/1701] Revert "llama.cpp: close the connection immediately
 on 'Stop'"

This reverts commit 5fdebc554b7ca46afb9695babf89397635e9f91d.
---
 modules/llama_cpp_server.py | 14 +++-----------
 1 file changed, 3 insertions(+), 11 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 3025aa7d72..5071c40c39 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -141,24 +141,16 @@ def generate_with_streaming(self, prompt, state):
             print()
 
         # Make a direct request with streaming enabled using a context manager
-        with self.session.post(url, json=payload, stream=True, timeout=(5, 0.1)) as response:
+        with self.session.post(url, json=payload, stream=True) as response:
             response.raise_for_status()  # Raise an exception for HTTP errors
 
             full_text = ""
-            iterator = response.iter_lines(decode_unicode=True)
 
-            while True:
+            # Process the streaming response
+            for line in response.iter_lines(decode_unicode=True):
                 if shared.stop_everything:
                     break
 
-                try:
-                    line = next(iterator)
-                except requests.exceptions.Timeout:
-                    # Check stop flag again on timeout
-                    continue
-                except StopIteration:
-                    break
-
                 if line:
                     try:
                         # Check if the line starts with "data: " and remove it

From ba976d1390b7151633d693b270604517fd14e701 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 19 Apr 2025 16:35:01 -0700
Subject: [PATCH 0320/1701] llama.cpp: avoid two 'encode' calls

---
 modules/llama_cpp_server.py | 55 +++++++++++++++++++++----------------
 modules/text_generation.py  |  9 ++++--
 2 files changed, 38 insertions(+), 26 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 5071c40c39..faf6e20eaf 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -27,6 +27,8 @@ def __init__(self, model_path, server_path=None):
         self.session = requests.Session()
         self.vocabulary_size = None
         self.bos_token = ""
+        self.last_input_length = 0
+        self.last_output_length = 0
 
         # Start the server
         self._start_server()
@@ -140,6 +142,9 @@ def generate_with_streaming(self, prompt, state):
             pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
             print()
 
+        self.last_input_length = len(token_ids)
+        self.last_output_length = 0
+
         # Make a direct request with streaming enabled using a context manager
         with self.session.post(url, json=payload, stream=True) as response:
             response.raise_for_status()  # Raise an exception for HTTP errors
@@ -151,30 +156,32 @@ def generate_with_streaming(self, prompt, state):
                 if shared.stop_everything:
                     break
 
-                if line:
-                    try:
-                        # Check if the line starts with "data: " and remove it
-                        if line.startswith('data: '):
-                            line = line[6:]  # Remove the "data: " prefix
-
-                        # Parse the JSON data
-                        data = json.loads(line)
-
-                        # Extract the token content
-                        if 'content' in data:
-                            token_text = data['content']
-                            full_text += token_text
-                            yield full_text
-
-                        # Check if generation is complete
-                        if data.get('stop', False):
-                            break
-
-                    except json.JSONDecodeError as e:
-                        # Log the error and the problematic line
-                        print(f"JSON decode error: {e}")
-                        print(f"Problematic line: {line}")
-                        continue
+                if not line:
+                    continue
+
+                try:
+                    # Check if the line starts with "data: " and remove it
+                    if line.startswith('data: '):
+                        line = line[6:]  # Remove the "data: " prefix
+
+                    # Parse the JSON data
+                    data = json.loads(line)
+
+                    # Extract the token content
+                    if data.get('content', ''):
+                        full_text += data['content']
+                        self.last_output_length += 1
+                        yield full_text
+
+                    # Check if generation is complete
+                    if data.get('stop', False):
+                        break
+
+                except json.JSONDecodeError as e:
+                    # Log the error and the problematic line
+                    print(f"JSON decode error: {e}")
+                    print(f"Problematic line: {line}")
+                    continue
 
     def generate(self, prompt, state):
         output = ""
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 16aba3cbb5..675eb379c9 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -481,8 +481,13 @@ def generate_reply_custom(question, original_question, seed, state, stopping_str
         traceback.print_exc()
     finally:
         t1 = time.time()
-        original_tokens = len(encode(original_question)[0])
-        new_tokens = len(encode(original_question + reply)[0]) - original_tokens
+        if shared.args.loader == 'llama.cpp':
+            original_tokens = shared.model.last_input_length
+            new_tokens = shared.model.last_output_length
+        else:
+            original_tokens = len(encode(original_question)[0])
+            new_tokens = len(encode(original_question + reply)[0]) - original_tokens
+
         print(f'Output generated in {(t1-t0):.2f} seconds ({new_tokens/(t1-t0):.2f} tokens/s, {new_tokens} tokens, context {original_tokens}, seed {seed})')
         return
 

From 9c9df2063f61d19fa1755b2906ba6804f88c4e68 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 19 Apr 2025 16:38:02 -0700
Subject: [PATCH 0321/1701] llama.cpp: fix unicode decoding (closes #6856)

---
 modules/llama_cpp_server.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index faf6e20eaf..9c97e00ba3 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -152,7 +152,7 @@ def generate_with_streaming(self, prompt, state):
             full_text = ""
 
             # Process the streaming response
-            for line in response.iter_lines(decode_unicode=True):
+            for line in response.iter_lines():
                 if shared.stop_everything:
                     break
 
@@ -160,6 +160,8 @@ def generate_with_streaming(self, prompt, state):
                     continue
 
                 try:
+                    line = line.decode('utf-8')
+
                     # Check if the line starts with "data: " and remove it
                     if line.startswith('data: '):
                         line = line[6:]  # Remove the "data: " prefix

From b9da5c7e3a34ef1a5ae2e372db7ff2c5299d523a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 19 Apr 2025 17:36:04 -0700
Subject: [PATCH 0322/1701] Use 127.0.0.1 instead of localhost for faster
 llama.cpp on Windows

---
 modules/llama_cpp_server.py | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 9c97e00ba3..ebce987a5a 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -37,7 +37,7 @@ def encode(self, text, add_bos_token=False, **kwargs):
         if self.bos_token and text.startswith(self.bos_token):
             add_bos_token = False
 
-        url = f"http://localhost:{self.port}/tokenize"
+        url = f"http://127.0.0.1:{self.port}/tokenize"
         payload = {
             "content": text,
             "add_special": add_bos_token,
@@ -48,7 +48,7 @@ def encode(self, text, add_bos_token=False, **kwargs):
         return result.get("tokens", [])
 
     def decode(self, token_ids, **kwargs):
-        url = f"http://localhost:{self.port}/detokenize"
+        url = f"http://127.0.0.1:{self.port}/detokenize"
         payload = {
             "tokens": token_ids,
         }
@@ -121,7 +121,7 @@ def prepare_payload(self, state):
         return payload
 
     def generate_with_streaming(self, prompt, state):
-        url = f"http://localhost:{self.port}/completion"
+        url = f"http://127.0.0.1:{self.port}/completion"
         payload = self.prepare_payload(state)
 
         token_ids = self.encode(prompt, add_bos_token=state["add_bos_token"])
@@ -194,7 +194,7 @@ def generate(self, prompt, state):
 
     def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
         """Get the logits/probabilities for the next token after a prompt"""
-        url = f"http://localhost:{self.port}/completion"
+        url = f"http://127.0.0.1:{self.port}/completion"
 
         payload = self.prepare_payload(state)
         payload.update({
@@ -225,7 +225,7 @@ def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
 
     def _get_vocabulary_size(self):
         """Get and store the model's maximum context length."""
-        url = f"http://localhost:{self.port}/v1/models"
+        url = f"http://127.0.0.1:{self.port}/v1/models"
         response = self.session.get(url).json()
 
         if "data" in response and len(response["data"]) > 0:
@@ -235,7 +235,7 @@ def _get_vocabulary_size(self):
 
     def _get_bos_token(self):
         """Get and store the model's BOS token."""
-        url = f"http://localhost:{self.port}/props"
+        url = f"http://127.0.0.1:{self.port}/props"
         response = self.session.get(url).json()
         if "bos_token" in response:
             self.bos_token = response["bos_token"]
@@ -308,7 +308,7 @@ def filter_stderr(process_stderr):
         threading.Thread(target=filter_stderr, args=(self.process.stderr,), daemon=True).start()
 
         # Wait for server to be healthy
-        health_url = f"http://localhost:{self.port}/health"
+        health_url = f"http://127.0.0.1:{self.port}/health"
         start_time = time.time()
         timeout = 3600 * 8  # 8 hours
         while time.time() - start_time < timeout:

From 5ab069786befb5473142a2c03de3a3870ba4a151 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 19 Apr 2025 17:38:36 -0700
Subject: [PATCH 0323/1701] llama.cpp: add back the two encode calls (they are
 harmless now)

---
 modules/llama_cpp_server.py | 6 ------
 modules/text_generation.py  | 8 ++------
 2 files changed, 2 insertions(+), 12 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index ebce987a5a..02a56b3cd5 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -27,8 +27,6 @@ def __init__(self, model_path, server_path=None):
         self.session = requests.Session()
         self.vocabulary_size = None
         self.bos_token = ""
-        self.last_input_length = 0
-        self.last_output_length = 0
 
         # Start the server
         self._start_server()
@@ -142,9 +140,6 @@ def generate_with_streaming(self, prompt, state):
             pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
             print()
 
-        self.last_input_length = len(token_ids)
-        self.last_output_length = 0
-
         # Make a direct request with streaming enabled using a context manager
         with self.session.post(url, json=payload, stream=True) as response:
             response.raise_for_status()  # Raise an exception for HTTP errors
@@ -172,7 +167,6 @@ def generate_with_streaming(self, prompt, state):
                     # Extract the token content
                     if data.get('content', ''):
                         full_text += data['content']
-                        self.last_output_length += 1
                         yield full_text
 
                     # Check if generation is complete
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 675eb379c9..70f0344390 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -481,12 +481,8 @@ def generate_reply_custom(question, original_question, seed, state, stopping_str
         traceback.print_exc()
     finally:
         t1 = time.time()
-        if shared.args.loader == 'llama.cpp':
-            original_tokens = shared.model.last_input_length
-            new_tokens = shared.model.last_output_length
-        else:
-            original_tokens = len(encode(original_question)[0])
-            new_tokens = len(encode(original_question + reply)[0]) - original_tokens
+        original_tokens = len(encode(original_question)[0])
+        new_tokens = len(encode(original_question + reply)[0]) - original_tokens
 
         print(f'Output generated in {(t1-t0):.2f} seconds ({new_tokens/(t1-t0):.2f} tokens/s, {new_tokens} tokens, context {original_tokens}, seed {seed})')
         return

From 6ba0164c70173cb03584d5954793cd1bec7c593b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 19 Apr 2025 17:45:21 -0700
Subject: [PATCH 0324/1701] Lint

---
 modules/text_generation.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index 70f0344390..16aba3cbb5 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -483,7 +483,6 @@ def generate_reply_custom(question, original_question, seed, state, stopping_str
         t1 = time.time()
         original_tokens = len(encode(original_question)[0])
         new_tokens = len(encode(original_question + reply)[0]) - original_tokens
-
         print(f'Output generated in {(t1-t0):.2f} seconds ({new_tokens/(t1-t0):.2f} tokens/s, {new_tokens} tokens, context {original_tokens}, seed {seed})')
         return
 

From ae02ffc605d90b19a507132c09055c7ff11df52b Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Sun, 20 Apr 2025 13:33:47 -0300
Subject: [PATCH 0325/1701] Refactor the transformers loader (#6859)

---
 extensions/openai/completions.py |  86 +++-------
 extensions/openai/script.py      |   6 +-
 modules/LoRA.py                  |   4 +-
 modules/callbacks.py             |  22 ---
 modules/evaluate.py              |   7 +-
 modules/loaders.py               |  18 +-
 modules/logits.py                |  23 ++-
 modules/models.py                | 283 ++-----------------------------
 modules/models_settings.py       |  17 --
 modules/sampler_hijack.py        |   2 +-
 modules/shared.py                |   4 +-
 modules/text_generation.py       | 131 +++++++-------
 modules/torch_utils.py           |  37 ++++
 modules/training.py              |  17 +-
 modules/transformers_loader.py   | 281 ++++++++++++++++++++++++++++++
 modules/ui.py                    |  10 --
 modules/ui_model_menu.py         |  40 +----
 server.py                        |   6 +-
 18 files changed, 465 insertions(+), 529 deletions(-)
 create mode 100644 modules/torch_utils.py
 create mode 100644 modules/transformers_loader.py

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index f1a6064597..75e2cc118f 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -7,10 +7,7 @@
 
 import requests
 import tiktoken
-import torch
-import torch.nn.functional as F
 from PIL import Image
-from transformers import LogitsProcessor, LogitsProcessorList
 
 from extensions.openai.errors import InvalidRequestError
 from extensions.openai.utils import debug_msg
@@ -22,54 +19,7 @@
     load_instruction_template_memoized
 )
 from modules.presets import load_preset_memoized
-from modules.text_generation import (
-    decode,
-    encode,
-    generate_reply,
-    get_reply_from_output_ids
-)
-
-
-class LogitsBiasProcessor(LogitsProcessor):
-    def __init__(self, logit_bias={}):
-        self.logit_bias = logit_bias
-        if self.logit_bias:
-            self.keys = list([int(key) for key in self.logit_bias.keys()])
-            values = [self.logit_bias[str(key)] for key in self.keys]
-            self.values = torch.tensor(values, dtype=torch.float, device=shared.model.device)
-            debug_msg(f"{self})")
-
-    def __call__(self, input_ids: torch.LongTensor, logits: torch.FloatTensor) -> torch.FloatTensor:
-        if self.logit_bias:
-            debug_msg(logits[0, self.keys], " + ", self.values)
-            logits[0, self.keys] += self.values
-            debug_msg(" --> ", logits[0, self.keys])
-            debug_msg(" max/min ", float(torch.max(logits[0])), float(torch.min(logits[0])))
-
-        return logits
-
-    def __repr__(self):
-        return f"<{self.__class__.__name__}(logit_bias={self.logit_bias})>"
-
-
-class LogprobProcessor(LogitsProcessor):
-    def __init__(self, logprobs=None):
-        self.logprobs = logprobs
-        self.token_alternatives = {}
-
-    def __call__(self, input_ids: torch.LongTensor, logits: torch.FloatTensor) -> torch.FloatTensor:
-        if self.logprobs is not None:  # 0-5
-            log_e_probabilities = F.log_softmax(logits, dim=1)
-            top_values, top_indices = torch.topk(log_e_probabilities, k=self.logprobs + 1)
-            top_tokens = [get_reply_from_output_ids([tok]) for tok in top_indices[0]]
-            top_probs = [float(x) for x in top_values[0]]
-            self.token_alternatives = dict(zip(top_tokens, top_probs))
-            debug_msg(repr(self))
-
-        return logits
-
-    def __repr__(self):
-        return f"<{self.__class__.__name__}(logprobs={self.logprobs}, token_alternatives={self.token_alternatives})>"
+from modules.text_generation import decode, encode, generate_reply
 
 
 def convert_logprobs_to_tiktoken(model, logprobs):
@@ -107,21 +57,29 @@ def process_parameters(body, is_legacy=False):
         elif isinstance(body['stop'], list):
             generate_params['custom_stopping_strings'] = body['stop']
 
-    logits_processor = []
-    logit_bias = body.get('logit_bias', None)
-    if logit_bias:  # {str: float, ...}
-        logits_processor = [LogitsBiasProcessor(logit_bias)]
+    if shared.args.loader != 'llama.cpp':
+        from transformers import LogitsProcessorList
 
-    logprobs = None  # coming to chat eventually
-    if 'logprobs' in body:
-        logprobs = body.get('logprobs', 0)  # maybe cap at topk? don't clamp 0-5.
-        generate_params['logprob_proc'] = LogprobProcessor(logprobs)
-        logits_processor.extend([generate_params['logprob_proc']])
-    else:
-        logprobs = None
+        from modules.transformers_loader import (
+            LogitsBiasProcessor,
+            LogprobProcessor
+        )
+
+        logits_processor = []
+        logit_bias = body.get('logit_bias', None)
+        if logit_bias:  # {str: float, ...}
+            logits_processor = [LogitsBiasProcessor(logit_bias)]
+
+        logprobs = None  # coming to chat eventually
+        if 'logprobs' in body:
+            logprobs = body.get('logprobs', 0)  # maybe cap at topk? don't clamp 0-5.
+            generate_params['logprob_proc'] = LogprobProcessor(logprobs)
+            logits_processor.extend([generate_params['logprob_proc']])
+        else:
+            logprobs = None
 
-    if logits_processor:  # requires logits_processor support
-        generate_params['logits_processor'] = LogitsProcessorList(logits_processor)
+        if logits_processor:  # requires logits_processor support
+            generate_params['logits_processor'] = LogitsProcessorList(logits_processor)
 
     return generate_params
 
diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index f23caf9b15..2ce692c7fb 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -16,11 +16,9 @@
 from sse_starlette import EventSourceResponse
 
 import extensions.openai.completions as OAIcompletions
-import extensions.openai.embeddings as OAIembeddings
 import extensions.openai.images as OAIimages
 import extensions.openai.logits as OAIlogits
 import extensions.openai.models as OAImodels
-import extensions.openai.moderations as OAImoderations
 from extensions.openai.errors import ServiceUnavailableError
 from extensions.openai.tokens import token_count, token_decode, token_encode
 from extensions.openai.utils import _start_cloudflared
@@ -211,6 +209,8 @@ async def handle_image_generation(request: Request):
 
 @app.post("/v1/embeddings", response_model=EmbeddingsResponse, dependencies=check_key)
 async def handle_embeddings(request: Request, request_data: EmbeddingsRequest):
+    import extensions.openai.embeddings as OAIembeddings
+
     input = request_data.input
     if not input:
         raise HTTPException(status_code=400, detail="Missing required argument input")
@@ -224,6 +224,8 @@ async def handle_embeddings(request: Request, request_data: EmbeddingsRequest):
 
 @app.post("/v1/moderations", dependencies=check_key)
 async def handle_moderations(request: Request):
+    import extensions.openai.moderations as OAImoderations
+
     body = await request.json()
     input = body["input"]
     if not input:
diff --git a/modules/LoRA.py b/modules/LoRA.py
index 1f4883e25b..a9e9a895e5 100644
--- a/modules/LoRA.py
+++ b/modules/LoRA.py
@@ -2,7 +2,6 @@
 
 import modules.shared as shared
 from modules.logging_colors import logger
-from modules.models import get_device
 
 
 def add_lora_to_model(lora_names):
@@ -47,9 +46,10 @@ def add_lora_exllamav2(lora_names):
 
 
 def add_lora_transformers(lora_names):
-
     from peft import PeftModel
 
+    from modules.torch_utils import get_device
+
     prior_set = set(shared.lora_names)
     added_set = set(lora_names) - prior_set
     removed_set = prior_set - set(lora_names)
diff --git a/modules/callbacks.py b/modules/callbacks.py
index 0f918f3d54..082365f098 100644
--- a/modules/callbacks.py
+++ b/modules/callbacks.py
@@ -2,9 +2,6 @@
 from queue import Queue
 from threading import Thread
 
-import torch
-import transformers
-
 import modules.shared as shared
 
 
@@ -12,25 +9,6 @@ class StopNowException(Exception):
     pass
 
 
-class _StopEverythingStoppingCriteria(transformers.StoppingCriteria):
-    def __init__(self):
-        transformers.StoppingCriteria.__init__(self)
-
-    def __call__(self, input_ids: torch.LongTensor, _scores: torch.FloatTensor) -> bool:
-        return shared.stop_everything
-
-
-class Stream(transformers.StoppingCriteria):
-    def __init__(self, callback_func=None):
-        self.callback_func = callback_func
-
-    def __call__(self, input_ids, scores) -> bool:
-        if self.callback_func is not None:
-            self.callback_func(input_ids[0])
-
-        return False
-
-
 class Iteratorize:
 
     """
diff --git a/modules/evaluate.py b/modules/evaluate.py
index dd7ef9a70b..bbf27da1fa 100644
--- a/modules/evaluate.py
+++ b/modules/evaluate.py
@@ -2,13 +2,12 @@
 from pathlib import Path
 
 import pandas as pd
-import torch
 from datasets import load_dataset
 from tqdm import tqdm
 
 from modules import shared
 from modules.logging_colors import logger
-from modules.models import clear_torch_cache, load_model, unload_model
+from modules.models import load_model, unload_model
 from modules.models_settings import get_model_metadata, update_model_parameters
 from modules.text_generation import encode
 
@@ -39,6 +38,10 @@ def calculate_perplexity(models, input_dataset, stride, _max_length):
     https://huggingface.co/docs/transformers/perplexity#calculating-ppl-with-fixedlength-models
     '''
 
+    import torch
+
+    from modules.torch_utils import clear_torch_cache
+
     if shared.args.loader == "llama.cpp":
         logger.error("Perplexity evaluation is not implemented for the llama.cpp loader.")
         raise ValueError
diff --git a/modules/loaders.py b/modules/loaders.py
index aaf4c8ed78..f69f17203c 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -3,11 +3,9 @@
 
 import gradio as gr
 
-from modules import shared
-
 loaders_and_params = OrderedDict({
     'Transformers': [
-        'gpu_memory',
+        'gpu_split',
         'cpu_memory',
         'alpha_value',
         'compress_pos_emb',
@@ -17,7 +15,6 @@
         'load_in_4bit',
         'torch_compile',
         'use_flash_attention_2',
-        'auto_devices',
         'cpu',
         'disk',
         'use_double_quant',
@@ -346,10 +343,6 @@ def blacklist_samplers(loader, dynamic_temperature):
     return output
 
 
-def get_gpu_memory_keys():
-    return [k for k in shared.gradio if k.startswith('gpu_memory')]
-
-
 @functools.cache
 def get_all_params():
     all_params = set()
@@ -357,11 +350,6 @@ def get_all_params():
         for el in loaders_and_params[k]:
             all_params.add(el)
 
-    if 'gpu_memory' in all_params:
-        all_params.remove('gpu_memory')
-        for k in get_gpu_memory_keys():
-            all_params.add(k)
-
     return sorted(all_params)
 
 
@@ -371,8 +359,4 @@ def make_loader_params_visible(loader):
     if loader in loaders_and_params:
         params = loaders_and_params[loader]
 
-        if 'gpu_memory' in params:
-            params.remove('gpu_memory')
-            params += get_gpu_memory_keys()
-
     return [gr.update(visible=True) if k in params else gr.update(visible=False) for k in all_params]
diff --git a/modules/logits.py b/modules/logits.py
index 020e742458..9a4243ff8a 100644
--- a/modules/logits.py
+++ b/modules/logits.py
@@ -2,11 +2,10 @@
 import traceback
 
 import numpy as np
-import torch
 
-from modules import models, sampler_hijack, shared
+from modules import models, shared
 from modules.logging_colors import logger
-from modules.models import get_device, load_model
+from modules.models import load_model
 from modules.text_generation import generate_reply
 
 global_scores = None
@@ -38,18 +37,16 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
         logger.error("No model is loaded! Select one in the Model tab.")
         return 'Error: No model is loaded1 Select one in the Model tab.', previous
 
-    is_non_hf_exllamav2 = shared.model.__class__.__name__ == 'Exllamav2Model'
-    is_llamacpp = shared.model.__class__.__name__ == 'LlamaServer'
-
-    if is_llamacpp:
+    # llama.cpp case
+    if shared.model.__class__.__name__ == 'LlamaServer':
         logprobs = shared.model.get_logits(prompt, state, n_probs=top_logits, use_samplers=use_samplers)
+
         if return_dict:
             output = {}
             for entry in logprobs:
                 token = repr(entry['token'])
                 prob = entry['prob'] if use_samplers else np.exp(entry['logprob'])
                 output[token] = prob
-
             return output
         else:
             output = ''
@@ -57,9 +54,17 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
                 token = repr(entry['token'])
                 prob = entry['prob'] if use_samplers else np.exp(entry['logprob'])
                 output += f"{prob:.5f}  -  {token}\n"
-
             return output, previous
+
+    # All other model types
     else:
+        import torch
+
+        from modules import sampler_hijack
+        from modules.torch_utils import get_device
+
+        is_non_hf_exllamav2 = shared.model.__class__.__name__ == 'Exllamav2Model'
+
         if not use_samplers:
             state = {'stream': True}
 
diff --git a/modules/models.py b/modules/models.py
index 48d92bd539..c4dfa1497e 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -1,61 +1,10 @@
-import gc
-import os
-import pprint
-import re
 import time
 from pathlib import Path
 
-import torch
-import transformers
-from accelerate import infer_auto_device_map, init_empty_weights
-from accelerate.utils import (
-    is_ccl_available,
-    is_npu_available,
-    is_xpu_available
-)
-from transformers import (
-    AutoConfig,
-    AutoModel,
-    AutoModelForCausalLM,
-    AutoModelForSeq2SeqLM,
-    AutoTokenizer,
-    BitsAndBytesConfig,
-    is_torch_npu_available,
-    is_torch_xpu_available
-)
-
 import modules.shared as shared
 from modules.logging_colors import logger
 from modules.models_settings import get_model_metadata
 
-transformers.logging.set_verbosity_error()
-
-local_rank = None
-if shared.args.deepspeed:
-    import deepspeed
-    from transformers.integrations.deepspeed import (
-        HfDeepSpeedConfig,
-        is_deepspeed_zero3_enabled
-    )
-
-    from modules.deepspeed_parameters import generate_ds_config
-
-    # Distributed setup
-    local_rank = shared.args.local_rank if shared.args.local_rank is not None else int(os.getenv("LOCAL_RANK", "0"))
-    world_size = int(os.getenv("WORLD_SIZE", "1"))
-    if is_xpu_available() and is_ccl_available():
-        torch.xpu.set_device(local_rank)
-        deepspeed.init_distributed(backend="ccl")
-    elif is_npu_available():
-        torch.npu.set_device(local_rank)
-        deepspeed.init_distributed(dist_backend="hccl")
-    else:
-        torch.cuda.set_device(local_rank)
-        deepspeed.init_distributed()
-    ds_config = generate_ds_config(shared.args.bf16, 1 * world_size, shared.args.nvme_offload_dir)
-    dschf = HfDeepSpeedConfig(ds_config)  # Keep this object alive for the Transformers integration
-
-
 last_generation_time = time.time()
 
 
@@ -66,8 +15,8 @@ def load_model(model_name, loader=None):
     shared.is_seq2seq = False
     shared.model_name = model_name
     load_func_map = {
-        'Transformers': huggingface_loader,
         'llama.cpp': llama_cpp_server_loader,
+        'Transformers': transformers_loader,
         'ExLlamav3_HF': ExLlamav3_HF_loader,
         'ExLlamav2_HF': ExLlamav2_HF_loader,
         'ExLlamav2': ExLlamav2_loader,
@@ -86,7 +35,6 @@ def load_model(model_name, loader=None):
                 raise ValueError
 
     shared.args.loader = loader
-    clear_torch_cache()
     output = load_func_map[loader](model_name)
     if type(output) is tuple:
         model, tokenizer = output
@@ -95,6 +43,7 @@ def load_model(model_name, loader=None):
         if model is None:
             return None, None
         else:
+            from modules.transformers_loader import load_tokenizer
             tokenizer = load_tokenizer(model_name)
 
     shared.settings.update({k: v for k, v in metadata.items() if k in shared.settings})
@@ -110,163 +59,6 @@ def load_model(model_name, loader=None):
     return model, tokenizer
 
 
-def load_tokenizer(model_name, tokenizer_dir=None):
-    if tokenizer_dir:
-        path_to_model = Path(tokenizer_dir)
-    else:
-        path_to_model = Path(f"{shared.args.model_dir}/{model_name}/")
-
-    tokenizer = None
-    if path_to_model.exists():
-        if shared.args.no_use_fast:
-            logger.info('Loading the tokenizer with use_fast=False.')
-
-        tokenizer = AutoTokenizer.from_pretrained(
-            path_to_model,
-            trust_remote_code=shared.args.trust_remote_code,
-            use_fast=not shared.args.no_use_fast
-        )
-
-    return tokenizer
-
-
-def huggingface_loader(model_name):
-    path_to_model = Path(f'{shared.args.model_dir}/{model_name}')
-    params = {
-        'low_cpu_mem_usage': True,
-        'torch_dtype': torch.bfloat16 if shared.args.bf16 else torch.float16,
-    }
-
-    if shared.args.trust_remote_code:
-        params['trust_remote_code'] = True
-
-    if shared.args.use_flash_attention_2:
-        params['use_flash_attention_2'] = True
-
-    if shared.args.force_safetensors:
-        params['force_safetensors'] = True
-
-    if shared.args.use_eager_attention:
-        params['attn_implementation'] = 'eager'
-
-    config = AutoConfig.from_pretrained(path_to_model, trust_remote_code=shared.args.trust_remote_code)
-
-    if 'chatglm' in model_name.lower():
-        LoaderClass = AutoModel
-    else:
-        if config.to_dict().get('is_encoder_decoder', False):
-            LoaderClass = AutoModelForSeq2SeqLM
-            shared.is_seq2seq = True
-        else:
-            LoaderClass = AutoModelForCausalLM
-
-    # Determine if we should use default loading
-    should_use_default_loading = not any([
-        shared.args.cpu,
-        shared.args.load_in_8bit,
-        shared.args.load_in_4bit,
-        shared.args.auto_devices,
-        shared.args.disk,
-        shared.args.deepspeed,
-        shared.args.gpu_memory is not None,
-        shared.args.cpu_memory is not None,
-        shared.args.compress_pos_emb > 1,
-        shared.args.alpha_value > 1,
-    ])
-
-    # Load the model without any special settings
-    if should_use_default_loading:
-        logger.info("TRANSFORMERS_PARAMS=")
-        pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(params)
-        print()
-
-        model = LoaderClass.from_pretrained(path_to_model, **params)
-        if not (hasattr(model, 'is_loaded_in_4bit') and model.is_loaded_in_4bit):
-            device = get_device()
-            if device:
-                model = model.to(device)
-
-    # DeepSpeed ZeRO-3
-    elif shared.args.deepspeed:
-        model = LoaderClass.from_pretrained(
-            path_to_model,
-            torch_dtype=params['torch_dtype'],
-            trust_remote_code=params.get('trust_remote_code')
-        )
-
-        model = deepspeed.initialize(
-            model=model,
-            config_params=ds_config,
-            model_parameters=None,
-            optimizer=None,
-            lr_scheduler=None
-        )[0]
-
-        model.module.eval()  # Inference
-        logger.info(f'DeepSpeed ZeRO-3 is enabled: {is_deepspeed_zero3_enabled()}')
-
-    # Load with quantization and/or offloading
-    else:
-        if not any((shared.args.cpu, torch.cuda.is_available(), is_xpu_available(), torch.backends.mps.is_available())):
-            logger.warning('torch.cuda.is_available() and is_xpu_available() returned False. This means that no GPU has been detected. Falling back to CPU mode.')
-            shared.args.cpu = True
-
-        if shared.args.cpu:
-            params['torch_dtype'] = torch.float32
-        else:
-            params['device_map'] = 'auto'
-            if x := get_max_memory_dict():
-                params['max_memory'] = x
-
-            if shared.args.load_in_4bit:
-                # See https://github.com/huggingface/transformers/pull/23479/files
-                # and https://huggingface.co/blog/4bit-transformers-bitsandbytes
-                quantization_config_params = {
-                    'load_in_4bit': True,
-                    'bnb_4bit_compute_dtype': eval(f"torch.{shared.args.compute_dtype}") if shared.args.compute_dtype in ["bfloat16", "float16", "float32"] else None,
-                    'bnb_4bit_quant_type': shared.args.quant_type,
-                    'bnb_4bit_use_double_quant': shared.args.use_double_quant,
-                    'llm_int8_enable_fp32_cpu_offload': True
-                }
-                params['quantization_config'] = BitsAndBytesConfig(**quantization_config_params)
-
-            elif shared.args.load_in_8bit:
-                if shared.args.auto_devices or shared.args.gpu_memory:
-                    params['quantization_config'] = BitsAndBytesConfig(load_in_8bit=True, llm_int8_enable_fp32_cpu_offload=True)
-                else:
-                    params['quantization_config'] = BitsAndBytesConfig(load_in_8bit=True)
-
-                if params.get('max_memory') is not None:
-                    with init_empty_weights():
-                        model = LoaderClass.from_config(config, trust_remote_code=params.get('trust_remote_code'))
-
-                    model.tie_weights()
-                    params['device_map'] = infer_auto_device_map(
-                        model,
-                        dtype=torch.int8,
-                        max_memory=params.get('max_memory'),
-                        no_split_module_classes=model._no_split_modules
-                    )
-
-            if shared.args.disk:
-                params['offload_folder'] = shared.args.disk_cache_dir
-
-        if shared.args.compress_pos_emb > 1:
-            params['rope_scaling'] = {'type': 'linear', 'factor': shared.args.compress_pos_emb}
-        elif shared.args.alpha_value > 1:
-            params['rope_scaling'] = {'type': 'dynamic', 'factor': shared.args.alpha_value}
-
-        logger.info("TRANSFORMERS_PARAMS=")
-        pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(params)
-        print()
-        model = LoaderClass.from_pretrained(path_to_model, **params)
-
-    if shared.args.torch_compile:
-        model = torch.compile(model)
-
-    return model
-
-
 def llama_cpp_server_loader(model_name):
     from modules.llama_cpp_server import LlamaServer
 
@@ -284,6 +76,11 @@ def llama_cpp_server_loader(model_name):
         logger.error(f"Error loading the model with llama.cpp: {str(e)}")
 
 
+def transformers_loader(model_name):
+    from modules.transformers_loader import load_model_HF
+    return load_model_HF(model_name)
+
+
 def ExLlamav3_HF_loader(model_name):
     from modules.exllamav3_hf import Exllamav3HF
 
@@ -328,71 +125,15 @@ def TensorRT_LLM_loader(model_name):
     return model
 
 
-def get_max_memory_dict():
-    max_memory = {}
-    max_cpu_memory = shared.args.cpu_memory.strip() if shared.args.cpu_memory is not None else '99GiB'
-    if shared.args.gpu_memory:
-        memory_map = list(map(lambda x: x.strip(), shared.args.gpu_memory))
-        for i in range(len(memory_map)):
-            max_memory[i] = f'{memory_map[i]}GiB' if not re.match('.*ib$', memory_map[i].lower()) else memory_map[i]
-
-        max_memory['cpu'] = f'{max_cpu_memory}GiB' if not re.match('.*ib$', max_cpu_memory.lower()) else max_cpu_memory
-
-    # If --auto-devices is provided standalone, try to get a reasonable value
-    # for the maximum memory of device :0
-    elif shared.args.auto_devices:
-        if is_xpu_available():
-            total_mem = (torch.xpu.get_device_properties(0).total_memory / (1024 * 1024))
-        else:
-            total_mem = (torch.cuda.get_device_properties(0).total_memory / (1024 * 1024))
-
-        suggestion = round((total_mem - 1000) / 1000) * 1000
-        if total_mem - suggestion < 800:
-            suggestion -= 1000
-
-        suggestion = int(round(suggestion / 1000))
-        logger.warning(f"Auto-assiging --gpu-memory {suggestion} for your GPU to try to prevent out-of-memory errors. You can manually set other values.")
-        max_memory[0] = f'{suggestion}GiB'
-        max_memory['cpu'] = f'{max_cpu_memory}GiB' if not re.match('.*ib$', max_cpu_memory.lower()) else max_cpu_memory
-
-    return max_memory if len(max_memory) > 0 else None
-
-
-def get_device():
-    if torch.cuda.is_available():
-        return torch.device('cuda')
-    elif shared.args.deepspeed:
-        import deepspeed
-        return deepspeed.get_accelerator().current_device_name()
-    elif torch.backends.mps.is_available():
-        return torch.device('mps')
-    elif is_torch_xpu_available():
-        return torch.device('xpu:0')
-    elif is_torch_npu_available():
-        return torch.device('npu:0')
-    else:
-        return None
-
-
-def clear_torch_cache():
-    gc.collect()
-    if not shared.args.cpu:
-        if torch.cuda.is_available():
-            torch.cuda.empty_cache()
-        elif is_xpu_available():
-            torch.xpu.empty_cache()
-        elif is_npu_available():
-            torch.npu.empty_cache()
-        elif torch.backends.mps.is_available():
-            if hasattr(torch.backends.mps, 'empty_cache'):
-                torch.backends.mps.empty_cache()
-
-
 def unload_model(keep_model_name=False):
+    is_llamacpp = (shared.model.__class__.__name__ == 'LlamaServer')
+
     shared.model = shared.tokenizer = None
     shared.lora_names = []
     shared.model_dirty_from_training = False
-    clear_torch_cache()
+    if not is_llamacpp:
+        from modules.torch_utils import clear_torch_cache
+        clear_torch_cache()
 
     if not keep_model_name:
         shared.model_name = 'None'
diff --git a/modules/models_settings.py b/modules/models_settings.py
index 0af89d2cc1..f5f3be7a2b 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -188,17 +188,12 @@ def update_model_parameters(state, initial=False):
     UI: update the command-line arguments based on the interface values
     '''
     elements = ui.list_model_elements()  # the names of the parameters
-    gpu_memories = []
 
     for i, element in enumerate(elements):
         if element not in state:
             continue
 
         value = state[element]
-        if element.startswith('gpu_memory'):
-            gpu_memories.append(value)
-            continue
-
         if initial and element in shared.provided_arguments:
             continue
 
@@ -211,18 +206,6 @@ def update_model_parameters(state, initial=False):
 
         setattr(shared.args, element, value)
 
-    found_positive = False
-    for i in gpu_memories:
-        if i > 0:
-            found_positive = True
-            break
-
-    if not (initial and vars(shared.args)['gpu_memory'] != vars(shared.args_defaults)['gpu_memory']):
-        if found_positive:
-            shared.args.gpu_memory = [f"{i}MiB" for i in gpu_memories]
-        else:
-            shared.args.gpu_memory = None
-
 
 def apply_model_settings_to_state(model, state):
     '''
diff --git a/modules/sampler_hijack.py b/modules/sampler_hijack.py
index e6883289e4..ee871a6eac 100644
--- a/modules/sampler_hijack.py
+++ b/modules/sampler_hijack.py
@@ -13,7 +13,7 @@
 
 from modules import shared
 from modules.logging_colors import logger
-from modules.models import get_device
+from modules.torch_utils import get_device
 
 global_scores = None
 
diff --git a/modules/shared.py b/modules/shared.py
index 641b3f4843..fd0f226ea5 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -91,9 +91,7 @@
 # Transformers/Accelerate
 group = parser.add_argument_group('Transformers/Accelerate')
 group.add_argument('--cpu', action='store_true', help='Use the CPU to generate text. Warning: Training on CPU is extremely slow.')
-group.add_argument('--auto-devices', action='store_true', help='Automatically split the model across the available GPU(s) and CPU.')
-group.add_argument('--gpu-memory', type=str, nargs='+', help='Maximum GPU memory in GiB to be allocated per GPU. Example: --gpu-memory 10 for a single GPU, --gpu-memory 10 5 for two GPUs. You can also set values in MiB like --gpu-memory 3500MiB.')
-group.add_argument('--cpu-memory', type=str, help='Maximum CPU memory in GiB to allocate for offloaded weights. Same as above.')
+group.add_argument('--cpu-memory', type=float, default=0, help='Maximum CPU memory in GiB. Use this for CPU offloading.')
 group.add_argument('--disk', action='store_true', help='If the model is too large for your GPU(s) and CPU combined, send the remaining layers to the disk.')
 group.add_argument('--disk-cache-dir', type=str, default='cache', help='Directory to save the disk cache to. Defaults to "cache".')
 group.add_argument('--load-in-8bit', action='store_true', help='Load the model with 8-bit precision (using bitsandbytes).')
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 16aba3cbb5..585e4d9d30 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -7,33 +7,18 @@
 import traceback
 
 import numpy as np
-import torch
-import transformers
-from transformers import (
-    LogitsProcessorList,
-    is_torch_npu_available,
-    is_torch_xpu_available
-)
 
 import modules.shared as shared
-from modules import models, sampler_hijack
-from modules.callbacks import (
-    Iteratorize,
-    Stream,
-    _StopEverythingStoppingCriteria
-)
+from modules import models
+from modules.callbacks import Iteratorize
 from modules.extensions import apply_extensions
-from modules.grammar.grammar_utils import initialize_grammar
-from modules.grammar.logits_process import GrammarConstrainedLogitsProcessor
 from modules.html_generator import generate_basic_html
 from modules.logging_colors import logger
-from modules.models import clear_torch_cache, get_device, load_model
-
-sampler_hijack.hijack_samplers()
 
 
 def generate_reply(*args, **kwargs):
     if shared.args.idle_timeout > 0 and shared.model is None and shared.model_name not in [None, 'None']:
+        from modules.models import load_model
         shared.model, shared.tokenizer = load_model(shared.model_name)
 
     shared.generation_lock.acquire()
@@ -46,7 +31,6 @@ def generate_reply(*args, **kwargs):
 
 
 def _generate_reply(question, state, stopping_strings=None, is_chat=False, escape_html=False, for_ui=False):
-
     # Find the appropriate generation function
     generate_func = apply_extensions('custom_generate_reply')
     if generate_func is None:
@@ -80,7 +64,6 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
             all_stop_strings += st
 
     shared.stop_everything = False
-    seed = set_manual_seed(state['seed'])
     last_update = -1
     reply = ''
     is_stream = state['stream']
@@ -93,7 +76,7 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
         min_update_interval = 1 / state['max_updates_second']
 
     # Generate
-    for reply in generate_func(question, original_question, seed, state, stopping_strings, is_chat=is_chat):
+    for reply in generate_func(question, original_question, state, stopping_strings, is_chat=is_chat):
         reply, stop_found = apply_stopping_strings(reply, all_stop_strings)
         if escape_html:
             reply = html.escape(reply)
@@ -132,44 +115,55 @@ def encode(prompt, add_special_tokens=True, add_bos_token=True, truncation_lengt
     if shared.tokenizer is None:
         raise ValueError('No tokenizer is loaded')
 
-    if shared.model.__class__.__name__ in ['LlamaServer', 'Exllamav2Model', 'TensorRTLLMModel']:
-        if shared.model.__class__.__name__ == 'LlamaServer':
-            input_ids = shared.tokenizer.encode(str(prompt), add_bos_token=add_bos_token)
-        else:
-            input_ids = shared.tokenizer.encode(str(prompt))
+    # llama.cpp case
+    if shared.model.__class__.__name__ == 'LlamaServer':
+        input_ids = shared.tokenizer.encode(str(prompt), add_bos_token=add_bos_token)
+        input_ids = np.array(input_ids).reshape(1, len(input_ids))
 
-        if shared.model.__class__.__name__ not in ['Exllamav2Model']:
-            input_ids = np.array(input_ids).reshape(1, len(input_ids))
-    else:
-        input_ids = shared.tokenizer.encode(str(prompt), return_tensors='pt', add_special_tokens=add_special_tokens)
-
-        if hasattr(shared.tokenizer, 'bos_token_id') and shared.tokenizer.bos_token_id is not None:
-            if add_bos_token:
-                if (len(input_ids[0]) > 0 and input_ids[0][0] != shared.tokenizer.bos_token_id) or len(input_ids[0]) == 0:
-                    # Add a missing bos token (it may not have been added due to faulty model metadata)
-                    bos_tensor = torch.tensor([[shared.tokenizer.bos_token_id]])
-                    input_ids = torch.cat((bos_tensor, input_ids), 1)
-
-                # Prevent double bos token due to jinja templates with  somewhere
-                while len(input_ids[0]) > 1 and input_ids[0][0] == shared.tokenizer.bos_token_id and input_ids[0][1] == shared.tokenizer.bos_token_id:
-                    input_ids = input_ids[:, 1:]
-            else:
-                # Remove any bos token that may have been added
-                while len(input_ids[0]) > 0 and input_ids[0][0] == shared.tokenizer.bos_token_id:
-                    input_ids = input_ids[:, 1:]
-
-    # Handling truncation
-    if truncation_length is not None:
-        input_ids = input_ids[:, -truncation_length:]
+        if truncation_length is not None:
+            input_ids = input_ids[:, -truncation_length:]
 
-    if shared.model.__class__.__name__ in ['LlamaServer', 'Exllamav2Model', 'TensorRTLLMModel'] or shared.args.cpu:
         return input_ids
+
+    # All other model types
     else:
-        device = get_device()
-        if device:
-            return input_ids.to(device)
+        import torch
 
-        return input_ids
+        from modules.torch_utils import get_device
+
+        if shared.model.__class__.__name__ in ['Exllamav2Model', 'TensorRTLLMModel']:
+            input_ids = shared.tokenizer.encode(str(prompt))
+            if shared.model.__class__.__name__ != 'Exllamav2Model':
+                input_ids = np.array(input_ids).reshape(1, len(input_ids))
+        else:
+            input_ids = shared.tokenizer.encode(str(prompt), return_tensors='pt', add_special_tokens=add_special_tokens)
+
+            if hasattr(shared.tokenizer, 'bos_token_id') and shared.tokenizer.bos_token_id is not None:
+                if add_bos_token:
+                    # Add BOS token if missing
+                    if (len(input_ids[0]) > 0 and input_ids[0][0] != shared.tokenizer.bos_token_id) or len(input_ids[0]) == 0:
+                        bos_tensor = torch.tensor([[shared.tokenizer.bos_token_id]])
+                        input_ids = torch.cat((bos_tensor, input_ids), 1)
+
+                    # Prevent double BOS tokens from jinja templates
+                    while len(input_ids[0]) > 1 and input_ids[0][0] == shared.tokenizer.bos_token_id and input_ids[0][1] == shared.tokenizer.bos_token_id:
+                        input_ids = input_ids[:, 1:]
+                else:
+                    # Remove BOS tokens when not wanted
+                    while len(input_ids[0]) > 0 and input_ids[0][0] == shared.tokenizer.bos_token_id:
+                        input_ids = input_ids[:, 1:]
+
+        if truncation_length is not None:
+            input_ids = input_ids[:, -truncation_length:]
+
+        if shared.model.__class__.__name__ in ['Exllamav2Model', 'TensorRTLLMModel'] or shared.args.cpu:
+            return input_ids
+        else:
+            device = get_device()
+            if device:
+                return input_ids.to(device)
+
+            return input_ids
 
 
 def decode(output_ids, skip_special_tokens=True):
@@ -221,6 +215,9 @@ def formatted_outputs(reply, model_name):
 
 
 def set_manual_seed(seed):
+    import torch
+    from transformers import is_torch_npu_available, is_torch_xpu_available
+
     seed = int(seed)
     if seed == -1:
         seed = random.randint(1, 2**31)
@@ -285,10 +282,26 @@ def get_reply_from_output_ids(output_ids, state=None, starting_from=0):
     return reply
 
 
-def generate_reply_HF(question, original_question, seed, state, stopping_strings=None, is_chat=False):
+def generate_reply_HF(question, original_question, state, stopping_strings=None, is_chat=False):
+    import torch
+    import transformers
+    from transformers import LogitsProcessorList
+
+    from modules.grammar.grammar_utils import initialize_grammar
+    from modules.grammar.logits_process import (
+        GrammarConstrainedLogitsProcessor
+    )
+    from modules.torch_utils import clear_torch_cache, get_device
+    from modules.transformers_loader import (
+        Stream,
+        _StopEverythingStoppingCriteria
+    )
+
     if shared.args.loader == 'Transformers':
         clear_torch_cache()
 
+    seed = set_manual_seed(state['seed'])
+
     generate_params = {}
     for k in [
         'temperature',
@@ -458,11 +471,15 @@ def generate_with_streaming(**kwargs):
         return
 
 
-def generate_reply_custom(question, original_question, seed, state, stopping_strings=None, is_chat=False):
+def generate_reply_custom(question, original_question, state, stopping_strings=None, is_chat=False):
     """
     For models that do not use the transformers library for sampling
     """
-    seed = set_manual_seed(state['seed'])
+
+    seed = state['seed']
+    if shared.args.loader != 'llama.cpp':
+        print(shared.args.loader)
+        seed = set_manual_seed(seed)
 
     t0 = time.time()
     reply = ''
diff --git a/modules/torch_utils.py b/modules/torch_utils.py
new file mode 100644
index 0000000000..ad9b26ade7
--- /dev/null
+++ b/modules/torch_utils.py
@@ -0,0 +1,37 @@
+import gc
+
+import torch
+from accelerate.utils import is_npu_available, is_xpu_available
+from transformers import is_torch_npu_available, is_torch_xpu_available
+
+from modules import shared
+
+
+def get_device():
+    if torch.cuda.is_available():
+        return torch.device('cuda')
+    elif shared.args.deepspeed:
+        import deepspeed
+        return deepspeed.get_accelerator().current_device_name()
+    elif torch.backends.mps.is_available():
+        return torch.device('mps')
+    elif is_torch_xpu_available():
+        return torch.device('xpu:0')
+    elif is_torch_npu_available():
+        return torch.device('npu:0')
+    else:
+        return None
+
+
+def clear_torch_cache():
+    gc.collect()
+    if not shared.args.cpu:
+        if torch.cuda.is_available():
+            torch.cuda.empty_cache()
+        elif is_xpu_available():
+            torch.xpu.empty_cache()
+        elif is_npu_available():
+            torch.npu.empty_cache()
+        elif torch.backends.mps.is_available():
+            if hasattr(torch.backends.mps, 'empty_cache'):
+                torch.backends.mps.empty_cache()
diff --git a/modules/training.py b/modules/training.py
index 11c4b8c5d7..c6c380a387 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -15,13 +15,6 @@
 from pathlib import Path
 
 import gradio as gr
-import torch
-import transformers
-from datasets import Dataset, load_dataset
-from transformers import is_torch_xpu_available
-from transformers.models.auto.modeling_auto import (
-    MODEL_FOR_CAUSAL_LM_MAPPING_NAMES
-)
 
 from modules import shared, ui, utils
 from modules.evaluate import (
@@ -33,7 +26,6 @@
 from modules.models import reload_model
 from modules.utils import natural_keys
 
-MODEL_CLASSES = {v[1]: v[0] for v in MODEL_FOR_CAUSAL_LM_MAPPING_NAMES.items()}
 PARAMETERS = ["lora_name", "always_override", "q_proj_en", "v_proj_en", "k_proj_en", "o_proj_en", "gate_proj_en", "down_proj_en", "up_proj_en", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "raw_text_file", "overlap_len", "newline_favor_len", "higher_rank_limit", "warmup_steps", "optimizer", "hard_cut_string", "train_only_after", "stop_at_loss", "add_eos_token", "min_chars", "report_to"]
 WANT_INTERRUPT = False
 
@@ -284,6 +276,9 @@ def calc_trainable_parameters(model):
 
 def do_train(lora_name: str, always_override: bool, q_proj_en: bool, v_proj_en: bool, k_proj_en: bool, o_proj_en: bool, gate_proj_en: bool, down_proj_en: bool, up_proj_en: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, raw_text_file: str, overlap_len: int, newline_favor_len: int, higher_rank_limit: bool, warmup_steps: int, optimizer: str, hard_cut_string: str, train_only_after: str, stop_at_loss: float, add_eos_token: bool, min_chars: int, report_to: str):
 
+    import torch
+    import transformers
+    from datasets import Dataset, load_dataset
     from peft import (
         LoraConfig,
         get_peft_model,
@@ -293,6 +288,12 @@ def do_train(lora_name: str, always_override: bool, q_proj_en: bool, v_proj_en:
     from peft.utils.other import \
         TRANSFORMERS_MODELS_TO_LORA_TARGET_MODULES_MAPPING as \
         model_to_lora_modules
+    from transformers import is_torch_xpu_available
+    from transformers.models.auto.modeling_auto import (
+        MODEL_FOR_CAUSAL_LM_MAPPING_NAMES
+    )
+
+    MODEL_CLASSES = {v[1]: v[0] for v in MODEL_FOR_CAUSAL_LM_MAPPING_NAMES.items()}
 
     global WANT_INTERRUPT
     WANT_INTERRUPT = False
diff --git a/modules/transformers_loader.py b/modules/transformers_loader.py
new file mode 100644
index 0000000000..5512f061ee
--- /dev/null
+++ b/modules/transformers_loader.py
@@ -0,0 +1,281 @@
+import os
+import pprint
+from pathlib import Path
+
+import torch
+import torch.nn.functional as F
+import transformers
+from accelerate import infer_auto_device_map, init_empty_weights
+from accelerate.utils import (
+    is_ccl_available,
+    is_npu_available,
+    is_xpu_available
+)
+from transformers import (
+    AutoConfig,
+    AutoModel,
+    AutoModelForCausalLM,
+    AutoModelForSeq2SeqLM,
+    AutoTokenizer,
+    BitsAndBytesConfig,
+    LogitsProcessor
+)
+
+import modules.shared as shared
+from modules import sampler_hijack
+from modules.logging_colors import logger
+from modules.text_generation import get_reply_from_output_ids
+from modules.torch_utils import get_device
+
+transformers.logging.set_verbosity_error()
+sampler_hijack.hijack_samplers()
+
+local_rank = None
+if shared.args.deepspeed:
+    import deepspeed
+    from transformers.integrations.deepspeed import (
+        HfDeepSpeedConfig,
+        is_deepspeed_zero3_enabled
+    )
+
+    from modules.deepspeed_parameters import generate_ds_config
+
+    # Distributed setup
+    local_rank = shared.args.local_rank if shared.args.local_rank is not None else int(os.getenv("LOCAL_RANK", "0"))
+    world_size = int(os.getenv("WORLD_SIZE", "1"))
+    if is_xpu_available() and is_ccl_available():
+        torch.xpu.set_device(local_rank)
+        deepspeed.init_distributed(backend="ccl")
+    elif is_npu_available():
+        torch.npu.set_device(local_rank)
+        deepspeed.init_distributed(dist_backend="hccl")
+    else:
+        torch.cuda.set_device(local_rank)
+        deepspeed.init_distributed()
+    ds_config = generate_ds_config(shared.args.bf16, 1 * world_size, shared.args.nvme_offload_dir)
+    dschf = HfDeepSpeedConfig(ds_config)  # Keep this object alive for the Transformers integration
+
+
+class _StopEverythingStoppingCriteria(transformers.StoppingCriteria):
+    def __init__(self):
+        transformers.StoppingCriteria.__init__(self)
+
+    def __call__(self, input_ids: torch.LongTensor, _scores: torch.FloatTensor) -> bool:
+        return shared.stop_everything
+
+
+class Stream(transformers.StoppingCriteria):
+    def __init__(self, callback_func=None):
+        self.callback_func = callback_func
+
+    def __call__(self, input_ids, scores) -> bool:
+        if self.callback_func is not None:
+            self.callback_func(input_ids[0])
+
+        return False
+
+
+class LogitsBiasProcessor(LogitsProcessor):
+    def __init__(self, logit_bias={}):
+        self.logit_bias = logit_bias
+        if self.logit_bias:
+            self.keys = list([int(key) for key in self.logit_bias.keys()])
+            values = [self.logit_bias[str(key)] for key in self.keys]
+            self.values = torch.tensor(values, dtype=torch.float, device=shared.model.device)
+
+    def __call__(self, input_ids: torch.LongTensor, logits: torch.FloatTensor) -> torch.FloatTensor:
+        if self.logit_bias:
+            logits[0, self.keys] += self.values
+
+        return logits
+
+    def __repr__(self):
+        return f"<{self.__class__.__name__}(logit_bias={self.logit_bias})>"
+
+
+class LogprobProcessor(LogitsProcessor):
+    def __init__(self, logprobs=None):
+        self.logprobs = logprobs
+        self.token_alternatives = {}
+
+    def __call__(self, input_ids: torch.LongTensor, logits: torch.FloatTensor) -> torch.FloatTensor:
+        if self.logprobs is not None:  # 0-5
+            log_e_probabilities = F.log_softmax(logits, dim=1)
+            top_values, top_indices = torch.topk(log_e_probabilities, k=self.logprobs + 1)
+            top_tokens = [get_reply_from_output_ids([tok]) for tok in top_indices[0]]
+            top_probs = [float(x) for x in top_values[0]]
+            self.token_alternatives = dict(zip(top_tokens, top_probs))
+
+        return logits
+
+    def __repr__(self):
+        return f"<{self.__class__.__name__}(logprobs={self.logprobs}, token_alternatives={self.token_alternatives})>"
+
+
+def load_tokenizer(model_name, tokenizer_dir=None):
+    if tokenizer_dir:
+        path_to_model = Path(tokenizer_dir)
+    else:
+        path_to_model = Path(f"{shared.args.model_dir}/{model_name}/")
+
+    tokenizer = None
+    if path_to_model.exists():
+        if shared.args.no_use_fast:
+            logger.info('Loading the tokenizer with use_fast=False.')
+
+        tokenizer = AutoTokenizer.from_pretrained(
+            path_to_model,
+            trust_remote_code=shared.args.trust_remote_code,
+            use_fast=not shared.args.no_use_fast
+        )
+
+    return tokenizer
+
+
+def load_model_HF(model_name):
+    path_to_model = Path(f'{shared.args.model_dir}/{model_name}')
+    params = {
+        'low_cpu_mem_usage': True,
+        'torch_dtype': torch.bfloat16 if shared.args.bf16 else torch.float16,
+    }
+
+    if shared.args.trust_remote_code:
+        params['trust_remote_code'] = True
+
+    if shared.args.use_flash_attention_2:
+        params['use_flash_attention_2'] = True
+
+    if shared.args.force_safetensors:
+        params['force_safetensors'] = True
+
+    if shared.args.use_eager_attention:
+        params['attn_implementation'] = 'eager'
+
+    config = AutoConfig.from_pretrained(path_to_model, trust_remote_code=shared.args.trust_remote_code)
+
+    if 'chatglm' in model_name.lower():
+        LoaderClass = AutoModel
+    else:
+        if config.to_dict().get('is_encoder_decoder', False):
+            LoaderClass = AutoModelForSeq2SeqLM
+            shared.is_seq2seq = True
+        else:
+            LoaderClass = AutoModelForCausalLM
+
+    # Determine if we should use default loading
+    should_use_default_loading = not any([
+        shared.args.cpu,
+        shared.args.load_in_8bit,
+        shared.args.load_in_4bit,
+        shared.args.disk,
+        shared.args.deepspeed,
+        shared.args.cpu_memory is not None,
+        shared.args.compress_pos_emb > 1,
+        shared.args.alpha_value > 1,
+    ])
+
+    # Load the model without any special settings
+    if should_use_default_loading:
+        params['device_map'] = 'auto'
+
+        logger.info("TRANSFORMERS_PARAMS=")
+        pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(params)
+        print()
+
+        model = LoaderClass.from_pretrained(path_to_model, **params)
+        if not (hasattr(model, 'is_loaded_in_4bit') and model.is_loaded_in_4bit):
+            device = get_device()
+            if device:
+                model = model.to(device)
+
+    # DeepSpeed ZeRO-3
+    elif shared.args.deepspeed:
+        model = LoaderClass.from_pretrained(
+            path_to_model,
+            torch_dtype=params['torch_dtype'],
+            trust_remote_code=params.get('trust_remote_code')
+        )
+
+        model = deepspeed.initialize(
+            model=model,
+            config_params=ds_config,
+            model_parameters=None,
+            optimizer=None,
+            lr_scheduler=None
+        )[0]
+
+        model.module.eval()  # Inference
+        logger.info(f'DeepSpeed ZeRO-3 is enabled: {is_deepspeed_zero3_enabled()}')
+
+    # Load with quantization and/or offloading
+    else:
+        if not any((shared.args.cpu, torch.cuda.is_available(), is_xpu_available(), torch.backends.mps.is_available())):
+            logger.warning('torch.cuda.is_available() and is_xpu_available() returned False. This means that no GPU has been detected. Falling back to CPU mode.')
+            shared.args.cpu = True
+
+        if shared.args.cpu:
+            params['torch_dtype'] = torch.float32
+        else:
+            params['device_map'] = 'auto'
+            if x := get_max_memory_dict():
+                params['max_memory'] = x
+
+            if shared.args.load_in_4bit:
+                # See https://github.com/huggingface/transformers/pull/23479/files
+                # and https://huggingface.co/blog/4bit-transformers-bitsandbytes
+                quantization_config_params = {
+                    'load_in_4bit': True,
+                    'bnb_4bit_compute_dtype': eval(f"torch.{shared.args.compute_dtype}") if shared.args.compute_dtype in ["bfloat16", "float16", "float32"] else None,
+                    'bnb_4bit_quant_type': shared.args.quant_type,
+                    'bnb_4bit_use_double_quant': shared.args.use_double_quant,
+                    'llm_int8_enable_fp32_cpu_offload': True
+                }
+                params['quantization_config'] = BitsAndBytesConfig(**quantization_config_params)
+
+            elif shared.args.load_in_8bit:
+                if shared.args.gpu_split:
+                    params['quantization_config'] = BitsAndBytesConfig(load_in_8bit=True, llm_int8_enable_fp32_cpu_offload=True)
+                else:
+                    params['quantization_config'] = BitsAndBytesConfig(load_in_8bit=True)
+
+                if params.get('max_memory') is not None:
+                    with init_empty_weights():
+                        model = LoaderClass.from_config(config, trust_remote_code=params.get('trust_remote_code'))
+
+                    model.tie_weights()
+                    params['device_map'] = infer_auto_device_map(
+                        model,
+                        dtype=torch.int8,
+                        max_memory=params.get('max_memory'),
+                        no_split_module_classes=model._no_split_modules
+                    )
+
+            if shared.args.disk:
+                params['offload_folder'] = shared.args.disk_cache_dir
+
+        if shared.args.compress_pos_emb > 1:
+            params['rope_scaling'] = {'type': 'linear', 'factor': shared.args.compress_pos_emb}
+        elif shared.args.alpha_value > 1:
+            params['rope_scaling'] = {'type': 'dynamic', 'factor': shared.args.alpha_value}
+
+        logger.info("TRANSFORMERS_PARAMS=")
+        pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(params)
+        print()
+        model = LoaderClass.from_pretrained(path_to_model, **params)
+
+    if shared.args.torch_compile:
+        model = torch.compile(model)
+
+    return model
+
+
+def get_max_memory_dict():
+    max_memory = {}
+    if shared.args.cpu_memory > 0:
+        max_memory['cpu'] = f'{shared.args.cpu_memory}GiB'
+
+    if shared.args.gpu_split:
+        for i, memory in enumerate(shared.args.gpu_split.split(',')):
+            max_memory[i] = f'{memory}GiB'
+
+    return max_memory if len(max_memory) > 0 else None
diff --git a/modules/ui.py b/modules/ui.py
index c36fe55351..d5caaeaaaf 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -2,9 +2,7 @@
 from pathlib import Path
 
 import gradio as gr
-import torch
 import yaml
-from transformers import is_torch_xpu_available
 
 import extensions
 from modules import shared
@@ -128,7 +126,6 @@ def list_model_elements():
         'torch_compile',
         'flash_attn',
         'use_flash_attention_2',
-        'auto_devices',
         'cpu',
         'disk',
         'row_split',
@@ -150,13 +147,6 @@ def list_model_elements():
         'no_use_fast',
     ]
 
-    if is_torch_xpu_available():
-        for i in range(torch.xpu.device_count()):
-            elements.append(f'gpu_memory_{i}')
-    else:
-        for i in range(torch.cuda.device_count()):
-            elements.append(f'gpu_memory_{i}')
-
     return elements
 
 
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 8db486219c..b4af771cfb 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -1,14 +1,9 @@
 import importlib
-import math
-import re
 import traceback
 from functools import partial
 from pathlib import Path
 
 import gradio as gr
-import psutil
-import torch
-from transformers import is_torch_npu_available, is_torch_xpu_available
 
 from modules import loaders, shared, ui, utils
 from modules.logging_colors import logger
@@ -27,35 +22,6 @@
 def create_ui():
     mu = shared.args.multi_user
 
-    # Finding the default values for the GPU and CPU memories
-    total_mem = []
-    if is_torch_xpu_available():
-        for i in range(torch.xpu.device_count()):
-            total_mem.append(math.floor(torch.xpu.get_device_properties(i).total_memory / (1024 * 1024)))
-    elif is_torch_npu_available():
-        for i in range(torch.npu.device_count()):
-            total_mem.append(math.floor(torch.npu.get_device_properties(i).total_memory / (1024 * 1024)))
-    else:
-        for i in range(torch.cuda.device_count()):
-            total_mem.append(math.floor(torch.cuda.get_device_properties(i).total_memory / (1024 * 1024)))
-
-    default_gpu_mem = []
-    if shared.args.gpu_memory is not None and len(shared.args.gpu_memory) > 0:
-        for i in shared.args.gpu_memory:
-            if 'mib' in i.lower():
-                default_gpu_mem.append(int(re.sub('[a-zA-Z ]', '', i)))
-            else:
-                default_gpu_mem.append(int(re.sub('[a-zA-Z ]', '', i)) * 1000)
-
-    while len(default_gpu_mem) < len(total_mem):
-        default_gpu_mem.append(0)
-
-    total_cpu_mem = math.floor(psutil.virtual_memory().total / (1024 * 1024))
-    if shared.args.cpu_memory is not None:
-        default_cpu_mem = re.sub('[a-zA-Z ]', '', shared.args.cpu_memory)
-    else:
-        default_cpu_mem = 0
-
     with gr.Tab("Model", elem_id="model-tab"):
         with gr.Row():
             with gr.Column():
@@ -80,10 +46,6 @@ def create_ui():
                 with gr.Blocks():
                     with gr.Row():
                         with gr.Column():
-                            for i in range(len(total_mem)):
-                                shared.gradio[f'gpu_memory_{i}'] = gr.Slider(label=f"gpu-memory in MiB for device :{i}", maximum=total_mem[i], value=default_gpu_mem[i])
-
-                            shared.gradio['cpu_memory'] = gr.Slider(label="cpu-memory in MiB", maximum=total_cpu_mem, value=default_cpu_mem)
                             shared.gradio['n_gpu_layers'] = gr.Slider(label="n-gpu-layers", minimum=0, maximum=256, value=shared.args.n_gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['threads'] = gr.Slider(label="threads", minimum=0, step=1, maximum=256, value=shared.args.threads)
                             shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
@@ -94,6 +56,7 @@ def create_ui():
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q6', 'q4'], value=shared.args.cache_type, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.')
                             shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
+                            shared.gradio['cpu_memory'] = gr.Number(label="Maximum CPU memory in GiB. Use this for CPU offloading.", value=shared.args.cpu_memory)
                             shared.gradio['alpha_value'] = gr.Number(label='alpha_value', value=shared.args.alpha_value, precision=2, info='Positional embeddings alpha factor for NTK RoPE scaling. Recommended values (NTKv1): 1.75 for 1.5x context, 2.5 for 2x context. Use either this or compress_pos_emb, not both.')
                             shared.gradio['rope_freq_base'] = gr.Number(label='rope_freq_base', value=shared.args.rope_freq_base, precision=0, info='Positional embeddings frequency base for NTK RoPE scaling. Related to alpha_value by rope_freq_base = 10000 * alpha_value ^ (64 / 63). 0 = from model.')
                             shared.gradio['compress_pos_emb'] = gr.Number(label='compress_pos_emb', value=shared.args.compress_pos_emb, precision=2, info='Positional embeddings compression factor. Should be set to (context length) / (model\'s original context length). Equal to 1/rope_freq_scale.')
@@ -107,7 +70,6 @@ def create_ui():
                             shared.gradio['torch_compile'] = gr.Checkbox(label="torch-compile", value=shared.args.torch_compile, info='Compile the model with torch.compile for improved performance.')
                             shared.gradio['flash_attn'] = gr.Checkbox(label="flash_attn", value=shared.args.flash_attn, info='Use flash-attention.')
                             shared.gradio['use_flash_attention_2'] = gr.Checkbox(label="use_flash_attention_2", value=shared.args.use_flash_attention_2, info='Set use_flash_attention_2=True while loading the model.')
-                            shared.gradio['auto_devices'] = gr.Checkbox(label="auto-devices", value=shared.args.auto_devices)
                             shared.gradio['cpu'] = gr.Checkbox(label="cpu", value=shared.args.cpu, info='llama.cpp: Use llama-cpp-python compiled without GPU acceleration. Transformers: use PyTorch in CPU mode.')
                             shared.gradio['disk'] = gr.Checkbox(label="disk", value=shared.args.disk)
                             shared.gradio['row_split'] = gr.Checkbox(label="row_split", value=shared.args.row_split, info='Split the model by rows across GPUs. This may improve multi-gpu performance.')
diff --git a/server.py b/server.py
index 1f227350fd..a4b6e3e388 100644
--- a/server.py
+++ b/server.py
@@ -1,11 +1,8 @@
 import os
 import warnings
 
-from modules import shared
-
-import accelerate  # This early import makes Intel GPUs happy
-
 import modules.one_click_installer_check
+from modules import shared
 from modules.block_requests import OpenMonkeyPatch, RequestBlocker
 from modules.logging_colors import logger
 
@@ -38,7 +35,6 @@
 
 import modules.extensions as extensions_module
 from modules import (
-    chat,
     training,
     ui,
     ui_chat,

From 9c59acf8208cb0f70e739cbd05659fff37599369 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 20 Apr 2025 10:01:31 -0700
Subject: [PATCH 0326/1701] Remove the numba requirement (it's no longer used)

---
 requirements.txt                 | 1 -
 requirements_amd.txt             | 1 -
 requirements_amd_noavx2.txt      | 1 -
 requirements_apple_intel.txt     | 1 -
 requirements_apple_silicon.txt   | 1 -
 requirements_cpu_only.txt        | 1 -
 requirements_cpu_only_noavx2.txt | 1 -
 requirements_noavx2.txt          | 1 -
 requirements_nowheels.txt        | 1 -
 9 files changed, 9 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 607efda0ec..c2f1a8bdc4 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -7,7 +7,6 @@ fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
 markdown
-numba==0.59.*
 numpy==1.26.*
 pandas
 peft==0.15.*
diff --git a/requirements_amd.txt b/requirements_amd.txt
index b242d4ad3c..a1920f0d95 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -6,7 +6,6 @@ fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
 markdown
-numba==0.59.*
 numpy==1.26.*
 pandas
 peft==0.15.*
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index b6105209f3..92d4f15230 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -6,7 +6,6 @@ fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
 markdown
-numba==0.59.*
 numpy==1.26.*
 pandas
 peft==0.15.*
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index ce730f635c..6b6707abb1 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -6,7 +6,6 @@ fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
 markdown
-numba==0.59.*
 numpy==1.26.*
 pandas
 peft==0.15.*
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index a7be282d0d..910fae3e7d 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -6,7 +6,6 @@ fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
 markdown
-numba==0.59.*
 numpy==1.26.*
 pandas
 peft==0.15.*
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index 2437c2ae4b..d34a4312bd 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -6,7 +6,6 @@ fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
 markdown
-numba==0.59.*
 numpy==1.26.*
 pandas
 peft==0.15.*
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index cbaa8e96c7..cb3f3d6b26 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -6,7 +6,6 @@ fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
 markdown
-numba==0.59.*
 numpy==1.26.*
 pandas
 peft==0.15.*
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index cce27aa22e..06352dc3ab 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -7,7 +7,6 @@ fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
 markdown
-numba==0.59.*
 numpy==1.26.*
 pandas
 peft==0.15.*
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index 3b61ca39f6..7b43e08d84 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -6,7 +6,6 @@ fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
 markdown
-numba==0.59.*
 numpy==1.26.*
 pandas
 peft==0.15.*

From b3bf7a885d6c3d7216c1d996c0eb7581e1413d81 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 20 Apr 2025 11:32:48 -0700
Subject: [PATCH 0327/1701] Fix ExLlamaV2_HF and ExLlamaV3_HF after
 ae02ffc605d90b19a507132c09055c7ff11df52b

---
 modules/exllamav2_hf.py        | 15 ++++++++++-----
 modules/exllamav3_hf.py        |  9 +++++++--
 modules/models.py              |  5 +++++
 modules/transformers_loader.py |  2 --
 4 files changed, 22 insertions(+), 9 deletions(-)

diff --git a/modules/exllamav2_hf.py b/modules/exllamav2_hf.py
index 6486e438fa..b159d9ce44 100644
--- a/modules/exllamav2_hf.py
+++ b/modules/exllamav2_hf.py
@@ -4,10 +4,6 @@
 from typing import Any, Dict, Optional, Union
 
 import torch
-from torch.nn import CrossEntropyLoss
-from transformers import GenerationConfig, PretrainedConfig, PreTrainedModel
-from transformers.modeling_outputs import CausalLMOutputWithPast
-
 from exllamav2 import (
     ExLlamaV2,
     ExLlamaV2Cache,
@@ -18,6 +14,15 @@
     ExLlamaV2Cache_TP,
     ExLlamaV2Config
 )
+from torch.nn import CrossEntropyLoss
+from transformers import (
+    GenerationConfig,
+    GenerationMixin,
+    PretrainedConfig,
+    PreTrainedModel
+)
+from transformers.modeling_outputs import CausalLMOutputWithPast
+
 from modules import shared
 from modules.logging_colors import logger
 
@@ -28,7 +33,7 @@
     traceback.print_exc()
 
 
-class Exllamav2HF(PreTrainedModel):
+class Exllamav2HF(PreTrainedModel, GenerationMixin):
     def __init__(self, config: ExLlamaV2Config):
         super().__init__(PretrainedConfig())
         self.ex_config = config
diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index 0f742fa23f..2d9c493a63 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -6,7 +6,12 @@
 import torch
 from exllamav3 import Cache, Config, Model
 from torch.nn import CrossEntropyLoss
-from transformers import GenerationConfig, PretrainedConfig, PreTrainedModel
+from transformers import (
+    GenerationConfig,
+    GenerationMixin,
+    PretrainedConfig,
+    PreTrainedModel
+)
 from transformers.modeling_outputs import CausalLMOutputWithPast
 
 from modules import shared
@@ -19,7 +24,7 @@
     traceback.print_exc()
 
 
-class Exllamav3HF(PreTrainedModel):
+class Exllamav3HF(PreTrainedModel, GenerationMixin):
     def __init__(self, model_dir):
         super().__init__(PretrainedConfig())
         self.generation_config = GenerationConfig()
diff --git a/modules/models.py b/modules/models.py
index c4dfa1497e..2c23462aec 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -1,3 +1,4 @@
+import sys
 import time
 from pathlib import Path
 
@@ -34,6 +35,10 @@ def load_model(model_name, loader=None):
                 logger.error('The path to the model does not exist. Exiting.')
                 raise ValueError
 
+    if loader != 'llama.cpp' and 'sampler_hijack' not in sys.modules:
+        from modules import sampler_hijack
+        sampler_hijack.hijack_samplers()
+
     shared.args.loader = loader
     output = load_func_map[loader](model_name)
     if type(output) is tuple:
diff --git a/modules/transformers_loader.py b/modules/transformers_loader.py
index 5512f061ee..add3be66cf 100644
--- a/modules/transformers_loader.py
+++ b/modules/transformers_loader.py
@@ -22,13 +22,11 @@
 )
 
 import modules.shared as shared
-from modules import sampler_hijack
 from modules.logging_colors import logger
 from modules.text_generation import get_reply_from_output_ids
 from modules.torch_utils import get_device
 
 transformers.logging.set_verbosity_error()
-sampler_hijack.hijack_samplers()
 
 local_rank = None
 if shared.args.deepspeed:

From d5e1bccef91466918c43e291d8bd90ede5ae5d9f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 20 Apr 2025 11:47:28 -0700
Subject: [PATCH 0328/1701] Remove the SpeechRecognition requirement

---
 extensions/openai/script.py      | 3 ++-
 requirements.txt                 | 1 -
 requirements_amd.txt             | 1 -
 requirements_amd_noavx2.txt      | 1 -
 requirements_apple_intel.txt     | 1 -
 requirements_apple_silicon.txt   | 1 -
 requirements_cpu_only.txt        | 1 -
 requirements_cpu_only_noavx2.txt | 1 -
 requirements_noavx2.txt          | 1 -
 requirements_nowheels.txt        | 1 -
 10 files changed, 2 insertions(+), 10 deletions(-)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 2ce692c7fb..f907cdbb70 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -6,7 +6,6 @@
 from collections import deque
 from threading import Thread
 
-import speech_recognition as sr
 import uvicorn
 from fastapi import Depends, FastAPI, Header, HTTPException
 from fastapi.middleware.cors import CORSMiddleware
@@ -163,6 +162,8 @@ def handle_billing_usage():
 
 @app.post('/v1/audio/transcriptions', dependencies=check_key)
 async def handle_audio_transcription(request: Request):
+    import speech_recognition as sr
+
     r = sr.Recognizer()
 
     form = await request.form()
diff --git a/requirements.txt b/requirements.txt
index c2f1a8bdc4..9e55bb3310 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -25,7 +25,6 @@ tqdm
 wandb
 
 # API
-SpeechRecognition==3.10.0
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
diff --git a/requirements_amd.txt b/requirements_amd.txt
index a1920f0d95..ba8bc80ce6 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -24,7 +24,6 @@ tqdm
 wandb
 
 # API
-SpeechRecognition==3.10.0
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 92d4f15230..d921fbb0ba 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -24,7 +24,6 @@ tqdm
 wandb
 
 # API
-SpeechRecognition==3.10.0
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 6b6707abb1..0b92376aaf 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -24,7 +24,6 @@ tqdm
 wandb
 
 # API
-SpeechRecognition==3.10.0
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 910fae3e7d..3e7fcd020e 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -24,7 +24,6 @@ tqdm
 wandb
 
 # API
-SpeechRecognition==3.10.0
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index d34a4312bd..43d54534e6 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -24,7 +24,6 @@ tqdm
 wandb
 
 # API
-SpeechRecognition==3.10.0
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index cb3f3d6b26..b444444365 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -24,7 +24,6 @@ tqdm
 wandb
 
 # API
-SpeechRecognition==3.10.0
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 06352dc3ab..351ea83e45 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -25,7 +25,6 @@ tqdm
 wandb
 
 # API
-SpeechRecognition==3.10.0
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index 7b43e08d84..2e631bf0be 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -24,7 +24,6 @@ tqdm
 wandb
 
 # API
-SpeechRecognition==3.10.0
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken

From 8cfd7f976befb9c0780ae7db47a696a61bac089f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 20 Apr 2025 13:35:42 -0700
Subject: [PATCH 0329/1701] Revert "Remove the old --model-menu flag"

This reverts commit 109de34e3b3187eb3f463bf463086a48444013a0.
---
 modules/shared.py |  1 +
 server.py         | 18 ++++++++++++++++++
 2 files changed, 19 insertions(+)

diff --git a/modules/shared.py b/modules/shared.py
index fd0f226ea5..08268ae0bc 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -79,6 +79,7 @@
 group.add_argument('--lora', type=str, nargs='+', help='The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.')
 group.add_argument('--model-dir', type=str, default='models/', help='Path to directory with all the models.')
 group.add_argument('--lora-dir', type=str, default='loras/', help='Path to directory with all the loras.')
+group.add_argument('--model-menu', action='store_true', help='Show a model menu in the terminal when the web UI is first launched.')
 group.add_argument('--settings', type=str, help='Load the default interface settings from this yaml file. See settings-template.yaml for an example. If you create a file called settings.yaml, this file will be loaded by default without the need to use the --settings flag.')
 group.add_argument('--extensions', type=str, nargs='+', help='The list of extensions to load. If you want to load more than one extension, write the names separated by spaces.')
 group.add_argument('--verbose', action='store_true', help='Print the prompts to the terminal.')
diff --git a/server.py b/server.py
index a4b6e3e388..1d26156639 100644
--- a/server.py
+++ b/server.py
@@ -214,10 +214,28 @@ def create_interface():
         if extension not in shared.args.extensions:
             shared.args.extensions.append(extension)
 
+    available_models = utils.get_available_models()
+
     # Model defined through --model
     if shared.args.model is not None:
         shared.model_name = shared.args.model
 
+    # Select the model from a command-line menu
+    elif shared.args.model_menu:
+        if len(available_models) == 0:
+            logger.error('No models are available! Please download at least one.')
+            sys.exit(0)
+        else:
+            print('The following models are available:\n')
+            for i, model in enumerate(available_models):
+                print(f'{i+1}. {model}')
+
+            print(f'\nWhich one do you want to load? 1-{len(available_models)}\n')
+            i = int(input()) - 1
+            print()
+
+        shared.model_name = available_models[i]
+
     # If any model has been selected, load it
     if shared.model_name != 'None':
         p = Path(shared.model_name)

From e243424ba12c1fe030dd5086f67571ad8a6970fa Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 20 Apr 2025 17:51:28 -0700
Subject: [PATCH 0330/1701] Fix an import

---
 modules/evaluate.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/evaluate.py b/modules/evaluate.py
index bbf27da1fa..ba0de378b3 100644
--- a/modules/evaluate.py
+++ b/modules/evaluate.py
@@ -2,7 +2,6 @@
 from pathlib import Path
 
 import pandas as pd
-from datasets import load_dataset
 from tqdm import tqdm
 
 from modules import shared
@@ -39,6 +38,7 @@ def calculate_perplexity(models, input_dataset, stride, _max_length):
     '''
 
     import torch
+    from datasets import load_dataset
 
     from modules.torch_utils import clear_torch_cache
 

From 99588be5760702d413435c9167935f29832a6d06 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 20 Apr 2025 18:57:26 -0700
Subject: [PATCH 0331/1701] Organize one_click.py

---
 one_click.py | 183 +++++++++++++++++++++++++--------------------------
 1 file changed, 91 insertions(+), 92 deletions(-)

diff --git a/one_click.py b/one_click.py
index 99bdc41e80..eff2ed9f3c 100644
--- a/one_click.py
+++ b/one_click.py
@@ -15,7 +15,6 @@
 # os.environ["HSA_OVERRIDE_GFX_VERSION"] = '10.3.0'
 # os.environ["HCC_AMDGPU_TARGET"] = 'gfx1030'
 
-
 # Define the required versions
 TORCH_VERSION = "2.6.0"
 TORCHVISION_VERSION = "0.21.0"
@@ -62,6 +61,19 @@ def is_x86_64():
     return platform.machine() == "x86_64"
 
 
+def is_installed():
+    site_packages_path = None
+    for sitedir in site.getsitepackages():
+        if "site-packages" in sitedir and conda_env_path in sitedir:
+            site_packages_path = sitedir
+            break
+
+    if site_packages_path:
+        return os.path.isfile(os.path.join(site_packages_path, 'torch', '__init__.py'))
+    else:
+        return os.path.isdir(conda_env_path)
+
+
 def cpu_has_avx2():
     try:
         import cpuinfo
@@ -104,44 +116,13 @@ def torch_version():
     return torver
 
 
-def update_pytorch_and_python():
-    print_big_message("Checking for PyTorch updates.")
-
-    # Update the Python version. Left here for future reference in case this becomes necessary.
-    # print_big_message("Checking for PyTorch and Python updates.")
-    # current_python_version = f"{sys.version_info.major}.{sys.version_info.minor}"
-    # if current_python_version != PYTHON_VERSION:
-    #     run_cmd(f"conda install -y python={PYTHON_VERSION}", assert_success=True, environment=True)
-
-    torver = torch_version()
-    base_cmd = f"python -m pip install --upgrade torch=={TORCH_VERSION} torchvision=={TORCHVISION_VERSION} torchaudio=={TORCHAUDIO_VERSION}"
-
-    if "+cu" in torver:
-        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/cu124"
-    elif "+rocm" in torver:
-        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.1"
-    elif "+cpu" in torver:
-        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/cpu"
-    elif "+cxx11" in torver:
-        intel_extension = "intel-extension-for-pytorch==2.1.10+xpu" if is_linux() else "intel-extension-for-pytorch==2.1.10"
-        install_cmd = f"{base_cmd} {intel_extension} --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
-    else:
-        install_cmd = base_cmd
-
-    run_cmd(install_cmd, assert_success=True, environment=True)
-
+def get_current_commit():
+    result = run_cmd("git rev-parse HEAD", capture_output=True, environment=True)
+    return result.stdout.decode('utf-8').strip()
 
-def is_installed():
-    site_packages_path = None
-    for sitedir in site.getsitepackages():
-        if "site-packages" in sitedir and conda_env_path in sitedir:
-            site_packages_path = sitedir
-            break
 
-    if site_packages_path:
-        return os.path.isfile(os.path.join(site_packages_path, 'torch', '__init__.py'))
-    else:
-        return os.path.isdir(conda_env_path)
+def get_extensions_names():
+    return [foldername for foldername in os.listdir('extensions') if os.path.isfile(os.path.join('extensions', foldername, 'requirements.txt'))]
 
 
 def check_env():
@@ -157,35 +138,11 @@ def check_env():
         sys.exit(1)
 
 
-def get_current_commit():
-    result = run_cmd("git rev-parse HEAD", capture_output=True, environment=True)
-    return result.stdout.decode('utf-8').strip()
-
-
 def clear_cache():
     run_cmd("conda clean -a -y", environment=True)
     run_cmd("python -m pip cache purge", environment=True)
 
 
-def print_big_message(message):
-    message = message.strip()
-    lines = message.split('\n')
-    print("\n\n*******************************************************************")
-    for line in lines:
-        print("*", line)
-
-    print("*******************************************************************\n\n")
-
-
-def calculate_file_hash(file_path):
-    p = os.path.join(script_dir, file_path)
-    if os.path.isfile(p):
-        with open(p, 'rb') as f:
-            return hashlib.sha256(f.read()).hexdigest()
-    else:
-        return ''
-
-
 def run_cmd(cmd, assert_success=False, environment=False, capture_output=False, env=None):
     # Use the conda environment
     if environment:
@@ -210,6 +167,25 @@ def run_cmd(cmd, assert_success=False, environment=False, capture_output=False,
     return result
 
 
+def print_big_message(message):
+    message = message.strip()
+    lines = message.split('\n')
+    print("\n\n*******************************************************************")
+    for line in lines:
+        print("*", line)
+
+    print("*******************************************************************\n\n")
+
+
+def calculate_file_hash(file_path):
+    p = os.path.join(script_dir, file_path)
+    if os.path.isfile(p):
+        with open(p, 'rb') as f:
+            return hashlib.sha256(f.read()).hexdigest()
+    else:
+        return ''
+
+
 def generate_alphabetic_sequence(index):
     result = ''
     while index >= 0:
@@ -238,6 +214,51 @@ def get_user_choice(question, options_dict):
     return choice
 
 
+def update_pytorch_and_python():
+    print_big_message("Checking for PyTorch updates.")
+
+    # Update the Python version. Left here for future reference in case this becomes necessary.
+    # print_big_message("Checking for PyTorch and Python updates.")
+    # current_python_version = f"{sys.version_info.major}.{sys.version_info.minor}"
+    # if current_python_version != PYTHON_VERSION:
+    #     run_cmd(f"conda install -y python={PYTHON_VERSION}", assert_success=True, environment=True)
+
+    torver = torch_version()
+    base_cmd = f"python -m pip install --upgrade torch=={TORCH_VERSION} torchvision=={TORCHVISION_VERSION} torchaudio=={TORCHAUDIO_VERSION}"
+
+    if "+cu" in torver:
+        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/cu124"
+    elif "+rocm" in torver:
+        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.1"
+    elif "+cpu" in torver:
+        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/cpu"
+    elif "+cxx11" in torver:
+        intel_extension = "intel-extension-for-pytorch==2.1.10+xpu" if is_linux() else "intel-extension-for-pytorch==2.1.10"
+        install_cmd = f"{base_cmd} {intel_extension} --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
+    else:
+        install_cmd = base_cmd
+
+    run_cmd(install_cmd, assert_success=True, environment=True)
+
+
+def clean_outdated_pytorch_cuda_dependencies():
+    patterns = ["cu121", "cu122", "torch2.4"]
+    result = run_cmd("python -m pip list --format=freeze", capture_output=True, environment=True)
+    matching_packages = []
+
+    for line in result.stdout.decode('utf-8').splitlines():
+        if "==" in line:
+            pkg_name, version = line.split('==', 1)
+            if any(pattern in version for pattern in patterns):
+                matching_packages.append(pkg_name)
+
+    if matching_packages:
+        print(f"\nUninstalling: {', '.join(matching_packages)}\n")
+        run_cmd(f"python -m pip uninstall -y {' '.join(matching_packages)}", assert_success=True, environment=True)
+
+    return matching_packages
+
+
 def install_webui():
     if os.path.isfile(state_file):
         os.remove(state_file)
@@ -323,37 +344,6 @@ def install_webui():
     update_requirements(initial_installation=True, pull=False)
 
 
-def get_extensions_names():
-    return [foldername for foldername in os.listdir('extensions') if os.path.isfile(os.path.join('extensions', foldername, 'requirements.txt'))]
-
-
-def install_extensions_requirements():
-    print_big_message("Installing extensions requirements.\nSome of these may fail on Windows.\nDon\'t worry if you see error messages, as they will not affect the main program.")
-    extensions = get_extensions_names()
-    for i, extension in enumerate(extensions):
-        print(f"\n\n--- [{i + 1}/{len(extensions)}]: {extension}\n\n")
-        extension_req_path = os.path.join("extensions", extension, "requirements.txt")
-        run_cmd(f"python -m pip install -r {extension_req_path} --upgrade", assert_success=False, environment=True)
-
-
-def clean_outdated_pytorch_cuda_dependencies():
-    patterns = ["cu121", "cu122", "torch2.4"]
-    result = run_cmd("python -m pip list --format=freeze", capture_output=True, environment=True)
-    matching_packages = []
-
-    for line in result.stdout.decode('utf-8').splitlines():
-        if "==" in line:
-            pkg_name, version = line.split('==', 1)
-            if any(pattern in version for pattern in patterns):
-                matching_packages.append(pkg_name)
-
-    if matching_packages:
-        print(f"\nUninstalling: {', '.join(matching_packages)}\n")
-        run_cmd(f"python -m pip uninstall -y {' '.join(matching_packages)}", assert_success=True, environment=True)
-
-    return matching_packages
-
-
 def update_requirements(initial_installation=False, pull=True):
     # Create .git directory if missing
     if not os.path.exists(os.path.join(script_dir, ".git")):
@@ -475,6 +465,15 @@ def update_requirements(initial_installation=False, pull=True):
     clear_cache()
 
 
+def install_extensions_requirements():
+    print_big_message("Installing extensions requirements.\nSome of these may fail on Windows.\nDon\'t worry if you see error messages, as they will not affect the main program.")
+    extensions = get_extensions_names()
+    for i, extension in enumerate(extensions):
+        print(f"\n\n--- [{i + 1}/{len(extensions)}]: {extension}\n\n")
+        extension_req_path = os.path.join("extensions", extension, "requirements.txt")
+        run_cmd(f"python -m pip install -r {extension_req_path} --upgrade", assert_success=False, environment=True)
+
+
 def launch_webui():
     run_cmd(f"python server.py {flags}", environment=True)
 

From d3e7c655e5b4ce3fb15ecd7e27064f8df727ce14 Mon Sep 17 00:00:00 2001
From: Matthew Jenkins <40323108+Matthew-Jenkins@users.noreply.github.com>
Date: Sun, 20 Apr 2025 22:06:24 -0400
Subject: [PATCH 0332/1701] Add support for llama-cpp builds from
 https://github.com/ggml-org/llama.cpp (#6862)

---
 modules/llama_cpp_server.py | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 02a56b3cd5..c88f945d66 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -1,4 +1,5 @@
 import json
+import os
 import pprint
 import socket
 import subprocess
@@ -281,12 +282,21 @@ def _start_server(self):
         if shared.args.rope_freq_base > 0:
             cmd += ["--rope-freq-base", str(shared.args.rope_freq_base)]
 
+        env = os.environ.copy()
+        if os.name == 'posix':
+            current_path = env.get('LD_LIBRARY_PATH', '')
+            if current_path:
+                env['LD_LIBRARY_PATH'] = f"{current_path}:{os.path.dirname(self.server_path)}"
+            else:
+                env['LD_LIBRARY_PATH'] = os.path.dirname(self.server_path)
+
         # Start the server with pipes for output
         self.process = subprocess.Popen(
             cmd,
             stderr=subprocess.PIPE,
             text=True,
-            bufsize=1
+            bufsize=1,
+            env=env
         )
 
         def filter_stderr(process_stderr):

From ff1c00bdd993f1610add8b0d2eff208809506444 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 20 Apr 2025 19:08:44 -0700
Subject: [PATCH 0333/1701] llama.cpp: set the random seed manually

---
 modules/text_generation.py | 27 ++++++++++++---------------
 1 file changed, 12 insertions(+), 15 deletions(-)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index 585e4d9d30..40046eb29c 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -215,20 +215,21 @@ def formatted_outputs(reply, model_name):
 
 
 def set_manual_seed(seed):
-    import torch
-    from transformers import is_torch_npu_available, is_torch_xpu_available
-
     seed = int(seed)
     if seed == -1:
         seed = random.randint(1, 2**31)
 
-    torch.manual_seed(seed)
-    if torch.cuda.is_available():
-        torch.cuda.manual_seed_all(seed)
-    elif is_torch_xpu_available():
-        torch.xpu.manual_seed_all(seed)
-    elif is_torch_npu_available():
-        torch.npu.manual_seed_all(seed)
+    if shared.args.loader != 'llama.cpp':
+        import torch
+        from transformers import is_torch_npu_available, is_torch_xpu_available
+
+        torch.manual_seed(seed)
+        if torch.cuda.is_available():
+            torch.cuda.manual_seed_all(seed)
+        elif is_torch_xpu_available():
+            torch.xpu.manual_seed_all(seed)
+        elif is_torch_npu_available():
+            torch.npu.manual_seed_all(seed)
 
     return seed
 
@@ -476,11 +477,7 @@ def generate_reply_custom(question, original_question, state, stopping_strings=N
     For models that do not use the transformers library for sampling
     """
 
-    seed = state['seed']
-    if shared.args.loader != 'llama.cpp':
-        print(shared.args.loader)
-        seed = set_manual_seed(seed)
-
+    seed = set_manual_seed(state['seed'])
     t0 = time.time()
     reply = ''
     try:

From 6117ef7d64543f337abe6f0b33dca6b8e9294f9e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 20 Apr 2025 19:12:04 -0700
Subject: [PATCH 0334/1701] Move the requirements*.txt to a requirements folder

---
 requirements.txt => requirements/requirements.txt                 | 0
 requirements_amd.txt => requirements/requirements_amd.txt         | 0
 .../requirements_amd_noavx2.txt                                   | 0
 .../requirements_apple_intel.txt                                  | 0
 .../requirements_apple_silicon.txt                                | 0
 .../requirements_cpu_only.txt                                     | 0
 .../requirements_cpu_only_noavx2.txt                              | 0
 requirements_noavx2.txt => requirements/requirements_noavx2.txt   | 0
 .../requirements_nowheels.txt                                     | 0
 9 files changed, 0 insertions(+), 0 deletions(-)
 rename requirements.txt => requirements/requirements.txt (100%)
 rename requirements_amd.txt => requirements/requirements_amd.txt (100%)
 rename requirements_amd_noavx2.txt => requirements/requirements_amd_noavx2.txt (100%)
 rename requirements_apple_intel.txt => requirements/requirements_apple_intel.txt (100%)
 rename requirements_apple_silicon.txt => requirements/requirements_apple_silicon.txt (100%)
 rename requirements_cpu_only.txt => requirements/requirements_cpu_only.txt (100%)
 rename requirements_cpu_only_noavx2.txt => requirements/requirements_cpu_only_noavx2.txt (100%)
 rename requirements_noavx2.txt => requirements/requirements_noavx2.txt (100%)
 rename requirements_nowheels.txt => requirements/requirements_nowheels.txt (100%)

diff --git a/requirements.txt b/requirements/requirements.txt
similarity index 100%
rename from requirements.txt
rename to requirements/requirements.txt
diff --git a/requirements_amd.txt b/requirements/requirements_amd.txt
similarity index 100%
rename from requirements_amd.txt
rename to requirements/requirements_amd.txt
diff --git a/requirements_amd_noavx2.txt b/requirements/requirements_amd_noavx2.txt
similarity index 100%
rename from requirements_amd_noavx2.txt
rename to requirements/requirements_amd_noavx2.txt
diff --git a/requirements_apple_intel.txt b/requirements/requirements_apple_intel.txt
similarity index 100%
rename from requirements_apple_intel.txt
rename to requirements/requirements_apple_intel.txt
diff --git a/requirements_apple_silicon.txt b/requirements/requirements_apple_silicon.txt
similarity index 100%
rename from requirements_apple_silicon.txt
rename to requirements/requirements_apple_silicon.txt
diff --git a/requirements_cpu_only.txt b/requirements/requirements_cpu_only.txt
similarity index 100%
rename from requirements_cpu_only.txt
rename to requirements/requirements_cpu_only.txt
diff --git a/requirements_cpu_only_noavx2.txt b/requirements/requirements_cpu_only_noavx2.txt
similarity index 100%
rename from requirements_cpu_only_noavx2.txt
rename to requirements/requirements_cpu_only_noavx2.txt
diff --git a/requirements_noavx2.txt b/requirements/requirements_noavx2.txt
similarity index 100%
rename from requirements_noavx2.txt
rename to requirements/requirements_noavx2.txt
diff --git a/requirements_nowheels.txt b/requirements/requirements_nowheels.txt
similarity index 100%
rename from requirements_nowheels.txt
rename to requirements/requirements_nowheels.txt

From c178ea02fe962e9094cec077b5432dd3a51c777a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 20 Apr 2025 19:27:38 -0700
Subject: [PATCH 0335/1701] Revert "Move the requirements*.txt to a
 requirements folder"

This reverts commit 6117ef7d64543f337abe6f0b33dca6b8e9294f9e.
---
 requirements/requirements.txt => requirements.txt                 | 0
 requirements/requirements_amd.txt => requirements_amd.txt         | 0
 .../requirements_amd_noavx2.txt => requirements_amd_noavx2.txt    | 0
 .../requirements_apple_intel.txt => requirements_apple_intel.txt  | 0
 ...quirements_apple_silicon.txt => requirements_apple_silicon.txt | 0
 .../requirements_cpu_only.txt => requirements_cpu_only.txt        | 0
 ...ements_cpu_only_noavx2.txt => requirements_cpu_only_noavx2.txt | 0
 requirements/requirements_noavx2.txt => requirements_noavx2.txt   | 0
 .../requirements_nowheels.txt => requirements_nowheels.txt        | 0
 9 files changed, 0 insertions(+), 0 deletions(-)
 rename requirements/requirements.txt => requirements.txt (100%)
 rename requirements/requirements_amd.txt => requirements_amd.txt (100%)
 rename requirements/requirements_amd_noavx2.txt => requirements_amd_noavx2.txt (100%)
 rename requirements/requirements_apple_intel.txt => requirements_apple_intel.txt (100%)
 rename requirements/requirements_apple_silicon.txt => requirements_apple_silicon.txt (100%)
 rename requirements/requirements_cpu_only.txt => requirements_cpu_only.txt (100%)
 rename requirements/requirements_cpu_only_noavx2.txt => requirements_cpu_only_noavx2.txt (100%)
 rename requirements/requirements_noavx2.txt => requirements_noavx2.txt (100%)
 rename requirements/requirements_nowheels.txt => requirements_nowheels.txt (100%)

diff --git a/requirements/requirements.txt b/requirements.txt
similarity index 100%
rename from requirements/requirements.txt
rename to requirements.txt
diff --git a/requirements/requirements_amd.txt b/requirements_amd.txt
similarity index 100%
rename from requirements/requirements_amd.txt
rename to requirements_amd.txt
diff --git a/requirements/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
similarity index 100%
rename from requirements/requirements_amd_noavx2.txt
rename to requirements_amd_noavx2.txt
diff --git a/requirements/requirements_apple_intel.txt b/requirements_apple_intel.txt
similarity index 100%
rename from requirements/requirements_apple_intel.txt
rename to requirements_apple_intel.txt
diff --git a/requirements/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
similarity index 100%
rename from requirements/requirements_apple_silicon.txt
rename to requirements_apple_silicon.txt
diff --git a/requirements/requirements_cpu_only.txt b/requirements_cpu_only.txt
similarity index 100%
rename from requirements/requirements_cpu_only.txt
rename to requirements_cpu_only.txt
diff --git a/requirements/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
similarity index 100%
rename from requirements/requirements_cpu_only_noavx2.txt
rename to requirements_cpu_only_noavx2.txt
diff --git a/requirements/requirements_noavx2.txt b/requirements_noavx2.txt
similarity index 100%
rename from requirements/requirements_noavx2.txt
rename to requirements_noavx2.txt
diff --git a/requirements/requirements_nowheels.txt b/requirements_nowheels.txt
similarity index 100%
rename from requirements/requirements_nowheels.txt
rename to requirements_nowheels.txt

From 86c3ed3218e0d2033855816bc3bb1054b2071bbc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 20 Apr 2025 20:00:56 -0700
Subject: [PATCH 0336/1701] Small change to the unload_model() function

---
 modules/models.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/modules/models.py b/modules/models.py
index 2c23462aec..99b068aa1d 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -131,6 +131,9 @@ def TensorRT_LLM_loader(model_name):
 
 
 def unload_model(keep_model_name=False):
+    if shared.model is None:
+        return
+
     is_llamacpp = (shared.model.__class__.__name__ == 'LlamaServer')
 
     shared.model = shared.tokenizer = None

From 15989c2ed8fb53a0aafcbd294155ee3c61d15ed1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 21 Apr 2025 16:36:35 -0700
Subject: [PATCH 0337/1701] Make llama.cpp the default loader

---
 modules/loaders.py | 34 +++++++++++++++++-----------------
 server.py          |  2 +-
 2 files changed, 18 insertions(+), 18 deletions(-)

diff --git a/modules/loaders.py b/modules/loaders.py
index f69f17203c..7d6afe80dc 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -4,6 +4,23 @@
 import gradio as gr
 
 loaders_and_params = OrderedDict({
+    'llama.cpp': [
+        'n_gpu_layers',
+        'threads',
+        'threads_batch',
+        'batch_size',
+        'n_ctx',
+        'cache_type',
+        'tensor_split',
+        'rope_freq_base',
+        'compress_pos_emb',
+        'flash_attn',
+        'row_split',
+        'no_kv_offload',
+        'no_mmap',
+        'mlock',
+        'numa',
+    ],
     'Transformers': [
         'gpu_split',
         'cpu_memory',
@@ -23,23 +40,6 @@
         'trust_remote_code',
         'no_use_fast',
     ],
-    'llama.cpp': [
-        'n_gpu_layers',
-        'threads',
-        'threads_batch',
-        'batch_size',
-        'n_ctx',
-        'cache_type',
-        'tensor_split',
-        'rope_freq_base',
-        'compress_pos_emb',
-        'flash_attn',
-        'row_split',
-        'no_kv_offload',
-        'no_mmap',
-        'mlock',
-        'numa',
-    ],
     'ExLlamav3_HF': [
         'max_seq_len',
         'gpu_split',
diff --git a/server.py b/server.py
index 1d26156639..41a5660dd6 100644
--- a/server.py
+++ b/server.py
@@ -85,7 +85,7 @@ def create_interface():
 
     # Force some events to be triggered on page load
     shared.persistent_interface_state.update({
-        'loader': shared.args.loader or 'Transformers',
+        'loader': shared.args.loader or 'llama.cpp',
         'mode': shared.settings['mode'] if shared.settings['mode'] == 'instruct' else gr.update(),
         'character_menu': shared.args.character or shared.settings['character'],
         'instruction_template_str': shared.settings['instruction_template_str'],

From 8320190184c285f6a909c11ee040962a498e6e4f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 21 Apr 2025 18:32:23 -0700
Subject: [PATCH 0338/1701] Fix the exllamav2_HF and exllamav3_HF loaders

---
 modules/sampler_hijack.py | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/modules/sampler_hijack.py b/modules/sampler_hijack.py
index ee871a6eac..dfdb691440 100644
--- a/modules/sampler_hijack.py
+++ b/modules/sampler_hijack.py
@@ -15,6 +15,9 @@
 from modules.logging_colors import logger
 from modules.torch_utils import get_device
 
+original_init = transformers.GenerationConfig.__init__
+original_get_logits_processor = transformers.GenerationMixin._get_logits_processor
+
 global_scores = None
 
 
@@ -484,7 +487,7 @@ def get_logits_processor_patch(self, **kwargs):
         generation_config.temperature = float(generation_config.temperature)  # Must be float
 
     # Get the original warpers
-    warpers = self._get_logits_processor_old(**kwargs)
+    warpers = original_get_logits_processor(self, **kwargs)
 
     for i in range(len(warpers) - 1, -1, -1):
         # Replace temperature with our modified class.
@@ -674,7 +677,7 @@ def custom_sort_key(obj):
 
 
 def generation_config_init_patch(self, **kwargs):
-    self.__init___old(**kwargs)
+    original_init(self, **kwargs)
     self.min_p = kwargs.pop("min_p", 0.0)
     self.dynamic_temperature = kwargs.pop("dynamic_temperature", False)
     self.dynatemp_low = kwargs.pop("dynatemp_low", 1)
@@ -702,8 +705,5 @@ def generation_config_init_patch(self, **kwargs):
 
 
 def hijack_samplers():
-    transformers.GenerationMixin._get_logits_processor_old = transformers.GenerationMixin._get_logits_processor
     transformers.GenerationMixin._get_logits_processor = get_logits_processor_patch
-
-    transformers.GenerationConfig.__init___old = transformers.GenerationConfig.__init__
     transformers.GenerationConfig.__init__ = generation_config_init_patch

From 78aeabca89275939f66c58dff7530966d9886a18 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 21 Apr 2025 18:33:14 -0700
Subject: [PATCH 0339/1701] Fix the transformers loader

---
 modules/models_settings.py | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index f5f3be7a2b..ee2ed71bde 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -197,13 +197,9 @@ def update_model_parameters(state, initial=False):
         if initial and element in shared.provided_arguments:
             continue
 
-        if element in ['cpu_memory'] and value == 0:
+        if element == 'cpu_memory' and value == 0:
             value = vars(shared.args_defaults)[element]
 
-        # Making some simple conversions
-        if element == 'cpu_memory' and value is not None:
-            value = f"{value}MiB"
-
         setattr(shared.args, element, value)
 
 

From ee09e44c85543025def97a990888f4086b80b190 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Tue, 22 Apr 2025 09:25:57 -0300
Subject: [PATCH 0340/1701] Portable version (#6868)

---
 .github/dependabot.yml                        |  10 +-
 .github/workflows/build-everything-tgw.yml    |  49 +++++
 .../workflows/build-portable-release-cuda.yml | 182 +++++++++++++++++
 .github/workflows/build-portable-release.yml  | 192 ++++++++++++++++++
 one_click.py                                  |  12 +-
 .../full/requirements.txt                     |   4 +-
 .../full/requirements_amd.txt                 |   2 +-
 .../full/requirements_amd_noavx2.txt          |   2 +-
 .../full/requirements_apple_intel.txt         |   5 +-
 .../full/requirements_apple_silicon.txt       |   5 +-
 .../full/requirements_cpu_only.txt            |   4 +-
 .../full/requirements_cpu_only_noavx2.txt     |   4 +-
 .../full/requirements_noavx2.txt              |   4 +-
 .../full/requirements_nowheels.txt            |   0
 requirements/portable/requirements.txt        |  19 ++
 requirements/portable/requirements_amd.txt    |  18 ++
 .../portable/requirements_amd_noavx2.txt      |  18 ++
 .../portable/requirements_apple_intel.txt     |  19 ++
 .../portable/requirements_apple_silicon.txt   |  20 ++
 .../portable/requirements_cpu_only.txt        |  19 ++
 .../portable/requirements_cpu_only_noavx2.txt |  19 ++
 requirements/portable/requirements_noavx2.txt |  19 ++
 .../portable/requirements_nowheels.txt        |  15 ++
 start_linux.sh                                |   6 +
 start_macos.sh                                |   6 +
 start_windows.bat                             |   6 +
 26 files changed, 638 insertions(+), 21 deletions(-)
 create mode 100644 .github/workflows/build-everything-tgw.yml
 create mode 100644 .github/workflows/build-portable-release-cuda.yml
 create mode 100644 .github/workflows/build-portable-release.yml
 rename requirements.txt => requirements/full/requirements.txt (86%)
 rename requirements_amd.txt => requirements/full/requirements_amd.txt (82%)
 rename requirements_amd_noavx2.txt => requirements/full/requirements_amd_noavx2.txt (82%)
 rename requirements_apple_silicon.txt => requirements/full/requirements_apple_intel.txt (51%)
 rename requirements_apple_intel.txt => requirements/full/requirements_apple_silicon.txt (70%)
 rename requirements_cpu_only.txt => requirements/full/requirements_cpu_only.txt (63%)
 rename requirements_cpu_only_noavx2.txt => requirements/full/requirements_cpu_only_noavx2.txt (63%)
 rename requirements_noavx2.txt => requirements/full/requirements_noavx2.txt (86%)
 rename requirements_nowheels.txt => requirements/full/requirements_nowheels.txt (100%)
 create mode 100644 requirements/portable/requirements.txt
 create mode 100644 requirements/portable/requirements_amd.txt
 create mode 100644 requirements/portable/requirements_amd_noavx2.txt
 create mode 100644 requirements/portable/requirements_apple_intel.txt
 create mode 100644 requirements/portable/requirements_apple_silicon.txt
 create mode 100644 requirements/portable/requirements_cpu_only.txt
 create mode 100644 requirements/portable/requirements_cpu_only_noavx2.txt
 create mode 100644 requirements/portable/requirements_noavx2.txt
 create mode 100644 requirements/portable/requirements_nowheels.txt

diff --git a/.github/dependabot.yml b/.github/dependabot.yml
index 93aaf445f0..69918a57f1 100644
--- a/.github/dependabot.yml
+++ b/.github/dependabot.yml
@@ -5,8 +5,14 @@
 
 version: 2
 updates:
-  - package-ecosystem: "pip" # See documentation for possible values
-    directory: "/" # Location of package manifests
+  - package-ecosystem: "pip"
+    directory: "/requirements/full/"
+    target-branch: "dev"
+    schedule:
+      interval: "weekly"
+
+  - package-ecosystem: "pip"
+    directory: "/requirements/portable/"
     target-branch: "dev"
     schedule:
       interval: "weekly"
diff --git a/.github/workflows/build-everything-tgw.yml b/.github/workflows/build-everything-tgw.yml
new file mode 100644
index 0000000000..91ca5f6b21
--- /dev/null
+++ b/.github/workflows/build-everything-tgw.yml
@@ -0,0 +1,49 @@
+name: Build Everything TGW
+
+on:
+  workflow_dispatch:
+    inputs:
+      version:
+        description: 'Version tag of text-generation-webui to build: v3.0'
+        default: 'v3.0'
+        required: true
+        type: string
+
+permissions:
+  contents: write
+
+jobs:
+  build_release_cuda_windows:
+    name: CUDA Windows
+    uses: ./.github/workflows/build-portable-release-cuda.yml
+    with:
+      version: ${{ inputs.version }}
+      config: 'os:windows-2019'
+
+  build_release_cuda_linux:
+    name: CUDA Linux
+    uses: ./.github/workflows/build-portable-release-cuda.yml
+    with:
+      version: ${{ inputs.version }}
+      config: 'os:ubuntu-22.04'
+
+  build_release_cpu_windows:
+    name: CPU Windows
+    uses: ./.github/workflows/build-portable-release.yml
+    with:
+      version: ${{ inputs.version }}
+      config: 'os:windows-2019'
+
+  build_release_cpu_linux:
+    name: CPU Linux
+    uses: ./.github/workflows/build-portable-release.yml
+    with:
+      version: ${{ inputs.version }}
+      config: 'os:ubuntu-22.04'
+
+  build_release_macos:
+    name: macOS
+    uses: ./.github/workflows/build-portable-release.yml
+    with:
+      version: ${{ inputs.version }}
+      config: 'os:macos-13,macos-14'
diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
new file mode 100644
index 0000000000..b3cc587a68
--- /dev/null
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -0,0 +1,182 @@
+name: Build CUDA
+
+on:
+  workflow_dispatch:
+    inputs:
+      version:
+        description: 'Version tag of text-generation-webui to build: v3.0'
+        default: 'v3.0'
+        required: true
+        type: string
+      config:
+        description: 'Override configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2'
+        default: 'Default'
+        required: false
+        type: string
+      exclude:
+        description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
+        default: 'None'
+        required: false
+        type: string
+  workflow_call:
+    inputs:
+      version:
+        description: 'Version tag of text-generation-webui to build: v3.0'
+        default: 'v3.0'
+        required: true
+        type: string
+      config:
+        description: 'Configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2'
+        default: 'Default'
+        required: false
+        type: string
+      exclude:
+        description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
+        default: 'None'
+        required: false
+        type: string
+
+permissions:
+  contents: write
+
+jobs:
+  define_matrix:
+    name: Define Build Matrix
+    runs-on: ubuntu-latest
+    outputs:
+      matrix: ${{ steps.set-matrix.outputs.matrix }}
+    defaults:
+      run:
+        shell: pwsh
+    env:
+      CONFIGIN: ${{ inputs.config }}
+      EXCLUDEIN: ${{ inputs.exclude }}
+
+    steps:
+      - name: Define Job Output
+        id: set-matrix
+        run: |
+          $matrix = @{
+              'os' = @('ubuntu-22.04', 'windows-2019')
+              'pyver' = @("3.11")
+              'avx' = @("AVX2")
+              'cuda' = @("11.7", "12.4")
+          }
+
+          if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})}
+
+          if ($env:EXCLUDEIN -ne 'None') {
+              $exclusions = @()
+              $exclusions += $env:EXCLUDEIN.split(';').replace(':','=').replace(',',"`n") | ConvertFrom-StringData
+              $matrix['exclude'] = $exclusions
+          }
+
+          $matrixOut = ConvertTo-Json $matrix -Compress
+          Write-Output ('matrix=' + $matrixOut) >> $env:GITHUB_OUTPUT
+
+  build_wheels:
+    name: ${{ matrix.os }} ${{ matrix.pyver }} CPU ${{ matrix.avx }} CUDA ${{ matrix.cuda }}
+    needs: define_matrix
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix: ${{ fromJSON(needs.define_matrix.outputs.matrix) }}
+    defaults:
+      run:
+        shell: pwsh
+    env:
+      AVXVER: ${{ matrix.avx }}
+      PCKGVER: ${{ inputs.version }}
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          repository: 'oobabooga/text-generation-webui'
+          ref: ${{ inputs.version }}
+          submodules: 'recursive'
+
+      - uses: actions/setup-python@v4
+        with:
+          python-version: ${{ matrix.pyver }}
+
+      - name: Build Package
+        shell: bash
+        run: |
+            rm -rf .git cmd* update_wizard* start_wsl.bat wsl.sh Colab-TextGen-GPU.ipynb docker
+
+            # Define common variables
+            CUDA_VERSION="${{ matrix.cuda }}"
+            AVX_SUPPORT="${{ matrix.avx }}"
+            VERSION="${{ inputs.version }}"
+
+            # 1. Set platform-specific variables
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                PLATFORM="windows"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-pc-windows-msvc-install_only.tar.gz"
+                PIP_PATH="portable_env/python.exe -m pip"
+                PACKAGES_PATH="portable_env/Lib/site-packages"
+                ZIP_CMD="powershell -Command \"Compress-Archive -Path text-generation-webui -DestinationPath"
+                rm start_linux.sh start_macos.sh
+            else
+                PLATFORM="linux"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-unknown-linux-gnu-install_only.tar.gz"
+                PIP_PATH="portable_env/bin/python -m pip"
+                PACKAGES_PATH="portable_env/lib/python3.11/site-packages"
+                ZIP_CMD="zip -r"
+                rm start_macos.sh start_windows.bat
+            fi
+
+            # 2. Download and extract Python
+            cd ..
+            echo "Downloading Python for $PLATFORM..."
+            curl -L -o python-build.tar.gz "$PYTHON_URL"
+            tar -xzf python-build.tar.gz
+            mv python text-generation-webui/portable_env
+
+            # 3. Prepare requirements file based on AVX and CUDA
+            if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
+                BASE_REQ_FILE="requirements/portable/requirements.txt"
+            else
+                BASE_REQ_FILE="requirements/portable/requirements_noavx2.txt"
+            fi
+
+            # Create CUDA-specific requirements file if needed
+            cd text-generation-webui
+            if [[ "$CUDA_VERSION" == "11.7" ]]; then
+                echo "Creating CUDA 11.7 specific requirements file"
+                sed 's/cu124/cu117/g' "$BASE_REQ_FILE" > requirements_cuda_temp.txt
+                REQ_FILE="requirements_cuda_temp.txt"
+            else
+                REQ_FILE="$BASE_REQ_FILE"
+            fi
+
+            # 4. Install packages
+            echo "Installing Python packages from $REQ_FILE..."
+            $PIP_PATH install --target="./$PACKAGES_PATH" -r "$REQ_FILE"
+
+            # 5. Clean up
+            if [[ "$CUDA_VERSION" == "11.7" ]]; then
+                rm requirements_cuda_temp.txt
+            fi
+
+            # 6. Create ZIP file
+            cd ..
+            ZIP_NAME="textgen-portable-${VERSION}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
+            echo "Creating archive: $ZIP_NAME"
+
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                powershell -Command "Compress-Archive -Path text-generation-webui -DestinationPath $ZIP_NAME"
+            else
+                zip -r "$ZIP_NAME" text-generation-webui
+            fi
+
+      - name: Upload files to a GitHub release
+        id: upload-release
+        uses: svenstaro/upload-release-action@2.7.0
+        continue-on-error: true
+        with:
+          repo_token: ${{ secrets.GITHUB_TOKEN }}
+          file: ../textgen-portable-${{ inputs.version }}*.zip
+          tag: ${{ inputs.version }}
+          file_glob: true
+          make_latest: false
+          overwrite: true
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
new file mode 100644
index 0000000000..bb61eb8fa9
--- /dev/null
+++ b/.github/workflows/build-portable-release.yml
@@ -0,0 +1,192 @@
+name: Build CPU and macOS
+
+on:
+  workflow_dispatch:
+    inputs:
+      version:
+        description: 'Version tag of text-generation-webui to build: v3.0'
+        default: 'v3.0'
+        required: true
+        type: string
+      config:
+        description: 'Override configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2'
+        default: 'Default'
+        required: false
+        type: string
+      exclude:
+        description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
+        default: 'None'
+        required: false
+        type: string
+  workflow_call:
+    inputs:
+      version:
+        description: 'Version tag of text-generation-webui to build: v3.0'
+        default: 'v3.0'
+        required: true
+        type: string
+      config:
+        description: 'Configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2'
+        default: 'Default'
+        required: false
+        type: string
+      exclude:
+        description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
+        default: 'None'
+        required: false
+        type: string
+
+permissions:
+  contents: write
+
+jobs:
+  define_matrix:
+    name: Define Build Matrix
+    runs-on: ubuntu-latest
+    outputs:
+      matrix: ${{ steps.set-matrix.outputs.matrix }}
+    defaults:
+      run:
+        shell: pwsh
+    env:
+      CONFIGIN: ${{ inputs.config }}
+      EXCLUDEIN: ${{ inputs.exclude }}
+
+    steps:
+      - name: Define Job Output
+        id: set-matrix
+        run: |
+          $matrix = @{
+              'os' = @('ubuntu-22.04', 'windows-2019', 'macos-13', 'macos-14')
+              'pyver' = @("3.11")
+              'avx' = @("AVX2")
+          }
+
+          if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})}
+
+          if ($env:EXCLUDEIN -ne 'None') {
+              $exclusions = @()
+              $exclusions += $env:EXCLUDEIN.split(';').replace(':','=').replace(',',"`n") | ConvertFrom-StringData
+              $matrix['exclude'] = $exclusions
+          }
+
+          $matrixOut = ConvertTo-Json $matrix -Compress
+          Write-Output ('matrix=' + $matrixOut) >> $env:GITHUB_OUTPUT
+
+  build_wheels:
+    name: ${{ matrix.os }} ${{ matrix.pyver }} CPU ${{ matrix.avx }}
+    needs: define_matrix
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix: ${{ fromJSON(needs.define_matrix.outputs.matrix) }}
+    defaults:
+      run:
+        shell: pwsh
+    env:
+      AVXVER: ${{ matrix.avx }}
+      PCKGVER: ${{ inputs.version }}
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          repository: 'oobabooga/text-generation-webui'
+          ref: ${{ inputs.version }}
+          submodules: 'recursive'
+
+      - uses: actions/setup-python@v4
+        with:
+          python-version: ${{ matrix.pyver }}
+
+      - name: Build Package
+        shell: bash
+        run: |
+            rm -rf .git cmd* update_wizard* start_wsl.bat wsl.sh Colab-TextGen-GPU.ipynb docker
+
+            # Define common variables
+            AVX_SUPPORT="${{ matrix.avx }}"
+            VERSION="${{ inputs.version }}"
+            OS_TYPE="${{ matrix.os }}"
+
+            # 1. Set platform-specific variables
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                PLATFORM="windows-cpu"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-pc-windows-msvc-install_only.tar.gz"
+                PIP_PATH="portable_env/python.exe -m pip"
+                PACKAGES_PATH="portable_env/Lib/site-packages"
+                rm start_linux.sh start_macos.sh
+            elif [[ "$RUNNER_OS" == "macOS" ]]; then
+                if [[ "$OS_TYPE" == "macos-13" ]]; then
+                    PLATFORM="macos-x86_64"
+                    PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-apple-darwin-install_only.tar.gz"
+                    REQ_TYPE="apple_intel"
+                else
+                    PLATFORM="macos-arm64"
+                    PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-aarch64-apple-darwin-install_only.tar.gz"
+                    REQ_TYPE="apple_silicon"
+                fi
+                PIP_PATH="portable_env/bin/python -m pip"
+                PACKAGES_PATH="portable_env/lib/python3.11/site-packages"
+                rm start_linux.sh start_windows.bat
+            else
+                # Linux case
+                PLATFORM="linux-cpu"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-unknown-linux-gnu-install_only.tar.gz"
+                PIP_PATH="portable_env/bin/python -m pip"
+                PACKAGES_PATH="portable_env/lib/python3.11/site-packages"
+                rm start_macos.sh start_windows.bat
+            fi
+
+            # 2. Download and extract Python
+            echo "Downloading Python for $PLATFORM..."
+            cd ..
+            curl -L -o python-build.tar.gz "$PYTHON_URL"
+            tar -xzf python-build.tar.gz
+            mv python text-generation-webui/portable_env
+
+            # 3. Prepare requirements file based on platform and AVX
+            cd text-generation-webui
+
+            # Select requirements file based on platform
+            if [[ "$RUNNER_OS" == "macOS" ]]; then
+                if [[ "$OS_TYPE" == "macos-13" ]]; then
+                    REQ_FILE="requirements/portable/requirements_apple_intel.txt"
+                else
+                    REQ_FILE="requirements/portable/requirements_apple_silicon.txt"
+                fi
+            else
+                # For Windows and Linux, check AVX support
+                if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
+                    REQ_FILE="requirements/portable/requirements_cpu_only.txt"
+                else
+                    REQ_FILE="requirements/portable/requirements_cpu_only_noavx2.txt"
+                fi
+            fi
+
+            echo "Using requirements file: $REQ_FILE"
+
+            # 4. Install packages
+            echo "Installing Python packages from $REQ_FILE..."
+            $PIP_PATH install --target="./$PACKAGES_PATH" -r "$REQ_FILE"
+
+            # 5. Create ZIP file
+            cd ..
+            ZIP_NAME="textgen-portable-${VERSION}-${PLATFORM}.zip"
+            echo "Creating archive: $ZIP_NAME"
+
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                powershell -Command "Compress-Archive -Path text-generation-webui -DestinationPath $ZIP_NAME"
+            else
+                zip -r "$ZIP_NAME" text-generation-webui
+            fi
+
+      - name: Upload files to a GitHub release
+        id: upload-release
+        uses: svenstaro/upload-release-action@2.7.0
+        continue-on-error: true
+        with:
+          repo_token: ${{ secrets.GITHUB_TOKEN }}
+          file: ../textgen-portable-${{ inputs.version }}*.zip
+          tag: ${{ inputs.version }}
+          file_glob: true
+          make_latest: false
+          overwrite: true
diff --git a/one_click.py b/one_click.py
index eff2ed9f3c..04b729eb54 100644
--- a/one_click.py
+++ b/one_click.py
@@ -356,14 +356,18 @@ def update_requirements(initial_installation=False, pull=True):
         )
 
     torver = torch_version()
+    requirements_base = os.path.join("requirements", "full")
+
     if "+rocm" in torver:
-        requirements_file = "requirements_amd" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
+        file_name = f"requirements_amd{'_noavx2' if not cpu_has_avx2() else ''}.txt"
     elif "+cpu" in torver or "+cxx11" in torver:
-        requirements_file = "requirements_cpu_only" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
+        file_name = f"requirements_cpu_only{'_noavx2' if not cpu_has_avx2() else ''}.txt"
     elif is_macos():
-        requirements_file = "requirements_apple_" + ("intel" if is_x86_64() else "silicon") + ".txt"
+        file_name = f"requirements_apple_{'intel' if is_x86_64() else 'silicon'}.txt"
     else:
-        requirements_file = "requirements" + ("_noavx2" if not cpu_has_avx2() else "") + ".txt"
+        file_name = f"requirements{'_noavx2' if not cpu_has_avx2() else ''}.txt"
+
+    requirements_file = os.path.join(requirements_base, file_name)
 
     # Load state from JSON file
     current_commit = get_current_commit()
diff --git a/requirements.txt b/requirements/full/requirements.txt
similarity index 86%
rename from requirements.txt
rename to requirements/full/requirements.txt
index 9e55bb3310..b9afaa0795 100644
--- a/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.2.0+cu124-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.2.0+cu124-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements_amd.txt b/requirements/full/requirements_amd.txt
similarity index 82%
rename from requirements_amd.txt
rename to requirements/full/requirements_amd.txt
index ba8bc80ce6..96cb299d8e 100644
--- a/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -29,6 +29,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/rocm/llama_cpp_binaries-0.2.0+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
similarity index 82%
rename from requirements_amd_noavx2.txt
rename to requirements/full/requirements_amd_noavx2.txt
index d921fbb0ba..0f1a4fc2bd 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -29,6 +29,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/rocm/llama_cpp_binaries-0.2.0+rocm6.1.2avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+rocm6.1.2avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements_apple_silicon.txt b/requirements/full/requirements_apple_intel.txt
similarity index 51%
rename from requirements_apple_silicon.txt
rename to requirements/full/requirements_apple_intel.txt
index 3e7fcd020e..8d1e529478 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -29,8 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.2.0-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.2.0-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.2.0-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_apple_intel.txt b/requirements/full/requirements_apple_silicon.txt
similarity index 70%
rename from requirements_apple_intel.txt
rename to requirements/full/requirements_apple_silicon.txt
index 0b92376aaf..a44ff3cb30 100644
--- a/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -29,7 +29,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.2.0-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/metal/llama_cpp_binaries-0.2.0-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
similarity index 63%
rename from requirements_cpu_only.txt
rename to requirements/full/requirements_cpu_only.txt
index 43d54534e6..358551629b 100644
--- a/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.2.0+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.2.0+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
similarity index 63%
rename from requirements_cpu_only_noavx2.txt
rename to requirements/full/requirements_cpu_only_noavx2.txt
index b444444365..0716455eb3 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.2.0+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/cpu/llama_cpp_binaries-0.2.0+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
similarity index 86%
rename from requirements_noavx2.txt
rename to requirements/full/requirements_noavx2.txt
index 351ea83e45..98c43b88d2 100644
--- a/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.2.0+cu124avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/textgen-webui/llama_cpp_binaries-0.2.0+cu124avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
similarity index 100%
rename from requirements_nowheels.txt
rename to requirements/full/requirements_nowheels.txt
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
new file mode 100644
index 0000000000..c3336fc7b6
--- /dev/null
+++ b/requirements/portable/requirements.txt
@@ -0,0 +1,19 @@
+fastapi==0.112.4
+gradio==4.37.*
+jinja2==3.1.6
+markdown
+numpy==1.26.*
+pydantic==2.8.2
+pyyaml
+requests
+rich
+tqdm
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
+
+# CUDA wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
new file mode 100644
index 0000000000..4855225ff1
--- /dev/null
+++ b/requirements/portable/requirements_amd.txt
@@ -0,0 +1,18 @@
+fastapi==0.112.4
+gradio==4.37.*
+jinja2==3.1.6
+markdown
+numpy==1.26.*
+pydantic==2.8.2
+pyyaml
+requests
+rich
+tqdm
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
+
+# AMD wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
new file mode 100644
index 0000000000..f40daa8a0a
--- /dev/null
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -0,0 +1,18 @@
+fastapi==0.112.4
+gradio==4.37.*
+jinja2==3.1.6
+markdown
+numpy==1.26.*
+pydantic==2.8.2
+pyyaml
+requests
+rich
+tqdm
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
+
+# AMD wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+rocm6.1.2avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
new file mode 100644
index 0000000000..1ede251e4b
--- /dev/null
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -0,0 +1,19 @@
+fastapi==0.112.4
+gradio==4.37.*
+jinja2==3.1.6
+markdown
+numpy==1.26.*
+pydantic==2.8.2
+pyyaml
+requests
+rich
+tqdm
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
+
+# Mac wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
new file mode 100644
index 0000000000..26b68bff9c
--- /dev/null
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -0,0 +1,20 @@
+fastapi==0.112.4
+gradio==4.37.*
+jinja2==3.1.6
+markdown
+numpy==1.26.*
+pydantic==2.8.2
+pyyaml
+requests
+rich
+tqdm
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
+
+# Mac wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
new file mode 100644
index 0000000000..456a0499a4
--- /dev/null
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -0,0 +1,19 @@
+fastapi==0.112.4
+gradio==4.37.*
+jinja2==3.1.6
+markdown
+numpy==1.26.*
+pydantic==2.8.2
+pyyaml
+requests
+rich
+tqdm
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
+
+# llama.cpp (CPU only, AVX2)
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
new file mode 100644
index 0000000000..7cd2dd348d
--- /dev/null
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -0,0 +1,19 @@
+fastapi==0.112.4
+gradio==4.37.*
+jinja2==3.1.6
+markdown
+numpy==1.26.*
+pydantic==2.8.2
+pyyaml
+requests
+rich
+tqdm
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
+
+# llama.cpp (CPU only, no AVX2)
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
new file mode 100644
index 0000000000..b47b8bbc3f
--- /dev/null
+++ b/requirements/portable/requirements_noavx2.txt
@@ -0,0 +1,19 @@
+fastapi==0.112.4
+gradio==4.37.*
+jinja2==3.1.6
+markdown
+numpy==1.26.*
+pydantic==2.8.2
+pyyaml
+requests
+rich
+tqdm
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
+
+# CUDA wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
new file mode 100644
index 0000000000..6f9566baed
--- /dev/null
+++ b/requirements/portable/requirements_nowheels.txt
@@ -0,0 +1,15 @@
+fastapi==0.112.4
+gradio==4.37.*
+jinja2==3.1.6
+markdown
+numpy==1.26.*
+pydantic==2.8.2
+pyyaml
+requests
+rich
+tqdm
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
diff --git a/start_linux.sh b/start_linux.sh
index 256604cb7e..00082f0721 100755
--- a/start_linux.sh
+++ b/start_linux.sh
@@ -2,6 +2,12 @@
 
 cd "$(dirname "${BASH_SOURCE[0]}")"
 
+# Portable install case
+if [ -d "portable_env" ]; then
+    ./portable_env/bin/python3 server.py --api --auto-launch "$@"
+    exit $?
+fi
+
 if [[ "$(pwd)" =~ " " ]]; then echo This script relies on Miniconda which can not be silently installed under a path with spaces. && exit; fi
 
 # deactivate existing conda envs as needed to avoid conflicts
diff --git a/start_macos.sh b/start_macos.sh
index 02f1011afa..628f59ccfd 100755
--- a/start_macos.sh
+++ b/start_macos.sh
@@ -2,6 +2,12 @@
 
 cd "$(dirname "${BASH_SOURCE[0]}")"
 
+# Portable install case
+if [ -d "portable_env" ]; then
+    ./portable_env/bin/python3 server.py --api --auto-launch --api-port 5005 "$@"
+    exit $?
+fi
+
 if [[ "$(pwd)" =~ " " ]]; then echo This script relies on Miniconda which can not be silently installed under a path with spaces. && exit; fi
 
 # deactivate existing conda envs as needed to avoid conflicts
diff --git a/start_windows.bat b/start_windows.bat
index 2e42d6fa93..451b85e0d0 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -3,6 +3,12 @@ setlocal enabledelayedexpansion
 
 cd /D "%~dp0"
 
+@rem Portable install case
+if exist "portable_env" (
+    .\portable_env\python.exe server.py --api --auto-launch %*
+    exit /b %errorlevel%
+)
+
 set PATH=%PATH%;%SystemRoot%\system32
 
 echo "%CD%"| findstr /C:" " >nul && echo This script relies on Miniconda which can not be silently installed under a path with spaces. && goto end

From 008c6dd6820556b62e033a17119ec34d62e87c7a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 22 Apr 2025 05:32:41 -0700
Subject: [PATCH 0341/1701] Lint

---
 modules/exllamav2.py    |  2 +-
 modules/exllamav2_hf.py | 18 +++++++++---------
 2 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/modules/exllamav2.py b/modules/exllamav2.py
index 0289bb2109..92e95ac617 100644
--- a/modules/exllamav2.py
+++ b/modules/exllamav2.py
@@ -3,6 +3,7 @@
 from pathlib import Path
 
 import torch
+
 from exllamav2 import (
     ExLlamaV2,
     ExLlamaV2Cache,
@@ -15,7 +16,6 @@
     ExLlamaV2Tokenizer
 )
 from exllamav2.generator import ExLlamaV2Sampler, ExLlamaV2StreamingGenerator
-
 from modules import shared
 from modules.logging_colors import logger
 from modules.text_generation import get_max_prompt_length
diff --git a/modules/exllamav2_hf.py b/modules/exllamav2_hf.py
index b159d9ce44..5ee0bf6005 100644
--- a/modules/exllamav2_hf.py
+++ b/modules/exllamav2_hf.py
@@ -4,6 +4,15 @@
 from typing import Any, Dict, Optional, Union
 
 import torch
+from torch.nn import CrossEntropyLoss
+from transformers import (
+    GenerationConfig,
+    GenerationMixin,
+    PretrainedConfig,
+    PreTrainedModel
+)
+from transformers.modeling_outputs import CausalLMOutputWithPast
+
 from exllamav2 import (
     ExLlamaV2,
     ExLlamaV2Cache,
@@ -14,15 +23,6 @@
     ExLlamaV2Cache_TP,
     ExLlamaV2Config
 )
-from torch.nn import CrossEntropyLoss
-from transformers import (
-    GenerationConfig,
-    GenerationMixin,
-    PretrainedConfig,
-    PreTrainedModel
-)
-from transformers.modeling_outputs import CausalLMOutputWithPast
-
 from modules import shared
 from modules.logging_colors import logger
 

From a3031795a3ef5f77dff968cbf99cf3c558be3fe8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 22 Apr 2025 07:58:19 -0700
Subject: [PATCH 0342/1701] Update the zip filename

---
 .github/workflows/build-portable-release-cuda.yml | 3 ++-
 .github/workflows/build-portable-release.yml      | 3 ++-
 2 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index b3cc587a68..73be7f0ce8 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -160,7 +160,8 @@ jobs:
 
             # 6. Create ZIP file
             cd ..
-            ZIP_NAME="textgen-portable-${VERSION}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
+            VERSION_CLEAN="${VERSION#v}"
+            ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index bb61eb8fa9..0ea912860f 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -170,7 +170,8 @@ jobs:
 
             # 5. Create ZIP file
             cd ..
-            ZIP_NAME="textgen-portable-${VERSION}-${PLATFORM}.zip"
+            VERSION_CLEAN="${VERSION#v}"
+            ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then

From da1919baaefc6032a23472f0409aa13b90f15fc8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 22 Apr 2025 08:02:08 -0700
Subject: [PATCH 0343/1701] Update the README

---
 README.md | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/README.md b/README.md
index a15e974f58..f62e3508fc 100644
--- a/README.md
+++ b/README.md
@@ -27,6 +27,14 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 
 ## How to install
 
+#### Option 1: Portable builds
+
+Compatible with GGUF (llama.cpp) models, just unzip and run, no installation. Available for Windows, Linux, and macOS.
+
+Download from: https://github.com/oobabooga/text-generation-webui/releases
+
+#### Option 2: One-click installer
+
 1) Clone or [download the repository](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip).
 2) Run the script that matches your OS: `start_linux.sh`, `start_windows.bat`, `start_macos.sh`, or `start_wsl.bat`.
 3) Select your GPU vendor when asked.
@@ -352,6 +360,10 @@ Run `python download-model.py --help` to see all the options.
 
 https://colab.research.google.com/github/oobabooga/text-generation-webui/blob/main/Colab-TextGen-GPU.ipynb
 
+## Community
+
+https://www.reddit.com/r/Oobabooga/
+
 ## Acknowledgment
 
 In August 2023, [Andreessen Horowitz](https://a16z.com/) (a16z) provided a generous grant to encourage and support my independent work on this project. I am **extremely** grateful for their trust and recognition.

From 39cbb5fee038e731c8f0880d2c413fb022896eb4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 22 Apr 2025 08:03:25 -0700
Subject: [PATCH 0344/1701] Lint

---
 modules/exllamav2.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/exllamav2.py b/modules/exllamav2.py
index 92e95ac617..0289bb2109 100644
--- a/modules/exllamav2.py
+++ b/modules/exllamav2.py
@@ -3,7 +3,6 @@
 from pathlib import Path
 
 import torch
-
 from exllamav2 import (
     ExLlamaV2,
     ExLlamaV2Cache,
@@ -16,6 +15,7 @@
     ExLlamaV2Tokenizer
 )
 from exllamav2.generator import ExLlamaV2Sampler, ExLlamaV2StreamingGenerator
+
 from modules import shared
 from modules.logging_colors import logger
 from modules.text_generation import get_max_prompt_length

From 25cf3600aa22eb367aee5314ec31291ae637a8df Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 22 Apr 2025 08:04:02 -0700
Subject: [PATCH 0345/1701] Lint

---
 modules/exllamav2_hf.py | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

diff --git a/modules/exllamav2_hf.py b/modules/exllamav2_hf.py
index 5ee0bf6005..b159d9ce44 100644
--- a/modules/exllamav2_hf.py
+++ b/modules/exllamav2_hf.py
@@ -4,15 +4,6 @@
 from typing import Any, Dict, Optional, Union
 
 import torch
-from torch.nn import CrossEntropyLoss
-from transformers import (
-    GenerationConfig,
-    GenerationMixin,
-    PretrainedConfig,
-    PreTrainedModel
-)
-from transformers.modeling_outputs import CausalLMOutputWithPast
-
 from exllamav2 import (
     ExLlamaV2,
     ExLlamaV2Cache,
@@ -23,6 +14,15 @@
     ExLlamaV2Cache_TP,
     ExLlamaV2Config
 )
+from torch.nn import CrossEntropyLoss
+from transformers import (
+    GenerationConfig,
+    GenerationMixin,
+    PretrainedConfig,
+    PreTrainedModel
+)
+from transformers.modeling_outputs import CausalLMOutputWithPast
+
 from modules import shared
 from modules.logging_colors import logger
 

From 4335a24ff8bfef400ebd1d22c2e15db5c18f781a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 22 Apr 2025 08:14:13 -0700
Subject: [PATCH 0346/1701] Fix the workflow

---
 .github/workflows/build-portable-release-cuda.yml | 2 +-
 .github/workflows/build-portable-release.yml      | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index 73be7f0ce8..b9647b1627 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -176,7 +176,7 @@ jobs:
         continue-on-error: true
         with:
           repo_token: ${{ secrets.GITHUB_TOKEN }}
-          file: ../textgen-portable-${{ inputs.version }}*.zip
+          file: ../textgen-portable-*.zip
           tag: ${{ inputs.version }}
           file_glob: true
           make_latest: false
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 0ea912860f..6e041966fa 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -186,7 +186,7 @@ jobs:
         continue-on-error: true
         with:
           repo_token: ${{ secrets.GITHUB_TOKEN }}
-          file: ../textgen-portable-${{ inputs.version }}*.zip
+          file: ../textgen-portable-*.zip
           tag: ${{ inputs.version }}
           file_glob: true
           make_latest: false

From 06619e5f0389bc7dd65aaf877ddae294d4f68563 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 22 Apr 2025 17:46:54 -0700
Subject: [PATCH 0347/1701] Add vulkan requirements.txt files

---
 requirements/portable/requirements_vulkan.txt | 19 +++++++++++++++++++
 .../portable/requirements_vulkan_noavx2.txt   | 19 +++++++++++++++++++
 2 files changed, 38 insertions(+)
 create mode 100644 requirements/portable/requirements_vulkan.txt
 create mode 100644 requirements/portable/requirements_vulkan_noavx2.txt

diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
new file mode 100644
index 0000000000..15834f8997
--- /dev/null
+++ b/requirements/portable/requirements_vulkan.txt
@@ -0,0 +1,19 @@
+fastapi==0.112.4
+gradio==4.37.*
+jinja2==3.1.6
+markdown
+numpy==1.26.*
+pydantic==2.8.2
+pyyaml
+requests
+rich
+tqdm
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
+
+# CUDA wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+vulkan-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+vulkan-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
new file mode 100644
index 0000000000..afb9e90f2f
--- /dev/null
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -0,0 +1,19 @@
+fastapi==0.112.4
+gradio==4.37.*
+jinja2==3.1.6
+markdown
+numpy==1.26.*
+pydantic==2.8.2
+pyyaml
+requests
+rich
+tqdm
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
+
+# CUDA wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+vulkanavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+vulkanavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

From 89ec4c9ba68bd3875cc2ee69a1d03beda1cea6a2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 22 Apr 2025 17:49:49 -0700
Subject: [PATCH 0348/1701] Add vulkan workflow

---
 .github/workflows/build-everything-tgw.yml    |  14 ++
 .../build-portable-release-vulkan.yml         | 168 ++++++++++++++++++
 2 files changed, 182 insertions(+)
 create mode 100644 .github/workflows/build-portable-release-vulkan.yml

diff --git a/.github/workflows/build-everything-tgw.yml b/.github/workflows/build-everything-tgw.yml
index 91ca5f6b21..b3a1e8dd49 100644
--- a/.github/workflows/build-everything-tgw.yml
+++ b/.github/workflows/build-everything-tgw.yml
@@ -27,6 +27,20 @@ jobs:
       version: ${{ inputs.version }}
       config: 'os:ubuntu-22.04'
 
+  build_release_vulkan_windows:
+    name: Vulkan Windows
+    uses: ./.github/workflows/build-portable-release-vulkan.yml
+    with:
+      version: ${{ inputs.version }}
+      config: 'os:windows-2019'
+
+  build_release_vulkan_linux:
+    name: Vulkan Linux
+    uses: ./.github/workflows/build-portable-release-vulkan.yml
+    with:
+      version: ${{ inputs.version }}
+      config: 'os:ubuntu-22.04'
+
   build_release_cpu_windows:
     name: CPU Windows
     uses: ./.github/workflows/build-portable-release.yml
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
new file mode 100644
index 0000000000..287635a304
--- /dev/null
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -0,0 +1,168 @@
+name: Build Vulkan
+
+on:
+  workflow_dispatch:
+    inputs:
+      version:
+        description: 'Version tag of text-generation-webui to build: v3.0'
+        default: 'v3.0'
+        required: true
+        type: string
+      config:
+        description: 'Override configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2'
+        default: 'Default'
+        required: false
+        type: string
+      exclude:
+        description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
+        default: 'None'
+        required: false
+        type: string
+  workflow_call:
+    inputs:
+      version:
+        description: 'Version tag of text-generation-webui to build: v3.0'
+        default: 'v3.0'
+        required: true
+        type: string
+      config:
+        description: 'Configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2'
+        default: 'Default'
+        required: false
+        type: string
+      exclude:
+        description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
+        default: 'None'
+        required: false
+        type: string
+
+permissions:
+  contents: write
+
+jobs:
+  define_matrix:
+    name: Define Build Matrix
+    runs-on: ubuntu-latest
+    outputs:
+      matrix: ${{ steps.set-matrix.outputs.matrix }}
+    defaults:
+      run:
+        shell: pwsh
+    env:
+      CONFIGIN: ${{ inputs.config }}
+      EXCLUDEIN: ${{ inputs.exclude }}
+
+    steps:
+      - name: Define Job Output
+        id: set-matrix
+        run: |
+          $matrix = @{
+              'os' = @('ubuntu-22.04', 'windows-2019')
+              'pyver' = @("3.11")
+              'avx' = @("AVX2")
+          }
+
+          if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})}
+
+          if ($env:EXCLUDEIN -ne 'None') {
+              $exclusions = @()
+              $exclusions += $env:EXCLUDEIN.split(';').replace(':','=').replace(',',"`n") | ConvertFrom-StringData
+              $matrix['exclude'] = $exclusions
+          }
+
+          $matrixOut = ConvertTo-Json $matrix -Compress
+          Write-Output ('matrix=' + $matrixOut) >> $env:GITHUB_OUTPUT
+
+  build_wheels:
+    name: ${{ matrix.os }} ${{ matrix.pyver }} CPU ${{ matrix.avx }}
+    needs: define_matrix
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix: ${{ fromJSON(needs.define_matrix.outputs.matrix) }}
+    defaults:
+      run:
+        shell: pwsh
+    env:
+      AVXVER: ${{ matrix.avx }}
+      PCKGVER: ${{ inputs.version }}
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          repository: 'oobabooga/text-generation-webui'
+          ref: ${{ inputs.version }}
+          submodules: 'recursive'
+
+      - uses: actions/setup-python@v4
+        with:
+          python-version: ${{ matrix.pyver }}
+
+      - name: Build Package
+        shell: bash
+        run: |
+            rm -rf .git cmd* update_wizard* start_wsl.bat wsl.sh Colab-TextGen-GPU.ipynb docker
+
+            # Define common variables
+            AVX_SUPPORT="${{ matrix.avx }}"
+            VERSION="${{ inputs.version }}"
+
+            # 1. Set platform-specific variables
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                PLATFORM="windows"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-pc-windows-msvc-install_only.tar.gz"
+                PIP_PATH="portable_env/python.exe -m pip"
+                PACKAGES_PATH="portable_env/Lib/site-packages"
+                ZIP_CMD="powershell -Command \"Compress-Archive -Path text-generation-webui -DestinationPath"
+                rm start_linux.sh start_macos.sh
+            else
+                PLATFORM="linux"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-unknown-linux-gnu-install_only.tar.gz"
+                PIP_PATH="portable_env/bin/python -m pip"
+                PACKAGES_PATH="portable_env/lib/python3.11/site-packages"
+                ZIP_CMD="zip -r"
+                rm start_macos.sh start_windows.bat
+            fi
+
+            # 2. Download and extract Python
+            cd ..
+            echo "Downloading Python for $PLATFORM..."
+            curl -L -o python-build.tar.gz "$PYTHON_URL"
+            tar -xzf python-build.tar.gz
+            mv python text-generation-webui/portable_env
+
+            # 3. Prepare requirements file based on AVX
+            if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
+                BASE_REQ_FILE="requirements/portable/requirements_vulkan.txt"
+            else
+                BASE_REQ_FILE="requirements/portable/requirements_vulkan_noavx2.txt"
+            fi
+
+            cd text-generation-webui
+
+            # 4. Install packages
+            echo "Installing Python packages from $REQ_FILE..."
+            $PIP_PATH install --target="./$PACKAGES_PATH" -r "$REQ_FILE"
+
+            # 6. Create ZIP file
+            cd ..
+            VERSION_CLEAN="${VERSION#v}"
+            ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.zip"
+            echo "Creating archive: $ZIP_NAME"
+
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                powershell -Command "Compress-Archive -Path text-generation-webui -DestinationPath $ZIP_NAME"
+            else
+                zip -r "$ZIP_NAME" text-generation-webui
+            fi
+
+      - name: Upload files to a GitHub release
+        id: upload-release
+        uses: svenstaro/upload-release-action@2.7.0
+        continue-on-error: true
+        with:
+          repo_token: ${{ secrets.GITHUB_TOKEN }}
+          file: ../textgen-portable-*.zip
+          tag: ${{ inputs.version }}
+          file_glob: true
+          make_latest: false
+          overwrite: true

From 765de6f6781e8e11f91a5fbd6ec04b126278f5af Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 22 Apr 2025 17:53:56 -0700
Subject: [PATCH 0349/1701] Temporary change

---
 .github/workflows/build-portable-release-vulkan.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 287635a304..6138ae4694 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -90,7 +90,7 @@ jobs:
       - uses: actions/checkout@v4
         with:
           repository: 'oobabooga/text-generation-webui'
-          ref: ${{ inputs.version }}
+          ref: 'main'
           submodules: 'recursive'
 
       - uses: actions/setup-python@v4

From 8228822a6ce3fe96d551b88739df646fc52a16b4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 22 Apr 2025 18:01:47 -0700
Subject: [PATCH 0350/1701] Revert "Temporary change"

This reverts commit 765de6f6781e8e11f91a5fbd6ec04b126278f5af.
---
 .github/workflows/build-portable-release-vulkan.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 6138ae4694..287635a304 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -90,7 +90,7 @@ jobs:
       - uses: actions/checkout@v4
         with:
           repository: 'oobabooga/text-generation-webui'
-          ref: 'main'
+          ref: ${{ inputs.version }}
           submodules: 'recursive'
 
       - uses: actions/setup-python@v4

From 812d878812e5527c6f575031f28d012e128012d6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 22 Apr 2025 18:35:22 -0700
Subject: [PATCH 0351/1701] Make the dependabot less spammy

---
 .github/dependabot.yml | 10 +++-------
 1 file changed, 3 insertions(+), 7 deletions(-)

diff --git a/.github/dependabot.yml b/.github/dependabot.yml
index 69918a57f1..6a4cb96afa 100644
--- a/.github/dependabot.yml
+++ b/.github/dependabot.yml
@@ -6,13 +6,9 @@
 version: 2
 updates:
   - package-ecosystem: "pip"
-    directory: "/requirements/full/"
-    target-branch: "dev"
-    schedule:
-      interval: "weekly"
-
-  - package-ecosystem: "pip"
-    directory: "/requirements/portable/"
+    directories: "/requirements/full/"
+      - "/requirements/full/"
+      - "/requirements/portable/"
     target-branch: "dev"
     schedule:
       interval: "weekly"

From bce1b68ca97eac64783a5efd05944b588ceb7db7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 22 Apr 2025 18:37:29 -0700
Subject: [PATCH 0352/1701] Minor fix after previous commit

---
 .github/dependabot.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/dependabot.yml b/.github/dependabot.yml
index 6a4cb96afa..9f7c99a63b 100644
--- a/.github/dependabot.yml
+++ b/.github/dependabot.yml
@@ -6,7 +6,7 @@
 version: 2
 updates:
   - package-ecosystem: "pip"
-    directories: "/requirements/full/"
+    directories:
       - "/requirements/full/"
       - "/requirements/portable/"
     target-branch: "dev"

From 9424ba17c8e5b8dcd5d89b5cd1735e5d210bea4e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 22 Apr 2025 19:56:42 -0700
Subject: [PATCH 0353/1701] UI: show only part 00001 of multipart GGUF models
 in the model menu

---
 modules/utils.py | 34 ++++++++++++++++++++++------------
 1 file changed, 22 insertions(+), 12 deletions(-)

diff --git a/modules/utils.py b/modules/utils.py
index f6be7541aa..269561aad9 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -76,44 +76,54 @@ def get_available_models():
     # Get all GGUF files
     gguf_files = get_available_ggufs()
 
+    # Filter out non-first parts of multipart GGUF files
+    filtered_gguf_files = []
+    for gguf_path in gguf_files:
+        filename = os.path.basename(gguf_path)
+
+        match = re.search(r'-(\d+)-of-\d+\.gguf$', filename)
+
+        if match:
+            part_number = match.group(1)
+            # Keep only if it's part 1
+            if part_number.lstrip("0") == "1":
+                filtered_gguf_files.append(gguf_path)
+        else:
+            # Not a multi-part file
+            filtered_gguf_files.append(gguf_path)
+
     model_dir = Path(shared.args.model_dir)
 
     # Find top-level directories containing GGUF files
     dirs_with_gguf = set()
     for gguf_path in gguf_files:
         path = Path(gguf_path)
-        if path.parts:  # If in a subdirectory
-            dirs_with_gguf.add(path.parts[0])  # Add top-level directory
+        if path.parts:
+            dirs_with_gguf.add(path.parts[0])
 
-    # Find directories with safetensors files directly under them
+    # Find directories with safetensors files
     dirs_with_safetensors = set()
     for item in os.listdir(model_dir):
         item_path = model_dir / item
         if item_path.is_dir():
-            # Check if there are safetensors files directly under this directory
             if any(file.lower().endswith(('.safetensors', '.pt')) for file in os.listdir(item_path) if (item_path / file).is_file()):
                 dirs_with_safetensors.add(item)
 
     # Find valid model directories
     model_dirs = []
-
     for item in os.listdir(model_dir):
         item_path = model_dir / item
-
-        # Skip if not a directory
         if not item_path.is_dir():
             continue
 
-        # Include directory if it either:
-        # 1. Doesn't contain GGUF files, OR
-        # 2. Contains both GGUF and safetensors files
+        # Include directory if it either doesn't contain GGUF files
+        # or contains both GGUF and safetensors files
         if item not in dirs_with_gguf or item in dirs_with_safetensors:
             model_dirs.append(item)
 
     model_dirs = sorted(model_dirs, key=natural_keys)
 
-    # Combine all models
-    return ['None'] + gguf_files + model_dirs
+    return ['None'] + filtered_gguf_files + model_dirs
 
 
 def get_available_ggufs():

From e99c20bcb036c47255d652026bb28f877b29651b Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Wed, 23 Apr 2025 20:10:16 -0300
Subject: [PATCH 0354/1701] llama.cpp: Add speculative decoding (#6891)

---
 css/main.css                |  8 ++++++++
 modules/llama_cpp_server.py | 20 ++++++++++++++++++++
 modules/loaders.py          |  6 ++++++
 modules/shared.py           |  9 +++++++++
 modules/training.py         |  4 ++--
 modules/ui.py               |  5 +++++
 modules/ui_model_menu.py    | 11 +++++++++++
 7 files changed, 61 insertions(+), 2 deletions(-)

diff --git a/css/main.css b/css/main.css
index a3fa9753f0..1545a74b12 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1291,3 +1291,11 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 .dark .footer-button:hover svg {
     stroke: rgb(209 213 219);
 }
+
+.tgw-accordion {
+    padding: 10px 12px !important;
+}
+
+.dark .tgw-accordion {
+    border: 1px solid var(--border-color-dark);
+}
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index c88f945d66..ecc543f33c 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -6,6 +6,7 @@
 import sys
 import threading
 import time
+from pathlib import Path
 
 import llama_cpp_binaries
 import requests
@@ -281,6 +282,25 @@ def _start_server(self):
             cmd += ["--rope-freq-scale", str(1.0 / shared.args.compress_pos_emb)]
         if shared.args.rope_freq_base > 0:
             cmd += ["--rope-freq-base", str(shared.args.rope_freq_base)]
+        if shared.args.model_draft not in [None, 'None']:
+            path = Path(shared.args.model_draft)
+            if not path.exists():
+                path = Path(f'{shared.args.model_dir}/{shared.args.model_draft}')
+
+            if path.is_file():
+                model_file = path
+            else:
+                model_file = sorted(Path(f'{shared.args.model_dir}/{shared.args.model_draft}').glob('*.gguf'))[0]
+
+            cmd += ["--model-draft", model_file]
+            if shared.args.draft_max > 0:
+                cmd += ["--draft-max", str(shared.args.draft_max)]
+            if shared.args.gpu_layers_draft > 0:
+                cmd += ["--gpu-layers-draft", str(shared.args.gpu_layers_draft)]
+            if shared.args.device_draft:
+                cmd += ["--device-draft", shared.args.device_draft]
+            if shared.args.ctx_size_draft > 0:
+                cmd += ["--ctx-size-draft", str(shared.args.ctx_size_draft)]
 
         env = os.environ.copy()
         if os.name == 'posix':
diff --git a/modules/loaders.py b/modules/loaders.py
index 7d6afe80dc..167b2c9869 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -20,6 +20,12 @@
         'no_mmap',
         'mlock',
         'numa',
+        'model_draft',
+        'draft_max',
+        'gpu_layers_draft',
+        'device_draft',
+        'ctx_size_draft',
+        'speculative_decoding_accordion'
     ],
     'Transformers': [
         'gpu_split',
diff --git a/modules/shared.py b/modules/shared.py
index 08268ae0bc..e531cd3cf2 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -13,6 +13,7 @@
 model = None
 tokenizer = None
 model_name = 'None'
+draft_model_name = 'None'
 is_seq2seq = False
 model_dirty_from_training = False
 lora_names = []
@@ -127,6 +128,14 @@
 group.add_argument('--no-kv-offload', action='store_true', help='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
 group.add_argument('--row-split', action='store_true', help='Split the model by rows across GPUs. This may improve multi-gpu performance.')
 
+# Speculative decoding
+group = parser.add_argument_group('Speculative decoding')
+group.add_argument('--model-draft', type=str, default=None, help='Path to the draft model for speculative decoding.')
+group.add_argument('--draft-max', type=int, default=4, help='Number of tokens to draft for speculative decoding.')
+group.add_argument('--gpu-layers-draft', type=int, default=0, help='Number of layers to offload to the GPU for the draft model.')
+group.add_argument('--device-draft', type=str, default=None, help='Comma-separated list of devices to use for offloading the draft model.')
+group.add_argument('--ctx-size-draft', type=int, default=0, help='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
+
 # ExLlamaV2
 group = parser.add_argument_group('ExLlamaV2')
 group.add_argument('--gpu-split', type=str, help='Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.')
diff --git a/modules/training.py b/modules/training.py
index c6c380a387..69142463e8 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -52,7 +52,7 @@ def create_ui():
                         with gr.Column():
                             always_override = gr.Checkbox(label='Override Existing Files', value=False, info='If the name is the same, checking will replace the existing file, and unchecking will load and continue from it (the rank must be the same).', elem_classes=['no-background'])
 
-                    with gr.Accordion(label='Target Modules', open=False):
+                    with gr.Accordion(label='Target Modules', open=False, elem_classes='tgw-accordion'):
                         gr.Markdown("Selects which modules to target in training. Targeting more modules is closer to a full fine-tune at the cost of increased VRAM requirements and adapter size.\nNOTE: Only works for model_id='llama', other types will retain default training behavior and not use these settings.")
                         with gr.Row():
                             with gr.Column():
@@ -86,7 +86,7 @@ def create_ui():
                             with gr.Row():
                                 lr_scheduler_type = gr.Dropdown(label='LR Scheduler', value='linear', choices=['linear', 'constant', 'constant_with_warmup', 'cosine', 'cosine_with_restarts', 'polynomial', 'inverse_sqrt'], info='Learning rate scheduler - defines how the learning rate changes over time. "Constant" means never change, "linear" means to go in a straight line from the learning rate down to 0, cosine follows a curve, etc.', elem_classes=['slim-dropdown'])
 
-                    with gr.Accordion(label='Advanced Options', open=False):
+                    with gr.Accordion(label='Advanced Options', open=False, elem_classes='tgw-accordion'):
                         with gr.Row():
                             with gr.Column():
                                 lora_dropout = gr.Slider(label='LoRA Dropout', minimum=0.0, maximum=1.0, step=0.025, value=0.05, info='Percentage probability for dropout of LoRA layers. This can help reduce overfitting. Most users should leave at default.')
diff --git a/modules/ui.py b/modules/ui.py
index d5caaeaaaf..6fc5e955a0 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -145,6 +145,11 @@ def list_model_elements():
         'cpp_runner',
         'trust_remote_code',
         'no_use_fast',
+        'model_draft',
+        'draft_max',
+        'gpu_layers_draft',
+        'device_draft',
+        'ctx_size_draft',
     ]
 
     return elements
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index b4af771cfb..1b0c25fad0 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -92,6 +92,17 @@ def create_ui():
                             shared.gradio['exllamav2_info'] = gr.Markdown("ExLlamav2_HF is recommended over ExLlamav2 for better integration with extensions and more consistent sampling behavior across loaders.")
                             shared.gradio['tensorrt_llm_info'] = gr.Markdown('* TensorRT-LLM has to be installed manually in a separate Python 3.10 environment at the moment. For a guide, consult the description of [this PR](https://github.com/oobabooga/text-generation-webui/pull/5715). \n\n* `max_seq_len` is only used when `cpp-runner` is checked.\n\n* `cpp_runner` does not support streaming at the moment.')
 
+                            # Speculative decoding
+                            with gr.Accordion("Speculative decoding", open=False, elem_classes='tgw-accordion') as shared.gradio['speculative_decoding_accordion']:
+                                with gr.Row():
+                                    shared.gradio['model_draft'] = gr.Dropdown(label="model-draft", choices=utils.get_available_models(), value=lambda: shared.draft_model_name, elem_classes='slim-dropdown', interactive=not mu)
+                                    ui.create_refresh_button(shared.gradio['model_draft'], lambda: None, lambda: {'choices': utils.get_available_models()}, 'refresh-button', interactive=not mu)
+
+                                shared.gradio['draft_max'] = gr.Number(label="draft-max", precision=0, step=1, value=shared.args.draft_max, info='Number of tokens to draft for speculative decoding.')
+                                shared.gradio['gpu_layers_draft'] = gr.Slider(label="gpu-layers-draft", minimum=0, maximum=256, value=shared.args.gpu_layers_draft, info='Number of layers to offload to the GPU for the draft model.')
+                                shared.gradio['device_draft'] = gr.Textbox(label="device-draft", value=shared.args.device_draft, info='Comma-separated list of devices to use for offloading the draft model.')
+                                shared.gradio['ctx_size_draft'] = gr.Number(label="ctx-size-draft", precision=0, step=256, value=shared.args.ctx_size_draft, info='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
+
             with gr.Column():
                 with gr.Row():
                     shared.gradio['autoload_model'] = gr.Checkbox(value=shared.settings['autoload_model'], label='Autoload the model', info='Whether to load the model as soon as it is selected in the Model dropdown.', interactive=not mu)

From bfbde73409b185f5743b173906cf69edaf4447d0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 24 Apr 2025 07:08:49 -0700
Subject: [PATCH 0355/1701] Make 'instruct' the default chat mode

---
 modules/shared.py      | 2 +-
 modules/ui_chat.py     | 2 +-
 settings-template.yaml | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index e531cd3cf2..356a2bb909 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -32,7 +32,7 @@
 settings = {
     'show_controls': True,
     'start_with': '',
-    'mode': 'chat-instruct',
+    'mode': 'instruct',
     'chat_style': 'cai-chat',
     'chat-instruct_command': 'Continue the chat dialogue below. Write a single reply for the character "<|character|>".\n\n<|prompt|>',
     'prompt-default': 'QA',
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index a830abfb10..b823b8e5f6 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -88,7 +88,7 @@ def create_ui():
                     shared.gradio['start_with'] = gr.Textbox(label='Start reply with', placeholder='Sure thing!', value=shared.settings['start_with'], elem_classes=['add_scrollbar'])
 
                 with gr.Row():
-                    shared.gradio['mode'] = gr.Radio(choices=['chat', 'chat-instruct', 'instruct'], value=shared.settings['mode'] if shared.settings['mode'] in ['chat', 'chat-instruct'] else None, label='Mode', info='Defines how the chat prompt is generated. In instruct and chat-instruct modes, the instruction template Parameters > Instruction template is used.', elem_id='chat-mode')
+                    shared.gradio['mode'] = gr.Radio(choices=['instruct', 'chat-instruct', 'chat'], value=shared.settings['mode'] if shared.settings['mode'] in ['chat', 'chat-instruct'] else None, label='Mode', info='Defines how the chat prompt is generated. In instruct and chat-instruct modes, the instruction template Parameters > Instruction template is used.', elem_id='chat-mode')
 
                 with gr.Row():
                     shared.gradio['chat_style'] = gr.Dropdown(choices=utils.get_available_chat_styles(), label='Chat style', value=shared.settings['chat_style'], visible=shared.settings['mode'] != 'instruct')
diff --git a/settings-template.yaml b/settings-template.yaml
index 0343df0abd..94a5c034c3 100644
--- a/settings-template.yaml
+++ b/settings-template.yaml
@@ -1,6 +1,6 @@
 show_controls: true
 start_with: ''
-mode: chat-instruct
+mode: instruct
 chat_style: cai-chat
 chat-instruct_command: |-
   Continue the chat dialogue below. Write a single reply for the character "<|character|>".

From c71a2af5ab3a558ca554d497219386e2d76bedfd Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 24 Apr 2025 08:19:46 -0700
Subject: [PATCH 0356/1701] Handle CMD_FLAGS.txt in the main code (closes
 #6896)

---
 modules/shared.py | 16 ++++++++++++++++
 one_click.py      | 10 ++--------
 2 files changed, 18 insertions(+), 8 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 356a2bb909..799259098b 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -1,6 +1,7 @@
 import argparse
 import copy
 import os
+import shlex
 import sys
 from collections import OrderedDict
 from pathlib import Path
@@ -201,6 +202,21 @@
 # Deprecated parameters
 group = parser.add_argument_group('Deprecated')
 
+# Handle CMD_FLAGS.txt
+cmd_flags_path = Path(__file__).parent.parent / "CMD_FLAGS.txt"
+if cmd_flags_path.exists():
+    with cmd_flags_path.open('r', encoding='utf-8') as f:
+        cmd_flags = ' '.join(
+            line.strip().rstrip('\\').strip()
+            for line in f
+            if line.strip().rstrip('\\').strip() and not line.strip().startswith('#')
+        )
+
+    if cmd_flags:
+        # Command-line takes precedence over CMD_FLAGS.txt
+        sys.argv = [sys.argv[0]] + shlex.split(cmd_flags) + sys.argv[1:]
+
+
 args = parser.parse_args()
 args_defaults = parser.parse_args([])
 provided_arguments = []
diff --git a/one_click.py b/one_click.py
index 04b729eb54..5e3d691b2e 100644
--- a/one_click.py
+++ b/one_click.py
@@ -28,14 +28,7 @@
 state_file = '.installer_state.json'
 
 # Command-line flags
-cmd_flags_path = os.path.join(script_dir, "CMD_FLAGS.txt")
-if os.path.exists(cmd_flags_path):
-    with open(cmd_flags_path, 'r') as f:
-        CMD_FLAGS = ' '.join(line.strip().rstrip('\\').strip() for line in f if line.strip().rstrip('\\').strip() and not line.strip().startswith('#'))
-else:
-    CMD_FLAGS = ''
-
-flags = f"{' '.join([flag for flag in sys.argv[1:] if flag != '--update-wizard'])} {CMD_FLAGS}"
+flags = f"{' '.join([flag for flag in sys.argv[1:] if flag != '--update-wizard'])}"
 
 
 def signal_handler(sig, frame):
@@ -300,6 +293,7 @@ def install_webui():
 
     # Write a flag to CMD_FLAGS.txt for CPU mode
     if selected_gpu == "NONE":
+        cmd_flags_path = os.path.join(script_dir, "CMD_FLAGS.txt")
         with open(cmd_flags_path, 'r+') as cmd_flags_file:
             if "--cpu" not in cmd_flags_file.read():
                 print_big_message("Adding the --cpu flag to CMD_FLAGS.txt.")

From b313adf653059b1e3bc58e1755f7dc99f9584859 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 24 Apr 2025 08:26:12 -0700
Subject: [PATCH 0357/1701] Bump llama.cpp, make the wheels work with any
 Python >= 3.7

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 2 +-
 requirements/portable/requirements_amd_noavx2.txt      | 2 +-
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 36 insertions(+), 36 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index b9afaa0795..74b1089a32 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 96cb299d8e..e75fcfd426 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -29,6 +29,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 0f1a4fc2bd..a57fac6ef1 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -29,6 +29,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+rocm6.1.2avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 8d1e529478..2c5b578003 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index a44ff3cb30..01f45dcbc6 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -29,8 +29,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 358551629b..07845af299 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 0716455eb3..ffa2b6eb0d 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 98c43b88d2..2152ff3edf 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index c3336fc7b6..fcd221e132 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 4855225ff1..3564b3bf91 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -15,4 +15,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+rocm6.1.2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index f40daa8a0a..98afdc67d2 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -15,4 +15,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+rocm6.1.2avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 1ede251e4b..433c1f17fa 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 26b68bff9c..9d9c6852aa 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -15,6 +15,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 456a0499a4..dd25a9d596 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx2-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 7cd2dd348d..dd7f740c27 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index b47b8bbc3f..7542e89766 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+cu124avx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 15834f8997..4810bd5009 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+vulkan-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+vulkan-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index afb9e90f2f..d1787e68e2 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+vulkanavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.3.0/llama_cpp_binaries-0.3.0+vulkanavx-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

From 60ac495d59c42d7d3d798910f00f7b922315b09f Mon Sep 17 00:00:00 2001
From: Ziya <32464406+ZiyaCu@users.noreply.github.com>
Date: Thu, 24 Apr 2025 18:42:05 +0300
Subject: [PATCH 0358/1701] extensions/superboogav2: existing embedding check
 bug fix (#6898)

---
 extensions/superboogav2/chromadb.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/extensions/superboogav2/chromadb.py b/extensions/superboogav2/chromadb.py
index c9e450e4b5..6e93dd9238 100644
--- a/extensions/superboogav2/chromadb.py
+++ b/extensions/superboogav2/chromadb.py
@@ -148,7 +148,7 @@ def _split_texts_by_cache_hit(self, texts: list[str], new_ids: list[str], metada
             id_ = new_ids[i]
             metadata = metadatas[i] if metadatas is not None else None
             embedding = self.embeddings_cache.get(text)
-            if embedding is not None and embedding.any():
+            if embedding is not None and any(embedding):
                 existing_texts.append(text)
                 existing_embeddings.append(embedding)
                 existing_ids.append(id_)

From f1b64df8ddd79833e685b22ed7447da86b5d7e46 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 24 Apr 2025 09:03:49 -0700
Subject: [PATCH 0359/1701] EXL2: add another torch.cuda.synchronize() call to
 prevent errors

---
 modules/text_generation.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index 40046eb29c..4e3d1d7adc 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -264,6 +264,11 @@ def apply_stopping_strings(reply, all_stop_strings):
 
 
 def get_reply_from_output_ids(output_ids, state=None, starting_from=0):
+    import torch
+
+    if torch.cuda.is_available():
+        torch.cuda.synchronize()
+
     reply = decode(output_ids[starting_from:], state['skip_special_tokens'] if state else True)
 
     # Handle tokenizers that do not add the leading space for the first token

From 93fd4ad25d4b98439c4ba2abecc7362ed9b8bd27 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 24 Apr 2025 09:20:11 -0700
Subject: [PATCH 0360/1701] llama.cpp: Document the --device-draft syntax

---
 modules/shared.py        | 2 +-
 modules/ui_model_menu.py | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 799259098b..eeaeb689af 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -134,7 +134,7 @@
 group.add_argument('--model-draft', type=str, default=None, help='Path to the draft model for speculative decoding.')
 group.add_argument('--draft-max', type=int, default=4, help='Number of tokens to draft for speculative decoding.')
 group.add_argument('--gpu-layers-draft', type=int, default=0, help='Number of layers to offload to the GPU for the draft model.')
-group.add_argument('--device-draft', type=str, default=None, help='Comma-separated list of devices to use for offloading the draft model.')
+group.add_argument('--device-draft', type=str, default=None, help='Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1')
 group.add_argument('--ctx-size-draft', type=int, default=0, help='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
 
 # ExLlamaV2
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 1b0c25fad0..56b6903fd9 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -100,7 +100,7 @@ def create_ui():
 
                                 shared.gradio['draft_max'] = gr.Number(label="draft-max", precision=0, step=1, value=shared.args.draft_max, info='Number of tokens to draft for speculative decoding.')
                                 shared.gradio['gpu_layers_draft'] = gr.Slider(label="gpu-layers-draft", minimum=0, maximum=256, value=shared.args.gpu_layers_draft, info='Number of layers to offload to the GPU for the draft model.')
-                                shared.gradio['device_draft'] = gr.Textbox(label="device-draft", value=shared.args.device_draft, info='Comma-separated list of devices to use for offloading the draft model.')
+                                shared.gradio['device_draft'] = gr.Textbox(label="device-draft", value=shared.args.device_draft, info='Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1')
                                 shared.gradio['ctx_size_draft'] = gr.Number(label="ctx-size-draft", precision=0, step=256, value=shared.args.ctx_size_draft, info='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
 
             with gr.Column():

From 8ebe8689163abe89668bbd7bf8a5b77c210cf6e0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 24 Apr 2025 09:31:15 -0700
Subject: [PATCH 0361/1701] Fix typos in
 b313adf653059b1e3bc58e1755f7dc99f9584859

---
 requirements/full/requirements_amd.txt        | 2 +-
 requirements/full/requirements_amd_noavx2.txt | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index e75fcfd426..c8e901904f 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -30,5 +30,5 @@ tiktoken
 
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index a57fac6ef1..d994f6e2c3 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -30,5 +30,5 @@ tiktoken
 
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"

From 8f2493cc60036648aedc38d9d8721993743f86e7 Mon Sep 17 00:00:00 2001
From: Matthew Jenkins <40323108+Matthew-Jenkins@users.noreply.github.com>
Date: Thu, 24 Apr 2025 22:38:57 -0400
Subject: [PATCH 0362/1701] Prevent llamacpp defaults from locking up consumer
 hardware (#6870)

---
 modules/shared.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/shared.py b/modules/shared.py
index eeaeb689af..98ec50b27c 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -120,7 +120,7 @@
 group.add_argument('--n_ctx', type=int, default=8192, help='Size of the prompt context.')
 group.add_argument('--threads', type=int, default=0, help='Number of threads to use.')
 group.add_argument('--threads-batch', type=int, default=0, help='Number of threads to use for batches/prompt processing.')
-group.add_argument('--batch-size', type=int, default=2048, help='Maximum number of prompt tokens to batch together when calling llama_eval.')
+group.add_argument('--batch-size', type=int, default=256, help='Maximum number of prompt tokens to batch together when calling llama_eval.')
 group.add_argument('--no-mmap', action='store_true', help='Prevent mmap from being used.')
 group.add_argument('--mlock', action='store_true', help='Force the system to keep the model in RAM.')
 group.add_argument('--n-gpu-layers', type=int, default=0, help='Number of layers to offload to the GPU.')

From ae1fe8736549b8bea2dee4daebad66f5828860c8 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Fri, 25 Apr 2025 00:11:04 -0300
Subject: [PATCH 0363/1701] ExLlamaV2: Add speculative decoding (#6899)

---
 modules/exllamav2.py | 50 +++++++++++++++++++++++++++++++++++++++++++-
 modules/loaders.py   |  5 ++++-
 2 files changed, 53 insertions(+), 2 deletions(-)

diff --git a/modules/exllamav2.py b/modules/exllamav2.py
index 0289bb2109..7d79e516b2 100644
--- a/modules/exllamav2.py
+++ b/modules/exllamav2.py
@@ -85,7 +85,44 @@ def from_pretrained(self, path_to_model):
             model.load_autosplit(cache)
 
         tokenizer = ExLlamaV2Tokenizer(config)
-        generator = ExLlamaV2StreamingGenerator(model, cache, tokenizer)
+
+        # Initialize draft model for speculative decoding
+        draft_model = None
+        draft_cache = None
+
+        if shared.args.model_draft and shared.args.model_draft.lower() not in ["none", ""]:
+            logger.info(f"Loading draft model for speculative decoding: {shared.args.model_draft}")
+
+            # Find the draft model path
+            draft_path = Path(shared.args.model_draft)
+            if not draft_path.exists():
+                draft_path = Path(f'{shared.args.model_dir}') / Path(shared.args.model_draft)
+
+            draft_config = ExLlamaV2Config()
+            draft_config.model_dir = str(draft_path)
+            draft_config.prepare()
+            draft_config.arch_compat_overrides()
+
+            # Set context size for draft model
+            if shared.args.ctx_size_draft > 0:
+                draft_config.max_seq_len = shared.args.ctx_size_draft
+            else:
+                draft_config.max_seq_len = config.max_seq_len
+
+            draft_model = ExLlamaV2(draft_config)
+            draft_cache = cache_type(draft_model, lazy=True)
+            draft_model.load_autosplit(draft_cache)
+
+            logger.info(f"Draft model loaded successfully with max_draft={shared.args.draft_max}")
+
+        generator = ExLlamaV2StreamingGenerator(
+            model,
+            cache,
+            tokenizer,
+            draft_model=draft_model,
+            draft_cache=draft_cache,
+            num_speculative_tokens=shared.args.draft_max if draft_model is not None else 0
+        )
 
         result = self()
         result.model = model
@@ -93,6 +130,8 @@ def from_pretrained(self, path_to_model):
         result.tokenizer = tokenizer
         result.generator = generator
         result.loras = None
+        result.draft_model = draft_model
+        result.draft_cache = draft_cache
         return result, result
 
     def encode(self, string, **kwargs):
@@ -179,6 +218,10 @@ def generate_with_streaming(self, prompt, state):
         else:
             max_new_tokens = state['max_new_tokens']
 
+        # Reset speculative decoding stats if using a draft model
+        if hasattr(self, 'draft_model') and self.draft_model is not None:
+            self.generator.reset_sd_stats()
+
         self.generator.begin_stream(ids, settings, loras=self.loras)
 
         decoded_text = ''
@@ -190,6 +233,11 @@ def generate_with_streaming(self, prompt, state):
             decoded_text += chunk
             yield decoded_text
 
+        # Log speculative decoding stats if using draft model
+        if hasattr(self, 'draft_model') and self.draft_model is not None:
+            efficiency, accuracy, total_tokens, total_draft_tokens, accepted_draft_tokens = self.generator.get_sd_stats()
+            logger.info(f"Speculative decoding: accepted={accepted_draft_tokens}/{total_draft_tokens} tokens")
+
     def generate(self, prompt, state):
         output = ''
         for output in self.generate_with_streaming(prompt, state):
diff --git a/modules/loaders.py b/modules/loaders.py
index 167b2c9869..d256e1e768 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -25,7 +25,7 @@
         'gpu_layers_draft',
         'device_draft',
         'ctx_size_draft',
-        'speculative_decoding_accordion'
+        'speculative_decoding_accordion',
     ],
     'Transformers': [
         'gpu_split',
@@ -82,6 +82,9 @@
         'no_xformers',
         'no_sdpa',
         'exllamav2_info',
+        'model_draft',
+        'ctx_size_draft',
+        'speculative_decoding_accordion',
     ],
     'HQQ': [
         'hqq_backend',

From a90df27ff59038e5ac55cca5b2b5962e6b79d855 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 24 Apr 2025 20:33:40 -0700
Subject: [PATCH 0364/1701] UI: Add a greeting when the chat history is empty

---
 css/main.css              |  9 +++++++++
 modules/html_generator.py | 16 +++++++++++++++-
 2 files changed, 24 insertions(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 1545a74b12..72adc0710f 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1299,3 +1299,12 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 .dark .tgw-accordion {
     border: 1px solid var(--border-color-dark);
 }
+
+.welcome-greeting {
+    text-align: center;
+    margin-top: 35vh;
+    font-size: 24px;
+    opacity: 0.6;
+    padding-left: 1rem;
+    padding-right: 1rem;
+}
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 144f259316..5227e87ec8 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -1,3 +1,4 @@
+import datetime
 import functools
 import html
 import os
@@ -389,8 +390,21 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
     return output
 
 
+def time_greeting():
+    current_hour = datetime.datetime.now().hour
+    if 5 <= current_hour < 12:
+        return "Good morning!"
+    elif 12 <= current_hour < 18:
+        return "Good afternoon!"
+    else:
+        return "Good evening!"
+
+
 def chat_html_wrapper(history, name1, name2, mode, style, character, reset_cache=False):
-    if mode == 'instruct':
+    if len(history['visible']) == 0:
+        greeting = f"
            {time_greeting()} How can I help you today?
            "
+        result = f'
            {greeting}
            '
+    elif mode == 'instruct':
         result = generate_instruct_html(history)
     elif style == 'wpp':
         result = generate_chat_html(history, name1, name2)

From 23399aff3cfad7cf8547ddde9bff4205e0193666 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 24 Apr 2025 20:39:00 -0700
Subject: [PATCH 0365/1701] UI: minor style change

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 72adc0710f..c15976006a 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1304,7 +1304,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     text-align: center;
     margin-top: 35vh;
     font-size: 24px;
-    opacity: 0.6;
+    opacity: 0.7;
     padding-left: 1rem;
     padding-right: 1rem;
 }

From 5993ebeb1bec1a0e673815b335da3d652ee09c56 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 25 Apr 2025 05:27:59 -0700
Subject: [PATCH 0366/1701] Bump exllamav2 to 0.2.9

---
 requirements/full/requirements.txt               | 6 +++---
 requirements/full/requirements_amd.txt           | 4 ++--
 requirements/full/requirements_amd_noavx2.txt    | 4 ++--
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_noavx2.txt        | 6 +++---
 6 files changed, 12 insertions(+), 12 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 74b1089a32..4a5cd36a30 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_c
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index c8e901904f..98df8e8aa2 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -30,5 +30,5 @@ tiktoken
 
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index d994f6e2c3..9307d73365 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -30,5 +30,5 @@ tiktoken
 
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 2c5b578003..ac9e17ace5 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -32,4 +32,4 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3-py3-none-any.whl
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 01f45dcbc6..ce401ee5a2 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,4 +33,4 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_c
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3-py3-none-any.whl
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 2152ff3edf..6fee3a510f 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_c
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.8/exllamav2-0.2.8-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

From 2c7ff86015023c1ff7e5586ab9475573706e0680 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 25 Apr 2025 05:28:22 -0700
Subject: [PATCH 0367/1701] Bump exllamav3 to
 https://github.com/turboderp-org/exllamav3/commit/de83084184c34cae0aaeca5c8b024de7f148830d

---
 requirements/full/requirements.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_noavx2.txt        | 4 ++--
 4 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 4a5cd36a30..e30e5de79d 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -32,8 +32,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index ac9e17ace5..414e188554 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -31,5 +31,5 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index ce401ee5a2..5627b38e5d 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -32,5 +32,5 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 6fee3a510f..646aaaaf33 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -32,8 +32,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a3/exllamav3-0.0.1a3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From b6fffbd2164793b55196e4b3da6aae5668b62d94 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 25 Apr 2025 05:37:44 -0700
Subject: [PATCH 0368/1701] UI: minor style change

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index c15976006a..8ca5b33f7c 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1302,7 +1302,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 
 .welcome-greeting {
     text-align: center;
-    margin-top: 35vh;
+    margin-top: 40vh;
     font-size: 24px;
     opacity: 0.7;
     padding-left: 1rem;

From 98f4c694b9df06daf99e9cce14a35bf22328e46a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 25 Apr 2025 07:32:51 -0700
Subject: [PATCH 0369/1701] llama.cpp: Add --extra-flags parameter for passing
 additional flags to llama-server

---
 modules/llama_cpp_server.py | 14 ++++++++++++++
 modules/loaders.py          |  1 +
 modules/shared.py           |  1 +
 modules/ui.py               |  1 +
 modules/ui_model_menu.py    |  1 +
 5 files changed, 18 insertions(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index ecc543f33c..7199470dfe 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -301,6 +301,20 @@ def _start_server(self):
                 cmd += ["--device-draft", shared.args.device_draft]
             if shared.args.ctx_size_draft > 0:
                 cmd += ["--ctx-size-draft", str(shared.args.ctx_size_draft)]
+        if shared.args.extra_flags:
+            # Clean up the input
+            extra_flags = shared.args.extra_flags.strip()
+            if extra_flags.startswith('"') and extra_flags.endswith('"'):
+                extra_flags = extra_flags[1:-1].strip()
+            elif extra_flags.startswith("'") and extra_flags.endswith("'"):
+                extra_flags = extra_flags[1:-1].strip()
+
+            for flag_item in extra_flags.split(';'):
+                if '=' in flag_item:
+                    flag, value = flag_item.split('=', 1)
+                    cmd += [f"--{flag}", value]
+                else:
+                    cmd.append(f"--{flag_item}")
 
         env = os.environ.copy()
         if os.name == 'posix':
diff --git a/modules/loaders.py b/modules/loaders.py
index d256e1e768..9442a147e7 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -12,6 +12,7 @@
         'n_ctx',
         'cache_type',
         'tensor_split',
+        'extra_flags',
         'rope_freq_base',
         'compress_pos_emb',
         'flash_attn',
diff --git a/modules/shared.py b/modules/shared.py
index 98ec50b27c..b4f8708273 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -128,6 +128,7 @@
 group.add_argument('--numa', action='store_true', help='Activate NUMA task allocation for llama.cpp.')
 group.add_argument('--no-kv-offload', action='store_true', help='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
 group.add_argument('--row-split', action='store_true', help='Split the model by rows across GPUs. This may improve multi-gpu performance.')
+group.add_argument('--extra-flags', type=str, default=None, help='Extra flags to pass to llama-server. Format: "flag1=value1;flag2;flag3=value3". Example: "override-tensor=exps=CPU"')
 
 # Speculative decoding
 group = parser.add_argument_group('Speculative decoding')
diff --git a/modules/ui.py b/modules/ui.py
index 6fc5e955a0..19b76ceed2 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -114,6 +114,7 @@ def list_model_elements():
         'max_seq_len',
         'cache_type',
         'tensor_split',
+        'extra_flags',
         'gpu_split',
         'alpha_value',
         'rope_freq_base',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 56b6903fd9..f3319cfbb7 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -56,6 +56,7 @@ def create_ui():
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q6', 'q4'], value=shared.args.cache_type, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.')
                             shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
+                            shared.gradio['extra_flags'] = gr.Textbox(label='extra-flags', info='Additional flags to pass to llama-server. Format: "flag1=value1;flag2;flag3=value3". Example: "override-tensor=exps=CPU"')
                             shared.gradio['cpu_memory'] = gr.Number(label="Maximum CPU memory in GiB. Use this for CPU offloading.", value=shared.args.cpu_memory)
                             shared.gradio['alpha_value'] = gr.Number(label='alpha_value', value=shared.args.alpha_value, precision=2, info='Positional embeddings alpha factor for NTK RoPE scaling. Recommended values (NTKv1): 1.75 for 1.5x context, 2.5 for 2x context. Use either this or compress_pos_emb, not both.')
                             shared.gradio['rope_freq_base'] = gr.Number(label='rope_freq_base', value=shared.args.rope_freq_base, precision=0, info='Positional embeddings frequency base for NTK RoPE scaling. Related to alpha_value by rope_freq_base = 10000 * alpha_value ^ (64 / 63). 0 = from model.')

From 0dd71e78c965289c5bfccb3cbc36183f04be23c6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 25 Apr 2025 09:07:14 -0700
Subject: [PATCH 0370/1701] Add `-noavx2` portable builds

---
 .github/workflows/build-portable-release-cuda.yml   |  8 ++++++--
 .github/workflows/build-portable-release-vulkan.yml |  8 ++++++--
 .github/workflows/build-portable-release.yml        | 12 ++++++++----
 3 files changed, 20 insertions(+), 8 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index b9647b1627..fcc744084d 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -59,7 +59,7 @@ jobs:
           $matrix = @{
               'os' = @('ubuntu-22.04', 'windows-2019')
               'pyver' = @("3.11")
-              'avx' = @("AVX2")
+              'avx' = @("AVX2", "AVX")
               'cuda' = @("11.7", "12.4")
           }
 
@@ -161,7 +161,11 @@ jobs:
             # 6. Create ZIP file
             cd ..
             VERSION_CLEAN="${VERSION#v}"
-            ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
+            if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
+                ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
+            else
+                ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}-noavx2.zip"
+            fi
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 287635a304..d5aa764cb9 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -59,7 +59,7 @@ jobs:
           $matrix = @{
               'os' = @('ubuntu-22.04', 'windows-2019')
               'pyver' = @("3.11")
-              'avx' = @("AVX2")
+              'avx' = @("AVX2", "AVX")
           }
 
           if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})}
@@ -146,7 +146,11 @@ jobs:
             # 6. Create ZIP file
             cd ..
             VERSION_CLEAN="${VERSION#v}"
-            ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.zip"
+            if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
+                ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.zip"
+            else
+                ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan-noavx2.zip"
+            fi
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 6e041966fa..2424cc4454 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -15,7 +15,7 @@ on:
         type: string
       exclude:
         description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
-        default: 'None'
+        default: 'os:macos-13,avx:AVX;os:macos-14,avx:AVX'
         required: false
         type: string
   workflow_call:
@@ -32,7 +32,7 @@ on:
         type: string
       exclude:
         description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
-        default: 'None'
+        default: 'os:macos-13,avx:AVX;os:macos-14,avx:AVX'
         required: false
         type: string
 
@@ -59,7 +59,7 @@ jobs:
           $matrix = @{
               'os' = @('ubuntu-22.04', 'windows-2019', 'macos-13', 'macos-14')
               'pyver' = @("3.11")
-              'avx' = @("AVX2")
+              'avx' = @("AVX2", "AVX")
           }
 
           if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})}
@@ -171,7 +171,11 @@ jobs:
             # 5. Create ZIP file
             cd ..
             VERSION_CLEAN="${VERSION#v}"
-            ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.zip"
+            if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
+                ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.zip"
+            else
+                ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-noavx2.zip"
+            fi
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then

From d35818f4e1f51b86d07c7b3c92af9c96facb3e8a Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Fri, 25 Apr 2025 18:02:02 -0300
Subject: [PATCH 0371/1701] UI: Add a collapsible thinking block to messages
 with `` steps (#6902)

---
 css/main.css              | 82 ++++++++++++++++++++++++++++++++--
 js/global_scope_js.js     | 76 ++++++++++++++++++++++++++++++--
 modules/chat.py           | 10 +----
 modules/html_generator.py | 93 ++++++++++++++++++++++++++++++++++++---
 modules/shared.py         |  1 -
 modules/ui.py             |  1 -
 modules/ui_parameters.py  |  1 -
 settings-template.yaml    |  1 -
 8 files changed, 238 insertions(+), 27 deletions(-)

diff --git a/css/main.css b/css/main.css
index 8ca5b33f7c..9dce4d0e0c 100644
--- a/css/main.css
+++ b/css/main.css
@@ -625,19 +625,19 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
         width: 100%;
         overflow-y: visible;
     }
-    
+
     .message {
         break-inside: avoid;
     }
-    
+
     .gradio-container {
         overflow: visible;
     }
-    
+
     .tab-nav {
         display: none !important;
     }
-    
+
     #chat-tab > :first-child {
         max-width: unset;
     }
@@ -1308,3 +1308,77 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     padding-left: 1rem;
     padding-right: 1rem;
 }
+
+/* Thinking blocks styling */
+.thinking-block {
+  margin-bottom: 12px;
+  border-radius: 8px;
+  border: 1px solid rgba(0, 0, 0, 0.1);
+  background-color: var(--light-theme-gray);
+  overflow: hidden;
+}
+
+.dark .thinking-block {
+  background-color: var(--darker-gray);
+}
+
+.thinking-header {
+  display: flex;
+  align-items: center;
+  padding: 10px 16px;
+  cursor: pointer;
+  user-select: none;
+  font-size: 14px;
+  color: rgba(0, 0, 0, 0.7);
+  transition: background-color 0.2s;
+}
+
+.thinking-header:hover {
+  background-color: rgba(0, 0, 0, 0.03);
+}
+
+.thinking-header::-webkit-details-marker {
+  display: none;
+}
+
+.thinking-icon {
+  margin-right: 8px;
+  color: rgba(0, 0, 0, 0.5);
+}
+
+.thinking-title {
+  font-weight: 500;
+}
+
+.thinking-content {
+  padding: 12px 16px;
+  border-top: 1px solid rgba(0, 0, 0, 0.07);
+  color: rgba(0, 0, 0, 0.7);
+  font-size: 14px;
+  line-height: 1.5;
+  overflow-wrap: break-word;
+  max-height: 300px;
+  overflow-y: scroll;
+  contain: layout;
+}
+
+/* Animation for opening thinking blocks */
+@keyframes fadeIn {
+  from { opacity: 0; }
+  to { opacity: 1; }
+}
+
+.thinking-block[open] .thinking-content {
+  animation: fadeIn 0.3s ease-out;
+}
+
+/* Additional style for in-progress thinking */
+.thinking-block[data-streaming="true"] .thinking-title {
+  animation: pulse 1.5s infinite;
+}
+
+@keyframes pulse {
+  0% { opacity: 0.6; }
+  50% { opacity: 1; }
+  100% { opacity: 0.6; }
+}
diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index f308edb961..e808c473c0 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -31,24 +31,94 @@ function removeLastClick() {
 }
 
 function handleMorphdomUpdate(text) {
+  // Track closed blocks
+  const closedBlocks = new Set();
+  document.querySelectorAll(".thinking-block").forEach(block => {
+    const blockId = block.getAttribute("data-block-id");
+    // If block exists and is not open, add to closed set
+    if (blockId && !block.hasAttribute("open")) {
+      closedBlocks.add(blockId);
+    }
+  });
+
+  // Store scroll positions for any open blocks
+  const scrollPositions = {};
+  document.querySelectorAll(".thinking-block[open]").forEach(block => {
+    const content = block.querySelector(".thinking-content");
+    const blockId = block.getAttribute("data-block-id");
+    if (content && blockId) {
+      const isAtBottom = Math.abs((content.scrollHeight - content.scrollTop) - content.clientHeight) < 5;
+      scrollPositions[blockId] = {
+        position: content.scrollTop,
+        isAtBottom: isAtBottom
+      };
+    }
+  });
+
   morphdom(
     document.getElementById("chat").parentNode,
     "
            " + text + "
            ",
     {
       onBeforeElUpdated: function(fromEl, toEl) {
+        // Preserve code highlighting
         if (fromEl.tagName === "PRE" && fromEl.querySelector("code[data-highlighted]")) {
           const fromCode = fromEl.querySelector("code");
           const toCode = toEl.querySelector("code");
 
           if (fromCode && toCode && fromCode.textContent === toCode.textContent) {
-            // If the  content is the same, preserve the entire 
             element
             toEl.className = fromEl.className;
             toEl.innerHTML = fromEl.innerHTML;
-            return false; // Skip updating the 
             element
+            return false;
+          }
+        }
+
+        // For thinking blocks, respect closed state
+        if (fromEl.classList && fromEl.classList.contains("thinking-block") &&
+            toEl.classList && toEl.classList.contains("thinking-block")) {
+          const blockId = toEl.getAttribute("data-block-id");
+          // If this block was closed by user, keep it closed
+          if (blockId && closedBlocks.has(blockId)) {
+            toEl.removeAttribute("open");
+          }
+        }
+
+        return !fromEl.isEqualNode(toEl);
+      },
+
+      onElUpdated: function(el) {
+        // Restore scroll positions for open thinking blocks
+        if (el.classList && el.classList.contains("thinking-block") && el.hasAttribute("open")) {
+          const blockId = el.getAttribute("data-block-id");
+          const content = el.querySelector(".thinking-content");
+
+          if (content && blockId && scrollPositions[blockId]) {
+            setTimeout(() => {
+              if (scrollPositions[blockId].isAtBottom) {
+                content.scrollTop = content.scrollHeight;
+              } else {
+                content.scrollTop = scrollPositions[blockId].position;
+              }
+            }, 0);
           }
         }
-        return !fromEl.isEqualNode(toEl); // Update only if nodes differ
       }
     }
   );
+
+  // Add toggle listeners for new blocks
+  document.querySelectorAll(".thinking-block").forEach(block => {
+    if (!block._hasToggleListener) {
+      block.addEventListener("toggle", function(e) {
+        if (this.open) {
+          const content = this.querySelector(".thinking-content");
+          if (content) {
+            setTimeout(() => {
+              content.scrollTop = content.scrollHeight;
+            }, 0);
+          }
+        }
+      });
+      block._hasToggleListener = true;
+    }
+  });
 }
diff --git a/modules/chat.py b/modules/chat.py
index fd94990750..94d90bdc3d 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -417,16 +417,8 @@ def generate_chat_reply(text, state, regenerate=False, _continue=False, loading_
             yield history
             return
 
-    show_after = html.escape(state.get("show_after")) if state.get("show_after") else None
     for history in chatbot_wrapper(text, state, regenerate=regenerate, _continue=_continue, loading_message=loading_message, for_ui=for_ui):
-        if show_after:
-            after = history["visible"][-1][1].partition(show_after)[2] or "*Is thinking...*"
-            yield {
-                'internal': history['internal'],
-                'visible': history['visible'][:-1] + [[history['visible'][-1][0], after]]
-            }
-        else:
-            yield history
+        yield history
 
 
 def character_is_loaded(state, raise_exception=False):
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 5227e87ec8..a72e485973 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -107,8 +107,87 @@ def replace_blockquote(m):
     return m.group().replace('\n', '\n> ').replace('\\begin{blockquote}', '').replace('\\end{blockquote}', '')
 
 
+def extract_thinking_block(string):
+    """Extract thinking blocks from the beginning of a string."""
+    if not string:
+        return None, string
+
+    THINK_START_TAG = "<think>"
+    THINK_END_TAG = "</think>"
+
+    # Look for opening tag
+    start_pos = string.lstrip().find(THINK_START_TAG)
+    if start_pos == -1:
+        return None, string
+
+    # Adjust start position to account for any leading whitespace
+    start_pos = string.find(THINK_START_TAG)
+
+    # Find the content after the opening tag
+    content_start = start_pos + len(THINK_START_TAG)
+
+    # Look for closing tag
+    end_pos = string.find(THINK_END_TAG, content_start)
+
+    if end_pos != -1:
+        # Both tags found - extract content between them
+        thinking_content = string[content_start:end_pos]
+        remaining_content = string[end_pos + len(THINK_END_TAG):]
+        return thinking_content, remaining_content
+    else:
+        # Only opening tag found - everything else is thinking content
+        thinking_content = string[content_start:]
+        return thinking_content, ""
+
+
 @functools.lru_cache(maxsize=None)
-def convert_to_markdown(string):
+def convert_to_markdown(string, message_id=None):
+    if not string:
+        return ""
+
+    # Use a default message ID if none provided
+    if message_id is None:
+        message_id = "unknown"
+
+    # Extract thinking block if present
+    thinking_content, remaining_content = extract_thinking_block(string)
+
+    # Process the main content
+    html_output = process_markdown_content(remaining_content)
+
+    # If thinking content was found, process it using the same function
+    if thinking_content is not None:
+        thinking_html = process_markdown_content(thinking_content)
+
+        # Generate unique ID for the thinking block
+        block_id = f"thinking-{message_id}-0"
+
+        # Check if thinking is complete or still in progress
+        is_streaming = not remaining_content
+        title_text = "Thinking..." if is_streaming else "Thought"
+
+        thinking_block = f'''
+        
            
+            
+                
+                    
+                    
+                    
+                
+                {title_text}
+            
+            
            {thinking_html}
            
+        
            
+        '''
+
+        # Prepend the thinking block to the message HTML
+        html_output = thinking_block + html_output
+
+    return html_output
+
+
+def process_markdown_content(string):
+    """Process a string through the markdown conversion pipeline."""
     if not string:
         return ""
 
@@ -209,15 +288,15 @@ def convert_to_markdown(string):
     return html_output
 
 
-def convert_to_markdown_wrapped(string, use_cache=True):
+def convert_to_markdown_wrapped(string, message_id=None, use_cache=True):
     '''
     Used to avoid caching convert_to_markdown calls during streaming.
     '''
 
     if use_cache:
-        return convert_to_markdown(string)
+        return convert_to_markdown(string, message_id=message_id)
 
-    return convert_to_markdown.__wrapped__(string)
+    return convert_to_markdown.__wrapped__(string, message_id=message_id)
 
 
 def generate_basic_html(string):
@@ -273,7 +352,7 @@ def generate_instruct_html(history):
     for i in range(len(history['visible'])):
         row_visible = history['visible'][i]
         row_internal = history['internal'][i]
-        converted_visible = [convert_to_markdown_wrapped(entry, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
+        converted_visible = [convert_to_markdown_wrapped(entry, message_id=i, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
 
         if converted_visible[0]:  # Don't display empty user messages
             output += (
@@ -320,7 +399,7 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
     for i in range(len(history['visible'])):
         row_visible = history['visible'][i]
         row_internal = history['internal'][i]
-        converted_visible = [convert_to_markdown_wrapped(entry, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
+        converted_visible = [convert_to_markdown_wrapped(entry, message_id=i, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
 
         if converted_visible[0]:  # Don't display empty user messages
             output += (
@@ -360,7 +439,7 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
     for i in range(len(history['visible'])):
         row_visible = history['visible'][i]
         row_internal = history['internal'][i]
-        converted_visible = [convert_to_markdown_wrapped(entry, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
+        converted_visible = [convert_to_markdown_wrapped(entry, message_id=i, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
 
         if converted_visible[0]:  # Don't display empty user messages
             output += (
diff --git a/modules/shared.py b/modules/shared.py
index b4f8708273..5177ac671b 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -59,7 +59,6 @@
     'seed': -1,
     'custom_stopping_strings': '',
     'custom_token_bans': '',
-    'show_after': '',
     'negative_prompt': '',
     'autoload_model': False,
     'dark_theme': True,
diff --git a/modules/ui.py b/modules/ui.py
index 19b76ceed2..3e1bf6d8c0 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -207,7 +207,6 @@ def list_interface_input_elements():
         'sampler_priority',
         'custom_stopping_strings',
         'custom_token_bans',
-        'show_after',
         'negative_prompt',
         'dry_sequence_breakers',
         'grammar_string',
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index c3245a9dc2..b494a758ff 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -93,7 +93,6 @@ def create_ui(default_preset):
                             shared.gradio['sampler_priority'] = gr.Textbox(value=generate_params['sampler_priority'], lines=12, label='Sampler priority', info='Parameter names separated by new lines or commas.', elem_classes=['add_scrollbar'])
                             shared.gradio['custom_stopping_strings'] = gr.Textbox(lines=2, value=shared.settings["custom_stopping_strings"] or None, label='Custom stopping strings', info='Written between "" and separated by commas.', placeholder='"\\n", "\\nYou:"')
                             shared.gradio['custom_token_bans'] = gr.Textbox(value=shared.settings['custom_token_bans'] or None, label='Token bans', info='Token IDs to ban, separated by commas. The IDs can be found in the Default or Notebook tab.')
-                            shared.gradio['show_after'] = gr.Textbox(value=shared.settings['show_after'] or None, label='Show after', info='Hide the reply before this text.', placeholder="")
                             shared.gradio['negative_prompt'] = gr.Textbox(value=shared.settings['negative_prompt'], label='Negative prompt', info='For CFG. Only used when guidance_scale is different than 1.', lines=3, elem_classes=['add_scrollbar'])
                             shared.gradio['dry_sequence_breakers'] = gr.Textbox(value=generate_params['dry_sequence_breakers'], label='dry_sequence_breakers', info='Tokens across which sequence matching is not continued. Specified as a comma-separated list of quoted strings.')
                             with gr.Row() as shared.gradio['grammar_file_row']:
diff --git a/settings-template.yaml b/settings-template.yaml
index 94a5c034c3..83764f97ac 100644
--- a/settings-template.yaml
+++ b/settings-template.yaml
@@ -29,7 +29,6 @@ truncation_length: 8192
 seed: -1
 custom_stopping_strings: ''
 custom_token_bans: ''
-show_after: ''
 negative_prompt: ''
 autoload_model: false
 dark_theme: true

From 877cf44c08dc98066118d1472844808b528fb778 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 25 Apr 2025 16:21:35 -0700
Subject: [PATCH 0372/1701] llama.cpp: Add StreamingLLM (`--streaming-llm`)

---
 modules/llama_cpp_server.py | 2 ++
 modules/shared.py           | 1 +
 modules/ui_model_menu.py    | 1 +
 3 files changed, 4 insertions(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 7199470dfe..7e5e3a4b7d 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -301,6 +301,8 @@ def _start_server(self):
                 cmd += ["--device-draft", shared.args.device_draft]
             if shared.args.ctx_size_draft > 0:
                 cmd += ["--ctx-size-draft", str(shared.args.ctx_size_draft)]
+        if shared.args.streaming_llm:
+            cmd += ["--cache-reuse", "1"]
         if shared.args.extra_flags:
             # Clean up the input
             extra_flags = shared.args.extra_flags.strip()
diff --git a/modules/shared.py b/modules/shared.py
index 5177ac671b..c40f8e2104 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -128,6 +128,7 @@
 group.add_argument('--no-kv-offload', action='store_true', help='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
 group.add_argument('--row-split', action='store_true', help='Split the model by rows across GPUs. This may improve multi-gpu performance.')
 group.add_argument('--extra-flags', type=str, default=None, help='Extra flags to pass to llama-server. Format: "flag1=value1;flag2;flag3=value3". Example: "override-tensor=exps=CPU"')
+group.add_argument('--streaming-llm', action='store_true', help='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
 
 # Speculative decoding
 group = parser.add_argument_group('Speculative decoding')
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index f3319cfbb7..1460dfec2f 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -71,6 +71,7 @@ def create_ui():
                             shared.gradio['torch_compile'] = gr.Checkbox(label="torch-compile", value=shared.args.torch_compile, info='Compile the model with torch.compile for improved performance.')
                             shared.gradio['flash_attn'] = gr.Checkbox(label="flash_attn", value=shared.args.flash_attn, info='Use flash-attention.')
                             shared.gradio['use_flash_attention_2'] = gr.Checkbox(label="use_flash_attention_2", value=shared.args.use_flash_attention_2, info='Set use_flash_attention_2=True while loading the model.')
+                            shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming_llm", value=shared.args.streaming_llm, info='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
                             shared.gradio['cpu'] = gr.Checkbox(label="cpu", value=shared.args.cpu, info='llama.cpp: Use llama-cpp-python compiled without GPU acceleration. Transformers: use PyTorch in CPU mode.')
                             shared.gradio['disk'] = gr.Checkbox(label="disk", value=shared.args.disk)
                             shared.gradio['row_split'] = gr.Checkbox(label="row_split", value=shared.args.row_split, info='Split the model by rows across GPUs. This may improve multi-gpu performance.')

From faababc4ea5e4548bebc13b50509587343b4c2db Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 25 Apr 2025 16:42:30 -0700
Subject: [PATCH 0373/1701] llama.cpp: Add a prompt processing progress bar

---
 modules/llama_cpp_server.py | 45 ++++++++++++++++++++++++++++---------
 1 file changed, 34 insertions(+), 11 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 7e5e3a4b7d..85743705f6 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -1,6 +1,7 @@
 import json
 import os
 import pprint
+import re
 import socket
 import subprocess
 import sys
@@ -10,6 +11,7 @@
 
 import llama_cpp_binaries
 import requests
+from tqdm import tqdm
 
 from modules import shared
 from modules.logging_colors import logger
@@ -335,17 +337,7 @@ def _start_server(self):
             env=env
         )
 
-        def filter_stderr(process_stderr):
-            try:
-                for line in iter(process_stderr.readline, ''):
-                    if not line.startswith(('srv ', 'slot ')) and 'log_server_r: request: GET /health' not in line:
-                        sys.stderr.write(line)
-                        sys.stderr.flush()
-            except (ValueError, IOError):
-                # Handle pipe closed exceptions
-                pass
-
-        threading.Thread(target=filter_stderr, args=(self.process.stderr,), daemon=True).start()
+        threading.Thread(target=filter_stderr_with_progress, args=(self.process.stderr,), daemon=True).start()
 
         # Wait for server to be healthy
         health_url = f"http://127.0.0.1:{self.port}/health"
@@ -396,3 +388,34 @@ def stop(self):
                 self.process.kill()
 
             self.process = None
+
+
+def filter_stderr_with_progress(process_stderr):
+    progress_bar = None
+    progress_pattern = re.compile(r'slot update_slots: id.*progress = (\d+\.\d+)')
+
+    try:
+        for line in iter(process_stderr.readline, ''):
+            progress_match = progress_pattern.search(line)
+
+            if progress_match:
+                progress = float(progress_match.group(1))
+
+                # Create progress bar on first progress message
+                if progress_bar is None:
+                    progress_bar = tqdm(total=1.0, desc="Prompt Processing", leave=False)
+
+                progress_bar.update(progress - progress_bar.n)
+
+                # Clean up when complete
+                if progress >= 1.0:
+                    progress_bar.close()
+                    progress_bar = None
+
+            if not line.startswith(('srv ', 'slot ')) and 'log_server_r: request: GET /health' not in line:
+                sys.stderr.write(line)
+                sys.stderr.flush()
+    except (ValueError, IOError):
+        if progress_bar:
+            progress_bar.close()
+        pass

From d4b1e31c493090424e3a38a7b115777dac943eac Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 25 Apr 2025 16:59:03 -0700
Subject: [PATCH 0374/1701] Use `--ctx-size` to specify the context size for
 all loaders

Old flags are still recognized as alternatives.
---
 modules/exllamav2.py        |  4 ++--
 modules/exllamav2_hf.py     | 20 ++++++++++----------
 modules/exllamav3_hf.py     |  2 +-
 modules/llama_cpp_server.py |  2 +-
 modules/loaders.py          | 10 +++++-----
 modules/models.py           |  6 ++----
 modules/models_settings.py  |  7 +++----
 modules/shared.py           |  7 +++++--
 modules/tensorrt_llm.py     |  6 +++---
 modules/ui.py               |  3 +--
 modules/ui_model_menu.py    | 11 ++++-------
 modules/ui_parameters.py    |  6 ++----
 12 files changed, 39 insertions(+), 45 deletions(-)

diff --git a/modules/exllamav2.py b/modules/exllamav2.py
index 7d79e516b2..952b73b876 100644
--- a/modules/exllamav2.py
+++ b/modules/exllamav2.py
@@ -3,6 +3,7 @@
 from pathlib import Path
 
 import torch
+
 from exllamav2 import (
     ExLlamaV2,
     ExLlamaV2Cache,
@@ -15,7 +16,6 @@
     ExLlamaV2Tokenizer
 )
 from exllamav2.generator import ExLlamaV2Sampler, ExLlamaV2StreamingGenerator
-
 from modules import shared
 from modules.logging_colors import logger
 from modules.text_generation import get_max_prompt_length
@@ -40,7 +40,7 @@ def from_pretrained(self, path_to_model):
         config.model_dir = str(path_to_model)
         config.prepare()
 
-        config.max_seq_len = shared.args.max_seq_len
+        config.max_seq_len = shared.args.ctx_size
         config.scale_pos_emb = shared.args.compress_pos_emb
         config.scale_alpha_value = shared.args.alpha_value
         config.no_flash_attn = shared.args.no_flash_attn
diff --git a/modules/exllamav2_hf.py b/modules/exllamav2_hf.py
index b159d9ce44..d6c3bf6ec3 100644
--- a/modules/exllamav2_hf.py
+++ b/modules/exllamav2_hf.py
@@ -4,6 +4,15 @@
 from typing import Any, Dict, Optional, Union
 
 import torch
+from torch.nn import CrossEntropyLoss
+from transformers import (
+    GenerationConfig,
+    GenerationMixin,
+    PretrainedConfig,
+    PreTrainedModel
+)
+from transformers.modeling_outputs import CausalLMOutputWithPast
+
 from exllamav2 import (
     ExLlamaV2,
     ExLlamaV2Cache,
@@ -14,15 +23,6 @@
     ExLlamaV2Cache_TP,
     ExLlamaV2Config
 )
-from torch.nn import CrossEntropyLoss
-from transformers import (
-    GenerationConfig,
-    GenerationMixin,
-    PretrainedConfig,
-    PreTrainedModel
-)
-from transformers.modeling_outputs import CausalLMOutputWithPast
-
 from modules import shared
 from modules.logging_colors import logger
 
@@ -192,7 +192,7 @@ def from_pretrained(cls, pretrained_model_name_or_path: Optional[Union[str, os.P
         config.model_dir = str(pretrained_model_name_or_path)
         config.prepare()
 
-        config.max_seq_len = shared.args.max_seq_len
+        config.max_seq_len = shared.args.ctx_size
         config.scale_pos_emb = shared.args.compress_pos_emb
         config.scale_alpha_value = shared.args.alpha_value
         config.no_flash_attn = shared.args.no_flash_attn
diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index 2d9c493a63..24ba9e136e 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -33,7 +33,7 @@ def __init__(self, model_dir):
         self.ex_model = Model.from_config(config)
 
         # Calculate the closest multiple of 256 at or above the chosen value
-        max_tokens = shared.args.max_seq_len
+        max_tokens = shared.args.ctx_size
         if max_tokens % 256 != 0:
             adjusted_tokens = ((max_tokens // 256) + 1) * 256
             logger.warning(f"max_num_tokens must be a multiple of 256. Adjusting from {max_tokens} to {adjusted_tokens}")
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 85743705f6..fb972a32b6 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -254,7 +254,7 @@ def _start_server(self):
         cmd = [
             self.server_path,
             "--model", self.model_path,
-            "--ctx-size", str(shared.args.n_ctx),
+            "--ctx-size", str(shared.args.ctx_size),
             "--n-gpu-layers", str(shared.args.n_gpu_layers),
             "--batch-size", str(shared.args.batch_size),
             "--port", str(self.port),
diff --git a/modules/loaders.py b/modules/loaders.py
index 9442a147e7..d8d62bf93f 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -9,7 +9,7 @@
         'threads',
         'threads_batch',
         'batch_size',
-        'n_ctx',
+        'ctx_size',
         'cache_type',
         'tensor_split',
         'extra_flags',
@@ -48,14 +48,14 @@
         'no_use_fast',
     ],
     'ExLlamav3_HF': [
-        'max_seq_len',
+        'ctx_size',
         'gpu_split',
         'cfg_cache',
         'trust_remote_code',
         'no_use_fast',
     ],
     'ExLlamav2_HF': [
-        'max_seq_len',
+        'ctx_size',
         'cache_type',
         'gpu_split',
         'alpha_value',
@@ -71,7 +71,7 @@
         'no_use_fast',
     ],
     'ExLlamav2': [
-        'max_seq_len',
+        'ctx_size',
         'cache_type',
         'gpu_split',
         'alpha_value',
@@ -93,7 +93,7 @@
         'no_use_fast',
     ],
     'TensorRT-LLM': [
-        'max_seq_len',
+        'ctx_size',
         'cpp_runner',
         'tensorrt_llm_info',
     ]
diff --git a/modules/models.py b/modules/models.py
index 99b068aa1d..d0b0402a82 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -52,10 +52,8 @@ def load_model(model_name, loader=None):
             tokenizer = load_tokenizer(model_name)
 
     shared.settings.update({k: v for k, v in metadata.items() if k in shared.settings})
-    if loader.lower().startswith('exllama') or loader.lower().startswith('tensorrt'):
-        shared.settings['truncation_length'] = shared.args.max_seq_len
-    elif loader == 'llama.cpp':
-        shared.settings['truncation_length'] = shared.args.n_ctx
+    if loader.lower().startswith('exllama') or loader.lower().startswith('tensorrt') or loader == 'llama.cpp':
+        shared.settings['truncation_length'] = shared.args.ctx_size
 
     logger.info(f"Loaded \"{model_name}\" in {(time.time()-t0):.2f} seconds.")
     logger.info(f"LOADER: \"{loader}\"")
diff --git a/modules/models_settings.py b/modules/models_settings.py
index ee2ed71bde..d3ecd51feb 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -11,8 +11,7 @@ def get_fallback_settings():
     return {
         'bf16': False,
         'use_eager_attention': False,
-        'max_seq_len': 2048,
-        'n_ctx': 2048,
+        'ctx_size': 2048,
         'rope_freq_base': 0,
         'compress_pos_emb': 1,
         'alpha_value': 1,
@@ -59,7 +58,7 @@ def get_model_metadata(model):
 
         for k in metadata:
             if k.endswith('context_length'):
-                model_settings['n_ctx'] = min(metadata[k], 8192)
+                model_settings['ctx_size'] = min(metadata[k], 8192)
                 model_settings['truncation_length_info'] = metadata[k]
             elif k.endswith('rope.freq_base'):
                 model_settings['rope_freq_base'] = metadata[k]
@@ -97,7 +96,7 @@ def get_model_metadata(model):
                 if k in metadata:
                     model_settings['truncation_length'] = metadata[k]
                     model_settings['truncation_length_info'] = metadata[k]
-                    model_settings['max_seq_len'] = min(metadata[k], 8192)
+                    model_settings['ctx_size'] = min(metadata[k], 8192)
 
             if 'rope_theta' in metadata:
                 model_settings['rope_freq_base'] = metadata['rope_theta']
diff --git a/modules/shared.py b/modules/shared.py
index c40f8e2104..572bfc09bc 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -116,7 +116,6 @@
 # llama.cpp
 group = parser.add_argument_group('llama.cpp')
 group.add_argument('--flash-attn', action='store_true', help='Use flash-attention.')
-group.add_argument('--n_ctx', type=int, default=8192, help='Size of the prompt context.')
 group.add_argument('--threads', type=int, default=0, help='Number of threads to use.')
 group.add_argument('--threads-batch', type=int, default=0, help='Number of threads to use for batches/prompt processing.')
 group.add_argument('--batch-size', type=int, default=256, help='Maximum number of prompt tokens to batch together when calling llama_eval.')
@@ -130,6 +129,11 @@
 group.add_argument('--extra-flags', type=str, default=None, help='Extra flags to pass to llama-server. Format: "flag1=value1;flag2;flag3=value3". Example: "override-tensor=exps=CPU"')
 group.add_argument('--streaming-llm', action='store_true', help='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
 
+# Cache
+group = parser.add_argument_group('Context and cache management')
+group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=8192, help='Context size in tokens.')
+group.add_argument('--cache_type', type=str, default='fp16', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.')
+
 # Speculative decoding
 group = parser.add_argument_group('Speculative decoding')
 group.add_argument('--model-draft', type=str, default=None, help='Path to the draft model for speculative decoding.')
@@ -142,7 +146,6 @@
 group = parser.add_argument_group('ExLlamaV2')
 group.add_argument('--gpu-split', type=str, help='Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.')
 group.add_argument('--autosplit', action='store_true', help='Autosplit the model tensors across the available GPUs. This causes --gpu-split to be ignored.')
-group.add_argument('--max_seq_len', type=int, default=8192, help='Maximum sequence length.')
 group.add_argument('--cfg-cache', action='store_true', help='ExLlamav2_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.')
 group.add_argument('--no_flash_attn', action='store_true', help='Force flash-attention to not be used.')
 group.add_argument('--no_xformers', action='store_true', help='Force xformers to not be used.')
diff --git a/modules/tensorrt_llm.py b/modules/tensorrt_llm.py
index c2685b7598..73178c395e 100644
--- a/modules/tensorrt_llm.py
+++ b/modules/tensorrt_llm.py
@@ -1,15 +1,15 @@
 from pathlib import Path
 
-import tensorrt_llm
 import torch
-from tensorrt_llm.runtime import ModelRunner, ModelRunnerCpp
 
+import tensorrt_llm
 from modules import shared
 from modules.logging_colors import logger
 from modules.text_generation import (
     get_max_prompt_length,
     get_reply_from_output_ids
 )
+from tensorrt_llm.runtime import ModelRunner, ModelRunnerCpp
 
 
 class TensorRTLLMModel:
@@ -35,7 +35,7 @@ def from_pretrained(self, path_to_model):
             logger.info("TensorRT-LLM: Using \"ModelRunnerCpp\"")
             runner_kwargs.update(
                 max_batch_size=1,
-                max_input_len=shared.args.max_seq_len - 512,
+                max_input_len=shared.args.ctx_size - 512,
                 max_output_len=512,
                 max_beam_width=1,
                 max_attention_window_size=None,
diff --git a/modules/ui.py b/modules/ui.py
index 3e1bf6d8c0..68cb76a615 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -110,8 +110,7 @@ def list_model_elements():
         'threads_batch',
         'batch_size',
         'hqq_backend',
-        'n_ctx',
-        'max_seq_len',
+        'ctx_size',
         'cache_type',
         'tensor_split',
         'extra_flags',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 1460dfec2f..9aeb02d167 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -51,8 +51,7 @@ def create_ui():
                             shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
                             shared.gradio['batch_size'] = gr.Slider(label="batch_size", minimum=1, maximum=4096, step=1, value=shared.args.batch_size)
                             shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
-                            shared.gradio['n_ctx'] = gr.Number(label="n_ctx", precision=0, step=256, value=shared.args.n_ctx, info='Context length. ⚠️ Lower this value if you can\'t load the model. Common values: 2048, 4096, 8192, 16384, 32768.')
-                            shared.gradio['max_seq_len'] = gr.Number(label='max_seq_len', precision=0, step=256, value=shared.args.max_seq_len, info='Context length. ⚠️ Lower this value if you can\'t load the model. Common values: 2048, 4096, 8192, 16384, 32768.')
+                            shared.gradio['ctx_size'] = gr.Number(label='ctx_size', precision=0, step=256, value=shared.args.ctx_size, info='Context length. ⚠️ Lower this value if you can\'t load the model. Common values: 2048, 4096, 8192, 16384, 32768, 65536.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q6', 'q4'], value=shared.args.cache_type, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.')
                             shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
@@ -92,7 +91,7 @@ def create_ui():
                             shared.gradio['trust_remote_code'] = gr.Checkbox(label="trust-remote-code", value=shared.args.trust_remote_code, info='Set trust_remote_code=True while loading the tokenizer/model. To enable this option, start the web UI with the --trust-remote-code flag.', interactive=shared.args.trust_remote_code)
                             shared.gradio['no_use_fast'] = gr.Checkbox(label="no_use_fast", value=shared.args.no_use_fast, info='Set use_fast=False while loading the tokenizer.')
                             shared.gradio['exllamav2_info'] = gr.Markdown("ExLlamav2_HF is recommended over ExLlamav2 for better integration with extensions and more consistent sampling behavior across loaders.")
-                            shared.gradio['tensorrt_llm_info'] = gr.Markdown('* TensorRT-LLM has to be installed manually in a separate Python 3.10 environment at the moment. For a guide, consult the description of [this PR](https://github.com/oobabooga/text-generation-webui/pull/5715). \n\n* `max_seq_len` is only used when `cpp-runner` is checked.\n\n* `cpp_runner` does not support streaming at the moment.')
+                            shared.gradio['tensorrt_llm_info'] = gr.Markdown('* TensorRT-LLM has to be installed manually in a separate Python 3.10 environment at the moment. For a guide, consult the description of [this PR](https://github.com/oobabooga/text-generation-webui/pull/5715). \n\n* `ctx_size` is only used when `cpp-runner` is checked.\n\n* `cpp_runner` does not support streaming at the moment.')
 
                             # Speculative decoding
                             with gr.Accordion("Speculative decoding", open=False, elem_classes='tgw-accordion') as shared.gradio['speculative_decoding_accordion']:
@@ -247,10 +246,8 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
 
 def update_truncation_length(current_length, state):
     if 'loader' in state:
-        if state['loader'].lower().startswith('exllama'):
-            return state['max_seq_len']
-        elif state['loader'] == 'llama.cpp':
-            return state['n_ctx']
+        if state['loader'].lower().startswith('exllama') or state['loader'] == 'llama.cpp':
+            return state['ctx_size']
 
     return current_length
 
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index b494a758ff..156e4128ed 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -121,10 +121,8 @@ def create_event_handlers():
 
 
 def get_truncation_length():
-    if 'max_seq_len' in shared.provided_arguments or shared.args.max_seq_len != shared.args_defaults.max_seq_len:
-        return shared.args.max_seq_len
-    elif 'n_ctx' in shared.provided_arguments or shared.args.n_ctx != shared.args_defaults.n_ctx:
-        return shared.args.n_ctx
+    if 'ctx_size' in shared.provided_arguments or shared.args.ctx_size != shared.args_defaults.ctx_size:
+        return shared.args.ctx_size
     else:
         return shared.settings['truncation_length']
 

From d4017fbb6d0b9be7a8964ad3fa03db0b373e453d Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Fri, 25 Apr 2025 21:32:00 -0300
Subject: [PATCH 0375/1701] ExLlamaV3: Add kv cache quantization (#6903)

---
 modules/exllamav3_hf.py  | 29 ++++++++++++++++++++++++++++-
 modules/loaders.py       |  2 ++
 modules/shared.py        |  2 +-
 modules/ui_model_menu.py |  2 +-
 4 files changed, 32 insertions(+), 3 deletions(-)

diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index 24ba9e136e..f15fc0b228 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -5,6 +5,7 @@
 
 import torch
 from exllamav3 import Cache, Config, Model
+from exllamav3.cache import CacheLayer_fp16, CacheLayer_quant
 from torch.nn import CrossEntropyLoss
 from transformers import (
     GenerationConfig,
@@ -39,7 +40,33 @@ def __init__(self, model_dir):
             logger.warning(f"max_num_tokens must be a multiple of 256. Adjusting from {max_tokens} to {adjusted_tokens}")
             max_tokens = adjusted_tokens
 
-        self.ex_cache = Cache(self.ex_model, max_num_tokens=max_tokens)
+        # Parse cache type
+        cache_type = shared.args.cache_type.lower()
+        cache_kwargs = {}
+        if cache_type == 'fp16':
+            layer_type = CacheLayer_fp16
+        elif cache_type.startswith('q'):
+            layer_type = CacheLayer_quant
+            if '_' in cache_type:
+                # Different bits for k and v (e.g., q4_q8)
+                k_part, v_part = cache_type.split('_')
+                k_bits = int(k_part[1:])
+                v_bits = int(v_part[1:])
+            else:
+                # Same bits for k and v (e.g., q4)
+                k_bits = v_bits = int(cache_type[1:])
+
+            # Validate bit ranges
+            if not (2 <= k_bits <= 8 and 2 <= v_bits <= 8):
+                logger.warning(f"Invalid quantization bits: k_bits={k_bits}, v_bits={v_bits}. Must be between 2 and 8. Falling back to fp16.")
+                layer_type = CacheLayer_fp16
+            else:
+                cache_kwargs = {'k_bits': k_bits, 'v_bits': v_bits}
+        else:
+            logger.warning(f"Unrecognized cache type: {cache_type}. Falling back to fp16.")
+            layer_type = CacheLayer_fp16
+
+        self.ex_cache = Cache(self.ex_model, max_num_tokens=max_tokens, layer_type=layer_type, **cache_kwargs)
 
         # Create load parameters dictionary
         load_params = {'progressbar': True}
diff --git a/modules/loaders.py b/modules/loaders.py
index d8d62bf93f..062e48377c 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -13,6 +13,7 @@
         'cache_type',
         'tensor_split',
         'extra_flags',
+        'streaming_llm',
         'rope_freq_base',
         'compress_pos_emb',
         'flash_attn',
@@ -49,6 +50,7 @@
     ],
     'ExLlamav3_HF': [
         'ctx_size',
+        'cache_type',
         'gpu_split',
         'cfg_cache',
         'trust_remote_code',
diff --git a/modules/shared.py b/modules/shared.py
index 572bfc09bc..96f659297f 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -132,7 +132,7 @@
 # Cache
 group = parser.add_argument_group('Context and cache management')
 group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=8192, help='Context size in tokens.')
-group.add_argument('--cache_type', type=str, default='fp16', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.')
+group.add_argument('--cache_type', type=str, default='fp16', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).')
 
 # Speculative decoding
 group = parser.add_argument_group('Speculative decoding')
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 9aeb02d167..6bd647c6f7 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -52,7 +52,7 @@ def create_ui():
                             shared.gradio['batch_size'] = gr.Slider(label="batch_size", minimum=1, maximum=4096, step=1, value=shared.args.batch_size)
                             shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
                             shared.gradio['ctx_size'] = gr.Number(label='ctx_size', precision=0, step=256, value=shared.args.ctx_size, info='Context length. ⚠️ Lower this value if you can\'t load the model. Common values: 2048, 4096, 8192, 16384, 32768, 65536.')
-                            shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q6', 'q4'], value=shared.args.cache_type, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.')
+                            shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
                             shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['extra_flags'] = gr.Textbox(label='extra-flags', info='Additional flags to pass to llama-server. Format: "flag1=value1;flag2;flag3=value3". Example: "override-tensor=exps=CPU"')

From d9de14d1f7e81d6e18711431acb9579dbb8a5ddf Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Sat, 26 Apr 2025 08:56:54 -0300
Subject: [PATCH 0376/1701] Restructure the repository (#6904)

---
 .gitignore                                    |  20 +-
 README.md                                     | 239 +++++++++---------
 download-model.py                             |   6 +-
 extensions/Training_PRO/matplotgraph.py       |   2 +-
 extensions/Training_PRO/script.py             |  48 ++--
 extensions/Training_PRO/train_utils.py        |  12 +-
 extensions/gallery/script.py                  |   4 +-
 extensions/openai/typing.py                   |   6 +-
 js/main.js                                    |   2 +-
 js/update_big_picture.js                      |   2 +-
 modules/chat.py                               |  40 +--
 modules/evaluate.py                           |   8 +-
 modules/html_generator.py                     |   8 +-
 modules/models_settings.py                    |   6 +-
 modules/presets.py                            |   2 +-
 modules/prompts.py                            |   2 +-
 modules/shared.py                             |   8 +-
 modules/training.py                           |  34 +--
 modules/transformers_loader.py                |   2 +-
 modules/ui.py                                 |   2 +-
 modules/ui_chat.py                            |   2 +-
 modules/ui_default.py                         |   4 +-
 modules/ui_file_saving.py                     |  12 +-
 modules/ui_model_menu.py                      |   4 +-
 modules/ui_parameters.py                      |   2 +-
 modules/ui_session.py                         |   2 +-
 modules/utils.py                              |  12 +-
 one_click.py                                  |   6 +-
 server.py                                     |  18 +-
 CMD_FLAGS.txt => user_data/CMD_FLAGS.txt      |   0
 .../characters}/Assistant.yaml                |   0
 .../characters}/Example.png                   | Bin
 .../characters}/Example.yaml                  |   0
 .../grammars}/arithmetic.gbnf                 |   0
 {grammars => user_data/grammars}/c.gbnf       |   0
 {grammars => user_data/grammars}/chess.gbnf   |   0
 {grammars => user_data/grammars}/json.gbnf    |   0
 .../grammars}/json_w_trailing_space.gbnf      |   0
 {grammars => user_data/grammars}/list.gbnf    |   0
 .../grammars}/roleplay.gbnf                   |   0
 .../grammars}/simple_arithmetic.gbnf          |   0
 .../Airoboros-v1.2.yaml                       |   0
 .../instruction-templates}/Alpaca.yaml        |   0
 .../instruction-templates}/Bactrian.yaml      |   0
 .../instruction-templates}/Baichuan Chat.yaml |   0
 .../instruction-templates}/Baize.yaml         |   0
 .../instruction-templates}/Bluemoon.yaml      |   0
 .../instruction-templates}/ChatGLM.yaml       |   0
 .../instruction-templates}/ChatML.yaml        |   0
 .../Chinese-Vicuna-Chat.yaml                  |   0
 .../instruction-templates}/Command-R.yaml     |   0
 .../Galactica Cite.yaml                       |   0
 .../Galactica Finetuned.yaml                  |   0
 .../instruction-templates}/Galactica Q.yaml   |   0
 .../Galactica Summary.yaml                    |   0
 .../Galactica Work.yaml                       |   0
 .../instruction-templates}/Galactica v2.yaml  |   0
 .../instruction-templates}/Galactica.yaml     |   0
 .../instruction-templates}/Gorilla.yaml       |   0
 .../Guanaco non-chat.yaml                     |   0
 .../instruction-templates}/Guanaco-QLoRA.yaml |   0
 .../H2O-prompt_answer.yaml                    |   0
 .../instruction-templates}/Hippogriff.yaml    |   0
 .../instruction-templates}/INCITE-Chat.yaml   |   0
 .../INCITE-Instruct.yaml                      |   0
 .../instruction-templates}/KoAlpaca.yaml      |   0
 .../instruction-templates}/Koala.yaml         |   0
 .../instruction-templates}/LLaVA.yaml         |   0
 .../instruction-templates}/Llama-v2.yaml      |   0
 .../instruction-templates}/Llama-v3.yaml      |   0
 .../instruction-templates}/MOSS.yaml          |   0
 .../Manticore Chat.yaml                       |   0
 .../instruction-templates}/Metharme.yaml      |   0
 .../instruction-templates}/Mistral.yaml       |   0
 .../instruction-templates}/NVIDIA-ChatQA.yaml |   0
 .../instruction-templates}/NewHope.yaml       |   0
 .../Open Assistant.yaml                       |   0
 .../instruction-templates}/OpenBuddy.yaml     |   0
 .../instruction-templates}/OpenChat.yaml      |   0
 .../OpenOrca-Platypus2.yaml                   |   0
 .../instruction-templates}/Orca Mini.yaml     |   0
 .../instruction-templates}/Orca-Vicuna.yaml   |   0
 .../instruction-templates}/RWKV-Raven.yaml    |   0
 .../instruction-templates}/RWKV-World.yaml    |   0
 .../instruction-templates}/Samantha.yaml      |   0
 .../instruction-templates}/StableBeluga2.yaml |   0
 .../instruction-templates}/StableLM.yaml      |   0
 .../instruction-templates}/StableVicuna.yaml  |   0
 .../instruction-templates}/Starchat-Beta.yaml |   0
 .../instruction-templates}/Synthia-CoT.yaml   |   0
 .../instruction-templates}/Synthia.yaml       |   0
 .../instruction-templates}/Tulu.yaml          |   0
 .../instruction-templates}/Vicuna-v0.yaml     |   0
 .../instruction-templates}/Vicuna-v1.1.yaml   |   0
 .../instruction-templates}/Vigogne-Chat.yaml  |   0
 .../Vigogne-Instruct.yaml                     |   0
 .../Wizard-Mega ShareGPT.yaml                 |   0
 .../instruction-templates}/Wizard-Mega.yaml   |   0
 .../instruction-templates}/Ziya.yaml          |   0
 .../loras}/place-your-loras-here.txt          |   0
 {models => user_data/models}/config.yaml      |   0
 .../models}/place-your-models-here.txt        |   0
 .../presets}/Contrastive Search.yaml          |   0
 {presets => user_data/presets}/Creative.yaml  |   0
 .../presets}/Deterministic.yaml               |   0
 {presets => user_data/presets}/Instruct.yaml  |   0
 .../presets}/Null preset.yaml                 |   0
 {presets => user_data/presets}/min_p.yaml     |   0
 .../prompts}/Alpaca-with-Input.txt            |   0
 {prompts => user_data/prompts}/QA.txt         |   0
 .../datasets/put-trainer-datasets-here.txt    |   0
 .../training}/formats/ChatML-format.json      |   0
 .../formats/alpaca-chatbot-format.json        |   0
 .../training}/formats/alpaca-format.json      |   0
 .../training}/formats/llama2-chat-format.json |   0
 .../training}/formats/vicuna-format.json      |   0
 116 files changed, 254 insertions(+), 261 deletions(-)
 rename CMD_FLAGS.txt => user_data/CMD_FLAGS.txt (100%)
 rename {characters => user_data/characters}/Assistant.yaml (100%)
 rename {characters => user_data/characters}/Example.png (100%)
 rename {characters => user_data/characters}/Example.yaml (100%)
 rename {grammars => user_data/grammars}/arithmetic.gbnf (100%)
 rename {grammars => user_data/grammars}/c.gbnf (100%)
 rename {grammars => user_data/grammars}/chess.gbnf (100%)
 rename {grammars => user_data/grammars}/json.gbnf (100%)
 rename {grammars => user_data/grammars}/json_w_trailing_space.gbnf (100%)
 rename {grammars => user_data/grammars}/list.gbnf (100%)
 rename {grammars => user_data/grammars}/roleplay.gbnf (100%)
 rename {grammars => user_data/grammars}/simple_arithmetic.gbnf (100%)
 rename {instruction-templates => user_data/instruction-templates}/Airoboros-v1.2.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Alpaca.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Bactrian.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Baichuan Chat.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Baize.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Bluemoon.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/ChatGLM.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/ChatML.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Chinese-Vicuna-Chat.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Command-R.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Galactica Cite.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Galactica Finetuned.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Galactica Q.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Galactica Summary.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Galactica Work.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Galactica v2.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Galactica.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Gorilla.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Guanaco non-chat.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Guanaco-QLoRA.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/H2O-prompt_answer.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Hippogriff.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/INCITE-Chat.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/INCITE-Instruct.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/KoAlpaca.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Koala.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/LLaVA.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Llama-v2.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Llama-v3.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/MOSS.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Manticore Chat.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Metharme.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Mistral.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/NVIDIA-ChatQA.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/NewHope.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Open Assistant.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/OpenBuddy.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/OpenChat.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/OpenOrca-Platypus2.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Orca Mini.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Orca-Vicuna.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/RWKV-Raven.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/RWKV-World.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Samantha.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/StableBeluga2.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/StableLM.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/StableVicuna.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Starchat-Beta.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Synthia-CoT.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Synthia.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Tulu.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Vicuna-v0.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Vicuna-v1.1.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Vigogne-Chat.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Vigogne-Instruct.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Wizard-Mega ShareGPT.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Wizard-Mega.yaml (100%)
 rename {instruction-templates => user_data/instruction-templates}/Ziya.yaml (100%)
 rename {loras => user_data/loras}/place-your-loras-here.txt (100%)
 rename {models => user_data/models}/config.yaml (100%)
 rename {models => user_data/models}/place-your-models-here.txt (100%)
 rename {presets => user_data/presets}/Contrastive Search.yaml (100%)
 rename {presets => user_data/presets}/Creative.yaml (100%)
 rename {presets => user_data/presets}/Deterministic.yaml (100%)
 rename {presets => user_data/presets}/Instruct.yaml (100%)
 rename {presets => user_data/presets}/Null preset.yaml (100%)
 rename {presets => user_data/presets}/min_p.yaml (100%)
 rename {prompts => user_data/prompts}/Alpaca-with-Input.txt (100%)
 rename {prompts => user_data/prompts}/QA.txt (100%)
 rename {training => user_data/training}/datasets/put-trainer-datasets-here.txt (100%)
 rename {training => user_data/training}/formats/ChatML-format.json (100%)
 rename {training => user_data/training}/formats/alpaca-chatbot-format.json (100%)
 rename {training => user_data/training}/formats/alpaca-format.json (100%)
 rename {training => user_data/training}/formats/llama2-chat-format.json (100%)
 rename {training => user_data/training}/formats/vicuna-format.json (100%)

diff --git a/.gitignore b/.gitignore
index 318e147d1b..bd69c9416e 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,26 +1,8 @@
-/cache
-/characters
 /css
 /extensions
-/grammars
 /installer_files
-/logs
-/loras
-/models
-/presets
-/prompts
 /repositories
-/softprompts
-/torch-dumps
-/training/datasets
-
-/CMD_FLAGS.txt
-/img_bot*
-/img_me*
-/models/config-user.yaml
-/notification.mp3
-/settings*.json
-/settings*.yaml
+/user_data
 
 .chroma
 .DS_Store
diff --git a/README.md b/README.md
index f62e3508fc..58f77786b2 100644
--- a/README.md
+++ b/README.md
@@ -182,131 +182,140 @@ List of command-line flags
 
 
 ```txt
-usage: server.py [-h] [--multi-user] [--character CHARACTER] [--model MODEL] [--lora LORA [LORA ...]] [--model-dir MODEL_DIR] [--lora-dir LORA_DIR] [--settings SETTINGS]
-                 [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--cpu] [--auto-devices] [--gpu-memory GPU_MEMORY [GPU_MEMORY ...]]
-                 [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast]
-                 [--use_flash_attention_2] [--use_eager_attention] [--torch-compile] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn]
-                 [--n_ctx N_CTX] [--threads THREADS] [--threads-batch THREADS_BATCH] [--batch-size BATCH_SIZE] [--no-mmap] [--mlock] [--n-gpu-layers N_GPU_LAYERS] [--tensor-split TENSOR_SPLIT]
-                 [--numa] [--no-kv-offload] [--row-split] [--gpu-split GPU_SPLIT] [--autosplit] [--max_seq_len MAX_SEQ_LEN] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa]
-                 [--num_experts_per_token NUM_EXPERTS_PER_TOKEN] [--enable_tp] [--hqq-backend HQQ_BACKEND] [--cpp-runner] [--cache_type CACHE_TYPE] [--deepspeed] [--nvme-offload-dir NVME_OFFLOAD_DIR]
-                 [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT]
-                 [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE]
-                 [--subpath SUBPATH] [--old-colors] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6]
-                 [--api-disable-ipv4] [--nowebui]
+usage: server.py [-h] [--multi-user] [--character CHARACTER] [--model MODEL] [--lora LORA [LORA ...]] [--model-dir MODEL_DIR] [--lora-dir LORA_DIR] [--model-menu] [--settings SETTINGS]
+                 [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR]
+                 [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--use_flash_attention_2] [--use_eager_attention] [--torch-compile] [--load-in-4bit]
+                 [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn] [--threads THREADS] [--threads-batch THREADS_BATCH] [--batch-size BATCH_SIZE] [--no-mmap]
+                 [--mlock] [--n-gpu-layers N_GPU_LAYERS] [--tensor-split TENSOR_SPLIT] [--numa] [--no-kv-offload] [--row-split] [--extra-flags EXTRA_FLAGS] [--streaming-llm] [--ctx-size CTX_SIZE]
+                 [--model-draft MODEL_DRAFT] [--draft-max DRAFT_MAX] [--gpu-layers-draft GPU_LAYERS_DRAFT] [--device-draft DEVICE_DRAFT] [--ctx-size-draft CTX_SIZE_DRAFT] [--gpu-split GPU_SPLIT]
+                 [--autosplit] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa] [--num_experts_per_token NUM_EXPERTS_PER_TOKEN] [--enable_tp] [--hqq-backend HQQ_BACKEND] [--cpp-runner]
+                 [--cache_type CACHE_TYPE] [--deepspeed] [--nvme-offload-dir NVME_OFFLOAD_DIR] [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE]
+                 [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH]
+                 [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors] [--api] [--public-api]
+                 [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4] [--nowebui]
 
 Text generation web UI
 
 options:
-  -h, --help                                     show this help message and exit
+  -h, --help                                           show this help message and exit
 
 Basic settings:
-  --multi-user                                   Multi-user mode. Chat histories are not saved or automatically loaded. Warning: this is likely not safe for sharing publicly.
-  --character CHARACTER                          The name of the character to load in chat mode by default.
-  --model MODEL                                  Name of the model to load by default.
-  --lora LORA [LORA ...]                         The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.
-  --model-dir MODEL_DIR                          Path to directory with all the models.
-  --lora-dir LORA_DIR                            Path to directory with all the loras.
-  --settings SETTINGS                            Load the default interface settings from this yaml file. See settings-template.yaml for an example. If you create a file called settings.yaml, this
-                                                 file will be loaded by default without the need to use the --settings flag.
-  --extensions EXTENSIONS [EXTENSIONS ...]       The list of extensions to load. If you want to load more than one extension, write the names separated by spaces.
-  --verbose                                      Print the prompts to the terminal.
-  --idle-timeout IDLE_TIMEOUT                    Unload model after this many minutes of inactivity. It will be automatically reloaded when you try to use it again.
+  --multi-user                                         Multi-user mode. Chat histories are not saved or automatically loaded. Warning: this is likely not safe for sharing publicly.
+  --character CHARACTER                                The name of the character to load in chat mode by default.
+  --model MODEL                                        Name of the model to load by default.
+  --lora LORA [LORA ...]                               The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.
+  --model-dir MODEL_DIR                                Path to directory with all the models.
+  --lora-dir LORA_DIR                                  Path to directory with all the loras.
+  --model-menu                                         Show a model menu in the terminal when the web UI is first launched.
+  --settings SETTINGS                                  Load the default interface settings from this yaml file. See settings-template.yaml for an example. If you create a file called settings.yaml,
+                                                       this file will be loaded by default without the need to use the --settings flag.
+  --extensions EXTENSIONS [EXTENSIONS ...]             The list of extensions to load. If you want to load more than one extension, write the names separated by spaces.
+  --verbose                                            Print the prompts to the terminal.
+  --idle-timeout IDLE_TIMEOUT                          Unload model after this many minutes of inactivity. It will be automatically reloaded when you try to use it again.
 
 Model loader:
-  --loader LOADER                                Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF, ExLlamav2,
-                                                 HQQ, TensorRT-LLM.
+  --loader LOADER                                      Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF,
+                                                       ExLlamav2, HQQ, TensorRT-LLM.
 
 Transformers/Accelerate:
-  --cpu                                          Use the CPU to generate text. Warning: Training on CPU is extremely slow.
-  --auto-devices                                 Automatically split the model across the available GPU(s) and CPU.
-  --gpu-memory GPU_MEMORY [GPU_MEMORY ...]       Maximum GPU memory in GiB to be allocated per GPU. Example: --gpu-memory 10 for a single GPU, --gpu-memory 10 5 for two GPUs. You can also set values
-                                                 in MiB like --gpu-memory 3500MiB.
-  --cpu-memory CPU_MEMORY                        Maximum CPU memory in GiB to allocate for offloaded weights. Same as above.
-  --disk                                         If the model is too large for your GPU(s) and CPU combined, send the remaining layers to the disk.
-  --disk-cache-dir DISK_CACHE_DIR                Directory to save the disk cache to. Defaults to "cache".
-  --load-in-8bit                                 Load the model with 8-bit precision (using bitsandbytes).
-  --bf16                                         Load the model with bfloat16 precision. Requires NVIDIA Ampere GPU.
-  --no-cache                                     Set use_cache to False while generating text. This reduces VRAM usage slightly, but it comes at a performance cost.
-  --trust-remote-code                            Set trust_remote_code=True while loading the model. Necessary for some models.
-  --force-safetensors                            Set use_safetensors=True while loading the model. This prevents arbitrary code execution.
-  --no_use_fast                                  Set use_fast=False while loading the tokenizer (it's True by default). Use this if you have any problems related to use_fast.
-  --use_flash_attention_2                        Set use_flash_attention_2=True while loading the model.
-  --use_eager_attention                          Set attn_implementation= eager while loading the model.
-  --torch-compile                                Compile the model with torch.compile for improved performance.
+  --cpu                                                Use the CPU to generate text. Warning: Training on CPU is extremely slow.
+  --cpu-memory CPU_MEMORY                              Maximum CPU memory in GiB. Use this for CPU offloading.
+  --disk                                               If the model is too large for your GPU(s) and CPU combined, send the remaining layers to the disk.
+  --disk-cache-dir DISK_CACHE_DIR                      Directory to save the disk cache to. Defaults to "user_data/cache".
+  --load-in-8bit                                       Load the model with 8-bit precision (using bitsandbytes).
+  --bf16                                               Load the model with bfloat16 precision. Requires NVIDIA Ampere GPU.
+  --no-cache                                           Set use_cache to False while generating text. This reduces VRAM usage slightly, but it comes at a performance cost.
+  --trust-remote-code                                  Set trust_remote_code=True while loading the model. Necessary for some models.
+  --force-safetensors                                  Set use_safetensors=True while loading the model. This prevents arbitrary code execution.
+  --no_use_fast                                        Set use_fast=False while loading the tokenizer (it's True by default). Use this if you have any problems related to use_fast.
+  --use_flash_attention_2                              Set use_flash_attention_2=True while loading the model.
+  --use_eager_attention                                Set attn_implementation= eager while loading the model.
+  --torch-compile                                      Compile the model with torch.compile for improved performance.
 
 bitsandbytes 4-bit:
-  --load-in-4bit                                 Load the model with 4-bit precision (using bitsandbytes).
-  --use_double_quant                             use_double_quant for 4-bit.
-  --compute_dtype COMPUTE_DTYPE                  compute dtype for 4-bit. Valid options: bfloat16, float16, float32.
-  --quant_type QUANT_TYPE                        quant_type for 4-bit. Valid options: nf4, fp4.
+  --load-in-4bit                                       Load the model with 4-bit precision (using bitsandbytes).
+  --use_double_quant                                   use_double_quant for 4-bit.
+  --compute_dtype COMPUTE_DTYPE                        compute dtype for 4-bit. Valid options: bfloat16, float16, float32.
+  --quant_type QUANT_TYPE                              quant_type for 4-bit. Valid options: nf4, fp4.
 
 llama.cpp:
-  --flash-attn                                   Use flash-attention.
-  --n_ctx N_CTX                                  Size of the prompt context.
-  --threads THREADS                              Number of threads to use.
-  --threads-batch THREADS_BATCH                  Number of threads to use for batches/prompt processing.
-  --batch-size BATCH_SIZE                        Maximum number of prompt tokens to batch together when calling llama_eval.
-  --no-mmap                                      Prevent mmap from being used.
-  --mlock                                        Force the system to keep the model in RAM.
-  --n-gpu-layers N_GPU_LAYERS                    Number of layers to offload to the GPU.
-  --tensor-split TENSOR_SPLIT                    Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.
-  --numa                                         Activate NUMA task allocation for llama.cpp.
-  --no-kv-offload                                Do not offload the K, Q, V to the GPU. This saves VRAM but reduces the performance.
-  --row-split                                    Split the model by rows across GPUs. This may improve multi-gpu performance.
+  --flash-attn                                         Use flash-attention.
+  --threads THREADS                                    Number of threads to use.
+  --threads-batch THREADS_BATCH                        Number of threads to use for batches/prompt processing.
+  --batch-size BATCH_SIZE                              Maximum number of prompt tokens to batch together when calling llama_eval.
+  --no-mmap                                            Prevent mmap from being used.
+  --mlock                                              Force the system to keep the model in RAM.
+  --n-gpu-layers N_GPU_LAYERS                          Number of layers to offload to the GPU.
+  --tensor-split TENSOR_SPLIT                          Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.
+  --numa                                               Activate NUMA task allocation for llama.cpp.
+  --no-kv-offload                                      Do not offload the K, Q, V to the GPU. This saves VRAM but reduces the performance.
+  --row-split                                          Split the model by rows across GPUs. This may improve multi-gpu performance.
+  --extra-flags EXTRA_FLAGS                            Extra flags to pass to llama-server. Format: "flag1=value1;flag2;flag3=value3". Example: "override-tensor=exps=CPU"
+  --streaming-llm                                      Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.
+
+Context and cache management:
+  --ctx-size CTX_SIZE, --n_ctx CTX_SIZE, --max_seq_len CTX_SIZE
+                                                       Context size in tokens.
+
+Speculative decoding:
+  --model-draft MODEL_DRAFT                            Path to the draft model for speculative decoding.
+  --draft-max DRAFT_MAX                                Number of tokens to draft for speculative decoding.
+  --gpu-layers-draft GPU_LAYERS_DRAFT                  Number of layers to offload to the GPU for the draft model.
+  --device-draft DEVICE_DRAFT                          Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1
+  --ctx-size-draft CTX_SIZE_DRAFT                      Size of the prompt context for the draft model. If 0, uses the same as the main model.
 
 ExLlamaV2:
-  --gpu-split GPU_SPLIT                          Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.
-  --autosplit                                    Autosplit the model tensors across the available GPUs. This causes --gpu-split to be ignored.
-  --max_seq_len MAX_SEQ_LEN                      Maximum sequence length.
-  --cfg-cache                                    ExLlamav2_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.
-  --no_flash_attn                                Force flash-attention to not be used.
-  --no_xformers                                  Force xformers to not be used.
-  --no_sdpa                                      Force Torch SDPA to not be used.
-  --num_experts_per_token NUM_EXPERTS_PER_TOKEN  Number of experts to use for generation. Applies to MoE models like Mixtral.
-  --enable_tp                                    Enable Tensor Parallelism (TP) in ExLlamaV2.
+  --gpu-split GPU_SPLIT                                Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.
+  --autosplit                                          Autosplit the model tensors across the available GPUs. This causes --gpu-split to be ignored.
+  --cfg-cache                                          ExLlamav2_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.
+  --no_flash_attn                                      Force flash-attention to not be used.
+  --no_xformers                                        Force xformers to not be used.
+  --no_sdpa                                            Force Torch SDPA to not be used.
+  --num_experts_per_token NUM_EXPERTS_PER_TOKEN        Number of experts to use for generation. Applies to MoE models like Mixtral.
+  --enable_tp                                          Enable Tensor Parallelism (TP) in ExLlamaV2.
 
 HQQ:
-  --hqq-backend HQQ_BACKEND                      Backend for the HQQ loader. Valid options: PYTORCH, PYTORCH_COMPILE, ATEN.
+  --hqq-backend HQQ_BACKEND                            Backend for the HQQ loader. Valid options: PYTORCH, PYTORCH_COMPILE, ATEN.
 
 TensorRT-LLM:
-  --cpp-runner                                   Use the ModelRunnerCpp runner, which is faster than the default ModelRunner but doesn't support streaming yet.
+  --cpp-runner                                         Use the ModelRunnerCpp runner, which is faster than the default ModelRunner but doesn't support streaming yet.
 
 Cache:
-  --cache_type CACHE_TYPE                        KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.
+  --cache_type CACHE_TYPE                              KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.
 
 DeepSpeed:
-  --deepspeed                                    Enable the use of DeepSpeed ZeRO-3 for inference via the Transformers integration.
-  --nvme-offload-dir NVME_OFFLOAD_DIR            DeepSpeed: Directory to use for ZeRO-3 NVME offloading.
-  --local_rank LOCAL_RANK                        DeepSpeed: Optional argument for distributed setups.
+  --deepspeed                                          Enable the use of DeepSpeed ZeRO-3 for inference via the Transformers integration.
+  --nvme-offload-dir NVME_OFFLOAD_DIR                  DeepSpeed: Directory to use for ZeRO-3 NVME offloading.
+  --local_rank LOCAL_RANK                              DeepSpeed: Optional argument for distributed setups.
 
 RoPE:
-  --alpha_value ALPHA_VALUE                      Positional embeddings alpha factor for NTK RoPE scaling. Use either this or compress_pos_emb, not both.
-  --rope_freq_base ROPE_FREQ_BASE                If greater than 0, will be used instead of alpha_value. Those two are related by rope_freq_base = 10000 * alpha_value ^ (64 / 63).
-  --compress_pos_emb COMPRESS_POS_EMB            Positional embeddings compression factor. Should be set to (context length) / (model's original context length). Equal to 1/rope_freq_scale.
+  --alpha_value ALPHA_VALUE                            Positional embeddings alpha factor for NTK RoPE scaling. Use either this or compress_pos_emb, not both.
+  --rope_freq_base ROPE_FREQ_BASE                      If greater than 0, will be used instead of alpha_value. Those two are related by rope_freq_base = 10000 * alpha_value ^ (64 / 63).
+  --compress_pos_emb COMPRESS_POS_EMB                  Positional embeddings compression factor. Should be set to (context length) / (model's original context length). Equal to 1/rope_freq_scale.
 
 Gradio:
-  --listen                                       Make the web UI reachable from your local network.
-  --listen-port LISTEN_PORT                      The listening port that the server will use.
-  --listen-host LISTEN_HOST                      The hostname that the server will use.
-  --share                                        Create a public URL. This is useful for running the web UI on Google Colab or similar.
-  --auto-launch                                  Open the web UI in the default browser upon launch.
-  --gradio-auth GRADIO_AUTH                      Set Gradio authentication password in the format "username:password". Multiple credentials can also be supplied with "u1:p1,u2:p2,u3:p3".
-  --gradio-auth-path GRADIO_AUTH_PATH            Set the Gradio authentication file path. The file should contain one or more user:password pairs in the same format as above.
-  --ssl-keyfile SSL_KEYFILE                      The path to the SSL certificate key file.
-  --ssl-certfile SSL_CERTFILE                    The path to the SSL certificate cert file.
-  --subpath SUBPATH                              Customize the subpath for gradio, use with reverse proxy
-  --old-colors                                   Use the legacy Gradio colors, before the December/2024 update.
+  --listen                                             Make the web UI reachable from your local network.
+  --listen-port LISTEN_PORT                            The listening port that the server will use.
+  --listen-host LISTEN_HOST                            The hostname that the server will use.
+  --share                                              Create a public URL. This is useful for running the web UI on Google Colab or similar.
+  --auto-launch                                        Open the web UI in the default browser upon launch.
+  --gradio-auth GRADIO_AUTH                            Set Gradio authentication password in the format "username:password". Multiple credentials can also be supplied with "u1:p1,u2:p2,u3:p3".
+  --gradio-auth-path GRADIO_AUTH_PATH                  Set the Gradio authentication file path. The file should contain one or more user:password pairs in the same format as above.
+  --ssl-keyfile SSL_KEYFILE                            The path to the SSL certificate key file.
+  --ssl-certfile SSL_CERTFILE                          The path to the SSL certificate cert file.
+  --subpath SUBPATH                                    Customize the subpath for gradio, use with reverse proxy
+  --old-colors                                         Use the legacy Gradio colors, before the December/2024 update.
 
 API:
-  --api                                          Enable the API extension.
-  --public-api                                   Create a public URL for the API using Cloudfare.
-  --public-api-id PUBLIC_API_ID                  Tunnel ID for named Cloudflare Tunnel. Use together with public-api option.
-  --api-port API_PORT                            The listening port for the API.
-  --api-key API_KEY                              API authentication key.
-  --admin-key ADMIN_KEY                          API authentication key for admin tasks like loading and unloading models. If not set, will be the same as --api-key.
-  --api-enable-ipv6                              Enable IPv6 for the API
-  --api-disable-ipv4                             Disable IPv4 for the API
-  --nowebui                                      Do not launch the Gradio UI. Useful for launching the API in standalone mode.
+  --api                                                Enable the API extension.
+  --public-api                                         Create a public URL for the API using Cloudfare.
+  --public-api-id PUBLIC_API_ID                        Tunnel ID for named Cloudflare Tunnel. Use together with public-api option.
+  --api-port API_PORT                                  The listening port for the API.
+  --api-key API_KEY                                    API authentication key.
+  --admin-key ADMIN_KEY                                API authentication key for admin tasks like loading and unloading models. If not set, will be the same as --api-key.
+  --api-enable-ipv6                                    Enable IPv6 for the API
+  --api-disable-ipv4                                   Disable IPv4 for the API
+  --nowebui                                            Do not launch the Gradio UI. Useful for launching the API in standalone mode.
 ```
 
 
@@ -317,35 +326,37 @@ https://github.com/oobabooga/text-generation-webui/wiki
 
 ## Downloading models
 
-Models should be placed in the folder `text-generation-webui/models`. They are usually downloaded from [Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads).
+Models should be placed in the folder `text-generation-webui/user_data/models`. They are usually downloaded from [Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads).
 
-* GGUF models are a single file and should be placed directly into `models`. Example:
+* GGUF models are a single file and should be placed directly into `user_data/models`. Example:
 
 ```
 text-generation-webui
-└── models
-    └── llama-2-13b-chat.Q4_K_M.gguf
+└── user_data
+    └── models
+        └── llama-2-13b-chat.Q4_K_M.gguf
 ```
 
 * The remaining model types (like 16-bit Transformers models and EXL2 models) are made of several files and must be placed in a subfolder. Example:
 
 ```
 text-generation-webui
-├── models
-│   ├── lmsys_vicuna-33b-v1.3
-│   │   ├── config.json
-│   │   ├── generation_config.json
-│   │   ├── pytorch_model-00001-of-00007.bin
-│   │   ├── pytorch_model-00002-of-00007.bin
-│   │   ├── pytorch_model-00003-of-00007.bin
-│   │   ├── pytorch_model-00004-of-00007.bin
-│   │   ├── pytorch_model-00005-of-00007.bin
-│   │   ├── pytorch_model-00006-of-00007.bin
-│   │   ├── pytorch_model-00007-of-00007.bin
-│   │   ├── pytorch_model.bin.index.json
-│   │   ├── special_tokens_map.json
-│   │   ├── tokenizer_config.json
-│   │   └── tokenizer.model
+└── user_data
+    └── models
+        └── lmsys_vicuna-33b-v1.3
+            ├── config.json
+            ├── generation_config.json
+            ├── pytorch_model-00001-of-00007.bin
+            ├── pytorch_model-00002-of-00007.bin
+            ├── pytorch_model-00003-of-00007.bin
+            ├── pytorch_model-00004-of-00007.bin
+            ├── pytorch_model-00005-of-00007.bin
+            ├── pytorch_model-00006-of-00007.bin
+            ├── pytorch_model-00007-of-00007.bin
+            ├── pytorch_model.bin.index.json
+            ├── special_tokens_map.json
+            ├── tokenizer_config.json
+            └── tokenizer.model
 ```
 
 In both cases, you can use the "Model" tab of the UI to download the model from Hugging Face automatically. It is also possible to download it via the command-line with:
diff --git a/download-model.py b/download-model.py
index 8ff1d69c77..255174916c 100644
--- a/download-model.py
+++ b/download-model.py
@@ -1,5 +1,5 @@
 '''
-Downloads models from Hugging Face to models/username_modelname.
+Downloads models from Hugging Face to user_data/models/username_modelname.
 
 Example:
 python download-model.py facebook/opt-1.3b
@@ -175,7 +175,7 @@ def get_output_folder(self, model, branch, is_lora, is_llamacpp=False, model_dir
         if model_dir:
             base_folder = model_dir
         else:
-            base_folder = 'models' if not is_lora else 'loras'
+            base_folder = 'user_data/models' if not is_lora else 'user_data/loras'
 
         # If the model is of type GGUF, save directly in the base_folder
         if is_llamacpp:
@@ -356,7 +356,7 @@ def check_model_files(self, model, branch, links, sha256, output_folder):
     parser.add_argument('--specific-file', type=str, default=None, help='Name of the specific file to download (if not provided, downloads all).')
     parser.add_argument('--exclude-pattern', type=str, default=None, help='Regex pattern to exclude files from download.')
     parser.add_argument('--output', type=str, default=None, help='Save the model files to this folder.')
-    parser.add_argument('--model-dir', type=str, default=None, help='Save the model files to a subfolder of this folder instead of the default one (text-generation-webui/models).')
+    parser.add_argument('--model-dir', type=str, default=None, help='Save the model files to a subfolder of this folder instead of the default one (text-generation-webui/user_data/models).')
     parser.add_argument('--clean', action='store_true', help='Does not resume the previous download.')
     parser.add_argument('--check', action='store_true', help='Validates the checksums of model files.')
     parser.add_argument('--max-retries', type=int, default=7, help='Max retries count when get error in download time.')
diff --git a/extensions/Training_PRO/matplotgraph.py b/extensions/Training_PRO/matplotgraph.py
index 348fc01a4a..b30bee8365 100644
--- a/extensions/Training_PRO/matplotgraph.py
+++ b/extensions/Training_PRO/matplotgraph.py
@@ -59,4 +59,4 @@ def create_graph(lora_path, lora_name):
             print(f"File 'training_graph.json' does not exist in the {lora_path}")
       
     except ImportError:
-        print("matplotlib is not installed. Please install matplotlib to create PNG graphs")
\ No newline at end of file
+        print("matplotlib is not installed. Please install matplotlib to create PNG graphs")
diff --git a/extensions/Training_PRO/script.py b/extensions/Training_PRO/script.py
index f553e482d7..cb11a8df0a 100644
--- a/extensions/Training_PRO/script.py
+++ b/extensions/Training_PRO/script.py
@@ -175,23 +175,23 @@ def ui():
                     with gr.Row():
                         with gr.Column():
                             with gr.Row():
-                                dataset = gr.Dropdown(choices=get_datasets('training/datasets', 'json'), value='None', label='Dataset', info='The dataset file to use for training.', elem_classes=['slim-dropdown'])
-                                create_refresh_button(dataset, lambda: None, lambda: {'choices': get_datasets('training/datasets', 'json')}, 'refresh-button')
+                                dataset = gr.Dropdown(choices=get_datasets('user_data/training/datasets', 'json'), value='None', label='Dataset', info='The dataset file to use for training.', elem_classes=['slim-dropdown'])
+                                create_refresh_button(dataset, lambda: None, lambda: {'choices': get_datasets('user_data/training/datasets', 'json')}, 'refresh-button')
                             with gr.Row():
-                                eval_dataset = gr.Dropdown(choices=get_datasets('training/datasets', 'json'), value='None', label='Evaluation Dataset', info='The (optional) dataset file used to evaluate the model after training.', elem_classes=['slim-dropdown'])
-                                create_refresh_button(eval_dataset, lambda: None, lambda: {'choices': get_datasets('training/datasets', 'json')}, 'refresh-button')
+                                eval_dataset = gr.Dropdown(choices=get_datasets('user_data/training/datasets', 'json'), value='None', label='Evaluation Dataset', info='The (optional) dataset file used to evaluate the model after training.', elem_classes=['slim-dropdown'])
+                                create_refresh_button(eval_dataset, lambda: None, lambda: {'choices': get_datasets('user_data/training/datasets', 'json')}, 'refresh-button')
 
                         with gr.Column():
                             with gr.Row():
-                                format = gr.Dropdown(choices=get_datasets('training/formats', 'json'), value='None', label='Data Format', info='The format file used to decide how to format the dataset input.', elem_classes=['slim-dropdown'])
-                                create_refresh_button(format, lambda: None, lambda: {'choices': get_datasets('training/formats', 'json')}, 'refresh-button')
+                                format = gr.Dropdown(choices=get_datasets('user_data/training/formats', 'json'), value='None', label='Data Format', info='The format file used to decide how to format the dataset input.', elem_classes=['slim-dropdown'])
+                                create_refresh_button(format, lambda: None, lambda: {'choices': get_datasets('user_data/training/formats', 'json')}, 'refresh-button')
                             with gr.Row():
                                 eval_steps = gr.Number(label='Evaluate every n steps', value=100, info='If an evaluation dataset is given, test it every time this many steps pass.')
 
                 with gr.Tab(label="Text file"):
                     with gr.Row():
-                        raw_text_file = gr.Dropdown(choices=get_datasets('training/datasets', 'txt'), value='None', label='Text file', info='The text file to use for training.', elem_classes=['slim-dropdown'])
-                        create_refresh_button(raw_text_file, lambda: None, lambda: {'choices': get_datasets('training/datasets', 'txt')}, 'refresh-button')
+                        raw_text_file = gr.Dropdown(choices=get_datasets('user_data/training/datasets', 'txt'), value='None', label='Text file', info='The text file to use for training.', elem_classes=['slim-dropdown'])
+                        create_refresh_button(raw_text_file, lambda: None, lambda: {'choices': get_datasets('user_data/training/datasets', 'txt')}, 'refresh-button')
 
                     with gr.Row():
                         with gr.Column():
@@ -208,7 +208,7 @@ def ui():
                             download_file_url = gr.Textbox(label='Download JSON or txt file to datasets (or formats) folder', value='',info='The URL of a file to download. If on github, make sure you get url of the raw file (https://raw.githubusercontent.com/...). If huggin face, make sure the url has /resolve/ in it not /blob/')
                             with gr.Row():
                                 download_check_overwrite = gr.Checkbox(label='Overwrite', value=False, info='Overwrite if file exist')
-                                download_folder = gr.Radio(label="Destination", value='training/datasets', choices=['training/datasets', 'training/formats'], interactive=True)
+                                download_folder = gr.Radio(label="Destination", value='user_data/training/datasets', choices=['user_data/training/datasets', 'user_data/training/formats'], interactive=True)
                             download_button = gr.Button('Download')
                             download_status = gr.Textbox(label='Download Status', value='', interactive=False)
                 with gr.Row():
@@ -235,7 +235,7 @@ def ui():
         with gr.Row():
             with gr.Column():
                 models = gr.Dropdown(utils.get_available_models(), label='Models', multiselect=True)
-                evaluate_text_file = gr.Dropdown(choices=['wikitext', 'ptb', 'ptb_new'] + get_datasets('training/datasets', 'txt')[1:], value='wikitext', label='Input dataset', info='The text file on which the model will be evaluated. The first options are automatically downloaded: wikitext, ptb, and ptb_new. The next options are your local text files under training/datasets.')
+                evaluate_text_file = gr.Dropdown(choices=['wikitext', 'ptb', 'ptb_new'] + get_datasets('user_data/training/datasets', 'txt')[1:], value='wikitext', label='Input dataset', info='The text file on which the model will be evaluated. The first options are automatically downloaded: wikitext, ptb, and ptb_new. The next options are your local text files under user_data/training/datasets.')
                 with gr.Row():
                     with gr.Column():
                         stride_length = gr.Slider(label='Stride', minimum=1, maximum=2048, value=512, step=1, info='Used to make the evaluation faster at the cost of accuracy. 1 = slowest but most accurate. 512 is a common value.')
@@ -310,7 +310,7 @@ def check_dataset(save_steps:int, micro_batch_size: int, epochs: int, cutoff_len
         
         if raw_text_file not in ['None', '']:
             logger.info("Loading Text file...")
-            fullpath = clean_path('training/datasets', f'{raw_text_file}')
+            fullpath = clean_path('user_data/training/datasets', f'{raw_text_file}')
             fullpath = Path(fullpath)
             if fullpath.is_dir():
                 logger.info('Training path directory {}'.format(raw_text_file))
@@ -324,10 +324,10 @@ def check_dataset(save_steps:int, micro_batch_size: int, epochs: int, cutoff_len
                         logger.info(f"Loaded training file: {file_path.name}")
             else:
                 try:
-                    with open(clean_path('training/datasets', f'{raw_text_file}.txt'), 'r', encoding='utf-8') as file:
+                    with open(clean_path('user_data/training/datasets', f'{raw_text_file}.txt'), 'r', encoding='utf-8') as file:
                         raw_text = file.read().replace('\r', '')
                 except:
-                    yield f"{raw_text_file}.txt doesn't seem to exsist anymore... check your training/datasets folder"
+                    yield f"{raw_text_file}.txt doesn't seem to exsist anymore... check your user_data/training/datasets folder"
                     return
             
  
@@ -353,7 +353,7 @@ def check_dataset(save_steps:int, micro_batch_size: int, epochs: int, cutoff_len
                 yield "Select format choice for dataset."
                 return
 
-            with open(clean_path('training/formats', f'{format}.json'), 'r', encoding='utf-8-sig') as formatFile:
+            with open(clean_path('user_data/training/formats', f'{format}.json'), 'r', encoding='utf-8-sig') as formatFile:
                 format_data: dict[str, str] = json.load(formatFile)
 
             def generate_prompt(data_point: dict[str, str]):
@@ -381,7 +381,7 @@ def generate_and_tokenize_prompt(data_point):
                 return tokenize_dummy(prompt)
 
             logger.info("Loading JSON datasets...")
-            data = load_dataset("json", data_files=clean_path('training/datasets', f'{dataset}.json'))
+            data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
             
             data_keys = [] 
 
@@ -456,7 +456,7 @@ def reload_lora():
     #debug_slicer.change(lambda x: non_serialized_params.update({"debug_slicer": x}), debug_slicer, None)
 
     def update_dataset():
-        return gr.update(choices=get_datasets('training/datasets', 'json')), gr.update(choices=get_datasets('training/datasets', 'txt'))
+        return gr.update(choices=get_datasets('user_data/training/datasets', 'json')), gr.update(choices=get_datasets('user_data/training/datasets', 'txt'))
 
     download_button.click(download_file_from_url, [download_file_url,download_check_overwrite,download_folder] , download_status).then(update_dataset,None,[dataset , raw_text_file])
 
@@ -670,7 +670,7 @@ def tokenize(prompt, append_eos_token=False, prepend_bos_token = False):
     if raw_text_file not in ['None', '']:
         train_template["template_type"] = "raw_text"
         logger.info("Loading text file...")
-        fullpath = clean_path('training/datasets', f'{raw_text_file}')
+        fullpath = clean_path('user_data/training/datasets', f'{raw_text_file}')
         fullpath = Path(fullpath)
         if fullpath.is_dir():
             logger.info('Training path directory {}'.format(raw_text_file))
@@ -683,7 +683,7 @@ def tokenize(prompt, append_eos_token=False, prepend_bos_token = False):
 
                     logger.info(f"Loaded training file: {file_path.name}")
         else:
-            with open(clean_path('training/datasets', f'{raw_text_file}.txt'), 'r', encoding='utf-8') as file:
+            with open(clean_path('user_data/training/datasets', f'{raw_text_file}.txt'), 'r', encoding='utf-8') as file:
                 raw_text = file.read().replace('\r', '')
         
         # FPHAM PRECISE SLICING        
@@ -720,7 +720,7 @@ def tokenize(prompt, append_eos_token=False, prepend_bos_token = False):
 
         train_template["template_type"] = "dataset"
 
-        with open(clean_path('training/formats', f'{format}.json'), 'r', encoding='utf-8-sig') as formatFile:
+        with open(clean_path('user_data/training/formats', f'{format}.json'), 'r', encoding='utf-8-sig') as formatFile:
             format_data: dict[str, str] = json.load(formatFile)
 
         # == store training prompt ==
@@ -742,7 +742,7 @@ def generate_and_tokenize_prompt(data_point):
             return tokenize(prompt, add_eos_token, add_bos_token)
 
         logger.info("Loading JSON datasets...")
-        data = load_dataset("json", data_files=clean_path('training/datasets', f'{dataset}.json'))
+        data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
         train_data = data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
 
         print(f"BOS: {add_bos_token} EOS: {add_eos_token}") 
@@ -751,7 +751,7 @@ def generate_and_tokenize_prompt(data_point):
         if eval_dataset == 'None':
             eval_data = None
         else:
-            eval_data = load_dataset("json", data_files=clean_path('training/datasets', f'{eval_dataset}.json'))
+            eval_data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{eval_dataset}.json'))
             eval_data = eval_data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
 
     # == We MUST reload model if it went through any previous training, even failed one ==
@@ -1157,11 +1157,11 @@ def log_train_dataset(trainer):
                 decoded_entries.append({"value": decoded_text})
 
             # Write the log file
-            Path('logs').mkdir(exist_ok=True)
-            with open(Path('logs/train_dataset_sample.json'), 'w') as json_file:
+            Path('user_data/logs').mkdir(exist_ok=True)
+            with open(Path('user_data/logs/train_dataset_sample.json'), 'w') as json_file:
                 json.dump(decoded_entries, json_file, indent=4)
 
-            logger.info("Log file 'train_dataset_sample.json' created in the 'logs' directory.")
+            logger.info("Log file 'train_dataset_sample.json' created in the 'user_data/logs' directory.")
         except Exception as e:
             logger.error(f"Failed to create log file due to error: {e}")
 
diff --git a/extensions/Training_PRO/train_utils.py b/extensions/Training_PRO/train_utils.py
index 1868614441..799948802b 100644
--- a/extensions/Training_PRO/train_utils.py
+++ b/extensions/Training_PRO/train_utils.py
@@ -194,13 +194,13 @@ def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, c
 
     if debug_slicer:
                     # Write the log file
-        Path('logs').mkdir(exist_ok=True)
+        Path('user_data/logs').mkdir(exist_ok=True)
         sentencelist_dict = {index: sentence for index, sentence in enumerate(sentencelist)}
-        output_file = "logs/sentencelist.json"
+        output_file = "user_data/logs/sentencelist.json"
         with open(output_file, 'w') as f:
             json.dump(sentencelist_dict, f,indent=2)
         
-        print("Saved sentencelist.json in logs folder")
+        print("Saved sentencelist.json in user_data/logs folder")
     
     return sentencelist   
 
@@ -281,13 +281,13 @@ def sliding_block_cut(text: str, min_chars_cut: int, eos_to_hc: bool, cutoff_len
 
     if debug_slicer:
                     # Write the log file
-        Path('logs').mkdir(exist_ok=True)
+        Path('user_data/logs').mkdir(exist_ok=True)
         sentencelist_dict = {index: sentence for index, sentence in enumerate(sentencelist)}
-        output_file = "logs/sentencelist.json"
+        output_file = "user_data/logs/sentencelist.json"
         with open(output_file, 'w') as f:
             json.dump(sentencelist_dict, f,indent=2)
         
-        print("Saved sentencelist.json in logs folder")
+        print("Saved sentencelist.json in user_data/logs folder")
     
     return sentencelist   
 
diff --git a/extensions/gallery/script.py b/extensions/gallery/script.py
index 76be4a5838..8b242fb603 100644
--- a/extensions/gallery/script.py
+++ b/extensions/gallery/script.py
@@ -72,13 +72,13 @@ def generate_html():
     global cards
     cards = []
     # Iterate through files in image folder
-    for file in sorted(Path("characters").glob("*")):
+    for file in sorted(Path("user_data/characters").glob("*")):
         if file.suffix in [".json", ".yml", ".yaml"]:
             character = file.stem
             container_html = '
            '
             image_html = "
            "
 
-            for path in [Path(f"characters/{character}.{extension}") for extension in ['png', 'jpg', 'jpeg']]:
+            for path in [Path(f"user_data/characters/{character}.{extension}") for extension in ['png', 'jpg', 'jpeg']]:
                 if path.exists():
                     image_html = f''
                     break
diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index ea68889704..4d6018f9ec 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -6,7 +6,7 @@
 
 
 class GenerationOptions(BaseModel):
-    preset: str | None = Field(default=None, description="The name of a file under text-generation-webui/presets (without the .yaml extension). The sampling parameters that get overwritten by this option are the keys in the default_preset() function in modules/presets.py.")
+    preset: str | None = Field(default=None, description="The name of a file under text-generation-webui/user_data/presets (without the .yaml extension). The sampling parameters that get overwritten by this option are the keys in the default_preset() function in modules/presets.py.")
     dynatemp_low: float = 1
     dynatemp_high: float = 1
     dynatemp_exponent: float = 1
@@ -103,10 +103,10 @@ class ChatCompletionRequestParams(BaseModel):
 
     mode: str = Field(default='instruct', description="Valid options: instruct, chat, chat-instruct.")
 
-    instruction_template: str | None = Field(default=None, description="An instruction template defined under text-generation-webui/instruction-templates. If not set, the correct template will be automatically obtained from the model metadata.")
+    instruction_template: str | None = Field(default=None, description="An instruction template defined under text-generation-webui/user_data/instruction-templates. If not set, the correct template will be automatically obtained from the model metadata.")
     instruction_template_str: str | None = Field(default=None, description="A Jinja2 instruction template. If set, will take precedence over everything else.")
 
-    character: str | None = Field(default=None, description="A character defined under text-generation-webui/characters. If not set, the default \"Assistant\" character will be used.")
+    character: str | None = Field(default=None, description="A character defined under text-generation-webui/user_data/characters. If not set, the default \"Assistant\" character will be used.")
     bot_name: str | None = Field(default=None, description="Overwrites the value set by character field.", alias="name2")
     context: str | None = Field(default=None, description="Overwrites the value set by character field.")
     greeting: str | None = Field(default=None, description="Overwrites the value set by character field.")
diff --git a/js/main.js b/js/main.js
index c5c47d0472..33b7d6bdc1 100644
--- a/js/main.js
+++ b/js/main.js
@@ -395,7 +395,7 @@ let bigPictureVisible = false;
 function addBigPicture() {
   var imgElement = document.createElement("img");
   var timestamp = new Date().getTime();
-  imgElement.src = "/file/cache/pfp_character.png?time=" + timestamp;
+  imgElement.src = "/file/user_data/cache/pfp_character.png?time=" + timestamp;
   imgElement.classList.add("bigProfilePicture");
   imgElement.addEventListener("load", function () {
     this.style.visibility = "visible";
diff --git a/js/update_big_picture.js b/js/update_big_picture.js
index 4c094776b9..ec51d63b07 100644
--- a/js/update_big_picture.js
+++ b/js/update_big_picture.js
@@ -2,6 +2,6 @@ function updateBigPicture() {
   var existingElement = document.querySelector(".bigProfilePicture");
   if (existingElement) {
     var timestamp = new Date().getTime();
-    existingElement.src = "/file/cache/pfp_character.png?time=" + timestamp;
+    existingElement.src = "/file/user_data/cache/pfp_character.png?time=" + timestamp;
   }
 }
diff --git a/modules/chat.py b/modules/chat.py
index 94d90bdc3d..e117e6eeef 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -525,9 +525,9 @@ def start_new_chat(state):
 
 def get_history_file_path(unique_id, character, mode):
     if mode == 'instruct':
-        p = Path(f'logs/instruct/{unique_id}.json')
+        p = Path(f'user_data/logs/instruct/{unique_id}.json')
     else:
-        p = Path(f'logs/chat/{character}/{unique_id}.json')
+        p = Path(f'user_data/logs/chat/{character}/{unique_id}.json')
 
     return p
 
@@ -563,13 +563,13 @@ def rename_history(old_id, new_id, character, mode):
 
 def get_paths(state):
     if state['mode'] == 'instruct':
-        return Path('logs/instruct').glob('*.json')
+        return Path('user_data/logs/instruct').glob('*.json')
     else:
         character = state['character_menu']
 
         # Handle obsolete filenames and paths
-        old_p = Path(f'logs/{character}_persistent.json')
-        new_p = Path(f'logs/persistent_{character}.json')
+        old_p = Path(f'user_data/logs/{character}_persistent.json')
+        new_p = Path(f'user_data/logs/persistent_{character}.json')
         if old_p.exists():
             logger.warning(f"Renaming \"{old_p}\" to \"{new_p}\"")
             old_p.rename(new_p)
@@ -581,7 +581,7 @@ def get_paths(state):
             p.parent.mkdir(exist_ok=True)
             new_p.rename(p)
 
-        return Path(f'logs/chat/{character}').glob('*.json')
+        return Path(f'user_data/logs/chat/{character}').glob('*.json')
 
 
 def find_all_histories(state):
@@ -732,7 +732,7 @@ def generate_pfp_cache(character):
     if not cache_folder.exists():
         cache_folder.mkdir()
 
-    for path in [Path(f"characters/{character}.{extension}") for extension in ['png', 'jpg', 'jpeg']]:
+    for path in [Path(f"user_data/characters/{character}.{extension}") for extension in ['png', 'jpg', 'jpeg']]:
         if path.exists():
             original_img = Image.open(path)
             original_img.save(Path(f'{cache_folder}/pfp_character.png'), format='PNG')
@@ -752,12 +752,12 @@ def load_character(character, name1, name2):
 
     filepath = None
     for extension in ["yml", "yaml", "json"]:
-        filepath = Path(f'characters/{character}.{extension}')
+        filepath = Path(f'user_data/characters/{character}.{extension}')
         if filepath.exists():
             break
 
     if filepath is None or not filepath.exists():
-        logger.error(f"Could not find the character \"{character}\" inside characters/. No character has been loaded.")
+        logger.error(f"Could not find the character \"{character}\" inside user_data/characters. No character has been loaded.")
         raise ValueError
 
     file_contents = open(filepath, 'r', encoding='utf-8').read()
@@ -796,7 +796,7 @@ def load_instruction_template(template):
     if template == 'None':
         return ''
 
-    for filepath in [Path(f'instruction-templates/{template}.yaml'), Path('instruction-templates/Alpaca.yaml')]:
+    for filepath in [Path(f'user_data/instruction-templates/{template}.yaml'), Path('user_data/instruction-templates/Alpaca.yaml')]:
         if filepath.exists():
             break
     else:
@@ -838,17 +838,17 @@ def upload_character(file, img, tavern=False):
 
     outfile_name = name
     i = 1
-    while Path(f'characters/{outfile_name}.yaml').exists():
+    while Path(f'user_data/characters/{outfile_name}.yaml').exists():
         outfile_name = f'{name}_{i:03d}'
         i += 1
 
-    with open(Path(f'characters/{outfile_name}.yaml'), 'w', encoding='utf-8') as f:
+    with open(Path(f'user_data/characters/{outfile_name}.yaml'), 'w', encoding='utf-8') as f:
         f.write(yaml_data)
 
     if img is not None:
-        img.save(Path(f'characters/{outfile_name}.png'))
+        img.save(Path(f'user_data/characters/{outfile_name}.png'))
 
-    logger.info(f'New character saved to "characters/{outfile_name}.yaml".')
+    logger.info(f'New character saved to "user_data/characters/{outfile_name}.yaml".')
     return gr.update(value=outfile_name, choices=get_available_characters())
 
 
@@ -923,9 +923,9 @@ def save_character(name, greeting, context, picture, filename):
         return
 
     data = generate_character_yaml(name, greeting, context)
-    filepath = Path(f'characters/{filename}.yaml')
+    filepath = Path(f'user_data/characters/{filename}.yaml')
     save_file(filepath, data)
-    path_to_img = Path(f'characters/{filename}.png')
+    path_to_img = Path(f'user_data/characters/{filename}.png')
     if picture is not None:
         picture.save(path_to_img)
         logger.info(f'Saved {path_to_img}.')
@@ -933,9 +933,9 @@ def save_character(name, greeting, context, picture, filename):
 
 def delete_character(name, instruct=False):
     for extension in ["yml", "yaml", "json"]:
-        delete_file(Path(f'characters/{name}.{extension}'))
+        delete_file(Path(f'user_data/characters/{name}.{extension}'))
 
-    delete_file(Path(f'characters/{name}.png'))
+    delete_file(Path(f'user_data/characters/{name}.png'))
 
 
 def jinja_template_from_old_format(params, verbose=False):
@@ -1238,7 +1238,7 @@ def handle_save_template_click(instruction_template_str):
     contents = generate_instruction_template_yaml(instruction_template_str)
     return [
         "My Template.yaml",
-        "instruction-templates/",
+        "user_data/instruction-templates/",
         contents,
         gr.update(visible=True)
     ]
@@ -1247,7 +1247,7 @@ def handle_save_template_click(instruction_template_str):
 def handle_delete_template_click(template):
     return [
         f"{template}.yaml",
-        "instruction-templates/",
+        "user_data/instruction-templates/",
         gr.update(visible=False)
     ]
 
diff --git a/modules/evaluate.py b/modules/evaluate.py
index ba0de378b3..4f41c1fc14 100644
--- a/modules/evaluate.py
+++ b/modules/evaluate.py
@@ -12,8 +12,8 @@
 
 
 def load_past_evaluations():
-    if Path('logs/evaluations.csv').exists():
-        df = pd.read_csv(Path('logs/evaluations.csv'), dtype=str)
+    if Path('user_data/logs/evaluations.csv').exists():
+        df = pd.read_csv(Path('user_data/logs/evaluations.csv'), dtype=str)
         df['Perplexity'] = pd.to_numeric(df['Perplexity'])
         return df
     else:
@@ -26,7 +26,7 @@ def load_past_evaluations():
 def save_past_evaluations(df):
     global past_evaluations
     past_evaluations = df
-    filepath = Path('logs/evaluations.csv')
+    filepath = Path('user_data/logs/evaluations.csv')
     filepath.parent.mkdir(parents=True, exist_ok=True)
     df.to_csv(filepath, index=False)
 
@@ -69,7 +69,7 @@ def calculate_perplexity(models, input_dataset, stride, _max_length):
         data = load_dataset('ptb_text_only', 'penn_treebank', split='test')
         text = " ".join(data['sentence'])
     else:
-        with open(Path(f'training/datasets/{input_dataset}.txt'), 'r', encoding='utf-8') as f:
+        with open(Path(f'user_data/training/datasets/{input_dataset}.txt'), 'r', encoding='utf-8') as f:
             text = f.read()
 
     for model in models:
diff --git a/modules/html_generator.py b/modules/html_generator.py
index a72e485973..c5252c26be 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -387,13 +387,13 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
 
     # We use ?character and ?time.time() to force the browser to reset caches
     img_bot = (
-        f''
-        if Path("cache/pfp_character_thumb.png").exists() else ''
+        f''
+        if Path("user_data/cache/pfp_character_thumb.png").exists() else ''
     )
 
     img_me = (
-        f''
-        if Path("cache/pfp_me.png").exists() else ''
+        f''
+        if Path("user_data/cache/pfp_me.png").exists() else ''
     )
 
     for i in range(len(history['visible'])):
diff --git a/modules/models_settings.py b/modules/models_settings.py
index d3ecd51feb..ae589bb3a5 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -25,7 +25,7 @@ def get_fallback_settings():
 def get_model_metadata(model):
     model_settings = {}
 
-    # Get settings from models/config.yaml and models/config-user.yaml
+    # Get settings from user_data/models/config.yaml and user_data/models/config-user.yaml
     settings = shared.model_config
     for pat in settings:
         if re.match(pat.lower(), Path(model).name.lower()):
@@ -144,7 +144,7 @@ def get_model_metadata(model):
     if 'rope_freq_base' in model_settings and model_settings['rope_freq_base'] == 10000:
         model_settings.pop('rope_freq_base')
 
-    # Apply user settings from models/config-user.yaml
+    # Apply user settings from user_data/models/config-user.yaml
     settings = shared.user_config
     for pat in settings:
         if re.match(pat.lower(), Path(model).name.lower()):
@@ -223,7 +223,7 @@ def apply_model_settings_to_state(model, state):
 
 def save_model_settings(model, state):
     '''
-    Save the settings for this model to models/config-user.yaml
+    Save the settings for this model to user_data/models/config-user.yaml
     '''
     if model == 'None':
         yield ("Not saving the settings because no model is selected in the menu.")
diff --git a/modules/presets.py b/modules/presets.py
index 7cab2af00b..a432bf5233 100644
--- a/modules/presets.py
+++ b/modules/presets.py
@@ -58,7 +58,7 @@ def presets_params():
 def load_preset(name, verbose=False):
     generate_params = default_preset()
     if name not in ['None', None, '']:
-        path = Path(f'presets/{name}.yaml')
+        path = Path(f'user_data/presets/{name}.yaml')
         if path.exists():
             with open(path, 'r') as infile:
                 preset = yaml.safe_load(infile)
diff --git a/modules/prompts.py b/modules/prompts.py
index 565c245079..8f00cac242 100644
--- a/modules/prompts.py
+++ b/modules/prompts.py
@@ -7,7 +7,7 @@ def load_prompt(fname):
     if fname in ['None', '']:
         return ''
     else:
-        file_path = Path(f'prompts/{fname}.txt')
+        file_path = Path(f'user_data/prompts/{fname}.txt')
         if not file_path.exists():
             return ''
 
diff --git a/modules/shared.py b/modules/shared.py
index 96f659297f..21e6dd005d 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -78,8 +78,8 @@
 group.add_argument('--character', type=str, help='The name of the character to load in chat mode by default.')
 group.add_argument('--model', type=str, help='Name of the model to load by default.')
 group.add_argument('--lora', type=str, nargs='+', help='The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.')
-group.add_argument('--model-dir', type=str, default='models/', help='Path to directory with all the models.')
-group.add_argument('--lora-dir', type=str, default='loras/', help='Path to directory with all the loras.')
+group.add_argument('--model-dir', type=str, default='user_data/models', help='Path to directory with all the models.')
+group.add_argument('--lora-dir', type=str, default='user_data/loras', help='Path to directory with all the loras.')
 group.add_argument('--model-menu', action='store_true', help='Show a model menu in the terminal when the web UI is first launched.')
 group.add_argument('--settings', type=str, help='Load the default interface settings from this yaml file. See settings-template.yaml for an example. If you create a file called settings.yaml, this file will be loaded by default without the need to use the --settings flag.')
 group.add_argument('--extensions', type=str, nargs='+', help='The list of extensions to load. If you want to load more than one extension, write the names separated by spaces.')
@@ -95,7 +95,7 @@
 group.add_argument('--cpu', action='store_true', help='Use the CPU to generate text. Warning: Training on CPU is extremely slow.')
 group.add_argument('--cpu-memory', type=float, default=0, help='Maximum CPU memory in GiB. Use this for CPU offloading.')
 group.add_argument('--disk', action='store_true', help='If the model is too large for your GPU(s) and CPU combined, send the remaining layers to the disk.')
-group.add_argument('--disk-cache-dir', type=str, default='cache', help='Directory to save the disk cache to. Defaults to "cache".')
+group.add_argument('--disk-cache-dir', type=str, default='user_data/cache', help='Directory to save the disk cache to. Defaults to "user_data/cache".')
 group.add_argument('--load-in-8bit', action='store_true', help='Load the model with 8-bit precision (using bitsandbytes).')
 group.add_argument('--bf16', action='store_true', help='Load the model with bfloat16 precision. Requires NVIDIA Ampere GPU.')
 group.add_argument('--no-cache', action='store_true', help='Set use_cache to False while generating text. This reduces VRAM usage slightly, but it comes at a performance cost.')
@@ -207,7 +207,7 @@
 group = parser.add_argument_group('Deprecated')
 
 # Handle CMD_FLAGS.txt
-cmd_flags_path = Path(__file__).parent.parent / "CMD_FLAGS.txt"
+cmd_flags_path = Path(__file__).parent.parent / "user_data" / "CMD_FLAGS.txt"
 if cmd_flags_path.exists():
     with cmd_flags_path.open('r', encoding='utf-8') as f:
         cmd_flags = ' '.join(
diff --git a/modules/training.py b/modules/training.py
index 69142463e8..2354c39df9 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -106,23 +106,23 @@ def create_ui():
                 with gr.Column():
                     with gr.Tab(label='Formatted Dataset'):
                         with gr.Row():
-                            format = gr.Dropdown(choices=utils.get_datasets('training/formats', 'json'), value='None', label='Data Format', info='The format file used to decide how to format the dataset input.', elem_classes=['slim-dropdown'], interactive=not mu)
-                            ui.create_refresh_button(format, lambda: None, lambda: {'choices': utils.get_datasets('training/formats', 'json')}, 'refresh-button', interactive=not mu)
+                            format = gr.Dropdown(choices=utils.get_datasets('user_data/training/formats', 'json'), value='None', label='Data Format', info='The format file used to decide how to format the dataset input.', elem_classes=['slim-dropdown'], interactive=not mu)
+                            ui.create_refresh_button(format, lambda: None, lambda: {'choices': utils.get_datasets('user_data/training/formats', 'json')}, 'refresh-button', interactive=not mu)
 
                         with gr.Row():
-                            dataset = gr.Dropdown(choices=utils.get_datasets('training/datasets', 'json'), value='None', label='Dataset', info='The dataset file to use for training.', elem_classes=['slim-dropdown'], interactive=not mu)
-                            ui.create_refresh_button(dataset, lambda: None, lambda: {'choices': utils.get_datasets('training/datasets', 'json')}, 'refresh-button', interactive=not mu)
+                            dataset = gr.Dropdown(choices=utils.get_datasets('user_data/training/datasets', 'json'), value='None', label='Dataset', info='The dataset file to use for training.', elem_classes=['slim-dropdown'], interactive=not mu)
+                            ui.create_refresh_button(dataset, lambda: None, lambda: {'choices': utils.get_datasets('user_data/training/datasets', 'json')}, 'refresh-button', interactive=not mu)
 
                         with gr.Row():
-                            eval_dataset = gr.Dropdown(choices=utils.get_datasets('training/datasets', 'json'), value='None', label='Evaluation Dataset', info='The (optional) dataset file used to evaluate the model after training.', elem_classes=['slim-dropdown'], interactive=not mu)
-                            ui.create_refresh_button(eval_dataset, lambda: None, lambda: {'choices': utils.get_datasets('training/datasets', 'json')}, 'refresh-button', interactive=not mu)
+                            eval_dataset = gr.Dropdown(choices=utils.get_datasets('user_data/training/datasets', 'json'), value='None', label='Evaluation Dataset', info='The (optional) dataset file used to evaluate the model after training.', elem_classes=['slim-dropdown'], interactive=not mu)
+                            ui.create_refresh_button(eval_dataset, lambda: None, lambda: {'choices': utils.get_datasets('user_data/training/datasets', 'json')}, 'refresh-button', interactive=not mu)
 
                         eval_steps = gr.Number(label='Evaluate every n steps', value=100, info='If an evaluation dataset is given, test it every time this many steps pass.')
 
                     with gr.Tab(label="Raw text file"):
                         with gr.Row():
-                            raw_text_file = gr.Dropdown(choices=utils.get_datasets('training/datasets', 'txt'), value='None', label='Text file', info='The raw text file to use for training.', elem_classes=['slim-dropdown'], interactive=not mu)
-                            ui.create_refresh_button(raw_text_file, lambda: None, lambda: {'choices': utils.get_datasets('training/datasets', 'txt')}, 'refresh-button', interactive=not mu)
+                            raw_text_file = gr.Dropdown(choices=utils.get_datasets('user_data/training/datasets', 'txt'), value='None', label='Text file', info='The raw text file to use for training.', elem_classes=['slim-dropdown'], interactive=not mu)
+                            ui.create_refresh_button(raw_text_file, lambda: None, lambda: {'choices': utils.get_datasets('user_data/training/datasets', 'txt')}, 'refresh-button', interactive=not mu)
 
                         with gr.Row():
                             with gr.Column():
@@ -143,7 +143,7 @@ def create_ui():
             with gr.Row():
                 with gr.Column():
                     models = gr.Dropdown(utils.get_available_models(), label='Models', multiselect=True, interactive=not mu)
-                    evaluate_text_file = gr.Dropdown(choices=['wikitext', 'ptb', 'ptb_new'] + utils.get_datasets('training/datasets', 'txt')[1:], value='wikitext', label='Input dataset', info='The raw text file on which the model will be evaluated. The first options are automatically downloaded: wikitext, ptb, and ptb_new. The next options are your local text files under training/datasets.', interactive=not mu)
+                    evaluate_text_file = gr.Dropdown(choices=['wikitext', 'ptb', 'ptb_new'] + utils.get_datasets('user_data/training/datasets', 'txt')[1:], value='wikitext', label='Input dataset', info='The raw text file on which the model will be evaluated. The first options are automatically downloaded: wikitext, ptb, and ptb_new. The next options are your local text files under user_data/training/datasets.', interactive=not mu)
                     with gr.Row():
                         with gr.Column():
                             stride_length = gr.Slider(label='Stride', minimum=0, maximum=32768, value=512, step=256, info='Used to make the evaluation faster at the cost of accuracy. 1 = slowest but most accurate. 512 is a common value.')
@@ -402,7 +402,7 @@ def tokenize(prompt, append_eos_token=False):
     if raw_text_file not in ['None', '']:
         train_template["template_type"] = "raw_text"
         logger.info("Loading raw text file dataset")
-        fullpath = clean_path('training/datasets', f'{raw_text_file}')
+        fullpath = clean_path('user_data/training/datasets', f'{raw_text_file}')
         fullpath = Path(fullpath)
         if fullpath.is_dir():
             logger.info('Training path directory {}'.format(raw_text_file))
@@ -415,7 +415,7 @@ def tokenize(prompt, append_eos_token=False):
 
                     logger.info(f"Loaded training file: {file_path.name}")
         else:
-            with open(clean_path('training/datasets', f'{raw_text_file}.txt'), 'r', encoding='utf-8') as file:
+            with open(clean_path('user_data/training/datasets', f'{raw_text_file}.txt'), 'r', encoding='utf-8') as file:
                 raw_text = file.read().replace('\r', '')
 
         cut_string = hard_cut_string.replace('\\n', '\n')
@@ -460,7 +460,7 @@ def tokenize(prompt, append_eos_token=False):
 
         train_template["template_type"] = "dataset"
 
-        with open(clean_path('training/formats', f'{format}.json'), 'r', encoding='utf-8-sig') as formatFile:
+        with open(clean_path('user_data/training/formats', f'{format}.json'), 'r', encoding='utf-8-sig') as formatFile:
             format_data: dict[str, str] = json.load(formatFile)
 
         # == store training prompt ==
@@ -482,13 +482,13 @@ def generate_and_tokenize_prompt(data_point):
             return tokenize(prompt, add_eos_token)
 
         logger.info("Loading JSON datasets")
-        data = load_dataset("json", data_files=clean_path('training/datasets', f'{dataset}.json'))
+        data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
         train_data = data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
 
         if eval_dataset == 'None':
             eval_data = None
         else:
-            eval_data = load_dataset("json", data_files=clean_path('training/datasets', f'{eval_dataset}.json'))
+            eval_data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{eval_dataset}.json'))
             eval_data = eval_data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
 
     # == We MUST reload model if it went through any previous training, even failed one ==
@@ -676,11 +676,11 @@ def log_train_dataset(trainer):
                 decoded_entries.append({"value": decoded_text})
 
             # Write the log file
-            Path('logs').mkdir(exist_ok=True)
-            with open(Path('logs/train_dataset_sample.json'), 'w') as json_file:
+            Path('user_data/logs').mkdir(exist_ok=True)
+            with open(Path('user_data/logs/train_dataset_sample.json'), 'w') as json_file:
                 json.dump(decoded_entries, json_file, indent=4)
 
-            logger.info("Log file 'train_dataset_sample.json' created in the 'logs' directory.")
+            logger.info("Log file 'train_dataset_sample.json' created in the 'user_data/logs' directory.")
         except Exception as e:
             logger.error(f"Failed to create log file due to error: {e}")
 
diff --git a/modules/transformers_loader.py b/modules/transformers_loader.py
index add3be66cf..905f5c4728 100644
--- a/modules/transformers_loader.py
+++ b/modules/transformers_loader.py
@@ -249,7 +249,7 @@ def load_model_HF(model_name):
                     )
 
             if shared.args.disk:
-                params['offload_folder'] = shared.args.disk_cache_dir
+                params['offload_folder'] = str(Path(shared.args.disk_cache_dir))
 
         if shared.args.compress_pos_emb > 1:
             params['rope_scaling'] = {'type': 'linear', 'factor': shared.args.compress_pos_emb}
diff --git a/modules/ui.py b/modules/ui.py
index 68cb76a615..ef5ed0e653 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -94,7 +94,7 @@
         input_radius='0.375rem',
     )
 
-if Path("notification.mp3").exists():
+if Path("user_data/notification.mp3").exists():
     audio_notification_js = "document.querySelector('#audio_notification audio')?.play();"
 else:
     audio_notification_js = ""
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index b823b8e5f6..0d58854945 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -146,7 +146,7 @@ def create_chat_settings_ui():
 
             with gr.Column(scale=1):
                 shared.gradio['character_picture'] = gr.Image(label='Character picture', type='pil', interactive=not mu)
-                shared.gradio['your_picture'] = gr.Image(label='Your picture', type='pil', value=Image.open(Path('cache/pfp_me.png')) if Path('cache/pfp_me.png').exists() else None, interactive=not mu)
+                shared.gradio['your_picture'] = gr.Image(label='Your picture', type='pil', value=Image.open(Path('user_data/cache/pfp_me.png')) if Path('user_data/cache/pfp_me.png').exists() else None, interactive=not mu)
 
     with gr.Tab('Instruction template'):
         with gr.Row():
diff --git a/modules/ui_default.py b/modules/ui_default.py
index ccae9a5ec3..c2946b3789 100644
--- a/modules/ui_default.py
+++ b/modules/ui_default.py
@@ -102,7 +102,7 @@ def handle_save_prompt(text):
     return [
         text,
         utils.current_time() + ".txt",
-        "prompts/",
+        "user_data/prompts/",
         gr.update(visible=True)
     ]
 
@@ -110,6 +110,6 @@ def handle_save_prompt(text):
 def handle_delete_prompt(prompt):
     return [
         prompt + ".txt",
-        "prompts/",
+        "user_data/prompts/",
         gr.update(visible=True)
     ]
diff --git a/modules/ui_file_saving.py b/modules/ui_file_saving.py
index 3a27e1b957..d1f9379b5a 100644
--- a/modules/ui_file_saving.py
+++ b/modules/ui_file_saving.py
@@ -28,7 +28,7 @@ def create_ui():
 
     # Character saver/deleter
     with gr.Group(visible=False, elem_classes='file-saver') as shared.gradio['character_saver']:
-        shared.gradio['save_character_filename'] = gr.Textbox(lines=1, label='File name', info='The character will be saved to your characters/ folder with this base filename.')
+        shared.gradio['save_character_filename'] = gr.Textbox(lines=1, label='File name', info='The character will be saved to your user_data/characters folder with this base filename.')
         with gr.Row():
             shared.gradio['save_character_cancel'] = gr.Button('Cancel', elem_classes="small-button")
             shared.gradio['save_character_confirm'] = gr.Button('Save', elem_classes="small-button", variant='primary', interactive=not mu)
@@ -41,7 +41,7 @@ def create_ui():
 
     # Preset saver
     with gr.Group(visible=False, elem_classes='file-saver') as shared.gradio['preset_saver']:
-        shared.gradio['save_preset_filename'] = gr.Textbox(lines=1, label='File name', info='The preset will be saved to your presets/ folder with this base filename.')
+        shared.gradio['save_preset_filename'] = gr.Textbox(lines=1, label='File name', info='The preset will be saved to your user_data/presets folder with this base filename.')
         shared.gradio['save_preset_contents'] = gr.Textbox(lines=10, label='File contents')
         with gr.Row():
             shared.gradio['save_preset_cancel'] = gr.Button('Cancel', elem_classes="small-button")
@@ -72,7 +72,7 @@ def create_event_handlers():
 
 def handle_save_preset_confirm_click(filename, contents):
     try:
-        utils.save_file(f"presets/{filename}.yaml", contents)
+        utils.save_file(f"user_data/presets/{filename}.yaml", contents)
         available_presets = utils.get_available_presets()
         output = gr.update(choices=available_presets, value=filename)
     except Exception:
@@ -145,7 +145,7 @@ def handle_save_preset_click(state):
 def handle_delete_preset_click(preset):
     return [
         f"{preset}.yaml",
-        "presets/",
+        "user_data/presets/",
         gr.update(visible=True)
     ]
 
@@ -154,7 +154,7 @@ def handle_save_grammar_click(grammar_string):
     return [
         grammar_string,
         "My Fancy Grammar.gbnf",
-        "grammars/",
+        "user_data/grammars/",
         gr.update(visible=True)
     ]
 
@@ -162,6 +162,6 @@ def handle_save_grammar_click(grammar_string):
 def handle_delete_grammar_click(grammar_file):
     return [
         grammar_file,
-        "grammars/",
+        "user_data/grammars/",
         gr.update(visible=True)
     ]
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 6bd647c6f7..dc09c899c5 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -223,9 +223,9 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
             model_dir=shared.args.model_dir if shared.args.model_dir != shared.args_defaults.model_dir else None
         )
 
-        if output_folder == Path("models"):
+        if output_folder == Path("user_data/models"):
             output_folder = Path(shared.args.model_dir)
-        elif output_folder == Path("loras"):
+        elif output_folder == Path("user_data/loras"):
             output_folder = Path(shared.args.lora_dir)
 
         if check:
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 156e4128ed..6c2715af41 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -128,7 +128,7 @@ def get_truncation_length():
 
 
 def load_grammar(name):
-    p = Path(f'grammars/{name}')
+    p = Path(f'user_data/grammars/{name}')
     if p.exists():
         return open(p, 'r', encoding='utf-8').read()
     else:
diff --git a/modules/ui_session.py b/modules/ui_session.py
index 66386d1203..42434e5166 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -48,7 +48,7 @@ def handle_save_settings(state, preset, extensions, show_controls, theme):
     return [
         contents,
         "settings.yaml",
-        "./",
+        "./user_data",
         gr.update(visible=True)
     ]
 
diff --git a/modules/utils.py b/modules/utils.py
index 269561aad9..7732413906 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -98,7 +98,7 @@ def get_available_models():
     dirs_with_gguf = set()
     for gguf_path in gguf_files:
         path = Path(gguf_path)
-        if path.parts:
+        if len(path.parts) > 0:
             dirs_with_gguf.add(path.parts[0])
 
     # Find directories with safetensors files
@@ -141,11 +141,11 @@ def get_available_ggufs():
 
 
 def get_available_presets():
-    return sorted(set((k.stem for k in Path('presets').glob('*.yaml'))), key=natural_keys)
+    return sorted(set((k.stem for k in Path('user_data/presets').glob('*.yaml'))), key=natural_keys)
 
 
 def get_available_prompts():
-    prompt_files = list(Path('prompts').glob('*.txt'))
+    prompt_files = list(Path('user_data/prompts').glob('*.txt'))
     sorted_files = sorted(prompt_files, key=lambda x: x.stat().st_mtime, reverse=True)
     prompts = [file.stem for file in sorted_files]
     prompts.append('None')
@@ -153,12 +153,12 @@ def get_available_prompts():
 
 
 def get_available_characters():
-    paths = (x for x in Path('characters').iterdir() if x.suffix in ('.json', '.yaml', '.yml'))
+    paths = (x for x in Path('user_data/characters').iterdir() if x.suffix in ('.json', '.yaml', '.yml'))
     return sorted(set((k.stem for k in paths)), key=natural_keys)
 
 
 def get_available_instruction_templates():
-    path = "instruction-templates"
+    path = "user_data/instruction-templates"
     paths = []
     if os.path.exists(path):
         paths = (x for x in Path(path).iterdir() if x.suffix in ('.json', '.yaml', '.yml'))
@@ -189,4 +189,4 @@ def get_available_chat_styles():
 
 
 def get_available_grammars():
-    return ['None'] + sorted([item.name for item in list(Path('grammars').glob('*.gbnf'))], key=natural_keys)
+    return ['None'] + sorted([item.name for item in list(Path('user_data/grammars').glob('*.gbnf'))], key=natural_keys)
diff --git a/one_click.py b/one_click.py
index 5e3d691b2e..065afd999f 100644
--- a/one_click.py
+++ b/one_click.py
@@ -293,10 +293,10 @@ def install_webui():
 
     # Write a flag to CMD_FLAGS.txt for CPU mode
     if selected_gpu == "NONE":
-        cmd_flags_path = os.path.join(script_dir, "CMD_FLAGS.txt")
+        cmd_flags_path = os.path.join(script_dir, "user_data", "CMD_FLAGS.txt")
         with open(cmd_flags_path, 'r+') as cmd_flags_file:
             if "--cpu" not in cmd_flags_file.read():
-                print_big_message("Adding the --cpu flag to CMD_FLAGS.txt.")
+                print_big_message("Adding the --cpu flag to user_data/CMD_FLAGS.txt.")
                 cmd_flags_file.write("\n--cpu\n")
 
     # Handle CUDA version display
@@ -532,7 +532,7 @@ def launch_webui():
             flags_list = re.split(' +(?=(?:[^\"]*\"[^\"]*\")*[^\"]*$)|=', flags)
             model_dir = [flags_list[(flags_list.index(flag) + 1)] for flag in flags_list if flag == '--model-dir'][0].strip('"\'')
         else:
-            model_dir = 'models'
+            model_dir = 'user_data/models'
 
         if len([item for item in glob.glob(f'{model_dir}/*') if not item.endswith(('.txt', '.yaml'))]) == 0:
             print_big_message("You haven't downloaded any model yet.\nOnce the web UI launches, head over to the \"Model\" tab and download one.")
diff --git a/server.py b/server.py
index 41a5660dd6..01d40ac408 100644
--- a/server.py
+++ b/server.py
@@ -94,8 +94,8 @@ def create_interface():
         'filter_by_loader': shared.args.loader or 'All'
     })
 
-    if Path("cache/pfp_character.png").exists():
-        Path("cache/pfp_character.png").unlink()
+    if Path("user_data/cache/pfp_character.png").exists():
+        Path("user_data/cache/pfp_character.png").unlink()
 
     # css/js strings
     css = ui.css
@@ -112,8 +112,8 @@ def create_interface():
         shared.gradio['interface_state'] = gr.State({k: None for k in shared.input_elements})
 
         # Audio notification
-        if Path("notification.mp3").exists():
-            shared.gradio['audio_notification'] = gr.Audio(interactive=False, value="notification.mp3", elem_id="audio_notification", visible=False)
+        if Path("user_data/notification.mp3").exists():
+            shared.gradio['audio_notification'] = gr.Audio(interactive=False, value="user_data/notification.mp3", elem_id="audio_notification", visible=False)
 
         # Floating menus for saving/deleting files
         ui_file_saving.create_ui()
@@ -179,7 +179,7 @@ def create_interface():
             ssl_keyfile=shared.args.ssl_keyfile,
             ssl_certfile=shared.args.ssl_certfile,
             root_path=shared.args.subpath,
-            allowed_paths=["cache", "css", "extensions", "js"]
+            allowed_paths=["css", "js", "extensions", "user_data/cache"]
         )
 
 
@@ -192,10 +192,10 @@ def create_interface():
     settings_file = None
     if shared.args.settings is not None and Path(shared.args.settings).exists():
         settings_file = Path(shared.args.settings)
-    elif Path('settings.yaml').exists():
-        settings_file = Path('settings.yaml')
-    elif Path('settings.json').exists():
-        settings_file = Path('settings.json')
+    elif Path('user_data/settings.yaml').exists():
+        settings_file = Path('user_data/settings.yaml')
+    elif Path('user_data/settings.json').exists():
+        settings_file = Path('user_data/settings.json')
 
     if settings_file is not None:
         logger.info(f"Loading settings from \"{settings_file}\"")
diff --git a/CMD_FLAGS.txt b/user_data/CMD_FLAGS.txt
similarity index 100%
rename from CMD_FLAGS.txt
rename to user_data/CMD_FLAGS.txt
diff --git a/characters/Assistant.yaml b/user_data/characters/Assistant.yaml
similarity index 100%
rename from characters/Assistant.yaml
rename to user_data/characters/Assistant.yaml
diff --git a/characters/Example.png b/user_data/characters/Example.png
similarity index 100%
rename from characters/Example.png
rename to user_data/characters/Example.png
diff --git a/characters/Example.yaml b/user_data/characters/Example.yaml
similarity index 100%
rename from characters/Example.yaml
rename to user_data/characters/Example.yaml
diff --git a/grammars/arithmetic.gbnf b/user_data/grammars/arithmetic.gbnf
similarity index 100%
rename from grammars/arithmetic.gbnf
rename to user_data/grammars/arithmetic.gbnf
diff --git a/grammars/c.gbnf b/user_data/grammars/c.gbnf
similarity index 100%
rename from grammars/c.gbnf
rename to user_data/grammars/c.gbnf
diff --git a/grammars/chess.gbnf b/user_data/grammars/chess.gbnf
similarity index 100%
rename from grammars/chess.gbnf
rename to user_data/grammars/chess.gbnf
diff --git a/grammars/json.gbnf b/user_data/grammars/json.gbnf
similarity index 100%
rename from grammars/json.gbnf
rename to user_data/grammars/json.gbnf
diff --git a/grammars/json_w_trailing_space.gbnf b/user_data/grammars/json_w_trailing_space.gbnf
similarity index 100%
rename from grammars/json_w_trailing_space.gbnf
rename to user_data/grammars/json_w_trailing_space.gbnf
diff --git a/grammars/list.gbnf b/user_data/grammars/list.gbnf
similarity index 100%
rename from grammars/list.gbnf
rename to user_data/grammars/list.gbnf
diff --git a/grammars/roleplay.gbnf b/user_data/grammars/roleplay.gbnf
similarity index 100%
rename from grammars/roleplay.gbnf
rename to user_data/grammars/roleplay.gbnf
diff --git a/grammars/simple_arithmetic.gbnf b/user_data/grammars/simple_arithmetic.gbnf
similarity index 100%
rename from grammars/simple_arithmetic.gbnf
rename to user_data/grammars/simple_arithmetic.gbnf
diff --git a/instruction-templates/Airoboros-v1.2.yaml b/user_data/instruction-templates/Airoboros-v1.2.yaml
similarity index 100%
rename from instruction-templates/Airoboros-v1.2.yaml
rename to user_data/instruction-templates/Airoboros-v1.2.yaml
diff --git a/instruction-templates/Alpaca.yaml b/user_data/instruction-templates/Alpaca.yaml
similarity index 100%
rename from instruction-templates/Alpaca.yaml
rename to user_data/instruction-templates/Alpaca.yaml
diff --git a/instruction-templates/Bactrian.yaml b/user_data/instruction-templates/Bactrian.yaml
similarity index 100%
rename from instruction-templates/Bactrian.yaml
rename to user_data/instruction-templates/Bactrian.yaml
diff --git a/instruction-templates/Baichuan Chat.yaml b/user_data/instruction-templates/Baichuan Chat.yaml
similarity index 100%
rename from instruction-templates/Baichuan Chat.yaml
rename to user_data/instruction-templates/Baichuan Chat.yaml
diff --git a/instruction-templates/Baize.yaml b/user_data/instruction-templates/Baize.yaml
similarity index 100%
rename from instruction-templates/Baize.yaml
rename to user_data/instruction-templates/Baize.yaml
diff --git a/instruction-templates/Bluemoon.yaml b/user_data/instruction-templates/Bluemoon.yaml
similarity index 100%
rename from instruction-templates/Bluemoon.yaml
rename to user_data/instruction-templates/Bluemoon.yaml
diff --git a/instruction-templates/ChatGLM.yaml b/user_data/instruction-templates/ChatGLM.yaml
similarity index 100%
rename from instruction-templates/ChatGLM.yaml
rename to user_data/instruction-templates/ChatGLM.yaml
diff --git a/instruction-templates/ChatML.yaml b/user_data/instruction-templates/ChatML.yaml
similarity index 100%
rename from instruction-templates/ChatML.yaml
rename to user_data/instruction-templates/ChatML.yaml
diff --git a/instruction-templates/Chinese-Vicuna-Chat.yaml b/user_data/instruction-templates/Chinese-Vicuna-Chat.yaml
similarity index 100%
rename from instruction-templates/Chinese-Vicuna-Chat.yaml
rename to user_data/instruction-templates/Chinese-Vicuna-Chat.yaml
diff --git a/instruction-templates/Command-R.yaml b/user_data/instruction-templates/Command-R.yaml
similarity index 100%
rename from instruction-templates/Command-R.yaml
rename to user_data/instruction-templates/Command-R.yaml
diff --git a/instruction-templates/Galactica Cite.yaml b/user_data/instruction-templates/Galactica Cite.yaml
similarity index 100%
rename from instruction-templates/Galactica Cite.yaml
rename to user_data/instruction-templates/Galactica Cite.yaml
diff --git a/instruction-templates/Galactica Finetuned.yaml b/user_data/instruction-templates/Galactica Finetuned.yaml
similarity index 100%
rename from instruction-templates/Galactica Finetuned.yaml
rename to user_data/instruction-templates/Galactica Finetuned.yaml
diff --git a/instruction-templates/Galactica Q.yaml b/user_data/instruction-templates/Galactica Q.yaml
similarity index 100%
rename from instruction-templates/Galactica Q.yaml
rename to user_data/instruction-templates/Galactica Q.yaml
diff --git a/instruction-templates/Galactica Summary.yaml b/user_data/instruction-templates/Galactica Summary.yaml
similarity index 100%
rename from instruction-templates/Galactica Summary.yaml
rename to user_data/instruction-templates/Galactica Summary.yaml
diff --git a/instruction-templates/Galactica Work.yaml b/user_data/instruction-templates/Galactica Work.yaml
similarity index 100%
rename from instruction-templates/Galactica Work.yaml
rename to user_data/instruction-templates/Galactica Work.yaml
diff --git a/instruction-templates/Galactica v2.yaml b/user_data/instruction-templates/Galactica v2.yaml
similarity index 100%
rename from instruction-templates/Galactica v2.yaml
rename to user_data/instruction-templates/Galactica v2.yaml
diff --git a/instruction-templates/Galactica.yaml b/user_data/instruction-templates/Galactica.yaml
similarity index 100%
rename from instruction-templates/Galactica.yaml
rename to user_data/instruction-templates/Galactica.yaml
diff --git a/instruction-templates/Gorilla.yaml b/user_data/instruction-templates/Gorilla.yaml
similarity index 100%
rename from instruction-templates/Gorilla.yaml
rename to user_data/instruction-templates/Gorilla.yaml
diff --git a/instruction-templates/Guanaco non-chat.yaml b/user_data/instruction-templates/Guanaco non-chat.yaml
similarity index 100%
rename from instruction-templates/Guanaco non-chat.yaml
rename to user_data/instruction-templates/Guanaco non-chat.yaml
diff --git a/instruction-templates/Guanaco-QLoRA.yaml b/user_data/instruction-templates/Guanaco-QLoRA.yaml
similarity index 100%
rename from instruction-templates/Guanaco-QLoRA.yaml
rename to user_data/instruction-templates/Guanaco-QLoRA.yaml
diff --git a/instruction-templates/H2O-prompt_answer.yaml b/user_data/instruction-templates/H2O-prompt_answer.yaml
similarity index 100%
rename from instruction-templates/H2O-prompt_answer.yaml
rename to user_data/instruction-templates/H2O-prompt_answer.yaml
diff --git a/instruction-templates/Hippogriff.yaml b/user_data/instruction-templates/Hippogriff.yaml
similarity index 100%
rename from instruction-templates/Hippogriff.yaml
rename to user_data/instruction-templates/Hippogriff.yaml
diff --git a/instruction-templates/INCITE-Chat.yaml b/user_data/instruction-templates/INCITE-Chat.yaml
similarity index 100%
rename from instruction-templates/INCITE-Chat.yaml
rename to user_data/instruction-templates/INCITE-Chat.yaml
diff --git a/instruction-templates/INCITE-Instruct.yaml b/user_data/instruction-templates/INCITE-Instruct.yaml
similarity index 100%
rename from instruction-templates/INCITE-Instruct.yaml
rename to user_data/instruction-templates/INCITE-Instruct.yaml
diff --git a/instruction-templates/KoAlpaca.yaml b/user_data/instruction-templates/KoAlpaca.yaml
similarity index 100%
rename from instruction-templates/KoAlpaca.yaml
rename to user_data/instruction-templates/KoAlpaca.yaml
diff --git a/instruction-templates/Koala.yaml b/user_data/instruction-templates/Koala.yaml
similarity index 100%
rename from instruction-templates/Koala.yaml
rename to user_data/instruction-templates/Koala.yaml
diff --git a/instruction-templates/LLaVA.yaml b/user_data/instruction-templates/LLaVA.yaml
similarity index 100%
rename from instruction-templates/LLaVA.yaml
rename to user_data/instruction-templates/LLaVA.yaml
diff --git a/instruction-templates/Llama-v2.yaml b/user_data/instruction-templates/Llama-v2.yaml
similarity index 100%
rename from instruction-templates/Llama-v2.yaml
rename to user_data/instruction-templates/Llama-v2.yaml
diff --git a/instruction-templates/Llama-v3.yaml b/user_data/instruction-templates/Llama-v3.yaml
similarity index 100%
rename from instruction-templates/Llama-v3.yaml
rename to user_data/instruction-templates/Llama-v3.yaml
diff --git a/instruction-templates/MOSS.yaml b/user_data/instruction-templates/MOSS.yaml
similarity index 100%
rename from instruction-templates/MOSS.yaml
rename to user_data/instruction-templates/MOSS.yaml
diff --git a/instruction-templates/Manticore Chat.yaml b/user_data/instruction-templates/Manticore Chat.yaml
similarity index 100%
rename from instruction-templates/Manticore Chat.yaml
rename to user_data/instruction-templates/Manticore Chat.yaml
diff --git a/instruction-templates/Metharme.yaml b/user_data/instruction-templates/Metharme.yaml
similarity index 100%
rename from instruction-templates/Metharme.yaml
rename to user_data/instruction-templates/Metharme.yaml
diff --git a/instruction-templates/Mistral.yaml b/user_data/instruction-templates/Mistral.yaml
similarity index 100%
rename from instruction-templates/Mistral.yaml
rename to user_data/instruction-templates/Mistral.yaml
diff --git a/instruction-templates/NVIDIA-ChatQA.yaml b/user_data/instruction-templates/NVIDIA-ChatQA.yaml
similarity index 100%
rename from instruction-templates/NVIDIA-ChatQA.yaml
rename to user_data/instruction-templates/NVIDIA-ChatQA.yaml
diff --git a/instruction-templates/NewHope.yaml b/user_data/instruction-templates/NewHope.yaml
similarity index 100%
rename from instruction-templates/NewHope.yaml
rename to user_data/instruction-templates/NewHope.yaml
diff --git a/instruction-templates/Open Assistant.yaml b/user_data/instruction-templates/Open Assistant.yaml
similarity index 100%
rename from instruction-templates/Open Assistant.yaml
rename to user_data/instruction-templates/Open Assistant.yaml
diff --git a/instruction-templates/OpenBuddy.yaml b/user_data/instruction-templates/OpenBuddy.yaml
similarity index 100%
rename from instruction-templates/OpenBuddy.yaml
rename to user_data/instruction-templates/OpenBuddy.yaml
diff --git a/instruction-templates/OpenChat.yaml b/user_data/instruction-templates/OpenChat.yaml
similarity index 100%
rename from instruction-templates/OpenChat.yaml
rename to user_data/instruction-templates/OpenChat.yaml
diff --git a/instruction-templates/OpenOrca-Platypus2.yaml b/user_data/instruction-templates/OpenOrca-Platypus2.yaml
similarity index 100%
rename from instruction-templates/OpenOrca-Platypus2.yaml
rename to user_data/instruction-templates/OpenOrca-Platypus2.yaml
diff --git a/instruction-templates/Orca Mini.yaml b/user_data/instruction-templates/Orca Mini.yaml
similarity index 100%
rename from instruction-templates/Orca Mini.yaml
rename to user_data/instruction-templates/Orca Mini.yaml
diff --git a/instruction-templates/Orca-Vicuna.yaml b/user_data/instruction-templates/Orca-Vicuna.yaml
similarity index 100%
rename from instruction-templates/Orca-Vicuna.yaml
rename to user_data/instruction-templates/Orca-Vicuna.yaml
diff --git a/instruction-templates/RWKV-Raven.yaml b/user_data/instruction-templates/RWKV-Raven.yaml
similarity index 100%
rename from instruction-templates/RWKV-Raven.yaml
rename to user_data/instruction-templates/RWKV-Raven.yaml
diff --git a/instruction-templates/RWKV-World.yaml b/user_data/instruction-templates/RWKV-World.yaml
similarity index 100%
rename from instruction-templates/RWKV-World.yaml
rename to user_data/instruction-templates/RWKV-World.yaml
diff --git a/instruction-templates/Samantha.yaml b/user_data/instruction-templates/Samantha.yaml
similarity index 100%
rename from instruction-templates/Samantha.yaml
rename to user_data/instruction-templates/Samantha.yaml
diff --git a/instruction-templates/StableBeluga2.yaml b/user_data/instruction-templates/StableBeluga2.yaml
similarity index 100%
rename from instruction-templates/StableBeluga2.yaml
rename to user_data/instruction-templates/StableBeluga2.yaml
diff --git a/instruction-templates/StableLM.yaml b/user_data/instruction-templates/StableLM.yaml
similarity index 100%
rename from instruction-templates/StableLM.yaml
rename to user_data/instruction-templates/StableLM.yaml
diff --git a/instruction-templates/StableVicuna.yaml b/user_data/instruction-templates/StableVicuna.yaml
similarity index 100%
rename from instruction-templates/StableVicuna.yaml
rename to user_data/instruction-templates/StableVicuna.yaml
diff --git a/instruction-templates/Starchat-Beta.yaml b/user_data/instruction-templates/Starchat-Beta.yaml
similarity index 100%
rename from instruction-templates/Starchat-Beta.yaml
rename to user_data/instruction-templates/Starchat-Beta.yaml
diff --git a/instruction-templates/Synthia-CoT.yaml b/user_data/instruction-templates/Synthia-CoT.yaml
similarity index 100%
rename from instruction-templates/Synthia-CoT.yaml
rename to user_data/instruction-templates/Synthia-CoT.yaml
diff --git a/instruction-templates/Synthia.yaml b/user_data/instruction-templates/Synthia.yaml
similarity index 100%
rename from instruction-templates/Synthia.yaml
rename to user_data/instruction-templates/Synthia.yaml
diff --git a/instruction-templates/Tulu.yaml b/user_data/instruction-templates/Tulu.yaml
similarity index 100%
rename from instruction-templates/Tulu.yaml
rename to user_data/instruction-templates/Tulu.yaml
diff --git a/instruction-templates/Vicuna-v0.yaml b/user_data/instruction-templates/Vicuna-v0.yaml
similarity index 100%
rename from instruction-templates/Vicuna-v0.yaml
rename to user_data/instruction-templates/Vicuna-v0.yaml
diff --git a/instruction-templates/Vicuna-v1.1.yaml b/user_data/instruction-templates/Vicuna-v1.1.yaml
similarity index 100%
rename from instruction-templates/Vicuna-v1.1.yaml
rename to user_data/instruction-templates/Vicuna-v1.1.yaml
diff --git a/instruction-templates/Vigogne-Chat.yaml b/user_data/instruction-templates/Vigogne-Chat.yaml
similarity index 100%
rename from instruction-templates/Vigogne-Chat.yaml
rename to user_data/instruction-templates/Vigogne-Chat.yaml
diff --git a/instruction-templates/Vigogne-Instruct.yaml b/user_data/instruction-templates/Vigogne-Instruct.yaml
similarity index 100%
rename from instruction-templates/Vigogne-Instruct.yaml
rename to user_data/instruction-templates/Vigogne-Instruct.yaml
diff --git a/instruction-templates/Wizard-Mega ShareGPT.yaml b/user_data/instruction-templates/Wizard-Mega ShareGPT.yaml
similarity index 100%
rename from instruction-templates/Wizard-Mega ShareGPT.yaml
rename to user_data/instruction-templates/Wizard-Mega ShareGPT.yaml
diff --git a/instruction-templates/Wizard-Mega.yaml b/user_data/instruction-templates/Wizard-Mega.yaml
similarity index 100%
rename from instruction-templates/Wizard-Mega.yaml
rename to user_data/instruction-templates/Wizard-Mega.yaml
diff --git a/instruction-templates/Ziya.yaml b/user_data/instruction-templates/Ziya.yaml
similarity index 100%
rename from instruction-templates/Ziya.yaml
rename to user_data/instruction-templates/Ziya.yaml
diff --git a/loras/place-your-loras-here.txt b/user_data/loras/place-your-loras-here.txt
similarity index 100%
rename from loras/place-your-loras-here.txt
rename to user_data/loras/place-your-loras-here.txt
diff --git a/models/config.yaml b/user_data/models/config.yaml
similarity index 100%
rename from models/config.yaml
rename to user_data/models/config.yaml
diff --git a/models/place-your-models-here.txt b/user_data/models/place-your-models-here.txt
similarity index 100%
rename from models/place-your-models-here.txt
rename to user_data/models/place-your-models-here.txt
diff --git a/presets/Contrastive Search.yaml b/user_data/presets/Contrastive Search.yaml
similarity index 100%
rename from presets/Contrastive Search.yaml
rename to user_data/presets/Contrastive Search.yaml
diff --git a/presets/Creative.yaml b/user_data/presets/Creative.yaml
similarity index 100%
rename from presets/Creative.yaml
rename to user_data/presets/Creative.yaml
diff --git a/presets/Deterministic.yaml b/user_data/presets/Deterministic.yaml
similarity index 100%
rename from presets/Deterministic.yaml
rename to user_data/presets/Deterministic.yaml
diff --git a/presets/Instruct.yaml b/user_data/presets/Instruct.yaml
similarity index 100%
rename from presets/Instruct.yaml
rename to user_data/presets/Instruct.yaml
diff --git a/presets/Null preset.yaml b/user_data/presets/Null preset.yaml
similarity index 100%
rename from presets/Null preset.yaml
rename to user_data/presets/Null preset.yaml
diff --git a/presets/min_p.yaml b/user_data/presets/min_p.yaml
similarity index 100%
rename from presets/min_p.yaml
rename to user_data/presets/min_p.yaml
diff --git a/prompts/Alpaca-with-Input.txt b/user_data/prompts/Alpaca-with-Input.txt
similarity index 100%
rename from prompts/Alpaca-with-Input.txt
rename to user_data/prompts/Alpaca-with-Input.txt
diff --git a/prompts/QA.txt b/user_data/prompts/QA.txt
similarity index 100%
rename from prompts/QA.txt
rename to user_data/prompts/QA.txt
diff --git a/training/datasets/put-trainer-datasets-here.txt b/user_data/training/datasets/put-trainer-datasets-here.txt
similarity index 100%
rename from training/datasets/put-trainer-datasets-here.txt
rename to user_data/training/datasets/put-trainer-datasets-here.txt
diff --git a/training/formats/ChatML-format.json b/user_data/training/formats/ChatML-format.json
similarity index 100%
rename from training/formats/ChatML-format.json
rename to user_data/training/formats/ChatML-format.json
diff --git a/training/formats/alpaca-chatbot-format.json b/user_data/training/formats/alpaca-chatbot-format.json
similarity index 100%
rename from training/formats/alpaca-chatbot-format.json
rename to user_data/training/formats/alpaca-chatbot-format.json
diff --git a/training/formats/alpaca-format.json b/user_data/training/formats/alpaca-format.json
similarity index 100%
rename from training/formats/alpaca-format.json
rename to user_data/training/formats/alpaca-format.json
diff --git a/training/formats/llama2-chat-format.json b/user_data/training/formats/llama2-chat-format.json
similarity index 100%
rename from training/formats/llama2-chat-format.json
rename to user_data/training/formats/llama2-chat-format.json
diff --git a/training/formats/vicuna-format.json b/user_data/training/formats/vicuna-format.json
similarity index 100%
rename from training/formats/vicuna-format.json
rename to user_data/training/formats/vicuna-format.json

From 763a7011c05567bcde6c30f637a22dd826bacb82 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 04:57:36 -0700
Subject: [PATCH 0377/1701] Remove an ancient/obsolete migration check

---
 modules/one_click_installer_check.py | 9 ---------
 server.py                            | 1 -
 2 files changed, 10 deletions(-)
 delete mode 100644 modules/one_click_installer_check.py

diff --git a/modules/one_click_installer_check.py b/modules/one_click_installer_check.py
deleted file mode 100644
index 4bde860094..0000000000
--- a/modules/one_click_installer_check.py
+++ /dev/null
@@ -1,9 +0,0 @@
-from pathlib import Path
-
-from modules.logging_colors import logger
-
-if Path('../webui.py').exists():
-    logger.warning('\nIt looks like you are running an outdated version of '
-                   'the one-click-installers.\n'
-                   'Please migrate your installation following the instructions here:\n'
-                   'https://github.com/oobabooga/text-generation-webui/wiki/Migrating-an-old-one%E2%80%90click-install')
diff --git a/server.py b/server.py
index 01d40ac408..169578a561 100644
--- a/server.py
+++ b/server.py
@@ -1,7 +1,6 @@
 import os
 import warnings
 
-import modules.one_click_installer_check
 from modules import shared
 from modules.block_requests import OpenMonkeyPatch, RequestBlocker
 from modules.logging_colors import logger

From b976112539b586fed6334a54b2a261a7c3569622 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 05:00:37 -0700
Subject: [PATCH 0378/1701] Remove the WSL installation scripts

They were useful in 2023 but now everything runs natively on Windows.
---
 .../workflows/build-portable-release-cuda.yml |   2 +-
 .../build-portable-release-vulkan.yml         |   2 +-
 .github/workflows/build-portable-release.yml  |   2 +-
 cmd_wsl.bat                                   |  11 --
 docs/10 - WSL.md                              | 146 ------------------
 start_wsl.bat                                 |  11 --
 update_wizard_wsl.bat                         |  11 --
 wsl.sh                                        | 115 --------------
 8 files changed, 3 insertions(+), 297 deletions(-)
 delete mode 100755 cmd_wsl.bat
 delete mode 100644 docs/10 - WSL.md
 delete mode 100755 start_wsl.bat
 delete mode 100755 update_wizard_wsl.bat
 delete mode 100755 wsl.sh

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index fcc744084d..aacd59f91f 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -101,7 +101,7 @@ jobs:
       - name: Build Package
         shell: bash
         run: |
-            rm -rf .git cmd* update_wizard* start_wsl.bat wsl.sh Colab-TextGen-GPU.ipynb docker
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker
 
             # Define common variables
             CUDA_VERSION="${{ matrix.cuda }}"
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index d5aa764cb9..6f1e5ec887 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -100,7 +100,7 @@ jobs:
       - name: Build Package
         shell: bash
         run: |
-            rm -rf .git cmd* update_wizard* start_wsl.bat wsl.sh Colab-TextGen-GPU.ipynb docker
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker
 
             # Define common variables
             AVX_SUPPORT="${{ matrix.avx }}"
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 2424cc4454..af8866523f 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -100,7 +100,7 @@ jobs:
       - name: Build Package
         shell: bash
         run: |
-            rm -rf .git cmd* update_wizard* start_wsl.bat wsl.sh Colab-TextGen-GPU.ipynb docker
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker
 
             # Define common variables
             AVX_SUPPORT="${{ matrix.avx }}"
diff --git a/cmd_wsl.bat b/cmd_wsl.bat
deleted file mode 100755
index f9f4348a46..0000000000
--- a/cmd_wsl.bat
+++ /dev/null
@@ -1,11 +0,0 @@
-@echo off
-
-cd /D "%~dp0"
-
-set PATH=%PATH%;%SystemRoot%\system32
-
-@rem sed -i 's/\x0D$//' ./wsl.sh converts newlines to unix format in the wsl script
-call wsl -e bash -lic "sed -i 's/\x0D$//' ./wsl.sh; source ./wsl.sh cmd"
-
-:end
-pause
diff --git a/docs/10 - WSL.md b/docs/10 - WSL.md
deleted file mode 100644
index e0d66393cb..0000000000
--- a/docs/10 - WSL.md	
+++ /dev/null
@@ -1,146 +0,0 @@
-## WSL instructions
-
-If you do not have WSL installed, follow the [instructions below](https://github.com/oobabooga/text-generation-webui/wiki/10-%E2%80%90-WSL#wsl-installation) first.
-
-### Additional WSL setup info
-
-If you want to install Linux to a drive other than C, open powershell and enter these commands:
-
-```
-cd D:\Path\To\Linux
-$ProgressPreference = 'SilentlyContinue'
-Invoke-WebRequest -Uri  -OutFile Linux.appx -UseBasicParsing
-mv Linux.appx Linux.zip
-```
-
-Then open Linux.zip and you should see several .appx files inside.
-
-The one with _x64.appx contains the exe installer that you need.
-
-Extract the contents of that _x64.appx file and run .exe to install.
-
-Linux Distro URLs: https://learn.microsoft.com/en-us/windows/wsl/install-manual#downloading-distributions
-
-**ENSURE THAT THE WSL LINUX DISTRO THAT YOU WISH TO USE IS SET AS THE DEFAULT!**
-
-Do this by using these commands:
-
-```
-wsl -l
-wsl -s 
-```
-
-### Web UI Installation
-
-Run the "start" script. By default it will install the web UI in WSL:
-/home/{username}/text-gen-install
-
-To launch the web UI in the future after it is already installed, run
-the same "start" script. Ensure that one_click.py and wsl.sh are next to it!
-
-### Updating the web UI
-
-As an alternative to running the "update" script, you can also run "wsl.sh update" in WSL.
-
-### Running an interactive shell
-
-As an alternative to running the "cmd" script, you can also run "wsl.sh cmd" in WSL.
-
-### Changing the default install location
-
-To change this, you will need to edit the scripts as follows:
-wsl.sh: line ~22   INSTALL_DIR="/path/to/install/dir"
-
-Keep in mind that there is a long-standing bug in WSL that significantly
-slows drive read/write speeds when using a physical drive as opposed to
-the virtual one that Linux is installed in.
-
-## WSL installation
-
-Guide created by [@jfryton](https://github.com/jfryton). Thank you jfryton.
-
------
-
-Here's an easy-to-follow, step-by-step guide for installing Windows Subsystem for Linux (WSL) with Ubuntu on Windows 10/11:
-
-### Step 1: Enable WSL
-
-1. Press the Windows key + X and click on "Windows PowerShell (Admin)" or "Windows Terminal (Admin)" to open PowerShell or Terminal with administrator privileges.
-2. In the PowerShell window, type the following command and press Enter:
-
-```
-wsl --install
-```
-
-If this command doesn't work, you can enable WSL with the following command for Windows 10:
-
-```
-wsl --set-default-version 1
-```
-
-For Windows 11, you can use:
-
-```
-wsl --set-default-version 2
-```
-
-You may be prompted to restart your computer. If so, save your work and restart.
-
-### Step 2: Install Ubuntu
-
-1. Open the Microsoft Store.
-2. Search for "Ubuntu" in the search bar.
-3. Choose the desired Ubuntu version (e.g., Ubuntu 20.04 LTS) and click "Get" or "Install" to download and install the Ubuntu app.
-4. Once the installation is complete, click "Launch" or search for "Ubuntu" in the Start menu and open the app.
-
-### Step 3: Set up Ubuntu
-
-1. When you first launch the Ubuntu app, it will take a few minutes to set up. Be patient as it installs the necessary files and sets up your environment.
-2. Once the setup is complete, you will be prompted to create a new UNIX username and password. Choose a username and password, and make sure to remember them, as you will need them for future administrative tasks within the Ubuntu environment.
-
-### Step 4: Update and upgrade packages
-
-1. After setting up your username and password, it's a good idea to update and upgrade your Ubuntu system. Run the following commands in the Ubuntu terminal:
-
-```
-sudo apt update
-sudo apt upgrade
-```
-
-2. Enter your password when prompted. This will update the package list and upgrade any outdated packages.
-
-Congratulations! You have now installed WSL with Ubuntu on your Windows 10/11 system. You can use the Ubuntu terminal for various tasks, like running Linux commands, installing packages, or managing files.
-
-You can launch your WSL Ubuntu installation by selecting the Ubuntu app (like any other program installed on your computer) or typing 'ubuntu' into Powershell or Terminal.
-
-### Step 5: Proceed with Linux instructions
-
-1. You can now follow the Linux setup instructions. If you receive any error messages about a missing tool or package, just install them using apt:
-
-```
-sudo apt install [missing package]
-```
-
-You will probably need to install build-essential
-
-```
-sudo apt install build-essential
-```
-
-If you face any issues or need to troubleshoot, you can always refer to the official Microsoft documentation for WSL: https://docs.microsoft.com/en-us/windows/wsl/
-
-### WSL2 performance using /mnt: 
-
-When you git clone a repository, put it inside WSL and not outside. To understand more, take a look at this [issue](https://github.com/microsoft/WSL/issues/4197#issuecomment-604592340)
-
-### Bonus: Port Forwarding
-
-By default, you won't be able to access the webui from another device on your local network. You will need to setup the appropriate port forwarding using the following steps:
-
-1. First, get the IP address of the WSL by typing `wsl hostname -I`. This will output the IP address, for example `172.20.134.111`.
-2. Then, use the following command (using PowerShell or Terminal with administrator privileges) to set up port forwarding, replacing `172.20.134.111` with the IP address you obtained in step 1:
-
-```
-netsh interface portproxy add v4tov4 listenaddress=0.0.0.0 listenport=7860 connectaddress=172.20.134.111 connectport=7860
-```
-
diff --git a/start_wsl.bat b/start_wsl.bat
deleted file mode 100755
index d7bacead6b..0000000000
--- a/start_wsl.bat
+++ /dev/null
@@ -1,11 +0,0 @@
-@echo off
-
-cd /D "%~dp0"
-
-set PATH=%PATH%;%SystemRoot%\system32
-
-@rem sed -i 's/\x0D$//' ./wsl.sh converts newlines to unix format in the wsl script
-call wsl -e bash -lic "sed -i 's/\x0D$//' ./wsl.sh; source ./wsl.sh %*"
-
-:end
-pause
diff --git a/update_wizard_wsl.bat b/update_wizard_wsl.bat
deleted file mode 100755
index 35f0a349b3..0000000000
--- a/update_wizard_wsl.bat
+++ /dev/null
@@ -1,11 +0,0 @@
-@echo off
-
-cd /D "%~dp0"
-
-set PATH=%PATH%;%SystemRoot%\system32
-
-@rem sed -i 's/\x0D$//' ./wsl.sh converts newlines to unix format in the wsl script   calling wsl.sh with 'update' will run updater
-call wsl -e bash -lic "sed -i 's/\x0D$//' ./wsl.sh; source ./wsl.sh update-wizard"
-
-:end
-pause
diff --git a/wsl.sh b/wsl.sh
deleted file mode 100755
index c5d28b1694..0000000000
--- a/wsl.sh
+++ /dev/null
@@ -1,115 +0,0 @@
-#!/bin/bash
-
-# detect if build-essential is missing or broken
-if ! dpkg-query -W -f'${Status}' "build-essential" 2>/dev/null | grep -q "ok installed"; then
-echo "build-essential not found or broken!
-
-A C++ compiler is required to build needed Python packages!
-To install one, run cmd_wsl.bat and enter these commands:
-
-sudo apt-get update
-sudo apt-get install build-essential
-"
-read -n1 -p "Continue the installer anyway? [y,n]" EXIT_PROMPT
-# only continue if user inputs 'y' else exit
-if ! [[ $EXIT_PROMPT == "Y" || $EXIT_PROMPT == "y" ]]; then exit; fi
-fi
-
-# deactivate existing conda envs as needed to avoid conflicts
-{ conda deactivate && conda deactivate && conda deactivate; } 2> /dev/null
-
-# config   unlike other scripts, can't use current directory due to file IO bug in WSL, needs to be in virtual drive
-INSTALL_DIR_PREFIX="$HOME/text-gen-install"
-if [[ ! $(realpath "$(pwd)/..") = /mnt/* ]]; then
-    INSTALL_DIR_PREFIX="$(realpath "$(pwd)/..")" && INSTALL_INPLACE=1
-fi
-INSTALL_DIR="$INSTALL_DIR_PREFIX/text-generation-webui"
-CONDA_ROOT_PREFIX="$INSTALL_DIR/installer_files/conda"
-INSTALL_ENV_DIR="$INSTALL_DIR/installer_files/env"
-MINICONDA_DOWNLOAD_URL="https://repo.anaconda.com/miniconda/Miniconda3-py311_24.11.1-0-Linux-x86_64.sh"
-conda_exists="F"
-
-# environment isolation
-export PYTHONNOUSERSITE=1
-unset PYTHONPATH
-unset PYTHONHOME
-export CUDA_PATH="$INSTALL_ENV_DIR"
-export CUDA_HOME="$CUDA_PATH"
-
-# /usr/lib/wsl/lib needs to be added to LD_LIBRARY_PATH to fix years-old bug in WSL where GPU drivers aren't linked properly
-export LD_LIBRARY_PATH="$CUDA_HOME/lib:/usr/lib/wsl/lib:$LD_LIBRARY_PATH"
-
-# open bash cli if called with 'wsl.sh cmd' with workarounds for existing conda
-if [ "$1" == "cmd" ]; then
-    exec bash --init-file <(echo ". ~/.bashrc; conda deactivate 2> /dev/null; cd $INSTALL_DIR || cd $HOME; source $CONDA_ROOT_PREFIX/etc/profile.d/conda.sh; conda activate $INSTALL_ENV_DIR")
-    exit
-fi
-
-if [[ "$INSTALL_DIR" =~ " " ]]; then echo This script relies on Miniconda which can not be silently installed under a path with spaces. && exit; fi
-
-# create install dir if missing
-if [ ! -d "$INSTALL_DIR" ]; then mkdir -p "$INSTALL_DIR" || exit; fi
-
-# figure out whether git and conda needs to be installed
-if "$CONDA_ROOT_PREFIX/bin/conda" --version &>/dev/null; then conda_exists="T"; fi
-
-# (if necessary) install git and conda into a contained environment
-# download miniconda
-if [ "$conda_exists" == "F" ]; then
-    echo "Downloading Miniconda from $MINICONDA_DOWNLOAD_URL to $INSTALL_DIR/miniconda_installer.sh"
-
-    curl -L "$MINICONDA_DOWNLOAD_URL" > "$INSTALL_DIR/miniconda_installer.sh"
-
-    chmod u+x "$INSTALL_DIR/miniconda_installer.sh"
-    bash "$INSTALL_DIR/miniconda_installer.sh" -b -p $CONDA_ROOT_PREFIX
-
-    # test the conda binary
-    echo "Miniconda version:"
-    "$CONDA_ROOT_PREFIX/bin/conda" --version
-
-    # delete the Miniconda installer
-    rm "$INSTALL_DIR/miniconda_installer.sh"
-fi
-
-# create the installer env
-if [ ! -e "$INSTALL_ENV_DIR" ]; then
-    "$CONDA_ROOT_PREFIX/bin/conda" create -y -k --prefix "$INSTALL_ENV_DIR" python=3.11 git
-fi
-
-# check if conda environment was actually created
-if [ ! -e "$INSTALL_ENV_DIR/bin/python" ]; then
-    echo "Conda environment is empty."
-    exit
-fi
-
-# activate installer env
-source "$CONDA_ROOT_PREFIX/etc/profile.d/conda.sh" # otherwise conda complains about 'shell not initialized' (needed when running in a script)
-conda activate "$INSTALL_ENV_DIR"
-
-pushd $INSTALL_DIR 1> /dev/null || exit
-
-if [ ! -f "./server.py" ]; then
-    git init -b main
-    git remote add origin https://github.com/oobabooga/text-generation-webui
-    git fetch
-    git remote set-head origin -a
-    git reset origin/HEAD --hard
-    git branch --set-upstream-to=origin/HEAD
-    git restore -- . :!./CMD_FLAGS.txt
-fi
-
-# copy CMD_FLAGS.txt to install dir to allow edits within Windows
-if [[ $INSTALL_INPLACE != 1 ]]; then
-    # workaround for old install migration
-    if [ ! -f "./wsl.sh" ]; then
-        git pull || exit
-        [ -f "../webui.py" ] && mv "../webui.py" "../webui-old.py"
-    fi
-    if [ -f "$(dirs +1)/CMD_FLAGS.txt" ] && [ -f "./CMD_FLAGS.txt" ]; then cp -u "$(dirs +1)/CMD_FLAGS.txt" "$INSTALL_DIR"; fi
-fi
-
-# setup installer env   update env if called with 'wsl.sh update'
-case "$1" in
-("update-wizard") python one_click.py --update-wizard;;
-(*) python one_click.py $@;;
-esac

From 19c8dced672d6f10ee6214abd364716a18e68d0a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 05:03:23 -0700
Subject: [PATCH 0379/1701] Move settings-template.yaml into user_data

---
 settings-template.yaml => user_data/settings-template.yaml | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 rename settings-template.yaml => user_data/settings-template.yaml (100%)

diff --git a/settings-template.yaml b/user_data/settings-template.yaml
similarity index 100%
rename from settings-template.yaml
rename to user_data/settings-template.yaml

From cbd4d967cc6216fc0d465ce2d46be43fe7155cd7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 05:09:52 -0700
Subject: [PATCH 0380/1701] Update a --help message

---
 modules/shared.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/shared.py b/modules/shared.py
index 21e6dd005d..6bcbdd46c5 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -81,7 +81,7 @@
 group.add_argument('--model-dir', type=str, default='user_data/models', help='Path to directory with all the models.')
 group.add_argument('--lora-dir', type=str, default='user_data/loras', help='Path to directory with all the loras.')
 group.add_argument('--model-menu', action='store_true', help='Show a model menu in the terminal when the web UI is first launched.')
-group.add_argument('--settings', type=str, help='Load the default interface settings from this yaml file. See settings-template.yaml for an example. If you create a file called settings.yaml, this file will be loaded by default without the need to use the --settings flag.')
+group.add_argument('--settings', type=str, help='Load the default interface settings from this yaml file. See user_data/settings-template.yaml for an example. If you create a file called user_data/settings.yaml, this file will be loaded by default without the need to use the --settings flag.')
 group.add_argument('--extensions', type=str, nargs='+', help='The list of extensions to load. If you want to load more than one extension, write the names separated by spaces.')
 group.add_argument('--verbose', action='store_true', help='Print the prompts to the terminal.')
 group.add_argument('--idle-timeout', type=int, default=0, help='Unload model after this many minutes of inactivity. It will be automatically reloaded when you try to use it again.')

From 6acb0e1bee1d31dd848056a3a5d0d5ebacc1d1a3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 05:13:08 -0700
Subject: [PATCH 0381/1701] Change a UI description

---
 modules/ui_session.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_session.py b/modules/ui_session.py
index 42434e5166..a3f098215c 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -13,7 +13,7 @@ def create_ui():
                 shared.gradio['reset_interface'] = gr.Button("Apply flags/extensions and restart", interactive=not mu)
                 with gr.Row():
                     shared.gradio['toggle_dark_mode'] = gr.Button('Toggle 💡')
-                    shared.gradio['save_settings'] = gr.Button('Save UI defaults to settings.yaml', interactive=not mu)
+                    shared.gradio['save_settings'] = gr.Button('Save UI defaults to user_data/settings.yaml', interactive=not mu)
 
                 with gr.Row():
                     with gr.Column():

From 3a207e7a57387bd6bc13ed529acb4b72f0de8467 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 07:31:04 -0700
Subject: [PATCH 0382/1701] Improve the --help formatting a bit

---
 README.md         | 181 +++++++++++++++++++++++-----------------------
 modules/shared.py |   4 +-
 2 files changed, 92 insertions(+), 93 deletions(-)

diff --git a/README.md b/README.md
index 58f77786b2..a950af53d0 100644
--- a/README.md
+++ b/README.md
@@ -186,9 +186,9 @@ usage: server.py [-h] [--multi-user] [--character CHARACTER] [--model MODEL] [--
                  [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR]
                  [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--use_flash_attention_2] [--use_eager_attention] [--torch-compile] [--load-in-4bit]
                  [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn] [--threads THREADS] [--threads-batch THREADS_BATCH] [--batch-size BATCH_SIZE] [--no-mmap]
-                 [--mlock] [--n-gpu-layers N_GPU_LAYERS] [--tensor-split TENSOR_SPLIT] [--numa] [--no-kv-offload] [--row-split] [--extra-flags EXTRA_FLAGS] [--streaming-llm] [--ctx-size CTX_SIZE]
+                 [--mlock] [--n-gpu-layers N_GPU_LAYERS] [--tensor-split TENSOR_SPLIT] [--numa] [--no-kv-offload] [--row-split] [--extra-flags EXTRA_FLAGS] [--streaming-llm] [--ctx-size N]
                  [--model-draft MODEL_DRAFT] [--draft-max DRAFT_MAX] [--gpu-layers-draft GPU_LAYERS_DRAFT] [--device-draft DEVICE_DRAFT] [--ctx-size-draft CTX_SIZE_DRAFT] [--gpu-split GPU_SPLIT]
-                 [--autosplit] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa] [--num_experts_per_token NUM_EXPERTS_PER_TOKEN] [--enable_tp] [--hqq-backend HQQ_BACKEND] [--cpp-runner]
+                 [--autosplit] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa] [--num_experts_per_token N] [--enable_tp] [--hqq-backend HQQ_BACKEND] [--cpp-runner]
                  [--cache_type CACHE_TYPE] [--deepspeed] [--nvme-offload-dir NVME_OFFLOAD_DIR] [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE]
                  [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH]
                  [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors] [--api] [--public-api]
@@ -197,125 +197,124 @@ usage: server.py [-h] [--multi-user] [--character CHARACTER] [--model MODEL] [--
 Text generation web UI
 
 options:
-  -h, --help                                           show this help message and exit
+  -h, --help                                show this help message and exit
 
 Basic settings:
-  --multi-user                                         Multi-user mode. Chat histories are not saved or automatically loaded. Warning: this is likely not safe for sharing publicly.
-  --character CHARACTER                                The name of the character to load in chat mode by default.
-  --model MODEL                                        Name of the model to load by default.
-  --lora LORA [LORA ...]                               The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.
-  --model-dir MODEL_DIR                                Path to directory with all the models.
-  --lora-dir LORA_DIR                                  Path to directory with all the loras.
-  --model-menu                                         Show a model menu in the terminal when the web UI is first launched.
-  --settings SETTINGS                                  Load the default interface settings from this yaml file. See settings-template.yaml for an example. If you create a file called settings.yaml,
-                                                       this file will be loaded by default without the need to use the --settings flag.
-  --extensions EXTENSIONS [EXTENSIONS ...]             The list of extensions to load. If you want to load more than one extension, write the names separated by spaces.
-  --verbose                                            Print the prompts to the terminal.
-  --idle-timeout IDLE_TIMEOUT                          Unload model after this many minutes of inactivity. It will be automatically reloaded when you try to use it again.
+  --multi-user                              Multi-user mode. Chat histories are not saved or automatically loaded. Warning: this is likely not safe for sharing publicly.
+  --character CHARACTER                     The name of the character to load in chat mode by default.
+  --model MODEL                             Name of the model to load by default.
+  --lora LORA [LORA ...]                    The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.
+  --model-dir MODEL_DIR                     Path to directory with all the models.
+  --lora-dir LORA_DIR                       Path to directory with all the loras.
+  --model-menu                              Show a model menu in the terminal when the web UI is first launched.
+  --settings SETTINGS                       Load the default interface settings from this yaml file. See user_data/settings-template.yaml for an example. If you create a file called
+                                            user_data/settings.yaml, this file will be loaded by default without the need to use the --settings flag.
+  --extensions EXTENSIONS [EXTENSIONS ...]  The list of extensions to load. If you want to load more than one extension, write the names separated by spaces.
+  --verbose                                 Print the prompts to the terminal.
+  --idle-timeout IDLE_TIMEOUT               Unload model after this many minutes of inactivity. It will be automatically reloaded when you try to use it again.
 
 Model loader:
-  --loader LOADER                                      Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF,
-                                                       ExLlamav2, HQQ, TensorRT-LLM.
+  --loader LOADER                           Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF, ExLlamav2, HQQ,
+                                            TensorRT-LLM.
 
 Transformers/Accelerate:
-  --cpu                                                Use the CPU to generate text. Warning: Training on CPU is extremely slow.
-  --cpu-memory CPU_MEMORY                              Maximum CPU memory in GiB. Use this for CPU offloading.
-  --disk                                               If the model is too large for your GPU(s) and CPU combined, send the remaining layers to the disk.
-  --disk-cache-dir DISK_CACHE_DIR                      Directory to save the disk cache to. Defaults to "user_data/cache".
-  --load-in-8bit                                       Load the model with 8-bit precision (using bitsandbytes).
-  --bf16                                               Load the model with bfloat16 precision. Requires NVIDIA Ampere GPU.
-  --no-cache                                           Set use_cache to False while generating text. This reduces VRAM usage slightly, but it comes at a performance cost.
-  --trust-remote-code                                  Set trust_remote_code=True while loading the model. Necessary for some models.
-  --force-safetensors                                  Set use_safetensors=True while loading the model. This prevents arbitrary code execution.
-  --no_use_fast                                        Set use_fast=False while loading the tokenizer (it's True by default). Use this if you have any problems related to use_fast.
-  --use_flash_attention_2                              Set use_flash_attention_2=True while loading the model.
-  --use_eager_attention                                Set attn_implementation= eager while loading the model.
-  --torch-compile                                      Compile the model with torch.compile for improved performance.
+  --cpu                                     Use the CPU to generate text. Warning: Training on CPU is extremely slow.
+  --cpu-memory CPU_MEMORY                   Maximum CPU memory in GiB. Use this for CPU offloading.
+  --disk                                    If the model is too large for your GPU(s) and CPU combined, send the remaining layers to the disk.
+  --disk-cache-dir DISK_CACHE_DIR           Directory to save the disk cache to. Defaults to "user_data/cache".
+  --load-in-8bit                            Load the model with 8-bit precision (using bitsandbytes).
+  --bf16                                    Load the model with bfloat16 precision. Requires NVIDIA Ampere GPU.
+  --no-cache                                Set use_cache to False while generating text. This reduces VRAM usage slightly, but it comes at a performance cost.
+  --trust-remote-code                       Set trust_remote_code=True while loading the model. Necessary for some models.
+  --force-safetensors                       Set use_safetensors=True while loading the model. This prevents arbitrary code execution.
+  --no_use_fast                             Set use_fast=False while loading the tokenizer (it's True by default). Use this if you have any problems related to use_fast.
+  --use_flash_attention_2                   Set use_flash_attention_2=True while loading the model.
+  --use_eager_attention                     Set attn_implementation= eager while loading the model.
+  --torch-compile                           Compile the model with torch.compile for improved performance.
 
 bitsandbytes 4-bit:
-  --load-in-4bit                                       Load the model with 4-bit precision (using bitsandbytes).
-  --use_double_quant                                   use_double_quant for 4-bit.
-  --compute_dtype COMPUTE_DTYPE                        compute dtype for 4-bit. Valid options: bfloat16, float16, float32.
-  --quant_type QUANT_TYPE                              quant_type for 4-bit. Valid options: nf4, fp4.
+  --load-in-4bit                            Load the model with 4-bit precision (using bitsandbytes).
+  --use_double_quant                        use_double_quant for 4-bit.
+  --compute_dtype COMPUTE_DTYPE             compute dtype for 4-bit. Valid options: bfloat16, float16, float32.
+  --quant_type QUANT_TYPE                   quant_type for 4-bit. Valid options: nf4, fp4.
 
 llama.cpp:
-  --flash-attn                                         Use flash-attention.
-  --threads THREADS                                    Number of threads to use.
-  --threads-batch THREADS_BATCH                        Number of threads to use for batches/prompt processing.
-  --batch-size BATCH_SIZE                              Maximum number of prompt tokens to batch together when calling llama_eval.
-  --no-mmap                                            Prevent mmap from being used.
-  --mlock                                              Force the system to keep the model in RAM.
-  --n-gpu-layers N_GPU_LAYERS                          Number of layers to offload to the GPU.
-  --tensor-split TENSOR_SPLIT                          Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.
-  --numa                                               Activate NUMA task allocation for llama.cpp.
-  --no-kv-offload                                      Do not offload the K, Q, V to the GPU. This saves VRAM but reduces the performance.
-  --row-split                                          Split the model by rows across GPUs. This may improve multi-gpu performance.
-  --extra-flags EXTRA_FLAGS                            Extra flags to pass to llama-server. Format: "flag1=value1;flag2;flag3=value3". Example: "override-tensor=exps=CPU"
-  --streaming-llm                                      Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.
+  --flash-attn                              Use flash-attention.
+  --threads THREADS                         Number of threads to use.
+  --threads-batch THREADS_BATCH             Number of threads to use for batches/prompt processing.
+  --batch-size BATCH_SIZE                   Maximum number of prompt tokens to batch together when calling llama_eval.
+  --no-mmap                                 Prevent mmap from being used.
+  --mlock                                   Force the system to keep the model in RAM.
+  --n-gpu-layers N_GPU_LAYERS               Number of layers to offload to the GPU.
+  --tensor-split TENSOR_SPLIT               Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.
+  --numa                                    Activate NUMA task allocation for llama.cpp.
+  --no-kv-offload                           Do not offload the K, Q, V to the GPU. This saves VRAM but reduces the performance.
+  --row-split                               Split the model by rows across GPUs. This may improve multi-gpu performance.
+  --extra-flags EXTRA_FLAGS                 Extra flags to pass to llama-server. Format: "flag1=value1;flag2;flag3=value3". Example: "override-tensor=exps=CPU"
+  --streaming-llm                           Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.
 
 Context and cache management:
-  --ctx-size CTX_SIZE, --n_ctx CTX_SIZE, --max_seq_len CTX_SIZE
-                                                       Context size in tokens.
+  --ctx-size N, --n_ctx N, --max_seq_len N  Context size in tokens.
 
 Speculative decoding:
-  --model-draft MODEL_DRAFT                            Path to the draft model for speculative decoding.
-  --draft-max DRAFT_MAX                                Number of tokens to draft for speculative decoding.
-  --gpu-layers-draft GPU_LAYERS_DRAFT                  Number of layers to offload to the GPU for the draft model.
-  --device-draft DEVICE_DRAFT                          Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1
-  --ctx-size-draft CTX_SIZE_DRAFT                      Size of the prompt context for the draft model. If 0, uses the same as the main model.
+  --model-draft MODEL_DRAFT                 Path to the draft model for speculative decoding.
+  --draft-max DRAFT_MAX                     Number of tokens to draft for speculative decoding.
+  --gpu-layers-draft GPU_LAYERS_DRAFT       Number of layers to offload to the GPU for the draft model.
+  --device-draft DEVICE_DRAFT               Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1
+  --ctx-size-draft CTX_SIZE_DRAFT           Size of the prompt context for the draft model. If 0, uses the same as the main model.
 
 ExLlamaV2:
-  --gpu-split GPU_SPLIT                                Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.
-  --autosplit                                          Autosplit the model tensors across the available GPUs. This causes --gpu-split to be ignored.
-  --cfg-cache                                          ExLlamav2_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.
-  --no_flash_attn                                      Force flash-attention to not be used.
-  --no_xformers                                        Force xformers to not be used.
-  --no_sdpa                                            Force Torch SDPA to not be used.
-  --num_experts_per_token NUM_EXPERTS_PER_TOKEN        Number of experts to use for generation. Applies to MoE models like Mixtral.
-  --enable_tp                                          Enable Tensor Parallelism (TP) in ExLlamaV2.
+  --gpu-split GPU_SPLIT                     Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.
+  --autosplit                               Autosplit the model tensors across the available GPUs. This causes --gpu-split to be ignored.
+  --cfg-cache                               ExLlamav2_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.
+  --no_flash_attn                           Force flash-attention to not be used.
+  --no_xformers                             Force xformers to not be used.
+  --no_sdpa                                 Force Torch SDPA to not be used.
+  --num_experts_per_token N                 Number of experts to use for generation. Applies to MoE models like Mixtral.
+  --enable_tp                               Enable Tensor Parallelism (TP) in ExLlamaV2.
 
 HQQ:
-  --hqq-backend HQQ_BACKEND                            Backend for the HQQ loader. Valid options: PYTORCH, PYTORCH_COMPILE, ATEN.
+  --hqq-backend HQQ_BACKEND                 Backend for the HQQ loader. Valid options: PYTORCH, PYTORCH_COMPILE, ATEN.
 
 TensorRT-LLM:
-  --cpp-runner                                         Use the ModelRunnerCpp runner, which is faster than the default ModelRunner but doesn't support streaming yet.
+  --cpp-runner                              Use the ModelRunnerCpp runner, which is faster than the default ModelRunner but doesn't support streaming yet.
 
 Cache:
-  --cache_type CACHE_TYPE                              KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.
+  --cache_type CACHE_TYPE                   KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.
 
 DeepSpeed:
-  --deepspeed                                          Enable the use of DeepSpeed ZeRO-3 for inference via the Transformers integration.
-  --nvme-offload-dir NVME_OFFLOAD_DIR                  DeepSpeed: Directory to use for ZeRO-3 NVME offloading.
-  --local_rank LOCAL_RANK                              DeepSpeed: Optional argument for distributed setups.
+  --deepspeed                               Enable the use of DeepSpeed ZeRO-3 for inference via the Transformers integration.
+  --nvme-offload-dir NVME_OFFLOAD_DIR       DeepSpeed: Directory to use for ZeRO-3 NVME offloading.
+  --local_rank LOCAL_RANK                   DeepSpeed: Optional argument for distributed setups.
 
 RoPE:
-  --alpha_value ALPHA_VALUE                            Positional embeddings alpha factor for NTK RoPE scaling. Use either this or compress_pos_emb, not both.
-  --rope_freq_base ROPE_FREQ_BASE                      If greater than 0, will be used instead of alpha_value. Those two are related by rope_freq_base = 10000 * alpha_value ^ (64 / 63).
-  --compress_pos_emb COMPRESS_POS_EMB                  Positional embeddings compression factor. Should be set to (context length) / (model's original context length). Equal to 1/rope_freq_scale.
+  --alpha_value ALPHA_VALUE                 Positional embeddings alpha factor for NTK RoPE scaling. Use either this or compress_pos_emb, not both.
+  --rope_freq_base ROPE_FREQ_BASE           If greater than 0, will be used instead of alpha_value. Those two are related by rope_freq_base = 10000 * alpha_value ^ (64 / 63).
+  --compress_pos_emb COMPRESS_POS_EMB       Positional embeddings compression factor. Should be set to (context length) / (model's original context length). Equal to 1/rope_freq_scale.
 
 Gradio:
-  --listen                                             Make the web UI reachable from your local network.
-  --listen-port LISTEN_PORT                            The listening port that the server will use.
-  --listen-host LISTEN_HOST                            The hostname that the server will use.
-  --share                                              Create a public URL. This is useful for running the web UI on Google Colab or similar.
-  --auto-launch                                        Open the web UI in the default browser upon launch.
-  --gradio-auth GRADIO_AUTH                            Set Gradio authentication password in the format "username:password". Multiple credentials can also be supplied with "u1:p1,u2:p2,u3:p3".
-  --gradio-auth-path GRADIO_AUTH_PATH                  Set the Gradio authentication file path. The file should contain one or more user:password pairs in the same format as above.
-  --ssl-keyfile SSL_KEYFILE                            The path to the SSL certificate key file.
-  --ssl-certfile SSL_CERTFILE                          The path to the SSL certificate cert file.
-  --subpath SUBPATH                                    Customize the subpath for gradio, use with reverse proxy
-  --old-colors                                         Use the legacy Gradio colors, before the December/2024 update.
+  --listen                                  Make the web UI reachable from your local network.
+  --listen-port LISTEN_PORT                 The listening port that the server will use.
+  --listen-host LISTEN_HOST                 The hostname that the server will use.
+  --share                                   Create a public URL. This is useful for running the web UI on Google Colab or similar.
+  --auto-launch                             Open the web UI in the default browser upon launch.
+  --gradio-auth GRADIO_AUTH                 Set Gradio authentication password in the format "username:password". Multiple credentials can also be supplied with "u1:p1,u2:p2,u3:p3".
+  --gradio-auth-path GRADIO_AUTH_PATH       Set the Gradio authentication file path. The file should contain one or more user:password pairs in the same format as above.
+  --ssl-keyfile SSL_KEYFILE                 The path to the SSL certificate key file.
+  --ssl-certfile SSL_CERTFILE               The path to the SSL certificate cert file.
+  --subpath SUBPATH                         Customize the subpath for gradio, use with reverse proxy
+  --old-colors                              Use the legacy Gradio colors, before the December/2024 update.
 
 API:
-  --api                                                Enable the API extension.
-  --public-api                                         Create a public URL for the API using Cloudfare.
-  --public-api-id PUBLIC_API_ID                        Tunnel ID for named Cloudflare Tunnel. Use together with public-api option.
-  --api-port API_PORT                                  The listening port for the API.
-  --api-key API_KEY                                    API authentication key.
-  --admin-key ADMIN_KEY                                API authentication key for admin tasks like loading and unloading models. If not set, will be the same as --api-key.
-  --api-enable-ipv6                                    Enable IPv6 for the API
-  --api-disable-ipv4                                   Disable IPv4 for the API
-  --nowebui                                            Do not launch the Gradio UI. Useful for launching the API in standalone mode.
+  --api                                     Enable the API extension.
+  --public-api                              Create a public URL for the API using Cloudfare.
+  --public-api-id PUBLIC_API_ID             Tunnel ID for named Cloudflare Tunnel. Use together with public-api option.
+  --api-port API_PORT                       The listening port for the API.
+  --api-key API_KEY                         API authentication key.
+  --admin-key ADMIN_KEY                     API authentication key for admin tasks like loading and unloading models. If not set, will be the same as --api-key.
+  --api-enable-ipv6                         Enable IPv6 for the API
+  --api-disable-ipv4                        Disable IPv4 for the API
+  --nowebui                                 Do not launch the Gradio UI. Useful for launching the API in standalone mode.
 ```
 
 
diff --git a/modules/shared.py b/modules/shared.py
index 6bcbdd46c5..63bdb53691 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -131,7 +131,7 @@
 
 # Cache
 group = parser.add_argument_group('Context and cache management')
-group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=8192, help='Context size in tokens.')
+group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=8192, metavar='N', help='Context size in tokens.')
 group.add_argument('--cache_type', type=str, default='fp16', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).')
 
 # Speculative decoding
@@ -150,7 +150,7 @@
 group.add_argument('--no_flash_attn', action='store_true', help='Force flash-attention to not be used.')
 group.add_argument('--no_xformers', action='store_true', help='Force xformers to not be used.')
 group.add_argument('--no_sdpa', action='store_true', help='Force Torch SDPA to not be used.')
-group.add_argument('--num_experts_per_token', type=int, default=2, help='Number of experts to use for generation. Applies to MoE models like Mixtral.')
+group.add_argument('--num_experts_per_token', type=int, default=2, metavar='N', help='Number of experts to use for generation. Applies to MoE models like Mixtral.')
 group.add_argument('--enable_tp', action='store_true', help='Enable Tensor Parallelism (TP) in ExLlamaV2.')
 
 # HQQ

From d1e7d9c5d5e083f7fd3c4ce4915a9d995bdc182a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 09:00:56 -0700
Subject: [PATCH 0383/1701] Update CMD_FLAGS.txt

---
 user_data/CMD_FLAGS.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/user_data/CMD_FLAGS.txt b/user_data/CMD_FLAGS.txt
index c2d63d9e8f..b0f667b01f 100644
--- a/user_data/CMD_FLAGS.txt
+++ b/user_data/CMD_FLAGS.txt
@@ -1,3 +1,3 @@
-# Only used by the one-click installer.
+# Add persistent flags here to use every time you launch the web UI.
 # Example:
 # --listen --api

From a317450dfa62df456c5bae2463954a3e77dd445b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 14:59:29 -0700
Subject: [PATCH 0384/1701] Update README

---
 README.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index a950af53d0..2f92ed06d4 100644
--- a/README.md
+++ b/README.md
@@ -43,7 +43,7 @@ Download from: https://github.com/oobabooga/text-generation-webui/releases
 
 To restart the web UI later, just run the same `start_` script. If you need to reinstall, delete the `installer_files` folder created during setup and run the script again.
 
-You can use command-line flags, like `./start_linux.sh --help`, or add them to `CMD_FLAGS.txt` (such as `--api` to enable API use). To update the project, run `update_wizard_linux.sh`, `update_wizard_windows.bat`, `update_wizard_macos.sh`, or `update_wizard_wsl.bat`.
+You can use command-line flags, like `./start_linux.sh --help`, or add them to `user_data/CMD_FLAGS.txt` (such as `--api` to enable API use). To update the project, run `update_wizard_linux.sh`, `update_wizard_windows.bat`, `update_wizard_macos.sh`, or `update_wizard_wsl.bat`.
 
 
            
 
@@ -157,7 +157,7 @@ mkdir -p logs cache
 #   TORCH_CUDA_ARCH_LIST based on your GPU model
 #   APP_RUNTIME_GID      your host user's group id (run `id -g` in a terminal)
 #   BUILD_EXTENIONS      optionally add comma separated list of extensions to build
-# Edit CMD_FLAGS.txt and add in it the options you want to execute (like --listen --cpu)
+# Edit user_data/CMD_FLAGS.txt and add in it the options you want to execute (like --listen --cpu)
 # 
 docker compose up --build
 ```

From bc55feaf3e528951296e9eaacb66d63e7b14a171 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 15:07:35 -0700
Subject: [PATCH 0385/1701] Improve host header validation in local mode

---
 extensions/openai/script.py | 14 ++++++++++++++
 modules/gradio_hijack.py    | 25 ++++++++++++++++++++++++-
 2 files changed, 38 insertions(+), 1 deletion(-)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index f907cdbb70..0a887de288 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -86,6 +86,20 @@ def verify_admin_key(authorization: str = Header(None)) -> None:
 )
 
 
+@app.middleware("http")
+async def validate_host_header(request: Request, call_next):
+    # Be strict about only approving access to localhost by default
+    if not (shared.args.listen or shared.args.public_api):
+        host = request.headers.get("host", "").split(":")[0]
+        if host not in ["localhost", "127.0.0.1"]:
+            return JSONResponse(
+                status_code=400,
+                content={"detail": "Invalid host header"}
+            )
+
+    return await call_next(request)
+
+
 @app.options("/", dependencies=check_key)
 async def options_route():
     return JSONResponse(content="OK")
diff --git a/modules/gradio_hijack.py b/modules/gradio_hijack.py
index 2ddd983a02..8e3bb0d94f 100644
--- a/modules/gradio_hijack.py
+++ b/modules/gradio_hijack.py
@@ -1,5 +1,6 @@
 '''
-Copied from: https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14184
+Most of the code here was adapted from:
+https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14184
 '''
 
 import inspect
@@ -7,6 +8,28 @@
 from functools import wraps
 
 import gradio as gr
+import gradio.routes
+from starlette.middleware.trustedhost import TrustedHostMiddleware
+
+from modules import shared
+
+orig_create_app = gradio.routes.App.create_app
+
+
+# Be strict about only approving access to localhost by default
+def create_app_with_trustedhost(*args, **kwargs):
+    app = orig_create_app(*args, **kwargs)
+
+    if not (shared.args.listen or shared.args.share):
+        app.add_middleware(
+            TrustedHostMiddleware,
+            allowed_hosts=["localhost", "127.0.0.1"]
+        )
+
+    return app
+
+
+gradio.routes.App.create_app = create_app_with_trustedhost
 
 
 class GradioDeprecationWarning(DeprecationWarning):

From 35717a088c0487b774fdf6550aaca68476ffe135 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 15:42:27 -0700
Subject: [PATCH 0386/1701] API: Add an /v1/internal/health endpoint

---
 extensions/openai/script.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 0a887de288..c2dc337be9 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -250,6 +250,11 @@ async def handle_moderations(request: Request):
     return JSONResponse(response)
 
 
+@app.get("/v1/internal/health", dependencies=check_key)
+async def handle_health_check():
+    return JSONResponse(content={"status": "ok"})
+
+
 @app.post("/v1/internal/encode", response_model=EncodeResponse, dependencies=check_key)
 async def handle_token_encode(request_data: EncodeRequest):
     response = token_encode(request_data.text)

From 029aab64048a6297b1710933db27beee072bc2fa Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 16:38:13 -0700
Subject: [PATCH 0387/1701] Revert "Add `-noavx2` portable builds"

This reverts commit 0dd71e78c965289c5bfccb3cbc36183f04be23c6.
---
 .github/workflows/build-portable-release-cuda.yml   |  8 ++------
 .github/workflows/build-portable-release-vulkan.yml |  8 ++------
 .github/workflows/build-portable-release.yml        | 12 ++++--------
 3 files changed, 8 insertions(+), 20 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index aacd59f91f..fb9e61b008 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -59,7 +59,7 @@ jobs:
           $matrix = @{
               'os' = @('ubuntu-22.04', 'windows-2019')
               'pyver' = @("3.11")
-              'avx' = @("AVX2", "AVX")
+              'avx' = @("AVX2")
               'cuda' = @("11.7", "12.4")
           }
 
@@ -161,11 +161,7 @@ jobs:
             # 6. Create ZIP file
             cd ..
             VERSION_CLEAN="${VERSION#v}"
-            if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
-                ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
-            else
-                ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}-noavx2.zip"
-            fi
+            ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 6f1e5ec887..3724e384e0 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -59,7 +59,7 @@ jobs:
           $matrix = @{
               'os' = @('ubuntu-22.04', 'windows-2019')
               'pyver' = @("3.11")
-              'avx' = @("AVX2", "AVX")
+              'avx' = @("AVX2")
           }
 
           if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})}
@@ -146,11 +146,7 @@ jobs:
             # 6. Create ZIP file
             cd ..
             VERSION_CLEAN="${VERSION#v}"
-            if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
-                ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.zip"
-            else
-                ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan-noavx2.zip"
-            fi
+            ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index af8866523f..bdf96cecb2 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -15,7 +15,7 @@ on:
         type: string
       exclude:
         description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
-        default: 'os:macos-13,avx:AVX;os:macos-14,avx:AVX'
+        default: 'None'
         required: false
         type: string
   workflow_call:
@@ -32,7 +32,7 @@ on:
         type: string
       exclude:
         description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
-        default: 'os:macos-13,avx:AVX;os:macos-14,avx:AVX'
+        default: 'None'
         required: false
         type: string
 
@@ -59,7 +59,7 @@ jobs:
           $matrix = @{
               'os' = @('ubuntu-22.04', 'windows-2019', 'macos-13', 'macos-14')
               'pyver' = @("3.11")
-              'avx' = @("AVX2", "AVX")
+              'avx' = @("AVX2")
           }
 
           if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})}
@@ -171,11 +171,7 @@ jobs:
             # 5. Create ZIP file
             cd ..
             VERSION_CLEAN="${VERSION#v}"
-            if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
-                ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.zip"
-            else
-                ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-noavx2.zip"
-            fi
+            ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then

From bf2aa19b2155b4fc611475eafc0e1abaa213281e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 16:39:22 -0700
Subject: [PATCH 0388/1701] Bump llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 2 +-
 requirements/full/requirements_amd_noavx2.txt          | 2 +-
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 2 +-
 requirements/portable/requirements_amd_noavx2.txt      | 2 +-
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index e30e5de79d..c20c161e73 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 98df8e8aa2..437da5b595 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -29,6 +29,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 9307d73365..b1c879906a 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -29,6 +29,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 414e188554..e62987b061 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 5627b38e5d..f7a9f114be 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -29,8 +29,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 07845af299..b8cd839082 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index ffa2b6eb0d..3b52d59b61 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 646aaaaf33..a04e89792b 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index fcd221e132..e128b285a0 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 3564b3bf91..fd79677c93 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -15,4 +15,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index 98afdc67d2..431b50a16e 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -15,4 +15,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 433c1f17fa..f68de3ebea 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 9d9c6852aa..fa1e90146d 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -15,6 +15,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index dd25a9d596..22b3adf49d 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index dd7f740c27..e266794aff 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 7542e89766..14f6fbbb0a 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 4810bd5009..0a619193c3 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index d1787e68e2..997bc1d9bc 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.5.0/llama_cpp_binaries-0.5.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

From 4ff91b6588deb358a802fb28caee189fa442785f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 17:24:40 -0700
Subject: [PATCH 0389/1701] Better default settings for Speculative Decoding

---
 modules/shared.py        | 3 +--
 modules/ui_model_menu.py | 2 +-
 2 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 63bdb53691..a2ff61e2de 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -14,7 +14,6 @@
 model = None
 tokenizer = None
 model_name = 'None'
-draft_model_name = 'None'
 is_seq2seq = False
 model_dirty_from_training = False
 lora_names = []
@@ -138,7 +137,7 @@
 group = parser.add_argument_group('Speculative decoding')
 group.add_argument('--model-draft', type=str, default=None, help='Path to the draft model for speculative decoding.')
 group.add_argument('--draft-max', type=int, default=4, help='Number of tokens to draft for speculative decoding.')
-group.add_argument('--gpu-layers-draft', type=int, default=0, help='Number of layers to offload to the GPU for the draft model.')
+group.add_argument('--gpu-layers-draft', type=int, default=256, help='Number of layers to offload to the GPU for the draft model.')
 group.add_argument('--device-draft', type=str, default=None, help='Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1')
 group.add_argument('--ctx-size-draft', type=int, default=0, help='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
 
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index dc09c899c5..546200f98c 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -96,7 +96,7 @@ def create_ui():
                             # Speculative decoding
                             with gr.Accordion("Speculative decoding", open=False, elem_classes='tgw-accordion') as shared.gradio['speculative_decoding_accordion']:
                                 with gr.Row():
-                                    shared.gradio['model_draft'] = gr.Dropdown(label="model-draft", choices=utils.get_available_models(), value=lambda: shared.draft_model_name, elem_classes='slim-dropdown', interactive=not mu)
+                                    shared.gradio['model_draft'] = gr.Dropdown(label="model-draft", choices=utils.get_available_models(), value=lambda: shared.args.model_draft, elem_classes='slim-dropdown', interactive=not mu)
                                     ui.create_refresh_button(shared.gradio['model_draft'], lambda: None, lambda: {'choices': utils.get_available_models()}, 'refresh-button', interactive=not mu)
 
                                 shared.gradio['draft_max'] = gr.Number(label="draft-max", precision=0, step=1, value=shared.args.draft_max, info='Number of tokens to draft for speculative decoding.')

From 234aba1c50374dee8d342d4df12e4de9428a544f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 17:33:47 -0700
Subject: [PATCH 0390/1701] llama.cpp: Simplify the prompt processing progress
 indicator

The progress bar was unreliable
---
 modules/llama_cpp_server.py | 23 +++--------------------
 1 file changed, 3 insertions(+), 20 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index fb972a32b6..895f3b1eaa 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -11,7 +11,6 @@
 
 import llama_cpp_binaries
 import requests
-from tqdm import tqdm
 
 from modules import shared
 from modules.logging_colors import logger
@@ -391,31 +390,15 @@ def stop(self):
 
 
 def filter_stderr_with_progress(process_stderr):
-    progress_bar = None
     progress_pattern = re.compile(r'slot update_slots: id.*progress = (\d+\.\d+)')
-
     try:
         for line in iter(process_stderr.readline, ''):
             progress_match = progress_pattern.search(line)
-
             if progress_match:
-                progress = float(progress_match.group(1))
-
-                # Create progress bar on first progress message
-                if progress_bar is None:
-                    progress_bar = tqdm(total=1.0, desc="Prompt Processing", leave=False)
-
-                progress_bar.update(progress - progress_bar.n)
-
-                # Clean up when complete
-                if progress >= 1.0:
-                    progress_bar.close()
-                    progress_bar = None
-
-            if not line.startswith(('srv ', 'slot ')) and 'log_server_r: request: GET /health' not in line:
+                sys.stderr.write(line)
+                sys.stderr.flush()
+            elif not line.startswith(('srv ', 'slot ')) and 'log_server_r: request: GET /health' not in line:
                 sys.stderr.write(line)
                 sys.stderr.flush()
     except (ValueError, IOError):
-        if progress_bar:
-            progress_bar.close()
         pass

From c4afc0421d9141610e059007f41592b2223d9aae Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 17:43:53 -0700
Subject: [PATCH 0391/1701] Fix parsing of --n_ctx and --max_seq_len

---
 modules/shared.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/shared.py b/modules/shared.py
index a2ff61e2de..2a08db2b98 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -130,7 +130,7 @@
 
 # Cache
 group = parser.add_argument_group('Context and cache management')
-group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=8192, metavar='N', help='Context size in tokens.')
+group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=8192, metavar='N', dest='ctx_size', help='Context size in tokens.')
 group.add_argument('--cache_type', type=str, default='fp16', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).')
 
 # Speculative decoding

From 0fe3b033d01efd2dfb62c1057a101192c779956a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 17:52:21 -0700
Subject: [PATCH 0392/1701] Fix parsing of --n_ctx and --max_seq_len (2nd
 attempt)

---
 modules/shared.py | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 2a08db2b98..7a3368f530 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -130,7 +130,7 @@
 
 # Cache
 group = parser.add_argument_group('Context and cache management')
-group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=8192, metavar='N', dest='ctx_size', help='Context size in tokens.')
+group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=8192, metavar='N', help='Context size in tokens.')
 group.add_argument('--cache_type', type=str, default='fp16', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).')
 
 # Speculative decoding
@@ -222,10 +222,19 @@
 
 args = parser.parse_args()
 args_defaults = parser.parse_args([])
+
+# Create a mapping of all argument aliases to their canonical names
+alias_to_dest = {}
+for action in parser._actions:
+    for opt in action.option_strings:
+        alias_to_dest[opt.lstrip('-').replace('-', '_')] = action.dest
+
 provided_arguments = []
 for arg in sys.argv[1:]:
     arg = arg.lstrip('-').replace('-', '_')
-    if hasattr(args, arg):
+    if arg in alias_to_dest:
+        provided_arguments.append(alias_to_dest[arg])
+    elif hasattr(args, arg):
         provided_arguments.append(arg)
 
 

From 4a32e1f80caa26445aded2f7856569c03d31be98 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 18:01:33 -0700
Subject: [PATCH 0393/1701] UI: show draft_max for ExLlamaV2

---
 modules/loaders.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/loaders.py b/modules/loaders.py
index 062e48377c..b8ae82d7f6 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -86,6 +86,7 @@
         'no_sdpa',
         'exllamav2_info',
         'model_draft',
+        'draft_max',
         'ctx_size_draft',
         'speculative_decoding_accordion',
     ],

From 8b83e6f843767d6423a9017e37c4fdbd6c7cb3ac Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 18:14:57 -0700
Subject: [PATCH 0394/1701] Prevent Gradio from saying 'Thank you for being a
 Gradio user!'

---
 modules/gradio_hijack.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/modules/gradio_hijack.py b/modules/gradio_hijack.py
index 8e3bb0d94f..817da40c6c 100644
--- a/modules/gradio_hijack.py
+++ b/modules/gradio_hijack.py
@@ -9,6 +9,7 @@
 
 import gradio as gr
 import gradio.routes
+import gradio.utils
 from starlette.middleware.trustedhost import TrustedHostMiddleware
 
 from modules import shared
@@ -30,6 +31,7 @@ def create_app_with_trustedhost(*args, **kwargs):
 
 
 gradio.routes.App.create_app = create_app_with_trustedhost
+gradio.utils.launch_counter = lambda: None
 
 
 class GradioDeprecationWarning(DeprecationWarning):

From 511eb6aa94965fcfbc3d39ae1c62c7069ba3e4f2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 18:20:00 -0700
Subject: [PATCH 0395/1701] Fix saving settings to settings.yaml

---
 modules/ui_session.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_session.py b/modules/ui_session.py
index a3f098215c..7cf9f6e66e 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -48,7 +48,7 @@ def handle_save_settings(state, preset, extensions, show_controls, theme):
     return [
         contents,
         "settings.yaml",
-        "./user_data",
+        "user_data/",
         gr.update(visible=True)
     ]
 

From 943451284fa7ba28f9768a9efb06e0e5736c8714 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 18:25:06 -0700
Subject: [PATCH 0396/1701] Fix the Notebook tab not loading its default prompt

---
 modules/ui.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui.py b/modules/ui.py
index ef5ed0e653..f137e62d69 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -266,7 +266,7 @@ def apply_interface_values(state, use_persistent=False):
         if 'textbox-default' in state and 'prompt_menu-default' in state:
             state.pop('prompt_menu-default')
 
-        if 'textbox-notebook' and 'prompt_menu-notebook' in state:
+        if 'textbox-notebook' in state and 'prompt_menu-notebook' in state:
             state.pop('prompt_menu-notebook')
 
     elements = list_interface_input_elements()

From 7b80acd524a1a37c007a86bc9404a39a9c188252 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 18:40:03 -0700
Subject: [PATCH 0397/1701] Fix parsing --extra-flags

---
 modules/llama_cpp_server.py | 2 +-
 modules/shared.py           | 2 +-
 modules/ui_model_menu.py    | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 895f3b1eaa..9572d5aad9 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -312,7 +312,7 @@ def _start_server(self):
             elif extra_flags.startswith("'") and extra_flags.endswith("'"):
                 extra_flags = extra_flags[1:-1].strip()
 
-            for flag_item in extra_flags.split(';'):
+            for flag_item in extra_flags.split(','):
                 if '=' in flag_item:
                     flag, value = flag_item.split('=', 1)
                     cmd += [f"--{flag}", value]
diff --git a/modules/shared.py b/modules/shared.py
index 7a3368f530..5d9dd36230 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -125,7 +125,7 @@
 group.add_argument('--numa', action='store_true', help='Activate NUMA task allocation for llama.cpp.')
 group.add_argument('--no-kv-offload', action='store_true', help='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
 group.add_argument('--row-split', action='store_true', help='Split the model by rows across GPUs. This may improve multi-gpu performance.')
-group.add_argument('--extra-flags', type=str, default=None, help='Extra flags to pass to llama-server. Format: "flag1=value1;flag2;flag3=value3". Example: "override-tensor=exps=CPU"')
+group.add_argument('--extra-flags', type=str, default=None, help='Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"')
 group.add_argument('--streaming-llm', action='store_true', help='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
 
 # Cache
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 546200f98c..e3cf2ba633 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -55,7 +55,7 @@ def create_ui():
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
                             shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
-                            shared.gradio['extra_flags'] = gr.Textbox(label='extra-flags', info='Additional flags to pass to llama-server. Format: "flag1=value1;flag2;flag3=value3". Example: "override-tensor=exps=CPU"')
+                            shared.gradio['extra_flags'] = gr.Textbox(label='extra-flags', info='Additional flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"', value=shared.args.extra_flags)
                             shared.gradio['cpu_memory'] = gr.Number(label="Maximum CPU memory in GiB. Use this for CPU offloading.", value=shared.args.cpu_memory)
                             shared.gradio['alpha_value'] = gr.Number(label='alpha_value', value=shared.args.alpha_value, precision=2, info='Positional embeddings alpha factor for NTK RoPE scaling. Recommended values (NTKv1): 1.75 for 1.5x context, 2.5 for 2x context. Use either this or compress_pos_emb, not both.')
                             shared.gradio['rope_freq_base'] = gr.Number(label='rope_freq_base', value=shared.args.rope_freq_base, precision=0, info='Positional embeddings frequency base for NTK RoPE scaling. Related to alpha_value by rope_freq_base = 10000 * alpha_value ^ (64 / 63). 0 = from model.')

From fa861de05b478260dfc129d0a8eacba0a274b5ee Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 18:52:44 -0700
Subject: [PATCH 0398/1701] Fix portable builds with Python 3.12

---
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 2 +-
 requirements/portable/requirements_amd_noavx2.txt      | 2 +-
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 10 files changed, 19 insertions(+), 19 deletions(-)

diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index e128b285a0..5c717343d8 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index fd79677c93..b616193d4d 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -15,4 +15,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index 431b50a16e..de4740c909 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -15,4 +15,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index f68de3ebea..6310327d37 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index fa1e90146d..f69b58e7e0 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -15,6 +15,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 22b3adf49d..dafa6bbe41 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index e266794aff..c02191eb13 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 14f6fbbb0a..456188b4aa 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 0a619193c3..7e733967ad 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 997bc1d9bc..0329a59892 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 363b632a0db8315ddc4223d4989df7757d80d66b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 19:22:36 -0700
Subject: [PATCH 0399/1701] Lint

---
 css/main.css | 70 ++++++++++++++++++++++++++--------------------------
 1 file changed, 35 insertions(+), 35 deletions(-)

diff --git a/css/main.css b/css/main.css
index 9dce4d0e0c..6e329b9a42 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1311,74 +1311,74 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 
 /* Thinking blocks styling */
 .thinking-block {
-  margin-bottom: 12px;
-  border-radius: 8px;
-  border: 1px solid rgba(0, 0, 0, 0.1);
-  background-color: var(--light-theme-gray);
-  overflow: hidden;
+    margin-bottom: 12px;
+    border-radius: 8px;
+    border: 1px solid rgb(0 0 0 / 10%);
+    background-color: var(--light-theme-gray);
+    overflow: hidden;
 }
 
 .dark .thinking-block {
-  background-color: var(--darker-gray);
+    background-color: var(--darker-gray);
 }
 
 .thinking-header {
-  display: flex;
-  align-items: center;
-  padding: 10px 16px;
-  cursor: pointer;
-  user-select: none;
-  font-size: 14px;
-  color: rgba(0, 0, 0, 0.7);
-  transition: background-color 0.2s;
+    display: flex;
+    align-items: center;
+    padding: 10px 16px;
+    cursor: pointer;
+    user-select: none;
+    font-size: 14px;
+    color: rgb(0 0 0 / 70%);
+    transition: background-color 0.2s;
 }
 
 .thinking-header:hover {
-  background-color: rgba(0, 0, 0, 0.03);
+    background-color: rgb(0 0 0 / 3%);
 }
 
 .thinking-header::-webkit-details-marker {
-  display: none;
+    display: none;
 }
 
 .thinking-icon {
-  margin-right: 8px;
-  color: rgba(0, 0, 0, 0.5);
+    margin-right: 8px;
+    color: rgb(0 0 0 / 50%);
 }
 
 .thinking-title {
-  font-weight: 500;
+    font-weight: 500;
 }
 
 .thinking-content {
-  padding: 12px 16px;
-  border-top: 1px solid rgba(0, 0, 0, 0.07);
-  color: rgba(0, 0, 0, 0.7);
-  font-size: 14px;
-  line-height: 1.5;
-  overflow-wrap: break-word;
-  max-height: 300px;
-  overflow-y: scroll;
-  contain: layout;
+    padding: 12px 16px;
+    border-top: 1px solid rgb(0 0 0 / 7%);
+    color: rgb(0 0 0 / 70%);
+    font-size: 14px;
+    line-height: 1.5;
+    overflow-wrap: break-word;
+    max-height: 300px;
+    overflow-y: scroll;
+    contain: layout;
 }
 
 /* Animation for opening thinking blocks */
 @keyframes fadeIn {
-  from { opacity: 0; }
-  to { opacity: 1; }
+    from { opacity: 0; }
+    to { opacity: 1; }
 }
 
 .thinking-block[open] .thinking-content {
-  animation: fadeIn 0.3s ease-out;
+    animation: fadeIn 0.3s ease-out;
 }
 
 /* Additional style for in-progress thinking */
 .thinking-block[data-streaming="true"] .thinking-title {
-  animation: pulse 1.5s infinite;
+    animation: pulse 1.5s infinite;
 }
 
 @keyframes pulse {
-  0% { opacity: 0.6; }
-  50% { opacity: 1; }
-  100% { opacity: 0.6; }
+    0% { opacity: 0.6; }
+    50% { opacity: 1; }
+    100% { opacity: 0.6; }
 }

From 70952553c7f3fd9fdab8e752260d18a2fff6d00e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 19:29:08 -0700
Subject: [PATCH 0400/1701] Lint

---
 modules/exllamav2.py    |  2 +-
 modules/exllamav2_hf.py | 18 +++++++++---------
 2 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/modules/exllamav2.py b/modules/exllamav2.py
index 952b73b876..6bb422eaec 100644
--- a/modules/exllamav2.py
+++ b/modules/exllamav2.py
@@ -3,7 +3,6 @@
 from pathlib import Path
 
 import torch
-
 from exllamav2 import (
     ExLlamaV2,
     ExLlamaV2Cache,
@@ -16,6 +15,7 @@
     ExLlamaV2Tokenizer
 )
 from exllamav2.generator import ExLlamaV2Sampler, ExLlamaV2StreamingGenerator
+
 from modules import shared
 from modules.logging_colors import logger
 from modules.text_generation import get_max_prompt_length
diff --git a/modules/exllamav2_hf.py b/modules/exllamav2_hf.py
index d6c3bf6ec3..eb80194079 100644
--- a/modules/exllamav2_hf.py
+++ b/modules/exllamav2_hf.py
@@ -4,15 +4,6 @@
 from typing import Any, Dict, Optional, Union
 
 import torch
-from torch.nn import CrossEntropyLoss
-from transformers import (
-    GenerationConfig,
-    GenerationMixin,
-    PretrainedConfig,
-    PreTrainedModel
-)
-from transformers.modeling_outputs import CausalLMOutputWithPast
-
 from exllamav2 import (
     ExLlamaV2,
     ExLlamaV2Cache,
@@ -23,6 +14,15 @@
     ExLlamaV2Cache_TP,
     ExLlamaV2Config
 )
+from torch.nn import CrossEntropyLoss
+from transformers import (
+    GenerationConfig,
+    GenerationMixin,
+    PretrainedConfig,
+    PreTrainedModel
+)
+from transformers.modeling_outputs import CausalLMOutputWithPast
+
 from modules import shared
 from modules.logging_colors import logger
 

From 765fea5e36ee8089387c87478b8cedccd9c8b23e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 19:33:46 -0700
Subject: [PATCH 0401/1701] UI: minor style change

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 6e329b9a42..d6e5ac83e3 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1357,7 +1357,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     font-size: 14px;
     line-height: 1.5;
     overflow-wrap: break-word;
-    max-height: 300px;
+    max-height: 250px;
     overflow-y: scroll;
     contain: layout;
 }

From f5b59d2b0b9d5e580a6c44cd269fbfcb2a286949 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 26 Apr 2025 20:11:24 -0700
Subject: [PATCH 0402/1701] Fix the vulkan workflow

---
 .github/workflows/build-portable-release-vulkan.yml | 1 +
 1 file changed, 1 insertion(+)

diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 3724e384e0..8de29791af 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -136,6 +136,7 @@ jobs:
             else
                 BASE_REQ_FILE="requirements/portable/requirements_vulkan_noavx2.txt"
             fi
+            REQ_FILE="$BASE_REQ_FILE"
 
             cd text-generation-webui
 

From 965ca7948fa88a79b8ae8c66d5d2aaf1d41d043e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 27 Apr 2025 07:33:08 -0700
Subject: [PATCH 0403/1701] Update README

---
 README.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index 2f92ed06d4..c8ac8a8685 100644
--- a/README.md
+++ b/README.md
@@ -29,7 +29,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 
 #### Option 1: Portable builds
 
-Compatible with GGUF (llama.cpp) models, just unzip and run, no installation. Available for Windows, Linux, and macOS.
+No installation needed – just unzip and run. Compatible with GGUF (llama.cpp) models on Windows, Linux, and macOS.
 
 Download from: https://github.com/oobabooga/text-generation-webui/releases
 
@@ -38,7 +38,7 @@ Download from: https://github.com/oobabooga/text-generation-webui/releases
 1) Clone or [download the repository](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip).
 2) Run the script that matches your OS: `start_linux.sh`, `start_windows.bat`, `start_macos.sh`, or `start_wsl.bat`.
 3) Select your GPU vendor when asked.
-4) Once the installation ends, browse to `http://localhost:7860`.
+4) After installation completes, browse to `http://localhost:7860`.
 5) Have fun!
 
 To restart the web UI later, just run the same `start_` script. If you need to reinstall, delete the `installer_files` folder created during setup and run the script again.

From ee0592473c16db1917c98a24ec24dce36fd00bed Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 27 Apr 2025 21:04:02 -0700
Subject: [PATCH 0404/1701] Fix ExLlamaV3_HF leaking memory (attempt)

---
 modules/exllamav3_hf.py | 93 ++++++++++++++++++++++++++++-------------
 1 file changed, 63 insertions(+), 30 deletions(-)

diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index f15fc0b228..12b22f646e 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -118,6 +118,9 @@ def __call__(self, *args, **kwargs):
         seq_tensor = torch.tensor(seq)
         reset = True
 
+        # Maximum number of tokens to process in a single forward pass
+        max_chunk_size = 2048
+
         # Make the forward call
         if labels is None:
             if past_seq is not None:
@@ -131,54 +134,84 @@ def __call__(self, *args, **kwargs):
                 if longest_prefix > 0:
                     reset = False
                     current_len = longest_prefix
-                    if len(seq_tensor) - longest_prefix > 1:
+                    remaining_tokens = len(seq_tensor) - longest_prefix - 1
+
+                    if remaining_tokens > 0:
+                        # Process tokens from longest_prefix to second-to-last token
+                        tokens_to_process = seq_tensor[longest_prefix:-1]
+
+                        # Process in chunks if the number of tokens is large
+                        for i in range(0, tokens_to_process.shape[0], max_chunk_size):
+                            chunk = tokens_to_process[i:i + max_chunk_size]
+                            self.ex_model.forward(
+                                input_ids=chunk.view(1, -1),
+                                params={
+                                    "attn_mode": "flash_attn",
+                                    "cache": ex_cache,
+                                    "past_len": longest_prefix + i,
+                                    "batch_shape": (1, self.max_tokens),
+                                    "reconstruct": False  # Force memory-efficient path
+                                }
+                            )
+
+                        current_len = longest_prefix + remaining_tokens
+
+            if reset:
+                if len(seq_tensor) > 1:
+                    # Process all tokens except the last one
+                    tokens_to_process = seq_tensor[:-1]
+
+                    # Process in chunks if the number of tokens is large
+                    current_len = 0
+                    for i in range(0, tokens_to_process.shape[0], max_chunk_size):
+                        chunk = tokens_to_process[i:i + max_chunk_size]
                         self.ex_model.forward(
-                            input_ids=seq_tensor[longest_prefix:-1].view(1, -1),
+                            input_ids=chunk.view(1, -1),
                             params={
                                 "attn_mode": "flash_attn",
                                 "cache": ex_cache,
-                                "past_len": longest_prefix,
-                                "batch_shape": (1, self.max_tokens)
+                                "past_len": current_len,
+                                "batch_shape": (1, self.max_tokens),
+                                "reconstruct": False  # Force memory-efficient path
                             }
                         )
-
-                        current_len = longest_prefix + len(seq_tensor) - longest_prefix - 1
-
-            if reset:
-                if len(seq_tensor) > 1:
-                    self.ex_model.forward(
-                        input_ids=seq_tensor[:-1].view(1, -1),
-                        params={
-                            "attn_mode": "flash_attn",
-                            "cache": ex_cache,
-                            "past_len": 0,
-                            "batch_shape": (1, self.max_tokens)
-                        }
-                    )
-
-                    current_len = len(seq_tensor) - 1
+                        current_len += chunk.shape[0]
                 else:
                     current_len = 0
 
+            # Process the last token and get logits
             logits = self.ex_model.forward(
                 input_ids=seq_tensor[-1:].view(1, -1),
                 params={
                     "attn_mode": "flash_attn",
                     "cache": ex_cache,
                     "past_len": current_len,
-                    "batch_shape": (1, self.max_tokens)
+                    "batch_shape": (1, self.max_tokens),
+                    "reconstruct": False  # Force memory-efficient path
                 }
             ).to(input_ids.device).float()
         else:
-            logits = self.ex_model.forward(
-                input_ids=seq_tensor.view(1, -1),
-                params={
-                    "attn_mode": "flash_attn",
-                    "cache": ex_cache,
-                    "past_len": 0,
-                    "batch_shape": (1, self.max_tokens)
-                }
-            ).float()
+            # When processing with labels, handle as a complete sequence
+            # Process in chunks if the number of tokens is large
+            tokens_to_process = seq_tensor
+            all_logits = None
+
+            for i in range(0, tokens_to_process.shape[0], max_chunk_size):
+                chunk = tokens_to_process[i:i + max_chunk_size]
+                chunk_logits = self.ex_model.forward(
+                    input_ids=chunk.view(1, -1),
+                    params={
+                        "attn_mode": "flash_attn_nc",  # No caching for training
+                        "reconstruct": False  # Force memory-efficient path
+                    }
+                ).float()
+
+                if all_logits is None:
+                    all_logits = chunk_logits
+                else:
+                    all_logits = torch.cat([all_logits, chunk_logits], dim=1)
+
+            logits = all_logits
 
         if is_negative:
             self.past_seq_negative = seq_tensor

From bbcaec75b47b1b2495623ff38181478c8ba659f8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 27 Apr 2025 21:13:16 -0700
Subject: [PATCH 0405/1701] API: Find a new port if the default one is taken
 (closes #6918)

---
 extensions/openai/script.py | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index c2dc337be9..a995da9d45 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -2,6 +2,7 @@
 import json
 import logging
 import os
+import socket
 import traceback
 from collections import deque
 from threading import Thread
@@ -374,9 +375,26 @@ async def handle_unload_loras():
     return JSONResponse(content="OK")
 
 
+def find_available_port(starting_port):
+    """Try the starting port, then find an available one if it's taken."""
+    try:
+        # Try to create a socket with the starting port
+        with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
+            s.bind(('', starting_port))
+            return starting_port
+    except OSError:
+        # Port is already in use, so find a new one
+        with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
+            s.bind(('', 0))  # Bind to port 0 to get an available port
+            new_port = s.getsockname()[1]
+            logger.warning(f"Port {starting_port} is already in use. Using port {new_port} instead.")
+            return new_port
+
+
 def run_server():
     # Parse configuration
     port = int(os.environ.get('OPENEDAI_PORT', shared.args.api_port))
+    port = find_available_port(port)
     ssl_certfile = os.environ.get('OPENEDAI_CERT_PATH', shared.args.ssl_certfile)
     ssl_keyfile = os.environ.get('OPENEDAI_KEY_PATH', shared.args.ssl_keyfile)
 

From c6c2855c80e6cf6ebd55602f6fe43d80aaca7fae Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 27 Apr 2025 21:22:21 -0700
Subject: [PATCH 0406/1701] llama.cpp: Remove the timeout while loading models
 (closes #6907)

---
 modules/llama_cpp_server.py | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 9572d5aad9..4b01c228d0 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -340,9 +340,7 @@ def _start_server(self):
 
         # Wait for server to be healthy
         health_url = f"http://127.0.0.1:{self.port}/health"
-        start_time = time.time()
-        timeout = 3600 * 8  # 8 hours
-        while time.time() - start_time < timeout:
+        while True:
             # Check if process is still alive
             if self.process.poll() is not None:
                 # Process has terminated
@@ -357,8 +355,6 @@ def _start_server(self):
                 pass
 
             time.sleep(1)
-        else:
-            raise TimeoutError(f"Server health check timed out after {timeout} seconds")
 
         # Server is now healthy, get model info
         self._get_vocabulary_size()

From be13f5199be5059d9c17a4008642c66f2700fe69 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 27 Apr 2025 21:40:38 -0700
Subject: [PATCH 0407/1701] UI: Add an info message about how to use
 Speculative Decoding

---
 modules/ui_model_menu.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index e3cf2ba633..c9679fadf2 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -96,7 +96,7 @@ def create_ui():
                             # Speculative decoding
                             with gr.Accordion("Speculative decoding", open=False, elem_classes='tgw-accordion') as shared.gradio['speculative_decoding_accordion']:
                                 with gr.Row():
-                                    shared.gradio['model_draft'] = gr.Dropdown(label="model-draft", choices=utils.get_available_models(), value=lambda: shared.args.model_draft, elem_classes='slim-dropdown', interactive=not mu)
+                                    shared.gradio['model_draft'] = gr.Dropdown(label="model-draft", choices=utils.get_available_models(), value=lambda: shared.args.model_draft, elem_classes='slim-dropdown',     info='Draft model. Speculative decoding only works with models sharing the same vocabulary (e.g., same model family).', interactive=not mu)
                                     ui.create_refresh_button(shared.gradio['model_draft'], lambda: None, lambda: {'choices': utils.get_available_models()}, 'refresh-button', interactive=not mu)
 
                                 shared.gradio['draft_max'] = gr.Number(label="draft-max", precision=0, step=1, value=shared.args.draft_max, info='Number of tokens to draft for speculative decoding.')

From 15a29e99f8501b229f0f975281e4973870e229df Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 27 Apr 2025 21:41:34 -0700
Subject: [PATCH 0408/1701] Lint

---
 modules/ui_model_menu.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index c9679fadf2..e0ae06f951 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -96,7 +96,7 @@ def create_ui():
                             # Speculative decoding
                             with gr.Accordion("Speculative decoding", open=False, elem_classes='tgw-accordion') as shared.gradio['speculative_decoding_accordion']:
                                 with gr.Row():
-                                    shared.gradio['model_draft'] = gr.Dropdown(label="model-draft", choices=utils.get_available_models(), value=lambda: shared.args.model_draft, elem_classes='slim-dropdown',     info='Draft model. Speculative decoding only works with models sharing the same vocabulary (e.g., same model family).', interactive=not mu)
+                                    shared.gradio['model_draft'] = gr.Dropdown(label="model-draft", choices=utils.get_available_models(), value=lambda: shared.args.model_draft, elem_classes='slim-dropdown', info='Draft model. Speculative decoding only works with models sharing the same vocabulary (e.g., same model family).', interactive=not mu)
                                     ui.create_refresh_button(shared.gradio['model_draft'], lambda: None, lambda: {'choices': utils.get_available_models()}, 'refresh-button', interactive=not mu)
 
                                 shared.gradio['draft_max'] = gr.Number(label="draft-max", precision=0, step=1, value=shared.args.draft_max, info='Number of tokens to draft for speculative decoding.')

From 1ee0acc852b7ac3e93cd1996414736bb75cfee59 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 28 Apr 2025 15:56:25 -0700
Subject: [PATCH 0409/1701] llama.cpp: Make --verbose print the llama-server
 command

---
 modules/llama_cpp_server.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 4b01c228d0..1f2db670fe 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -327,6 +327,11 @@ def _start_server(self):
             else:
                 env['LD_LIBRARY_PATH'] = os.path.dirname(self.server_path)
 
+        if shared.args.verbose:
+            logger.info(f"llama-server command-line flags:")
+            print(' '.join(str(item) for item in cmd[1:]))
+            print()
+
         # Start the server with pipes for output
         self.process = subprocess.Popen(
             cmd,

From d10bded7f807ad3eff1e5dc43370d8f9e09faf1d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 28 Apr 2025 22:37:01 -0700
Subject: [PATCH 0410/1701] UI: Add an `enable_thinking` option to
 enable/disable Qwen3 thinking

---
 extensions/openai/typing.py      |  1 +
 modules/chat.py                  | 53 +++++++++++++++++++++++++-------
 modules/loaders.py               |  5 +++
 modules/shared.py                |  1 +
 modules/ui.py                    |  1 +
 modules/ui_parameters.py         |  1 +
 user_data/settings-template.yaml |  1 +
 7 files changed, 52 insertions(+), 11 deletions(-)

diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 4d6018f9ec..b1979cbc6a 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -42,6 +42,7 @@ class GenerationOptions(BaseModel):
     auto_max_new_tokens: bool = False
     ban_eos_token: bool = False
     add_bos_token: bool = True
+    enable_thinking: bool = True
     skip_special_tokens: bool = True
     static_cache: bool = False
     truncation_length: int = 0
diff --git a/modules/chat.py b/modules/chat.py
index e117e6eeef..98913d5c8f 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -90,6 +90,44 @@ def get_generation_prompt(renderer, impersonate=False, strip_trailing_spaces=Tru
     return prefix, suffix
 
 
+def get_thinking_suppression_string(template):
+    """
+    Determines what string needs to be added to suppress thinking mode
+    by comparing template renderings with thinking enabled vs disabled.
+    """
+
+    # Render with thinking enabled
+    with_thinking = template.render(
+        messages=[{'role': 'user', 'content': ''}],
+        builtin_tools=None,
+        tools=None,
+        tools_in_user_message=False,
+        add_generation_prompt=True,
+        enable_thinking=True
+    )
+
+    # Render with thinking disabled
+    without_thinking = template.render(
+        messages=[{'role': 'user', 'content': ''}],
+        builtin_tools=None,
+        tools=None,
+        tools_in_user_message=False,
+        add_generation_prompt=True,
+        enable_thinking=False
+    )
+
+    # Find the difference (what gets added to suppress thinking)
+    i = 0
+    while i < min(len(with_thinking), len(without_thinking)) and with_thinking[i] == without_thinking[i]:
+        i += 1
+
+    j = 0
+    while j < min(len(with_thinking), len(without_thinking)) - i and with_thinking[-1 - j] == without_thinking[-1 - j]:
+        j += 1
+
+    return without_thinking[i:len(without_thinking) - j if j else None]
+
+
 def generate_chat_prompt(user_input, state, **kwargs):
     impersonate = kwargs.get('impersonate', False)
     _continue = kwargs.get('_continue', False)
@@ -147,13 +185,6 @@ def generate_chat_prompt(user_input, state, **kwargs):
     if user_input and not impersonate and not _continue:
         messages.append({"role": "user", "content": user_input})
 
-    def remove_extra_bos(prompt):
-        for bos_token in ['', '<|startoftext|>', '', '<|endoftext|>']:
-            while prompt.startswith(bos_token):
-                prompt = prompt[len(bos_token):]
-
-        return prompt
-
     def make_prompt(messages):
         if state['mode'] == 'chat-instruct' and _continue:
             prompt = renderer(messages=messages[:-1])
@@ -165,7 +196,6 @@ def make_prompt(messages):
             if state['custom_system_message'].strip() != '':
                 outer_messages.append({"role": "system", "content": state['custom_system_message']})
 
-            prompt = remove_extra_bos(prompt)
             command = state['chat-instruct_command']
             command = command.replace('<|character|>', state['name2'] if not impersonate else state['name1'])
             command = command.replace('<|prompt|>', prompt)
@@ -182,11 +212,10 @@ def make_prompt(messages):
             outer_messages.append({"role": "user", "content": command})
             outer_messages.append({"role": "assistant", "content": prefix})
 
-            prompt = instruction_template.render(messages=outer_messages)
+            prompt = instruct_renderer(messages=outer_messages)
             suffix = get_generation_prompt(instruct_renderer, impersonate=False)[1]
             if len(suffix) > 0:
                 prompt = prompt[:-len(suffix)]
-
         else:
             if _continue:
                 suffix = get_generation_prompt(renderer, impersonate=impersonate)[1]
@@ -199,7 +228,9 @@ def make_prompt(messages):
 
                 prompt += prefix
 
-        prompt = remove_extra_bos(prompt)
+        if state['mode'] == 'instruct' and not any((_continue, impersonate, state['enable_thinking'])):
+            prompt += get_thinking_suppression_string(instruction_template)
+
         return prompt
 
     prompt = make_prompt(messages)
diff --git a/modules/loaders.py b/modules/loaders.py
index b8ae82d7f6..738198b1e0 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -143,6 +143,7 @@ def transformers_samplers():
         'auto_max_new_tokens',
         'ban_eos_token',
         'add_bos_token',
+        'enable_thinking',
         'skip_special_tokens',
         'static_cache',
         'seed',
@@ -195,6 +196,7 @@ def transformers_samplers():
         'auto_max_new_tokens',
         'ban_eos_token',
         'add_bos_token',
+        'enable_thinking',
         'skip_special_tokens',
         'seed',
         'sampler_priority',
@@ -241,6 +243,7 @@ def transformers_samplers():
         'auto_max_new_tokens',
         'ban_eos_token',
         'add_bos_token',
+        'enable_thinking',
         'skip_special_tokens',
         'seed',
         'sampler_priority',
@@ -279,6 +282,7 @@ def transformers_samplers():
         'auto_max_new_tokens',
         'ban_eos_token',
         'add_bos_token',
+        'enable_thinking',
         'skip_special_tokens',
         'seed',
         'custom_token_bans',
@@ -311,6 +315,7 @@ def transformers_samplers():
         'auto_max_new_tokens',
         'ban_eos_token',
         'add_bos_token',
+        'enable_thinking',
         'seed',
         'sampler_priority',
         'dry_sequence_breakers',
diff --git a/modules/shared.py b/modules/shared.py
index 5d9dd36230..4c1179e345 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -51,6 +51,7 @@
     'auto_max_new_tokens': True,
     'ban_eos_token': False,
     'add_bos_token': True,
+    'enable_thinking': True,
     'skip_special_tokens': True,
     'stream': True,
     'static_cache': False,
diff --git a/modules/ui.py b/modules/ui.py
index f137e62d69..4105f53b75 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -198,6 +198,7 @@ def list_interface_input_elements():
         'auto_max_new_tokens',
         'ban_eos_token',
         'add_bos_token',
+        'enable_thinking',
         'skip_special_tokens',
         'stream',
         'static_cache',
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 6c2715af41..3f609d7157 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -82,6 +82,7 @@ def create_ui(default_preset):
                             shared.gradio['auto_max_new_tokens'] = gr.Checkbox(value=shared.settings['auto_max_new_tokens'], label='auto_max_new_tokens', info='Expand max_new_tokens to the available context length.')
                             shared.gradio['ban_eos_token'] = gr.Checkbox(value=shared.settings['ban_eos_token'], label='Ban the eos_token', info='Forces the model to never end the generation prematurely.')
                             shared.gradio['add_bos_token'] = gr.Checkbox(value=shared.settings['add_bos_token'], label='Add the bos_token to the beginning of prompts', info='Disabling this can make the replies more creative.')
+                            shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='enable_thinking', info='Used by Qwen3 to toggle  mode.')
                             shared.gradio['skip_special_tokens'] = gr.Checkbox(value=shared.settings['skip_special_tokens'], label='Skip special tokens', info='Some specific models need this unset.')
                             shared.gradio['stream'] = gr.Checkbox(value=shared.settings['stream'], label='Activate text streaming')
                             shared.gradio['static_cache'] = gr.Checkbox(value=shared.settings['static_cache'], label='Static KV cache', info='Use a static cache for improved performance.')
diff --git a/user_data/settings-template.yaml b/user_data/settings-template.yaml
index 83764f97ac..8092327659 100644
--- a/user_data/settings-template.yaml
+++ b/user_data/settings-template.yaml
@@ -22,6 +22,7 @@ max_updates_second: 0
 auto_max_new_tokens: true
 ban_eos_token: false
 add_bos_token: true
+enable_thinking: true
 skip_special_tokens: true
 stream: true
 static_cache: false

From c5fb51e5d12322bfb08458060913d9d1c90c5318 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 28 Apr 2025 22:39:42 -0700
Subject: [PATCH 0411/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index c8ac8a8685..08ffe37db5 100644
--- a/README.md
+++ b/README.md
@@ -36,7 +36,7 @@ Download from: https://github.com/oobabooga/text-generation-webui/releases
 #### Option 2: One-click installer
 
 1) Clone or [download the repository](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip).
-2) Run the script that matches your OS: `start_linux.sh`, `start_windows.bat`, `start_macos.sh`, or `start_wsl.bat`.
+2) Run the script that matches your OS: `start_linux.sh`, `start_windows.bat`, or `start_macos.sh`.
 3) Select your GPU vendor when asked.
 4) After installation completes, browse to `http://localhost:7860`.
 5) Have fun!

From 1dd4aedbe1edcc8fbfd7e7be07f170dbfaa7f0cf Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 29 Apr 2025 05:28:46 -0700
Subject: [PATCH 0412/1701] Fix the streaming_llm UI checkbox not being
 interactive

---
 modules/ui.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/ui.py b/modules/ui.py
index 4105f53b75..fb016f8769 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -114,6 +114,7 @@ def list_model_elements():
         'cache_type',
         'tensor_split',
         'extra_flags',
+        'streaming_llm',
         'gpu_split',
         'alpha_value',
         'rope_freq_base',

From c5fe92d1526cfe30825194f603acf2db38c964ec Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 30 Apr 2025 05:24:58 -0700
Subject: [PATCH 0413/1701] Bump llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 2 +-
 requirements/full/requirements_amd_noavx2.txt          | 2 +-
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 2 +-
 requirements/portable/requirements_amd_noavx2.txt      | 2 +-
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index c20c161e73..ac441e637c 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 437da5b595..b0da6c646b 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -29,6 +29,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index b1c879906a..1a5191f515 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -29,6 +29,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index e62987b061..ff2799f666 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index f7a9f114be..a5ca209e8d 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -29,8 +29,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index b8cd839082..dfef9addec 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 3b52d59b61..5b6e1124c5 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index a04e89792b..baecb103bc 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 5c717343d8..c6855bc818 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index b616193d4d..a7754ecec2 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -15,4 +15,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index de4740c909..72b1e57a0d 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -15,4 +15,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 6310327d37..3e0401c965 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index f69b58e7e0..b7efcaf069 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -15,6 +15,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index dafa6bbe41..26ebcd43b3 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index c02191eb13..54ebf5b4bf 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 456188b4aa..31b2bb3c45 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 7e733967ad..7f1b4c5157 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 0329a59892..29bb1e0609 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.6.0/llama_cpp_binaries-0.6.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 7f49e3c3ce6a09ffdd52526e6d98b3a886ccb47e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 30 Apr 2025 05:25:09 -0700
Subject: [PATCH 0414/1701] Bump ExLlamaV3

---
 requirements/full/requirements.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_noavx2.txt        | 4 ++--
 4 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index ac441e637c..038c384b7a 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -32,8 +32,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index ff2799f666..e9a54c33b6 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -31,5 +31,5 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index a5ca209e8d..cf8b3d7e2b 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -32,5 +32,5 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index baecb103bc..8769207570 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -32,8 +32,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a4/exllamav3-0.0.1a4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 771d3d8ed640bd412edfd597420213ffb1e6fc78 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 30 Apr 2025 06:48:32 -0700
Subject: [PATCH 0415/1701] Fix getting the llama.cpp logprobs for
 Qwen3-30B-A3B

---
 modules/logits.py | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/modules/logits.py b/modules/logits.py
index 9a4243ff8a..32aef7ae5d 100644
--- a/modules/logits.py
+++ b/modules/logits.py
@@ -45,6 +45,9 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
             output = {}
             for entry in logprobs:
                 token = repr(entry['token'])
+                if len(token) > 2 and token.startswith("'") and token.endswith("'"):
+                    token = token[1:-1]
+
                 prob = entry['prob'] if use_samplers else np.exp(entry['logprob'])
                 output[token] = prob
             return output
@@ -52,6 +55,9 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
             output = ''
             for entry in logprobs:
                 token = repr(entry['token'])
+                if len(token) > 2 and token.startswith("'") and token.endswith("'"):
+                    token = token[1:-1]
+
                 prob = entry['prob'] if use_samplers else np.exp(entry['logprob'])
                 output += f"{prob:.5f}  -  {token}\n"
             return output, previous

From e9569c3984c4f439d32425727cc1b2b7fc1178de Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 30 Apr 2025 06:57:23 -0700
Subject: [PATCH 0416/1701] Fixes after
 c5fe92d1526cfe30825194f603acf2db38c964ec

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 2 +-
 requirements/full/requirements_amd_noavx2.txt          | 2 +-
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 2 +-
 requirements/portable/requirements_amd_noavx2.txt      | 2 +-
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 038c384b7a..532b6f9d11 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index b0da6c646b..e887a5eb65 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -29,6 +29,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 1a5191f515..53a036de1b 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -29,6 +29,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index e9a54c33b6..0c7f8671ed 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index cf8b3d7e2b..c9682d0175 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -29,8 +29,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index dfef9addec..f758fe8eb0 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 5b6e1124c5..7a319ccf7c 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 8769207570..5909c25115 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index c6855bc818..8f03b4858a 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index a7754ecec2..c91d998c7b 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -15,4 +15,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index 72b1e57a0d..b2154c9fdb 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -15,4 +15,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 3e0401c965..81ad7b3668 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index b7efcaf069..f17dccf5f6 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -15,6 +15,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 26ebcd43b3..076f1b9c48 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 54ebf5b4bf..38134ccfdd 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 31b2bb3c45..4bbebc4ebb 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 7f1b4c5157..58f9f985d8 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 29bb1e0609..7e5080d527 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.6.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From a4bf3397240e3f52594186918da37a3224450221 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 30 Apr 2025 11:13:14 -0700
Subject: [PATCH 0417/1701] Bump llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 2 +-
 requirements/full/requirements_amd_noavx2.txt          | 2 +-
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 2 +-
 requirements/portable/requirements_amd_noavx2.txt      | 2 +-
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 532b6f9d11..6f265eba75 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index e887a5eb65..c8e75ee797 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -29,6 +29,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 53a036de1b..e54d6d9c51 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -29,6 +29,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 0c7f8671ed..d714ea3df6 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index c9682d0175..89f4f57639 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -29,8 +29,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6-py3-none-any.whl
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index f758fe8eb0..47ad5759a4 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 7a319ccf7c..334f11df64 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 5909c25115..e216c9cd13 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 8f03b4858a..c720daa7ef 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index c91d998c7b..7d9c00c079 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -15,4 +15,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index b2154c9fdb..d718c1b1c7 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -15,4 +15,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 81ad7b3668..9e184b5359 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index f17dccf5f6..ec059716b9 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -15,6 +15,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 076f1b9c48..d473b8241d 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 38134ccfdd..d3fffb4341 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 4bbebc4ebb..cdfa6a015f 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 58f9f985d8..1a7ce6ed01 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 7e5080d527..4737321dd5 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.7.0/llama_cpp_binaries-0.7.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From b46ca01340dfc096e88147e4f9dd3f971708d635 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 30 Apr 2025 14:53:15 -0700
Subject: [PATCH 0418/1701] UI: Set max_updates_second to 12 by default

When the tokens/second at at ~50 and the model is a thinking model,
the markdown rendering for the streaming message becomes a CPU
bottleneck.
---
 modules/shared.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/shared.py b/modules/shared.py
index 4c1179e345..fb10c0140c 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -47,7 +47,7 @@
     'max_new_tokens_max': 4096,
     'prompt_lookup_num_tokens': 0,
     'max_tokens_second': 0,
-    'max_updates_second': 0,
+    'max_updates_second': 12,
     'auto_max_new_tokens': True,
     'ban_eos_token': False,
     'add_bos_token': True,

From cd5c32dc197a6f62c7e328a36507eb202ff6b6a5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 30 Apr 2025 14:54:05 -0700
Subject: [PATCH 0419/1701] UI: Fix max_updates_second not working

---
 modules/text_generation.py | 2 --
 1 file changed, 2 deletions(-)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index 4e3d1d7adc..8d0918683d 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -100,8 +100,6 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
                     last_update = cur_time
                     yield reply
 
-                yield reply
-
         if stop_found or (state['max_tokens_second'] > 0 and shared.stop_everything):
             break
 

From 195a45c6e15c2af2075ea859caadbb6076497695 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 30 Apr 2025 15:12:46 -0700
Subject: [PATCH 0420/1701] UI: Make thinking blocks closed by default

---
 js/global_scope_js.js     | 22 ++++++++++++----------
 modules/html_generator.py |  2 +-
 2 files changed, 13 insertions(+), 11 deletions(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index e808c473c0..29d2d8bd27 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -31,13 +31,13 @@ function removeLastClick() {
 }
 
 function handleMorphdomUpdate(text) {
-  // Track closed blocks
-  const closedBlocks = new Set();
+  // Track open blocks
+  const openBlocks = new Set();
   document.querySelectorAll(".thinking-block").forEach(block => {
     const blockId = block.getAttribute("data-block-id");
-    // If block exists and is not open, add to closed set
-    if (blockId && !block.hasAttribute("open")) {
-      closedBlocks.add(blockId);
+    // If block exists and is open, add to open set
+    if (blockId && block.hasAttribute("open")) {
+      openBlocks.add(blockId);
     }
   });
 
@@ -72,13 +72,15 @@ function handleMorphdomUpdate(text) {
           }
         }
 
-        // For thinking blocks, respect closed state
+        // For thinking blocks, assume closed by default
         if (fromEl.classList && fromEl.classList.contains("thinking-block") &&
-            toEl.classList && toEl.classList.contains("thinking-block")) {
+           toEl.classList && toEl.classList.contains("thinking-block")) {
           const blockId = toEl.getAttribute("data-block-id");
-          // If this block was closed by user, keep it closed
-          if (blockId && closedBlocks.has(blockId)) {
-            toEl.removeAttribute("open");
+          // Remove open attribute by default
+          toEl.removeAttribute("open");
+          // If this block was explicitly opened by user, keep it open
+          if (blockId && openBlocks.has(blockId)) {
+            toEl.setAttribute("open", "");
           }
         }
 
diff --git a/modules/html_generator.py b/modules/html_generator.py
index c5252c26be..67d15b6e98 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -167,7 +167,7 @@ def convert_to_markdown(string, message_id=None):
         title_text = "Thinking..." if is_streaming else "Thought"
 
         thinking_block = f'''
-        
            
+        
            
             
                 
                     

From a6c3ec2299fe6dc766bf661ab8171781c8167245 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 30 Apr 2025 15:24:07 -0700
Subject: [PATCH 0421/1701] llama.cpp: Explicitly send cache_prompt = True

---
 modules/llama_cpp_server.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 1f2db670fe..71d498b102 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -135,6 +135,7 @@ def generate_with_streaming(self, prompt, state):
             "prompt": token_ids,
             "n_predict": max_new_tokens,
             "stream": True,
+            "cache_prompt": True
         })
 
         if shared.args.verbose:

From ec2e641749f3e48cb721cf4e860caf206038c711 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 30 Apr 2025 15:25:26 -0700
Subject: [PATCH 0422/1701] Update settings-template.yaml

---
 user_data/settings-template.yaml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/user_data/settings-template.yaml b/user_data/settings-template.yaml
index 8092327659..20896da327 100644
--- a/user_data/settings-template.yaml
+++ b/user_data/settings-template.yaml
@@ -18,7 +18,7 @@ max_new_tokens_min: 1
 max_new_tokens_max: 4096
 prompt_lookup_num_tokens: 0
 max_tokens_second: 0
-max_updates_second: 0
+max_updates_second: 12
 auto_max_new_tokens: true
 ban_eos_token: false
 add_bos_token: true

From 55283bb8f1a1f7fcef65b8f35ee81658e35a46af Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 30 Apr 2025 18:43:45 -0700
Subject: [PATCH 0423/1701] Fix CFG with ExLlamaV2_HF (closes #6937)

---
 modules/exllamav2_hf.py | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/modules/exllamav2_hf.py b/modules/exllamav2_hf.py
index eb80194079..4aa46375b6 100644
--- a/modules/exllamav2_hf.py
+++ b/modules/exllamav2_hf.py
@@ -65,7 +65,7 @@ def __init__(self, config: ExLlamaV2Config):
         elif kv_cache_type == 'q4':
             cache_type = ExLlamaV2Cache_Q4
         else:
-            raise ValueError(f"Invalid cache type for ExLlamaV2: {cache_type}. Valid options are: fp16, fp8, q8, q6, q4.")
+            raise ValueError(f"Invalid cache type for ExLlamaV2: {kv_cache_type}. Valid options are: fp16, fp8, q8, q6, q4.")
 
         # Use TP if specified
         if shared.args.enable_tp:
@@ -78,12 +78,10 @@ def __init__(self, config: ExLlamaV2Config):
 
         self.past_seq = None
         if shared.args.cfg_cache:
-            if shared.args.cache_8bit:
-                self.ex_cache_negative = ExLlamaV2Cache_8bit(self.ex_model)
-            elif shared.args.cache_4bit:
-                self.ex_cache_negative = ExLlamaV2Cache_Q4(self.ex_model)
+            if shared.args.enable_tp:
+                self.ex_cache_negative = ExLlamaV2Cache_TP(self.ex_model, base=cache_type)
             else:
-                self.ex_cache_negative = ExLlamaV2Cache(self.ex_model)
+                self.ex_cache_negative = cache_type(self.ex_model, lazy=shared.args.autosplit)
 
             self.past_seq_negative = None
 

From 307d13b540a0a32a93ed6b8b68c3a828a2576ddf Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 30 Apr 2025 18:58:14 -0700
Subject: [PATCH 0424/1701] UI: Minor label change

---
 modules/ui_model_menu.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index e0ae06f951..d13bcff72a 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -51,7 +51,7 @@ def create_ui():
                             shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
                             shared.gradio['batch_size'] = gr.Slider(label="batch_size", minimum=1, maximum=4096, step=1, value=shared.args.batch_size)
                             shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
-                            shared.gradio['ctx_size'] = gr.Number(label='ctx_size', precision=0, step=256, value=shared.args.ctx_size, info='Context length. ⚠️ Lower this value if you can\'t load the model. Common values: 2048, 4096, 8192, 16384, 32768, 65536.')
+                            shared.gradio['ctx_size'] = gr.Number(label='ctx-size', precision=0, step=256, value=shared.args.ctx_size, info='Context length. ⚠️ Lower this value if you can\'t load the model. Common values: 2048, 4096, 8192, 16384, 32768, 65536.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
                             shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')

From b950a0c6dbc6c61d2dcfdd61c38a3885e3c072a3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 30 Apr 2025 20:02:10 -0700
Subject: [PATCH 0425/1701] Lint

---
 modules/llama_cpp_server.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 71d498b102..d9187db8f4 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -329,7 +329,7 @@ def _start_server(self):
                 env['LD_LIBRARY_PATH'] = os.path.dirname(self.server_path)
 
         if shared.args.verbose:
-            logger.info(f"llama-server command-line flags:")
+            logger.info("llama-server command-line flags:")
             print(' '.join(str(item) for item in cmd[1:]))
             print()
 

From 89090d9a6127a92806f6e87469259d31259500ab Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 1 May 2025 08:22:54 -0700
Subject: [PATCH 0426/1701] Update README

---
 README.md | 19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)

diff --git a/README.md b/README.md
index 08ffe37db5..4b541b9e48 100644
--- a/README.md
+++ b/README.md
@@ -12,22 +12,23 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 
 ## Features
 
-- Supports multiple text generation backends in one UI/API, including [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [ExLlamaV2](https://github.com/turboderp-org/exllamav2). [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) is supported via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile), and the Transformers loader is compatible with libraries like [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM), but they must be installed manually.
-- OpenAI-compatible API with Chat and Completions endpoints – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples).
-- Automatic prompt formatting using Jinja2 templates.
+- Supports multiple text generation backends in one UI/API, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [ExLlamaV2](https://github.com/turboderp-org/exllamav2).
+  - [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) is also supported via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile).
+  - Additional quantization libraries like [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM) can be used with the Transformers loader if you install them manually.
+- Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for llama.cpp GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory that doesn't interfere with your system environment.
+- UI that resembles the original ChatGPT style.
+- Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
 - Three chat modes: `instruct`, `chat-instruct`, and `chat`, with automatic prompt templates in `chat-instruct`.
-- "Past chats" menu to quickly switch between conversations.
 - Free-form text generation in the Default/Notebook tabs without being limited to chat turns. You can send formatted conversations from the Chat tab to these.
 - Multiple sampling parameters and generation options for sophisticated text generation control.
-- Switch between different models easily in the UI without restarting.
-- Simple LoRA fine-tuning tool.
-- Requirements installed in a self-contained `installer_files` directory that doesn't interfere with the system environment.
+- Switch between different models easily in the UI without restarting, with fine control over settings.
+- OpenAI-compatible API with Chat and Completions endpoints – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples).
+- 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - Extension support, with numerous built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details.
-- Completely private: no telemetry, no tracking, no remote connections.
 
 ## How to install
 
-#### Option 1: Portable builds
+#### Option 1: Portable builds (start here)
 
 No installation needed – just unzip and run. Compatible with GGUF (llama.cpp) models on Windows, Linux, and macOS.
 

From c12a53c998ce39ec762b9f7895861f1d94c2d827 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 1 May 2025 19:46:56 -0700
Subject: [PATCH 0427/1701] Use turboderp's exllamav2 wheels

---
 requirements/full/requirements.txt               | 6 +++---
 requirements/full/requirements_amd.txt           | 4 ++--
 requirements/full/requirements_amd_noavx2.txt    | 4 ++--
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_noavx2.txt        | 6 +++---
 6 files changed, 12 insertions(+), 12 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 6f265eba75..c0ace41b0a 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_c
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index c8e75ee797..91582eb341 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -30,5 +30,5 @@ tiktoken
 
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index e54d6d9c51..7b86050e96 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -30,5 +30,5 @@ tiktoken
 
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.1.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index d714ea3df6..cc747edb81 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -32,4 +32,4 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6-py3-none-any.whl
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
+https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 89f4f57639..67b3260edc 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,4 +33,4 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_c
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6-py3-none-any.whl
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
+https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index e216c9cd13..3575d35269 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_c
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

From f8aaf3c23a793b60ce7452213304acb493be98af Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 1 May 2025 19:50:46 -0700
Subject: [PATCH 0428/1701] Use ROCm 6.2.4 on AMD

---
 README.md    | 2 +-
 one_click.py | 6 +++---
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/README.md b/README.md
index 4b541b9e48..3280186c16 100644
--- a/README.md
+++ b/README.md
@@ -90,7 +90,7 @@ conda activate textgen
 |--------|---------|---------|
 | Linux/WSL | NVIDIA | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/cu124` |
 | Linux/WSL | CPU only | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/cpu` |
-| Linux | AMD | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/rocm6.1` |
+| Linux | AMD | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/rocm6.2.4` |
 | MacOS + MPS | Any | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0` |
 | Windows | NVIDIA | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/cu124` |
 | Windows | CPU only | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0` |
diff --git a/one_click.py b/one_click.py
index 065afd999f..cb16b813ee 100644
--- a/one_click.py
+++ b/one_click.py
@@ -222,7 +222,7 @@ def update_pytorch_and_python():
     if "+cu" in torver:
         install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/cu124"
     elif "+rocm" in torver:
-        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.1"
+        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.2.4"
     elif "+cpu" in torver:
         install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/cpu"
     elif "+cxx11" in torver:
@@ -273,7 +273,7 @@ def install_webui():
             "What is your GPU?",
             {
                 'A': 'NVIDIA - CUDA 12.4',
-                'B': 'AMD - Linux/macOS only, requires ROCm 6.1',
+                'B': 'AMD - Linux/macOS only, requires ROCm 6.2.4',
                 'C': 'Apple M Series',
                 'D': 'Intel Arc (beta)',
                 'N': 'CPU mode'
@@ -314,7 +314,7 @@ def install_webui():
     if selected_gpu == "NVIDIA":
         install_pytorch += "--index-url https://download.pytorch.org/whl/cu124"
     elif selected_gpu == "AMD":
-        install_pytorch += "--index-url https://download.pytorch.org/whl/rocm6.1"
+        install_pytorch += "--index-url https://download.pytorch.org/whl/rocm6.2.4"
     elif selected_gpu in ["APPLE", "NONE"]:
         install_pytorch += "--index-url https://download.pytorch.org/whl/cpu"
     elif selected_gpu == "INTEL":

From d5c407cf35453ba2d06eea942942ff11cdc7993b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 1 May 2025 20:05:36 -0700
Subject: [PATCH 0429/1701] Use Vulkan instead of ROCm for llama.cpp on AMD

---
 requirements/full/requirements_amd.txt        | 3 ++-
 requirements/full/requirements_amd_noavx2.txt | 3 ++-
 2 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 91582eb341..24eeee6aa9 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -29,6 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 7b86050e96..99716f3cfb 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -29,6 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"

From 9e3867dc8358baf153d6f7c182496dad158696a4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 2 May 2025 09:36:15 -0700
Subject: [PATCH 0430/1701] llama.cpp: Fix manual random seeds

---
 modules/text_generation.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index 8d0918683d..b9bf9b16ca 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -480,7 +480,7 @@ def generate_reply_custom(question, original_question, state, stopping_strings=N
     For models that do not use the transformers library for sampling
     """
 
-    seed = set_manual_seed(state['seed'])
+    state['seed'] = set_manual_seed(state['seed'])
     t0 = time.time()
     reply = ''
     try:

From 3f26b0408bd02f500acc8c090a7e50ee286051b5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 2 May 2025 16:17:22 -0700
Subject: [PATCH 0431/1701] Fix after 9e3867dc8358baf153d6f7c182496dad158696a4

---
 modules/text_generation.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index b9bf9b16ca..8fd65dc423 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -500,7 +500,7 @@ def generate_reply_custom(question, original_question, state, stopping_strings=N
         t1 = time.time()
         original_tokens = len(encode(original_question)[0])
         new_tokens = len(encode(original_question + reply)[0]) - original_tokens
-        print(f'Output generated in {(t1-t0):.2f} seconds ({new_tokens/(t1-t0):.2f} tokens/s, {new_tokens} tokens, context {original_tokens}, seed {seed})')
+        print(f'Output generated in {(t1-t0):.2f} seconds ({new_tokens/(t1-t0):.2f} tokens/s, {new_tokens} tokens, context {original_tokens}, seed {state["seed"]})')
         return
 
 

From 905afced1c8339833280de254cd597b389a3dade Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 2 May 2025 16:32:22 -0700
Subject: [PATCH 0432/1701] Add a --portable flag to hide things in portable
 mode

---
 modules/presets.py       |  9 ++++++++-
 modules/shared.py        |  1 +
 modules/ui_model_menu.py | 17 +++++++++++------
 modules/ui_parameters.py |  2 +-
 server.py                |  5 +++--
 start_linux.sh           |  2 +-
 start_macos.sh           |  2 +-
 start_windows.bat        |  2 +-
 8 files changed, 27 insertions(+), 13 deletions(-)

diff --git a/modules/presets.py b/modules/presets.py
index a432bf5233..50d0f9850a 100644
--- a/modules/presets.py
+++ b/modules/presets.py
@@ -11,7 +11,7 @@
 
 
 def default_preset():
-    return {
+    result = {
         'temperature': 1,
         'dynatemp_low': 1,
         'dynatemp_high': 1,
@@ -50,6 +50,13 @@ def default_preset():
         'dry_sequence_breakers': '"\\n", ":", "\\"", "*"',
     }
 
+    if shared.args.portable:
+        samplers = result['sampler_priority'].split('\n')
+        samplers = [sampler for sampler in samplers if sampler in ["dry", "top_k", "typ_p", "top_p", "min_p", "xtc", "temperature", "repetition_penalty"]]
+        result['sampler_priority'] = '\n'.join(samplers)
+
+    return result
+
 
 def presets_params():
     return [k for k in default_preset()]
diff --git a/modules/shared.py b/modules/shared.py
index fb10c0140c..39b0bdaaab 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -190,6 +190,7 @@
 group.add_argument('--ssl-certfile', type=str, help='The path to the SSL certificate cert file.', default=None)
 group.add_argument('--subpath', type=str, help='Customize the subpath for gradio, use with reverse proxy')
 group.add_argument('--old-colors', action='store_true', help='Use the legacy Gradio colors, before the December/2024 update.')
+group.add_argument('--portable', action='store_true', help='Hide features not available in portable mode like training.')
 
 # API
 group = parser.add_argument_group('API')
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index d13bcff72a..4a49d20934 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -35,14 +35,17 @@ def create_ui():
                             shared.gradio['save_model_settings'] = gr.Button("Save settings", elem_classes='refresh-button', interactive=not mu)
 
                     with gr.Column():
-                        with gr.Row():
-                            shared.gradio['lora_menu'] = gr.Dropdown(multiselect=True, choices=utils.get_available_loras(), value=shared.lora_names, label='LoRA(s)', elem_classes='slim-dropdown', interactive=not mu)
-                            ui.create_refresh_button(shared.gradio['lora_menu'], lambda: None, lambda: {'choices': utils.get_available_loras(), 'value': shared.lora_names}, 'refresh-button', interactive=not mu)
-                            shared.gradio['lora_menu_apply'] = gr.Button(value='Apply LoRAs', elem_classes='refresh-button', interactive=not mu)
+                        if shared.args.portable:
+                            pass
+                        else:
+                            with gr.Row():
+                                shared.gradio['lora_menu'] = gr.Dropdown(multiselect=True, choices=utils.get_available_loras(), value=shared.lora_names, label='LoRA(s)', elem_classes='slim-dropdown', interactive=not mu)
+                                ui.create_refresh_button(shared.gradio['lora_menu'], lambda: None, lambda: {'choices': utils.get_available_loras(), 'value': shared.lora_names}, 'refresh-button', interactive=not mu)
+                                shared.gradio['lora_menu_apply'] = gr.Button(value='Apply LoRAs', elem_classes='refresh-button', interactive=not mu)
 
         with gr.Row():
             with gr.Column():
-                shared.gradio['loader'] = gr.Dropdown(label="Model loader", choices=loaders.loaders_and_params.keys(), value=None)
+                shared.gradio['loader'] = gr.Dropdown(label="Model loader", choices=loaders.loaders_and_params.keys() if not shared.args.portable else ['llama.cpp'], value=None)
                 with gr.Blocks():
                     with gr.Row():
                         with gr.Column():
@@ -150,7 +153,9 @@ def create_event_handlers():
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         save_model_settings, gradio('model_menu', 'interface_state'), gradio('model_status'), show_progress=False)
 
-    shared.gradio['lora_menu_apply'].click(load_lora_wrapper, gradio('lora_menu'), gradio('model_status'), show_progress=False)
+    if not shared.args.portable:
+        shared.gradio['lora_menu_apply'].click(load_lora_wrapper, gradio('lora_menu'), gradio('model_status'), show_progress=False)
+
     shared.gradio['download_model_button'].click(download_model_wrapper, gradio('custom_model_menu', 'download_specific_file'), gradio('model_status'), show_progress=True)
     shared.gradio['get_file_list'].click(partial(download_model_wrapper, return_links=True), gradio('custom_model_menu', 'download_specific_file'), gradio('model_status'), show_progress=True)
     shared.gradio['autoload_model'].change(lambda x: gr.update(visible=not x), gradio('autoload_model'), gradio('load_model'))
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 3f609d7157..071b30b685 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -21,7 +21,7 @@ def create_ui(default_preset):
                         shared.gradio['random_preset'] = gr.Button('🎲', elem_classes='refresh-button')
 
                 with gr.Column():
-                    shared.gradio['filter_by_loader'] = gr.Dropdown(label="Filter by loader", choices=["All"] + list(loaders.loaders_and_params.keys()), value="All", elem_classes='slim-dropdown')
+                    shared.gradio['filter_by_loader'] = gr.Dropdown(label="Filter by loader", choices=["All"] + list(loaders.loaders_and_params.keys()) if not shared.args.portable else ['llama.cpp'], value="All", elem_classes='slim-dropdown')
 
             with gr.Row():
                 with gr.Column():
diff --git a/server.py b/server.py
index 169578a561..b0b9e63317 100644
--- a/server.py
+++ b/server.py
@@ -90,7 +90,7 @@ def create_interface():
         'instruction_template_str': shared.settings['instruction_template_str'],
         'prompt_menu-default': shared.settings['prompt-default'],
         'prompt_menu-notebook': shared.settings['prompt-notebook'],
-        'filter_by_loader': shared.args.loader or 'All'
+        'filter_by_loader': (shared.args.loader or 'All') if not shared.args.portable else 'llama.cpp'
     })
 
     if Path("user_data/cache/pfp_character.png").exists():
@@ -127,7 +127,8 @@ def create_interface():
 
         ui_parameters.create_ui(shared.settings['preset'])  # Parameters tab
         ui_model_menu.create_ui()  # Model tab
-        training.create_ui()  # Training tab
+        if not shared.args.portable:
+            training.create_ui()  # Training tab
         ui_session.create_ui()  # Session tab
 
         # Generation events
diff --git a/start_linux.sh b/start_linux.sh
index 00082f0721..c74f127258 100755
--- a/start_linux.sh
+++ b/start_linux.sh
@@ -4,7 +4,7 @@ cd "$(dirname "${BASH_SOURCE[0]}")"
 
 # Portable install case
 if [ -d "portable_env" ]; then
-    ./portable_env/bin/python3 server.py --api --auto-launch "$@"
+    ./portable_env/bin/python3 server.py --portable --api --auto-launch "$@"
     exit $?
 fi
 
diff --git a/start_macos.sh b/start_macos.sh
index 628f59ccfd..7a060ba601 100755
--- a/start_macos.sh
+++ b/start_macos.sh
@@ -4,7 +4,7 @@ cd "$(dirname "${BASH_SOURCE[0]}")"
 
 # Portable install case
 if [ -d "portable_env" ]; then
-    ./portable_env/bin/python3 server.py --api --auto-launch --api-port 5005 "$@"
+    ./portable_env/bin/python3 server.py --portable --api --auto-launch --api-port 5005 "$@"
     exit $?
 fi
 
diff --git a/start_windows.bat b/start_windows.bat
index 451b85e0d0..1616ee27db 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -5,7 +5,7 @@ cd /D "%~dp0"
 
 @rem Portable install case
 if exist "portable_env" (
-    .\portable_env\python.exe server.py --api --auto-launch %*
+    .\portable_env\python.exe server.py --portable --api --auto-launch %*
     exit /b %errorlevel%
 )
 

From 4cea720da8cca27cbb5e8ac560019a55e6afb73a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 2 May 2025 16:38:28 -0700
Subject: [PATCH 0433/1701] UI: Remove the "Autoload the model" feature

---
 modules/shared.py                | 1 -
 modules/ui_model_menu.py         | 9 ++-------
 user_data/settings-template.yaml | 1 -
 3 files changed, 2 insertions(+), 9 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 39b0bdaaab..cfedb992ee 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -60,7 +60,6 @@
     'custom_stopping_strings': '',
     'custom_token_bans': '',
     'negative_prompt': '',
-    'autoload_model': False,
     'dark_theme': True,
     'default_extensions': [],
     'instruction_template_str': "{%- set ns = namespace(found=false) -%}\n{%- for message in messages -%}\n    {%- if message['role'] == 'system' -%}\n        {%- set ns.found = true -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if not ns.found -%}\n    {{- '' + 'Below is an instruction that describes a task. Write a response that appropriately completes the request.' + '\\n\\n' -}}\n{%- endif %}\n{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {{- '' + message['content'] + '\\n\\n' -}}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{-'### Instruction:\\n' + message['content'] + '\\n\\n'-}}\n        {%- else -%}\n            {{-'### Response:\\n' + message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{-'### Response:\\n'-}}\n{%- endif -%}",
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 4a49d20934..9361ef910a 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -30,7 +30,7 @@ def create_ui():
                         with gr.Row():
                             shared.gradio['model_menu'] = gr.Dropdown(choices=utils.get_available_models(), value=lambda: shared.model_name, label='Model', elem_classes='slim-dropdown', interactive=not mu)
                             ui.create_refresh_button(shared.gradio['model_menu'], lambda: None, lambda: {'choices': utils.get_available_models()}, 'refresh-button', interactive=not mu)
-                            shared.gradio['load_model'] = gr.Button("Load", visible=not shared.settings['autoload_model'], elem_classes='refresh-button', interactive=not mu)
+                            shared.gradio['load_model'] = gr.Button("Load", elem_classes='refresh-button', interactive=not mu)
                             shared.gradio['unload_model'] = gr.Button("Unload", elem_classes='refresh-button', interactive=not mu)
                             shared.gradio['save_model_settings'] = gr.Button("Save settings", elem_classes='refresh-button', interactive=not mu)
 
@@ -108,9 +108,6 @@ def create_ui():
                                 shared.gradio['ctx_size_draft'] = gr.Number(label="ctx-size-draft", precision=0, step=256, value=shared.args.ctx_size_draft, info='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
 
             with gr.Column():
-                with gr.Row():
-                    shared.gradio['autoload_model'] = gr.Checkbox(value=shared.settings['autoload_model'], label='Autoload the model', info='Whether to load the model as soon as it is selected in the Model dropdown.', interactive=not mu)
-
                 with gr.Tab("Download"):
                     shared.gradio['custom_model_menu'] = gr.Textbox(label="Download model or LoRA", info="Enter the Hugging Face username/model path, for instance: facebook/galactica-125m. To specify a branch, add it at the end after a \":\" character like this: facebook/galactica-125m:main. To download a single file, enter its name in the second box.", interactive=not mu)
                     shared.gradio['download_specific_file'] = gr.Textbox(placeholder="File name (for GGUF models)", show_label=False, max_lines=1, interactive=not mu)
@@ -135,11 +132,10 @@ def create_event_handlers():
 
     # In this event handler, the interface state is read and updated
     # with the model defaults (if any), and then the model is loaded
-    # unless "autoload_model" is unchecked
     shared.gradio['model_menu'].change(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         handle_load_model_event_initial, gradio('model_menu', 'interface_state'), gradio(ui.list_interface_input_elements()) + gradio('interface_state'), show_progress=False).then(
-        load_model_wrapper, gradio('model_menu', 'loader', 'autoload_model'), gradio('model_status'), show_progress=True).success(
+        partial(load_model_wrapper, autoload=False), gradio('model_menu', 'loader'), gradio('model_status'), show_progress=True).success(
         handle_load_model_event_final, gradio('truncation_length', 'loader', 'interface_state'), gradio('truncation_length', 'filter_by_loader'), show_progress=False)
 
     shared.gradio['load_model'].click(
@@ -158,7 +154,6 @@ def create_event_handlers():
 
     shared.gradio['download_model_button'].click(download_model_wrapper, gradio('custom_model_menu', 'download_specific_file'), gradio('model_status'), show_progress=True)
     shared.gradio['get_file_list'].click(partial(download_model_wrapper, return_links=True), gradio('custom_model_menu', 'download_specific_file'), gradio('model_status'), show_progress=True)
-    shared.gradio['autoload_model'].change(lambda x: gr.update(visible=not x), gradio('autoload_model'), gradio('load_model'))
     shared.gradio['customized_template_submit'].click(save_instruction_template, gradio('model_menu', 'customized_template'), gradio('model_status'), show_progress=True)
 
 
diff --git a/user_data/settings-template.yaml b/user_data/settings-template.yaml
index 20896da327..ce0f77e142 100644
--- a/user_data/settings-template.yaml
+++ b/user_data/settings-template.yaml
@@ -31,7 +31,6 @@ seed: -1
 custom_stopping_strings: ''
 custom_token_bans: ''
 negative_prompt: ''
-autoload_model: false
 dark_theme: true
 default_extensions: []
 instruction_template_str: |-

From 3526b7923c9f5a3b3ba55056e445a660a03d2bc6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 2 May 2025 17:40:53 -0700
Subject: [PATCH 0434/1701] Remove extensions with requirements from portable
 builds

---
 .github/workflows/build-portable-release-cuda.yml   | 2 ++
 .github/workflows/build-portable-release-vulkan.yml | 2 ++
 .github/workflows/build-portable-release.yml        | 2 ++
 3 files changed, 6 insertions(+)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index fb9e61b008..571cbac0c5 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -102,6 +102,8 @@ jobs:
         shell: bash
         run: |
             rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker
+            allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
+            find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
             # Define common variables
             CUDA_VERSION="${{ matrix.cuda }}"
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 8de29791af..4e88d4d9ab 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -101,6 +101,8 @@ jobs:
         shell: bash
         run: |
             rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker
+            allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
+            find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
             # Define common variables
             AVX_SUPPORT="${{ matrix.avx }}"
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index bdf96cecb2..6910ce2c3b 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -101,6 +101,8 @@ jobs:
         shell: bash
         run: |
             rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker
+            allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
+            find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
             # Define common variables
             AVX_SUPPORT="${{ matrix.avx }}"

From d08acb4af9c2a4f4d0f7fd97babb217c0890e1c8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 2 May 2025 20:50:52 -0700
Subject: [PATCH 0435/1701] UI: Rename enable_thinking -> Enable thinking

---
 modules/ui_parameters.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 071b30b685..733d090101 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -82,7 +82,7 @@ def create_ui(default_preset):
                             shared.gradio['auto_max_new_tokens'] = gr.Checkbox(value=shared.settings['auto_max_new_tokens'], label='auto_max_new_tokens', info='Expand max_new_tokens to the available context length.')
                             shared.gradio['ban_eos_token'] = gr.Checkbox(value=shared.settings['ban_eos_token'], label='Ban the eos_token', info='Forces the model to never end the generation prematurely.')
                             shared.gradio['add_bos_token'] = gr.Checkbox(value=shared.settings['add_bos_token'], label='Add the bos_token to the beginning of prompts', info='Disabling this can make the replies more creative.')
-                            shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='enable_thinking', info='Used by Qwen3 to toggle  mode.')
+                            shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', info='Used by Qwen3 to toggle  mode.')
                             shared.gradio['skip_special_tokens'] = gr.Checkbox(value=shared.settings['skip_special_tokens'], label='Skip special tokens', info='Some specific models need this unset.')
                             shared.gradio['stream'] = gr.Checkbox(value=shared.settings['stream'], label='Activate text streaming')
                             shared.gradio['static_cache'] = gr.Checkbox(value=shared.settings['static_cache'], label='Static KV cache', info='Use a static cache for improved performance.')

From b21bd8bb1e79466be945abfe417e92e52b63ec6f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 2 May 2025 22:41:49 -0700
Subject: [PATCH 0436/1701] UI: Invert user/assistant message colors in
 instruct mode

The goal is to make assistant messages more readable.
---
 css/html_instruct_style.css | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index 4613b38055..b98544a1dc 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -61,11 +61,11 @@
 }
 
 .dark .chat .user-message {
-    background: transparent;
+    background: var(--light-gray);
 }
 
 .dark .chat .assistant-message {
-    background: var(--light-gray);
+    background: transparent;
 }
 
 .chat .user-message .text,

From b71ef50e9d01c15a09c67b95e2032fed535c63ba Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 2 May 2025 23:45:58 -0700
Subject: [PATCH 0437/1701] UI: Add a min-height to prevent constant scrolling
 during chat streaming

---
 css/chat_style-Dark.css            | 2 ++
 css/chat_style-TheEncrypted777.css | 2 ++
 css/chat_style-cai-chat.css        | 1 +
 css/main.css                       | 5 +++++
 modules/html_generator.py          | 9 ++++++---
 5 files changed, 16 insertions(+), 3 deletions(-)

diff --git a/css/chat_style-Dark.css b/css/chat_style-Dark.css
index 368a2a16b1..3b2bd38556 100644
--- a/css/chat_style-Dark.css
+++ b/css/chat_style-Dark.css
@@ -1,5 +1,6 @@
 .message {
     display: grid;
+    align-items: start;
     grid-template-columns: 60px minmax(0, 1fr);
     padding-bottom: 28px;
     font-size: 18px;
@@ -102,6 +103,7 @@
 @media screen and (width <= 688px) {
     .message {
         display: grid;
+        align-items: start;
         grid-template-columns: 60px minmax(0, 1fr);
         padding-bottom: 25px;
         font-size: 15px;
diff --git a/css/chat_style-TheEncrypted777.css b/css/chat_style-TheEncrypted777.css
index 6404f41d7a..25d26db81a 100644
--- a/css/chat_style-TheEncrypted777.css
+++ b/css/chat_style-TheEncrypted777.css
@@ -2,6 +2,7 @@
 
 .message {
     display: grid;
+    align-items: start;
     grid-template-columns: 60px minmax(0, 1fr);
     padding-bottom: 28px;
     font-size: 18px;
@@ -100,6 +101,7 @@
 @media screen and (width <= 688px) {
     .message {
         display: grid;
+        align-items: start;
         grid-template-columns: 60px minmax(0, 1fr);
         padding-bottom: 25px;
         font-size: 15px;
diff --git a/css/chat_style-cai-chat.css b/css/chat_style-cai-chat.css
index 93276bd347..223f615087 100644
--- a/css/chat_style-cai-chat.css
+++ b/css/chat_style-cai-chat.css
@@ -1,5 +1,6 @@
 .message {
     display: grid;
+    align-items: start;
     grid-template-columns: 60px minmax(0, 1fr);
     padding-bottom: 2em;
     font-size: 15px;
diff --git a/css/main.css b/css/main.css
index d6e5ac83e3..cf0dfde70b 100644
--- a/css/main.css
+++ b/css/main.css
@@ -403,6 +403,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 .chat-parent {
     height: calc(100dvh - 98px - var(--input-delta));
     overflow: auto !important;
+    /* scroll-behavior: smooth; */
     border-radius: 0 !important;
     margin-bottom: var(--input-delta) !important;
 }
@@ -1382,3 +1383,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     50% { opacity: 1; }
     100% { opacity: 0.6; }
 }
+
+.streaming {
+    min-height: 70vh;
+}
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 67d15b6e98..a6f5f930f0 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -365,8 +365,9 @@ def generate_instruct_html(history):
                 f'
            '
             )
 
+        streaming_class = " streaming" if i == len(history["visible"]) - 1 else ""
         output += (
-            f'
            '
             f'
            '
             f'
            {converted_visible[1]}
            '
@@ -414,8 +415,9 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
                 f'
            '
             )
 
+        streaming_class = " streaming" if i == len(history["visible"]) - 1 else ""
         output += (
-            f'
            '
             f'
            {img_bot}
            '
             f'
            '
@@ -452,8 +454,9 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
                 f'
            '
             )
 
+        streaming_class = " streaming" if i == len(history["visible"]) - 1 else ""
         output += (
-            f'
            '
             f'
            '
             f'
            {converted_visible[1]}
            '

From ea60f14674a89d3a71e5504edacb8f64f148b57c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 3 May 2025 06:06:50 -0700
Subject: [PATCH 0438/1701] UI: Show the list of files if the user tries to
 download a GGUF repository

---
 modules/ui_model_menu.py | 32 ++++++++++++++++++++++++++++++++
 1 file changed, 32 insertions(+)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 9361ef910a..2c593df66f 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -192,6 +192,26 @@ def load_lora_wrapper(selected_loras):
 
 def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), return_links=False, check=False):
     try:
+        # Handle direct GGUF URLs
+        if repo_id.startswith("https://") and ("huggingface.co" in repo_id) and (repo_id.endswith(".gguf") or repo_id.endswith(".gguf?download=true")):
+            try:
+                path = repo_id.split("huggingface.co/")[1]
+
+                # Extract the repository ID (first two parts of the path)
+                parts = path.split("/")
+                if len(parts) >= 2:
+                    extracted_repo_id = f"{parts[0]}/{parts[1]}"
+
+                    # Extract the filename (last part of the path)
+                    filename = repo_id.split("/")[-1]
+                    if "?download=true" in filename:
+                        filename = filename.replace("?download=true", "")
+
+                    repo_id = extracted_repo_id
+                    specific_file = filename
+            except:
+                pass
+
         if repo_id == "":
             yield ("Please enter a model path")
             return
@@ -205,6 +225,18 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
 
         yield ("Getting the download links from Hugging Face")
         links, sha256, is_lora, is_llamacpp = downloader.get_download_links_from_huggingface(model, branch, text_only=False, specific_file=specific_file)
+
+        # Check for multiple GGUF files
+        gguf_files = [link for link in links if link.lower().endswith('.gguf')]
+        if len(gguf_files) > 1 and not specific_file:
+            output = "Multiple GGUF files found. Please copy one of the following filenames to the 'File name' field:\n\n```\n"
+            for link in gguf_files:
+                output += f"{Path(link).name}\n"
+
+            output += "```"
+            yield output
+            return
+
         if return_links:
             output = "```\n"
             for link in links:

From 4c2e3b168bc1751dbb3f1b222fdd749ad7a5d36e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 3 May 2025 06:51:20 -0700
Subject: [PATCH 0439/1701] llama.cpp: Add a retry mechanism when getting the
 logits (sometimes it fails)

---
 modules/llama_cpp_server.py | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index d9187db8f4..2ebeb56032 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -210,14 +210,15 @@ def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
             pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
             print()
 
-        response = self.session.post(url, json=payload)
-        result = response.json()
+        for retry in range(5):
+            response = self.session.post(url, json=payload)
+            result = response.json()
 
-        if "completion_probabilities" in result:
-            if use_samplers:
-                return result["completion_probabilities"][0]["top_probs"]
-            else:
-                return result["completion_probabilities"][0]["top_logprobs"]
+            if "completion_probabilities" in result:
+                if use_samplers:
+                    return result["completion_probabilities"][0]["top_probs"]
+                else:
+                    return result["completion_probabilities"][0]["top_logprobs"]
         else:
             raise Exception(f"Unexpected response format: 'completion_probabilities' not found in {result}")
 

From 5f5569e9ac21ffdcb335b0557909ad102104fc8f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 4 May 2025 06:20:24 -0700
Subject: [PATCH 0440/1701] Update README

---
 README.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index 3280186c16..8a7b24674a 100644
--- a/README.md
+++ b/README.md
@@ -44,7 +44,7 @@ Download from: https://github.com/oobabooga/text-generation-webui/releases
 
 To restart the web UI later, just run the same `start_` script. If you need to reinstall, delete the `installer_files` folder created during setup and run the script again.
 
-You can use command-line flags, like `./start_linux.sh --help`, or add them to `user_data/CMD_FLAGS.txt` (such as `--api` to enable API use). To update the project, run `update_wizard_linux.sh`, `update_wizard_windows.bat`, `update_wizard_macos.sh`, or `update_wizard_wsl.bat`.
+You can use command-line flags, like `./start_linux.sh --help`, or add them to `user_data/CMD_FLAGS.txt` (such as `--api` to enable API use). To update the project, run `update_wizard_linux.sh`, `update_wizard_windows.bat`, or `update_wizard_macos.sh`.
 
 
            
 
@@ -55,7 +55,7 @@ Setup details and information about installing manually
 
 The script uses Miniconda to set up a Conda environment in the `installer_files` folder.
 
-If you ever need to install something manually in the `installer_files` environment, you can launch an interactive shell using the cmd script: `cmd_linux.sh`, `cmd_windows.bat`, `cmd_macos.sh`, or `cmd_wsl.bat`.
+If you ever need to install something manually in the `installer_files` environment, you can launch an interactive shell using the cmd script: `cmd_linux.sh`, `cmd_windows.bat`, or `cmd_macos.sh`.
 
 * There is no need to run any of those scripts (`start_`, `update_wizard_`, or `cmd_`) as admin/root.
 * To install the requirements for extensions, you can use the `extensions_reqs` script for your OS. At the end, this script will install the main requirements for the project to make sure that they take precedence in case of version conflicts.

From b7a5c7db8de89159144fadb59920045efc3fe544 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 4 May 2025 07:14:42 -0700
Subject: [PATCH 0441/1701] llama.cpp: Handle short arguments in --extra-flags

---
 modules/llama_cpp_server.py | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 2ebeb56032..7244001a80 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -317,9 +317,15 @@ def _start_server(self):
             for flag_item in extra_flags.split(','):
                 if '=' in flag_item:
                     flag, value = flag_item.split('=', 1)
-                    cmd += [f"--{flag}", value]
+                    if len(flag) <= 3:
+                        cmd += [f"-{flag}", value]
+                    else:
+                        cmd += [f"--{flag}", value]
                 else:
-                    cmd.append(f"--{flag_item}")
+                    if len(flag_item) <= 3:
+                        cmd.append(f"-{flag_item}")
+                    else:
+                        cmd.append(f"--{flag_item}")
 
         env = os.environ.copy()
         if os.name == 'posix':

From 7853fb1c8d701bb8b720b3907bdc50017911d6a6 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Sun, 4 May 2025 18:58:37 -0300
Subject: [PATCH 0442/1701] Optimize the Chat tab (#6948)

---
 css/main.css       | 34 +++++++++++-----------------------
 js/main.js         |  8 +-------
 modules/ui_chat.py |  2 +-
 3 files changed, 13 insertions(+), 31 deletions(-)

diff --git a/css/main.css b/css/main.css
index cf0dfde70b..64e96ccc31 100644
--- a/css/main.css
+++ b/css/main.css
@@ -389,7 +389,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 .chat {
     margin-left: auto;
     margin-right: auto;
-    min-height: var(--chat-height);
+    flex: 1;
     overflow-y: auto;
     display: flex;
     flex-direction: column;
@@ -401,11 +401,9 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 .chat-parent {
-    height: calc(100dvh - 98px - var(--input-delta));
+    flex: 1;
     overflow: auto !important;
-    /* scroll-behavior: smooth; */
     border-radius: 0 !important;
-    margin-bottom: var(--input-delta) !important;
 }
 
 .chat-parent .prose {
@@ -422,8 +420,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 .chat-parent.bigchat {
-    height: calc(100dvh - 98px - var(--input-delta)) !important;
-    margin-bottom: var(--input-delta) !important;
+    flex: 1;
 }
 
 .chat > .messages {
@@ -604,8 +601,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 .chat-input-positioned {
-    position: absolute;
-    bottom: 0;
     max-width: 54rem;
     left: 50%;
     transform: translateX(-50%);
@@ -790,7 +785,8 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 #chat-input-container {
-    min-width: 0 !important;
+    display: flex;
+    flex-direction: column;
 }
 
 #chat-input-container > .form {
@@ -799,9 +795,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 #chat-input-row {
-    padding-bottom: 1.5em;
-    padding-left: 1rem;
-    padding-right: 1rem;
+    padding: 1rem;
 }
 
 #chat-input-row.bigchat {
@@ -809,22 +803,16 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 #chat-col {
-    padding-bottom: 100px;
+    height: 100dvh;
+    display: flex;
+    flex-direction: column;
+    padding-bottom: 0;
 }
 
 @media screen and (width <= 924px) {
     #chat-col {
-        padding-bottom: 100px;
+        height: calc(100dvh - 132px);
         margin-top: 32px;
-        position: relative; /* Ensure positioning for the pseudo-element */
-    }
-
-    .chat-parent {
-        height: calc(100dvh - 98px - var(--input-delta) - 32px);
-    }
-
-    .chat-parent.bigchat {
-        height: calc(100dvh - 98px - var(--input-delta) - 32px) !important;
     }
 }
 
diff --git a/js/main.js b/js/main.js
index 33b7d6bdc1..408815db66 100644
--- a/js/main.js
+++ b/js/main.js
@@ -442,12 +442,6 @@ function updateCssProperties() {
 
   // Check if the chat container is visible
   if (chatContainer.clientHeight > 0) {
-    const chatContainerParentHeight = chatContainer.parentNode.clientHeight;
-    const newChatHeight = `${chatContainerParentHeight - chatInputHeight - 80}px`;
-
-    document.documentElement.style.setProperty("--chat-height", newChatHeight);
-    document.documentElement.style.setProperty("--input-delta", `${chatInputHeight - 40}px`);
-
     // Adjust scrollTop based on input height change
     if (chatInputHeight !== currentChatInputHeight) {
       const deltaHeight = chatInputHeight - currentChatInputHeight;
@@ -720,7 +714,7 @@ function isMobile() {
 // Function to initialize sidebars
 function initializeSidebars() {
   const isOnMobile = isMobile();
-  
+
   if (isOnMobile) {
     // Mobile state: Hide sidebars and set closed states
     [pastChatsRow, chatControlsRow, headerBar].forEach(el => {
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 0d58854945..0856cfab05 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -46,8 +46,8 @@ def create_ui():
 
         with gr.Row():
             with gr.Column(elem_id='chat-col'):
-                shared.gradio['html_display'] = gr.HTML(value=chat_html_wrapper({'internal': [], 'visible': []}, '', '', 'chat', 'cai-chat', '')['html'], visible=True)
                 shared.gradio['display'] = gr.JSON(value={}, visible=False)  # Hidden buffer
+                shared.gradio['html_display'] = gr.HTML(value=chat_html_wrapper({'internal': [], 'visible': []}, '', '', 'chat', 'cai-chat', '')['html'], visible=True)
                 with gr.Row(elem_id="chat-input-row"):
                     with gr.Column(scale=1, elem_id='gr-hover-container'):
                         gr.HTML(value='
            ☰
            ', elem_id='gr-hover')

From d1866219261c5fa1e9d8c0a9c6c380b965ca7cc7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 4 May 2025 15:19:46 -0700
Subject: [PATCH 0443/1701] UI: Fixes after previous commit

---
 css/main.css | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 64e96ccc31..f76a27871e 100644
--- a/css/main.css
+++ b/css/main.css
@@ -787,6 +787,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 #chat-input-container {
     display: flex;
     flex-direction: column;
+    min-width: 0 !important;
 }
 
 #chat-input-container > .form {
@@ -807,12 +808,13 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     display: flex;
     flex-direction: column;
     padding-bottom: 0;
+    gap: 0;
 }
 
 @media screen and (width <= 924px) {
     #chat-col {
-        height: calc(100dvh - 132px);
         margin-top: 32px;
+        height: calc(100dvh - 32px);
     }
 }
 

From 84ab1f95bedd2433cec165502375901fdaa56a98 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 4 May 2025 15:21:52 -0700
Subject: [PATCH 0444/1701] UI: Increase the chat area a bit

---
 css/main.css | 1 +
 1 file changed, 1 insertion(+)

diff --git a/css/main.css b/css/main.css
index f76a27871e..d5d5e77165 100644
--- a/css/main.css
+++ b/css/main.css
@@ -797,6 +797,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 
 #chat-input-row {
     padding: 1rem;
+    padding-top: 0;
 }
 
 #chat-input-row.bigchat {

From d9da16edba88b09dcbfe97a7be302c65ed244ebb Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 4 May 2025 16:53:52 -0700
Subject: [PATCH 0445/1701] UI: Remove the chat input textarea border

---
 css/main.css | 1 +
 1 file changed, 1 insertion(+)

diff --git a/css/main.css b/css/main.css
index d5d5e77165..b3e699fa8a 100644
--- a/css/main.css
+++ b/css/main.css
@@ -581,6 +581,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 
 #chat-input textarea {
     padding: 0.65rem 2.5rem;
+    border: 0;
 }
 
 #chat-input textarea::placeholder {

From 690d693913f68d25d08fd74db902495766c12e5e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 4 May 2025 18:04:52 -0700
Subject: [PATCH 0446/1701] UI: Add padding to only show the last message/reply
 after sending a message

To avoid scrolling
---
 css/chat_style-Dark.css            |  3 ++-
 css/chat_style-TheEncrypted777.css |  3 ++-
 css/chat_style-cai-chat-square.css |  3 ++-
 css/chat_style-cai-chat.css        |  3 ++-
 css/chat_style-messenger.css       |  3 ++-
 css/chat_style-wpp.css             |  3 ++-
 css/html_instruct_style.css        |  4 ----
 css/main.css                       |  4 ----
 js/main.js                         | 10 ++++++++++
 modules/html_generator.py          |  9 +++------
 10 files changed, 25 insertions(+), 20 deletions(-)

diff --git a/css/chat_style-Dark.css b/css/chat_style-Dark.css
index 3b2bd38556..1ad46bc0dc 100644
--- a/css/chat_style-Dark.css
+++ b/css/chat_style-Dark.css
@@ -2,7 +2,8 @@
     display: grid;
     align-items: start;
     grid-template-columns: 60px minmax(0, 1fr);
-    padding-bottom: 28px;
+    padding-bottom: 14px;
+    padding-top: 14px;
     font-size: 18px;
     font-family: Roboto, Arial, sans-serif; /* Modern font */
     line-height: 1.5;
diff --git a/css/chat_style-TheEncrypted777.css b/css/chat_style-TheEncrypted777.css
index 25d26db81a..9e1230b7d2 100644
--- a/css/chat_style-TheEncrypted777.css
+++ b/css/chat_style-TheEncrypted777.css
@@ -4,7 +4,8 @@
     display: grid;
     align-items: start;
     grid-template-columns: 60px minmax(0, 1fr);
-    padding-bottom: 28px;
+    padding-bottom: 14px;
+    padding-top: 14px;
     font-size: 18px;
     font-family: 'Noto Sans', Arial, sans-serif;
     line-height: 1.428571429;
diff --git a/css/chat_style-cai-chat-square.css b/css/chat_style-cai-chat-square.css
index 854fff607c..015f6927c6 100644
--- a/css/chat_style-cai-chat-square.css
+++ b/css/chat_style-cai-chat-square.css
@@ -16,6 +16,7 @@
 }
 
 .message {
-    padding-bottom: 2em;
+    padding-bottom: 1em;
+    padding-top: 1em;
     grid-template-columns: 70px minmax(0, 1fr);
 }
diff --git a/css/chat_style-cai-chat.css b/css/chat_style-cai-chat.css
index 223f615087..0e91101f00 100644
--- a/css/chat_style-cai-chat.css
+++ b/css/chat_style-cai-chat.css
@@ -2,7 +2,8 @@
     display: grid;
     align-items: start;
     grid-template-columns: 60px minmax(0, 1fr);
-    padding-bottom: 2em;
+    padding-bottom: 1em;
+    padding-top: 1em;
     font-size: 15px;
     font-family: 'Noto Sans', Helvetica, Arial, sans-serif;
     line-height: 22.5px !important;
diff --git a/css/chat_style-messenger.css b/css/chat_style-messenger.css
index f0fd15788b..6518d6ca71 100644
--- a/css/chat_style-messenger.css
+++ b/css/chat_style-messenger.css
@@ -1,5 +1,6 @@
 .message {
-    padding-bottom: 25px;
+    padding-bottom: 12.5px;
+    padding-top: 12.5px;
     font-size: 15px;
     font-family: 'Noto Sans', Helvetica, Arial, sans-serif;
     line-height: 1.428571429;
diff --git a/css/chat_style-wpp.css b/css/chat_style-wpp.css
index 30ca61f335..1442dd0a43 100644
--- a/css/chat_style-wpp.css
+++ b/css/chat_style-wpp.css
@@ -1,5 +1,6 @@
 .message {
-    padding-bottom: 25px;
+    padding-bottom: 12.5px;
+    padding-top: 12.5px;
     font-size: 15px;
     font-family: 'Noto Sans', Helvetica, Arial, sans-serif;
     line-height: 1.428571429;
diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index b98544a1dc..f4339311db 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -8,10 +8,6 @@
     padding-top: 0 !important;
 }
 
-.chat > .messages > :last-child {
-    margin-bottom: 1.7rem !important;
-}
-
 .chat .message-body p, .chat .message-body li {
     font-size: 1rem !important;
     line-height: 28px !important;
diff --git a/css/main.css b/css/main.css
index b3e699fa8a..9915735df4 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1375,7 +1375,3 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     50% { opacity: 1; }
     100% { opacity: 0.6; }
 }
-
-.streaming {
-    min-height: 70vh;
-}
diff --git a/js/main.js b/js/main.js
index 408815db66..e661178863 100644
--- a/js/main.js
+++ b/js/main.js
@@ -150,6 +150,16 @@ const observer = new MutationObserver(function(mutations) {
   if (!isScrolled && targetElement.scrollTop !== targetElement.scrollHeight) {
     targetElement.scrollTop = targetElement.scrollHeight;
   }
+
+  const chatElement = document.getElementById("chat");
+  if (chatElement) {
+    const messagesContainer = chatElement.querySelector(".messages");
+    const lastChild = messagesContainer?.lastElementChild;
+    const prevSibling = lastChild?.previousElementSibling;
+    if (lastChild && prevSibling) {
+      lastChild.style.minHeight = `calc(100vh - ${prevSibling.offsetHeight}px - 102px)`;
+    }
+  }
 });
 
 // Configure the observer to watch for changes in the subtree and attributes
diff --git a/modules/html_generator.py b/modules/html_generator.py
index a6f5f930f0..67d15b6e98 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -365,9 +365,8 @@ def generate_instruct_html(history):
                 f'
            '
             )
 
-        streaming_class = " streaming" if i == len(history["visible"]) - 1 else ""
         output += (
-            f'
            '
             f'
            '
             f'
            {converted_visible[1]}
            '
@@ -415,9 +414,8 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
                 f'
            '
             )
 
-        streaming_class = " streaming" if i == len(history["visible"]) - 1 else ""
         output += (
-            f'
            '
             f'
            {img_bot}
            '
             f'
            '
@@ -454,9 +452,8 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
                 f'
            '
             )
 
-        streaming_class = " streaming" if i == len(history["visible"]) - 1 else ""
         output += (
-            f'
            '
             f'
            '
             f'
            {converted_visible[1]}
            '

From 2da197bba4b1547d51086e312d877d942e810be2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 4 May 2025 18:29:05 -0700
Subject: [PATCH 0447/1701] Refinement after previous commit

---
 js/main.js | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index e661178863..205cf88e8d 100644
--- a/js/main.js
+++ b/js/main.js
@@ -157,7 +157,7 @@ const observer = new MutationObserver(function(mutations) {
     const lastChild = messagesContainer?.lastElementChild;
     const prevSibling = lastChild?.previousElementSibling;
     if (lastChild && prevSibling) {
-      lastChild.style.minHeight = `calc(100vh - ${prevSibling.offsetHeight}px - 102px)`;
+      lastChild.style.minHeight = `calc(max(70vh, 100vh - ${prevSibling.offsetHeight}px - 102px))`;
     }
   }
 });

From d0211afb3c513bde0d8662bd686ddb0dc87354cd Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 4 May 2025 18:52:01 -0700
Subject: [PATCH 0448/1701] Save the chat history right after sending a message

---
 modules/chat.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index 98913d5c8f..feac6bddc6 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -483,6 +483,8 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
     history = state['history']
     for i, history in enumerate(generate_chat_reply(text, state, regenerate, _continue, loading_message=True, for_ui=True)):
         yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu']), history
+        if i == 0:
+            save_history(history, state['unique_id'], state['character_menu'], state['mode'])
 
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
 

From df7bb0db1fe6478d037debf73272b10cef1f75c7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 4 May 2025 20:03:55 -0700
Subject: [PATCH 0449/1701] Rename --n-gpu-layers to --gpu-layers

---
 modules/shared.py        | 2 +-
 modules/ui_model_menu.py | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index cfedb992ee..b952c4a1d0 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -120,7 +120,7 @@
 group.add_argument('--batch-size', type=int, default=256, help='Maximum number of prompt tokens to batch together when calling llama_eval.')
 group.add_argument('--no-mmap', action='store_true', help='Prevent mmap from being used.')
 group.add_argument('--mlock', action='store_true', help='Force the system to keep the model in RAM.')
-group.add_argument('--n-gpu-layers', type=int, default=0, help='Number of layers to offload to the GPU.')
+group.add_argument('--gpu-layers', '--n-gpu-layers', type=int, default=0, metavar='N', help='Number of layers to offload to the GPU.')
 group.add_argument('--tensor-split', type=str, default=None, help='Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.')
 group.add_argument('--numa', action='store_true', help='Activate NUMA task allocation for llama.cpp.')
 group.add_argument('--no-kv-offload', action='store_true', help='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 2c593df66f..943645cf69 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -49,7 +49,7 @@ def create_ui():
                 with gr.Blocks():
                     with gr.Row():
                         with gr.Column():
-                            shared.gradio['n_gpu_layers'] = gr.Slider(label="n-gpu-layers", minimum=0, maximum=256, value=shared.args.n_gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
+                            shared.gradio['n_gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=256, value=shared.args.n_gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['threads'] = gr.Slider(label="threads", minimum=0, step=1, maximum=256, value=shared.args.threads)
                             shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
                             shared.gradio['batch_size'] = gr.Slider(label="batch_size", minimum=1, maximum=4096, step=1, value=shared.args.batch_size)

From f3da45f65d76f8c48fd95678ecc841afb0ddd04e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 4 May 2025 20:37:15 -0700
Subject: [PATCH 0450/1701] ExLlamaV3_HF: Change max_chunk_size to 256

---
 modules/exllamav3_hf.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index 12b22f646e..417df4734a 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -119,7 +119,7 @@ def __call__(self, *args, **kwargs):
         reset = True
 
         # Maximum number of tokens to process in a single forward pass
-        max_chunk_size = 2048
+        max_chunk_size = 256
 
         # Make the forward call
         if labels is None:

From b817bb33fd7b26a24c81798dabb36af4620d4a53 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 5 May 2025 04:54:25 -0700
Subject: [PATCH 0451/1701] Minor fix after
 df7bb0db1fe6478d037debf73272b10cef1f75c7

---
 modules/llama_cpp_server.py | 2 +-
 modules/ui_model_menu.py    | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 7244001a80..0ddb3fff2f 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -256,7 +256,7 @@ def _start_server(self):
             self.server_path,
             "--model", self.model_path,
             "--ctx-size", str(shared.args.ctx_size),
-            "--n-gpu-layers", str(shared.args.n_gpu_layers),
+            "--gpu-layers", str(shared.args.gpu_layers),
             "--batch-size", str(shared.args.batch_size),
             "--port", str(self.port),
         ]
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 943645cf69..e05d225664 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -49,7 +49,7 @@ def create_ui():
                 with gr.Blocks():
                     with gr.Row():
                         with gr.Column():
-                            shared.gradio['n_gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=256, value=shared.args.n_gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
+                            shared.gradio['n_gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=256, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['threads'] = gr.Slider(label="threads", minimum=0, step=1, maximum=256, value=shared.args.threads)
                             shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
                             shared.gradio['batch_size'] = gr.Slider(label="batch_size", minimum=1, maximum=4096, step=1, value=shared.args.batch_size)

From 475e012ee8e0cbeb53fade01359cd649b9b5d470 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 5 May 2025 06:16:11 -0700
Subject: [PATCH 0452/1701] UI: Improve the light theme colors

---
 css/main.css  | 3 ++-
 modules/ui.py | 2 +-
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/css/main.css b/css/main.css
index 9915735df4..b1d6e3456f 100644
--- a/css/main.css
+++ b/css/main.css
@@ -2,7 +2,7 @@
     --darker-gray: #202123;
     --dark-gray: #343541;
     --light-gray: #444654;
-    --light-theme-gray: #f5f5f5;
+    --light-theme-gray: #f3f4f6;
     --border-color-dark: #525252;
     --header-width: 112px;
     --selected-item-color-dark: #32333e;
@@ -580,6 +580,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 #chat-input textarea {
+    background: var(--light-theme-gray);
     padding: 0.65rem 2.5rem;
     border: 0;
 }
diff --git a/modules/ui.py b/modules/ui.py
index fb016f8769..d08c1435e5 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -61,7 +61,7 @@
         background_fill_primary_dark='var(--darker-gray)',
         body_background_fill="white",
         block_background_fill="transparent",
-        body_text_color="#333",
+        body_text_color='rgb(64, 64, 64)',
         button_secondary_background_fill="#f4f4f4",
         button_secondary_border_color="var(--border-color-primary)",
 

From 6001d279c64d92c1f2a312142e41119807694729 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 5 May 2025 07:42:13 -0700
Subject: [PATCH 0453/1701] Light theme improvement

---
 css/main.css | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index b1d6e3456f..38585a1c9a 100644
--- a/css/main.css
+++ b/css/main.css
@@ -979,6 +979,11 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     cursor: pointer;
 }
 
+#past-chats .selected,
+#past-chats label:hover {
+    background-color: rgb(224, 224, 224) !important;
+}
+
 #past-chats-buttons,
 #delete-chat-row,
 #rename-row {
@@ -987,7 +992,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     gap: 9px;
 }
 
-
 #past-chats-row,
 #chat-controls {
     width: 260px;

From 967b70327ea10a9c5cc7c932583993687b9d4ba7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 5 May 2025 07:56:47 -0700
Subject: [PATCH 0454/1701] Light theme improvement

---
 css/html_instruct_style.css |  2 +-
 css/main.css                | 12 ++++++------
 2 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index f4339311db..fb98433845 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -42,7 +42,7 @@
 }
 
 .chat .user-message {
-    background: #f5f5f5;
+    background: #f3f4f6;
     padding: 1.5rem 1rem;
     padding-bottom: 2rem;
     border-radius: 0;
diff --git a/css/main.css b/css/main.css
index 38585a1c9a..d6a0d22054 100644
--- a/css/main.css
+++ b/css/main.css
@@ -2,7 +2,7 @@
     --darker-gray: #202123;
     --dark-gray: #343541;
     --light-gray: #444654;
-    --light-theme-gray: #f3f4f6;
+    --light-theme-gray: #f9fbff;
     --border-color-dark: #525252;
     --header-width: 112px;
     --selected-item-color-dark: #32333e;
@@ -580,7 +580,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 #chat-input textarea {
-    background: var(--light-theme-gray);
+    background: #f3f4f6;
     padding: 0.65rem 2.5rem;
     border: 0;
 }
@@ -981,7 +981,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 
 #past-chats .selected,
 #past-chats label:hover {
-    background-color: rgb(224, 224, 224) !important;
+    background-color: #dbeafe !important;
 }
 
 #past-chats-buttons,
@@ -1123,8 +1123,8 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     border: 0 !important;
 }
 
-.dark #past-chats .selected,
-.dark #past-chats label:hover {
+.dark gradio-app .gradio-container.gradio-container-4-37-2 .contain #past-chats .selected,
+.dark gradio-app .gradio-container.gradio-container-4-37-2 .contain #past-chats label:hover {
     background-color: var(--selected-item-color-dark) !important;
 }
 
@@ -1161,7 +1161,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 .header_bar button.selected {
-    background: #E0E0E0;
+    background: #dbeafe;
 }
 
 #chat-controls,

From bf5290bc0ff15f6894a4eb5785e8df60831ecb25 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 5 May 2025 08:04:12 -0700
Subject: [PATCH 0455/1701] Fix the hover menu in light theme

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index d6a0d22054..59165a6229 100644
--- a/css/main.css
+++ b/css/main.css
@@ -742,7 +742,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 
 .hover-menu button {
     width: 100%;
-    background: transparent !important;
+    background: white !important;
     border-radius: 0 !important;
     justify-content: space-between;
     margin: 0 !important;

From 53d8e4650202f5891364197011098b3af34fe6ac Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 5 May 2025 12:28:17 -0700
Subject: [PATCH 0456/1701] Ensure environment isolation in portable installs

---
 start_linux.sh    | 9 +++++----
 start_macos.sh    | 9 +++++----
 start_windows.bat | 9 +++++----
 3 files changed, 15 insertions(+), 12 deletions(-)

diff --git a/start_linux.sh b/start_linux.sh
index c74f127258..e2b005587e 100755
--- a/start_linux.sh
+++ b/start_linux.sh
@@ -1,5 +1,10 @@
 #!/usr/bin/env bash
 
+# environment isolation
+export PYTHONNOUSERSITE=1
+unset PYTHONPATH
+unset PYTHONHOME
+
 cd "$(dirname "${BASH_SOURCE[0]}")"
 
 # Portable install case
@@ -61,10 +66,6 @@ if [ ! -e "$INSTALL_ENV_DIR/bin/python" ]; then
     exit
 fi
 
-# environment isolation
-export PYTHONNOUSERSITE=1
-unset PYTHONPATH
-unset PYTHONHOME
 export CUDA_PATH="$INSTALL_ENV_DIR"
 export CUDA_HOME="$CUDA_PATH"
 
diff --git a/start_macos.sh b/start_macos.sh
index 7a060ba601..bff11bc177 100755
--- a/start_macos.sh
+++ b/start_macos.sh
@@ -1,5 +1,10 @@
 #!/bin/bash
 
+# environment isolation
+export PYTHONNOUSERSITE=1
+unset PYTHONPATH
+unset PYTHONHOME
+
 cd "$(dirname "${BASH_SOURCE[0]}")"
 
 # Portable install case
@@ -61,10 +66,6 @@ if [ ! -e "$INSTALL_ENV_DIR/bin/python" ]; then
     exit
 fi
 
-# environment isolation
-export PYTHONNOUSERSITE=1
-unset PYTHONPATH
-unset PYTHONHOME
 export CUDA_PATH="$INSTALL_ENV_DIR"
 export CUDA_HOME="$CUDA_PATH"
 
diff --git a/start_windows.bat b/start_windows.bat
index 1616ee27db..f5e66ec281 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -1,6 +1,11 @@
 @echo off
 setlocal enabledelayedexpansion
 
+@rem environment isolation
+set PYTHONNOUSERSITE=1
+set PYTHONPATH=
+set PYTHONHOME=
+
 cd /D "%~dp0"
 
 @rem Portable install case
@@ -87,10 +92,6 @@ if not exist "%INSTALL_ENV_DIR%" (
 @rem check if conda environment was actually created
 if not exist "%INSTALL_ENV_DIR%\python.exe" ( echo. && echo Conda environment is empty. && goto end )
 
-@rem environment isolation
-set PYTHONNOUSERSITE=1
-set PYTHONPATH=
-set PYTHONHOME=
 set "CUDA_PATH=%INSTALL_ENV_DIR%"
 set "CUDA_HOME=%CUDA_PATH%"
 

From 8137eb8ef46ac6950cb96094e3cc30b0a72dee76 Mon Sep 17 00:00:00 2001
From: mamei16 
Date: Mon, 5 May 2025 23:05:23 +0200
Subject: [PATCH 0457/1701] Dynamic Chat Message UI Update Speed (#6952)

---
 modules/shared.py                |  1 -
 modules/text_generation.py       | 18 ++++++++----------
 modules/ui.py                    |  1 -
 modules/ui_parameters.py         |  2 --
 user_data/settings-template.yaml |  1 -
 5 files changed, 8 insertions(+), 15 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index b952c4a1d0..b4dfbfd1e1 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -47,7 +47,6 @@
     'max_new_tokens_max': 4096,
     'prompt_lookup_num_tokens': 0,
     'max_tokens_second': 0,
-    'max_updates_second': 12,
     'auto_max_new_tokens': True,
     'ban_eos_token': False,
     'add_bos_token': True,
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 8fd65dc423..7e48a2f680 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -64,41 +64,39 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
             all_stop_strings += st
 
     shared.stop_everything = False
-    last_update = -1
     reply = ''
     is_stream = state['stream']
     if len(all_stop_strings) > 0 and not state['stream']:
         state = copy.deepcopy(state)
         state['stream'] = True
 
-    min_update_interval = 0
-    if state.get('max_updates_second', 0) > 0:
-        min_update_interval = 1 / state['max_updates_second']
-
     # Generate
+    last_update = -1
+    latency_threshold = 1 / 1000
     for reply in generate_func(question, original_question, state, stopping_strings, is_chat=is_chat):
+        cur_time = time.monotonic()
         reply, stop_found = apply_stopping_strings(reply, all_stop_strings)
         if escape_html:
             reply = html.escape(reply)
 
         if is_stream:
-            cur_time = time.time()
-
             # Limit number of tokens/second to make text readable in real time
             if state['max_tokens_second'] > 0:
                 diff = 1 / state['max_tokens_second'] - (cur_time - last_update)
                 if diff > 0:
                     time.sleep(diff)
 
-                last_update = time.time()
+                last_update = time.monotonic()
                 yield reply
 
             # Limit updates to avoid lag in the Gradio UI
             # API updates are not limited
             else:
-                if cur_time - last_update > min_update_interval:
-                    last_update = cur_time
+                # If 'generate_func' takes less than 0.001 seconds to yield the next token
+                # (equivalent to more than 1000 tok/s), assume that the UI is lagging behind and skip yielding
+                if (cur_time - last_update) > latency_threshold:
                     yield reply
+                last_update = time.monotonic()
 
         if stop_found or (state['max_tokens_second'] > 0 and shared.stop_everything):
             break
diff --git a/modules/ui.py b/modules/ui.py
index d08c1435e5..b3d4bccf07 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -192,7 +192,6 @@ def list_interface_input_elements():
         'max_new_tokens',
         'prompt_lookup_num_tokens',
         'max_tokens_second',
-        'max_updates_second',
         'do_sample',
         'dynamic_temperature',
         'temperature_last',
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 733d090101..84f9fbfc24 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -71,8 +71,6 @@ def create_ui(default_preset):
                             shared.gradio['max_new_tokens'] = gr.Slider(minimum=shared.settings['max_new_tokens_min'], maximum=shared.settings['max_new_tokens_max'], value=shared.settings['max_new_tokens'], step=1, label='max_new_tokens', info='⚠️ Setting this too high can cause prompt truncation.')
                             shared.gradio['prompt_lookup_num_tokens'] = gr.Slider(value=shared.settings['prompt_lookup_num_tokens'], minimum=0, maximum=10, step=1, label='prompt_lookup_num_tokens', info='Activates Prompt Lookup Decoding.')
                             shared.gradio['max_tokens_second'] = gr.Slider(value=shared.settings['max_tokens_second'], minimum=0, maximum=20, step=1, label='Maximum tokens/second', info='To make text readable in real time.')
-                            shared.gradio['max_updates_second'] = gr.Slider(value=shared.settings['max_updates_second'], minimum=0, maximum=24, step=1, label='Maximum UI updates/second', info='Set this if you experience lag in the UI during streaming.')
-
                 with gr.Column():
                     with gr.Row():
                         with gr.Column():
diff --git a/user_data/settings-template.yaml b/user_data/settings-template.yaml
index ce0f77e142..db481e8499 100644
--- a/user_data/settings-template.yaml
+++ b/user_data/settings-template.yaml
@@ -18,7 +18,6 @@ max_new_tokens_min: 1
 max_new_tokens_max: 4096
 prompt_lookup_num_tokens: 0
 max_tokens_second: 0
-max_updates_second: 12
 auto_max_new_tokens: true
 ban_eos_token: false
 add_bos_token: true

From 85bf2e15b98117ef5630e81bf4a002440fffe2c2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 5 May 2025 14:14:48 -0700
Subject: [PATCH 0458/1701] API: Remove obsolete multimodal extension handling

Multimodal support will be added back once it's implemented in llama-server.
---
 extensions/openai/completions.py | 24 ------------------------
 1 file changed, 24 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 75e2cc118f..46c761996b 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -96,30 +96,6 @@ def convert_history(history):
     user_input_last = True
     system_message = ""
 
-    # Multimodal: convert OpenAI format to multimodal extension format
-    if any('content' in entry and isinstance(entry['content'], list) for entry in history):
-        new_history = []
-        for entry in history:
-            if isinstance(entry['content'], list):
-                for item in entry['content']:
-                    if not isinstance(item, dict):
-                        continue
-
-                    image_url = None
-                    content = None
-                    if item['type'] == 'image_url' and isinstance(item['image_url'], dict):
-                        image_url = item['image_url']['url']
-                    elif item['type'] == 'text' and isinstance(item['text'], str):
-                        content = item['text']
-                    if image_url:
-                        new_history.append({"image_url": image_url, "role": "user"})
-                    if content:
-                        new_history.append({"content": content, "role": "user"})
-            else:
-                new_history.append(entry)
-
-        history = new_history
-
     for entry in history:
         if "image_url" in entry:
             image_url = entry['image_url']

From f82667f0b4c0824420a6637efee3c680ddbe25f3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 5 May 2025 14:17:00 -0700
Subject: [PATCH 0459/1701] Remove more multimodal extension references

---
 extensions/openai/completions.py | 28 +---------------------------
 1 file changed, 1 insertion(+), 27 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 46c761996b..a7d8b4e493 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -1,13 +1,8 @@
-import base64
 import copy
-import re
 import time
 from collections import deque
-from io import BytesIO
 
-import requests
 import tiktoken
-from PIL import Image
 
 from extensions.openai.errors import InvalidRequestError
 from extensions.openai.utils import debug_msg
@@ -97,28 +92,7 @@ def convert_history(history):
     system_message = ""
 
     for entry in history:
-        if "image_url" in entry:
-            image_url = entry['image_url']
-            if "base64" in image_url:
-                image_url = re.sub('^data:image/.+;base64,', '', image_url)
-                img = Image.open(BytesIO(base64.b64decode(image_url)))
-            else:
-                try:
-                    my_res = requests.get(image_url)
-                    img = Image.open(BytesIO(my_res.content))
-                except Exception:
-                    raise 'Image cannot be loaded from the URL!'
-
-            buffered = BytesIO()
-            if img.mode in ("RGBA", "P"):
-                img = img.convert("RGB")
-
-            img.save(buffered, format="JPEG")
-            img_str = base64.b64encode(buffered.getvalue()).decode('utf-8')
-            content = f''
-        else:
-            content = entry["content"]
-
+        content = entry["content"]
         role = entry["role"]
 
         if role == "user":

From 941e0663da48345150ae77d7c6b6eb54e21d671d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 5 May 2025 14:18:05 -0700
Subject: [PATCH 0460/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 8a7b24674a..6cc84c5005 100644
--- a/README.md
+++ b/README.md
@@ -60,7 +60,7 @@ If you ever need to install something manually in the `installer_files` environm
 * There is no need to run any of those scripts (`start_`, `update_wizard_`, or `cmd_`) as admin/root.
 * To install the requirements for extensions, you can use the `extensions_reqs` script for your OS. At the end, this script will install the main requirements for the project to make sure that they take precedence in case of version conflicts.
 * For additional instructions about AMD and WSL setup, consult [the documentation](https://github.com/oobabooga/text-generation-webui/wiki).
-* For automated installation, you can use the `GPU_CHOICE`, `USE_CUDA118`, `LAUNCH_AFTER_INSTALL`, and `INSTALL_EXTENSIONS` environment variables. For instance: `GPU_CHOICE=A USE_CUDA118=FALSE LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh`.
+* For automated installation, you can use the `GPU_CHOICE`, `LAUNCH_AFTER_INSTALL`, and `INSTALL_EXTENSIONS` environment variables. For instance: `GPU_CHOICE=A LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh`.
 
 ### Manual installation using Conda
 

From 987505ead345b0e113d636311f6a5faa4fcbe986 Mon Sep 17 00:00:00 2001
From: Evgenii Novikov 
Date: Tue, 6 May 2025 00:03:33 +0200
Subject: [PATCH 0461/1701] docker: Fix app uid typo in cpu docker compose
 (#6957)

---
 docker/cpu/docker-compose.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/docker/cpu/docker-compose.yml b/docker/cpu/docker-compose.yml
index c9d415ae22..9aba314a4e 100644
--- a/docker/cpu/docker-compose.yml
+++ b/docker/cpu/docker-compose.yml
@@ -22,7 +22,7 @@ services:
         TORCH_CUDA_ARCH_LIST: ${TORCH_CUDA_ARCH_LIST:-7.5} 
         BUILD_EXTENSIONS: ${BUILD_EXTENSIONS:-}
         APP_GID: ${APP_GID:-6972} 
-        APP_UID: ${APP_UID-6972} 
+        APP_UID: ${APP_UID:-6972} 
     env_file: .env
     user: "${APP_RUNTIME_UID:-6972}:${APP_RUNTIME_GID:-6972}"
     ports:

From 99bd66445f90df58a3f3832b35cca94dc397d1be Mon Sep 17 00:00:00 2001
From: Alireza Ghasemi 
Date: Tue, 6 May 2025 00:04:06 +0200
Subject: [PATCH 0462/1701] SuperboogaV2: minor update to avoid json
 serialization errors #6945

---
 extensions/superboogav2/chromadb.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/extensions/superboogav2/chromadb.py b/extensions/superboogav2/chromadb.py
index 6e93dd9238..f4f7782112 100644
--- a/extensions/superboogav2/chromadb.py
+++ b/extensions/superboogav2/chromadb.py
@@ -292,6 +292,8 @@ def _get_documents_up_to_token_count(self, documents: list[str], max_token_count
 
         for doc in documents:
             doc_tokens = encode(doc)[0]
+            if isinstance(doc_tokens, np.ndarray):
+                doc_tokens = doc_tokens.tolist()
             doc_token_count = len(doc_tokens)
             if current_token_count + doc_token_count > max_token_count:
                 # If adding this document would exceed the max token count,

From 76f947e3cf1c71e4105f708f02b2ca163a69987c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 5 May 2025 15:58:29 -0700
Subject: [PATCH 0463/1701] UI: Minor style change

---
 css/main.css | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/css/main.css b/css/main.css
index 59165a6229..520ff97212 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1380,3 +1380,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     50% { opacity: 1; }
     100% { opacity: 0.6; }
 }
+
+strong {
+    font-weight: bold;
+}

From 530223bf0b196257e41ec948c2e92e1c3e507e9f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 5 May 2025 16:00:49 -0700
Subject: [PATCH 0464/1701] UI: Fix the hover menu colors

---
 css/main.css | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/css/main.css b/css/main.css
index 520ff97212..b8ba82566c 100644
--- a/css/main.css
+++ b/css/main.css
@@ -761,6 +761,10 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     background: var(--button-secondary-background-fill-hover) !important;
 }
 
+.dark .hover-menu button:hover {
+    background: var(--selected-item-color-dark) !important;
+}
+
 .transparent-substring {
     opacity: 0.333;
 }
@@ -1109,12 +1113,9 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     color: #9ca3af;
 }
 
-.dark .hover-menu {
-    background-color: var(--darker-gray);
-}
-
 .dark .hover-menu button {
     border-color: var(--border-color-primary);
+    background-color: var(--darker-gray) !important;
 }
 
 .dark #chat-controls,

From 4e8f628d3c206e8362cea5b5f7557abe33351bc0 Mon Sep 17 00:00:00 2001
From: Evgenii Novikov 
Date: Tue, 6 May 2025 01:05:15 +0200
Subject: [PATCH 0465/1701] docker: App uid typo in other docker composes
 (#6958)

---
 docker/amd/docker-compose.yml    | 2 +-
 docker/intel/docker-compose.yml  | 2 +-
 docker/nvidia/docker-compose.yml | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/docker/amd/docker-compose.yml b/docker/amd/docker-compose.yml
index 4709ae941c..8866e9edb8 100644
--- a/docker/amd/docker-compose.yml
+++ b/docker/amd/docker-compose.yml
@@ -22,7 +22,7 @@ services:
         TORCH_CUDA_ARCH_LIST: ${TORCH_CUDA_ARCH_LIST:-7.5} 
         BUILD_EXTENSIONS: ${BUILD_EXTENSIONS:-}
         APP_GID: ${APP_GID:-6972} 
-        APP_UID: ${APP_UID-6972} 
+        APP_UID: ${APP_UID:-6972} 
     env_file: .env
     user: "${APP_RUNTIME_UID:-6972}:${APP_RUNTIME_GID:-6972}"
     ports:
diff --git a/docker/intel/docker-compose.yml b/docker/intel/docker-compose.yml
index 31e9dde015..78e0669804 100644
--- a/docker/intel/docker-compose.yml
+++ b/docker/intel/docker-compose.yml
@@ -22,7 +22,7 @@ services:
         TORCH_CUDA_ARCH_LIST: ${TORCH_CUDA_ARCH_LIST:-7.5} 
         BUILD_EXTENSIONS: ${BUILD_EXTENSIONS:-}
         APP_GID: ${APP_GID:-6972} 
-        APP_UID: ${APP_UID-6972} 
+        APP_UID: ${APP_UID:-6972} 
     env_file: .env
     user: "${APP_RUNTIME_UID:-6972}:${APP_RUNTIME_GID:-6972}"
     ports:
diff --git a/docker/nvidia/docker-compose.yml b/docker/nvidia/docker-compose.yml
index 835dd8384b..0392078e06 100644
--- a/docker/nvidia/docker-compose.yml
+++ b/docker/nvidia/docker-compose.yml
@@ -22,7 +22,7 @@ services:
         TORCH_CUDA_ARCH_LIST: ${TORCH_CUDA_ARCH_LIST:-7.5} 
         BUILD_EXTENSIONS: ${BUILD_EXTENSIONS:-}
         APP_GID: ${APP_GID:-6972} 
-        APP_UID: ${APP_UID-6972} 
+        APP_UID: ${APP_UID:-6972} 
     env_file: .env
     user: "${APP_RUNTIME_UID:-6972}:${APP_RUNTIME_GID:-6972}"
     ports:

From cbef35054cb598b033e17b0442e8dad2da6873c4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 5 May 2025 17:46:09 -0700
Subject: [PATCH 0466/1701] UI: CSS fix

---
 css/main.css | 1 +
 1 file changed, 1 insertion(+)

diff --git a/css/main.css b/css/main.css
index b8ba82566c..746f1f9eb9 100644
--- a/css/main.css
+++ b/css/main.css
@@ -426,6 +426,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 .chat > .messages {
     display: flex;
     flex-direction: column;
+    min-height: calc(100vh - 102px);
 }
 
 .chat > .messages > :first-child {

From d1c0154d664e51d1ee6ea82d9c0e799d96367d4a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 6 May 2025 06:38:39 -0700
Subject: [PATCH 0467/1701] llama.cpp: Add top_n_sigma, fix typical_p in
 sampler priority

---
 modules/llama_cpp_server.py | 5 ++++-
 modules/presets.py          | 2 +-
 2 files changed, 5 insertions(+), 2 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 0ddb3fff2f..b9902cd7d3 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -66,6 +66,7 @@ def prepare_payload(self, state):
             "top_k": state["top_k"],
             "top_p": state["top_p"],
             "min_p": state["min_p"],
+            "top_n_sigma": state["top_n_sigma"] if state["top_n_sigma"] > 0 else -1,
             "tfs_z": state["tfs"],
             "typical_p": state["typical_p"],
             "repeat_penalty": state["repetition_penalty"],
@@ -102,8 +103,10 @@ def prepare_payload(self, state):
 
             penalty_found = False
             for s in samplers:
-                if s.strip() in ["dry", "top_k", "typ_p", "top_p", "min_p", "xtc", "temperature"]:
+                if s.strip() in ["dry", "top_k", "top_p", "top_n_sigma", "min_p", "temperature", "xtc"]:
                     filtered_samplers.append(s.strip())
+                elif s.strip() == "typical_p":
+                    filtered_samplers.append("typ_p")
                 elif not penalty_found and s.strip() == "repetition_penalty":
                     filtered_samplers.append("penalties")
                     penalty_found = True
diff --git a/modules/presets.py b/modules/presets.py
index 50d0f9850a..5a9a587315 100644
--- a/modules/presets.py
+++ b/modules/presets.py
@@ -52,7 +52,7 @@ def default_preset():
 
     if shared.args.portable:
         samplers = result['sampler_priority'].split('\n')
-        samplers = [sampler for sampler in samplers if sampler in ["dry", "top_k", "typ_p", "top_p", "min_p", "xtc", "temperature", "repetition_penalty"]]
+        samplers = [sampler for sampler in samplers if sampler in ["dry", "top_k", "top_p", "top_n_sigma", "min_p", "temperature", "xtc", "typical_p", "repetition_penalty"]]
         result['sampler_priority'] = '\n'.join(samplers)
 
     return result

From 89590adc14c941814c2d54795cfc78fab959d9e7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 6 May 2025 06:41:17 -0700
Subject: [PATCH 0468/1701] Update llama.cpp

---
 requirements/full/requirements.txt             |  4 ++--
 requirements/full/requirements_amd.txt         |  4 ++--
 requirements/full/requirements_amd_noavx2.txt  |  4 ++--
 requirements/full/requirements_apple_intel.txt |  4 ++--
 .../full/requirements_apple_silicon.txt        |  6 +++---
 requirements/full/requirements_cpu_only.txt    |  4 ++--
 .../full/requirements_cpu_only_noavx2.txt      |  4 ++--
 requirements/full/requirements_noavx2.txt      |  4 ++--
 requirements/portable/requirements.txt         |  4 ++--
 requirements/portable/requirements_amd.txt     | 18 ------------------
 .../portable/requirements_amd_noavx2.txt       | 18 ------------------
 .../portable/requirements_apple_intel.txt      |  4 ++--
 .../portable/requirements_apple_silicon.txt    |  6 +++---
 .../portable/requirements_cpu_only.txt         |  4 ++--
 .../portable/requirements_cpu_only_noavx2.txt  |  4 ++--
 requirements/portable/requirements_noavx2.txt  |  4 ++--
 requirements/portable/requirements_vulkan.txt  |  4 ++--
 .../portable/requirements_vulkan_noavx2.txt    |  4 ++--
 18 files changed, 34 insertions(+), 70 deletions(-)
 delete mode 100644 requirements/portable/requirements_amd.txt
 delete mode 100644 requirements/portable/requirements_amd_noavx2.txt

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index c0ace41b0a..a60ea7b4c6 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 24eeee6aa9..431cd74032 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 99716f3cfb..0c581f869e 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index cc747edb81..f7213efef4 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 67b3260edc..4aac3dea31 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -29,8 +29,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 47ad5759a4..ac277d6103 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 334f11df64..cc412d3300 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 3575d35269..78265f1a83 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index c720daa7ef..1240d335c0 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
deleted file mode 100644
index 7d9c00c079..0000000000
--- a/requirements/portable/requirements_amd.txt
+++ /dev/null
@@ -1,18 +0,0 @@
-fastapi==0.112.4
-gradio==4.37.*
-jinja2==3.1.6
-markdown
-numpy==1.26.*
-pydantic==2.8.2
-pyyaml
-requests
-rich
-tqdm
-
-# API
-flask_cloudflared==0.0.14
-sse-starlette==1.6.5
-tiktoken
-
-# AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+rocm6.1.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
deleted file mode 100644
index d718c1b1c7..0000000000
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ /dev/null
@@ -1,18 +0,0 @@
-fastapi==0.112.4
-gradio==4.37.*
-jinja2==3.1.6
-markdown
-numpy==1.26.*
-pydantic==2.8.2
-pyyaml
-requests
-rich
-tqdm
-
-# API
-flask_cloudflared==0.0.14
-sse-starlette==1.6.5
-tiktoken
-
-# AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+rocm6.1.2avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 9e184b5359..6b165b7cd1 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index ec059716b9..1b2b5cf2b3 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -15,6 +15,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index d473b8241d..2793d74395 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index d3fffb4341..6d7316a6f5 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index cdfa6a015f..e56eba0886 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 1a7ce6ed01..a7f8c703db 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 4737321dd5..5b427fd208 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.8.0/llama_cpp_binaries-0.8.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 605cc9ab14533dd20cc11363f020fb9947cfb723 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 6 May 2025 06:42:15 -0700
Subject: [PATCH 0469/1701] Update exllamav3

---
 requirements/full/requirements.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_noavx2.txt        | 4 ++--
 4 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index a60ea7b4c6..3b50c6744d 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -32,8 +32,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a7/exllamav3-0.0.1a7+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a7/exllamav3-0.0.1a7+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index f7213efef4..ba23ea9c34 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -31,5 +31,5 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a7/exllamav3-0.0.1a7-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 4aac3dea31..c245ab747b 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -32,5 +32,5 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a7/exllamav3-0.0.1a7-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 78265f1a83..d8bbf6d1ed 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -32,8 +32,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a6/exllamav3-0.0.1a6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a7/exllamav3-0.0.1a7+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a7/exllamav3-0.0.1a7+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 1927afe89457dce8eb805b2275ba7c8a9680a967 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 6 May 2025 08:18:49 -0700
Subject: [PATCH 0470/1701] Fix top_n_sigma not showing for llama.cpp

---
 modules/loaders.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/loaders.py b/modules/loaders.py
index 738198b1e0..217d569ce6 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -300,6 +300,7 @@ def transformers_samplers():
         'xtc_threshold',
         'xtc_probability',
         'tfs',
+        'top_n_sigma',
         'dry_multiplier',
         'dry_allowed_length',
         'dry_base',

From 05115e42ee1ab7a2848b883e469885ce9504f04a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 6 May 2025 08:27:21 -0700
Subject: [PATCH 0471/1701] Set top_n_sigma before temperature by default

---
 modules/presets.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/presets.py b/modules/presets.py
index 5a9a587315..cf70660551 100644
--- a/modules/presets.py
+++ b/modules/presets.py
@@ -46,7 +46,7 @@ def default_preset():
         'do_sample': True,
         'dynamic_temperature': False,
         'temperature_last': False,
-        'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ndry\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_n_sigma\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nmirostat\nxtc\nencoder_repetition_penalty\nno_repeat_ngram',
+        'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ndry\ntop_n_sigma\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nmirostat\nxtc\nencoder_repetition_penalty\nno_repeat_ngram',
         'dry_sequence_breakers': '"\\n", ":", "\\"", "*"',
     }
 

From c4f36db0d859e1819550e576a7fbd513c990c64d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 6 May 2025 08:41:13 -0700
Subject: [PATCH 0472/1701] llama.cpp: remove tfs (it doesn't get used)

---
 modules/llama_cpp_server.py | 1 -
 modules/loaders.py          | 1 -
 2 files changed, 2 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index b9902cd7d3..d8d2f61bdd 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -67,7 +67,6 @@ def prepare_payload(self, state):
             "top_p": state["top_p"],
             "min_p": state["min_p"],
             "top_n_sigma": state["top_n_sigma"] if state["top_n_sigma"] > 0 else -1,
-            "tfs_z": state["tfs"],
             "typical_p": state["typical_p"],
             "repeat_penalty": state["repetition_penalty"],
             "repeat_last_n": state["repetition_penalty_range"],
diff --git a/modules/loaders.py b/modules/loaders.py
index 217d569ce6..b29679bd47 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -299,7 +299,6 @@ def transformers_samplers():
         'typical_p',
         'xtc_threshold',
         'xtc_probability',
-        'tfs',
         'top_n_sigma',
         'dry_multiplier',
         'dry_allowed_length',

From 5ef564a22e8df21a7480d5c8d6e32919f35f14c7 Mon Sep 17 00:00:00 2001
From: Downtown-Case 
Date: Tue, 6 May 2025 15:03:33 -0500
Subject: [PATCH 0473/1701] Fix model config loading in shared.py for Python
 3.13 (#6961)

---
 modules/shared.py | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index b4dfbfd1e1..6fd4604c98 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -310,11 +310,13 @@ def load_user_config():
     add_extension('openai', last=True)
 
 # Load model-specific settings
-with Path(f'{args.model_dir}/config.yaml') as p:
-    if p.exists():
-        model_config = yaml.safe_load(open(p, 'r').read())
-    else:
-        model_config = {}
+p = Path(f'{args.model_dir}/config.yaml')
+if p.exists():
+    model_config = yaml.safe_load(open(p, 'r').read())
+else:
+    model_config = {}
+del p
+
 
 # Load custom model-specific settings
 user_config = load_user_config()

From e4fb2475d25e1dccfa39f5d943bcde61ef517245 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 6 May 2025 14:02:01 -0700
Subject: [PATCH 0474/1701] UI: Multiple small style improvements (light/dark
 themes)

---
 css/html_instruct_style.css |  2 +-
 css/main.css                | 22 +++++++++++-----------
 2 files changed, 12 insertions(+), 12 deletions(-)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index fb98433845..6ad250aa34 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -66,7 +66,7 @@
 
 .chat .user-message .text,
 .chat .assistant-message .text {
-    max-width: 645px;
+    max-width: 700px;
     margin-left: auto;
     margin-right: auto;
 }
diff --git a/css/main.css b/css/main.css
index 746f1f9eb9..30089aca9b 100644
--- a/css/main.css
+++ b/css/main.css
@@ -545,7 +545,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     border-radius: 5px;
     font-size: 82%;
     padding: 1px 3px;
-    background: white !important;
+    background: #f3f4f6 !important;
     color: #1f2328;
 }
 
@@ -559,18 +559,17 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     padding: 15px;
 }
 
-.message-body :not(pre) > code::before {
-    content: "`";
-}
-
-.message-body :not(pre) > code::after {
-    content: "`";
-}
-
 .message-body :not(pre) > code {
     white-space: normal !important;
     font-weight: bold;
-    font-family: unset;
+    font-size: 0.95em;
+    font-family: Menlo,"Roboto Mono","Courier New",Courier,monospace,Inter,sans-serif;
+    padding: .15rem .3rem;
+    background-color: #ececec;
+}
+
+.dark .message-body :not(pre) > code {
+    background-color: rgb(255 255 255 / 12.5%);
 }
 
 #chat-input {
@@ -584,6 +583,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     background: #f3f4f6;
     padding: 0.65rem 2.5rem;
     border: 0;
+    box-shadow: 0;
 }
 
 #chat-input textarea::placeholder {
@@ -759,7 +759,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 .hover-menu button:hover {
-    background: var(--button-secondary-background-fill-hover) !important;
+    background: #dbeafe !important;
 }
 
 .dark .hover-menu button:hover {

From b28fa86db6921adc8a42038f7062b72a27cb68b1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 6 May 2025 17:51:55 -0700
Subject: [PATCH 0475/1701] Default --gpu-layers to 256

---
 modules/shared.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/shared.py b/modules/shared.py
index 6fd4604c98..f2698bd20e 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -119,7 +119,7 @@
 group.add_argument('--batch-size', type=int, default=256, help='Maximum number of prompt tokens to batch together when calling llama_eval.')
 group.add_argument('--no-mmap', action='store_true', help='Prevent mmap from being used.')
 group.add_argument('--mlock', action='store_true', help='Force the system to keep the model in RAM.')
-group.add_argument('--gpu-layers', '--n-gpu-layers', type=int, default=0, metavar='N', help='Number of layers to offload to the GPU.')
+group.add_argument('--gpu-layers', '--n-gpu-layers', type=int, default=256, metavar='N', help='Number of layers to offload to the GPU.')
 group.add_argument('--tensor-split', type=str, default=None, help='Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.')
 group.add_argument('--numa', action='store_true', help='Activate NUMA task allocation for llama.cpp.')
 group.add_argument('--no-kv-offload', action='store_true', help='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')

From d2bae7694c0798f9f51bc61a1f7b20d93059f106 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 8 May 2025 07:26:23 -0700
Subject: [PATCH 0476/1701] UI: Change the ctx-size description

---
 modules/ui_model_menu.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index e05d225664..8dea457ea8 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -54,7 +54,7 @@ def create_ui():
                             shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
                             shared.gradio['batch_size'] = gr.Slider(label="batch_size", minimum=1, maximum=4096, step=1, value=shared.args.batch_size)
                             shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
-                            shared.gradio['ctx_size'] = gr.Number(label='ctx-size', precision=0, step=256, value=shared.args.ctx_size, info='Context length. ⚠️ Lower this value if you can\'t load the model. Common values: 2048, 4096, 8192, 16384, 32768, 65536.')
+                            shared.gradio['ctx_size'] = gr.Number(label='ctx-size', precision=0, step=256, value=shared.args.ctx_size, info='Context length. ⚠️ Lower this value if you can\'t load the model. Common values: 2048, 4096, 8192, 16384, 32768, 65536, 131072.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
                             shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')

From 348d4860c278eda1dedff15c05082e2d3358c3f3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 8 May 2025 07:58:59 -0700
Subject: [PATCH 0477/1701] UI: Create a "Main options" section in the Model
 tab

---
 modules/ui_model_menu.py | 70 ++++++++++++++++++++++------------------
 1 file changed, 39 insertions(+), 31 deletions(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 8dea457ea8..28b7222d25 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -47,52 +47,27 @@ def create_ui():
             with gr.Column():
                 shared.gradio['loader'] = gr.Dropdown(label="Model loader", choices=loaders.loaders_and_params.keys() if not shared.args.portable else ['llama.cpp'], value=None)
                 with gr.Blocks():
+                    gr.Markdown("## Main options")
                     with gr.Row():
                         with gr.Column():
                             shared.gradio['n_gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=256, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
-                            shared.gradio['threads'] = gr.Slider(label="threads", minimum=0, step=1, maximum=256, value=shared.args.threads)
-                            shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
-                            shared.gradio['batch_size'] = gr.Slider(label="batch_size", minimum=1, maximum=4096, step=1, value=shared.args.batch_size)
-                            shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
+                            shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['ctx_size'] = gr.Number(label='ctx-size', precision=0, step=256, value=shared.args.ctx_size, info='Context length. ⚠️ Lower this value if you can\'t load the model. Common values: 2048, 4096, 8192, 16384, 32768, 65536, 131072.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
-                            shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
-                            shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
-                            shared.gradio['extra_flags'] = gr.Textbox(label='extra-flags', info='Additional flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"', value=shared.args.extra_flags)
-                            shared.gradio['cpu_memory'] = gr.Number(label="Maximum CPU memory in GiB. Use this for CPU offloading.", value=shared.args.cpu_memory)
-                            shared.gradio['alpha_value'] = gr.Number(label='alpha_value', value=shared.args.alpha_value, precision=2, info='Positional embeddings alpha factor for NTK RoPE scaling. Recommended values (NTKv1): 1.75 for 1.5x context, 2.5 for 2x context. Use either this or compress_pos_emb, not both.')
-                            shared.gradio['rope_freq_base'] = gr.Number(label='rope_freq_base', value=shared.args.rope_freq_base, precision=0, info='Positional embeddings frequency base for NTK RoPE scaling. Related to alpha_value by rope_freq_base = 10000 * alpha_value ^ (64 / 63). 0 = from model.')
-                            shared.gradio['compress_pos_emb'] = gr.Number(label='compress_pos_emb', value=shared.args.compress_pos_emb, precision=2, info='Positional embeddings compression factor. Should be set to (context length) / (model\'s original context length). Equal to 1/rope_freq_scale.')
-                            shared.gradio['compute_dtype'] = gr.Dropdown(label="compute_dtype", choices=["bfloat16", "float16", "float32"], value=shared.args.compute_dtype, info='Used by load-in-4bit.')
-                            shared.gradio['quant_type'] = gr.Dropdown(label="quant_type", choices=["nf4", "fp4"], value=shared.args.quant_type, info='Used by load-in-4bit.')
-                            shared.gradio['num_experts_per_token'] = gr.Number(label="Number of experts per token", value=shared.args.num_experts_per_token, info='Only applies to MoE models like Mixtral.')
+                            shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
 
                         with gr.Column():
+                            shared.gradio['flash_attn'] = gr.Checkbox(label="flash_attn", value=shared.args.flash_attn, info='Use flash-attention.')
+                            shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming_llm", value=shared.args.streaming_llm, info='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
                             shared.gradio['load_in_8bit'] = gr.Checkbox(label="load-in-8bit", value=shared.args.load_in_8bit)
                             shared.gradio['load_in_4bit'] = gr.Checkbox(label="load-in-4bit", value=shared.args.load_in_4bit)
                             shared.gradio['torch_compile'] = gr.Checkbox(label="torch-compile", value=shared.args.torch_compile, info='Compile the model with torch.compile for improved performance.')
-                            shared.gradio['flash_attn'] = gr.Checkbox(label="flash_attn", value=shared.args.flash_attn, info='Use flash-attention.')
                             shared.gradio['use_flash_attention_2'] = gr.Checkbox(label="use_flash_attention_2", value=shared.args.use_flash_attention_2, info='Set use_flash_attention_2=True while loading the model.')
-                            shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming_llm", value=shared.args.streaming_llm, info='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
-                            shared.gradio['cpu'] = gr.Checkbox(label="cpu", value=shared.args.cpu, info='llama.cpp: Use llama-cpp-python compiled without GPU acceleration. Transformers: use PyTorch in CPU mode.')
-                            shared.gradio['disk'] = gr.Checkbox(label="disk", value=shared.args.disk)
-                            shared.gradio['row_split'] = gr.Checkbox(label="row_split", value=shared.args.row_split, info='Split the model by rows across GPUs. This may improve multi-gpu performance.')
-                            shared.gradio['no_kv_offload'] = gr.Checkbox(label="no_kv_offload", value=shared.args.no_kv_offload, info='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
-                            shared.gradio['no_mmap'] = gr.Checkbox(label="no-mmap", value=shared.args.no_mmap)
-                            shared.gradio['mlock'] = gr.Checkbox(label="mlock", value=shared.args.mlock)
-                            shared.gradio['numa'] = gr.Checkbox(label="numa", value=shared.args.numa, info='NUMA support can help on some systems with non-uniform memory access.')
                             shared.gradio['use_double_quant'] = gr.Checkbox(label="use_double_quant", value=shared.args.use_double_quant, info='Used by load-in-4bit.')
-                            shared.gradio['use_eager_attention'] = gr.Checkbox(label="use_eager_attention", value=shared.args.use_eager_attention, info='Set attn_implementation= eager while loading the model.')
-                            shared.gradio['bf16'] = gr.Checkbox(label="bf16", value=shared.args.bf16)
                             shared.gradio['autosplit'] = gr.Checkbox(label="autosplit", value=shared.args.autosplit, info='Automatically split the model tensors across the available GPUs.')
                             shared.gradio['enable_tp'] = gr.Checkbox(label="enable_tp", value=shared.args.enable_tp, info='Enable Tensor Parallelism (TP).')
-                            shared.gradio['no_flash_attn'] = gr.Checkbox(label="no_flash_attn", value=shared.args.no_flash_attn)
-                            shared.gradio['no_xformers'] = gr.Checkbox(label="no_xformers", value=shared.args.no_xformers)
-                            shared.gradio['no_sdpa'] = gr.Checkbox(label="no_sdpa", value=shared.args.no_sdpa)
-                            shared.gradio['cfg_cache'] = gr.Checkbox(label="cfg-cache", value=shared.args.cfg_cache, info='Necessary to use CFG with this loader.')
                             shared.gradio['cpp_runner'] = gr.Checkbox(label="cpp-runner", value=shared.args.cpp_runner, info='Enable inference with ModelRunnerCpp, which is faster than the default ModelRunner.')
                             shared.gradio['trust_remote_code'] = gr.Checkbox(label="trust-remote-code", value=shared.args.trust_remote_code, info='Set trust_remote_code=True while loading the tokenizer/model. To enable this option, start the web UI with the --trust-remote-code flag.', interactive=shared.args.trust_remote_code)
-                            shared.gradio['no_use_fast'] = gr.Checkbox(label="no_use_fast", value=shared.args.no_use_fast, info='Set use_fast=False while loading the tokenizer.')
                             shared.gradio['exllamav2_info'] = gr.Markdown("ExLlamav2_HF is recommended over ExLlamav2 for better integration with extensions and more consistent sampling behavior across loaders.")
                             shared.gradio['tensorrt_llm_info'] = gr.Markdown('* TensorRT-LLM has to be installed manually in a separate Python 3.10 environment at the moment. For a guide, consult the description of [this PR](https://github.com/oobabooga/text-generation-webui/pull/5715). \n\n* `ctx_size` is only used when `cpp-runner` is checked.\n\n* `cpp_runner` does not support streaming at the moment.')
 
@@ -102,11 +77,44 @@ def create_ui():
                                     shared.gradio['model_draft'] = gr.Dropdown(label="model-draft", choices=utils.get_available_models(), value=lambda: shared.args.model_draft, elem_classes='slim-dropdown', info='Draft model. Speculative decoding only works with models sharing the same vocabulary (e.g., same model family).', interactive=not mu)
                                     ui.create_refresh_button(shared.gradio['model_draft'], lambda: None, lambda: {'choices': utils.get_available_models()}, 'refresh-button', interactive=not mu)
 
-                                shared.gradio['draft_max'] = gr.Number(label="draft-max", precision=0, step=1, value=shared.args.draft_max, info='Number of tokens to draft for speculative decoding.')
                                 shared.gradio['gpu_layers_draft'] = gr.Slider(label="gpu-layers-draft", minimum=0, maximum=256, value=shared.args.gpu_layers_draft, info='Number of layers to offload to the GPU for the draft model.')
+                                shared.gradio['draft_max'] = gr.Number(label="draft-max", precision=0, step=1, value=shared.args.draft_max, info='Number of tokens to draft for speculative decoding. Recommended value: 4.')
                                 shared.gradio['device_draft'] = gr.Textbox(label="device-draft", value=shared.args.device_draft, info='Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1')
                                 shared.gradio['ctx_size_draft'] = gr.Number(label="ctx-size-draft", precision=0, step=256, value=shared.args.ctx_size_draft, info='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
 
+                    gr.Markdown("## Other options")
+                    with gr.Accordion("See more options", open=False, elem_classes='tgw-accordion'):
+                        with gr.Row():
+                            with gr.Column():
+                                shared.gradio['threads'] = gr.Slider(label="threads", minimum=0, step=1, maximum=256, value=shared.args.threads)
+                                shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
+                                shared.gradio['batch_size'] = gr.Slider(label="batch_size", minimum=1, maximum=4096, step=1, value=shared.args.batch_size)
+                                shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
+                                shared.gradio['extra_flags'] = gr.Textbox(label='extra-flags', info='Additional flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"', value=shared.args.extra_flags)
+                                shared.gradio['cpu_memory'] = gr.Number(label="Maximum CPU memory in GiB. Use this for CPU offloading.", value=shared.args.cpu_memory)
+                                shared.gradio['alpha_value'] = gr.Number(label='alpha_value', value=shared.args.alpha_value, precision=2, info='Positional embeddings alpha factor for NTK RoPE scaling. Recommended values (NTKv1): 1.75 for 1.5x context, 2.5 for 2x context. Use either this or compress_pos_emb, not both.')
+                                shared.gradio['rope_freq_base'] = gr.Number(label='rope_freq_base', value=shared.args.rope_freq_base, precision=0, info='Positional embeddings frequency base for NTK RoPE scaling. Related to alpha_value by rope_freq_base = 10000 * alpha_value ^ (64 / 63). 0 = from model.')
+                                shared.gradio['compress_pos_emb'] = gr.Number(label='compress_pos_emb', value=shared.args.compress_pos_emb, precision=2, info='Positional embeddings compression factor. Should be set to (context length) / (model\'s original context length). Equal to 1/rope_freq_scale.')
+                                shared.gradio['compute_dtype'] = gr.Dropdown(label="compute_dtype", choices=["bfloat16", "float16", "float32"], value=shared.args.compute_dtype, info='Used by load-in-4bit.')
+                                shared.gradio['quant_type'] = gr.Dropdown(label="quant_type", choices=["nf4", "fp4"], value=shared.args.quant_type, info='Used by load-in-4bit.')
+                                shared.gradio['num_experts_per_token'] = gr.Number(label="Number of experts per token", value=shared.args.num_experts_per_token, info='Only applies to MoE models like Mixtral.')
+
+                            with gr.Column():
+                                shared.gradio['cpu'] = gr.Checkbox(label="cpu", value=shared.args.cpu, info='llama.cpp: Use llama-cpp-python compiled without GPU acceleration. Transformers: use PyTorch in CPU mode.')
+                                shared.gradio['disk'] = gr.Checkbox(label="disk", value=shared.args.disk)
+                                shared.gradio['row_split'] = gr.Checkbox(label="row_split", value=shared.args.row_split, info='Split the model by rows across GPUs. This may improve multi-gpu performance.')
+                                shared.gradio['no_kv_offload'] = gr.Checkbox(label="no_kv_offload", value=shared.args.no_kv_offload, info='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
+                                shared.gradio['no_mmap'] = gr.Checkbox(label="no-mmap", value=shared.args.no_mmap)
+                                shared.gradio['mlock'] = gr.Checkbox(label="mlock", value=shared.args.mlock)
+                                shared.gradio['numa'] = gr.Checkbox(label="numa", value=shared.args.numa, info='NUMA support can help on some systems with non-uniform memory access.')
+                                shared.gradio['use_eager_attention'] = gr.Checkbox(label="use_eager_attention", value=shared.args.use_eager_attention, info='Set attn_implementation= eager while loading the model.')
+                                shared.gradio['bf16'] = gr.Checkbox(label="bf16", value=shared.args.bf16)
+                                shared.gradio['no_flash_attn'] = gr.Checkbox(label="no_flash_attn", value=shared.args.no_flash_attn)
+                                shared.gradio['no_xformers'] = gr.Checkbox(label="no_xformers", value=shared.args.no_xformers)
+                                shared.gradio['no_sdpa'] = gr.Checkbox(label="no_sdpa", value=shared.args.no_sdpa)
+                                shared.gradio['cfg_cache'] = gr.Checkbox(label="cfg-cache", value=shared.args.cfg_cache, info='Necessary to use CFG with this loader.')
+                                shared.gradio['no_use_fast'] = gr.Checkbox(label="no_use_fast", value=shared.args.no_use_fast, info='Set use_fast=False while loading the tokenizer.')
+
             with gr.Column():
                 with gr.Tab("Download"):
                     shared.gradio['custom_model_menu'] = gr.Textbox(label="Download model or LoRA", info="Enter the Hugging Face username/model path, for instance: facebook/galactica-125m. To specify a branch, add it at the end after a \":\" character like this: facebook/galactica-125m:main. To download a single file, enter its name in the second box.", interactive=not mu)

From a2ab42d39099d89543a8e5c5753350e51905fa36 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 8 May 2025 08:00:38 -0700
Subject: [PATCH 0478/1701] UI: Remove the exllamav2 info message

---
 modules/loaders.py       | 1 -
 modules/ui_model_menu.py | 1 -
 2 files changed, 2 deletions(-)

diff --git a/modules/loaders.py b/modules/loaders.py
index b29679bd47..4b76549b84 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -84,7 +84,6 @@
         'no_flash_attn',
         'no_xformers',
         'no_sdpa',
-        'exllamav2_info',
         'model_draft',
         'draft_max',
         'ctx_size_draft',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 28b7222d25..33e152a088 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -68,7 +68,6 @@ def create_ui():
                             shared.gradio['enable_tp'] = gr.Checkbox(label="enable_tp", value=shared.args.enable_tp, info='Enable Tensor Parallelism (TP).')
                             shared.gradio['cpp_runner'] = gr.Checkbox(label="cpp-runner", value=shared.args.cpp_runner, info='Enable inference with ModelRunnerCpp, which is faster than the default ModelRunner.')
                             shared.gradio['trust_remote_code'] = gr.Checkbox(label="trust-remote-code", value=shared.args.trust_remote_code, info='Set trust_remote_code=True while loading the tokenizer/model. To enable this option, start the web UI with the --trust-remote-code flag.', interactive=shared.args.trust_remote_code)
-                            shared.gradio['exllamav2_info'] = gr.Markdown("ExLlamav2_HF is recommended over ExLlamav2 for better integration with extensions and more consistent sampling behavior across loaders.")
                             shared.gradio['tensorrt_llm_info'] = gr.Markdown('* TensorRT-LLM has to be installed manually in a separate Python 3.10 environment at the moment. For a guide, consult the description of [this PR](https://github.com/oobabooga/text-generation-webui/pull/5715). \n\n* `ctx_size` is only used when `cpp-runner` is checked.\n\n* `cpp_runner` does not support streaming at the moment.')
 
                             # Speculative decoding

From 13a434f3518e381d04acd869ed3c0ba3d3823d34 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 8 May 2025 08:06:07 -0700
Subject: [PATCH 0479/1701] Bump exllamav3

---
 requirements/full/requirements.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_noavx2.txt        | 4 ++--
 4 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 3b50c6744d..ac89f45b4f 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -32,8 +32,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a7/exllamav3-0.0.1a7+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a7/exllamav3-0.0.1a7+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index ba23ea9c34..6abdb1a4f5 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -31,5 +31,5 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a7/exllamav3-0.0.1a7-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index c245ab747b..682c6a47dc 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -32,5 +32,5 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a7/exllamav3-0.0.1a7-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index d8bbf6d1ed..1e185079c9 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -32,8 +32,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a7/exllamav3-0.0.1a7+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a7/exllamav3-0.0.1a7+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From ed6e16191da79523c5cabfd927130b307b3b54b9 Mon Sep 17 00:00:00 2001
From: Scott Z 
Date: Thu, 8 May 2025 11:21:52 -0400
Subject: [PATCH 0480/1701] Docker fix for NVIDIA (#6964)

---
 docker/nvidia/Dockerfile         |  2 +-
 docker/nvidia/docker-compose.yml | 12 +-----------
 2 files changed, 2 insertions(+), 12 deletions(-)

diff --git a/docker/nvidia/Dockerfile b/docker/nvidia/Dockerfile
index 900a432967..82594a26f2 100644
--- a/docker/nvidia/Dockerfile
+++ b/docker/nvidia/Dockerfile
@@ -14,7 +14,7 @@ WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
 RUN GPU_CHOICE=A LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
-COPY CMD_FLAGS.txt /home/app/text-generation-webui/
+COPY /user_data/CMD_FLAGS.txt /home/app/text-generation-webui/user_data
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
 WORKDIR /home/app/text-generation-webui
 # set umask to ensure group read / write at runtime
diff --git a/docker/nvidia/docker-compose.yml b/docker/nvidia/docker-compose.yml
index 0392078e06..23d5cacca5 100644
--- a/docker/nvidia/docker-compose.yml
+++ b/docker/nvidia/docker-compose.yml
@@ -31,17 +31,7 @@ services:
     stdin_open: true
     tty: true
     volumes:
-      - ./cache:/home/app/text-generation-webui/cache
-      - ./characters:/home/app/text-generation-webui/characters
-      - ./extensions:/home/app/text-generation-webui/extensions
-      - ./loras:/home/app/text-generation-webui/loras
-      - ./logs:/home/app/text-generation-webui/logs
-      - ./models:/home/app/text-generation-webui/models
-      - ./presets:/home/app/text-generation-webui/presets
-      - ./prompts:/home/app/text-generation-webui/prompts
-      - ./softprompts:/home/app/text-generation-webui/softprompts
-      - ./training:/home/app/text-generation-webui/training
-      - ./cloudflared:/etc/cloudflared
+      - ./user_data:/home/app/text-generation-webui/user_data
     deploy:
       resources:
         reservations:

From fa960496d554ece24c06088607692fa7b874ff5b Mon Sep 17 00:00:00 2001
From: Jonas 
Date: Thu, 8 May 2025 17:30:27 +0200
Subject: [PATCH 0481/1701] Tools support for OpenAI compatible API (#6827)

---
 extensions/openai/completions.py | 73 +++++++++++++++++++++----
 extensions/openai/typing.py      | 47 +++++++++++++++-
 extensions/openai/utils.py       | 94 ++++++++++++++++++++++++++++++++
 modules/chat.py                  | 12 ++--
 4 files changed, 209 insertions(+), 17 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index a7d8b4e493..ed0bcc40ff 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -1,11 +1,14 @@
 import copy
 import time
+import json
 from collections import deque
 
 import tiktoken
 
 from extensions.openai.errors import InvalidRequestError
-from extensions.openai.utils import debug_msg
+from extensions.openai.utils import debug_msg, getToolCallId, parseToolCall
+from extensions.openai.typing import ToolDefinition
+from pydantic import ValidationError
 from modules import shared
 from modules.chat import (
     generate_chat_prompt,
@@ -99,19 +102,24 @@ def convert_history(history):
             user_input = content
             user_input_last = True
             if current_message:
-                chat_dialogue.append([current_message, ''])
+                chat_dialogue.append([current_message, '', ''])
                 current_message = ""
 
             current_message = content
         elif role == "assistant":
+            if "tool_calls" in entry and isinstance(entry["tool_calls"], list) and len(entry["tool_calls"]) > 0 and content.strip() == "":
+                continue  # skip tool calls
             current_reply = content
             user_input_last = False
             if current_message:
-                chat_dialogue.append([current_message, current_reply])
+                chat_dialogue.append([current_message, current_reply, ''])
                 current_message = ""
                 current_reply = ""
             else:
-                chat_dialogue.append(['', current_reply])
+                chat_dialogue.append(['', current_reply, ''])
+        elif role == "tool":
+            user_input_last = False
+            chat_dialogue.append(['', '', content])
         elif role == "system":
             system_message += f"\n{content}" if system_message else content
 
@@ -131,6 +139,10 @@ def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, p
     if 'messages' not in body:
         raise InvalidRequestError(message="messages is required", param='messages')
 
+    tools = None
+    if 'tools' in body and body['tools'] is not None and isinstance(body['tools'], list) and len(body['tools']) > 0:
+        tools = validateTools(body['tools']) # raises InvalidRequestError if validation fails
+
     messages = body['messages']
     for m in messages:
         if 'role' not in m:
@@ -188,6 +200,7 @@ def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, p
         'custom_system_message': custom_system_message,
         'chat_template_str': chat_template_str,
         'chat-instruct_command': chat_instruct_command,
+        'tools': tools,
         'history': history,
         'stream': stream
     })
@@ -200,7 +213,7 @@ def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, p
     requested_model = generate_params.pop('model')
     logprob_proc = generate_params.pop('logprob_proc', None)
 
-    def chat_streaming_chunk(content):
+    def chat_streaming_chunk(content, chunk_tool_calls=None):
         # begin streaming
         chunk = {
             "id": cmpl_id,
@@ -210,7 +223,7 @@ def chat_streaming_chunk(content):
             resp_list: [{
                 "index": 0,
                 "finish_reason": None,
-                "delta": {'role': 'assistant', 'content': content},
+                "delta": {'role': 'assistant', 'content': content, 'tool_calls': chunk_tool_calls},
             }],
         }
 
@@ -219,6 +232,7 @@ def chat_streaming_chunk(content):
             chunk[resp_list][0]["logprobs"] = {'top_logprobs': [top_logprobs]}
         # else:
         #    chunk[resp_list][0]["logprobs"] = None
+
         return chunk
 
     # generate reply #######################################
@@ -227,8 +241,6 @@ def chat_streaming_chunk(content):
         yield {'prompt': prompt}
         return
 
-    debug_msg({'prompt': prompt, 'generate_params': generate_params})
-
     if stream:
         yield chat_streaming_chunk('')
 
@@ -238,8 +250,23 @@ def chat_streaming_chunk(content):
     answer = ''
     seen_content = ''
 
+    tool_calls = []
+    end_last_tool_call = 0
+    supported_tools = [x["function"]["name"] for x in tools] if tools is not None else None
+
     for a in generator:
         answer = a['internal'][-1][1]
+
+        if supported_tools is not None:
+            tool_call = parseToolCall(answer[end_last_tool_call:], supported_tools) if len(answer) > 0 else []
+            if len(tool_call) > 0:
+                for tc in tool_call:
+                    tc["id"] = getToolCallId()
+                    tc["index"] = str(len(tool_calls))
+                    tc["function"]["arguments"] = json.dumps(tc["function"]["arguments"])
+                    tool_calls.append(tc)
+                end_last_tool_call = len(answer)
+
         if stream:
             len_seen = len(seen_content)
             new_content = answer[len_seen:]
@@ -247,18 +274,25 @@ def chat_streaming_chunk(content):
             if not new_content or chr(0xfffd) in new_content:  # partial unicode character, don't send it yet.
                 continue
 
-            seen_content = answer
             chunk = chat_streaming_chunk(new_content)
+
+            seen_content = answer
             yield chunk
 
+        # stop generation if tool_calls were generated previously
+        if len(tool_calls) > 0:
+            break
+
     token_count = len(encode(prompt)[0])
     completion_token_count = len(encode(answer)[0])
     stop_reason = "stop"
+    if len(tool_calls) > 0:
+        stop_reason = "tool_calls"
     if token_count + completion_token_count >= generate_params['truncation_length'] or completion_token_count >= generate_params['max_new_tokens']:
         stop_reason = "length"
 
     if stream:
-        chunk = chat_streaming_chunk('')
+        chunk = chat_streaming_chunk('', tool_calls)
         chunk[resp_list][0]['finish_reason'] = stop_reason
         chunk['usage'] = {
             "prompt_tokens": token_count,
@@ -276,7 +310,8 @@ def chat_streaming_chunk(content):
             resp_list: [{
                 "index": 0,
                 "finish_reason": stop_reason,
-                "message": {"role": "assistant", "content": answer}
+                "message": {"role": "assistant", "content": answer},
+                "tool_calls": tool_calls
             }],
             "usage": {
                 "prompt_tokens": token_count,
@@ -465,3 +500,19 @@ def completions(body: dict, is_legacy: bool = False) -> dict:
 def stream_completions(body: dict, is_legacy: bool = False):
     for resp in completions_common(body, is_legacy, stream=True):
         yield resp
+
+
+def validateTools(tools: list[dict]):
+    # Validate each tool definition in the JSON array
+    valid_tools = None
+    for idx in range(len(tools)):
+        tool = tools[idx]
+        try:
+            tool_definition = ToolDefinition(**tool)
+            if valid_tools is None:
+                valid_tools = []
+            valid_tools.append(tool)
+        except ValidationError:
+            raise InvalidRequestError(message=f"Invalid tool specification at index {idx}.", param='tools')
+
+    return valid_tools
diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index b1979cbc6a..b28ebb4ec7 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -1,8 +1,8 @@
 import json
 import time
-from typing import Dict, List
+from typing import Dict, List, Optional
 
-from pydantic import BaseModel, Field
+from pydantic import BaseModel, Field, validator
 
 
 class GenerationOptions(BaseModel):
@@ -54,6 +54,48 @@ class GenerationOptions(BaseModel):
     grammar_string: str = ""
 
 
+class ToolDefinition(BaseModel):
+    function: 'ToolFunction'
+    type: str
+
+
+class ToolFunction(BaseModel):
+    description: str
+    name: str
+    parameters: 'ToolParameters'
+
+
+class ToolParameters(BaseModel):
+    properties: Optional[Dict[str, 'ToolProperty']] = None
+    required: Optional[list[str]] = None
+    type: str
+    description: Optional[str] = None
+
+
+class ToolProperty(BaseModel):
+    description: Optional[str] = None
+    type: Optional[str] = None  # we are faced with definitions like anyOf, e.g. {'type': 'function', 'function': {'name': 'git_create_branch', 'description': 'Creates a new branch from an optional base branch', 'parameters': {'type': 'object', 'properties': {'repo_path': {'title': 'Repo Path', 'type': 'string'}, 'branch_name': {'title': 'Branch Name', 'type': 'string'}, 'base_branch': {'anyOf': [{'type': 'string'}, {'type': 'null'}], 'default': None, 'title': 'Base Branch'}}, 'required': ['repo_path', 'branch_name'], 'title': 'GitCreateBranch'}}}
+
+
+class FunctionCall(BaseModel):
+    name: str
+    arguments: Optional[str] = None
+    parameters: Optional[str] = None
+
+    @validator('arguments', allow_reuse=True)
+    def checkPropertyArgsOrParams(cls, v, values, **kwargs):
+        if not v and not values.get('parameters'):
+            raise ValueError("At least one of 'arguments' or 'parameters' must be provided as property in FunctionCall type")
+        return v
+
+
+class ToolCall(BaseModel):
+    id: str
+    index: int
+    type: str
+    function: FunctionCall
+
+
 class CompletionRequestParams(BaseModel):
     model: str | None = Field(default=None, description="Unused parameter. To change the model, use the /v1/internal/model/load endpoint.")
     prompt: str | List[str]
@@ -92,6 +134,7 @@ class ChatCompletionRequestParams(BaseModel):
     frequency_penalty: float | None = 0
     function_call: str | dict | None = Field(default=None, description="Unused parameter.")
     functions: List[dict] | None = Field(default=None, description="Unused parameter.")
+    tools: List[dict] | None = Field(default=None, description="Tools signatures passed via MCP.")
     logit_bias: dict | None = None
     max_tokens: int | None = None
     n: int | None = Field(default=1, description="Unused parameter.")
diff --git a/extensions/openai/utils.py b/extensions/openai/utils.py
index 2b4147690e..8cb856ff32 100644
--- a/extensions/openai/utils.py
+++ b/extensions/openai/utils.py
@@ -1,6 +1,9 @@
 import base64
 import os
 import time
+import json
+import random
+import re
 import traceback
 from typing import Callable, Optional
 
@@ -52,3 +55,94 @@ def _start_cloudflared(port: int, tunnel_id: str, max_attempts: int = 3, on_star
             time.sleep(3)
 
         raise Exception('Could not start cloudflared.')
+
+
+def getToolCallId() -> str:
+    letter_bytes = "abcdefghijklmnopqrstuvwxyz0123456789"
+    b = [random.choice(letter_bytes) for _ in range(8)]
+    return "call_" + "".join(b).lower()
+
+
+def checkAndSanitizeToolCallCandidate(candidate_dict: dict, tool_names: list[str]):
+    # check if property 'function' exists and is a dictionary, otherwise adapt dict
+    if 'function' not in candidate_dict and 'name' in candidate_dict and isinstance(candidate_dict['name'], str):
+        candidate_dict = {"type": "function", "function": candidate_dict}
+    if 'function' in candidate_dict and isinstance(candidate_dict['function'], str):
+        candidate_dict['name'] = candidate_dict['function']
+        del candidate_dict['function']
+        candidate_dict = {"type": "function", "function": candidate_dict}
+    if 'function' in candidate_dict and isinstance(candidate_dict['function'], dict):
+        # check if 'name' exists within 'function' and is part of known tools
+        if 'name' in candidate_dict['function'] and candidate_dict['function']['name'] in tool_names:
+            candidate_dict["type"] = "function"  # ensure required property 'type' exists and has the right value
+            # map property 'parameters' used by some older models to 'arguments'
+            if "arguments" not in candidate_dict["function"] and "parameters" in candidate_dict["function"]:
+                candidate_dict["function"]["arguments"] = candidate_dict["function"]["parameters"]
+                del candidate_dict["function"]["parameters"]
+            return candidate_dict
+    return None
+
+
+def parseToolCall(answer: str, tool_names: list[str]):
+    matches = []
+
+    # abort on very short answers to save computation cycles
+    if len(answer) < 10:
+        return matches
+
+    # Define the regex pattern to find the JSON content wrapped in , , , and other tags observed from various models
+    patterns = [ r"(```[^\n]*)\n(.*?)```", r"<([^>]+)>(.*?)" ]
+
+    for pattern in patterns:
+        for match in re.finditer(pattern, answer, re.DOTALL):
+            # print(match.group(2))
+            if match.group(2) is None:
+                continue
+            # remove backtick wraps if present
+            candidate = re.sub(r"^```(json|xml|python[^\n]*)\n", "", match.group(2).strip())
+            candidate = re.sub(r"```$", "", candidate.strip())
+            # unwrap inner tags
+            candidate = re.sub(pattern, r"\2", candidate.strip(), flags=re.DOTALL)
+            # llm might have generated multiple json objects separated by linebreaks, check for this pattern and try parsing each object individually
+            if re.search(r"\}\s*\n\s*\{", candidate) is not None:
+                candidate = re.sub(r"\}\s*\n\s*\{", "},\n{", candidate)
+            if not candidate.strip().startswith("["):
+                candidate = "[" + candidate + "]"
+
+            candidates = []
+            try:
+                # parse the candidate JSON into a dictionary
+                candidates = json.loads(candidate)
+                if not isinstance(candidates, list):
+                    candidates = [candidates]
+            except json.JSONDecodeError:
+                # Ignore invalid JSON silently
+                continue
+
+            for candidate_dict in candidates:
+                checked_candidate = checkAndSanitizeToolCallCandidate(candidate_dict, tool_names)
+                if checked_candidate is not None:
+                    matches.append(checked_candidate)
+
+        # last resort if nothing has been mapped: LLM might have produced plain json tool call without xml-like tags
+        if len(matches) == 0:
+            try:
+                candidate = answer
+                # llm might have generated multiple json objects separated by linebreaks, check for this pattern and try parsing each object individually
+                if re.search(r"\}\s*\n\s*\{", candidate) is not None:
+                    candidate = re.sub(r"\}\s*\n\s*\{", "},\n{", candidate)
+                if not candidate.strip().startswith("["):
+                    candidate = "[" + candidate + "]"
+                # parse the candidate JSON into a dictionary
+                candidates = json.loads(candidate)
+                if not isinstance(candidates, list):
+                    candidates = [candidates]
+                for candidate_dict in candidates:
+                    checked_candidate = checkAndSanitizeToolCallCandidate(candidate_dict, tool_names)
+                    if checked_candidate is not None:
+                        matches.append(checked_candidate)
+            except json.JSONDecodeError:
+                # Ignore invalid JSON silently
+                pass
+
+    return matches
diff --git a/modules/chat.py b/modules/chat.py
index feac6bddc6..b524b1b962 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -145,7 +145,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
     instruct_renderer = partial(
         instruction_template.render,
         builtin_tools=None,
-        tools=None,
+        tools=state['tools'] if 'tools' in state else None,
         tools_in_user_message=False,
         add_generation_prompt=False
     )
@@ -171,9 +171,13 @@ def generate_chat_prompt(user_input, state, **kwargs):
             messages.append({"role": "system", "content": context})
 
     insert_pos = len(messages)
-    for user_msg, assistant_msg in reversed(history):
-        user_msg = user_msg.strip()
-        assistant_msg = assistant_msg.strip()
+    for entry in reversed(history):
+        user_msg = entry[0].strip()
+        assistant_msg = entry[1].strip()
+        tool_msg = entry[2].strip() if len(entry) > 2 else ''
+
+        if tool_msg:
+            messages.insert(insert_pos, {"role": "tool", "content": tool_msg})
 
         if assistant_msg:
             messages.insert(insert_pos, {"role": "assistant", "content": assistant_msg})

From a1b3307b6636b13373e8f399690cb3b782854d2c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 8 May 2025 08:58:14 -0700
Subject: [PATCH 0482/1701] Bump llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index ac89f45b4f..3a059c9129 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 431cd74032..ebc33216a5 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 0c581f869e..8ec6898f6a 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 6abdb1a4f5..afc869c888 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 682c6a47dc..8d7d29b785 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -29,8 +29,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index ac277d6103..d69aae1827 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index cc412d3300..540c9ac8e6 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 1e185079c9..3bb5a74a23 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 1240d335c0..95319d7588 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 6b165b7cd1..4b49b4e17d 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 1b2b5cf2b3..a6ebda30be 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -15,6 +15,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 2793d74395..bb5ba8ad39 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 6d7316a6f5..3d17dd49e6 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index e56eba0886..ff9fa04cbb 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index a7f8c703db..e17f8ce7ed 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 5b427fd208..dd01b3a85a 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.9.0/llama_cpp_binaries-0.9.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 1c7209a725c8811f2d4d2325007b2e871c5af020 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 8 May 2025 09:46:43 -0700
Subject: [PATCH 0483/1701] Save the chat history periodically during streaming

---
 modules/chat.py | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index b524b1b962..403d05e112 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -5,6 +5,7 @@
 import json
 import pprint
 import re
+import time
 from datetime import datetime
 from functools import partial
 from pathlib import Path
@@ -485,10 +486,16 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
         send_dummy_reply(state['start_with'], state)
 
     history = state['history']
+    last_save_time = time.monotonic()
+    save_interval = 8
     for i, history in enumerate(generate_chat_reply(text, state, regenerate, _continue, loading_message=True, for_ui=True)):
         yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu']), history
-        if i == 0:
+
+        current_time = time.monotonic()
+        # Save on first iteration or if save_interval seconds have passed
+        if i == 0 or (current_time - last_save_time) >= save_interval:
             save_history(history, state['unique_id'], state['character_menu'], state['mode'])
+            last_save_time = current_time
 
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
 

From 3bc2ec2b119c058446f9e9600213c75302a4ac4f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 8 May 2025 10:34:09 -0700
Subject: [PATCH 0484/1701] Fix #6965

---
 one_click.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/one_click.py b/one_click.py
index cb16b813ee..482a6aa9f1 100644
--- a/one_click.py
+++ b/one_click.py
@@ -126,7 +126,7 @@ def check_env():
         sys.exit(1)
 
     # Ensure this is a new environment and not the base environment
-    if os.environ["CONDA_DEFAULT_ENV"] == "base":
+    if os.environ.get("CONDA_DEFAULT_ENV", "") == "base":
         print("Create an environment for this project and activate it. Exiting...")
         sys.exit(1)
 

From 9ea2a69210ab5658ba8daf6d7d604589de5fc741 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 8 May 2025 10:41:25 -0700
Subject: [PATCH 0485/1701] llama.cpp: Add --no-webui to the llama-server
 command

---
 modules/llama_cpp_server.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index d8d2f61bdd..1046969a7e 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -261,6 +261,7 @@ def _start_server(self):
             "--gpu-layers", str(shared.args.gpu_layers),
             "--batch-size", str(shared.args.batch_size),
             "--port", str(self.port),
+            "--no-webui",
         ]
 
         if shared.args.flash_attn:

From bf7e4a4597b6492b4c440d32a8afbda59d4ef035 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 8 May 2025 16:12:07 -0700
Subject: [PATCH 0486/1701] Docs: Add a tool/function calling example (from
 https://github.com/oobabooga/text-generation-webui/pull/6827#issuecomment-2854716960)

---
 docs/12 - OpenAI API.md | 79 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 79 insertions(+)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index 364c6b0926..db9befed93 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -257,6 +257,85 @@ headers = {
 
 in any of the examples above.
 
+#### Tool/Function Calling Example
+
+You need to use a model with tools support. The prompt will be automatically formatted using the model's Jinja2 template.
+
+Request:
+
+```
+curl http://127.0.0.1:5000/v1/chat/completions \
+  -H "Content-Type: application/json" \
+  -d '{
+    "messages": [
+      {
+        "role": "system",
+        "content": "You are a helpful assistant."
+      },
+      {
+        "role": "user",
+        "content": "What time is it currently in New York City?"
+      }
+    ],
+    "tools": [
+      {
+        "type": "function",
+        "function": {
+          "name": "get_current_time",
+          "description": "Get current time in a specific timezones",
+          "parameters": {
+            "type": "object",
+            "required": ["timezone"],
+            "properties": {
+              "timezone": {
+                "type": "string",
+                "description": "IANA timezone name (e.g., America/New_York, Europe/London). Use Europe/Berlin as local timezone if no timezone provided by the user."
+              }
+            }
+          }
+        }
+      }
+    ]
+  }'
+```
+
+Sample response:
+
+```
+{
+    "id": "chatcmpl-1746532051477984256",
+    "object": "chat.completion",
+    "created": 1746532051,
+    "model": "qwen2.5-coder-14b-instruct-q4_k_m.gguf",
+    "choices": [
+        {
+            "index": 0,
+            "finish_reason": "tool_calls",
+            "message": {
+                "role": "assistant",
+                "content": "```xml\n\n{\n  \"name\": \"get_current_time\",\n  \"arguments\": {\n    \"timezone\": \"America/New_York\"\n  }\n}\n\n```"
+            },
+            "tool_calls": [
+                {
+                    "type": "function",
+                    "function": {
+                        "name": "get_current_time",
+                        "arguments": "{\"timezone\": \"America/New_York\"}"
+                    },
+                    "id": "call_52ij07mh",
+                    "index": "0"
+                }
+            ]
+        }
+    ],
+    "usage": {
+        "prompt_tokens": 224,
+        "completion_tokens": 38,
+        "total_tokens": 262
+    }
+}
+```
+
 ### Environment variables
 
 The following environment variables can be used (they take precedence over everything else):

From f8ef6e09af5d2e28cf67d1eea165591e156ac9d2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 8 May 2025 18:19:04 -0700
Subject: [PATCH 0487/1701] UI: Make ctx-size a slider

---
 modules/ui_model_menu.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 33e152a088..d4d9b8b12f 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -51,8 +51,8 @@ def create_ui():
                     with gr.Row():
                         with gr.Column():
                             shared.gradio['n_gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=256, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
+                            shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
-                            shared.gradio['ctx_size'] = gr.Number(label='ctx-size', precision=0, step=256, value=shared.args.ctx_size, info='Context length. ⚠️ Lower this value if you can\'t load the model. Common values: 2048, 4096, 8192, 16384, 32768, 65536, 131072.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
                             shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
 

From 512bc2d0e02bef2434370c2317bcf56e50f0513f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 8 May 2025 23:43:55 -0700
Subject: [PATCH 0488/1701] UI: Update some labels

---
 modules/ui_model_menu.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index d4d9b8b12f..1e27255b12 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -53,12 +53,12 @@ def create_ui():
                             shared.gradio['n_gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=256, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
-                            shared.gradio['cache_type'] = gr.Dropdown(label="cache_type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
+                            shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
                             shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
 
                         with gr.Column():
-                            shared.gradio['flash_attn'] = gr.Checkbox(label="flash_attn", value=shared.args.flash_attn, info='Use flash-attention.')
-                            shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming_llm", value=shared.args.streaming_llm, info='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
+                            shared.gradio['flash_attn'] = gr.Checkbox(label="flash-attn", value=shared.args.flash_attn, info='Use flash-attention.')
+                            shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming-llm", value=shared.args.streaming_llm, info='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
                             shared.gradio['load_in_8bit'] = gr.Checkbox(label="load-in-8bit", value=shared.args.load_in_8bit)
                             shared.gradio['load_in_4bit'] = gr.Checkbox(label="load-in-4bit", value=shared.args.load_in_4bit)
                             shared.gradio['torch_compile'] = gr.Checkbox(label="torch-compile", value=shared.args.torch_compile, info='Compile the model with torch.compile for improved performance.')

From 2bde625d5716355b30fdd414c9b104812b101ed1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 9 May 2025 00:19:25 -0700
Subject: [PATCH 0489/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 6cc84c5005..0833f9b0c9 100644
--- a/README.md
+++ b/README.md
@@ -22,7 +22,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 - Free-form text generation in the Default/Notebook tabs without being limited to chat turns. You can send formatted conversations from the Chat tab to these.
 - Multiple sampling parameters and generation options for sophisticated text generation control.
 - Switch between different models easily in the UI without restarting, with fine control over settings.
-- OpenAI-compatible API with Chat and Completions endpoints – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples).
+- OpenAI-compatible API with Chat and Completions endpoints, including tool-calling support – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples).
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - Extension support, with numerous built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details.
 

From 8984e95c671c262b1667805895d317a9ffe9cd0a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 9 May 2025 07:21:05 -0700
Subject: [PATCH 0490/1701] UI: More friendly message when no model is loaded

---
 modules/logits.py          |  7 ++++---
 modules/text_generation.py |  5 +++--
 modules/utils.py           | 14 ++++++++++++++
 3 files changed, 21 insertions(+), 5 deletions(-)

diff --git a/modules/logits.py b/modules/logits.py
index 32aef7ae5d..56a2057291 100644
--- a/modules/logits.py
+++ b/modules/logits.py
@@ -7,6 +7,7 @@
 from modules.logging_colors import logger
 from modules.models import load_model
 from modules.text_generation import generate_reply
+from modules.utils import check_model_loaded
 
 global_scores = None
 
@@ -33,9 +34,9 @@ def get_next_logits(*args, **kwargs):
 
 
 def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, return_dict=False):
-    if shared.model is None:
-        logger.error("No model is loaded! Select one in the Model tab.")
-        return 'Error: No model is loaded1 Select one in the Model tab.', previous
+    model_is_loaded, error_message = check_model_loaded()
+    if not model_is_loaded:
+        return error_message, previous
 
     # llama.cpp case
     if shared.model.__class__.__name__ == 'LlamaServer':
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 7e48a2f680..c0c0350d42 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -14,6 +14,7 @@
 from modules.extensions import apply_extensions
 from modules.html_generator import generate_basic_html
 from modules.logging_colors import logger
+from modules.utils import check_model_loaded
 
 
 def generate_reply(*args, **kwargs):
@@ -34,8 +35,8 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
     # Find the appropriate generation function
     generate_func = apply_extensions('custom_generate_reply')
     if generate_func is None:
-        if shared.model_name == 'None' or shared.model is None:
-            logger.error("No model is loaded! Select one in the Model tab.")
+        model_is_loaded, error_message = check_model_loaded()
+        if not model_is_loaded:
             yield ''
             return
 
diff --git a/modules/utils.py b/modules/utils.py
index 7732413906..0e390d0899 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -72,6 +72,20 @@ def natural_keys(text):
     return [atoi(c) for c in re.split(r'(\d+)', text)]
 
 
+def check_model_loaded():
+    if shared.model_name == 'None' or shared.model is None:
+        if len(get_available_models()) <= 1:
+            error_msg = "No model is loaded.\n\nTo get started:\n1) Place a GGUF file in your user_data/models folder\n2) Go to the Model tab and select it"
+            logger.error(error_msg)
+            return False, error_msg
+        else:
+            error_msg = "No model is loaded. Please select one in the Model tab."
+            logger.error(error_msg)
+            return False, error_msg
+
+    return True, None
+
+
 def get_available_models():
     # Get all GGUF files
     gguf_files = get_available_ggufs()

From 4920981b140862f3b085f614b83269e6ac228605 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 9 May 2025 20:35:38 -0700
Subject: [PATCH 0491/1701] UI: Remove the typing cursor

---
 modules/chat.py | 10 ++--------
 1 file changed, 2 insertions(+), 8 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 403d05e112..b83c4bfecf 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -399,16 +399,13 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
 
         # Extract the reply
         if state['mode'] in ['chat', 'chat-instruct']:
-            visible_reply = re.sub("(||{{user}})", state['name1'], reply + '▍')
+            visible_reply = re.sub("(||{{user}})", state['name1'], reply)
         else:
-            visible_reply = reply + '▍'
+            visible_reply = reply
 
         visible_reply = html.escape(visible_reply)
 
         if shared.stop_everything:
-            if output['visible'][-1][1].endswith('▍'):
-                output['visible'][-1][1] = output['visible'][-1][1][:-1]
-
             output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
             yield output
             return
@@ -424,9 +421,6 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             if is_stream:
                 yield output
 
-    if output['visible'][-1][1].endswith('▍'):
-        output['visible'][-1][1] = output['visible'][-1][1][:-1]
-
     output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
     yield output
 

From 47d47585095da3a76988eabe52765a332a668d55 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 10 May 2025 17:46:00 -0700
Subject: [PATCH 0492/1701] Fix #6970

---
 modules/shared.py | 8 ++------
 1 file changed, 2 insertions(+), 6 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index f2698bd20e..4e0a20db94 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -128,9 +128,9 @@
 group.add_argument('--streaming-llm', action='store_true', help='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
 
 # Cache
-group = parser.add_argument_group('Context and cache management')
+group = parser.add_argument_group('Context and cache')
 group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=8192, metavar='N', help='Context size in tokens.')
-group.add_argument('--cache_type', type=str, default='fp16', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).')
+group.add_argument('--cache-type', '--cache_type', type=str, default='fp16', metavar='N', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).')
 
 # Speculative decoding
 group = parser.add_argument_group('Speculative decoding')
@@ -159,10 +159,6 @@
 group = parser.add_argument_group('TensorRT-LLM')
 group.add_argument('--cpp-runner', action='store_true', help='Use the ModelRunnerCpp runner, which is faster than the default ModelRunner but doesn\'t support streaming yet.')
 
-# Cache
-group = parser.add_argument_group('Cache')
-group.add_argument('--cache_type', type=str, default='fp16', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.')
-
 # DeepSpeed
 group = parser.add_argument_group('DeepSpeed')
 group.add_argument('--deepspeed', action='store_true', help='Enable the use of DeepSpeed ZeRO-3 for inference via the Transformers integration.')

From 006a866079d4a719f4405efdbbe18c03e106c541 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 10 May 2025 17:55:48 -0700
Subject: [PATCH 0493/1701] Fix API failing to cancel streams (attempt), closes
 #6966

---
 extensions/openai/script.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index a995da9d45..66f38501af 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -118,6 +118,7 @@ async def generator():
                 for resp in response:
                     disconnected = await request.is_disconnected()
                     if disconnected:
+                        stop_everything_event()
                         break
 
                     yield {"data": json.dumps(resp)}
@@ -141,6 +142,7 @@ async def generator():
                 for resp in response:
                     disconnected = await request.is_disconnected()
                     if disconnected:
+                        stop_everything_event()
                         break
 
                     yield {"data": json.dumps(resp)}

From 0c5fa3728e8f0505692966f7a296e6561566c7bd Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 10 May 2025 19:12:40 -0700
Subject: [PATCH 0494/1701] Revert "Fix API failing to cancel streams
 (attempt), closes #6966"

This reverts commit 006a866079d4a719f4405efdbbe18c03e106c541.
---
 extensions/openai/script.py | 2 --
 1 file changed, 2 deletions(-)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 66f38501af..a995da9d45 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -118,7 +118,6 @@ async def generator():
                 for resp in response:
                     disconnected = await request.is_disconnected()
                     if disconnected:
-                        stop_everything_event()
                         break
 
                     yield {"data": json.dumps(resp)}
@@ -142,7 +141,6 @@ async def generator():
                 for resp in response:
                     disconnected = await request.is_disconnected()
                     if disconnected:
-                        stop_everything_event()
                         break
 
                     yield {"data": json.dumps(resp)}

From e7ac06c1694024594450437f3b899e32ab2ce6e4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 10 May 2025 19:20:04 -0700
Subject: [PATCH 0495/1701] New attempt

---
 modules/llama_cpp_server.py | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 1046969a7e..615f29adc3 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -146,8 +146,9 @@ def generate_with_streaming(self, prompt, state):
             pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
             print()
 
-        # Make a direct request with streaming enabled using a context manager
-        with self.session.post(url, json=payload, stream=True) as response:
+        # Make a request with streaming enabled
+        response = self.session.post(url, json=payload, stream=True)
+        try:
             response.raise_for_status()  # Raise an exception for HTTP errors
 
             full_text = ""
@@ -185,6 +186,9 @@ def generate_with_streaming(self, prompt, state):
                     print(f"Problematic line: {line}")
                     continue
 
+        finally:
+            response.close()
+
     def generate(self, prompt, state):
         output = ""
         for output in self.generate_with_streaming(prompt, state):

From 62c774bf24d35a1ebdcdb9927f8a6c6ae3949c82 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 13 May 2025 06:42:25 -0700
Subject: [PATCH 0496/1701] Revert "New attempt"

This reverts commit e7ac06c1694024594450437f3b899e32ab2ce6e4.
---
 modules/llama_cpp_server.py | 8 ++------
 1 file changed, 2 insertions(+), 6 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 615f29adc3..1046969a7e 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -146,9 +146,8 @@ def generate_with_streaming(self, prompt, state):
             pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
             print()
 
-        # Make a request with streaming enabled
-        response = self.session.post(url, json=payload, stream=True)
-        try:
+        # Make a direct request with streaming enabled using a context manager
+        with self.session.post(url, json=payload, stream=True) as response:
             response.raise_for_status()  # Raise an exception for HTTP errors
 
             full_text = ""
@@ -186,9 +185,6 @@ def generate_with_streaming(self, prompt, state):
                     print(f"Problematic line: {line}")
                     continue
 
-        finally:
-            response.close()
-
     def generate(self, prompt, state):
         output = ""
         for output in self.generate_with_streaming(prompt, state):

From c375b6941395454bef52d9ac0e102c0de3f4d3ee Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 13 May 2025 11:23:33 -0700
Subject: [PATCH 0497/1701] API: Fix llama.cpp generating after disconnect,
 improve disconnect detection, fix deadlock on simultaneous requests

---
 extensions/openai/script.py | 19 +++++++++++++++----
 1 file changed, 15 insertions(+), 4 deletions(-)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index a995da9d45..2b4f274f6d 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -14,6 +14,7 @@
 from fastapi.responses import JSONResponse
 from pydub import AudioSegment
 from sse_starlette import EventSourceResponse
+from starlette.concurrency import iterate_in_threadpool
 
 import extensions.openai.completions as OAIcompletions
 import extensions.openai.images as OAIimages
@@ -115,7 +116,7 @@ async def openai_completions(request: Request, request_data: CompletionRequest):
         async def generator():
             async with streaming_semaphore:
                 response = OAIcompletions.stream_completions(to_dict(request_data), is_legacy=is_legacy)
-                for resp in response:
+                async for resp in iterate_in_threadpool(response):
                     disconnected = await request.is_disconnected()
                     if disconnected:
                         break
@@ -125,7 +126,12 @@ async def generator():
         return EventSourceResponse(generator())  # SSE streaming
 
     else:
-        response = OAIcompletions.completions(to_dict(request_data), is_legacy=is_legacy)
+        response = await asyncio.to_thread(
+            OAIcompletions.completions,
+            to_dict(request_data),
+            is_legacy=is_legacy
+        )
+
         return JSONResponse(response)
 
 
@@ -138,7 +144,7 @@ async def openai_chat_completions(request: Request, request_data: ChatCompletion
         async def generator():
             async with streaming_semaphore:
                 response = OAIcompletions.stream_chat_completions(to_dict(request_data), is_legacy=is_legacy)
-                for resp in response:
+                async for resp in iterate_in_threadpool(response):
                     disconnected = await request.is_disconnected()
                     if disconnected:
                         break
@@ -148,7 +154,12 @@ async def generator():
         return EventSourceResponse(generator())  # SSE streaming
 
     else:
-        response = OAIcompletions.chat_completions(to_dict(request_data), is_legacy=is_legacy)
+        response = await asyncio.to_thread(
+            OAIcompletions.chat_completions,
+            to_dict(request_data),
+            is_legacy=is_legacy
+        )
+
         return JSONResponse(response)
 
 

From 3fa1a899aea3ff2700a20a8bc2da17202d3065e5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 13 May 2025 12:07:59 -0700
Subject: [PATCH 0498/1701] UI: Fix gpu-layers being ignored (closes #6973)

---
 modules/loaders.py         | 2 +-
 modules/models_settings.py | 2 +-
 modules/ui.py              | 2 +-
 modules/ui_model_menu.py   | 2 +-
 4 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/modules/loaders.py b/modules/loaders.py
index 4b76549b84..583b65c218 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -5,7 +5,7 @@
 
 loaders_and_params = OrderedDict({
     'llama.cpp': [
-        'n_gpu_layers',
+        'gpu_layers',
         'threads',
         'threads_batch',
         'batch_size',
diff --git a/modules/models_settings.py b/modules/models_settings.py
index ae589bb3a5..4418e3fb88 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -67,7 +67,7 @@ def get_model_metadata(model):
             elif k.endswith('rope.scaling.factor'):
                 model_settings['compress_pos_emb'] = metadata[k]
             elif k.endswith('block_count'):
-                model_settings['n_gpu_layers'] = metadata[k] + 1
+                model_settings['gpu_layers'] = metadata[k] + 1
 
         if 'tokenizer.chat_template' in metadata:
             template = metadata['tokenizer.chat_template']
diff --git a/modules/ui.py b/modules/ui.py
index b3d4bccf07..eeb6ce92cd 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -105,7 +105,7 @@ def list_model_elements():
         'filter_by_loader',
         'loader',
         'cpu_memory',
-        'n_gpu_layers',
+        'gpu_layers',
         'threads',
         'threads_batch',
         'batch_size',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 1e27255b12..b63a127c06 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -50,7 +50,7 @@ def create_ui():
                     gr.Markdown("## Main options")
                     with gr.Row():
                         with gr.Column():
-                            shared.gradio['n_gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=256, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
+                            shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=256, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')

From 2826c60044a05f316510ef93546b5dbff59b3864 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 13 May 2025 14:45:46 -0700
Subject: [PATCH 0499/1701] Use logger for "Output generated in ..." messages

---
 modules/text_generation.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index c0c0350d42..00b9275a0d 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -470,7 +470,7 @@ def generate_with_streaming(**kwargs):
         t1 = time.time()
         original_tokens = len(original_input_ids[0])
         new_tokens = len(output) - (original_tokens if not shared.is_seq2seq else 0)
-        print(f'Output generated in {(t1-t0):.2f} seconds ({new_tokens/(t1-t0):.2f} tokens/s, {new_tokens} tokens, context {original_tokens}, seed {seed})')
+        logger.info(f'Output generated in {(t1-t0):.2f} seconds ({new_tokens/(t1-t0):.2f} tokens/s, {new_tokens} tokens, context {original_tokens}, seed {seed})')
         return
 
 
@@ -499,7 +499,7 @@ def generate_reply_custom(question, original_question, state, stopping_strings=N
         t1 = time.time()
         original_tokens = len(encode(original_question)[0])
         new_tokens = len(encode(original_question + reply)[0]) - original_tokens
-        print(f'Output generated in {(t1-t0):.2f} seconds ({new_tokens/(t1-t0):.2f} tokens/s, {new_tokens} tokens, context {original_tokens}, seed {state["seed"]})')
+        logger.info(f'Output generated in {(t1-t0):.2f} seconds ({new_tokens/(t1-t0):.2f} tokens/s, {new_tokens} tokens, context {original_tokens}, seed {state["seed"]})')
         return
 
 

From 035cd3e2a906a6094d0f1f298df49c0152f1a2ee Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 13 May 2025 20:09:22 -0700
Subject: [PATCH 0500/1701] UI: Hide the extension install menu in portable
 builds

---
 modules/ui_session.py | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/modules/ui_session.py b/modules/ui_session.py
index 7cf9f6e66e..a4eba66721 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -23,11 +23,15 @@ def create_ui():
                         shared.gradio['bool_menu'] = gr.CheckboxGroup(choices=get_boolean_arguments(), value=get_boolean_arguments(active=True), label="Boolean command-line flags", elem_classes='checkboxgroup-table')
 
             with gr.Column():
-                extension_name = gr.Textbox(lines=1, label='Install or update an extension', info='Enter the GitHub URL below and press Enter. For a list of extensions, see: https://github.com/oobabooga/text-generation-webui-extensions ⚠️  WARNING ⚠️ : extensions can execute arbitrary code. Make sure to inspect their source code before activating them.', interactive=not mu)
-                extension_status = gr.Markdown()
+                if not shared.args.portable:
+                    extension_name = gr.Textbox(lines=1, label='Install or update an extension', info='Enter the GitHub URL below and press Enter. For a list of extensions, see: https://github.com/oobabooga/text-generation-webui-extensions ⚠️  WARNING ⚠️ : extensions can execute arbitrary code. Make sure to inspect their source code before activating them.', interactive=not mu)
+                    extension_status = gr.Markdown()
+                else:
+                    pass
 
         shared.gradio['theme_state'] = gr.Textbox(visible=False, value='dark' if shared.settings['dark_theme'] else 'light')
-        extension_name.submit(clone_or_pull_repository, extension_name, extension_status, show_progress=False)
+        if not shared.args.portable:
+            extension_name.submit(clone_or_pull_repository, extension_name, extension_status, show_progress=False)
 
         # Reset interface event
         shared.gradio['reset_interface'].click(

From c4a715fd1e86e52e3350f8126847524b488a04e2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 13 May 2025 20:14:09 -0700
Subject: [PATCH 0501/1701] UI: Move the LoRA menu under "Other options"

---
 modules/ui_model_menu.py | 28 ++++++++++------------------
 1 file changed, 10 insertions(+), 18 deletions(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index b63a127c06..81ad1a5328 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -26,25 +26,12 @@ def create_ui():
         with gr.Row():
             with gr.Column():
                 with gr.Row():
-                    with gr.Column():
-                        with gr.Row():
-                            shared.gradio['model_menu'] = gr.Dropdown(choices=utils.get_available_models(), value=lambda: shared.model_name, label='Model', elem_classes='slim-dropdown', interactive=not mu)
-                            ui.create_refresh_button(shared.gradio['model_menu'], lambda: None, lambda: {'choices': utils.get_available_models()}, 'refresh-button', interactive=not mu)
-                            shared.gradio['load_model'] = gr.Button("Load", elem_classes='refresh-button', interactive=not mu)
-                            shared.gradio['unload_model'] = gr.Button("Unload", elem_classes='refresh-button', interactive=not mu)
-                            shared.gradio['save_model_settings'] = gr.Button("Save settings", elem_classes='refresh-button', interactive=not mu)
-
-                    with gr.Column():
-                        if shared.args.portable:
-                            pass
-                        else:
-                            with gr.Row():
-                                shared.gradio['lora_menu'] = gr.Dropdown(multiselect=True, choices=utils.get_available_loras(), value=shared.lora_names, label='LoRA(s)', elem_classes='slim-dropdown', interactive=not mu)
-                                ui.create_refresh_button(shared.gradio['lora_menu'], lambda: None, lambda: {'choices': utils.get_available_loras(), 'value': shared.lora_names}, 'refresh-button', interactive=not mu)
-                                shared.gradio['lora_menu_apply'] = gr.Button(value='Apply LoRAs', elem_classes='refresh-button', interactive=not mu)
+                    shared.gradio['model_menu'] = gr.Dropdown(choices=utils.get_available_models(), value=lambda: shared.model_name, label='Model', elem_classes='slim-dropdown', interactive=not mu)
+                    ui.create_refresh_button(shared.gradio['model_menu'], lambda: None, lambda: {'choices': utils.get_available_models()}, 'refresh-button', interactive=not mu)
+                    shared.gradio['load_model'] = gr.Button("Load", elem_classes='refresh-button', interactive=not mu)
+                    shared.gradio['unload_model'] = gr.Button("Unload", elem_classes='refresh-button', interactive=not mu)
+                    shared.gradio['save_model_settings'] = gr.Button("Save settings", elem_classes='refresh-button', interactive=not mu)
 
-        with gr.Row():
-            with gr.Column():
                 shared.gradio['loader'] = gr.Dropdown(label="Model loader", choices=loaders.loaders_and_params.keys() if not shared.args.portable else ['llama.cpp'], value=None)
                 with gr.Blocks():
                     gr.Markdown("## Main options")
@@ -113,6 +100,11 @@ def create_ui():
                                 shared.gradio['no_sdpa'] = gr.Checkbox(label="no_sdpa", value=shared.args.no_sdpa)
                                 shared.gradio['cfg_cache'] = gr.Checkbox(label="cfg-cache", value=shared.args.cfg_cache, info='Necessary to use CFG with this loader.')
                                 shared.gradio['no_use_fast'] = gr.Checkbox(label="no_use_fast", value=shared.args.no_use_fast, info='Set use_fast=False while loading the tokenizer.')
+                                with gr.Row():
+                                    shared.gradio['lora_menu'] = gr.Dropdown(multiselect=True, choices=utils.get_available_loras(), value=shared.lora_names, label='LoRA(s)', elem_classes='slim-dropdown', interactive=not mu)
+                                    ui.create_refresh_button(shared.gradio['lora_menu'], lambda: None, lambda: {'choices': utils.get_available_loras(), 'value': shared.lora_names}, 'refresh-button', interactive=not mu)
+                                    shared.gradio['lora_menu_apply'] = gr.Button(value='Apply LoRAs', elem_classes='refresh-button', interactive=not mu)
+
 
             with gr.Column():
                 with gr.Tab("Download"):

From 5534d01da0913d315709a6adacd075639a6cffec Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Fri, 16 May 2025 00:07:37 -0300
Subject: [PATCH 0502/1701] Estimate the VRAM for GGUF models + autoset
 `gpu-layers` (#6980)

---
 css/main.css                |  14 +++-
 modules/llama_cpp_server.py |   3 +
 modules/models.py           |   1 -
 modules/models_settings.py  | 151 +++++++++++++++++++++++++++++++++++-
 modules/ui_model_menu.py    |  16 +++-
 server.py                   |  12 +++
 6 files changed, 193 insertions(+), 4 deletions(-)

diff --git a/css/main.css b/css/main.css
index 30089aca9b..0902b18466 100644
--- a/css/main.css
+++ b/css/main.css
@@ -569,7 +569,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 .dark .message-body :not(pre) > code {
-    background-color: rgb(255 255 255 / 12.5%);
+    background-color: rgb(255 255 255 / 10%);
 }
 
 #chat-input {
@@ -1386,3 +1386,15 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 strong {
     font-weight: bold;
 }
+
+.min.svelte-1ybaih5 {
+    min-height: 0;
+}
+
+#vram-info .value {
+    color: #008d00;
+}
+
+.dark #vram-info .value {
+    color: #07ff07;
+}
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 1046969a7e..3fc7a0cc0a 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -282,8 +282,10 @@ def _start_server(self):
             cmd.append("--no-kv-offload")
         if shared.args.row_split:
             cmd += ["--split-mode", "row"]
+        cache_type = "fp16"
         if shared.args.cache_type != "fp16" and shared.args.cache_type in llamacpp_valid_cache_types:
             cmd += ["--cache-type-k", shared.args.cache_type, "--cache-type-v", shared.args.cache_type]
+            cache_type = shared.args.cache_type
         if shared.args.compress_pos_emb != 1:
             cmd += ["--rope-freq-scale", str(1.0 / shared.args.compress_pos_emb)]
         if shared.args.rope_freq_base > 0:
@@ -343,6 +345,7 @@ def _start_server(self):
             print(' '.join(str(item) for item in cmd[1:]))
             print()
 
+        logger.info(f"Using gpu_layers={shared.args.gpu_layers} | ctx_size={shared.args.ctx_size} | cache_type={cache_type}")
         # Start the server with pipes for output
         self.process = subprocess.Popen(
             cmd,
diff --git a/modules/models.py b/modules/models.py
index d0b0402a82..9ecee803a1 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -71,7 +71,6 @@ def llama_cpp_server_loader(model_name):
     else:
         model_file = sorted(Path(f'{shared.args.model_dir}/{model_name}').glob('*.gguf'))[0]
 
-    logger.info(f"llama.cpp weights detected: \"{model_file}\"")
     try:
         model = LlamaServer(model_file)
         return model, model
diff --git a/modules/models_settings.py b/modules/models_settings.py
index 4418e3fb88..a8e175947f 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -1,7 +1,11 @@
+import functools
 import json
 import re
+import subprocess
+from math import exp
 from pathlib import Path
 
+import gradio as gr
 import yaml
 
 from modules import chat, loaders, metadata_gguf, shared, ui
@@ -216,7 +220,17 @@ def apply_model_settings_to_state(model, state):
 
     for k in model_settings:
         if k in state:
-            state[k] = model_settings[k]
+            if k == 'gpu_layers':
+                available_vram = get_nvidia_free_vram()
+                n_layers = model_settings[k]
+                if available_vram > 0:
+                    tolerance = 906
+                    while n_layers > 0 and estimate_vram(model, n_layers, state['ctx_size'], state['cache_type']) > available_vram - tolerance:
+                        n_layers -= 1
+
+                state[k] = gr.update(value=n_layers, maximum=model_settings[k])
+            else:
+                state[k] = model_settings[k]
 
     return state
 
@@ -277,3 +291,138 @@ def save_instruction_template(model, template):
         yield (f"Instruction template for `{model}` unset in `{p}`, as the value for template was `{template}`.")
     else:
         yield (f"Instruction template for `{model}` saved to `{p}` as `{template}`.")
+
+
+@functools.lru_cache(maxsize=None)
+def get_gguf_metadata_cached(model_file):
+    return metadata_gguf.load_metadata(model_file)
+
+
+def get_model_size_mb(model_file: Path) -> float:
+    filename = model_file.name
+
+    # Check for multipart pattern
+    match = re.match(r'(.+)-\d+-of-\d+\.gguf$', filename)
+
+    if match:
+        # It's a multipart file, find all matching parts
+        base_pattern = match.group(1)
+        part_files = sorted(model_file.parent.glob(f'{base_pattern}-*-of-*.gguf'))
+        total_size = sum(p.stat().st_size for p in part_files)
+    else:
+        # Single part
+        total_size = model_file.stat().st_size
+
+    return total_size / (1024 ** 2)  # Return size in MB
+
+
+def estimate_vram(gguf_file, gpu_layers, ctx_size, cache_type):
+    model_file = Path(f'{shared.args.model_dir}/{gguf_file}')
+    metadata = get_gguf_metadata_cached(model_file)
+    size_in_mb = get_model_size_mb(model_file)
+
+    # Extract values from metadata
+    n_layers = None
+    n_kv_heads = None
+    embedding_dim = None
+    context_length = None
+    feed_forward_dim = None
+
+    for key, value in metadata.items():
+        if key.endswith('.block_count'):
+            n_layers = value
+        elif key.endswith('.attention.head_count_kv'):
+            n_kv_heads = value
+        elif key.endswith('.embedding_length'):
+            embedding_dim = value
+        elif key.endswith('.context_length'):
+            context_length = value
+        elif key.endswith('.feed_forward_length'):
+            feed_forward_dim = value
+
+    if gpu_layers > n_layers:
+        gpu_layers = n_layers
+
+    # Convert cache_type to numeric
+    if cache_type == 'q4_0':
+        cache_type = 4
+    elif cache_type == 'q8_0':
+        cache_type = 8
+    else:
+        cache_type = 16
+
+    # Derived features
+    size_per_layer = size_in_mb / max(n_layers, 1e-6)
+    context_per_layer = context_length / max(n_layers, 1e-6)
+    ffn_per_embedding = feed_forward_dim / max(embedding_dim, 1e-6)
+    kv_cache_factor = n_kv_heads * cache_type * ctx_size
+
+    # Helper function for smaller
+    def smaller(x, y):
+        return 1 if x < y else 0
+
+    # Calculate VRAM using the model
+    # Details: https://oobabooga.github.io/blog/posts/gguf-vram-formula/
+    vram = (
+        (size_per_layer - 21.19195204848197)
+        * exp(0.0001047328491557063 * size_in_mb * smaller(ffn_per_embedding, 2.671096993407845))
+        + 0.0006621544775632052 * context_per_layer
+        + 3.34664386576376e-05 * kv_cache_factor
+    ) * (1.363306170123392 + gpu_layers) + 1255.163594536052
+
+    return vram
+
+
+def get_nvidia_free_vram():
+    """
+    Calculates the total free VRAM across all NVIDIA GPUs by parsing nvidia-smi output.
+
+    Returns:
+        int: The total free VRAM in MiB summed across all detected NVIDIA GPUs.
+             Returns -1 if nvidia-smi command fails (not found, error, etc.).
+             Returns 0 if nvidia-smi succeeds but no GPU memory info found.
+    """
+    try:
+        # Execute nvidia-smi command
+        result = subprocess.run(
+            ['nvidia-smi'],
+            capture_output=True,
+            text=True,
+            check=False
+        )
+
+        # Check if nvidia-smi returned an error
+        if result.returncode != 0:
+            return -1
+
+        # Parse the output for memory usage patterns
+        output = result.stdout
+
+        # Find memory usage like "XXXXMiB / YYYYMiB"
+        # Captures used and total memory for each GPU
+        matches = re.findall(r"(\d+)\s*MiB\s*/\s*(\d+)\s*MiB", output)
+
+        if not matches:
+            # No GPUs found in expected format
+            return 0
+
+        total_free_vram_mib = 0
+        for used_mem_str, total_mem_str in matches:
+            try:
+                used_mib = int(used_mem_str)
+                total_mib = int(total_mem_str)
+                total_free_vram_mib += (total_mib - used_mib)
+            except ValueError:
+                # Skip malformed entries
+                pass
+
+        return total_free_vram_mib
+
+    except FileNotFoundError:
+        raise
+        # nvidia-smi not found (likely no NVIDIA drivers installed)
+        return -1
+    except Exception:
+        raise
+        # Handle any other unexpected exceptions
+        return -1
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 81ad1a5328..2353f39c51 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -11,6 +11,7 @@
 from modules.models import load_model, unload_model
 from modules.models_settings import (
     apply_model_settings_to_state,
+    estimate_vram,
     get_model_metadata,
     save_instruction_template,
     save_model_settings,
@@ -44,6 +45,7 @@ def create_ui():
                             shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
 
                         with gr.Column():
+                            shared.gradio['vram_info'] = gr.HTML(value=lambda: estimate_vram_wrapper(shared.args.model, shared.args.gpu_layers, shared.args.ctx_size, shared.args.cache_type))
                             shared.gradio['flash_attn'] = gr.Checkbox(label="flash-attn", value=shared.args.flash_attn, info='Use flash-attention.')
                             shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming-llm", value=shared.args.streaming_llm, info='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
                             shared.gradio['load_in_8bit'] = gr.Checkbox(label="load-in-8bit", value=shared.args.load_in_8bit)
@@ -105,7 +107,6 @@ def create_ui():
                                     ui.create_refresh_button(shared.gradio['lora_menu'], lambda: None, lambda: {'choices': utils.get_available_loras(), 'value': shared.lora_names}, 'refresh-button', interactive=not mu)
                                     shared.gradio['lora_menu_apply'] = gr.Button(value='Apply LoRAs', elem_classes='refresh-button', interactive=not mu)
 
-
             with gr.Column():
                 with gr.Tab("Download"):
                     shared.gradio['custom_model_menu'] = gr.Textbox(label="Download model or LoRA", info="Enter the Hugging Face username/model path, for instance: facebook/galactica-125m. To specify a branch, add it at the end after a \":\" character like this: facebook/galactica-125m:main. To download a single file, enter its name in the second box.", interactive=not mu)
@@ -148,6 +149,11 @@ def create_event_handlers():
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         save_model_settings, gradio('model_menu', 'interface_state'), gradio('model_status'), show_progress=False)
 
+    shared.gradio['model_menu'].change(estimate_vram_wrapper, gradio('model_menu', 'gpu_layers', 'ctx_size', 'cache_type'), gradio('vram_info'), show_progress=False)
+    shared.gradio['gpu_layers'].change(estimate_vram_wrapper, gradio('model_menu', 'gpu_layers', 'ctx_size', 'cache_type'), gradio('vram_info'), show_progress=False)
+    shared.gradio['ctx_size'].change(estimate_vram_wrapper, gradio('model_menu', 'gpu_layers', 'ctx_size', 'cache_type'), gradio('vram_info'), show_progress=False)
+    shared.gradio['cache_type'].change(estimate_vram_wrapper, gradio('model_menu', 'gpu_layers', 'ctx_size', 'cache_type'), gradio('vram_info'), show_progress=False)
+
     if not shared.args.portable:
         shared.gradio['lora_menu_apply'].click(load_lora_wrapper, gradio('lora_menu'), gradio('model_status'), show_progress=False)
 
@@ -275,6 +281,14 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
         yield traceback.format_exc().replace('\n', '\n\n')
 
 
+def estimate_vram_wrapper(model, gpu_layers, ctx_size, cache_type):
+    if model in ["None", None]:
+        return "
            Estimated VRAM to load the model:"
+
+    result = estimate_vram(model, gpu_layers, ctx_size, cache_type)
+    return f"
            Estimated VRAM to load the model: {result:.0f} MiB"
+
+
 def update_truncation_length(current_length, state):
     if 'loader' in state:
         if state['loader'].lower().startswith('exllama') or state['loader'] == 'llama.cpp':
diff --git a/server.py b/server.py
index b0b9e63317..c35d65a888 100644
--- a/server.py
+++ b/server.py
@@ -49,8 +49,10 @@
 from modules.LoRA import add_lora_to_model
 from modules.models import load_model, unload_model_if_idle
 from modules.models_settings import (
+    estimate_vram,
     get_fallback_settings,
     get_model_metadata,
+    get_nvidia_free_vram,
     update_model_parameters
 )
 from modules.shared import do_cmd_flags_warnings
@@ -248,6 +250,16 @@ def create_interface():
         model_settings = get_model_metadata(model_name)
         update_model_parameters(model_settings, initial=True)  # hijack the command-line arguments
 
+        if 'gpu_layers' not in shared.provided_arguments:
+            available_vram = get_nvidia_free_vram()
+            if available_vram > 0:
+                n_layers = model_settings['gpu_layers']
+                tolerance = 906
+                while n_layers > 0 and estimate_vram(model_name, n_layers, shared.args.ctx_size, shared.args.cache_type) > available_vram - tolerance:
+                    n_layers -= 1
+
+            shared.args.gpu_layers = n_layers
+
         # Load the model
         shared.model, shared.tokenizer = load_model(model_name)
         if shared.args.lora:

From 041248cc9f321aa6ff2e706083cdb776e3bf8d21 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 15 May 2025 20:10:02 -0700
Subject: [PATCH 0503/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 3a059c9129..45bb5c85d3 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index ebc33216a5..4e01198981 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 8ec6898f6a..a3bd1350a5 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index afc869c888..a52f2d6489 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 8d7d29b785..929b1d8609 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -29,8 +29,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index d69aae1827..bd7c4a4f25 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 540c9ac8e6..b5aa1cf798 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 3bb5a74a23..bc320c2731 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 95319d7588..7995939874 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 4b49b4e17d..ca16e4c7fd 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index a6ebda30be..18e1c506a4 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -15,6 +15,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index bb5ba8ad39..693f471200 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 3d17dd49e6..8635d11e40 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index ff9fa04cbb..e844596e6b 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index e17f8ce7ed..9b7435d161 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index dd01b3a85a..513b7a159e 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.11.0/llama_cpp_binaries-0.11.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 8cb73b78e14154040ac4a2f7dd33dc7d46121108 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 15 May 2025 20:10:34 -0700
Subject: [PATCH 0504/1701] Update ExLlamaV3

---
 requirements/full/requirements.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_noavx2.txt        | 4 ++--
 4 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 45bb5c85d3..af5f7d8a88 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -32,8 +32,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index a52f2d6489..363365bf17 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -31,5 +31,5 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 929b1d8609..2843fed27a 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -32,5 +32,5 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index bc320c2731..89947cbe5f 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -32,8 +32,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a8/exllamav3-0.0.1a8+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From fd612979330ee0009ccbb14ac5bff894b675bb82 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 15 May 2025 21:19:19 -0700
Subject: [PATCH 0505/1701] Lint

---
 extensions/openai/completions.py    | 8 ++++----
 extensions/openai/utils.py          | 6 +++---
 extensions/superboogav2/chromadb.py | 3 ++-
 modules/tensorrt_llm.py             | 4 ++--
 4 files changed, 11 insertions(+), 10 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index ed0bcc40ff..5181b18bff 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -1,14 +1,14 @@
 import copy
-import time
 import json
+import time
 from collections import deque
 
 import tiktoken
+from pydantic import ValidationError
 
 from extensions.openai.errors import InvalidRequestError
-from extensions.openai.utils import debug_msg, getToolCallId, parseToolCall
 from extensions.openai.typing import ToolDefinition
-from pydantic import ValidationError
+from extensions.openai.utils import debug_msg, getToolCallId, parseToolCall
 from modules import shared
 from modules.chat import (
     generate_chat_prompt,
@@ -141,7 +141,7 @@ def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, p
 
     tools = None
     if 'tools' in body and body['tools'] is not None and isinstance(body['tools'], list) and len(body['tools']) > 0:
-        tools = validateTools(body['tools']) # raises InvalidRequestError if validation fails
+        tools = validateTools(body['tools'])  # raises InvalidRequestError if validation fails
 
     messages = body['messages']
     for m in messages:
diff --git a/extensions/openai/utils.py b/extensions/openai/utils.py
index 8cb856ff32..9a1de2e769 100644
--- a/extensions/openai/utils.py
+++ b/extensions/openai/utils.py
@@ -1,9 +1,9 @@
 import base64
-import os
-import time
 import json
+import os
 import random
 import re
+import time
 import traceback
 from typing import Callable, Optional
 
@@ -91,7 +91,7 @@ def parseToolCall(answer: str, tool_names: list[str]):
         return matches
 
     # Define the regex pattern to find the JSON content wrapped in , , , and other tags observed from various models
-    patterns = [ r"(```[^\n]*)\n(.*?)```", r"<([^>]+)>(.*?)" ]
+    patterns = [r"(```[^\n]*)\n(.*?)```", r"<([^>]+)>(.*?)"]
 
     for pattern in patterns:
         for match in re.finditer(pattern, answer, re.DOTALL):
diff --git a/extensions/superboogav2/chromadb.py b/extensions/superboogav2/chromadb.py
index f4f7782112..9344e25c67 100644
--- a/extensions/superboogav2/chromadb.py
+++ b/extensions/superboogav2/chromadb.py
@@ -1,10 +1,11 @@
 import math
 import random
 import threading
-import torch
+
 import chromadb
 import numpy as np
 import posthog
+import torch
 from chromadb.config import Settings
 from chromadb.utils import embedding_functions
 
diff --git a/modules/tensorrt_llm.py b/modules/tensorrt_llm.py
index 73178c395e..0527d493f8 100644
--- a/modules/tensorrt_llm.py
+++ b/modules/tensorrt_llm.py
@@ -1,15 +1,15 @@
 from pathlib import Path
 
+import tensorrt_llm
 import torch
+from tensorrt_llm.runtime import ModelRunner, ModelRunnerCpp
 
-import tensorrt_llm
 from modules import shared
 from modules.logging_colors import logger
 from modules.text_generation import (
     get_max_prompt_length,
     get_reply_from_output_ids
 )
-from tensorrt_llm.runtime import ModelRunner, ModelRunnerCpp
 
 
 class TensorRTLLMModel:

From cbf4daf1c8d149206da80892ced0220cf858ebb7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 15 May 2025 21:21:54 -0700
Subject: [PATCH 0506/1701] Hide the LoRA menu in portable mode

---
 modules/ui_model_menu.py | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 2353f39c51..a19111249d 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -102,10 +102,11 @@ def create_ui():
                                 shared.gradio['no_sdpa'] = gr.Checkbox(label="no_sdpa", value=shared.args.no_sdpa)
                                 shared.gradio['cfg_cache'] = gr.Checkbox(label="cfg-cache", value=shared.args.cfg_cache, info='Necessary to use CFG with this loader.')
                                 shared.gradio['no_use_fast'] = gr.Checkbox(label="no_use_fast", value=shared.args.no_use_fast, info='Set use_fast=False while loading the tokenizer.')
-                                with gr.Row():
-                                    shared.gradio['lora_menu'] = gr.Dropdown(multiselect=True, choices=utils.get_available_loras(), value=shared.lora_names, label='LoRA(s)', elem_classes='slim-dropdown', interactive=not mu)
-                                    ui.create_refresh_button(shared.gradio['lora_menu'], lambda: None, lambda: {'choices': utils.get_available_loras(), 'value': shared.lora_names}, 'refresh-button', interactive=not mu)
-                                    shared.gradio['lora_menu_apply'] = gr.Button(value='Apply LoRAs', elem_classes='refresh-button', interactive=not mu)
+                                if not shared.args.portable:
+                                    with gr.Row():
+                                        shared.gradio['lora_menu'] = gr.Dropdown(multiselect=True, choices=utils.get_available_loras(), value=shared.lora_names, label='LoRA(s)', elem_classes='slim-dropdown', interactive=not mu)
+                                        ui.create_refresh_button(shared.gradio['lora_menu'], lambda: None, lambda: {'choices': utils.get_available_loras(), 'value': shared.lora_names}, 'refresh-button', interactive=not mu)
+                                        shared.gradio['lora_menu_apply'] = gr.Button(value='Apply LoRAs', elem_classes='refresh-button', interactive=not mu)
 
             with gr.Column():
                 with gr.Tab("Download"):

From 93e1850a2c1eef8fe914bd020dde3e94d6b54f6c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 15 May 2025 21:42:15 -0700
Subject: [PATCH 0507/1701] Only show the VRAM info for llama.cpp

---
 modules/loaders.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/loaders.py b/modules/loaders.py
index 583b65c218..79a7a4a35e 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -28,6 +28,7 @@
         'device_draft',
         'ctx_size_draft',
         'speculative_decoding_accordion',
+        'vram_info',
     ],
     'Transformers': [
         'gpu_split',

From 4925c307cfc97c1ca549b71db6f1aaaf82fd4fb2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 09:07:38 -0700
Subject: [PATCH 0508/1701] Auto-adjust GPU layers on context size and cache
 type changes + many fixes

---
 modules/models_settings.py | 80 +++++++++++++++++++++++++++++++-------
 modules/ui_model_menu.py   | 46 +++++++++++++++-------
 server.py                  | 25 ++++++------
 3 files changed, 111 insertions(+), 40 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index a8e175947f..6ea6660c45 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -213,24 +213,26 @@ def apply_model_settings_to_state(model, state):
     model_settings = get_model_metadata(model)
     if 'loader' in model_settings:
         loader = model_settings.pop('loader')
-
-        # If the user is using an alternative loader for the same model type, let them keep using it
         if not (loader == 'ExLlamav2_HF' and state['loader'] in ['ExLlamav2']):
             state['loader'] = loader
 
     for k in model_settings:
-        if k in state:
-            if k == 'gpu_layers':
-                available_vram = get_nvidia_free_vram()
-                n_layers = model_settings[k]
-                if available_vram > 0:
-                    tolerance = 906
-                    while n_layers > 0 and estimate_vram(model, n_layers, state['ctx_size'], state['cache_type']) > available_vram - tolerance:
-                        n_layers -= 1
-
-                state[k] = gr.update(value=n_layers, maximum=model_settings[k])
-            else:
-                state[k] = model_settings[k]
+        if k in state and k != 'gpu_layers':  # Skip gpu_layers, handle separately
+            state[k] = model_settings[k]
+
+    # Handle GPU layers and VRAM update for llama.cpp
+    if state['loader'] == 'llama.cpp' and 'gpu_layers' in model_settings:
+        vram_info, gpu_layers_update = update_gpu_layers_and_vram(
+            state['loader'],
+            model,
+            model_settings['gpu_layers'],
+            state['ctx_size'],
+            state['cache_type'],
+            auto_adjust=True
+        )
+
+        state['gpu_layers'] = gpu_layers_update
+        state['vram_info'] = vram_info
 
     return state
 
@@ -426,3 +428,53 @@ def get_nvidia_free_vram():
         raise
         # Handle any other unexpected exceptions
         return -1
+
+
+def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type, auto_adjust=False, for_ui=True):
+    """
+    Unified function to handle GPU layers and VRAM updates.
+
+    Args:
+        for_ui: If True, returns Gradio updates. If False, returns raw values.
+
+    Returns:
+        - If for_ui=True: (vram_info_update, gpu_layers_update) or just vram_info_update
+        - If for_ui=False: (vram_usage, adjusted_layers) or just vram_usage
+    """
+    if loader != 'llama.cpp' or model in ["None", None]:
+        vram_info = "
            Estimated VRAM to load the model:"
+        if for_ui:
+            return (vram_info, gr.update()) if auto_adjust else vram_info
+        else:
+            return (0, gpu_layers) if auto_adjust else 0
+
+    current_layers = gpu_layers
+    max_layers = gpu_layers
+
+    if auto_adjust:
+        # Get max layers from model metadata
+        model_settings = get_model_metadata(model)
+        max_layers = model_settings.get('gpu_layers', gpu_layers)
+
+        # Auto-adjust based on available VRAM
+        available_vram = get_nvidia_free_vram()
+        if available_vram > 0:
+            tolerance = 906
+            current_layers = max_layers
+            while current_layers > 0 and estimate_vram(model, current_layers, ctx_size, cache_type) > available_vram - tolerance:
+                current_layers -= 1
+
+    # Calculate VRAM with current layers
+    vram_usage = estimate_vram(model, current_layers, ctx_size, cache_type)
+
+    if for_ui:
+        vram_info = f"
            Estimated VRAM to load the model: {vram_usage:.0f} MiB"
+        if auto_adjust:
+            return vram_info, gr.update(value=current_layers, maximum=max_layers)
+        else:
+            return vram_info
+    else:
+        if auto_adjust:
+            return vram_usage, current_layers
+        else:
+            return vram_usage
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index a19111249d..b6febb50f6 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -11,10 +11,10 @@
 from modules.models import load_model, unload_model
 from modules.models_settings import (
     apply_model_settings_to_state,
-    estimate_vram,
     get_model_metadata,
     save_instruction_template,
     save_model_settings,
+    update_gpu_layers_and_vram,
     update_model_parameters
 )
 from modules.utils import gradio
@@ -45,7 +45,7 @@ def create_ui():
                             shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
 
                         with gr.Column():
-                            shared.gradio['vram_info'] = gr.HTML(value=lambda: estimate_vram_wrapper(shared.args.model, shared.args.gpu_layers, shared.args.ctx_size, shared.args.cache_type))
+                            shared.gradio['vram_info'] = gr.HTML(value=get_initial_vram_info())
                             shared.gradio['flash_attn'] = gr.Checkbox(label="flash-attn", value=shared.args.flash_attn, info='Use flash-attention.')
                             shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming-llm", value=shared.args.streaming_llm, info='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
                             shared.gradio['load_in_8bit'] = gr.Checkbox(label="load-in-8bit", value=shared.args.load_in_8bit)
@@ -150,10 +150,18 @@ def create_event_handlers():
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         save_model_settings, gradio('model_menu', 'interface_state'), gradio('model_status'), show_progress=False)
 
-    shared.gradio['model_menu'].change(estimate_vram_wrapper, gradio('model_menu', 'gpu_layers', 'ctx_size', 'cache_type'), gradio('vram_info'), show_progress=False)
-    shared.gradio['gpu_layers'].change(estimate_vram_wrapper, gradio('model_menu', 'gpu_layers', 'ctx_size', 'cache_type'), gradio('vram_info'), show_progress=False)
-    shared.gradio['ctx_size'].change(estimate_vram_wrapper, gradio('model_menu', 'gpu_layers', 'ctx_size', 'cache_type'), gradio('vram_info'), show_progress=False)
-    shared.gradio['cache_type'].change(estimate_vram_wrapper, gradio('model_menu', 'gpu_layers', 'ctx_size', 'cache_type'), gradio('vram_info'), show_progress=False)
+    # For ctx_size and cache_type - auto-adjust GPU layers
+    for param in ['ctx_size', 'cache_type']:
+        shared.gradio[param].change(
+            partial(update_gpu_layers_and_vram, auto_adjust=True),
+            gradio('loader', 'model_menu', 'gpu_layers', 'ctx_size', 'cache_type'),
+            gradio('vram_info', 'gpu_layers'), show_progress=False)
+
+    # For manual gpu_layers changes - only update VRAM
+    shared.gradio['gpu_layers'].change(
+        partial(update_gpu_layers_and_vram, auto_adjust=False),
+        gradio('loader', 'model_menu', 'gpu_layers', 'ctx_size', 'cache_type'),
+        gradio('vram_info'), show_progress=False)
 
     if not shared.args.portable:
         shared.gradio['lora_menu_apply'].click(load_lora_wrapper, gradio('lora_menu'), gradio('model_status'), show_progress=False)
@@ -282,14 +290,6 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
         yield traceback.format_exc().replace('\n', '\n\n')
 
 
-def estimate_vram_wrapper(model, gpu_layers, ctx_size, cache_type):
-    if model in ["None", None]:
-        return "
            Estimated VRAM to load the model:"
-
-    result = estimate_vram(model, gpu_layers, ctx_size, cache_type)
-    return f"
            Estimated VRAM to load the model: {result:.0f} MiB"
-
-
 def update_truncation_length(current_length, state):
     if 'loader' in state:
         if state['loader'].lower().startswith('exllama') or state['loader'] == 'llama.cpp':
@@ -298,10 +298,26 @@ def update_truncation_length(current_length, state):
     return current_length
 
 
+def get_initial_vram_info():
+    if shared.model_name != 'None' and shared.args.loader == 'llama.cpp':
+        return update_gpu_layers_and_vram(
+            shared.args.loader,
+            shared.model_name,
+            shared.args.gpu_layers,
+            shared.args.ctx_size,
+            shared.args.cache_type,
+            auto_adjust=False,
+            for_ui=True
+        )
+
+    return "
            Estimated VRAM to load the model:"
+
+
 def handle_load_model_event_initial(model, state):
     state = apply_model_settings_to_state(model, state)
     output = ui.apply_interface_values(state)
-    update_model_parameters(state)
+    update_model_parameters(state)  # This updates the command-line flags
+
     return output + [state]
 
 
diff --git a/server.py b/server.py
index c35d65a888..c22ed1f1b8 100644
--- a/server.py
+++ b/server.py
@@ -49,10 +49,9 @@
 from modules.LoRA import add_lora_to_model
 from modules.models import load_model, unload_model_if_idle
 from modules.models_settings import (
-    estimate_vram,
     get_fallback_settings,
     get_model_metadata,
-    get_nvidia_free_vram,
+    update_gpu_layers_and_vram,
     update_model_parameters
 )
 from modules.shared import do_cmd_flags_warnings
@@ -250,15 +249,19 @@ def create_interface():
         model_settings = get_model_metadata(model_name)
         update_model_parameters(model_settings, initial=True)  # hijack the command-line arguments
 
-        if 'gpu_layers' not in shared.provided_arguments:
-            available_vram = get_nvidia_free_vram()
-            if available_vram > 0:
-                n_layers = model_settings['gpu_layers']
-                tolerance = 906
-                while n_layers > 0 and estimate_vram(model_name, n_layers, shared.args.ctx_size, shared.args.cache_type) > available_vram - tolerance:
-                    n_layers -= 1
-
-            shared.args.gpu_layers = n_layers
+        # Auto-adjust GPU layers if not provided by user and it's a llama.cpp model
+        if 'gpu_layers' not in shared.provided_arguments and shared.args.loader == 'llama.cpp' and 'gpu_layers' in model_settings:
+            vram_usage, adjusted_layers = update_gpu_layers_and_vram(
+                shared.args.loader,
+                model_name,
+                model_settings['gpu_layers'],
+                shared.args.ctx_size,
+                shared.args.cache_type,
+                auto_adjust=True,
+                for_ui=False
+            )
+
+            shared.args.gpu_layers = adjusted_layers
 
         # Load the model
         shared.model, shared.tokenizer = load_model(model_name)

From ee7b3028acaa38399272e78bb05272a420e72f05 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 09:12:36 -0700
Subject: [PATCH 0509/1701] Always cache GGUF metadata calls

---
 modules/models_settings.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 6ea6660c45..8ecd226748 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -58,7 +58,7 @@ def get_model_metadata(model):
         else:
             model_file = list(path.glob('*.gguf'))[0]
 
-        metadata = metadata_gguf.load_metadata(model_file)
+        metadata = load_gguf_metadata_with_cache(model_file)
 
         for k in metadata:
             if k.endswith('context_length'):
@@ -295,8 +295,8 @@ def save_instruction_template(model, template):
         yield (f"Instruction template for `{model}` saved to `{p}` as `{template}`.")
 
 
-@functools.lru_cache(maxsize=None)
-def get_gguf_metadata_cached(model_file):
+@functools.lru_cache(maxsize=1)
+def load_gguf_metadata_with_cache(model_file):
     return metadata_gguf.load_metadata(model_file)
 
 
@@ -320,7 +320,7 @@ def get_model_size_mb(model_file: Path) -> float:
 
 def estimate_vram(gguf_file, gpu_layers, ctx_size, cache_type):
     model_file = Path(f'{shared.args.model_dir}/{gguf_file}')
-    metadata = get_gguf_metadata_cached(model_file)
+    metadata = load_gguf_metadata_with_cache(model_file)
     size_in_mb = get_model_size_mb(model_file)
 
     # Extract values from metadata

From 9ec9b1bf837a995af8f203c8d05897510ab77c3d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 09:56:23 -0700
Subject: [PATCH 0510/1701] Auto-adjust GPU layers after model unload to
 utilize freed VRAM

---
 modules/ui_model_menu.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index b6febb50f6..39c57bf37d 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -145,7 +145,9 @@ def create_event_handlers():
         partial(load_model_wrapper, autoload=True), gradio('model_menu', 'loader'), gradio('model_status'), show_progress=True).success(
         handle_load_model_event_final, gradio('truncation_length', 'loader', 'interface_state'), gradio('truncation_length', 'filter_by_loader'), show_progress=False)
 
-    shared.gradio['unload_model'].click(handle_unload_model_click, None, gradio('model_status'), show_progress=False)
+    shared.gradio['unload_model'].click(handle_unload_model_click, None, gradio('model_status'), show_progress=False).then(
+        partial(update_gpu_layers_and_vram, auto_adjust=True), gradio('loader', 'model_menu', 'gpu_layers', 'ctx_size', 'cache_type'), gradio('vram_info', 'gpu_layers'), show_progress=False)
+
     shared.gradio['save_model_settings'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         save_model_settings, gradio('model_menu', 'interface_state'), gradio('model_status'), show_progress=False)

From 253e85a519219385668aabeabc82633c8e734ff9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 10:02:30 -0700
Subject: [PATCH 0511/1701] Only compute VRAM/GPU layers for llama.cpp models

---
 modules/models_settings.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 8ecd226748..0eb179d7da 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -441,7 +441,7 @@ def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type,
         - If for_ui=True: (vram_info_update, gpu_layers_update) or just vram_info_update
         - If for_ui=False: (vram_usage, adjusted_layers) or just vram_usage
     """
-    if loader != 'llama.cpp' or model in ["None", None]:
+    if loader != 'llama.cpp' or model in ["None", None] or not model.endswith(".gguf"):
         vram_info = "
            Estimated VRAM to load the model:"
         if for_ui:
             return (vram_info, gr.update()) if auto_adjust else vram_info

From 38c50087feb11af41fed7a944ab0d7ef45a3bc44 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 11:55:06 -0700
Subject: [PATCH 0512/1701] Prevent a crash on systems without an NVIDIA GPU

---
 modules/models_settings.py | 2 --
 1 file changed, 2 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 0eb179d7da..3fdf3c84c9 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -421,11 +421,9 @@ def get_nvidia_free_vram():
         return total_free_vram_mib
 
     except FileNotFoundError:
-        raise
         # nvidia-smi not found (likely no NVIDIA drivers installed)
         return -1
     except Exception:
-        raise
         # Handle any other unexpected exceptions
         return -1
 

From fc483650b5e8c4933ac20b647cb822cf45856596 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 11:58:17 -0700
Subject: [PATCH 0513/1701] Set the maximum gpu_layers value automatically when
 the model is loaded with --model

---
 modules/ui_model_menu.py | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 39c57bf37d..cd101c4ab4 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -38,7 +38,7 @@ def create_ui():
                     gr.Markdown("## Main options")
                     with gr.Row():
                         with gr.Column():
-                            shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=256, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
+                            shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=get_initial_gpu_layers_max(), value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
@@ -315,6 +315,14 @@ def get_initial_vram_info():
     return "
            Estimated VRAM to load the model:"
 
 
+def get_initial_gpu_layers_max():
+    if shared.model_name != 'None' and shared.args.loader == 'llama.cpp':
+        model_settings = get_model_metadata(shared.model_name)
+        return model_settings.get('gpu_layers', 256)
+
+    return 256
+
+
 def handle_load_model_event_initial(model, state):
     state = apply_model_settings_to_state(model, state)
     output = ui.apply_interface_values(state)

From adb975a380b219bbe14bbd7a19c83eaebc15cd55 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 12:52:43 -0700
Subject: [PATCH 0514/1701] Prevent fractional gpu-layers in the UI

---
 modules/ui_model_menu.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index cd101c4ab4..59bb675932 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -38,7 +38,7 @@ def create_ui():
                     gr.Markdown("## Main options")
                     with gr.Row():
                         with gr.Column():
-                            shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=get_initial_gpu_layers_max(), value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
+                            shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')

From 470c822f44dce2269dfaa8e3b37989195982b975 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 12:54:39 -0700
Subject: [PATCH 0515/1701] API: Hide the uvicorn access logs from the terminal

---
 extensions/openai/script.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 2b4f274f6d..2c98ee78cf 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -447,7 +447,7 @@ def run_server():
 
     # Start server
     logging.getLogger("uvicorn.error").propagate = False
-    uvicorn.run(app, host=server_addrs, port=port, ssl_certfile=ssl_certfile, ssl_keyfile=ssl_keyfile)
+    uvicorn.run(app, host=server_addrs, port=port, ssl_certfile=ssl_certfile, ssl_keyfile=ssl_keyfile, access_log=False)
 
 
 def setup():

From e4d3f4449d75ea1b1f7f3438dbed8c910a970cec Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 13:02:27 -0700
Subject: [PATCH 0516/1701] API: Fix a regression

---
 modules/llama_cpp_server.py | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 3fc7a0cc0a..d695c74e9c 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -146,8 +146,9 @@ def generate_with_streaming(self, prompt, state):
             pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
             print()
 
-        # Make a direct request with streaming enabled using a context manager
-        with self.session.post(url, json=payload, stream=True) as response:
+        # Make the generation request
+        response = self.session.post(url, json=payload, stream=True)
+        try:
             response.raise_for_status()  # Raise an exception for HTTP errors
 
             full_text = ""
@@ -184,6 +185,8 @@ def generate_with_streaming(self, prompt, state):
                     print(f"JSON decode error: {e}")
                     print(f"Problematic line: {line}")
                     continue
+        finally:
+            response.close()
 
     def generate(self, prompt, state):
         output = ""

From 1c549d176b27233daf0ef6992bf5b5d8215784f9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 17:24:06 -0700
Subject: [PATCH 0517/1701] Fix GPU layers slider: honor saved settings and
 show true maximum

---
 modules/models_settings.py | 32 ++++++++++++++++++++++----------
 modules/ui_model_menu.py   |  2 +-
 2 files changed, 23 insertions(+), 11 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 3fdf3c84c9..6715d49492 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -72,6 +72,7 @@ def get_model_metadata(model):
                 model_settings['compress_pos_emb'] = metadata[k]
             elif k.endswith('block_count'):
                 model_settings['gpu_layers'] = metadata[k] + 1
+                model_settings['max_gpu_layers'] = metadata[k] + 1
 
         if 'tokenizer.chat_template' in metadata:
             template = metadata['tokenizer.chat_template']
@@ -450,17 +451,28 @@ def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type,
     max_layers = gpu_layers
 
     if auto_adjust:
-        # Get max layers from model metadata
+        # Get model settings including user preferences
         model_settings = get_model_metadata(model)
-        max_layers = model_settings.get('gpu_layers', gpu_layers)
-
-        # Auto-adjust based on available VRAM
-        available_vram = get_nvidia_free_vram()
-        if available_vram > 0:
-            tolerance = 906
-            current_layers = max_layers
-            while current_layers > 0 and estimate_vram(model, current_layers, ctx_size, cache_type) > available_vram - tolerance:
-                current_layers -= 1
+
+        # Check if the value is from user config-user.yaml
+        user_config = shared.user_config
+        model_regex = Path(model).name + '$'
+        has_user_setting = model_regex in user_config and 'gpu_layers' in user_config[model_regex]
+
+        if has_user_setting:
+            # Just return the current user value without adjustment
+            max_layers = model_settings.get('max_gpu_layers', 256)
+        else:
+            # No user setting, use model's max and auto-adjust
+            max_layers = model_settings.get('max_gpu_layers', model_settings.get('gpu_layers', gpu_layers))
+            current_layers = max_layers  # Start from max
+
+            # Auto-adjust based on available VRAM
+            available_vram = get_nvidia_free_vram()
+            if available_vram > 0:
+                tolerance = 906
+                while current_layers > 0 and estimate_vram(model, current_layers, ctx_size, cache_type) > available_vram - tolerance:
+                    current_layers -= 1
 
     # Calculate VRAM with current layers
     vram_usage = estimate_vram(model, current_layers, ctx_size, cache_type)
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 59bb675932..5b7dfdd876 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -318,7 +318,7 @@ def get_initial_vram_info():
 def get_initial_gpu_layers_max():
     if shared.model_name != 'None' and shared.args.loader == 'llama.cpp':
         model_settings = get_model_metadata(shared.model_name)
-        return model_settings.get('gpu_layers', 256)
+        return model_settings.get('max_gpu_layers', model_settings.get('gpu_layers', 256))
 
     return 256
 

From d99fb0a22a44dc4fb4d695647ba07cbf55e044c6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 17:29:18 -0700
Subject: [PATCH 0518/1701] Add backward compatibility with saved n_gpu_layers
 values

---
 modules/models_settings.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 6715d49492..76bce7a989 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -154,6 +154,9 @@ def get_model_metadata(model):
     for pat in settings:
         if re.match(pat.lower(), Path(model).name.lower()):
             for k in settings[pat]:
+                if k == 'n_gpu_layers':
+                    k = 'gpu_layers'
+
                 model_settings[k] = settings[pat][k]
 
     # Load instruction template if defined by name rather than by value

From 71fa046c1708a235853c359ef95b363a20c762d3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 17:38:08 -0700
Subject: [PATCH 0519/1701] Minor changes after
 1c549d176b27233daf0ef6992bf5b5d8215784f9

---
 modules/models_settings.py | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 76bce7a989..3a2400d4d5 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -457,17 +457,20 @@ def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type,
         # Get model settings including user preferences
         model_settings = get_model_metadata(model)
 
-        # Check if the value is from user config-user.yaml
+        # Get the true maximum layers
+        max_layers = model_settings.get('max_gpu_layers', model_settings.get('gpu_layers', gpu_layers))
+
+        # Check if this is a user-saved setting
         user_config = shared.user_config
         model_regex = Path(model).name + '$'
         has_user_setting = model_regex in user_config and 'gpu_layers' in user_config[model_regex]
 
         if has_user_setting:
-            # Just return the current user value without adjustment
-            max_layers = model_settings.get('max_gpu_layers', 256)
+            # For user settings, just use the current value (which already has user pref)
+            # but ensure the slider maximum is correct
+            current_layers = gpu_layers  # Already has user setting
         else:
-            # No user setting, use model's max and auto-adjust
-            max_layers = model_settings.get('max_gpu_layers', model_settings.get('gpu_layers', gpu_layers))
+            # No user setting, auto-adjust from the maximum
             current_layers = max_layers  # Start from max
 
             # Auto-adjust based on available VRAM

From e3bba510d443a0a447f85083a2dff4a116a50848 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 17:48:54 -0700
Subject: [PATCH 0520/1701] UI: Only add a blank space to streaming messages in
 instruct mode

---
 css/main.css              | 2 +-
 js/main.js                | 2 +-
 modules/html_generator.py | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/css/main.css b/css/main.css
index 0902b18466..3fec7bb065 100644
--- a/css/main.css
+++ b/css/main.css
@@ -390,7 +390,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     margin-left: auto;
     margin-right: auto;
     flex: 1;
-    overflow-y: auto;
+    overflow-y: hidden;
     display: flex;
     flex-direction: column;
     word-break: break-word;
diff --git a/js/main.js b/js/main.js
index 205cf88e8d..6cecd341d0 100644
--- a/js/main.js
+++ b/js/main.js
@@ -152,7 +152,7 @@ const observer = new MutationObserver(function(mutations) {
   }
 
   const chatElement = document.getElementById("chat");
-  if (chatElement) {
+  if (chatElement && chatElement.getAttribute("data-mode") === "instruct") {
     const messagesContainer = chatElement.querySelector(".messages");
     const lastChild = messagesContainer?.lastElementChild;
     const prevSibling = lastChild?.previousElementSibling;
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 67d15b6e98..39659476e1 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -347,7 +347,7 @@ def get_image_cache(path):
 
 
 def generate_instruct_html(history):
-    output = f'
            '
+    output = f'
            '
 
     for i in range(len(history['visible'])):
         row_visible = history['visible'][i]

From c0e295dd1d8f06ea889e8a4f8ec68ca755c3727b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 17:53:20 -0700
Subject: [PATCH 0521/1701] Remove the 'None' option from the model menu

---
 modules/utils.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/utils.py b/modules/utils.py
index 0e390d0899..0e8bdd1867 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -137,7 +137,7 @@ def get_available_models():
 
     model_dirs = sorted(model_dirs, key=natural_keys)
 
-    return ['None'] + filtered_gguf_files + model_dirs
+    return filtered_gguf_files + model_dirs
 
 
 def get_available_ggufs():

From 4bf763e1d94e91c1d016b597df642f076c06f849 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 18:22:43 -0700
Subject: [PATCH 0522/1701] Multiple small CSS fixes

---
 css/chat_style-Dark.css            | 4 ++--
 css/chat_style-TheEncrypted777.css | 4 ++--
 css/chat_style-cai-chat-square.css | 4 ++--
 css/chat_style-cai-chat.css        | 4 ++--
 css/chat_style-messenger.css       | 4 ++--
 css/chat_style-wpp.css             | 4 ++--
 6 files changed, 12 insertions(+), 12 deletions(-)

diff --git a/css/chat_style-Dark.css b/css/chat_style-Dark.css
index 1ad46bc0dc..6a4784ccd1 100644
--- a/css/chat_style-Dark.css
+++ b/css/chat_style-Dark.css
@@ -2,8 +2,8 @@
     display: grid;
     align-items: start;
     grid-template-columns: 60px minmax(0, 1fr);
-    padding-bottom: 14px;
-    padding-top: 14px;
+    padding-bottom: 22px;
+    padding-top: 6px;
     font-size: 18px;
     font-family: Roboto, Arial, sans-serif; /* Modern font */
     line-height: 1.5;
diff --git a/css/chat_style-TheEncrypted777.css b/css/chat_style-TheEncrypted777.css
index 9e1230b7d2..fbd47072fb 100644
--- a/css/chat_style-TheEncrypted777.css
+++ b/css/chat_style-TheEncrypted777.css
@@ -4,8 +4,8 @@
     display: grid;
     align-items: start;
     grid-template-columns: 60px minmax(0, 1fr);
-    padding-bottom: 14px;
-    padding-top: 14px;
+    padding-bottom: 21px;
+    padding-top: 7px;
     font-size: 18px;
     font-family: 'Noto Sans', Arial, sans-serif;
     line-height: 1.428571429;
diff --git a/css/chat_style-cai-chat-square.css b/css/chat_style-cai-chat-square.css
index 015f6927c6..291a120965 100644
--- a/css/chat_style-cai-chat-square.css
+++ b/css/chat_style-cai-chat-square.css
@@ -16,7 +16,7 @@
 }
 
 .message {
-    padding-bottom: 1em;
-    padding-top: 1em;
+    padding-bottom: 1.5em;
+    padding-top: 0.5em;
     grid-template-columns: 70px minmax(0, 1fr);
 }
diff --git a/css/chat_style-cai-chat.css b/css/chat_style-cai-chat.css
index 0e91101f00..b06b1269c7 100644
--- a/css/chat_style-cai-chat.css
+++ b/css/chat_style-cai-chat.css
@@ -2,8 +2,8 @@
     display: grid;
     align-items: start;
     grid-template-columns: 60px minmax(0, 1fr);
-    padding-bottom: 1em;
-    padding-top: 1em;
+    padding-bottom: 1.5em;
+    padding-top: 0.5em;
     font-size: 15px;
     font-family: 'Noto Sans', Helvetica, Arial, sans-serif;
     line-height: 22.5px !important;
diff --git a/css/chat_style-messenger.css b/css/chat_style-messenger.css
index 6518d6ca71..65af5f7a5f 100644
--- a/css/chat_style-messenger.css
+++ b/css/chat_style-messenger.css
@@ -1,6 +1,6 @@
 .message {
-    padding-bottom: 12.5px;
-    padding-top: 12.5px;
+    padding-bottom: 22px;
+    padding-top: 3px;
     font-size: 15px;
     font-family: 'Noto Sans', Helvetica, Arial, sans-serif;
     line-height: 1.428571429;
diff --git a/css/chat_style-wpp.css b/css/chat_style-wpp.css
index 1442dd0a43..353201c228 100644
--- a/css/chat_style-wpp.css
+++ b/css/chat_style-wpp.css
@@ -1,6 +1,6 @@
 .message {
-    padding-bottom: 12.5px;
-    padding-top: 12.5px;
+    padding-bottom: 22px;
+    padding-top: 3px;
     font-size: 15px;
     font-family: 'Noto Sans', Helvetica, Arial, sans-serif;
     line-height: 1.428571429;

From 0f77ff9670364d605ad40bd1addd153213ce65f0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 19:19:22 -0700
Subject: [PATCH 0523/1701] UI: Use total VRAM (not free) for layers
 calculation when a model is loaded

---
 modules/models_settings.py | 21 +++++++++++++++------
 1 file changed, 15 insertions(+), 6 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 3a2400d4d5..6b9493cae1 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -379,12 +379,15 @@ def smaller(x, y):
     return vram
 
 
-def get_nvidia_free_vram():
+def get_nvidia_vram(return_free=True):
     """
-    Calculates the total free VRAM across all NVIDIA GPUs by parsing nvidia-smi output.
+    Calculates VRAM statistics across all NVIDIA GPUs by parsing nvidia-smi output.
+
+    Args:
+        return_free (bool): If True, returns free VRAM. If False, returns total VRAM.
 
     Returns:
-        int: The total free VRAM in MiB summed across all detected NVIDIA GPUs.
+        int: Either the total free VRAM or total VRAM in MiB summed across all detected NVIDIA GPUs.
              Returns -1 if nvidia-smi command fails (not found, error, etc.).
              Returns 0 if nvidia-smi succeeds but no GPU memory info found.
     """
@@ -412,17 +415,21 @@ def get_nvidia_free_vram():
             # No GPUs found in expected format
             return 0
 
+        total_vram_mib = 0
         total_free_vram_mib = 0
+
         for used_mem_str, total_mem_str in matches:
             try:
                 used_mib = int(used_mem_str)
                 total_mib = int(total_mem_str)
+                total_vram_mib += total_mib
                 total_free_vram_mib += (total_mib - used_mib)
             except ValueError:
                 # Skip malformed entries
                 pass
 
-        return total_free_vram_mib
+        # Return either free or total VRAM based on the flag
+        return total_free_vram_mib if return_free else total_vram_mib
 
     except FileNotFoundError:
         # nvidia-smi not found (likely no NVIDIA drivers installed)
@@ -473,8 +480,10 @@ def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type,
             # No user setting, auto-adjust from the maximum
             current_layers = max_layers  # Start from max
 
-            # Auto-adjust based on available VRAM
-            available_vram = get_nvidia_free_vram()
+            # Auto-adjust based on available/total VRAM
+            # If a model is loaded and it's for the UI, use the total VRAM to avoid confusion
+            return_free = False if (for_ui and shared.model_name not in [None, 'None']) else True
+            available_vram = get_nvidia_vram(return_free=return_free)
             if available_vram > 0:
                 tolerance = 906
                 while current_layers > 0 and estimate_vram(model, current_layers, ctx_size, cache_type) > available_vram - tolerance:

From 052c82b664faf109d2ded6c0250ad777259adb94 Mon Sep 17 00:00:00 2001
From: mamei16 
Date: Sat, 17 May 2025 16:19:13 +0200
Subject: [PATCH 0524/1701] Fix KeyError: 'gpu_layers' when loading existing
 model settings (#6991)

---
 modules/models_settings.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 6b9493cae1..10234b4b6b 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -154,10 +154,11 @@ def get_model_metadata(model):
     for pat in settings:
         if re.match(pat.lower(), Path(model).name.lower()):
             for k in settings[pat]:
+                new_k = k
                 if k == 'n_gpu_layers':
-                    k = 'gpu_layers'
+                    new_k = 'gpu_layers'
 
-                model_settings[k] = settings[pat][k]
+                model_settings[new_k] = settings[pat][k]
 
     # Load instruction template if defined by name rather than by value
     if model_settings['instruction_template'] != 'Custom (obtained from model metadata)':

From 4800d1d522f84efaa50f4222aefd6fcae7e19e0c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 16 May 2025 22:33:32 -0700
Subject: [PATCH 0525/1701] More robust VRAM calculation

---
 modules/models_settings.py | 26 +++++++-------------------
 1 file changed, 7 insertions(+), 19 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 6b9493cae1..81a7a00ebb 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -2,7 +2,7 @@
 import json
 import re
 import subprocess
-from math import exp
+from math import floor
 from pathlib import Path
 
 import gradio as gr
@@ -331,8 +331,6 @@ def estimate_vram(gguf_file, gpu_layers, ctx_size, cache_type):
     n_layers = None
     n_kv_heads = None
     embedding_dim = None
-    context_length = None
-    feed_forward_dim = None
 
     for key, value in metadata.items():
         if key.endswith('.block_count'):
@@ -341,10 +339,6 @@ def estimate_vram(gguf_file, gpu_layers, ctx_size, cache_type):
             n_kv_heads = value
         elif key.endswith('.embedding_length'):
             embedding_dim = value
-        elif key.endswith('.context_length'):
-            context_length = value
-        elif key.endswith('.feed_forward_length'):
-            feed_forward_dim = value
 
     if gpu_layers > n_layers:
         gpu_layers = n_layers
@@ -359,22 +353,16 @@ def estimate_vram(gguf_file, gpu_layers, ctx_size, cache_type):
 
     # Derived features
     size_per_layer = size_in_mb / max(n_layers, 1e-6)
-    context_per_layer = context_length / max(n_layers, 1e-6)
-    ffn_per_embedding = feed_forward_dim / max(embedding_dim, 1e-6)
     kv_cache_factor = n_kv_heads * cache_type * ctx_size
-
-    # Helper function for smaller
-    def smaller(x, y):
-        return 1 if x < y else 0
+    embedding_per_context = embedding_dim / ctx_size
 
     # Calculate VRAM using the model
     # Details: https://oobabooga.github.io/blog/posts/gguf-vram-formula/
     vram = (
-        (size_per_layer - 21.19195204848197)
-        * exp(0.0001047328491557063 * size_in_mb * smaller(ffn_per_embedding, 2.671096993407845))
-        + 0.0006621544775632052 * context_per_layer
-        + 3.34664386576376e-05 * kv_cache_factor
-    ) * (1.363306170123392 + gpu_layers) + 1255.163594536052
+        (size_per_layer - 17.99552795246051 + 3.148552680382576e-05 * kv_cache_factor)
+        * (gpu_layers + max(0.9690636483914102, cache_type - (floor(50.77817218646521 * embedding_per_context) + 9.987899908205632)))
+        + 1516.522943869404
+    )
 
     return vram
 
@@ -485,7 +473,7 @@ def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type,
             return_free = False if (for_ui and shared.model_name not in [None, 'None']) else True
             available_vram = get_nvidia_vram(return_free=return_free)
             if available_vram > 0:
-                tolerance = 906
+                tolerance = 577
                 while current_layers > 0 and estimate_vram(model, current_layers, ctx_size, cache_type) > available_vram - tolerance:
                     current_layers -= 1
 

From 366de4b5611a344c708d76aeebb3b2d8c42a55c7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 17 May 2025 17:11:38 -0700
Subject: [PATCH 0526/1701] UI: Fix the chat area height when "Show controls"
 is unchecked

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 3fec7bb065..1ad7012232 100644
--- a/css/main.css
+++ b/css/main.css
@@ -827,7 +827,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 #chat-col.bigchat {
-    padding-bottom: 80px !important;
+    padding-bottom: 15px !important;
 }
 
 .message-body ol, .message-body ul {

From 076aa67963cd080837679662d79cb73326efb2ba Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 17 May 2025 22:14:14 -0700
Subject: [PATCH 0527/1701] Fix API issues

---
 extensions/openai/script.py | 38 +++++++++++++++++++++----------------
 1 file changed, 22 insertions(+), 16 deletions(-)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 2c98ee78cf..b7394bc5c6 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -114,14 +114,17 @@ async def openai_completions(request: Request, request_data: CompletionRequest):
 
     if request_data.stream:
         async def generator():
-            async with streaming_semaphore:
-                response = OAIcompletions.stream_completions(to_dict(request_data), is_legacy=is_legacy)
-                async for resp in iterate_in_threadpool(response):
-                    disconnected = await request.is_disconnected()
-                    if disconnected:
-                        break
-
-                    yield {"data": json.dumps(resp)}
+            try:
+                async with streaming_semaphore:
+                    response = OAIcompletions.stream_completions(to_dict(request_data), is_legacy=is_legacy)
+                    async for resp in iterate_in_threadpool(response):
+                        disconnected = await request.is_disconnected()
+                        if disconnected:
+                            break
+
+                        yield {"data": json.dumps(resp)}
+            finally:
+                return
 
         return EventSourceResponse(generator())  # SSE streaming
 
@@ -142,14 +145,17 @@ async def openai_chat_completions(request: Request, request_data: ChatCompletion
 
     if request_data.stream:
         async def generator():
-            async with streaming_semaphore:
-                response = OAIcompletions.stream_chat_completions(to_dict(request_data), is_legacy=is_legacy)
-                async for resp in iterate_in_threadpool(response):
-                    disconnected = await request.is_disconnected()
-                    if disconnected:
-                        break
-
-                    yield {"data": json.dumps(resp)}
+            try:
+                async with streaming_semaphore:
+                    response = OAIcompletions.stream_chat_completions(to_dict(request_data), is_legacy=is_legacy)
+                    async for resp in iterate_in_threadpool(response):
+                        disconnected = await request.is_disconnected()
+                        if disconnected:
+                            break
+
+                        yield {"data": json.dumps(resp)}
+            finally:
+                return
 
         return EventSourceResponse(generator())  # SSE streaming
 

From bd13a8f255ce7b637ee9e00fa7077752a6e56ca4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 17 May 2025 22:31:55 -0700
Subject: [PATCH 0528/1701] UI: Light theme improvement

---
 css/main.css | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/css/main.css b/css/main.css
index 1ad7012232..dc0ba817cb 100644
--- a/css/main.css
+++ b/css/main.css
@@ -131,7 +131,7 @@ gradio-app > :first-child {
 }
 
 .header_bar {
-    box-shadow: 0 0 3px rgba(22 22 22 / 35%);
+    border-right: var(--input-border-width) solid var(--input-border-color);
     margin-bottom: 0;
     overflow-x: scroll;
     text-wrap: nowrap;
@@ -1171,11 +1171,11 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     background-color: var(--light-theme-gray);
 }
 
-#chat-controls {
+.dark #chat-controls {
     border-left: 1px solid #d9d9d0;
 }
 
-#past-chats-row {
+.dark #past-chats-row {
     border-right: 1px solid #d9d9d0;
 }
 

From f1ec6c8662c0c8a744c827aa8d99036983aca8cc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 18 May 2025 09:04:51 -0700
Subject: [PATCH 0529/1701] Minor label changes

---
 modules/ui_chat.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 0856cfab05..a0c37dadb2 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -79,8 +79,8 @@ def create_ui():
                 shared.gradio['Send dummy reply'] = gr.Button('Send dummy reply')
 
             with gr.Row():
-                shared.gradio['send-chat-to-default'] = gr.Button('Send to default')
-                shared.gradio['send-chat-to-notebook'] = gr.Button('Send to notebook')
+                shared.gradio['send-chat-to-default'] = gr.Button('Send to Default')
+                shared.gradio['send-chat-to-notebook'] = gr.Button('Send to Notebook')
 
         with gr.Row(elem_id='chat-controls', elem_classes=['pretty_scrollbar']):
             with gr.Column():

From 2faaf18f1f9f4d29933017add849f8579021618c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 18 May 2025 09:06:20 -0700
Subject: [PATCH 0530/1701] Add back the "Common values" to the ctx-size slider

---
 modules/ui_model_menu.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 5b7dfdd876..85cf418963 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -39,7 +39,7 @@ def create_ui():
                     with gr.Row():
                         with gr.Column():
                             shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
-                            shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. ⚠️ Lower this value if you can\'t load the model.')
+                            shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. Common values: 4096, 8192, 16384, 32768, 65536, 131072. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
                             shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)

From 9d7a36356d7de6b7557928d199ccc94aa9c8f99e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 18 May 2025 10:56:16 -0700
Subject: [PATCH 0531/1701] Remove unnecessary js that was causing scrolling
 issues

---
 js/main.js | 28 ----------------------------
 1 file changed, 28 deletions(-)

diff --git a/js/main.js b/js/main.js
index 6cecd341d0..01c346a74c 100644
--- a/js/main.js
+++ b/js/main.js
@@ -132,8 +132,6 @@ targetElement.addEventListener("scroll", function() {
 
 // Create a MutationObserver instance
 const observer = new MutationObserver(function(mutations) {
-  updateCssProperties();
-
   if (targetElement.classList.contains("_generating")) {
     typing.parentNode.classList.add("visible-dots");
     document.getElementById("stop").style.display = "flex";
@@ -446,32 +444,6 @@ const chatInput = document.querySelector("#chat-input textarea");
 // Variables to store current dimensions
 let currentChatInputHeight = chatInput.clientHeight;
 
-// Update chat layout based on chat and input dimensions
-function updateCssProperties() {
-  const chatInputHeight = chatInput.clientHeight;
-
-  // Check if the chat container is visible
-  if (chatContainer.clientHeight > 0) {
-    // Adjust scrollTop based on input height change
-    if (chatInputHeight !== currentChatInputHeight) {
-      const deltaHeight = chatInputHeight - currentChatInputHeight;
-      if (!isScrolled && deltaHeight < 0) {
-        chatContainer.scrollTop = chatContainer.scrollHeight;
-      } else {
-        chatContainer.scrollTop += deltaHeight;
-      }
-
-      currentChatInputHeight = chatInputHeight;
-    }
-  }
-}
-
-// Observe textarea size changes and call update function
-new ResizeObserver(updateCssProperties).observe(document.querySelector("#chat-input textarea"));
-
-// Handle changes in window size
-window.addEventListener("resize", updateCssProperties);
-
 //------------------------------------------------
 // Focus on the rename text area when it becomes visible
 //------------------------------------------------

From 126b3a768fa9af7f5318dbfd70b7e6ad00defc68 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 18 May 2025 12:38:36 -0700
Subject: [PATCH 0532/1701] Revert "Dynamic Chat Message UI Update Speed
 (#6952)" (for now)

This reverts commit 8137eb8ef46ac6950cb96094e3cc30b0a72dee76.
---
 modules/shared.py                |  1 +
 modules/text_generation.py       | 18 ++++++++++--------
 modules/ui.py                    |  1 +
 modules/ui_parameters.py         |  2 ++
 user_data/settings-template.yaml |  1 +
 5 files changed, 15 insertions(+), 8 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 4e0a20db94..a6c0cbe9f7 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -47,6 +47,7 @@
     'max_new_tokens_max': 4096,
     'prompt_lookup_num_tokens': 0,
     'max_tokens_second': 0,
+    'max_updates_second': 12,
     'auto_max_new_tokens': True,
     'ban_eos_token': False,
     'add_bos_token': True,
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 00b9275a0d..962311df3c 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -65,39 +65,41 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
             all_stop_strings += st
 
     shared.stop_everything = False
+    last_update = -1
     reply = ''
     is_stream = state['stream']
     if len(all_stop_strings) > 0 and not state['stream']:
         state = copy.deepcopy(state)
         state['stream'] = True
 
+    min_update_interval = 0
+    if state.get('max_updates_second', 0) > 0:
+        min_update_interval = 1 / state['max_updates_second']
+
     # Generate
-    last_update = -1
-    latency_threshold = 1 / 1000
     for reply in generate_func(question, original_question, state, stopping_strings, is_chat=is_chat):
-        cur_time = time.monotonic()
         reply, stop_found = apply_stopping_strings(reply, all_stop_strings)
         if escape_html:
             reply = html.escape(reply)
 
         if is_stream:
+            cur_time = time.time()
+
             # Limit number of tokens/second to make text readable in real time
             if state['max_tokens_second'] > 0:
                 diff = 1 / state['max_tokens_second'] - (cur_time - last_update)
                 if diff > 0:
                     time.sleep(diff)
 
-                last_update = time.monotonic()
+                last_update = time.time()
                 yield reply
 
             # Limit updates to avoid lag in the Gradio UI
             # API updates are not limited
             else:
-                # If 'generate_func' takes less than 0.001 seconds to yield the next token
-                # (equivalent to more than 1000 tok/s), assume that the UI is lagging behind and skip yielding
-                if (cur_time - last_update) > latency_threshold:
+                if cur_time - last_update > min_update_interval:
+                    last_update = cur_time
                     yield reply
-                last_update = time.monotonic()
 
         if stop_found or (state['max_tokens_second'] > 0 and shared.stop_everything):
             break
diff --git a/modules/ui.py b/modules/ui.py
index eeb6ce92cd..25f9361273 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -192,6 +192,7 @@ def list_interface_input_elements():
         'max_new_tokens',
         'prompt_lookup_num_tokens',
         'max_tokens_second',
+        'max_updates_second',
         'do_sample',
         'dynamic_temperature',
         'temperature_last',
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 84f9fbfc24..733d090101 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -71,6 +71,8 @@ def create_ui(default_preset):
                             shared.gradio['max_new_tokens'] = gr.Slider(minimum=shared.settings['max_new_tokens_min'], maximum=shared.settings['max_new_tokens_max'], value=shared.settings['max_new_tokens'], step=1, label='max_new_tokens', info='⚠️ Setting this too high can cause prompt truncation.')
                             shared.gradio['prompt_lookup_num_tokens'] = gr.Slider(value=shared.settings['prompt_lookup_num_tokens'], minimum=0, maximum=10, step=1, label='prompt_lookup_num_tokens', info='Activates Prompt Lookup Decoding.')
                             shared.gradio['max_tokens_second'] = gr.Slider(value=shared.settings['max_tokens_second'], minimum=0, maximum=20, step=1, label='Maximum tokens/second', info='To make text readable in real time.')
+                            shared.gradio['max_updates_second'] = gr.Slider(value=shared.settings['max_updates_second'], minimum=0, maximum=24, step=1, label='Maximum UI updates/second', info='Set this if you experience lag in the UI during streaming.')
+
                 with gr.Column():
                     with gr.Row():
                         with gr.Column():
diff --git a/user_data/settings-template.yaml b/user_data/settings-template.yaml
index db481e8499..ce0f77e142 100644
--- a/user_data/settings-template.yaml
+++ b/user_data/settings-template.yaml
@@ -18,6 +18,7 @@ max_new_tokens_min: 1
 max_new_tokens_max: 4096
 prompt_lookup_num_tokens: 0
 max_tokens_second: 0
+max_updates_second: 12
 auto_max_new_tokens: true
 ban_eos_token: false
 add_bos_token: true

From 83bfd5c64b44e9eada63963e8aff05a608a7e90c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 18 May 2025 12:45:01 -0700
Subject: [PATCH 0533/1701] Fix API issues

---
 extensions/openai/script.py | 18 ++++++++++--------
 1 file changed, 10 insertions(+), 8 deletions(-)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index b7394bc5c6..b6abae20e8 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -114,8 +114,8 @@ async def openai_completions(request: Request, request_data: CompletionRequest):
 
     if request_data.stream:
         async def generator():
-            try:
-                async with streaming_semaphore:
+            async with streaming_semaphore:
+                try:
                     response = OAIcompletions.stream_completions(to_dict(request_data), is_legacy=is_legacy)
                     async for resp in iterate_in_threadpool(response):
                         disconnected = await request.is_disconnected()
@@ -123,8 +123,9 @@ async def generator():
                             break
 
                         yield {"data": json.dumps(resp)}
-            finally:
-                return
+                finally:
+                    stop_everything_event()
+                    return
 
         return EventSourceResponse(generator())  # SSE streaming
 
@@ -145,8 +146,8 @@ async def openai_chat_completions(request: Request, request_data: ChatCompletion
 
     if request_data.stream:
         async def generator():
-            try:
-                async with streaming_semaphore:
+            async with streaming_semaphore:
+                try:
                     response = OAIcompletions.stream_chat_completions(to_dict(request_data), is_legacy=is_legacy)
                     async for resp in iterate_in_threadpool(response):
                         disconnected = await request.is_disconnected()
@@ -154,8 +155,9 @@ async def generator():
                             break
 
                         yield {"data": json.dumps(resp)}
-            finally:
-                return
+                finally:
+                    stop_everything_event()
+                    return
 
         return EventSourceResponse(generator())  # SSE streaming
 

From 9cd6ea6c0b4b8fb99cf73d8fc0d1064db64fc2e8 Mon Sep 17 00:00:00 2001
From: Tiago Silva 
Date: Sun, 18 May 2025 22:07:16 +0100
Subject: [PATCH 0534/1701] Fix Dockerfile in AMD and Intel (#6995)

---
 README.md                       |  4 ++--
 docker/amd/Dockerfile           |  2 +-
 docker/amd/docker-compose.yml   | 12 +-----------
 docker/intel/Dockerfile         |  2 +-
 docker/intel/docker-compose.yml | 10 +---------
 5 files changed, 6 insertions(+), 24 deletions(-)

diff --git a/README.md b/README.md
index 0833f9b0c9..041513ac1d 100644
--- a/README.md
+++ b/README.md
@@ -146,14 +146,14 @@ The `requirements*.txt` above contain various wheels precompiled through GitHub
 For NVIDIA GPU:
 ln -s docker/{nvidia/Dockerfile,nvidia/docker-compose.yml,.dockerignore} .
 For AMD GPU: 
-ln -s docker/{amd/Dockerfile,intel/docker-compose.yml,.dockerignore} .
+ln -s docker/{amd/Dockerfile,amd/docker-compose.yml,.dockerignore} .
 For Intel GPU:
 ln -s docker/{intel/Dockerfile,amd/docker-compose.yml,.dockerignore} .
 For CPU only
 ln -s docker/{cpu/Dockerfile,cpu/docker-compose.yml,.dockerignore} .
 cp docker/.env.example .env
 #Create logs/cache dir : 
-mkdir -p logs cache
+mkdir -p user_data/logs user_data/cache
 # Edit .env and set: 
 #   TORCH_CUDA_ARCH_LIST based on your GPU model
 #   APP_RUNTIME_GID      your host user's group id (run `id -g` in a terminal)
diff --git a/docker/amd/Dockerfile b/docker/amd/Dockerfile
index 66e5863c90..c23083f7d4 100644
--- a/docker/amd/Dockerfile
+++ b/docker/amd/Dockerfile
@@ -14,7 +14,7 @@ WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
 RUN GPU_CHOICE=B LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
-COPY CMD_FLAGS.txt /home/app/text-generation-webui/
+COPY /user_data/CMD_FLAGS.txt /home/app/text-generation-webui/user_data
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
 WORKDIR /home/app/text-generation-webui
 # set umask to ensure group read / write at runtime
diff --git a/docker/amd/docker-compose.yml b/docker/amd/docker-compose.yml
index 8866e9edb8..a727ca3e67 100644
--- a/docker/amd/docker-compose.yml
+++ b/docker/amd/docker-compose.yml
@@ -41,14 +41,4 @@ services:
     security_opt:
       - seccomp=unconfined
     volumes:
-      - ./cache:/home/app/text-generation-webui/cache
-      - ./characters:/home/app/text-generation-webui/characters
-      - ./extensions:/home/app/text-generation-webui/extensions
-      - ./loras:/home/app/text-generation-webui/loras
-      - ./logs:/home/app/text-generation-webui/logs
-      - ./models:/home/app/text-generation-webui/models
-      - ./presets:/home/app/text-generation-webui/presets
-      - ./prompts:/home/app/text-generation-webui/prompts
-      - ./softprompts:/home/app/text-generation-webui/softprompts
-      - ./training:/home/app/text-generation-webui/training
-      - ./cloudflared:/etc/cloudflared
+      - ./user_data:/home/app/text-generation-webui/user_data
diff --git a/docker/intel/Dockerfile b/docker/intel/Dockerfile
index cab62442d7..4a70980358 100644
--- a/docker/intel/Dockerfile
+++ b/docker/intel/Dockerfile
@@ -14,7 +14,7 @@ WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
 RUN GPU_CHOICE=D LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
-COPY CMD_FLAGS.txt /home/app/text-generation-webui/
+COPY /user_data/CMD_FLAGS.txt /home/app/text-generation-webui/user_data
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
 # set umask to ensure group read / write at runtime
 WORKDIR /home/app/text-generation-webui
diff --git a/docker/intel/docker-compose.yml b/docker/intel/docker-compose.yml
index 78e0669804..bb48dd227d 100644
--- a/docker/intel/docker-compose.yml
+++ b/docker/intel/docker-compose.yml
@@ -41,12 +41,4 @@ services:
     security_opt:
       - seccomp=unconfined
     volumes:
-      - ./characters:/home/app/text-generation-webui/characters
-      - ./extensions:/home/app/text-generation-webui/extensions
-      - ./loras:/home/app/text-generation-webui/loras
-      - ./models:/home/app/text-generation-webui/models
-      - ./presets:/home/app/text-generation-webui/presets
-      - ./prompts:/home/app/text-generation-webui/prompts
-      - ./softprompts:/home/app/text-generation-webui/softprompts
-      - ./training:/home/app/text-generation-webui/training
-      - ./cloudflared:/etc/cloudflared
+      - ./user_data:/home/app/text-generation-webui/user_data

From 0c1bc6d1d07559e6518786948e728e5899a3471e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 18 May 2025 14:08:54 -0700
Subject: [PATCH 0535/1701] Bump llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index af5f7d8a88..1dcf8c934f 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 4e01198981..4a1702e912 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index a3bd1350a5..0caca6311e 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 363365bf17..9a43979833 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 2843fed27a..16e7726430 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -29,8 +29,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index bd7c4a4f25..468f97fa47 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index b5aa1cf798..eb7872ed45 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 89947cbe5f..3ba42c0b7e 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 7995939874..6831c4615a 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index ca16e4c7fd..fbb77ec06c 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 18e1c506a4..71575b281f 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -15,6 +15,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 693f471200..d093ab1404 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 8635d11e40..064d8e6c9a 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index e844596e6b..342239e8c2 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 9b7435d161..4ef3e97bf0 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 513b7a159e..7b39feb122 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.12.0/llama_cpp_binaries-0.12.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 0c7237e4b7c3de52f1de279134c12dcd0a41dcc9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 18 May 2025 20:01:29 -0700
Subject: [PATCH 0536/1701] Update README

---
 README.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index 041513ac1d..ee5a04bf9c 100644
--- a/README.md
+++ b/README.md
@@ -14,8 +14,8 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 
 - Supports multiple text generation backends in one UI/API, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [ExLlamaV2](https://github.com/turboderp-org/exllamav2).
   - [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) is also supported via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile).
-  - Additional quantization libraries like [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [HQQ](https://github.com/mobiusml/hqq), and [AQLM](https://github.com/Vahe1994/AQLM) can be used with the Transformers loader if you install them manually.
-- Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for llama.cpp GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory that doesn't interfere with your system environment.
+  - Additional quantization libraries like [HQQ](https://github.com/mobiusml/hqq) and [AQLM](https://github.com/Vahe1994/AQLM) can be used with the Transformers loader if you install them manually.
+- Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory that doesn't interfere with your system environment.
 - UI that resembles the original ChatGPT style.
 - Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
 - Three chat modes: `instruct`, `chat-instruct`, and `chat`, with automatic prompt templates in `chat-instruct`.

From 9ec46b8c4485742140cdbe118354155b88b99019 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 19 May 2025 09:23:24 -0700
Subject: [PATCH 0537/1701] Remove the HQQ loader (HQQ models can be loaded
 through Transformers)

---
 README.md                  |  4 +---
 modules/loaders.py         |  6 ------
 modules/models.py          | 16 ----------------
 modules/models_settings.py |  2 --
 modules/shared.py          |  8 +-------
 modules/ui.py              |  1 -
 modules/ui_model_menu.py   |  2 --
 7 files changed, 2 insertions(+), 37 deletions(-)

diff --git a/README.md b/README.md
index ee5a04bf9c..7105ce23ba 100644
--- a/README.md
+++ b/README.md
@@ -12,9 +12,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 
 ## Features
 
-- Supports multiple text generation backends in one UI/API, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [ExLlamaV2](https://github.com/turboderp-org/exllamav2).
-  - [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) is also supported via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile).
-  - Additional quantization libraries like [HQQ](https://github.com/mobiusml/hqq) and [AQLM](https://github.com/Vahe1994/AQLM) can be used with the Transformers loader if you install them manually.
+- Supports multiple text generation backends in one UI/API, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).
 - Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory that doesn't interfere with your system environment.
 - UI that resembles the original ChatGPT style.
 - Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
diff --git a/modules/loaders.py b/modules/loaders.py
index 79a7a4a35e..6fbd2198c3 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -90,11 +90,6 @@
         'ctx_size_draft',
         'speculative_decoding_accordion',
     ],
-    'HQQ': [
-        'hqq_backend',
-        'trust_remote_code',
-        'no_use_fast',
-    ],
     'TensorRT-LLM': [
         'ctx_size',
         'cpp_runner',
@@ -158,7 +153,6 @@ def transformers_samplers():
 
 loaders_samplers = {
     'Transformers': transformers_samplers(),
-    'HQQ': transformers_samplers(),
     'ExLlamav3_HF': {
         'temperature',
         'dynatemp_low',
diff --git a/modules/models.py b/modules/models.py
index 9ecee803a1..4218d58c9f 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -21,7 +21,6 @@ def load_model(model_name, loader=None):
         'ExLlamav3_HF': ExLlamav3_HF_loader,
         'ExLlamav2_HF': ExLlamav2_HF_loader,
         'ExLlamav2': ExLlamav2_loader,
-        'HQQ': HQQ_loader,
         'TensorRT-LLM': TensorRT_LLM_loader,
     }
 
@@ -102,21 +101,6 @@ def ExLlamav2_loader(model_name):
     return model, tokenizer
 
 
-def HQQ_loader(model_name):
-    try:
-        from hqq.core.quantize import HQQBackend, HQQLinear
-        from hqq.models.hf.base import AutoHQQHFModel
-    except ModuleNotFoundError:
-        raise ModuleNotFoundError("Failed to import 'hqq'. Please install it manually following the instructions in the HQQ GitHub repository.")
-
-    logger.info(f"Loading HQQ model with backend: \"{shared.args.hqq_backend}\"")
-
-    model_dir = Path(f'{shared.args.model_dir}/{model_name}')
-    model = AutoHQQHFModel.from_quantized(str(model_dir))
-    HQQLinear.set_backend(getattr(HQQBackend, shared.args.hqq_backend))
-    return model
-
-
 def TensorRT_LLM_loader(model_name):
     try:
         from modules.tensorrt_llm import TensorRTLLMModel
diff --git a/modules/models_settings.py b/modules/models_settings.py
index 47dbc020ec..e742e0d87a 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -183,8 +183,6 @@ def infer_loader(model_name, model_settings, hf_quant_method=None):
         loader = 'ExLlamav3_HF'
     elif re.match(r'.*exl2', model_name.lower()):
         loader = 'ExLlamav2_HF'
-    elif re.match(r'.*-hqq', model_name.lower()):
-        return 'HQQ'
     else:
         loader = 'Transformers'
 
diff --git a/modules/shared.py b/modules/shared.py
index a6c0cbe9f7..d2305f305b 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -87,7 +87,7 @@
 
 # Model loader
 group = parser.add_argument_group('Model loader')
-group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF, ExLlamav2, HQQ, TensorRT-LLM.')
+group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF, ExLlamav2, TensorRT-LLM.')
 
 # Transformers/Accelerate
 group = parser.add_argument_group('Transformers/Accelerate')
@@ -152,10 +152,6 @@
 group.add_argument('--num_experts_per_token', type=int, default=2, metavar='N', help='Number of experts to use for generation. Applies to MoE models like Mixtral.')
 group.add_argument('--enable_tp', action='store_true', help='Enable Tensor Parallelism (TP) in ExLlamaV2.')
 
-# HQQ
-group = parser.add_argument_group('HQQ')
-group.add_argument('--hqq-backend', type=str, default='PYTORCH_COMPILE', help='Backend for the HQQ loader. Valid options: PYTORCH, PYTORCH_COMPILE, ATEN.')
-
 # TensorRT-LLM
 group = parser.add_argument_group('TensorRT-LLM')
 group.add_argument('--cpp-runner', action='store_true', help='Use the ModelRunnerCpp runner, which is faster than the default ModelRunner but doesn\'t support streaming yet.')
@@ -263,8 +259,6 @@ def fix_loader_name(name):
         return 'ExLlamav2_HF'
     elif name in ['exllamav3-hf', 'exllamav3_hf', 'exllama-v3-hf', 'exllama_v3_hf', 'exllama-v3_hf', 'exllama3-hf', 'exllama3_hf', 'exllama-3-hf', 'exllama_3_hf', 'exllama-3_hf']:
         return 'ExLlamav3_HF'
-    elif name in ['hqq']:
-        return 'HQQ'
     elif name in ['tensorrt', 'tensorrtllm', 'tensorrt_llm', 'tensorrt-llm', 'tensort', 'tensortllm']:
         return 'TensorRT-LLM'
 
diff --git a/modules/ui.py b/modules/ui.py
index 25f9361273..f5dc063273 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -109,7 +109,6 @@ def list_model_elements():
         'threads',
         'threads_batch',
         'batch_size',
-        'hqq_backend',
         'ctx_size',
         'cache_type',
         'tensor_split',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 85cf418963..d361f69256 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -42,8 +42,6 @@ def create_ui():
                             shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. Common values: 4096, 8192, 16384, 32768, 65536, 131072. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
-                            shared.gradio['hqq_backend'] = gr.Dropdown(label="hqq_backend", choices=["PYTORCH", "PYTORCH_COMPILE", "ATEN"], value=shared.args.hqq_backend)
-
                         with gr.Column():
                             shared.gradio['vram_info'] = gr.HTML(value=get_initial_vram_info())
                             shared.gradio['flash_attn'] = gr.Checkbox(label="flash-attn", value=shared.args.flash_attn, info='Use flash-attention.')

From 8e10f9894aee28088b176adc5bed91f80cc11b5c Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Tue, 20 May 2025 10:48:46 -0300
Subject: [PATCH 0538/1701] Add a metadata field to the chat history & add
 date/time to chat messages (#7003)

---
 css/main.css              |  33 +++++++++++
 modules/chat.py           | 114 ++++++++++++++++++++++++++++++++++++--
 modules/html_generator.py |  65 +++++++++++++++++++---
 modules/ui_chat.py        |   2 +-
 4 files changed, 200 insertions(+), 14 deletions(-)

diff --git a/css/main.css b/css/main.css
index dc0ba817cb..319c177831 100644
--- a/css/main.css
+++ b/css/main.css
@@ -419,6 +419,14 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     padding-right: 1rem;
 }
 
+.chat .message .timestamp {
+    font-size: 0.7em;
+    display: inline-block;
+    font-weight: normal;
+    opacity: 0.7;
+    margin-left: 5px;
+}
+
 .chat-parent.bigchat {
     flex: 1;
 }
@@ -1269,6 +1277,31 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     left: 75px;
 }
 
+.footer-button.footer-info-button {
+    bottom: -23px;
+}
+
+.user-message .footer-button.footer-info-button {
+    left: 25px;
+}
+
+.assistant-message:not(:last-child) .footer-button.footer-info-button {
+    left: 25px;
+}
+
+.assistant-message:last-child .footer-button.footer-info-button {
+    left: 100px;
+}
+
+.message:not(:last-child) .text-bot .footer-button.footer-info-button,
+.message .text-you .footer-button.footer-info-button {
+    left: 25px;
+}
+
+.message:last-child .text-bot .footer-button.footer-info-button {
+    left: 100px;
+}
+
 .message:hover .footer-button,
 .user-message:hover .footer-button,
 .assistant-message:hover .footer-button {
diff --git a/modules/chat.py b/modules/chat.py
index b83c4bfecf..cbcde21290 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -37,6 +37,30 @@ def strftime_now(format):
     return datetime.now().strftime(format)
 
 
+def get_current_timestamp():
+    """Returns the current time in 24-hour format"""
+    return datetime.now().strftime('%b %d, %Y %H:%M')
+
+
+def update_message_metadata(metadata_dict, role, index, **fields):
+    """
+    Updates or adds metadata fields for a specific message.
+
+    Args:
+        metadata_dict: The metadata dictionary
+        role: The role (user, assistant, etc)
+        index: The message index
+        **fields: Arbitrary metadata fields to update/add
+    """
+    key = f"{role}_{index}"
+    if key not in metadata_dict:
+        metadata_dict[key] = {}
+
+    # Update with provided fields
+    for field_name, field_value in fields.items():
+        metadata_dict[key][field_name] = field_value
+
+
 jinja_env = ImmutableSandboxedEnvironment(
     trim_blocks=True,
     lstrip_blocks=True,
@@ -347,6 +371,10 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
     output = apply_extensions('history', output)
     state = apply_extensions('state', state)
 
+    # Initialize metadata if not present
+    if 'metadata' not in output:
+        output['metadata'] = {}
+
     visible_text = None
     stopping_strings = get_stopping_strings(state)
     is_stream = state['stream']
@@ -359,40 +387,56 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
         text, visible_text = apply_extensions('chat_input', text, visible_text, state)
         text = apply_extensions('input', text, state, is_chat=True)
 
+        # Current row index
+        row_idx = len(output['internal'])
         output['internal'].append([text, ''])
         output['visible'].append([visible_text, ''])
+        # Add metadata with timestamp
+        update_message_metadata(output['metadata'], "user", row_idx, timestamp=get_current_timestamp())
 
         # *Is typing...*
         if loading_message:
             yield {
                 'visible': output['visible'][:-1] + [[output['visible'][-1][0], shared.processing_message]],
-                'internal': output['internal']
+                'internal': output['internal'],
+                'metadata': output['metadata']
             }
     else:
         text, visible_text = output['internal'][-1][0], output['visible'][-1][0]
         if regenerate:
+            row_idx = len(output['internal']) - 1
             if loading_message:
                 yield {
                     'visible': output['visible'][:-1] + [[visible_text, shared.processing_message]],
-                    'internal': output['internal'][:-1] + [[text, '']]
+                    'internal': output['internal'][:-1] + [[text, '']],
+                    'metadata': output['metadata']
                 }
         elif _continue:
             last_reply = [output['internal'][-1][1], output['visible'][-1][1]]
             if loading_message:
                 yield {
                     'visible': output['visible'][:-1] + [[visible_text, last_reply[1] + '...']],
-                    'internal': output['internal']
+                    'internal': output['internal'],
+                    'metadata': output['metadata']
                 }
 
     # Generate the prompt
     kwargs = {
         '_continue': _continue,
-        'history': output if _continue else {k: v[:-1] for k, v in output.items()}
+        'history': output if _continue else {
+            k: (v[:-1] if k in ['internal', 'visible'] else v)
+            for k, v in output.items()
+        }
     }
+
     prompt = apply_extensions('custom_generate_chat_prompt', text, state, **kwargs)
     if prompt is None:
         prompt = generate_chat_prompt(text, state, **kwargs)
 
+    # Add timestamp for assistant's response at the start of generation
+    row_idx = len(output['internal']) - 1
+    update_message_metadata(output['metadata'], "assistant", row_idx, timestamp=get_current_timestamp())
+
     # Generate
     reply = None
     for j, reply in enumerate(generate_reply(prompt, state, stopping_strings=stopping_strings, is_chat=True, for_ui=for_ui)):
@@ -495,9 +539,19 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
 
 
 def remove_last_message(history):
+    if 'metadata' not in history:
+        history['metadata'] = {}
+
     if len(history['visible']) > 0 and history['internal'][-1][0] != '<|BEGIN-VISIBLE-CHAT|>':
+        row_idx = len(history['internal']) - 1
         last = history['visible'].pop()
         history['internal'].pop()
+
+        # Remove metadata directly by known keys
+        if f"user_{row_idx}" in history['metadata']:
+            del history['metadata'][f"user_{row_idx}"]
+        if f"assistant_{row_idx}" in history['metadata']:
+            del history['metadata'][f"assistant_{row_idx}"]
     else:
         last = ['', '']
 
@@ -514,30 +568,54 @@ def send_last_reply_to_input(history):
 def replace_last_reply(text, state):
     history = state['history']
 
+    # Initialize metadata if not present
+    if 'metadata' not in history:
+        history['metadata'] = {}
+
     if len(text.strip()) == 0:
         return history
     elif len(history['visible']) > 0:
+        row_idx = len(history['internal']) - 1
         history['visible'][-1][1] = html.escape(text)
         history['internal'][-1][1] = apply_extensions('input', text, state, is_chat=True)
+        update_message_metadata(history['metadata'], "assistant", row_idx, timestamp=get_current_timestamp())
 
     return history
 
 
 def send_dummy_message(text, state):
     history = state['history']
+
+    # Initialize metadata if not present
+    if 'metadata' not in history:
+        history['metadata'] = {}
+
+    row_idx = len(history['internal'])
     history['visible'].append([html.escape(text), ''])
     history['internal'].append([apply_extensions('input', text, state, is_chat=True), ''])
+    update_message_metadata(history['metadata'], "user", row_idx, timestamp=get_current_timestamp())
+
     return history
 
 
 def send_dummy_reply(text, state):
     history = state['history']
+
+    # Initialize metadata if not present
+    if 'metadata' not in history:
+        history['metadata'] = {}
+
     if len(history['visible']) > 0 and not history['visible'][-1][1] == '':
+        row_idx = len(history['internal'])
         history['visible'].append(['', ''])
         history['internal'].append(['', ''])
+        # We don't need to add system metadata
 
+    row_idx = len(history['internal']) - 1
     history['visible'][-1][1] = html.escape(text)
     history['internal'][-1][1] = apply_extensions('input', text, state, is_chat=True)
+    update_message_metadata(history['metadata'], "assistant", row_idx, timestamp=get_current_timestamp())
+
     return history
 
 
@@ -547,7 +625,8 @@ def redraw_html(history, name1, name2, mode, style, character, reset_cache=False
 
 def start_new_chat(state):
     mode = state['mode']
-    history = {'internal': [], 'visible': []}
+    # Initialize with empty metadata dictionary
+    history = {'internal': [], 'visible': [], 'metadata': {}}
 
     if mode != 'instruct':
         greeting = replace_character_names(state['greeting'], state['name1'], state['name2'])
@@ -555,6 +634,9 @@ def start_new_chat(state):
             history['internal'] += [['<|BEGIN-VISIBLE-CHAT|>', greeting]]
             history['visible'] += [['', apply_extensions('output', html.escape(greeting), state, is_chat=True)]]
 
+            # Add timestamp for assistant's greeting
+            update_message_metadata(history['metadata'], "assistant", 0, timestamp=get_current_timestamp())
+
     unique_id = datetime.now().strftime('%Y%m%d-%H-%M-%S')
     save_history(history, unique_id, state['character_menu'], state['mode'])
 
@@ -735,6 +817,16 @@ def load_history(unique_id, character, mode):
             'visible': f['data_visible']
         }
 
+    # Add metadata if it doesn't exist
+    if 'metadata' not in history:
+        history['metadata'] = {}
+        # Add placeholder timestamps for existing messages
+        for i, (user_msg, asst_msg) in enumerate(history['internal']):
+            if user_msg and user_msg != '<|BEGIN-VISIBLE-CHAT|>':
+                update_message_metadata(history['metadata'], "user", i, timestamp="")
+            if asst_msg:
+                update_message_metadata(history['metadata'], "assistant", i, timestamp="")
+
     return history
 
 
@@ -750,6 +842,16 @@ def load_history_json(file, history):
                 'visible': f['data_visible']
             }
 
+        # Add metadata if it doesn't exist
+        if 'metadata' not in history:
+            history['metadata'] = {}
+            # Add placeholder timestamps
+            for i, (user_msg, asst_msg) in enumerate(history['internal']):
+                if user_msg and user_msg != '<|BEGIN-VISIBLE-CHAT|>':
+                    update_message_metadata(history['metadata'], "user", i, timestamp="")
+                if asst_msg:
+                    update_message_metadata(history['metadata'], "assistant", i, timestamp="")
+
         return history
     except:
         return history
@@ -1299,7 +1401,7 @@ def handle_your_picture_change(picture, state):
 
 def handle_send_instruction_click(state):
     state['mode'] = 'instruct'
-    state['history'] = {'internal': [], 'visible': []}
+    state['history'] = {'internal': [], 'visible': [], 'metadata': {}}
 
     output = generate_chat_prompt("Input", state)
 
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 39659476e1..5dbde6dae3 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -169,11 +169,7 @@ def convert_to_markdown(string, message_id=None):
         thinking_block = f'''
         
            
             
-                
-                    
-                    
-                    
-                
+                {info_svg_small}
                 {title_text}
             
             
            {thinking_html}
            
@@ -339,11 +335,24 @@ def get_image_cache(path):
 refresh_svg = ''''''
 continue_svg = ''''''
 remove_svg = ''''''
+info_svg = ''''''
+info_svg_small = ''''''
 
 copy_button = f''
 refresh_button = f''
 continue_button = f''
 remove_button = f''
+info_button = f''
+
+
+def format_message_timestamp(history, role, index):
+    """Get a formatted timestamp HTML span for a message if available"""
+    key = f"{role}_{index}"
+    if 'metadata' in history and key in history['metadata'] and history['metadata'][key].get('timestamp'):
+        timestamp = history['metadata'][key]['timestamp']
+        return f"{timestamp}"
+
+    return ""
 
 
 def generate_instruct_html(history):
@@ -354,6 +363,23 @@ def generate_instruct_html(history):
         row_internal = history['internal'][i]
         converted_visible = [convert_to_markdown_wrapped(entry, message_id=i, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
 
+        # Get timestamps
+        user_timestamp = format_message_timestamp(history, "user", i)
+        assistant_timestamp = format_message_timestamp(history, "assistant", i)
+
+        # Create info buttons for timestamps if they exist
+        info_message_user = ""
+        if user_timestamp != "":
+            # Extract the timestamp value from the span
+            user_timestamp_value = user_timestamp.split('>', 1)[1].split('<', 1)[0]
+            info_message_user = info_button.replace("message", user_timestamp_value)
+
+        info_message_assistant = ""
+        if assistant_timestamp != "":
+            # Extract the timestamp value from the span
+            assistant_timestamp_value = assistant_timestamp.split('>', 1)[1].split('<', 1)[0]
+            info_message_assistant = info_button.replace("message", assistant_timestamp_value)
+
         if converted_visible[0]:  # Don't display empty user messages
             output += (
                 f'
            '
                 f'
            {converted_visible[0]}
            '
                 f'{copy_button}'
+                f'{info_message_user}'
                 f'
            '
                 f'
            '
             )
@@ -374,6 +401,7 @@ def generate_instruct_html(history):
             f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
             f'{continue_button if i == len(history["visible"]) - 1 else ""}'
             f'{remove_button if i == len(history["visible"]) - 1 else ""}'
+            f'{info_message_assistant}'
             f'
            '
             f'
            '
         )
@@ -401,13 +429,17 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
         row_internal = history['internal'][i]
         converted_visible = [convert_to_markdown_wrapped(entry, message_id=i, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
 
+        # Get timestamps
+        user_timestamp = format_message_timestamp(history, "user", i)
+        assistant_timestamp = format_message_timestamp(history, "assistant", i)
+
         if converted_visible[0]:  # Don't display empty user messages
             output += (
                 f'
            '
                 f'
            {img_me}
            '
                 f'
            '
-                f'
            {name1}
            '
+                f'
            {name1}{user_timestamp}
            '
                 f'
            {converted_visible[0]}
            '
                 f'{copy_button}'
                 f'
            '
@@ -419,7 +451,7 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
             f'data-raw="{html.escape(row_internal[1], quote=True)}">'
             f'
            {img_bot}
            '
             f'
            '
-            f'
            {name2}
            '
+            f'
            {name2}{assistant_timestamp}
            '
             f'
            {converted_visible[1]}
            '
             f'{copy_button}'
             f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
@@ -441,6 +473,23 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
         row_internal = history['internal'][i]
         converted_visible = [convert_to_markdown_wrapped(entry, message_id=i, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
 
+        # Get timestamps
+        user_timestamp = format_message_timestamp(history, "user", i)
+        assistant_timestamp = format_message_timestamp(history, "assistant", i)
+
+        # Create info buttons for timestamps if they exist
+        info_message_user = ""
+        if user_timestamp != "":
+            # Extract the timestamp value from the span
+            user_timestamp_value = user_timestamp.split('>', 1)[1].split('<', 1)[0]
+            info_message_user = info_button.replace("message", user_timestamp_value)
+
+        info_message_assistant = ""
+        if assistant_timestamp != "":
+            # Extract the timestamp value from the span
+            assistant_timestamp_value = assistant_timestamp.split('>', 1)[1].split('<', 1)[0]
+            info_message_assistant = info_button.replace("message", assistant_timestamp_value)
+
         if converted_visible[0]:  # Don't display empty user messages
             output += (
                 f'
            '
                 f'
            {converted_visible[0]}
            '
                 f'{copy_button}'
+                f'{info_message_user}'
                 f'
            '
                 f'
            '
             )
@@ -461,6 +511,7 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
             f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
             f'{continue_button if i == len(history["visible"]) - 1 else ""}'
             f'{remove_button if i == len(history["visible"]) - 1 else ""}'
+            f'{info_message_assistant}'
             f'
            '
             f'
            '
         )
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index a0c37dadb2..7a5430ca70 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -47,7 +47,7 @@ def create_ui():
         with gr.Row():
             with gr.Column(elem_id='chat-col'):
                 shared.gradio['display'] = gr.JSON(value={}, visible=False)  # Hidden buffer
-                shared.gradio['html_display'] = gr.HTML(value=chat_html_wrapper({'internal': [], 'visible': []}, '', '', 'chat', 'cai-chat', '')['html'], visible=True)
+                shared.gradio['html_display'] = gr.HTML(value=chat_html_wrapper({'internal': [], 'visible': [], 'metadata': {}}, '', '', 'chat', 'cai-chat', '')['html'], visible=True)
                 with gr.Row(elem_id="chat-input-row"):
                     with gr.Column(scale=1, elem_id='gr-hover-container'):
                         gr.HTML(value='
            ☰
            ', elem_id='gr-hover')

From c25a381540eb8c40e945730b058ca3e83fe0674c Mon Sep 17 00:00:00 2001
From: Daniel Dengler 
Date: Tue, 20 May 2025 16:07:40 +0200
Subject: [PATCH 0539/1701] Add a "Branch here" footer button to chat messages
 (#6967)

---
 css/main.css              | 66 +++++++++------------------------------
 js/global_scope_js.js     | 31 ++++++++++++++++++
 modules/chat.py           | 10 ++++--
 modules/html_generator.py | 47 +++++++++++++++-------------
 modules/ui.py             |  1 +
 modules/ui_chat.py        |  5 +--
 6 files changed, 83 insertions(+), 77 deletions(-)

diff --git a/css/main.css b/css/main.css
index 319c177831..d7142336bd 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1244,67 +1244,31 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     position: relative;
 }
 
-.footer-button {
+/* New container for the buttons */
+.message-actions {
     position: absolute;
+    bottom: -23px;
+    left: 0;
+    display: flex;   
+    gap: 5px;
+    opacity: 0;
+    transition: opacity 0.2s;
+}
+
+.footer-button {
     padding: 0;
     margin: 0;
     border: none;
     border-radius: 3px;
     cursor: pointer;
-    opacity: 0;
     display: flex;
     align-items: center;
-    transition: opacity 0.2s;
-}
-
-.footer-button.footer-copy-button {
-    bottom: -23px;
-    left: 0;
-}
-
-.footer-button.footer-refresh-button {
-    bottom: -23px;
-    left: 25px;
-}
-
-.footer-button.footer-continue-button {
-    bottom: -23px;
-    left: 50px;
-}
-
-.footer-button.footer-remove-button {
-    bottom: -23px;
-    left: 75px;
-}
-
-.footer-button.footer-info-button {
-    bottom: -23px;
-}
-
-.user-message .footer-button.footer-info-button {
-    left: 25px;
-}
-
-.assistant-message:not(:last-child) .footer-button.footer-info-button {
-    left: 25px;
-}
-
-.assistant-message:last-child .footer-button.footer-info-button {
-    left: 100px;
-}
-
-.message:not(:last-child) .text-bot .footer-button.footer-info-button,
-.message .text-you .footer-button.footer-info-button {
-    left: 25px;
-}
-
-.message:last-child .text-bot .footer-button.footer-info-button {
-    left: 100px;
+    justify-content: center;
 }
 
-.message:hover .footer-button,
-.user-message:hover .footer-button,
-.assistant-message:hover .footer-button {
+.message:hover .message-actions,
+.user-message:hover .message-actions,
+.assistant-message:hover .message-actions {
     opacity: 1;
 }
 
diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 29d2d8bd27..285d82f91b 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -18,6 +18,37 @@ function copyToClipboard(element) {
   });
 }
 
+function branchHere(element) {
+  if (!element) return;
+
+  const messageElement = element.closest(".message, .user-message, .assistant-message");
+  if (!messageElement) return;
+
+  const index = messageElement.getAttribute("data-index");
+  if (!index) return;
+
+  const branchIndexInput = document.getElementById("Branch-index").querySelector("input");
+  if (!branchIndexInput) {
+    console.error("Element with ID 'Branch-index' not found.");
+    return;
+  }
+  const branchButton = document.getElementById("Branch");
+
+  if (!branchButton) {
+    console.error("Required element 'Branch' not found.");
+    return;
+  }
+
+  branchIndexInput.value = index;
+
+  // Trigger any 'change' or 'input' events Gradio might be listening for
+  const event = new Event("input", { bubbles: true }); // 'change' might also work
+  branchIndexInput.dispatchEvent(event);
+
+  branchButton.click(); // Gradio will now pick up the 'index'
+
+}
+
 function regenerateClick() {
   document.getElementById("Regenerate").click();
 }
diff --git a/modules/chat.py b/modules/chat.py
index cbcde21290..13f733e96a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1248,7 +1248,13 @@ def handle_delete_chat_confirm_click(state):
 
 
 def handle_branch_chat_click(state):
-    history = state['history']
+    branch_from_index = state['branch_index']
+    if branch_from_index == -1:
+        history = state['history']
+    else:
+        history = state['history']
+        history['visible'] = history['visible'][:branch_from_index + 1]
+        history['internal'] = history['internal'][:branch_from_index + 1]
     new_unique_id = datetime.now().strftime('%Y%m%d-%H-%M-%S')
     save_history(history, new_unique_id, state['character_menu'], state['mode'])
 
@@ -1259,7 +1265,7 @@ def handle_branch_chat_click(state):
 
     past_chats_update = gr.update(choices=histories, value=new_unique_id)
 
-    return [history, html, past_chats_update]
+    return [history, html, past_chats_update, -1]
 
 
 def handle_rename_chat_click():
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 5dbde6dae3..36b31ac599 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -335,10 +335,12 @@ def get_image_cache(path):
 refresh_svg = ''''''
 continue_svg = ''''''
 remove_svg = ''''''
+branch_svg = ''''''
 info_svg = ''''''
 info_svg_small = ''''''
 
 copy_button = f''
+branch_button = f''
 refresh_button = f''
 continue_button = f''
 remove_button = f''
@@ -355,6 +357,17 @@ def format_message_timestamp(history, role, index):
     return ""
 
 
+def actions_html(history, i, info_message=""):
+    return (f'
            '
+            f'{copy_button}'
+            f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
+            f'{continue_button if i == len(history["visible"]) - 1 else ""}'
+            f'{remove_button if i == len(history["visible"]) - 1 else ""}'
+            f'{branch_button}'
+            f'{info_message}'
+            f'
            ')
+
+
 def generate_instruct_html(history):
     output = f'
            '
 
@@ -386,22 +399,18 @@ def generate_instruct_html(history):
                 f'data-raw="{html.escape(row_internal[0], quote=True)}">'
                 f'
            '
                 f'
            {converted_visible[0]}
            '
-                f'{copy_button}'
-                f'{info_message_user}'
+                f'
            {copy_button}{info_message_user}
            '
                 f'
            '
                 f'
            '
             )
 
         output += (
             f'
            '
+            f'data-raw="{html.escape(row_internal[1], quote=True)}"'
+            f'data-index={i}>'
             f'
            '
             f'
            {converted_visible[1]}
            '
-            f'{copy_button}'
-            f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
-            f'{continue_button if i == len(history["visible"]) - 1 else ""}'
-            f'{remove_button if i == len(history["visible"]) - 1 else ""}'
-            f'{info_message_assistant}'
+            f'{actions_html(history, i, info_message_assistant)}'
             f'
            '
             f'
            '
         )
@@ -441,22 +450,20 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
                 f'
            '
                 f'
            {name1}{user_timestamp}
            '
                 f'
            {converted_visible[0]}
            '
-                f'{copy_button}'
+                f'
            {copy_button}
            '
                 f'
            '
                 f'
            '
             )
 
         output += (
             f'
            '
+            f'data-raw="{html.escape(row_internal[1], quote=True)}"'
+            f'data-index={i}>'
             f'
            {img_bot}
            '
             f'
            '
             f'
            {name2}{assistant_timestamp}
            '
             f'
            {converted_visible[1]}
            '
-            f'{copy_button}'
-            f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
-            f'{continue_button if i == len(history["visible"]) - 1 else ""}'
-            f'{remove_button if i == len(history["visible"]) - 1 else ""}'
+            f'{actions_html(history, i)}'
             f'
            '
             f'
            '
         )
@@ -496,22 +503,18 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
                 f'data-raw="{html.escape(row_internal[0], quote=True)}">'
                 f'
            '
                 f'
            {converted_visible[0]}
            '
-                f'{copy_button}'
-                f'{info_message_user}'
+                f'
            {copy_button}{info_message_user}
            '
                 f'
            '
                 f'
            '
             )
 
         output += (
             f'
            '
+            f'data-raw="{html.escape(row_internal[1], quote=True)}"'
+            f'data-index={i}>'
             f'
            '
             f'
            {converted_visible[1]}
            '
-            f'{copy_button}'
-            f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
-            f'{continue_button if i == len(history["visible"]) - 1 else ""}'
-            f'{remove_button if i == len(history["visible"]) - 1 else ""}'
-            f'{info_message_assistant}'
+            f'{actions_html(history, i, info_message_assistant)}'
             f'
            '
             f'
            '
         )
diff --git a/modules/ui.py b/modules/ui.py
index f5dc063273..5e8fa14e9b 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -210,6 +210,7 @@ def list_interface_input_elements():
         'negative_prompt',
         'dry_sequence_breakers',
         'grammar_string',
+        'branch_index'
     ]
 
     # Chat elements
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 7a5430ca70..513a632bb4 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -24,7 +24,8 @@ def create_ui():
         with gr.Row(elem_id='past-chats-row', elem_classes=['pretty_scrollbar']):
             with gr.Column():
                 with gr.Row(elem_id='past-chats-buttons'):
-                    shared.gradio['branch_chat'] = gr.Button('Branch', elem_classes='refresh-button', interactive=not mu)
+                    shared.gradio['branch_chat'] = gr.Button('Branch', elem_classes='refresh-button', elem_id='Branch', interactive=not mu)
+                    shared.gradio['branch_index'] = gr.Number(value=-1, precision=0, visible=False, elem_id="Branch-index", interactive=True)
                     shared.gradio['rename_chat'] = gr.Button('Rename', elem_classes='refresh-button', interactive=not mu)
                     shared.gradio['delete_chat'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu)
                     shared.gradio['Start new chat'] = gr.Button('New chat', elem_classes=['refresh-button', 'focus-on-chat-input'])
@@ -258,7 +259,7 @@ def create_event_handlers():
 
     shared.gradio['branch_chat'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        chat.handle_branch_chat_click, gradio('interface_state'), gradio('history', 'display', 'unique_id'), show_progress=False)
+        chat.handle_branch_chat_click, gradio('interface_state'), gradio('history', 'display', 'unique_id', 'branch_index'), show_progress=False)
 
     shared.gradio['rename_chat'].click(chat.handle_rename_chat_click, None, gradio('rename_to', 'rename-row'), show_progress=False)
     shared.gradio['rename_to-cancel'].click(lambda: gr.update(visible=False), None, gradio('rename-row'), show_progress=False)

From 616ea6966d4821357076ff0c3b0a37967b736dd1 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Tue, 20 May 2025 12:51:28 -0300
Subject: [PATCH 0540/1701] Store previous reply versions on regenerate (#7004)

---
 modules/chat.py | 37 +++++++++++++++++++++++++++++++++++++
 1 file changed, 37 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index 13f733e96a..3efc55db89 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -365,6 +365,34 @@ def get_stopping_strings(state):
     return result
 
 
+def add_message_version(history, row_idx, is_current=True):
+    """Add the current message as a version in the history metadata"""
+    if 'metadata' not in history:
+        history['metadata'] = {}
+
+    if row_idx >= len(history['internal']) or not history['internal'][row_idx][1].strip():
+        return  # Skip if row doesn't exist or message is empty
+
+    key = f"assistant_{row_idx}"
+
+    # Initialize metadata structures if needed
+    if key not in history['metadata']:
+        history['metadata'][key] = {"timestamp": get_current_timestamp()}
+    if "versions" not in history['metadata'][key]:
+        history['metadata'][key]["versions"] = []
+
+    # Add current message as a version
+    history['metadata'][key]["versions"].append({
+        "content": history['internal'][row_idx][1],
+        "visible_content": history['visible'][row_idx][1],
+        "timestamp": get_current_timestamp()
+    })
+
+    # Update index if this is the current version
+    if is_current:
+        history['metadata'][key]["current_version_index"] = len(history['metadata'][key]["versions"]) - 1
+
+
 def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_message=True, for_ui=False):
     history = state['history']
     output = copy.deepcopy(history)
@@ -405,6 +433,10 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
         text, visible_text = output['internal'][-1][0], output['visible'][-1][0]
         if regenerate:
             row_idx = len(output['internal']) - 1
+
+            # Store the existing response as a version before regenerating
+            add_message_version(output, row_idx, is_current=False)
+
             if loading_message:
                 yield {
                     'visible': output['visible'][:-1] + [[visible_text, shared.processing_message]],
@@ -465,6 +497,11 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             if is_stream:
                 yield output
 
+    # Add the newly generated response as a version (only for regeneration)
+    if regenerate:
+        row_idx = len(output['internal']) - 1
+        add_message_version(output, row_idx, is_current=True)
+
     output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
     yield output
 

From 51c50b265d50a46b345b1b1d4afa55b5c94d5063 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 20 May 2025 11:15:38 -0700
Subject: [PATCH 0541/1701] Update llama.cpp to
 https://github.com/ggml-org/llama.cpp/commit/b7a17463ec190aeee7b9077c606c910fb4688b84

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 1dcf8c934f..c65ab8a2d2 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 4a1702e912..3da16d3ea0 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 0caca6311e..271b4bd051 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 9a43979833..15df937cf8 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -29,7 +29,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 16e7726430..bd2f833971 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -29,8 +29,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 468f97fa47..98c2564946 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index eb7872ed45..6e13c1d223 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -29,5 +29,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 3ba42c0b7e..67a5cb735c 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -30,8 +30,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 6831c4615a..409252f6c0 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index fbb77ec06c..89adbabf52 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 71575b281f..0b1c03fa25 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -15,6 +15,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index d093ab1404..eb4319b7e2 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 064d8e6c9a..0a60d4de4b 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 342239e8c2..652e990047 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 4ef3e97bf0..c83d61c705 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 7b39feb122..e69f3bdf37 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -15,5 +15,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.13.0/llama_cpp_binaries-0.13.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 5d00574a566ac8c66af16f76c9cbda6696e46e00 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 20 May 2025 16:20:49 -0700
Subject: [PATCH 0542/1701] Minor UI fixes

---
 modules/models_settings.py | 4 ++--
 modules/ui_model_menu.py   | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index e742e0d87a..df5a8e8ddf 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -438,7 +438,7 @@ def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type,
         - If for_ui=False: (vram_usage, adjusted_layers) or just vram_usage
     """
     if loader != 'llama.cpp' or model in ["None", None] or not model.endswith(".gguf"):
-        vram_info = "
            Estimated VRAM to load the model:"
+        vram_info = "
            Estimated VRAM to load the model:
            "
         if for_ui:
             return (vram_info, gr.update()) if auto_adjust else vram_info
         else:
@@ -480,7 +480,7 @@ def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type,
     vram_usage = estimate_vram(model, current_layers, ctx_size, cache_type)
 
     if for_ui:
-        vram_info = f"
            Estimated VRAM to load the model: {vram_usage:.0f} MiB"
+        vram_info = f"
            Estimated VRAM to load the model: {vram_usage:.0f} MiB
            "
         if auto_adjust:
             return vram_info, gr.update(value=current_layers, maximum=max_layers)
         else:
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index d361f69256..862b389359 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -310,7 +310,7 @@ def get_initial_vram_info():
             for_ui=True
         )
 
-    return "
            Estimated VRAM to load the model:"
+    return "
            Estimated VRAM to load the model:
            "
 
 
 def get_initial_gpu_layers_max():

From 409a48d6bdd0f2bc861fc459cdd701d697bdd188 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Wed, 21 May 2025 00:36:20 -0300
Subject: [PATCH 0543/1701] Add attachments support (text files, PDF documents)
 (#7005)

---
 css/main.css                                  |  56 ++++++++
 modules/chat.py                               | 124 ++++++++++++++++--
 modules/html_generator.py                     |  41 ++++++
 modules/ui_chat.py                            |   6 +-
 requirements/full/requirements.txt            |   1 +
 requirements/full/requirements_amd.txt        |   1 +
 requirements/full/requirements_amd_noavx2.txt |   1 +
 .../full/requirements_apple_intel.txt         |   1 +
 .../full/requirements_apple_silicon.txt       |   1 +
 requirements/full/requirements_cpu_only.txt   |   1 +
 .../full/requirements_cpu_only_noavx2.txt     |   1 +
 requirements/full/requirements_noavx2.txt     |   1 +
 requirements/full/requirements_nowheels.txt   |   1 +
 requirements/portable/requirements.txt        |   1 +
 .../portable/requirements_apple_intel.txt     |   1 +
 .../portable/requirements_apple_silicon.txt   |   1 +
 .../portable/requirements_cpu_only.txt        |   1 +
 .../portable/requirements_cpu_only_noavx2.txt |   1 +
 requirements/portable/requirements_noavx2.txt |   1 +
 .../portable/requirements_nowheels.txt        |   1 +
 requirements/portable/requirements_vulkan.txt |   1 +
 .../portable/requirements_vulkan_noavx2.txt   |   1 +
 22 files changed, 233 insertions(+), 12 deletions(-)

diff --git a/css/main.css b/css/main.css
index d7142336bd..6cb99fc3dc 100644
--- a/css/main.css
+++ b/css/main.css
@@ -592,6 +592,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     padding: 0.65rem 2.5rem;
     border: 0;
     box-shadow: 0;
+    border-radius: 8px;
 }
 
 #chat-input textarea::placeholder {
@@ -611,6 +612,16 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     display: none;
 }
 
+#chat-input .submit-button {
+    display: none;
+}
+
+#chat-input .upload-button {
+    margin-right: 16px;
+    margin-bottom: 7px;
+    background: transparent;
+}
+
 .chat-input-positioned {
     max-width: 54rem;
     left: 50%;
@@ -1395,3 +1406,48 @@ strong {
 .dark #vram-info .value {
     color: #07ff07;
 }
+
+.message-attachments {
+    display: flex;
+    flex-wrap: wrap;
+    gap: 8px;
+    margin-top: 8px;
+}
+
+.attachment-box {
+    display: flex;
+    flex-direction: column;
+    align-items: center;
+    justify-content: center;
+    padding: 8px;
+    background: rgb(0 0 0 / 5%);
+    border-radius: 6px;
+    border: 1px solid rgb(0 0 0 / 10%);
+    min-width: 80px;
+    max-width: 120px;
+}
+
+.attachment-icon {
+    margin-bottom: 4px;
+    color: #555;
+}
+
+.attachment-name {
+    font-size: 0.8em;
+    text-align: center;
+    word-break: break-word;
+    overflow: hidden;
+    text-overflow: ellipsis;
+    display: -webkit-box;
+    -webkit-line-clamp: 2;
+    -webkit-box-orient: vertical;
+}
+
+.dark .attachment-box {
+    background: rgb(255 255 255 / 5%);
+    border: 1px solid rgb(255 255 255 / 10%);
+}
+
+.dark .attachment-icon {
+    color: #ccc;
+}
diff --git a/modules/chat.py b/modules/chat.py
index 3efc55db89..cdd50c9220 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -157,7 +157,9 @@ def generate_chat_prompt(user_input, state, **kwargs):
     impersonate = kwargs.get('impersonate', False)
     _continue = kwargs.get('_continue', False)
     also_return_rows = kwargs.get('also_return_rows', False)
-    history = kwargs.get('history', state['history'])['internal']
+    history_data = kwargs.get('history', state['history'])
+    history = history_data['internal']
+    metadata = history_data.get('metadata', {})
 
     # Templates
     chat_template_str = state['chat_template_str']
@@ -196,11 +198,13 @@ def generate_chat_prompt(user_input, state, **kwargs):
             messages.append({"role": "system", "content": context})
 
     insert_pos = len(messages)
-    for entry in reversed(history):
+    for i, entry in enumerate(reversed(history)):
         user_msg = entry[0].strip()
         assistant_msg = entry[1].strip()
         tool_msg = entry[2].strip() if len(entry) > 2 else ''
 
+        row_idx = len(history) - i - 1
+
         if tool_msg:
             messages.insert(insert_pos, {"role": "tool", "content": tool_msg})
 
@@ -208,10 +212,40 @@ def generate_chat_prompt(user_input, state, **kwargs):
             messages.insert(insert_pos, {"role": "assistant", "content": assistant_msg})
 
         if user_msg not in ['', '<|BEGIN-VISIBLE-CHAT|>']:
-            messages.insert(insert_pos, {"role": "user", "content": user_msg})
+            # Check for user message attachments in metadata
+            user_key = f"user_{row_idx}"
+            enhanced_user_msg = user_msg
+
+            # Add attachment content if present
+            if user_key in metadata and "attachments" in metadata[user_key]:
+                attachments_text = ""
+                for attachment in metadata[user_key]["attachments"]:
+                    filename = attachment.get("name", "file")
+                    content = attachment.get("content", "")
+                    attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
+
+                if attachments_text:
+                    enhanced_user_msg = f"{user_msg}\n\nATTACHMENTS:{attachments_text}"
+
+            messages.insert(insert_pos, {"role": "user", "content": enhanced_user_msg})
 
     user_input = user_input.strip()
     if user_input and not impersonate and not _continue:
+        # For the current user input being processed, check if we need to add attachments
+        if not impersonate and not _continue and len(history_data.get('metadata', {})) > 0:
+            current_row_idx = len(history)
+            user_key = f"user_{current_row_idx}"
+
+            if user_key in metadata and "attachments" in metadata[user_key]:
+                attachments_text = ""
+                for attachment in metadata[user_key]["attachments"]:
+                    filename = attachment.get("name", "file")
+                    content = attachment.get("content", "")
+                    attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
+
+                if attachments_text:
+                    user_input = f"{user_input}\n\nATTACHMENTS:{attachments_text}"
+
         messages.append({"role": "user", "content": user_input})
 
     def make_prompt(messages):
@@ -280,7 +314,6 @@ def make_prompt(messages):
 
             # Resort to truncating the user input
             else:
-
                 user_message = messages[-1]['content']
 
                 # Bisect the truncation point
@@ -393,7 +426,74 @@ def add_message_version(history, row_idx, is_current=True):
         history['metadata'][key]["current_version_index"] = len(history['metadata'][key]["versions"]) - 1
 
 
+def add_message_attachment(history, row_idx, file_path, is_user=True):
+    """Add a file attachment to a message in history metadata"""
+    if 'metadata' not in history:
+        history['metadata'] = {}
+
+    key = f"{'user' if is_user else 'assistant'}_{row_idx}"
+
+    if key not in history['metadata']:
+        history['metadata'][key] = {"timestamp": get_current_timestamp()}
+    if "attachments" not in history['metadata'][key]:
+        history['metadata'][key]["attachments"] = []
+
+    # Get file info using pathlib
+    path = Path(file_path)
+    filename = path.name
+    file_extension = path.suffix.lower()
+
+    try:
+        # Handle different file types
+        if file_extension == '.pdf':
+            # Process PDF file
+            content = extract_pdf_text(path)
+            file_type = "application/pdf"
+        else:
+            # Default handling for text files
+            with open(path, 'r', encoding='utf-8') as f:
+                content = f.read()
+            file_type = "text/plain"
+
+        # Add attachment
+        attachment = {
+            "name": filename,
+            "type": file_type,
+            "content": content,
+        }
+
+        history['metadata'][key]["attachments"].append(attachment)
+        return content  # Return the content for reuse
+    except Exception as e:
+        logger.error(f"Error processing attachment {filename}: {e}")
+        return None
+
+
+def extract_pdf_text(pdf_path):
+    """Extract text from a PDF file"""
+    import PyPDF2
+
+    text = ""
+    try:
+        with open(pdf_path, 'rb') as file:
+            pdf_reader = PyPDF2.PdfReader(file)
+            for page_num in range(len(pdf_reader.pages)):
+                page = pdf_reader.pages[page_num]
+                text += page.extract_text() + "\n\n"
+
+        return text.strip()
+    except Exception as e:
+        logger.error(f"Error extracting text from PDF: {e}")
+        return f"[Error extracting PDF text: {str(e)}]"
+
+
 def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_message=True, for_ui=False):
+    # Handle dict format with text and files
+    files = []
+    if isinstance(text, dict):
+        files = text.get('files', [])
+        text = text.get('text', '')
+
     history = state['history']
     output = copy.deepcopy(history)
     output = apply_extensions('history', output)
@@ -411,12 +511,18 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
     if not (regenerate or _continue):
         visible_text = html.escape(text)
 
+        # Process file attachments and store in metadata
+        row_idx = len(output['internal'])
+
+        # Add attachments to metadata only, not modifying the message text
+        for file_path in files:
+            add_message_attachment(output, row_idx, file_path, is_user=True)
+
         # Apply extensions
         text, visible_text = apply_extensions('chat_input', text, visible_text, state)
         text = apply_extensions('input', text, state, is_chat=True)
 
         # Current row index
-        row_idx = len(output['internal'])
         output['internal'].append([text, ''])
         output['visible'].append([visible_text, ''])
         # Add metadata with timestamp
@@ -1215,7 +1321,7 @@ def handle_replace_last_reply_click(text, state):
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
 
-    return [history, html, ""]
+    return [history, html, {"text": "", "files": []}]
 
 
 def handle_send_dummy_message_click(text, state):
@@ -1223,7 +1329,7 @@ def handle_send_dummy_message_click(text, state):
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
 
-    return [history, html, ""]
+    return [history, html, {"text": "", "files": []}]
 
 
 def handle_send_dummy_reply_click(text, state):
@@ -1231,7 +1337,7 @@ def handle_send_dummy_reply_click(text, state):
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
 
-    return [history, html, ""]
+    return [history, html, {"text": "", "files": []}]
 
 
 def handle_remove_last_click(state):
@@ -1239,7 +1345,7 @@ def handle_remove_last_click(state):
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
 
-    return [history, html, last_input]
+    return [history, html, {"text": last_input, "files": []}]
 
 
 def handle_unique_id_select(state):
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 36b31ac599..f5e0b28fe7 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -338,6 +338,7 @@ def get_image_cache(path):
 branch_svg = ''''''
 info_svg = ''''''
 info_svg_small = ''''''
+attachment_svg = ''''''
 
 copy_button = f''
 branch_button = f''
@@ -357,6 +358,28 @@ def format_message_timestamp(history, role, index):
     return ""
 
 
+def format_message_attachments(history, role, index):
+    """Get formatted HTML for message attachments if available"""
+    key = f"{role}_{index}"
+    if 'metadata' in history and key in history['metadata'] and 'attachments' in history['metadata'][key]:
+        attachments = history['metadata'][key]['attachments']
+        if not attachments:
+            return ""
+
+        attachments_html = '
            '
+        for attachment in attachments:
+            attachments_html += (
+                f'
            '
+                f'
            {attachment_svg}
            '
+                f'
            {html.escape(attachment["name"])}
            '
+                f'
            '
+            )
+        attachments_html += '
            '
+        return attachments_html
+
+    return ""
+
+
 def actions_html(history, i, info_message=""):
     return (f'
            '
             f'{copy_button}'
@@ -380,6 +403,10 @@ def generate_instruct_html(history):
         user_timestamp = format_message_timestamp(history, "user", i)
         assistant_timestamp = format_message_timestamp(history, "assistant", i)
 
+        # Get attachments
+        user_attachments = format_message_attachments(history, "user", i)
+        assistant_attachments = format_message_attachments(history, "assistant", i)
+
         # Create info buttons for timestamps if they exist
         info_message_user = ""
         if user_timestamp != "":
@@ -399,6 +426,7 @@ def generate_instruct_html(history):
                 f'data-raw="{html.escape(row_internal[0], quote=True)}">'
                 f'
            '
                 f'
            {converted_visible[0]}
            '
+                f'{user_attachments}'
                 f'
            {copy_button}{info_message_user}
            '
                 f'
            '
                 f'
            '
@@ -410,6 +438,7 @@ def generate_instruct_html(history):
             f'data-index={i}>'
             f'
            '
             f'
            {converted_visible[1]}
            '
+            f'{assistant_attachments}'
             f'{actions_html(history, i, info_message_assistant)}'
             f'
            '
             f'
            '
@@ -442,6 +471,10 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
         user_timestamp = format_message_timestamp(history, "user", i)
         assistant_timestamp = format_message_timestamp(history, "assistant", i)
 
+        # Get attachments
+        user_attachments = format_message_attachments(history, "user", i)
+        assistant_attachments = format_message_attachments(history, "assistant", i)
+
         if converted_visible[0]:  # Don't display empty user messages
             output += (
                 f'
            '
                 f'
            {name1}{user_timestamp}
            '
                 f'
            {converted_visible[0]}
            '
+                f'{user_attachments}'
                 f'
            {copy_button}
            '
                 f'
            '
                 f'
            '
@@ -463,6 +497,7 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
             f'
            '
             f'
            {name2}{assistant_timestamp}
            '
             f'
            {converted_visible[1]}
            '
+            f'{assistant_attachments}'
             f'{actions_html(history, i)}'
             f'
            '
             f'
            '
@@ -484,6 +519,10 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
         user_timestamp = format_message_timestamp(history, "user", i)
         assistant_timestamp = format_message_timestamp(history, "assistant", i)
 
+        # Get attachments
+        user_attachments = format_message_attachments(history, "user", i)
+        assistant_attachments = format_message_attachments(history, "assistant", i)
+
         # Create info buttons for timestamps if they exist
         info_message_user = ""
         if user_timestamp != "":
@@ -503,6 +542,7 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
                 f'data-raw="{html.escape(row_internal[0], quote=True)}">'
                 f'
            '
                 f'
            {converted_visible[0]}
            '
+                f'{user_attachments}'
                 f'
            {copy_button}{info_message_user}
            '
                 f'
            '
                 f'
            '
@@ -514,6 +554,7 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
             f'data-index={i}>'
             f'
            '
             f'
            {converted_visible[1]}
            '
+            f'{assistant_attachments}'
             f'{actions_html(history, i, info_message_assistant)}'
             f'
            '
             f'
            '
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 513a632bb4..f244113c37 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -54,7 +54,7 @@ def create_ui():
                         gr.HTML(value='
            ☰
            ', elem_id='gr-hover')
 
                     with gr.Column(scale=10, elem_id='chat-input-container'):
-                        shared.gradio['textbox'] = gr.Textbox(label='', placeholder='Send a message', elem_id='chat-input', elem_classes=['add_scrollbar'])
+                        shared.gradio['textbox'] = gr.MultimodalTextbox(label='', placeholder='Send a message', file_types=['text', '.pdf'], elem_id='chat-input', elem_classes=['add_scrollbar'])
                         shared.gradio['show_controls'] = gr.Checkbox(value=shared.settings['show_controls'], label='Show controls (Ctrl+S)', elem_id='show-controls')
                         shared.gradio['typing-dots'] = gr.HTML(value='
            ', label='typing', elem_id='typing-container')
 
@@ -186,7 +186,7 @@ def create_event_handlers():
 
     shared.gradio['Generate'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        lambda x: (x, ''), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False).then(
+        lambda x: (x, {"text": "", "files": []}), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False).then(
         lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
         chat.generate_chat_reply_wrapper, gradio(inputs), gradio('display', 'history'), show_progress=False).then(
         None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then(
@@ -194,7 +194,7 @@ def create_event_handlers():
 
     shared.gradio['textbox'].submit(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        lambda x: (x, ''), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False).then(
+        lambda x: (x, {"text": "", "files": []}), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False).then(
         lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
         chat.generate_chat_reply_wrapper, gradio(inputs), gradio('display', 'history'), show_progress=False).then(
         None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then(
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index c65ab8a2d2..afb5f9d483 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -13,6 +13,7 @@ peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 3da16d3ea0..46c330344a 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -12,6 +12,7 @@ peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 271b4bd051..c8e94cbd9a 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -12,6 +12,7 @@ peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 15df937cf8..dc403ae2ae 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -12,6 +12,7 @@ peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index bd2f833971..5c643c4ca6 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -12,6 +12,7 @@ peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 98c2564946..ccabea84e3 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -12,6 +12,7 @@ peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 6e13c1d223..7e9da47f59 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -12,6 +12,7 @@ peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 67a5cb735c..fdf5cd0e32 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -13,6 +13,7 @@ peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 2e631bf0be..22d39ded65 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -12,6 +12,7 @@ peft==0.15.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 409252f6c0..ec9bafc629 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -4,6 +4,7 @@ jinja2==3.1.6
 markdown
 numpy==1.26.*
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 89adbabf52..025a737eb0 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -4,6 +4,7 @@ jinja2==3.1.6
 markdown
 numpy==1.26.*
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 0b1c03fa25..32644e87e6 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -4,6 +4,7 @@ jinja2==3.1.6
 markdown
 numpy==1.26.*
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index eb4319b7e2..bd5c1d9bdc 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -4,6 +4,7 @@ jinja2==3.1.6
 markdown
 numpy==1.26.*
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 0a60d4de4b..51f2b7d9c7 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -4,6 +4,7 @@ jinja2==3.1.6
 markdown
 numpy==1.26.*
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 652e990047..aad6bf5a8d 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -4,6 +4,7 @@ jinja2==3.1.6
 markdown
 numpy==1.26.*
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 6f9566baed..4c05542622 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -4,6 +4,7 @@ jinja2==3.1.6
 markdown
 numpy==1.26.*
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index c83d61c705..3d98d1b035 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -4,6 +4,7 @@ jinja2==3.1.6
 markdown
 numpy==1.26.*
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index e69f3bdf37..f954b8d248 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -4,6 +4,7 @@ jinja2==3.1.6
 markdown
 numpy==1.26.*
 pydantic==2.8.2
+PyPDF2==3.0.1
 pyyaml
 requests
 rich

From cc8a4fdcb114bfd068c42cea267e34daaf901a30 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 20 May 2025 21:31:18 -0700
Subject: [PATCH 0544/1701] Minor improvement to attachments prompt format

---
 modules/chat.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index cdd50c9220..715f4327bc 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -225,7 +225,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
                     attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
 
                 if attachments_text:
-                    enhanced_user_msg = f"{user_msg}\n\nATTACHMENTS:{attachments_text}"
+                    enhanced_user_msg = f"{user_msg}\n\nATTACHMENTS:\n{attachments_text}"
 
             messages.insert(insert_pos, {"role": "user", "content": enhanced_user_msg})
 
@@ -244,7 +244,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
                     attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
 
                 if attachments_text:
-                    user_input = f"{user_input}\n\nATTACHMENTS:{attachments_text}"
+                    user_input = f"{user_input}\n\nATTACHMENTS:\n{attachments_text}"
 
         messages.append({"role": "user", "content": user_input})
 

From 8620d6ffe73048932594494752f82cc4a20f8f92 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 20 May 2025 21:34:07 -0700
Subject: [PATCH 0545/1701] Make it possible to upload multiple text files/pdfs
 at once

---
 modules/ui_chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index f244113c37..ab4b4e609e 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -54,7 +54,7 @@ def create_ui():
                         gr.HTML(value='
            ☰
            ', elem_id='gr-hover')
 
                     with gr.Column(scale=10, elem_id='chat-input-container'):
-                        shared.gradio['textbox'] = gr.MultimodalTextbox(label='', placeholder='Send a message', file_types=['text', '.pdf'], elem_id='chat-input', elem_classes=['add_scrollbar'])
+                        shared.gradio['textbox'] = gr.MultimodalTextbox(label='', placeholder='Send a message', file_types=['text', '.pdf'], file_count="multiple", elem_id='chat-input', elem_classes=['add_scrollbar'])
                         shared.gradio['show_controls'] = gr.Checkbox(value=shared.settings['show_controls'], label='Show controls (Ctrl+S)', elem_id='show-controls')
                         shared.gradio['typing-dots'] = gr.HTML(value='
            ', label='typing', elem_id='typing-container')
 

From 0d3f85477897c2999f456713ce998b59b26a6a22 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 20 May 2025 21:40:42 -0700
Subject: [PATCH 0546/1701] Improve the style of thinking blocks

---
 css/main.css | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/css/main.css b/css/main.css
index 6cb99fc3dc..8444cae894 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1370,6 +1370,11 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     contain: layout;
 }
 
+.chat .message-body .thinking-content p,
+.chat .message-body .thinking-content li {
+    font-size: 14px !important;
+}
+
 /* Animation for opening thinking blocks */
 @keyframes fadeIn {
     from { opacity: 0; }

From 7f6579ab20d8fd215e81f3b766f3aa9d83066bdb Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 20 May 2025 21:49:44 -0700
Subject: [PATCH 0547/1701] Minor style change

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 8444cae894..d1be8eb176 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1372,7 +1372,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 
 .chat .message-body .thinking-content p,
 .chat .message-body .thinking-content li {
-    font-size: 14px !important;
+    font-size: 15px !important;
 }
 
 /* Animation for opening thinking blocks */

From bae1aa34aa020aa749f942708b96e28e2b85c4a4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 25 May 2025 17:19:26 -0700
Subject: [PATCH 0548/1701] Fix loading `Llama-3_3-Nemotron-Super-49B-v1` and
 similar models (closes #7012)

---
 modules/models_settings.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index df5a8e8ddf..c914bdeae3 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -335,7 +335,7 @@ def estimate_vram(gguf_file, gpu_layers, ctx_size, cache_type):
         if key.endswith('.block_count'):
             n_layers = value
         elif key.endswith('.attention.head_count_kv'):
-            n_kv_heads = value
+            n_kv_heads = max(value) if isinstance(value, list) else value
         elif key.endswith('.embedding_length'):
             embedding_dim = value
 

From 73bfc936a078ce428cc10b590a83e0391b6aed58 Mon Sep 17 00:00:00 2001
From: djholtby 
Date: Mon, 26 May 2025 21:39:03 -0400
Subject: [PATCH 0549/1701] Close response generator when stopping API
 generation (#7014)

---
 extensions/openai/script.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index b6abae20e8..24bcd69d24 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -125,6 +125,7 @@ async def generator():
                         yield {"data": json.dumps(resp)}
                 finally:
                     stop_everything_event()
+                    response.close()
                     return
 
         return EventSourceResponse(generator())  # SSE streaming
@@ -157,6 +158,7 @@ async def generator():
                         yield {"data": json.dumps(resp)}
                 finally:
                     stop_everything_event()
+                    response.close()
                     return
 
         return EventSourceResponse(generator())  # SSE streaming

From 8531100109ecc4a5bed41cc2f3adaddf9d7157f8 Mon Sep 17 00:00:00 2001
From: Underscore <47636331+Th-Underscore@users.noreply.github.com>
Date: Mon, 26 May 2025 21:40:09 -0400
Subject: [PATCH 0550/1701] Fix textbox text usage in methods (#7009)

---
 modules/chat.py | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 715f4327bc..36a078363d 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -708,8 +708,9 @@ def send_last_reply_to_input(history):
         return ''
 
 
-def replace_last_reply(text, state):
+def replace_last_reply(textbox, state):
     history = state['history']
+    text = textbox['text']
 
     # Initialize metadata if not present
     if 'metadata' not in history:
@@ -726,8 +727,9 @@ def replace_last_reply(text, state):
     return history
 
 
-def send_dummy_message(text, state):
+def send_dummy_message(textbox, state):
     history = state['history']
+    text = textbox['text']
 
     # Initialize metadata if not present
     if 'metadata' not in history:
@@ -741,8 +743,9 @@ def send_dummy_message(text, state):
     return history
 
 
-def send_dummy_reply(text, state):
+def send_dummy_reply(textbox, state):
     history = state['history']
+    text = textbox['text']
 
     # Initialize metadata if not present
     if 'metadata' not in history:

From cc9b7253c1216e5340da85cba9b65a13cf3526e9 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Mon, 26 May 2025 23:13:10 -0300
Subject: [PATCH 0551/1701] Update transformers requirement in
 /requirements/full (#7017)

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index afb5f9d483..3d18f5fd04 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.52.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 46c330344a..82b199645a 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.52.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index c8e94cbd9a..a8b0301454 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.52.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index dc403ae2ae..5a61ac7dc1 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.52.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 5c643c4ca6..6862c3b466 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.52.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index ccabea84e3..e698277993 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.52.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 7e9da47f59..97bff786ba 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.52.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index fdf5cd0e32..17c7e24624 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -21,7 +21,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.52.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 22d39ded65..89b32caf58 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -20,7 +20,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.52.*
 tqdm
 wandb
 

From 355b5f6c8b5552ccdae1aa363931724306bdbb16 Mon Sep 17 00:00:00 2001
From: Underscore <47636331+Th-Underscore@users.noreply.github.com>
Date: Tue, 27 May 2025 21:54:18 -0400
Subject: [PATCH 0552/1701] UI: Add message version navigation (#6947)

---------

Co-authored-by: oobabooga <112222186+oobabooga@users.noreply.github.com>
---
 css/main.css              | 41 ++++++++++++++++-
 js/global_scope_js.js     | 38 ++++++++++++++++
 js/main.js                | 93 ++++++++++++++++++++++++++++++++++++++-
 modules/chat.py           | 59 +++++++++++++++++++++++--
 modules/html_generator.py | 27 +++++++++++-
 modules/ui.py             |  2 +
 modules/ui_chat.py        | 10 +++++
 7 files changed, 262 insertions(+), 8 deletions(-)

diff --git a/css/main.css b/css/main.css
index d1be8eb176..be27544c2f 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1260,7 +1260,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     position: absolute;
     bottom: -23px;
     left: 0;
-    display: flex;   
+    display: flex;
     gap: 5px;
     opacity: 0;
     transition: opacity 0.2s;
@@ -1456,3 +1456,42 @@ strong {
 .dark .attachment-icon {
     color: #ccc;
 }
+
+/* --- Simple Version Navigation --- */
+.version-navigation {
+    position: absolute;
+    bottom: -23px;
+    right: 0;
+    display: flex;
+    align-items: center;
+    gap: 5px;
+    opacity: 0;
+    transition: opacity 0.2s;
+}
+
+.message:hover .version-navigation,
+.user-message:hover .version-navigation,
+.assistant-message:hover .version-navigation {
+    opacity: 1;
+}
+
+.version-nav-button {
+    padding: 2px 6px;
+    font-size: 12px;
+    min-width: auto;
+}
+
+.version-nav-button[disabled] {
+    opacity: 0.3;
+    cursor: not-allowed;
+}
+
+.version-position {
+    font-size: 11px;
+    color: currentColor;
+    font-family: monospace;
+    min-width: 35px;
+    text-align: center;
+    opacity: 0.8;
+    user-select: none;
+}
diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 285d82f91b..9174622ed1 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -49,6 +49,44 @@ function branchHere(element) {
 
 }
 
+function navigateVersion(element, direction) {
+  if (!element) return;
+
+  const messageElement = element.closest(".message, .user-message, .assistant-message");
+  if (!messageElement) return;
+
+  const index = messageElement.getAttribute("data-index");
+  if (!index) return;
+
+  const indexInput = document.getElementById("Navigate-message-index").querySelector("input");
+  if (!indexInput) {
+    console.error("Element with ID 'Navigate-message-index' not found.");
+    return;
+  }
+
+  const directionInput = document.getElementById("Navigate-direction").querySelector("textarea");
+  if (!directionInput) {
+    console.error("Element with ID 'Navigate-direction' not found.");
+    return;
+  }
+
+  const navigateButton = document.getElementById("Navigate-version");
+  if (!navigateButton) {
+    console.error("Required element 'Navigate-version' not found.");
+    return;
+  }
+
+  indexInput.value = index;
+  directionInput.value = direction;
+
+  // Trigger any 'change' or 'input' events Gradio might be listening for
+  const event = new Event("input", { bubbles: true });
+  indexInput.dispatchEvent(event);
+  directionInput.dispatchEvent(event);
+
+  navigateButton.click();
+}
+
 function regenerateClick() {
   document.getElementById("Regenerate").click();
 }
diff --git a/js/main.js b/js/main.js
index 01c346a74c..d90e8adeb0 100644
--- a/js/main.js
+++ b/js/main.js
@@ -39,9 +39,24 @@ document.querySelector(".header_bar").addEventListener("click", function(event)
 //------------------------------------------------
 // Keyboard shortcuts
 //------------------------------------------------
+
+// --- Helper functions --- //
+function isModifiedKeyboardEvent() {
+  return (event instanceof KeyboardEvent &&
+    event.shiftKey ||
+    event.ctrlKey ||
+    event.altKey ||
+    event.metaKey);
+}
+
+function isFocusedOnEditableTextbox() {
+  if (event.target.tagName === "INPUT" || event.target.tagName === "TEXTAREA") {
+    return !!event.target.value;
+  }
+}
+
 let previousTabId = "chat-tab-button";
 document.addEventListener("keydown", function(event) {
-
   // Stop generation on Esc pressed
   if (event.key === "Escape") {
     // Find the element with id 'stop' and click it
@@ -49,10 +64,15 @@ document.addEventListener("keydown", function(event) {
     if (stopButton) {
       stopButton.click();
     }
+    return;
+  }
+
+  if (!document.querySelector("#chat-tab").checkVisibility() ) {
+    return;
   }
 
   // Show chat controls on Ctrl + S
-  else if (event.ctrlKey && event.key == "s") {
+  if (event.ctrlKey && event.key == "s") {
     event.preventDefault();
 
     var showControlsElement = document.getElementById("show-controls");
@@ -100,6 +120,23 @@ document.addEventListener("keydown", function(event) {
     document.getElementById("Impersonate").click();
   }
 
+  // --- Simple version navigation --- //
+  if (!isFocusedOnEditableTextbox()) {
+    // Version navigation on Arrow keys (horizontal)
+    if (!isModifiedKeyboardEvent() && event.key === "ArrowLeft") {
+      event.preventDefault();
+      navigateLastAssistantMessage("left");
+    }
+
+    else if (!isModifiedKeyboardEvent() && event.key === "ArrowRight") {
+      event.preventDefault();
+      if (!navigateLastAssistantMessage("right")) {
+        // If can't navigate right (last version), regenerate
+        document.getElementById("Regenerate").click();
+      }
+    }
+  }
+
 });
 
 //------------------------------------------------
@@ -789,3 +826,55 @@ function createMobileTopBar() {
 }
 
 createMobileTopBar();
+
+//------------------------------------------------
+// Simple Navigation Functions
+//------------------------------------------------
+
+function navigateLastAssistantMessage(direction) {
+  const chat = document.querySelector("#chat");
+  if (!chat) return false;
+
+  const messages = chat.querySelectorAll("[data-index]");
+  if (messages.length === 0) return false;
+
+  // Find the last assistant message (starting from the end)
+  let lastAssistantMessage = null;
+  for (let i = messages.length - 1; i >= 0; i--) {
+    const msg = messages[i];
+    if (
+      msg.classList.contains("assistant-message") ||
+      msg.querySelector(".circle-bot") ||
+      msg.querySelector(".text-bot")
+    ) {
+      lastAssistantMessage = msg;
+      break;
+    }
+  }
+
+  if (!lastAssistantMessage) return false;
+
+  const buttons = lastAssistantMessage.querySelectorAll(".version-nav-button");
+
+  for (let i = 0; i < buttons.length; i++) {
+    const button = buttons[i];
+    const onclick = button.getAttribute("onclick");
+    const disabled = button.hasAttribute("disabled");
+
+    const isLeft = onclick && onclick.includes("'left'");
+    const isRight = onclick && onclick.includes("'right'");
+
+    if (!disabled) {
+      if (direction === "left" && isLeft) {
+        navigateVersion(button, direction);
+        return true;
+      }
+      if (direction === "right" && isRight) {
+        navigateVersion(button, direction);
+        return true;
+      }
+    }
+  }
+
+  return false;
+}
diff --git a/modules/chat.py b/modules/chat.py
index 36a078363d..6eed47ee80 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -414,10 +414,20 @@ def add_message_version(history, row_idx, is_current=True):
     if "versions" not in history['metadata'][key]:
         history['metadata'][key]["versions"] = []
 
+    # Check if this version already exists
+    current_content = history['internal'][row_idx][1]
+    current_visible = history['visible'][row_idx][1]
+
+    for i, version in enumerate(history['metadata'][key]["versions"]):
+        if version['content'] == current_content and version['visible_content'] == current_visible:
+            if is_current:
+                history['metadata'][key]["current_version_index"] = i
+            return
+
     # Add current message as a version
     history['metadata'][key]["versions"].append({
-        "content": history['internal'][row_idx][1],
-        "visible_content": history['visible'][row_idx][1],
+        "content": current_content,
+        "visible_content": current_visible,
         "timestamp": get_current_timestamp()
     })
 
@@ -540,8 +550,9 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
         if regenerate:
             row_idx = len(output['internal']) - 1
 
-            # Store the existing response as a version before regenerating
-            add_message_version(output, row_idx, is_current=False)
+            # Store the first response as a version before regenerating
+            if not output['metadata'].get(f"assistant_{row_idx}", {}).get('versions'):
+                add_message_version(output, row_idx, is_current=False)
 
             if loading_message:
                 yield {
@@ -1414,6 +1425,46 @@ def handle_branch_chat_click(state):
     return [history, html, past_chats_update, -1]
 
 
+def handle_navigate_version_click(state):
+    history = state['history']
+    message_index = int(state['navigate_message_index'])
+    direction = state['navigate_direction']
+
+    # Get assistant message metadata
+    key = f"assistant_{message_index}"
+    if key not in history['metadata'] or 'versions' not in history['metadata'][key]:
+        # No versions to navigate
+        html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
+        return [history, html]
+
+    metadata = history['metadata'][key]
+    current_idx = metadata.get('current_version_index', 0)
+    versions = metadata['versions']
+
+    # Calculate new index
+    if direction == 'left':
+        new_idx = max(0, current_idx - 1)
+    else:  # right
+        new_idx = min(len(versions) - 1, current_idx + 1)
+
+    if new_idx == current_idx:
+        # No change needed
+        html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
+        return [history, html]
+
+    # Update history with new version
+    version = versions[new_idx]
+    history['internal'][message_index][1] = version['content']
+    history['visible'][message_index][1] = version['visible_content']
+    metadata['current_version_index'] = new_idx
+
+    # Redraw and save
+    html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
+    save_history(history, state['unique_id'], state['character_menu'], state['mode'])
+
+    return [history, html]
+
+
 def handle_rename_chat_click():
     return [
         gr.update(value="My New Chat"),
diff --git a/modules/html_generator.py b/modules/html_generator.py
index f5e0b28fe7..1dfeb44519 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -380,6 +380,30 @@ def format_message_attachments(history, role, index):
     return ""
 
 
+def get_version_navigation_html(history, i):
+    """Generate simple navigation arrows for message versions"""
+    key = f"assistant_{i}"
+    metadata = history.get('metadata', {})
+
+    if key not in metadata or 'versions' not in metadata[key]:
+        return ""
+
+    versions = metadata[key]['versions']
+    current_idx = metadata[key].get('current_version_index', 0)
+
+    if len(versions) <= 1:
+        return ""
+
+    left_disabled = ' disabled' if current_idx == 0 else ''
+    right_disabled = ' disabled' if current_idx >= len(versions) - 1 else ''
+
+    left_arrow = f''
+    right_arrow = f''
+    position = f'{current_idx + 1}/{len(versions)}'
+
+    return f'
            {left_arrow}{position}{right_arrow}
            '
+
+
 def actions_html(history, i, info_message=""):
     return (f'
            '
             f'{copy_button}'
@@ -388,7 +412,8 @@ def actions_html(history, i, info_message=""):
             f'{remove_button if i == len(history["visible"]) - 1 else ""}'
             f'{branch_button}'
             f'{info_message}'
-            f'
            ')
+            f'
            '
+            f'{get_version_navigation_html(history, i)}')
 
 
 def generate_instruct_html(history):
diff --git a/modules/ui.py b/modules/ui.py
index 5e8fa14e9b..52c095a271 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -157,6 +157,8 @@ def list_model_elements():
 
 def list_interface_input_elements():
     elements = [
+        'navigate_message_index',
+        'navigate_direction',
         'temperature',
         'dynatemp_low',
         'dynatemp_high',
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index ab4b4e609e..7a9f6f76ab 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -97,6 +97,12 @@ def create_ui():
                 with gr.Row():
                     shared.gradio['chat-instruct_command'] = gr.Textbox(value=shared.settings['chat-instruct_command'], lines=12, label='Command for chat-instruct mode', info='<|character|> and <|prompt|> get replaced with the bot name and the regular chat prompt respectively.', visible=shared.settings['mode'] == 'chat-instruct', elem_classes=['add_scrollbar'])
 
+        # Hidden elements for version navigation (similar to branch)
+        with gr.Row(visible=False):
+            shared.gradio['navigate_message_index'] = gr.Number(value=-1, precision=0, elem_id="Navigate-message-index")
+            shared.gradio['navigate_direction'] = gr.Textbox(value="", elem_id="Navigate-direction")
+            shared.gradio['navigate_version'] = gr.Button(elem_id="Navigate-version")
+
 
 def create_chat_settings_ui():
     mu = shared.args.multi_user
@@ -293,6 +299,10 @@ def create_event_handlers():
     shared.gradio['chat_style'].change(chat.redraw_html, gradio(reload_arr), gradio('display'), show_progress=False)
     shared.gradio['Copy last reply'].click(chat.send_last_reply_to_input, gradio('history'), gradio('textbox'), show_progress=False)
 
+    shared.gradio['navigate_version'].click(
+        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
+        chat.handle_navigate_version_click, gradio('interface_state'), gradio('history', 'display'), show_progress=False)
+
     # Save/delete a character
     shared.gradio['save_character'].click(chat.handle_save_character_click, gradio('name2'), gradio('save_character_filename', 'character_saver'), show_progress=False)
     shared.gradio['delete_character'].click(lambda: gr.update(visible=True), None, gradio('character_deleter'), show_progress=False)

From 5028480ebabf26ec44778588b4fbd019cd9456ed Mon Sep 17 00:00:00 2001
From: Underscore <47636331+Th-Underscore@users.noreply.github.com>
Date: Tue, 27 May 2025 23:55:27 -0400
Subject: [PATCH 0553/1701] UI: Add footer buttons for editing messages (#7019)

---------

Co-authored-by: oobabooga <112222186+oobabooga@users.noreply.github.com>
---
 css/main.css              |  49 +++++++++++++-
 js/global_scope_js.js     | 132 +++++++++++++++++++++++++++++++++++++-
 js/main.js                |  26 ++++----
 modules/chat.py           |  80 +++++++++++++----------
 modules/html_generator.py |  44 ++++++++-----
 modules/ui.py             |   7 +-
 modules/ui_chat.py        |  18 +++---
 7 files changed, 282 insertions(+), 74 deletions(-)

diff --git a/css/main.css b/css/main.css
index be27544c2f..9d68ba02a9 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1457,6 +1457,53 @@ strong {
     color: #ccc;
 }
 
+/* Message Editing Styles */
+.editing-textarea {
+    width: 100%;
+    min-height: 200px;
+    padding: 10px;
+    border-radius: 5px;
+    border: 1px solid #ccc;
+    background-color: var(--light-theme-gray);
+    font-family: inherit;
+    font-size: inherit;
+    resize: vertical;
+}
+
+.dark .editing-textarea {
+    border: 1px solid var(--border-color-dark);
+    background-color: var(--darker-gray);
+}
+
+.editing-textarea:focus {
+    outline: none;
+    border-color: var(--selected-item-color-dark);
+}
+
+.edit-controls-container {
+    margin-top: 0;
+    display: flex;
+    gap: 8px;
+    padding-bottom: 8px;
+}
+
+.edit-control-button {
+    padding: 6px 12px;
+    border: 1px solid #ccc;
+    border-radius: 4px;
+    cursor: pointer;
+    background-color: #f8f9fa;
+    color: #212529;
+    font-size: 12px;
+    margin: 0;
+}
+
+.dark .edit-control-button {
+    border: 1px solid var(--border-color-dark);
+    background-color: var(--light-gray);
+    color: #efefef;
+}
+
 /* --- Simple Version Navigation --- */
 .version-navigation {
     position: absolute;
@@ -1488,7 +1535,7 @@ strong {
 
 .version-position {
     font-size: 11px;
-    color: currentColor;
+    color: currentcolor;
     font-family: monospace;
     min-width: 35px;
     text-align: center;
diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 9174622ed1..0e86d450fe 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -1,3 +1,7 @@
+// -------------------------------------------------
+// Event handlers
+// -------------------------------------------------
+
 function copyToClipboard(element) {
   if (!element) return;
 
@@ -42,11 +46,135 @@ function branchHere(element) {
   branchIndexInput.value = index;
 
   // Trigger any 'change' or 'input' events Gradio might be listening for
-  const event = new Event("input", { bubbles: true }); // 'change' might also work
+  const event = new Event("input", { bubbles: true });
   branchIndexInput.dispatchEvent(event);
 
-  branchButton.click(); // Gradio will now pick up the 'index'
+  branchButton.click();
+}
+
+// -------------------------------------------------
+// Message Editing Functions
+// -------------------------------------------------
+
+function editHere(buttonElement) {
+  if (!buttonElement) return;
+
+  const messageElement = buttonElement.closest(".message, .user-message, .assistant-message");
+  if (!messageElement) return;
+
+  const messageBody = messageElement.querySelector(".message-body");
+  if (!messageBody) return;
+
+  // If already editing, focus the textarea
+  const existingTextarea = messageBody.querySelector(".editing-textarea");
+  if (existingTextarea) {
+    existingTextarea.focus();
+    return;
+  }
+
+  // Determine role based on message element - handle different chat modes
+  const isUserMessage = messageElement.classList.contains("user-message") ||
+                       messageElement.querySelector(".text-you") !== null ||
+                       messageElement.querySelector(".circle-you") !== null;
+
+  startEditing(messageElement, messageBody, isUserMessage);
+}
+
+function startEditing(messageElement, messageBody, isUserMessage) {
+  const rawText = messageElement.getAttribute("data-raw") || messageBody.textContent;
+  const originalHTML = messageBody.innerHTML;
+
+  // Create editing interface
+  const editingInterface = createEditingInterface(rawText);
+
+  // Replace message content
+  messageBody.innerHTML = "";
+  messageBody.appendChild(editingInterface.textarea);
+  messageBody.appendChild(editingInterface.controls);
+
+  editingInterface.textarea.focus();
+  editingInterface.textarea.setSelectionRange(rawText.length, rawText.length);
+
+  // Setup event handlers
+  setupEditingHandlers(editingInterface.textarea, messageElement, originalHTML, messageBody, isUserMessage);
+}
+
+function createEditingInterface(text) {
+  const textarea = document.createElement("textarea");
+  textarea.value = text;
+  textarea.className = "editing-textarea";
+  textarea.rows = Math.max(3, text.split("\n").length);
+
+  const controls = document.createElement("div");
+  controls.className = "edit-controls-container";
+
+  const saveButton = document.createElement("button");
+  saveButton.textContent = "Save";
+  saveButton.className = "edit-control-button";
+  saveButton.type = "button";
+
+  const cancelButton = document.createElement("button");
+  cancelButton.textContent = "Cancel";
+  cancelButton.className = "edit-control-button edit-cancel-button";
+  cancelButton.type = "button";
+
+  controls.appendChild(saveButton);
+  controls.appendChild(cancelButton);
+
+  return { textarea, controls, saveButton, cancelButton };
+}
+
+function setupEditingHandlers(textarea, messageElement, originalHTML, messageBody, isUserMessage) {
+  const saveButton = messageBody.querySelector(".edit-control-button:not(.edit-cancel-button)");
+  const cancelButton = messageBody.querySelector(".edit-cancel-button");
+
+  const submitEdit = () => {
+    const index = messageElement.getAttribute("data-index");
+    if (!index || !submitMessageEdit(index, textarea.value, isUserMessage)) {
+      cancelEdit();
+    }
+  };
+
+  const cancelEdit = () => {
+    messageBody.innerHTML = originalHTML;
+  };
+
+  // Event handlers
+  saveButton.onclick = submitEdit;
+  cancelButton.onclick = cancelEdit;
+
+  textarea.onkeydown = (e) => {
+    if (e.key === "Enter" && !e.shiftKey) {
+      e.preventDefault();
+      submitEdit();
+    } else if (e.key === "Escape") {
+      e.preventDefault();
+      cancelEdit();
+    }
+  };
+}
+
+function submitMessageEdit(index, newText, isUserMessage) {
+  const editIndexInput = document.getElementById("Edit-message-index")?.querySelector("input");
+  const editTextInput = document.getElementById("Edit-message-text")?.querySelector("textarea");
+  const editRoleInput = document.getElementById("Edit-message-role")?.querySelector("textarea");
+  const editButton = document.getElementById("Edit-message");
+
+  if (!editIndexInput || !editTextInput || !editRoleInput || !editButton) {
+    console.error("Edit elements not found");
+    return false;
+  }
+
+  editIndexInput.value = index;
+  editTextInput.value = newText;
+  editRoleInput.value = isUserMessage ? "user" : "assistant";
+
+  editIndexInput.dispatchEvent(new Event("input", { bubbles: true }));
+  editTextInput.dispatchEvent(new Event("input", { bubbles: true }));
+  editRoleInput.dispatchEvent(new Event("input", { bubbles: true }));
 
+  editButton.click();
+  return true;
 }
 
 function navigateVersion(element, direction) {
diff --git a/js/main.js b/js/main.js
index d90e8adeb0..fc014f66ac 100644
--- a/js/main.js
+++ b/js/main.js
@@ -1,3 +1,7 @@
+// ------------------------------------------------
+// Main
+// ------------------------------------------------
+
 let main_parent = document.getElementById("chat-tab").parentNode;
 let extensions = document.getElementById("extensions");
 
@@ -102,18 +106,6 @@ document.addEventListener("keydown", function(event) {
     document.getElementById("Remove-last").click();
   }
 
-  // Copy last on Ctrl + Shift + K
-  else if (event.ctrlKey && event.shiftKey && event.key === "K") {
-    event.preventDefault();
-    document.getElementById("Copy-last").click();
-  }
-
-  // Replace last on Ctrl + Shift + L
-  else if (event.ctrlKey && event.shiftKey && event.key === "L") {
-    event.preventDefault();
-    document.getElementById("Replace-last").click();
-  }
-
   // Impersonate on Ctrl + Shift + M
   else if (event.ctrlKey && event.shiftKey && event.key === "M") {
     event.preventDefault();
@@ -388,6 +380,16 @@ document.addEventListener("click", function (event) {
   }
 });
 
+document.addEventListener("dblclick", (event) => {
+  const messageElement = event.target.closest(".message, .user-message, .assistant-message");
+  if (!messageElement) return;
+
+  const editButton = messageElement.querySelector(".footer-edit-button");
+  if (editButton) {
+    editButton.click();
+  }
+});
+
 //------------------------------------------------
 // Relocate the "Show controls" checkbox
 //------------------------------------------------
diff --git a/modules/chat.py b/modules/chat.py
index 6eed47ee80..9598efa75d 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -712,32 +712,6 @@ def remove_last_message(history):
     return html.unescape(last[0]), history
 
 
-def send_last_reply_to_input(history):
-    if len(history['visible']) > 0:
-        return html.unescape(history['visible'][-1][1])
-    else:
-        return ''
-
-
-def replace_last_reply(textbox, state):
-    history = state['history']
-    text = textbox['text']
-
-    # Initialize metadata if not present
-    if 'metadata' not in history:
-        history['metadata'] = {}
-
-    if len(text.strip()) == 0:
-        return history
-    elif len(history['visible']) > 0:
-        row_idx = len(history['internal']) - 1
-        history['visible'][-1][1] = html.escape(text)
-        history['internal'][-1][1] = apply_extensions('input', text, state, is_chat=True)
-        update_message_metadata(history['metadata'], "assistant", row_idx, timestamp=get_current_timestamp())
-
-    return history
-
-
 def send_dummy_message(textbox, state):
     history = state['history']
     text = textbox['text']
@@ -1330,14 +1304,6 @@ def my_yaml_output(data):
     return result
 
 
-def handle_replace_last_reply_click(text, state):
-    history = replace_last_reply(text, state)
-    save_history(history, state['unique_id'], state['character_menu'], state['mode'])
-    html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
-
-    return [history, html, {"text": "", "files": []}]
-
-
 def handle_send_dummy_message_click(text, state):
     history = send_dummy_message(text, state)
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
@@ -1425,6 +1391,52 @@ def handle_branch_chat_click(state):
     return [history, html, past_chats_update, -1]
 
 
+def handle_edit_message_click(state):
+    history = state['history']
+    message_index = int(state['edit_message_index'])
+    new_text = state['edit_message_text']
+    role = state['edit_message_role']  # "user" or "assistant"
+
+    if message_index >= len(history['internal']):
+        html_output = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
+        return [history, html_output, gr.update()]
+
+    # Use the role passed from frontend
+    is_user_msg = (role == "user")
+    role_idx = 0 if is_user_msg else 1
+
+    # For assistant messages, save the original version BEFORE updating content
+    if not is_user_msg:
+        if not history['metadata'].get(f"assistant_{message_index}", {}).get('versions'):
+            add_message_version(history, message_index, is_current=False)
+
+    # NOW update the message content
+    history['internal'][message_index][role_idx] = apply_extensions('input', new_text, state, is_chat=True)
+    history['visible'][message_index][role_idx] = html.escape(new_text)
+
+    # Branch if editing user message, add version if editing assistant message
+    if is_user_msg:
+        # Branch like branch-here
+        history['visible'] = history['visible'][:message_index + 1]
+        history['internal'] = history['internal'][:message_index + 1]
+        new_unique_id = datetime.now().strftime('%Y%m%d-%H-%M-%S')
+        save_history(history, new_unique_id, state['character_menu'], state['mode'])
+        histories = find_all_histories_with_first_prompts(state)
+        past_chats_update = gr.update(choices=histories, value=new_unique_id)
+        state['unique_id'] = new_unique_id
+    elif not is_user_msg:
+        # Add the new version as current
+        add_message_version(history, message_index, is_current=True)
+        past_chats_update = gr.update()
+    else:
+        past_chats_update = gr.update()
+
+    save_history(history, state['unique_id'], state['character_menu'], state['mode'])
+    html_output = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
+
+    return [history, html_output, past_chats_update]
+
+
 def handle_navigate_version_click(state):
     history = state['history']
     message_index = int(state['navigate_message_index'])
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 1dfeb44519..9a93555f9c 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -336,12 +336,14 @@ def get_image_cache(path):
 continue_svg = ''''''
 remove_svg = ''''''
 branch_svg = ''''''
+edit_svg = ''''''
 info_svg = ''''''
 info_svg_small = ''''''
 attachment_svg = ''''''
 
 copy_button = f''
 branch_button = f''
+edit_button = f''
 refresh_button = f''
 continue_button = f''
 remove_button = f''
@@ -404,16 +406,23 @@ def get_version_navigation_html(history, i):
     return f'
            {left_arrow}{position}{right_arrow}
            '
 
 
-def actions_html(history, i, info_message=""):
+def actions_html(history, i, role, info_message=""):
+    if role == "assistant":
+        return (f'
            '
+                f'{copy_button}'
+                f'{edit_button}'
+                f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
+                f'{continue_button if i == len(history["visible"]) - 1 else ""}'
+                f'{remove_button if i == len(history["visible"]) - 1 else ""}'
+                f'{branch_button}'
+                f'{info_message}'
+                f'
            '
+                f'{get_version_navigation_html(history, i)}')
     return (f'
            '
             f'{copy_button}'
-            f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
-            f'{continue_button if i == len(history["visible"]) - 1 else ""}'
-            f'{remove_button if i == len(history["visible"]) - 1 else ""}'
-            f'{branch_button}'
+            f'{edit_button}'
             f'{info_message}'
-            f'
            '
-            f'{get_version_navigation_html(history, i)}')
+            f'
            ')
 
 
 def generate_instruct_html(history):
@@ -448,11 +457,12 @@ def generate_instruct_html(history):
         if converted_visible[0]:  # Don't display empty user messages
             output += (
                 f'
            '
+                f'data-raw="{html.escape(row_internal[0], quote=True)}"'
+                f'data-index={i}>'
                 f'
            '
                 f'
            {converted_visible[0]}
            '
                 f'{user_attachments}'
-                f'
            {copy_button}{info_message_user}
            '
+                f'{actions_html(history, i, "user", info_message_user)}'
                 f'
            '
                 f'
            '
             )
@@ -464,7 +474,7 @@ def generate_instruct_html(history):
             f'
            '
             f'
            {converted_visible[1]}
            '
             f'{assistant_attachments}'
-            f'{actions_html(history, i, info_message_assistant)}'
+            f'{actions_html(history, i, "assistant", info_message_assistant)}'
             f'
            '
             f'
            '
         )
@@ -503,13 +513,14 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
         if converted_visible[0]:  # Don't display empty user messages
             output += (
                 f'
            '
+                f'data-raw="{html.escape(row_internal[0], quote=True)}"'
+                f'data-index={i}>'
                 f'
            {img_me}
            '
                 f'
            '
                 f'
            {name1}{user_timestamp}
            '
                 f'
            {converted_visible[0]}
            '
                 f'{user_attachments}'
-                f'
            {copy_button}
            '
+                f'{actions_html(history, i, "user")}'
                 f'
            '
                 f'
            '
             )
@@ -523,7 +534,7 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
             f'
            {name2}{assistant_timestamp}
            '
             f'
            {converted_visible[1]}
            '
             f'{assistant_attachments}'
-            f'{actions_html(history, i)}'
+            f'{actions_html(history, i, "assistant")}'
             f'
            '
             f'
            '
         )
@@ -564,11 +575,12 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
         if converted_visible[0]:  # Don't display empty user messages
             output += (
                 f'
            '
+                f'data-raw="{html.escape(row_internal[0], quote=True)}"'
+                f'data-index={i}>'
                 f'
            '
                 f'
            {converted_visible[0]}
            '
                 f'{user_attachments}'
-                f'
            {copy_button}{info_message_user}
            '
+                f'{actions_html(history, i, "user", info_message_user)}'
                 f'
            '
                 f'
            '
             )
@@ -580,7 +592,7 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
             f'
            '
             f'
            {converted_visible[1]}
            '
             f'{assistant_attachments}'
-            f'{actions_html(history, i, info_message_assistant)}'
+            f'{actions_html(history, i, "assistant", info_message_assistant)}'
             f'
            '
             f'
            '
         )
diff --git a/modules/ui.py b/modules/ui.py
index 52c095a271..00393b530a 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -212,7 +212,12 @@ def list_interface_input_elements():
         'negative_prompt',
         'dry_sequence_breakers',
         'grammar_string',
-        'branch_index'
+        'navigate_message_index',
+        'navigate_direction',
+        'edit_message_index',
+        'edit_message_text',
+        'edit_message_role',
+        'branch_index',
     ]
 
     # Chat elements
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 7a9f6f76ab..2856ce1f74 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -71,8 +71,6 @@ def create_ui():
                 shared.gradio['Remove last'] = gr.Button('Remove last reply (Ctrl + Shift + Backspace)', elem_id='Remove-last')
 
             with gr.Row():
-                shared.gradio['Replace last reply'] = gr.Button('Replace last reply (Ctrl + Shift + L)', elem_id='Replace-last')
-                shared.gradio['Copy last reply'] = gr.Button('Copy last reply (Ctrl + Shift + K)', elem_id='Copy-last')
                 shared.gradio['Impersonate'] = gr.Button('Impersonate (Ctrl + Shift + M)', elem_id='Impersonate')
 
             with gr.Row():
@@ -97,11 +95,15 @@ def create_ui():
                 with gr.Row():
                     shared.gradio['chat-instruct_command'] = gr.Textbox(value=shared.settings['chat-instruct_command'], lines=12, label='Command for chat-instruct mode', info='<|character|> and <|prompt|> get replaced with the bot name and the regular chat prompt respectively.', visible=shared.settings['mode'] == 'chat-instruct', elem_classes=['add_scrollbar'])
 
-        # Hidden elements for version navigation (similar to branch)
+        # Hidden elements for version navigation and editing
         with gr.Row(visible=False):
             shared.gradio['navigate_message_index'] = gr.Number(value=-1, precision=0, elem_id="Navigate-message-index")
             shared.gradio['navigate_direction'] = gr.Textbox(value="", elem_id="Navigate-direction")
             shared.gradio['navigate_version'] = gr.Button(elem_id="Navigate-version")
+            shared.gradio['edit_message_index'] = gr.Number(value=-1, precision=0, elem_id="Edit-message-index")
+            shared.gradio['edit_message_text'] = gr.Textbox(value="", elem_id="Edit-message-text")
+            shared.gradio['edit_message_role'] = gr.Textbox(value="", elem_id="Edit-message-role")
+            shared.gradio['edit_message'] = gr.Button(elem_id="Edit-message")
 
 
 def create_chat_settings_ui():
@@ -228,10 +230,6 @@ def create_event_handlers():
         None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
 
-    shared.gradio['Replace last reply'].click(
-        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        chat.handle_replace_last_reply_click, gradio('textbox', 'interface_state'), gradio('history', 'display', 'textbox'), show_progress=False)
-
     shared.gradio['Send dummy message'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_send_dummy_message_click, gradio('textbox', 'interface_state'), gradio('history', 'display', 'textbox'), show_progress=False)
@@ -297,12 +295,16 @@ def create_event_handlers():
         None, gradio('mode'), None, js="(mode) => {mode === 'instruct' ? document.getElementById('character-menu').parentNode.parentNode.style.display = 'none' : document.getElementById('character-menu').parentNode.parentNode.style.display = ''}")
 
     shared.gradio['chat_style'].change(chat.redraw_html, gradio(reload_arr), gradio('display'), show_progress=False)
-    shared.gradio['Copy last reply'].click(chat.send_last_reply_to_input, gradio('history'), gradio('textbox'), show_progress=False)
 
     shared.gradio['navigate_version'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_navigate_version_click, gradio('interface_state'), gradio('history', 'display'), show_progress=False)
 
+    shared.gradio['edit_message'].click(
+        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
+        chat.handle_edit_message_click, gradio('interface_state'), gradio('history', 'display', 'unique_id'), show_progress=False).then(
+        lambda: None, None, None, js='() => { const role = document.getElementById("Edit-message-role").querySelector("textarea").value; if (role === "user") document.getElementById("Regenerate").click(); }')
+
     # Save/delete a character
     shared.gradio['save_character'].click(chat.handle_save_character_click, gradio('name2'), gradio('save_character_filename', 'character_saver'), show_progress=False)
     shared.gradio['delete_character'].click(lambda: gr.update(visible=True), None, gradio('character_deleter'), show_progress=False)

From 2db36da979b539263deacbd3ac8b3f6dbba7f97f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 27 May 2025 21:00:11 -0700
Subject: [PATCH 0554/1701] UI: Make scrollbars more discrete in dark mode

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 9d68ba02a9..90dd51bc9c 100644
--- a/css/main.css
+++ b/css/main.css
@@ -265,7 +265,7 @@ button {
 
 .dark .pretty_scrollbar::-webkit-scrollbar-thumb,
 .dark .pretty_scrollbar::-webkit-scrollbar-thumb:hover {
-    background: #ccc;
+    background: rgba(255, 255, 255, 0.2);
     border-radius: 10px;
 }
 

From f6ca0ee0727bceac867d5a5bbea0c6d61fea35ea Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 27 May 2025 21:20:51 -0700
Subject: [PATCH 0555/1701] Fix regenerate sometimes not creating a new message
 version

---
 modules/chat.py | 24 +++++-------------------
 1 file changed, 5 insertions(+), 19 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 9598efa75d..59ca4d345a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -399,40 +399,26 @@ def get_stopping_strings(state):
 
 
 def add_message_version(history, row_idx, is_current=True):
-    """Add the current message as a version in the history metadata"""
-    if 'metadata' not in history:
-        history['metadata'] = {}
-
-    if row_idx >= len(history['internal']) or not history['internal'][row_idx][1].strip():
-        return  # Skip if row doesn't exist or message is empty
-
     key = f"assistant_{row_idx}"
-
-    # Initialize metadata structures if needed
     if key not in history['metadata']:
-        history['metadata'][key] = {"timestamp": get_current_timestamp()}
+        history['metadata'][key] = {}
+
     if "versions" not in history['metadata'][key]:
         history['metadata'][key]["versions"] = []
 
-    # Check if this version already exists
     current_content = history['internal'][row_idx][1]
     current_visible = history['visible'][row_idx][1]
 
-    for i, version in enumerate(history['metadata'][key]["versions"]):
-        if version['content'] == current_content and version['visible_content'] == current_visible:
-            if is_current:
-                history['metadata'][key]["current_version_index"] = i
-            return
-
-    # Add current message as a version
+    # Always add the current message as a new version entry.
+    # The timestamp will differentiate it even if content is identical to a previous version.
     history['metadata'][key]["versions"].append({
         "content": current_content,
         "visible_content": current_visible,
         "timestamp": get_current_timestamp()
     })
 
-    # Update index if this is the current version
     if is_current:
+        # Set the current_version_index to the newly added version (which is now the last one).
         history['metadata'][key]["current_version_index"] = len(history['metadata'][key]["versions"]) - 1
 
 

From 1b0e2d8750ee315086acb2738fab76ad28abadb8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 27 May 2025 22:36:24 -0700
Subject: [PATCH 0556/1701] UI: Add a token counter to the chat tab (counts
 input + history)

---
 css/main.css       |  7 ++++++
 modules/chat.py    | 54 +++++++++++++++++++++++++++++++++++++++++++++-
 modules/ui_chat.py |  9 ++++++++
 3 files changed, 69 insertions(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 90dd51bc9c..6e030453eb 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1542,3 +1542,10 @@ strong {
     opacity: 0.8;
     user-select: none;
 }
+
+.token-display {
+    font-family: monospace;
+    font-size: 13px;
+    color: var(--body-text-color-subdued);
+    margin-top: 4px;
+}
diff --git a/modules/chat.py b/modules/chat.py
index 59ca4d345a..498c0d883d 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -230,7 +230,15 @@ def generate_chat_prompt(user_input, state, **kwargs):
             messages.insert(insert_pos, {"role": "user", "content": enhanced_user_msg})
 
     user_input = user_input.strip()
-    if user_input and not impersonate and not _continue:
+
+    # Check if we have attachments even with empty input
+    has_attachments = False
+    if not impersonate and not _continue and len(history_data.get('metadata', {})) > 0:
+        current_row_idx = len(history)
+        user_key = f"user_{current_row_idx}"
+        has_attachments = user_key in metadata and "attachments" in metadata[user_key]
+
+    if (user_input or has_attachments) and not impersonate and not _continue:
         # For the current user input being processed, check if we need to add attachments
         if not impersonate and not _continue and len(history_data.get('metadata', {})) > 0:
             current_row_idx = len(history)
@@ -350,6 +358,50 @@ def make_prompt(messages):
         return prompt
 
 
+def count_prompt_tokens(text_input, state):
+    """Count tokens for current history + input including attachments"""
+    if shared.tokenizer is None:
+        return "Tokenizer not available"
+
+    try:
+        # Handle dict format with text and files
+        files = []
+        if isinstance(text_input, dict):
+            files = text_input.get('files', [])
+            text = text_input.get('text', '')
+        else:
+            text = text_input
+            files = []
+
+        # Create temporary history copy to add attachments
+        temp_history = copy.deepcopy(state['history'])
+        if 'metadata' not in temp_history:
+            temp_history['metadata'] = {}
+
+        # Process attachments if any
+        if files:
+            row_idx = len(temp_history['internal'])
+            for file_path in files:
+                add_message_attachment(temp_history, row_idx, file_path, is_user=True)
+
+        # Create temp state with modified history
+        temp_state = copy.deepcopy(state)
+        temp_state['history'] = temp_history
+
+        # Build prompt using existing logic
+        prompt = generate_chat_prompt(text, temp_state)
+        current_tokens = get_encoded_length(prompt)
+        max_tokens = temp_state['truncation_length']
+
+        percentage = (current_tokens / max_tokens) * 100 if max_tokens > 0 else 0
+
+        return f"History + Input:
            {current_tokens:,} / {max_tokens:,} tokens ({percentage:.1f}%)"
+
+    except Exception as e:
+        logger.error(f"Error counting tokens: {e}")
+        return f"Error: {str(e)}"
+
+
 def get_stopping_strings(state):
     stopping_strings = []
     renderers = []
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 2856ce1f74..952a40a575 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -95,6 +95,11 @@ def create_ui():
                 with gr.Row():
                     shared.gradio['chat-instruct_command'] = gr.Textbox(value=shared.settings['chat-instruct_command'], lines=12, label='Command for chat-instruct mode', info='<|character|> and <|prompt|> get replaced with the bot name and the regular chat prompt respectively.', visible=shared.settings['mode'] == 'chat-instruct', elem_classes=['add_scrollbar'])
 
+                with gr.Row():
+                    shared.gradio['count_tokens'] = gr.Button('Count tokens', size='sm')
+
+                shared.gradio['token_display'] = gr.HTML(value='', elem_classes='token-display')
+
         # Hidden elements for version navigation and editing
         with gr.Row(visible=False):
             shared.gradio['navigate_message_index'] = gr.Number(value=-1, precision=0, elem_id="Navigate-message-index")
@@ -360,3 +365,7 @@ def create_event_handlers():
         None, None, None, js=f'() => {{{ui.switch_tabs_js}; switch_to_notebook()}}')
 
     shared.gradio['show_controls'].change(None, gradio('show_controls'), None, js=f'(x) => {{{ui.show_controls_js}; toggle_controls(x)}}')
+
+    shared.gradio['count_tokens'].click(
+        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
+        chat.count_prompt_tokens, gradio('textbox', 'interface_state'), gradio('token_display'), show_progress=False)

From 077bbc6b101f8f6045b95369bc82373187741d12 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Wed, 28 May 2025 04:27:28 -0300
Subject: [PATCH 0557/1701] Add web search support (#7023)

---
 modules/chat.py                               |   4 +
 modules/ui.py                                 |   6 +-
 modules/ui_chat.py                            |  12 ++
 modules/web_search.py                         | 125 ++++++++++++++++++
 requirements/full/requirements.txt            |   2 +
 requirements/full/requirements_amd.txt        |   2 +
 requirements/full/requirements_amd_noavx2.txt |   2 +
 .../full/requirements_apple_intel.txt         |   2 +
 .../full/requirements_apple_silicon.txt       |   2 +
 requirements/full/requirements_cpu_only.txt   |   2 +
 .../full/requirements_cpu_only_noavx2.txt     |   2 +
 requirements/full/requirements_noavx2.txt     |   2 +
 requirements/full/requirements_nowheels.txt   |   2 +
 requirements/portable/requirements.txt        |   2 +
 .../portable/requirements_apple_intel.txt     |   2 +
 .../portable/requirements_apple_silicon.txt   |   2 +
 .../portable/requirements_cpu_only.txt        |   2 +
 .../portable/requirements_cpu_only_noavx2.txt |   2 +
 requirements/portable/requirements_noavx2.txt |   2 +
 .../portable/requirements_nowheels.txt        |   2 +
 requirements/portable/requirements_vulkan.txt |   2 +
 .../portable/requirements_vulkan_noavx2.txt   |   2 +
 22 files changed, 181 insertions(+), 2 deletions(-)
 create mode 100644 modules/web_search.py

diff --git a/modules/chat.py b/modules/chat.py
index 498c0d883d..b2aacd5cc5 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -31,6 +31,7 @@
     get_max_prompt_length
 )
 from modules.utils import delete_file, get_available_characters, save_file
+from modules.web_search import add_web_search_attachments
 
 
 def strftime_now(format):
@@ -566,6 +567,9 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
         for file_path in files:
             add_message_attachment(output, row_idx, file_path, is_user=True)
 
+        # Add web search results as attachments if enabled
+        add_web_search_attachments(output, row_idx, text, state)
+
         # Apply extensions
         text, visible_text = apply_extensions('chat_input', text, visible_text, state)
         text = apply_extensions('input', text, state, is_chat=True)
diff --git a/modules/ui.py b/modules/ui.py
index 00393b530a..e24e6402a1 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -157,8 +157,6 @@ def list_model_elements():
 
 def list_interface_input_elements():
     elements = [
-        'navigate_message_index',
-        'navigate_direction',
         'temperature',
         'dynatemp_low',
         'dynatemp_high',
@@ -218,6 +216,10 @@ def list_interface_input_elements():
         'edit_message_text',
         'edit_message_role',
         'branch_index',
+        'enable_web_search',
+        'web_search_pages',
+        'navigate_message_index',
+        'navigate_direction',
     ]
 
     # Chat elements
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 952a40a575..719af85ae7 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -86,6 +86,12 @@ def create_ui():
                 with gr.Row():
                     shared.gradio['start_with'] = gr.Textbox(label='Start reply with', placeholder='Sure thing!', value=shared.settings['start_with'], elem_classes=['add_scrollbar'])
 
+                with gr.Row():
+                    shared.gradio['enable_web_search'] = gr.Checkbox(value=shared.settings.get('enable_web_search', False), label='Activate web search')
+
+                with gr.Row(visible=shared.settings.get('enable_web_search', False)) as shared.gradio['web_search_row']:
+                    shared.gradio['web_search_pages'] = gr.Number(value=shared.settings.get('web_search_pages', 3), precision=0, label='Number of pages to download', minimum=1, maximum=10)
+
                 with gr.Row():
                     shared.gradio['mode'] = gr.Radio(choices=['instruct', 'chat-instruct', 'chat'], value=shared.settings['mode'] if shared.settings['mode'] in ['chat', 'chat-instruct'] else None, label='Mode', info='Defines how the chat prompt is generated. In instruct and chat-instruct modes, the instruction template Parameters > Instruction template is used.', elem_id='chat-mode')
 
@@ -369,3 +375,9 @@ def create_event_handlers():
     shared.gradio['count_tokens'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.count_prompt_tokens, gradio('textbox', 'interface_state'), gradio('token_display'), show_progress=False)
+
+    shared.gradio['enable_web_search'].change(
+        lambda x: gr.update(visible=x),
+        gradio('enable_web_search'),
+        gradio('web_search_row')
+    )
diff --git a/modules/web_search.py b/modules/web_search.py
new file mode 100644
index 0000000000..e7688ba490
--- /dev/null
+++ b/modules/web_search.py
@@ -0,0 +1,125 @@
+from datetime import datetime
+
+import requests
+from bs4 import BeautifulSoup
+from duckduckgo_search import DDGS
+
+from modules.logging_colors import logger
+from modules.text_generation import generate_reply
+
+
+def get_current_timestamp():
+    """Returns the current time in 24-hour format"""
+    return datetime.now().strftime('%b %d, %Y %H:%M')
+
+
+def generate_search_query(user_message, state):
+    """Generate a search query from user message using the LLM"""
+    search_prompt = f"{user_message}\n\n=====\n\nPlease turn the message above into a short web search query in the same language as the message. Respond with only the search query, nothing else."
+
+    # Use a minimal state for search query generation
+    search_state = state.copy()
+    search_state['max_new_tokens'] = 64
+    search_state['temperature'] = 0.1
+
+    query = ""
+    for reply in generate_reply(search_prompt, search_state, stopping_strings=[], is_chat=False):
+        query = reply.strip()
+
+    return query
+
+
+def download_web_page(url, timeout=10):
+    """Download and extract text from a web page"""
+    try:
+        headers = {
+            'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'
+        }
+        response = requests.get(url, headers=headers, timeout=timeout)
+        response.raise_for_status()
+
+        soup = BeautifulSoup(response.content, 'html.parser')
+
+        # Remove script and style elements
+        for script in soup(["script", "style"]):
+            script.decompose()
+
+        # Get text and clean it up
+        text = soup.get_text()
+        lines = (line.strip() for line in text.splitlines())
+        chunks = (phrase.strip() for line in lines for phrase in line.split("  "))
+        text = ' '.join(chunk for chunk in chunks if chunk)
+
+        return text
+    except Exception as e:
+        logger.error(f"Error downloading {url}: {e}")
+        return f"[Error downloading content from {url}: {str(e)}]"
+
+
+def perform_web_search(query, num_pages=3):
+    """Perform web search and return results with content"""
+    try:
+        with DDGS() as ddgs:
+            results = list(ddgs.text(query, max_results=num_pages))
+
+        search_results = []
+        for i, result in enumerate(results):
+            url = result.get('href', '')
+            title = result.get('title', f'Search Result {i+1}')
+
+            # Download page content
+            content = download_web_page(url)
+
+            search_results.append({
+                'title': title,
+                'url': url,
+                'content': content
+            })
+
+        return search_results
+    except Exception as e:
+        logger.error(f"Error performing web search: {e}")
+        return []
+
+
+def add_web_search_attachments(history, row_idx, user_message, state):
+    """Perform web search and add results as attachments"""
+    if not state.get('enable_web_search', False):
+        return
+
+    try:
+        # Generate search query
+        search_query = generate_search_query(user_message, state)
+        if not search_query:
+            logger.warning("Failed to generate search query")
+            return
+
+        logger.info(f"Generated search query: {search_query}")
+
+        # Perform web search
+        num_pages = int(state.get('web_search_pages', 3))
+        search_results = perform_web_search(search_query, num_pages)
+
+        if not search_results:
+            logger.warning("No search results found")
+            return
+
+        # Add search results as attachments
+        key = f"user_{row_idx}"
+        if key not in history['metadata']:
+            history['metadata'][key] = {"timestamp": get_current_timestamp()}
+        if "attachments" not in history['metadata'][key]:
+            history['metadata'][key]["attachments"] = []
+
+        for result in search_results:
+            attachment = {
+                "name": f"{result['title']}",
+                "type": "text/html",
+                "content": f"URL: {result['url']}\n\n{result['content']}"
+            }
+            history['metadata'][key]["attachments"].append(attachment)
+
+        logger.info(f"Added {len(search_results)} web search results as attachments")
+
+    except Exception as e:
+        logger.error(f"Error in web search: {e}")
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 3d18f5fd04..0eaf10da6a 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -1,7 +1,9 @@
 accelerate==1.5.*
+beautifulsoup4==4.13.4
 bitsandbytes==0.45.*
 colorama
 datasets
+duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 82b199645a..65f184bfa6 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -1,6 +1,8 @@
 accelerate==1.5.*
+beautifulsoup4==4.13.4
 colorama
 datasets
+duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index a8b0301454..d20b2ec3fb 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -1,6 +1,8 @@
 accelerate==1.5.*
+beautifulsoup4==4.13.4
 colorama
 datasets
+duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 5a61ac7dc1..2613d78765 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -1,6 +1,8 @@
 accelerate==1.5.*
+beautifulsoup4==4.13.4
 colorama
 datasets
+duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 6862c3b466..af583b0070 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -1,6 +1,8 @@
 accelerate==1.5.*
+beautifulsoup4==4.13.4
 colorama
 datasets
+duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index e698277993..9bf2a37dd2 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -1,6 +1,8 @@
 accelerate==1.5.*
+beautifulsoup4==4.13.4
 colorama
 datasets
+duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 97bff786ba..1731448ea9 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -1,6 +1,8 @@
 accelerate==1.5.*
+beautifulsoup4==4.13.4
 colorama
 datasets
+duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 17c7e24624..fc481a1a71 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -1,7 +1,9 @@
 accelerate==1.5.*
+beautifulsoup4==4.13.4
 bitsandbytes==0.45.*
 colorama
 datasets
+duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 89b32caf58..2ed8affa47 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -1,6 +1,8 @@
 accelerate==1.5.*
+beautifulsoup4==4.13.4
 colorama
 datasets
+duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index ec9bafc629..fdae681df4 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -1,3 +1,5 @@
+beautifulsoup4==4.13.4
+duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 025a737eb0..a58f39f7f9 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -1,3 +1,5 @@
+beautifulsoup4==4.13.4
+duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 32644e87e6..91ea3a6df8 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -1,3 +1,5 @@
+beautifulsoup4==4.13.4
+duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index bd5c1d9bdc..37e5aa40b4 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -1,3 +1,5 @@
+beautifulsoup4==4.13.4
+duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 51f2b7d9c7..dcb2884ba9 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -1,3 +1,5 @@
+beautifulsoup4==4.13.4
+duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index aad6bf5a8d..8f1295bbac 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -1,3 +1,5 @@
+beautifulsoup4==4.13.4
+duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 4c05542622..21805fe241 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -1,3 +1,5 @@
+beautifulsoup4==4.13.4
+duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 3d98d1b035..858b4488d7 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -1,3 +1,5 @@
+beautifulsoup4==4.13.4
+duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index f954b8d248..569bae9962 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -1,3 +1,5 @@
+beautifulsoup4==4.13.4
+duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 jinja2==3.1.6

From 75c6ae8502cae60bd8dabef1e2af4aec5766ca35 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 28 May 2025 00:29:17 -0700
Subject: [PATCH 0558/1701] UI: Don't edit messages on double click

---
 js/main.js | 10 ----------
 1 file changed, 10 deletions(-)

diff --git a/js/main.js b/js/main.js
index fc014f66ac..48bb863201 100644
--- a/js/main.js
+++ b/js/main.js
@@ -380,16 +380,6 @@ document.addEventListener("click", function (event) {
   }
 });
 
-document.addEventListener("dblclick", (event) => {
-  const messageElement = event.target.closest(".message, .user-message, .assistant-message");
-  if (!messageElement) return;
-
-  const editButton = messageElement.querySelector(".footer-edit-button");
-  if (editButton) {
-    editButton.click();
-  }
-});
-
 //------------------------------------------------
 // Relocate the "Show controls" checkbox
 //------------------------------------------------

From 0aedb8992165b386dac244baeb5fb5967513869e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 28 May 2025 00:35:20 -0700
Subject: [PATCH 0559/1701] UI: Small style improvement to attachments

---
 css/main.css | 1 +
 1 file changed, 1 insertion(+)

diff --git a/css/main.css b/css/main.css
index 6e030453eb..181a19b8fa 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1417,6 +1417,7 @@ strong {
     flex-wrap: wrap;
     gap: 8px;
     margin-top: 8px;
+    padding-bottom: 6px;
 }
 
 .attachment-box {

From 6c3590ba9ab0bd540097a50986a59f0099d11d92 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 28 May 2025 05:28:15 -0700
Subject: [PATCH 0560/1701] Make web search attachments clickable

---
 modules/html_generator.py | 8 +++++++-
 modules/web_search.py     | 5 +++--
 2 files changed, 10 insertions(+), 3 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 9a93555f9c..bfb278cdf1 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -370,10 +370,16 @@ def format_message_attachments(history, role, index):
 
         attachments_html = '
            '
         for attachment in attachments:
+            name = html.escape(attachment["name"])
+
+            # Make clickable if URL exists
+            if "url" in attachment:
+                name = f'{name}'
+
             attachments_html += (
                 f'
            '
                 f'
            {attachment_svg}
            '
-                f'
            {html.escape(attachment["name"])}
            '
+                f'
            {name}
            '
                 f'
            '
             )
         attachments_html += '
            '
diff --git a/modules/web_search.py b/modules/web_search.py
index e7688ba490..d3387ac9a8 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -113,9 +113,10 @@ def add_web_search_attachments(history, row_idx, user_message, state):
 
         for result in search_results:
             attachment = {
-                "name": f"{result['title']}",
+                "name": result['title'],
                 "type": "text/html",
-                "content": f"URL: {result['url']}\n\n{result['content']}"
+                "url": result['url'],
+                "content": result['content']
             }
             history['metadata'][key]["attachments"].append(attachment)
 

From 27641ac1823751165615a1a53b62ae24977e37a0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 28 May 2025 17:09:05 -0700
Subject: [PATCH 0561/1701] UI: Make message editing work the same for user and
 assistant messages

---
 js/global_scope_js.js     | 28 ++++++------
 modules/chat.py           | 94 ++++++++++++++++++++-------------------
 modules/html_generator.py | 42 ++++++++++-------
 modules/ui.py             |  3 +-
 modules/ui_chat.py        |  4 +-
 5 files changed, 94 insertions(+), 77 deletions(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 0e86d450fe..3274f47eea 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -186,31 +186,33 @@ function navigateVersion(element, direction) {
   const index = messageElement.getAttribute("data-index");
   if (!index) return;
 
-  const indexInput = document.getElementById("Navigate-message-index").querySelector("input");
-  if (!indexInput) {
-    console.error("Element with ID 'Navigate-message-index' not found.");
-    return;
-  }
-
-  const directionInput = document.getElementById("Navigate-direction").querySelector("textarea");
-  if (!directionInput) {
-    console.error("Element with ID 'Navigate-direction' not found.");
-    return;
+  // Determine role based on message element classes
+  let role = "assistant"; // Default role
+  if (messageElement.classList.contains("user-message") ||
+      messageElement.querySelector(".text-you") ||
+      messageElement.querySelector(".circle-you")) {
+    role = "user";
   }
 
+  const indexInput = document.getElementById("Navigate-message-index")?.querySelector("input");
+  const directionInput = document.getElementById("Navigate-direction")?.querySelector("textarea");
+  const roleInput = document.getElementById("Navigate-message-role")?.querySelector("textarea");
   const navigateButton = document.getElementById("Navigate-version");
-  if (!navigateButton) {
-    console.error("Required element 'Navigate-version' not found.");
+
+  if (!indexInput || !directionInput || !roleInput || !navigateButton) {
+    console.error("Navigation control elements (index, direction, role, or button) not found.");
     return;
   }
 
   indexInput.value = index;
   directionInput.value = direction;
+  roleInput.value = role;
 
-  // Trigger any 'change' or 'input' events Gradio might be listening for
+  // Trigger 'input' events for Gradio to pick up changes
   const event = new Event("input", { bubbles: true });
   indexInput.dispatchEvent(event);
   directionInput.dispatchEvent(event);
+  roleInput.dispatchEvent(event);
 
   navigateButton.click();
 }
diff --git a/modules/chat.py b/modules/chat.py
index b2aacd5cc5..8bac680c3e 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -451,19 +451,21 @@ def get_stopping_strings(state):
     return result
 
 
-def add_message_version(history, row_idx, is_current=True):
-    key = f"assistant_{row_idx}"
+def add_message_version(history, role, row_idx, is_current=True):
+    key = f"{role}_{row_idx}"
+    if 'metadata' not in history:
+        history['metadata'] = {}
     if key not in history['metadata']:
         history['metadata'][key] = {}
 
     if "versions" not in history['metadata'][key]:
         history['metadata'][key]["versions"] = []
 
-    current_content = history['internal'][row_idx][1]
-    current_visible = history['visible'][row_idx][1]
+    # Determine which index to use for content based on role
+    content_idx = 0 if role == 'user' else 1
+    current_content = history['internal'][row_idx][content_idx]
+    current_visible = history['visible'][row_idx][content_idx]
 
-    # Always add the current message as a new version entry.
-    # The timestamp will differentiate it even if content is identical to a previous version.
     history['metadata'][key]["versions"].append({
         "content": current_content,
         "visible_content": current_visible,
@@ -594,7 +596,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
 
             # Store the first response as a version before regenerating
             if not output['metadata'].get(f"assistant_{row_idx}", {}).get('versions'):
-                add_message_version(output, row_idx, is_current=False)
+                add_message_version(output, "assistant", row_idx, is_current=False)
 
             if loading_message:
                 yield {
@@ -656,12 +658,13 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             if is_stream:
                 yield output
 
+    output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
+
     # Add the newly generated response as a version (only for regeneration)
     if regenerate:
         row_idx = len(output['internal']) - 1
-        add_message_version(output, row_idx, is_current=True)
+        add_message_version(output, "assistant", row_idx, is_current=True)
 
-    output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
     yield output
 
 
@@ -1441,37 +1444,35 @@ def handle_edit_message_click(state):
 
     if message_index >= len(history['internal']):
         html_output = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
-        return [history, html_output, gr.update()]
+        return [history, html_output, gr.update()]  # No unique_id change
+
+    role_idx = 0 if role == "user" else 1
+
+    if 'metadata' not in history:
+        history['metadata'] = {}
 
-    # Use the role passed from frontend
-    is_user_msg = (role == "user")
-    role_idx = 0 if is_user_msg else 1
+    key = f"{role}_{message_index}"
+    if key not in history['metadata']:
+        history['metadata'][key] = {}
+
+    # If no versions exist yet for this message, store the current (pre-edit) content as the first version.
+    if "versions" not in history['metadata'][key] or not history['metadata'][key]["versions"]:
+        original_content = history['internal'][message_index][role_idx]
+        original_visible = history['visible'][message_index][role_idx]
 
-    # For assistant messages, save the original version BEFORE updating content
-    if not is_user_msg:
-        if not history['metadata'].get(f"assistant_{message_index}", {}).get('versions'):
-            add_message_version(history, message_index, is_current=False)
+        history['metadata'][key]["versions"] = [{
+            "content": original_content,
+            "visible_content": original_visible,
+            "timestamp": get_current_timestamp()
+        }]
 
-    # NOW update the message content
     history['internal'][message_index][role_idx] = apply_extensions('input', new_text, state, is_chat=True)
     history['visible'][message_index][role_idx] = html.escape(new_text)
 
-    # Branch if editing user message, add version if editing assistant message
-    if is_user_msg:
-        # Branch like branch-here
-        history['visible'] = history['visible'][:message_index + 1]
-        history['internal'] = history['internal'][:message_index + 1]
-        new_unique_id = datetime.now().strftime('%Y%m%d-%H-%M-%S')
-        save_history(history, new_unique_id, state['character_menu'], state['mode'])
-        histories = find_all_histories_with_first_prompts(state)
-        past_chats_update = gr.update(choices=histories, value=new_unique_id)
-        state['unique_id'] = new_unique_id
-    elif not is_user_msg:
-        # Add the new version as current
-        add_message_version(history, message_index, is_current=True)
-        past_chats_update = gr.update()
-    else:
-        past_chats_update = gr.update()
+    add_message_version(history, role, message_index, is_current=True)
+
+    # Since we are not branching, unique_id does not change.
+    past_chats_update = gr.update()
 
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
     html_output = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
@@ -1483,33 +1484,36 @@ def handle_navigate_version_click(state):
     history = state['history']
     message_index = int(state['navigate_message_index'])
     direction = state['navigate_direction']
+    role = state['navigate_message_role']
+
+    if not role:
+        logger.error("Role not provided for version navigation.")
+        html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
+        return [history, html]
 
-    # Get assistant message metadata
-    key = f"assistant_{message_index}"
-    if key not in history['metadata'] or 'versions' not in history['metadata'][key]:
-        # No versions to navigate
+    key = f"{role}_{message_index}"
+    if 'metadata' not in history or key not in history['metadata'] or 'versions' not in history['metadata'][key]:
         html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
         return [history, html]
 
     metadata = history['metadata'][key]
-    current_idx = metadata.get('current_version_index', 0)
     versions = metadata['versions']
+    # Default to the last version if current_version_index is not set
+    current_idx = metadata.get('current_version_index', len(versions) - 1 if versions else 0)
 
-    # Calculate new index
     if direction == 'left':
         new_idx = max(0, current_idx - 1)
     else:  # right
         new_idx = min(len(versions) - 1, current_idx + 1)
 
     if new_idx == current_idx:
-        # No change needed
         html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
         return [history, html]
 
-    # Update history with new version
-    version = versions[new_idx]
-    history['internal'][message_index][1] = version['content']
-    history['visible'][message_index][1] = version['visible_content']
+    msg_content_idx = 0 if role == 'user' else 1  # 0 for user content, 1 for assistant content in the pair
+    version_to_load = versions[new_idx]
+    history['internal'][message_index][msg_content_idx] = version_to_load['content']
+    history['visible'][message_index][msg_content_idx] = version_to_load['visible_content']
     metadata['current_version_index'] = new_idx
 
     # Redraw and save
diff --git a/modules/html_generator.py b/modules/html_generator.py
index bfb278cdf1..cbf3e19ca2 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -388,16 +388,17 @@ def format_message_attachments(history, role, index):
     return ""
 
 
-def get_version_navigation_html(history, i):
+def get_version_navigation_html(history, i, role):
     """Generate simple navigation arrows for message versions"""
-    key = f"assistant_{i}"
+    key = f"{role}_{i}"
     metadata = history.get('metadata', {})
 
     if key not in metadata or 'versions' not in metadata[key]:
         return ""
 
     versions = metadata[key]['versions']
-    current_idx = metadata[key].get('current_version_index', 0)
+    # Default to the last version if current_version_index isn't set in metadata
+    current_idx = metadata[key].get('current_version_index', len(versions) - 1 if versions else 0)
 
     if len(versions) <= 1:
         return ""
@@ -413,22 +414,33 @@ def get_version_navigation_html(history, i):
 
 
 def actions_html(history, i, role, info_message=""):
+    action_buttons = ""
+    version_nav_html = ""
+
     if role == "assistant":
-        return (f'
            '
-                f'{copy_button}'
-                f'{edit_button}'
-                f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
-                f'{continue_button if i == len(history["visible"]) - 1 else ""}'
-                f'{remove_button if i == len(history["visible"]) - 1 else ""}'
-                f'{branch_button}'
-                f'{info_message}'
-                f'
            '
-                f'{get_version_navigation_html(history, i)}')
-    return (f'
            '
+        action_buttons = (
+            f'{copy_button}'
+            f'{edit_button}'
+            f'{refresh_button if i == len(history["visible"]) - 1 else ""}'
+            f'{continue_button if i == len(history["visible"]) - 1 else ""}'
+            f'{remove_button if i == len(history["visible"]) - 1 else ""}'
+            f'{branch_button}'
+        )
+
+        version_nav_html = get_version_navigation_html(history, i, "assistant")
+    elif role == "user":
+        action_buttons = (
             f'{copy_button}'
             f'{edit_button}'
+        )
+
+        version_nav_html = get_version_navigation_html(history, i, "user")
+
+    return (f'
            '
+            f'{action_buttons}'
             f'{info_message}'
-            f'
            ')
+            f'
            '
+            f'{version_nav_html}')
 
 
 def generate_instruct_html(history):
diff --git a/modules/ui.py b/modules/ui.py
index e24e6402a1..a2662e1403 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -212,14 +212,13 @@ def list_interface_input_elements():
         'grammar_string',
         'navigate_message_index',
         'navigate_direction',
+        'navigate_message_role',
         'edit_message_index',
         'edit_message_text',
         'edit_message_role',
         'branch_index',
         'enable_web_search',
         'web_search_pages',
-        'navigate_message_index',
-        'navigate_direction',
     ]
 
     # Chat elements
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 719af85ae7..df3d3929b7 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -110,6 +110,7 @@ def create_ui():
         with gr.Row(visible=False):
             shared.gradio['navigate_message_index'] = gr.Number(value=-1, precision=0, elem_id="Navigate-message-index")
             shared.gradio['navigate_direction'] = gr.Textbox(value="", elem_id="Navigate-direction")
+            shared.gradio['navigate_message_role'] = gr.Textbox(value="", elem_id="Navigate-message-role")
             shared.gradio['navigate_version'] = gr.Button(elem_id="Navigate-version")
             shared.gradio['edit_message_index'] = gr.Number(value=-1, precision=0, elem_id="Edit-message-index")
             shared.gradio['edit_message_text'] = gr.Textbox(value="", elem_id="Edit-message-text")
@@ -313,8 +314,7 @@ def create_event_handlers():
 
     shared.gradio['edit_message'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        chat.handle_edit_message_click, gradio('interface_state'), gradio('history', 'display', 'unique_id'), show_progress=False).then(
-        lambda: None, None, None, js='() => { const role = document.getElementById("Edit-message-role").querySelector("textarea").value; if (role === "user") document.getElementById("Regenerate").click(); }')
+        chat.handle_edit_message_click, gradio('interface_state'), gradio('history', 'display', 'unique_id'), show_progress=False)
 
     # Save/delete a character
     shared.gradio['save_character'].click(chat.handle_save_character_click, gradio('name2'), gradio('save_character_filename', 'character_saver'), show_progress=False)

From 3eb0b77427ad7b87c128999fd915f97b22104819 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 28 May 2025 18:14:51 -0700
Subject: [PATCH 0562/1701] Improve the web search query generation

---
 modules/chat.py       | 25 ++++++++++++++++++++++++-
 modules/web_search.py | 29 ++++-------------------------
 2 files changed, 28 insertions(+), 26 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 8bac680c3e..495fe93403 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -538,6 +538,27 @@ def extract_pdf_text(pdf_path):
         return f"[Error extracting PDF text: {str(e)}]"
 
 
+def generate_search_query(user_message, state):
+    """Generate a search query from user message using the LLM"""
+    # Augment the user message with search instruction
+    augmented_message = f"{user_message}\n\n=====\n\nPlease turn the message above into a short web search query in the same language as the message. Respond with only the search query, nothing else."
+
+    # Use a minimal state for search query generation but keep the full history
+    search_state = state.copy()
+    search_state['max_new_tokens'] = 64
+    search_state['auto_max_new_tokens'] = False
+    search_state['enable_thinking'] = False
+
+    # Generate the full prompt using existing history + augmented message
+    formatted_prompt = generate_chat_prompt(augmented_message, search_state)
+
+    query = ""
+    for reply in generate_reply(formatted_prompt, search_state, stopping_strings=[], is_chat=True):
+        query = reply.strip()
+
+    return query
+
+
 def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_message=True, for_ui=False):
     # Handle dict format with text and files
     files = []
@@ -570,7 +591,9 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             add_message_attachment(output, row_idx, file_path, is_user=True)
 
         # Add web search results as attachments if enabled
-        add_web_search_attachments(output, row_idx, text, state)
+        if state.get('enable_web_search', False):
+            search_query = generate_search_query(text, state)
+            add_web_search_attachments(output, row_idx, text, search_query, state)
 
         # Apply extensions
         text, visible_text = apply_extensions('chat_input', text, visible_text, state)
diff --git a/modules/web_search.py b/modules/web_search.py
index d3387ac9a8..667178c5c3 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -13,22 +13,6 @@ def get_current_timestamp():
     return datetime.now().strftime('%b %d, %Y %H:%M')
 
 
-def generate_search_query(user_message, state):
-    """Generate a search query from user message using the LLM"""
-    search_prompt = f"{user_message}\n\n=====\n\nPlease turn the message above into a short web search query in the same language as the message. Respond with only the search query, nothing else."
-
-    # Use a minimal state for search query generation
-    search_state = state.copy()
-    search_state['max_new_tokens'] = 64
-    search_state['temperature'] = 0.1
-
-    query = ""
-    for reply in generate_reply(search_prompt, search_state, stopping_strings=[], is_chat=False):
-        query = reply.strip()
-
-    return query
-
-
 def download_web_page(url, timeout=10):
     """Download and extract text from a web page"""
     try:
@@ -82,19 +66,14 @@ def perform_web_search(query, num_pages=3):
         return []
 
 
-def add_web_search_attachments(history, row_idx, user_message, state):
+def add_web_search_attachments(history, row_idx, user_message, search_query, state):
     """Perform web search and add results as attachments"""
-    if not state.get('enable_web_search', False):
+    if not search_query:
+        logger.warning("No search query provided")
         return
 
     try:
-        # Generate search query
-        search_query = generate_search_query(user_message, state)
-        if not search_query:
-            logger.warning("Failed to generate search query")
-            return
-
-        logger.info(f"Generated search query: {search_query}")
+        logger.info(f"Using search query: {search_query}")
 
         # Perform web search
         num_pages = int(state.get('web_search_pages', 3))

From 7080a02252b9949297950ef3669361d21f4a6bcf Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 28 May 2025 18:15:21 -0700
Subject: [PATCH 0563/1701] Reduce the timeout for downloading web pages

---
 modules/web_search.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/web_search.py b/modules/web_search.py
index 667178c5c3..070f850c87 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -13,7 +13,7 @@ def get_current_timestamp():
     return datetime.now().strftime('%b %d, %Y %H:%M')
 
 
-def download_web_page(url, timeout=10):
+def download_web_page(url, timeout=5):
     """Download and extract text from a web page"""
     try:
         headers = {

From 75d6cfd14d1aed5ba19bd747479794cbd34212d0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 28 May 2025 20:34:14 -0700
Subject: [PATCH 0564/1701] Download fetched web search results in parallel

---
 modules/web_search.py | 44 +++++++++++++++++++++++++++++++++----------
 1 file changed, 34 insertions(+), 10 deletions(-)

diff --git a/modules/web_search.py b/modules/web_search.py
index 070f850c87..1f67034968 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -1,3 +1,5 @@
+import concurrent.futures
+from concurrent.futures import as_completed
 from datetime import datetime
 
 import requests
@@ -5,7 +7,6 @@
 from duckduckgo_search import DDGS
 
 from modules.logging_colors import logger
-from modules.text_generation import generate_reply
 
 
 def get_current_timestamp():
@@ -40,27 +41,50 @@ def download_web_page(url, timeout=5):
         return f"[Error downloading content from {url}: {str(e)}]"
 
 
-def perform_web_search(query, num_pages=3):
+def perform_web_search(query, num_pages=3, max_workers=5):
     """Perform web search and return results with content"""
     try:
         with DDGS() as ddgs:
             results = list(ddgs.text(query, max_results=num_pages))
 
-        search_results = []
+        # Prepare download tasks
+        download_tasks = []
         for i, result in enumerate(results):
             url = result.get('href', '')
             title = result.get('title', f'Search Result {i+1}')
+            download_tasks.append((url, title, i))
 
-            # Download page content
-            content = download_web_page(url)
+        search_results = [None] * len(download_tasks)  # Pre-allocate to maintain order
 
-            search_results.append({
-                'title': title,
-                'url': url,
-                'content': content
-            })
+        # Download pages in parallel
+        with concurrent.futures.ThreadPoolExecutor(max_workers=max_workers) as executor:
+            # Submit all download tasks
+            future_to_task = {
+                executor.submit(download_web_page, task[0]): task
+                for task in download_tasks
+            }
+
+            # Collect results as they complete
+            for future in as_completed(future_to_task):
+                url, title, index = future_to_task[future]
+                try:
+                    content = future.result()
+                    search_results[index] = {
+                        'title': title,
+                        'url': url,
+                        'content': content
+                    }
+                except Exception as e:
+                    logger.error(f"Error downloading {url}: {e}")
+                    # Include failed downloads with empty content
+                    search_results[index] = {
+                        'title': title,
+                        'url': url,
+                        'content': ''
+                    }
 
         return search_results
+
     except Exception as e:
         logger.error(f"Error performing web search: {e}")
         return []

From 63234b9b6f60ec4f276480b4e7f9d4cd1395dcaf Mon Sep 17 00:00:00 2001
From: Underscore <47636331+Th-Underscore@users.noreply.github.com>
Date: Thu, 29 May 2025 07:22:03 -0400
Subject: [PATCH 0565/1701] UI: Fix impersonate (#7025)

---
 modules/chat.py | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 495fe93403..7afd906de1 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -691,16 +691,19 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
     yield output
 
 
-def impersonate_wrapper(text, state):
+def impersonate_wrapper(textbox, state):
+    text = textbox['text']
     static_output = chat_html_wrapper(state['history'], state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
 
     prompt = generate_chat_prompt('', state, impersonate=True)
     stopping_strings = get_stopping_strings(state)
 
-    yield text + '...', static_output
+    textbox['text'] = text + '...'
+    yield textbox, static_output
     reply = None
     for reply in generate_reply(prompt + text, state, stopping_strings=stopping_strings, is_chat=True):
-        yield (text + reply).lstrip(' '), static_output
+        textbox['text'] = (text + reply).lstrip(' ')
+        yield textbox, static_output
         if shared.stop_everything:
             return
 

From a8d02dec8f5e6a054a153b3b09425b51e090ae11 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 04:24:21 -0700
Subject: [PATCH 0566/1701] Bump llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 0eaf10da6a..5f61aff9b9 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 65f184bfa6..a718b6cab4 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -32,7 +32,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index d20b2ec3fb..5fddc62309 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -32,7 +32,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 2613d78765..8e0144451d 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -32,7 +32,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index af583b0070..77779f3dc2 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -32,8 +32,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 9bf2a37dd2..79efc607c5 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -32,5 +32,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 1731448ea9..8b29453ec1 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -32,5 +32,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index fc481a1a71..f1f4a02e4c 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index fdae681df4..adf50d9af8 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index a58f39f7f9..46b3679164 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 91ea3a6df8..660527113b 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -18,6 +18,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 37e5aa40b4..4013abcc7a 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index dcb2884ba9..41808854bf 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 8f1295bbac..cff79ec600 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 858b4488d7..762b3fa347 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 569bae9962..b425d30525 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 685cfe254036111711de027f6d3a8198d02e7545 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 04:26:43 -0700
Subject: [PATCH 0567/1701] Lint

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 181a19b8fa..8af87b42f7 100644
--- a/css/main.css
+++ b/css/main.css
@@ -265,7 +265,7 @@ button {
 
 .dark .pretty_scrollbar::-webkit-scrollbar-thumb,
 .dark .pretty_scrollbar::-webkit-scrollbar-thumb:hover {
-    background: rgba(255, 255, 255, 0.2);
+    background: rgb(255 255 255 / 20%);
     border-radius: 10px;
 }
 

From f2ee917d4f600ebbc5fa9d5fcf65cf5feef27fc1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 04:55:05 -0700
Subject: [PATCH 0568/1701] Update README

---
 README.md | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index 7105ce23ba..afb21cb07e 100644
--- a/README.md
+++ b/README.md
@@ -14,14 +14,17 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 
 - Supports multiple text generation backends in one UI/API, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).
 - Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory that doesn't interfere with your system environment.
+- **File attachments**: Upload text files and PDF documents directly in conversations to talk about their contents.
+- **Web search**: Optionally search the internet with LLM-generated queries based on your input to add context to the conversation.
+- **Advanced chat management**: Edit messages, navigate between message versions (like "swipes"), and branch conversations at any point.
+- **Automatic prompt formatting** using Jinja2 templates. You don't need to ever worry about prompt formats.
 - UI that resembles the original ChatGPT style.
-- Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
 - Three chat modes: `instruct`, `chat-instruct`, and `chat`, with automatic prompt templates in `chat-instruct`.
 - Free-form text generation in the Default/Notebook tabs without being limited to chat turns. You can send formatted conversations from the Chat tab to these.
 - Multiple sampling parameters and generation options for sophisticated text generation control.
 - Switch between different models easily in the UI without restarting, with fine control over settings.
 - OpenAI-compatible API with Chat and Completions endpoints, including tool-calling support – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples).
-- 100% offline and private, with zero telemetry, external resources, or remote update requests.
+- 100% offline and private, with zero telemetry, external resources, or remote update requests. Web search is optional and user-controlled.
 - Extension support, with numerous built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details.
 
 ## How to install

From 2a9699033d90f4ffedfb22cbba7003c6441d08dc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 04:55:59 -0700
Subject: [PATCH 0569/1701] Update README

---
 README.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index afb21cb07e..058094361a 100644
--- a/README.md
+++ b/README.md
@@ -16,8 +16,8 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 - Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory that doesn't interfere with your system environment.
 - **File attachments**: Upload text files and PDF documents directly in conversations to talk about their contents.
 - **Web search**: Optionally search the internet with LLM-generated queries based on your input to add context to the conversation.
-- **Advanced chat management**: Edit messages, navigate between message versions (like "swipes"), and branch conversations at any point.
-- **Automatic prompt formatting** using Jinja2 templates. You don't need to ever worry about prompt formats.
+- Advanced chat management: Edit messages, navigate between message versions (like "swipes"), and branch conversations at any point.
+- Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
 - UI that resembles the original ChatGPT style.
 - Three chat modes: `instruct`, `chat-instruct`, and `chat`, with automatic prompt templates in `chat-instruct`.
 - Free-form text generation in the Default/Notebook tabs without being limited to chat turns. You can send formatted conversations from the Chat tab to these.

From 9a94d7b4f6ae95b6b4b2fc521b5b25c300915dc9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 05:02:52 -0700
Subject: [PATCH 0570/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 058094361a..900d5fbd73 100644
--- a/README.md
+++ b/README.md
@@ -13,7 +13,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 ## Features
 
 - Supports multiple text generation backends in one UI/API, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).
-- Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory that doesn't interfere with your system environment.
+- Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory without affecting your system.
 - **File attachments**: Upload text files and PDF documents directly in conversations to talk about their contents.
 - **Web search**: Optionally search the internet with LLM-generated queries based on your input to add context to the conversation.
 - Advanced chat management: Edit messages, navigate between message versions (like "swipes"), and branch conversations at any point.

From 0986d075fb22dc5aa582bbefdfdb0ebdb6ee92c8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 05:03:59 -0700
Subject: [PATCH 0571/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 900d5fbd73..ec01c0aaeb 100644
--- a/README.md
+++ b/README.md
@@ -16,7 +16,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 - Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory without affecting your system.
 - **File attachments**: Upload text files and PDF documents directly in conversations to talk about their contents.
 - **Web search**: Optionally search the internet with LLM-generated queries based on your input to add context to the conversation.
-- Advanced chat management: Edit messages, navigate between message versions (like "swipes"), and branch conversations at any point.
+- Advanced chat management: Edit messages, navigate between message versions, and branch conversations at any point.
 - Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
 - UI that resembles the original ChatGPT style.
 - Three chat modes: `instruct`, `chat-instruct`, and `chat`, with automatic prompt templates in `chat-instruct`.

From 36bc2760058ed4e6998f4c55176c7311b0facabe Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 05:39:26 -0700
Subject: [PATCH 0572/1701] Update README

---
 README.md | 1 +
 1 file changed, 1 insertion(+)

diff --git a/README.md b/README.md
index ec01c0aaeb..9accffb7bd 100644
--- a/README.md
+++ b/README.md
@@ -18,6 +18,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 - **Web search**: Optionally search the internet with LLM-generated queries based on your input to add context to the conversation.
 - Advanced chat management: Edit messages, navigate between message versions, and branch conversations at any point.
 - Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
+- Automatic GPU layers for GGUF models (on NVIDIA GPUs).
 - UI that resembles the original ChatGPT style.
 - Three chat modes: `instruct`, `chat-instruct`, and `chat`, with automatic prompt templates in `chat-instruct`.
 - Free-form text generation in the Default/Notebook tabs without being limited to chat turns. You can send formatted conversations from the Chat tab to these.

From 81794692ab6fbc0ef24c7484b6571de090984dde Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 08:07:14 -0700
Subject: [PATCH 0573/1701] UI: Make the dark theme darker

---
 css/main.css | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/css/main.css b/css/main.css
index 8af87b42f7..0d0a13cfda 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1,11 +1,11 @@
 :root {
     --darker-gray: #202123;
-    --dark-gray: #343541;
-    --light-gray: #444654;
+    --dark-gray: #2A2B32;
+    --light-gray: #373943;
     --light-theme-gray: #f9fbff;
     --border-color-dark: #525252;
     --header-width: 112px;
-    --selected-item-color-dark: #32333e;
+    --selected-item-color-dark: #2E2F38;
 }
 
 @font-face {

From c970c5f1665c3966c84ba50a05a45d2598038ea6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 08:15:13 -0700
Subject: [PATCH 0574/1701] Make scrollbars darker in dark theme

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 0d0a13cfda..7f9d4618d6 100644
--- a/css/main.css
+++ b/css/main.css
@@ -265,7 +265,7 @@ button {
 
 .dark .pretty_scrollbar::-webkit-scrollbar-thumb,
 .dark .pretty_scrollbar::-webkit-scrollbar-thumb:hover {
-    background: rgb(255 255 255 / 20%);
+    background: rgb(255 255 255 / 10%);
     border-radius: 10px;
 }
 

From 3f37a2e915a31b273caddd12a80412a199d753a7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 08:49:31 -0700
Subject: [PATCH 0575/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 9accffb7bd..361584f852 100644
--- a/README.md
+++ b/README.md
@@ -13,7 +13,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 ## Features
 
 - Supports multiple text generation backends in one UI/API, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).
-- Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory without affecting your system.
+- Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory.
 - **File attachments**: Upload text files and PDF documents directly in conversations to talk about their contents.
 - **Web search**: Optionally search the internet with LLM-generated queries based on your input to add context to the conversation.
 - Advanced chat management: Edit messages, navigate between message versions, and branch conversations at any point.

From faa5c82c64e2036762ed3ff60a38fc5b37dac36d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 09:02:34 -0700
Subject: [PATCH 0576/1701] Fix message version count not updating during
 regeneration streaming

---
 modules/chat.py | 37 ++++++++++++++++++++++++++++++-------
 1 file changed, 30 insertions(+), 7 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 7afd906de1..90d66687dc 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -617,10 +617,19 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
         if regenerate:
             row_idx = len(output['internal']) - 1
 
-            # Store the first response as a version before regenerating
+            # Store the old response as a version before regenerating
             if not output['metadata'].get(f"assistant_{row_idx}", {}).get('versions'):
                 add_message_version(output, "assistant", row_idx, is_current=False)
 
+            # Add new empty version (will be filled during streaming)
+            key = f"assistant_{row_idx}"
+            output['metadata'][key]["versions"].append({
+                "content": "",
+                "visible_content": "",
+                "timestamp": get_current_timestamp()
+            })
+            output['metadata'][key]["current_version_index"] = len(output['metadata'][key]["versions"]) - 1
+
             if loading_message:
                 yield {
                     'visible': output['visible'][:-1] + [[visible_text, shared.processing_message]],
@@ -673,20 +682,34 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
         if _continue:
             output['internal'][-1] = [text, last_reply[0] + reply]
             output['visible'][-1] = [visible_text, last_reply[1] + visible_reply]
-            if is_stream:
-                yield output
         elif not (j == 0 and visible_reply.strip() == ''):
             output['internal'][-1] = [text, reply.lstrip(' ')]
             output['visible'][-1] = [visible_text, visible_reply.lstrip(' ')]
-            if is_stream:
-                yield output
+
+        # Keep version metadata in sync during streaming (for regeneration)
+        if regenerate:
+            row_idx = len(output['internal']) - 1
+            key = f"assistant_{row_idx}"
+            current_idx = output['metadata'][key]['current_version_index']
+            output['metadata'][key]['versions'][current_idx].update({
+                'content': output['internal'][row_idx][1],
+                'visible_content': output['visible'][row_idx][1]
+            })
+
+        if is_stream:
+            yield output
 
     output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
 
-    # Add the newly generated response as a version (only for regeneration)
+    # Final sync for version metadata (in case streaming was disabled)
     if regenerate:
         row_idx = len(output['internal']) - 1
-        add_message_version(output, "assistant", row_idx, is_current=True)
+        key = f"assistant_{row_idx}"
+        current_idx = output['metadata'][key]['current_version_index']
+        output['metadata'][key]['versions'][current_idx].update({
+            'content': output['internal'][row_idx][1],
+            'visible_content': output['visible'][row_idx][1]
+        })
 
     yield output
 

From 724147ffabce95b5d20528b83b6e44c1523d58f0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 10:49:29 -0700
Subject: [PATCH 0577/1701] Better detect when no model is available

---
 modules/utils.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/utils.py b/modules/utils.py
index 0e8bdd1867..577c55b8d1 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -74,7 +74,7 @@ def natural_keys(text):
 
 def check_model_loaded():
     if shared.model_name == 'None' or shared.model is None:
-        if len(get_available_models()) <= 1:
+        if len(get_available_models()) == 0:
             error_msg = "No model is loaded.\n\nTo get started:\n1) Place a GGUF file in your user_data/models folder\n2) Go to the Model tab and select it"
             logger.error(error_msg)
             return False, error_msg

From e7129f9dbefbe87fa4c425b5873f80cbddaf7cf0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 12:45:53 -0700
Subject: [PATCH 0578/1701] Prevent footer buttons below last assistant message
 from always appearing

---
 js/main.js | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/js/main.js b/js/main.js
index 48bb863201..ea3ff46a90 100644
--- a/js/main.js
+++ b/js/main.js
@@ -171,7 +171,6 @@ const observer = new MutationObserver(function(mutations) {
     document.getElementById("Generate").style.display = "flex";
   }
 
-
   doSyntaxHighlighting();
 
   if (!isScrolled && targetElement.scrollTop !== targetElement.scrollHeight) {
@@ -184,7 +183,7 @@ const observer = new MutationObserver(function(mutations) {
     const lastChild = messagesContainer?.lastElementChild;
     const prevSibling = lastChild?.previousElementSibling;
     if (lastChild && prevSibling) {
-      lastChild.style.minHeight = `calc(max(70vh, 100vh - ${prevSibling.offsetHeight}px - 102px))`;
+      lastChild.style.setProperty("margin-bottom", `calc(max(70vh, 100vh - ${prevSibling.offsetHeight}px - 102px) - ${lastChild.offsetHeight}px)`, "important");
     }
   }
 });

From aff41f3482bc7045334b0d81ac514723fdbd4f97 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 12:53:41 -0700
Subject: [PATCH 0579/1701] Update README

---
 README.md | 29 +++++++++++++----------------
 1 file changed, 13 insertions(+), 16 deletions(-)

diff --git a/README.md b/README.md
index 361584f852..daf409d0e4 100644
--- a/README.md
+++ b/README.md
@@ -189,13 +189,13 @@ usage: server.py [-h] [--multi-user] [--character CHARACTER] [--model MODEL] [--
                  [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR]
                  [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--use_flash_attention_2] [--use_eager_attention] [--torch-compile] [--load-in-4bit]
                  [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn] [--threads THREADS] [--threads-batch THREADS_BATCH] [--batch-size BATCH_SIZE] [--no-mmap]
-                 [--mlock] [--n-gpu-layers N_GPU_LAYERS] [--tensor-split TENSOR_SPLIT] [--numa] [--no-kv-offload] [--row-split] [--extra-flags EXTRA_FLAGS] [--streaming-llm] [--ctx-size N]
+                 [--mlock] [--gpu-layers N] [--tensor-split TENSOR_SPLIT] [--numa] [--no-kv-offload] [--row-split] [--extra-flags EXTRA_FLAGS] [--streaming-llm] [--ctx-size N] [--cache-type N]
                  [--model-draft MODEL_DRAFT] [--draft-max DRAFT_MAX] [--gpu-layers-draft GPU_LAYERS_DRAFT] [--device-draft DEVICE_DRAFT] [--ctx-size-draft CTX_SIZE_DRAFT] [--gpu-split GPU_SPLIT]
-                 [--autosplit] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa] [--num_experts_per_token N] [--enable_tp] [--hqq-backend HQQ_BACKEND] [--cpp-runner]
-                 [--cache_type CACHE_TYPE] [--deepspeed] [--nvme-offload-dir NVME_OFFLOAD_DIR] [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE]
-                 [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH]
-                 [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors] [--api] [--public-api]
-                 [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4] [--nowebui]
+                 [--autosplit] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa] [--num_experts_per_token N] [--enable_tp] [--cpp-runner] [--deepspeed] [--nvme-offload-dir NVME_OFFLOAD_DIR]
+                 [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT]
+                 [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE]
+                 [--subpath SUBPATH] [--old-colors] [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY]
+                 [--api-enable-ipv6] [--api-disable-ipv4] [--nowebui]
 
 Text generation web UI
 
@@ -217,7 +217,7 @@ Basic settings:
   --idle-timeout IDLE_TIMEOUT               Unload model after this many minutes of inactivity. It will be automatically reloaded when you try to use it again.
 
 Model loader:
-  --loader LOADER                           Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF, ExLlamav2, HQQ,
+  --loader LOADER                           Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF, ExLlamav2,
                                             TensorRT-LLM.
 
 Transformers/Accelerate:
@@ -248,16 +248,18 @@ llama.cpp:
   --batch-size BATCH_SIZE                   Maximum number of prompt tokens to batch together when calling llama_eval.
   --no-mmap                                 Prevent mmap from being used.
   --mlock                                   Force the system to keep the model in RAM.
-  --n-gpu-layers N_GPU_LAYERS               Number of layers to offload to the GPU.
+  --gpu-layers N, --n-gpu-layers N          Number of layers to offload to the GPU.
   --tensor-split TENSOR_SPLIT               Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.
   --numa                                    Activate NUMA task allocation for llama.cpp.
   --no-kv-offload                           Do not offload the K, Q, V to the GPU. This saves VRAM but reduces the performance.
   --row-split                               Split the model by rows across GPUs. This may improve multi-gpu performance.
-  --extra-flags EXTRA_FLAGS                 Extra flags to pass to llama-server. Format: "flag1=value1;flag2;flag3=value3". Example: "override-tensor=exps=CPU"
+  --extra-flags EXTRA_FLAGS                 Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"
   --streaming-llm                           Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.
 
-Context and cache management:
+Context and cache:
   --ctx-size N, --n_ctx N, --max_seq_len N  Context size in tokens.
+  --cache-type N, --cache_type N            KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits
+                                            separately, e.g. q4_q8).
 
 Speculative decoding:
   --model-draft MODEL_DRAFT                 Path to the draft model for speculative decoding.
@@ -276,15 +278,9 @@ ExLlamaV2:
   --num_experts_per_token N                 Number of experts to use for generation. Applies to MoE models like Mixtral.
   --enable_tp                               Enable Tensor Parallelism (TP) in ExLlamaV2.
 
-HQQ:
-  --hqq-backend HQQ_BACKEND                 Backend for the HQQ loader. Valid options: PYTORCH, PYTORCH_COMPILE, ATEN.
-
 TensorRT-LLM:
   --cpp-runner                              Use the ModelRunnerCpp runner, which is faster than the default ModelRunner but doesn't support streaming yet.
 
-Cache:
-  --cache_type CACHE_TYPE                   KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4.
-
 DeepSpeed:
   --deepspeed                               Enable the use of DeepSpeed ZeRO-3 for inference via the Transformers integration.
   --nvme-offload-dir NVME_OFFLOAD_DIR       DeepSpeed: Directory to use for ZeRO-3 NVME offloading.
@@ -307,6 +303,7 @@ Gradio:
   --ssl-certfile SSL_CERTFILE               The path to the SSL certificate cert file.
   --subpath SUBPATH                         Customize the subpath for gradio, use with reverse proxy
   --old-colors                              Use the legacy Gradio colors, before the December/2024 update.
+  --portable                                Hide features not available in portable mode like training.
 
 API:
   --api                                     Enable the API extension.

From f59998d2680f346038320b536617c4738c393947 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 13:08:48 -0700
Subject: [PATCH 0580/1701] Don't limit the number of prompt characters printed
 with --verbose

---
 modules/text_generation.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index 962311df3c..1fd6d810f7 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -505,11 +505,11 @@ def generate_reply_custom(question, original_question, state, stopping_strings=N
         return
 
 
-def print_prompt(prompt, max_chars=2000):
+def print_prompt(prompt, max_chars=-1):
     DARK_YELLOW = "\033[38;5;3m"
     RESET = "\033[0m"
 
-    if len(prompt) > max_chars:
+    if max_chars > 0 and len(prompt) > max_chars:
         half_chars = max_chars // 2
         hidden_len = len(prompt[half_chars:-half_chars])
         hidden_msg = f"{DARK_YELLOW}[...{hidden_len} characters hidden...]{RESET}"

From a45a65213052dad02d696ed54af1b9f2ea82cd4a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 13:28:51 -0700
Subject: [PATCH 0581/1701] CSS fix

---
 js/main.js | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index ea3ff46a90..f23dc246ee 100644
--- a/js/main.js
+++ b/js/main.js
@@ -183,7 +183,10 @@ const observer = new MutationObserver(function(mutations) {
     const lastChild = messagesContainer?.lastElementChild;
     const prevSibling = lastChild?.previousElementSibling;
     if (lastChild && prevSibling) {
-      lastChild.style.setProperty("margin-bottom", `calc(max(70vh, 100vh - ${prevSibling.offsetHeight}px - 102px) - ${lastChild.offsetHeight}px)`, "important");
+      lastChild.style.setProperty("margin-bottom",
+        `max(0px, calc(max(70vh, 100vh - ${prevSibling.offsetHeight}px - 102px) - ${lastChild.offsetHeight}px))`,
+        "important"
+      );
     }
   }
 });

From 8078c41ec67b96656d7e96128d915290b319e4f5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 13:32:19 -0700
Subject: [PATCH 0582/1701] Revert "Bump llama.cpp"

This reverts commit a8d02dec8f5e6a054a153b3b09425b51e090ae11.
---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 5f61aff9b9..0eaf10da6a 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index a718b6cab4..65f184bfa6 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -32,7 +32,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 5fddc62309..d20b2ec3fb 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -32,7 +32,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 8e0144451d..2613d78765 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -32,7 +32,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 77779f3dc2..af583b0070 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -32,8 +32,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 79efc607c5..9bf2a37dd2 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -32,5 +32,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 8b29453ec1..1731448ea9 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -32,5 +32,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index f1f4a02e4c..fc481a1a71 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index adf50d9af8..fdae681df4 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 46b3679164..a58f39f7f9 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 660527113b..91ea3a6df8 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -18,6 +18,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 4013abcc7a..37e5aa40b4 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 41808854bf..dcb2884ba9 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index cff79ec600..8f1295bbac 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 762b3fa347..858b4488d7 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index b425d30525..569bae9962 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.15.0/llama_cpp_binaries-0.15.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From dce02732a4caef16157ffbc288dfe079053e0bb4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 14:08:48 -0700
Subject: [PATCH 0583/1701] Fix timestamp issues when editing/swiping messages

---
 modules/chat.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 90d66687dc..6b3ff4fcb7 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1508,11 +1508,12 @@ def handle_edit_message_click(state):
     if "versions" not in history['metadata'][key] or not history['metadata'][key]["versions"]:
         original_content = history['internal'][message_index][role_idx]
         original_visible = history['visible'][message_index][role_idx]
+        original_timestamp = history['metadata'][key].get('timestamp', get_current_timestamp())
 
         history['metadata'][key]["versions"] = [{
             "content": original_content,
             "visible_content": original_visible,
-            "timestamp": get_current_timestamp()
+            "timestamp": original_timestamp
         }]
 
     history['internal'][message_index][role_idx] = apply_extensions('input', new_text, state, is_chat=True)
@@ -1564,6 +1565,7 @@ def handle_navigate_version_click(state):
     history['internal'][message_index][msg_content_idx] = version_to_load['content']
     history['visible'][message_index][msg_content_idx] = version_to_load['visible_content']
     metadata['current_version_index'] = new_idx
+    update_message_metadata(history['metadata'], role, message_index, timestamp=version_to_load['timestamp'])
 
     # Redraw and save
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])

From acbcc12e7b19cc9f540d32b8d601ceefde77b7a1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 14:11:21 -0700
Subject: [PATCH 0584/1701] Clean up

---
 modules/chat.py    | 7 ++-----
 modules/ui_chat.py | 2 +-
 2 files changed, 3 insertions(+), 6 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 6b3ff4fcb7..e526a9a0c4 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1493,7 +1493,7 @@ def handle_edit_message_click(state):
 
     if message_index >= len(history['internal']):
         html_output = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
-        return [history, html_output, gr.update()]  # No unique_id change
+        return [history, html_output]
 
     role_idx = 0 if role == "user" else 1
 
@@ -1521,13 +1521,10 @@ def handle_edit_message_click(state):
 
     add_message_version(history, role, message_index, is_current=True)
 
-    # Since we are not branching, unique_id does not change.
-    past_chats_update = gr.update()
-
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
     html_output = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
 
-    return [history, html_output, past_chats_update]
+    return [history, html_output]
 
 
 def handle_navigate_version_click(state):
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index df3d3929b7..d79aa5235b 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -314,7 +314,7 @@ def create_event_handlers():
 
     shared.gradio['edit_message'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        chat.handle_edit_message_click, gradio('interface_state'), gradio('history', 'display', 'unique_id'), show_progress=False)
+        chat.handle_edit_message_click, gradio('interface_state'), gradio('history', 'display'), show_progress=False)
 
     # Save/delete a character
     shared.gradio['save_character'].click(chat.handle_save_character_click, gradio('name2'), gradio('save_character_filename', 'character_saver'), show_progress=False)

From d1bfb08e8d4bab174e6b4467eff20f8a01a2a613 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 14:27:47 -0700
Subject: [PATCH 0585/1701] Improve the style of message editing

---
 css/main.css | 1 +
 1 file changed, 1 insertion(+)

diff --git a/css/main.css b/css/main.css
index 7f9d4618d6..9685c86317 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1462,6 +1462,7 @@ strong {
 .editing-textarea {
     width: 100%;
     min-height: 200px;
+    max-height: 65vh;
     padding: 10px;
     border-radius: 5px;
     border: 1px solid #ccc;

From 28e6bd4fcd8cd385cc92cc56c0c49fc474006147 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 29 May 2025 14:49:07 -0700
Subject: [PATCH 0586/1701] Revert "Update transformers requirement in
 /requirements/full (#7017)"

This reverts commit cc9b7253c1216e5340da85cba9b65a13cf3526e9.
---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 0eaf10da6a..2c32271538 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -23,7 +23,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.52.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 65f184bfa6..6aeb325e0a 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.52.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index d20b2ec3fb..3b052423f7 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.52.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 2613d78765..8c51459e50 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.52.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index af583b0070..b9f15d4502 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.52.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 9bf2a37dd2..0877d968f4 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.52.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 1731448ea9..cab7823743 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.52.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index fc481a1a71..dfd4257797 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -23,7 +23,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.52.*
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 2ed8affa47..5d9f84ce49 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.52.*
+transformers==4.50.*
 tqdm
 wandb
 

From 7c29879e795776ceb742a8ddb47fd3843069cf34 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 30 May 2025 11:17:47 -0700
Subject: [PATCH 0587/1701] Fix 'Start reply with' (closes #7033)

---
 modules/chat.py | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index e526a9a0c4..881f73302a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -806,9 +806,12 @@ def remove_last_message(history):
     return html.unescape(last[0]), history
 
 
-def send_dummy_message(textbox, state):
+def send_dummy_message(text, state):
     history = state['history']
-    text = textbox['text']
+
+    # Handle both dict and string inputs
+    if isinstance(text, dict):
+        text = text['text']
 
     # Initialize metadata if not present
     if 'metadata' not in history:
@@ -822,9 +825,12 @@ def send_dummy_message(textbox, state):
     return history
 
 
-def send_dummy_reply(textbox, state):
+def send_dummy_reply(text, state):
     history = state['history']
-    text = textbox['text']
+
+    # Handle both dict and string inputs
+    if isinstance(text, dict):
+        text = text['text']
 
     # Initialize metadata if not present
     if 'metadata' not in history:

From 298d4719c6c9545a701a9cc9e8f4efceb108599a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 30 May 2025 11:32:24 -0700
Subject: [PATCH 0588/1701] Multiple small style improvements

---
 css/main.css  | 4 ++++
 modules/ui.py | 3 +++
 2 files changed, 7 insertions(+)

diff --git a/css/main.css b/css/main.css
index 9685c86317..967d94ed3c 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1551,3 +1551,7 @@ strong {
     color: var(--body-text-color-subdued);
     margin-top: 4px;
 }
+
+button:focus {
+    outline: none;
+}
diff --git a/modules/ui.py b/modules/ui.py
index a2662e1403..9f4d67cb73 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -71,6 +71,7 @@
         block_background_fill_dark='transparent',
         block_border_color_dark='transparent',
         input_border_color_dark='var(--border-color-dark)',
+        input_border_color_focus_dark='var(--border-color-dark)',
         checkbox_border_color_dark='var(--border-color-dark)',
         border_color_primary_dark='var(--border-color-dark)',
         button_secondary_border_color_dark='var(--border-color-dark)',
@@ -89,6 +90,8 @@
         checkbox_label_shadow='none',
         block_shadow='none',
         block_shadow_dark='none',
+        input_shadow_focus='none',
+        input_shadow_focus_dark='none',
         button_large_radius='0.375rem',
         button_large_padding='6px 12px',
         input_radius='0.375rem',

From 219f0a773166deeb0326c2874b29e66e382df524 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 30 May 2025 12:05:49 -0700
Subject: [PATCH 0589/1701] Fix exllamav3_hf models failing to unload (closes
 #7031)

---
 modules/exllamav3_hf.py | 17 +++++++++++++++++
 modules/models.py       |  3 +++
 2 files changed, 20 insertions(+)

diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index 417df4734a..1254ff5dad 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -245,3 +245,20 @@ def from_pretrained(cls, pretrained_model_name_or_path: Optional[Union[str, os.P
         pretrained_model_name_or_path = Path(f'{shared.args.model_dir}') / Path(pretrained_model_name_or_path)
 
         return Exllamav3HF(pretrained_model_name_or_path)
+
+    def unload(self):
+        """Properly unload the ExllamaV3 model and free GPU memory."""
+        if hasattr(self, 'ex_model') and self.ex_model is not None:
+            self.ex_model.unload()
+            self.ex_model = None
+
+        if hasattr(self, 'ex_cache') and self.ex_cache is not None:
+            self.ex_cache = None
+
+        # Clean up any additional ExllamaV3 resources
+        if hasattr(self, 'past_seq'):
+            self.past_seq = None
+        if hasattr(self, 'past_seq_negative'):
+            self.past_seq_negative = None
+        if hasattr(self, 'ex_cache_negative'):
+            self.ex_cache_negative = None
diff --git a/modules/models.py b/modules/models.py
index 4218d58c9f..d329ae3c32 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -116,10 +116,13 @@ def unload_model(keep_model_name=False):
         return
 
     is_llamacpp = (shared.model.__class__.__name__ == 'LlamaServer')
+    if shared.args.loader == 'ExLlamav3_HF':
+        shared.model.unload()
 
     shared.model = shared.tokenizer = None
     shared.lora_names = []
     shared.model_dirty_from_training = False
+
     if not is_llamacpp:
         from modules.torch_utils import clear_torch_cache
         clear_torch_cache()

From 15f466ca3f8255f2566f016db8d7b8fd9ebef3f4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 30 May 2025 15:49:57 -0700
Subject: [PATCH 0590/1701] Update README

---
 README.md | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

diff --git a/README.md b/README.md
index daf409d0e4..55df33d2ca 100644
--- a/README.md
+++ b/README.md
@@ -14,18 +14,18 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 
 - Supports multiple text generation backends in one UI/API, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).
 - Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory.
-- **File attachments**: Upload text files and PDF documents directly in conversations to talk about their contents.
-- **Web search**: Optionally search the internet with LLM-generated queries based on your input to add context to the conversation.
-- Advanced chat management: Edit messages, navigate between message versions, and branch conversations at any point.
+- 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
-- Automatic GPU layers for GGUF models (on NVIDIA GPUs).
-- UI that resembles the original ChatGPT style.
-- Three chat modes: `instruct`, `chat-instruct`, and `chat`, with automatic prompt templates in `chat-instruct`.
-- Free-form text generation in the Default/Notebook tabs without being limited to chat turns. You can send formatted conversations from the Chat tab to these.
+- **File attachments**: Upload text files and PDF documents to talk about their contents.
+- **Web search**: Optionally search the internet with LLM-generated queries to add context to the conversation.
+- Aesthetic UI with dark and light themes.
+- `instruct` mode for instruction-following (like ChatGPT), and `chat-instruct`/`chat` modes for talking to custom characters.
+- Edit messages, navigate between message versions, and branch conversations at any point.
 - Multiple sampling parameters and generation options for sophisticated text generation control.
-- Switch between different models easily in the UI without restarting, with fine control over settings.
+- Switch between different models in the UI without restarting.
+- Automatic GPU layers for GGUF models (on NVIDIA GPUs).
+- Free-form text generation in the Default/Notebook tabs without being limited to chat turns.
 - OpenAI-compatible API with Chat and Completions endpoints, including tool-calling support – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples).
-- 100% offline and private, with zero telemetry, external resources, or remote update requests. Web search is optional and user-controlled.
 - Extension support, with numerous built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details.
 
 ## How to install

From c55d3c61c6e44712e90fa60c1e434d7687e90947 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 31 May 2025 14:21:42 -0700
Subject: [PATCH 0591/1701] Bump exllamav2 to 0.3.1

---
 requirements/full/requirements.txt               | 6 +++---
 requirements/full/requirements_amd.txt           | 4 ++--
 requirements/full/requirements_amd_noavx2.txt    | 4 ++--
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_noavx2.txt        | 6 +++---
 6 files changed, 12 insertions(+), 12 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 2c32271538..dd6313413a 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -37,8 +37,8 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 6aeb325e0a..acdbd45565 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -34,5 +34,5 @@ tiktoken
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 3b052423f7..a478d7d38a 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -34,5 +34,5 @@ tiktoken
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 8c51459e50..98ed90a2f7 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -35,4 +35,4 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9-py3-none-any.whl
-https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index b9f15d4502..cb72d03652 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -36,4 +36,4 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9-py3-none-any.whl
-https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index dfd4257797..f6982134a5 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -37,8 +37,8 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.2.9/exllamav2-0.2.9-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/oobabooga/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

From dc8ed6dbe769457b3a2758780abefab0ab04c8a4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 31 May 2025 14:27:33 -0700
Subject: [PATCH 0592/1701] Bump exllamav3 to 0.0.3

---
 requirements/full/requirements.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_noavx2.txt        | 4 ++--
 4 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index dd6313413a..ec055876f9 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -35,8 +35,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 98ed90a2f7..96a48f325d 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -34,5 +34,5 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index cb72d03652..14b740815d 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -35,5 +35,5 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index f6982134a5..de50730860 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -35,8 +35,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.1a9/exllamav3-0.0.1a9+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 1d88456659d8e71800f6fb732b8cad7d36fa4c20 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 31 May 2025 20:15:07 -0700
Subject: [PATCH 0593/1701] Add support for .docx attachments

---
 README.md                                     |  2 +-
 modules/chat.py                               | 50 +++++++++++++++++++
 requirements/full/requirements.txt            |  1 +
 requirements/full/requirements_amd.txt        |  1 +
 requirements/full/requirements_amd_noavx2.txt |  1 +
 .../full/requirements_apple_intel.txt         |  1 +
 .../full/requirements_apple_silicon.txt       |  1 +
 requirements/full/requirements_cpu_only.txt   |  1 +
 .../full/requirements_cpu_only_noavx2.txt     |  1 +
 requirements/full/requirements_noavx2.txt     |  1 +
 requirements/full/requirements_nowheels.txt   |  1 +
 requirements/portable/requirements.txt        |  1 +
 .../portable/requirements_apple_intel.txt     |  1 +
 .../portable/requirements_apple_silicon.txt   |  1 +
 .../portable/requirements_cpu_only.txt        |  1 +
 .../portable/requirements_cpu_only_noavx2.txt |  1 +
 requirements/portable/requirements_noavx2.txt |  1 +
 .../portable/requirements_nowheels.txt        |  1 +
 requirements/portable/requirements_vulkan.txt |  1 +
 .../portable/requirements_vulkan_noavx2.txt   |  1 +
 20 files changed, 69 insertions(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 55df33d2ca..16b0253919 100644
--- a/README.md
+++ b/README.md
@@ -16,7 +16,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 - Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory.
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
-- **File attachments**: Upload text files and PDF documents to talk about their contents.
+- **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
 - **Web search**: Optionally search the internet with LLM-generated queries to add context to the conversation.
 - Aesthetic UI with dark and light themes.
 - `instruct` mode for instruction-following (like ChatGPT), and `chat-instruct`/`chat` modes for talking to custom characters.
diff --git a/modules/chat.py b/modules/chat.py
index 881f73302a..ba61c7a97d 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -500,6 +500,9 @@ def add_message_attachment(history, row_idx, file_path, is_user=True):
             # Process PDF file
             content = extract_pdf_text(path)
             file_type = "application/pdf"
+        elif file_extension == '.docx':
+            content = extract_docx_text(path)
+            file_type = "application/docx"
         else:
             # Default handling for text files
             with open(path, 'r', encoding='utf-8') as f:
@@ -538,6 +541,53 @@ def extract_pdf_text(pdf_path):
         return f"[Error extracting PDF text: {str(e)}]"
 
 
+def extract_docx_text(docx_path):
+    """
+    Extract text from a .docx file, including headers,
+    body (paragraphs and tables), and footers.
+    """
+    try:
+        import docx
+
+        doc = docx.Document(docx_path)
+        parts = []
+
+        # 1) Extract non-empty header paragraphs from each section
+        for section in doc.sections:
+            for para in section.header.paragraphs:
+                text = para.text.strip()
+                if text:
+                    parts.append(text)
+
+        # 2) Extract body blocks (paragraphs and tables) in document order
+        parent_elm = doc.element.body
+        for child in parent_elm.iterchildren():
+            if isinstance(child, docx.oxml.text.paragraph.CT_P):
+                para = docx.text.paragraph.Paragraph(child, doc)
+                text = para.text.strip()
+                if text:
+                    parts.append(text)
+
+            elif isinstance(child, docx.oxml.table.CT_Tbl):
+                table = docx.table.Table(child, doc)
+                for row in table.rows:
+                    cells = [cell.text.strip() for cell in row.cells]
+                    parts.append("\t".join(cells))
+
+        # 3) Extract non-empty footer paragraphs from each section
+        for section in doc.sections:
+            for para in section.footer.paragraphs:
+                text = para.text.strip()
+                if text:
+                    parts.append(text)
+
+        return "\n".join(parts)
+
+    except Exception as e:
+        logger.error(f"Error extracting text from DOCX: {e}")
+        return f"[Error extracting DOCX text: {str(e)}]"
+
+
 def generate_search_query(user_message, state):
     """Generate a search query from user message using the LLM"""
     # Augment the user message with search instruction
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index ec055876f9..e61677a618 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -16,6 +16,7 @@ Pillow>=9.5.0
 psutil
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index acdbd45565..f807199d44 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -15,6 +15,7 @@ Pillow>=9.5.0
 psutil
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index a478d7d38a..4fb70eb134 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -15,6 +15,7 @@ Pillow>=9.5.0
 psutil
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 96a48f325d..a311ab9b48 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -15,6 +15,7 @@ Pillow>=9.5.0
 psutil
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 14b740815d..30e8409a32 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -15,6 +15,7 @@ Pillow>=9.5.0
 psutil
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 0877d968f4..7094994920 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -15,6 +15,7 @@ Pillow>=9.5.0
 psutil
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index cab7823743..318bb93af3 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -15,6 +15,7 @@ Pillow>=9.5.0
 psutil
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index de50730860..e0cb84b480 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -16,6 +16,7 @@ Pillow>=9.5.0
 psutil
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 5d9f84ce49..a412367cf2 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -15,6 +15,7 @@ Pillow>=9.5.0
 psutil
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index fdae681df4..bde310e174 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -7,6 +7,7 @@ markdown
 numpy==1.26.*
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index a58f39f7f9..521edc0c38 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -7,6 +7,7 @@ markdown
 numpy==1.26.*
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 91ea3a6df8..ef7946ff62 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -7,6 +7,7 @@ markdown
 numpy==1.26.*
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 37e5aa40b4..a3ad743ed9 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -7,6 +7,7 @@ markdown
 numpy==1.26.*
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index dcb2884ba9..eec052d36f 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -7,6 +7,7 @@ markdown
 numpy==1.26.*
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 8f1295bbac..c9898a05d7 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -7,6 +7,7 @@ markdown
 numpy==1.26.*
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 21805fe241..f6c866cf34 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -7,6 +7,7 @@ markdown
 numpy==1.26.*
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 858b4488d7..0de9c7cbc7 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -7,6 +7,7 @@ markdown
 numpy==1.26.*
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 569bae9962..2bfb4d519f 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -7,6 +7,7 @@ markdown
 numpy==1.26.*
 pydantic==2.8.2
 PyPDF2==3.0.1
+python-docx==1.1.2
 pyyaml
 requests
 rich

From 4a2727b71d8976366cc35e18048ad9742ccb1898 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 31 May 2025 20:24:31 -0700
Subject: [PATCH 0594/1701] Add a tooltip to the file upload button

---
 js/main.js | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/js/main.js b/js/main.js
index f23dc246ee..0fdd7ffdb4 100644
--- a/js/main.js
+++ b/js/main.js
@@ -872,3 +872,10 @@ function navigateLastAssistantMessage(direction) {
 
   return false;
 }
+
+//------------------------------------------------
+// Tooltips
+//------------------------------------------------
+
+// File upload button
+document.querySelector("#chat-input .upload-button").title = "Upload text files, PDFs, and DOCX documents";

From f8d220c1e6c0263e76797b0e34dc9ce20335875b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 31 May 2025 21:22:36 -0700
Subject: [PATCH 0595/1701] Add a tooltip to the web search checkbox

---
 js/main.js         | 3 +++
 modules/ui_chat.py | 2 +-
 2 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index 0fdd7ffdb4..b9cb3cddb1 100644
--- a/js/main.js
+++ b/js/main.js
@@ -879,3 +879,6 @@ function navigateLastAssistantMessage(direction) {
 
 // File upload button
 document.querySelector("#chat-input .upload-button").title = "Upload text files, PDFs, and DOCX documents";
+
+// Activate web search
+document.getElementById("web-search").title = "Search the internet with DuckDuckGo";
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index d79aa5235b..73528a92c9 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -87,7 +87,7 @@ def create_ui():
                     shared.gradio['start_with'] = gr.Textbox(label='Start reply with', placeholder='Sure thing!', value=shared.settings['start_with'], elem_classes=['add_scrollbar'])
 
                 with gr.Row():
-                    shared.gradio['enable_web_search'] = gr.Checkbox(value=shared.settings.get('enable_web_search', False), label='Activate web search')
+                    shared.gradio['enable_web_search'] = gr.Checkbox(value=shared.settings.get('enable_web_search', False), label='Activate web search', elem_id='web-search')
 
                 with gr.Row(visible=shared.settings.get('enable_web_search', False)) as shared.gradio['web_search_row']:
                     shared.gradio['web_search_pages'] = gr.Number(value=shared.settings.get('web_search_pages', 3), precision=0, label='Number of pages to download', minimum=1, maximum=10)

From 85f2f01a3a78dc85bce9eeded71d9ff9f5bd4ab3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 31 May 2025 21:29:25 -0700
Subject: [PATCH 0596/1701] UI: Fix extra gaps on the right sidebar

---
 css/main.css | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/css/main.css b/css/main.css
index 967d94ed3c..bdaacd4fcf 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1555,3 +1555,8 @@ strong {
 button:focus {
     outline: none;
 }
+
+/* Fix extra gaps for hidden elements on the right sidebar */
+.svelte-sa48pu.stretch:has(> .hidden:only-child) {
+    display: none;
+}

From 98a7508a99f2c3bcb2139f7ef975b692f004c695 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 31 May 2025 22:18:17 -0700
Subject: [PATCH 0597/1701] UI: Move 'Show controls' inside the hover menu

---
 css/main.css       | 52 +++++++++++++++++++++++-----------------------
 js/main.js         | 42 +++++++++++++++----------------------
 modules/ui_chat.py | 25 ++++++++--------------
 3 files changed, 52 insertions(+), 67 deletions(-)

diff --git a/css/main.css b/css/main.css
index bdaacd4fcf..adc59fbaaa 100644
--- a/css/main.css
+++ b/css/main.css
@@ -582,7 +582,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 
 #chat-input {
     padding: 0;
-    padding-top: 18px;
     background: transparent;
     border: none;
 }
@@ -661,31 +660,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     }
 }
 
-#show-controls {
-    position: absolute;
-    background-color: transparent;
-    border: 0 !important;
-    border-radius: 0;
-}
-
-#show-controls label {
-    z-index: 1000;
-    position: absolute;
-    right: 30px;
-    top: 10px;
-    white-space: nowrap;
-    overflow: hidden;
-    text-overflow: ellipsis;
-}
-
-.dark #show-controls span {
-    color: var(--neutral-400);
-}
-
-#show-controls span {
-    color: var(--neutral-600);
-}
-
 #typing-container {
     display: none;
     position: absolute;
@@ -785,6 +759,32 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     background: var(--selected-item-color-dark) !important;
 }
 
+#show-controls {
+    height: 36px;
+    border-top: 1px solid var(--border-color-dark) !important;
+    border-left: 1px solid var(--border-color-dark) !important;
+    border-right: 1px solid var(--border-color-dark) !important;
+    border-radius: 0;
+    border-bottom: 0 !important;
+    background-color: var(--darker-gray);
+    padding-top: 3px;
+    padding-left: 4px;
+    display: flex;
+}
+
+#show-controls label {
+    display: flex;
+    flex-direction: row-reverse;
+    font-weight: bold;
+    justify-content: space-between;
+    width: 100%;
+    padding-right: 12px;
+}
+
+#show-controls label input {
+    margin-top: 4px;
+}
+
 .transparent-substring {
     opacity: 0.333;
 }
diff --git a/js/main.js b/js/main.js
index b9cb3cddb1..3652daa030 100644
--- a/js/main.js
+++ b/js/main.js
@@ -277,7 +277,7 @@ for (i = 0; i < slimDropdownElements.length; i++) {
 // The show/hide events were adapted from:
 // https://github.com/SillyTavern/SillyTavern/blob/6c8bd06308c69d51e2eb174541792a870a83d2d6/public/script.js
 //------------------------------------------------
-var buttonsInChat = document.querySelectorAll("#chat-tab #chat-buttons button");
+var buttonsInChat = document.querySelectorAll("#chat-tab #chat-buttons button, #chat-tab #chat-buttons #show-controls");
 var button = document.getElementById("hover-element-button");
 var menu = document.getElementById("hover-menu");
 var istouchscreen = (navigator.maxTouchPoints > 0) || "ontouchstart" in document.documentElement;
@@ -298,18 +298,21 @@ if (buttonsInChat.length > 0) {
     const thisButton = buttonsInChat[i];
     menu.appendChild(thisButton);
 
-    thisButton.addEventListener("click", () => {
-      hideMenu();
-    });
-
-    const buttonText = thisButton.textContent;
-    const matches = buttonText.match(/(\(.*?\))/);
-
-    if (matches && matches.length > 1) {
-      // Apply the transparent-substring class to the matched substring
-      const substring = matches[1];
-      const newText = buttonText.replace(substring, ` ${substring.slice(1, -1)}`);
-      thisButton.innerHTML = newText;
+    // Only apply transformations to button elements
+    if (thisButton.tagName.toLowerCase() === 'button') {
+      thisButton.addEventListener("click", () => {
+        hideMenu();
+      });
+      
+      const buttonText = thisButton.textContent;
+      const matches = buttonText.match(/(\(.*?\))/);
+
+      if (matches && matches.length > 1) {
+        // Apply the transparent-substring class to the matched substring
+        const substring = matches[1];
+        const newText = buttonText.replace(substring, ` ${substring.slice(1, -1)}`);
+        thisButton.innerHTML = newText;
+      }
     }
   }
 }
@@ -382,21 +385,10 @@ document.addEventListener("click", function (event) {
   }
 });
 
-//------------------------------------------------
-// Relocate the "Show controls" checkbox
-//------------------------------------------------
-var elementToMove = document.getElementById("show-controls");
-var parent = elementToMove.parentNode;
-for (var i = 0; i < 2; i++) {
-  parent = parent.parentNode;
-}
-
-parent.insertBefore(elementToMove, parent.firstChild);
-
 //------------------------------------------------
 // Position the chat input
 //------------------------------------------------
-document.getElementById("show-controls").parentNode.classList.add("chat-input-positioned");
+document.getElementById("chat-input-row").classList.add("chat-input-positioned");
 
 //------------------------------------------------
 // Focus on the chat input
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 73528a92c9..822b77b8e5 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -55,7 +55,6 @@ def create_ui():
 
                     with gr.Column(scale=10, elem_id='chat-input-container'):
                         shared.gradio['textbox'] = gr.MultimodalTextbox(label='', placeholder='Send a message', file_types=['text', '.pdf'], file_count="multiple", elem_id='chat-input', elem_classes=['add_scrollbar'])
-                        shared.gradio['show_controls'] = gr.Checkbox(value=shared.settings['show_controls'], label='Show controls (Ctrl+S)', elem_id='show-controls')
                         shared.gradio['typing-dots'] = gr.HTML(value='
            ', label='typing', elem_id='typing-container')
 
                     with gr.Column(scale=1, elem_id='generate-stop-container'):
@@ -65,21 +64,15 @@ def create_ui():
 
         # Hover menu buttons
         with gr.Column(elem_id='chat-buttons'):
-            with gr.Row():
-                shared.gradio['Regenerate'] = gr.Button('Regenerate (Ctrl + Enter)', elem_id='Regenerate')
-                shared.gradio['Continue'] = gr.Button('Continue (Alt + Enter)', elem_id='Continue')
-                shared.gradio['Remove last'] = gr.Button('Remove last reply (Ctrl + Shift + Backspace)', elem_id='Remove-last')
-
-            with gr.Row():
-                shared.gradio['Impersonate'] = gr.Button('Impersonate (Ctrl + Shift + M)', elem_id='Impersonate')
-
-            with gr.Row():
-                shared.gradio['Send dummy message'] = gr.Button('Send dummy message')
-                shared.gradio['Send dummy reply'] = gr.Button('Send dummy reply')
-
-            with gr.Row():
-                shared.gradio['send-chat-to-default'] = gr.Button('Send to Default')
-                shared.gradio['send-chat-to-notebook'] = gr.Button('Send to Notebook')
+            shared.gradio['Regenerate'] = gr.Button('Regenerate (Ctrl + Enter)', elem_id='Regenerate')
+            shared.gradio['Continue'] = gr.Button('Continue (Alt + Enter)', elem_id='Continue')
+            shared.gradio['Remove last'] = gr.Button('Remove last reply (Ctrl + Shift + Backspace)', elem_id='Remove-last')
+            shared.gradio['Impersonate'] = gr.Button('Impersonate (Ctrl + Shift + M)', elem_id='Impersonate')
+            shared.gradio['Send dummy message'] = gr.Button('Send dummy message')
+            shared.gradio['Send dummy reply'] = gr.Button('Send dummy reply')
+            shared.gradio['send-chat-to-default'] = gr.Button('Send to Default')
+            shared.gradio['send-chat-to-notebook'] = gr.Button('Send to Notebook')
+            shared.gradio['show_controls'] = gr.Checkbox(value=shared.settings['show_controls'], label='Show controls (Ctrl+S)', elem_id='show-controls')
 
         with gr.Row(elem_id='chat-controls', elem_classes=['pretty_scrollbar']):
             with gr.Column():

From 0816ecedb75add2dd1a61c9bd9a477e5d847c88a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 31 May 2025 22:24:39 -0700
Subject: [PATCH 0598/1701] Lint

---
 js/main.js | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/js/main.js b/js/main.js
index 3652daa030..d152a572ef 100644
--- a/js/main.js
+++ b/js/main.js
@@ -299,11 +299,11 @@ if (buttonsInChat.length > 0) {
     menu.appendChild(thisButton);
 
     // Only apply transformations to button elements
-    if (thisButton.tagName.toLowerCase() === 'button') {
+    if (thisButton.tagName.toLowerCase() === "button") {
       thisButton.addEventListener("click", () => {
         hideMenu();
       });
-      
+
       const buttonText = thisButton.textContent;
       const matches = buttonText.match(/(\(.*?\))/);
 

From 9e801930087170bb24628e680ad4cbd4f6a5b098 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 31 May 2025 22:39:07 -0700
Subject: [PATCH 0599/1701] Add the model name to each message's metadata

---
 modules/chat.py           |  2 +-
 modules/html_generator.py | 47 ++++++++++++++++++++++++++-------------
 2 files changed, 32 insertions(+), 17 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index ba61c7a97d..1222d2bb95 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -710,7 +710,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
 
     # Add timestamp for assistant's response at the start of generation
     row_idx = len(output['internal']) - 1
-    update_message_metadata(output['metadata'], "assistant", row_idx, timestamp=get_current_timestamp())
+    update_message_metadata(output['metadata'], "assistant", row_idx, timestamp=get_current_timestamp(), model_name=shared.model_name)
 
     # Generate
     reply = None
diff --git a/modules/html_generator.py b/modules/html_generator.py
index cbf3e19ca2..03b5d485d8 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -350,12 +350,14 @@ def get_image_cache(path):
 info_button = f''
 
 
-def format_message_timestamp(history, role, index):
+def format_message_timestamp(history, role, index, tooltip_include_timestamp=True):
     """Get a formatted timestamp HTML span for a message if available"""
     key = f"{role}_{index}"
     if 'metadata' in history and key in history['metadata'] and history['metadata'][key].get('timestamp'):
         timestamp = history['metadata'][key]['timestamp']
-        return f"{timestamp}"
+        tooltip_text = get_message_tooltip(history, role, index, include_timestamp=tooltip_include_timestamp)
+        title_attr = f' title="{html.escape(tooltip_text)}"' if tooltip_text else ''
+        return f"{timestamp}"
 
     return ""
 
@@ -388,6 +390,23 @@ def format_message_attachments(history, role, index):
     return ""
 
 
+def get_message_tooltip(history, role, index, include_timestamp=True):
+    """Get tooltip text combining timestamp and model name for a message"""
+    key = f"{role}_{index}"
+    if 'metadata' not in history or key not in history['metadata']:
+        return ""
+
+    meta = history['metadata'][key]
+    tooltip_parts = []
+
+    if include_timestamp and meta.get('timestamp'):
+        tooltip_parts.append(meta['timestamp'])
+    if meta.get('model_name'):
+        tooltip_parts.append(f"Model: {meta['model_name']}")
+
+    return " | ".join(tooltip_parts)
+
+
 def get_version_navigation_html(history, i, role):
     """Generate simple navigation arrows for message versions"""
     key = f"{role}_{i}"
@@ -462,15 +481,13 @@ def generate_instruct_html(history):
         # Create info buttons for timestamps if they exist
         info_message_user = ""
         if user_timestamp != "":
-            # Extract the timestamp value from the span
-            user_timestamp_value = user_timestamp.split('>', 1)[1].split('<', 1)[0]
-            info_message_user = info_button.replace("message", user_timestamp_value)
+            tooltip_text = get_message_tooltip(history, "user", i)
+            info_message_user = info_button.replace('title="message"', f'title="{html.escape(tooltip_text)}"')
 
         info_message_assistant = ""
         if assistant_timestamp != "":
-            # Extract the timestamp value from the span
-            assistant_timestamp_value = assistant_timestamp.split('>', 1)[1].split('<', 1)[0]
-            info_message_assistant = info_button.replace("message", assistant_timestamp_value)
+            tooltip_text = get_message_tooltip(history, "assistant", i)
+            info_message_assistant = info_button.replace('title="message"', f'title="{html.escape(tooltip_text)}"')
 
         if converted_visible[0]:  # Don't display empty user messages
             output += (
@@ -521,8 +538,8 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
         converted_visible = [convert_to_markdown_wrapped(entry, message_id=i, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
 
         # Get timestamps
-        user_timestamp = format_message_timestamp(history, "user", i)
-        assistant_timestamp = format_message_timestamp(history, "assistant", i)
+        user_timestamp = format_message_timestamp(history, "user", i, tooltip_include_timestamp=False)
+        assistant_timestamp = format_message_timestamp(history, "assistant", i, tooltip_include_timestamp=False)
 
         # Get attachments
         user_attachments = format_message_attachments(history, "user", i)
@@ -580,15 +597,13 @@ def generate_chat_html(history, name1, name2, reset_cache=False):
         # Create info buttons for timestamps if they exist
         info_message_user = ""
         if user_timestamp != "":
-            # Extract the timestamp value from the span
-            user_timestamp_value = user_timestamp.split('>', 1)[1].split('<', 1)[0]
-            info_message_user = info_button.replace("message", user_timestamp_value)
+            tooltip_text = get_message_tooltip(history, "user", i)
+            info_message_user = info_button.replace('title="message"', f'title="{html.escape(tooltip_text)}"')
 
         info_message_assistant = ""
         if assistant_timestamp != "":
-            # Extract the timestamp value from the span
-            assistant_timestamp_value = assistant_timestamp.split('>', 1)[1].split('<', 1)[0]
-            info_message_assistant = info_button.replace("message", assistant_timestamp_value)
+            tooltip_text = get_message_tooltip(history, "assistant", i)
+            info_message_assistant = info_button.replace('title="message"', f'title="{html.escape(tooltip_text)}"')
 
         if converted_visible[0]:  # Don't display empty user messages
             output += (

From 88ff3e6ad8ddf96aabf6d7ceb4c228ed6fb08980 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 1 Jun 2025 08:00:37 -0700
Subject: [PATCH 0600/1701] CSS fixes after
 98a7508a99f2c3bcb2139f7ef975b692f004c695

---
 css/main.css | 2 +-
 js/main.js   | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/css/main.css b/css/main.css
index adc59fbaaa..0c6dc16ee5 100644
--- a/css/main.css
+++ b/css/main.css
@@ -665,7 +665,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     position: absolute;
     background-color: transparent;
     left: -2px;
-    top: 4px;
+    top: -14px;
     padding: var(--block-padding);
 }
 
diff --git a/js/main.js b/js/main.js
index d152a572ef..05c19571fc 100644
--- a/js/main.js
+++ b/js/main.js
@@ -184,7 +184,7 @@ const observer = new MutationObserver(function(mutations) {
     const prevSibling = lastChild?.previousElementSibling;
     if (lastChild && prevSibling) {
       lastChild.style.setProperty("margin-bottom",
-        `max(0px, calc(max(70vh, 100vh - ${prevSibling.offsetHeight}px - 102px) - ${lastChild.offsetHeight}px))`,
+        `max(0px, calc(max(70vh, 100vh - ${prevSibling.offsetHeight}px - 84px) - ${lastChild.offsetHeight}px))`,
         "important"
       );
     }

From 3e3746283cd60409f83b6cf5549ba08d12612bde Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 1 Jun 2025 10:55:31 -0700
Subject: [PATCH 0601/1701] Improve the typing dots position

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 0c6dc16ee5..296476cd15 100644
--- a/css/main.css
+++ b/css/main.css
@@ -665,7 +665,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     position: absolute;
     background-color: transparent;
     left: -2px;
-    top: -14px;
+    top: -5px;
     padding: var(--block-padding);
 }
 

From 83849336d8efcae0340b768a39c83106ee406264 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 1 Jun 2025 10:58:28 -0700
Subject: [PATCH 0602/1701] Improve how Show controls looks in the hover menu

---
 css/main.css | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 296476cd15..71d67ff48f 100644
--- a/css/main.css
+++ b/css/main.css
@@ -776,9 +776,10 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     display: flex;
     flex-direction: row-reverse;
     font-weight: bold;
-    justify-content: space-between;
+    justify-content: start;
     width: 100%;
     padding-right: 12px;
+    gap: 10px;
 }
 
 #show-controls label input {

From bf42b2c3a1175266dcc7c481f589d53805d956f3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 1 Jun 2025 11:02:04 -0700
Subject: [PATCH 0603/1701] Fix thinking blocks sometimes showing a white
 outline

---
 css/main.css | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/css/main.css b/css/main.css
index 71d67ff48f..a9cb36ab8e 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1327,6 +1327,10 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     overflow: hidden;
 }
 
+.thinking-content:focus, .thinking-header:focus {
+    outline: 0 !important;
+}
+
 .dark .thinking-block {
     background-color: var(--darker-gray);
 }

From 7a81beb0c16ff51a90fbe77e6300076714af1fd0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 1 Jun 2025 18:23:23 -0700
Subject: [PATCH 0604/1701] Turn long pasted text into an attachment
 automatically

---
 js/main.js | 40 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 40 insertions(+)

diff --git a/js/main.js b/js/main.js
index 05c19571fc..8090937fa9 100644
--- a/js/main.js
+++ b/js/main.js
@@ -865,6 +865,46 @@ function navigateLastAssistantMessage(direction) {
   return false;
 }
 
+//------------------------------------------------
+// Paste Handler for Long Text
+//------------------------------------------------
+
+const MAX_PLAIN_TEXT_LENGTH = 2500;
+
+function setupPasteHandler() {
+  const textbox = document.querySelector("#chat-input textarea[data-testid=\"textbox\"]");
+  const fileInput = document.querySelector("#chat-input input[data-testid=\"file-upload\"]");
+
+  if (!textbox || !fileInput) {
+    setTimeout(setupPasteHandler, 500);
+    return;
+  }
+
+  textbox.addEventListener("paste", async (event) => {
+    const text = event.clipboardData?.getData("text");
+
+    if (text && text.length > MAX_PLAIN_TEXT_LENGTH) {
+      event.preventDefault();
+
+      const file = new File([text], "pasted_text.txt", {
+        type: "text/plain",
+        lastModified: Date.now()
+      });
+
+      const dataTransfer = new DataTransfer();
+      dataTransfer.items.add(file);
+      fileInput.files = dataTransfer.files;
+      fileInput.dispatchEvent(new Event("change", { bubbles: true }));
+    }
+  });
+}
+
+if (document.readyState === "loading") {
+  document.addEventListener("DOMContentLoaded", setupPasteHandler);
+} else {
+  setupPasteHandler();
+}
+
 //------------------------------------------------
 // Tooltips
 //------------------------------------------------

From 92adceb7b57464ef03886cba5324a32e7d8f8b67 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 1 Jun 2025 19:22:21 -0700
Subject: [PATCH 0605/1701] UI: Fix the model downloader progress bar

---
 download-model.py        |  52 +++++++++++------
 modules/ui_model_menu.py | 119 +++++++++++++++++++++++++++------------
 2 files changed, 117 insertions(+), 54 deletions(-)

diff --git a/download-model.py b/download-model.py
index 255174916c..576a8b7910 100644
--- a/download-model.py
+++ b/download-model.py
@@ -32,6 +32,7 @@ def __init__(self, max_retries=7):
         self.max_retries = max_retries
         self.session = self.get_session()
         self._progress_bar_slots = None
+        self.progress_queue = None
 
     def get_session(self):
         session = requests.Session()
@@ -218,33 +219,45 @@ def get_single_file(self, url, output_folder, start_from_scratch=False):
 
         max_retries = self.max_retries
         attempt = 0
+        file_downloaded_count_for_progress = 0
+
         try:
             while attempt < max_retries:
                 attempt += 1
                 session = self.session
                 headers = {}
                 mode = 'wb'
+                current_file_size_on_disk = 0
 
                 try:
                     if output_path.exists() and not start_from_scratch:
-                        # Resume download
-                        r = session.get(url, stream=True, timeout=20)
-                        total_size = int(r.headers.get('content-length', 0))
-                        if output_path.stat().st_size >= total_size:
+                        current_file_size_on_disk = output_path.stat().st_size
+                        r_head = session.head(url, timeout=20)
+                        r_head.raise_for_status()
+                        total_size = int(r_head.headers.get('content-length', 0))
+
+                        if current_file_size_on_disk >= total_size and total_size > 0:
+                            if self.progress_queue is not None and total_size > 0:
+                                self.progress_queue.put((1.0, str(filename)))
                             return
 
-                        headers = {'Range': f'bytes={output_path.stat().st_size}-'}
+                        headers = {'Range': f'bytes={current_file_size_on_disk}-'}
                         mode = 'ab'
 
                     with session.get(url, stream=True, headers=headers, timeout=30) as r:
-                        r.raise_for_status()  # If status is not 2xx, raise an error
-                        total_size = int(r.headers.get('content-length', 0))
-                        block_size = 1024 * 1024  # 1MB
+                        r.raise_for_status()
+                        total_size_from_stream = int(r.headers.get('content-length', 0))
+                        if mode == 'ab':
+                            effective_total_size = current_file_size_on_disk + total_size_from_stream
+                        else:
+                            effective_total_size = total_size_from_stream
 
-                        filename_str = str(filename)  # Convert PosixPath to string if necessary
+                        block_size = 1024 * 1024
+                        filename_str = str(filename)
 
                         tqdm_kwargs = {
-                            'total': total_size,
+                            'total': effective_total_size,
+                            'initial': current_file_size_on_disk if mode == 'ab' else 0,
                             'unit': 'B',
                             'unit_scale': True,
                             'unit_divisor': 1024,
@@ -261,16 +274,20 @@ def get_single_file(self, url, output_folder, start_from_scratch=False):
                             })
 
                         with open(output_path, mode) as f:
+                            if mode == 'ab':
+                                f.seek(current_file_size_on_disk)
+
                             with tqdm.tqdm(**tqdm_kwargs) as t:
-                                count = 0
+                                file_downloaded_count_for_progress = current_file_size_on_disk
                                 for data in r.iter_content(block_size):
                                     f.write(data)
                                     t.update(len(data))
-                                    if total_size != 0 and self.progress_bar is not None:
-                                        count += len(data)
-                                        self.progress_bar(float(count) / float(total_size), f"{filename_str}")
+                                    if effective_total_size != 0 and self.progress_queue is not None:
+                                        file_downloaded_count_for_progress += len(data)
+                                        progress_fraction = float(file_downloaded_count_for_progress) / float(effective_total_size)
+                                        self.progress_queue.put((progress_fraction, filename_str))
+                        break
 
-                        break  # Exit loop if successful
                 except (RequestException, ConnectionError, Timeout) as e:
                     print(f"Error downloading {filename}: {e}.")
                     print(f"That was attempt {attempt}/{max_retries}.", end=' ')
@@ -295,10 +312,9 @@ def start_download_threads(self, file_list, output_folder, start_from_scratch=Fa
         finally:
             print(f"\nDownload of {len(file_list)} files to {output_folder} completed.")
 
-    def download_model_files(self, model, branch, links, sha256, output_folder, progress_bar=None, start_from_scratch=False, threads=4, specific_file=None, is_llamacpp=False):
-        self.progress_bar = progress_bar
+    def download_model_files(self, model, branch, links, sha256, output_folder, progress_queue=None, start_from_scratch=False, threads=4, specific_file=None, is_llamacpp=False):
+        self.progress_queue = progress_queue
 
-        # Create the folder and writing the metadata
         output_folder.mkdir(parents=True, exist_ok=True)
 
         if not is_llamacpp:
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 862b389359..2a7d3d9df3 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -1,4 +1,6 @@
 import importlib
+import queue
+import threading
 import traceback
 from functools import partial
 from pathlib import Path
@@ -205,48 +207,51 @@ def load_lora_wrapper(selected_loras):
 
 
 def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), return_links=False, check=False):
+    downloader_module = importlib.import_module("download-model")
+    downloader = downloader_module.ModelDownloader()
+    update_queue = queue.Queue()
+
     try:
         # Handle direct GGUF URLs
         if repo_id.startswith("https://") and ("huggingface.co" in repo_id) and (repo_id.endswith(".gguf") or repo_id.endswith(".gguf?download=true")):
             try:
                 path = repo_id.split("huggingface.co/")[1]
-
-                # Extract the repository ID (first two parts of the path)
                 parts = path.split("/")
                 if len(parts) >= 2:
                     extracted_repo_id = f"{parts[0]}/{parts[1]}"
-
-                    # Extract the filename (last part of the path)
-                    filename = repo_id.split("/")[-1]
-                    if "?download=true" in filename:
-                        filename = filename.replace("?download=true", "")
-
+                    filename = repo_id.split("/")[-1].replace("?download=true", "")
                     repo_id = extracted_repo_id
                     specific_file = filename
-            except:
-                pass
-
-        if repo_id == "":
-            yield ("Please enter a model path")
+            except Exception as e:
+                yield f"Error parsing GGUF URL: {e}"
+                progress(0.0)
+                return
+
+        if not repo_id:
+            yield "Please enter a model path."
+            progress(0.0)
             return
 
         repo_id = repo_id.strip()
         specific_file = specific_file.strip()
-        downloader = importlib.import_module("download-model").ModelDownloader()
 
-        progress(0.0)
-        model, branch = downloader.sanitize_model_and_branch_names(repo_id, None)
+        progress(0.0, "Preparing download...")
 
-        yield ("Getting the download links from Hugging Face")
+        model, branch = downloader.sanitize_model_and_branch_names(repo_id, None)
+        yield "Getting download links from Hugging Face..."
         links, sha256, is_lora, is_llamacpp = downloader.get_download_links_from_huggingface(model, branch, text_only=False, specific_file=specific_file)
 
+        if not links:
+            yield "No files found to download for the given model/criteria."
+            progress(0.0)
+            return
+
         # Check for multiple GGUF files
         gguf_files = [link for link in links if link.lower().endswith('.gguf')]
         if len(gguf_files) > 1 and not specific_file:
             output = "Multiple GGUF files found. Please copy one of the following filenames to the 'File name' field:\n\n```\n"
             for link in gguf_files:
                 output += f"{Path(link).name}\n"
-
             output += "```"
             yield output
             return
@@ -255,17 +260,13 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
             output = "```\n"
             for link in links:
                 output += f"{Path(link).name}" + "\n"
-
             output += "```"
             yield output
             return
 
-        yield ("Getting the output folder")
+        yield "Determining output folder..."
         output_folder = downloader.get_output_folder(
-            model,
-            branch,
-            is_lora,
-            is_llamacpp=is_llamacpp,
+            model, branch, is_lora, is_llamacpp=is_llamacpp,
             model_dir=shared.args.model_dir if shared.args.model_dir != shared.args_defaults.model_dir else None
         )
 
@@ -275,19 +276,65 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
             output_folder = Path(shared.args.lora_dir)
 
         if check:
-            progress(0.5)
-
-            yield ("Checking previously downloaded files")
+            yield "Checking previously downloaded files..."
+            progress(0.5, "Verifying files...")
             downloader.check_model_files(model, branch, links, sha256, output_folder)
-            progress(1.0)
-        else:
-            yield (f"Downloading file{'s' if len(links) > 1 else ''} to `{output_folder}/`")
-            downloader.download_model_files(model, branch, links, sha256, output_folder, progress_bar=progress, threads=4, is_llamacpp=is_llamacpp)
-
-            yield (f"Model successfully saved to `{output_folder}/`.")
-    except:
-        progress(1.0)
-        yield traceback.format_exc().replace('\n', '\n\n')
+            progress(1.0, "Verification complete.")
+            yield "File check complete."
+            return
+
+        yield ""
+        progress(0.0, "Download starting...")
+
+        def downloader_thread_target():
+            try:
+                downloader.download_model_files(
+                    model, branch, links, sha256, output_folder,
+                    progress_queue=update_queue,
+                    threads=4,
+                    is_llamacpp=is_llamacpp,
+                    specific_file=specific_file
+                )
+                update_queue.put(("COMPLETED", f"Model successfully saved to `{output_folder}/`."))
+            except Exception as e:
+                tb_str = traceback.format_exc().replace('\n', '\n\n')
+                update_queue.put(("ERROR", tb_str))
+
+        download_thread = threading.Thread(target=downloader_thread_target)
+        download_thread.start()
+
+        while True:
+            try:
+                message = update_queue.get(timeout=0.2)
+                if not isinstance(message, tuple) or len(message) != 2:
+                    continue
+
+                msg_identifier, data = message
+
+                if msg_identifier == "COMPLETED":
+                    progress(1.0, "Download complete!")
+                    yield data
+                    break
+                elif msg_identifier == "ERROR":
+                    progress(0.0, "Error occurred")
+                    yield data
+                    break
+                elif isinstance(msg_identifier, float):
+                    progress_value = msg_identifier
+                    description_str = data
+                    progress(progress_value, f"Downloading: {description_str}")
+
+            except queue.Empty:
+                if not download_thread.is_alive():
+                    yield "Download process finished."
+                    break
+
+        download_thread.join()
+
+    except Exception as e:
+        progress(0.0)
+        tb_str = traceback.format_exc().replace('\n', '\n\n')
+        yield tb_str
 
 
 def update_truncation_length(current_length, state):

From ad6d0218ae0c015694bef7a43f5f628d281a1c36 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 1 Jun 2025 19:27:14 -0700
Subject: [PATCH 0606/1701] Fix after 219f0a773166deeb0326c2874b29e66e382df524

---
 modules/models.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/models.py b/modules/models.py
index d329ae3c32..c1e7fb5695 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -116,7 +116,7 @@ def unload_model(keep_model_name=False):
         return
 
     is_llamacpp = (shared.model.__class__.__name__ == 'LlamaServer')
-    if shared.args.loader == 'ExLlamav3_HF':
+    if shared.model.__class__.__name__ == 'Exllamav3HF':
         shared.model.unload()
 
     shared.model = shared.tokenizer = None

From 2db7745cbde543d7e1abd81c0389c544c84621db Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 1 Jun 2025 22:12:24 -0700
Subject: [PATCH 0607/1701] Show llama.cpp prompt processing on one line
 instead of many lines

---
 modules/llama_cpp_server.py | 21 +++++++++++++++++++--
 1 file changed, 19 insertions(+), 2 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index d695c74e9c..aa7125415a 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -409,14 +409,31 @@ def stop(self):
 
 def filter_stderr_with_progress(process_stderr):
     progress_pattern = re.compile(r'slot update_slots: id.*progress = (\d+\.\d+)')
+    last_was_progress = False
+
     try:
         for line in iter(process_stderr.readline, ''):
+            line = line.rstrip('\n\r')  # Remove existing newlines
             progress_match = progress_pattern.search(line)
+
             if progress_match:
-                sys.stderr.write(line)
+                if last_was_progress:
+                    # Overwrite the previous progress line using carriage return
+                    sys.stderr.write(f'\r{line}')
+                else:
+                    # First progress line - print normally
+                    sys.stderr.write(line)
                 sys.stderr.flush()
+                last_was_progress = True
             elif not line.startswith(('srv ', 'slot ')) and 'log_server_r: request: GET /health' not in line:
-                sys.stderr.write(line)
+                if last_was_progress:
+                    # Finish the progress line with a newline, then print the new line
+                    sys.stderr.write(f'\n{line}\n')
+                else:
+                    # Normal line - print with newline
+                    sys.stderr.write(f'{line}\n')
                 sys.stderr.flush()
+                last_was_progress = False
+            # For filtered lines, don't change last_was_progress state
     except (ValueError, IOError):
         pass

From 45c9ae312c1ff60ce13c721d1290b65f01bf9660 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 1 Jun 2025 22:17:22 -0700
Subject: [PATCH 0608/1701] Use the flash-attention wheels in
 https://github.com/kingbri1/flash-attention

---
 requirements/full/requirements.txt        | 2 +-
 requirements/full/requirements_noavx2.txt | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index e61677a618..04d972201f 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -41,5 +41,5 @@ https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/oobabooga/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index e0cb84b480..7c3635cc9d 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -41,5 +41,5 @@ https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/oobabooga/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

From bb409c926e986e57b8c3eea3582abb466f32ad08 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Mon, 2 Jun 2025 09:50:17 -0300
Subject: [PATCH 0609/1701] Update only the last message during streaming + add
 back dynamic UI update speed (#7038)

---
 js/global_scope_js.js            |  25 ++-
 modules/chat.py                  |   4 +-
 modules/html_generator.py        | 284 ++++++++++++++++---------------
 modules/shared.py                |   3 +-
 modules/text_generation.py       |  18 +-
 modules/ui.py                    |   6 +-
 modules/ui_chat.py               |   4 +-
 modules/ui_parameters.py         |   2 -
 user_data/settings-template.yaml |   1 -
 9 files changed, 186 insertions(+), 161 deletions(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 3274f47eea..d5140c9377 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -229,10 +229,23 @@ function removeLastClick() {
   document.getElementById("Remove-last").click();
 }
 
-function handleMorphdomUpdate(text) {
+function handleMorphdomUpdate(data) {
+  // Determine target element and use it as query scope
+  var target_element, target_html;
+  if (data.last_message_only) {
+    const childNodes = document.getElementsByClassName("messages")[0].childNodes;
+    target_element = childNodes[childNodes.length - 1];
+    target_html = data.html;
+  } else {
+    target_element = document.getElementById("chat").parentNode;
+    target_html =  "
            " + data.html + "
            ";
+  }
+
+  const queryScope = target_element;
+
   // Track open blocks
   const openBlocks = new Set();
-  document.querySelectorAll(".thinking-block").forEach(block => {
+  queryScope.querySelectorAll(".thinking-block").forEach(block => {
     const blockId = block.getAttribute("data-block-id");
     // If block exists and is open, add to open set
     if (blockId && block.hasAttribute("open")) {
@@ -242,7 +255,7 @@ function handleMorphdomUpdate(text) {
 
   // Store scroll positions for any open blocks
   const scrollPositions = {};
-  document.querySelectorAll(".thinking-block[open]").forEach(block => {
+  queryScope.querySelectorAll(".thinking-block[open]").forEach(block => {
     const content = block.querySelector(".thinking-content");
     const blockId = block.getAttribute("data-block-id");
     if (content && blockId) {
@@ -255,8 +268,8 @@ function handleMorphdomUpdate(text) {
   });
 
   morphdom(
-    document.getElementById("chat").parentNode,
-    "
            " + text + "
            ",
+    target_element,
+    target_html,
     {
       onBeforeElUpdated: function(fromEl, toEl) {
         // Preserve code highlighting
@@ -307,7 +320,7 @@ function handleMorphdomUpdate(text) {
   );
 
   // Add toggle listeners for new blocks
-  document.querySelectorAll(".thinking-block").forEach(block => {
+  queryScope.querySelectorAll(".thinking-block").forEach(block => {
     if (!block._hasToggleListener) {
       block.addEventListener("toggle", function(e) {
         if (this.open) {
diff --git a/modules/chat.py b/modules/chat.py
index 1222d2bb95..f1ea16f10a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -825,7 +825,9 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
     last_save_time = time.monotonic()
     save_interval = 8
     for i, history in enumerate(generate_chat_reply(text, state, regenerate, _continue, loading_message=True, for_ui=True)):
-        yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu']), history
+        yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'], last_message_only=(i > 0)), history
+        if i == 0:
+            time.sleep(0.125)  # We need this to make sure the first update goes through
 
         current_time = time.monotonic()
         # Save on first iteration or if save_interval seconds have passed
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 03b5d485d8..f90e3b04db 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -462,64 +462,69 @@ def actions_html(history, i, role, info_message=""):
             f'{version_nav_html}')
 
 
-def generate_instruct_html(history):
-    output = f'
            '
+def generate_instruct_html(history, last_message_only=False):
+    if not last_message_only:
+        output = f'
            '
+    else:
+        output = ""
 
-    for i in range(len(history['visible'])):
-        row_visible = history['visible'][i]
-        row_internal = history['internal'][i]
-        converted_visible = [convert_to_markdown_wrapped(entry, message_id=i, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
-
-        # Get timestamps
-        user_timestamp = format_message_timestamp(history, "user", i)
-        assistant_timestamp = format_message_timestamp(history, "assistant", i)
-
-        # Get attachments
-        user_attachments = format_message_attachments(history, "user", i)
-        assistant_attachments = format_message_attachments(history, "assistant", i)
-
-        # Create info buttons for timestamps if they exist
-        info_message_user = ""
-        if user_timestamp != "":
-            tooltip_text = get_message_tooltip(history, "user", i)
-            info_message_user = info_button.replace('title="message"', f'title="{html.escape(tooltip_text)}"')
-
-        info_message_assistant = ""
-        if assistant_timestamp != "":
-            tooltip_text = get_message_tooltip(history, "assistant", i)
-            info_message_assistant = info_button.replace('title="message"', f'title="{html.escape(tooltip_text)}"')
-
-        if converted_visible[0]:  # Don't display empty user messages
-            output += (
-                f'
            '
-                f'
            '
-                f'
            {converted_visible[0]}
            '
-                f'{user_attachments}'
-                f'{actions_html(history, i, "user", info_message_user)}'
-                f'
            '
-                f'
            '
-            )
+    def create_message(role, content, raw_content):
+        """Inner function that captures variables from outer scope."""
+        class_name = "user-message" if role == "user" else "assistant-message"
+
+        # Get role-specific data
+        timestamp = format_message_timestamp(history, role, i)
+        attachments = format_message_attachments(history, role, i)
 
-        output += (
-            f'
            '
             f'
            '
-            f'
            {converted_visible[1]}
            '
-            f'{assistant_attachments}'
-            f'{actions_html(history, i, "assistant", info_message_assistant)}'
+            f'
            {content}
            '
+            f'{attachments}'
+            f'{actions_html(history, i, role, info_message)}'
             f'
            '
             f'
            '
         )
 
-    output += "
            "
+    # Determine range
+    start_idx = len(history['visible']) - 1 if last_message_only else 0
+    end_idx = len(history['visible'])
+
+    for i in range(start_idx, end_idx):
+        row_visible = history['visible'][i]
+        row_internal = history['internal'][i]
+
+        # Convert content
+        if last_message_only:
+            converted_visible = [None, convert_to_markdown_wrapped(row_visible[1], message_id=i, use_cache=i != len(history['visible']) - 1)]
+        else:
+            converted_visible = [convert_to_markdown_wrapped(entry, message_id=i, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
+
+        # Generate messages
+        if not last_message_only and converted_visible[0]:
+            output += create_message("user", converted_visible[0], row_internal[0])
+
+        output += create_message("assistant", converted_visible[1], row_internal[1])
+
+    if not last_message_only:
+        output += "
            "
+
     return output
 
 
-def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=False):
-    output = f'
            '
+def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=False, last_message_only=False):
+    if not last_message_only:
+        output = f'
            '
+    else:
+        output = ""
 
     # We use ?character and ?time.time() to force the browser to reset caches
     img_bot = (
@@ -527,110 +532,117 @@ def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=
         if Path("user_data/cache/pfp_character_thumb.png").exists() else ''
     )
 
-    img_me = (
-        f''
-        if Path("user_data/cache/pfp_me.png").exists() else ''
-    )
+    def create_message(role, content, raw_content):
+        """Inner function for CAI-style messages."""
+        circle_class = "circle-you" if role == "user" else "circle-bot"
+        name = name1 if role == "user" else name2
 
-    for i in range(len(history['visible'])):
-        row_visible = history['visible'][i]
-        row_internal = history['internal'][i]
-        converted_visible = [convert_to_markdown_wrapped(entry, message_id=i, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
-
-        # Get timestamps
-        user_timestamp = format_message_timestamp(history, "user", i, tooltip_include_timestamp=False)
-        assistant_timestamp = format_message_timestamp(history, "assistant", i, tooltip_include_timestamp=False)
-
-        # Get attachments
-        user_attachments = format_message_attachments(history, "user", i)
-        assistant_attachments = format_message_attachments(history, "assistant", i)
-
-        if converted_visible[0]:  # Don't display empty user messages
-            output += (
-                f'
            '
-                f'
            {img_me}
            '
-                f'
            '
-                f'
            {name1}{user_timestamp}
            '
-                f'
            {converted_visible[0]}
            '
-                f'{user_attachments}'
-                f'{actions_html(history, i, "user")}'
-                f'
            '
-                f'
            '
-            )
+        # Get role-specific data
+        timestamp = format_message_timestamp(history, role, i, tooltip_include_timestamp=False)
+        attachments = format_message_attachments(history, role, i)
+
+        # Get appropriate image
+        if role == "user":
+            img = (f''
+                   if Path("user_data/cache/pfp_me.png").exists() else '')
+        else:
+            img = img_bot
 
-        output += (
+        return (
             f'
            '
-            f'
            {img_bot}
            '
+            f'
            {img}
            '
             f'
            '
-            f'
            {name2}{assistant_timestamp}
            '
-            f'
            {converted_visible[1]}
            '
-            f'{assistant_attachments}'
-            f'{actions_html(history, i, "assistant")}'
+            f'
            {name}{timestamp}
            '
+            f'
            {content}
            '
+            f'{attachments}'
+            f'{actions_html(history, i, role)}'
             f'
            '
             f'
            '
         )
 
-    output += "
            "
+    # Determine range
+    start_idx = len(history['visible']) - 1 if last_message_only else 0
+    end_idx = len(history['visible'])
+
+    for i in range(start_idx, end_idx):
+        row_visible = history['visible'][i]
+        row_internal = history['internal'][i]
+
+        # Convert content
+        if last_message_only:
+            converted_visible = [None, convert_to_markdown_wrapped(row_visible[1], message_id=i, use_cache=i != len(history['visible']) - 1)]
+        else:
+            converted_visible = [convert_to_markdown_wrapped(entry, message_id=i, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
+
+        # Generate messages
+        if not last_message_only and converted_visible[0]:
+            output += create_message("user", converted_visible[0], row_internal[0])
+
+        output += create_message("assistant", converted_visible[1], row_internal[1])
+
+    if not last_message_only:
+        output += "
            "
+
     return output
 
 
-def generate_chat_html(history, name1, name2, reset_cache=False):
-    output = f'
            '
+def generate_chat_html(history, name1, name2, reset_cache=False, last_message_only=False):
+    if not last_message_only:
+        output = f'
            '
+    else:
+        output = ""
 
-    for i in range(len(history['visible'])):
-        row_visible = history['visible'][i]
-        row_internal = history['internal'][i]
-        converted_visible = [convert_to_markdown_wrapped(entry, message_id=i, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
-
-        # Get timestamps
-        user_timestamp = format_message_timestamp(history, "user", i)
-        assistant_timestamp = format_message_timestamp(history, "assistant", i)
-
-        # Get attachments
-        user_attachments = format_message_attachments(history, "user", i)
-        assistant_attachments = format_message_attachments(history, "assistant", i)
-
-        # Create info buttons for timestamps if they exist
-        info_message_user = ""
-        if user_timestamp != "":
-            tooltip_text = get_message_tooltip(history, "user", i)
-            info_message_user = info_button.replace('title="message"', f'title="{html.escape(tooltip_text)}"')
-
-        info_message_assistant = ""
-        if assistant_timestamp != "":
-            tooltip_text = get_message_tooltip(history, "assistant", i)
-            info_message_assistant = info_button.replace('title="message"', f'title="{html.escape(tooltip_text)}"')
-
-        if converted_visible[0]:  # Don't display empty user messages
-            output += (
-                f'
            '
-                f'
            '
-                f'
            {converted_visible[0]}
            '
-                f'{user_attachments}'
-                f'{actions_html(history, i, "user", info_message_user)}'
-                f'
            '
-                f'
            '
-            )
+    def create_message(role, content, raw_content):
+        """Inner function for WPP-style messages."""
+        text_class = "text-you" if role == "user" else "text-bot"
+
+        # Get role-specific data
+        timestamp = format_message_timestamp(history, role, i)
+        attachments = format_message_attachments(history, role, i)
+
+        # Create info button if timestamp exists
+        info_message = ""
+        if timestamp:
+            tooltip_text = get_message_tooltip(history, role, i)
+            info_message = info_button.replace('title="message"', f'title="{html.escape(tooltip_text)}"')
 
-        output += (
+        return (
             f'
            '
-            f'
            '
-            f'
            {converted_visible[1]}
            '
-            f'{assistant_attachments}'
-            f'{actions_html(history, i, "assistant", info_message_assistant)}'
+            f'
            '
+            f'
            {content}
            '
+            f'{attachments}'
+            f'{actions_html(history, i, role, info_message)}'
             f'
            '
             f'
            '
         )
 
-    output += "
            "
+    # Determine range
+    start_idx = len(history['visible']) - 1 if last_message_only else 0
+    end_idx = len(history['visible'])
+
+    for i in range(start_idx, end_idx):
+        row_visible = history['visible'][i]
+        row_internal = history['internal'][i]
+
+        # Convert content
+        if last_message_only:
+            converted_visible = [None, convert_to_markdown_wrapped(row_visible[1], message_id=i, use_cache=i != len(history['visible']) - 1)]
+        else:
+            converted_visible = [convert_to_markdown_wrapped(entry, message_id=i, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
+
+        # Generate messages
+        if not last_message_only and converted_visible[0]:
+            output += create_message("user", converted_visible[0], row_internal[0])
+
+        output += create_message("assistant", converted_visible[1], row_internal[1])
+
+    if not last_message_only:
+        output += "
            "
+
     return output
 
 
@@ -644,15 +656,15 @@ def time_greeting():
         return "Good evening!"
 
 
-def chat_html_wrapper(history, name1, name2, mode, style, character, reset_cache=False):
+def chat_html_wrapper(history, name1, name2, mode, style, character, reset_cache=False, last_message_only=False):
     if len(history['visible']) == 0:
         greeting = f"
            {time_greeting()} How can I help you today?
            "
         result = f'
            {greeting}
            '
     elif mode == 'instruct':
-        result = generate_instruct_html(history)
+        result = generate_instruct_html(history, last_message_only=last_message_only)
     elif style == 'wpp':
-        result = generate_chat_html(history, name1, name2)
+        result = generate_chat_html(history, name1, name2, last_message_only=last_message_only)
     else:
-        result = generate_cai_chat_html(history, name1, name2, style, character, reset_cache)
+        result = generate_cai_chat_html(history, name1, name2, style, character, reset_cache=reset_cache, last_message_only=last_message_only)
 
-    return {'html': result}
+    return {'html': result, 'last_message_only': last_message_only}
diff --git a/modules/shared.py b/modules/shared.py
index d2305f305b..f712f7f8ac 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -21,7 +21,7 @@
 # Generation variables
 stop_everything = False
 generation_lock = None
-processing_message = '*Is typing...*'
+processing_message = ''
 
 # UI variables
 gradio = {}
@@ -47,7 +47,6 @@
     'max_new_tokens_max': 4096,
     'prompt_lookup_num_tokens': 0,
     'max_tokens_second': 0,
-    'max_updates_second': 12,
     'auto_max_new_tokens': True,
     'ban_eos_token': False,
     'add_bos_token': True,
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 1fd6d810f7..0d499d5070 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -65,41 +65,39 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
             all_stop_strings += st
 
     shared.stop_everything = False
-    last_update = -1
     reply = ''
     is_stream = state['stream']
     if len(all_stop_strings) > 0 and not state['stream']:
         state = copy.deepcopy(state)
         state['stream'] = True
 
-    min_update_interval = 0
-    if state.get('max_updates_second', 0) > 0:
-        min_update_interval = 1 / state['max_updates_second']
-
     # Generate
+    last_update = -1
+    latency_threshold = 1 / 1000
     for reply in generate_func(question, original_question, state, stopping_strings, is_chat=is_chat):
+        cur_time = time.monotonic()
         reply, stop_found = apply_stopping_strings(reply, all_stop_strings)
         if escape_html:
             reply = html.escape(reply)
 
         if is_stream:
-            cur_time = time.time()
-
             # Limit number of tokens/second to make text readable in real time
             if state['max_tokens_second'] > 0:
                 diff = 1 / state['max_tokens_second'] - (cur_time - last_update)
                 if diff > 0:
                     time.sleep(diff)
 
-                last_update = time.time()
+                last_update = time.monotonic()
                 yield reply
 
             # Limit updates to avoid lag in the Gradio UI
             # API updates are not limited
             else:
-                if cur_time - last_update > min_update_interval:
-                    last_update = cur_time
+                # If 'generate_func' takes less than 0.001 seconds to yield the next token
+                # (equivalent to more than 1000 tok/s), assume that the UI is lagging behind and skip yielding
+                if (cur_time - last_update) > latency_threshold:
                     yield reply
+                last_update = time.monotonic()
 
         if stop_found or (state['max_tokens_second'] > 0 and shared.stop_everything):
             break
diff --git a/modules/ui.py b/modules/ui.py
index 9f4d67cb73..14a09d2b7a 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -6,6 +6,7 @@
 
 import extensions
 from modules import shared
+from modules.chat import load_history
 
 with open(Path(__file__).resolve().parent / '../css/NotoSans/stylesheet.css', 'r') as f:
     css = f.read()
@@ -194,7 +195,6 @@ def list_interface_input_elements():
         'max_new_tokens',
         'prompt_lookup_num_tokens',
         'max_tokens_second',
-        'max_updates_second',
         'do_sample',
         'dynamic_temperature',
         'temperature_last',
@@ -270,6 +270,10 @@ def gather_interface_values(*args):
     if not shared.args.multi_user:
         shared.persistent_interface_state = output
 
+    # Prevent history loss if backend is restarted but UI is not refreshed
+    if output['history'] is None and output['unique_id'] is not None:
+        output['history'] = load_history(output['unique_id'], output['character_menu'], output['mode'])
+
     return output
 
 
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 822b77b8e5..0d5a2c1817 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -18,7 +18,7 @@ def create_ui():
     mu = shared.args.multi_user
 
     shared.gradio['Chat input'] = gr.State()
-    shared.gradio['history'] = gr.JSON(visible=False)
+    shared.gradio['history'] = gr.State({'internal': [], 'visible': [], 'metadata': {}})
 
     with gr.Tab('Chat', id='Chat', elem_id='chat-tab'):
         with gr.Row(elem_id='past-chats-row', elem_classes=['pretty_scrollbar']):
@@ -195,7 +195,7 @@ def create_event_handlers():
     shared.reload_inputs = gradio(reload_arr)
 
     # Morph HTML updates instead of updating everything
-    shared.gradio['display'].change(None, gradio('display'), None, js="(data) => handleMorphdomUpdate(data.html)")
+    shared.gradio['display'].change(None, gradio('display'), None, js="(data) => handleMorphdomUpdate(data)")
 
     shared.gradio['Generate'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 733d090101..84f9fbfc24 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -71,8 +71,6 @@ def create_ui(default_preset):
                             shared.gradio['max_new_tokens'] = gr.Slider(minimum=shared.settings['max_new_tokens_min'], maximum=shared.settings['max_new_tokens_max'], value=shared.settings['max_new_tokens'], step=1, label='max_new_tokens', info='⚠️ Setting this too high can cause prompt truncation.')
                             shared.gradio['prompt_lookup_num_tokens'] = gr.Slider(value=shared.settings['prompt_lookup_num_tokens'], minimum=0, maximum=10, step=1, label='prompt_lookup_num_tokens', info='Activates Prompt Lookup Decoding.')
                             shared.gradio['max_tokens_second'] = gr.Slider(value=shared.settings['max_tokens_second'], minimum=0, maximum=20, step=1, label='Maximum tokens/second', info='To make text readable in real time.')
-                            shared.gradio['max_updates_second'] = gr.Slider(value=shared.settings['max_updates_second'], minimum=0, maximum=24, step=1, label='Maximum UI updates/second', info='Set this if you experience lag in the UI during streaming.')
-
                 with gr.Column():
                     with gr.Row():
                         with gr.Column():
diff --git a/user_data/settings-template.yaml b/user_data/settings-template.yaml
index ce0f77e142..db481e8499 100644
--- a/user_data/settings-template.yaml
+++ b/user_data/settings-template.yaml
@@ -18,7 +18,6 @@ max_new_tokens_min: 1
 max_new_tokens_max: 4096
 prompt_lookup_num_tokens: 0
 max_tokens_second: 0
-max_updates_second: 12
 auto_max_new_tokens: true
 ban_eos_token: false
 add_bos_token: true

From 7278548cd18a9ba05062eb2db59d7f2965d8a9f6 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Mon, 2 Jun 2025 09:57:55 -0300
Subject: [PATCH 0610/1701] Simplify the one-click installer (#7039)

---
 one_click.py | 253 +++++++++++++++++++++++++--------------------------
 1 file changed, 123 insertions(+), 130 deletions(-)

diff --git a/one_click.py b/one_click.py
index 482a6aa9f1..cccb0dc94f 100644
--- a/one_click.py
+++ b/one_click.py
@@ -70,12 +70,8 @@ def is_installed():
 def cpu_has_avx2():
     try:
         import cpuinfo
-
         info = cpuinfo.get_cpu_info()
-        if 'avx2' in info['flags']:
-            return True
-        else:
-            return False
+        return 'avx2' in info['flags']
     except:
         return True
 
@@ -83,30 +79,112 @@ def cpu_has_avx2():
 def cpu_has_amx():
     try:
         import cpuinfo
-
         info = cpuinfo.get_cpu_info()
-        if 'amx' in info['flags']:
-            return True
-        else:
-            return False
+        return 'amx' in info['flags']
     except:
         return True
 
 
-def torch_version():
-    site_packages_path = None
-    for sitedir in site.getsitepackages():
-        if "site-packages" in sitedir and conda_env_path in sitedir:
-            site_packages_path = sitedir
-            break
+def load_state():
+    """Load installer state from JSON file"""
+    if os.path.exists(state_file):
+        try:
+            with open(state_file, 'r') as f:
+                return json.load(f)
+        except:
+            return {}
+    return {}
 
-    if site_packages_path:
-        torch_version_file = open(os.path.join(site_packages_path, 'torch', 'version.py')).read().splitlines()
-        torver = [line for line in torch_version_file if line.startswith('__version__')][0].split('__version__ = ')[1].strip("'")
+
+def save_state(state):
+    """Save installer state to JSON file"""
+    with open(state_file, 'w') as f:
+        json.dump(state, f)
+
+
+def get_gpu_choice():
+    """Get GPU choice from state file or ask user"""
+    state = load_state()
+    gpu_choice = state.get('gpu_choice')
+
+    if not gpu_choice:
+        if "GPU_CHOICE" in os.environ:
+            choice = os.environ["GPU_CHOICE"].upper()
+            print_big_message(f"Selected GPU choice \"{choice}\" based on the GPU_CHOICE environment variable.")
+        else:
+            choice = get_user_choice(
+                "What is your GPU?",
+                {
+                    'A': 'NVIDIA - CUDA 12.4',
+                    'B': 'AMD - Linux/macOS only, requires ROCm 6.2.4',
+                    'C': 'Apple M Series',
+                    'D': 'Intel Arc (beta)',
+                    'N': 'CPU mode'
+                },
+            )
+
+        # Convert choice to GPU name
+        gpu_choice = {"A": "NVIDIA", "B": "AMD", "C": "APPLE", "D": "INTEL", "N": "NONE"}[choice]
+
+        # Save choice to state
+        state['gpu_choice'] = gpu_choice
+        save_state(state)
+
+    return gpu_choice
+
+
+def get_pytorch_install_command(gpu_choice):
+    """Get PyTorch installation command based on GPU choice"""
+    base_cmd = f"python -m pip install torch=={TORCH_VERSION} torchvision=={TORCHVISION_VERSION} torchaudio=={TORCHAUDIO_VERSION} "
+
+    if gpu_choice == "NVIDIA":
+        return base_cmd + "--index-url https://download.pytorch.org/whl/cu124"
+    elif gpu_choice == "AMD":
+        return base_cmd + "--index-url https://download.pytorch.org/whl/rocm6.2.4"
+    elif gpu_choice in ["APPLE", "NONE"]:
+        return base_cmd + "--index-url https://download.pytorch.org/whl/cpu"
+    elif gpu_choice == "INTEL":
+        if is_linux():
+            return "python -m pip install torch==2.1.0a0 torchvision==0.16.0a0 torchaudio==2.1.0a0 intel-extension-for-pytorch==2.1.10+xpu --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
+        else:
+            return "python -m pip install torch==2.1.0a0 torchvision==0.16.0a0 torchaudio==2.1.0a0 intel-extension-for-pytorch==2.1.10 --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
     else:
-        from torch import __version__ as torver
+        return base_cmd
 
-    return torver
+
+def get_pytorch_update_command(gpu_choice):
+    """Get PyTorch update command based on GPU choice"""
+    base_cmd = f"python -m pip install --upgrade torch=={TORCH_VERSION} torchvision=={TORCHVISION_VERSION} torchaudio=={TORCHAUDIO_VERSION}"
+
+    if gpu_choice == "NVIDIA":
+        return f"{base_cmd} --index-url https://download.pytorch.org/whl/cu124"
+    elif gpu_choice == "AMD":
+        return f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.2.4"
+    elif gpu_choice in ["APPLE", "NONE"]:
+        return f"{base_cmd} --index-url https://download.pytorch.org/whl/cpu"
+    elif gpu_choice == "INTEL":
+        intel_extension = "intel-extension-for-pytorch==2.1.10+xpu" if is_linux() else "intel-extension-for-pytorch==2.1.10"
+        return f"{base_cmd} {intel_extension} --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
+    else:
+        return base_cmd
+
+
+def get_requirements_file(gpu_choice):
+    """Get requirements file path based on GPU choice"""
+    requirements_base = os.path.join("requirements", "full")
+
+    if gpu_choice == "AMD":
+        file_name = f"requirements_amd{'_noavx2' if not cpu_has_avx2() else ''}.txt"
+    elif gpu_choice == "APPLE":
+        file_name = f"requirements_apple_{'intel' if is_x86_64() else 'silicon'}.txt"
+    elif gpu_choice in ["INTEL", "NONE"]:
+        file_name = f"requirements_cpu_only{'_noavx2' if not cpu_has_avx2() else ''}.txt"
+    elif gpu_choice == "NVIDIA":
+        file_name = f"requirements{'_noavx2' if not cpu_has_avx2() else ''}.txt"
+    else:
+        raise ValueError(f"Unknown GPU choice: {gpu_choice}")
+
+    return os.path.join(requirements_base, file_name)
 
 
 def get_current_commit():
@@ -209,28 +287,8 @@ def get_user_choice(question, options_dict):
 
 def update_pytorch_and_python():
     print_big_message("Checking for PyTorch updates.")
-
-    # Update the Python version. Left here for future reference in case this becomes necessary.
-    # print_big_message("Checking for PyTorch and Python updates.")
-    # current_python_version = f"{sys.version_info.major}.{sys.version_info.minor}"
-    # if current_python_version != PYTHON_VERSION:
-    #     run_cmd(f"conda install -y python={PYTHON_VERSION}", assert_success=True, environment=True)
-
-    torver = torch_version()
-    base_cmd = f"python -m pip install --upgrade torch=={TORCH_VERSION} torchvision=={TORCHVISION_VERSION} torchaudio=={TORCHAUDIO_VERSION}"
-
-    if "+cu" in torver:
-        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/cu124"
-    elif "+rocm" in torver:
-        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.2.4"
-    elif "+cpu" in torver:
-        install_cmd = f"{base_cmd} --index-url https://download.pytorch.org/whl/cpu"
-    elif "+cxx11" in torver:
-        intel_extension = "intel-extension-for-pytorch==2.1.10+xpu" if is_linux() else "intel-extension-for-pytorch==2.1.10"
-        install_cmd = f"{base_cmd} {intel_extension} --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
-    else:
-        install_cmd = base_cmd
-
+    gpu_choice = get_gpu_choice()
+    install_cmd = get_pytorch_update_command(gpu_choice)
     run_cmd(install_cmd, assert_success=True, environment=True)
 
 
@@ -256,43 +314,11 @@ def install_webui():
     if os.path.isfile(state_file):
         os.remove(state_file)
 
-    # Ask the user for the GPU vendor
-    if "GPU_CHOICE" in os.environ:
-        choice = os.environ["GPU_CHOICE"].upper()
-        print_big_message(f"Selected GPU choice \"{choice}\" based on the GPU_CHOICE environment variable.")
-
-        # Warn about changed meanings and handle old choices
-        if choice == "B":
-            print_big_message("Warning: GPU_CHOICE='B' now means 'AMD' in the new version.")
-        elif choice == "C":
-            print_big_message("Warning: GPU_CHOICE='C' now means 'Apple M Series' in the new version.")
-        elif choice == "D":
-            print_big_message("Warning: GPU_CHOICE='D' now means 'Intel Arc' in the new version.")
-    else:
-        choice = get_user_choice(
-            "What is your GPU?",
-            {
-                'A': 'NVIDIA - CUDA 12.4',
-                'B': 'AMD - Linux/macOS only, requires ROCm 6.2.4',
-                'C': 'Apple M Series',
-                'D': 'Intel Arc (beta)',
-                'N': 'CPU mode'
-            },
-        )
-
-    # Convert choices to GPU names for compatibility
-    gpu_choice_to_name = {
-        "A": "NVIDIA",
-        "B": "AMD",
-        "C": "APPLE",
-        "D": "INTEL",
-        "N": "NONE"
-    }
-
-    selected_gpu = gpu_choice_to_name[choice]
+    # Get GPU choice and save it to state
+    gpu_choice = get_gpu_choice()
 
     # Write a flag to CMD_FLAGS.txt for CPU mode
-    if selected_gpu == "NONE":
+    if gpu_choice == "NONE":
         cmd_flags_path = os.path.join(script_dir, "user_data", "CMD_FLAGS.txt")
         with open(cmd_flags_path, 'r+') as cmd_flags_file:
             if "--cpu" not in cmd_flags_file.read():
@@ -300,34 +326,20 @@ def install_webui():
                 cmd_flags_file.write("\n--cpu\n")
 
     # Handle CUDA version display
-    elif any((is_windows(), is_linux())) and selected_gpu == "NVIDIA":
+    elif any((is_windows(), is_linux())) and gpu_choice == "NVIDIA":
         print("CUDA: 12.4")
 
     # No PyTorch for AMD on Windows (?)
-    elif is_windows() and selected_gpu == "AMD":
+    elif is_windows() and gpu_choice == "AMD":
         print("PyTorch setup on Windows is not implemented yet. Exiting...")
         sys.exit(1)
 
-    # Find the Pytorch installation command
-    install_pytorch = f"python -m pip install torch=={TORCH_VERSION} torchvision=={TORCHVISION_VERSION} torchaudio=={TORCHAUDIO_VERSION} "
-
-    if selected_gpu == "NVIDIA":
-        install_pytorch += "--index-url https://download.pytorch.org/whl/cu124"
-    elif selected_gpu == "AMD":
-        install_pytorch += "--index-url https://download.pytorch.org/whl/rocm6.2.4"
-    elif selected_gpu in ["APPLE", "NONE"]:
-        install_pytorch += "--index-url https://download.pytorch.org/whl/cpu"
-    elif selected_gpu == "INTEL":
-        if is_linux():
-            install_pytorch = "python -m pip install torch==2.1.0a0 torchvision==0.16.0a0 torchaudio==2.1.0a0 intel-extension-for-pytorch==2.1.10+xpu --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
-        else:
-            install_pytorch = "python -m pip install torch==2.1.0a0 torchvision==0.16.0a0 torchaudio==2.1.0a0 intel-extension-for-pytorch==2.1.10 --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
-
     # Install Git and then Pytorch
     print_big_message("Installing PyTorch.")
+    install_pytorch = get_pytorch_install_command(gpu_choice)
     run_cmd(f"conda install -y ninja git && {install_pytorch} && python -m pip install py-cpuinfo==9.0.0", assert_success=True, environment=True)
 
-    if selected_gpu == "INTEL":
+    if gpu_choice == "INTEL":
         # Install oneAPI dependencies via conda
         print_big_message("Installing Intel oneAPI runtime libraries.")
         run_cmd("conda install -y -c https://software.repos.intel.com/python/conda/ -c conda-forge dpcpp-cpp-rt=2024.0 mkl-dpcpp=2024.0", environment=True)
@@ -349,31 +361,15 @@ def update_requirements(initial_installation=False, pull=True):
             assert_success=True
         )
 
-    torver = torch_version()
-    requirements_base = os.path.join("requirements", "full")
-
-    if "+rocm" in torver:
-        file_name = f"requirements_amd{'_noavx2' if not cpu_has_avx2() else ''}.txt"
-    elif "+cpu" in torver or "+cxx11" in torver:
-        file_name = f"requirements_cpu_only{'_noavx2' if not cpu_has_avx2() else ''}.txt"
-    elif is_macos():
-        file_name = f"requirements_apple_{'intel' if is_x86_64() else 'silicon'}.txt"
-    else:
-        file_name = f"requirements{'_noavx2' if not cpu_has_avx2() else ''}.txt"
-
-    requirements_file = os.path.join(requirements_base, file_name)
-
-    # Load state from JSON file
     current_commit = get_current_commit()
-    wheels_changed = False
-    if os.path.exists(state_file):
-        with open(state_file, 'r') as f:
-            last_state = json.load(f)
-
-        if 'wheels_changed' in last_state or last_state.get('last_installed_commit') != current_commit:
+    wheels_changed = not os.path.exists(state_file)
+    if not wheels_changed:
+        state = load_state()
+        if 'wheels_changed' in state or state.get('last_installed_commit') != current_commit:
             wheels_changed = True
-    else:
-        wheels_changed = True
+
+    gpu_choice = get_gpu_choice()
+    requirements_file = get_requirements_file(gpu_choice)
 
     if pull:
         # Read .whl lines before pulling
@@ -409,19 +405,17 @@ def update_requirements(initial_installation=False, pull=True):
                 print_big_message(f"File '{file}' was updated during 'git pull'. Please run the script again.")
 
                 # Save state before exiting
-                current_state = {}
+                state = load_state()
                 if wheels_changed:
-                    current_state['wheels_changed'] = True
-
-                with open(state_file, 'w') as f:
-                    json.dump(current_state, f)
-
+                    state['wheels_changed'] = True
+                save_state(state)
                 sys.exit(1)
 
     # Save current state
-    current_state = {'last_installed_commit': current_commit}
-    with open(state_file, 'w') as f:
-        json.dump(current_state, f)
+    state = load_state()
+    state['last_installed_commit'] = current_commit
+    state.pop('wheels_changed', None)  # Remove wheels_changed flag
+    save_state(state)
 
     if os.environ.get("INSTALL_EXTENSIONS", "").lower() in ("yes", "y", "true", "1", "t", "on"):
         install_extensions_requirements()
@@ -432,11 +426,10 @@ def update_requirements(initial_installation=False, pull=True):
     # Update PyTorch
     if not initial_installation:
         update_pytorch_and_python()
-        torver = torch_version()
         clean_outdated_pytorch_cuda_dependencies()
 
     print_big_message(f"Installing webui requirements from file: {requirements_file}")
-    print(f"TORCH: {torver}\n")
+    print(f"GPU Choice: {gpu_choice}\n")
 
     # Prepare the requirements file
     textgen_requirements = open(requirements_file).read().splitlines()

From b30a73016d626e985e248de15fa65e5a531c8bd2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 2 Jun 2025 07:49:22 -0700
Subject: [PATCH 0611/1701] Remove the "Is typing..." yield by default

---
 modules/chat.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index f1ea16f10a..3c4c3636e5 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -656,7 +656,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
         update_message_metadata(output['metadata'], "user", row_idx, timestamp=get_current_timestamp())
 
         # *Is typing...*
-        if loading_message:
+        if loading_message and shared.processing_message:
             yield {
                 'visible': output['visible'][:-1] + [[output['visible'][-1][0], shared.processing_message]],
                 'internal': output['internal'],
@@ -680,7 +680,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             })
             output['metadata'][key]["current_version_index"] = len(output['metadata'][key]["versions"]) - 1
 
-            if loading_message:
+            if loading_message and shared.processing_message:
                 yield {
                     'visible': output['visible'][:-1] + [[visible_text, shared.processing_message]],
                     'internal': output['internal'][:-1] + [[text, '']],

From b38ec0ec385d44d49d3fe7adf2ad77ae62302214 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 2 Jun 2025 11:33:17 -0700
Subject: [PATCH 0612/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 04d972201f..277f8249bd 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index f807199d44..dbf35c3403 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 4fb70eb134..2e5eb6c929 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index a311ab9b48..9a19ab2911 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 30e8409a32..973d9bfbfc 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 7094994920..4a48a51fb5 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 318bb93af3..76bde864b4 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 7c3635cc9d..6cd0fa659e 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index bde310e174..60ce941e52 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 521edc0c38..b1649bc962 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index ef7946ff62..571eba52be 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index a3ad743ed9..88170cf34a 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index eec052d36f..e96cef491e 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index c9898a05d7..78f94aa53e 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 0de9c7cbc7..3e41427d03 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 2bfb4d519f..022ebb6161 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.14.0/llama_cpp_binaries-0.14.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 93b3752cdf9f43dd391462168e2e14dd2ab75643 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Jun 2025 09:40:30 -0700
Subject: [PATCH 0613/1701] Revert "Remove the "Is typing..." yield by default"

This reverts commit b30a73016d626e985e248de15fa65e5a531c8bd2.
---
 modules/chat.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 3c4c3636e5..f1ea16f10a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -656,7 +656,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
         update_message_metadata(output['metadata'], "user", row_idx, timestamp=get_current_timestamp())
 
         # *Is typing...*
-        if loading_message and shared.processing_message:
+        if loading_message:
             yield {
                 'visible': output['visible'][:-1] + [[output['visible'][-1][0], shared.processing_message]],
                 'internal': output['internal'],
@@ -680,7 +680,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             })
             output['metadata'][key]["current_version_index"] = len(output['metadata'][key]["versions"]) - 1
 
-            if loading_message and shared.processing_message:
+            if loading_message:
                 yield {
                     'visible': output['visible'][:-1] + [[visible_text, shared.processing_message]],
                     'internal': output['internal'][:-1] + [[text, '']],

From 9bd7359ffab5e434b7cdfdb43ee91cb3ad397c0d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Jun 2025 10:47:14 -0700
Subject: [PATCH 0614/1701] Scroll the textarea into view when editing a
 message

---
 js/global_scope_js.js | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index d5140c9377..801f1574cf 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -95,6 +95,12 @@ function startEditing(messageElement, messageBody, isUserMessage) {
   editingInterface.textarea.focus();
   editingInterface.textarea.setSelectionRange(rawText.length, rawText.length);
 
+  // Scroll the textarea into view
+  editingInterface.textarea.scrollIntoView({
+    behavior: "smooth",
+    block: "center"
+  });
+
   // Setup event handlers
   setupEditingHandlers(editingInterface.textarea, messageElement, originalHTML, messageBody, isUserMessage);
 }

From 66a75c899a4b0786cd8744886a189864923287b5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Jun 2025 10:59:43 -0700
Subject: [PATCH 0615/1701] Improve the scrollbars in code blocks

---
 js/main.js | 1 +
 1 file changed, 1 insertion(+)

diff --git a/js/main.js b/js/main.js
index 8090937fa9..2e72d82ea3 100644
--- a/js/main.js
+++ b/js/main.js
@@ -229,6 +229,7 @@ function doSyntaxHighlighting() {
         codeBlocks.forEach((codeBlock) => {
           hljs.highlightElement(codeBlock);
           codeBlock.setAttribute("data-highlighted", "true");
+          codeBlock.classList.add("pretty_scrollbar");
         });
 
         renderMathInElement(messageBody, {

From 3d676cd50f8661ca96a20a452611422acb47177c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Jun 2025 11:02:04 -0700
Subject: [PATCH 0616/1701] Optimize syntax highlighting

---
 js/main.js | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index 2e72d82ea3..9a620fa9f7 100644
--- a/js/main.js
+++ b/js/main.js
@@ -217,7 +217,7 @@ function isElementVisibleOnScreen(element) {
 }
 
 function doSyntaxHighlighting() {
-  const messageBodies = document.querySelectorAll(".message-body");
+  const messageBodies = document.getElementById("chat").querySelectorAll(".message-body");
 
   if (messageBodies.length > 0) {
     observer.disconnect();

From 3829507d0fd66eccc532b5d8d0e3d77c38143d0c Mon Sep 17 00:00:00 2001
From: Hanusz Leszek 
Date: Wed, 4 Jun 2025 20:13:36 +0200
Subject: [PATCH 0617/1701] Stop model during graceful shutdown (#7042)

---
 server.py | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/server.py b/server.py
index c22ed1f1b8..99d2e17137 100644
--- a/server.py
+++ b/server.py
@@ -60,6 +60,14 @@
 
 def signal_handler(sig, frame):
     logger.info("Received Ctrl+C. Shutting down Text generation web UI gracefully.")
+
+    # Explicitly stop LlamaServer to avoid __del__ cleanup issues during shutdown
+    if shared.model and shared.model.__class__.__name__ == 'LlamaServer':
+        try:
+            shared.model.stop()
+        except:
+            pass
+
     sys.exit(0)
 
 

From 977ec801b7682c3239fe3e6fdfcb8b90c1e802f3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Jun 2025 06:33:45 -0700
Subject: [PATCH 0618/1701] Improve table colors in instruct mode

---
 css/html_instruct_style.css | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index 6ad250aa34..9831ee8fb9 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -17,6 +17,14 @@
     color: #d1d5db !important;
 }
 
+.chat .message-body :is(th, td) {
+    border-color: #40404096 !important;
+}
+
+.dark .chat .message-body :is(th, td) {
+    border-color: #ffffff75 !important;
+}
+
 .chat .message-body :is(p, ul, ol) {
     margin: 1.25em 0 !important;
 }

From d47c8eb956a72ebc7c1f582718758697aef62118 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Jun 2025 06:56:24 -0700
Subject: [PATCH 0619/1701] Remove quotes from LLM-generated websearch query
 (closes #7045).

Fix by @Quiet-Joker
---
 modules/chat.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index f1ea16f10a..14f2a4f785 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -604,7 +604,12 @@ def generate_search_query(user_message, state):
 
     query = ""
     for reply in generate_reply(formatted_prompt, search_state, stopping_strings=[], is_chat=True):
-        query = reply.strip()
+        query = reply
+
+    # Strip and remove surrounding quotes if present
+    query = query.strip()
+    if len(query) >= 2 and query.startswith('"') and query.endswith('"'):
+        query = query[1:-1]
 
     return query
 

From 45f823ddf625cb914acb6620b0551bf6734b2838 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Jun 2025 22:23:34 -0700
Subject: [PATCH 0620/1701] Print \n after the llama.cpp progress bar reaches
 1.0

---
 modules/llama_cpp_server.py | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index aa7125415a..d8fcb89d47 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -417,14 +417,23 @@ def filter_stderr_with_progress(process_stderr):
             progress_match = progress_pattern.search(line)
 
             if progress_match:
+                progress_value = float(progress_match.group(1))
+
                 if last_was_progress:
                     # Overwrite the previous progress line using carriage return
                     sys.stderr.write(f'\r{line}')
                 else:
                     # First progress line - print normally
                     sys.stderr.write(line)
+
+                # Check if progress is 100% (1.0)
+                if progress_value >= 1.0:
+                    sys.stderr.write('\n')  # Add newline for completed progress
+                    last_was_progress = False
+                else:
+                    last_was_progress = True
+
                 sys.stderr.flush()
-                last_was_progress = True
             elif not line.startswith(('srv ', 'slot ')) and 'log_server_r: request: GET /health' not in line:
                 if last_was_progress:
                     # Finish the progress line with a newline, then print the new line

From f8f23b5489e70467ffc46a1b6ae2a406b3fcf9ff Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Jun 2025 22:25:13 -0700
Subject: [PATCH 0621/1701] Simplify the llama.cpp stderr filter code

---
 modules/llama_cpp_server.py | 53 ++++++++++++++++---------------------
 1 file changed, 23 insertions(+), 30 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index d8fcb89d47..f0a72de895 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -408,41 +408,34 @@ def stop(self):
 
 
 def filter_stderr_with_progress(process_stderr):
-    progress_pattern = re.compile(r'slot update_slots: id.*progress = (\d+\.\d+)')
+    """
+    Reads stderr lines from a process, filters out noise, and displays progress updates
+    inline (overwriting the same line) until completion.
+    """
+    progress_re = re.compile(r'slot update_slots: id.*progress = (\d+\.\d+)')
     last_was_progress = False
 
     try:
-        for line in iter(process_stderr.readline, ''):
-            line = line.rstrip('\n\r')  # Remove existing newlines
-            progress_match = progress_pattern.search(line)
-
-            if progress_match:
-                progress_value = float(progress_match.group(1))
-
+        for raw in iter(process_stderr.readline, ''):
+            line = raw.rstrip('\r\n')
+            match = progress_re.search(line)
+
+            if match:
+                progress = float(match.group(1))
+                # choose carriage return for in-progress or newline at completion
+                end_char = '\r' if progress < 1.0 else '\n'
+                print(line, end=end_char, file=sys.stderr, flush=True)
+                last_was_progress = (progress < 1.0)
+
+            # skip noise lines
+            elif not (line.startswith(('srv ', 'slot ')) or 'log_server_r: request: GET /health' in line):
+                # if we were in progress, finish that line first
                 if last_was_progress:
-                    # Overwrite the previous progress line using carriage return
-                    sys.stderr.write(f'\r{line}')
-                else:
-                    # First progress line - print normally
-                    sys.stderr.write(line)
-
-                # Check if progress is 100% (1.0)
-                if progress_value >= 1.0:
-                    sys.stderr.write('\n')  # Add newline for completed progress
-                    last_was_progress = False
-                else:
-                    last_was_progress = True
+                    print(file=sys.stderr)
 
-                sys.stderr.flush()
-            elif not line.startswith(('srv ', 'slot ')) and 'log_server_r: request: GET /health' not in line:
-                if last_was_progress:
-                    # Finish the progress line with a newline, then print the new line
-                    sys.stderr.write(f'\n{line}\n')
-                else:
-                    # Normal line - print with newline
-                    sys.stderr.write(f'{line}\n')
-                sys.stderr.flush()
+                print(line, file=sys.stderr, flush=True)
                 last_was_progress = False
-            # For filtered lines, don't change last_was_progress state
+
     except (ValueError, IOError):
+        # silently ignore broken output or IO errors
         pass

From 379dd01ca7770484bbd0798252665935a0bbfa05 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Jun 2025 22:32:07 -0700
Subject: [PATCH 0622/1701] Filter out failed web search downloads from
 attachments

---
 modules/web_search.py | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/modules/web_search.py b/modules/web_search.py
index 1f67034968..a1e47253b7 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -107,6 +107,13 @@ def add_web_search_attachments(history, row_idx, user_message, search_query, sta
             logger.warning("No search results found")
             return
 
+        # Filter out failed downloads before adding attachments
+        successful_results = [result for result in search_results if result['content'] and result['content'].strip()]
+
+        if not successful_results:
+            logger.warning("No successful downloads to add as attachments")
+            return
+
         # Add search results as attachments
         key = f"user_{row_idx}"
         if key not in history['metadata']:
@@ -114,7 +121,7 @@ def add_web_search_attachments(history, row_idx, user_message, search_query, sta
         if "attachments" not in history['metadata'][key]:
             history['metadata'][key]["attachments"] = []
 
-        for result in search_results:
+        for result in successful_results:
             attachment = {
                 "name": result['title'],
                 "type": "text/html",
@@ -123,7 +130,7 @@ def add_web_search_attachments(history, row_idx, user_message, search_query, sta
             }
             history['metadata'][key]["attachments"].append(attachment)
 
-        logger.info(f"Added {len(search_results)} web search results as attachments")
+        logger.info(f"Added {len(successful_results)} successful web search results as attachments")
 
     except Exception as e:
         logger.error(f"Error in web search: {e}")

From 2d263f227d090d2d5c0b152e045ab5cc3eae3073 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Jun 2025 22:38:20 -0700
Subject: [PATCH 0623/1701] Fix the chat input reappearing when the page is
 reloaded

---
 modules/ui.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/modules/ui.py b/modules/ui.py
index 14a09d2b7a..89bbbdb65e 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -270,6 +270,9 @@ def gather_interface_values(*args):
     if not shared.args.multi_user:
         shared.persistent_interface_state = output
 
+        # Remove the chat input, as it gets cleared after this function call
+        shared.persistent_interface_state.pop('textbox')
+
     # Prevent history loss if backend is restarted but UI is not refreshed
     if output['history'] is None and output['unique_id'] is not None:
         output['history'] = load_history(output['unique_id'], output['character_menu'], output['mode'])

From db847eed4c0693c735e364fcf48e79e4f001bb77 Mon Sep 17 00:00:00 2001
From: rakha abadi susilo 
Date: Sun, 8 Jun 2025 07:44:15 +0700
Subject: [PATCH 0624/1701] Add RTX 50XX Nvidia blackwell support (ExLlamaV2/V3
 and Transformers) (#7011)

---------

Co-authored-by: oobabooga <112222186+oobabooga@users.noreply.github.com>
---
 one_click.py                                  | 21 ++++++---
 requirements/full/requirements_cuda128.txt    | 46 +++++++++++++++++++
 .../full/requirements_cuda128_noavx2.txt      | 46 +++++++++++++++++++
 3 files changed, 106 insertions(+), 7 deletions(-)
 create mode 100644 requirements/full/requirements_cuda128.txt
 create mode 100644 requirements/full/requirements_cuda128_noavx2.txt

diff --git a/one_click.py b/one_click.py
index cccb0dc94f..94f2aab075 100644
--- a/one_click.py
+++ b/one_click.py
@@ -17,8 +17,6 @@
 
 # Define the required versions
 TORCH_VERSION = "2.6.0"
-TORCHVISION_VERSION = "0.21.0"
-TORCHAUDIO_VERSION = "2.6.0"
 PYTHON_VERSION = "3.11"
 LIBSTDCXX_VERSION_LINUX = "12.1.0"
 
@@ -119,12 +117,13 @@ def get_gpu_choice():
                     'B': 'AMD - Linux/macOS only, requires ROCm 6.2.4',
                     'C': 'Apple M Series',
                     'D': 'Intel Arc (beta)',
+                    'E': 'NVIDIA - CUDA 12.8',
                     'N': 'CPU mode'
                 },
             )
 
         # Convert choice to GPU name
-        gpu_choice = {"A": "NVIDIA", "B": "AMD", "C": "APPLE", "D": "INTEL", "N": "NONE"}[choice]
+        gpu_choice = {"A": "NVIDIA", "B": "AMD", "C": "APPLE", "D": "INTEL", "E": "NVIDIA_CUDA128", "N": "NONE"}[choice]
 
         # Save choice to state
         state['gpu_choice'] = gpu_choice
@@ -135,29 +134,33 @@ def get_gpu_choice():
 
 def get_pytorch_install_command(gpu_choice):
     """Get PyTorch installation command based on GPU choice"""
-    base_cmd = f"python -m pip install torch=={TORCH_VERSION} torchvision=={TORCHVISION_VERSION} torchaudio=={TORCHAUDIO_VERSION} "
+    base_cmd = f"python -m pip install torch=={TORCH_VERSION} "
 
     if gpu_choice == "NVIDIA":
         return base_cmd + "--index-url https://download.pytorch.org/whl/cu124"
+    elif gpu_choice == "NVIDIA_CUDA128":
+        return "python -m pip install torch==2.7.1 --index-url https://download.pytorch.org/whl/cu128"
     elif gpu_choice == "AMD":
         return base_cmd + "--index-url https://download.pytorch.org/whl/rocm6.2.4"
     elif gpu_choice in ["APPLE", "NONE"]:
         return base_cmd + "--index-url https://download.pytorch.org/whl/cpu"
     elif gpu_choice == "INTEL":
         if is_linux():
-            return "python -m pip install torch==2.1.0a0 torchvision==0.16.0a0 torchaudio==2.1.0a0 intel-extension-for-pytorch==2.1.10+xpu --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
+            return "python -m pip install torch==2.1.0a0 intel-extension-for-pytorch==2.1.10+xpu --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
         else:
-            return "python -m pip install torch==2.1.0a0 torchvision==0.16.0a0 torchaudio==2.1.0a0 intel-extension-for-pytorch==2.1.10 --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
+            return "python -m pip install torch==2.1.0a0 intel-extension-for-pytorch==2.1.10 --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
     else:
         return base_cmd
 
 
 def get_pytorch_update_command(gpu_choice):
     """Get PyTorch update command based on GPU choice"""
-    base_cmd = f"python -m pip install --upgrade torch=={TORCH_VERSION} torchvision=={TORCHVISION_VERSION} torchaudio=={TORCHAUDIO_VERSION}"
+    base_cmd = f"python -m pip install --upgrade torch=={TORCH_VERSION} "
 
     if gpu_choice == "NVIDIA":
         return f"{base_cmd} --index-url https://download.pytorch.org/whl/cu124"
+    elif gpu_choice == "NVIDIA_CUDA128":
+        return "python -m pip install --upgrade torch==2.7.1 --index-url https://download.pytorch.org/whl/cu128"
     elif gpu_choice == "AMD":
         return f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.2.4"
     elif gpu_choice in ["APPLE", "NONE"]:
@@ -181,6 +184,8 @@ def get_requirements_file(gpu_choice):
         file_name = f"requirements_cpu_only{'_noavx2' if not cpu_has_avx2() else ''}.txt"
     elif gpu_choice == "NVIDIA":
         file_name = f"requirements{'_noavx2' if not cpu_has_avx2() else ''}.txt"
+    elif gpu_choice == "NVIDIA_CUDA128":
+        file_name = f"requirements_cuda128{'_noavx2' if not cpu_has_avx2() else ''}.txt"
     else:
         raise ValueError(f"Unknown GPU choice: {gpu_choice}")
 
@@ -328,6 +333,8 @@ def install_webui():
     # Handle CUDA version display
     elif any((is_windows(), is_linux())) and gpu_choice == "NVIDIA":
         print("CUDA: 12.4")
+    elif any((is_windows(), is_linux())) and gpu_choice == "NVIDIA_CUDA128":
+        print("CUDA: 12.8")
 
     # No PyTorch for AMD on Windows (?)
     elif is_windows() and gpu_choice == "AMD":
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
new file mode 100644
index 0000000000..9fc9960667
--- /dev/null
+++ b/requirements/full/requirements_cuda128.txt
@@ -0,0 +1,46 @@
+accelerate==1.5.*
+beautifulsoup4==4.13.4
+bitsandbytes==0.45.*
+colorama
+datasets
+duckduckgo_search==8.0.2
+einops
+fastapi==0.112.4
+gradio==4.37.*
+jinja2==3.1.6
+markdown
+numpy==2.2.*
+pandas
+peft==0.15.*
+Pillow>=9.5.0
+psutil
+pydantic==2.8.2
+PyPDF2==3.0.1
+python-docx==1.1.2
+pyyaml
+requests
+rich
+safetensors==0.5.*
+scipy
+sentencepiece
+tensorboard
+transformers==4.52.*
+triton-windows; platform_system == "Windows"
+tqdm
+wandb
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
+
+# CUDA wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
new file mode 100644
index 0000000000..ff34673ab4
--- /dev/null
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -0,0 +1,46 @@
+accelerate==1.5.*
+beautifulsoup4==4.13.4
+bitsandbytes==0.45.*
+colorama
+datasets
+duckduckgo_search==8.0.2
+einops
+fastapi==0.112.4
+gradio==4.37.*
+jinja2==3.1.6
+markdown
+numpy==2.2.*
+pandas
+peft==0.15.*
+Pillow>=9.5.0
+psutil
+pydantic==2.8.2
+PyPDF2==3.0.1
+python-docx==1.1.2
+pyyaml
+requests
+rich
+safetensors==0.5.*
+scipy
+sentencepiece
+tensorboard
+transformers==4.52.*
+triton-windows; platform_system == "Windows"
+tqdm
+wandb
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
+
+# CUDA wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

From 35ed55d18f0d7acfe55af10d2086185f7141cc46 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Sat, 7 Jun 2025 22:46:52 -0300
Subject: [PATCH 0625/1701] UI persistence (#7050)

---
 modules/shared.py     |   2 +
 modules/ui.py         | 106 ++++++++++++++++++++++++++++++++++++++++++
 modules/ui_session.py |  17 +------
 server.py             |   3 ++
 4 files changed, 112 insertions(+), 16 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index f712f7f8ac..2e50077937 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -35,6 +35,8 @@
     'mode': 'instruct',
     'chat_style': 'cai-chat',
     'chat-instruct_command': 'Continue the chat dialogue below. Write a single reply for the character "<|character|>".\n\n<|prompt|>',
+    'enable_web_search': False,
+    'web_search_pages': 3,
     'prompt-default': 'QA',
     'prompt-notebook': 'QA',
     'character': 'Assistant',
diff --git a/modules/ui.py b/modules/ui.py
index 89bbbdb65e..db3adf0f00 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -1,4 +1,5 @@
 import copy
+import threading
 from pathlib import Path
 
 import gradio as gr
@@ -7,6 +8,16 @@
 import extensions
 from modules import shared
 from modules.chat import load_history
+from modules.utils import gradio
+
+# Global state for auto-saving UI settings with debouncing
+_auto_save_timer = None
+_auto_save_lock = threading.Lock()
+_last_interface_state = None
+_last_preset = None
+_last_extensions = None
+_last_show_controls = None
+_last_theme_state = None
 
 with open(Path(__file__).resolve().parent / '../css/NotoSans/stylesheet.css', 'r') as f:
     css = f.read()
@@ -334,6 +345,101 @@ def save_settings(state, preset, extensions_list, show_controls, theme_state):
     return yaml.dump(output, sort_keys=False, width=float("inf"), allow_unicode=True)
 
 
+def store_current_state_and_debounce(interface_state, preset, extensions, show_controls, theme_state):
+    """Store current state and trigger debounced save"""
+    global _auto_save_timer, _last_interface_state, _last_preset, _last_extensions, _last_show_controls, _last_theme_state
+
+    if shared.args.multi_user:
+        return
+
+    # Store the current state in global variables
+    _last_interface_state = interface_state
+    _last_preset = preset
+    _last_extensions = extensions
+    _last_show_controls = show_controls
+    _last_theme_state = theme_state
+
+    # Reset the debounce timer
+    with _auto_save_lock:
+        if _auto_save_timer is not None:
+            _auto_save_timer.cancel()
+
+        _auto_save_timer = threading.Timer(2.0, _perform_debounced_save)
+        _auto_save_timer.start()
+
+
+def _perform_debounced_save():
+    """Actually perform the save using the stored state"""
+    global _auto_save_timer
+
+    try:
+        if _last_interface_state is not None:
+            contents = save_settings(_last_interface_state, _last_preset, _last_extensions, _last_show_controls, _last_theme_state)
+            settings_path = Path('user_data') / 'settings.yaml'
+            settings_path.parent.mkdir(exist_ok=True)
+            with open(settings_path, 'w', encoding='utf-8') as f:
+                f.write(contents)
+    except Exception as e:
+        print(f"Auto-save failed: {e}")
+    finally:
+        with _auto_save_lock:
+            _auto_save_timer = None
+
+
+def setup_auto_save():
+    """Attach auto-save to key UI elements"""
+    if shared.args.multi_user:
+        return
+
+    change_elements = [
+        # Chat tab (ui_chat.py)
+        'start_with',
+        'enable_web_search',
+        'web_search_pages',
+        'mode',
+        'chat_style',
+        'chat-instruct_command',
+        'character_menu',
+        'name1',
+        'user_bio',
+        'custom_system_message',
+        'chat_template_str',
+
+        # Parameters tab (ui_parameters.py)
+        'preset_menu',
+        'max_new_tokens',
+        'prompt_lookup_num_tokens',
+        'max_tokens_second',
+        'auto_max_new_tokens',
+        'ban_eos_token',
+        'add_bos_token',
+        'enable_thinking',
+        'skip_special_tokens',
+        'stream',
+        'static_cache',
+        'seed',
+        'custom_stopping_strings',
+        'custom_token_bans',
+        'negative_prompt',
+
+        # Default tab (ui_default.py)
+        'prompt_menu-default',
+
+        # Notebook tab (ui_notebook.py)
+        'prompt_menu-notebook',
+
+        # Session tab (ui_session.py)
+        'show_controls',
+        'theme_state',
+    ]
+
+    for element_name in change_elements:
+        if element_name in shared.gradio:
+            shared.gradio[element_name].change(
+                gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
+                store_current_state_and_debounce, gradio('interface_state', 'preset_menu', 'extensions_menu', 'show_controls', 'theme_state'), None, show_progress=False)
+
+
 def create_refresh_button(refresh_component, refresh_method, refreshed_args, elem_class, interactive=True):
     """
     Copied from https://github.com/AUTOMATIC1111/stable-diffusion-webui
diff --git a/modules/ui_session.py b/modules/ui_session.py
index a4eba66721..4ed740cd70 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -10,11 +10,10 @@ def create_ui():
     with gr.Tab("Session", elem_id="session-tab"):
         with gr.Row():
             with gr.Column():
-                shared.gradio['reset_interface'] = gr.Button("Apply flags/extensions and restart", interactive=not mu)
                 with gr.Row():
                     shared.gradio['toggle_dark_mode'] = gr.Button('Toggle 💡')
-                    shared.gradio['save_settings'] = gr.Button('Save UI defaults to user_data/settings.yaml', interactive=not mu)
 
+                shared.gradio['reset_interface'] = gr.Button("Apply flags/extensions and restart", interactive=not mu)
                 with gr.Row():
                     with gr.Column():
                         shared.gradio['extensions_menu'] = gr.CheckboxGroup(choices=utils.get_available_extensions(), value=shared.args.extensions, label="Available extensions", info='Note that some of these extensions may require manually installing Python requirements through the command: pip install -r extensions/extension_name/requirements.txt', elem_classes='checkboxgroup-table')
@@ -42,20 +41,6 @@ def create_ui():
             lambda x: 'dark' if x == 'light' else 'light', gradio('theme_state'), gradio('theme_state')).then(
             None, None, None, js=f'() => {{{ui.dark_theme_js}; toggleDarkMode()}}')
 
-        shared.gradio['save_settings'].click(
-            ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-            handle_save_settings, gradio('interface_state', 'preset_menu', 'extensions_menu', 'show_controls', 'theme_state'), gradio('save_contents', 'save_filename', 'save_root', 'file_saver'), show_progress=False)
-
-
-def handle_save_settings(state, preset, extensions, show_controls, theme):
-    contents = ui.save_settings(state, preset, extensions, show_controls, theme)
-    return [
-        contents,
-        "settings.yaml",
-        "user_data/",
-        gr.update(visible=True)
-    ]
-
 
 def set_interface_arguments(extensions, bool_active):
     shared.args.extensions = extensions
diff --git a/server.py b/server.py
index 99d2e17137..3be4c27c80 100644
--- a/server.py
+++ b/server.py
@@ -150,6 +150,9 @@ def create_interface():
         ui_parameters.create_event_handlers()
         ui_model_menu.create_event_handlers()
 
+        # UI persistence events
+        ui.setup_auto_save()
+
         # Interface launch events
         shared.gradio['interface'].load(
             None,

From 6436bf1920b128b40ed2abb35da947af6844cca3 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Sun, 8 Jun 2025 01:58:02 -0300
Subject: [PATCH 0626/1701] More UI persistence: presets and characters (#7051)

---
 modules/chat.py           | 61 ++++++++++++++++++++++++++
 modules/html_generator.py | 16 ++++---
 modules/presets.py        | 68 ++++-------------------------
 modules/shared.py         | 61 +++++++++++++++++++++++---
 modules/ui.py             | 48 +++++++++++++++++++--
 modules/ui_chat.py        | 20 ++++++---
 modules/ui_default.py     |  4 +-
 modules/ui_notebook.py    |  4 +-
 modules/ui_parameters.py  | 90 ++++++++++++++++++++-------------------
 server.py                 | 20 +++++----
 10 files changed, 258 insertions(+), 134 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 14f2a4f785..f740db5528 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1220,6 +1220,45 @@ def load_character(character, name1, name2):
     return name1, name2, picture, greeting, context
 
 
+def reset_character_for_ui(state):
+    """Reset character fields to the currently loaded character's saved values"""
+    if state['character_menu'] and state['character_menu'] != 'None':
+        try:
+            name1, name2, picture, greeting, context = load_character(state['character_menu'], state['name1'], state['name2'])
+
+            state['name2'] = name2
+            state['greeting'] = greeting
+            state['context'] = context
+            state['character_picture'] = picture  # This triggers cache update via generate_pfp_cache
+
+            return state, name2, context, greeting, picture
+
+        except Exception as e:
+            logger.error(f"Failed to reset character '{state['character_menu']}': {e}")
+            return clear_character_for_ui(state)
+    else:
+        return clear_character_for_ui(state)
+
+
+def clear_character_for_ui(state):
+    """Clear all character fields and picture cache"""
+    state['name2'] = shared.settings['name2']
+    state['context'] = shared.settings['context']
+    state['greeting'] = shared.settings['greeting']
+    state['character_picture'] = None
+
+    # Clear the cache files
+    cache_folder = Path(shared.args.disk_cache_dir)
+    for cache_file in ['pfp_character.png', 'pfp_character_thumb.png']:
+        cache_path = Path(f'{cache_folder}/{cache_file}')
+        if cache_path.exists():
+            cache_path.unlink()
+
+    logger.info("Cleared character fields and picture cache")
+
+    return state, state['name2'], state['context'], state['greeting'], None
+
+
 def load_instruction_template(template):
     if template == 'None':
         return ''
@@ -1710,6 +1749,28 @@ def handle_character_menu_change(state):
     ]
 
 
+def handle_character_picture_change(picture):
+    """Update or clear cache when character picture changes"""
+    cache_folder = Path(shared.args.disk_cache_dir)
+    if not cache_folder.exists():
+        cache_folder.mkdir()
+
+    if picture is not None:
+        # Save to cache
+        picture.save(Path(f'{cache_folder}/pfp_character.png'), format='PNG')
+        thumb = make_thumbnail(picture)
+        thumb.save(Path(f'{cache_folder}/pfp_character_thumb.png'), format='PNG')
+        logger.info("Updated character picture cache")
+    else:
+        # Remove cache files when picture is cleared
+        for cache_file in ['pfp_character.png', 'pfp_character_thumb.png']:
+            cache_path = Path(f'{cache_folder}/{cache_file}')
+            if cache_path.exists():
+                cache_path.unlink()
+
+        logger.info("Cleared character picture cache")
+
+
 def handle_mode_change(state):
     history = load_latest_history(state)
     histories = find_all_histories_with_first_prompts(state)
diff --git a/modules/html_generator.py b/modules/html_generator.py
index f90e3b04db..eac7d91ac8 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -520,17 +520,23 @@ def create_message(role, content, raw_content):
     return output
 
 
+def get_character_image_with_cache_buster():
+    """Get character image URL with cache busting based on file modification time"""
+    cache_path = Path("user_data/cache/pfp_character_thumb.png")
+    if cache_path.exists():
+        mtime = int(cache_path.stat().st_mtime)
+        return f''
+
+    return ''
+
+
 def generate_cai_chat_html(history, name1, name2, style, character, reset_cache=False, last_message_only=False):
     if not last_message_only:
         output = f'
            '
     else:
         output = ""
 
-    # We use ?character and ?time.time() to force the browser to reset caches
-    img_bot = (
-        f''
-        if Path("user_data/cache/pfp_character_thumb.png").exists() else ''
-    )
+    img_bot = get_character_image_with_cache_buster()
 
     def create_message(role, content, raw_content):
         """Inner function for CAI-style messages."""
diff --git a/modules/presets.py b/modules/presets.py
index cf70660551..3eb1f5fc1f 100644
--- a/modules/presets.py
+++ b/modules/presets.py
@@ -1,6 +1,5 @@
 import functools
 import pprint
-import random
 from pathlib import Path
 
 import yaml
@@ -93,68 +92,17 @@ def load_preset_for_ui(name, state):
     return state, *[generate_params[k] for k in presets_params()]
 
 
-def random_preset(state):
-    params_and_values = {
-        'remove_tail_tokens': {
-            'top_p': [0.5, 0.8, 0.9, 0.95, 0.99],
-            'min_p': [0.5, 0.2, 0.1, 0.05, 0.01],
-            'top_k': [3, 5, 10, 20, 30, 40],
-            'typical_p': [0.2, 0.575, 0.95],
-            'tfs': [0.5, 0.8, 0.9, 0.95, 0.99],
-            'top_a': [0.5, 0.2, 0.1, 0.05, 0.01],
-            'epsilon_cutoff': [1, 3, 5, 7, 9],
-            'eta_cutoff': [3, 6, 9, 12, 15, 18],
-        },
-        'flatten_distribution': {
-            'temperature': [0.1, 0.5, 0.7, 0.8, 1, 1.2, 1.5, 2.0, 5.0],
-            'dynamic_temperature': [
-                [0.1, 1],
-                [0.1, 1.5],
-                [0.1, 2],
-                [0.1, 5],
-                [0.5, 1],
-                [0.5, 1.5],
-                [0.5, 2],
-                [0.5, 5],
-                [0.8, 1],
-                [0.8, 1.5],
-                [0.8, 2],
-                [0.8, 5],
-                [1, 1.5],
-                [1, 2],
-                [1, 5]
-            ],
-            'smoothing_factor': [0.2, 0.3, 0.6, 1.2],
-        },
-        'repetition': {
-            'repetition_penalty': [1, 1.05, 1.1, 1.15, 1.20, 1.25],
-            'presence_penalty': [0, 0.1, 0.2, 0.4, 0.6, 0.8, 1.0, 2.0],
-            'frequency_penalty': [0, 0.1, 0.2, 0.4, 0.6, 0.8, 1.0, 2.0],
-        },
-        'other': {
-            'temperature_last': [True, False],
-        }
-    }
+def reset_preset_for_ui(name, state):
+    """Reset current preset to its saved values from file"""
+    generate_params = load_preset(name, verbose=True)
+    state.update(generate_params)
+    return state, *[generate_params[k] for k in presets_params()]
 
-    generate_params = default_preset()
-    for cat in params_and_values:
-        choices = list(params_and_values[cat].keys())
-        if shared.args.loader is not None:
-            choices = [x for x in choices if loader_contains(x)]
-
-        if len(choices) > 0:
-            choice = random.choice(choices)
-            value = random.choice(params_and_values[cat][choice])
-            if choice == 'dynamic_temperature':
-                generate_params['dynamic_temperature'] = True
-                generate_params['dynatemp_low'] = value[0]
-                generate_params['dynatemp_high'] = value[1]
-            else:
-                generate_params[choice] = value
 
+def neutralize_samplers_for_ui(state):
+    """Set all samplers to their default/neutral values"""
+    generate_params = default_preset()
     state.update(generate_params)
-    logger.info("GENERATED_PRESET=")
-    pprint.PrettyPrinter(indent=4, width=1, sort_dicts=False).pprint(remove_defaults(state))
     return state, *[generate_params[k] for k in presets_params()]
 
 
diff --git a/modules/shared.py b/modules/shared.py
index 2e50077937..f4f8e18004 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -39,10 +39,6 @@
     'web_search_pages': 3,
     'prompt-default': 'QA',
     'prompt-notebook': 'QA',
-    'character': 'Assistant',
-    'name1': 'You',
-    'user_bio': '',
-    'custom_system_message': '',
     'preset': 'min_p',
     'max_new_tokens': 512,
     'max_new_tokens_min': 1,
@@ -63,8 +59,64 @@
     'negative_prompt': '',
     'dark_theme': True,
     'default_extensions': [],
+
+    # Character settings
+    'character': 'Assistant',
+    'name1': 'You',
+    'name2': 'AI',
+    'user_bio': '',
+    'context': 'The following is a conversation with an AI Large Language Model. The AI has been trained to answer questions, provide recommendations, and help with decision making. The AI follows user requests. The AI thinks outside the box.',
+    'greeting': 'How can I help you today?',
+    'custom_system_message': '',
     'instruction_template_str': "{%- set ns = namespace(found=false) -%}\n{%- for message in messages -%}\n    {%- if message['role'] == 'system' -%}\n        {%- set ns.found = true -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if not ns.found -%}\n    {{- '' + 'Below is an instruction that describes a task. Write a response that appropriately completes the request.' + '\\n\\n' -}}\n{%- endif %}\n{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {{- '' + message['content'] + '\\n\\n' -}}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{-'### Instruction:\\n' + message['content'] + '\\n\\n'-}}\n        {%- else -%}\n            {{-'### Response:\\n' + message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{-'### Response:\\n'-}}\n{%- endif -%}",
     'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{- name1 + ': ' + message['content'] + '\\n'-}}\n        {%- else -%}\n            {{- name2 + ': ' + message['content'] + '\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}",
+
+    # Generation parameters - Curve shape
+    'temperature': 1.0,
+    'dynatemp_low': 1.0,
+    'dynatemp_high': 1.0,
+    'dynatemp_exponent': 1.0,
+    'smoothing_factor': 0.0,
+    'smoothing_curve': 1.0,
+
+    # Generation parameters - Curve cutoff
+    'min_p': 0.0,
+    'top_p': 1.0,
+    'top_k': 0,
+    'typical_p': 1.0,
+    'xtc_threshold': 0.1,
+    'xtc_probability': 0.0,
+    'epsilon_cutoff': 0.0,
+    'eta_cutoff': 0.0,
+    'tfs': 1.0,
+    'top_a': 0.0,
+    'top_n_sigma': 0.0,
+
+    # Generation parameters - Repetition suppression
+    'dry_multiplier': 0.0,
+    'dry_allowed_length': 2,
+    'dry_base': 1.75,
+    'repetition_penalty': 1.0,
+    'frequency_penalty': 0.0,
+    'presence_penalty': 0.0,
+    'encoder_repetition_penalty': 1.0,
+    'no_repeat_ngram_size': 0,
+    'repetition_penalty_range': 1024,
+
+    # Generation parameters - Alternative sampling methods
+    'penalty_alpha': 0.0,
+    'guidance_scale': 1.0,
+    'mirostat_mode': 0,
+    'mirostat_tau': 5.0,
+    'mirostat_eta': 0.1,
+
+    # Generation parameters - Other options
+    'do_sample': True,
+    'dynamic_temperature': False,
+    'temperature_last': False,
+    'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ndry\ntop_n_sigma\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nmirostat\nxtc\nencoder_repetition_penalty\nno_repeat_ngram',
+    'dry_sequence_breakers': '"\\n", ":", "\\"", "*"',
+    'grammar_string': '',
 }
 
 default_settings = copy.deepcopy(settings)
@@ -75,7 +127,6 @@
 # Basic settings
 group = parser.add_argument_group('Basic settings')
 group.add_argument('--multi-user', action='store_true', help='Multi-user mode. Chat histories are not saved or automatically loaded. Warning: this is likely not safe for sharing publicly.')
-group.add_argument('--character', type=str, help='The name of the character to load in chat mode by default.')
 group.add_argument('--model', type=str, help='Name of the model to load by default.')
 group.add_argument('--lora', type=str, nargs='+', help='The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.')
 group.add_argument('--model-dir', type=str, default='user_data/models', help='Path to directory with all the models.')
diff --git a/modules/ui.py b/modules/ui.py
index db3adf0f00..8ec4b165e7 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -310,7 +310,7 @@ def apply_interface_values(state, use_persistent=False):
 
 def save_settings(state, preset, extensions_list, show_controls, theme_state):
     output = copy.deepcopy(shared.settings)
-    exclude = ['name2', 'greeting', 'context', 'truncation_length', 'instruction_template_str']
+    exclude = []
     for k in state:
         if k in shared.settings and k not in exclude:
             output[k] = state[k]
@@ -323,6 +323,7 @@ def save_settings(state, preset, extensions_list, show_controls, theme_state):
     output['seed'] = int(output['seed'])
     output['show_controls'] = show_controls
     output['dark_theme'] = True if theme_state == 'dark' else False
+    output.pop('instruction_template_str')
 
     # Save extension values in the UI
     for extension_name in extensions_list:
@@ -364,7 +365,7 @@ def store_current_state_and_debounce(interface_state, preset, extensions, show_c
         if _auto_save_timer is not None:
             _auto_save_timer.cancel()
 
-        _auto_save_timer = threading.Timer(2.0, _perform_debounced_save)
+        _auto_save_timer = threading.Timer(1.0, _perform_debounced_save)
         _auto_save_timer.start()
 
 
@@ -401,15 +402,52 @@ def setup_auto_save():
         'chat-instruct_command',
         'character_menu',
         'name1',
+        'name2',
+        'context',
+        'greeting',
         'user_bio',
         'custom_system_message',
         'chat_template_str',
 
-        # Parameters tab (ui_parameters.py)
+        # Parameters tab (ui_parameters.py) - Generation parameters
         'preset_menu',
+        'temperature',
+        'dynatemp_low',
+        'dynatemp_high',
+        'dynatemp_exponent',
+        'smoothing_factor',
+        'smoothing_curve',
+        'min_p',
+        'top_p',
+        'top_k',
+        'typical_p',
+        'xtc_threshold',
+        'xtc_probability',
+        'epsilon_cutoff',
+        'eta_cutoff',
+        'tfs',
+        'top_a',
+        'top_n_sigma',
+        'dry_multiplier',
+        'dry_allowed_length',
+        'dry_base',
+        'repetition_penalty',
+        'frequency_penalty',
+        'presence_penalty',
+        'encoder_repetition_penalty',
+        'no_repeat_ngram_size',
+        'repetition_penalty_range',
+        'penalty_alpha',
+        'guidance_scale',
+        'mirostat_mode',
+        'mirostat_tau',
+        'mirostat_eta',
         'max_new_tokens',
         'prompt_lookup_num_tokens',
         'max_tokens_second',
+        'do_sample',
+        'dynamic_temperature',
+        'temperature_last',
         'auto_max_new_tokens',
         'ban_eos_token',
         'add_bos_token',
@@ -417,10 +455,14 @@ def setup_auto_save():
         'skip_special_tokens',
         'stream',
         'static_cache',
+        'truncation_length',
         'seed',
+        'sampler_priority',
         'custom_stopping_strings',
         'custom_token_bans',
         'negative_prompt',
+        'dry_sequence_breakers',
+        'grammar_string',
 
         # Default tab (ui_default.py)
         'prompt_menu-default',
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 0d5a2c1817..712843d3c2 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -86,7 +86,7 @@ def create_ui():
                     shared.gradio['web_search_pages'] = gr.Number(value=shared.settings.get('web_search_pages', 3), precision=0, label='Number of pages to download', minimum=1, maximum=10)
 
                 with gr.Row():
-                    shared.gradio['mode'] = gr.Radio(choices=['instruct', 'chat-instruct', 'chat'], value=shared.settings['mode'] if shared.settings['mode'] in ['chat', 'chat-instruct'] else None, label='Mode', info='Defines how the chat prompt is generated. In instruct and chat-instruct modes, the instruction template Parameters > Instruction template is used.', elem_id='chat-mode')
+                    shared.gradio['mode'] = gr.Radio(choices=['instruct', 'chat-instruct', 'chat'], value=None, label='Mode', info='Defines how the chat prompt is generated. In instruct and chat-instruct modes, the instruction template Parameters > Instruction template is used.', elem_id='chat-mode')
 
                 with gr.Row():
                     shared.gradio['chat_style'] = gr.Dropdown(choices=utils.get_available_chat_styles(), label='Chat style', value=shared.settings['chat_style'], visible=shared.settings['mode'] != 'instruct')
@@ -118,14 +118,16 @@ def create_chat_settings_ui():
             with gr.Column(scale=8):
                 with gr.Tab("Character"):
                     with gr.Row():
-                        shared.gradio['character_menu'] = gr.Dropdown(value=None, choices=utils.get_available_characters(), label='Character', elem_id='character-menu', info='Used in chat and chat-instruct modes.', elem_classes='slim-dropdown')
+                        shared.gradio['character_menu'] = gr.Dropdown(value=shared.settings['character'], choices=utils.get_available_characters(), label='Character', elem_id='character-menu', info='Used in chat and chat-instruct modes.', elem_classes='slim-dropdown')
                         ui.create_refresh_button(shared.gradio['character_menu'], lambda: None, lambda: {'choices': utils.get_available_characters()}, 'refresh-button', interactive=not mu)
                         shared.gradio['save_character'] = gr.Button('💾', elem_classes='refresh-button', elem_id="save-character", interactive=not mu)
                         shared.gradio['delete_character'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu)
+                        shared.gradio['reset_character'] = gr.Button('↺', elem_classes='refresh-button', interactive=True)
+                        shared.gradio['clear_character'] = gr.Button('⚪', elem_classes='refresh-button', interactive=True)
 
-                    shared.gradio['name2'] = gr.Textbox(value='', lines=1, label='Character\'s name')
-                    shared.gradio['context'] = gr.Textbox(value='', lines=10, label='Context', elem_classes=['add_scrollbar'])
-                    shared.gradio['greeting'] = gr.Textbox(value='', lines=5, label='Greeting', elem_classes=['add_scrollbar'])
+                    shared.gradio['name2'] = gr.Textbox(value=shared.settings['name2'], lines=1, label='Character\'s name')
+                    shared.gradio['context'] = gr.Textbox(value=shared.settings['context'], lines=10, label='Context', elem_classes=['add_scrollbar'])
+                    shared.gradio['greeting'] = gr.Textbox(value=shared.settings['greeting'], lines=5, label='Greeting', elem_classes=['add_scrollbar'])
 
                 with gr.Tab("User"):
                     shared.gradio['name1'] = gr.Textbox(value=shared.settings['name1'], lines=1, label='Name')
@@ -178,7 +180,7 @@ def create_chat_settings_ui():
         with gr.Row():
             with gr.Column():
                 shared.gradio['custom_system_message'] = gr.Textbox(value=shared.settings['custom_system_message'], lines=2, label='Custom system message', info='If not empty, will be used instead of the default one.', elem_classes=['add_scrollbar'])
-                shared.gradio['instruction_template_str'] = gr.Textbox(value='', label='Instruction template', lines=24, info='This gets autodetected; you usually don\'t need to change it. Used in instruct and chat-instruct modes.', elem_classes=['add_scrollbar', 'monospace'])
+                shared.gradio['instruction_template_str'] = gr.Textbox(value=shared.settings['instruction_template_str'], label='Instruction template', lines=24, info='This gets autodetected; you usually don\'t need to change it. Used in instruct and chat-instruct modes.', elem_classes=['add_scrollbar', 'monospace'])
                 with gr.Row():
                     shared.gradio['send_instruction_to_default'] = gr.Button('Send to default', elem_classes=['small-button'])
                     shared.gradio['send_instruction_to_notebook'] = gr.Button('Send to notebook', elem_classes=['small-button'])
@@ -294,6 +296,8 @@ def create_event_handlers():
         chat.handle_character_menu_change, gradio('interface_state'), gradio('history', 'display', 'name1', 'name2', 'character_picture', 'greeting', 'context', 'unique_id'), show_progress=False).then(
         None, None, None, js=f'() => {{{ui.update_big_picture_js}; updateBigPicture()}}')
 
+    shared.gradio['character_picture'].change(chat.handle_character_picture_change, gradio('character_picture'), None, show_progress=False)
+
     shared.gradio['mode'].change(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_mode_change, gradio('interface_state'), gradio('history', 'display', 'chat_style', 'chat-instruct_command', 'unique_id'), show_progress=False).then(
@@ -317,6 +321,10 @@ def create_event_handlers():
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_save_template_click, gradio('instruction_template_str'), gradio('save_filename', 'save_root', 'save_contents', 'file_saver'), show_progress=False)
 
+    shared.gradio['reset_character'].click(
+        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
+        chat.reset_character_for_ui, gradio('interface_state'), gradio('interface_state', 'name2', 'context', 'greeting', 'character_picture'), show_progress=False)
+
     shared.gradio['delete_template'].click(chat.handle_delete_template_click, gradio('instruction_template'), gradio('delete_filename', 'delete_root', 'file_deleter'), show_progress=False)
     shared.gradio['save_chat_history'].click(
         lambda x: json.dumps(x, indent=4), gradio('history'), gradio('temporary_text')).then(
diff --git a/modules/ui_default.py b/modules/ui_default.py
index c2946b3789..8acc4b1093 100644
--- a/modules/ui_default.py
+++ b/modules/ui_default.py
@@ -19,7 +19,7 @@ def create_ui():
         with gr.Row():
             with gr.Column():
                 with gr.Row():
-                    shared.gradio['textbox-default'] = gr.Textbox(value='', lines=27, label='Input', elem_classes=['textbox_default', 'add_scrollbar'])
+                    shared.gradio['textbox-default'] = gr.Textbox(value=load_prompt(shared.settings['prompt-default']), lines=27, label='Input', elem_classes=['textbox_default', 'add_scrollbar'])
                     shared.gradio['token-counter-default'] = gr.HTML(value="0", elem_id="default-token-counter")
 
                 with gr.Row():
@@ -28,7 +28,7 @@ def create_ui():
                     shared.gradio['Generate-default'] = gr.Button('Generate', variant='primary')
 
                 with gr.Row():
-                    shared.gradio['prompt_menu-default'] = gr.Dropdown(choices=utils.get_available_prompts(), value='None', label='Prompt', elem_classes='slim-dropdown')
+                    shared.gradio['prompt_menu-default'] = gr.Dropdown(choices=utils.get_available_prompts(), value=shared.settings['prompt-default'], label='Prompt', elem_classes='slim-dropdown')
                     ui.create_refresh_button(shared.gradio['prompt_menu-default'], lambda: None, lambda: {'choices': utils.get_available_prompts()}, 'refresh-button', interactive=not mu)
                     shared.gradio['save_prompt-default'] = gr.Button('💾', elem_classes='refresh-button', interactive=not mu)
                     shared.gradio['delete_prompt-default'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu)
diff --git a/modules/ui_notebook.py b/modules/ui_notebook.py
index b234ac5753..3f79a93c36 100644
--- a/modules/ui_notebook.py
+++ b/modules/ui_notebook.py
@@ -22,7 +22,7 @@ def create_ui():
             with gr.Column(scale=4):
                 with gr.Tab('Raw'):
                     with gr.Row():
-                        shared.gradio['textbox-notebook'] = gr.Textbox(value='', lines=27, elem_id='textbox-notebook', elem_classes=['textbox', 'add_scrollbar'])
+                        shared.gradio['textbox-notebook'] = gr.Textbox(value=load_prompt(shared.settings['prompt-notebook']), lines=27, elem_id='textbox-notebook', elem_classes=['textbox', 'add_scrollbar'])
                         shared.gradio['token-counter-notebook'] = gr.HTML(value="0", elem_id="notebook-token-counter")
 
                 with gr.Tab('Markdown'):
@@ -56,7 +56,7 @@ def create_ui():
             with gr.Column(scale=1):
                 gr.HTML('
            ')
                 with gr.Row():
-                    shared.gradio['prompt_menu-notebook'] = gr.Dropdown(choices=utils.get_available_prompts(), value='None', label='Prompt', elem_classes='slim-dropdown')
+                    shared.gradio['prompt_menu-notebook'] = gr.Dropdown(choices=utils.get_available_prompts(), value=shared.settings['prompt-notebook'], label='Prompt', elem_classes='slim-dropdown')
                     ui.create_refresh_button(shared.gradio['prompt_menu-notebook'], lambda: None, lambda: {'choices': utils.get_available_prompts()}, ['refresh-button', 'refresh-button-small'], interactive=not mu)
                     shared.gradio['save_prompt-notebook'] = gr.Button('💾', elem_classes=['refresh-button', 'refresh-button-small'], interactive=not mu)
                     shared.gradio['delete_prompt-notebook'] = gr.Button('🗑️', elem_classes=['refresh-button', 'refresh-button-small'], interactive=not mu)
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 84f9fbfc24..ff965d3990 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -6,19 +6,19 @@
 from modules.utils import gradio
 
 
-def create_ui(default_preset):
+def create_ui():
     mu = shared.args.multi_user
-    generate_params = presets.load_preset(default_preset)
     with gr.Tab("Parameters", elem_id="parameters"):
         with gr.Tab("Generation"):
             with gr.Row():
                 with gr.Column():
                     with gr.Row():
-                        shared.gradio['preset_menu'] = gr.Dropdown(choices=utils.get_available_presets(), value=default_preset, label='Preset', elem_classes='slim-dropdown')
+                        shared.gradio['preset_menu'] = gr.Dropdown(choices=utils.get_available_presets(), value=shared.settings['preset'], label='Preset', elem_classes='slim-dropdown')
                         ui.create_refresh_button(shared.gradio['preset_menu'], lambda: None, lambda: {'choices': utils.get_available_presets()}, 'refresh-button', interactive=not mu)
                         shared.gradio['save_preset'] = gr.Button('💾', elem_classes='refresh-button', interactive=not mu)
                         shared.gradio['delete_preset'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu)
-                        shared.gradio['random_preset'] = gr.Button('🎲', elem_classes='refresh-button')
+                        shared.gradio['reset_preset'] = gr.Button('Reload preset', elem_classes='refresh-button', interactive=True)
+                        shared.gradio['neutralize_samplers'] = gr.Button('Neutralize samplers', elem_classes='refresh-button', interactive=True)
 
                 with gr.Column():
                     shared.gradio['filter_by_loader'] = gr.Dropdown(label="Filter by loader", choices=["All"] + list(loaders.loaders_and_params.keys()) if not shared.args.portable else ['llama.cpp'], value="All", elem_classes='slim-dropdown')
@@ -28,44 +28,44 @@ def create_ui(default_preset):
                     with gr.Row():
                         with gr.Column():
                             gr.Markdown('## Curve shape')
-                            shared.gradio['temperature'] = gr.Slider(0.01, 5, value=generate_params['temperature'], step=0.01, label='temperature')
-                            shared.gradio['dynatemp_low'] = gr.Slider(0.01, 5, value=generate_params['dynatemp_low'], step=0.01, label='dynatemp_low', visible=generate_params['dynamic_temperature'])
-                            shared.gradio['dynatemp_high'] = gr.Slider(0.01, 5, value=generate_params['dynatemp_high'], step=0.01, label='dynatemp_high', visible=generate_params['dynamic_temperature'])
-                            shared.gradio['dynatemp_exponent'] = gr.Slider(0.01, 5, value=generate_params['dynatemp_exponent'], step=0.01, label='dynatemp_exponent', visible=generate_params['dynamic_temperature'])
-                            shared.gradio['smoothing_factor'] = gr.Slider(0.0, 10.0, value=generate_params['smoothing_factor'], step=0.01, label='smoothing_factor', info='Activates Quadratic Sampling.')
-                            shared.gradio['smoothing_curve'] = gr.Slider(1.0, 10.0, value=generate_params['smoothing_curve'], step=0.01, label='smoothing_curve', info='Adjusts the dropoff curve of Quadratic Sampling.')
+                            shared.gradio['temperature'] = gr.Slider(0.01, 5, value=shared.settings['temperature'], step=0.01, label='temperature')
+                            shared.gradio['dynatemp_low'] = gr.Slider(0.01, 5, value=shared.settings['dynatemp_low'], step=0.01, label='dynatemp_low', visible=shared.settings['dynamic_temperature'])
+                            shared.gradio['dynatemp_high'] = gr.Slider(0.01, 5, value=shared.settings['dynatemp_high'], step=0.01, label='dynatemp_high', visible=shared.settings['dynamic_temperature'])
+                            shared.gradio['dynatemp_exponent'] = gr.Slider(0.01, 5, value=shared.settings['dynatemp_exponent'], step=0.01, label='dynatemp_exponent', visible=shared.settings['dynamic_temperature'])
+                            shared.gradio['smoothing_factor'] = gr.Slider(0.0, 10.0, value=shared.settings['smoothing_factor'], step=0.01, label='smoothing_factor', info='Activates Quadratic Sampling.')
+                            shared.gradio['smoothing_curve'] = gr.Slider(1.0, 10.0, value=shared.settings['smoothing_curve'], step=0.01, label='smoothing_curve', info='Adjusts the dropoff curve of Quadratic Sampling.')
 
                             gr.Markdown('## Curve cutoff')
-                            shared.gradio['min_p'] = gr.Slider(0.0, 1.0, value=generate_params['min_p'], step=0.01, label='min_p')
-                            shared.gradio['top_n_sigma'] = gr.Slider(0.0, 5.0, value=generate_params['top_n_sigma'], step=0.01, label='top_n_sigma')
-                            shared.gradio['top_p'] = gr.Slider(0.0, 1.0, value=generate_params['top_p'], step=0.01, label='top_p')
-                            shared.gradio['top_k'] = gr.Slider(0, 200, value=generate_params['top_k'], step=1, label='top_k')
-                            shared.gradio['typical_p'] = gr.Slider(0.0, 1.0, value=generate_params['typical_p'], step=0.01, label='typical_p')
-                            shared.gradio['xtc_threshold'] = gr.Slider(0, 0.5, value=generate_params['xtc_threshold'], step=0.01, label='xtc_threshold', info='If 2 or more tokens have probability above this threshold, consider removing all but the last one.')
-                            shared.gradio['xtc_probability'] = gr.Slider(0, 1, value=generate_params['xtc_probability'], step=0.01, label='xtc_probability', info='Probability that the removal will actually happen. 0 disables the sampler. 1 makes it always happen.')
-                            shared.gradio['epsilon_cutoff'] = gr.Slider(0, 9, value=generate_params['epsilon_cutoff'], step=0.01, label='epsilon_cutoff')
-                            shared.gradio['eta_cutoff'] = gr.Slider(0, 20, value=generate_params['eta_cutoff'], step=0.01, label='eta_cutoff')
-                            shared.gradio['tfs'] = gr.Slider(0.0, 1.0, value=generate_params['tfs'], step=0.01, label='tfs')
-                            shared.gradio['top_a'] = gr.Slider(0.0, 1.0, value=generate_params['top_a'], step=0.01, label='top_a')
+                            shared.gradio['min_p'] = gr.Slider(0.0, 1.0, value=shared.settings['min_p'], step=0.01, label='min_p')
+                            shared.gradio['top_n_sigma'] = gr.Slider(0.0, 5.0, value=shared.settings['top_n_sigma'], step=0.01, label='top_n_sigma')
+                            shared.gradio['top_p'] = gr.Slider(0.0, 1.0, value=shared.settings['top_p'], step=0.01, label='top_p')
+                            shared.gradio['top_k'] = gr.Slider(0, 200, value=shared.settings['top_k'], step=1, label='top_k')
+                            shared.gradio['typical_p'] = gr.Slider(0.0, 1.0, value=shared.settings['typical_p'], step=0.01, label='typical_p')
+                            shared.gradio['xtc_threshold'] = gr.Slider(0, 0.5, value=shared.settings['xtc_threshold'], step=0.01, label='xtc_threshold', info='If 2 or more tokens have probability above this threshold, consider removing all but the last one.')
+                            shared.gradio['xtc_probability'] = gr.Slider(0, 1, value=shared.settings['xtc_probability'], step=0.01, label='xtc_probability', info='Probability that the removal will actually happen. 0 disables the sampler. 1 makes it always happen.')
+                            shared.gradio['epsilon_cutoff'] = gr.Slider(0, 9, value=shared.settings['epsilon_cutoff'], step=0.01, label='epsilon_cutoff')
+                            shared.gradio['eta_cutoff'] = gr.Slider(0, 20, value=shared.settings['eta_cutoff'], step=0.01, label='eta_cutoff')
+                            shared.gradio['tfs'] = gr.Slider(0.0, 1.0, value=shared.settings['tfs'], step=0.01, label='tfs')
+                            shared.gradio['top_a'] = gr.Slider(0.0, 1.0, value=shared.settings['top_a'], step=0.01, label='top_a')
 
                             gr.Markdown('## Repetition suppression')
-                            shared.gradio['dry_multiplier'] = gr.Slider(0, 5, value=generate_params['dry_multiplier'], step=0.01, label='dry_multiplier', info='Set to greater than 0 to enable DRY. Recommended value: 0.8.')
-                            shared.gradio['dry_allowed_length'] = gr.Slider(1, 20, value=generate_params['dry_allowed_length'], step=1, label='dry_allowed_length', info='Longest sequence that can be repeated without being penalized.')
-                            shared.gradio['dry_base'] = gr.Slider(1, 4, value=generate_params['dry_base'], step=0.01, label='dry_base', info='Controls how fast the penalty grows with increasing sequence length.')
-                            shared.gradio['repetition_penalty'] = gr.Slider(1.0, 1.5, value=generate_params['repetition_penalty'], step=0.01, label='repetition_penalty')
-                            shared.gradio['frequency_penalty'] = gr.Slider(0, 2, value=generate_params['frequency_penalty'], step=0.05, label='frequency_penalty')
-                            shared.gradio['presence_penalty'] = gr.Slider(0, 2, value=generate_params['presence_penalty'], step=0.05, label='presence_penalty')
-                            shared.gradio['encoder_repetition_penalty'] = gr.Slider(0.8, 1.5, value=generate_params['encoder_repetition_penalty'], step=0.01, label='encoder_repetition_penalty')
-                            shared.gradio['no_repeat_ngram_size'] = gr.Slider(0, 20, step=1, value=generate_params['no_repeat_ngram_size'], label='no_repeat_ngram_size')
-                            shared.gradio['repetition_penalty_range'] = gr.Slider(0, 4096, step=64, value=generate_params['repetition_penalty_range'], label='repetition_penalty_range')
+                            shared.gradio['dry_multiplier'] = gr.Slider(0, 5, value=shared.settings['dry_multiplier'], step=0.01, label='dry_multiplier', info='Set to greater than 0 to enable DRY. Recommended value: 0.8.')
+                            shared.gradio['dry_allowed_length'] = gr.Slider(1, 20, value=shared.settings['dry_allowed_length'], step=1, label='dry_allowed_length', info='Longest sequence that can be repeated without being penalized.')
+                            shared.gradio['dry_base'] = gr.Slider(1, 4, value=shared.settings['dry_base'], step=0.01, label='dry_base', info='Controls how fast the penalty grows with increasing sequence length.')
+                            shared.gradio['repetition_penalty'] = gr.Slider(1.0, 1.5, value=shared.settings['repetition_penalty'], step=0.01, label='repetition_penalty')
+                            shared.gradio['frequency_penalty'] = gr.Slider(0, 2, value=shared.settings['frequency_penalty'], step=0.05, label='frequency_penalty')
+                            shared.gradio['presence_penalty'] = gr.Slider(0, 2, value=shared.settings['presence_penalty'], step=0.05, label='presence_penalty')
+                            shared.gradio['encoder_repetition_penalty'] = gr.Slider(0.8, 1.5, value=shared.settings['encoder_repetition_penalty'], step=0.01, label='encoder_repetition_penalty')
+                            shared.gradio['no_repeat_ngram_size'] = gr.Slider(0, 20, step=1, value=shared.settings['no_repeat_ngram_size'], label='no_repeat_ngram_size')
+                            shared.gradio['repetition_penalty_range'] = gr.Slider(0, 4096, step=64, value=shared.settings['repetition_penalty_range'], label='repetition_penalty_range')
 
                         with gr.Column():
                             gr.Markdown('## Alternative sampling methods')
-                            shared.gradio['penalty_alpha'] = gr.Slider(0, 5, value=generate_params['penalty_alpha'], label='penalty_alpha', info='For Contrastive Search. do_sample must be unchecked.')
-                            shared.gradio['guidance_scale'] = gr.Slider(-0.5, 2.5, step=0.05, value=generate_params['guidance_scale'], label='guidance_scale', info='For CFG. 1.5 is a good value.')
-                            shared.gradio['mirostat_mode'] = gr.Slider(0, 2, step=1, value=generate_params['mirostat_mode'], label='mirostat_mode', info='mode=1 is for llama.cpp only.')
-                            shared.gradio['mirostat_tau'] = gr.Slider(0, 10, step=0.01, value=generate_params['mirostat_tau'], label='mirostat_tau')
-                            shared.gradio['mirostat_eta'] = gr.Slider(0, 1, step=0.01, value=generate_params['mirostat_eta'], label='mirostat_eta')
+                            shared.gradio['penalty_alpha'] = gr.Slider(0, 5, value=shared.settings['penalty_alpha'], label='penalty_alpha', info='For Contrastive Search. do_sample must be unchecked.')
+                            shared.gradio['guidance_scale'] = gr.Slider(-0.5, 2.5, step=0.05, value=shared.settings['guidance_scale'], label='guidance_scale', info='For CFG. 1.5 is a good value.')
+                            shared.gradio['mirostat_mode'] = gr.Slider(0, 2, step=1, value=shared.settings['mirostat_mode'], label='mirostat_mode', info='mode=1 is for llama.cpp only.')
+                            shared.gradio['mirostat_tau'] = gr.Slider(0, 10, step=0.01, value=shared.settings['mirostat_tau'], label='mirostat_tau')
+                            shared.gradio['mirostat_eta'] = gr.Slider(0, 1, step=0.01, value=shared.settings['mirostat_eta'], label='mirostat_eta')
 
                             gr.Markdown('## Other options')
                             shared.gradio['max_new_tokens'] = gr.Slider(minimum=shared.settings['max_new_tokens_min'], maximum=shared.settings['max_new_tokens_max'], value=shared.settings['max_new_tokens'], step=1, label='max_new_tokens', info='⚠️ Setting this too high can cause prompt truncation.')
@@ -74,9 +74,9 @@ def create_ui(default_preset):
                 with gr.Column():
                     with gr.Row():
                         with gr.Column():
-                            shared.gradio['do_sample'] = gr.Checkbox(value=generate_params['do_sample'], label='do_sample')
-                            shared.gradio['dynamic_temperature'] = gr.Checkbox(value=generate_params['dynamic_temperature'], label='dynamic_temperature')
-                            shared.gradio['temperature_last'] = gr.Checkbox(value=generate_params['temperature_last'], label='temperature_last', info='Moves temperature/dynamic temperature/quadratic sampling to the end of the sampler stack, ignoring their positions in "Sampler priority".')
+                            shared.gradio['do_sample'] = gr.Checkbox(value=shared.settings['do_sample'], label='do_sample')
+                            shared.gradio['dynamic_temperature'] = gr.Checkbox(value=shared.settings['dynamic_temperature'], label='dynamic_temperature')
+                            shared.gradio['temperature_last'] = gr.Checkbox(value=shared.settings['temperature_last'], label='temperature_last', info='Moves temperature/dynamic temperature/quadratic sampling to the end of the sampler stack, ignoring their positions in "Sampler priority".')
                             shared.gradio['auto_max_new_tokens'] = gr.Checkbox(value=shared.settings['auto_max_new_tokens'], label='auto_max_new_tokens', info='Expand max_new_tokens to the available context length.')
                             shared.gradio['ban_eos_token'] = gr.Checkbox(value=shared.settings['ban_eos_token'], label='Ban the eos_token', info='Forces the model to never end the generation prematurely.')
                             shared.gradio['add_bos_token'] = gr.Checkbox(value=shared.settings['add_bos_token'], label='Add the bos_token to the beginning of prompts', info='Disabling this can make the replies more creative.')
@@ -89,18 +89,18 @@ def create_ui(default_preset):
                             shared.gradio['truncation_length'] = gr.Number(precision=0, step=256, value=get_truncation_length(), label='Truncate the prompt up to this length', info='The leftmost tokens are removed if the prompt exceeds this length.')
                             shared.gradio['seed'] = gr.Number(value=shared.settings['seed'], label='Seed (-1 for random)')
 
-                            shared.gradio['sampler_priority'] = gr.Textbox(value=generate_params['sampler_priority'], lines=12, label='Sampler priority', info='Parameter names separated by new lines or commas.', elem_classes=['add_scrollbar'])
+                            shared.gradio['sampler_priority'] = gr.Textbox(value=shared.settings['sampler_priority'], lines=12, label='Sampler priority', info='Parameter names separated by new lines or commas.', elem_classes=['add_scrollbar'])
                             shared.gradio['custom_stopping_strings'] = gr.Textbox(lines=2, value=shared.settings["custom_stopping_strings"] or None, label='Custom stopping strings', info='Written between "" and separated by commas.', placeholder='"\\n", "\\nYou:"')
                             shared.gradio['custom_token_bans'] = gr.Textbox(value=shared.settings['custom_token_bans'] or None, label='Token bans', info='Token IDs to ban, separated by commas. The IDs can be found in the Default or Notebook tab.')
                             shared.gradio['negative_prompt'] = gr.Textbox(value=shared.settings['negative_prompt'], label='Negative prompt', info='For CFG. Only used when guidance_scale is different than 1.', lines=3, elem_classes=['add_scrollbar'])
-                            shared.gradio['dry_sequence_breakers'] = gr.Textbox(value=generate_params['dry_sequence_breakers'], label='dry_sequence_breakers', info='Tokens across which sequence matching is not continued. Specified as a comma-separated list of quoted strings.')
+                            shared.gradio['dry_sequence_breakers'] = gr.Textbox(value=shared.settings['dry_sequence_breakers'], label='dry_sequence_breakers', info='Tokens across which sequence matching is not continued. Specified as a comma-separated list of quoted strings.')
                             with gr.Row() as shared.gradio['grammar_file_row']:
                                 shared.gradio['grammar_file'] = gr.Dropdown(value='None', choices=utils.get_available_grammars(), label='Load grammar from file (.gbnf)', elem_classes='slim-dropdown')
                                 ui.create_refresh_button(shared.gradio['grammar_file'], lambda: None, lambda: {'choices': utils.get_available_grammars()}, 'refresh-button', interactive=not mu)
                                 shared.gradio['save_grammar'] = gr.Button('💾', elem_classes='refresh-button', interactive=not mu)
                                 shared.gradio['delete_grammar'] = gr.Button('🗑️ ', elem_classes='refresh-button', interactive=not mu)
 
-                            shared.gradio['grammar_string'] = gr.Textbox(value='', label='Grammar', lines=16, elem_classes=['add_scrollbar', 'monospace'])
+                            shared.gradio['grammar_string'] = gr.Textbox(value=shared.settings['grammar_string'], label='Grammar', lines=16, elem_classes=['add_scrollbar', 'monospace'])
 
         ui_chat.create_chat_settings_ui()
 
@@ -111,9 +111,13 @@ def create_event_handlers():
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         presets.load_preset_for_ui, gradio('preset_menu', 'interface_state'), gradio('interface_state') + gradio(presets.presets_params()), show_progress=False)
 
-    shared.gradio['random_preset'].click(
+    shared.gradio['reset_preset'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        presets.random_preset, gradio('interface_state'), gradio('interface_state') + gradio(presets.presets_params()), show_progress=False)
+        presets.reset_preset_for_ui, gradio('preset_menu', 'interface_state'), gradio('interface_state') + gradio(presets.presets_params()), show_progress=False)
+
+    shared.gradio['neutralize_samplers'].click(
+        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
+        presets.neutralize_samplers_for_ui, gradio('interface_state'), gradio('interface_state') + gradio(presets.presets_params()), show_progress=False)
 
     shared.gradio['grammar_file'].change(load_grammar, gradio('grammar_file'), gradio('grammar_string'), show_progress=False)
     shared.gradio['dynamic_temperature'].change(lambda x: [gr.update(visible=x)] * 3, gradio('dynamic_temperature'), gradio('dynatemp_low', 'dynatemp_high', 'dynatemp_exponent'), show_progress=False)
diff --git a/server.py b/server.py
index 3be4c27c80..3dd0a3f310 100644
--- a/server.py
+++ b/server.py
@@ -45,6 +45,7 @@
     ui_session,
     utils
 )
+from modules.chat import generate_pfp_cache
 from modules.extensions import apply_extensions
 from modules.LoRA import add_lora_to_model
 from modules.models import load_model, unload_model_if_idle
@@ -93,17 +94,20 @@ def create_interface():
 
     # Force some events to be triggered on page load
     shared.persistent_interface_state.update({
+        'mode': shared.settings['mode'],
         'loader': shared.args.loader or 'llama.cpp',
-        'mode': shared.settings['mode'] if shared.settings['mode'] == 'instruct' else gr.update(),
-        'character_menu': shared.args.character or shared.settings['character'],
-        'instruction_template_str': shared.settings['instruction_template_str'],
-        'prompt_menu-default': shared.settings['prompt-default'],
-        'prompt_menu-notebook': shared.settings['prompt-notebook'],
         'filter_by_loader': (shared.args.loader or 'All') if not shared.args.portable else 'llama.cpp'
     })
 
-    if Path("user_data/cache/pfp_character.png").exists():
-        Path("user_data/cache/pfp_character.png").unlink()
+    # Clear existing cache files
+    for cache_file in ['pfp_character.png', 'pfp_character_thumb.png']:
+        cache_path = Path(f"user_data/cache/{cache_file}")
+        if cache_path.exists():
+            cache_path.unlink()
+
+    # Regenerate for default character
+    if shared.settings['mode'] != 'instruct':
+        generate_pfp_cache(shared.settings['character'])
 
     # css/js strings
     css = ui.css
@@ -134,7 +138,7 @@ def create_interface():
         ui_default.create_ui()
         ui_notebook.create_ui()
 
-        ui_parameters.create_ui(shared.settings['preset'])  # Parameters tab
+        ui_parameters.create_ui()  # Parameters tab
         ui_model_menu.create_ui()  # Model tab
         if not shared.args.portable:
             training.create_ui()  # Training tab

From 3650a6fd1fc332f462bc1aa08f7671a3d517d847 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Jun 2025 22:02:34 -0700
Subject: [PATCH 0627/1701] Small UI changes

---
 css/main.css             | 2 +-
 modules/ui_chat.py       | 3 +--
 modules/ui_parameters.py | 2 +-
 3 files changed, 3 insertions(+), 4 deletions(-)

diff --git a/css/main.css b/css/main.css
index a9cb36ab8e..4277519a91 100644
--- a/css/main.css
+++ b/css/main.css
@@ -53,7 +53,7 @@ div.svelte-iyf88w {
 }
 
 .refresh-button {
-    max-width: 4.4em;
+    max-width: none;
     min-width: 2.2em !important;
     height: 39.594px;
     align-self: end;
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 712843d3c2..f8e2fc32a1 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -122,8 +122,7 @@ def create_chat_settings_ui():
                         ui.create_refresh_button(shared.gradio['character_menu'], lambda: None, lambda: {'choices': utils.get_available_characters()}, 'refresh-button', interactive=not mu)
                         shared.gradio['save_character'] = gr.Button('💾', elem_classes='refresh-button', elem_id="save-character", interactive=not mu)
                         shared.gradio['delete_character'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu)
-                        shared.gradio['reset_character'] = gr.Button('↺', elem_classes='refresh-button', interactive=True)
-                        shared.gradio['clear_character'] = gr.Button('⚪', elem_classes='refresh-button', interactive=True)
+                        shared.gradio['reset_character'] = gr.Button('Restore character', elem_classes='refresh-button', interactive=True)
 
                     shared.gradio['name2'] = gr.Textbox(value=shared.settings['name2'], lines=1, label='Character\'s name')
                     shared.gradio['context'] = gr.Textbox(value=shared.settings['context'], lines=10, label='Context', elem_classes=['add_scrollbar'])
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index ff965d3990..9b5cb3abc3 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -17,7 +17,7 @@ def create_ui():
                         ui.create_refresh_button(shared.gradio['preset_menu'], lambda: None, lambda: {'choices': utils.get_available_presets()}, 'refresh-button', interactive=not mu)
                         shared.gradio['save_preset'] = gr.Button('💾', elem_classes='refresh-button', interactive=not mu)
                         shared.gradio['delete_preset'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu)
-                        shared.gradio['reset_preset'] = gr.Button('Reload preset', elem_classes='refresh-button', interactive=True)
+                        shared.gradio['reset_preset'] = gr.Button('Restore preset', elem_classes='refresh-button', interactive=True)
                         shared.gradio['neutralize_samplers'] = gr.Button('Neutralize samplers', elem_classes='refresh-button', interactive=True)
 
                 with gr.Column():

From caf9fca5f339e4018913608f8accb3897352853f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Jun 2025 22:11:35 -0700
Subject: [PATCH 0628/1701] Avoid some code repetition

---
 modules/shared.py | 190 +++++++++++++++++++++++-----------------------
 1 file changed, 97 insertions(+), 93 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index f4f8e18004..082003992f 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -9,6 +9,7 @@
 import yaml
 
 from modules.logging_colors import logger
+from modules.presets import default_preset
 
 # Model variables
 model = None
@@ -28,99 +29,6 @@
 persistent_interface_state = {}
 need_restart = False
 
-# UI defaults
-settings = {
-    'show_controls': True,
-    'start_with': '',
-    'mode': 'instruct',
-    'chat_style': 'cai-chat',
-    'chat-instruct_command': 'Continue the chat dialogue below. Write a single reply for the character "<|character|>".\n\n<|prompt|>',
-    'enable_web_search': False,
-    'web_search_pages': 3,
-    'prompt-default': 'QA',
-    'prompt-notebook': 'QA',
-    'preset': 'min_p',
-    'max_new_tokens': 512,
-    'max_new_tokens_min': 1,
-    'max_new_tokens_max': 4096,
-    'prompt_lookup_num_tokens': 0,
-    'max_tokens_second': 0,
-    'auto_max_new_tokens': True,
-    'ban_eos_token': False,
-    'add_bos_token': True,
-    'enable_thinking': True,
-    'skip_special_tokens': True,
-    'stream': True,
-    'static_cache': False,
-    'truncation_length': 8192,
-    'seed': -1,
-    'custom_stopping_strings': '',
-    'custom_token_bans': '',
-    'negative_prompt': '',
-    'dark_theme': True,
-    'default_extensions': [],
-
-    # Character settings
-    'character': 'Assistant',
-    'name1': 'You',
-    'name2': 'AI',
-    'user_bio': '',
-    'context': 'The following is a conversation with an AI Large Language Model. The AI has been trained to answer questions, provide recommendations, and help with decision making. The AI follows user requests. The AI thinks outside the box.',
-    'greeting': 'How can I help you today?',
-    'custom_system_message': '',
-    'instruction_template_str': "{%- set ns = namespace(found=false) -%}\n{%- for message in messages -%}\n    {%- if message['role'] == 'system' -%}\n        {%- set ns.found = true -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if not ns.found -%}\n    {{- '' + 'Below is an instruction that describes a task. Write a response that appropriately completes the request.' + '\\n\\n' -}}\n{%- endif %}\n{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {{- '' + message['content'] + '\\n\\n' -}}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{-'### Instruction:\\n' + message['content'] + '\\n\\n'-}}\n        {%- else -%}\n            {{-'### Response:\\n' + message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{-'### Response:\\n'-}}\n{%- endif -%}",
-    'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{- name1 + ': ' + message['content'] + '\\n'-}}\n        {%- else -%}\n            {{- name2 + ': ' + message['content'] + '\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}",
-
-    # Generation parameters - Curve shape
-    'temperature': 1.0,
-    'dynatemp_low': 1.0,
-    'dynatemp_high': 1.0,
-    'dynatemp_exponent': 1.0,
-    'smoothing_factor': 0.0,
-    'smoothing_curve': 1.0,
-
-    # Generation parameters - Curve cutoff
-    'min_p': 0.0,
-    'top_p': 1.0,
-    'top_k': 0,
-    'typical_p': 1.0,
-    'xtc_threshold': 0.1,
-    'xtc_probability': 0.0,
-    'epsilon_cutoff': 0.0,
-    'eta_cutoff': 0.0,
-    'tfs': 1.0,
-    'top_a': 0.0,
-    'top_n_sigma': 0.0,
-
-    # Generation parameters - Repetition suppression
-    'dry_multiplier': 0.0,
-    'dry_allowed_length': 2,
-    'dry_base': 1.75,
-    'repetition_penalty': 1.0,
-    'frequency_penalty': 0.0,
-    'presence_penalty': 0.0,
-    'encoder_repetition_penalty': 1.0,
-    'no_repeat_ngram_size': 0,
-    'repetition_penalty_range': 1024,
-
-    # Generation parameters - Alternative sampling methods
-    'penalty_alpha': 0.0,
-    'guidance_scale': 1.0,
-    'mirostat_mode': 0,
-    'mirostat_tau': 5.0,
-    'mirostat_eta': 0.1,
-
-    # Generation parameters - Other options
-    'do_sample': True,
-    'dynamic_temperature': False,
-    'temperature_last': False,
-    'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ndry\ntop_n_sigma\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nmirostat\nxtc\nencoder_repetition_penalty\nno_repeat_ngram',
-    'dry_sequence_breakers': '"\\n", ":", "\\"", "*"',
-    'grammar_string': '',
-}
-
-default_settings = copy.deepcopy(settings)
-
 # Parser copied from https://github.com/vladmandic/automatic
 parser = argparse.ArgumentParser(description="Text generation web UI", conflict_handler='resolve', add_help=True, formatter_class=lambda prog: argparse.HelpFormatter(prog, max_help_position=55, indent_increment=2, width=200))
 
@@ -282,6 +190,102 @@
     elif hasattr(args, arg):
         provided_arguments.append(arg)
 
+# Default generation parameters
+neutral_samplers = default_preset()
+
+# UI defaults
+settings = {
+    'show_controls': True,
+    'start_with': '',
+    'mode': 'instruct',
+    'chat_style': 'cai-chat',
+    'chat-instruct_command': 'Continue the chat dialogue below. Write a single reply for the character "<|character|>".\n\n<|prompt|>',
+    'enable_web_search': False,
+    'web_search_pages': 3,
+    'prompt-default': 'QA',
+    'prompt-notebook': 'QA',
+    'preset': 'min_p',
+    'max_new_tokens': 512,
+    'max_new_tokens_min': 1,
+    'max_new_tokens_max': 4096,
+    'prompt_lookup_num_tokens': 0,
+    'max_tokens_second': 0,
+    'auto_max_new_tokens': True,
+    'ban_eos_token': False,
+    'add_bos_token': True,
+    'enable_thinking': True,
+    'skip_special_tokens': True,
+    'stream': True,
+    'static_cache': False,
+    'truncation_length': 8192,
+    'seed': -1,
+    'custom_stopping_strings': '',
+    'custom_token_bans': '',
+    'negative_prompt': '',
+    'dark_theme': True,
+    'default_extensions': [],
+
+    # Character settings
+    'character': 'Assistant',
+    'name1': 'You',
+    'name2': 'AI',
+    'user_bio': '',
+    'context': 'The following is a conversation with an AI Large Language Model. The AI has been trained to answer questions, provide recommendations, and help with decision making. The AI follows user requests. The AI thinks outside the box.',
+    'greeting': 'How can I help you today?',
+    'custom_system_message': '',
+    'instruction_template_str': "{%- set ns = namespace(found=false) -%}\n{%- for message in messages -%}\n    {%- if message['role'] == 'system' -%}\n        {%- set ns.found = true -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if not ns.found -%}\n    {{- '' + 'Below is an instruction that describes a task. Write a response that appropriately completes the request.' + '\\n\\n' -}}\n{%- endif %}\n{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {{- '' + message['content'] + '\\n\\n' -}}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{-'### Instruction:\\n' + message['content'] + '\\n\\n'-}}\n        {%- else -%}\n            {{-'### Response:\\n' + message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{-'### Response:\\n'-}}\n{%- endif -%}",
+    'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{- name1 + ': ' + message['content'] + '\\n'-}}\n        {%- else -%}\n            {{- name2 + ': ' + message['content'] + '\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}",
+
+    # Generation parameters - Curve shape
+    'temperature': neutral_samplers['temperature'],
+    'dynatemp_low': neutral_samplers['dynatemp_low'],
+    'dynatemp_high': neutral_samplers['dynatemp_high'],
+    'dynatemp_exponent': neutral_samplers['dynatemp_exponent'],
+    'smoothing_factor': neutral_samplers['smoothing_factor'],
+    'smoothing_curve': neutral_samplers['smoothing_curve'],
+
+    # Generation parameters - Curve cutoff
+    'min_p': neutral_samplers['min_p'],
+    'top_p': neutral_samplers['top_p'],
+    'top_k': neutral_samplers['top_k'],
+    'typical_p': neutral_samplers['typical_p'],
+    'xtc_threshold': neutral_samplers['xtc_threshold'],
+    'xtc_probability': neutral_samplers['xtc_probability'],
+    'epsilon_cutoff': neutral_samplers['epsilon_cutoff'],
+    'eta_cutoff': neutral_samplers['eta_cutoff'],
+    'tfs': neutral_samplers['tfs'],
+    'top_a': neutral_samplers['top_a'],
+    'top_n_sigma': neutral_samplers['top_n_sigma'],
+
+    # Generation parameters - Repetition suppression
+    'dry_multiplier': neutral_samplers['dry_multiplier'],
+    'dry_allowed_length': neutral_samplers['dry_allowed_length'],
+    'dry_base': neutral_samplers['dry_base'],
+    'repetition_penalty': neutral_samplers['repetition_penalty'],
+    'frequency_penalty': neutral_samplers['frequency_penalty'],
+    'presence_penalty': neutral_samplers['presence_penalty'],
+    'encoder_repetition_penalty': neutral_samplers['encoder_repetition_penalty'],
+    'no_repeat_ngram_size': neutral_samplers['no_repeat_ngram_size'],
+    'repetition_penalty_range': neutral_samplers['repetition_penalty_range'],
+
+    # Generation parameters - Alternative sampling methods
+    'penalty_alpha': neutral_samplers['penalty_alpha'],
+    'guidance_scale': neutral_samplers['guidance_scale'],
+    'mirostat_mode': neutral_samplers['mirostat_mode'],
+    'mirostat_tau': neutral_samplers['mirostat_tau'],
+    'mirostat_eta': neutral_samplers['mirostat_eta'],
+
+    # Generation parameters - Other options
+    'do_sample': neutral_samplers['do_sample'],
+    'dynamic_temperature': neutral_samplers['dynamic_temperature'],
+    'temperature_last': neutral_samplers['temperature_last'],
+    'sampler_priority': neutral_samplers['sampler_priority'],
+    'dry_sequence_breakers': neutral_samplers['dry_sequence_breakers'],
+    'grammar_string': '',
+}
+
+default_settings = copy.deepcopy(settings)
+
 
 def do_cmd_flags_warnings():
     # Security warnings

From fe955cac1fd7298d820e42790e257d9ace1c5eb4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Jun 2025 22:15:19 -0700
Subject: [PATCH 0629/1701] Small UI changes

---
 css/main.css       | 4 ++++
 modules/ui_chat.py | 8 ++++----
 2 files changed, 8 insertions(+), 4 deletions(-)

diff --git a/css/main.css b/css/main.css
index 4277519a91..7d7b0cbd81 100644
--- a/css/main.css
+++ b/css/main.css
@@ -62,6 +62,10 @@ div.svelte-iyf88w {
     flex: none;
 }
 
+.refresh-button-medium {
+    max-width: 4.4em;
+}
+
 .refresh-button-small {
     max-width: 2.2em;
 }
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index f8e2fc32a1..ef9330e467 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -24,11 +24,11 @@ def create_ui():
         with gr.Row(elem_id='past-chats-row', elem_classes=['pretty_scrollbar']):
             with gr.Column():
                 with gr.Row(elem_id='past-chats-buttons'):
-                    shared.gradio['branch_chat'] = gr.Button('Branch', elem_classes='refresh-button', elem_id='Branch', interactive=not mu)
-                    shared.gradio['branch_index'] = gr.Number(value=-1, precision=0, visible=False, elem_id="Branch-index", interactive=True)
-                    shared.gradio['rename_chat'] = gr.Button('Rename', elem_classes='refresh-button', interactive=not mu)
+                    shared.gradio['branch_chat'] = gr.Button('Branch', elem_classes=['refresh-button', 'refresh-button-medium'], elem_id='Branch', interactive=not mu)
+                    shared.gradio['rename_chat'] = gr.Button('Rename', elem_classes=['refresh-button', 'refresh-button-medium'], interactive=not mu)
                     shared.gradio['delete_chat'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu)
-                    shared.gradio['Start new chat'] = gr.Button('New chat', elem_classes=['refresh-button', 'focus-on-chat-input'])
+                    shared.gradio['Start new chat'] = gr.Button('New chat', elem_classes=['refresh-button', 'refresh-button-medium', 'focus-on-chat-input'])
+                    shared.gradio['branch_index'] = gr.Number(value=-1, precision=0, visible=False, elem_id="Branch-index", interactive=True)
 
                 shared.gradio['search_chat'] = gr.Textbox(placeholder='Search chats...', max_lines=1, elem_id='search_chat')
 

From 0dbc4cbc71e7eed1d51fd9169048a54cb8d2d927 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Jun 2025 22:20:58 -0700
Subject: [PATCH 0630/1701] Add Qwen3 presets

---
 user_data/presets/Qwen3 - No Thinking.yaml | 3 +++
 user_data/presets/Qwen3 - Thinking.yaml    | 3 +++
 2 files changed, 6 insertions(+)
 create mode 100644 user_data/presets/Qwen3 - No Thinking.yaml
 create mode 100644 user_data/presets/Qwen3 - Thinking.yaml

diff --git a/user_data/presets/Qwen3 - No Thinking.yaml b/user_data/presets/Qwen3 - No Thinking.yaml
new file mode 100644
index 0000000000..b1c1e03c44
--- /dev/null
+++ b/user_data/presets/Qwen3 - No Thinking.yaml	
@@ -0,0 +1,3 @@
+temperature: 0.7
+top_p: 0.8
+top_k: 20
diff --git a/user_data/presets/Qwen3 - Thinking.yaml b/user_data/presets/Qwen3 - Thinking.yaml
new file mode 100644
index 0000000000..cb2942f905
--- /dev/null
+++ b/user_data/presets/Qwen3 - Thinking.yaml	
@@ -0,0 +1,3 @@
+temperature: 0.6
+top_p: 0.95
+top_k: 20

From 1bdf11b511b69f6eb10f9067847c64d39c8872f4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Jun 2025 22:23:09 -0700
Subject: [PATCH 0631/1701] Use the Qwen3 - Thinking preset by default

---
 modules/shared.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 082003992f..3794206b8d 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -204,7 +204,7 @@
     'web_search_pages': 3,
     'prompt-default': 'QA',
     'prompt-notebook': 'QA',
-    'preset': 'min_p',
+    'preset': 'Qwen3 - Thinking' if Path('user_data/presets/Qwen3 - Thinking.yaml').exists() else '',
     'max_new_tokens': 512,
     'max_new_tokens_min': 1,
     'max_new_tokens_max': 4096,
@@ -237,7 +237,7 @@
     'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{- name1 + ': ' + message['content'] + '\\n'-}}\n        {%- else -%}\n            {{- name2 + ': ' + message['content'] + '\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}",
 
     # Generation parameters - Curve shape
-    'temperature': neutral_samplers['temperature'],
+    'temperature': 0.6,
     'dynatemp_low': neutral_samplers['dynatemp_low'],
     'dynatemp_high': neutral_samplers['dynatemp_high'],
     'dynatemp_exponent': neutral_samplers['dynatemp_exponent'],
@@ -246,8 +246,8 @@
 
     # Generation parameters - Curve cutoff
     'min_p': neutral_samplers['min_p'],
-    'top_p': neutral_samplers['top_p'],
-    'top_k': neutral_samplers['top_k'],
+    'top_p': 0.95,
+    'top_k': 20,
     'typical_p': neutral_samplers['typical_p'],
     'xtc_threshold': neutral_samplers['xtc_threshold'],
     'xtc_probability': neutral_samplers['xtc_probability'],

From ae150fa24f204fb56c9fe4e2530435bfcf0ec1d7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Jun 2025 22:25:46 -0700
Subject: [PATCH 0632/1701] Remove the null preset

---
 user_data/presets/Null preset.yaml | 1 -
 1 file changed, 1 deletion(-)
 delete mode 100644 user_data/presets/Null preset.yaml

diff --git a/user_data/presets/Null preset.yaml b/user_data/presets/Null preset.yaml
deleted file mode 100644
index 714aa9a3ed..0000000000
--- a/user_data/presets/Null preset.yaml	
+++ /dev/null
@@ -1 +0,0 @@
-temperature: 1

From 1cab149c1a020c86f215bafd48b809cbc02a225b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Jun 2025 22:26:13 -0700
Subject: [PATCH 0633/1701] Remove the contrastive search preset

---
 user_data/presets/Contrastive Search.yaml | 3 ---
 1 file changed, 3 deletions(-)
 delete mode 100644 user_data/presets/Contrastive Search.yaml

diff --git a/user_data/presets/Contrastive Search.yaml b/user_data/presets/Contrastive Search.yaml
deleted file mode 100644
index d9a47a9f5b..0000000000
--- a/user_data/presets/Contrastive Search.yaml	
+++ /dev/null
@@ -1,3 +0,0 @@
-do_sample: false
-top_k: 4
-penalty_alpha: 0.3

From af6bb7513afc6a45cba1716a8cd665d4ee264779 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 09:09:36 -0700
Subject: [PATCH 0634/1701] Add back the "Save UI defaults" button

It's useful for saving extensions settings.
---
 modules/ui_session.py | 21 ++++++++++++++++++---
 1 file changed, 18 insertions(+), 3 deletions(-)

diff --git a/modules/ui_session.py b/modules/ui_session.py
index 4ed740cd70..db9ce3b0cf 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -12,6 +12,7 @@ def create_ui():
             with gr.Column():
                 with gr.Row():
                     shared.gradio['toggle_dark_mode'] = gr.Button('Toggle 💡')
+                    shared.gradio['save_settings'] = gr.Button('Save UI defaults to user_data/settings.yaml', interactive=not mu)
 
                 shared.gradio['reset_interface'] = gr.Button("Apply flags/extensions and restart", interactive=not mu)
                 with gr.Row():
@@ -32,14 +33,28 @@ def create_ui():
         if not shared.args.portable:
             extension_name.submit(clone_or_pull_repository, extension_name, extension_status, show_progress=False)
 
+        shared.gradio['save_settings'].click(
+            ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
+            handle_save_settings, gradio('interface_state', 'preset_menu', 'extensions_menu', 'show_controls', 'theme_state'), gradio('save_contents', 'save_filename', 'save_root', 'file_saver'), show_progress=False)
+
+        shared.gradio['toggle_dark_mode'].click(
+            lambda x: 'dark' if x == 'light' else 'light', gradio('theme_state'), gradio('theme_state')).then(
+            None, None, None, js=f'() => {{{ui.dark_theme_js}; toggleDarkMode()}}')
+
         # Reset interface event
         shared.gradio['reset_interface'].click(
             set_interface_arguments, gradio('extensions_menu', 'bool_menu'), None).then(
             None, None, None, js='() => {document.body.innerHTML=\'
            Reloading...
            \'; setTimeout(function(){location.reload()},2500); return []}')
 
-        shared.gradio['toggle_dark_mode'].click(
-            lambda x: 'dark' if x == 'light' else 'light', gradio('theme_state'), gradio('theme_state')).then(
-            None, None, None, js=f'() => {{{ui.dark_theme_js}; toggleDarkMode()}}')
+
+def handle_save_settings(state, preset, extensions, show_controls, theme):
+    contents = ui.save_settings(state, preset, extensions, show_controls, theme)
+    return [
+        contents,
+        "settings.yaml",
+        "user_data/",
+        gr.update(visible=True)
+    ]
 
 
 def set_interface_arguments(extensions, bool_active):

From 42e7864d6242745263074a08bb4696059650104e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 09:20:21 -0700
Subject: [PATCH 0635/1701] Reorganize the Session tab

---
 modules/github.py     | 38 --------------------------------------
 modules/ui_session.py | 19 +++++--------------
 modules/utils.py      |  3 +--
 3 files changed, 6 insertions(+), 54 deletions(-)
 delete mode 100644 modules/github.py

diff --git a/modules/github.py b/modules/github.py
deleted file mode 100644
index f3dc26e1bb..0000000000
--- a/modules/github.py
+++ /dev/null
@@ -1,38 +0,0 @@
-import subprocess
-from pathlib import Path
-
-new_extensions = set()
-
-
-def clone_or_pull_repository(github_url):
-    global new_extensions
-
-    repository_folder = Path("extensions")
-    repo_name = github_url.rstrip("/").split("/")[-1].split(".")[0]
-
-    # Check if the repository folder exists
-    if not repository_folder.exists():
-        repository_folder.mkdir(parents=True)
-
-    repo_path = repository_folder / repo_name
-
-    # Check if the repository is already cloned
-    if repo_path.exists():
-        yield f"Updating {github_url}..."
-        # Perform a 'git pull' to update the repository
-        try:
-            pull_output = subprocess.check_output(["git", "-C", repo_path, "pull"], stderr=subprocess.STDOUT)
-            yield "Done."
-            return pull_output.decode()
-        except subprocess.CalledProcessError as e:
-            return str(e)
-
-    # Clone the repository
-    try:
-        yield f"Cloning {github_url}..."
-        clone_output = subprocess.check_output(["git", "clone", github_url, repo_path], stderr=subprocess.STDOUT)
-        new_extensions.add(repo_name)
-        yield f"The extension `{repo_name}` has been downloaded.\n\nPlease close the web UI completely and launch it again to be able to load it."
-        return clone_output.decode()
-    except subprocess.CalledProcessError as e:
-        return str(e)
diff --git a/modules/ui_session.py b/modules/ui_session.py
index db9ce3b0cf..086a06bfae 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -1,7 +1,6 @@
 import gradio as gr
 
 from modules import shared, ui, utils
-from modules.github import clone_or_pull_repository
 from modules.utils import gradio
 
 
@@ -10,10 +9,12 @@ def create_ui():
     with gr.Tab("Session", elem_id="session-tab"):
         with gr.Row():
             with gr.Column():
-                with gr.Row():
-                    shared.gradio['toggle_dark_mode'] = gr.Button('Toggle 💡')
-                    shared.gradio['save_settings'] = gr.Button('Save UI defaults to user_data/settings.yaml', interactive=not mu)
+                gr.Markdown("## Settings")
+                shared.gradio['save_settings'] = gr.Button('Save settings to user_data/settings.yaml', elem_classes='refresh-button', interactive=not mu)
+                shared.gradio['toggle_dark_mode'] = gr.Button('Toggle light/dark theme 💡', elem_classes='refresh-button')
 
+            with gr.Column():
+                gr.Markdown("## Extensions & flags")
                 shared.gradio['reset_interface'] = gr.Button("Apply flags/extensions and restart", interactive=not mu)
                 with gr.Row():
                     with gr.Column():
@@ -22,17 +23,7 @@ def create_ui():
                     with gr.Column():
                         shared.gradio['bool_menu'] = gr.CheckboxGroup(choices=get_boolean_arguments(), value=get_boolean_arguments(active=True), label="Boolean command-line flags", elem_classes='checkboxgroup-table')
 
-            with gr.Column():
-                if not shared.args.portable:
-                    extension_name = gr.Textbox(lines=1, label='Install or update an extension', info='Enter the GitHub URL below and press Enter. For a list of extensions, see: https://github.com/oobabooga/text-generation-webui-extensions ⚠️  WARNING ⚠️ : extensions can execute arbitrary code. Make sure to inspect their source code before activating them.', interactive=not mu)
-                    extension_status = gr.Markdown()
-                else:
-                    pass
-
         shared.gradio['theme_state'] = gr.Textbox(visible=False, value='dark' if shared.settings['dark_theme'] else 'light')
-        if not shared.args.portable:
-            extension_name.submit(clone_or_pull_repository, extension_name, extension_status, show_progress=False)
-
         shared.gradio['save_settings'].click(
             ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
             handle_save_settings, gradio('interface_state', 'preset_menu', 'extensions_menu', 'show_controls', 'theme_state'), gradio('save_contents', 'save_filename', 'save_root', 'file_saver'), show_progress=False)
diff --git a/modules/utils.py b/modules/utils.py
index 577c55b8d1..218735415e 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -3,7 +3,7 @@
 from datetime import datetime
 from pathlib import Path
 
-from modules import github, shared
+from modules import shared
 from modules.logging_colors import logger
 
 
@@ -182,7 +182,6 @@ def get_available_instruction_templates():
 
 def get_available_extensions():
     extensions = sorted(set(map(lambda x: x.parts[1], Path('extensions').glob('*/script.py'))), key=natural_keys)
-    extensions = [v for v in extensions if v not in github.new_extensions]
     return extensions
 
 

From 84f66484c524c41d172abe341b206d31598ef40b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 09:31:38 -0700
Subject: [PATCH 0636/1701] Make it optional to paste long pasted content to an
 attachment

---
 js/main.js            | 2 +-
 modules/shared.py     | 1 +
 modules/ui.py         | 6 ++++++
 modules/ui_session.py | 1 +
 4 files changed, 9 insertions(+), 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index 9a620fa9f7..7e2a457c16 100644
--- a/js/main.js
+++ b/js/main.js
@@ -884,7 +884,7 @@ function setupPasteHandler() {
   textbox.addEventListener("paste", async (event) => {
     const text = event.clipboardData?.getData("text");
 
-    if (text && text.length > MAX_PLAIN_TEXT_LENGTH) {
+    if (text && text.length > MAX_PLAIN_TEXT_LENGTH && document.querySelector("#paste_to_attachment input[data-testid=\"checkbox\"]")?.checked) {
       event.preventDefault();
 
       const file = new File([text], "pasted_text.txt", {
diff --git a/modules/shared.py b/modules/shared.py
index 3794206b8d..59c7dbcdb9 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -223,6 +223,7 @@
     'custom_token_bans': '',
     'negative_prompt': '',
     'dark_theme': True,
+    'paste_to_attachment': False,
     'default_extensions': [],
 
     # Character settings
diff --git a/modules/ui.py b/modules/ui.py
index 8ec4b165e7..46403dd9e8 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -268,6 +268,11 @@ def list_interface_input_elements():
     # Model elements
     elements += list_model_elements()
 
+    # Other elements
+    elements += [
+        'paste_to_attachment'
+    ]
+
     return elements
 
 
@@ -473,6 +478,7 @@ def setup_auto_save():
         # Session tab (ui_session.py)
         'show_controls',
         'theme_state',
+        'paste_to_attachment'
     ]
 
     for element_name in change_elements:
diff --git a/modules/ui_session.py b/modules/ui_session.py
index 086a06bfae..2ece22519f 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -12,6 +12,7 @@ def create_ui():
                 gr.Markdown("## Settings")
                 shared.gradio['save_settings'] = gr.Button('Save settings to user_data/settings.yaml', elem_classes='refresh-button', interactive=not mu)
                 shared.gradio['toggle_dark_mode'] = gr.Button('Toggle light/dark theme 💡', elem_classes='refresh-button')
+                shared.gradio['paste_to_attachment'] = gr.Checkbox(label='Turn long pasted text into attachments in the Chat tab', value=shared.settings['paste_to_attachment'], elem_id='paste_to_attachment')
 
             with gr.Column():
                 gr.Markdown("## Extensions & flags")

From 1f1435997a65b92c498f6d5512683a6ee2f21b26 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 09:37:54 -0700
Subject: [PATCH 0637/1701] Don't show the new 'Restore character' button in
 the Chat tab

---
 js/main.js         | 2 ++
 modules/chat.py    | 2 +-
 modules/ui_chat.py | 6 +++---
 3 files changed, 6 insertions(+), 4 deletions(-)

diff --git a/js/main.js b/js/main.js
index 7e2a457c16..70afabe3b3 100644
--- a/js/main.js
+++ b/js/main.js
@@ -555,6 +555,7 @@ function moveToChatTab() {
 
   newParent.insertBefore(grandParent, newParent.children[newPosition]);
   document.getElementById("save-character").style.display = "none";
+  document.getElementById("restore-character").style.display = "none";
 }
 
 function restoreOriginalPosition() {
@@ -566,6 +567,7 @@ function restoreOriginalPosition() {
     }
 
     document.getElementById("save-character").style.display = "";
+    document.getElementById("restore-character").style.display = "";
     movedElement.style.display = "";
     movedElement.children[0].style.minWidth = "";
   }
diff --git a/modules/chat.py b/modules/chat.py
index f740db5528..49511af198 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1220,7 +1220,7 @@ def load_character(character, name1, name2):
     return name1, name2, picture, greeting, context
 
 
-def reset_character_for_ui(state):
+def restore_character_for_ui(state):
     """Reset character fields to the currently loaded character's saved values"""
     if state['character_menu'] and state['character_menu'] != 'None':
         try:
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index ef9330e467..d7a5ec69ce 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -122,7 +122,7 @@ def create_chat_settings_ui():
                         ui.create_refresh_button(shared.gradio['character_menu'], lambda: None, lambda: {'choices': utils.get_available_characters()}, 'refresh-button', interactive=not mu)
                         shared.gradio['save_character'] = gr.Button('💾', elem_classes='refresh-button', elem_id="save-character", interactive=not mu)
                         shared.gradio['delete_character'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu)
-                        shared.gradio['reset_character'] = gr.Button('Restore character', elem_classes='refresh-button', interactive=True)
+                        shared.gradio['restore_character'] = gr.Button('Restore character', elem_classes='refresh-button', interactive=True, elem_id='restore-character')
 
                     shared.gradio['name2'] = gr.Textbox(value=shared.settings['name2'], lines=1, label='Character\'s name')
                     shared.gradio['context'] = gr.Textbox(value=shared.settings['context'], lines=10, label='Context', elem_classes=['add_scrollbar'])
@@ -320,9 +320,9 @@ def create_event_handlers():
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_save_template_click, gradio('instruction_template_str'), gradio('save_filename', 'save_root', 'save_contents', 'file_saver'), show_progress=False)
 
-    shared.gradio['reset_character'].click(
+    shared.gradio['restore_character'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        chat.reset_character_for_ui, gradio('interface_state'), gradio('interface_state', 'name2', 'context', 'greeting', 'character_picture'), show_progress=False)
+        chat.restore_character_for_ui, gradio('interface_state'), gradio('interface_state', 'name2', 'context', 'greeting', 'character_picture'), show_progress=False)
 
     shared.gradio['delete_template'].click(chat.handle_delete_template_click, gradio('instruction_template'), gradio('delete_filename', 'delete_root', 'file_deleter'), show_progress=False)
     shared.gradio['save_chat_history'].click(

From 78899244d5e1e6504a5be88252b17e44a48b6ca7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 09:40:09 -0700
Subject: [PATCH 0638/1701] Remove settings-template.yaml

---
 user_data/settings-template.yaml | 76 --------------------------------
 1 file changed, 76 deletions(-)
 delete mode 100644 user_data/settings-template.yaml

diff --git a/user_data/settings-template.yaml b/user_data/settings-template.yaml
deleted file mode 100644
index db481e8499..0000000000
--- a/user_data/settings-template.yaml
+++ /dev/null
@@ -1,76 +0,0 @@
-show_controls: true
-start_with: ''
-mode: instruct
-chat_style: cai-chat
-chat-instruct_command: |-
-  Continue the chat dialogue below. Write a single reply for the character "<|character|>".
-
-  <|prompt|>
-prompt-default: QA
-prompt-notebook: QA
-character: Assistant
-name1: You
-user_bio: ''
-custom_system_message: ''
-preset: min_p
-max_new_tokens: 512
-max_new_tokens_min: 1
-max_new_tokens_max: 4096
-prompt_lookup_num_tokens: 0
-max_tokens_second: 0
-auto_max_new_tokens: true
-ban_eos_token: false
-add_bos_token: true
-enable_thinking: true
-skip_special_tokens: true
-stream: true
-static_cache: false
-truncation_length: 8192
-seed: -1
-custom_stopping_strings: ''
-custom_token_bans: ''
-negative_prompt: ''
-dark_theme: true
-default_extensions: []
-instruction_template_str: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + 'Below is an instruction that describes a task. Write a response that appropriately completes the request.' + '\n\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '\n\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'### Instruction:\n' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'### Response:\n' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'### Response:\n'-}}
-  {%- endif -%}
-chat_template_str: |-
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {%- if message['content'] -%}
-              {{- message['content'] + '\n\n' -}}
-          {%- endif -%}
-          {%- if user_bio -%}
-              {{- user_bio + '\n\n' -}}
-          {%- endif -%}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{- name1 + ': ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{- name2 + ': ' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-

From eb0ab9db1d285fb3be949c4180e63d7df30e61a0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 15:04:05 -0700
Subject: [PATCH 0639/1701] Fix light/dark theme persistence across page
 reloads

---
 modules/ui_session.py |  2 +-
 server.py             | 25 ++++++++++++++++++++-----
 2 files changed, 21 insertions(+), 6 deletions(-)

diff --git a/modules/ui_session.py b/modules/ui_session.py
index 2ece22519f..33d7dcb733 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -31,7 +31,7 @@ def create_ui():
 
         shared.gradio['toggle_dark_mode'].click(
             lambda x: 'dark' if x == 'light' else 'light', gradio('theme_state'), gradio('theme_state')).then(
-            None, None, None, js=f'() => {{{ui.dark_theme_js}; toggleDarkMode()}}')
+            None, None, None, js=f'() => {{{ui.dark_theme_js}; toggleDarkMode(); localStorage.setItem("theme", document.body.classList.contains("dark") ? "dark" : "light")}}')
 
         # Reset interface event
         shared.gradio['reset_interface'].click(
diff --git a/server.py b/server.py
index 3dd0a3f310..bd440fe0b9 100644
--- a/server.py
+++ b/server.py
@@ -163,11 +163,26 @@ def create_interface():
             gradio('show_controls'),
             None,
             js=f"""(x) => {{
-                if ({str(shared.settings['dark_theme']).lower()}) {{
-                    document.getElementsByTagName('body')[0].classList.add('dark');
-                }}
-                else {{
-                    document.getElementsByTagName('body')[0].classList.remove('dark');
+                // Check if this is first visit or if localStorage is out of sync
+                const savedTheme = localStorage.getItem('theme');
+                const serverTheme = {str(shared.settings['dark_theme']).lower()} ? 'dark' : 'light';
+
+                // If no saved theme or mismatch with server on first load, use server setting
+                if (!savedTheme || !sessionStorage.getItem('theme_synced')) {{
+                    localStorage.setItem('theme', serverTheme);
+                    sessionStorage.setItem('theme_synced', 'true');
+                    if (serverTheme === 'dark') {{
+                        document.getElementsByTagName('body')[0].classList.add('dark');
+                    }} else {{
+                        document.getElementsByTagName('body')[0].classList.remove('dark');
+                    }}
+                }} else {{
+                    // Use localStorage for subsequent reloads
+                    if (savedTheme === 'dark') {{
+                        document.getElementsByTagName('body')[0].classList.add('dark');
+                    }} else {{
+                        document.getElementsByTagName('body')[0].classList.remove('dark');
+                    }}
                 }}
                 {js}
                 {ui.show_controls_js}

From f81b1540caf324923eeb1bec75c9ce8774865e38 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 15:19:25 -0700
Subject: [PATCH 0640/1701] Small style improvements

---
 css/main.css  | 2 +-
 modules/ui.py | 4 +++-
 2 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/css/main.css b/css/main.css
index 7d7b0cbd81..e2ce48017f 100644
--- a/css/main.css
+++ b/css/main.css
@@ -269,7 +269,7 @@ button {
 
 .dark .pretty_scrollbar::-webkit-scrollbar-thumb,
 .dark .pretty_scrollbar::-webkit-scrollbar-thumb:hover {
-    background: rgb(255 255 255 / 10%);
+    background: rgb(255 255 255 / 6.25%);
     border-radius: 10px;
 }
 
diff --git a/modules/ui.py b/modules/ui.py
index 46403dd9e8..458b18edd3 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -74,8 +74,10 @@
         body_background_fill="white",
         block_background_fill="transparent",
         body_text_color='rgb(64, 64, 64)',
-        button_secondary_background_fill="#f4f4f4",
+        button_secondary_background_fill="white",
         button_secondary_border_color="var(--border-color-primary)",
+        input_shadow="none",
+        button_shadow_hover="none",
 
         # Dark Mode Colors
         input_background_fill_dark='var(--darker-gray)',

From ff01bcb870e6a5bca6b11f32d522911b5f2ccf40 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 15:33:05 -0700
Subject: [PATCH 0641/1701] Use user_data/cache/gradio for Gradio temp files

---
 server.py | 19 ++++++++++++++++---
 1 file changed, 16 insertions(+), 3 deletions(-)

diff --git a/server.py b/server.py
index bd440fe0b9..e178e8fedb 100644
--- a/server.py
+++ b/server.py
@@ -1,12 +1,26 @@
 import os
+import shutil
 import warnings
+from pathlib import Path
 
 from modules import shared
 from modules.block_requests import OpenMonkeyPatch, RequestBlocker
 from modules.logging_colors import logger
 
-os.environ['GRADIO_ANALYTICS_ENABLED'] = 'False'
-os.environ['BITSANDBYTES_NOWELCOME'] = '1'
+# Set up Gradio temp directory path
+gradio_temp_path = Path('user_data') / 'cache' / 'gradio'
+
+# Set environment variables
+os.environ.update({
+    'GRADIO_ANALYTICS_ENABLED': 'False',
+    'BITSANDBYTES_NOWELCOME': '1',
+    'GRADIO_TEMP_DIR': str(gradio_temp_path)
+})
+
+# Clear and recreate gradio temp directory
+shutil.rmtree(gradio_temp_path, ignore_errors=True)
+gradio_temp_path.mkdir(parents=True, exist_ok=True)
+
 warnings.filterwarnings('ignore', category=UserWarning, message='TypedStorage is deprecated')
 warnings.filterwarnings('ignore', category=UserWarning, message='Using the update method is deprecated')
 warnings.filterwarnings('ignore', category=UserWarning, message='Field "model_name" has conflict')
@@ -27,7 +41,6 @@
 import sys
 import time
 from functools import partial
-from pathlib import Path
 from threading import Lock, Thread
 
 import yaml

From 7ed1926ce7a37d1ccd6b6c8be399581a258759a5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 15:38:40 -0700
Subject: [PATCH 0642/1701] Small change after previous commit

---
 server.py | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/server.py b/server.py
index e178e8fedb..5cad1d8a82 100644
--- a/server.py
+++ b/server.py
@@ -9,6 +9,8 @@
 
 # Set up Gradio temp directory path
 gradio_temp_path = Path('user_data') / 'cache' / 'gradio'
+shutil.rmtree(gradio_temp_path, ignore_errors=True)
+gradio_temp_path.mkdir(parents=True, exist_ok=True)
 
 # Set environment variables
 os.environ.update({
@@ -17,10 +19,6 @@
     'GRADIO_TEMP_DIR': str(gradio_temp_path)
 })
 
-# Clear and recreate gradio temp directory
-shutil.rmtree(gradio_temp_path, ignore_errors=True)
-gradio_temp_path.mkdir(parents=True, exist_ok=True)
-
 warnings.filterwarnings('ignore', category=UserWarning, message='TypedStorage is deprecated')
 warnings.filterwarnings('ignore', category=UserWarning, message='Using the update method is deprecated')
 warnings.filterwarnings('ignore', category=UserWarning, message='Field "model_name" has conflict')

From e976a5ddc7598de49227f600515a7eee72fc1af6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 17:34:56 -0700
Subject: [PATCH 0643/1701] Re-highlight code blocks when switching light/dark
 themes

---
 js/dark_theme.js | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/js/dark_theme.js b/js/dark_theme.js
index b540fb113a..f61060cdf9 100644
--- a/js/dark_theme.js
+++ b/js/dark_theme.js
@@ -6,4 +6,11 @@ function toggleDarkMode() {
   } else {
     currentCSS.setAttribute("href", "file/css/highlightjs/github-dark.min.css");
   }
+
+  // Re-highlight all code blocks once stylesheet loads
+  currentCSS.onload = function() {
+    document.querySelectorAll("pre code").forEach(block => {
+      hljs.highlightElement(block);
+    });
+  };
 }

From b5e021fc49c840d1ac934f647b64317918ee1208 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 17:44:04 -0700
Subject: [PATCH 0644/1701] Make the dark theme darker

---
 css/main.css | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/css/main.css b/css/main.css
index e2ce48017f..d04ff62478 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1,11 +1,11 @@
 :root {
-    --darker-gray: #202123;
-    --dark-gray: #2A2B32;
-    --light-gray: #373943;
+    --darker-gray: #1C1C1D;
+    --dark-gray: #212125;
+    --light-gray: #2C2E34;
     --light-theme-gray: #f9fbff;
     --border-color-dark: #525252;
     --header-width: 112px;
-    --selected-item-color-dark: #2E2F38;
+    --selected-item-color-dark: #282930;
 }
 
 @font-face {

From 06dfb7e772a943c1bfdeff6d125d8cf056dc3b73 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 18:03:07 -0700
Subject: [PATCH 0645/1701] Improve the style of the hover menu

---
 css/main.css | 19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)

diff --git a/css/main.css b/css/main.css
index d04ff62478..8e57db89dd 100644
--- a/css/main.css
+++ b/css/main.css
@@ -745,16 +745,13 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     justify-content: space-between;
     margin: 0 !important;
     height: 36px;
+    border-color: transparent !important;
 }
 
 .hover-menu button:not(#clear-history-confirm) {
     border-bottom: 0 !important;
 }
 
-.hover-menu button:not(#clear-history-confirm):last-child {
-    border-bottom: var(--button-border-width) solid var(--border-color-primary) !important;
-}
-
 .hover-menu button:hover {
     background: #dbeafe !important;
 }
@@ -764,26 +761,30 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 #show-controls {
+    background-color: white;
+    border-color: transparent !important;
     height: 36px;
-    border-top: 1px solid var(--border-color-dark) !important;
-    border-left: 1px solid var(--border-color-dark) !important;
-    border-right: 1px solid var(--border-color-dark) !important;
     border-radius: 0;
     border-bottom: 0 !important;
-    background-color: var(--darker-gray);
     padding-top: 3px;
     padding-left: 4px;
     display: flex;
+    font-weight: normal;
+}
+
+.dark #show-controls {
+    background-color: var(--darker-gray);
 }
 
 #show-controls label {
     display: flex;
     flex-direction: row-reverse;
-    font-weight: bold;
     justify-content: start;
     width: 100%;
     padding-right: 12px;
     gap: 10px;
+    font-weight: 600;
+    color: var(--button-secondary-text-color);
 }
 
 #show-controls label input {

From 0b8d2d65a262271e74697410c6aff577cad2594a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 18:11:27 -0700
Subject: [PATCH 0646/1701] Minor style improvement

---
 modules/ui.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/ui.py b/modules/ui.py
index 458b18edd3..ea36c639b9 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -109,6 +109,7 @@
         button_large_radius='0.375rem',
         button_large_padding='6px 12px',
         input_radius='0.375rem',
+        block_radius='0',
     )
 
 if Path("user_data/notification.mp3").exists():

From 4a369e070aab9e540187bf456a83df5478565a0e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 18:47:48 -0700
Subject: [PATCH 0647/1701] Add buttons for easily deleting past chats

---
 css/main.css       | 53 +++++++++++++++++++++++++++++++++++
 js/main.js         | 70 ++++++++++++++++++++++++++++++++++++++++++++++
 modules/chat.py    |  1 -
 modules/ui_chat.py | 10 +++----
 4 files changed, 127 insertions(+), 7 deletions(-)

diff --git a/css/main.css b/css/main.css
index 8e57db89dd..c5460eb7fc 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1570,3 +1570,56 @@ button:focus {
 .svelte-sa48pu.stretch:has(> .hidden:only-child) {
     display: none;
 }
+
+.delete-container {
+    position: absolute;
+    right: 8px;
+    display: flex;
+    gap: 6px;
+    opacity: 0;
+    transition: opacity 0.2s;
+    margin-left: 0;
+}
+
+.chat-label-with-delete {
+    position: relative;
+    padding-right: 60px;
+}
+
+.trash-btn {
+    border: none;
+    background: none;
+    cursor: pointer;
+    padding: 2px;
+    opacity: 0.7;
+}
+
+.cancel-btn {
+    border: none;
+    background: #ef4444;
+    color: white;
+    cursor: pointer;
+    width: 20px;
+    height: 20px;
+    border-radius: 2px;
+    font-family: monospace;
+    font-size: 12px;
+    align-items: center;
+    justify-content: center;
+    display: none;
+}
+
+.confirm-btn {
+    border: none;
+    background: #22c55e;
+    color: white;
+    cursor: pointer;
+    width: 20px;
+    height: 20px;
+    border-radius: 2px;
+    font-family: monospace;
+    font-size: 12px;
+    align-items: center;
+    justify-content: center;
+    display: none;
+}
diff --git a/js/main.js b/js/main.js
index 70afabe3b3..e9ca5a0b78 100644
--- a/js/main.js
+++ b/js/main.js
@@ -917,3 +917,73 @@ document.querySelector("#chat-input .upload-button").title = "Upload text files,
 
 // Activate web search
 document.getElementById("web-search").title = "Search the internet with DuckDuckGo";
+
+//------------------------------------------------
+// Inline icons for deleting past chats
+//------------------------------------------------
+
+function addMiniDeletes() {
+  document.querySelectorAll("#past-chats label:not(.has-delete)").forEach(label => {
+    const container = document.createElement("span");
+    container.className = "delete-container";
+
+    label.classList.add("chat-label-with-delete");
+
+    const trashBtn = document.createElement("button");
+    trashBtn.innerHTML = "🗑️";
+    trashBtn.className = "trash-btn";
+
+    const cancelBtn = document.createElement("button");
+    cancelBtn.innerHTML = "✕";
+    cancelBtn.className = "cancel-btn";
+
+    const confirmBtn = document.createElement("button");
+    confirmBtn.innerHTML = "✓";
+    confirmBtn.className = "confirm-btn";
+
+    label.addEventListener("mouseenter", () => {
+      container.style.opacity = "1";
+    });
+
+    label.addEventListener("mouseleave", () => {
+      container.style.opacity = "0";
+    });
+
+    trashBtn.onclick = (e) => {
+      e.stopPropagation();
+      label.querySelector("input").click();
+      document.querySelector("#delete_chat").click();
+      trashBtn.style.display = "none";
+      cancelBtn.style.display = "flex";
+      confirmBtn.style.display = "flex";
+    };
+
+    cancelBtn.onclick = (e) => {
+      e.stopPropagation();
+      document.querySelector("#delete_chat-cancel").click();
+      resetButtons();
+    };
+
+    confirmBtn.onclick = (e) => {
+      e.stopPropagation();
+      document.querySelector("#delete_chat-confirm").click();
+      resetButtons();
+    };
+
+    function resetButtons() {
+      trashBtn.style.display = "inline";
+      cancelBtn.style.display = "none";
+      confirmBtn.style.display = "none";
+    }
+
+    container.append(trashBtn, cancelBtn, confirmBtn);
+    label.appendChild(container);
+    label.classList.add("has-delete");
+  });
+}
+
+new MutationObserver(() => addMiniDeletes()).observe(
+  document.querySelector("#past-chats"),
+  {childList: true, subtree: true}
+);
+addMiniDeletes();
diff --git a/modules/chat.py b/modules/chat.py
index 49511af198..ad5045e0b6 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1562,7 +1562,6 @@ def handle_delete_chat_confirm_click(state):
         unique_id,
         gr.update(visible=False),
         gr.update(visible=True),
-        gr.update(visible=False)
     ]
 
 
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index d7a5ec69ce..428b64c98c 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -26,15 +26,15 @@ def create_ui():
                 with gr.Row(elem_id='past-chats-buttons'):
                     shared.gradio['branch_chat'] = gr.Button('Branch', elem_classes=['refresh-button', 'refresh-button-medium'], elem_id='Branch', interactive=not mu)
                     shared.gradio['rename_chat'] = gr.Button('Rename', elem_classes=['refresh-button', 'refresh-button-medium'], interactive=not mu)
-                    shared.gradio['delete_chat'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu)
+                    shared.gradio['delete_chat'] = gr.Button('🗑️', visible=False, elem_classes='refresh-button', interactive=not mu, elem_id='delete_chat')
                     shared.gradio['Start new chat'] = gr.Button('New chat', elem_classes=['refresh-button', 'refresh-button-medium', 'focus-on-chat-input'])
                     shared.gradio['branch_index'] = gr.Number(value=-1, precision=0, visible=False, elem_id="Branch-index", interactive=True)
 
                 shared.gradio['search_chat'] = gr.Textbox(placeholder='Search chats...', max_lines=1, elem_id='search_chat')
 
                 with gr.Row(elem_id='delete-chat-row', visible=False) as shared.gradio['delete-chat-row']:
-                    shared.gradio['delete_chat-cancel'] = gr.Button('Cancel', elem_classes=['refresh-button', 'focus-on-chat-input'])
-                    shared.gradio['delete_chat-confirm'] = gr.Button('Confirm', variant='stop', elem_classes=['refresh-button', 'focus-on-chat-input'])
+                    shared.gradio['delete_chat-cancel'] = gr.Button('Cancel', elem_classes=['refresh-button', 'focus-on-chat-input'], elem_id='delete_chat-cancel')
+                    shared.gradio['delete_chat-confirm'] = gr.Button('Confirm', variant='stop', elem_classes=['refresh-button', 'focus-on-chat-input'], elem_id='delete_chat-confirm')
 
                 with gr.Row(elem_id='rename-row', visible=False) as shared.gradio['rename-row']:
                     shared.gradio['rename_to'] = gr.Textbox(label='Rename to:', placeholder='New name', elem_classes=['no-background'])
@@ -261,11 +261,9 @@ def create_event_handlers():
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_start_new_chat_click, gradio('interface_state'), gradio('history', 'display', 'unique_id'), show_progress=False)
 
-    shared.gradio['delete_chat'].click(lambda: gr.update(visible=True), None, gradio('delete-chat-row'))
-    shared.gradio['delete_chat-cancel'].click(lambda: gr.update(visible=False), None, gradio('delete-chat-row'))
     shared.gradio['delete_chat-confirm'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        chat.handle_delete_chat_confirm_click, gradio('interface_state'), gradio('history', 'display', 'unique_id', 'delete-chat-row'), show_progress=False)
+        chat.handle_delete_chat_confirm_click, gradio('interface_state'), gradio('history', 'display', 'unique_id'), show_progress=False)
 
     shared.gradio['branch_chat'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(

From f3388c2ab4cdf60cbd035d9a021b8e7de1e4665d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 18:53:04 -0700
Subject: [PATCH 0648/1701] Fix selecting next chat when deleting with active
 search

---
 modules/chat.py | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index ad5045e0b6..d62bbc4205 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1549,7 +1549,10 @@ def handle_start_new_chat_click(state):
 
 
 def handle_delete_chat_confirm_click(state):
-    index = str(find_all_histories(state).index(state['unique_id']))
+    filtered_histories = find_all_histories_with_first_prompts(state)
+    filtered_ids = [h[1] for h in filtered_histories]
+    index = str(filtered_ids.index(state['unique_id']))
+
     delete_history(state['unique_id'], state['character_menu'], state['mode'])
     history, unique_id = load_history_after_deletion(state, index)
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])

From f9a007c6a85899d31a9b38baf40fcad5daa0673d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 19:25:23 -0700
Subject: [PATCH 0649/1701] Properly filter out failed web search downloads
 from attachments

---
 modules/web_search.py | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/modules/web_search.py b/modules/web_search.py
index a1e47253b7..2b6c6c4016 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -38,7 +38,7 @@ def download_web_page(url, timeout=5):
         return text
     except Exception as e:
         logger.error(f"Error downloading {url}: {e}")
-        return f"[Error downloading content from {url}: {str(e)}]"
+        return ""
 
 
 def perform_web_search(query, num_pages=3, max_workers=5):
@@ -74,9 +74,7 @@ def perform_web_search(query, num_pages=3, max_workers=5):
                         'url': url,
                         'content': content
                     }
-                except Exception as e:
-                    logger.error(f"Error downloading {url}: {e}")
-                    # Include failed downloads with empty content
+                except Exception:
                     search_results[index] = {
                         'title': title,
                         'url': url,
@@ -108,7 +106,7 @@ def add_web_search_attachments(history, row_idx, user_message, search_query, sta
             return
 
         # Filter out failed downloads before adding attachments
-        successful_results = [result for result in search_results if result['content'] and result['content'].strip()]
+        successful_results = [result for result in search_results if result['content'].strip()]
 
         if not successful_results:
             logger.warning("No successful downloads to add as attachments")
@@ -130,7 +128,7 @@ def add_web_search_attachments(history, row_idx, user_message, search_query, sta
             }
             history['metadata'][key]["attachments"].append(attachment)
 
-        logger.info(f"Added {len(successful_results)} successful web search results as attachments")
+        logger.info(f"Added {len(successful_results)} successful web search results as attachments.")
 
     except Exception as e:
         logger.error(f"Error in web search: {e}")

From 80637cae288dfd218b1bf664e8caedbf473ef298 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 21:55:49 -0700
Subject: [PATCH 0650/1701] Add version to portable build folder names

---
 .github/workflows/build-portable-release-cuda.yml   | 9 ++++++---
 .github/workflows/build-portable-release-vulkan.yml | 9 ++++++---
 .github/workflows/build-portable-release.yml        | 9 ++++++---
 3 files changed, 18 insertions(+), 9 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index 571cbac0c5..283fdd72cb 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -160,16 +160,19 @@ jobs:
                 rm requirements_cuda_temp.txt
             fi
 
-            # 6. Create ZIP file
+            # 6. Move up and rename folder to include version
             cd ..
             VERSION_CLEAN="${VERSION#v}"
+            mv text-generation-webui text-generation-webui-${VERSION_CLEAN}
+
+            # 7. Create ZIP file
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui -DestinationPath $ZIP_NAME"
+                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
             else
-                zip -r "$ZIP_NAME" text-generation-webui
+                zip -r "$ZIP_NAME" text-generation-webui-${VERSION_CLEAN}
             fi
 
       - name: Upload files to a GitHub release
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 4e88d4d9ab..c6ab8fa75a 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -146,16 +146,19 @@ jobs:
             echo "Installing Python packages from $REQ_FILE..."
             $PIP_PATH install --target="./$PACKAGES_PATH" -r "$REQ_FILE"
 
-            # 6. Create ZIP file
+            # 5. Move up and rename folder to include version
             cd ..
             VERSION_CLEAN="${VERSION#v}"
+            mv text-generation-webui text-generation-webui-${VERSION_CLEAN}
+
+            # 6. Create ZIP file
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui -DestinationPath $ZIP_NAME"
+                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
             else
-                zip -r "$ZIP_NAME" text-generation-webui
+                zip -r "$ZIP_NAME" text-generation-webui-${VERSION_CLEAN}
             fi
 
       - name: Upload files to a GitHub release
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 6910ce2c3b..58bfdb256c 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -170,16 +170,19 @@ jobs:
             echo "Installing Python packages from $REQ_FILE..."
             $PIP_PATH install --target="./$PACKAGES_PATH" -r "$REQ_FILE"
 
-            # 5. Create ZIP file
+            # 5. Move up and rename folder to include version
             cd ..
             VERSION_CLEAN="${VERSION#v}"
+            mv text-generation-webui text-generation-webui-${VERSION_CLEAN}
+
+            # 6. Create ZIP file
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui -DestinationPath $ZIP_NAME"
+                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
             else
-                zip -r "$ZIP_NAME" text-generation-webui
+                zip -r "$ZIP_NAME" text-generation-webui-${VERSION_CLEAN}
             fi
 
       - name: Upload files to a GitHub release

From eefbf96f6a30ad63da6a494b78fbb3e496462d38 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Jun 2025 22:14:56 -0700
Subject: [PATCH 0651/1701] Don't save truncation_length to
 user_data/settings.yaml

---
 modules/ui.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/ui.py b/modules/ui.py
index ea36c639b9..38693da8e4 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -332,6 +332,7 @@ def save_settings(state, preset, extensions_list, show_controls, theme_state):
     output['show_controls'] = show_controls
     output['dark_theme'] = True if theme_state == 'dark' else False
     output.pop('instruction_template_str')
+    output.pop('truncation_length')
 
     # Save extension values in the UI
     for extension_name in extensions_list:

From 1602ac1c8ff640399a8a1eadd0305958a89dfccf Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Jun 2025 09:03:39 -0700
Subject: [PATCH 0652/1701] Improve the style of thinking blocks in dark mode

---
 css/main.css | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index c5460eb7fc..307d0437fb 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1337,7 +1337,8 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 }
 
 .dark .thinking-block {
-    background-color: var(--darker-gray);
+    background-color: transparent;
+    border: 1px solid var(--input-border-color);
 }
 
 .thinking-header {

From 14efd420845226706aefb92c74fae33d0a49816c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Jun 2025 11:25:36 -0700
Subject: [PATCH 0653/1701] Improve scroll performance by disabling hover
 effects during scroll

---
 css/main.css |  5 +++++
 js/main.js   | 11 ++++++++++-
 2 files changed, 15 insertions(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 307d0437fb..62dfc05fb4 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1624,3 +1624,8 @@ button:focus {
     justify-content: center;
     display: none;
 }
+
+/* Disable hover effects while scrolling */
+.chat-parent.scrolling * {
+    pointer-events: none !important;
+}
diff --git a/js/main.js b/js/main.js
index e9ca5a0b78..1953a6be74 100644
--- a/js/main.js
+++ b/js/main.js
@@ -146,8 +146,12 @@ const targetElement = document.getElementById("chat").parentNode.parentNode.pare
 targetElement.classList.add("pretty_scrollbar");
 targetElement.classList.add("chat-parent");
 let isScrolled = false;
+let scrollTimeout;
 
 targetElement.addEventListener("scroll", function() {
+  // Add scrolling class to disable hover effects
+  targetElement.classList.add("scrolling");
+
   let diff = targetElement.scrollHeight - targetElement.clientHeight;
   if(Math.abs(targetElement.scrollTop - diff) <= 10 || diff == 0) {
     isScrolled = false;
@@ -155,7 +159,12 @@ targetElement.addEventListener("scroll", function() {
     isScrolled = true;
   }
 
-  doSyntaxHighlighting();
+  // Clear previous timeout and set new one
+  clearTimeout(scrollTimeout);
+  scrollTimeout = setTimeout(() => {
+    targetElement.classList.remove("scrolling");
+    doSyntaxHighlighting(); // Only run after scrolling stops
+  }, 150);
 
 });
 

From 747a4a0e5647aac2aea40bdb1e8ae0d2705bc027 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Jun 2025 12:32:10 -0700
Subject: [PATCH 0654/1701] Reposition the ... typing dots

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 62dfc05fb4..a22fdd950f 100644
--- a/css/main.css
+++ b/css/main.css
@@ -668,7 +668,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     display: none;
     position: absolute;
     background-color: transparent;
-    left: -2px;
+    left: 23px;
     top: -5px;
     padding: var(--block-padding);
 }

From f5a5d0c0cbcca8e18a3a30f678d5d6ae2396c0d9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Jun 2025 17:32:25 -0700
Subject: [PATCH 0655/1701] Add the URL of web attachments to the prompt

---
 modules/chat.py | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index d62bbc4205..25a0607b44 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -223,7 +223,10 @@ def generate_chat_prompt(user_input, state, **kwargs):
                 for attachment in metadata[user_key]["attachments"]:
                     filename = attachment.get("name", "file")
                     content = attachment.get("content", "")
-                    attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
+                    if attachment.get("type") == "text/html" and attachment.get("url"):
+                        attachments_text += f"\nName: {filename}\nURL: {attachment['url']}\nContents:\n\n=====\n{content}\n=====\n\n"
+                    else:
+                        attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
 
                 if attachments_text:
                     enhanced_user_msg = f"{user_msg}\n\nATTACHMENTS:\n{attachments_text}"
@@ -250,7 +253,10 @@ def generate_chat_prompt(user_input, state, **kwargs):
                 for attachment in metadata[user_key]["attachments"]:
                     filename = attachment.get("name", "file")
                     content = attachment.get("content", "")
-                    attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
+                    if attachment.get("type") == "text/html" and attachment.get("url"):
+                        attachments_text += f"\nName: {filename}\nURL: {attachment['url']}\nContents:\n\n=====\n{content}\n=====\n\n"
+                    else:
+                        attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
 
                 if attachments_text:
                     user_input = f"{user_input}\n\nATTACHMENTS:\n{attachments_text}"

From 263b5d5557efd7632f0d02adb6f7f44020f19b41 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Jun 2025 17:55:26 -0700
Subject: [PATCH 0656/1701] Use html2text to extract the text of web searches
 without losing formatting

---
 modules/web_search.py                         | 36 ++++++++++---------
 requirements/full/requirements.txt            |  2 +-
 requirements/full/requirements_amd.txt        |  2 +-
 requirements/full/requirements_amd_noavx2.txt |  2 +-
 .../full/requirements_apple_intel.txt         |  2 +-
 .../full/requirements_apple_silicon.txt       |  2 +-
 requirements/full/requirements_cpu_only.txt   |  2 +-
 .../full/requirements_cpu_only_noavx2.txt     |  2 +-
 requirements/full/requirements_cuda128.txt    |  2 +-
 .../full/requirements_cuda128_noavx2.txt      |  2 +-
 requirements/full/requirements_noavx2.txt     |  2 +-
 requirements/full/requirements_nowheels.txt   |  2 +-
 requirements/portable/requirements.txt        |  2 +-
 .../portable/requirements_apple_intel.txt     |  2 +-
 .../portable/requirements_apple_silicon.txt   |  2 +-
 .../portable/requirements_cpu_only.txt        |  2 +-
 .../portable/requirements_cpu_only_noavx2.txt |  2 +-
 requirements/portable/requirements_noavx2.txt |  2 +-
 .../portable/requirements_nowheels.txt        |  2 +-
 requirements/portable/requirements_vulkan.txt |  2 +-
 .../portable/requirements_vulkan_noavx2.txt   |  2 +-
 21 files changed, 39 insertions(+), 37 deletions(-)

diff --git a/modules/web_search.py b/modules/web_search.py
index 2b6c6c4016..ffd7e4833a 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -3,8 +3,6 @@
 from datetime import datetime
 
 import requests
-from bs4 import BeautifulSoup
-from duckduckgo_search import DDGS
 
 from modules.logging_colors import logger
 
@@ -14,35 +12,39 @@ def get_current_timestamp():
     return datetime.now().strftime('%b %d, %Y %H:%M')
 
 
-def download_web_page(url, timeout=5):
-    """Download and extract text from a web page"""
+def download_web_page(url, timeout=10):
+    """
+    Download a web page and convert its HTML content to structured Markdown text.
+    """
+    import html2text
+
     try:
         headers = {
             'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'
         }
         response = requests.get(url, headers=headers, timeout=timeout)
-        response.raise_for_status()
-
-        soup = BeautifulSoup(response.content, 'html.parser')
+        response.raise_for_status()  # Raise an exception for bad status codes
 
-        # Remove script and style elements
-        for script in soup(["script", "style"]):
-            script.decompose()
+        # Initialize the HTML to Markdown converter
+        h = html2text.HTML2Text()
+        h.body_width = 0
 
-        # Get text and clean it up
-        text = soup.get_text()
-        lines = (line.strip() for line in text.splitlines())
-        chunks = (phrase.strip() for line in lines for phrase in line.split("  "))
-        text = ' '.join(chunk for chunk in chunks if chunk)
+        # Convert the HTML to Markdown
+        markdown_text = h.handle(response.text)
 
-        return text
-    except Exception as e:
+        return markdown_text
+    except requests.exceptions.RequestException as e:
         logger.error(f"Error downloading {url}: {e}")
         return ""
+    except Exception as e:
+        logger.error(f"An unexpected error occurred: {e}")
+        return ""
 
 
 def perform_web_search(query, num_pages=3, max_workers=5):
     """Perform web search and return results with content"""
+    from duckduckgo_search import DDGS
+
     try:
         with DDGS() as ddgs:
             results = list(ddgs.text(query, max_results=num_pages))
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 277f8249bd..b751482a2e 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -1,5 +1,4 @@
 accelerate==1.5.*
-beautifulsoup4==4.13.4
 bitsandbytes==0.45.*
 colorama
 datasets
@@ -7,6 +6,7 @@ duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index dbf35c3403..11bacf974e 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -1,11 +1,11 @@
 accelerate==1.5.*
-beautifulsoup4==4.13.4
 colorama
 datasets
 duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 2e5eb6c929..a64a93f032 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -1,11 +1,11 @@
 accelerate==1.5.*
-beautifulsoup4==4.13.4
 colorama
 datasets
 duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 9a19ab2911..62747ac4f8 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -1,11 +1,11 @@
 accelerate==1.5.*
-beautifulsoup4==4.13.4
 colorama
 datasets
 duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 973d9bfbfc..bc82f07aa0 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -1,11 +1,11 @@
 accelerate==1.5.*
-beautifulsoup4==4.13.4
 colorama
 datasets
 duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 4a48a51fb5..f880f40a48 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -1,11 +1,11 @@
 accelerate==1.5.*
-beautifulsoup4==4.13.4
 colorama
 datasets
 duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 76bde864b4..6d8875cbd1 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -1,11 +1,11 @@
 accelerate==1.5.*
-beautifulsoup4==4.13.4
 colorama
 datasets
 duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 9fc9960667..b2bcf91ceb 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -1,5 +1,4 @@
 accelerate==1.5.*
-beautifulsoup4==4.13.4
 bitsandbytes==0.45.*
 colorama
 datasets
@@ -7,6 +6,7 @@ duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index ff34673ab4..54496cd7cd 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -1,5 +1,4 @@
 accelerate==1.5.*
-beautifulsoup4==4.13.4
 bitsandbytes==0.45.*
 colorama
 datasets
@@ -7,6 +6,7 @@ duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 6cd0fa659e..eabcdbd063 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -1,5 +1,4 @@
 accelerate==1.5.*
-beautifulsoup4==4.13.4
 bitsandbytes==0.45.*
 colorama
 datasets
@@ -7,6 +6,7 @@ duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index a412367cf2..d26663a7a4 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -1,11 +1,11 @@
 accelerate==1.5.*
-beautifulsoup4==4.13.4
 colorama
 datasets
 duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 60ce941e52..5e5d4ba565 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -1,7 +1,7 @@
-beautifulsoup4==4.13.4
 duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index b1649bc962..4909f5a2eb 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -1,7 +1,7 @@
-beautifulsoup4==4.13.4
 duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 571eba52be..e54b2593f5 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -1,7 +1,7 @@
-beautifulsoup4==4.13.4
 duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 88170cf34a..74c0c5a722 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -1,7 +1,7 @@
-beautifulsoup4==4.13.4
 duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index e96cef491e..264bc37804 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -1,7 +1,7 @@
-beautifulsoup4==4.13.4
 duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 78f94aa53e..fcb8f05ec0 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -1,7 +1,7 @@
-beautifulsoup4==4.13.4
 duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index f6c866cf34..3d30e6d6bd 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -1,7 +1,7 @@
-beautifulsoup4==4.13.4
 duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 3e41427d03..395f225fd3 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -1,7 +1,7 @@
-beautifulsoup4==4.13.4
 duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 022ebb6161..0d41f541f4 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -1,7 +1,7 @@
-beautifulsoup4==4.13.4
 duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
+html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==1.26.*

From d085dc6a93577bbafbde529c497528965dbfc19b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Jun 2025 18:40:54 -0700
Subject: [PATCH 0657/1701] Minor optimization after
 e976a5ddc7598de49227f600515a7eee72fc1af6

---
 js/dark_theme.js | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/js/dark_theme.js b/js/dark_theme.js
index f61060cdf9..7136f5bf2e 100644
--- a/js/dark_theme.js
+++ b/js/dark_theme.js
@@ -9,8 +9,12 @@ function toggleDarkMode() {
 
   // Re-highlight all code blocks once stylesheet loads
   currentCSS.onload = function() {
-    document.querySelectorAll("pre code").forEach(block => {
-      hljs.highlightElement(block);
+    const messageBodies = document.getElementById("chat").querySelectorAll(".message-body");
+    messageBodies.forEach((messageBody) => {
+      const codeBlocks = messageBody.querySelectorAll("pre code");
+      codeBlocks.forEach((codeBlock) => {
+        hljs.highlightElement(codeBlock);
+      });
     });
   };
 }

From 1443612e72619e063d57226d05df3c159ec6d0ca Mon Sep 17 00:00:00 2001
From: Miriam 
Date: Mon, 9 Jun 2025 19:22:01 -0700
Subject: [PATCH 0658/1701] check .attention.head_count if
 .attention.head_count_kv doesn't exist (#7048)

---
 modules/models_settings.py | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index c914bdeae3..283a974477 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -329,6 +329,7 @@ def estimate_vram(gguf_file, gpu_layers, ctx_size, cache_type):
     # Extract values from metadata
     n_layers = None
     n_kv_heads = None
+    n_attention_heads = None  # Fallback for models without separate KV heads
     embedding_dim = None
 
     for key, value in metadata.items():
@@ -336,9 +337,14 @@ def estimate_vram(gguf_file, gpu_layers, ctx_size, cache_type):
             n_layers = value
         elif key.endswith('.attention.head_count_kv'):
             n_kv_heads = max(value) if isinstance(value, list) else value
+        elif key.endswith('.attention.head_count'):
+            n_attention_heads = max(value) if isinstance(value, list) else value
         elif key.endswith('.embedding_length'):
             embedding_dim = value
 
+    if n_kv_heads is None:
+        n_kv_heads = n_attention_heads
+
     if gpu_layers > n_layers:
         gpu_layers = n_layers
 

From 331d03c33fd69dd50b5c3b9f9ed8fb7d9da56bf0 Mon Sep 17 00:00:00 2001
From: Miriam 
Date: Mon, 9 Jun 2025 19:25:39 -0700
Subject: [PATCH 0659/1701] fix failure when --nowebui called without --api
 (#7055)

---
 server.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/server.py b/server.py
index 5cad1d8a82..80cc2f1875 100644
--- a/server.py
+++ b/server.py
@@ -318,8 +318,8 @@ def create_interface():
 
     if shared.args.nowebui:
         # Start the API in standalone mode
-        shared.args.extensions = [x for x in shared.args.extensions if x != 'gallery']
-        if shared.args.extensions is not None and len(shared.args.extensions) > 0:
+        shared.args.extensions = [x for x in (shared.args.extensions or []) if x != 'gallery']
+        if shared.args.extensions:
             extensions_module.load_extensions()
     else:
         # Launch the web UI

From ec731210200ab53dc26bf56e6616d4533f64c165 Mon Sep 17 00:00:00 2001
From: Mykeehu 
Date: Tue, 10 Jun 2025 05:17:05 +0200
Subject: [PATCH 0660/1701] Fix continue/start reply with when using
 translation extensions (#6944)

---------

Co-authored-by: oobabooga 
---
 modules/chat.py | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 25a0607b44..88ba33bb78 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -760,7 +760,18 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
         if is_stream:
             yield output
 
-    output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
+    if _continue:
+        # Reprocess the entire internal text for extensions (like translation)
+        full_internal = output['internal'][-1][1]
+        if state['mode'] in ['chat', 'chat-instruct']:
+            full_visible = re.sub("(||{{user}})", state['name1'], full_internal)
+        else:
+            full_visible = full_internal
+
+        full_visible = html.escape(full_visible)
+        output['visible'][-1][1] = apply_extensions('output', full_visible, state, is_chat=True)
+    else:
+        output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
 
     # Final sync for version metadata (in case streaming was disabled)
     if regenerate:

From df98f4b3312f84995d4dbdbe73681ed78c8e08d2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Jun 2025 20:28:16 -0700
Subject: [PATCH 0661/1701] Don't save active extensions through the UI

Prevents command-line activated extensions from becoming permanently active due to autosave.
---
 modules/shared.py | 1 -
 modules/ui.py     | 1 -
 server.py         | 6 ------
 3 files changed, 8 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 59c7dbcdb9..cc2884c1b3 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -224,7 +224,6 @@
     'negative_prompt': '',
     'dark_theme': True,
     'paste_to_attachment': False,
-    'default_extensions': [],
 
     # Character settings
     'character': 'Assistant',
diff --git a/modules/ui.py b/modules/ui.py
index 38693da8e4..ca1ff5284c 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -327,7 +327,6 @@ def save_settings(state, preset, extensions_list, show_controls, theme_state):
     output['prompt-default'] = state['prompt_menu-default']
     output['prompt-notebook'] = state['prompt_menu-notebook']
     output['character'] = state['character_menu']
-    output['default_extensions'] = extensions_list
     output['seed'] = int(output['seed'])
     output['show_controls'] = show_controls
     output['dark_theme'] = True if theme_state == 'dark' else False
diff --git a/server.py b/server.py
index 80cc2f1875..e0e3fbe550 100644
--- a/server.py
+++ b/server.py
@@ -249,13 +249,7 @@ def create_interface():
     shared.model_config['.*'] = get_fallback_settings()
     shared.model_config.move_to_end('.*', last=False)  # Move to the beginning
 
-    # Activate the extensions listed on settings.yaml
     extensions_module.available_extensions = utils.get_available_extensions()
-    for extension in shared.settings['default_extensions']:
-        shared.args.extensions = shared.args.extensions or []
-        if extension not in shared.args.extensions:
-            shared.args.extensions.append(extension)
-
     available_models = utils.get_available_models()
 
     # Model defined through --model

From efd9c9707b809754b2e467b53bbeb76688768e6d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Jun 2025 20:57:25 -0700
Subject: [PATCH 0662/1701] Fix random seeds being saved to settings.yaml

---
 modules/text_generation.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index 0d499d5070..55b538b069 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -479,6 +479,7 @@ def generate_reply_custom(question, original_question, state, stopping_strings=N
     For models that do not use the transformers library for sampling
     """
 
+    state = copy.deepcopy(state)
     state['seed'] = set_manual_seed(state['seed'])
     t0 = time.time()
     reply = ''

From c92eba0b0a0776119ce1912e25dbeeb7a1dfc749 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Jun 2025 22:03:23 -0700
Subject: [PATCH 0663/1701] Reorganize the Parameters tab (left: preset
 parameters, right: everything else)

---
 modules/ui_parameters.py | 19 +++++++++++--------
 1 file changed, 11 insertions(+), 8 deletions(-)

diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 9b5cb3abc3..e2b1055478 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -34,6 +34,7 @@ def create_ui():
                             shared.gradio['dynatemp_exponent'] = gr.Slider(0.01, 5, value=shared.settings['dynatemp_exponent'], step=0.01, label='dynatemp_exponent', visible=shared.settings['dynamic_temperature'])
                             shared.gradio['smoothing_factor'] = gr.Slider(0.0, 10.0, value=shared.settings['smoothing_factor'], step=0.01, label='smoothing_factor', info='Activates Quadratic Sampling.')
                             shared.gradio['smoothing_curve'] = gr.Slider(1.0, 10.0, value=shared.settings['smoothing_curve'], step=0.01, label='smoothing_curve', info='Adjusts the dropoff curve of Quadratic Sampling.')
+                            shared.gradio['dynamic_temperature'] = gr.Checkbox(value=shared.settings['dynamic_temperature'], label='dynamic_temperature')
 
                             gr.Markdown('## Curve cutoff')
                             shared.gradio['min_p'] = gr.Slider(0.0, 1.0, value=shared.settings['min_p'], step=0.01, label='min_p')
@@ -68,15 +69,19 @@ def create_ui():
                             shared.gradio['mirostat_eta'] = gr.Slider(0, 1, step=0.01, value=shared.settings['mirostat_eta'], label='mirostat_eta')
 
                             gr.Markdown('## Other options')
-                            shared.gradio['max_new_tokens'] = gr.Slider(minimum=shared.settings['max_new_tokens_min'], maximum=shared.settings['max_new_tokens_max'], value=shared.settings['max_new_tokens'], step=1, label='max_new_tokens', info='⚠️ Setting this too high can cause prompt truncation.')
-                            shared.gradio['prompt_lookup_num_tokens'] = gr.Slider(value=shared.settings['prompt_lookup_num_tokens'], minimum=0, maximum=10, step=1, label='prompt_lookup_num_tokens', info='Activates Prompt Lookup Decoding.')
-                            shared.gradio['max_tokens_second'] = gr.Slider(value=shared.settings['max_tokens_second'], minimum=0, maximum=20, step=1, label='Maximum tokens/second', info='To make text readable in real time.')
+                            shared.gradio['do_sample'] = gr.Checkbox(value=shared.settings['do_sample'], label='do_sample')
+                            shared.gradio['temperature_last'] = gr.Checkbox(value=shared.settings['temperature_last'], label='temperature_last', info='Moves temperature/dynamic temperature/quadratic sampling to the end of the sampler stack, ignoring their positions in "Sampler priority".')
+                            shared.gradio['sampler_priority'] = gr.Textbox(value=shared.settings['sampler_priority'], lines=10, label='Sampler priority', info='Parameter names separated by new lines or commas.', elem_classes=['add_scrollbar'])
+                            shared.gradio['dry_sequence_breakers'] = gr.Textbox(value=shared.settings['dry_sequence_breakers'], label='dry_sequence_breakers', info='Tokens across which sequence matching is not continued. Specified as a comma-separated list of quoted strings.')
+
                 with gr.Column():
                     with gr.Row():
                         with gr.Column():
-                            shared.gradio['do_sample'] = gr.Checkbox(value=shared.settings['do_sample'], label='do_sample')
-                            shared.gradio['dynamic_temperature'] = gr.Checkbox(value=shared.settings['dynamic_temperature'], label='dynamic_temperature')
-                            shared.gradio['temperature_last'] = gr.Checkbox(value=shared.settings['temperature_last'], label='temperature_last', info='Moves temperature/dynamic temperature/quadratic sampling to the end of the sampler stack, ignoring their positions in "Sampler priority".')
+                            with gr.Blocks():
+                                shared.gradio['max_new_tokens'] = gr.Slider(minimum=shared.settings['max_new_tokens_min'], maximum=shared.settings['max_new_tokens_max'], value=shared.settings['max_new_tokens'], step=1, label='max_new_tokens', info='⚠️ Setting this too high can cause prompt truncation.')
+                                shared.gradio['prompt_lookup_num_tokens'] = gr.Slider(value=shared.settings['prompt_lookup_num_tokens'], minimum=0, maximum=10, step=1, label='prompt_lookup_num_tokens', info='Activates Prompt Lookup Decoding.')
+                                shared.gradio['max_tokens_second'] = gr.Slider(value=shared.settings['max_tokens_second'], minimum=0, maximum=20, step=1, label='Maximum tokens/second', info='To make text readable in real time.')
+
                             shared.gradio['auto_max_new_tokens'] = gr.Checkbox(value=shared.settings['auto_max_new_tokens'], label='auto_max_new_tokens', info='Expand max_new_tokens to the available context length.')
                             shared.gradio['ban_eos_token'] = gr.Checkbox(value=shared.settings['ban_eos_token'], label='Ban the eos_token', info='Forces the model to never end the generation prematurely.')
                             shared.gradio['add_bos_token'] = gr.Checkbox(value=shared.settings['add_bos_token'], label='Add the bos_token to the beginning of prompts', info='Disabling this can make the replies more creative.')
@@ -89,11 +94,9 @@ def create_ui():
                             shared.gradio['truncation_length'] = gr.Number(precision=0, step=256, value=get_truncation_length(), label='Truncate the prompt up to this length', info='The leftmost tokens are removed if the prompt exceeds this length.')
                             shared.gradio['seed'] = gr.Number(value=shared.settings['seed'], label='Seed (-1 for random)')
 
-                            shared.gradio['sampler_priority'] = gr.Textbox(value=shared.settings['sampler_priority'], lines=12, label='Sampler priority', info='Parameter names separated by new lines or commas.', elem_classes=['add_scrollbar'])
                             shared.gradio['custom_stopping_strings'] = gr.Textbox(lines=2, value=shared.settings["custom_stopping_strings"] or None, label='Custom stopping strings', info='Written between "" and separated by commas.', placeholder='"\\n", "\\nYou:"')
                             shared.gradio['custom_token_bans'] = gr.Textbox(value=shared.settings['custom_token_bans'] or None, label='Token bans', info='Token IDs to ban, separated by commas. The IDs can be found in the Default or Notebook tab.')
                             shared.gradio['negative_prompt'] = gr.Textbox(value=shared.settings['negative_prompt'], label='Negative prompt', info='For CFG. Only used when guidance_scale is different than 1.', lines=3, elem_classes=['add_scrollbar'])
-                            shared.gradio['dry_sequence_breakers'] = gr.Textbox(value=shared.settings['dry_sequence_breakers'], label='dry_sequence_breakers', info='Tokens across which sequence matching is not continued. Specified as a comma-separated list of quoted strings.')
                             with gr.Row() as shared.gradio['grammar_file_row']:
                                 shared.gradio['grammar_file'] = gr.Dropdown(value='None', choices=utils.get_available_grammars(), label='Load grammar from file (.gbnf)', elem_classes='slim-dropdown')
                                 ui.create_refresh_button(shared.gradio['grammar_file'], lambda: None, lambda: {'choices': utils.get_available_grammars()}, 'refresh-button', interactive=not mu)

From 2dabdbc7da85f91ea9ddd74dfb49c2d161a6ef4d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Jun 2025 05:25:23 -0700
Subject: [PATCH 0664/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_cuda128.txt             | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt      | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index b751482a2e..a71e52406a 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 11bacf974e..db1ead1a59 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index a64a93f032..a08aa39201 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 62747ac4f8..fa217c3ead 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index bc82f07aa0..52581f1aa9 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index f880f40a48..b72f22aa20 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 6d8875cbd1..e8de605700 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index b2bcf91ceb..801009fb77 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 54496cd7cd..8b77e144fd 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index eabcdbd063..5e81ce1f7f 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 5e5d4ba565..4ddcf43f41 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 4909f5a2eb..38a21618a0 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index e54b2593f5..0b70c80058 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 74c0c5a722..510a20f45a 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 264bc37804..e6d9f0c5df 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index fcb8f05ec0..48f92e0a38 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 395f225fd3..9f93424f8b 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 0d41f541f4..9070b9a609 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.16.0/llama_cpp_binaries-0.16.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 889153952f5545fb2b94b7a853aa24ae754ef79f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Jun 2025 09:02:52 -0700
Subject: [PATCH 0665/1701] Lint

---
 modules/ui_model_menu.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 2a7d3d9df3..9e982f0ec4 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -260,6 +260,7 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
             output = "```\n"
             for link in links:
                 output += f"{Path(link).name}" + "\n"
+
             output += "```"
             yield output
             return

From 18bd78f1f038b2b69178d464184b937f7d2b15d6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Jun 2025 14:03:25 -0700
Subject: [PATCH 0666/1701] Make the llama.cpp prompt processing messages
 shorter

---
 modules/llama_cpp_server.py | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index f0a72de895..a79e24e4e8 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -422,9 +422,17 @@ def filter_stderr_with_progress(process_stderr):
 
             if match:
                 progress = float(match.group(1))
+
+                # Extract just the part from "prompt processing" onwards
+                prompt_processing_idx = line.find('prompt processing')
+                if prompt_processing_idx != -1:
+                    display_line = line[prompt_processing_idx:]
+                else:
+                    display_line = line  # fallback to full line
+
                 # choose carriage return for in-progress or newline at completion
                 end_char = '\r' if progress < 1.0 else '\n'
-                print(line, end=end_char, file=sys.stderr, flush=True)
+                print(display_line, end=end_char, file=sys.stderr, flush=True)
                 last_was_progress = (progress < 1.0)
 
             # skip noise lines

From 3f9eb3aad1d1a734791dd41e4817570d81b485eb Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Jun 2025 14:22:37 -0700
Subject: [PATCH 0667/1701] Fix the preset dropdown when the default preset
 file is not present

---
 modules/shared.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/shared.py b/modules/shared.py
index cc2884c1b3..b8ab24266f 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -204,7 +204,7 @@
     'web_search_pages': 3,
     'prompt-default': 'QA',
     'prompt-notebook': 'QA',
-    'preset': 'Qwen3 - Thinking' if Path('user_data/presets/Qwen3 - Thinking.yaml').exists() else '',
+    'preset': 'Qwen3 - Thinking' if Path('user_data/presets/Qwen3 - Thinking.yaml').exists() else None,
     'max_new_tokens': 512,
     'max_new_tokens_min': 1,
     'max_new_tokens_max': 4096,

From 1c1fd3be46f66b2f3110aed52a155817ca3392b7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Jun 2025 14:29:28 -0700
Subject: [PATCH 0668/1701] Remove some log messages

---
 modules/chat.py | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 88ba33bb78..dfc301df49 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1271,8 +1271,6 @@ def clear_character_for_ui(state):
         if cache_path.exists():
             cache_path.unlink()
 
-    logger.info("Cleared character fields and picture cache")
-
     return state, state['name2'], state['context'], state['greeting'], None
 
 
@@ -1779,7 +1777,6 @@ def handle_character_picture_change(picture):
         picture.save(Path(f'{cache_folder}/pfp_character.png'), format='PNG')
         thumb = make_thumbnail(picture)
         thumb.save(Path(f'{cache_folder}/pfp_character_thumb.png'), format='PNG')
-        logger.info("Updated character picture cache")
     else:
         # Remove cache files when picture is cleared
         for cache_file in ['pfp_character.png', 'pfp_character_thumb.png']:
@@ -1787,8 +1784,6 @@ def handle_character_picture_change(picture):
             if cache_path.exists():
                 cache_path.unlink()
 
-        logger.info("Cleared character picture cache")
-
 
 def handle_mode_change(state):
     history = load_latest_history(state)

From 75da90190f33317dd9731e14d844dd7238a2c870 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Jun 2025 17:34:54 -0700
Subject: [PATCH 0669/1701] Fix character dropdown sometimes disappearing in
 the Parameters tab

---
 modules/ui_chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 428b64c98c..3b841b8b24 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -298,7 +298,7 @@ def create_event_handlers():
     shared.gradio['mode'].change(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_mode_change, gradio('interface_state'), gradio('history', 'display', 'chat_style', 'chat-instruct_command', 'unique_id'), show_progress=False).then(
-        None, gradio('mode'), None, js="(mode) => {mode === 'instruct' ? document.getElementById('character-menu').parentNode.parentNode.style.display = 'none' : document.getElementById('character-menu').parentNode.parentNode.style.display = ''}")
+        None, gradio('mode'), None, js="(mode) => {const characterContainer = document.getElementById('character-menu').parentNode.parentNode; const isInChatTab = document.querySelector('#chat-controls').contains(characterContainer); if (isInChatTab) { characterContainer.style.display = mode === 'instruct' ? 'none' : ''; }}")
 
     shared.gradio['chat_style'].change(chat.redraw_html, gradio(reload_arr), gradio('display'), show_progress=False)
 

From 4cf39120fca88fc78aeaaa55946ab74e5db1a512 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Jun 2025 18:03:00 -0700
Subject: [PATCH 0670/1701] Fix chat area sometimes not scrolling up to edit
 message

---
 js/global_scope_js.js | 9 +++++++++
 js/main.js            | 8 ++++----
 2 files changed, 13 insertions(+), 4 deletions(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 801f1574cf..205d937535 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -95,12 +95,21 @@ function startEditing(messageElement, messageBody, isUserMessage) {
   editingInterface.textarea.focus();
   editingInterface.textarea.setSelectionRange(rawText.length, rawText.length);
 
+  // Temporarily mark as scrolled to prevent auto-scroll
+  const wasScrolled = window.isScrolled;
+  window.isScrolled = true;
+
   // Scroll the textarea into view
   editingInterface.textarea.scrollIntoView({
     behavior: "smooth",
     block: "center"
   });
 
+  // Restore the original scroll state after animation
+  setTimeout(() => {
+    window.isScrolled = wasScrolled;
+  }, 500);
+
   // Setup event handlers
   setupEditingHandlers(editingInterface.textarea, messageElement, originalHTML, messageBody, isUserMessage);
 }
diff --git a/js/main.js b/js/main.js
index 1953a6be74..e970884d5c 100644
--- a/js/main.js
+++ b/js/main.js
@@ -145,7 +145,7 @@ typingSibling.insertBefore(typing, typingSibling.childNodes[2]);
 const targetElement = document.getElementById("chat").parentNode.parentNode.parentNode;
 targetElement.classList.add("pretty_scrollbar");
 targetElement.classList.add("chat-parent");
-let isScrolled = false;
+window.isScrolled = false;
 let scrollTimeout;
 
 targetElement.addEventListener("scroll", function() {
@@ -154,9 +154,9 @@ targetElement.addEventListener("scroll", function() {
 
   let diff = targetElement.scrollHeight - targetElement.clientHeight;
   if(Math.abs(targetElement.scrollTop - diff) <= 10 || diff == 0) {
-    isScrolled = false;
+    window.isScrolled = false;
   } else {
-    isScrolled = true;
+    window.isScrolled = true;
   }
 
   // Clear previous timeout and set new one
@@ -182,7 +182,7 @@ const observer = new MutationObserver(function(mutations) {
 
   doSyntaxHighlighting();
 
-  if (!isScrolled && targetElement.scrollTop !== targetElement.scrollHeight) {
+  if (!window.isScrolled && targetElement.scrollTop !== targetElement.scrollHeight) {
     targetElement.scrollTop = targetElement.scrollHeight;
   }
 

From bc921c66e5ef2f2a14286702e564c1b22dbba5a7 Mon Sep 17 00:00:00 2001
From: LawnMauer 
Date: Wed, 11 Jun 2025 03:16:50 +0200
Subject: [PATCH 0671/1701] Load js and css sources in UTF-8 (#7059)

---
 modules/html_generator.py |  7 ++++---
 modules/ui.py             | 22 +++++++++++-----------
 2 files changed, 15 insertions(+), 14 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index eac7d91ac8..af64894e4d 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -39,15 +39,16 @@ def minify_css(css: str) -> str:
     return css
 
 
-with open(Path(__file__).resolve().parent / '../css/html_readable_style.css', 'r') as f:
+with open(Path(__file__).resolve().parent / '../css/html_readable_style.css', 'r', encoding='utf-8') as f:
     readable_css = f.read()
-with open(Path(__file__).resolve().parent / '../css/html_instruct_style.css', 'r') as f:
+with open(Path(__file__).resolve().parent / '../css/html_instruct_style.css', 'r', encoding='utf-8') as f:
     instruct_css = f.read()
 
 # Custom chat styles
 chat_styles = {}
 for k in get_available_chat_styles():
-    chat_styles[k] = open(Path(f'css/chat_style-{k}.css'), 'r').read()
+    with open(Path(f'css/chat_style-{k}.css'), 'r', encoding='utf-8') as f:
+        chat_styles[k] = f.read()
 
 # Handle styles that derive from other styles
 for k in chat_styles:
diff --git a/modules/ui.py b/modules/ui.py
index ca1ff5284c..59da511860 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -19,27 +19,27 @@
 _last_show_controls = None
 _last_theme_state = None
 
-with open(Path(__file__).resolve().parent / '../css/NotoSans/stylesheet.css', 'r') as f:
+with open(Path(__file__).resolve().parent / '../css/NotoSans/stylesheet.css', 'r', encoding='utf-8') as f:
     css = f.read()
-with open(Path(__file__).resolve().parent / '../css/main.css', 'r') as f:
+with open(Path(__file__).resolve().parent / '../css/main.css', 'r', encoding='utf-8') as f:
     css += f.read()
-with open(Path(__file__).resolve().parent / '../css/katex/katex.min.css', 'r') as f:
+with open(Path(__file__).resolve().parent / '../css/katex/katex.min.css', 'r', encoding='utf-8') as f:
     css += f.read()
-with open(Path(__file__).resolve().parent / '../css/highlightjs/highlightjs-copy.min.css', 'r') as f:
+with open(Path(__file__).resolve().parent / '../css/highlightjs/highlightjs-copy.min.css', 'r', encoding='utf-8') as f:
     css += f.read()
-with open(Path(__file__).resolve().parent / '../js/main.js', 'r') as f:
+with open(Path(__file__).resolve().parent / '../js/main.js', 'r', encoding='utf-8') as f:
     js = f.read()
-with open(Path(__file__).resolve().parent / '../js/global_scope_js.js', 'r') as f:
+with open(Path(__file__).resolve().parent / '../js/global_scope_js.js', 'r', encoding='utf-8') as f:
     global_scope_js = f.read()
-with open(Path(__file__).resolve().parent / '../js/save_files.js', 'r') as f:
+with open(Path(__file__).resolve().parent / '../js/save_files.js', 'r', encoding='utf-8') as f:
     save_files_js = f.read()
-with open(Path(__file__).resolve().parent / '../js/switch_tabs.js', 'r') as f:
+with open(Path(__file__).resolve().parent / '../js/switch_tabs.js', 'r', encoding='utf-8') as f:
     switch_tabs_js = f.read()
-with open(Path(__file__).resolve().parent / '../js/show_controls.js', 'r') as f:
+with open(Path(__file__).resolve().parent / '../js/show_controls.js', 'r', encoding='utf-8') as f:
     show_controls_js = f.read()
-with open(Path(__file__).resolve().parent / '../js/update_big_picture.js', 'r') as f:
+with open(Path(__file__).resolve().parent / '../js/update_big_picture.js', 'r', encoding='utf-8') as f:
     update_big_picture_js = f.read()
-with open(Path(__file__).resolve().parent / '../js/dark_theme.js', 'r') as f:
+with open(Path(__file__).resolve().parent / '../js/dark_theme.js', 'r', encoding='utf-8') as f:
     dark_theme_js = f.read()
 
 refresh_symbol = '🔄'

From 552cb09f09de0d3a9442334580fd4a66f95a1ce3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Jun 2025 18:45:42 -0700
Subject: [PATCH 0672/1701] Do not bump Transformers to 4.52 on CUDA 12.8

Performance is slow, and the older version works fine with torch 2.7.
---
 requirements/full/requirements_cuda128.txt        | 3 +--
 requirements/full/requirements_cuda128_noavx2.txt | 3 +--
 2 files changed, 2 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 801009fb77..7851041f1e 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -24,8 +24,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.52.*
-triton-windows; platform_system == "Windows"
+transformers==4.50.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 8b77e144fd..c801516608 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -24,8 +24,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.52.*
-triton-windows; platform_system == "Windows"
+transformers==4.50.*
 tqdm
 wandb
 

From a86a5a026e1d5c38b9d63900e877eda3f5631501 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Jun 2025 19:25:22 -0700
Subject: [PATCH 0673/1701] Fix the GitHub Actions workflows

---
 .github/workflows/build-portable-release-cuda.yml   | 3 +++
 .github/workflows/build-portable-release-vulkan.yml | 3 +++
 .github/workflows/build-portable-release.yml        | 3 +++
 3 files changed, 9 insertions(+)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index 283fdd72cb..0de2d1451f 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -175,6 +175,9 @@ jobs:
                 zip -r "$ZIP_NAME" text-generation-webui-${VERSION_CLEAN}
             fi
 
+            # Rename back after creating the zip (the next step assumes this folder exists)
+            mv text-generation-webui-${VERSION_CLEAN} text-generation-webui
+
       - name: Upload files to a GitHub release
         id: upload-release
         uses: svenstaro/upload-release-action@2.7.0
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index c6ab8fa75a..8437c56ec0 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -161,6 +161,9 @@ jobs:
                 zip -r "$ZIP_NAME" text-generation-webui-${VERSION_CLEAN}
             fi
 
+            # Rename back after creating the zip (the next step assumes this folder exists)
+            mv text-generation-webui-${VERSION_CLEAN} text-generation-webui
+
       - name: Upload files to a GitHub release
         id: upload-release
         uses: svenstaro/upload-release-action@2.7.0
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 58bfdb256c..f60837bf89 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -185,6 +185,9 @@ jobs:
                 zip -r "$ZIP_NAME" text-generation-webui-${VERSION_CLEAN}
             fi
 
+            # Rename back after creating the zip (the next step assumes this folder exists)
+            mv text-generation-webui-${VERSION_CLEAN} text-generation-webui
+
       - name: Upload files to a GitHub release
         id: upload-release
         uses: svenstaro/upload-release-action@2.7.0

From 43fc170224e0fb2ffd93c93f6a2b1c16e37344aa Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Jun 2025 19:34:41 -0700
Subject: [PATCH 0674/1701] Fix the Windows workflow

---
 .github/workflows/build-portable-release-cuda.yml   | 6 +++++-
 .github/workflows/build-portable-release-vulkan.yml | 6 +++++-
 .github/workflows/build-portable-release.yml        | 6 +++++-
 3 files changed, 15 insertions(+), 3 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index 0de2d1451f..f6d8b39cf5 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -176,7 +176,11 @@ jobs:
             fi
 
             # Rename back after creating the zip (the next step assumes this folder exists)
-            mv text-generation-webui-${VERSION_CLEAN} text-generation-webui
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                powershell -Command "Move-Item -Path text-generation-webui-${VERSION_CLEAN} -Destination text-generation-webui -Force"
+            else
+                mv text-generation-webui-${VERSION_CLEAN} text-generation-webui
+            fi
 
       - name: Upload files to a GitHub release
         id: upload-release
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 8437c56ec0..55ce13d502 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -162,7 +162,11 @@ jobs:
             fi
 
             # Rename back after creating the zip (the next step assumes this folder exists)
-            mv text-generation-webui-${VERSION_CLEAN} text-generation-webui
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                powershell -Command "Move-Item -Path text-generation-webui-${VERSION_CLEAN} -Destination text-generation-webui -Force"
+            else
+                mv text-generation-webui-${VERSION_CLEAN} text-generation-webui
+            fi
 
       - name: Upload files to a GitHub release
         id: upload-release
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index f60837bf89..5df2db3193 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -186,7 +186,11 @@ jobs:
             fi
 
             # Rename back after creating the zip (the next step assumes this folder exists)
-            mv text-generation-webui-${VERSION_CLEAN} text-generation-webui
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                powershell -Command "Move-Item -Path text-generation-webui-${VERSION_CLEAN} -Destination text-generation-webui -Force"
+            else
+                mv text-generation-webui-${VERSION_CLEAN} text-generation-webui
+            fi
 
       - name: Upload files to a GitHub release
         id: upload-release

From fe0685a7421075e1b9761f32ea034d5ba83e0a83 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Jun 2025 19:42:22 -0700
Subject: [PATCH 0675/1701] New attempt

---
 .github/workflows/build-portable-release-cuda.yml   | 7 ++++++-
 .github/workflows/build-portable-release-vulkan.yml | 7 ++++++-
 .github/workflows/build-portable-release.yml        | 7 ++++++-
 3 files changed, 18 insertions(+), 3 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index f6d8b39cf5..7ba1fc6b39 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -163,7 +163,12 @@ jobs:
             # 6. Move up and rename folder to include version
             cd ..
             VERSION_CLEAN="${VERSION#v}"
-            mv text-generation-webui text-generation-webui-${VERSION_CLEAN}
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                sleep 3
+                powershell -Command "Move-Item -Path text-generation-webui -Destination text-generation-webui-${VERSION_CLEAN} -Force"
+            else
+                mv text-generation-webui text-generation-webui-${VERSION_CLEAN}
+            fi
 
             # 7. Create ZIP file
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 55ce13d502..c02aa662c2 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -149,7 +149,12 @@ jobs:
             # 5. Move up and rename folder to include version
             cd ..
             VERSION_CLEAN="${VERSION#v}"
-            mv text-generation-webui text-generation-webui-${VERSION_CLEAN}
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                sleep 3
+                powershell -Command "Move-Item -Path text-generation-webui -Destination text-generation-webui-${VERSION_CLEAN} -Force"
+            else
+                mv text-generation-webui text-generation-webui-${VERSION_CLEAN}
+            fi
 
             # 6. Create ZIP file
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.zip"
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 5df2db3193..85179617ce 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -173,7 +173,12 @@ jobs:
             # 5. Move up and rename folder to include version
             cd ..
             VERSION_CLEAN="${VERSION#v}"
-            mv text-generation-webui text-generation-webui-${VERSION_CLEAN}
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                sleep 3  # Give time for any lingering file handles to close
+                powershell -Command "Move-Item -Path text-generation-webui -Destination text-generation-webui-${VERSION_CLEAN} -Force"
+            else
+                mv text-generation-webui text-generation-webui-${VERSION_CLEAN}
+            fi
 
             # 6. Create ZIP file
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.zip"

From 00fbbd6f57efe59f85ea61278300603aeea13541 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Jun 2025 19:46:28 -0700
Subject: [PATCH 0676/1701] Undo changes to portable builds

---
 .../workflows/build-portable-release-cuda.yml | 21 +++----------------
 .../build-portable-release-vulkan.yml         | 21 +++----------------
 .github/workflows/build-portable-release.yml  | 21 +++----------------
 3 files changed, 9 insertions(+), 54 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index 7ba1fc6b39..571cbac0c5 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -160,31 +160,16 @@ jobs:
                 rm requirements_cuda_temp.txt
             fi
 
-            # 6. Move up and rename folder to include version
+            # 6. Create ZIP file
             cd ..
             VERSION_CLEAN="${VERSION#v}"
-            if [[ "$RUNNER_OS" == "Windows" ]]; then
-                sleep 3
-                powershell -Command "Move-Item -Path text-generation-webui -Destination text-generation-webui-${VERSION_CLEAN} -Force"
-            else
-                mv text-generation-webui text-generation-webui-${VERSION_CLEAN}
-            fi
-
-            # 7. Create ZIP file
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
-            else
-                zip -r "$ZIP_NAME" text-generation-webui-${VERSION_CLEAN}
-            fi
-
-            # Rename back after creating the zip (the next step assumes this folder exists)
-            if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Move-Item -Path text-generation-webui-${VERSION_CLEAN} -Destination text-generation-webui -Force"
+                powershell -Command "Compress-Archive -Path text-generation-webui -DestinationPath $ZIP_NAME"
             else
-                mv text-generation-webui-${VERSION_CLEAN} text-generation-webui
+                zip -r "$ZIP_NAME" text-generation-webui
             fi
 
       - name: Upload files to a GitHub release
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index c02aa662c2..4e88d4d9ab 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -146,31 +146,16 @@ jobs:
             echo "Installing Python packages from $REQ_FILE..."
             $PIP_PATH install --target="./$PACKAGES_PATH" -r "$REQ_FILE"
 
-            # 5. Move up and rename folder to include version
+            # 6. Create ZIP file
             cd ..
             VERSION_CLEAN="${VERSION#v}"
-            if [[ "$RUNNER_OS" == "Windows" ]]; then
-                sleep 3
-                powershell -Command "Move-Item -Path text-generation-webui -Destination text-generation-webui-${VERSION_CLEAN} -Force"
-            else
-                mv text-generation-webui text-generation-webui-${VERSION_CLEAN}
-            fi
-
-            # 6. Create ZIP file
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
-            else
-                zip -r "$ZIP_NAME" text-generation-webui-${VERSION_CLEAN}
-            fi
-
-            # Rename back after creating the zip (the next step assumes this folder exists)
-            if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Move-Item -Path text-generation-webui-${VERSION_CLEAN} -Destination text-generation-webui -Force"
+                powershell -Command "Compress-Archive -Path text-generation-webui -DestinationPath $ZIP_NAME"
             else
-                mv text-generation-webui-${VERSION_CLEAN} text-generation-webui
+                zip -r "$ZIP_NAME" text-generation-webui
             fi
 
       - name: Upload files to a GitHub release
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 85179617ce..6910ce2c3b 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -170,31 +170,16 @@ jobs:
             echo "Installing Python packages from $REQ_FILE..."
             $PIP_PATH install --target="./$PACKAGES_PATH" -r "$REQ_FILE"
 
-            # 5. Move up and rename folder to include version
+            # 5. Create ZIP file
             cd ..
             VERSION_CLEAN="${VERSION#v}"
-            if [[ "$RUNNER_OS" == "Windows" ]]; then
-                sleep 3  # Give time for any lingering file handles to close
-                powershell -Command "Move-Item -Path text-generation-webui -Destination text-generation-webui-${VERSION_CLEAN} -Force"
-            else
-                mv text-generation-webui text-generation-webui-${VERSION_CLEAN}
-            fi
-
-            # 6. Create ZIP file
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
-            else
-                zip -r "$ZIP_NAME" text-generation-webui-${VERSION_CLEAN}
-            fi
-
-            # Rename back after creating the zip (the next step assumes this folder exists)
-            if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Move-Item -Path text-generation-webui-${VERSION_CLEAN} -Destination text-generation-webui -Force"
+                powershell -Command "Compress-Archive -Path text-generation-webui -DestinationPath $ZIP_NAME"
             else
-                mv text-generation-webui-${VERSION_CLEAN} text-generation-webui
+                zip -r "$ZIP_NAME" text-generation-webui
             fi
 
       - name: Upload files to a GitHub release

From 801db438b058acf05db14088fafd916435f8dcd6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Jun 2025 19:46:28 -0700
Subject: [PATCH 0677/1701] Undo changes to portable builds

---
 .../workflows/build-portable-release-cuda.yml | 21 +++----------------
 .../build-portable-release-vulkan.yml         | 21 +++----------------
 .github/workflows/build-portable-release.yml  | 21 +++----------------
 3 files changed, 9 insertions(+), 54 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index 7ba1fc6b39..571cbac0c5 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -160,31 +160,16 @@ jobs:
                 rm requirements_cuda_temp.txt
             fi
 
-            # 6. Move up and rename folder to include version
+            # 6. Create ZIP file
             cd ..
             VERSION_CLEAN="${VERSION#v}"
-            if [[ "$RUNNER_OS" == "Windows" ]]; then
-                sleep 3
-                powershell -Command "Move-Item -Path text-generation-webui -Destination text-generation-webui-${VERSION_CLEAN} -Force"
-            else
-                mv text-generation-webui text-generation-webui-${VERSION_CLEAN}
-            fi
-
-            # 7. Create ZIP file
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
-            else
-                zip -r "$ZIP_NAME" text-generation-webui-${VERSION_CLEAN}
-            fi
-
-            # Rename back after creating the zip (the next step assumes this folder exists)
-            if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Move-Item -Path text-generation-webui-${VERSION_CLEAN} -Destination text-generation-webui -Force"
+                powershell -Command "Compress-Archive -Path text-generation-webui -DestinationPath $ZIP_NAME"
             else
-                mv text-generation-webui-${VERSION_CLEAN} text-generation-webui
+                zip -r "$ZIP_NAME" text-generation-webui
             fi
 
       - name: Upload files to a GitHub release
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index c02aa662c2..4e88d4d9ab 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -146,31 +146,16 @@ jobs:
             echo "Installing Python packages from $REQ_FILE..."
             $PIP_PATH install --target="./$PACKAGES_PATH" -r "$REQ_FILE"
 
-            # 5. Move up and rename folder to include version
+            # 6. Create ZIP file
             cd ..
             VERSION_CLEAN="${VERSION#v}"
-            if [[ "$RUNNER_OS" == "Windows" ]]; then
-                sleep 3
-                powershell -Command "Move-Item -Path text-generation-webui -Destination text-generation-webui-${VERSION_CLEAN} -Force"
-            else
-                mv text-generation-webui text-generation-webui-${VERSION_CLEAN}
-            fi
-
-            # 6. Create ZIP file
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
-            else
-                zip -r "$ZIP_NAME" text-generation-webui-${VERSION_CLEAN}
-            fi
-
-            # Rename back after creating the zip (the next step assumes this folder exists)
-            if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Move-Item -Path text-generation-webui-${VERSION_CLEAN} -Destination text-generation-webui -Force"
+                powershell -Command "Compress-Archive -Path text-generation-webui -DestinationPath $ZIP_NAME"
             else
-                mv text-generation-webui-${VERSION_CLEAN} text-generation-webui
+                zip -r "$ZIP_NAME" text-generation-webui
             fi
 
       - name: Upload files to a GitHub release
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 85179617ce..6910ce2c3b 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -170,31 +170,16 @@ jobs:
             echo "Installing Python packages from $REQ_FILE..."
             $PIP_PATH install --target="./$PACKAGES_PATH" -r "$REQ_FILE"
 
-            # 5. Move up and rename folder to include version
+            # 5. Create ZIP file
             cd ..
             VERSION_CLEAN="${VERSION#v}"
-            if [[ "$RUNNER_OS" == "Windows" ]]; then
-                sleep 3  # Give time for any lingering file handles to close
-                powershell -Command "Move-Item -Path text-generation-webui -Destination text-generation-webui-${VERSION_CLEAN} -Force"
-            else
-                mv text-generation-webui text-generation-webui-${VERSION_CLEAN}
-            fi
-
-            # 6. Create ZIP file
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
-            else
-                zip -r "$ZIP_NAME" text-generation-webui-${VERSION_CLEAN}
-            fi
-
-            # Rename back after creating the zip (the next step assumes this folder exists)
-            if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Move-Item -Path text-generation-webui-${VERSION_CLEAN} -Destination text-generation-webui -Force"
+                powershell -Command "Compress-Archive -Path text-generation-webui -DestinationPath $ZIP_NAME"
             else
-                mv text-generation-webui-${VERSION_CLEAN} text-generation-webui
+                zip -r "$ZIP_NAME" text-generation-webui
             fi
 
       - name: Upload files to a GitHub release

From 13a5288d011495ad8a354ab7d2d0dbd1ccf19e1c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Jun 2025 21:08:18 -0700
Subject: [PATCH 0678/1701] Fix an error when upgrading from cuda 12.4 to cuda
 12.8

---
 one_click.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/one_click.py b/one_click.py
index 94f2aab075..050da76b10 100644
--- a/one_click.py
+++ b/one_click.py
@@ -298,7 +298,7 @@ def update_pytorch_and_python():
 
 
 def clean_outdated_pytorch_cuda_dependencies():
-    patterns = ["cu121", "cu122", "torch2.4"]
+    patterns = ["cu121", "cu122", "torch2.4", "torchvision", "torchaudio"]
     result = run_cmd("python -m pip list --format=freeze", capture_output=True, environment=True)
     matching_packages = []
 

From 27140f3563eda5fd3f0044f129df851830fb3eed Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 11 Jun 2025 07:25:27 -0700
Subject: [PATCH 0679/1701] Revert "Don't save active extensions through the
 UI"

This reverts commit df98f4b3312f84995d4dbdbe73681ed78c8e08d2.
---
 modules/shared.py | 1 +
 modules/ui.py     | 1 +
 server.py         | 6 ++++++
 3 files changed, 8 insertions(+)

diff --git a/modules/shared.py b/modules/shared.py
index b8ab24266f..a91b48b95e 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -224,6 +224,7 @@
     'negative_prompt': '',
     'dark_theme': True,
     'paste_to_attachment': False,
+    'default_extensions': [],
 
     # Character settings
     'character': 'Assistant',
diff --git a/modules/ui.py b/modules/ui.py
index 59da511860..e716672ee9 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -327,6 +327,7 @@ def save_settings(state, preset, extensions_list, show_controls, theme_state):
     output['prompt-default'] = state['prompt_menu-default']
     output['prompt-notebook'] = state['prompt_menu-notebook']
     output['character'] = state['character_menu']
+    output['default_extensions'] = extensions_list
     output['seed'] = int(output['seed'])
     output['show_controls'] = show_controls
     output['dark_theme'] = True if theme_state == 'dark' else False
diff --git a/server.py b/server.py
index e0e3fbe550..80cc2f1875 100644
--- a/server.py
+++ b/server.py
@@ -249,7 +249,13 @@ def create_interface():
     shared.model_config['.*'] = get_fallback_settings()
     shared.model_config.move_to_end('.*', last=False)  # Move to the beginning
 
+    # Activate the extensions listed on settings.yaml
     extensions_module.available_extensions = utils.get_available_extensions()
+    for extension in shared.settings['default_extensions']:
+        shared.args.extensions = shared.args.extensions or []
+        if extension not in shared.args.extensions:
+            shared.args.extensions.append(extension)
+
     available_models = utils.get_available_models()
 
     # Model defined through --model

From 570d5b89363dd0a1c6a63e74865365584eddd1e2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 11 Jun 2025 07:39:49 -0700
Subject: [PATCH 0680/1701] Only save extensions on manual save

---
 modules/ui.py         | 20 +++++++++++++++++---
 modules/ui_session.py |  2 +-
 2 files changed, 18 insertions(+), 4 deletions(-)

diff --git a/modules/ui.py b/modules/ui.py
index e716672ee9..52688128de 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -316,7 +316,7 @@ def apply_interface_values(state, use_persistent=False):
         return [state[k] if k in state else gr.update() for k in elements]
 
 
-def save_settings(state, preset, extensions_list, show_controls, theme_state):
+def save_settings(state, preset, extensions_list, show_controls, theme_state, manual_save=False):
     output = copy.deepcopy(shared.settings)
     exclude = []
     for k in state:
@@ -327,13 +327,27 @@ def save_settings(state, preset, extensions_list, show_controls, theme_state):
     output['prompt-default'] = state['prompt_menu-default']
     output['prompt-notebook'] = state['prompt_menu-notebook']
     output['character'] = state['character_menu']
-    output['default_extensions'] = extensions_list
     output['seed'] = int(output['seed'])
     output['show_controls'] = show_controls
     output['dark_theme'] = True if theme_state == 'dark' else False
     output.pop('instruction_template_str')
     output.pop('truncation_length')
 
+    # Only save extensions on manual save
+    if manual_save:
+        output['default_extensions'] = extensions_list
+    else:
+        # Preserve existing extensions from settings file during autosave
+        settings_path = Path('user_data') / 'settings.yaml'
+        if settings_path.exists():
+            try:
+                with open(settings_path, 'r', encoding='utf-8') as f:
+                    existing_settings = yaml.safe_load(f.read()) or {}
+                if 'default_extensions' in existing_settings:
+                    output['default_extensions'] = existing_settings['default_extensions']
+            except Exception:
+                pass  # If we can't read the file, just don't modify extensions
+
     # Save extension values in the UI
     for extension_name in extensions_list:
         extension = getattr(extensions, extension_name, None)
@@ -384,7 +398,7 @@ def _perform_debounced_save():
 
     try:
         if _last_interface_state is not None:
-            contents = save_settings(_last_interface_state, _last_preset, _last_extensions, _last_show_controls, _last_theme_state)
+            contents = save_settings(_last_interface_state, _last_preset, _last_extensions, _last_show_controls, _last_theme_state, manual_save=False)
             settings_path = Path('user_data') / 'settings.yaml'
             settings_path.parent.mkdir(exist_ok=True)
             with open(settings_path, 'w', encoding='utf-8') as f:
diff --git a/modules/ui_session.py b/modules/ui_session.py
index 33d7dcb733..42fcf89d79 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -40,7 +40,7 @@ def create_ui():
 
 
 def handle_save_settings(state, preset, extensions, show_controls, theme):
-    contents = ui.save_settings(state, preset, extensions, show_controls, theme)
+    contents = ui.save_settings(state, preset, extensions, show_controls, theme, manual_save=True)
     return [
         contents,
         "settings.yaml",

From 004fd8316c723f093bfd68099f8609477b4f623b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 11 Jun 2025 07:47:25 -0700
Subject: [PATCH 0681/1701] Minor changes

---
 modules/shared.py     | 26 ++++++++++++++------------
 modules/ui.py         |  1 +
 modules/ui_session.py |  2 +-
 3 files changed, 16 insertions(+), 13 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index a91b48b95e..83920df860 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -224,18 +224,6 @@
     'negative_prompt': '',
     'dark_theme': True,
     'paste_to_attachment': False,
-    'default_extensions': [],
-
-    # Character settings
-    'character': 'Assistant',
-    'name1': 'You',
-    'name2': 'AI',
-    'user_bio': '',
-    'context': 'The following is a conversation with an AI Large Language Model. The AI has been trained to answer questions, provide recommendations, and help with decision making. The AI follows user requests. The AI thinks outside the box.',
-    'greeting': 'How can I help you today?',
-    'custom_system_message': '',
-    'instruction_template_str': "{%- set ns = namespace(found=false) -%}\n{%- for message in messages -%}\n    {%- if message['role'] == 'system' -%}\n        {%- set ns.found = true -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if not ns.found -%}\n    {{- '' + 'Below is an instruction that describes a task. Write a response that appropriately completes the request.' + '\\n\\n' -}}\n{%- endif %}\n{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {{- '' + message['content'] + '\\n\\n' -}}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{-'### Instruction:\\n' + message['content'] + '\\n\\n'-}}\n        {%- else -%}\n            {{-'### Response:\\n' + message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{-'### Response:\\n'-}}\n{%- endif -%}",
-    'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{- name1 + ': ' + message['content'] + '\\n'-}}\n        {%- else -%}\n            {{- name2 + ': ' + message['content'] + '\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}",
 
     # Generation parameters - Curve shape
     'temperature': 0.6,
@@ -283,6 +271,20 @@
     'sampler_priority': neutral_samplers['sampler_priority'],
     'dry_sequence_breakers': neutral_samplers['dry_sequence_breakers'],
     'grammar_string': '',
+
+    # Character settings
+    'character': 'Assistant',
+    'name1': 'You',
+    'name2': 'AI',
+    'user_bio': '',
+    'context': 'The following is a conversation with an AI Large Language Model. The AI has been trained to answer questions, provide recommendations, and help with decision making. The AI follows user requests. The AI thinks outside the box.',
+    'greeting': 'How can I help you today?',
+    'custom_system_message': '',
+    'instruction_template_str': "{%- set ns = namespace(found=false) -%}\n{%- for message in messages -%}\n    {%- if message['role'] == 'system' -%}\n        {%- set ns.found = true -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if not ns.found -%}\n    {{- '' + 'Below is an instruction that describes a task. Write a response that appropriately completes the request.' + '\\n\\n' -}}\n{%- endif %}\n{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {{- '' + message['content'] + '\\n\\n' -}}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{-'### Instruction:\\n' + message['content'] + '\\n\\n'-}}\n        {%- else -%}\n            {{-'### Response:\\n' + message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{-'### Response:\\n'-}}\n{%- endif -%}",
+    'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{- name1 + ': ' + message['content'] + '\\n'-}}\n        {%- else -%}\n            {{- name2 + ': ' + message['content'] + '\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}",
+
+    # Extensions
+    'default_extensions': [],
 }
 
 default_settings = copy.deepcopy(settings)
diff --git a/modules/ui.py b/modules/ui.py
index 52688128de..2925faa57a 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -343,6 +343,7 @@ def save_settings(state, preset, extensions_list, show_controls, theme_state, ma
             try:
                 with open(settings_path, 'r', encoding='utf-8') as f:
                     existing_settings = yaml.safe_load(f.read()) or {}
+
                 if 'default_extensions' in existing_settings:
                     output['default_extensions'] = existing_settings['default_extensions']
             except Exception:
diff --git a/modules/ui_session.py b/modules/ui_session.py
index 42fcf89d79..0673828ec0 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -10,12 +10,12 @@ def create_ui():
         with gr.Row():
             with gr.Column():
                 gr.Markdown("## Settings")
-                shared.gradio['save_settings'] = gr.Button('Save settings to user_data/settings.yaml', elem_classes='refresh-button', interactive=not mu)
                 shared.gradio['toggle_dark_mode'] = gr.Button('Toggle light/dark theme 💡', elem_classes='refresh-button')
                 shared.gradio['paste_to_attachment'] = gr.Checkbox(label='Turn long pasted text into attachments in the Chat tab', value=shared.settings['paste_to_attachment'], elem_id='paste_to_attachment')
 
             with gr.Column():
                 gr.Markdown("## Extensions & flags")
+                shared.gradio['save_settings'] = gr.Button('Save settings to user_data/settings.yaml', elem_classes='refresh-button', interactive=not mu)
                 shared.gradio['reset_interface'] = gr.Button("Apply flags/extensions and restart", interactive=not mu)
                 with gr.Row():
                     with gr.Column():

From 9d6a7f1bcffd896580b02282821e0104b06f7cd3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 11 Jun 2025 07:55:35 -0700
Subject: [PATCH 0682/1701] Minor changes

---
 server.py | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/server.py b/server.py
index 80cc2f1875..cfb21a6e35 100644
--- a/server.py
+++ b/server.py
@@ -236,14 +236,14 @@ def create_interface():
         settings_file = Path(shared.args.settings)
     elif Path('user_data/settings.yaml').exists():
         settings_file = Path('user_data/settings.yaml')
-    elif Path('user_data/settings.json').exists():
-        settings_file = Path('user_data/settings.json')
 
     if settings_file is not None:
         logger.info(f"Loading settings from \"{settings_file}\"")
-        file_contents = open(settings_file, 'r', encoding='utf-8').read()
-        new_settings = json.loads(file_contents) if settings_file.suffix == "json" else yaml.safe_load(file_contents)
-        shared.settings.update(new_settings)
+        with open(settings_file, 'r', encoding='utf-8') as f:
+            new_settings = yaml.safe_load(f.read())
+
+        if new_settings:
+            shared.settings.update(new_settings)
 
     # Fallback settings for models
     shared.model_config['.*'] = get_fallback_settings()

From b4d2a00e207bacb20bbd1480d1681c96b71dd61a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Jun 2025 08:35:33 -0700
Subject: [PATCH 0683/1701] Update README

---
 README.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index 16b0253919..cfd354df47 100644
--- a/README.md
+++ b/README.md
@@ -6,9 +6,9 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 
 [Try the Deep Reason extension](https://oobabooga.gumroad.com/l/deep_reason)
 
-|![Image1](https://github.com/oobabooga/screenshots/raw/main/AFTER-INSTRUCT.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/AFTER-CHAT.png) |
+|![Image1](https://github.com/oobabooga/screenshots/raw/main/INSTRUCT-3.5.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/CHAT-3.5.png) |
 |:---:|:---:|
-|![Image1](https://github.com/oobabooga/screenshots/raw/main/AFTER-DEFAULT.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/AFTER-PARAMETERS.png) |
+|![Image1](https://github.com/oobabooga/screenshots/raw/main/DEFAULT-3.5.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/PARAMETERS-3.5.png) |
 
 ## Features
 

From 2dee3a66ffeb1778c70f960ade466fcd822949f6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Jun 2025 16:54:43 -0700
Subject: [PATCH 0684/1701] Add an option to include/exclude attachments from
 previous messages in the chat prompt

---
 modules/chat.py       | 4 ++--
 modules/shared.py     | 1 +
 modules/ui.py         | 6 ++++--
 modules/ui_session.py | 1 +
 4 files changed, 8 insertions(+), 4 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index dfc301df49..f080ef65bc 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -217,8 +217,8 @@ def generate_chat_prompt(user_input, state, **kwargs):
             user_key = f"user_{row_idx}"
             enhanced_user_msg = user_msg
 
-            # Add attachment content if present
-            if user_key in metadata and "attachments" in metadata[user_key]:
+            # Add attachment content if present AND if past attachments are enabled
+            if (state.get('include_past_attachments', True) and user_key in metadata and "attachments" in metadata[user_key]):
                 attachments_text = ""
                 for attachment in metadata[user_key]["attachments"]:
                     filename = attachment.get("name", "file")
diff --git a/modules/shared.py b/modules/shared.py
index 83920df860..9efc8daf9f 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -224,6 +224,7 @@
     'negative_prompt': '',
     'dark_theme': True,
     'paste_to_attachment': False,
+    'include_past_attachments': True,
 
     # Generation parameters - Curve shape
     'temperature': 0.6,
diff --git a/modules/ui.py b/modules/ui.py
index 2925faa57a..463e59deec 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -273,7 +273,8 @@ def list_interface_input_elements():
 
     # Other elements
     elements += [
-        'paste_to_attachment'
+        'paste_to_attachment',
+        'include_past_attachments',
     ]
 
     return elements
@@ -497,7 +498,8 @@ def setup_auto_save():
         # Session tab (ui_session.py)
         'show_controls',
         'theme_state',
-        'paste_to_attachment'
+        'paste_to_attachment',
+        'include_past_attachments'
     ]
 
     for element_name in change_elements:
diff --git a/modules/ui_session.py b/modules/ui_session.py
index 0673828ec0..dc7ca2fb09 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -12,6 +12,7 @@ def create_ui():
                 gr.Markdown("## Settings")
                 shared.gradio['toggle_dark_mode'] = gr.Button('Toggle light/dark theme 💡', elem_classes='refresh-button')
                 shared.gradio['paste_to_attachment'] = gr.Checkbox(label='Turn long pasted text into attachments in the Chat tab', value=shared.settings['paste_to_attachment'], elem_id='paste_to_attachment')
+                shared.gradio['include_past_attachments'] = gr.Checkbox(label='Include attachments/search results from previous messages in the chat prompt', value=shared.settings['include_past_attachments'])
 
             with gr.Column():
                 gr.Markdown("## Extensions & flags")

From a25a1fc8d031951ffe241c12ad8396ddd24d7289 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Jun 2025 22:01:02 -0700
Subject: [PATCH 0685/1701] Disable message action icons during streaming for
 better performance

---
 css/main.css | 19 ++++++++++++++-----
 1 file changed, 14 insertions(+), 5 deletions(-)

diff --git a/css/main.css b/css/main.css
index a22fdd950f..6d02582e29 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1289,6 +1289,20 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     opacity: 1;
 }
 
+/* Disable message action hover effects during generation */
+._generating .message:hover .message-actions,
+._generating .user-message:hover .message-actions,
+._generating .assistant-message:hover .message-actions {
+    opacity: 0 !important;
+}
+
+/* Disable message action hover effects during scrolling */
+.scrolling .message:hover .message-actions,
+.scrolling .user-message:hover .message-actions,
+.scrolling .assistant-message:hover .message-actions {
+    opacity: 0 !important;
+}
+
 .footer-button svg {
     stroke: rgb(156 163 175);
     transition: stroke 0.2s;
@@ -1624,8 +1638,3 @@ button:focus {
     justify-content: center;
     display: none;
 }
-
-/* Disable hover effects while scrolling */
-.chat-parent.scrolling * {
-    pointer-events: none !important;
-}

From f337767f36c60817d3d9f76417e7a1687ff2c564 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Jun 2025 22:17:39 -0700
Subject: [PATCH 0686/1701] Add error handling for non-llama.cpp models in
 portable mode

---
 modules/models_settings.py | 11 ++++++++++-
 modules/ui_model_menu.py   |  7 ++++++-
 2 files changed, 16 insertions(+), 2 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 283a974477..15985608ae 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -9,6 +9,7 @@
 import yaml
 
 from modules import chat, loaders, metadata_gguf, shared, ui
+from modules.logging_colors import logger
 
 
 def get_fallback_settings():
@@ -56,7 +57,13 @@ def get_model_metadata(model):
         if path.is_file():
             model_file = path
         else:
-            model_file = list(path.glob('*.gguf'))[0]
+            gguf_files = list(path.glob('*.gguf'))
+            if not gguf_files:
+                error_msg = f"No .gguf files found in directory: {path}"
+                logger.error(error_msg)
+                raise FileNotFoundError(error_msg)
+
+            model_file = gguf_files[0]
 
         metadata = load_gguf_metadata_with_cache(model_file)
 
@@ -171,6 +178,8 @@ def infer_loader(model_name, model_settings, hf_quant_method=None):
     path_to_model = Path(f'{shared.args.model_dir}/{model_name}')
     if not path_to_model.exists():
         loader = None
+    elif shared.args.portable:
+        loader = 'llama.cpp'
     elif len(list(path_to_model.glob('*.gguf'))) > 0:
         loader = 'llama.cpp'
     elif re.match(r'.*\.gguf', model_name.lower()):
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 9e982f0ec4..a15bba2b1a 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -174,7 +174,12 @@ def create_event_handlers():
 
 
 def load_model_wrapper(selected_model, loader, autoload=False):
-    settings = get_model_metadata(selected_model)
+    try:
+        settings = get_model_metadata(selected_model)
+    except FileNotFoundError:
+        exc = traceback.format_exc()
+        yield exc.replace('\n', '\n\n')
+        return
 
     if not autoload:
         yield "### {}\n\n- Settings updated: Click \"Load\" to load the model\n- Max sequence length: {}".format(selected_model, settings['truncation_length_info'])

From f4f621b215d16bce2fd1b0696e0271bee0d69568 Mon Sep 17 00:00:00 2001
From: Miriam 
Date: Thu, 12 Jun 2025 22:56:33 -0700
Subject: [PATCH 0687/1701] ensure estimated vram is updated when switching
 between different models (#7071)

---
 modules/ui_model_menu.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index a15bba2b1a..6b1062035d 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -135,7 +135,7 @@ def create_event_handlers():
     # with the model defaults (if any), and then the model is loaded
     shared.gradio['model_menu'].change(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        handle_load_model_event_initial, gradio('model_menu', 'interface_state'), gradio(ui.list_interface_input_elements()) + gradio('interface_state'), show_progress=False).then(
+        handle_load_model_event_initial, gradio('model_menu', 'interface_state'), gradio(ui.list_interface_input_elements()) + gradio('interface_state') + gradio('vram_info'), show_progress=False).then(
         partial(load_model_wrapper, autoload=False), gradio('model_menu', 'loader'), gradio('model_status'), show_progress=True).success(
         handle_load_model_event_final, gradio('truncation_length', 'loader', 'interface_state'), gradio('truncation_length', 'filter_by_loader'), show_progress=False)
 
@@ -379,7 +379,8 @@ def handle_load_model_event_initial(model, state):
     output = ui.apply_interface_values(state)
     update_model_parameters(state)  # This updates the command-line flags
 
-    return output + [state]
+    vram_info = state.get('vram_info', "
            Estimated VRAM to load the model:
            ")
+    return output + [state] + [vram_info]
 
 
 def handle_load_model_event_final(truncation_length, loader, state):

From b58e80cb99435e28f88932c0705f2e7eb5fee215 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Jun 2025 01:02:11 -0700
Subject: [PATCH 0688/1701] Update README

---
 README.md | 30 ++++++++++++++++++------------
 1 file changed, 18 insertions(+), 12 deletions(-)

diff --git a/README.md b/README.md
index cfd354df47..1163b19018 100644
--- a/README.md
+++ b/README.md
@@ -12,7 +12,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 
 ## Features
 
-- Supports multiple text generation backends in one UI/API, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).
+- Supports multiple local text generation backends, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).
 - Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory.
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
@@ -38,20 +38,26 @@ Download from: https://github.com/oobabooga/text-generation-webui/releases
 
 #### Option 2: One-click installer
 
-1) Clone or [download the repository](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip).
-2) Run the script that matches your OS: `start_linux.sh`, `start_windows.bat`, or `start_macos.sh`.
-3) Select your GPU vendor when asked.
-4) After installation completes, browse to `http://localhost:7860`.
-5) Have fun!
+1. Clone the repository, or [download its source code](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip) and extract it.
+2. Run the script for your operating system:
 
-To restart the web UI later, just run the same `start_` script. If you need to reinstall, delete the `installer_files` folder created during setup and run the script again.
+   * `start_windows.bat`
+   * `start_linux.sh`
+   * `start_macos.sh`
+3. When prompted, select your GPU vendor.
+4. Once setup is complete, open your browser and go to: `http://127.0.0.1:7860`
 
-You can use command-line flags, like `./start_linux.sh --help`, or add them to `user_data/CMD_FLAGS.txt` (such as `--api` to enable API use). To update the project, run `update_wizard_linux.sh`, `update_wizard_windows.bat`, or `update_wizard_macos.sh`.
+To **restart** the web UI later, run the same `start_` script.
 
-
            
-
-Setup details and information about installing manually
-
+To **reinstall**, delete the `installer_files` folder created during setup and run the script again.
+
+You can pass **command-line flags** directly (e.g., `./start_linux.sh --help`), or add them to `user_data/CMD_FLAGS.txt` (e.g., `--api` to enable the API).
+
+To **update** the project, run the appropriate update script:
+
+* `update_wizard_windows.bat`
+* `update_wizard_linux.sh`
+* `update_wizard_macos.sh`
 
 ### One-click-installer
 

From 5ba52967ac5613e311292ce89f23f23fe5932057 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Jun 2025 01:04:41 -0700
Subject: [PATCH 0689/1701] Update README

---
 README.md | 21 +++++++--------------
 1 file changed, 7 insertions(+), 14 deletions(-)

diff --git a/README.md b/README.md
index 1163b19018..ac3eac4b8a 100644
--- a/README.md
+++ b/README.md
@@ -39,25 +39,18 @@ Download from: https://github.com/oobabooga/text-generation-webui/releases
 #### Option 2: One-click installer
 
 1. Clone the repository, or [download its source code](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip) and extract it.
-2. Run the script for your operating system:
-
-   * `start_windows.bat`
-   * `start_linux.sh`
-   * `start_macos.sh`
+2. Run the startup script for your OS: `start_windows.bat`, `start_linux.sh`, or `start_macos.sh`.
 3. When prompted, select your GPU vendor.
-4. Once setup is complete, open your browser and go to: `http://127.0.0.1:7860`
-
-To **restart** the web UI later, run the same `start_` script.
+4. After installation, open your browser to `http://127.0.0.1:7860`.
 
-To **reinstall**, delete the `installer_files` folder created during setup and run the script again.
+To restart the web UI later, run the same `start_` script.
 
-You can pass **command-line flags** directly (e.g., `./start_linux.sh --help`), or add them to `user_data/CMD_FLAGS.txt` (e.g., `--api` to enable the API).
+To reinstall, delete the `installer_files` folder and run the script again.
 
-To **update** the project, run the appropriate update script:
+You can pass command-line flags directly (e.g., `./start_linux.sh --help`), or add them to `user_data/CMD_FLAGS.txt` (e.g., `--api` to enable the API).
 
-* `update_wizard_windows.bat`
-* `update_wizard_linux.sh`
-* `update_wizard_macos.sh`
+To update, run one of the following scripts, depending on your OS:
+`update_wizard_windows.bat`, `update_wizard_linux.sh`, or `update_wizard_macos.sh`.
 
 ### One-click-installer
 

From aab28398efee8e18b2c7d487faf60ab7af559dcc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Jun 2025 01:06:44 -0700
Subject: [PATCH 0690/1701] Update README

---
 README.md | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index ac3eac4b8a..35830589e7 100644
--- a/README.md
+++ b/README.md
@@ -49,8 +49,12 @@ To reinstall, delete the `installer_files` folder and run the script again.
 
 You can pass command-line flags directly (e.g., `./start_linux.sh --help`), or add them to `user_data/CMD_FLAGS.txt` (e.g., `--api` to enable the API).
 
-To update, run one of the following scripts, depending on your OS:
-`update_wizard_windows.bat`, `update_wizard_linux.sh`, or `update_wizard_macos.sh`.
+To update, run the update script for your OS: `update_wizard_windows.bat`, `update_wizard_linux.sh`, or `update_wizard_macos.sh`.
+
+
            
+
+Setup details and information about installing manually
+
 
 ### One-click-installer
 

From 7cb650237cdd2082722a2f2157d80378cf16ee75 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Jun 2025 01:12:52 -0700
Subject: [PATCH 0691/1701] Update the README

---
 README.md | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index 35830589e7..43134bfce9 100644
--- a/README.md
+++ b/README.md
@@ -45,7 +45,7 @@ Download from: https://github.com/oobabooga/text-generation-webui/releases
 
 To restart the web UI later, run the same `start_` script.
 
-To reinstall, delete the `installer_files` folder and run the script again.
+To reinstall with a fresh Python environment, delete the `installer_files` folder and run the `start_` script again.
 
 You can pass command-line flags directly (e.g., `./start_linux.sh --help`), or add them to `user_data/CMD_FLAGS.txt` (e.g., `--api` to enable the API).
 
@@ -328,7 +328,11 @@ https://github.com/oobabooga/text-generation-webui/wiki
 
 ## Downloading models
 
-Models should be placed in the folder `text-generation-webui/user_data/models`. They are usually downloaded from [Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads).
+Models should be placed in the folder `text-generation-webui/user_data/models`. They are usually downloaded from [Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads&search=gguf).
+
+To check if a GGUF model will fit in your hardware before downloading it, you can use this tool I created:
+
+[Accurate GGUF VRAM Calculator](https://huggingface.co/spaces/oobabooga/accurate-gguf-vram-calculator)
 
 * GGUF models are a single file and should be placed directly into `user_data/models`. Example:
 

From 322cd28e24dc772827b4bcbf5c25cdaa7f4f4f62 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Jun 2025 01:27:33 -0700
Subject: [PATCH 0692/1701] Update README

---
 README.md | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/README.md b/README.md
index 43134bfce9..3b5008fdd4 100644
--- a/README.md
+++ b/README.md
@@ -322,10 +322,6 @@ API:
 
 
            
 
-## Documentation
-
-https://github.com/oobabooga/text-generation-webui/wiki
-
 ## Downloading models
 
 Models should be placed in the folder `text-generation-webui/user_data/models`. They are usually downloaded from [Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads&search=gguf).
@@ -373,6 +369,10 @@ python download-model.py organization/model
 
 Run `python download-model.py --help` to see all the options.
 
+## Documentation
+
+https://github.com/oobabooga/text-generation-webui/wiki
+
 ## Google Colab notebook
 
 https://colab.research.google.com/github/oobabooga/text-generation-webui/blob/main/Colab-TextGen-GPU.ipynb

From 9a2353f97bf18e597358747ca4853a5128bc97b1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Jun 2025 05:44:02 -0700
Subject: [PATCH 0693/1701] Better log message when the user input gets
 truncated

---
 modules/chat.py | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index f080ef65bc..d9ac652c49 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -353,7 +353,17 @@ def make_prompt(messages):
                     logger.error(f"Failed to build the chat prompt. The input is too long for the available context length.\n\nTruncation length: {state['truncation_length']}\nmax_new_tokens: {state['max_new_tokens']} (is it too high?)\nAvailable context length: {max_length}\n")
                     raise ValueError
                 else:
-                    logger.warning(f"The input has been truncated. Context length: {state['truncation_length']}, max_new_tokens: {state['max_new_tokens']}, available context length: {max_length}.")
+                    # Calculate token counts for the log message
+                    original_user_tokens = get_encoded_length(user_message)
+                    truncated_user_tokens = get_encoded_length(user_message[:left])
+                    total_context = max_length + state['max_new_tokens']
+
+                    logger.warning(
+                        f"User message truncated from {original_user_tokens} to {truncated_user_tokens} tokens. "
+                        f"Context full: {max_length} input tokens ({total_context} total, {state['max_new_tokens']} for output). "
+                        f"Increase ctx-size while loading the model to avoid truncation."
+                    )
+
                     break
 
             prompt = make_prompt(messages)

From dfab11f0b5f2866c1003ab2dd54797021e44e102 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Jun 2025 07:45:42 -0700
Subject: [PATCH 0694/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 3b5008fdd4..45ab48eb38 100644
--- a/README.md
+++ b/README.md
@@ -41,7 +41,7 @@ Download from: https://github.com/oobabooga/text-generation-webui/releases
 1. Clone the repository, or [download its source code](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip) and extract it.
 2. Run the startup script for your OS: `start_windows.bat`, `start_linux.sh`, or `start_macos.sh`.
 3. When prompted, select your GPU vendor.
-4. After installation, open your browser to `http://127.0.0.1:7860`.
+4. After installation, open `http://127.0.0.1:7860` in your browser.
 
 To restart the web UI later, run the same `start_` script.
 

From 09eb326486a9dfa874851fb17a193201e20b0e77 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Jun 2025 07:46:43 -0700
Subject: [PATCH 0695/1701] Merge README.md changes from dev branch

---
 README.md | 31 +++++++++++++++++++------------
 1 file changed, 19 insertions(+), 12 deletions(-)

diff --git a/README.md b/README.md
index cfd354df47..45ab48eb38 100644
--- a/README.md
+++ b/README.md
@@ -12,7 +12,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 
 ## Features
 
-- Supports multiple text generation backends in one UI/API, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).
+- Supports multiple local text generation backends, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).
 - Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory.
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
@@ -38,15 +38,18 @@ Download from: https://github.com/oobabooga/text-generation-webui/releases
 
 #### Option 2: One-click installer
 
-1) Clone or [download the repository](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip).
-2) Run the script that matches your OS: `start_linux.sh`, `start_windows.bat`, or `start_macos.sh`.
-3) Select your GPU vendor when asked.
-4) After installation completes, browse to `http://localhost:7860`.
-5) Have fun!
+1. Clone the repository, or [download its source code](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip) and extract it.
+2. Run the startup script for your OS: `start_windows.bat`, `start_linux.sh`, or `start_macos.sh`.
+3. When prompted, select your GPU vendor.
+4. After installation, open `http://127.0.0.1:7860` in your browser.
 
-To restart the web UI later, just run the same `start_` script. If you need to reinstall, delete the `installer_files` folder created during setup and run the script again.
+To restart the web UI later, run the same `start_` script.
 
-You can use command-line flags, like `./start_linux.sh --help`, or add them to `user_data/CMD_FLAGS.txt` (such as `--api` to enable API use). To update the project, run `update_wizard_linux.sh`, `update_wizard_windows.bat`, or `update_wizard_macos.sh`.
+To reinstall with a fresh Python environment, delete the `installer_files` folder and run the `start_` script again.
+
+You can pass command-line flags directly (e.g., `./start_linux.sh --help`), or add them to `user_data/CMD_FLAGS.txt` (e.g., `--api` to enable the API).
+
+To update, run the update script for your OS: `update_wizard_windows.bat`, `update_wizard_linux.sh`, or `update_wizard_macos.sh`.
 
 
            
 
@@ -319,13 +322,13 @@ API:
 
 
            
 
-## Documentation
+## Downloading models
 
-https://github.com/oobabooga/text-generation-webui/wiki
+Models should be placed in the folder `text-generation-webui/user_data/models`. They are usually downloaded from [Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads&search=gguf).
 
-## Downloading models
+To check if a GGUF model will fit in your hardware before downloading it, you can use this tool I created:
 
-Models should be placed in the folder `text-generation-webui/user_data/models`. They are usually downloaded from [Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads).
+[Accurate GGUF VRAM Calculator](https://huggingface.co/spaces/oobabooga/accurate-gguf-vram-calculator)
 
 * GGUF models are a single file and should be placed directly into `user_data/models`. Example:
 
@@ -366,6 +369,10 @@ python download-model.py organization/model
 
 Run `python download-model.py --help` to see all the options.
 
+## Documentation
+
+https://github.com/oobabooga/text-generation-webui/wiki
+
 ## Google Colab notebook
 
 https://colab.research.google.com/github/oobabooga/text-generation-webui/blob/main/Colab-TextGen-GPU.ipynb

From 879fa3d8c43ad1fab8bbb407a5ab6c8aa44aefc0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Jun 2025 07:14:22 -0700
Subject: [PATCH 0696/1701] Improve the wpp style & simplify the code

---
 css/chat_style-wpp.css    | 96 +++++++++++++++++++++++++++++----------
 modules/html_generator.py | 60 ------------------------
 2 files changed, 71 insertions(+), 85 deletions(-)

diff --git a/css/chat_style-wpp.css b/css/chat_style-wpp.css
index 353201c228..8ff82c6c98 100644
--- a/css/chat_style-wpp.css
+++ b/css/chat_style-wpp.css
@@ -1,57 +1,103 @@
 .message {
-    padding-bottom: 22px;
-    padding-top: 3px;
+    display: block;
+    padding-top: 0;
+    padding-bottom: 21px;
     font-size: 15px;
     font-family: 'Noto Sans', Helvetica, Arial, sans-serif;
     line-height: 1.428571429;
+    grid-template-columns: none;
 }
 
-.text-you {
-    background-color: #d9fdd3;
-    border-radius: 15px;
-    padding: 10px;
-    padding-top: 5px;
+.circle-you, .circle-bot {
+    display: none;
+}
+
+.text {
+    max-width: 65%;
+    border-radius: 18px;
+    padding: 12px 16px;
+    margin-bottom: 8px;
+    clear: both;
+    box-shadow: 0 1px 2px rgb(0 0 0 / 10%);
+}
+
+.username {
+    font-weight: 600;
+    margin-bottom: 8px;
+    opacity: 0.65;
+    padding-left: 4px;
+}
+
+/* User messages - right aligned, WhatsApp green */
+.circle-you + .text {
+    background-color: #dcf8c6;
     float: right;
+    margin-left: auto;
+    margin-right: 8px;
+}
+
+.circle-you + .text .username {
+    display: none;
+}
+
+/* Bot messages - left aligned, white */
+.circle-bot + .text {
+    background-color: #fff;
+    float: left;
+    margin-right: auto;
+    margin-left: 8px;
+    border: 1px solid #e5e5e5;
 }
 
-.text-bot {
-    background-color: #f2f2f2;
-    border-radius: 15px;
-    padding: 10px;
-    padding-top: 5px;
+.circle-bot + .text .message-actions {
+    bottom: -25px !important;
 }
 
-.dark .text-you {
-    background-color: #005c4b;
-    color: #111b21;
+/* Dark theme colors */
+.dark .circle-you + .text {
+    background-color: #056162;
+    color: #e4e6ea;
+    box-shadow: 0 1px 2px rgb(0 0 0 / 30%);
 }
 
-.dark .text-bot {
-    background-color: #1f2937;
-    color: #111b21;
+.dark .circle-bot + .text {
+    background-color: #202c33;
+    color: #e4e6ea;
+    border: 1px solid #3c4043;
+    box-shadow: 0 1px 2px rgb(0 0 0 / 30%);
 }
 
-.text-bot p, .text-you p {
-    margin-top: 5px;
+.dark .username {
+    opacity: 0.7;
 }
 
 .message-body img {
     max-width: 300px;
     max-height: 300px;
-    border-radius: 20px;
+    border-radius: 12px;
 }
 
 .message-body p {
     margin-bottom: 0 !important;
+    margin-top: 2px !important;
     font-size: 15px !important;
-    line-height: 1.428571429 !important;
-    font-weight: 500;
+    line-height: 1.4 !important;
+    font-weight: 400;
+}
+
+.message-body p:first-child {
+    margin-top: 0 !important;
 }
 
 .dark .message-body p em {
-    color: rgb(138 138 138) !important;
+    color: rgb(170 170 170) !important;
 }
 
 .message-body p em {
-    color: rgb(110 110 110) !important;
+    color: rgb(100 100 100) !important;
+}
+
+/* Message actions positioning */
+.message-actions {
+    margin-top: 8px;
 }
diff --git a/modules/html_generator.py b/modules/html_generator.py
index af64894e4d..11572fc6c0 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -595,64 +595,6 @@ def create_message(role, content, raw_content):
     return output
 
 
-def generate_chat_html(history, name1, name2, reset_cache=False, last_message_only=False):
-    if not last_message_only:
-        output = f'
            '
-    else:
-        output = ""
-
-    def create_message(role, content, raw_content):
-        """Inner function for WPP-style messages."""
-        text_class = "text-you" if role == "user" else "text-bot"
-
-        # Get role-specific data
-        timestamp = format_message_timestamp(history, role, i)
-        attachments = format_message_attachments(history, role, i)
-
-        # Create info button if timestamp exists
-        info_message = ""
-        if timestamp:
-            tooltip_text = get_message_tooltip(history, role, i)
-            info_message = info_button.replace('title="message"', f'title="{html.escape(tooltip_text)}"')
-
-        return (
-            f'
            '
-            f'
            '
-            f'
            {content}
            '
-            f'{attachments}'
-            f'{actions_html(history, i, role, info_message)}'
-            f'
            '
-            f'
            '
-        )
-
-    # Determine range
-    start_idx = len(history['visible']) - 1 if last_message_only else 0
-    end_idx = len(history['visible'])
-
-    for i in range(start_idx, end_idx):
-        row_visible = history['visible'][i]
-        row_internal = history['internal'][i]
-
-        # Convert content
-        if last_message_only:
-            converted_visible = [None, convert_to_markdown_wrapped(row_visible[1], message_id=i, use_cache=i != len(history['visible']) - 1)]
-        else:
-            converted_visible = [convert_to_markdown_wrapped(entry, message_id=i, use_cache=i != len(history['visible']) - 1) for entry in row_visible]
-
-        # Generate messages
-        if not last_message_only and converted_visible[0]:
-            output += create_message("user", converted_visible[0], row_internal[0])
-
-        output += create_message("assistant", converted_visible[1], row_internal[1])
-
-    if not last_message_only:
-        output += "
            "
-
-    return output
-
-
 def time_greeting():
     current_hour = datetime.datetime.now().hour
     if 5 <= current_hour < 12:
@@ -669,8 +611,6 @@ def chat_html_wrapper(history, name1, name2, mode, style, character, reset_cache
         result = f'
            {greeting}
            '
     elif mode == 'instruct':
         result = generate_instruct_html(history, last_message_only=last_message_only)
-    elif style == 'wpp':
-        result = generate_chat_html(history, name1, name2, last_message_only=last_message_only)
     else:
         result = generate_cai_chat_html(history, name1, name2, style, character, reset_cache=reset_cache, last_message_only=last_message_only)
 

From d2da40b0e4584bf11e6ab9f4c8f32847cf835940 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Jun 2025 08:23:57 -0700
Subject: [PATCH 0697/1701] Remember the last selected chat for each
 mode/character

---
 modules/chat.py | 70 ++++++++++++++++++++++++++++++++++++++++++-------
 1 file changed, 60 insertions(+), 10 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index d9ac652c49..e7cd5ad22a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1079,16 +1079,27 @@ def load_latest_history(state):
     '''
 
     if shared.args.multi_user:
-        return start_new_chat(state)
+        return start_new_chat(state), None
 
     histories = find_all_histories(state)
 
     if len(histories) > 0:
-        history = load_history(histories[0], state['character_menu'], state['mode'])
-    else:
-        history = start_new_chat(state)
+        # Try to load the last visited chat for this character/mode
+        chat_state = load_last_chat_state()
+        key = get_chat_state_key(state['character_menu'], state['mode'])
+        last_chat_id = chat_state.get("last_chats", {}).get(key)
+
+        # If we have a stored last chat and it still exists, use it
+        if last_chat_id and last_chat_id in histories:
+            unique_id = last_chat_id
+        else:
+            # Fall back to most recent (current behavior)
+            unique_id = histories[0]
 
-    return history
+        history = load_history(unique_id, state['character_menu'], state['mode'])
+        return history, unique_id
+    else:
+        return start_new_chat(state), None
 
 
 def load_history_after_deletion(state, idx):
@@ -1120,6 +1131,42 @@ def update_character_menu_after_deletion(idx):
     return gr.update(choices=characters, value=characters[idx])
 
 
+def get_chat_state_key(character, mode):
+    """Generate a key for storing last chat state"""
+    if mode == 'instruct':
+        return 'instruct'
+    else:
+        return f"chat_{character}"
+
+
+def load_last_chat_state():
+    """Load the last chat state from file"""
+    state_file = Path('user_data/logs/chat_state.json')
+    if state_file.exists():
+        try:
+            with open(state_file, 'r', encoding='utf-8') as f:
+                return json.loads(f.read())
+        except:
+            pass
+
+    return {"last_chats": {}}
+
+
+def save_last_chat_state(character, mode, unique_id):
+    """Save the last visited chat for a character/mode"""
+    if shared.args.multi_user:
+        return
+
+    state = load_last_chat_state()
+    key = get_chat_state_key(character, mode)
+    state["last_chats"][key] = unique_id
+
+    state_file = Path('user_data/logs/chat_state.json')
+    state_file.parent.mkdir(exist_ok=True)
+    with open(state_file, 'w', encoding='utf-8') as f:
+        f.write(json.dumps(state, indent=2))
+
+
 def load_history(unique_id, character, mode):
     p = get_history_file_path(unique_id, character, mode)
 
@@ -1553,6 +1600,9 @@ def handle_unique_id_select(state):
     history = load_history(state['unique_id'], state['character_menu'], state['mode'])
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
 
+    # Save this as the last visited chat
+    save_last_chat_state(state['character_menu'], state['mode'], state['unique_id'])
+
     convert_to_markdown.cache_clear()
 
     return [history, html]
@@ -1753,14 +1803,14 @@ def handle_character_menu_change(state):
     state['greeting'] = greeting
     state['context'] = context
 
-    history = load_latest_history(state)
+    history, loaded_unique_id = load_latest_history(state)
     histories = find_all_histories_with_first_prompts(state)
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
 
     convert_to_markdown.cache_clear()
 
     if len(histories) > 0:
-        past_chats_update = gr.update(choices=histories, value=histories[0][1])
+        past_chats_update = gr.update(choices=histories, value=loaded_unique_id or histories[0][1])
     else:
         past_chats_update = gr.update(choices=histories)
 
@@ -1772,7 +1822,7 @@ def handle_character_menu_change(state):
         picture,
         greeting,
         context,
-        past_chats_update,
+        past_chats_update
     ]
 
 
@@ -1796,14 +1846,14 @@ def handle_character_picture_change(picture):
 
 
 def handle_mode_change(state):
-    history = load_latest_history(state)
+    history, loaded_unique_id = load_latest_history(state)
     histories = find_all_histories_with_first_prompts(state)
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
 
     convert_to_markdown.cache_clear()
 
     if len(histories) > 0:
-        past_chats_update = gr.update(choices=histories, value=histories[0][1])
+        past_chats_update = gr.update(choices=histories, value=loaded_unique_id or histories[0][1])
     else:
         past_chats_update = gr.update(choices=histories)
 

From 1d2315983712e62d535dcca2d9f44b725c21d99e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Jun 2025 08:38:20 -0700
Subject: [PATCH 0698/1701] Increase the size of the enlarged character profile
 picture

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 6d02582e29..3e02c6529f 100644
--- a/css/main.css
+++ b/css/main.css
@@ -925,7 +925,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     position: fixed;
     bottom: 0;
     left: 0;
-    width: calc(100vw / 2 - 600px);
+    width: calc(0.5 * (100vw - min(100vw, 48rem) - (120px - var(--header-width))));
     z-index: 10000;
 }
 

From 8e0ef5b419e7fe15790af3494f030a9bf40058a2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Jun 2025 09:09:46 -0700
Subject: [PATCH 0699/1701] Hide the header bar on Ctrl+S

---
 js/show_controls.js | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/js/show_controls.js b/js/show_controls.js
index 1a87b52d96..06387211ab 100644
--- a/js/show_controls.js
+++ b/js/show_controls.js
@@ -4,6 +4,8 @@ const belowChatInput = document.querySelectorAll(
 const chatParent = document.querySelector(".chat-parent");
 
 function toggle_controls(value) {
+  const headerBar = document.querySelector(".header_bar");
+
   if (value) {
     belowChatInput.forEach(element => {
       element.style.display = "inherit";
@@ -19,6 +21,11 @@ function toggle_controls(value) {
       gallery_element.style.display = "block";
     }
 
+    // Show header bar
+    if (headerBar) {
+      headerBar.style.removeProperty("display");
+    }
+
   } else {
     belowChatInput.forEach(element => {
       element.style.display = "none";
@@ -28,5 +35,10 @@ function toggle_controls(value) {
     document.getElementById("chat-input-row").classList.add("bigchat");
     document.getElementById("chat-col").classList.add("bigchat");
     document.getElementById("chat-tab").style.paddingBottom = "0px";
+
+    // Hide header bar
+    if (headerBar) {
+      headerBar.style.setProperty("display", "none", "important");
+    }
   }
 }

From 8e9c0287aa4c815007a392f37de5d496c151ca70 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Jun 2025 10:12:11 -0700
Subject: [PATCH 0700/1701] UI: Fix edge case where gpu-layers slider maximum
 is incorrectly limited

---
 modules/models_settings.py | 17 +++++------------
 1 file changed, 5 insertions(+), 12 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 15985608ae..3d98568aaf 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -459,26 +459,19 @@ def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type,
         else:
             return (0, gpu_layers) if auto_adjust else 0
 
+    # Get model settings including user preferences
+    model_settings = get_model_metadata(model)
+
     current_layers = gpu_layers
-    max_layers = gpu_layers
+    max_layers = model_settings.get('max_gpu_layers', 256)
 
     if auto_adjust:
-        # Get model settings including user preferences
-        model_settings = get_model_metadata(model)
-
-        # Get the true maximum layers
-        max_layers = model_settings.get('max_gpu_layers', model_settings.get('gpu_layers', gpu_layers))
-
         # Check if this is a user-saved setting
         user_config = shared.user_config
         model_regex = Path(model).name + '$'
         has_user_setting = model_regex in user_config and 'gpu_layers' in user_config[model_regex]
 
-        if has_user_setting:
-            # For user settings, just use the current value (which already has user pref)
-            # but ensure the slider maximum is correct
-            current_layers = gpu_layers  # Already has user setting
-        else:
+        if not has_user_setting:
             # No user setting, auto-adjust from the maximum
             current_layers = max_layers  # Start from max
 

From 1c1cf09a59fe5389961b457eb6173ec2f6bc0168 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Jun 2025 14:52:49 -0700
Subject: [PATCH 0701/1701] Update workflows

---
 .../workflows/build-portable-release-cuda.yml | 19 +++++++++++--------
 .../build-portable-release-vulkan.yml         | 19 +++++++++++--------
 .github/workflows/build-portable-release.yml  | 17 +++++++++++------
 3 files changed, 33 insertions(+), 22 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index 571cbac0c5..a63ec16896 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -101,7 +101,13 @@ jobs:
       - name: Build Package
         shell: bash
         run: |
-            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker
+            VERSION_CLEAN="${{ inputs.version }}"
+            VERSION_CLEAN="${VERSION_CLEAN#v}"
+            cd ..
+            cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
+            cd "text-generation-webui-${VERSION_CLEAN}"
+
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg
             allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
@@ -116,14 +122,12 @@ jobs:
                 PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-pc-windows-msvc-install_only.tar.gz"
                 PIP_PATH="portable_env/python.exe -m pip"
                 PACKAGES_PATH="portable_env/Lib/site-packages"
-                ZIP_CMD="powershell -Command \"Compress-Archive -Path text-generation-webui -DestinationPath"
                 rm start_linux.sh start_macos.sh
             else
                 PLATFORM="linux"
                 PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-unknown-linux-gnu-install_only.tar.gz"
                 PIP_PATH="portable_env/bin/python -m pip"
                 PACKAGES_PATH="portable_env/lib/python3.11/site-packages"
-                ZIP_CMD="zip -r"
                 rm start_macos.sh start_windows.bat
             fi
 
@@ -132,7 +136,7 @@ jobs:
             echo "Downloading Python for $PLATFORM..."
             curl -L -o python-build.tar.gz "$PYTHON_URL"
             tar -xzf python-build.tar.gz
-            mv python text-generation-webui/portable_env
+            mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
 
             # 3. Prepare requirements file based on AVX and CUDA
             if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
@@ -142,7 +146,7 @@ jobs:
             fi
 
             # Create CUDA-specific requirements file if needed
-            cd text-generation-webui
+            cd "text-generation-webui-${VERSION_CLEAN}"
             if [[ "$CUDA_VERSION" == "11.7" ]]; then
                 echo "Creating CUDA 11.7 specific requirements file"
                 sed 's/cu124/cu117/g' "$BASE_REQ_FILE" > requirements_cuda_temp.txt
@@ -162,14 +166,13 @@ jobs:
 
             # 6. Create ZIP file
             cd ..
-            VERSION_CLEAN="${VERSION#v}"
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui -DestinationPath $ZIP_NAME"
+                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
             else
-                zip -r "$ZIP_NAME" text-generation-webui
+                zip -r "$ZIP_NAME" "text-generation-webui-${VERSION_CLEAN}"
             fi
 
       - name: Upload files to a GitHub release
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 4e88d4d9ab..b6217164fb 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -100,7 +100,13 @@ jobs:
       - name: Build Package
         shell: bash
         run: |
-            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker
+            VERSION_CLEAN="${{ inputs.version }}"
+            VERSION_CLEAN="${VERSION_CLEAN#v}"
+            cd ..
+            cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
+            cd "text-generation-webui-${VERSION_CLEAN}"
+
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg
             allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
@@ -114,14 +120,12 @@ jobs:
                 PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-pc-windows-msvc-install_only.tar.gz"
                 PIP_PATH="portable_env/python.exe -m pip"
                 PACKAGES_PATH="portable_env/Lib/site-packages"
-                ZIP_CMD="powershell -Command \"Compress-Archive -Path text-generation-webui -DestinationPath"
                 rm start_linux.sh start_macos.sh
             else
                 PLATFORM="linux"
                 PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-unknown-linux-gnu-install_only.tar.gz"
                 PIP_PATH="portable_env/bin/python -m pip"
                 PACKAGES_PATH="portable_env/lib/python3.11/site-packages"
-                ZIP_CMD="zip -r"
                 rm start_macos.sh start_windows.bat
             fi
 
@@ -130,7 +134,7 @@ jobs:
             echo "Downloading Python for $PLATFORM..."
             curl -L -o python-build.tar.gz "$PYTHON_URL"
             tar -xzf python-build.tar.gz
-            mv python text-generation-webui/portable_env
+            mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
 
             # 3. Prepare requirements file based on AVX
             if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
@@ -140,7 +144,7 @@ jobs:
             fi
             REQ_FILE="$BASE_REQ_FILE"
 
-            cd text-generation-webui
+            cd "text-generation-webui-${VERSION_CLEAN}"
 
             # 4. Install packages
             echo "Installing Python packages from $REQ_FILE..."
@@ -148,14 +152,13 @@ jobs:
 
             # 6. Create ZIP file
             cd ..
-            VERSION_CLEAN="${VERSION#v}"
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui -DestinationPath $ZIP_NAME"
+                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
             else
-                zip -r "$ZIP_NAME" text-generation-webui
+                zip -r "$ZIP_NAME" "text-generation-webui-${VERSION_CLEAN}"
             fi
 
       - name: Upload files to a GitHub release
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 6910ce2c3b..dd86f522dd 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -100,7 +100,13 @@ jobs:
       - name: Build Package
         shell: bash
         run: |
-            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker
+            VERSION_CLEAN="${{ inputs.version }}"
+            VERSION_CLEAN="${VERSION_CLEAN#v}"
+            cd ..
+            cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
+            cd "text-generation-webui-${VERSION_CLEAN}"
+
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg
             allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
@@ -143,10 +149,10 @@ jobs:
             cd ..
             curl -L -o python-build.tar.gz "$PYTHON_URL"
             tar -xzf python-build.tar.gz
-            mv python text-generation-webui/portable_env
+            mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
 
             # 3. Prepare requirements file based on platform and AVX
-            cd text-generation-webui
+            cd "text-generation-webui-${VERSION_CLEAN}"
 
             # Select requirements file based on platform
             if [[ "$RUNNER_OS" == "macOS" ]]; then
@@ -172,14 +178,13 @@ jobs:
 
             # 5. Create ZIP file
             cd ..
-            VERSION_CLEAN="${VERSION#v}"
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.zip"
             echo "Creating archive: $ZIP_NAME"
 
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui -DestinationPath $ZIP_NAME"
+                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
             else
-                zip -r "$ZIP_NAME" text-generation-webui
+                zip -r "$ZIP_NAME" "text-generation-webui-${VERSION_CLEAN}"
             fi
 
       - name: Upload files to a GitHub release

From ad0be25c4690918ab3024edb689952e60d941a29 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Jun 2025 15:00:14 -0700
Subject: [PATCH 0702/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_cuda128.txt             | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt      | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index a71e52406a..20ea39ab10 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index db1ead1a59..d0e9aa88b8 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index a08aa39201..609f449e72 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index fa217c3ead..e373306f1b 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 52581f1aa9..2098128f4c 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index b72f22aa20..08828865e3 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index e8de605700..db9e04358e 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 7851041f1e..15d19ffa80 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index c801516608..7275dd0201 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 5e81ce1f7f..f514eb2f88 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 4ddcf43f41..f64e41952b 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 38a21618a0..055d587775 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 0b70c80058..05f8eb63b2 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 510a20f45a..08af341e6a 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index e6d9f0c5df..236f7f9cfb 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 48f92e0a38..a8fb7d85d1 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 9f93424f8b..0811567d15 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 9070b9a609..d9fb09ea59 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.18.0/llama_cpp_binaries-0.18.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 09606a38d34faf980e7ba7c2f4484ce55cc0fc70 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Jun 2025 19:34:07 -0700
Subject: [PATCH 0703/1701] Truncate web search results to at most 8192 tokens

---
 modules/web_search.py | 19 ++++++++++++++++++-
 1 file changed, 18 insertions(+), 1 deletion(-)

diff --git a/modules/web_search.py b/modules/web_search.py
index ffd7e4833a..5b14be55af 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -4,6 +4,7 @@
 
 import requests
 
+from modules import shared
 from modules.logging_colors import logger
 
 
@@ -90,6 +91,22 @@ def perform_web_search(query, num_pages=3, max_workers=5):
         return []
 
 
+def truncate_content_by_tokens(content, max_tokens=8192):
+    """Truncate content to fit within token limit using binary search"""
+    if len(shared.tokenizer.encode(content)) <= max_tokens:
+        return content
+
+    left, right = 0, len(content)
+    while left < right:
+        mid = (left + right + 1) // 2
+        if len(shared.tokenizer.encode(content[:mid])) <= max_tokens:
+            left = mid
+        else:
+            right = mid - 1
+
+    return content[:left]
+
+
 def add_web_search_attachments(history, row_idx, user_message, search_query, state):
     """Perform web search and add results as attachments"""
     if not search_query:
@@ -126,7 +143,7 @@ def add_web_search_attachments(history, row_idx, user_message, search_query, sta
                 "name": result['title'],
                 "type": "text/html",
                 "url": result['url'],
-                "content": result['content']
+                "content": truncate_content_by_tokens(result['content'])
             }
             history['metadata'][key]["attachments"].append(attachment)
 

From e263dbf8522c0b48c34e0842c40032e5a2d2eba2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Jun 2025 19:43:36 -0700
Subject: [PATCH 0704/1701] Improve user input truncation

---
 modules/chat.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index e7cd5ad22a..81e7b16a82 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -332,10 +332,10 @@ def make_prompt(messages):
                 user_message = messages[-1]['content']
 
                 # Bisect the truncation point
-                left, right = 0, len(user_message) - 1
+                left, right = 0, len(user_message)
 
-                while right - left > 1:
-                    mid = (left + right) // 2
+                while left < right:
+                    mid = (left + right + 1) // 2
 
                     messages[-1]['content'] = user_message[:mid]
                     prompt = make_prompt(messages)
@@ -344,7 +344,7 @@ def make_prompt(messages):
                     if encoded_length <= max_length:
                         left = mid
                     else:
-                        right = mid
+                        right = mid - 1
 
                 messages[-1]['content'] = user_message[:left]
                 prompt = make_prompt(messages)

From db7d717df73f5025c3b852024b3f44c42a94dec0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Jun 2025 20:00:11 -0700
Subject: [PATCH 0705/1701] Remove images and links from websearch results

This reduces noise a lot
---
 modules/web_search.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/modules/web_search.py b/modules/web_search.py
index 5b14be55af..401a42bb36 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -29,6 +29,8 @@ def download_web_page(url, timeout=10):
         # Initialize the HTML to Markdown converter
         h = html2text.HTML2Text()
         h.body_width = 0
+        h.ignore_images = True
+        h.ignore_links = True
 
         # Convert the HTML to Markdown
         markdown_text = h.handle(response.text)

From 609c3ac8936092af1df48305a49af3f0a3da165e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Jun 2025 08:03:27 -0700
Subject: [PATCH 0706/1701] Optimize the end of generation with llama.cpp

---
 modules/llama_cpp_server.py |  2 ++
 modules/text_generation.py  | 10 ++++++++--
 2 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index a79e24e4e8..e64f16944a 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -30,6 +30,7 @@ def __init__(self, model_path, server_path=None):
         self.session = requests.Session()
         self.vocabulary_size = None
         self.bos_token = ""
+        self.last_prompt_token_count = 0
 
         # Start the server
         self._start_server()
@@ -128,6 +129,7 @@ def generate_with_streaming(self, prompt, state):
         payload = self.prepare_payload(state)
 
         token_ids = self.encode(prompt, add_bos_token=state["add_bos_token"])
+        self.last_prompt_token_count = len(token_ids)
         if state['auto_max_new_tokens']:
             max_new_tokens = state['truncation_length'] - len(token_ids)
         else:
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 55b538b069..a75141f198 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -498,8 +498,14 @@ def generate_reply_custom(question, original_question, state, stopping_strings=N
         traceback.print_exc()
     finally:
         t1 = time.time()
-        original_tokens = len(encode(original_question)[0])
-        new_tokens = len(encode(original_question + reply)[0]) - original_tokens
+
+        if hasattr(shared.model, 'last_prompt_token_count'):
+            original_tokens = shared.model.last_prompt_token_count
+            new_tokens = len(encode(reply)[0]) if reply else 0
+        else:
+            original_tokens = len(encode(original_question)[0])
+            new_tokens = len(encode(original_question + reply)[0]) - original_tokens
+
         logger.info(f'Output generated in {(t1-t0):.2f} seconds ({new_tokens/(t1-t0):.2f} tokens/s, {new_tokens} tokens, context {original_tokens}, seed {state["seed"]})')
         return
 

From 4fc254c1dd2292004e26c4ffafde31c4fd856747 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Jun 2025 08:11:32 -0700
Subject: [PATCH 0707/1701] Optimize syntax highlighting on long conversations

---
 js/main.js | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

diff --git a/js/main.js b/js/main.js
index e970884d5c..9f77e4a742 100644
--- a/js/main.js
+++ b/js/main.js
@@ -231,8 +231,15 @@ function doSyntaxHighlighting() {
   if (messageBodies.length > 0) {
     observer.disconnect();
 
-    messageBodies.forEach((messageBody) => {
+    let hasSeenVisible = false;
+
+    // Go from last message to first
+    for (let i = messageBodies.length - 1; i >= 0; i--) {
+      const messageBody = messageBodies[i];
+
       if (isElementVisibleOnScreen(messageBody)) {
+        hasSeenVisible = true;
+
         // Handle both code and math in a single pass through each message
         const codeBlocks = messageBody.querySelectorAll("pre code:not([data-highlighted])");
         codeBlocks.forEach((codeBlock) => {
@@ -249,8 +256,12 @@ function doSyntaxHighlighting() {
             { left: "\\[", right: "\\]", display: true },
           ],
         });
+      } else if (hasSeenVisible) {
+        // We've seen visible messages but this one is not visible
+        // Since we're going from last to first, we can break
+        break;
       }
-    });
+    }
 
     observer.observe(targetElement, config);
   }

From e8dc7b0ee9b810131b7b347268f8c4680c58ae06 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Jun 2025 08:15:29 -0700
Subject: [PATCH 0708/1701] Bump exllamav3 to 0.0.4

---
 requirements/full/requirements.txt                | 4 ++--
 requirements/full/requirements_apple_intel.txt    | 2 +-
 requirements/full/requirements_apple_silicon.txt  | 2 +-
 requirements/full/requirements_cuda128.txt        | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt | 4 ++--
 requirements/full/requirements_noavx2.txt         | 4 ++--
 6 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 20ea39ab10..adef3d6d3e 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -36,8 +36,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index e373306f1b..3eb9fc422b 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -35,5 +35,5 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 2098128f4c..0cb1163bad 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -36,5 +36,5 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 15d19ffa80..fa4a5ca936 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -36,8 +36,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 7275dd0201..72a0fd928e 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -36,8 +36,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index f514eb2f88..62fbd05087 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -36,8 +36,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.3/exllamav3-0.0.3+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From b279460a819f26da079e8ebe1e2e05c0fefe631f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Jun 2025 08:25:07 -0700
Subject: [PATCH 0709/1701] Improve the wpp style

---
 css/chat_style-wpp.css | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/css/chat_style-wpp.css b/css/chat_style-wpp.css
index 8ff82c6c98..57b90ee32a 100644
--- a/css/chat_style-wpp.css
+++ b/css/chat_style-wpp.css
@@ -78,8 +78,6 @@
 }
 
 .message-body p {
-    margin-bottom: 0 !important;
-    margin-top: 2px !important;
     font-size: 15px !important;
     line-height: 1.4 !important;
     font-weight: 400;
@@ -101,3 +99,7 @@
 .message-actions {
     margin-top: 8px;
 }
+
+.message-body p, .chat .message-body ul, .chat .message-body ol {
+    margin-bottom: 10px !important;
+}

From cc757f6226b596c409a4369fdf0df4e227e4b032 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Jun 2025 08:32:06 -0700
Subject: [PATCH 0710/1701] Small style improvements to the chat tab

---
 css/main.css | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/css/main.css b/css/main.css
index 3e02c6529f..8561988f32 100644
--- a/css/main.css
+++ b/css/main.css
@@ -905,6 +905,10 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     flex-shrink: 1;
 }
 
+#search_chat {
+    padding-right: 0.5rem;
+}
+
 #search_chat > :nth-child(2) > :first-child {
     display: none;
 }
@@ -1020,12 +1024,14 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     width: 100%;
     justify-content: center;
     gap: 9px;
+    padding-right: 0.5rem;
 }
 
 #past-chats-row,
 #chat-controls {
     width: 260px;
     padding: 0.5rem;
+    padding-right: 0;
     height: calc(100dvh - 16px);
     flex-shrink: 0;
     box-sizing: content-box;

From bc2b0f54e9a9f8ba4cd8be041bc5bcd228196320 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Jun 2025 15:53:16 -0700
Subject: [PATCH 0711/1701] Only save extensions settings on manual save

---
 modules/ui.py | 37 ++++++++++++++++++++++---------------
 1 file changed, 22 insertions(+), 15 deletions(-)

diff --git a/modules/ui.py b/modules/ui.py
index 463e59deec..57a1c3837d 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -6,6 +6,7 @@
 import yaml
 
 import extensions
+import modules.extensions as extensions_module
 from modules import shared
 from modules.chat import load_history
 from modules.utils import gradio
@@ -334,35 +335,41 @@ def save_settings(state, preset, extensions_list, show_controls, theme_state, ma
     output.pop('instruction_template_str')
     output.pop('truncation_length')
 
-    # Only save extensions on manual save
+    # Handle extensions and extension parameters
     if manual_save:
+        # Save current extensions and their parameter values
         output['default_extensions'] = extensions_list
+
+        for extension_name in extensions_list:
+            extension = getattr(extensions, extension_name, None)
+            if extension:
+                extension = extension.script
+                if hasattr(extension, 'params'):
+                    params = getattr(extension, 'params')
+                    for param in params:
+                        _id = f"{extension_name}-{param}"
+                        # Only save if different from default value
+                        if param not in shared.default_settings or params[param] != shared.default_settings[param]:
+                            output[_id] = params[param]
     else:
-        # Preserve existing extensions from settings file during autosave
+        # Preserve existing extensions and extension parameters during autosave
         settings_path = Path('user_data') / 'settings.yaml'
         if settings_path.exists():
             try:
                 with open(settings_path, 'r', encoding='utf-8') as f:
                     existing_settings = yaml.safe_load(f.read()) or {}
 
+                # Preserve default_extensions
                 if 'default_extensions' in existing_settings:
                     output['default_extensions'] = existing_settings['default_extensions']
+
+                # Preserve extension parameter values
+                for key, value in existing_settings.items():
+                    if any(key.startswith(f"{ext_name}-") for ext_name in extensions_module.available_extensions):
+                        output[key] = value
             except Exception:
                 pass  # If we can't read the file, just don't modify extensions
 
-    # Save extension values in the UI
-    for extension_name in extensions_list:
-        extension = getattr(extensions, extension_name, None)
-        if extension:
-            extension = extension.script
-            if hasattr(extension, 'params'):
-                params = getattr(extension, 'params')
-                for param in params:
-                    _id = f"{extension_name}-{param}"
-                    # Only save if different from default value
-                    if param not in shared.default_settings or params[param] != shared.default_settings[param]:
-                        output[_id] = params[param]
-
     # Do not save unchanged settings
     for key in list(output.keys()):
         if key in shared.default_settings and output[key] == shared.default_settings[key]:

From 9bcef8a6487097cb480f9265c9f0061f71cdc1e5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Jun 2025 17:57:41 -0700
Subject: [PATCH 0712/1701] Fix "show controls" conflicting with manually
 hiding the sidebars

---
 js/show_controls.js | 59 ++++++++++++++++++++++++++++++---------------
 1 file changed, 40 insertions(+), 19 deletions(-)

diff --git a/js/show_controls.js b/js/show_controls.js
index 06387211ab..c98102d3a4 100644
--- a/js/show_controls.js
+++ b/js/show_controls.js
@@ -1,16 +1,30 @@
-const belowChatInput = document.querySelectorAll(
-  "#chat-tab > div > :nth-child(1), #chat-tab > div > :nth-child(3), #chat-tab > div > :nth-child(4), #extensions"
-);
 const chatParent = document.querySelector(".chat-parent");
 
 function toggle_controls(value) {
-  const headerBar = document.querySelector(".header_bar");
+  const extensions = document.querySelector("#extensions");
 
   if (value) {
-    belowChatInput.forEach(element => {
-      element.style.display = "inherit";
-    });
+    // SHOW MODE: Click toggles to show hidden sidebars
+    const navToggle = document.getElementById("navigation-toggle");
+    const pastChatsToggle = document.getElementById("past-chats-toggle");
+    const chatControlsToggle = document.getElementById("chat-controls-toggle");
 
+    if (navToggle && document.querySelector(".header_bar")?.classList.contains("sidebar-hidden")) {
+      navToggle.click();
+    }
+    if (pastChatsToggle && document.getElementById("past-chats-row")?.classList.contains("sidebar-hidden")) {
+      pastChatsToggle.click();
+    }
+    if (chatControlsToggle && document.getElementById("chat-controls")?.classList.contains("sidebar-hidden")) {
+      chatControlsToggle.click();
+    }
+
+    // Show extensions only
+    if (extensions) {
+      extensions.style.display = "inherit";
+    }
+
+    // Remove bigchat classes
     chatParent.classList.remove("bigchat");
     document.getElementById("chat-input-row").classList.remove("bigchat");
     document.getElementById("chat-col").classList.remove("bigchat");
@@ -21,24 +35,31 @@ function toggle_controls(value) {
       gallery_element.style.display = "block";
     }
 
-    // Show header bar
-    if (headerBar) {
-      headerBar.style.removeProperty("display");
+  } else {
+    // HIDE MODE: Click toggles to hide visible sidebars
+    const navToggle = document.getElementById("navigation-toggle");
+    const pastChatsToggle = document.getElementById("past-chats-toggle");
+    const chatControlsToggle = document.getElementById("chat-controls-toggle");
+
+    if (navToggle && !document.querySelector(".header_bar")?.classList.contains("sidebar-hidden")) {
+      navToggle.click();
+    }
+    if (pastChatsToggle && !document.getElementById("past-chats-row")?.classList.contains("sidebar-hidden")) {
+      pastChatsToggle.click();
+    }
+    if (chatControlsToggle && !document.getElementById("chat-controls")?.classList.contains("sidebar-hidden")) {
+      chatControlsToggle.click();
     }
 
-  } else {
-    belowChatInput.forEach(element => {
-      element.style.display = "none";
-    });
+    // Hide extensions only
+    if (extensions) {
+      extensions.style.display = "none";
+    }
 
+    // Add bigchat classes
     chatParent.classList.add("bigchat");
     document.getElementById("chat-input-row").classList.add("bigchat");
     document.getElementById("chat-col").classList.add("bigchat");
     document.getElementById("chat-tab").style.paddingBottom = "0px";
-
-    // Hide header bar
-    if (headerBar) {
-      headerBar.style.setProperty("display", "none", "important");
-    }
   }
 }

From d347b056e3901878c7c95b7ed0414725f05ae45b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Jun 2025 18:12:11 -0700
Subject: [PATCH 0713/1701] Always close/open the two sidebars at the same time

---
 js/main.js          | 26 ++++++++++++++++++++++++++
 js/show_controls.js |  8 --------
 2 files changed, 26 insertions(+), 8 deletions(-)

diff --git a/js/main.js b/js/main.js
index 9f77e4a742..8b31bbaa49 100644
--- a/js/main.js
+++ b/js/main.js
@@ -788,11 +788,37 @@ initializeSidebars();
 
 // Add click event listeners to toggle buttons
 pastChatsToggle.addEventListener("click", () => {
+  const isCurrentlyOpen = !pastChatsRow.classList.contains("sidebar-hidden");
   toggleSidebar(pastChatsRow, pastChatsToggle);
+
+  if (isCurrentlyOpen) {
+    // If we just closed the left sidebar, also close the right sidebar
+    if (!chatControlsRow.classList.contains("sidebar-hidden")) {
+      toggleSidebar(chatControlsRow, chatControlsToggle, true);
+    }
+  } else {
+    // If we just opened the left sidebar, also open the right sidebar
+    if (chatControlsRow.classList.contains("sidebar-hidden")) {
+      toggleSidebar(chatControlsRow, chatControlsToggle, false);
+    }
+  }
 });
 
 chatControlsToggle.addEventListener("click", () => {
+  const isCurrentlyOpen = !chatControlsRow.classList.contains("sidebar-hidden");
   toggleSidebar(chatControlsRow, chatControlsToggle);
+
+  if (isCurrentlyOpen) {
+    // If we just closed the right sidebar, also close the left sidebar
+    if (!pastChatsRow.classList.contains("sidebar-hidden")) {
+      toggleSidebar(pastChatsRow, pastChatsToggle, true);
+    }
+  } else {
+    // If we just opened the right sidebar, also open the left sidebar
+    if (pastChatsRow.classList.contains("sidebar-hidden")) {
+      toggleSidebar(pastChatsRow, pastChatsToggle, false);
+    }
+  }
 });
 
 navigationToggle.addEventListener("click", () => {
diff --git a/js/show_controls.js b/js/show_controls.js
index c98102d3a4..f974d412fa 100644
--- a/js/show_controls.js
+++ b/js/show_controls.js
@@ -7,7 +7,6 @@ function toggle_controls(value) {
     // SHOW MODE: Click toggles to show hidden sidebars
     const navToggle = document.getElementById("navigation-toggle");
     const pastChatsToggle = document.getElementById("past-chats-toggle");
-    const chatControlsToggle = document.getElementById("chat-controls-toggle");
 
     if (navToggle && document.querySelector(".header_bar")?.classList.contains("sidebar-hidden")) {
       navToggle.click();
@@ -15,9 +14,6 @@ function toggle_controls(value) {
     if (pastChatsToggle && document.getElementById("past-chats-row")?.classList.contains("sidebar-hidden")) {
       pastChatsToggle.click();
     }
-    if (chatControlsToggle && document.getElementById("chat-controls")?.classList.contains("sidebar-hidden")) {
-      chatControlsToggle.click();
-    }
 
     // Show extensions only
     if (extensions) {
@@ -39,7 +35,6 @@ function toggle_controls(value) {
     // HIDE MODE: Click toggles to hide visible sidebars
     const navToggle = document.getElementById("navigation-toggle");
     const pastChatsToggle = document.getElementById("past-chats-toggle");
-    const chatControlsToggle = document.getElementById("chat-controls-toggle");
 
     if (navToggle && !document.querySelector(".header_bar")?.classList.contains("sidebar-hidden")) {
       navToggle.click();
@@ -47,9 +42,6 @@ function toggle_controls(value) {
     if (pastChatsToggle && !document.getElementById("past-chats-row")?.classList.contains("sidebar-hidden")) {
       pastChatsToggle.click();
     }
-    if (chatControlsToggle && !document.getElementById("chat-controls")?.classList.contains("sidebar-hidden")) {
-      chatControlsToggle.click();
-    }
 
     // Hide extensions only
     if (extensions) {

From 949b7ec9cf0840e48e39abe7181a69ec8a53fa6a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Jun 2025 18:50:21 -0700
Subject: [PATCH 0714/1701] Further optimize scrolling in the chat tab

---
 css/main.css | 1 +
 1 file changed, 1 insertion(+)

diff --git a/css/main.css b/css/main.css
index 8561988f32..2e8d31ad7d 100644
--- a/css/main.css
+++ b/css/main.css
@@ -421,6 +421,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     text-align: start;
     padding-left: 1rem;
     padding-right: 1rem;
+    contain: layout;
 }
 
 .chat .message .timestamp {

From f77f1504f5c75f6c4a0c6697c77073c5f25b18f5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Jun 2025 06:12:37 -0700
Subject: [PATCH 0715/1701] Improve the style of the Character and User tabs

---
 css/main.css       | 12 ++++++++++++
 modules/ui_chat.py |  6 +++---
 2 files changed, 15 insertions(+), 3 deletions(-)

diff --git a/css/main.css b/css/main.css
index 2e8d31ad7d..93950e2e7b 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1645,3 +1645,15 @@ button:focus {
     justify-content: center;
     display: none;
 }
+
+#character-context textarea {
+    height: calc((100vh - 400px) * 2/3) !important;
+}
+
+#character-greeting textarea {
+    height: calc((100vh - 400px) * 1/3) !important;
+}
+
+#user-description textarea {
+    height: calc(100vh - 281px) !important;
+}
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 3b841b8b24..9bbe250722 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -125,12 +125,12 @@ def create_chat_settings_ui():
                         shared.gradio['restore_character'] = gr.Button('Restore character', elem_classes='refresh-button', interactive=True, elem_id='restore-character')
 
                     shared.gradio['name2'] = gr.Textbox(value=shared.settings['name2'], lines=1, label='Character\'s name')
-                    shared.gradio['context'] = gr.Textbox(value=shared.settings['context'], lines=10, label='Context', elem_classes=['add_scrollbar'])
-                    shared.gradio['greeting'] = gr.Textbox(value=shared.settings['greeting'], lines=5, label='Greeting', elem_classes=['add_scrollbar'])
+                    shared.gradio['context'] = gr.Textbox(value=shared.settings['context'], lines=10, label='Context', elem_classes=['add_scrollbar'], elem_id="character-context")
+                    shared.gradio['greeting'] = gr.Textbox(value=shared.settings['greeting'], lines=5, label='Greeting', elem_classes=['add_scrollbar'], elem_id="character-greeting")
 
                 with gr.Tab("User"):
                     shared.gradio['name1'] = gr.Textbox(value=shared.settings['name1'], lines=1, label='Name')
-                    shared.gradio['user_bio'] = gr.Textbox(value=shared.settings['user_bio'], lines=10, label='Description', info='Here you can optionally write a description of yourself.', placeholder='{{user}}\'s personality: ...', elem_classes=['add_scrollbar'])
+                    shared.gradio['user_bio'] = gr.Textbox(value=shared.settings['user_bio'], lines=10, label='Description', info='Here you can optionally write a description of yourself.', placeholder='{{user}}\'s personality: ...', elem_classes=['add_scrollbar'], elem_id="user-description")
 
                 with gr.Tab('Chat history'):
                     with gr.Row():

From c9c3b716fbc416cc900df87d81de81440f65dad4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Jun 2025 07:21:25 -0700
Subject: [PATCH 0716/1701] Move character settings to a new 'Character' main
 tab

---
 css/main.css             | 10 +++++++---
 modules/ui_chat.py       |  9 ++++++---
 modules/ui_parameters.py |  1 -
 server.py                |  1 +
 4 files changed, 14 insertions(+), 7 deletions(-)

diff --git a/css/main.css b/css/main.css
index 93950e2e7b..2a6f87152d 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1647,13 +1647,17 @@ button:focus {
 }
 
 #character-context textarea {
-    height: calc((100vh - 400px) * 2/3) !important;
+    height: calc((100vh - 350px) * 2/3) !important;
 }
 
 #character-greeting textarea {
-    height: calc((100vh - 400px) * 1/3) !important;
+    height: calc((100vh - 350px) * 1/3) !important;
 }
 
 #user-description textarea {
-    height: calc(100vh - 281px) !important;
+    height: calc(100vh - 231px) !important;
+}
+
+#character-tab {
+    border: 0;
 }
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 9bbe250722..c0dc05ae2b 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -111,9 +111,9 @@ def create_ui():
             shared.gradio['edit_message'] = gr.Button(elem_id="Edit-message")
 
 
-def create_chat_settings_ui():
+def create_character_settings_ui():
     mu = shared.args.multi_user
-    with gr.Tab('Chat'):
+    with gr.Tab('Character', elem_id="character-tab"):
         with gr.Row():
             with gr.Column(scale=8):
                 with gr.Tab("Character"):
@@ -163,6 +163,9 @@ def create_chat_settings_ui():
                 shared.gradio['character_picture'] = gr.Image(label='Character picture', type='pil', interactive=not mu)
                 shared.gradio['your_picture'] = gr.Image(label='Your picture', type='pil', value=Image.open(Path('user_data/cache/pfp_me.png')) if Path('user_data/cache/pfp_me.png').exists() else None, interactive=not mu)
 
+
+def create_chat_settings_ui():
+    mu = shared.args.multi_user
     with gr.Tab('Instruction template'):
         with gr.Row():
             with gr.Column():
@@ -186,7 +189,7 @@ def create_chat_settings_ui():
                     shared.gradio['send_instruction_to_negative_prompt'] = gr.Button('Send to negative prompt', elem_classes=['small-button'])
 
             with gr.Column():
-                shared.gradio['chat_template_str'] = gr.Textbox(value=shared.settings['chat_template_str'], label='Chat template', lines=22, elem_classes=['add_scrollbar', 'monospace'])
+                shared.gradio['chat_template_str'] = gr.Textbox(value=shared.settings['chat_template_str'], label='Chat template', lines=22, elem_classes=['add_scrollbar', 'monospace'], info='Defines how the chat prompt in chat/chat-instruct modes is generated.')
 
 
 def create_event_handlers():
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index e2b1055478..26b58a0938 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -93,7 +93,6 @@ def create_ui():
                         with gr.Column():
                             shared.gradio['truncation_length'] = gr.Number(precision=0, step=256, value=get_truncation_length(), label='Truncate the prompt up to this length', info='The leftmost tokens are removed if the prompt exceeds this length.')
                             shared.gradio['seed'] = gr.Number(value=shared.settings['seed'], label='Seed (-1 for random)')
-
                             shared.gradio['custom_stopping_strings'] = gr.Textbox(lines=2, value=shared.settings["custom_stopping_strings"] or None, label='Custom stopping strings', info='Written between "" and separated by commas.', placeholder='"\\n", "\\nYou:"')
                             shared.gradio['custom_token_bans'] = gr.Textbox(value=shared.settings['custom_token_bans'] or None, label='Token bans', info='Token IDs to ban, separated by commas. The IDs can be found in the Default or Notebook tab.')
                             shared.gradio['negative_prompt'] = gr.Textbox(value=shared.settings['negative_prompt'], label='Negative prompt', info='For CFG. Only used when guidance_scale is different than 1.', lines=3, elem_classes=['add_scrollbar'])
diff --git a/server.py b/server.py
index cfb21a6e35..ceee8a00f6 100644
--- a/server.py
+++ b/server.py
@@ -150,6 +150,7 @@ def create_interface():
         ui_notebook.create_ui()
 
         ui_parameters.create_ui()  # Parameters tab
+        ui_chat.create_character_settings_ui()  # Character tab
         ui_model_menu.create_ui()  # Model tab
         if not shared.args.portable:
             training.create_ui()  # Training tab

From 34bf93ef47325fced325edcf0aaf9fe6434096a7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Jun 2025 07:22:14 -0700
Subject: [PATCH 0717/1701] Move 'Custom system message' to the Parameters tab

---
 modules/ui_chat.py       | 1 -
 modules/ui_parameters.py | 1 +
 2 files changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index c0dc05ae2b..7e8f0273ae 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -181,7 +181,6 @@ def create_chat_settings_ui():
 
         with gr.Row():
             with gr.Column():
-                shared.gradio['custom_system_message'] = gr.Textbox(value=shared.settings['custom_system_message'], lines=2, label='Custom system message', info='If not empty, will be used instead of the default one.', elem_classes=['add_scrollbar'])
                 shared.gradio['instruction_template_str'] = gr.Textbox(value=shared.settings['instruction_template_str'], label='Instruction template', lines=24, info='This gets autodetected; you usually don\'t need to change it. Used in instruct and chat-instruct modes.', elem_classes=['add_scrollbar', 'monospace'])
                 with gr.Row():
                     shared.gradio['send_instruction_to_default'] = gr.Button('Send to default', elem_classes=['small-button'])
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 26b58a0938..e42e4c0cbe 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -93,6 +93,7 @@ def create_ui():
                         with gr.Column():
                             shared.gradio['truncation_length'] = gr.Number(precision=0, step=256, value=get_truncation_length(), label='Truncate the prompt up to this length', info='The leftmost tokens are removed if the prompt exceeds this length.')
                             shared.gradio['seed'] = gr.Number(value=shared.settings['seed'], label='Seed (-1 for random)')
+                            shared.gradio['custom_system_message'] = gr.Textbox(value=shared.settings['custom_system_message'], lines=2, label='Custom system message', info='If not empty, will be used instead of the default one.', elem_classes=['add_scrollbar'])
                             shared.gradio['custom_stopping_strings'] = gr.Textbox(lines=2, value=shared.settings["custom_stopping_strings"] or None, label='Custom stopping strings', info='Written between "" and separated by commas.', placeholder='"\\n", "\\nYou:"')
                             shared.gradio['custom_token_bans'] = gr.Textbox(value=shared.settings['custom_token_bans'] or None, label='Token bans', info='Token IDs to ban, separated by commas. The IDs can be found in the Default or Notebook tab.')
                             shared.gradio['negative_prompt'] = gr.Textbox(value=shared.settings['negative_prompt'], label='Negative prompt', info='For CFG. Only used when guidance_scale is different than 1.', lines=3, elem_classes=['add_scrollbar'])

From 7ba3d4425fe06acff0928eea117c7e68d163e81b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Jun 2025 07:23:09 -0700
Subject: [PATCH 0718/1701] Remove the 'Send to negative prompt' button

---
 modules/ui_chat.py | 6 ------
 1 file changed, 6 deletions(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 7e8f0273ae..3ad59a401d 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -185,7 +185,6 @@ def create_chat_settings_ui():
                 with gr.Row():
                     shared.gradio['send_instruction_to_default'] = gr.Button('Send to default', elem_classes=['small-button'])
                     shared.gradio['send_instruction_to_notebook'] = gr.Button('Send to notebook', elem_classes=['small-button'])
-                    shared.gradio['send_instruction_to_negative_prompt'] = gr.Button('Send to negative prompt', elem_classes=['small-button'])
 
             with gr.Column():
                 shared.gradio['chat_template_str'] = gr.Textbox(value=shared.settings['chat_template_str'], label='Chat template', lines=22, elem_classes=['add_scrollbar', 'monospace'], info='Defines how the chat prompt in chat/chat-instruct modes is generated.')
@@ -355,11 +354,6 @@ def create_event_handlers():
         chat.handle_send_instruction_click, gradio('interface_state'), gradio('textbox-notebook'), show_progress=False).then(
         None, None, None, js=f'() => {{{ui.switch_tabs_js}; switch_to_notebook()}}')
 
-    shared.gradio['send_instruction_to_negative_prompt'].click(
-        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        chat.handle_send_instruction_click, gradio('interface_state'), gradio('negative_prompt'), show_progress=False).then(
-        None, None, None, js=f'() => {{{ui.switch_tabs_js}; switch_to_generation_parameters()}}')
-
     shared.gradio['send-chat-to-default'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_send_chat_click, gradio('interface_state'), gradio('textbox-default'), show_progress=False).then(

From cac225b5895650e1d387e4a91aeeb7b061443c3d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Jun 2025 07:26:39 -0700
Subject: [PATCH 0719/1701] Small style improvements

---
 css/main.css       | 9 +++++----
 modules/ui_chat.py | 4 ++--
 2 files changed, 7 insertions(+), 6 deletions(-)

diff --git a/css/main.css b/css/main.css
index 2a6f87152d..13105bf085 100644
--- a/css/main.css
+++ b/css/main.css
@@ -97,11 +97,11 @@ ol li p, ul li p {
     display: inline-block;
 }
 
-#chat-tab, #default-tab, #notebook-tab, #parameters, #chat-settings, #lora, #training-tab, #model-tab, #session-tab {
+#chat-tab, #default-tab, #notebook-tab, #parameters, #chat-settings, #lora, #training-tab, #model-tab, #session-tab, #character-tab {
     border: 0;
 }
 
-#default-tab, #notebook-tab, #parameters, #chat-settings, #lora, #training-tab, #model-tab, #session-tab {
+#default-tab, #notebook-tab, #parameters, #chat-settings, #lora, #training-tab, #model-tab, #session-tab, #character-tab {
     padding: 1rem;
 }
 
@@ -1658,6 +1658,7 @@ button:focus {
     height: calc(100vh - 231px) !important;
 }
 
-#character-tab {
-    border: 0;
+#instruction-template-str textarea,
+#chat-template-str textarea {
+    height: calc(100vh - 300px) !important;
 }
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 3ad59a401d..d1447286c2 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -181,13 +181,13 @@ def create_chat_settings_ui():
 
         with gr.Row():
             with gr.Column():
-                shared.gradio['instruction_template_str'] = gr.Textbox(value=shared.settings['instruction_template_str'], label='Instruction template', lines=24, info='This gets autodetected; you usually don\'t need to change it. Used in instruct and chat-instruct modes.', elem_classes=['add_scrollbar', 'monospace'])
+                shared.gradio['instruction_template_str'] = gr.Textbox(value=shared.settings['instruction_template_str'], label='Instruction template', lines=24, info='This gets autodetected; you usually don\'t need to change it. Used in instruct and chat-instruct modes.', elem_classes=['add_scrollbar', 'monospace'], elem_id='instruction-template-str')
                 with gr.Row():
                     shared.gradio['send_instruction_to_default'] = gr.Button('Send to default', elem_classes=['small-button'])
                     shared.gradio['send_instruction_to_notebook'] = gr.Button('Send to notebook', elem_classes=['small-button'])
 
             with gr.Column():
-                shared.gradio['chat_template_str'] = gr.Textbox(value=shared.settings['chat_template_str'], label='Chat template', lines=22, elem_classes=['add_scrollbar', 'monospace'], info='Defines how the chat prompt in chat/chat-instruct modes is generated.')
+                shared.gradio['chat_template_str'] = gr.Textbox(value=shared.settings['chat_template_str'], label='Chat template', lines=22, elem_classes=['add_scrollbar', 'monospace'], info='Defines how the chat prompt in chat/chat-instruct modes is generated.', elem_id='chat-template-str')
 
 
 def create_event_handlers():

From db67d69ddca355127cd315ea911eea27d30a2177 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Jun 2025 07:28:14 -0700
Subject: [PATCH 0720/1701] Lint

---
 server.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/server.py b/server.py
index ceee8a00f6..b28d30bc50 100644
--- a/server.py
+++ b/server.py
@@ -33,7 +33,6 @@
 
 matplotlib.use('Agg')  # This fixes LaTeX rendering on some systems
 
-import json
 import os
 import signal
 import sys

From de24b3bb31874c1d6336510651084c6abc71d69c Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Mon, 16 Jun 2025 13:19:29 -0300
Subject: [PATCH 0721/1701] Merge the Default and Notebook tabs into a single
 Notebook tab (#7078)

---
 README.md              |  2 +-
 css/main.css           |  4 ++--
 modules/shared.py      |  2 +-
 modules/ui.py          |  5 +++--
 modules/ui_default.py  |  6 +++---
 modules/ui_notebook.py |  2 +-
 modules/ui_session.py  | 32 ++++++++++++++++++++++++++++++++
 server.py              |  5 +++--
 8 files changed, 46 insertions(+), 12 deletions(-)

diff --git a/README.md b/README.md
index 45ab48eb38..9fe4fcd4b3 100644
--- a/README.md
+++ b/README.md
@@ -24,7 +24,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 - Multiple sampling parameters and generation options for sophisticated text generation control.
 - Switch between different models in the UI without restarting.
 - Automatic GPU layers for GGUF models (on NVIDIA GPUs).
-- Free-form text generation in the Default/Notebook tabs without being limited to chat turns.
+- Free-form text generation in the Notebook tab without being limited to chat turns.
 - OpenAI-compatible API with Chat and Completions endpoints, including tool-calling support – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples).
 - Extension support, with numerous built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details.
 
diff --git a/css/main.css b/css/main.css
index 13105bf085..ee50ac6a8a 100644
--- a/css/main.css
+++ b/css/main.css
@@ -97,11 +97,11 @@ ol li p, ul li p {
     display: inline-block;
 }
 
-#chat-tab, #default-tab, #notebook-tab, #parameters, #chat-settings, #lora, #training-tab, #model-tab, #session-tab, #character-tab {
+#notebook-parent-tab, #chat-tab, #parameters, #chat-settings, #lora, #training-tab, #model-tab, #session-tab, #character-tab {
     border: 0;
 }
 
-#default-tab, #notebook-tab, #parameters, #chat-settings, #lora, #training-tab, #model-tab, #session-tab, #character-tab {
+#notebook-parent-tab, #parameters, #chat-settings, #lora, #training-tab, #model-tab, #session-tab, #character-tab {
     padding: 1rem;
 }
 
diff --git a/modules/shared.py b/modules/shared.py
index 9efc8daf9f..4c9e34a7a7 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -202,7 +202,6 @@
     'chat-instruct_command': 'Continue the chat dialogue below. Write a single reply for the character "<|character|>".\n\n<|prompt|>',
     'enable_web_search': False,
     'web_search_pages': 3,
-    'prompt-default': 'QA',
     'prompt-notebook': 'QA',
     'preset': 'Qwen3 - Thinking' if Path('user_data/presets/Qwen3 - Thinking.yaml').exists() else None,
     'max_new_tokens': 512,
@@ -223,6 +222,7 @@
     'custom_token_bans': '',
     'negative_prompt': '',
     'dark_theme': True,
+    'show_two_notebook_columns': False,
     'paste_to_attachment': False,
     'include_past_attachments': True,
 
diff --git a/modules/ui.py b/modules/ui.py
index 57a1c3837d..0e8afa8f5a 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -274,6 +274,7 @@ def list_interface_input_elements():
 
     # Other elements
     elements += [
+        'show_two_notebook_columns',
         'paste_to_attachment',
         'include_past_attachments',
     ]
@@ -326,8 +327,7 @@ def save_settings(state, preset, extensions_list, show_controls, theme_state, ma
             output[k] = state[k]
 
     output['preset'] = preset
-    output['prompt-default'] = state['prompt_menu-default']
-    output['prompt-notebook'] = state['prompt_menu-notebook']
+    output['prompt-notebook'] = state['prompt_menu-default'] if state['show_two_notebook_columns'] else state['prompt_menu-notebook']
     output['character'] = state['character_menu']
     output['seed'] = int(output['seed'])
     output['show_controls'] = show_controls
@@ -505,6 +505,7 @@ def setup_auto_save():
         # Session tab (ui_session.py)
         'show_controls',
         'theme_state',
+        'show_two_notebook_columns',
         'paste_to_attachment',
         'include_past_attachments'
     ]
diff --git a/modules/ui_default.py b/modules/ui_default.py
index 8acc4b1093..e6d2f92d9a 100644
--- a/modules/ui_default.py
+++ b/modules/ui_default.py
@@ -15,11 +15,11 @@
 
 def create_ui():
     mu = shared.args.multi_user
-    with gr.Tab('Default', elem_id='default-tab'):
+    with gr.Row(visible=shared.settings['show_two_notebook_columns']) as shared.gradio['default-tab']:
         with gr.Row():
             with gr.Column():
                 with gr.Row():
-                    shared.gradio['textbox-default'] = gr.Textbox(value=load_prompt(shared.settings['prompt-default']), lines=27, label='Input', elem_classes=['textbox_default', 'add_scrollbar'])
+                    shared.gradio['textbox-default'] = gr.Textbox(value=load_prompt(shared.settings['prompt-notebook']), lines=27, label='Input', elem_classes=['textbox_default', 'add_scrollbar'])
                     shared.gradio['token-counter-default'] = gr.HTML(value="0", elem_id="default-token-counter")
 
                 with gr.Row():
@@ -28,7 +28,7 @@ def create_ui():
                     shared.gradio['Generate-default'] = gr.Button('Generate', variant='primary')
 
                 with gr.Row():
-                    shared.gradio['prompt_menu-default'] = gr.Dropdown(choices=utils.get_available_prompts(), value=shared.settings['prompt-default'], label='Prompt', elem_classes='slim-dropdown')
+                    shared.gradio['prompt_menu-default'] = gr.Dropdown(choices=utils.get_available_prompts(), value=shared.settings['prompt-notebook'], label='Prompt', elem_classes='slim-dropdown')
                     ui.create_refresh_button(shared.gradio['prompt_menu-default'], lambda: None, lambda: {'choices': utils.get_available_prompts()}, 'refresh-button', interactive=not mu)
                     shared.gradio['save_prompt-default'] = gr.Button('💾', elem_classes='refresh-button', interactive=not mu)
                     shared.gradio['delete_prompt-default'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu)
diff --git a/modules/ui_notebook.py b/modules/ui_notebook.py
index 3f79a93c36..e6b87342b6 100644
--- a/modules/ui_notebook.py
+++ b/modules/ui_notebook.py
@@ -16,7 +16,7 @@
 
 def create_ui():
     mu = shared.args.multi_user
-    with gr.Tab('Notebook', elem_id='notebook-tab'):
+    with gr.Row(visible=not shared.settings['show_two_notebook_columns']) as shared.gradio['notebook-tab']:
         shared.gradio['last_input-notebook'] = gr.State('')
         with gr.Row():
             with gr.Column(scale=4):
diff --git a/modules/ui_session.py b/modules/ui_session.py
index dc7ca2fb09..c0a34228cf 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -11,6 +11,7 @@ def create_ui():
             with gr.Column():
                 gr.Markdown("## Settings")
                 shared.gradio['toggle_dark_mode'] = gr.Button('Toggle light/dark theme 💡', elem_classes='refresh-button')
+                shared.gradio['show_two_notebook_columns'] = gr.Checkbox(label='Show two columns in the Notebook tab', value=shared.settings['show_two_notebook_columns'])
                 shared.gradio['paste_to_attachment'] = gr.Checkbox(label='Turn long pasted text into attachments in the Chat tab', value=shared.settings['paste_to_attachment'], elem_id='paste_to_attachment')
                 shared.gradio['include_past_attachments'] = gr.Checkbox(label='Include attachments/search results from previous messages in the chat prompt', value=shared.settings['include_past_attachments'])
 
@@ -34,6 +35,12 @@ def create_ui():
             lambda x: 'dark' if x == 'light' else 'light', gradio('theme_state'), gradio('theme_state')).then(
             None, None, None, js=f'() => {{{ui.dark_theme_js}; toggleDarkMode(); localStorage.setItem("theme", document.body.classList.contains("dark") ? "dark" : "light")}}')
 
+        shared.gradio['show_two_notebook_columns'].change(
+            handle_default_to_notebook_change,
+            gradio('show_two_notebook_columns', 'textbox-default', 'output_textbox', 'prompt_menu-default', 'textbox-notebook', 'prompt_menu-notebook'),
+            gradio('default-tab', 'notebook-tab', 'textbox-default', 'output_textbox', 'prompt_menu-default', 'textbox-notebook', 'prompt_menu-notebook')
+        )
+
         # Reset interface event
         shared.gradio['reset_interface'].click(
             set_interface_arguments, gradio('extensions_menu', 'bool_menu'), None).then(
@@ -50,6 +57,31 @@ def handle_save_settings(state, preset, extensions, show_controls, theme):
     ]
 
 
+def handle_default_to_notebook_change(show_two_columns, default_input, default_output, default_prompt, notebook_input, notebook_prompt):
+    if show_two_columns:
+        # Notebook to default
+        return [
+            gr.update(visible=True),
+            gr.update(visible=False),
+            notebook_input,
+            notebook_input,
+            gr.update(value=notebook_prompt, choices=utils.get_available_prompts()),
+            gr.update(),
+            gr.update(),
+        ]
+    else:
+        # Default to notebook
+        return [
+            gr.update(visible=False),
+            gr.update(visible=True),
+            gr.update(),
+            gr.update(),
+            gr.update(),
+            default_input,
+            gr.update(value=default_prompt, choices=utils.get_available_prompts())
+        ]
+
+
 def set_interface_arguments(extensions, bool_active):
     shared.args.extensions = extensions
 
diff --git a/server.py b/server.py
index b28d30bc50..16f6888b89 100644
--- a/server.py
+++ b/server.py
@@ -145,8 +145,9 @@ def create_interface():
 
         # Text Generation tab
         ui_chat.create_ui()
-        ui_default.create_ui()
-        ui_notebook.create_ui()
+        with gr.Tab("Notebook", elem_id='notebook-parent-tab'):
+            ui_default.create_ui()
+            ui_notebook.create_ui()
 
         ui_parameters.create_ui()  # Parameters tab
         ui_chat.create_character_settings_ui()  # Character tab

From d0befe0729dc49d8988cb14f0f8cc57782aad4f1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Jun 2025 09:22:22 -0700
Subject: [PATCH 0722/1701] Add a comment

---
 server.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/server.py b/server.py
index 16f6888b89..7ce3c208f9 100644
--- a/server.py
+++ b/server.py
@@ -143,8 +143,10 @@ def create_interface():
         # Temporary clipboard for saving files
         shared.gradio['temporary_text'] = gr.Textbox(visible=False)
 
-        # Text Generation tab
+        # Chat tab
         ui_chat.create_ui()
+
+        # Notebook tab
         with gr.Tab("Notebook", elem_id='notebook-parent-tab'):
             ui_default.create_ui()
             ui_notebook.create_ui()

From faae4dc1b08e9b6b4bd01eb4ef5fdf7b60e5c0d2 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Mon, 16 Jun 2025 17:36:05 -0300
Subject: [PATCH 0723/1701] Autosave generated text in the Notebook tab (#7079)

---
 css/main.css                            |   4 +-
 modules/prompts.py                      |  25 +++-
 modules/shared.py                       |   2 +-
 modules/ui_default.py                   | 175 ++++++++++++++++++++++--
 modules/ui_notebook.py                  | 149 ++++++++++++++++++--
 modules/ui_session.py                   |   2 +-
 modules/utils.py                        |   6 +-
 user_data/prompts/Alpaca-with-Input.txt |  10 --
 user_data/prompts/QA.txt                |   4 -
 9 files changed, 325 insertions(+), 52 deletions(-)
 delete mode 100644 user_data/prompts/Alpaca-with-Input.txt
 delete mode 100644 user_data/prompts/QA.txt

diff --git a/css/main.css b/css/main.css
index ee50ac6a8a..d85fe600b2 100644
--- a/css/main.css
+++ b/css/main.css
@@ -167,11 +167,11 @@ gradio-app > :first-child {
 }
 
 .textbox_default textarea {
-    height: calc(100dvh - 201px);
+    height: calc(100dvh - 202px);
 }
 
 .textbox_default_output textarea {
-    height: calc(100dvh - 117px);
+    height: calc(100dvh - 118px);
 }
 
 .textbox textarea {
diff --git a/modules/prompts.py b/modules/prompts.py
index 8f00cac242..79d9b56e46 100644
--- a/modules/prompts.py
+++ b/modules/prompts.py
@@ -1,22 +1,33 @@
 from pathlib import Path
 
+from modules import shared, utils
 from modules.text_generation import get_encoded_length
 
 
 def load_prompt(fname):
-    if fname in ['None', '']:
-        return ''
-    else:
-        file_path = Path(f'user_data/prompts/{fname}.txt')
-        if not file_path.exists():
-            return ''
+    if not fname:
+        # Create new file
+        new_name = utils.current_time()
+        prompt_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
+        prompt_path.parent.mkdir(parents=True, exist_ok=True)
+        initial_content = "In this story,"
+        prompt_path.write_text(initial_content, encoding='utf-8')
+
+        # Update settings to point to new file
+        shared.settings['prompt-notebook'] = new_name
 
+        return initial_content
+
+    file_path = Path(f'user_data/logs/notebook/{fname}.txt')
+    if file_path.exists():
         with open(file_path, 'r', encoding='utf-8') as f:
             text = f.read()
-            if text[-1] == '\n':
+            if len(text) > 0 and text[-1] == '\n':
                 text = text[:-1]
 
             return text
+    else:
+        return ''
 
 
 def count_tokens(text):
diff --git a/modules/shared.py b/modules/shared.py
index 4c9e34a7a7..5333ec4f4d 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -202,7 +202,7 @@
     'chat-instruct_command': 'Continue the chat dialogue below. Write a single reply for the character "<|character|>".\n\n<|prompt|>',
     'enable_web_search': False,
     'web_search_pages': 3,
-    'prompt-notebook': 'QA',
+    'prompt-notebook': '',
     'preset': 'Qwen3 - Thinking' if Path('user_data/presets/Qwen3 - Thinking.yaml').exists() else None,
     'max_new_tokens': 512,
     'max_new_tokens_min': 1,
diff --git a/modules/ui_default.py b/modules/ui_default.py
index e6d2f92d9a..66174e359e 100644
--- a/modules/ui_default.py
+++ b/modules/ui_default.py
@@ -1,3 +1,6 @@
+import time
+from pathlib import Path
+
 import gradio as gr
 
 from modules import logits, shared, ui, utils
@@ -19,7 +22,8 @@ def create_ui():
         with gr.Row():
             with gr.Column():
                 with gr.Row():
-                    shared.gradio['textbox-default'] = gr.Textbox(value=load_prompt(shared.settings['prompt-notebook']), lines=27, label='Input', elem_classes=['textbox_default', 'add_scrollbar'])
+                    initial_text = load_prompt(shared.settings['prompt-notebook'])
+                    shared.gradio['textbox-default'] = gr.Textbox(value=initial_text, lines=27, label='Input', elem_classes=['textbox_default', 'add_scrollbar'])
                     shared.gradio['token-counter-default'] = gr.HTML(value="0", elem_id="default-token-counter")
 
                 with gr.Row():
@@ -30,9 +34,19 @@ def create_ui():
                 with gr.Row():
                     shared.gradio['prompt_menu-default'] = gr.Dropdown(choices=utils.get_available_prompts(), value=shared.settings['prompt-notebook'], label='Prompt', elem_classes='slim-dropdown')
                     ui.create_refresh_button(shared.gradio['prompt_menu-default'], lambda: None, lambda: {'choices': utils.get_available_prompts()}, 'refresh-button', interactive=not mu)
-                    shared.gradio['save_prompt-default'] = gr.Button('💾', elem_classes='refresh-button', interactive=not mu)
+                    shared.gradio['new_prompt-default'] = gr.Button('New', elem_classes='refresh-button', interactive=not mu)
+                    shared.gradio['rename_prompt-default'] = gr.Button('Rename', elem_classes='refresh-button', interactive=not mu)
                     shared.gradio['delete_prompt-default'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu)
 
+                    # Rename elements (initially hidden)
+                    shared.gradio['rename_prompt_to-default'] = gr.Textbox(label="New name", elem_classes=['no-background'], visible=False)
+                    shared.gradio['rename_prompt-cancel-default'] = gr.Button('Cancel', elem_classes=['refresh-button'], visible=False)
+                    shared.gradio['rename_prompt-confirm-default'] = gr.Button('Confirm', elem_classes=['refresh-button'], variant='primary', visible=False)
+
+                    # Delete confirmation elements (initially hidden)
+                    shared.gradio['delete_prompt-cancel-default'] = gr.Button('Cancel', elem_classes=['refresh-button'], visible=False)
+                    shared.gradio['delete_prompt-confirm-default'] = gr.Button('Confirm', variant='stop', elem_classes=['refresh-button'], visible=False)
+
             with gr.Column():
                 with gr.Tab('Raw'):
                     shared.gradio['output_textbox'] = gr.Textbox(lines=27, label='Output', elem_id='textbox-default', elem_classes=['textbox_default_output', 'add_scrollbar'])
@@ -64,7 +78,7 @@ def create_event_handlers():
     shared.gradio['Generate-default'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-default', 'Generate-default')).then(
-        generate_reply_wrapper, gradio(inputs), gradio(outputs), show_progress=False).then(
+        generate_and_save_wrapper, gradio('textbox-default', 'interface_state', 'prompt_menu-default'), gradio(outputs), show_progress=False).then(
         lambda state, left, right: state.update({'textbox-default': left, 'output_textbox': right}), gradio('interface_state', 'textbox-default', 'output_textbox'), None).then(
         lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-default', 'Generate-default')).then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
@@ -72,7 +86,7 @@ def create_event_handlers():
     shared.gradio['textbox-default'].submit(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-default', 'Generate-default')).then(
-        generate_reply_wrapper, gradio(inputs), gradio(outputs), show_progress=False).then(
+        generate_and_save_wrapper, gradio('textbox-default', 'interface_state', 'prompt_menu-default'), gradio(outputs), show_progress=False).then(
         lambda state, left, right: state.update({'textbox-default': left, 'output_textbox': right}), gradio('interface_state', 'textbox-default', 'output_textbox'), None).then(
         lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-default', 'Generate-default')).then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
@@ -80,7 +94,7 @@ def create_event_handlers():
     shared.gradio['Continue-default'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-default', 'Generate-default')).then(
-        generate_reply_wrapper, [shared.gradio['output_textbox']] + gradio(inputs)[1:], gradio(outputs), show_progress=False).then(
+        continue_and_save_wrapper, gradio('output_textbox', 'textbox-default', 'interface_state', 'prompt_menu-default'), gradio(outputs), show_progress=False).then(
         lambda state, left, right: state.update({'textbox-default': left, 'output_textbox': right}), gradio('interface_state', 'textbox-default', 'output_textbox'), None).then(
         lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-default', 'Generate-default')).then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
@@ -88,8 +102,44 @@ def create_event_handlers():
     shared.gradio['Stop-default'].click(stop_everything_event, None, None, queue=False)
     shared.gradio['markdown_render-default'].click(lambda x: x, gradio('output_textbox'), gradio('markdown-default'), queue=False)
     shared.gradio['prompt_menu-default'].change(load_prompt, gradio('prompt_menu-default'), gradio('textbox-default'), show_progress=False)
-    shared.gradio['save_prompt-default'].click(handle_save_prompt, gradio('textbox-default'), gradio('save_contents', 'save_filename', 'save_root', 'file_saver'), show_progress=False)
-    shared.gradio['delete_prompt-default'].click(handle_delete_prompt, gradio('prompt_menu-default'), gradio('delete_filename', 'delete_root', 'file_deleter'), show_progress=False)
+    shared.gradio['new_prompt-default'].click(handle_new_prompt, None, gradio('prompt_menu-default'), show_progress=False)
+
+    shared.gradio['delete_prompt-default'].click(
+        lambda: [gr.update(visible=False), gr.update(visible=True), gr.update(visible=True)],
+        None,
+        gradio('delete_prompt-default', 'delete_prompt-cancel-default', 'delete_prompt-confirm-default'),
+        show_progress=False)
+
+    shared.gradio['delete_prompt-cancel-default'].click(
+        lambda: [gr.update(visible=True), gr.update(visible=False), gr.update(visible=False)],
+        None,
+        gradio('delete_prompt-default', 'delete_prompt-cancel-default', 'delete_prompt-confirm-default'),
+        show_progress=False)
+
+    shared.gradio['delete_prompt-confirm-default'].click(
+        handle_delete_prompt_confirm_default,
+        gradio('prompt_menu-default'),
+        gradio('prompt_menu-default', 'delete_prompt-default', 'delete_prompt-cancel-default', 'delete_prompt-confirm-default'),
+        show_progress=False)
+
+    shared.gradio['rename_prompt-default'].click(
+        handle_rename_prompt_click_default,
+        gradio('prompt_menu-default'),
+        gradio('rename_prompt_to-default', 'rename_prompt-default', 'rename_prompt-cancel-default', 'rename_prompt-confirm-default'),
+        show_progress=False)
+
+    shared.gradio['rename_prompt-cancel-default'].click(
+        lambda: [gr.update(visible=False), gr.update(visible=True), gr.update(visible=False), gr.update(visible=False)],
+        None,
+        gradio('rename_prompt_to-default', 'rename_prompt-default', 'rename_prompt-cancel-default', 'rename_prompt-confirm-default'),
+        show_progress=False)
+
+    shared.gradio['rename_prompt-confirm-default'].click(
+        handle_rename_prompt_confirm_default,
+        gradio('rename_prompt_to-default', 'prompt_menu-default'),
+        gradio('prompt_menu-default', 'rename_prompt_to-default', 'rename_prompt-default', 'rename_prompt-cancel-default', 'rename_prompt-confirm-default'),
+        show_progress=False)
+
     shared.gradio['textbox-default'].change(lambda x: f"{count_tokens(x)}", gradio('textbox-default'), gradio('token-counter-default'), show_progress=False)
     shared.gradio['get_logits-default'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
@@ -98,18 +148,113 @@ def create_event_handlers():
     shared.gradio['get_tokens-default'].click(get_token_ids, gradio('textbox-default'), gradio('tokens-default'), show_progress=False)
 
 
-def handle_save_prompt(text):
+def autosave_prompt(text, prompt_name):
+    """Automatically save the text to the selected prompt file"""
+    if prompt_name and text.strip():
+        prompt_path = Path("user_data/logs/notebook") / f"{prompt_name}.txt"
+        prompt_path.parent.mkdir(parents=True, exist_ok=True)
+        prompt_path.write_text(text, encoding='utf-8')
+
+
+def generate_and_save_wrapper(textbox_content, interface_state, prompt_name):
+    """Generate reply and automatically save the result with periodic saves"""
+    last_save_time = time.monotonic()
+    save_interval = 8
+    output = textbox_content
+
+    # Initial autosave
+    autosave_prompt(output, prompt_name)
+
+    for i, (output, html_output) in enumerate(generate_reply_wrapper(textbox_content, interface_state)):
+        yield output, html_output
+
+        current_time = time.monotonic()
+        # Save on first iteration or if save_interval seconds have passed
+        if i == 0 or (current_time - last_save_time) >= save_interval:
+            autosave_prompt(output, prompt_name)
+            last_save_time = current_time
+
+    # Final autosave
+    autosave_prompt(output, prompt_name)
+
+
+def continue_and_save_wrapper(output_textbox, textbox_content, interface_state, prompt_name):
+    """Continue generation and automatically save the result with periodic saves"""
+    last_save_time = time.monotonic()
+    save_interval = 8
+    output = output_textbox
+
+    # Initial autosave
+    autosave_prompt(output, prompt_name)
+
+    for i, (output, html_output) in enumerate(generate_reply_wrapper(output_textbox, interface_state)):
+        yield output, html_output
+
+        current_time = time.monotonic()
+        # Save on first iteration or if save_interval seconds have passed
+        if i == 0 or (current_time - last_save_time) >= save_interval:
+            autosave_prompt(output, prompt_name)
+            last_save_time = current_time
+
+    # Final autosave
+    autosave_prompt(output, prompt_name)
+
+
+def handle_new_prompt():
+    new_name = utils.current_time()
+
+    # Create the new prompt file
+    prompt_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
+    prompt_path.parent.mkdir(parents=True, exist_ok=True)
+    prompt_path.write_text("In this story,", encoding='utf-8')
+
+    return gr.update(choices=utils.get_available_prompts(), value=new_name)
+
+
+def handle_delete_prompt_confirm_default(prompt_name):
+    available_prompts = utils.get_available_prompts()
+    current_index = available_prompts.index(prompt_name) if prompt_name in available_prompts else 0
+
+    (Path("user_data/logs/notebook") / f"{prompt_name}.txt").unlink(missing_ok=True)
+    available_prompts = utils.get_available_prompts()
+
+    if available_prompts:
+        new_value = available_prompts[min(current_index, len(available_prompts) - 1)]
+    else:
+        new_value = utils.current_time()
+        Path("user_data/logs/notebook").mkdir(parents=True, exist_ok=True)
+        (Path("user_data/logs/notebook") / f"{new_value}.txt").write_text("In this story,")
+        available_prompts = [new_value]
+
     return [
-        text,
-        utils.current_time() + ".txt",
-        "user_data/prompts/",
-        gr.update(visible=True)
+        gr.update(choices=available_prompts, value=new_value),
+        gr.update(visible=True),
+        gr.update(visible=False),
+        gr.update(visible=False)
     ]
 
 
-def handle_delete_prompt(prompt):
+def handle_rename_prompt_click_default(current_name):
     return [
-        prompt + ".txt",
-        "user_data/prompts/",
+        gr.update(value=current_name, visible=True),
+        gr.update(visible=False),
+        gr.update(visible=True),
         gr.update(visible=True)
     ]
+
+
+def handle_rename_prompt_confirm_default(new_name, current_name):
+    old_path = Path("user_data/logs/notebook") / f"{current_name}.txt"
+    new_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
+
+    if old_path.exists() and not new_path.exists():
+        old_path.rename(new_path)
+
+    available_prompts = utils.get_available_prompts()
+    return [
+        gr.update(choices=available_prompts, value=new_name),
+        gr.update(visible=False),
+        gr.update(visible=True),
+        gr.update(visible=False),
+        gr.update(visible=False)
+    ]
diff --git a/modules/ui_notebook.py b/modules/ui_notebook.py
index e6b87342b6..e154a0ba6e 100644
--- a/modules/ui_notebook.py
+++ b/modules/ui_notebook.py
@@ -1,3 +1,6 @@
+import time
+from pathlib import Path
+
 import gradio as gr
 
 from modules import logits, shared, ui, utils
@@ -7,7 +10,7 @@
     get_token_ids,
     stop_everything_event
 )
-from modules.ui_default import handle_delete_prompt, handle_save_prompt
+from modules.ui_default import autosave_prompt
 from modules.utils import gradio
 
 inputs = ('textbox-notebook', 'interface_state')
@@ -22,7 +25,8 @@ def create_ui():
             with gr.Column(scale=4):
                 with gr.Tab('Raw'):
                     with gr.Row():
-                        shared.gradio['textbox-notebook'] = gr.Textbox(value=load_prompt(shared.settings['prompt-notebook']), lines=27, elem_id='textbox-notebook', elem_classes=['textbox', 'add_scrollbar'])
+                        initial_text = load_prompt(shared.settings['prompt-notebook'])
+                        shared.gradio['textbox-notebook'] = gr.Textbox(value=initial_text, lines=27, elem_id='textbox-notebook', elem_classes=['textbox', 'add_scrollbar'])
                         shared.gradio['token-counter-notebook'] = gr.HTML(value="0", elem_id="notebook-token-counter")
 
                 with gr.Tab('Markdown'):
@@ -57,9 +61,19 @@ def create_ui():
                 gr.HTML('
            ')
                 with gr.Row():
                     shared.gradio['prompt_menu-notebook'] = gr.Dropdown(choices=utils.get_available_prompts(), value=shared.settings['prompt-notebook'], label='Prompt', elem_classes='slim-dropdown')
-                    ui.create_refresh_button(shared.gradio['prompt_menu-notebook'], lambda: None, lambda: {'choices': utils.get_available_prompts()}, ['refresh-button', 'refresh-button-small'], interactive=not mu)
-                    shared.gradio['save_prompt-notebook'] = gr.Button('💾', elem_classes=['refresh-button', 'refresh-button-small'], interactive=not mu)
-                    shared.gradio['delete_prompt-notebook'] = gr.Button('🗑️', elem_classes=['refresh-button', 'refresh-button-small'], interactive=not mu)
+
+                with gr.Row():
+                    ui.create_refresh_button(shared.gradio['prompt_menu-notebook'], lambda: None, lambda: {'choices': utils.get_available_prompts()}, ['refresh-button'], interactive=not mu)
+                    shared.gradio['new_prompt-notebook'] = gr.Button('New', elem_classes=['refresh-button'], interactive=not mu)
+                    shared.gradio['rename_prompt-notebook'] = gr.Button('Rename', elem_classes=['refresh-button'], interactive=not mu)
+                    shared.gradio['delete_prompt-notebook'] = gr.Button('🗑️', elem_classes=['refresh-button'], interactive=not mu)
+                    shared.gradio['delete_prompt-confirm-notebook'] = gr.Button('Confirm', variant='stop', elem_classes=['refresh-button'], visible=False)
+                    shared.gradio['delete_prompt-cancel-notebook'] = gr.Button('Cancel', elem_classes=['refresh-button'], visible=False)
+
+                with gr.Row(visible=False) as shared.gradio['rename-row-notebook']:
+                    shared.gradio['rename_prompt_to-notebook'] = gr.Textbox(label="New name", elem_classes=['no-background'])
+                    shared.gradio['rename_prompt-cancel-notebook'] = gr.Button('Cancel', elem_classes=['refresh-button'])
+                    shared.gradio['rename_prompt-confirm-notebook'] = gr.Button('Confirm', elem_classes=['refresh-button'], variant='primary')
 
 
 def create_event_handlers():
@@ -67,7 +81,7 @@ def create_event_handlers():
         lambda x: x, gradio('textbox-notebook'), gradio('last_input-notebook')).then(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-notebook', 'Generate-notebook')).then(
-        generate_reply_wrapper, gradio(inputs), gradio(outputs), show_progress=False).then(
+        generate_and_save_wrapper_notebook, gradio('textbox-notebook', 'interface_state', 'prompt_menu-notebook'), gradio(outputs), show_progress=False).then(
         lambda state, text: state.update({'textbox-notebook': text}), gradio('interface_state', 'textbox-notebook'), None).then(
         lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-notebook', 'Generate-notebook')).then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
@@ -76,7 +90,7 @@ def create_event_handlers():
         lambda x: x, gradio('textbox-notebook'), gradio('last_input-notebook')).then(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-notebook', 'Generate-notebook')).then(
-        generate_reply_wrapper, gradio(inputs), gradio(outputs), show_progress=False).then(
+        generate_and_save_wrapper_notebook, gradio('textbox-notebook', 'interface_state', 'prompt_menu-notebook'), gradio(outputs), show_progress=False).then(
         lambda state, text: state.update({'textbox-notebook': text}), gradio('interface_state', 'textbox-notebook'), None).then(
         lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-notebook', 'Generate-notebook')).then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
@@ -85,7 +99,7 @@ def create_event_handlers():
         lambda x: x, gradio('last_input-notebook'), gradio('textbox-notebook'), show_progress=False).then(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-notebook', 'Generate-notebook')).then(
-        generate_reply_wrapper, gradio(inputs), gradio(outputs), show_progress=False).then(
+        generate_and_save_wrapper_notebook, gradio('textbox-notebook', 'interface_state', 'prompt_menu-notebook'), gradio(outputs), show_progress=False).then(
         lambda state, text: state.update({'textbox-notebook': text}), gradio('interface_state', 'textbox-notebook'), None).then(
         lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-notebook', 'Generate-notebook')).then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
@@ -97,11 +111,126 @@ def create_event_handlers():
     shared.gradio['markdown_render-notebook'].click(lambda x: x, gradio('textbox-notebook'), gradio('markdown-notebook'), queue=False)
     shared.gradio['Stop-notebook'].click(stop_everything_event, None, None, queue=False)
     shared.gradio['prompt_menu-notebook'].change(load_prompt, gradio('prompt_menu-notebook'), gradio('textbox-notebook'), show_progress=False)
-    shared.gradio['save_prompt-notebook'].click(handle_save_prompt, gradio('textbox-notebook'), gradio('save_contents', 'save_filename', 'save_root', 'file_saver'), show_progress=False)
-    shared.gradio['delete_prompt-notebook'].click(handle_delete_prompt, gradio('prompt_menu-notebook'), gradio('delete_filename', 'delete_root', 'file_deleter'), show_progress=False)
+    shared.gradio['new_prompt-notebook'].click(handle_new_prompt, None, gradio('prompt_menu-notebook'), show_progress=False)
+
+    shared.gradio['delete_prompt-notebook'].click(
+        lambda: [gr.update(visible=False), gr.update(visible=True), gr.update(visible=True)],
+        None,
+        gradio('delete_prompt-notebook', 'delete_prompt-cancel-notebook', 'delete_prompt-confirm-notebook'),
+        show_progress=False)
+
+    shared.gradio['delete_prompt-cancel-notebook'].click(
+        lambda: [gr.update(visible=True), gr.update(visible=False), gr.update(visible=False)],
+        None,
+        gradio('delete_prompt-notebook', 'delete_prompt-cancel-notebook', 'delete_prompt-confirm-notebook'),
+        show_progress=False)
+
+    shared.gradio['delete_prompt-confirm-notebook'].click(
+        handle_delete_prompt_confirm_notebook,
+        gradio('prompt_menu-notebook'),
+        gradio('prompt_menu-notebook', 'delete_prompt-notebook', 'delete_prompt-cancel-notebook', 'delete_prompt-confirm-notebook'),
+        show_progress=False)
+
+    shared.gradio['rename_prompt-notebook'].click(
+        handle_rename_prompt_click_notebook,
+        gradio('prompt_menu-notebook'),
+        gradio('rename_prompt_to-notebook', 'rename_prompt-notebook', 'rename-row-notebook'),
+        show_progress=False)
+
+    shared.gradio['rename_prompt-cancel-notebook'].click(
+        lambda: [gr.update(visible=True), gr.update(visible=False)],
+        None,
+        gradio('rename_prompt-notebook', 'rename-row-notebook'),
+        show_progress=False)
+
+    shared.gradio['rename_prompt-confirm-notebook'].click(
+        handle_rename_prompt_confirm_notebook,
+        gradio('rename_prompt_to-notebook', 'prompt_menu-notebook'),
+        gradio('prompt_menu-notebook', 'rename_prompt-notebook', 'rename-row-notebook'),
+        show_progress=False)
+
     shared.gradio['textbox-notebook'].input(lambda x: f"{count_tokens(x)}", gradio('textbox-notebook'), gradio('token-counter-notebook'), show_progress=False)
     shared.gradio['get_logits-notebook'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         logits.get_next_logits, gradio('textbox-notebook', 'interface_state', 'use_samplers-notebook', 'logits-notebook'), gradio('logits-notebook', 'logits-notebook-previous'), show_progress=False)
 
     shared.gradio['get_tokens-notebook'].click(get_token_ids, gradio('textbox-notebook'), gradio('tokens-notebook'), show_progress=False)
+
+
+def generate_and_save_wrapper_notebook(textbox_content, interface_state, prompt_name):
+    """Generate reply and automatically save the result for notebook mode with periodic saves"""
+    last_save_time = time.monotonic()
+    save_interval = 8
+    output = textbox_content
+
+    # Initial autosave
+    autosave_prompt(output, prompt_name)
+
+    for i, (output, html_output) in enumerate(generate_reply_wrapper(textbox_content, interface_state)):
+        yield output, html_output
+
+        current_time = time.monotonic()
+        # Save on first iteration or if save_interval seconds have passed
+        if i == 0 or (current_time - last_save_time) >= save_interval:
+            autosave_prompt(output, prompt_name)
+            last_save_time = current_time
+
+    # Final autosave
+    autosave_prompt(output, prompt_name)
+
+
+def handle_new_prompt():
+    new_name = utils.current_time()
+
+    # Create the new prompt file
+    prompt_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
+    prompt_path.parent.mkdir(parents=True, exist_ok=True)
+    prompt_path.write_text("In this story,", encoding='utf-8')
+
+    return gr.update(choices=utils.get_available_prompts(), value=new_name)
+
+
+def handle_delete_prompt_confirm_notebook(prompt_name):
+    available_prompts = utils.get_available_prompts()
+    current_index = available_prompts.index(prompt_name) if prompt_name in available_prompts else 0
+
+    (Path("user_data/logs/notebook") / f"{prompt_name}.txt").unlink(missing_ok=True)
+    available_prompts = utils.get_available_prompts()
+
+    if available_prompts:
+        new_value = available_prompts[min(current_index, len(available_prompts) - 1)]
+    else:
+        new_value = utils.current_time()
+        Path("user_data/logs/notebook").mkdir(parents=True, exist_ok=True)
+        (Path("user_data/logs/notebook") / f"{new_value}.txt").write_text("In this story,")
+        available_prompts = [new_value]
+
+    return [
+        gr.update(choices=available_prompts, value=new_value),
+        gr.update(visible=True),
+        gr.update(visible=False),
+        gr.update(visible=False)
+    ]
+
+
+def handle_rename_prompt_click_notebook(current_name):
+    return [
+        gr.update(value=current_name),
+        gr.update(visible=False),
+        gr.update(visible=True)
+    ]
+
+
+def handle_rename_prompt_confirm_notebook(new_name, current_name):
+    old_path = Path("user_data/logs/notebook") / f"{current_name}.txt"
+    new_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
+
+    if old_path.exists() and not new_path.exists():
+        old_path.rename(new_path)
+
+    available_prompts = utils.get_available_prompts()
+    return [
+        gr.update(choices=available_prompts, value=new_name),
+        gr.update(visible=True),
+        gr.update(visible=False)
+    ]
diff --git a/modules/ui_session.py b/modules/ui_session.py
index c0a34228cf..431aeeb02f 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -77,7 +77,7 @@ def handle_default_to_notebook_change(show_two_columns, default_input, default_o
             gr.update(),
             gr.update(),
             gr.update(),
-            default_input,
+            default_output,
             gr.update(value=default_prompt, choices=utils.get_available_prompts())
         ]
 
diff --git a/modules/utils.py b/modules/utils.py
index 218735415e..da5ac9780d 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -159,10 +159,12 @@ def get_available_presets():
 
 
 def get_available_prompts():
-    prompt_files = list(Path('user_data/prompts').glob('*.txt'))
+    notebook_dir = Path('user_data/logs/notebook')
+    notebook_dir.mkdir(parents=True, exist_ok=True)
+
+    prompt_files = list(notebook_dir.glob('*.txt'))
     sorted_files = sorted(prompt_files, key=lambda x: x.stat().st_mtime, reverse=True)
     prompts = [file.stem for file in sorted_files]
-    prompts.append('None')
     return prompts
 
 
diff --git a/user_data/prompts/Alpaca-with-Input.txt b/user_data/prompts/Alpaca-with-Input.txt
deleted file mode 100644
index 56df0e285b..0000000000
--- a/user_data/prompts/Alpaca-with-Input.txt
+++ /dev/null
@@ -1,10 +0,0 @@
-Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.
-
-### Instruction:
-Instruction
-
-### Input:
-Input
-
-### Response:
-
diff --git a/user_data/prompts/QA.txt b/user_data/prompts/QA.txt
deleted file mode 100644
index 32b0e2350f..0000000000
--- a/user_data/prompts/QA.txt
+++ /dev/null
@@ -1,4 +0,0 @@
-Common sense questions and answers
-
-Question: 
-Factual answer:

From 97a539cab690254ac5b51bdf8e7afb181b7992e2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Jun 2025 13:55:45 -0700
Subject: [PATCH 0724/1701] Minor style change

---
 css/main.css | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/css/main.css b/css/main.css
index d85fe600b2..3e51e920a5 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1648,17 +1648,21 @@ button:focus {
 
 #character-context textarea {
     height: calc((100vh - 350px) * 2/3) !important;
+    min-height: 90px !important;
 }
 
 #character-greeting textarea {
     height: calc((100vh - 350px) * 1/3) !important;
+    min-height: 90px !important;
 }
 
 #user-description textarea {
     height: calc(100vh - 231px) !important;
+    min-height: 90px !important;
 }
 
 #instruction-template-str textarea,
 #chat-template-str textarea {
     height: calc(100vh - 300px) !important;
+    min-height: 90px !important;
 }

From 88c020435716ebe4453f1b393cc7b3fc72dbf231 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Jun 2025 14:53:05 -0700
Subject: [PATCH 0725/1701] Disable start_with when generating the websearch
 query

---
 modules/chat.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/chat.py b/modules/chat.py
index 81e7b16a82..6b7722b7e2 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -614,6 +614,7 @@ def generate_search_query(user_message, state):
     search_state['max_new_tokens'] = 64
     search_state['auto_max_new_tokens'] = False
     search_state['enable_thinking'] = False
+    search_state['start_with'] = ""
 
     # Generate the full prompt using existing history + augmented message
     formatted_prompt = generate_chat_prompt(augmented_message, search_state)

From 26eda537f0d1689f1c989b6add8c434530bcfada Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Jun 2025 17:48:23 -0700
Subject: [PATCH 0726/1701] Add auto-save for notebook textbox while typing

---
 modules/ui_notebook.py | 51 +++++++++++++++++++++++++++++++++++++++---
 1 file changed, 48 insertions(+), 3 deletions(-)

diff --git a/modules/ui_notebook.py b/modules/ui_notebook.py
index e154a0ba6e..3a8c9df2f4 100644
--- a/modules/ui_notebook.py
+++ b/modules/ui_notebook.py
@@ -1,3 +1,4 @@
+import threading
 import time
 from pathlib import Path
 
@@ -13,6 +14,11 @@
 from modules.ui_default import autosave_prompt
 from modules.utils import gradio
 
+_notebook_file_lock = threading.Lock()
+_notebook_auto_save_timer = None
+_last_notebook_text = None
+_last_notebook_prompt = None
+
 inputs = ('textbox-notebook', 'interface_state')
 outputs = ('textbox-notebook', 'html-notebook')
 
@@ -150,6 +156,13 @@ def create_event_handlers():
         show_progress=False)
 
     shared.gradio['textbox-notebook'].input(lambda x: f"{count_tokens(x)}", gradio('textbox-notebook'), gradio('token-counter-notebook'), show_progress=False)
+    shared.gradio['textbox-notebook'].change(
+        store_notebook_state_and_debounce,
+        gradio('textbox-notebook', 'prompt_menu-notebook'),
+        None,
+        show_progress=False
+    )
+
     shared.gradio['get_logits-notebook'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         logits.get_next_logits, gradio('textbox-notebook', 'interface_state', 'use_samplers-notebook', 'logits-notebook'), gradio('logits-notebook', 'logits-notebook-previous'), show_progress=False)
@@ -164,7 +177,7 @@ def generate_and_save_wrapper_notebook(textbox_content, interface_state, prompt_
     output = textbox_content
 
     # Initial autosave
-    autosave_prompt(output, prompt_name)
+    safe_autosave_prompt(output, prompt_name)
 
     for i, (output, html_output) in enumerate(generate_reply_wrapper(textbox_content, interface_state)):
         yield output, html_output
@@ -172,11 +185,11 @@ def generate_and_save_wrapper_notebook(textbox_content, interface_state, prompt_
         current_time = time.monotonic()
         # Save on first iteration or if save_interval seconds have passed
         if i == 0 or (current_time - last_save_time) >= save_interval:
-            autosave_prompt(output, prompt_name)
+            safe_autosave_prompt(output, prompt_name)
             last_save_time = current_time
 
     # Final autosave
-    autosave_prompt(output, prompt_name)
+    safe_autosave_prompt(output, prompt_name)
 
 
 def handle_new_prompt():
@@ -234,3 +247,35 @@ def handle_rename_prompt_confirm_notebook(new_name, current_name):
         gr.update(visible=True),
         gr.update(visible=False)
     ]
+
+
+def safe_autosave_prompt(content, prompt_name):
+    """Thread-safe wrapper for autosave_prompt to prevent file corruption"""
+    with _notebook_file_lock:
+        autosave_prompt(content, prompt_name)
+
+
+def store_notebook_state_and_debounce(text, prompt_name):
+    """Store current notebook state and trigger debounced save"""
+    global _notebook_auto_save_timer, _last_notebook_text, _last_notebook_prompt
+
+    if shared.args.multi_user:
+        return
+
+    _last_notebook_text = text
+    _last_notebook_prompt = prompt_name
+
+    if _notebook_auto_save_timer is not None:
+        _notebook_auto_save_timer.cancel()
+
+    _notebook_auto_save_timer = threading.Timer(1.0, _perform_notebook_debounced_save)
+    _notebook_auto_save_timer.start()
+
+
+def _perform_notebook_debounced_save():
+    """Actually perform the notebook save using the stored state"""
+    try:
+        if _last_notebook_text is not None and _last_notebook_prompt is not None:
+            safe_autosave_prompt(_last_notebook_text, _last_notebook_prompt)
+    except Exception as e:
+        print(f"Notebook auto-save failed: {e}")

From be3d371290ed68010bc07fdf424270a270a3436b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Jun 2025 18:42:17 -0700
Subject: [PATCH 0727/1701] Close the big profile picture when switching to
 instruct mode

---
 modules/ui_chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index d1447286c2..da634f072c 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -299,7 +299,7 @@ def create_event_handlers():
     shared.gradio['mode'].change(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_mode_change, gradio('interface_state'), gradio('history', 'display', 'chat_style', 'chat-instruct_command', 'unique_id'), show_progress=False).then(
-        None, gradio('mode'), None, js="(mode) => {const characterContainer = document.getElementById('character-menu').parentNode.parentNode; const isInChatTab = document.querySelector('#chat-controls').contains(characterContainer); if (isInChatTab) { characterContainer.style.display = mode === 'instruct' ? 'none' : ''; }}")
+        None, gradio('mode'), None, js="(mode) => {const characterContainer = document.getElementById('character-menu').parentNode.parentNode; const isInChatTab = document.querySelector('#chat-controls').contains(characterContainer); if (isInChatTab) { characterContainer.style.display = mode === 'instruct' ? 'none' : ''; } if (mode === 'instruct') document.querySelectorAll('.bigProfilePicture').forEach(el => el.remove());}")
 
     shared.gradio['chat_style'].change(chat.redraw_html, gradio(reload_arr), gradio('display'), show_progress=False)
 

From 66e991841a2e318cf88deeec4294f8df438ac80b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Jun 2025 18:45:44 -0700
Subject: [PATCH 0728/1701] Fix the character pfp not appearing when switching
 from instruct to chat modes

---
 modules/chat.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index 6b7722b7e2..d76835665f 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1849,6 +1849,11 @@ def handle_character_picture_change(picture):
 def handle_mode_change(state):
     history, loaded_unique_id = load_latest_history(state)
     histories = find_all_histories_with_first_prompts(state)
+
+    # Ensure character picture cache exists
+    if state['mode'] in ['chat', 'chat-instruct'] and state['character_menu'] and state['character_menu'] != 'None':
+        generate_pfp_cache(state['character_menu'])
+
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
 
     convert_to_markdown.cache_clear()

From 8f49e6144e39eda43513d981bdfb42012bb2408a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Jun 2025 21:06:06 -0700
Subject: [PATCH 0729/1701] Update README

---
 README.md | 18 +++++++++++-------
 1 file changed, 11 insertions(+), 7 deletions(-)

diff --git a/README.md b/README.md
index 9fe4fcd4b3..a73adaa0eb 100644
--- a/README.md
+++ b/README.md
@@ -30,14 +30,18 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 
 ## How to install
 
-#### Option 1: Portable builds (start here)
+#### Option 1: Portable builds (get started in 1 minute)
 
-No installation needed – just unzip and run. Compatible with GGUF (llama.cpp) models on Windows, Linux, and macOS.
+No installation needed – just download, unzip and run. All dependencies included.
 
-Download from: https://github.com/oobabooga/text-generation-webui/releases
+Compatible with GGUF (llama.cpp) models on Windows, Linux, and macOS.
+
+Download from here: https://github.com/oobabooga/text-generation-webui/releases
 
 #### Option 2: One-click installer
 
+For users who need additional backends (ExLlamaV3, Transformers) or extensions (TTS, voice input, translation, etc). Requires ~10GB disk space and downloads PyTorch.
+
 1. Clone the repository, or [download its source code](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip) and extract it.
 2. Run the startup script for your OS: `start_windows.bat`, `start_linux.sh`, or `start_macos.sh`.
 3. When prompted, select your GPU vendor.
@@ -150,21 +154,21 @@ The `requirements*.txt` above contain various wheels precompiled through GitHub
 ```
 For NVIDIA GPU:
 ln -s docker/{nvidia/Dockerfile,nvidia/docker-compose.yml,.dockerignore} .
-For AMD GPU: 
+For AMD GPU:
 ln -s docker/{amd/Dockerfile,amd/docker-compose.yml,.dockerignore} .
 For Intel GPU:
 ln -s docker/{intel/Dockerfile,amd/docker-compose.yml,.dockerignore} .
 For CPU only
 ln -s docker/{cpu/Dockerfile,cpu/docker-compose.yml,.dockerignore} .
 cp docker/.env.example .env
-#Create logs/cache dir : 
+#Create logs/cache dir :
 mkdir -p user_data/logs user_data/cache
-# Edit .env and set: 
+# Edit .env and set:
 #   TORCH_CUDA_ARCH_LIST based on your GPU model
 #   APP_RUNTIME_GID      your host user's group id (run `id -g` in a terminal)
 #   BUILD_EXTENIONS      optionally add comma separated list of extensions to build
 # Edit user_data/CMD_FLAGS.txt and add in it the options you want to execute (like --listen --cpu)
-# 
+#
 docker compose up --build
 ```
 

From 8689d7ecea8f10fb08f082f50eb1817a7494157d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Jun 2025 21:21:39 -0700
Subject: [PATCH 0730/1701] Update README

---
 README.md | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/README.md b/README.md
index a73adaa0eb..6e7c05b18e 100644
--- a/README.md
+++ b/README.md
@@ -192,7 +192,7 @@ List of command-line flags
 
            
 
 ```txt
-usage: server.py [-h] [--multi-user] [--character CHARACTER] [--model MODEL] [--lora LORA [LORA ...]] [--model-dir MODEL_DIR] [--lora-dir LORA_DIR] [--model-menu] [--settings SETTINGS]
+usage: server.py [-h] [--multi-user] [--model MODEL] [--lora LORA [LORA ...]] [--model-dir MODEL_DIR] [--lora-dir LORA_DIR] [--model-menu] [--settings SETTINGS]
                  [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR]
                  [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--use_flash_attention_2] [--use_eager_attention] [--torch-compile] [--load-in-4bit]
                  [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn] [--threads THREADS] [--threads-batch THREADS_BATCH] [--batch-size BATCH_SIZE] [--no-mmap]
@@ -211,7 +211,6 @@ options:
 
 Basic settings:
   --multi-user                              Multi-user mode. Chat histories are not saved or automatically loaded. Warning: this is likely not safe for sharing publicly.
-  --character CHARACTER                     The name of the character to load in chat mode by default.
   --model MODEL                             Name of the model to load by default.
   --lora LORA [LORA ...]                    The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.
   --model-dir MODEL_DIR                     Path to directory with all the models.

From 0d1597616ff9be86d70a5eae49b021f65fc4457f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Jun 2025 07:09:33 -0700
Subject: [PATCH 0731/1701] Safer usage of mkdir across the project

---
 download-model.py                           |   4 +-
 extensions/Training_PRO/custom_scheduler.py | 116 +++++++-------
 extensions/Training_PRO/matplotgraph.py     |   6 +-
 extensions/Training_PRO/script.py           | 161 ++++++++++----------
 extensions/Training_PRO/train_utils.py      |  96 ++++++------
 extensions/coqui_tts/script.py              |   3 +-
 extensions/sd_api_pictures/script.py        |   3 +-
 modules/chat.py                             |  12 +-
 modules/evaluate.py                         |   4 +-
 modules/prompts.py                          |   4 +-
 modules/training.py                         |   7 +-
 modules/ui.py                               |   4 +-
 modules/ui_default.py                       |  12 +-
 modules/ui_notebook.py                      |   8 +-
 modules/utils.py                            |   3 +-
 server.py                                   |   3 +-
 16 files changed, 240 insertions(+), 206 deletions(-)

diff --git a/download-model.py b/download-model.py
index 576a8b7910..0c6307a1cd 100644
--- a/download-model.py
+++ b/download-model.py
@@ -314,8 +314,8 @@ def start_download_threads(self, file_list, output_folder, start_from_scratch=Fa
 
     def download_model_files(self, model, branch, links, sha256, output_folder, progress_queue=None, start_from_scratch=False, threads=4, specific_file=None, is_llamacpp=False):
         self.progress_queue = progress_queue
-
-        output_folder.mkdir(parents=True, exist_ok=True)
+        if not output_folder.exists():
+            output_folder.mkdir(parents=True, exist_ok=True)
 
         if not is_llamacpp:
             metadata = f'url: https://huggingface.co/{model}\n' \
diff --git a/extensions/Training_PRO/custom_scheduler.py b/extensions/Training_PRO/custom_scheduler.py
index 1e80daed1a..59e7706ed3 100644
--- a/extensions/Training_PRO/custom_scheduler.py
+++ b/extensions/Training_PRO/custom_scheduler.py
@@ -20,7 +20,7 @@
 
 def custom_scheduler_global_update(current_loss: float):
     custom_scheduler_params.update({'current_loss': current_loss})
-  
+
 def custom_scheduler_global_setup(trigger_loss: float, ramp_down_ratio: float):
     custom_scheduler_params.update({'trigger_loss': trigger_loss})
     custom_scheduler_params.update({'ramp_down_ratio': ramp_down_ratio})
@@ -35,12 +35,12 @@ def custom_scheduler_global_setup(trigger_loss: float, ramp_down_ratio: float):
 
 # hold constant to the half of epochs then cosine down to 0
 def _get_fp_half_schedule_with_warmup_lr_lambda(current_step: int, *, num_warmup_steps: int, num_training_steps: int, num_firstepoch_steps: int):
-    
+
     global last_print_label
     print_label = ''
 
     half_steps = num_training_steps//2
-    
+
     num_warmup_steps = min(num_warmup_steps,half_steps)
 
     if current_step < num_warmup_steps:
@@ -49,57 +49,57 @@ def _get_fp_half_schedule_with_warmup_lr_lambda(current_step: int, *, num_warmup
         print_label = 'Scheduler: Hold'
     else:
         print_label = 'Scheduler: Annealing'
-    
+
     if print_label != last_print_label:
         print(print_label)
-    
+
     last_print_label = print_label
 
     if current_step < num_warmup_steps:
         return float(current_step) / float(max(1, num_warmup_steps))
-    
+
     if current_step < half_steps:
-        return 1.0 
-    
+        return 1.0
+
     progress = float(current_step - half_steps) / float(max(1, num_training_steps - half_steps))
     num_cycles = 0.5
-    return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))    
- 
+    return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))
+
 
 # raise up in cosine, then fall back in cosine
 def _get_fp_cosine_raise_and_fall_lr_lambda(current_step: int, *, num_warmup_steps: int, num_training_steps: int, num_firstepoch_steps: int):
-    
+
     global last_print_label
     print_label = ''
 
     half_steps = num_training_steps//2
-    
+
     #num_warmup_steps = min(num_warmup_steps,half_steps)
 
     if current_step < half_steps:
         print_label = 'Scheduler: Raise'
     else:
         print_label = 'Scheduler: Fall'
-    
+
     if print_label != last_print_label:
         print(print_label)
-    
+
     last_print_label = print_label
 
-    
+
     # linear
     #    return float(current_step) / float(max(1, num_warmup_steps))
-    
+
     progress = float(current_step - half_steps) / float(max(1, num_training_steps - half_steps))
     num_cycles = 0.5
-    return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))    
- 
+    return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))
+
 # constant to the first epochs then cosine down to 0 over the rest epochs
 def _get_fp_cosine_schedule_with_warmup_lr_lambda(current_step: int, *, num_warmup_steps: int, num_training_steps: int, num_firstepoch_steps: int):
-    
+
     global last_print_label
     print_label = ''
-    
+
     num_warmup_steps = min(num_warmup_steps,num_firstepoch_steps)
 
     if current_step < num_warmup_steps:
@@ -108,56 +108,56 @@ def _get_fp_cosine_schedule_with_warmup_lr_lambda(current_step: int, *, num_warm
         print_label = 'Scheduler: Hold'
     else:
         print_label = 'Scheduler: Annealing'
-    
+
     if print_label != last_print_label:
         print(print_label)
-    
+
     last_print_label = print_label
 
     if current_step < num_warmup_steps:
         return float(current_step) / float(max(1, num_warmup_steps))
-    
+
     if current_step < num_firstepoch_steps:
-        return 1.0 
-    
+        return 1.0
+
     progress = float(current_step - num_firstepoch_steps) / float(max(1, num_training_steps - num_firstepoch_steps))
     num_cycles = 0.5
-    return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))    
-    
-# halve lr each epoch   
+    return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))
+
+# halve lr each epoch
 
 def _get_fp_cdrop_rate_schedule_with_warmup_lr_lambda(current_step: int, *, num_warmup_steps: int, num_training_steps: int, num_firstepoch_steps: int):
-    
+
     global last_print_label
     print_label = ''
-    
+
     num_warmup_steps = min(num_warmup_steps, num_firstepoch_steps)
 
     current_epoch = (current_step // num_firstepoch_steps) + 1
-    
-    
+
+
     if current_step < num_warmup_steps:
         print_label = 'Scheduler: Warmup'
     elif current_step < num_firstepoch_steps:
         print_label = 'Scheduler: Hold'
     else:
         print_label = 'Scheduler: Drop Rate'
-    
+
     if print_label != last_print_label:
         print(print_label)
-    
+
     last_print_label = print_label
 
     if current_step < num_warmup_steps:
         return float(current_step) / float(max(1, num_warmup_steps))
-    
+
     if current_step < num_firstepoch_steps:
-        return 1.0 
+        return 1.0
 
     # Compute the learning rate for the annealing phase
-    
+
     learning_rate = 1.0 / float(2 ** (current_epoch - 1))
-   
+
     return learning_rate
 
 # epoch decay: 1/(1 + decay * epoch)
@@ -177,7 +177,7 @@ def custom_cosine_scheduler_with_warmup(optimizer, num_warmup_steps, num_trainin
     Return:
         `torch.optim.lr_scheduler.LambdaLR` with the appropriate schedule.
     """
-    
+
     lr_lambda = partial(
         _get_fp_cosine_schedule_with_warmup_lr_lambda,
         num_warmup_steps=num_warmup_steps,
@@ -201,7 +201,7 @@ def custom_half_scheduler_with_warmup(optimizer, num_warmup_steps, num_training_
     Return:
         `torch.optim.lr_scheduler.LambdaLR` with the appropriate schedule.
     """
-    
+
     lr_lambda = partial(
         _get_fp_half_schedule_with_warmup_lr_lambda,
         num_warmup_steps=num_warmup_steps,
@@ -225,7 +225,7 @@ def custom_raise_fall_scheduler_with_warmup(optimizer, num_warmup_steps, num_tra
     Return:
         `torch.optim.lr_scheduler.LambdaLR` with the appropriate schedule.
     """
-    
+
     lr_lambda = partial(
         _get_fp_cosine_raise_and_fall_lr_lambda,
         num_warmup_steps=num_warmup_steps,
@@ -257,7 +257,7 @@ def neftune_forward(self, input: torch.Tensor):
         mag_norm = self.neftune_noise_alpha / torch.sqrt(dims)
         embeddings = embeddings + torch.zeros_like(embeddings).uniform_(-mag_norm, mag_norm)
 
-    return embeddings    
+    return embeddings
 
 
 class FPNEFtuneTrainer(transformers.Trainer):
@@ -267,7 +267,7 @@ def __init__(self,neftune_noise_alpha:float = 0.0, model = None, *args, **kwargs
             model = self._activate_neftune(model)
         super().__init__(model = model, *args, **kwargs)
 
-   
+
     def _activate_neftune(self, model):
         r"""
         Activates the neftune as presented in this code: https://github.com/neelsjain/NEFTune and paper: https://arxiv.org/abs/2310.05914
@@ -290,7 +290,7 @@ def _activate_neftune(self, model):
         embeddings._trl_old_forward = old_forward
 
         return model
-    
+
     def train(self, *args, **kwargs):
         output = super().train(*args, **kwargs)
 
@@ -318,7 +318,7 @@ def __init__(self,neftune_noise_alpha:float = 0.0, model = None, *args, **kwargs
             model = self._activate_neftune(model)
         super().__init__(model = model, *args, **kwargs)
 
-   
+
     def _activate_neftune(self, model):
         r"""
         Activates the neftune as presented in this code: https://github.com/neelsjain/NEFTune and paper: https://arxiv.org/abs/2310.05914
@@ -341,7 +341,7 @@ def _activate_neftune(self, model):
         embeddings._trl_old_forward = old_forward
 
         return model
-    
+
     def train(self, *args, **kwargs):
         output = super().train(*args, **kwargs)
 
@@ -364,19 +364,19 @@ def train(self, *args, **kwargs):
 
     def create_scheduler(self, num_training_steps: int, optimizer: torch.optim.Optimizer = None):
         #Setup the scheduler. The optimizer of the trainer must have been set up either before this method is called or passed as an argument.
-        
+
         num_train_epochs = self.args.num_train_epochs
         num_warmup_steps=self.args.get_warmup_steps(num_training_steps)
         num_firstepoch_steps = math.ceil(num_training_steps/num_train_epochs)
-        num_warmup_acc = num_warmup_steps*self.args.gradient_accumulation_steps 
+        num_warmup_acc = num_warmup_steps*self.args.gradient_accumulation_steps
         num_firstepoch_steps_acc = num_firstepoch_steps*self.args.gradient_accumulation_steps
         num_training_steps_acc = num_training_steps*self.args.gradient_accumulation_steps
 
         custom_scheduler_params.update({'dynamic_scheduler_stop': False})
- 
+
         print (f"Warm-up steps aligned to Gradient accumulation ({self.args.gradient_accumulation_steps}) = {num_warmup_acc} actual warmup steps")
         if self.args.lr_scheduler_type == 'cosine':
-            
+
             num_warmup_acc_min = min(num_warmup_acc, num_firstepoch_steps_acc)
 
             if num_warmup_acc>num_firstepoch_steps_acc:
@@ -388,13 +388,13 @@ def create_scheduler(self, num_training_steps: int, optimizer: torch.optim.Optim
             self.lr_scheduler = custom_cosine_scheduler_with_warmup(
                     optimizer=self.optimizer if optimizer is None else optimizer,
                     num_warmup_steps=num_warmup_steps,
-                    num_training_steps=num_training_steps, 
+                    num_training_steps=num_training_steps,
                     num_firstepoch_steps = num_firstepoch_steps,
                 )
             self._created_lr_scheduler = True
             return self.lr_scheduler
         elif self.args.lr_scheduler_type == 'constant':
-           
+
             half_step_acc = num_training_steps_acc//2
             num_warmup_acc_min = min(num_warmup_acc, half_step_acc)
 
@@ -407,15 +407,15 @@ def create_scheduler(self, num_training_steps: int, optimizer: torch.optim.Optim
             self.lr_scheduler = custom_half_scheduler_with_warmup(
                     optimizer=self.optimizer if optimizer is None else optimizer,
                     num_warmup_steps=num_warmup_steps,
-                    num_training_steps=num_training_steps, 
+                    num_training_steps=num_training_steps,
                     num_firstepoch_steps = num_firstepoch_steps,
                 )
             self._created_lr_scheduler = True
             return self.lr_scheduler
         elif self.args.lr_scheduler_type == 'constant_with_warmup':
-           
+
             half_step_acc = num_training_steps_acc//2
-            
+
             if num_warmup_steps>0:
                 print(f"Warmup doesn't apply to this scheduler [Raise-Fall]")
 
@@ -424,10 +424,10 @@ def create_scheduler(self, num_training_steps: int, optimizer: torch.optim.Optim
             self.lr_scheduler = custom_raise_fall_scheduler_with_warmup(
                     optimizer=self.optimizer if optimizer is None else optimizer,
                     num_warmup_steps=num_warmup_steps,
-                    num_training_steps=num_training_steps, 
+                    num_training_steps=num_training_steps,
                     num_firstepoch_steps = num_firstepoch_steps,
                 )
             self._created_lr_scheduler = True
-            return self.lr_scheduler        
+            return self.lr_scheduler
         else:
-            return  super().create_scheduler(num_training_steps=num_training_steps, optimizer=optimizer)
\ No newline at end of file
+            return  super().create_scheduler(num_training_steps=num_training_steps, optimizer=optimizer)
diff --git a/extensions/Training_PRO/matplotgraph.py b/extensions/Training_PRO/matplotgraph.py
index b30bee8365..10199ab69a 100644
--- a/extensions/Training_PRO/matplotgraph.py
+++ b/extensions/Training_PRO/matplotgraph.py
@@ -5,7 +5,7 @@ def create_graph(lora_path, lora_name):
     try:
         import matplotlib.pyplot as plt
         from matplotlib.ticker import ScalarFormatter
-        
+
         peft_model_path = f'{lora_path}/training_graph.json'
         image_model_path = f'{lora_path}/training_graph.png'
         # Check if the JSON file exists
@@ -20,7 +20,7 @@ def create_graph(lora_path, lora_name):
 
             # Create the line chart
             fig, ax1 = plt.subplots(figsize=(10, 6))
-        
+
 
             # Plot y1 (learning rate) on the first y-axis
             ax1.plot(x, y1, 'b-', label='Learning Rate')
@@ -57,6 +57,6 @@ def create_graph(lora_path, lora_name):
             print(f"Graph saved in {image_model_path}")
         else:
             print(f"File 'training_graph.json' does not exist in the {lora_path}")
-      
+
     except ImportError:
         print("matplotlib is not installed. Please install matplotlib to create PNG graphs")
diff --git a/extensions/Training_PRO/script.py b/extensions/Training_PRO/script.py
index cb11a8df0a..05a879f6cc 100644
--- a/extensions/Training_PRO/script.py
+++ b/extensions/Training_PRO/script.py
@@ -109,12 +109,12 @@ def ui():
                             copy_from = gr.Dropdown(label='Copy parameters from', value='None', choices=get_available_loras_local(non_serialized_params['Lora_sortedByTime']), elem_classes=['slim-dropdown'])
                             create_refresh_button(copy_from, lambda: None, lambda: {'choices': get_available_loras_local(non_serialized_params['Lora_sortedByTime'])}, 'refresh-button')
                     with gr.Column():
-                        sort_byTime = gr.Checkbox(label='Sort list by Date', value=False, info='Sorts Loras by date created.', elem_classes=['no-background'])                        
+                        sort_byTime = gr.Checkbox(label='Sort list by Date', value=False, info='Sorts Loras by date created.', elem_classes=['no-background'])
 
                 with gr.Row():
                     with gr.Column(scale=5):
                         lora_name = gr.Textbox(label='Name', info='The name of your new LoRA file')
-    
+
                     with gr.Column():
                         always_override = gr.Checkbox(label='Override Existing Files', value=False, info='If the name is the same, checking will replace the existing file, and unchecking will load and continue from it (the rank must be the same).', elem_classes=['no-background'])
 
@@ -132,14 +132,14 @@ def ui():
                         epochs = gr.Number(label='Epochs', value=3, info='Number of times every entry in the dataset should be fed into training. So 1 means feed each item in once, 5 means feed it in five times, etc.')
                         learning_rate = gr.Textbox(label='Learning Rate', value='3e-4', info='In scientific notation. 3e-4 is a good starting base point. 1e-2 is extremely high, 1e-6 is extremely low.')
                         lr_scheduler_type = gr.Dropdown(label='LR Scheduler', value='linear', choices=['linear', 'constant', 'constant_with_warmup', 'cosine', 'cosine_with_restarts', 'polynomial', 'inverse_sqrt', 'FP_low_epoch_annealing', 'FP_half_time_annealing','FP_raise_fall_creative'], info='Learning rate scheduler - defines how the learning rate changes over time. Custom schedulers: FP_low_epoch_annealing, FP_half_time_annealing, FP_raise_fall_creative (see README)', elem_classes=['slim-dropdown'])
-                        
+
                 with gr.Accordion(label='Checkpoints', open=True):
                     with gr.Row():
                         with gr.Column():
                             save_steps = gr.Number(label='Save every n steps', value=0, info='A checkpoint will be saved every n steps and at each Epoch boundary. (0 = OFF)')
-                        with gr.Column():    
-                            save_steps_under_loss = gr.Slider(label='Save at 10% Loss change', value=1.8, minimum=0.0, maximum=3.0, step=0.1, info="Saves checkpoints at (or bellow) this loss and then each time loss falls by at least 10% This works independently from 'Save every n steps'")    
-                    with gr.Row():        
+                        with gr.Column():
+                            save_steps_under_loss = gr.Slider(label='Save at 10% Loss change', value=1.8, minimum=0.0, maximum=3.0, step=0.1, info="Saves checkpoints at (or bellow) this loss and then each time loss falls by at least 10% This works independently from 'Save every n steps'")
+                    with gr.Row():
                         save_chackpoint_now = gr.Button('Queue Checkpoint Now')
 
                 with gr.Accordion(label='Advanced Options', open=True):
@@ -148,7 +148,7 @@ def ui():
                             warmup_steps = gr.Number(label='Warmup Steps', value=100, info='Number of max steps used for a linear warmup. Reduces early over-fitting by the first training blocks. Value has precedent over Warmup Ratio. Aligns to the closest multiple of graddient accumulation')
                             warmup_ratio = gr.Slider(label='Warmup Ratio', minimum=0.0, maximum=0.2, step=0.025, value=0.0, info='Ratio of total training steps that will be used for a linear warmup. It applies only if Warmup Step is 0.')
                             neft_noise_alpha = gr.Slider(label='NEFtune noise scale', minimum=0.0, maximum=15, step=1, value=0.0, info='Add noise to the training to improve generalization. [0 - OFF, Starting value to experiment: 5]')
-                            training_projection = gr.Radio(value = train_choices[4], label='LLaMA Target Projections', info='Change the targets (LORA is typically q-v)', choices=train_choices)    
+                            training_projection = gr.Radio(value = train_choices[4], label='LLaMA Target Projections', info='Change the targets (LORA is typically q-v)', choices=train_choices)
                             lora_dropout = gr.Slider(label='LoRA Dropout', minimum=0.0, maximum=1.0, step=0.025, value=0.05, info='Percentage probability for dropout of LoRA layers. This can help reduce overfitting. Most users should leave at default.')
                             optimizer = gr.Dropdown(label='Optimizer', value='adamw_torch', choices=['adamw_hf', 'adamw_torch', 'adamw_torch_fused', 'adamw_torch_xla', 'adamw_apex_fused', 'adafactor', 'adamw_bnb_8bit', 'adamw_anyprecision', 'sgd', 'adagrad'], info='Different optimizer implementation options, for advanced users. Effects of different options are not well documented yet.', elem_classes=['slim-dropdown'])
 
@@ -157,10 +157,10 @@ def ui():
                             add_bos_token = gr.Checkbox(label='Add BOS token', value=True, info="Adds BOS token for each dataset item")
                             add_eos_token = gr.Checkbox(label='Add EOS token', value=False, info="Adds EOS token for each dataset item")
                             add_eos_token_type = gr.Dropdown(label='EOS placement (Text file)', choices=['Every Block', 'Hard Cut Blocks Only'], value='Every Block', info='', allow_custom_value = False)
-                            
+
                             higher_rank_limit = gr.Checkbox(label='Enable higher ranks', value=False, info='If checked, changes Rank/Alpha slider above to go much higher. This will not work without a datacenter-class GPU.')
                             report_to = gr.Radio(label="Save detailed logs with", value="None", choices=["None", "wandb", "tensorboard"], interactive=True)
-                # for future            
+                # for future
                 #with gr.Accordion(label='Dynamic Scheduler', open = False):
                 #    ds_min_epochs = gr.Number(label='Minimum Epochs', value='1', info='Minimum epochs that will be always performed before ramp down can be triggered')
                 #    ds_max_epochs = gr.Number(label='Maximum Epochs (fallback)', value='50', info='Maximum Epochs before the training will bail out completely (should be a large number)')
@@ -168,7 +168,7 @@ def ui():
                 #    ds_loss_rolling_window = gr.Number(label='Loss rolling average', value='4', info='Calculate loss by averaging last x numbers to avoid jumps and noise')
                 #    ds_epochs_to_ramp = gr.Slider(label='Ramp down ratio', minimum=0.0, maximum=2.0, step=0.1, value=1.00, info='How long the ramp down will last relative to ellapsed steps (before trigger)')
                 #    gr.Markdown('These are settings for FP_dynamic_loss_trigger scheduler. The scheduler will do warm up, then hold constant untill a loss falls under Trigger Loss, then it will commence linear ramp down schedule and stop. The length of ramp down is set by Ramp down ratio where (ramp down steps) = ratio * (elapsed steps). (The time to completition shown will be very high untill ramp down is triggered.)')
-                        
+
 
             with gr.Column():
                 with gr.Tab(label='Formatted Dataset'):
@@ -217,7 +217,7 @@ def ui():
                             cutoff_len = gr.Slider(label='Chunk Length (Cutoff Length)', minimum=32, maximum=2048, value=256, step=32, info='The maximum length of a chunk (in tokens). Applies to both JSON dataset and text files. Higher values require much more VRAM.')
                 with gr.Row():
                     with gr.Column():
-                        check_dataset_btn = gr.Button('Verify Dataset/Text File and suggest data entries')    
+                        check_dataset_btn = gr.Button('Verify Dataset/Text File and suggest data entries')
                         check_dataset_txt = gr.Textbox(label='Dataset info', value='')
 
                 with gr.Row():
@@ -227,8 +227,8 @@ def ui():
                 with gr.Accordion(label="Graph", open=True):
                     with gr.Row():
                         # show_actions_button = False - we use old gradio
-                        plot_graph = gr.LinePlot(x="epoch", y="value", title="Loss Metrics", overlay_point=True, tooltip=["epoch", "value"], x_lim=[0, 1], y_lim=[0, 3.5], width=500, height=250) 
- 
+                        plot_graph = gr.LinePlot(x="epoch", y="value", title="Loss Metrics", overlay_point=True, tooltip=["epoch", "value"], x_lim=[0, 1], y_lim=[0, 3.5], width=500, height=250)
+
                 output = gr.Markdown(value="Ready")
 
     with gr.Tab('Perplexity evaluation', elem_id='evaluate-tab'):
@@ -267,7 +267,7 @@ def fix_old_version(batch_size_val,micro_batch_size_val, grad_accumulation_val):
 
         return grad_accumulation_val
 
-    
+
     copy_from.change(partial(do_copy_params, all_params= all_params), copy_from, all_params).then(fix_old_version,[batch_size,micro_batch_size, grad_accumulation],grad_accumulation)
     start_button.click(do_train, all_params, [output,plot_graph])
     stop_button.click(do_interrupt, None, None, queue=False)
@@ -306,8 +306,8 @@ def check_dataset(save_steps:int, micro_batch_size: int, epochs: int, cutoff_len
         if shared.tokenizer is None:
             yield "Tokenizer is not available. Please Load some Model first."
             return
-        
-        
+
+
         if raw_text_file not in ['None', '']:
             logger.info("Loading Text file...")
             fullpath = clean_path('user_data/training/datasets', f'{raw_text_file}')
@@ -329,8 +329,8 @@ def check_dataset(save_steps:int, micro_batch_size: int, epochs: int, cutoff_len
                 except:
                     yield f"{raw_text_file}.txt doesn't seem to exsist anymore... check your user_data/training/datasets folder"
                     return
-            
- 
+
+
             if min_chars<0:
                 min_chars = 0
 
@@ -343,11 +343,11 @@ def check_dataset(save_steps:int, micro_batch_size: int, epochs: int, cutoff_len
             total_blocks = len(text_chunks)
             result = f"Text: ({raw_text_file}.txt) has {total_blocks} blocks (Block Size {cutoff_len} tokens)"
             del text_chunks
-       
+
         else:
             if dataset in ['None', '']:
                 yield "Select dataset or text file."
-                return 
+                return
 
             if format in ['None', '']:
                 yield "Select format choice for dataset."
@@ -382,8 +382,8 @@ def generate_and_tokenize_prompt(data_point):
 
             logger.info("Loading JSON datasets...")
             data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
-            
-            data_keys = [] 
+
+            data_keys = []
 
             if data:
                 if 'train' in data:  # Check if the 'train' split exists in the dataset
@@ -400,11 +400,11 @@ def generate_and_tokenize_prompt(data_point):
             #for options, data in format_data.items():
             #    format_keys = options.split(',')
             #    result += f"{format_keys}, "
-            #result = result.rstrip()    
-            #result = result.rstrip(',')  
+            #result = result.rstrip()
+            #result = result.rstrip(',')
 
         if total_blocks>0:
-            number_ofSteps = int(math.ceil(total_blocks / micro_batch_size) * epochs) 
+            number_ofSteps = int(math.ceil(total_blocks / micro_batch_size) * epochs)
             num_stepsPer_epoch = int(math.ceil(number_ofSteps/epochs))
             min_warm = math.ceil(100 / grad_accumulation)
 
@@ -415,20 +415,20 @@ def generate_and_tokenize_prompt(data_point):
             save_each_n_max = int(math.ceil(number_ofSteps/5))
             gradient_accumulation_max = int(total_blocks)//micro_batch_size
 
- 
+
             result += f"\n[Batch Size: {micro_batch_size}, Epochs: {epochs}, Gradient Accumulation: {grad_accumulation}]\n"
             result += f"Total number of steps: {number_ofSteps}\n"
             result += f"Steps per each Epoch: {num_stepsPer_epoch}\n"
             result += f"Suggestions:\n"
             result += f"Checkpoints: Save every {save_each_n_min} - {save_each_n_max} steps (Current: {int(save_steps)})\n"
             result += f"Warmup steps: {warmup_steps_suggest} (Current: {int(warmup_steps)})"
-            if gradient_accumulation_max < grad_accumulation: 
+            if gradient_accumulation_max < grad_accumulation:
                 result += f"\n\nWARNING: Gradient Accumulation {grad_accumulation} is too high: It should be below {gradient_accumulation_max}"
 
 
         yield result
         return
-    
+
     check_dataset_btn.click(check_dataset, dataset_calc_params ,check_dataset_txt)
 
     # Evaluation events. For some reason, the interrupt event
@@ -449,10 +449,10 @@ def generate_and_tokenize_prompt(data_point):
 
     def reload_lora():
         return gr.Dropdown.update(choices=get_available_loras_local(non_serialized_params['Lora_sortedByTime']))
- 
+
     # nonserialized items
 
-    sort_byTime.change(lambda x: non_serialized_params.update({"Lora_sortedByTime": x}), sort_byTime, None).then(reload_lora,None,copy_from) 
+    sort_byTime.change(lambda x: non_serialized_params.update({"Lora_sortedByTime": x}), sort_byTime, None).then(reload_lora,None,copy_from)
     #debug_slicer.change(lambda x: non_serialized_params.update({"debug_slicer": x}), debug_slicer, None)
 
     def update_dataset():
@@ -482,7 +482,7 @@ def do_copy_params(lora_name: str, all_params):
         else:
             params = {}
     else:
-        params = {}        
+        params = {}
 
     result = list()
     for i in range(0, len(PARAMETERS)):
@@ -521,7 +521,8 @@ def backup_adapter(input_folder):
 
             # Create the new subfolder
             subfolder_path = Path(f"{input_folder}/{creation_date_str}")
-            subfolder_path.mkdir(parents=True, exist_ok=True)
+            if not subfolder_path.exists():
+                subfolder_path.mkdir(parents=True, exist_ok=True)
 
             # Check if the file already exists in the subfolder
             backup_adapter_file = Path(f"{input_folder}/{creation_date_str}/adapter_model.bin")
@@ -607,7 +608,7 @@ def do_train(lora_name: str, always_override: bool, save_steps: int, micro_batch
     shared.tokenizer.padding_side = "left"
 
     def encode(text, prepend_bos_token):
-       
+
         result = shared.tokenizer.encode(text, truncation=True, max_length=cutoff_len)
         # Check if the first two tokens are BOS
         if len(result) >= 2 and result[:2] == [shared.tokenizer.bos_token_id, shared.tokenizer.bos_token_id]:
@@ -626,7 +627,7 @@ def tokenize(prompt, append_eos_token=False, prepend_bos_token = False):
                 input_ids.append(shared.tokenizer.eos_token_id)
 
             input_ids = [shared.tokenizer.pad_token_id] * (cutoff_len - len(input_ids)) + input_ids
-            
+
             labels = [1] * len(input_ids)
         else:
             ind = prompt.index(train_only_after) + len(train_only_after)
@@ -653,7 +654,7 @@ def tokenize(prompt, append_eos_token=False, prepend_bos_token = False):
         }
 
     train_template.clear()
-            
+
     #reset stuff
     print(f"*** LoRA: {lora_name} ***")
     non_serialized_params.update({"stop_at_loss": stop_at_loss})
@@ -665,7 +666,7 @@ def tokenize(prompt, append_eos_token=False, prepend_bos_token = False):
     non_serialized_params.update({"checkpoint_offset": 0})
     non_serialized_params.update({"epoch_offset": 0})
     train_log_graph.clear()
-  
+
      # == Prep the dataset, format, etc ==
     if raw_text_file not in ['None', '']:
         train_template["template_type"] = "raw_text"
@@ -685,8 +686,8 @@ def tokenize(prompt, append_eos_token=False, prepend_bos_token = False):
         else:
             with open(clean_path('user_data/training/datasets', f'{raw_text_file}.txt'), 'r', encoding='utf-8') as file:
                 raw_text = file.read().replace('\r', '')
-        
-        # FPHAM PRECISE SLICING        
+
+        # FPHAM PRECISE SLICING
         if min_chars<0:
             min_chars = 0
 
@@ -703,7 +704,7 @@ def tokenize(prompt, append_eos_token=False, prepend_bos_token = False):
 
         train_data = Dataset.from_list([tokenize(x, add_EOS_to_all, add_bos_token) for x in text_chunks])
         if add_EOS_to_all:
-            print(f"Added EOS to {len(text_chunks)} blocks") 
+            print(f"Added EOS to {len(text_chunks)} blocks")
 
         print(f"All Data Blocks: {len(text_chunks)}")
 
@@ -745,7 +746,7 @@ def generate_and_tokenize_prompt(data_point):
         data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
         train_data = data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
 
-        print(f"BOS: {add_bos_token} EOS: {add_eos_token}") 
+        print(f"BOS: {add_bos_token} EOS: {add_eos_token}")
         print(f"Data Blocks: {train_data.num_rows}")
 
         if eval_dataset == 'None':
@@ -783,7 +784,7 @@ def generate_and_tokenize_prompt(data_point):
             print(f"Method: {RED}QLORA{RESET}")
             prepare_model_for_kbit_training(shared.model)
         else:
-            print(f"Method: {RED}LoRA{RESET}")    
+            print(f"Method: {RED}LoRA{RESET}")
 
     # base model is now frozen and should not be reused for any other LoRA training than this one
     shared.model_dirty_from_training = True
@@ -796,7 +797,7 @@ def generate_and_tokenize_prompt(data_point):
     elif training_projection==train_choices[2]:
         model_to_lora_modules[model_id] = ["q_proj","k_proj", "v_proj"]
     elif training_projection==train_choices[3]:
-        model_to_lora_modules[model_id] = ["k_proj", "v_proj", "down_proj"]        
+        model_to_lora_modules[model_id] = ["k_proj", "v_proj", "down_proj"]
     else:
         model_to_lora_modules[model_id] = ["q_proj", "v_proj"]
 
@@ -827,9 +828,9 @@ def generate_and_tokenize_prompt(data_point):
             set_peft_model_state_dict(lora_model, state_dict_peft)
 
             print(f" + Continue Training on {RED}{lora_file_path}/adapter_model.bin{RESET}")
-            
+
             #load training_log.json if exist
-           
+
             if Path(f"{lora_file_path}/training_log.json").is_file():
                 with open(f"{lora_file_path}/training_log.json", 'r') as json_file:
                     json_ilog = json.load(json_file)
@@ -840,13 +841,13 @@ def generate_and_tokenize_prompt(data_point):
                         if key=='epoch':
                             non_serialized_params.update({"epoch_offset": value})
                             print(f" + Epoch offset: {RED}{non_serialized_params['epoch_offset']}{RESET}")
-           
+
 
             if Path(f"{lora_file_path}/training_graph.json").is_file():
                 try:
                     with open(f"{lora_file_path}/training_graph.json", 'r') as json_file:
                         train_log_graph = json.load(json_file)
-                        print(" + Training Graph loaded")   
+                        print(" + Training Graph loaded")
                 except:
                     print(f"Can't read training_graph")
 
@@ -876,72 +877,72 @@ def on_step_begin(self, args: transformers.TrainingArguments, state: transformer
             else:
                 current_loss = float(train_log.get('loss', 0.0))
                 current_epoch_int = int(float(train_log.get('epoch', 0.0)))
-              
+
                 force_save = False
 
                 current_steps_offset = tracked.current_steps + non_serialized_params['checkpoint_offset']
 
-                folder_save = f"checkpoint-{current_steps_offset}"    
+                folder_save = f"checkpoint-{current_steps_offset}"
 
                 # save if triggered by user
                 if non_serialized_params['save_checkpoint_now']:
                     force_save = True
                     non_serialized_params.update({"save_checkpoint_now": False})
                     print(f"\033[1;31;1mSave Checkpoint manually trigerred.\033[0;37;0m")
-                    folder_save = f"checkpoint-{current_steps_offset}-user"  
+                    folder_save = f"checkpoint-{current_steps_offset}-user"
 
                 patience = 3     # Set the number of consecutive steps for tracking stability
-                
+
                 if gradient_accumulation_steps==1:
                     patience = 4
 
                 min_steps = ssteps10
 
-                # Save each time the loss is below the threshold 
+                # Save each time the loss is below the threshold
                 if current_loss < non_serialized_params['save_steps_under_loss'] and current_loss > 0 and state.global_step > min_steps:
                     current_stability = non_serialized_params['current_stability']
                     current_stability += 1
-                    non_serialized_params.update({"current_stability": current_stability}) 
+                    non_serialized_params.update({"current_stability": current_stability})
 
                     if current_stability >= patience:
                         current_stability = 0
-                        non_serialized_params.update({"current_stability": current_stability})     
+                        non_serialized_params.update({"current_stability": current_stability})
                         current_loss_dec = round(current_loss, 2)
                         loss_str = f"{current_loss_dec:.2f}"
                         loss_str = loss_str.replace('.', '_')
                         new_save = (current_loss_dec-0.1) + 0.01
                         non_serialized_params.update({"save_steps_under_loss": new_save})
 
-                        folder_save = f"checkpoint-{current_steps_offset}-loss-{loss_str}" 
-                        force_save = True   
+                        folder_save = f"checkpoint-{current_steps_offset}-loss-{loss_str}"
+                        force_save = True
+
 
-                   
                 else:
                     # Reset stability if the loss goes above the threshold
-                    non_serialized_params.update({"current_stability": 0})   
+                    non_serialized_params.update({"current_stability": 0})
 
                 # Save full epochs
-                if actual_save_steps>0 and current_epoch_int > non_serialized_params['save_epochs'] and state.global_step > min_steps: 
+                if actual_save_steps>0 and current_epoch_int > non_serialized_params['save_epochs'] and state.global_step > min_steps:
+
 
-                    
                     current_epoch_offset = current_epoch_int
-                    
+
                     if non_serialized_params['epoch_offset'] > 0:
                         current_epoch_offset = current_epoch_int + round(non_serialized_params['epoch_offset'], 2)
-                    
+
                     ep_off_str = f"{current_epoch_offset}"
                     ep_off_str = ep_off_str.replace('.', '_')
-                    folder_save = f"checkpoint-{current_steps_offset}-epoch-{ep_off_str}" 
+                    folder_save = f"checkpoint-{current_steps_offset}-epoch-{ep_off_str}"
 
                     non_serialized_params.update({"save_epochs": current_epoch_int})
                     force_save = True
 
                 # save each actual_save_steps
                 if state.global_step > 0 and actual_save_steps > 0 and state.global_step % actual_save_steps == 0:
-                    folder_save = f"checkpoint-{current_steps_offset}"  
-                    force_save = True   
+                    folder_save = f"checkpoint-{current_steps_offset}"
+                    force_save = True
 
-                if force_save:       
+                if force_save:
                     lora_model.save_pretrained(f"{lora_file_path}/{folder_save}/", safe_serialization = non_serialized_params['safe_serialization'])
                     print(f"\033[1;30;40mStep: {tracked.current_steps:6} \033[0;37;0m Saved: [{folder_save}]")
                     # Save log
@@ -950,7 +951,7 @@ def on_step_begin(self, args: transformers.TrainingArguments, state: transformer
                     # == Save training prompt ==
                     with open(f"{lora_file_path}/{folder_save}/training_prompt.json", 'w', encoding='utf-8') as file:
                         json.dump(train_template, file, indent=2)
-                
+
 
         def on_substep_end(self, args: transformers.TrainingArguments, state: transformers.TrainerState, control: transformers.TrainerControl, **kwargs):
             tracked.current_steps += 1
@@ -975,7 +976,7 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
                 print(f"\033[1;30;40mStep: {tracked.current_steps:6} [+{non_serialized_params['checkpoint_offset']}] \033[0;37;0m", end='')
             else:
                 print(f"\033[1;30;40mStep: {tracked.current_steps:6} \033[0;37;0m", end='')
-            
+
             graphentry = {
                 'current_steps': int(train_log.get('current_steps_adjusted',0)),
                 'loss': float(train_log.get('loss', 0.0)),
@@ -986,7 +987,7 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
             cur_loss = float(train_log.get('loss', 0.0))
             cur_lr = float(train_log.get('learning_rate', 0.0))
             cur_epoch = float(train_log.get('epoch', 0.0))
-            
+
             if len(statistics['loss']) == 1:
                 first_epoch = statistics['loss'][0]['epoch']
                 first_value = statistics['loss'][0]['value']
@@ -1013,7 +1014,7 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
 
     # FPHAM SAMPLE REQ Transformers error handling
     gradient_accumulation_max = int(train_data.num_rows)//micro_batch_size
-    
+
     if gradient_accumulation_max < gradient_accumulation_steps:
         print(f"{RED}WARNING:{RESET} Current gradient accumulation is {RED}too high{RESET} for the amount of training data.")
         print(f"Gradient accumulation: {gradient_accumulation_steps} should be less than: {gradient_accumulation_max}. {RED}This could crash Accelerate/Transformers{RESET}")
@@ -1041,9 +1042,9 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
     elif lr_scheduler_type =='FP_raise_fall_creative':
         custom_scheduller = True
         lr_scheduler_type_arg = 'constant_with_warmup'
-    
+
     #gradient_checkpointing=True
-    
+
     args=transformers.TrainingArguments(
             report_to=report_to if report_to != "None" else None,
             per_device_train_batch_size=micro_batch_size,
@@ -1095,7 +1096,7 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
             data_collator=transformers.DataCollatorForLanguageModeling(shared.tokenizer, mlm=False),
             callbacks=list([Callbacks()])
         )
-    
+
     # END OF FPHAM CUSTOM SCHEDULER
 
     lora_model.config.use_cache = False
@@ -1141,7 +1142,7 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
     if stop_at_loss > 0:
         print(f"Monitoring loss {RED}(Auto-Stop at: {stop_at_loss}){RESET}")
 
-    
+
 
     if WANT_INTERRUPT:
         yield "Interrupted before start.", zero_pd
@@ -1157,7 +1158,9 @@ def log_train_dataset(trainer):
                 decoded_entries.append({"value": decoded_text})
 
             # Write the log file
-            Path('user_data/logs').mkdir(exist_ok=True)
+            if not Path('user_data/logs').exists():
+                Path('user_data/logs').mkdir(exist_ok=True)
+
             with open(Path('user_data/logs/train_dataset_sample.json'), 'w') as json_file:
                 json.dump(decoded_entries, json_file, indent=4)
 
@@ -1191,7 +1194,7 @@ def threaded_run():
         else:
             max_value = 3.5
             last_epoch = 0
-            first_epoch = 0           
+            first_epoch = 0
 
         if WANT_INTERRUPT:
 
@@ -1210,7 +1213,7 @@ def threaded_run():
             time_elapsed = time.perf_counter() - start_time
             lastloss = float(train_log.get('loss', 0.0))
 
-            non_serialized_params.update({"training_loop": True})               
+            non_serialized_params.update({"training_loop": True})
 
             if lastloss > 0:
                 lastloss_str = f", ... Current Loss: `{lastloss:.2f}`"
@@ -1232,7 +1235,7 @@ def threaded_run():
             if stop_at_loss != non_serialized_params['stop_at_loss']:
                 stop_at_loss = non_serialized_params['stop_at_loss']
                 print(f"Stop at loss changed {RED}(Auto-Stop at: {stop_at_loss}){RESET}")
-            
+
             losses = gr.LinePlot.update(
 				value = pd.DataFrame(statistics['loss']),
                 x="epoch", y="value",
@@ -1240,7 +1243,7 @@ def threaded_run():
                 overlay_point=True, tooltip=["epoch", "value"],
 				x_lim=[first_epoch,last_epoch], y_lim=[0,max_value],
                 width=500, height=250 )
-				
+
 
             yield f"Running... **{tracked.current_steps}** / **{tracked.max_steps}** ... {timer_info}, {format_time(time_elapsed)} / {format_time(total_time_estimate)} ... {format_time(total_time_estimate - time_elapsed)} remaining {lastloss_str}", losses
 
@@ -1256,7 +1259,7 @@ def threaded_run():
     else:
         max_value = 3.5
         last_epoch = 0
-        first_epoch = 0 
+        first_epoch = 0
 
     return_pd = gr.LinePlot.update(
         value = pd.DataFrame(statistics['loss']),
diff --git a/extensions/Training_PRO/train_utils.py b/extensions/Training_PRO/train_utils.py
index 799948802b..d1a45c1795 100644
--- a/extensions/Training_PRO/train_utils.py
+++ b/extensions/Training_PRO/train_utils.py
@@ -20,7 +20,7 @@ def list_subfoldersByTime(directory):
     if not directory.endswith('/'):
         directory += '/'
     subfolders = []
-    subfolders.append('None') 
+    subfolders.append('None')
     path = directory
     name_list = os.listdir(path)
     full_list = [os.path.join(path,i) for i in name_list]
@@ -37,19 +37,19 @@ def list_subfoldersByTime(directory):
     return subfolders
 
 def get_available_loras_local(_sortedByTime):
-    
+
     model_dir = shared.args.lora_dir  # Update with the appropriate directory path
     subfolders = []
     if _sortedByTime:
         subfolders = list_subfoldersByTime(model_dir)
     else:
-        subfolders = utils.get_available_loras()        
+        subfolders = utils.get_available_loras()
 
     return subfolders
 
 
 # FPHAM SPLIT BY SENTENCE BLOCK ===============
-     
+
 def split_sentences(text: str, cutoff_len: int):
     sentences = []
     sentence = ''
@@ -57,24 +57,24 @@ def split_sentences(text: str, cutoff_len: int):
     abbreviations = ['Mr. ', 'Mrs. ', 'Dr. ', 'Ms. ', 'St. ', 'Prof. ', 'Jr. ', 'Ltd. ', 'Capt. ', 'Col. ', 'Gen. ', 'Ave. ', 'Blvd. ', 'Co. ', 'Corp. ', 'Dept. ', 'Est. ', 'Gov. ', 'Inc. ', 'Ph.D. ', 'Univ. ']
     errors = 0
     max_cut = cutoff_len-1
-    prev_char = ''  
+    prev_char = ''
 
     for char in text:
         sentence += char
 
-    
+
         if (any(sentence.endswith(delimiter) for delimiter in delimiters) and
-            not (prev_char.isupper() and len(sentence) >= 3 and sentence[-3] != ' ') and 
+            not (prev_char.isupper() and len(sentence) >= 3 and sentence[-3] != ' ') and
             not any(sentence.endswith(abbreviation) for abbreviation in abbreviations)):
             tokens = shared.tokenizer.encode(sentence)
-            
+
             if len(tokens) > max_cut:
                 tokens = tokens[:max_cut]
                 sentence = shared.tokenizer.decode(tokens, skip_special_tokens=True)
                 errors = errors + 1
 
             sentences.append({'text': sentence, 'size': len(tokens)})
-            
+
             sentence = ''
 
         prev_char = char
@@ -83,7 +83,7 @@ def split_sentences(text: str, cutoff_len: int):
         tokens = shared.tokenizer.encode(sentence)
         if len(tokens) > max_cut:
             tokens = tokens[:max_cut]
-            sentence = shared.tokenizer.decode(tokens, skip_special_tokens=True)  
+            sentence = shared.tokenizer.decode(tokens, skip_special_tokens=True)
             errors = errors + 1
 
         sentences.append({'text': sentence, 'size': len(tokens)})
@@ -95,16 +95,16 @@ def split_sentences(text: str, cutoff_len: int):
 
 # The goal of following code is to create blocks of text + overlapping blocks while:
 # respects sentence boundaries
-# always uses all the text 
+# always uses all the text
 # hard cut defined by hard_cut_string or  will always end at the end of data block
 # no overlapping blocks will be created across hard cut or across              token
 
 def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, cutoff_len: int, hard_cut_string: str, debug_slicer:bool):
 
     EOSX_str = '' #hardcut placeholder
-    EOS_str = '            ' 
+    EOS_str = ''
     print("Precise raw text slicer: ON")
-    
+
     cut_string = hard_cut_string.replace('\\n', '\n')
     text = text.replace(cut_string, EOSX_str)
     sentences = split_sentences(text, cutoff_len)
@@ -121,7 +121,7 @@ def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, c
     half_index = 0
 
     for index, item in enumerate(sentences):
-        
+
         if halfcut_length+ item['size'] < half_cut:
             halfcut_length += item['size']
             half_index = index
@@ -130,7 +130,7 @@ def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, c
             halfcut_length = -2 * max_cut
 
 
-        if totalLength + item['size'] < max_cut and not currentSentence.endswith(EOSX_str): 
+        if totalLength + item['size'] < max_cut and not currentSentence.endswith(EOSX_str):
             currentSentence += item['text']
             totalLength += item['size']
         else:
@@ -141,14 +141,14 @@ def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, c
             currentSentence = item['text']
             totalLength = item['size']
             halfcut_length = item['size']
-            
-    if len(currentSentence.strip()) > min_chars_cut:    
+
+    if len(currentSentence.strip()) > min_chars_cut:
         sentencelist.append(currentSentence.strip())
 
     unique_blocks = len(sentencelist)
     print(f"Text Blocks: {unique_blocks}")
 
-    #overlap strategies: 
+    #overlap strategies:
     # don't overlap across HARD CUT (EOSX)
     if overlap:
         for edge_idx in edgeindex:
@@ -162,15 +162,15 @@ def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, c
                 else:
                     #if by chance EOSX is at the end then it's acceptable
                     if currentSentence.endswith(EOSX_str) and len(currentSentence.strip()) > min_chars_cut:
-                            sentencelist.append(currentSentence.strip())    
-                    # otherwise don't cross hard cut    
+                            sentencelist.append(currentSentence.strip())
+                    # otherwise don't cross hard cut
                     elif EOSX_str not in currentSentence and len(currentSentence.strip()) > min_chars_cut:
                         sentencelist.append(currentSentence.strip())
-                    
+
                     currentSentence = ''
                     totalLength = 0
                     break
-        
+
         print(f"+ Overlapping blocks: {len(sentencelist)-unique_blocks}")
 
     num_EOS = 0
@@ -179,7 +179,7 @@ def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, c
             sentencelist[i] = sentencelist[i].replace(EOSX_str, EOS_str)
         else:
             sentencelist[i] = sentencelist[i].replace(EOSX_str, '')
-        
+
         #someone may have had stop strings in the raw text...
         sentencelist[i] = sentencelist[i].replace("", EOS_str)
         num_EOS += sentencelist[i].count(EOS_str)
@@ -193,47 +193,49 @@ def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, c
 
 
     if debug_slicer:
-                    # Write the log file
-        Path('user_data/logs').mkdir(exist_ok=True)
+        # Write the log file
+        if not Path('user_data/logs').exists():
+            Path('user_data/logs').mkdir(exist_ok=True)
+
         sentencelist_dict = {index: sentence for index, sentence in enumerate(sentencelist)}
         output_file = "user_data/logs/sentencelist.json"
         with open(output_file, 'w') as f:
             json.dump(sentencelist_dict, f,indent=2)
-        
+
         print("Saved sentencelist.json in user_data/logs folder")
-    
-    return sentencelist   
+
+    return sentencelist
 
 
 def sliding_block_cut(text: str, min_chars_cut: int, eos_to_hc: bool, cutoff_len: int, hard_cut_string: str, debug_slicer:bool):
 
     EOSX_str = '' #hardcut placeholder
-    EOS_str = '' 
+    EOS_str = ''
     print("Mega Block Overlap: ON")
-    
+
     cut_string = hard_cut_string.replace('\\n', '\n')
     text = text.replace(cut_string, EOSX_str)
     sentences = split_sentences(text, cutoff_len)
 
     print(f"Sentences: {len(sentences)}")
     sentencelist = []
-    
+
     max_cut = cutoff_len-1
 
     #print(f"max_cut: {max_cut}")
     advancing_to = 0
 
     prev_block_lastsentence = ""
-    
+
 
     for i in range(len(sentences)):
         totalLength = 0
         currentSentence = ''
         lastsentence = ""
-        
+
         if i >= advancing_to:
             for k in range(i, len(sentences)):
-                
+
                 current_length = sentences[k]['size']
 
                 if totalLength + current_length <= max_cut and not currentSentence.endswith(EOSX_str):
@@ -245,7 +247,7 @@ def sliding_block_cut(text: str, min_chars_cut: int, eos_to_hc: bool, cutoff_len
                         if prev_block_lastsentence!=lastsentence:
                             sentencelist.append(currentSentence.strip())
                             prev_block_lastsentence = lastsentence
-                        
+
                     advancing_to = 0
                     if currentSentence.endswith(EOSX_str):
                         advancing_to = k
@@ -253,7 +255,7 @@ def sliding_block_cut(text: str, min_chars_cut: int, eos_to_hc: bool, cutoff_len
                     currentSentence = ""
                     totalLength = 0
                     break
-            
+
             if currentSentence != "":
                 if len(currentSentence.strip()) > min_chars_cut:
                     sentencelist.append(currentSentence.strip())
@@ -266,7 +268,7 @@ def sliding_block_cut(text: str, min_chars_cut: int, eos_to_hc: bool, cutoff_len
             sentencelist[i] = sentencelist[i].replace(EOSX_str, EOS_str)
         else:
             sentencelist[i] = sentencelist[i].replace(EOSX_str, '')
-        
+
         #someone may have had stop strings in the raw text...
         sentencelist[i] = sentencelist[i].replace("", EOS_str)
         num_EOS += sentencelist[i].count(EOS_str)
@@ -280,16 +282,18 @@ def sliding_block_cut(text: str, min_chars_cut: int, eos_to_hc: bool, cutoff_len
 
 
     if debug_slicer:
-                    # Write the log file
-        Path('user_data/logs').mkdir(exist_ok=True)
+        # Write the log file
+        if not Path('user_data/logs').exists():
+            Path('user_data/logs').mkdir(exist_ok=True)
+
         sentencelist_dict = {index: sentence for index, sentence in enumerate(sentencelist)}
         output_file = "user_data/logs/sentencelist.json"
         with open(output_file, 'w') as f:
             json.dump(sentencelist_dict, f,indent=2)
-        
+
         print("Saved sentencelist.json in user_data/logs folder")
-    
-    return sentencelist   
+
+    return sentencelist
 
 # Example usage:
 # download_file_from_url('https://example.com/path/to/your/file.ext', '/output/directory')
@@ -326,17 +330,17 @@ def download_file_from_url(url, overwrite, output_dir_in, valid_extensions = {'.
 
         # Send an HTTP GET request to the URL with a timeout
         file_extension = os.path.splitext(filename_lower)[-1]
-        
+
         if file_extension not in valid_extensions:
             yield f"Invalid file extension: {file_extension}. Only {valid_extensions} files are supported."
             return
 
         with session.get(url, stream=True, headers=headers, timeout=10) as r:
-            r.raise_for_status() 
+            r.raise_for_status()
             # total size can be wildly inaccurate
             #total_size = int(r.headers.get('content-length', 0))
-            
-            block_size = 1024 * 4  
+
+            block_size = 1024 * 4
             with open(local_filename, mode) as f:
                 count = 0
                 for data in r.iter_content(block_size):
diff --git a/extensions/coqui_tts/script.py b/extensions/coqui_tts/script.py
index fe6364b9f2..4f5700d1de 100644
--- a/extensions/coqui_tts/script.py
+++ b/extensions/coqui_tts/script.py
@@ -168,7 +168,8 @@ def setup():
     print("[XTTS] Loading XTTS...")
     model = load_model()
     print("[XTTS] Done!")
-    Path(f"{this_dir}/outputs").mkdir(parents=True, exist_ok=True)
+    if not Path(f"{this_dir}/outputs").exists():
+        Path(f"{this_dir}/outputs").mkdir(parents=True, exist_ok=True)
 
 
 def ui():
diff --git a/extensions/sd_api_pictures/script.py b/extensions/sd_api_pictures/script.py
index f216da3849..d6853664a5 100644
--- a/extensions/sd_api_pictures/script.py
+++ b/extensions/sd_api_pictures/script.py
@@ -168,7 +168,8 @@ def get_SD_pictures(description, character):
 
             variadic = f'{date.today().strftime("%Y_%m_%d")}/{character}_{int(time.time())}'
             output_file = Path(f'extensions/sd_api_pictures/outputs/{variadic}.png')
-            output_file.parent.mkdir(parents=True, exist_ok=True)
+            if not output_file.parent.exists():
+                output_file.parent.mkdir(parents=True, exist_ok=True)
 
             with open(output_file.as_posix(), 'wb') as f:
                 f.write(img_data)
diff --git a/modules/chat.py b/modules/chat.py
index d76835665f..7770b5cff8 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -973,8 +973,8 @@ def save_history(history, unique_id, character, mode):
         return
 
     p = get_history_file_path(unique_id, character, mode)
-    if not p.parent.is_dir():
-        p.parent.mkdir(parents=True)
+    if not p.parent.exists():
+        p.parent.mkdir(parents=True, exist_ok=True)
 
     with open(p, 'w', encoding='utf-8') as f:
         f.write(json.dumps(history, indent=4, ensure_ascii=False))
@@ -1014,7 +1014,9 @@ def get_paths(state):
             unique_id = datetime.now().strftime('%Y%m%d-%H-%M-%S')
             p = get_history_file_path(unique_id, character, state['mode'])
             logger.warning(f"Moving \"{new_p}\" to \"{p}\"")
-            p.parent.mkdir(exist_ok=True)
+            if not p.parent.exists():
+                p.parent.mkdir(exist_ok=True)
+
             new_p.rename(p)
 
         return Path(f'user_data/logs/chat/{character}').glob('*.json')
@@ -1163,7 +1165,9 @@ def save_last_chat_state(character, mode, unique_id):
     state["last_chats"][key] = unique_id
 
     state_file = Path('user_data/logs/chat_state.json')
-    state_file.parent.mkdir(exist_ok=True)
+    if not state_file.parent.exists():
+        state_file.parent.mkdir(exist_ok=True)
+
     with open(state_file, 'w', encoding='utf-8') as f:
         f.write(json.dumps(state, indent=2))
 
diff --git a/modules/evaluate.py b/modules/evaluate.py
index 4f41c1fc14..65cceda949 100644
--- a/modules/evaluate.py
+++ b/modules/evaluate.py
@@ -27,7 +27,9 @@ def save_past_evaluations(df):
     global past_evaluations
     past_evaluations = df
     filepath = Path('user_data/logs/evaluations.csv')
-    filepath.parent.mkdir(parents=True, exist_ok=True)
+    if not filepath.parent.exists():
+        filepath.parent.mkdir(parents=True, exist_ok=True)
+
     df.to_csv(filepath, index=False)
 
 
diff --git a/modules/prompts.py b/modules/prompts.py
index 79d9b56e46..1b154cca1e 100644
--- a/modules/prompts.py
+++ b/modules/prompts.py
@@ -9,7 +9,9 @@ def load_prompt(fname):
         # Create new file
         new_name = utils.current_time()
         prompt_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
-        prompt_path.parent.mkdir(parents=True, exist_ok=True)
+        if not prompt_path.parent.exists():
+            prompt_path.parent.mkdir(parents=True, exist_ok=True)
+
         initial_content = "In this story,"
         prompt_path.write_text(initial_content, encoding='utf-8')
 
diff --git a/modules/training.py b/modules/training.py
index 2354c39df9..2689977f10 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -241,7 +241,8 @@ def backup_adapter(input_folder):
 
             # Create the new subfolder
             subfolder_path = Path(f"{input_folder}/{creation_date_str}")
-            subfolder_path.mkdir(parents=True, exist_ok=True)
+            if not subfolder_path.exists():
+                subfolder_path.mkdir(parents=True, exist_ok=True)
 
             # Check if the file already exists in the subfolder
             backup_adapter_file = Path(f"{input_folder}/{creation_date_str}/adapter_model.bin")
@@ -676,7 +677,9 @@ def log_train_dataset(trainer):
                 decoded_entries.append({"value": decoded_text})
 
             # Write the log file
-            Path('user_data/logs').mkdir(exist_ok=True)
+            if not Path('user_data/logs').exists():
+                Path('user_data/logs').mkdir(exist_ok=True)
+
             with open(Path('user_data/logs/train_dataset_sample.json'), 'w') as json_file:
                 json.dump(decoded_entries, json_file, indent=4)
 
diff --git a/modules/ui.py b/modules/ui.py
index 0e8afa8f5a..1247b3652a 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -409,7 +409,9 @@ def _perform_debounced_save():
         if _last_interface_state is not None:
             contents = save_settings(_last_interface_state, _last_preset, _last_extensions, _last_show_controls, _last_theme_state, manual_save=False)
             settings_path = Path('user_data') / 'settings.yaml'
-            settings_path.parent.mkdir(exist_ok=True)
+            if not settings_path.parent.exists():
+                settings_path.parent.mkdir(exist_ok=True)
+
             with open(settings_path, 'w', encoding='utf-8') as f:
                 f.write(contents)
     except Exception as e:
diff --git a/modules/ui_default.py b/modules/ui_default.py
index 66174e359e..21a638925d 100644
--- a/modules/ui_default.py
+++ b/modules/ui_default.py
@@ -152,7 +152,9 @@ def autosave_prompt(text, prompt_name):
     """Automatically save the text to the selected prompt file"""
     if prompt_name and text.strip():
         prompt_path = Path("user_data/logs/notebook") / f"{prompt_name}.txt"
-        prompt_path.parent.mkdir(parents=True, exist_ok=True)
+        if not prompt_path.parent.exists():
+            prompt_path.parent.mkdir(parents=True, exist_ok=True)
+
         prompt_path.write_text(text, encoding='utf-8')
 
 
@@ -205,7 +207,9 @@ def handle_new_prompt():
 
     # Create the new prompt file
     prompt_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
-    prompt_path.parent.mkdir(parents=True, exist_ok=True)
+    if not prompt_path.parent.exists():
+        prompt_path.parent.mkdir(parents=True, exist_ok=True)
+
     prompt_path.write_text("In this story,", encoding='utf-8')
 
     return gr.update(choices=utils.get_available_prompts(), value=new_name)
@@ -222,7 +226,9 @@ def handle_delete_prompt_confirm_default(prompt_name):
         new_value = available_prompts[min(current_index, len(available_prompts) - 1)]
     else:
         new_value = utils.current_time()
-        Path("user_data/logs/notebook").mkdir(parents=True, exist_ok=True)
+        if not Path("user_data/logs/notebook").exists():
+            Path("user_data/logs/notebook").mkdir(parents=True, exist_ok=True)
+
         (Path("user_data/logs/notebook") / f"{new_value}.txt").write_text("In this story,")
         available_prompts = [new_value]
 
diff --git a/modules/ui_notebook.py b/modules/ui_notebook.py
index 3a8c9df2f4..916e613f80 100644
--- a/modules/ui_notebook.py
+++ b/modules/ui_notebook.py
@@ -197,7 +197,9 @@ def handle_new_prompt():
 
     # Create the new prompt file
     prompt_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
-    prompt_path.parent.mkdir(parents=True, exist_ok=True)
+    if not prompt_path.parent.exists():
+        prompt_path.parent.mkdir(parents=True, exist_ok=True)
+
     prompt_path.write_text("In this story,", encoding='utf-8')
 
     return gr.update(choices=utils.get_available_prompts(), value=new_name)
@@ -214,7 +216,9 @@ def handle_delete_prompt_confirm_notebook(prompt_name):
         new_value = available_prompts[min(current_index, len(available_prompts) - 1)]
     else:
         new_value = utils.current_time()
-        Path("user_data/logs/notebook").mkdir(parents=True, exist_ok=True)
+        if not Path("user_data/logs/notebook").exists():
+            Path("user_data/logs/notebook").mkdir(parents=True, exist_ok=True)
+
         (Path("user_data/logs/notebook") / f"{new_value}.txt").write_text("In this story,")
         available_prompts = [new_value]
 
diff --git a/modules/utils.py b/modules/utils.py
index da5ac9780d..b3bac583f6 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -160,7 +160,8 @@ def get_available_presets():
 
 def get_available_prompts():
     notebook_dir = Path('user_data/logs/notebook')
-    notebook_dir.mkdir(parents=True, exist_ok=True)
+    if not notebook_dir.exists():
+        notebook_dir.mkdir(parents=True, exist_ok=True)
 
     prompt_files = list(notebook_dir.glob('*.txt'))
     sorted_files = sorted(prompt_files, key=lambda x: x.stat().st_mtime, reverse=True)
diff --git a/server.py b/server.py
index 7ce3c208f9..2b998792bf 100644
--- a/server.py
+++ b/server.py
@@ -10,7 +10,8 @@
 # Set up Gradio temp directory path
 gradio_temp_path = Path('user_data') / 'cache' / 'gradio'
 shutil.rmtree(gradio_temp_path, ignore_errors=True)
-gradio_temp_path.mkdir(parents=True, exist_ok=True)
+if not gradio_temp_path.exists():
+    gradio_temp_path.mkdir(parents=True, exist_ok=True)
 
 # Set environment variables
 os.environ.update({

From aa44e542cbaef727592bb55504fed55774fc2eba Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Jun 2025 07:11:59 -0700
Subject: [PATCH 0732/1701] Revert "Safer usage of mkdir across the project"

This reverts commit 0d1597616ff9be86d70a5eae49b021f65fc4457f.
---
 download-model.py                           |   4 +-
 extensions/Training_PRO/custom_scheduler.py | 116 +++++++-------
 extensions/Training_PRO/matplotgraph.py     |   6 +-
 extensions/Training_PRO/script.py           | 161 ++++++++++----------
 extensions/Training_PRO/train_utils.py      |  96 ++++++------
 extensions/coqui_tts/script.py              |   3 +-
 extensions/sd_api_pictures/script.py        |   3 +-
 modules/chat.py                             |  12 +-
 modules/evaluate.py                         |   4 +-
 modules/prompts.py                          |   4 +-
 modules/training.py                         |   7 +-
 modules/ui.py                               |   4 +-
 modules/ui_default.py                       |  12 +-
 modules/ui_notebook.py                      |   8 +-
 modules/utils.py                            |   3 +-
 server.py                                   |   3 +-
 16 files changed, 206 insertions(+), 240 deletions(-)

diff --git a/download-model.py b/download-model.py
index 0c6307a1cd..576a8b7910 100644
--- a/download-model.py
+++ b/download-model.py
@@ -314,8 +314,8 @@ def start_download_threads(self, file_list, output_folder, start_from_scratch=Fa
 
     def download_model_files(self, model, branch, links, sha256, output_folder, progress_queue=None, start_from_scratch=False, threads=4, specific_file=None, is_llamacpp=False):
         self.progress_queue = progress_queue
-        if not output_folder.exists():
-            output_folder.mkdir(parents=True, exist_ok=True)
+
+        output_folder.mkdir(parents=True, exist_ok=True)
 
         if not is_llamacpp:
             metadata = f'url: https://huggingface.co/{model}\n' \
diff --git a/extensions/Training_PRO/custom_scheduler.py b/extensions/Training_PRO/custom_scheduler.py
index 59e7706ed3..1e80daed1a 100644
--- a/extensions/Training_PRO/custom_scheduler.py
+++ b/extensions/Training_PRO/custom_scheduler.py
@@ -20,7 +20,7 @@
 
 def custom_scheduler_global_update(current_loss: float):
     custom_scheduler_params.update({'current_loss': current_loss})
-
+  
 def custom_scheduler_global_setup(trigger_loss: float, ramp_down_ratio: float):
     custom_scheduler_params.update({'trigger_loss': trigger_loss})
     custom_scheduler_params.update({'ramp_down_ratio': ramp_down_ratio})
@@ -35,12 +35,12 @@ def custom_scheduler_global_setup(trigger_loss: float, ramp_down_ratio: float):
 
 # hold constant to the half of epochs then cosine down to 0
 def _get_fp_half_schedule_with_warmup_lr_lambda(current_step: int, *, num_warmup_steps: int, num_training_steps: int, num_firstepoch_steps: int):
-
+    
     global last_print_label
     print_label = ''
 
     half_steps = num_training_steps//2
-
+    
     num_warmup_steps = min(num_warmup_steps,half_steps)
 
     if current_step < num_warmup_steps:
@@ -49,57 +49,57 @@ def _get_fp_half_schedule_with_warmup_lr_lambda(current_step: int, *, num_warmup
         print_label = 'Scheduler: Hold'
     else:
         print_label = 'Scheduler: Annealing'
-
+    
     if print_label != last_print_label:
         print(print_label)
-
+    
     last_print_label = print_label
 
     if current_step < num_warmup_steps:
         return float(current_step) / float(max(1, num_warmup_steps))
-
+    
     if current_step < half_steps:
-        return 1.0
-
+        return 1.0 
+    
     progress = float(current_step - half_steps) / float(max(1, num_training_steps - half_steps))
     num_cycles = 0.5
-    return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))
-
+    return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))    
+ 
 
 # raise up in cosine, then fall back in cosine
 def _get_fp_cosine_raise_and_fall_lr_lambda(current_step: int, *, num_warmup_steps: int, num_training_steps: int, num_firstepoch_steps: int):
-
+    
     global last_print_label
     print_label = ''
 
     half_steps = num_training_steps//2
-
+    
     #num_warmup_steps = min(num_warmup_steps,half_steps)
 
     if current_step < half_steps:
         print_label = 'Scheduler: Raise'
     else:
         print_label = 'Scheduler: Fall'
-
+    
     if print_label != last_print_label:
         print(print_label)
-
+    
     last_print_label = print_label
 
-
+    
     # linear
     #    return float(current_step) / float(max(1, num_warmup_steps))
-
+    
     progress = float(current_step - half_steps) / float(max(1, num_training_steps - half_steps))
     num_cycles = 0.5
-    return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))
-
+    return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))    
+ 
 # constant to the first epochs then cosine down to 0 over the rest epochs
 def _get_fp_cosine_schedule_with_warmup_lr_lambda(current_step: int, *, num_warmup_steps: int, num_training_steps: int, num_firstepoch_steps: int):
-
+    
     global last_print_label
     print_label = ''
-
+    
     num_warmup_steps = min(num_warmup_steps,num_firstepoch_steps)
 
     if current_step < num_warmup_steps:
@@ -108,56 +108,56 @@ def _get_fp_cosine_schedule_with_warmup_lr_lambda(current_step: int, *, num_warm
         print_label = 'Scheduler: Hold'
     else:
         print_label = 'Scheduler: Annealing'
-
+    
     if print_label != last_print_label:
         print(print_label)
-
+    
     last_print_label = print_label
 
     if current_step < num_warmup_steps:
         return float(current_step) / float(max(1, num_warmup_steps))
-
+    
     if current_step < num_firstepoch_steps:
-        return 1.0
-
+        return 1.0 
+    
     progress = float(current_step - num_firstepoch_steps) / float(max(1, num_training_steps - num_firstepoch_steps))
     num_cycles = 0.5
-    return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))
-
-# halve lr each epoch
+    return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))    
+    
+# halve lr each epoch   
 
 def _get_fp_cdrop_rate_schedule_with_warmup_lr_lambda(current_step: int, *, num_warmup_steps: int, num_training_steps: int, num_firstepoch_steps: int):
-
+    
     global last_print_label
     print_label = ''
-
+    
     num_warmup_steps = min(num_warmup_steps, num_firstepoch_steps)
 
     current_epoch = (current_step // num_firstepoch_steps) + 1
-
-
+    
+    
     if current_step < num_warmup_steps:
         print_label = 'Scheduler: Warmup'
     elif current_step < num_firstepoch_steps:
         print_label = 'Scheduler: Hold'
     else:
         print_label = 'Scheduler: Drop Rate'
-
+    
     if print_label != last_print_label:
         print(print_label)
-
+    
     last_print_label = print_label
 
     if current_step < num_warmup_steps:
         return float(current_step) / float(max(1, num_warmup_steps))
-
+    
     if current_step < num_firstepoch_steps:
-        return 1.0
+        return 1.0 
 
     # Compute the learning rate for the annealing phase
-
+    
     learning_rate = 1.0 / float(2 ** (current_epoch - 1))
-
+   
     return learning_rate
 
 # epoch decay: 1/(1 + decay * epoch)
@@ -177,7 +177,7 @@ def custom_cosine_scheduler_with_warmup(optimizer, num_warmup_steps, num_trainin
     Return:
         `torch.optim.lr_scheduler.LambdaLR` with the appropriate schedule.
     """
-
+    
     lr_lambda = partial(
         _get_fp_cosine_schedule_with_warmup_lr_lambda,
         num_warmup_steps=num_warmup_steps,
@@ -201,7 +201,7 @@ def custom_half_scheduler_with_warmup(optimizer, num_warmup_steps, num_training_
     Return:
         `torch.optim.lr_scheduler.LambdaLR` with the appropriate schedule.
     """
-
+    
     lr_lambda = partial(
         _get_fp_half_schedule_with_warmup_lr_lambda,
         num_warmup_steps=num_warmup_steps,
@@ -225,7 +225,7 @@ def custom_raise_fall_scheduler_with_warmup(optimizer, num_warmup_steps, num_tra
     Return:
         `torch.optim.lr_scheduler.LambdaLR` with the appropriate schedule.
     """
-
+    
     lr_lambda = partial(
         _get_fp_cosine_raise_and_fall_lr_lambda,
         num_warmup_steps=num_warmup_steps,
@@ -257,7 +257,7 @@ def neftune_forward(self, input: torch.Tensor):
         mag_norm = self.neftune_noise_alpha / torch.sqrt(dims)
         embeddings = embeddings + torch.zeros_like(embeddings).uniform_(-mag_norm, mag_norm)
 
-    return embeddings
+    return embeddings    
 
 
 class FPNEFtuneTrainer(transformers.Trainer):
@@ -267,7 +267,7 @@ def __init__(self,neftune_noise_alpha:float = 0.0, model = None, *args, **kwargs
             model = self._activate_neftune(model)
         super().__init__(model = model, *args, **kwargs)
 
-
+   
     def _activate_neftune(self, model):
         r"""
         Activates the neftune as presented in this code: https://github.com/neelsjain/NEFTune and paper: https://arxiv.org/abs/2310.05914
@@ -290,7 +290,7 @@ def _activate_neftune(self, model):
         embeddings._trl_old_forward = old_forward
 
         return model
-
+    
     def train(self, *args, **kwargs):
         output = super().train(*args, **kwargs)
 
@@ -318,7 +318,7 @@ def __init__(self,neftune_noise_alpha:float = 0.0, model = None, *args, **kwargs
             model = self._activate_neftune(model)
         super().__init__(model = model, *args, **kwargs)
 
-
+   
     def _activate_neftune(self, model):
         r"""
         Activates the neftune as presented in this code: https://github.com/neelsjain/NEFTune and paper: https://arxiv.org/abs/2310.05914
@@ -341,7 +341,7 @@ def _activate_neftune(self, model):
         embeddings._trl_old_forward = old_forward
 
         return model
-
+    
     def train(self, *args, **kwargs):
         output = super().train(*args, **kwargs)
 
@@ -364,19 +364,19 @@ def train(self, *args, **kwargs):
 
     def create_scheduler(self, num_training_steps: int, optimizer: torch.optim.Optimizer = None):
         #Setup the scheduler. The optimizer of the trainer must have been set up either before this method is called or passed as an argument.
-
+        
         num_train_epochs = self.args.num_train_epochs
         num_warmup_steps=self.args.get_warmup_steps(num_training_steps)
         num_firstepoch_steps = math.ceil(num_training_steps/num_train_epochs)
-        num_warmup_acc = num_warmup_steps*self.args.gradient_accumulation_steps
+        num_warmup_acc = num_warmup_steps*self.args.gradient_accumulation_steps 
         num_firstepoch_steps_acc = num_firstepoch_steps*self.args.gradient_accumulation_steps
         num_training_steps_acc = num_training_steps*self.args.gradient_accumulation_steps
 
         custom_scheduler_params.update({'dynamic_scheduler_stop': False})
-
+ 
         print (f"Warm-up steps aligned to Gradient accumulation ({self.args.gradient_accumulation_steps}) = {num_warmup_acc} actual warmup steps")
         if self.args.lr_scheduler_type == 'cosine':
-
+            
             num_warmup_acc_min = min(num_warmup_acc, num_firstepoch_steps_acc)
 
             if num_warmup_acc>num_firstepoch_steps_acc:
@@ -388,13 +388,13 @@ def create_scheduler(self, num_training_steps: int, optimizer: torch.optim.Optim
             self.lr_scheduler = custom_cosine_scheduler_with_warmup(
                     optimizer=self.optimizer if optimizer is None else optimizer,
                     num_warmup_steps=num_warmup_steps,
-                    num_training_steps=num_training_steps,
+                    num_training_steps=num_training_steps, 
                     num_firstepoch_steps = num_firstepoch_steps,
                 )
             self._created_lr_scheduler = True
             return self.lr_scheduler
         elif self.args.lr_scheduler_type == 'constant':
-
+           
             half_step_acc = num_training_steps_acc//2
             num_warmup_acc_min = min(num_warmup_acc, half_step_acc)
 
@@ -407,15 +407,15 @@ def create_scheduler(self, num_training_steps: int, optimizer: torch.optim.Optim
             self.lr_scheduler = custom_half_scheduler_with_warmup(
                     optimizer=self.optimizer if optimizer is None else optimizer,
                     num_warmup_steps=num_warmup_steps,
-                    num_training_steps=num_training_steps,
+                    num_training_steps=num_training_steps, 
                     num_firstepoch_steps = num_firstepoch_steps,
                 )
             self._created_lr_scheduler = True
             return self.lr_scheduler
         elif self.args.lr_scheduler_type == 'constant_with_warmup':
-
+           
             half_step_acc = num_training_steps_acc//2
-
+            
             if num_warmup_steps>0:
                 print(f"Warmup doesn't apply to this scheduler [Raise-Fall]")
 
@@ -424,10 +424,10 @@ def create_scheduler(self, num_training_steps: int, optimizer: torch.optim.Optim
             self.lr_scheduler = custom_raise_fall_scheduler_with_warmup(
                     optimizer=self.optimizer if optimizer is None else optimizer,
                     num_warmup_steps=num_warmup_steps,
-                    num_training_steps=num_training_steps,
+                    num_training_steps=num_training_steps, 
                     num_firstepoch_steps = num_firstepoch_steps,
                 )
             self._created_lr_scheduler = True
-            return self.lr_scheduler
+            return self.lr_scheduler        
         else:
-            return  super().create_scheduler(num_training_steps=num_training_steps, optimizer=optimizer)
+            return  super().create_scheduler(num_training_steps=num_training_steps, optimizer=optimizer)
\ No newline at end of file
diff --git a/extensions/Training_PRO/matplotgraph.py b/extensions/Training_PRO/matplotgraph.py
index 10199ab69a..b30bee8365 100644
--- a/extensions/Training_PRO/matplotgraph.py
+++ b/extensions/Training_PRO/matplotgraph.py
@@ -5,7 +5,7 @@ def create_graph(lora_path, lora_name):
     try:
         import matplotlib.pyplot as plt
         from matplotlib.ticker import ScalarFormatter
-
+        
         peft_model_path = f'{lora_path}/training_graph.json'
         image_model_path = f'{lora_path}/training_graph.png'
         # Check if the JSON file exists
@@ -20,7 +20,7 @@ def create_graph(lora_path, lora_name):
 
             # Create the line chart
             fig, ax1 = plt.subplots(figsize=(10, 6))
-
+        
 
             # Plot y1 (learning rate) on the first y-axis
             ax1.plot(x, y1, 'b-', label='Learning Rate')
@@ -57,6 +57,6 @@ def create_graph(lora_path, lora_name):
             print(f"Graph saved in {image_model_path}")
         else:
             print(f"File 'training_graph.json' does not exist in the {lora_path}")
-
+      
     except ImportError:
         print("matplotlib is not installed. Please install matplotlib to create PNG graphs")
diff --git a/extensions/Training_PRO/script.py b/extensions/Training_PRO/script.py
index 05a879f6cc..cb11a8df0a 100644
--- a/extensions/Training_PRO/script.py
+++ b/extensions/Training_PRO/script.py
@@ -109,12 +109,12 @@ def ui():
                             copy_from = gr.Dropdown(label='Copy parameters from', value='None', choices=get_available_loras_local(non_serialized_params['Lora_sortedByTime']), elem_classes=['slim-dropdown'])
                             create_refresh_button(copy_from, lambda: None, lambda: {'choices': get_available_loras_local(non_serialized_params['Lora_sortedByTime'])}, 'refresh-button')
                     with gr.Column():
-                        sort_byTime = gr.Checkbox(label='Sort list by Date', value=False, info='Sorts Loras by date created.', elem_classes=['no-background'])
+                        sort_byTime = gr.Checkbox(label='Sort list by Date', value=False, info='Sorts Loras by date created.', elem_classes=['no-background'])                        
 
                 with gr.Row():
                     with gr.Column(scale=5):
                         lora_name = gr.Textbox(label='Name', info='The name of your new LoRA file')
-
+    
                     with gr.Column():
                         always_override = gr.Checkbox(label='Override Existing Files', value=False, info='If the name is the same, checking will replace the existing file, and unchecking will load and continue from it (the rank must be the same).', elem_classes=['no-background'])
 
@@ -132,14 +132,14 @@ def ui():
                         epochs = gr.Number(label='Epochs', value=3, info='Number of times every entry in the dataset should be fed into training. So 1 means feed each item in once, 5 means feed it in five times, etc.')
                         learning_rate = gr.Textbox(label='Learning Rate', value='3e-4', info='In scientific notation. 3e-4 is a good starting base point. 1e-2 is extremely high, 1e-6 is extremely low.')
                         lr_scheduler_type = gr.Dropdown(label='LR Scheduler', value='linear', choices=['linear', 'constant', 'constant_with_warmup', 'cosine', 'cosine_with_restarts', 'polynomial', 'inverse_sqrt', 'FP_low_epoch_annealing', 'FP_half_time_annealing','FP_raise_fall_creative'], info='Learning rate scheduler - defines how the learning rate changes over time. Custom schedulers: FP_low_epoch_annealing, FP_half_time_annealing, FP_raise_fall_creative (see README)', elem_classes=['slim-dropdown'])
-
+                        
                 with gr.Accordion(label='Checkpoints', open=True):
                     with gr.Row():
                         with gr.Column():
                             save_steps = gr.Number(label='Save every n steps', value=0, info='A checkpoint will be saved every n steps and at each Epoch boundary. (0 = OFF)')
-                        with gr.Column():
-                            save_steps_under_loss = gr.Slider(label='Save at 10% Loss change', value=1.8, minimum=0.0, maximum=3.0, step=0.1, info="Saves checkpoints at (or bellow) this loss and then each time loss falls by at least 10% This works independently from 'Save every n steps'")
-                    with gr.Row():
+                        with gr.Column():    
+                            save_steps_under_loss = gr.Slider(label='Save at 10% Loss change', value=1.8, minimum=0.0, maximum=3.0, step=0.1, info="Saves checkpoints at (or bellow) this loss and then each time loss falls by at least 10% This works independently from 'Save every n steps'")    
+                    with gr.Row():        
                         save_chackpoint_now = gr.Button('Queue Checkpoint Now')
 
                 with gr.Accordion(label='Advanced Options', open=True):
@@ -148,7 +148,7 @@ def ui():
                             warmup_steps = gr.Number(label='Warmup Steps', value=100, info='Number of max steps used for a linear warmup. Reduces early over-fitting by the first training blocks. Value has precedent over Warmup Ratio. Aligns to the closest multiple of graddient accumulation')
                             warmup_ratio = gr.Slider(label='Warmup Ratio', minimum=0.0, maximum=0.2, step=0.025, value=0.0, info='Ratio of total training steps that will be used for a linear warmup. It applies only if Warmup Step is 0.')
                             neft_noise_alpha = gr.Slider(label='NEFtune noise scale', minimum=0.0, maximum=15, step=1, value=0.0, info='Add noise to the training to improve generalization. [0 - OFF, Starting value to experiment: 5]')
-                            training_projection = gr.Radio(value = train_choices[4], label='LLaMA Target Projections', info='Change the targets (LORA is typically q-v)', choices=train_choices)
+                            training_projection = gr.Radio(value = train_choices[4], label='LLaMA Target Projections', info='Change the targets (LORA is typically q-v)', choices=train_choices)    
                             lora_dropout = gr.Slider(label='LoRA Dropout', minimum=0.0, maximum=1.0, step=0.025, value=0.05, info='Percentage probability for dropout of LoRA layers. This can help reduce overfitting. Most users should leave at default.')
                             optimizer = gr.Dropdown(label='Optimizer', value='adamw_torch', choices=['adamw_hf', 'adamw_torch', 'adamw_torch_fused', 'adamw_torch_xla', 'adamw_apex_fused', 'adafactor', 'adamw_bnb_8bit', 'adamw_anyprecision', 'sgd', 'adagrad'], info='Different optimizer implementation options, for advanced users. Effects of different options are not well documented yet.', elem_classes=['slim-dropdown'])
 
@@ -157,10 +157,10 @@ def ui():
                             add_bos_token = gr.Checkbox(label='Add BOS token', value=True, info="Adds BOS token for each dataset item")
                             add_eos_token = gr.Checkbox(label='Add EOS token', value=False, info="Adds EOS token for each dataset item")
                             add_eos_token_type = gr.Dropdown(label='EOS placement (Text file)', choices=['Every Block', 'Hard Cut Blocks Only'], value='Every Block', info='', allow_custom_value = False)
-
+                            
                             higher_rank_limit = gr.Checkbox(label='Enable higher ranks', value=False, info='If checked, changes Rank/Alpha slider above to go much higher. This will not work without a datacenter-class GPU.')
                             report_to = gr.Radio(label="Save detailed logs with", value="None", choices=["None", "wandb", "tensorboard"], interactive=True)
-                # for future
+                # for future            
                 #with gr.Accordion(label='Dynamic Scheduler', open = False):
                 #    ds_min_epochs = gr.Number(label='Minimum Epochs', value='1', info='Minimum epochs that will be always performed before ramp down can be triggered')
                 #    ds_max_epochs = gr.Number(label='Maximum Epochs (fallback)', value='50', info='Maximum Epochs before the training will bail out completely (should be a large number)')
@@ -168,7 +168,7 @@ def ui():
                 #    ds_loss_rolling_window = gr.Number(label='Loss rolling average', value='4', info='Calculate loss by averaging last x numbers to avoid jumps and noise')
                 #    ds_epochs_to_ramp = gr.Slider(label='Ramp down ratio', minimum=0.0, maximum=2.0, step=0.1, value=1.00, info='How long the ramp down will last relative to ellapsed steps (before trigger)')
                 #    gr.Markdown('These are settings for FP_dynamic_loss_trigger scheduler. The scheduler will do warm up, then hold constant untill a loss falls under Trigger Loss, then it will commence linear ramp down schedule and stop. The length of ramp down is set by Ramp down ratio where (ramp down steps) = ratio * (elapsed steps). (The time to completition shown will be very high untill ramp down is triggered.)')
-
+                        
 
             with gr.Column():
                 with gr.Tab(label='Formatted Dataset'):
@@ -217,7 +217,7 @@ def ui():
                             cutoff_len = gr.Slider(label='Chunk Length (Cutoff Length)', minimum=32, maximum=2048, value=256, step=32, info='The maximum length of a chunk (in tokens). Applies to both JSON dataset and text files. Higher values require much more VRAM.')
                 with gr.Row():
                     with gr.Column():
-                        check_dataset_btn = gr.Button('Verify Dataset/Text File and suggest data entries')
+                        check_dataset_btn = gr.Button('Verify Dataset/Text File and suggest data entries')    
                         check_dataset_txt = gr.Textbox(label='Dataset info', value='')
 
                 with gr.Row():
@@ -227,8 +227,8 @@ def ui():
                 with gr.Accordion(label="Graph", open=True):
                     with gr.Row():
                         # show_actions_button = False - we use old gradio
-                        plot_graph = gr.LinePlot(x="epoch", y="value", title="Loss Metrics", overlay_point=True, tooltip=["epoch", "value"], x_lim=[0, 1], y_lim=[0, 3.5], width=500, height=250)
-
+                        plot_graph = gr.LinePlot(x="epoch", y="value", title="Loss Metrics", overlay_point=True, tooltip=["epoch", "value"], x_lim=[0, 1], y_lim=[0, 3.5], width=500, height=250) 
+ 
                 output = gr.Markdown(value="Ready")
 
     with gr.Tab('Perplexity evaluation', elem_id='evaluate-tab'):
@@ -267,7 +267,7 @@ def fix_old_version(batch_size_val,micro_batch_size_val, grad_accumulation_val):
 
         return grad_accumulation_val
 
-
+    
     copy_from.change(partial(do_copy_params, all_params= all_params), copy_from, all_params).then(fix_old_version,[batch_size,micro_batch_size, grad_accumulation],grad_accumulation)
     start_button.click(do_train, all_params, [output,plot_graph])
     stop_button.click(do_interrupt, None, None, queue=False)
@@ -306,8 +306,8 @@ def check_dataset(save_steps:int, micro_batch_size: int, epochs: int, cutoff_len
         if shared.tokenizer is None:
             yield "Tokenizer is not available. Please Load some Model first."
             return
-
-
+        
+        
         if raw_text_file not in ['None', '']:
             logger.info("Loading Text file...")
             fullpath = clean_path('user_data/training/datasets', f'{raw_text_file}')
@@ -329,8 +329,8 @@ def check_dataset(save_steps:int, micro_batch_size: int, epochs: int, cutoff_len
                 except:
                     yield f"{raw_text_file}.txt doesn't seem to exsist anymore... check your user_data/training/datasets folder"
                     return
-
-
+            
+ 
             if min_chars<0:
                 min_chars = 0
 
@@ -343,11 +343,11 @@ def check_dataset(save_steps:int, micro_batch_size: int, epochs: int, cutoff_len
             total_blocks = len(text_chunks)
             result = f"Text: ({raw_text_file}.txt) has {total_blocks} blocks (Block Size {cutoff_len} tokens)"
             del text_chunks
-
+       
         else:
             if dataset in ['None', '']:
                 yield "Select dataset or text file."
-                return
+                return 
 
             if format in ['None', '']:
                 yield "Select format choice for dataset."
@@ -382,8 +382,8 @@ def generate_and_tokenize_prompt(data_point):
 
             logger.info("Loading JSON datasets...")
             data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
-
-            data_keys = []
+            
+            data_keys = [] 
 
             if data:
                 if 'train' in data:  # Check if the 'train' split exists in the dataset
@@ -400,11 +400,11 @@ def generate_and_tokenize_prompt(data_point):
             #for options, data in format_data.items():
             #    format_keys = options.split(',')
             #    result += f"{format_keys}, "
-            #result = result.rstrip()
-            #result = result.rstrip(',')
+            #result = result.rstrip()    
+            #result = result.rstrip(',')  
 
         if total_blocks>0:
-            number_ofSteps = int(math.ceil(total_blocks / micro_batch_size) * epochs)
+            number_ofSteps = int(math.ceil(total_blocks / micro_batch_size) * epochs) 
             num_stepsPer_epoch = int(math.ceil(number_ofSteps/epochs))
             min_warm = math.ceil(100 / grad_accumulation)
 
@@ -415,20 +415,20 @@ def generate_and_tokenize_prompt(data_point):
             save_each_n_max = int(math.ceil(number_ofSteps/5))
             gradient_accumulation_max = int(total_blocks)//micro_batch_size
 
-
+ 
             result += f"\n[Batch Size: {micro_batch_size}, Epochs: {epochs}, Gradient Accumulation: {grad_accumulation}]\n"
             result += f"Total number of steps: {number_ofSteps}\n"
             result += f"Steps per each Epoch: {num_stepsPer_epoch}\n"
             result += f"Suggestions:\n"
             result += f"Checkpoints: Save every {save_each_n_min} - {save_each_n_max} steps (Current: {int(save_steps)})\n"
             result += f"Warmup steps: {warmup_steps_suggest} (Current: {int(warmup_steps)})"
-            if gradient_accumulation_max < grad_accumulation:
+            if gradient_accumulation_max < grad_accumulation: 
                 result += f"\n\nWARNING: Gradient Accumulation {grad_accumulation} is too high: It should be below {gradient_accumulation_max}"
 
 
         yield result
         return
-
+    
     check_dataset_btn.click(check_dataset, dataset_calc_params ,check_dataset_txt)
 
     # Evaluation events. For some reason, the interrupt event
@@ -449,10 +449,10 @@ def generate_and_tokenize_prompt(data_point):
 
     def reload_lora():
         return gr.Dropdown.update(choices=get_available_loras_local(non_serialized_params['Lora_sortedByTime']))
-
+ 
     # nonserialized items
 
-    sort_byTime.change(lambda x: non_serialized_params.update({"Lora_sortedByTime": x}), sort_byTime, None).then(reload_lora,None,copy_from)
+    sort_byTime.change(lambda x: non_serialized_params.update({"Lora_sortedByTime": x}), sort_byTime, None).then(reload_lora,None,copy_from) 
     #debug_slicer.change(lambda x: non_serialized_params.update({"debug_slicer": x}), debug_slicer, None)
 
     def update_dataset():
@@ -482,7 +482,7 @@ def do_copy_params(lora_name: str, all_params):
         else:
             params = {}
     else:
-        params = {}
+        params = {}        
 
     result = list()
     for i in range(0, len(PARAMETERS)):
@@ -521,8 +521,7 @@ def backup_adapter(input_folder):
 
             # Create the new subfolder
             subfolder_path = Path(f"{input_folder}/{creation_date_str}")
-            if not subfolder_path.exists():
-                subfolder_path.mkdir(parents=True, exist_ok=True)
+            subfolder_path.mkdir(parents=True, exist_ok=True)
 
             # Check if the file already exists in the subfolder
             backup_adapter_file = Path(f"{input_folder}/{creation_date_str}/adapter_model.bin")
@@ -608,7 +607,7 @@ def do_train(lora_name: str, always_override: bool, save_steps: int, micro_batch
     shared.tokenizer.padding_side = "left"
 
     def encode(text, prepend_bos_token):
-
+       
         result = shared.tokenizer.encode(text, truncation=True, max_length=cutoff_len)
         # Check if the first two tokens are BOS
         if len(result) >= 2 and result[:2] == [shared.tokenizer.bos_token_id, shared.tokenizer.bos_token_id]:
@@ -627,7 +626,7 @@ def tokenize(prompt, append_eos_token=False, prepend_bos_token = False):
                 input_ids.append(shared.tokenizer.eos_token_id)
 
             input_ids = [shared.tokenizer.pad_token_id] * (cutoff_len - len(input_ids)) + input_ids
-
+            
             labels = [1] * len(input_ids)
         else:
             ind = prompt.index(train_only_after) + len(train_only_after)
@@ -654,7 +653,7 @@ def tokenize(prompt, append_eos_token=False, prepend_bos_token = False):
         }
 
     train_template.clear()
-
+            
     #reset stuff
     print(f"*** LoRA: {lora_name} ***")
     non_serialized_params.update({"stop_at_loss": stop_at_loss})
@@ -666,7 +665,7 @@ def tokenize(prompt, append_eos_token=False, prepend_bos_token = False):
     non_serialized_params.update({"checkpoint_offset": 0})
     non_serialized_params.update({"epoch_offset": 0})
     train_log_graph.clear()
-
+  
      # == Prep the dataset, format, etc ==
     if raw_text_file not in ['None', '']:
         train_template["template_type"] = "raw_text"
@@ -686,8 +685,8 @@ def tokenize(prompt, append_eos_token=False, prepend_bos_token = False):
         else:
             with open(clean_path('user_data/training/datasets', f'{raw_text_file}.txt'), 'r', encoding='utf-8') as file:
                 raw_text = file.read().replace('\r', '')
-
-        # FPHAM PRECISE SLICING
+        
+        # FPHAM PRECISE SLICING        
         if min_chars<0:
             min_chars = 0
 
@@ -704,7 +703,7 @@ def tokenize(prompt, append_eos_token=False, prepend_bos_token = False):
 
         train_data = Dataset.from_list([tokenize(x, add_EOS_to_all, add_bos_token) for x in text_chunks])
         if add_EOS_to_all:
-            print(f"Added EOS to {len(text_chunks)} blocks")
+            print(f"Added EOS to {len(text_chunks)} blocks") 
 
         print(f"All Data Blocks: {len(text_chunks)}")
 
@@ -746,7 +745,7 @@ def generate_and_tokenize_prompt(data_point):
         data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
         train_data = data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
 
-        print(f"BOS: {add_bos_token} EOS: {add_eos_token}")
+        print(f"BOS: {add_bos_token} EOS: {add_eos_token}") 
         print(f"Data Blocks: {train_data.num_rows}")
 
         if eval_dataset == 'None':
@@ -784,7 +783,7 @@ def generate_and_tokenize_prompt(data_point):
             print(f"Method: {RED}QLORA{RESET}")
             prepare_model_for_kbit_training(shared.model)
         else:
-            print(f"Method: {RED}LoRA{RESET}")
+            print(f"Method: {RED}LoRA{RESET}")    
 
     # base model is now frozen and should not be reused for any other LoRA training than this one
     shared.model_dirty_from_training = True
@@ -797,7 +796,7 @@ def generate_and_tokenize_prompt(data_point):
     elif training_projection==train_choices[2]:
         model_to_lora_modules[model_id] = ["q_proj","k_proj", "v_proj"]
     elif training_projection==train_choices[3]:
-        model_to_lora_modules[model_id] = ["k_proj", "v_proj", "down_proj"]
+        model_to_lora_modules[model_id] = ["k_proj", "v_proj", "down_proj"]        
     else:
         model_to_lora_modules[model_id] = ["q_proj", "v_proj"]
 
@@ -828,9 +827,9 @@ def generate_and_tokenize_prompt(data_point):
             set_peft_model_state_dict(lora_model, state_dict_peft)
 
             print(f" + Continue Training on {RED}{lora_file_path}/adapter_model.bin{RESET}")
-
+            
             #load training_log.json if exist
-
+           
             if Path(f"{lora_file_path}/training_log.json").is_file():
                 with open(f"{lora_file_path}/training_log.json", 'r') as json_file:
                     json_ilog = json.load(json_file)
@@ -841,13 +840,13 @@ def generate_and_tokenize_prompt(data_point):
                         if key=='epoch':
                             non_serialized_params.update({"epoch_offset": value})
                             print(f" + Epoch offset: {RED}{non_serialized_params['epoch_offset']}{RESET}")
-
+           
 
             if Path(f"{lora_file_path}/training_graph.json").is_file():
                 try:
                     with open(f"{lora_file_path}/training_graph.json", 'r') as json_file:
                         train_log_graph = json.load(json_file)
-                        print(" + Training Graph loaded")
+                        print(" + Training Graph loaded")   
                 except:
                     print(f"Can't read training_graph")
 
@@ -877,72 +876,72 @@ def on_step_begin(self, args: transformers.TrainingArguments, state: transformer
             else:
                 current_loss = float(train_log.get('loss', 0.0))
                 current_epoch_int = int(float(train_log.get('epoch', 0.0)))
-
+              
                 force_save = False
 
                 current_steps_offset = tracked.current_steps + non_serialized_params['checkpoint_offset']
 
-                folder_save = f"checkpoint-{current_steps_offset}"
+                folder_save = f"checkpoint-{current_steps_offset}"    
 
                 # save if triggered by user
                 if non_serialized_params['save_checkpoint_now']:
                     force_save = True
                     non_serialized_params.update({"save_checkpoint_now": False})
                     print(f"\033[1;31;1mSave Checkpoint manually trigerred.\033[0;37;0m")
-                    folder_save = f"checkpoint-{current_steps_offset}-user"
+                    folder_save = f"checkpoint-{current_steps_offset}-user"  
 
                 patience = 3     # Set the number of consecutive steps for tracking stability
-
+                
                 if gradient_accumulation_steps==1:
                     patience = 4
 
                 min_steps = ssteps10
 
-                # Save each time the loss is below the threshold
+                # Save each time the loss is below the threshold 
                 if current_loss < non_serialized_params['save_steps_under_loss'] and current_loss > 0 and state.global_step > min_steps:
                     current_stability = non_serialized_params['current_stability']
                     current_stability += 1
-                    non_serialized_params.update({"current_stability": current_stability})
+                    non_serialized_params.update({"current_stability": current_stability}) 
 
                     if current_stability >= patience:
                         current_stability = 0
-                        non_serialized_params.update({"current_stability": current_stability})
+                        non_serialized_params.update({"current_stability": current_stability})     
                         current_loss_dec = round(current_loss, 2)
                         loss_str = f"{current_loss_dec:.2f}"
                         loss_str = loss_str.replace('.', '_')
                         new_save = (current_loss_dec-0.1) + 0.01
                         non_serialized_params.update({"save_steps_under_loss": new_save})
 
-                        folder_save = f"checkpoint-{current_steps_offset}-loss-{loss_str}"
-                        force_save = True
-
+                        folder_save = f"checkpoint-{current_steps_offset}-loss-{loss_str}" 
+                        force_save = True   
 
+                   
                 else:
                     # Reset stability if the loss goes above the threshold
-                    non_serialized_params.update({"current_stability": 0})
+                    non_serialized_params.update({"current_stability": 0})   
 
                 # Save full epochs
-                if actual_save_steps>0 and current_epoch_int > non_serialized_params['save_epochs'] and state.global_step > min_steps:
-
+                if actual_save_steps>0 and current_epoch_int > non_serialized_params['save_epochs'] and state.global_step > min_steps: 
 
+                    
                     current_epoch_offset = current_epoch_int
-
+                    
                     if non_serialized_params['epoch_offset'] > 0:
                         current_epoch_offset = current_epoch_int + round(non_serialized_params['epoch_offset'], 2)
-
+                    
                     ep_off_str = f"{current_epoch_offset}"
                     ep_off_str = ep_off_str.replace('.', '_')
-                    folder_save = f"checkpoint-{current_steps_offset}-epoch-{ep_off_str}"
+                    folder_save = f"checkpoint-{current_steps_offset}-epoch-{ep_off_str}" 
 
                     non_serialized_params.update({"save_epochs": current_epoch_int})
                     force_save = True
 
                 # save each actual_save_steps
                 if state.global_step > 0 and actual_save_steps > 0 and state.global_step % actual_save_steps == 0:
-                    folder_save = f"checkpoint-{current_steps_offset}"
-                    force_save = True
+                    folder_save = f"checkpoint-{current_steps_offset}"  
+                    force_save = True   
 
-                if force_save:
+                if force_save:       
                     lora_model.save_pretrained(f"{lora_file_path}/{folder_save}/", safe_serialization = non_serialized_params['safe_serialization'])
                     print(f"\033[1;30;40mStep: {tracked.current_steps:6} \033[0;37;0m Saved: [{folder_save}]")
                     # Save log
@@ -951,7 +950,7 @@ def on_step_begin(self, args: transformers.TrainingArguments, state: transformer
                     # == Save training prompt ==
                     with open(f"{lora_file_path}/{folder_save}/training_prompt.json", 'w', encoding='utf-8') as file:
                         json.dump(train_template, file, indent=2)
-
+                
 
         def on_substep_end(self, args: transformers.TrainingArguments, state: transformers.TrainerState, control: transformers.TrainerControl, **kwargs):
             tracked.current_steps += 1
@@ -976,7 +975,7 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
                 print(f"\033[1;30;40mStep: {tracked.current_steps:6} [+{non_serialized_params['checkpoint_offset']}] \033[0;37;0m", end='')
             else:
                 print(f"\033[1;30;40mStep: {tracked.current_steps:6} \033[0;37;0m", end='')
-
+            
             graphentry = {
                 'current_steps': int(train_log.get('current_steps_adjusted',0)),
                 'loss': float(train_log.get('loss', 0.0)),
@@ -987,7 +986,7 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
             cur_loss = float(train_log.get('loss', 0.0))
             cur_lr = float(train_log.get('learning_rate', 0.0))
             cur_epoch = float(train_log.get('epoch', 0.0))
-
+            
             if len(statistics['loss']) == 1:
                 first_epoch = statistics['loss'][0]['epoch']
                 first_value = statistics['loss'][0]['value']
@@ -1014,7 +1013,7 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
 
     # FPHAM SAMPLE REQ Transformers error handling
     gradient_accumulation_max = int(train_data.num_rows)//micro_batch_size
-
+    
     if gradient_accumulation_max < gradient_accumulation_steps:
         print(f"{RED}WARNING:{RESET} Current gradient accumulation is {RED}too high{RESET} for the amount of training data.")
         print(f"Gradient accumulation: {gradient_accumulation_steps} should be less than: {gradient_accumulation_max}. {RED}This could crash Accelerate/Transformers{RESET}")
@@ -1042,9 +1041,9 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
     elif lr_scheduler_type =='FP_raise_fall_creative':
         custom_scheduller = True
         lr_scheduler_type_arg = 'constant_with_warmup'
-
+    
     #gradient_checkpointing=True
-
+    
     args=transformers.TrainingArguments(
             report_to=report_to if report_to != "None" else None,
             per_device_train_batch_size=micro_batch_size,
@@ -1096,7 +1095,7 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
             data_collator=transformers.DataCollatorForLanguageModeling(shared.tokenizer, mlm=False),
             callbacks=list([Callbacks()])
         )
-
+    
     # END OF FPHAM CUSTOM SCHEDULER
 
     lora_model.config.use_cache = False
@@ -1142,7 +1141,7 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
     if stop_at_loss > 0:
         print(f"Monitoring loss {RED}(Auto-Stop at: {stop_at_loss}){RESET}")
 
-
+    
 
     if WANT_INTERRUPT:
         yield "Interrupted before start.", zero_pd
@@ -1158,9 +1157,7 @@ def log_train_dataset(trainer):
                 decoded_entries.append({"value": decoded_text})
 
             # Write the log file
-            if not Path('user_data/logs').exists():
-                Path('user_data/logs').mkdir(exist_ok=True)
-
+            Path('user_data/logs').mkdir(exist_ok=True)
             with open(Path('user_data/logs/train_dataset_sample.json'), 'w') as json_file:
                 json.dump(decoded_entries, json_file, indent=4)
 
@@ -1194,7 +1191,7 @@ def threaded_run():
         else:
             max_value = 3.5
             last_epoch = 0
-            first_epoch = 0
+            first_epoch = 0           
 
         if WANT_INTERRUPT:
 
@@ -1213,7 +1210,7 @@ def threaded_run():
             time_elapsed = time.perf_counter() - start_time
             lastloss = float(train_log.get('loss', 0.0))
 
-            non_serialized_params.update({"training_loop": True})
+            non_serialized_params.update({"training_loop": True})               
 
             if lastloss > 0:
                 lastloss_str = f", ... Current Loss: `{lastloss:.2f}`"
@@ -1235,7 +1232,7 @@ def threaded_run():
             if stop_at_loss != non_serialized_params['stop_at_loss']:
                 stop_at_loss = non_serialized_params['stop_at_loss']
                 print(f"Stop at loss changed {RED}(Auto-Stop at: {stop_at_loss}){RESET}")
-
+            
             losses = gr.LinePlot.update(
 				value = pd.DataFrame(statistics['loss']),
                 x="epoch", y="value",
@@ -1243,7 +1240,7 @@ def threaded_run():
                 overlay_point=True, tooltip=["epoch", "value"],
 				x_lim=[first_epoch,last_epoch], y_lim=[0,max_value],
                 width=500, height=250 )
-
+				
 
             yield f"Running... **{tracked.current_steps}** / **{tracked.max_steps}** ... {timer_info}, {format_time(time_elapsed)} / {format_time(total_time_estimate)} ... {format_time(total_time_estimate - time_elapsed)} remaining {lastloss_str}", losses
 
@@ -1259,7 +1256,7 @@ def threaded_run():
     else:
         max_value = 3.5
         last_epoch = 0
-        first_epoch = 0
+        first_epoch = 0 
 
     return_pd = gr.LinePlot.update(
         value = pd.DataFrame(statistics['loss']),
diff --git a/extensions/Training_PRO/train_utils.py b/extensions/Training_PRO/train_utils.py
index d1a45c1795..799948802b 100644
--- a/extensions/Training_PRO/train_utils.py
+++ b/extensions/Training_PRO/train_utils.py
@@ -20,7 +20,7 @@ def list_subfoldersByTime(directory):
     if not directory.endswith('/'):
         directory += '/'
     subfolders = []
-    subfolders.append('None')
+    subfolders.append('None') 
     path = directory
     name_list = os.listdir(path)
     full_list = [os.path.join(path,i) for i in name_list]
@@ -37,19 +37,19 @@ def list_subfoldersByTime(directory):
     return subfolders
 
 def get_available_loras_local(_sortedByTime):
-
+    
     model_dir = shared.args.lora_dir  # Update with the appropriate directory path
     subfolders = []
     if _sortedByTime:
         subfolders = list_subfoldersByTime(model_dir)
     else:
-        subfolders = utils.get_available_loras()
+        subfolders = utils.get_available_loras()        
 
     return subfolders
 
 
 # FPHAM SPLIT BY SENTENCE BLOCK ===============
-
+     
 def split_sentences(text: str, cutoff_len: int):
     sentences = []
     sentence = ''
@@ -57,24 +57,24 @@ def split_sentences(text: str, cutoff_len: int):
     abbreviations = ['Mr. ', 'Mrs. ', 'Dr. ', 'Ms. ', 'St. ', 'Prof. ', 'Jr. ', 'Ltd. ', 'Capt. ', 'Col. ', 'Gen. ', 'Ave. ', 'Blvd. ', 'Co. ', 'Corp. ', 'Dept. ', 'Est. ', 'Gov. ', 'Inc. ', 'Ph.D. ', 'Univ. ']
     errors = 0
     max_cut = cutoff_len-1
-    prev_char = ''
+    prev_char = ''  
 
     for char in text:
         sentence += char
 
-
+    
         if (any(sentence.endswith(delimiter) for delimiter in delimiters) and
-            not (prev_char.isupper() and len(sentence) >= 3 and sentence[-3] != ' ') and
+            not (prev_char.isupper() and len(sentence) >= 3 and sentence[-3] != ' ') and 
             not any(sentence.endswith(abbreviation) for abbreviation in abbreviations)):
             tokens = shared.tokenizer.encode(sentence)
-
+            
             if len(tokens) > max_cut:
                 tokens = tokens[:max_cut]
                 sentence = shared.tokenizer.decode(tokens, skip_special_tokens=True)
                 errors = errors + 1
 
             sentences.append({'text': sentence, 'size': len(tokens)})
-
+            
             sentence = ''
 
         prev_char = char
@@ -83,7 +83,7 @@ def split_sentences(text: str, cutoff_len: int):
         tokens = shared.tokenizer.encode(sentence)
         if len(tokens) > max_cut:
             tokens = tokens[:max_cut]
-            sentence = shared.tokenizer.decode(tokens, skip_special_tokens=True)
+            sentence = shared.tokenizer.decode(tokens, skip_special_tokens=True)  
             errors = errors + 1
 
         sentences.append({'text': sentence, 'size': len(tokens)})
@@ -95,16 +95,16 @@ def split_sentences(text: str, cutoff_len: int):
 
 # The goal of following code is to create blocks of text + overlapping blocks while:
 # respects sentence boundaries
-# always uses all the text
+# always uses all the text 
 # hard cut defined by hard_cut_string or  will always end at the end of data block
 # no overlapping blocks will be created across hard cut or across  token
 
 def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, cutoff_len: int, hard_cut_string: str, debug_slicer:bool):
 
     EOSX_str = '' #hardcut placeholder
-    EOS_str = ''
+    EOS_str = '' 
     print("Precise raw text slicer: ON")
-
+    
     cut_string = hard_cut_string.replace('\\n', '\n')
     text = text.replace(cut_string, EOSX_str)
     sentences = split_sentences(text, cutoff_len)
@@ -121,7 +121,7 @@ def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, c
     half_index = 0
 
     for index, item in enumerate(sentences):
-
+        
         if halfcut_length+ item['size'] < half_cut:
             halfcut_length += item['size']
             half_index = index
@@ -130,7 +130,7 @@ def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, c
             halfcut_length = -2 * max_cut
 
 
-        if totalLength + item['size'] < max_cut and not currentSentence.endswith(EOSX_str):
+        if totalLength + item['size'] < max_cut and not currentSentence.endswith(EOSX_str): 
             currentSentence += item['text']
             totalLength += item['size']
         else:
@@ -141,14 +141,14 @@ def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, c
             currentSentence = item['text']
             totalLength = item['size']
             halfcut_length = item['size']
-
-    if len(currentSentence.strip()) > min_chars_cut:
+            
+    if len(currentSentence.strip()) > min_chars_cut:    
         sentencelist.append(currentSentence.strip())
 
     unique_blocks = len(sentencelist)
     print(f"Text Blocks: {unique_blocks}")
 
-    #overlap strategies:
+    #overlap strategies: 
     # don't overlap across HARD CUT (EOSX)
     if overlap:
         for edge_idx in edgeindex:
@@ -162,15 +162,15 @@ def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, c
                 else:
                     #if by chance EOSX is at the end then it's acceptable
                     if currentSentence.endswith(EOSX_str) and len(currentSentence.strip()) > min_chars_cut:
-                            sentencelist.append(currentSentence.strip())
-                    # otherwise don't cross hard cut
+                            sentencelist.append(currentSentence.strip())    
+                    # otherwise don't cross hard cut    
                     elif EOSX_str not in currentSentence and len(currentSentence.strip()) > min_chars_cut:
                         sentencelist.append(currentSentence.strip())
-
+                    
                     currentSentence = ''
                     totalLength = 0
                     break
-
+        
         print(f"+ Overlapping blocks: {len(sentencelist)-unique_blocks}")
 
     num_EOS = 0
@@ -179,7 +179,7 @@ def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, c
             sentencelist[i] = sentencelist[i].replace(EOSX_str, EOS_str)
         else:
             sentencelist[i] = sentencelist[i].replace(EOSX_str, '')
-
+        
         #someone may have had stop strings in the raw text...
         sentencelist[i] = sentencelist[i].replace("", EOS_str)
         num_EOS += sentencelist[i].count(EOS_str)
@@ -193,49 +193,47 @@ def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, c
 
 
     if debug_slicer:
-        # Write the log file
-        if not Path('user_data/logs').exists():
-            Path('user_data/logs').mkdir(exist_ok=True)
-
+                    # Write the log file
+        Path('user_data/logs').mkdir(exist_ok=True)
         sentencelist_dict = {index: sentence for index, sentence in enumerate(sentencelist)}
         output_file = "user_data/logs/sentencelist.json"
         with open(output_file, 'w') as f:
             json.dump(sentencelist_dict, f,indent=2)
-
+        
         print("Saved sentencelist.json in user_data/logs folder")
-
-    return sentencelist
+    
+    return sentencelist   
 
 
 def sliding_block_cut(text: str, min_chars_cut: int, eos_to_hc: bool, cutoff_len: int, hard_cut_string: str, debug_slicer:bool):
 
     EOSX_str = '' #hardcut placeholder
-    EOS_str = ''
+    EOS_str = '' 
     print("Mega Block Overlap: ON")
-
+    
     cut_string = hard_cut_string.replace('\\n', '\n')
     text = text.replace(cut_string, EOSX_str)
     sentences = split_sentences(text, cutoff_len)
 
     print(f"Sentences: {len(sentences)}")
     sentencelist = []
-
+    
     max_cut = cutoff_len-1
 
     #print(f"max_cut: {max_cut}")
     advancing_to = 0
 
     prev_block_lastsentence = ""
-
+    
 
     for i in range(len(sentences)):
         totalLength = 0
         currentSentence = ''
         lastsentence = ""
-
+        
         if i >= advancing_to:
             for k in range(i, len(sentences)):
-
+                
                 current_length = sentences[k]['size']
 
                 if totalLength + current_length <= max_cut and not currentSentence.endswith(EOSX_str):
@@ -247,7 +245,7 @@ def sliding_block_cut(text: str, min_chars_cut: int, eos_to_hc: bool, cutoff_len
                         if prev_block_lastsentence!=lastsentence:
                             sentencelist.append(currentSentence.strip())
                             prev_block_lastsentence = lastsentence
-
+                        
                     advancing_to = 0
                     if currentSentence.endswith(EOSX_str):
                         advancing_to = k
@@ -255,7 +253,7 @@ def sliding_block_cut(text: str, min_chars_cut: int, eos_to_hc: bool, cutoff_len
                     currentSentence = ""
                     totalLength = 0
                     break
-
+            
             if currentSentence != "":
                 if len(currentSentence.strip()) > min_chars_cut:
                     sentencelist.append(currentSentence.strip())
@@ -268,7 +266,7 @@ def sliding_block_cut(text: str, min_chars_cut: int, eos_to_hc: bool, cutoff_len
             sentencelist[i] = sentencelist[i].replace(EOSX_str, EOS_str)
         else:
             sentencelist[i] = sentencelist[i].replace(EOSX_str, '')
-
+        
         #someone may have had stop strings in the raw text...
         sentencelist[i] = sentencelist[i].replace("", EOS_str)
         num_EOS += sentencelist[i].count(EOS_str)
@@ -282,18 +280,16 @@ def sliding_block_cut(text: str, min_chars_cut: int, eos_to_hc: bool, cutoff_len
 
 
     if debug_slicer:
-        # Write the log file
-        if not Path('user_data/logs').exists():
-            Path('user_data/logs').mkdir(exist_ok=True)
-
+                    # Write the log file
+        Path('user_data/logs').mkdir(exist_ok=True)
         sentencelist_dict = {index: sentence for index, sentence in enumerate(sentencelist)}
         output_file = "user_data/logs/sentencelist.json"
         with open(output_file, 'w') as f:
             json.dump(sentencelist_dict, f,indent=2)
-
+        
         print("Saved sentencelist.json in user_data/logs folder")
-
-    return sentencelist
+    
+    return sentencelist   
 
 # Example usage:
 # download_file_from_url('https://example.com/path/to/your/file.ext', '/output/directory')
@@ -330,17 +326,17 @@ def download_file_from_url(url, overwrite, output_dir_in, valid_extensions = {'.
 
         # Send an HTTP GET request to the URL with a timeout
         file_extension = os.path.splitext(filename_lower)[-1]
-
+        
         if file_extension not in valid_extensions:
             yield f"Invalid file extension: {file_extension}. Only {valid_extensions} files are supported."
             return
 
         with session.get(url, stream=True, headers=headers, timeout=10) as r:
-            r.raise_for_status()
+            r.raise_for_status() 
             # total size can be wildly inaccurate
             #total_size = int(r.headers.get('content-length', 0))
-
-            block_size = 1024 * 4
+            
+            block_size = 1024 * 4  
             with open(local_filename, mode) as f:
                 count = 0
                 for data in r.iter_content(block_size):
diff --git a/extensions/coqui_tts/script.py b/extensions/coqui_tts/script.py
index 4f5700d1de..fe6364b9f2 100644
--- a/extensions/coqui_tts/script.py
+++ b/extensions/coqui_tts/script.py
@@ -168,8 +168,7 @@ def setup():
     print("[XTTS] Loading XTTS...")
     model = load_model()
     print("[XTTS] Done!")
-    if not Path(f"{this_dir}/outputs").exists():
-        Path(f"{this_dir}/outputs").mkdir(parents=True, exist_ok=True)
+    Path(f"{this_dir}/outputs").mkdir(parents=True, exist_ok=True)
 
 
 def ui():
diff --git a/extensions/sd_api_pictures/script.py b/extensions/sd_api_pictures/script.py
index d6853664a5..f216da3849 100644
--- a/extensions/sd_api_pictures/script.py
+++ b/extensions/sd_api_pictures/script.py
@@ -168,8 +168,7 @@ def get_SD_pictures(description, character):
 
             variadic = f'{date.today().strftime("%Y_%m_%d")}/{character}_{int(time.time())}'
             output_file = Path(f'extensions/sd_api_pictures/outputs/{variadic}.png')
-            if not output_file.parent.exists():
-                output_file.parent.mkdir(parents=True, exist_ok=True)
+            output_file.parent.mkdir(parents=True, exist_ok=True)
 
             with open(output_file.as_posix(), 'wb') as f:
                 f.write(img_data)
diff --git a/modules/chat.py b/modules/chat.py
index 7770b5cff8..d76835665f 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -973,8 +973,8 @@ def save_history(history, unique_id, character, mode):
         return
 
     p = get_history_file_path(unique_id, character, mode)
-    if not p.parent.exists():
-        p.parent.mkdir(parents=True, exist_ok=True)
+    if not p.parent.is_dir():
+        p.parent.mkdir(parents=True)
 
     with open(p, 'w', encoding='utf-8') as f:
         f.write(json.dumps(history, indent=4, ensure_ascii=False))
@@ -1014,9 +1014,7 @@ def get_paths(state):
             unique_id = datetime.now().strftime('%Y%m%d-%H-%M-%S')
             p = get_history_file_path(unique_id, character, state['mode'])
             logger.warning(f"Moving \"{new_p}\" to \"{p}\"")
-            if not p.parent.exists():
-                p.parent.mkdir(exist_ok=True)
-
+            p.parent.mkdir(exist_ok=True)
             new_p.rename(p)
 
         return Path(f'user_data/logs/chat/{character}').glob('*.json')
@@ -1165,9 +1163,7 @@ def save_last_chat_state(character, mode, unique_id):
     state["last_chats"][key] = unique_id
 
     state_file = Path('user_data/logs/chat_state.json')
-    if not state_file.parent.exists():
-        state_file.parent.mkdir(exist_ok=True)
-
+    state_file.parent.mkdir(exist_ok=True)
     with open(state_file, 'w', encoding='utf-8') as f:
         f.write(json.dumps(state, indent=2))
 
diff --git a/modules/evaluate.py b/modules/evaluate.py
index 65cceda949..4f41c1fc14 100644
--- a/modules/evaluate.py
+++ b/modules/evaluate.py
@@ -27,9 +27,7 @@ def save_past_evaluations(df):
     global past_evaluations
     past_evaluations = df
     filepath = Path('user_data/logs/evaluations.csv')
-    if not filepath.parent.exists():
-        filepath.parent.mkdir(parents=True, exist_ok=True)
-
+    filepath.parent.mkdir(parents=True, exist_ok=True)
     df.to_csv(filepath, index=False)
 
 
diff --git a/modules/prompts.py b/modules/prompts.py
index 1b154cca1e..79d9b56e46 100644
--- a/modules/prompts.py
+++ b/modules/prompts.py
@@ -9,9 +9,7 @@ def load_prompt(fname):
         # Create new file
         new_name = utils.current_time()
         prompt_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
-        if not prompt_path.parent.exists():
-            prompt_path.parent.mkdir(parents=True, exist_ok=True)
-
+        prompt_path.parent.mkdir(parents=True, exist_ok=True)
         initial_content = "In this story,"
         prompt_path.write_text(initial_content, encoding='utf-8')
 
diff --git a/modules/training.py b/modules/training.py
index 2689977f10..2354c39df9 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -241,8 +241,7 @@ def backup_adapter(input_folder):
 
             # Create the new subfolder
             subfolder_path = Path(f"{input_folder}/{creation_date_str}")
-            if not subfolder_path.exists():
-                subfolder_path.mkdir(parents=True, exist_ok=True)
+            subfolder_path.mkdir(parents=True, exist_ok=True)
 
             # Check if the file already exists in the subfolder
             backup_adapter_file = Path(f"{input_folder}/{creation_date_str}/adapter_model.bin")
@@ -677,9 +676,7 @@ def log_train_dataset(trainer):
                 decoded_entries.append({"value": decoded_text})
 
             # Write the log file
-            if not Path('user_data/logs').exists():
-                Path('user_data/logs').mkdir(exist_ok=True)
-
+            Path('user_data/logs').mkdir(exist_ok=True)
             with open(Path('user_data/logs/train_dataset_sample.json'), 'w') as json_file:
                 json.dump(decoded_entries, json_file, indent=4)
 
diff --git a/modules/ui.py b/modules/ui.py
index 1247b3652a..0e8afa8f5a 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -409,9 +409,7 @@ def _perform_debounced_save():
         if _last_interface_state is not None:
             contents = save_settings(_last_interface_state, _last_preset, _last_extensions, _last_show_controls, _last_theme_state, manual_save=False)
             settings_path = Path('user_data') / 'settings.yaml'
-            if not settings_path.parent.exists():
-                settings_path.parent.mkdir(exist_ok=True)
-
+            settings_path.parent.mkdir(exist_ok=True)
             with open(settings_path, 'w', encoding='utf-8') as f:
                 f.write(contents)
     except Exception as e:
diff --git a/modules/ui_default.py b/modules/ui_default.py
index 21a638925d..66174e359e 100644
--- a/modules/ui_default.py
+++ b/modules/ui_default.py
@@ -152,9 +152,7 @@ def autosave_prompt(text, prompt_name):
     """Automatically save the text to the selected prompt file"""
     if prompt_name and text.strip():
         prompt_path = Path("user_data/logs/notebook") / f"{prompt_name}.txt"
-        if not prompt_path.parent.exists():
-            prompt_path.parent.mkdir(parents=True, exist_ok=True)
-
+        prompt_path.parent.mkdir(parents=True, exist_ok=True)
         prompt_path.write_text(text, encoding='utf-8')
 
 
@@ -207,9 +205,7 @@ def handle_new_prompt():
 
     # Create the new prompt file
     prompt_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
-    if not prompt_path.parent.exists():
-        prompt_path.parent.mkdir(parents=True, exist_ok=True)
-
+    prompt_path.parent.mkdir(parents=True, exist_ok=True)
     prompt_path.write_text("In this story,", encoding='utf-8')
 
     return gr.update(choices=utils.get_available_prompts(), value=new_name)
@@ -226,9 +222,7 @@ def handle_delete_prompt_confirm_default(prompt_name):
         new_value = available_prompts[min(current_index, len(available_prompts) - 1)]
     else:
         new_value = utils.current_time()
-        if not Path("user_data/logs/notebook").exists():
-            Path("user_data/logs/notebook").mkdir(parents=True, exist_ok=True)
-
+        Path("user_data/logs/notebook").mkdir(parents=True, exist_ok=True)
         (Path("user_data/logs/notebook") / f"{new_value}.txt").write_text("In this story,")
         available_prompts = [new_value]
 
diff --git a/modules/ui_notebook.py b/modules/ui_notebook.py
index 916e613f80..3a8c9df2f4 100644
--- a/modules/ui_notebook.py
+++ b/modules/ui_notebook.py
@@ -197,9 +197,7 @@ def handle_new_prompt():
 
     # Create the new prompt file
     prompt_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
-    if not prompt_path.parent.exists():
-        prompt_path.parent.mkdir(parents=True, exist_ok=True)
-
+    prompt_path.parent.mkdir(parents=True, exist_ok=True)
     prompt_path.write_text("In this story,", encoding='utf-8')
 
     return gr.update(choices=utils.get_available_prompts(), value=new_name)
@@ -216,9 +214,7 @@ def handle_delete_prompt_confirm_notebook(prompt_name):
         new_value = available_prompts[min(current_index, len(available_prompts) - 1)]
     else:
         new_value = utils.current_time()
-        if not Path("user_data/logs/notebook").exists():
-            Path("user_data/logs/notebook").mkdir(parents=True, exist_ok=True)
-
+        Path("user_data/logs/notebook").mkdir(parents=True, exist_ok=True)
         (Path("user_data/logs/notebook") / f"{new_value}.txt").write_text("In this story,")
         available_prompts = [new_value]
 
diff --git a/modules/utils.py b/modules/utils.py
index b3bac583f6..da5ac9780d 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -160,8 +160,7 @@ def get_available_presets():
 
 def get_available_prompts():
     notebook_dir = Path('user_data/logs/notebook')
-    if not notebook_dir.exists():
-        notebook_dir.mkdir(parents=True, exist_ok=True)
+    notebook_dir.mkdir(parents=True, exist_ok=True)
 
     prompt_files = list(notebook_dir.glob('*.txt'))
     sorted_files = sorted(prompt_files, key=lambda x: x.stat().st_mtime, reverse=True)
diff --git a/server.py b/server.py
index 2b998792bf..7ce3c208f9 100644
--- a/server.py
+++ b/server.py
@@ -10,8 +10,7 @@
 # Set up Gradio temp directory path
 gradio_temp_path = Path('user_data') / 'cache' / 'gradio'
 shutil.rmtree(gradio_temp_path, ignore_errors=True)
-if not gradio_temp_path.exists():
-    gradio_temp_path.mkdir(parents=True, exist_ok=True)
+gradio_temp_path.mkdir(parents=True, exist_ok=True)
 
 # Set environment variables
 os.environ.update({

From 87ae09ecd6517185ff8a470052701698c2c03605 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Jun 2025 07:46:58 -0700
Subject: [PATCH 0733/1701] Improve the basic API examples

---
 docs/12 - OpenAI API.md     | 53 ++++++++++++++++++++++---------------
 extensions/openai/typing.py |  2 +-
 2 files changed, 33 insertions(+), 22 deletions(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index db9befed93..6ff58b1e09 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -1,6 +1,6 @@
 ## OpenAI compatible API
 
-The main API for this project is meant to be a drop-in replacement to the OpenAI API, including Chat and Completions endpoints. 
+The main API for this project is meant to be a drop-in replacement to the OpenAI API, including Chat and Completions endpoints.
 
 * It is 100% offline and private.
 * It doesn't create any logs.
@@ -30,10 +30,10 @@ curl http://127.0.0.1:5000/v1/completions \
   -H "Content-Type: application/json" \
   -d '{
     "prompt": "This is a cake recipe:\n\n1.",
-    "max_tokens": 200,
-    "temperature": 1,
-    "top_p": 0.9,
-    "seed": 10
+    "max_tokens": 512,
+    "temperature": 0.6,
+    "top_p": 0.95,
+    "top_k": 20
   }'
 ```
 
@@ -51,7 +51,9 @@ curl http://127.0.0.1:5000/v1/chat/completions \
         "content": "Hello!"
       }
     ],
-    "mode": "instruct"
+    "temperature": 0.6,
+    "top_p": 0.95,
+    "top_k": 20
   }'
 ```
 
@@ -67,8 +69,11 @@ curl http://127.0.0.1:5000/v1/chat/completions \
         "content": "Hello! Who are you?"
       }
     ],
-    "mode": "chat",
-    "character": "Example"
+    "mode": "chat-instruct",
+    "character": "Example",
+    "temperature": 0.6,
+    "top_p": 0.95,
+    "top_k": 20
   }'
 ```
 
@@ -84,7 +89,9 @@ curl http://127.0.0.1:5000/v1/chat/completions \
         "content": "Hello!"
       }
     ],
-    "mode": "instruct",
+    "temperature": 0.6,
+    "top_p": 0.95,
+    "top_k": 20,
     "stream": true
   }'
 ```
@@ -125,10 +132,11 @@ curl -k http://127.0.0.1:5000/v1/internal/model/list \
 curl -k http://127.0.0.1:5000/v1/internal/model/load \
   -H "Content-Type: application/json" \
   -d '{
-    "model_name": "model_name",
+    "model_name": "Qwen_Qwen3-0.6B-Q4_K_M.gguf",
     "args": {
-      "load_in_4bit": true,
-      "n_gpu_layers": 12
+        "ctx_size": 32768,
+        "flash_attn": true,
+        "cache_type": "q8_0"
     }
   }'
 ```
@@ -150,9 +158,10 @@ while True:
     user_message = input("> ")
     history.append({"role": "user", "content": user_message})
     data = {
-        "mode": "chat",
-        "character": "Example",
-        "messages": history
+        "messages": history,
+        "temperature": 0.6,
+        "top_p": 0.95,
+        "top_k": 20
     }
 
     response = requests.post(url, headers=headers, json=data, verify=False)
@@ -182,9 +191,11 @@ while True:
     user_message = input("> ")
     history.append({"role": "user", "content": user_message})
     data = {
-        "mode": "instruct",
         "stream": True,
-        "messages": history
+        "messages": history,
+        "temperature": 0.6,
+        "top_p": 0.95,
+        "top_k": 20
     }
 
     stream_response = requests.post(url, headers=headers, json=data, verify=False, stream=True)
@@ -218,10 +229,10 @@ headers = {
 
 data = {
     "prompt": "This is a cake recipe:\n\n1.",
-    "max_tokens": 200,
-    "temperature": 1,
-    "top_p": 0.9,
-    "seed": 10,
+    "max_tokens": 512,
+    "temperature": 0.6,
+    "top_p": 0.95,
+    "top_k": 20,
     "stream": True,
 }
 
diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index b28ebb4ec7..6643ed16fb 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -158,7 +158,7 @@ class ChatCompletionRequestParams(BaseModel):
     user_bio: str | None = Field(default=None, description="The user description/personality.")
     chat_template_str: str | None = Field(default=None, description="Jinja2 template for chat.")
 
-    chat_instruct_command: str | None = None
+    chat_instruct_command: str | None = "Continue the chat dialogue below. Write a single reply for the character \"<|character|>\".\n\n<|prompt|>"
 
     continue_: bool = Field(default=False, description="Makes the last bot message in the history be continued instead of starting a new message.")
 

From 73138a29fa5b3b40b9bb5c062a01d0da7bf06c25 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Jun 2025 07:49:24 -0700
Subject: [PATCH 0734/1701] Small change

---
 docs/12 - OpenAI API.md | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index 6ff58b1e09..ec99939764 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -134,9 +134,9 @@ curl -k http://127.0.0.1:5000/v1/internal/model/load \
   -d '{
     "model_name": "Qwen_Qwen3-0.6B-Q4_K_M.gguf",
     "args": {
-        "ctx_size": 32768,
-        "flash_attn": true,
-        "cache_type": "q8_0"
+      "ctx_size": 32768,
+      "flash_attn": true,
+      "cache_type": "q8_0"
     }
   }'
 ```

From 315e06f695d2dc59c292f978c2983cefb7376456 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Jun 2025 07:51:16 -0700
Subject: [PATCH 0735/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_cuda128.txt             | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt      | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index adef3d6d3e..19e5e0fecb 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index d0e9aa88b8..ebef87a656 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 609f449e72..f1fccc930c 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 3eb9fc422b..734f22c78a 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 0cb1163bad..f837aade12 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 08828865e3..9ec8a720c1 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index db9e04358e..3a3fcde97e 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index fa4a5ca936..84ffa3276f 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 72a0fd928e..da995438e2 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 62fbd05087..e68e818743 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index f64e41952b..f596675ced 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 055d587775..e472e42865 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 05f8eb63b2..b60eccf5c1 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 08af341e6a..c658684808 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 236f7f9cfb..d0f113a76b 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index a8fb7d85d1..df1c576211 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 0811567d15..2da3a81a1a 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index d9fb09ea59..f53432d8f0 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.19.0/llama_cpp_binaries-0.19.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 01ef4c61bd906230708fc63887f35d76b3fef72f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Jun 2025 08:45:11 -0700
Subject: [PATCH 0736/1701] Only open/close both sidebars at the same time on
 desktop

---
 js/main.js | 42 ++++++++++++++++++++++++------------------
 1 file changed, 24 insertions(+), 18 deletions(-)

diff --git a/js/main.js b/js/main.js
index 8b31bbaa49..f018e332a2 100644
--- a/js/main.js
+++ b/js/main.js
@@ -791,15 +791,18 @@ pastChatsToggle.addEventListener("click", () => {
   const isCurrentlyOpen = !pastChatsRow.classList.contains("sidebar-hidden");
   toggleSidebar(pastChatsRow, pastChatsToggle);
 
-  if (isCurrentlyOpen) {
-    // If we just closed the left sidebar, also close the right sidebar
-    if (!chatControlsRow.classList.contains("sidebar-hidden")) {
-      toggleSidebar(chatControlsRow, chatControlsToggle, true);
-    }
-  } else {
-    // If we just opened the left sidebar, also open the right sidebar
-    if (chatControlsRow.classList.contains("sidebar-hidden")) {
-      toggleSidebar(chatControlsRow, chatControlsToggle, false);
+  // On desktop, open/close both sidebars at the same time
+  if (!isMobile()) {
+    if (isCurrentlyOpen) {
+      // If we just closed the left sidebar, also close the right sidebar
+      if (!chatControlsRow.classList.contains("sidebar-hidden")) {
+        toggleSidebar(chatControlsRow, chatControlsToggle, true);
+      }
+    } else {
+      // If we just opened the left sidebar, also open the right sidebar
+      if (chatControlsRow.classList.contains("sidebar-hidden")) {
+        toggleSidebar(chatControlsRow, chatControlsToggle, false);
+      }
     }
   }
 });
@@ -808,15 +811,18 @@ chatControlsToggle.addEventListener("click", () => {
   const isCurrentlyOpen = !chatControlsRow.classList.contains("sidebar-hidden");
   toggleSidebar(chatControlsRow, chatControlsToggle);
 
-  if (isCurrentlyOpen) {
-    // If we just closed the right sidebar, also close the left sidebar
-    if (!pastChatsRow.classList.contains("sidebar-hidden")) {
-      toggleSidebar(pastChatsRow, pastChatsToggle, true);
-    }
-  } else {
-    // If we just opened the right sidebar, also open the left sidebar
-    if (pastChatsRow.classList.contains("sidebar-hidden")) {
-      toggleSidebar(pastChatsRow, pastChatsToggle, false);
+  // On desktop, open/close both sidebars at the same time
+  if (!isMobile()) {
+    if (isCurrentlyOpen) {
+      // If we just closed the right sidebar, also close the left sidebar
+      if (!pastChatsRow.classList.contains("sidebar-hidden")) {
+        toggleSidebar(pastChatsRow, pastChatsToggle, true);
+      }
+    } else {
+      // If we just opened the right sidebar, also open the left sidebar
+      if (pastChatsRow.classList.contains("sidebar-hidden")) {
+        toggleSidebar(pastChatsRow, pastChatsToggle, false);
+      }
     }
   }
 });

From b4edfce9939c1453ec7c00d946416f348af7acc1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Jun 2025 09:30:03 -0700
Subject: [PATCH 0737/1701] Workaround for jittering while typing on firefox

---
 js/main.js | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index f018e332a2..a370159d96 100644
--- a/js/main.js
+++ b/js/main.js
@@ -182,7 +182,8 @@ const observer = new MutationObserver(function(mutations) {
 
   doSyntaxHighlighting();
 
-  if (!window.isScrolled && targetElement.scrollTop !== targetElement.scrollHeight) {
+  const isUserTyping = document.activeElement && document.activeElement.closest("#chat-input") && !targetElement.classList.contains("_generating");
+  if (!window.isScrolled && !isUserTyping && targetElement.scrollTop !== targetElement.scrollHeight) {
     targetElement.scrollTop = targetElement.scrollHeight;
   }
 

From 75217d37138b5d12a21358d0617b02e7ea8da4d6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Jun 2025 09:37:24 -0700
Subject: [PATCH 0738/1701] Change issue template

---
 .github/ISSUE_TEMPLATE/bug_report_template.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/ISSUE_TEMPLATE/bug_report_template.yml b/.github/ISSUE_TEMPLATE/bug_report_template.yml
index bd30a0c9c1..ad22b6565a 100644
--- a/.github/ISSUE_TEMPLATE/bug_report_template.yml
+++ b/.github/ISSUE_TEMPLATE/bug_report_template.yml
@@ -46,7 +46,7 @@ body:
     id: system-info
     attributes:
       label: System Info
-      description: "Please share your system info with us: operating system, GPU brand, and GPU model. If you are using a Google Colab notebook, mention that instead."
+      description: "Please share your operating system and GPU type (NVIDIA/AMD/Intel/Apple). If you are using a Google Colab notebook, mention that instead."
       render: shell
       placeholder: 
     validations:

From fc23345c6d23447417d89c4b45fbb58c40c434f1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Jun 2025 14:59:03 -0700
Subject: [PATCH 0739/1701] Send the default input to the notebook textbox when
 switching 2 columns to 1 (instead of the output)

---
 modules/ui_session.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/ui_session.py b/modules/ui_session.py
index 431aeeb02f..a69e155bdd 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -64,7 +64,7 @@ def handle_default_to_notebook_change(show_two_columns, default_input, default_o
             gr.update(visible=True),
             gr.update(visible=False),
             notebook_input,
-            notebook_input,
+            "",
             gr.update(value=notebook_prompt, choices=utils.get_available_prompts()),
             gr.update(),
             gr.update(),
@@ -77,7 +77,7 @@ def handle_default_to_notebook_change(show_two_columns, default_input, default_o
             gr.update(),
             gr.update(),
             gr.update(),
-            default_output,
+            default_input,
             gr.update(value=default_prompt, choices=utils.get_available_prompts())
         ]
 

From da148232eb6221c129a6838196ee63de18d755e8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Jun 2025 15:10:44 -0700
Subject: [PATCH 0740/1701] Better filenames for new prompts in the Notebook
 tab

---
 modules/utils.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/utils.py b/modules/utils.py
index da5ac9780d..c285d40161 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -53,7 +53,7 @@ def delete_file(fname):
 
 
 def current_time():
-    return f"{datetime.now().strftime('%Y-%m-%d-%H%M%S')}"
+    return f"{datetime.now().strftime('%Y-%m-%d_%Hh%Mm%Ss')}"
 
 
 def atoi(text):

From 2d37602382e6dd1335cddb50c14f60e5be6351e6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Jun 2025 15:26:59 -0700
Subject: [PATCH 0741/1701] Small improvements to wpp style

---
 css/chat_style-wpp.css | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/css/chat_style-wpp.css b/css/chat_style-wpp.css
index 57b90ee32a..b2ac4d2a38 100644
--- a/css/chat_style-wpp.css
+++ b/css/chat_style-wpp.css
@@ -25,12 +25,12 @@
     font-weight: 600;
     margin-bottom: 8px;
     opacity: 0.65;
-    padding-left: 4px;
+    padding-left: 0;
 }
 
 /* User messages - right aligned, WhatsApp green */
 .circle-you + .text {
-    background-color: #dcf8c6;
+    background-color: #d9fdd3;
     float: right;
     margin-left: auto;
     margin-right: 8px;
@@ -55,7 +55,7 @@
 
 /* Dark theme colors */
 .dark .circle-you + .text {
-    background-color: #056162;
+    background-color: #144d37;
     color: #e4e6ea;
     box-shadow: 0 1px 2px rgb(0 0 0 / 30%);
 }

From a2cdd06afc4610ee97db24c8f2b84e234473e827 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Jun 2025 15:29:40 -0700
Subject: [PATCH 0742/1701] Revert "Workaround for jittering while typing on
 firefox"

This reverts commit b4edfce9939c1453ec7c00d946416f348af7acc1.
---
 js/main.js | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/js/main.js b/js/main.js
index a370159d96..f018e332a2 100644
--- a/js/main.js
+++ b/js/main.js
@@ -182,8 +182,7 @@ const observer = new MutationObserver(function(mutations) {
 
   doSyntaxHighlighting();
 
-  const isUserTyping = document.activeElement && document.activeElement.closest("#chat-input") && !targetElement.classList.contains("_generating");
-  if (!window.isScrolled && !isUserTyping && targetElement.scrollTop !== targetElement.scrollHeight) {
+  if (!window.isScrolled && targetElement.scrollTop !== targetElement.scrollHeight) {
     targetElement.scrollTop = targetElement.scrollHeight;
   }
 

From 678f40297b311e7d9b660619729cd5902e7c3917 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Jun 2025 17:40:48 -0700
Subject: [PATCH 0743/1701] Clear the default tab output when switching prompts

---
 modules/ui_default.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_default.py b/modules/ui_default.py
index 66174e359e..b159cc9931 100644
--- a/modules/ui_default.py
+++ b/modules/ui_default.py
@@ -101,7 +101,7 @@ def create_event_handlers():
 
     shared.gradio['Stop-default'].click(stop_everything_event, None, None, queue=False)
     shared.gradio['markdown_render-default'].click(lambda x: x, gradio('output_textbox'), gradio('markdown-default'), queue=False)
-    shared.gradio['prompt_menu-default'].change(load_prompt, gradio('prompt_menu-default'), gradio('textbox-default'), show_progress=False)
+    shared.gradio['prompt_menu-default'].change(lambda x : (load_prompt(x), ""), gradio('prompt_menu-default'), gradio('textbox-default', 'output_textbox'), show_progress=False)
     shared.gradio['new_prompt-default'].click(handle_new_prompt, None, gradio('prompt_menu-default'), show_progress=False)
 
     shared.gradio['delete_prompt-default'].click(

From 22cc9e01155e410349808c002969ae7afe702ad0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Jun 2025 10:18:39 -0700
Subject: [PATCH 0744/1701] Remove 'Send to Default'

---
 js/switch_tabs.js  |  6 ------
 modules/chat.py    |  2 +-
 modules/ui_chat.py | 14 +-------------
 3 files changed, 2 insertions(+), 20 deletions(-)

diff --git a/js/switch_tabs.js b/js/switch_tabs.js
index 0564f89178..22cef9d080 100644
--- a/js/switch_tabs.js
+++ b/js/switch_tabs.js
@@ -31,12 +31,6 @@ function switch_to_chat() {
   scrollToTop();
 }
 
-function switch_to_default() {
-  let default_tab_button = main_parent.childNodes[0].childNodes[5];
-  default_tab_button.click();
-  scrollToTop();
-}
-
 function switch_to_notebook() {
   let notebook_tab_button = main_parent.childNodes[0].childNodes[9];
   notebook_tab_button.click();
diff --git a/modules/chat.py b/modules/chat.py
index d76835665f..4c2a8ced8d 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1924,4 +1924,4 @@ def handle_send_instruction_click(state):
 def handle_send_chat_click(state):
     output = generate_chat_prompt("", state, _continue=True)
 
-    return output
+    return output, output, ""
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index da634f072c..588ea3d66f 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -70,7 +70,6 @@ def create_ui():
             shared.gradio['Impersonate'] = gr.Button('Impersonate (Ctrl + Shift + M)', elem_id='Impersonate')
             shared.gradio['Send dummy message'] = gr.Button('Send dummy message')
             shared.gradio['Send dummy reply'] = gr.Button('Send dummy reply')
-            shared.gradio['send-chat-to-default'] = gr.Button('Send to Default')
             shared.gradio['send-chat-to-notebook'] = gr.Button('Send to Notebook')
             shared.gradio['show_controls'] = gr.Checkbox(value=shared.settings['show_controls'], label='Show controls (Ctrl+S)', elem_id='show-controls')
 
@@ -183,7 +182,6 @@ def create_chat_settings_ui():
             with gr.Column():
                 shared.gradio['instruction_template_str'] = gr.Textbox(value=shared.settings['instruction_template_str'], label='Instruction template', lines=24, info='This gets autodetected; you usually don\'t need to change it. Used in instruct and chat-instruct modes.', elem_classes=['add_scrollbar', 'monospace'], elem_id='instruction-template-str')
                 with gr.Row():
-                    shared.gradio['send_instruction_to_default'] = gr.Button('Send to default', elem_classes=['small-button'])
                     shared.gradio['send_instruction_to_notebook'] = gr.Button('Send to notebook', elem_classes=['small-button'])
 
             with gr.Column():
@@ -344,24 +342,14 @@ def create_event_handlers():
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_your_picture_change, gradio('your_picture', 'interface_state'), gradio('display'), show_progress=False)
 
-    shared.gradio['send_instruction_to_default'].click(
-        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        chat.handle_send_instruction_click, gradio('interface_state'), gradio('textbox-default'), show_progress=False).then(
-        None, None, None, js=f'() => {{{ui.switch_tabs_js}; switch_to_default()}}')
-
     shared.gradio['send_instruction_to_notebook'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_send_instruction_click, gradio('interface_state'), gradio('textbox-notebook'), show_progress=False).then(
         None, None, None, js=f'() => {{{ui.switch_tabs_js}; switch_to_notebook()}}')
 
-    shared.gradio['send-chat-to-default'].click(
-        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        chat.handle_send_chat_click, gradio('interface_state'), gradio('textbox-default'), show_progress=False).then(
-        None, None, None, js=f'() => {{{ui.switch_tabs_js}; switch_to_default()}}')
-
     shared.gradio['send-chat-to-notebook'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        chat.handle_send_chat_click, gradio('interface_state'), gradio('textbox-notebook'), show_progress=False).then(
+        chat.handle_send_chat_click, gradio('interface_state'), gradio('textbox-notebook', 'textbox-default', 'output_textbox'), show_progress=False).then(
         None, None, None, js=f'() => {{{ui.switch_tabs_js}; switch_to_notebook()}}')
 
     shared.gradio['show_controls'].change(None, gradio('show_controls'), None, js=f'(x) => {{{ui.show_controls_js}; toggle_controls(x)}}')

From 8b7eb5c87cc0ffb2d28205c32d158e13ea0ba739 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Jun 2025 10:22:36 -0700
Subject: [PATCH 0745/1701] Code simplification

---
 js/switch_tabs.js | 29 +++++++----------------------
 1 file changed, 7 insertions(+), 22 deletions(-)

diff --git a/js/switch_tabs.js b/js/switch_tabs.js
index 22cef9d080..7fb78aea85 100644
--- a/js/switch_tabs.js
+++ b/js/switch_tabs.js
@@ -1,24 +1,14 @@
-let chat_tab = document.getElementById("chat-tab");
-let main_parent = chat_tab.parentNode;
-
 function scrollToTop() {
-  window.scrollTo({
-    top: 0,
-    // behavior: 'smooth'
-  });
+  window.scrollTo({ top: 0 });
 }
 
 function findButtonsByText(buttonText) {
   const buttons = document.getElementsByTagName("button");
   const matchingButtons = [];
-  buttonText = buttonText.trim();
 
   for (let i = 0; i < buttons.length; i++) {
-    const button = buttons[i];
-    const buttonInnerText = button.textContent.trim();
-
-    if (buttonInnerText === buttonText) {
-      matchingButtons.push(button);
+    if (buttons[i].textContent.trim() === buttonText) {
+      matchingButtons.push(buttons[i]);
     }
   }
 
@@ -26,28 +16,23 @@ function findButtonsByText(buttonText) {
 }
 
 function switch_to_chat() {
-  let chat_tab_button = main_parent.childNodes[0].childNodes[1];
-  chat_tab_button.click();
+  document.getElementById("chat-tab-button").click();
   scrollToTop();
 }
 
 function switch_to_notebook() {
-  let notebook_tab_button = main_parent.childNodes[0].childNodes[9];
-  notebook_tab_button.click();
+  document.getElementById("notebook-parent-tab-button").click();
   findButtonsByText("Raw")[1].click();
   scrollToTop();
 }
 
 function switch_to_generation_parameters() {
-  let parameters_tab_button = main_parent.childNodes[0].childNodes[13];
-  parameters_tab_button.click();
+  document.getElementById("parameters-button").click();
   findButtonsByText("Generation")[0].click();
   scrollToTop();
 }
 
 function switch_to_character() {
-  let parameters_tab_button = main_parent.childNodes[0].childNodes[13];
-  parameters_tab_button.click();
-  findButtonsByText("Character")[0].click();
+  document.getElementById("character-tab-button").click();
   scrollToTop();
 }

From 7cb2b1bfdb4775142aca692226dc9b778f7832b8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Jun 2025 10:27:38 -0700
Subject: [PATCH 0746/1701] Fix some events

---
 modules/chat.py    | 10 ++++++++--
 modules/ui_chat.py |  2 +-
 2 files changed, 9 insertions(+), 3 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 4c2a8ced8d..9290dd6223 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1918,10 +1918,16 @@ def handle_send_instruction_click(state):
 
     output = generate_chat_prompt("Input", state)
 
-    return output
+    if state["show_two_notebook_columns"]:
+        return gr.update(), output, ""
+    else:
+        return output, gr.update(), gr.update()
 
 
 def handle_send_chat_click(state):
     output = generate_chat_prompt("", state, _continue=True)
 
-    return output, output, ""
+    if state["show_two_notebook_columns"]:
+        return gr.update(), output, ""
+    else:
+        return output, gr.update(), gr.update()
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 588ea3d66f..8a90608fa6 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -344,7 +344,7 @@ def create_event_handlers():
 
     shared.gradio['send_instruction_to_notebook'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        chat.handle_send_instruction_click, gradio('interface_state'), gradio('textbox-notebook'), show_progress=False).then(
+        chat.handle_send_instruction_click, gradio('interface_state'), gradio('textbox-notebook', 'textbox-default', 'output_textbox'), show_progress=False).then(
         None, None, None, js=f'() => {{{ui.switch_tabs_js}; switch_to_notebook()}}')
 
     shared.gradio['send-chat-to-notebook'].click(

From 2f45d7530956516437df792f6077972ade7f665e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Jun 2025 13:22:06 -0700
Subject: [PATCH 0747/1701] Increase the area of the notebook textbox

---
 css/main.css           | 8 ++++++--
 modules/ui_notebook.py | 2 +-
 2 files changed, 7 insertions(+), 3 deletions(-)

diff --git a/css/main.css b/css/main.css
index 3e51e920a5..bc59f8339a 100644
--- a/css/main.css
+++ b/css/main.css
@@ -175,7 +175,7 @@ gradio-app > :first-child {
 }
 
 .textbox textarea {
-    height: calc(100dvh - 172px);
+    height: calc(100dvh - 145px)
 }
 
 .textbox_logits textarea {
@@ -307,7 +307,7 @@ audio {
 }
 
 #notebook-token-counter {
-    top: calc(100dvh - 171px) !important;
+    top: calc(100dvh - 172px) !important;
 }
 
 #default-token-counter span, #notebook-token-counter span {
@@ -1666,3 +1666,7 @@ button:focus {
     height: calc(100vh - 300px) !important;
     min-height: 90px !important;
 }
+
+#textbox-notebook span {
+    display: none;
+}
diff --git a/modules/ui_notebook.py b/modules/ui_notebook.py
index 3a8c9df2f4..f15684ccf7 100644
--- a/modules/ui_notebook.py
+++ b/modules/ui_notebook.py
@@ -32,7 +32,7 @@ def create_ui():
                 with gr.Tab('Raw'):
                     with gr.Row():
                         initial_text = load_prompt(shared.settings['prompt-notebook'])
-                        shared.gradio['textbox-notebook'] = gr.Textbox(value=initial_text, lines=27, elem_id='textbox-notebook', elem_classes=['textbox', 'add_scrollbar'])
+                        shared.gradio['textbox-notebook'] = gr.Textbox(label="", value=initial_text, lines=27, elem_id='textbox-notebook', elem_classes=['textbox', 'add_scrollbar'])
                         shared.gradio['token-counter-notebook'] = gr.HTML(value="0", elem_id="notebook-token-counter")
 
                 with gr.Tab('Markdown'):

From 197b327374df61478b68b57f83a918d80424d240 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Jun 2025 13:36:54 -0700
Subject: [PATCH 0748/1701] Minor log message change

---
 modules/models_settings.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 3d98568aaf..37aa37cf52 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -59,7 +59,7 @@ def get_model_metadata(model):
         else:
             gguf_files = list(path.glob('*.gguf'))
             if not gguf_files:
-                error_msg = f"No .gguf files found in directory: {path}"
+                error_msg = f"No .gguf models found in directory: {path}"
                 logger.error(error_msg)
                 raise FileNotFoundError(error_msg)
 

From 6cc7bbf0092d96cb8e37f7c480a87d74884e6a71 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Jun 2025 15:54:32 -0700
Subject: [PATCH 0749/1701] Better autosave behavior for notebook tab when
 there are 2 columns

---
 modules/ui_default.py  | 68 ++++++++----------------------------------
 modules/ui_notebook.py |  9 +++++-
 2 files changed, 20 insertions(+), 57 deletions(-)

diff --git a/modules/ui_default.py b/modules/ui_default.py
index b159cc9931..9697cc21d9 100644
--- a/modules/ui_default.py
+++ b/modules/ui_default.py
@@ -1,4 +1,3 @@
-import time
 from pathlib import Path
 
 import gradio as gr
@@ -10,6 +9,7 @@
     get_token_ids,
     stop_everything_event
 )
+from modules.ui_notebook import store_notebook_state_and_debounce
 from modules.utils import gradio
 
 inputs = ('textbox-default', 'interface_state')
@@ -78,7 +78,7 @@ def create_event_handlers():
     shared.gradio['Generate-default'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-default', 'Generate-default')).then(
-        generate_and_save_wrapper, gradio('textbox-default', 'interface_state', 'prompt_menu-default'), gradio(outputs), show_progress=False).then(
+        generate_reply_wrapper, gradio('textbox-default', 'interface_state'), gradio(outputs), show_progress=False).then(
         lambda state, left, right: state.update({'textbox-default': left, 'output_textbox': right}), gradio('interface_state', 'textbox-default', 'output_textbox'), None).then(
         lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-default', 'Generate-default')).then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
@@ -86,7 +86,7 @@ def create_event_handlers():
     shared.gradio['textbox-default'].submit(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-default', 'Generate-default')).then(
-        generate_and_save_wrapper, gradio('textbox-default', 'interface_state', 'prompt_menu-default'), gradio(outputs), show_progress=False).then(
+        generate_reply_wrapper, gradio('textbox-default', 'interface_state'), gradio(outputs), show_progress=False).then(
         lambda state, left, right: state.update({'textbox-default': left, 'output_textbox': right}), gradio('interface_state', 'textbox-default', 'output_textbox'), None).then(
         lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-default', 'Generate-default')).then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
@@ -94,7 +94,7 @@ def create_event_handlers():
     shared.gradio['Continue-default'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('Stop-default', 'Generate-default')).then(
-        continue_and_save_wrapper, gradio('output_textbox', 'textbox-default', 'interface_state', 'prompt_menu-default'), gradio(outputs), show_progress=False).then(
+        generate_reply_wrapper, gradio('output_textbox', 'interface_state'), gradio(outputs), show_progress=False).then(
         lambda state, left, right: state.update({'textbox-default': left, 'output_textbox': right}), gradio('interface_state', 'textbox-default', 'output_textbox'), None).then(
         lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('Stop-default', 'Generate-default')).then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
@@ -104,6 +104,14 @@ def create_event_handlers():
     shared.gradio['prompt_menu-default'].change(lambda x : (load_prompt(x), ""), gradio('prompt_menu-default'), gradio('textbox-default', 'output_textbox'), show_progress=False)
     shared.gradio['new_prompt-default'].click(handle_new_prompt, None, gradio('prompt_menu-default'), show_progress=False)
 
+    # Input change handler to save input (reusing notebook's debounced saving)
+    shared.gradio['textbox-default'].change(
+        store_notebook_state_and_debounce,
+        gradio('textbox-default', 'prompt_menu-default'),
+        None,
+        show_progress=False
+    )
+
     shared.gradio['delete_prompt-default'].click(
         lambda: [gr.update(visible=False), gr.update(visible=True), gr.update(visible=True)],
         None,
@@ -148,58 +156,6 @@ def create_event_handlers():
     shared.gradio['get_tokens-default'].click(get_token_ids, gradio('textbox-default'), gradio('tokens-default'), show_progress=False)
 
 
-def autosave_prompt(text, prompt_name):
-    """Automatically save the text to the selected prompt file"""
-    if prompt_name and text.strip():
-        prompt_path = Path("user_data/logs/notebook") / f"{prompt_name}.txt"
-        prompt_path.parent.mkdir(parents=True, exist_ok=True)
-        prompt_path.write_text(text, encoding='utf-8')
-
-
-def generate_and_save_wrapper(textbox_content, interface_state, prompt_name):
-    """Generate reply and automatically save the result with periodic saves"""
-    last_save_time = time.monotonic()
-    save_interval = 8
-    output = textbox_content
-
-    # Initial autosave
-    autosave_prompt(output, prompt_name)
-
-    for i, (output, html_output) in enumerate(generate_reply_wrapper(textbox_content, interface_state)):
-        yield output, html_output
-
-        current_time = time.monotonic()
-        # Save on first iteration or if save_interval seconds have passed
-        if i == 0 or (current_time - last_save_time) >= save_interval:
-            autosave_prompt(output, prompt_name)
-            last_save_time = current_time
-
-    # Final autosave
-    autosave_prompt(output, prompt_name)
-
-
-def continue_and_save_wrapper(output_textbox, textbox_content, interface_state, prompt_name):
-    """Continue generation and automatically save the result with periodic saves"""
-    last_save_time = time.monotonic()
-    save_interval = 8
-    output = output_textbox
-
-    # Initial autosave
-    autosave_prompt(output, prompt_name)
-
-    for i, (output, html_output) in enumerate(generate_reply_wrapper(output_textbox, interface_state)):
-        yield output, html_output
-
-        current_time = time.monotonic()
-        # Save on first iteration or if save_interval seconds have passed
-        if i == 0 or (current_time - last_save_time) >= save_interval:
-            autosave_prompt(output, prompt_name)
-            last_save_time = current_time
-
-    # Final autosave
-    autosave_prompt(output, prompt_name)
-
-
 def handle_new_prompt():
     new_name = utils.current_time()
 
diff --git a/modules/ui_notebook.py b/modules/ui_notebook.py
index f15684ccf7..939d81f70b 100644
--- a/modules/ui_notebook.py
+++ b/modules/ui_notebook.py
@@ -11,7 +11,6 @@
     get_token_ids,
     stop_everything_event
 )
-from modules.ui_default import autosave_prompt
 from modules.utils import gradio
 
 _notebook_file_lock = threading.Lock()
@@ -249,6 +248,14 @@ def handle_rename_prompt_confirm_notebook(new_name, current_name):
     ]
 
 
+def autosave_prompt(text, prompt_name):
+    """Automatically save the text to the selected prompt file"""
+    if prompt_name and text.strip():
+        prompt_path = Path("user_data/logs/notebook") / f"{prompt_name}.txt"
+        prompt_path.parent.mkdir(parents=True, exist_ok=True)
+        prompt_path.write_text(text, encoding='utf-8')
+
+
 def safe_autosave_prompt(content, prompt_name):
     """Thread-safe wrapper for autosave_prompt to prevent file corruption"""
     with _notebook_file_lock:

From 0b26650f471318e7255ad70e91402a130215bf63 Mon Sep 17 00:00:00 2001
From: NoxWorld2660 <57005290+NoxWorld2660@users.noreply.github.com>
Date: Thu, 19 Jun 2025 01:14:24 +0200
Subject: [PATCH 0750/1701] Expose real model list via /v1/models endpoint
 (#7088)

---
 extensions/openai/script.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 24bcd69d24..3d8d5f7328 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -180,7 +180,7 @@ async def handle_models(request: Request):
     is_list = request.url.path.split('?')[0].split('#')[0] == '/v1/models'
 
     if is_list:
-        response = OAImodels.list_dummy_models()
+        response = OAImodels.list_models()
     else:
         model_name = path[len('/v1/models/'):]
         response = OAImodels.model_info_dict(model_name)

From 6af3598cfa808045731a687a492d6cf73ae4e0ae Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Jun 2025 16:15:29 -0700
Subject: [PATCH 0751/1701] API: Remove obsolete list_dummy_models function

---
 extensions/openai/models.py | 13 -------------
 1 file changed, 13 deletions(-)

diff --git a/extensions/openai/models.py b/extensions/openai/models.py
index a7e67df6f6..f8d9a1e84d 100644
--- a/extensions/openai/models.py
+++ b/extensions/openai/models.py
@@ -18,19 +18,6 @@ def list_models():
     return {'model_names': get_available_models()[1:]}
 
 
-def list_dummy_models():
-    result = {
-        "object": "list",
-        "data": []
-    }
-
-    # these are expected by so much, so include some here as a dummy
-    for model in ['gpt-3.5-turbo', 'text-embedding-ada-002']:
-        result["data"].append(model_info_dict(model))
-
-    return result
-
-
 def model_info_dict(model_name: str) -> dict:
     return {
         "id": model_name,

From e33921a629f2047bfbca01f401bbee11eb5dc45d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Jun 2025 17:54:34 -0700
Subject: [PATCH 0752/1701] Fix jittering while typing on Firefox (closes
 #7086)

---
 js/main.js | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index f018e332a2..3ff4bf06c7 100644
--- a/js/main.js
+++ b/js/main.js
@@ -170,6 +170,13 @@ targetElement.addEventListener("scroll", function() {
 
 // Create a MutationObserver instance
 const observer = new MutationObserver(function(mutations) {
+  // Check if this is just the scrolling class being toggled
+  const isScrollingClassOnly = mutations.every(mutation =>
+    mutation.type === "attributes" &&
+    mutation.attributeName === "class" &&
+    mutation.target === targetElement
+  );
+
   if (targetElement.classList.contains("_generating")) {
     typing.parentNode.classList.add("visible-dots");
     document.getElementById("stop").style.display = "flex";
@@ -182,7 +189,7 @@ const observer = new MutationObserver(function(mutations) {
 
   doSyntaxHighlighting();
 
-  if (!window.isScrolled && targetElement.scrollTop !== targetElement.scrollHeight) {
+  if (!window.isScrolled && !isScrollingClassOnly && targetElement.scrollTop !== targetElement.scrollHeight) {
     targetElement.scrollTop = targetElement.scrollHeight;
   }
 

From 0cb82483eff7302203d2f095d7c811453019fa03 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Jun 2025 18:26:59 -0700
Subject: [PATCH 0753/1701] Lint

---
 modules/ui_default.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_default.py b/modules/ui_default.py
index 9697cc21d9..44af48a306 100644
--- a/modules/ui_default.py
+++ b/modules/ui_default.py
@@ -101,7 +101,7 @@ def create_event_handlers():
 
     shared.gradio['Stop-default'].click(stop_everything_event, None, None, queue=False)
     shared.gradio['markdown_render-default'].click(lambda x: x, gradio('output_textbox'), gradio('markdown-default'), queue=False)
-    shared.gradio['prompt_menu-default'].change(lambda x : (load_prompt(x), ""), gradio('prompt_menu-default'), gradio('textbox-default', 'output_textbox'), show_progress=False)
+    shared.gradio['prompt_menu-default'].change(lambda x: (load_prompt(x), ""), gradio('prompt_menu-default'), gradio('textbox-default', 'output_textbox'), show_progress=False)
     shared.gradio['new_prompt-default'].click(handle_new_prompt, None, gradio('prompt_menu-default'), show_progress=False)
 
     # Input change handler to save input (reusing notebook's debounced saving)

From 4e0dfbdde3d534f8a0464107829013a13e0a42e3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Jun 2025 21:02:57 -0700
Subject: [PATCH 0754/1701] Remove .github and .gitignore folders from portable
 builds

---
 .github/workflows/build-portable-release-cuda.yml   | 2 +-
 .github/workflows/build-portable-release-vulkan.yml | 2 +-
 .github/workflows/build-portable-release.yml        | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index a63ec16896..729ce695d7 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -107,7 +107,7 @@ jobs:
             cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
             cd "text-generation-webui-${VERSION_CLEAN}"
 
-            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore
             allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index b6217164fb..98eb363522 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -106,7 +106,7 @@ jobs:
             cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
             cd "text-generation-webui-${VERSION_CLEAN}"
 
-            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore
             allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index dd86f522dd..8863d67367 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -106,7 +106,7 @@ jobs:
             cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
             cd "text-generation-webui-${VERSION_CLEAN}"
 
-            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore
             allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 

From 76a722dc903506116d09724d9b65601a4594b072 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Jun 2025 21:02:57 -0700
Subject: [PATCH 0755/1701] Remove .github and .gitignore folders from portable
 builds

---
 .github/workflows/build-portable-release-cuda.yml   | 2 +-
 .github/workflows/build-portable-release-vulkan.yml | 2 +-
 .github/workflows/build-portable-release.yml        | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index a63ec16896..729ce695d7 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -107,7 +107,7 @@ jobs:
             cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
             cd "text-generation-webui-${VERSION_CLEAN}"
 
-            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore
             allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index b6217164fb..98eb363522 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -106,7 +106,7 @@ jobs:
             cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
             cd "text-generation-webui-${VERSION_CLEAN}"
 
-            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore
             allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index dd86f522dd..8863d67367 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -106,7 +106,7 @@ jobs:
             cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
             cd "text-generation-webui-${VERSION_CLEAN}"
 
-            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore
             allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 

From 9c6913ad61793ed5a990dc9529c4dc46d4d1dbc4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Jun 2025 21:35:07 -0700
Subject: [PATCH 0756/1701] Show file sizes on "Get file list"

---
 download-model.py        | 14 +++++++++++--
 modules/ui_model_menu.py | 43 +++++++++++++++++++++++++++++++++++-----
 2 files changed, 50 insertions(+), 7 deletions(-)

diff --git a/download-model.py b/download-model.py
index 576a8b7910..c0a3aa36b5 100644
--- a/download-model.py
+++ b/download-model.py
@@ -82,6 +82,7 @@ def get_download_links_from_huggingface(self, model, branch, text_only=False, sp
 
         links = []
         sha256 = []
+        file_sizes = []
         classifications = []
         has_pytorch = False
         has_pt = False
@@ -118,8 +119,14 @@ def get_download_links_from_huggingface(self, model, branch, text_only=False, sp
                 is_tokenizer = re.match(r"(tokenizer|ice|spiece).*\.model", fname) or is_tiktoken
                 is_text = re.match(r".*\.(txt|json|py|md)", fname) or is_tokenizer
                 if any((is_pytorch, is_safetensors, is_pt, is_gguf, is_tokenizer, is_text)):
+                    file_size = 0
                     if 'lfs' in dict[i]:
                         sha256.append([fname, dict[i]['lfs']['oid']])
+                        file_size = dict[i]['lfs'].get('size', 0)
+                    elif 'size' in dict[i]:
+                        file_size = dict[i]['size']
+
+                    file_sizes.append(file_size)
 
                     if is_text:
                         links.append(f"{base}/{model}/resolve/{branch}/{fname}")
@@ -152,6 +159,7 @@ def get_download_links_from_huggingface(self, model, branch, text_only=False, sp
             for i in range(len(classifications) - 1, -1, -1):
                 if classifications[i] in ['pytorch', 'pt', 'gguf']:
                     links.pop(i)
+                    file_sizes.pop(i)
 
         # For GGUF, try to download only the Q4_K_M if no specific file is specified.
         if has_gguf and specific_file is None:
@@ -164,13 +172,15 @@ def get_download_links_from_huggingface(self, model, branch, text_only=False, sp
                 for i in range(len(classifications) - 1, -1, -1):
                     if 'q4_k_m' not in links[i].lower():
                         links.pop(i)
+                        file_sizes.pop(i)
             else:
                 for i in range(len(classifications) - 1, -1, -1):
                     if links[i].lower().endswith('.gguf'):
                         links.pop(i)
+                        file_sizes.pop(i)
 
         is_llamacpp = has_gguf and specific_file is not None
-        return links, sha256, is_lora, is_llamacpp
+        return links, sha256, is_lora, is_llamacpp, file_sizes
 
     def get_output_folder(self, model, branch, is_lora, is_llamacpp=False, model_dir=None):
         if model_dir:
@@ -396,7 +406,7 @@ def check_model_files(self, model, branch, links, sha256, output_folder):
         sys.exit()
 
     # Get the download links from Hugging Face
-    links, sha256, is_lora, is_llamacpp = downloader.get_download_links_from_huggingface(
+    links, sha256, is_lora, is_llamacpp, file_sizes = downloader.get_download_links_from_huggingface(
         model, branch, text_only=args.text_only, specific_file=specific_file, exclude_pattern=exclude_pattern
     )
 
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 6b1062035d..1ee8089e33 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -1,4 +1,5 @@
 import importlib
+import math
 import queue
 import threading
 import traceback
@@ -244,7 +245,7 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
 
         model, branch = downloader.sanitize_model_and_branch_names(repo_id, None)
         yield "Getting download links from Hugging Face..."
-        links, sha256, is_lora, is_llamacpp = downloader.get_download_links_from_huggingface(model, branch, text_only=False, specific_file=specific_file)
+        links, sha256, is_lora, is_llamacpp, file_sizes = downloader.get_download_links_from_huggingface(model, branch, text_only=False, specific_file=specific_file)
 
         if not links:
             yield "No files found to download for the given model/criteria."
@@ -254,17 +255,33 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
         # Check for multiple GGUF files
         gguf_files = [link for link in links if link.lower().endswith('.gguf')]
         if len(gguf_files) > 1 and not specific_file:
+            # Sort GGUF files by size (ascending - smallest first)
+            gguf_data = []
+            for i, link in enumerate(links):
+                if link.lower().endswith('.gguf'):
+                    file_size = file_sizes[i]
+                    gguf_data.append((file_size, link))
+
+            gguf_data.sort(key=lambda x: x[0])
+
             output = "Multiple GGUF files found. Please copy one of the following filenames to the 'File name' field:\n\n```\n"
-            for link in gguf_files:
-                output += f"{Path(link).name}\n"
+            for file_size, link in gguf_data:
+                size_str = format_file_size(file_size)
+                output += f"{size_str} - {Path(link).name}\n"
+
             output += "```"
             yield output
             return
 
         if return_links:
+            # Sort files by size (ascending - smallest first)
+            file_data = list(zip(file_sizes, links))
+            file_data.sort(key=lambda x: x[0])
+
             output = "```\n"
-            for link in links:
-                output += f"{Path(link).name}" + "\n"
+            for file_size, link in file_data:
+                size_str = format_file_size(file_size)
+                output += f"{size_str} - {Path(link).name}\n"
 
             output += "```"
             yield output
@@ -391,3 +408,19 @@ def handle_load_model_event_final(truncation_length, loader, state):
 def handle_unload_model_click():
     unload_model()
     return "Model unloaded"
+
+
+def format_file_size(size_bytes):
+    """Convert bytes to human readable format with 2 decimal places for GB and above"""
+    if size_bytes == 0:
+        return "0 B"
+
+    size_names = ["B", "KB", "MB", "GB", "TB"]
+    i = int(math.floor(math.log(size_bytes, 1024)))
+    p = math.pow(1024, i)
+    s = size_bytes / p
+
+    if i >= 3:  # GB or TB
+        return f"{s:.2f} {size_names[i]}"
+    else:
+        return f"{s:.1f} {size_names[i]}"

From dcdc42fa06ba56eec5ca09b305147a27ee08ff39 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Jun 2025 07:57:17 -0700
Subject: [PATCH 0757/1701] Fix the /v1/models output format (closes #7089)

---
 extensions/openai/models.py | 2 +-
 extensions/openai/script.py | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/extensions/openai/models.py b/extensions/openai/models.py
index f8d9a1e84d..9c086a170c 100644
--- a/extensions/openai/models.py
+++ b/extensions/openai/models.py
@@ -15,7 +15,7 @@ def get_current_model_info():
 
 
 def list_models():
-    return {'model_names': get_available_models()[1:]}
+    return {'model_names': get_available_models()}
 
 
 def model_info_dict(model_name: str) -> dict:
diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 3d8d5f7328..f3b5b093ef 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -180,7 +180,7 @@ async def handle_models(request: Request):
     is_list = request.url.path.split('?')[0].split('#')[0] == '/v1/models'
 
     if is_list:
-        response = OAImodels.list_models()
+        response = OAImodels.list_models()['model_names']
     else:
         model_name = path[len('/v1/models/'):]
         response = OAImodels.model_info_dict(model_name)

From 93cd47c94861032344bb7dcd5df322ca1a5749d6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Jun 2025 08:00:30 -0700
Subject: [PATCH 0758/1701] Bump numpy to 2.2 (loses #7090)

---
 requirements/full/requirements.txt                     | 2 +-
 requirements/full/requirements_amd.txt                 | 2 +-
 requirements/full/requirements_amd_noavx2.txt          | 2 +-
 requirements/full/requirements_apple_intel.txt         | 2 +-
 requirements/full/requirements_apple_silicon.txt       | 2 +-
 requirements/full/requirements_cpu_only.txt            | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt     | 2 +-
 requirements/full/requirements_noavx2.txt              | 2 +-
 requirements/full/requirements_nowheels.txt            | 2 +-
 requirements/portable/requirements.txt                 | 2 +-
 requirements/portable/requirements_apple_intel.txt     | 2 +-
 requirements/portable/requirements_apple_silicon.txt   | 2 +-
 requirements/portable/requirements_cpu_only.txt        | 2 +-
 requirements/portable/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/portable/requirements_noavx2.txt          | 2 +-
 requirements/portable/requirements_nowheels.txt        | 2 +-
 requirements/portable/requirements_vulkan.txt          | 2 +-
 requirements/portable/requirements_vulkan_noavx2.txt   | 2 +-
 18 files changed, 18 insertions(+), 18 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 19e5e0fecb..14828e8db6 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -9,7 +9,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pandas
 peft==0.15.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index ebef87a656..ec2040d0de 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -8,7 +8,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pandas
 peft==0.15.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index f1fccc930c..4b50c69326 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -8,7 +8,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pandas
 peft==0.15.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 734f22c78a..a6e2925e61 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -8,7 +8,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pandas
 peft==0.15.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index f837aade12..1b97d8448e 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -8,7 +8,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pandas
 peft==0.15.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 9ec8a720c1..fa553f7cf8 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -8,7 +8,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pandas
 peft==0.15.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 3a3fcde97e..505f8ae8aa 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -8,7 +8,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pandas
 peft==0.15.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index e68e818743..13ecaf879e 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -9,7 +9,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pandas
 peft==0.15.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index d26663a7a4..b1c3c6eaf1 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -8,7 +8,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pandas
 peft==0.15.*
 Pillow>=9.5.0
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index f596675ced..49cd626484 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -4,7 +4,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pydantic==2.8.2
 PyPDF2==3.0.1
 python-docx==1.1.2
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index e472e42865..fca8208f03 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -4,7 +4,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pydantic==2.8.2
 PyPDF2==3.0.1
 python-docx==1.1.2
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index b60eccf5c1..86b00651f1 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -4,7 +4,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pydantic==2.8.2
 PyPDF2==3.0.1
 python-docx==1.1.2
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index c658684808..fca46fb524 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -4,7 +4,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pydantic==2.8.2
 PyPDF2==3.0.1
 python-docx==1.1.2
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index d0f113a76b..3ebc248838 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -4,7 +4,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pydantic==2.8.2
 PyPDF2==3.0.1
 python-docx==1.1.2
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index df1c576211..2bdf257e1b 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -4,7 +4,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pydantic==2.8.2
 PyPDF2==3.0.1
 python-docx==1.1.2
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 3d30e6d6bd..c22d6441e1 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -4,7 +4,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pydantic==2.8.2
 PyPDF2==3.0.1
 python-docx==1.1.2
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 2da3a81a1a..0b98ee5ab4 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -4,7 +4,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pydantic==2.8.2
 PyPDF2==3.0.1
 python-docx==1.1.2
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index f53432d8f0..188dc08523 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -4,7 +4,7 @@ gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
-numpy==1.26.*
+numpy==2.2.*
 pydantic==2.8.2
 PyPDF2==3.0.1
 python-docx==1.1.2

From 84617abdeb1c746a5b159e2044e6ed90ffd8b2d6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Jun 2025 10:25:55 -0700
Subject: [PATCH 0759/1701] Properly fix the /v1/models endpoint

---
 extensions/openai/models.py | 9 +++++++++
 extensions/openai/script.py | 2 +-
 2 files changed, 10 insertions(+), 1 deletion(-)

diff --git a/extensions/openai/models.py b/extensions/openai/models.py
index 9c086a170c..115149b07a 100644
--- a/extensions/openai/models.py
+++ b/extensions/openai/models.py
@@ -18,6 +18,15 @@ def list_models():
     return {'model_names': get_available_models()}
 
 
+def list_models_openai_format():
+    """Returns model list in OpenAI API format"""
+    model_names = get_available_models()
+    return {
+        "object": "list",
+        "data": [model_info_dict(name) for name in model_names]
+    }
+
+
 def model_info_dict(model_name: str) -> dict:
     return {
         "id": model_name,
diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index f3b5b093ef..9440cb1ebc 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -180,7 +180,7 @@ async def handle_models(request: Request):
     is_list = request.url.path.split('?')[0].split('#')[0] == '/v1/models'
 
     if is_list:
-        response = OAImodels.list_models()['model_names']
+        response = OAImodels.list_models_openai_format()
     else:
         model_name = path[len('/v1/models/'):]
         response = OAImodels.model_info_dict(model_name)

From c4029914e83688d8d390231b8be139d6b8223ca1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Jun 2025 10:48:33 -0700
Subject: [PATCH 0760/1701] Update README

---
 README.md | 41 ++++++++++++++++++++++++++++++++++++-----
 1 file changed, 36 insertions(+), 5 deletions(-)

diff --git a/README.md b/README.md
index 6e7c05b18e..e62e0dd288 100644
--- a/README.md
+++ b/README.md
@@ -57,7 +57,7 @@ To update, run the update script for your OS: `update_wizard_windows.bat`, `upda
 
 
            
 
-Setup details and information about installing manually
+One-click installer details
 
 
 ### One-click-installer
@@ -67,13 +67,44 @@ The script uses Miniconda to set up a Conda environment in the `installer_files`
 If you ever need to install something manually in the `installer_files` environment, you can launch an interactive shell using the cmd script: `cmd_linux.sh`, `cmd_windows.bat`, or `cmd_macos.sh`.
 
 * There is no need to run any of those scripts (`start_`, `update_wizard_`, or `cmd_`) as admin/root.
-* To install the requirements for extensions, you can use the `extensions_reqs` script for your OS. At the end, this script will install the main requirements for the project to make sure that they take precedence in case of version conflicts.
-* For additional instructions about AMD and WSL setup, consult [the documentation](https://github.com/oobabooga/text-generation-webui/wiki).
+* To install requirements for extensions, it is recommended to use the update wizard script with the "Install/update extensions requirements" option. At the end, this script will install the main requirements for the project to make sure that they take precedence in case of version conflicts.
 * For automated installation, you can use the `GPU_CHOICE`, `LAUNCH_AFTER_INSTALL`, and `INSTALL_EXTENSIONS` environment variables. For instance: `GPU_CHOICE=A LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh`.
 
-### Manual installation using Conda
+
            
+
+
            
+
+Manual portable installation with venv
+
+
+### Manual portable installation with venv
+
+Very fast setup that should work on any Python 3.9+:
+
+```bash
+# Create virtual environment
+python -m venv venv
+
+# Activate virtual environment
+# On Windows:
+venv\Scripts\activate
+# On macOS/Linux:
+source venv/bin/activate
+
+# Install dependencies (choose appropriate file under requirements/portable for your hardware)
+pip install -r requirements/portable/requirements.txt
+
+# When done working, deactivate
+deactivate
+```
+
            
+
+
            
+
+Manual full installation with conda or docker
+
 
-Recommended if you have some experience with the command-line.
+### Full installation with Conda
 
 #### 0. Install Conda
 

From 09cd1cb4e2704c321014f891a4dddd80cdb279db Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Jun 2025 10:51:45 -0700
Subject: [PATCH 0761/1701] Update README

---
 README.md | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/README.md b/README.md
index e62e0dd288..ce3602a1fc 100644
--- a/README.md
+++ b/README.md
@@ -91,9 +91,16 @@ venv\Scripts\activate
 # On macOS/Linux:
 source venv/bin/activate
 
+# Clone the repository
+git clone https://github.com/oobabooga/text-generation-webui
+cd text-generation-webui
+
 # Install dependencies (choose appropriate file under requirements/portable for your hardware)
 pip install -r requirements/portable/requirements.txt
 
+# Launch server (basic command)
+python server.py --portable --api --auto-launch
+
 # When done working, deactivate
 deactivate
 ```

From 645463b9f0e1029a67f089eb72681555da82c93d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Jun 2025 11:28:12 -0700
Subject: [PATCH 0762/1701] Add fallback values for theme colors

---
 modules/ui.py | 30 +++++++++++++++---------------
 1 file changed, 15 insertions(+), 15 deletions(-)

diff --git a/modules/ui.py b/modules/ui.py
index 0e8afa8f5a..fb719158c5 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -69,9 +69,9 @@
         border_color_primary='#c5c5d2',
         body_text_color_subdued='#484848',
         background_fill_secondary='#eaeaea',
-        background_fill_secondary_dark='var(--selected-item-color-dark)',
+        background_fill_secondary_dark='var(--selected-item-color-dark, #282930)',
         background_fill_primary='var(--neutral-50)',
-        background_fill_primary_dark='var(--darker-gray)',
+        background_fill_primary_dark='var(--darker-gray, #1C1C1D)',
         body_background_fill="white",
         block_background_fill="transparent",
         body_text_color='rgb(64, 64, 64)',
@@ -81,25 +81,25 @@
         button_shadow_hover="none",
 
         # Dark Mode Colors
-        input_background_fill_dark='var(--darker-gray)',
-        checkbox_background_color_dark='var(--darker-gray)',
+        input_background_fill_dark='var(--darker-gray, #1C1C1D)',
+        checkbox_background_color_dark='var(--darker-gray, #1C1C1D)',
         block_background_fill_dark='transparent',
         block_border_color_dark='transparent',
-        input_border_color_dark='var(--border-color-dark)',
-        input_border_color_focus_dark='var(--border-color-dark)',
-        checkbox_border_color_dark='var(--border-color-dark)',
-        border_color_primary_dark='var(--border-color-dark)',
-        button_secondary_border_color_dark='var(--border-color-dark)',
-        body_background_fill_dark='var(--dark-gray)',
+        input_border_color_dark='var(--border-color-dark, #525252)',
+        input_border_color_focus_dark='var(--border-color-dark, #525252)',
+        checkbox_border_color_dark='var(--border-color-dark, #525252)',
+        border_color_primary_dark='var(--border-color-dark, #525252)',
+        button_secondary_border_color_dark='var(--border-color-dark, #525252)',
+        body_background_fill_dark='var(--dark-gray, #212125)',
         button_primary_background_fill_dark='transparent',
         button_secondary_background_fill_dark='transparent',
         checkbox_label_background_fill_dark='transparent',
         button_cancel_background_fill_dark='transparent',
-        button_secondary_background_fill_hover_dark='var(--selected-item-color-dark)',
-        checkbox_label_background_fill_hover_dark='var(--selected-item-color-dark)',
-        table_even_background_fill_dark='var(--darker-gray)',
-        table_odd_background_fill_dark='var(--selected-item-color-dark)',
-        code_background_fill_dark='var(--darker-gray)',
+        button_secondary_background_fill_hover_dark='var(--selected-item-color-dark, #282930)',
+        checkbox_label_background_fill_hover_dark='var(--selected-item-color-dark, #282930)',
+        table_even_background_fill_dark='var(--darker-gray, #1C1C1D)',
+        table_odd_background_fill_dark='var(--selected-item-color-dark, #282930)',
+        code_background_fill_dark='var(--darker-gray, #1C1C1D)',
 
         # Shadows and Radius
         checkbox_label_shadow='none',

From 3344510553775c1b21c290ec499fb9b02c06396f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Jun 2025 12:03:19 -0700
Subject: [PATCH 0763/1701] Force dark theme on the Gradio login page

---
 js/global_scope_js.js     | 22 ++++++++++++++++++++++
 modules/block_requests.py |  8 ++++++++
 2 files changed, 30 insertions(+)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 205d937535..eff9115680 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -351,3 +351,25 @@ function handleMorphdomUpdate(data) {
     }
   });
 }
+
+// Wait for Gradio to finish setting its styles, then force dark theme
+const observer = new MutationObserver((mutations) => {
+    mutations.forEach((mutation) => {
+        if (mutation.type === 'attributes' &&
+            mutation.target.tagName === 'GRADIO-APP' &&
+            mutation.attributeName === 'style') {
+
+            // Gradio just set its styles, now force dark theme
+            document.body.classList.add('dark');
+
+            observer.disconnect();
+        }
+    });
+});
+
+// Start observing
+observer.observe(document.documentElement, {
+    attributes: true,
+    subtree: true,
+    attributeFilter: ['style']
+});
diff --git a/modules/block_requests.py b/modules/block_requests.py
index 5a4b533f74..dc1ee4671e 100644
--- a/modules/block_requests.py
+++ b/modules/block_requests.py
@@ -1,5 +1,6 @@
 import builtins
 import io
+import re
 
 import requests
 
@@ -62,6 +63,13 @@ def my_open(*args, **kwargs):
             '\n  '
         )
 
+        file_contents = re.sub(
+            r'@media \(prefers-color-scheme: dark\) \{\s*body \{([^}]*)\}\s*\}',
+            r'body.dark {\1}',
+            file_contents,
+            flags=re.DOTALL
+        )
+
         if len(args) > 1 and args[1] == 'rb':
             file_contents = file_contents.encode('utf-8')
             return io.BytesIO(file_contents)

From a1b606a6ac8016236b20963fdd73da30b9d41236 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Jun 2025 12:30:57 -0700
Subject: [PATCH 0764/1701] Fix obtaining the maximum number of GPU layers for
 DeepSeek-R1-0528-GGUF

---
 modules/models_settings.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 37aa37cf52..6b22499ea1 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -77,7 +77,7 @@ def get_model_metadata(model):
                 model_settings['compress_pos_emb'] = metadata[k]
             elif k.endswith('rope.scaling.factor'):
                 model_settings['compress_pos_emb'] = metadata[k]
-            elif k.endswith('block_count'):
+            elif k.endswith('.block_count'):
                 model_settings['gpu_layers'] = metadata[k] + 1
                 model_settings['max_gpu_layers'] = metadata[k] + 1
 

From ee945517ff7cc62f88c55d28f881a2a6096268ef Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Jun 2025 12:39:53 -0700
Subject: [PATCH 0765/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index ce3602a1fc..d75b85913f 100644
--- a/README.md
+++ b/README.md
@@ -15,11 +15,11 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 - Supports multiple local text generation backends, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).
 - Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory.
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
-- Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
 - **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
 - **Web search**: Optionally search the internet with LLM-generated queries to add context to the conversation.
 - Aesthetic UI with dark and light themes.
 - `instruct` mode for instruction-following (like ChatGPT), and `chat-instruct`/`chat` modes for talking to custom characters.
+- Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
 - Edit messages, navigate between message versions, and branch conversations at any point.
 - Multiple sampling parameters and generation options for sophisticated text generation control.
 - Switch between different models in the UI without restarting.

From 90f42f311a5f6ae41ae822c262b8bc1d1c7313ae Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Jun 2025 12:43:05 -0700
Subject: [PATCH 0766/1701] Update README

---
 README.md | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/README.md b/README.md
index d75b85913f..ac508f7854 100644
--- a/README.md
+++ b/README.md
@@ -82,6 +82,10 @@ Manual portable installation with venv
 Very fast setup that should work on any Python 3.9+:
 
 ```bash
+# Clone repository
+git clone https://github.com/oobabooga/text-generation-webui
+cd text-generation-webui
+
 # Create virtual environment
 python -m venv venv
 
@@ -91,10 +95,6 @@ venv\Scripts\activate
 # On macOS/Linux:
 source venv/bin/activate
 
-# Clone the repository
-git clone https://github.com/oobabooga/text-generation-webui
-cd text-generation-webui
-
 # Install dependencies (choose appropriate file under requirements/portable for your hardware)
 pip install -r requirements/portable/requirements.txt
 

From 2517ea9c9e661f7b808380ecdac2540f140017c7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Jun 2025 15:23:06 -0700
Subject: [PATCH 0767/1701] Lint

---
 js/global_scope_js.js | 27 +++++++++++++--------------
 1 file changed, 13 insertions(+), 14 deletions(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index eff9115680..ebed1f3dcd 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -354,22 +354,21 @@ function handleMorphdomUpdate(data) {
 
 // Wait for Gradio to finish setting its styles, then force dark theme
 const observer = new MutationObserver((mutations) => {
-    mutations.forEach((mutation) => {
-        if (mutation.type === 'attributes' &&
-            mutation.target.tagName === 'GRADIO-APP' &&
-            mutation.attributeName === 'style') {
-
-            // Gradio just set its styles, now force dark theme
-            document.body.classList.add('dark');
-
-            observer.disconnect();
-        }
-    });
+  mutations.forEach((mutation) => {
+    if (mutation.type === "attributes" &&
+        mutation.target.tagName === "GRADIO-APP" &&
+        mutation.attributeName === "style") {
+
+      // Gradio just set its styles, now force dark theme
+      document.body.classList.add("dark");
+      observer.disconnect();
+    }
+  });
 });
 
 // Start observing
 observer.observe(document.documentElement, {
-    attributes: true,
-    subtree: true,
-    attributeFilter: ['style']
+  attributes: true,
+  subtree: true,
+  attributeFilter: ["style"]
 });

From f08db63fbc800181625fe45813b34918010a6297 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Jun 2025 15:26:45 -0700
Subject: [PATCH 0768/1701] Change some comments

---
 modules/ui_model_menu.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 1ee8089e33..ba8c60515d 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -255,7 +255,7 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
         # Check for multiple GGUF files
         gguf_files = [link for link in links if link.lower().endswith('.gguf')]
         if len(gguf_files) > 1 and not specific_file:
-            # Sort GGUF files by size (ascending - smallest first)
+            # Sort by size in ascending order
             gguf_data = []
             for i, link in enumerate(links):
                 if link.lower().endswith('.gguf'):
@@ -274,7 +274,7 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
             return
 
         if return_links:
-            # Sort files by size (ascending - smallest first)
+            # Sort by size in ascending order
             file_data = list(zip(file_sizes, links))
             file_data.sort(key=lambda x: x[0])
 

From acd57b6a85fff1fef5152629b472070be2d38232 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Jun 2025 15:39:43 -0700
Subject: [PATCH 0769/1701] Minor UI change

---
 modules/ui_model_menu.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index ba8c60515d..a18d7d10d0 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -264,7 +264,7 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
 
             gguf_data.sort(key=lambda x: x[0])
 
-            output = "Multiple GGUF files found. Please copy one of the following filenames to the 'File name' field:\n\n```\n"
+            output = "Multiple GGUF files found. Please copy one of the following filenames to the 'File name' field above:\n\n```\n"
             for file_size, link in gguf_data:
                 size_str = format_file_size(file_size)
                 output += f"{size_str} - {Path(link).name}\n"

From f154aeafeafe131ca09802c2dfc12470f0fb68df Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Jun 2025 21:23:10 -0700
Subject: [PATCH 0770/1701] Optimize chat scrolling for the 40th time,
 hopefully the last one

---
 css/main.css | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/css/main.css b/css/main.css
index bc59f8339a..b9bf2ba554 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1670,3 +1670,15 @@ button:focus {
 #textbox-notebook span {
     display: none;
 }
+
+.chat-parent {
+    /* Optimize for scrolling performance */
+    will-change: scroll-position;
+    contain: layout style paint;
+
+    /* Ensure GPU acceleration */
+    transform: translateZ(0);
+
+    /* Prevent layout shifts */
+    overflow-anchor: none;
+}

From bb97ca1b22dfb6674c72e13b400b64ffaac59684 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Jun 2025 21:41:41 -0700
Subject: [PATCH 0771/1701] Fix a small issue with the chat input

---
 css/main.css | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/css/main.css b/css/main.css
index b9bf2ba554..531d622a12 100644
--- a/css/main.css
+++ b/css/main.css
@@ -670,7 +670,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     position: absolute;
     background-color: transparent;
     left: 23px;
-    top: -5px;
+    top: -25px;
     padding: var(--block-padding);
 }
 
@@ -830,7 +830,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
 
 #chat-input-row {
     padding: 1rem;
-    padding-top: 0;
 }
 
 #chat-input-row.bigchat {
@@ -1682,3 +1681,7 @@ button:focus {
     /* Prevent layout shifts */
     overflow-anchor: none;
 }
+
+#chat-input span {
+    display: none;
+}

From 58282f7107635d80d415651af764d590a1e5a9f1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 20 Jun 2025 06:59:48 -0700
Subject: [PATCH 0772/1701] Replace 'Generate' with 'Send' in the Chat tab

---
 css/main.css       | 4 ----
 modules/ui_chat.py | 2 +-
 2 files changed, 1 insertion(+), 5 deletions(-)

diff --git a/css/main.css b/css/main.css
index 531d622a12..036a2bb73e 100644
--- a/css/main.css
+++ b/css/main.css
@@ -365,10 +365,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     gap: 0;
 }
 
-#chat-tab button#Generate, #chat-tab button#stop {
-    width: 89.3438px !important;
-}
-
 #chat-tab button, #notebook-tab button, #default-tab button {
     min-width: 0 !important;
 }
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 8a90608fa6..8ed0aab195 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -60,7 +60,7 @@ def create_ui():
                     with gr.Column(scale=1, elem_id='generate-stop-container'):
                         with gr.Row():
                             shared.gradio['Stop'] = gr.Button('Stop', elem_id='stop', visible=False)
-                            shared.gradio['Generate'] = gr.Button('Generate', elem_id='Generate', variant='primary')
+                            shared.gradio['Generate'] = gr.Button('Send', elem_id='Generate', variant='primary')
 
         # Hover menu buttons
         with gr.Column(elem_id='chat-buttons'):

From 02f604479d492b0199bba260c5e8235ae9730d28 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 21 Jun 2025 14:02:17 -0700
Subject: [PATCH 0773/1701] Remove the pre-jinja2 custom stopping string
 handling (closes #7094)

---
 modules/models_settings.py   | 1 -
 user_data/models/config.yaml | 5 -----
 2 files changed, 6 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 6b22499ea1..bea5b4d64c 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -23,7 +23,6 @@ def get_fallback_settings():
         'truncation_length': shared.settings['truncation_length'],
         'truncation_length_info': shared.settings['truncation_length'],
         'skip_special_tokens': shared.settings['skip_special_tokens'],
-        'custom_stopping_strings': shared.settings['custom_stopping_strings'],
     }
 
 
diff --git a/user_data/models/config.yaml b/user_data/models/config.yaml
index 8521c4c697..038ebcf1ed 100644
--- a/user_data/models/config.yaml
+++ b/user_data/models/config.yaml
@@ -33,18 +33,14 @@
   skip_special_tokens: false
 .*alpaca-native-4bit:
   instruction_template: 'Alpaca'
-  custom_stopping_strings: '"### End"'
 .*llava:
   instruction_template: 'LLaVA'
-  custom_stopping_strings: '"\n###"'
 .*llava.*1.5:
   instruction_template: 'Vicuna-v1.1'
 .*wizard.*mega:
   instruction_template: 'Wizard-Mega'
-  custom_stopping_strings: '""'
 .*starchat-beta:
   instruction_template: 'Starchat-Beta'
-  custom_stopping_strings: '"<|end|>"'
 (?!.*v0)(?!.*1.1)(?!.*1_1)(?!.*stable)(?!.*chinese).*vicuna:
   instruction_template: 'Vicuna-v0'
 .*vicuna.*v0:
@@ -151,7 +147,6 @@
   instruction_template: 'Alpaca'
 .*(openorca-platypus2):
   instruction_template: 'OpenOrca-Platypus2'
-  custom_stopping_strings: '"### Instruction:", "### Response:"'
 .*longchat:
   instruction_template: 'Vicuna-v1.1'
 .*vicuna-33b:

From 1b19dd77a4f01684a6a436914c4dd37a0a7975a5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 22 Jun 2025 17:29:17 -0700
Subject: [PATCH 0774/1701] Move 'Enable thinking' to the Chat tab

---
 modules/ui_chat.py       | 5 ++---
 modules/ui_parameters.py | 1 -
 2 files changed, 2 insertions(+), 4 deletions(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 8ed0aab195..159df235c4 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -78,9 +78,8 @@ def create_ui():
                 with gr.Row():
                     shared.gradio['start_with'] = gr.Textbox(label='Start reply with', placeholder='Sure thing!', value=shared.settings['start_with'], elem_classes=['add_scrollbar'])
 
-                with gr.Row():
-                    shared.gradio['enable_web_search'] = gr.Checkbox(value=shared.settings.get('enable_web_search', False), label='Activate web search', elem_id='web-search')
-
+                shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', info='Used by Qwen3 to toggle  mode.')
+                shared.gradio['enable_web_search'] = gr.Checkbox(value=shared.settings.get('enable_web_search', False), label='Activate web search', elem_id='web-search')
                 with gr.Row(visible=shared.settings.get('enable_web_search', False)) as shared.gradio['web_search_row']:
                     shared.gradio['web_search_pages'] = gr.Number(value=shared.settings.get('web_search_pages', 3), precision=0, label='Number of pages to download', minimum=1, maximum=10)
 
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index e42e4c0cbe..5c92f32edb 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -85,7 +85,6 @@ def create_ui():
                             shared.gradio['auto_max_new_tokens'] = gr.Checkbox(value=shared.settings['auto_max_new_tokens'], label='auto_max_new_tokens', info='Expand max_new_tokens to the available context length.')
                             shared.gradio['ban_eos_token'] = gr.Checkbox(value=shared.settings['ban_eos_token'], label='Ban the eos_token', info='Forces the model to never end the generation prematurely.')
                             shared.gradio['add_bos_token'] = gr.Checkbox(value=shared.settings['add_bos_token'], label='Add the bos_token to the beginning of prompts', info='Disabling this can make the replies more creative.')
-                            shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', info='Used by Qwen3 to toggle  mode.')
                             shared.gradio['skip_special_tokens'] = gr.Checkbox(value=shared.settings['skip_special_tokens'], label='Skip special tokens', info='Some specific models need this unset.')
                             shared.gradio['stream'] = gr.Checkbox(value=shared.settings['stream'], label='Activate text streaming')
                             shared.gradio['static_cache'] = gr.Checkbox(value=shared.settings['static_cache'], label='Static KV cache', info='Use a static cache for improved performance.')

From c3faecfd2737c84609b5b758426e5f1546f3cc46 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 22 Jun 2025 17:51:09 -0700
Subject: [PATCH 0775/1701] Minor change

---
 modules/ui_chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 159df235c4..e2d8935a77 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -19,6 +19,7 @@ def create_ui():
 
     shared.gradio['Chat input'] = gr.State()
     shared.gradio['history'] = gr.State({'internal': [], 'visible': [], 'metadata': {}})
+    shared.gradio['display'] = gr.JSON(value={}, visible=False)  # Hidden buffer
 
     with gr.Tab('Chat', id='Chat', elem_id='chat-tab'):
         with gr.Row(elem_id='past-chats-row', elem_classes=['pretty_scrollbar']):
@@ -47,7 +48,6 @@ def create_ui():
 
         with gr.Row():
             with gr.Column(elem_id='chat-col'):
-                shared.gradio['display'] = gr.JSON(value={}, visible=False)  # Hidden buffer
                 shared.gradio['html_display'] = gr.HTML(value=chat_html_wrapper({'internal': [], 'visible': [], 'metadata': {}}, '', '', 'chat', 'cai-chat', '')['html'], visible=True)
                 with gr.Row(elem_id="chat-input-row"):
                     with gr.Column(scale=1, elem_id='gr-hover-container'):

From 9a58964834833e37fc830f06f4d4a87a142d4179 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 22 Jun 2025 20:25:53 -0700
Subject: [PATCH 0776/1701] Keep the last message visible when the input height
 changes

---
 js/main.js | 39 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)

diff --git a/js/main.js b/js/main.js
index 3ff4bf06c7..ea046e435f 100644
--- a/js/main.js
+++ b/js/main.js
@@ -1046,3 +1046,42 @@ new MutationObserver(() => addMiniDeletes()).observe(
   {childList: true, subtree: true}
 );
 addMiniDeletes();
+
+//------------------------------------------------
+// Maintain distance from bottom when input height changes
+//------------------------------------------------
+let wasAtBottom = false;
+let preservedDistance = 0;
+
+function checkIfAtBottom() {
+  const distanceFromBottom = targetElement.scrollHeight - targetElement.scrollTop - targetElement.clientHeight;
+  wasAtBottom = distanceFromBottom <= 1; // Allow for rounding errors
+}
+
+function preserveScrollPosition() {
+  preservedDistance = targetElement.scrollHeight - targetElement.scrollTop - targetElement.clientHeight;
+}
+
+function restoreScrollPosition() {
+  if (wasAtBottom) {
+    // Force to bottom
+    targetElement.scrollTop = targetElement.scrollHeight - targetElement.clientHeight;
+  } else {
+    // Restore original distance
+    targetElement.scrollTop = targetElement.scrollHeight - targetElement.clientHeight - preservedDistance;
+  }
+}
+
+// Check position before input
+chatInput.addEventListener("beforeinput", () => {
+  checkIfAtBottom();
+  preserveScrollPosition();
+});
+
+// Restore after input
+chatInput.addEventListener("input", () => {
+  requestAnimationFrame(() => restoreScrollPosition());
+});
+
+// Update wasAtBottom when user scrolls
+targetElement.addEventListener("scroll", checkIfAtBottom);

From ab162f976c8ef3622d061420a1f47312c727fd7a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 3 Jul 2025 11:31:52 -0700
Subject: [PATCH 0777/1701] Use miniforge instead of miniconda to avoid
 anaconda licensing issues

---
 README.md         | 12 +++++++-----
 start_linux.sh    |  2 +-
 start_macos.sh    |  2 +-
 start_windows.bat |  4 ++--
 4 files changed, 11 insertions(+), 9 deletions(-)

diff --git a/README.md b/README.md
index ac508f7854..ee4a326a02 100644
--- a/README.md
+++ b/README.md
@@ -62,7 +62,7 @@ One-click installer details
 
 ### One-click-installer
 
-The script uses Miniconda to set up a Conda environment in the `installer_files` folder.
+The script uses Miniforge to set up a Conda environment in the `installer_files` folder.
 
 If you ever need to install something manually in the `installer_files` environment, you can launch an interactive shell using the cmd script: `cmd_linux.sh`, `cmd_windows.bat`, or `cmd_macos.sh`.
 
@@ -115,15 +115,17 @@ Manual full installation with conda or docker
 
 #### 0. Install Conda
 
-https://docs.conda.io/en/latest/miniconda.html
+https://github.com/conda-forge/miniforge
 
-On Linux or WSL, it can be automatically installed with these two commands ([source](https://educe-ubc.github.io/conda.html)):
+On Linux or WSL, Miniforge can be automatically installed with these two commands:
 
 ```
-curl -sL "https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh" > "Miniconda3.sh"
-bash Miniconda3.sh
+curl -sL "https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-Linux-x86_64.sh" > "Miniforge3.sh"
+bash Miniforge3.sh
 ```
 
+For other platforms, download from: https://github.com/conda-forge/miniforge/releases/latest
+
 #### 1. Create a new conda environment
 
 ```
diff --git a/start_linux.sh b/start_linux.sh
index e2b005587e..0b9e65ab6b 100755
--- a/start_linux.sh
+++ b/start_linux.sh
@@ -30,7 +30,7 @@ esac
 INSTALL_DIR="$(pwd)/installer_files"
 CONDA_ROOT_PREFIX="$(pwd)/installer_files/conda"
 INSTALL_ENV_DIR="$(pwd)/installer_files/env"
-MINICONDA_DOWNLOAD_URL="https://repo.anaconda.com/miniconda/Miniconda3-py311_24.11.1-0-Linux-${OS_ARCH}.sh"
+MINICONDA_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-Linux-${OS_ARCH}.sh"
 conda_exists="F"
 
 # figure out whether git and conda needs to be installed
diff --git a/start_macos.sh b/start_macos.sh
index bff11bc177..2e1a784297 100755
--- a/start_macos.sh
+++ b/start_macos.sh
@@ -30,7 +30,7 @@ esac
 INSTALL_DIR="$(pwd)/installer_files"
 CONDA_ROOT_PREFIX="$(pwd)/installer_files/conda"
 INSTALL_ENV_DIR="$(pwd)/installer_files/env"
-MINICONDA_DOWNLOAD_URL="https://repo.anaconda.com/miniconda/Miniconda3-py311_24.11.1-0-MacOSX-${OS_ARCH}.sh"
+MINICONDA_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-MacOSX-${OS_ARCH}.sh"
 conda_exists="F"
 
 # figure out whether git and conda needs to be installed
diff --git a/start_windows.bat b/start_windows.bat
index f5e66ec281..020baeae0e 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -36,8 +36,8 @@ set TEMP=%cd%\installer_files
 set INSTALL_DIR=%cd%\installer_files
 set CONDA_ROOT_PREFIX=%cd%\installer_files\conda
 set INSTALL_ENV_DIR=%cd%\installer_files\env
-set MINICONDA_DOWNLOAD_URL=https://repo.anaconda.com/miniconda/Miniconda3-py311_24.11.1-0-Windows-x86_64.exe
-set MINICONDA_CHECKSUM=43dcbcc315ff91edf959e002cd2f1ede38c64b999fefcc951bccf2ed69c9e8bb
+set MINICONDA_DOWNLOAD_URL=https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-Windows-x86_64.exe
+set MINICONDA_CHECKSUM=b48cd98430170983076dfb51769a6d37668176f59bf3b59c4b21ac4c9bc24f39
 set conda_exists=F
 
 @rem figure out whether git and conda needs to be installed

From 13373391df5923aea218807ce1d0c863549958c2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 3 Jul 2025 14:13:22 -0700
Subject: [PATCH 0778/1701] Rename miniconda -> miniforge everywhere

---
 cmd_linux.sh              |  2 +-
 cmd_macos.sh              |  2 +-
 cmd_windows.bat           |  4 ++--
 start_linux.sh            | 20 ++++++++++----------
 start_macos.sh            | 20 ++++++++++----------
 start_windows.bat         | 34 +++++++++++++++++-----------------
 update_wizard_linux.sh    |  2 +-
 update_wizard_macos.sh    |  2 +-
 update_wizard_windows.bat |  4 ++--
 9 files changed, 45 insertions(+), 45 deletions(-)

diff --git a/cmd_linux.sh b/cmd_linux.sh
index 576dbf0223..e124aab9d6 100755
--- a/cmd_linux.sh
+++ b/cmd_linux.sh
@@ -2,7 +2,7 @@
 
 cd "$(dirname "${BASH_SOURCE[0]}")"
 
-if [[ "$(pwd)" =~ " " ]]; then echo This script relies on Miniconda which can not be silently installed under a path with spaces. && exit; fi
+if [[ "$(pwd)" =~ " " ]]; then echo This script relies on Miniforge which can not be silently installed under a path with spaces. && exit; fi
 
 # deactivate existing conda envs as needed to avoid conflicts
 { conda deactivate && conda deactivate && conda deactivate; } 2> /dev/null
diff --git a/cmd_macos.sh b/cmd_macos.sh
index 1b052e5c34..7e1a379fd6 100755
--- a/cmd_macos.sh
+++ b/cmd_macos.sh
@@ -2,7 +2,7 @@
 
 cd "$(dirname "${BASH_SOURCE[0]}")"
 
-if [[ "$(pwd)" =~ " " ]]; then echo This script relies on Miniconda which can not be silently installed under a path with spaces. && exit; fi
+if [[ "$(pwd)" =~ " " ]]; then echo This script relies on Miniforge which can not be silently installed under a path with spaces. && exit; fi
 
 # deactivate existing conda envs as needed to avoid conflicts
 { conda deactivate && conda deactivate && conda deactivate; } 2> /dev/null
diff --git a/cmd_windows.bat b/cmd_windows.bat
index 531a326158..787b433501 100755
--- a/cmd_windows.bat
+++ b/cmd_windows.bat
@@ -4,7 +4,7 @@ cd /D "%~dp0"
 
 set PATH=%PATH%;%SystemRoot%\system32
 
-echo "%CD%"| findstr /C:" " >nul && echo This script relies on Miniconda which can not be silently installed under a path with spaces. && goto end
+echo "%CD%"| findstr /C:" " >nul && echo This script relies on Miniforge which can not be silently installed under a path with spaces. && goto end
 
 @rem fix failed install when installing to a separate drive
 set TMP=%cd%\installer_files
@@ -25,7 +25,7 @@ set "CUDA_PATH=%INSTALL_ENV_DIR%"
 set "CUDA_HOME=%CUDA_PATH%"
 
 @rem activate installer env
-call "%CONDA_ROOT_PREFIX%\condabin\conda.bat" activate "%INSTALL_ENV_DIR%" || ( echo. && echo Miniconda hook not found. && goto end )
+call "%CONDA_ROOT_PREFIX%\condabin\conda.bat" activate "%INSTALL_ENV_DIR%" || ( echo. && echo Miniforge hook not found. && goto end )
 
 @rem enter commands
 cmd /k "%*"
diff --git a/start_linux.sh b/start_linux.sh
index 0b9e65ab6b..a991e1bc9f 100755
--- a/start_linux.sh
+++ b/start_linux.sh
@@ -13,7 +13,7 @@ if [ -d "portable_env" ]; then
     exit $?
 fi
 
-if [[ "$(pwd)" =~ " " ]]; then echo This script relies on Miniconda which can not be silently installed under a path with spaces. && exit; fi
+if [[ "$(pwd)" =~ " " ]]; then echo This script relies on Miniforge which can not be silently installed under a path with spaces. && exit; fi
 
 # deactivate existing conda envs as needed to avoid conflicts
 { conda deactivate && conda deactivate && conda deactivate; } 2> /dev/null
@@ -30,29 +30,29 @@ esac
 INSTALL_DIR="$(pwd)/installer_files"
 CONDA_ROOT_PREFIX="$(pwd)/installer_files/conda"
 INSTALL_ENV_DIR="$(pwd)/installer_files/env"
-MINICONDA_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-Linux-${OS_ARCH}.sh"
+MINIFORGE_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-Linux-${OS_ARCH}.sh"
 conda_exists="F"
 
 # figure out whether git and conda needs to be installed
 if "$CONDA_ROOT_PREFIX/bin/conda" --version &>/dev/null; then conda_exists="T"; fi
 
 # (if necessary) install git and conda into a contained environment
-# download miniconda
+# download miniforge
 if [ "$conda_exists" == "F" ]; then
-    echo "Downloading Miniconda from $MINICONDA_DOWNLOAD_URL to $INSTALL_DIR/miniconda_installer.sh"
+    echo "Downloading Miniforge from $MINIFORGE_DOWNLOAD_URL to $INSTALL_DIR/miniforge_installer.sh"
 
     mkdir -p "$INSTALL_DIR"
-    curl -L "$MINICONDA_DOWNLOAD_URL" > "$INSTALL_DIR/miniconda_installer.sh"
+    curl -L "$MINIFORGE_DOWNLOAD_URL" > "$INSTALL_DIR/miniforge_installer.sh"
 
-    chmod u+x "$INSTALL_DIR/miniconda_installer.sh"
-    bash "$INSTALL_DIR/miniconda_installer.sh" -b -p $CONDA_ROOT_PREFIX
+    chmod u+x "$INSTALL_DIR/miniforge_installer.sh"
+    bash "$INSTALL_DIR/miniforge_installer.sh" -b -p $CONDA_ROOT_PREFIX
 
     # test the conda binary
-    echo "Miniconda version:"
+    echo "Miniforge version:"
     "$CONDA_ROOT_PREFIX/bin/conda" --version
 
-    # delete the Miniconda installer
-    rm "$INSTALL_DIR/miniconda_installer.sh"
+    # delete the Miniforge installer
+    rm "$INSTALL_DIR/miniforge_installer.sh"
 fi
 
 # create the installer env
diff --git a/start_macos.sh b/start_macos.sh
index 2e1a784297..6426229abd 100755
--- a/start_macos.sh
+++ b/start_macos.sh
@@ -13,7 +13,7 @@ if [ -d "portable_env" ]; then
     exit $?
 fi
 
-if [[ "$(pwd)" =~ " " ]]; then echo This script relies on Miniconda which can not be silently installed under a path with spaces. && exit; fi
+if [[ "$(pwd)" =~ " " ]]; then echo This script relies on Miniforge which can not be silently installed under a path with spaces. && exit; fi
 
 # deactivate existing conda envs as needed to avoid conflicts
 { conda deactivate && conda deactivate && conda deactivate; } 2> /dev/null
@@ -30,29 +30,29 @@ esac
 INSTALL_DIR="$(pwd)/installer_files"
 CONDA_ROOT_PREFIX="$(pwd)/installer_files/conda"
 INSTALL_ENV_DIR="$(pwd)/installer_files/env"
-MINICONDA_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-MacOSX-${OS_ARCH}.sh"
+MINIFORGE_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-MacOSX-${OS_ARCH}.sh"
 conda_exists="F"
 
 # figure out whether git and conda needs to be installed
 if "$CONDA_ROOT_PREFIX/bin/conda" --version &>/dev/null; then conda_exists="T"; fi
 
 # (if necessary) install git and conda into a contained environment
-# download miniconda
+# download miniforge
 if [ "$conda_exists" == "F" ]; then
-    echo "Downloading Miniconda from $MINICONDA_DOWNLOAD_URL to $INSTALL_DIR/miniconda_installer.sh"
+    echo "Downloading Miniforge from $MINIFORGE_DOWNLOAD_URL to $INSTALL_DIR/miniforge_installer.sh"
 
     mkdir -p "$INSTALL_DIR"
-    curl -L "$MINICONDA_DOWNLOAD_URL" > "$INSTALL_DIR/miniconda_installer.sh"
+    curl -L "$MINIFORGE_DOWNLOAD_URL" > "$INSTALL_DIR/miniforge_installer.sh"
 
-    chmod u+x "$INSTALL_DIR/miniconda_installer.sh"
-    bash "$INSTALL_DIR/miniconda_installer.sh" -b -p $CONDA_ROOT_PREFIX
+    chmod u+x "$INSTALL_DIR/miniforge_installer.sh"
+    bash "$INSTALL_DIR/miniforge_installer.sh" -b -p $CONDA_ROOT_PREFIX
 
     # test the conda binary
-    echo "Miniconda version:"
+    echo "Miniforge version:"
     "$CONDA_ROOT_PREFIX/bin/conda" --version
 
-    # delete the Miniconda installer
-    rm "$INSTALL_DIR/miniconda_installer.sh"
+    # delete the Miniforge installer
+    rm "$INSTALL_DIR/miniforge_installer.sh"
 fi
 
 # create the installer env
diff --git a/start_windows.bat b/start_windows.bat
index 020baeae0e..68c6194e76 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -16,7 +16,7 @@ if exist "portable_env" (
 
 set PATH=%PATH%;%SystemRoot%\system32
 
-echo "%CD%"| findstr /C:" " >nul && echo This script relies on Miniconda which can not be silently installed under a path with spaces. && goto end
+echo "%CD%"| findstr /C:" " >nul && echo This script relies on Miniforge which can not be silently installed under a path with spaces. && goto end
 
 @rem Check for special characters in installation path
 set "SPCHARMESSAGE="WARNING: Special characters were detected in the installation path!" "         This can cause the installation to fail!""
@@ -36,8 +36,8 @@ set TEMP=%cd%\installer_files
 set INSTALL_DIR=%cd%\installer_files
 set CONDA_ROOT_PREFIX=%cd%\installer_files\conda
 set INSTALL_ENV_DIR=%cd%\installer_files\env
-set MINICONDA_DOWNLOAD_URL=https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-Windows-x86_64.exe
-set MINICONDA_CHECKSUM=b48cd98430170983076dfb51769a6d37668176f59bf3b59c4b21ac4c9bc24f39
+set MINIFORGE_DOWNLOAD_URL=https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-Windows-x86_64.exe
+set MINIFORGE_CHECKSUM=b48cd98430170983076dfb51769a6d37668176f59bf3b59c4b21ac4c9bc24f39
 set conda_exists=F
 
 @rem figure out whether git and conda needs to be installed
@@ -47,40 +47,40 @@ if "%ERRORLEVEL%" EQU "0" set conda_exists=T
 @rem (if necessary) install git and conda into a contained environment
 @rem download conda
 if "%conda_exists%" == "F" (
-	echo Downloading Miniconda from %MINICONDA_DOWNLOAD_URL% to %INSTALL_DIR%\miniconda_installer.exe
+	echo Downloading Miniforge from %MINIFORGE_DOWNLOAD_URL% to %INSTALL_DIR%\miniforge_installer.exe
 
 	mkdir "%INSTALL_DIR%"
-	call curl -Lk "%MINICONDA_DOWNLOAD_URL%" > "%INSTALL_DIR%\miniconda_installer.exe" || ( echo. && echo Miniconda failed to download. && goto end )
+	call curl -Lk "%MINIFORGE_DOWNLOAD_URL%" > "%INSTALL_DIR%\miniforge_installer.exe" || ( echo. && echo Miniforge failed to download. && goto end )
 
 	@rem Try CertUtil first
-	for /f %%a in ('CertUtil -hashfile "%INSTALL_DIR%\miniconda_installer.exe" SHA256 ^| find /i /v " " ^| find /i "%MINICONDA_CHECKSUM%"') do (
+	for /f %%a in ('CertUtil -hashfile "%INSTALL_DIR%\miniforge_installer.exe" SHA256 ^| find /i /v " " ^| find /i "%MINIFORGE_CHECKSUM%"') do (
 		set "output=%%a"
 	)
 
 	@rem If CertUtil fails, try PowerShell
 	if not defined output (
-		for /f %%a in ('powershell -Command "if((Get-FileHash \"%INSTALL_DIR%\miniconda_installer.exe\" -Algorithm SHA256).Hash -eq ''%MINICONDA_CHECKSUM%''){echo true}"') do (
+		for /f %%a in ('powershell -Command "if((Get-FileHash \"%INSTALL_DIR%\miniforge_installer.exe\" -Algorithm SHA256).Hash -eq ''%MINIFORGE_CHECKSUM%''){echo true}"') do (
 			set "output=%%a"
 		)
 	)
 
 	if not defined output (
-		echo The checksum verification for miniconda_installer.exe has failed.
-		del "%INSTALL_DIR%\miniconda_installer.exe"
+		echo The checksum verification for miniforge_installer.exe has failed.
+		del "%INSTALL_DIR%\miniforge_installer.exe"
 		goto end
 	) else (
-		echo The checksum verification for miniconda_installer.exe has passed successfully.
+		echo The checksum verification for miniforge_installer.exe has passed successfully.
 	)
 
-	echo Installing Miniconda to %CONDA_ROOT_PREFIX%
-	start /wait "" "%INSTALL_DIR%\miniconda_installer.exe" /InstallationType=JustMe /NoShortcuts=1 /AddToPath=0 /RegisterPython=0 /NoRegistry=1 /S /D=%CONDA_ROOT_PREFIX%
+	echo Installing Miniforge to %CONDA_ROOT_PREFIX%
+	start /wait "" "%INSTALL_DIR%\miniforge_installer.exe" /InstallationType=JustMe /NoShortcuts=1 /AddToPath=0 /RegisterPython=0 /NoRegistry=1 /S /D=%CONDA_ROOT_PREFIX%
 
 	@rem test the conda binary
-	echo Miniconda version:
-	call "%CONDA_ROOT_PREFIX%\_conda.exe" --version || ( echo. && echo Miniconda not found. && goto end )
+	echo Miniforge version:
+	call "%CONDA_ROOT_PREFIX%\_conda.exe" --version || ( echo. && echo Miniforge not found. && goto end )
 
-	@rem delete the Miniconda installer
-	del "%INSTALL_DIR%\miniconda_installer.exe"
+	@rem delete the Miniforge installer
+	del "%INSTALL_DIR%\miniforge_installer.exe"
 )
 
 @rem create the installer env
@@ -96,7 +96,7 @@ set "CUDA_PATH=%INSTALL_ENV_DIR%"
 set "CUDA_HOME=%CUDA_PATH%"
 
 @rem activate installer env
-call "%CONDA_ROOT_PREFIX%\condabin\conda.bat" activate "%INSTALL_ENV_DIR%" || ( echo. && echo Miniconda hook not found. && goto end )
+call "%CONDA_ROOT_PREFIX%\condabin\conda.bat" activate "%INSTALL_ENV_DIR%" || ( echo. && echo Miniforge hook not found. && goto end )
 
 @rem setup installer env
 call python one_click.py %*
diff --git a/update_wizard_linux.sh b/update_wizard_linux.sh
index c81d9d9b20..eb4a753caf 100755
--- a/update_wizard_linux.sh
+++ b/update_wizard_linux.sh
@@ -2,7 +2,7 @@
 
 cd "$(dirname "${BASH_SOURCE[0]}")"
 
-if [[ "$(pwd)" =~ " " ]]; then echo This script relies on Miniconda which can not be silently installed under a path with spaces. && exit; fi
+if [[ "$(pwd)" =~ " " ]]; then echo This script relies on Miniforge which can not be silently installed under a path with spaces. && exit; fi
 
 # deactivate existing conda envs as needed to avoid conflicts
 { conda deactivate && conda deactivate && conda deactivate; } 2> /dev/null
diff --git a/update_wizard_macos.sh b/update_wizard_macos.sh
index f58bb9e94b..61a8110c80 100755
--- a/update_wizard_macos.sh
+++ b/update_wizard_macos.sh
@@ -2,7 +2,7 @@
 
 cd "$(dirname "${BASH_SOURCE[0]}")"
 
-if [[ "$(pwd)" =~ " " ]]; then echo This script relies on Miniconda which can not be silently installed under a path with spaces. && exit; fi
+if [[ "$(pwd)" =~ " " ]]; then echo This script relies on Miniforge which can not be silently installed under a path with spaces. && exit; fi
 
 # deactivate existing conda envs as needed to avoid conflicts
 { conda deactivate && conda deactivate && conda deactivate; } 2> /dev/null
diff --git a/update_wizard_windows.bat b/update_wizard_windows.bat
index fac251a764..e9203c529b 100755
--- a/update_wizard_windows.bat
+++ b/update_wizard_windows.bat
@@ -4,7 +4,7 @@ cd /D "%~dp0"
 
 set PATH=%PATH%;%SystemRoot%\system32
 
-echo "%CD%"| findstr /C:" " >nul && echo This script relies on Miniconda which can not be silently installed under a path with spaces. && goto end
+echo "%CD%"| findstr /C:" " >nul && echo This script relies on Miniforge which can not be silently installed under a path with spaces. && goto end
 
 @rem fix failed install when installing to a separate drive
 set TMP=%cd%\installer_files
@@ -25,7 +25,7 @@ set "CUDA_PATH=%INSTALL_ENV_DIR%"
 set "CUDA_HOME=%CUDA_PATH%"
 
 @rem activate installer env
-call "%CONDA_ROOT_PREFIX%\condabin\conda.bat" activate "%INSTALL_ENV_DIR%" || ( echo. && echo Miniconda hook not found. && goto end )
+call "%CONDA_ROOT_PREFIX%\condabin\conda.bat" activate "%INSTALL_ENV_DIR%" || ( echo. && echo Miniforge hook not found. && goto end )
 
 @rem update installer env
 call python one_click.py --update-wizard && (

From cbba88f5655f949828af0e46b5a12e1a5ac835a2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 3 Jul 2025 18:14:52 -0700
Subject: [PATCH 0779/1701] Fix scrolling during streaming when thinking blocks
 are present

---
 js/main.js | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/js/main.js b/js/main.js
index ea046e435f..c44c144e38 100644
--- a/js/main.js
+++ b/js/main.js
@@ -199,10 +199,11 @@ const observer = new MutationObserver(function(mutations) {
     const lastChild = messagesContainer?.lastElementChild;
     const prevSibling = lastChild?.previousElementSibling;
     if (lastChild && prevSibling) {
-      lastChild.style.setProperty("margin-bottom",
-        `max(0px, calc(max(70vh, 100vh - ${prevSibling.offsetHeight}px - 84px) - ${lastChild.offsetHeight}px))`,
-        "important"
-      );
+      // Add padding to the messages container to create room for the last message.
+      // The purpose of this is to avoid constant scrolling during streaming in
+      // instruct mode.
+      const bufferHeight = Math.max(0, Math.max(0.7 * window.innerHeight, window.innerHeight - prevSibling.offsetHeight - 84) - lastChild.offsetHeight);
+      messagesContainer.style.paddingBottom = `${bufferHeight}px`;
     }
   }
 });

From 877c651c04490d8b13948395010fb53acc84e74d Mon Sep 17 00:00:00 2001
From: zombiegreedo <65383556+zombiegreedo@users.noreply.github.com>
Date: Thu, 3 Jul 2025 22:05:46 -0400
Subject: [PATCH 0780/1701] Handle either missing  start or  end
 tags (#7102)

---
 modules/html_generator.py | 38 +++++++++++++++++++-------------------
 1 file changed, 19 insertions(+), 19 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 11572fc6c0..6844c244d2 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -116,29 +116,29 @@ def extract_thinking_block(string):
     THINK_START_TAG = "<think>"
     THINK_END_TAG = "</think>"
 
-    # Look for opening tag
-    start_pos = string.lstrip().find(THINK_START_TAG)
-    if start_pos == -1:
-        return None, string
-
-    # Adjust start position to account for any leading whitespace
+    # Look for think tag
     start_pos = string.find(THINK_START_TAG)
+    end_pos = string.find(THINK_END_TAG)
 
-    # Find the content after the opening tag
-    content_start = start_pos + len(THINK_START_TAG)
-
-    # Look for closing tag
-    end_pos = string.find(THINK_END_TAG, content_start)
+    # Return if neither tag is in string
+    if start_pos == -1 and end_pos == -1:
+        return None, string
 
-    if end_pos != -1:
-        # Both tags found - extract content between them
-        thinking_content = string[content_start:end_pos]
-        remaining_content = string[end_pos + len(THINK_END_TAG):]
-        return thinking_content, remaining_content
+    # handle missing start or end tags
+    if start_pos == -1:
+        thought_start = 0
     else:
-        # Only opening tag found - everything else is thinking content
-        thinking_content = string[content_start:]
-        return thinking_content, ""
+        thought_start = start_pos + len(THINK_START_TAG)
+    if end_pos == -1:
+        thought_end = len(string)
+        content_start = len(string)
+    else:
+        thought_end = end_pos
+        content_start = end_pos + len(THINK_END_TAG)
+
+    thinking_content = string[thought_start:thought_end]
+    remaining_content = string[content_start:]
+    return thinking_content, remaining_content
 
 
 @functools.lru_cache(maxsize=None)

From 23bb94a5fbc83995bcadd234981991584391146b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 3 Jul 2025 20:36:54 -0700
Subject: [PATCH 0781/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_cuda128.txt             | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt      | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 14828e8db6..cc0bd517a7 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index ec2040d0de..ce7060dc41 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 4b50c69326..560347fbe4 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index a6e2925e61..bb1365920a 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 1b97d8448e..8bb217613c 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index fa553f7cf8..c068c8d8a1 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 505f8ae8aa..bf41929c2f 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 84ffa3276f..aabad9a867 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index da995438e2..ed6ab18670 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 13ecaf879e..ef9d5b80ac 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 49cd626484..77ccdfd6cf 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index fca8208f03..f6a42e55fe 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 86b00651f1..8be7742e39 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index fca46fb524..de5b66b742 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 3ebc248838..dfd71db7b4 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 2bdf257e1b..a19238309a 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 0b98ee5ab4..890b14ffe0 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 188dc08523..e2607dbb82 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.20.0/llama_cpp_binaries-0.20.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 92ec8dda03b6a64a0efb5b4dc6b8109b460f19ad Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 4 Jul 2025 06:04:04 -0700
Subject: [PATCH 0782/1701] Fix chat history getting lost if the UI is inactive
 for a long time (closes #7109)

---
 modules/ui.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui.py b/modules/ui.py
index fb719158c5..0030bb0231 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -296,7 +296,7 @@ def gather_interface_values(*args):
         shared.persistent_interface_state.pop('textbox')
 
     # Prevent history loss if backend is restarted but UI is not refreshed
-    if output['history'] is None and output['unique_id'] is not None:
+    if (output['history'] is None or (len(output['history'].get('visible', [])) == 0 and len(output['history'].get('internal', [])) == 0)) and output['unique_id'] is not None:
         output['history'] = load_history(output['unique_id'], output['character_menu'], output['mode'])
 
     return output

From c4d5331c03e223b84963b81b98e36a1d71fa3338 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 4 Jul 2025 13:21:52 -0700
Subject: [PATCH 0783/1701] Fix autoscroll after fonts load

---
 js/main.js | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/js/main.js b/js/main.js
index c44c144e38..c805fc9a7a 100644
--- a/js/main.js
+++ b/js/main.js
@@ -1086,3 +1086,17 @@ chatInput.addEventListener("input", () => {
 
 // Update wasAtBottom when user scrolls
 targetElement.addEventListener("scroll", checkIfAtBottom);
+
+//------------------------------------------------
+// Fix autoscroll after fonts load
+//------------------------------------------------
+document.fonts.addEventListener("loadingdone", (event) => {
+  setTimeout(() => {
+    if (!window.isScrolled) {
+      const maxScroll = targetElement.scrollHeight - targetElement.clientHeight;
+      if (targetElement.scrollTop < maxScroll - 5) {
+        targetElement.scrollTop = maxScroll;
+      }
+    }
+  }, 50);
+});

From c4d738f39fc5f3b998517ce1778eb8662c1a60c6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 5 Jul 2025 14:09:29 -0700
Subject: [PATCH 0784/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_cuda128.txt             | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt      | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index cc0bd517a7..1f7b8f9e82 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index ce7060dc41..ea2dbf6733 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 560347fbe4..251d5e24e0 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index bb1365920a..649eaaca11 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 8bb217613c..7f1895fc71 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index c068c8d8a1..6ec88d63de 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index bf41929c2f..7e6c780d0c 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index aabad9a867..cbac577ad3 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index ed6ab18670..3f7add1487 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index ef9d5b80ac..9903bf7a14 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 77ccdfd6cf..785ce62364 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index f6a42e55fe..d56d6a79f2 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 8be7742e39..3e00a781b4 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index de5b66b742..8006762e9f 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index dfd71db7b4..9ec080824a 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index a19238309a..8ca5441a2b 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 890b14ffe0..a164c879fa 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index e2607dbb82..3b6943b797 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.21.0/llama_cpp_binaries-0.21.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 4583924ce7ebf56b77fd7ca837a63faeb3d3b697 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 5 Jul 2025 14:24:15 -0700
Subject: [PATCH 0785/1701] Remove torchvision/torchaudio mentions from the
 README

---
 README.md | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/README.md b/README.md
index ee4a326a02..04449e1d95 100644
--- a/README.md
+++ b/README.md
@@ -137,12 +137,12 @@ conda activate textgen
 
 | System | GPU | Command |
 |--------|---------|---------|
-| Linux/WSL | NVIDIA | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/cu124` |
-| Linux/WSL | CPU only | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/cpu` |
-| Linux | AMD | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/rocm6.2.4` |
-| MacOS + MPS | Any | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0` |
-| Windows | NVIDIA | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/cu124` |
-| Windows | CPU only | `pip3 install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0` |
+| Linux/WSL | NVIDIA | `pip3 install torch==2.6.0 --index-url https://download.pytorch.org/whl/cu124` |
+| Linux/WSL | CPU only | `pip3 install torch==2.6.0 --index-url https://download.pytorch.org/whl/cpu` |
+| Linux | AMD | `pip3 install torch==2.6.0 --index-url https://download.pytorch.org/whl/rocm6.2.4` |
+| MacOS + MPS | Any | `pip3 install torch==2.6.0` |
+| Windows | NVIDIA | `pip3 install torch==2.6.0 --index-url https://download.pytorch.org/whl/cu124` |
+| Windows | CPU only | `pip3 install torch==2.6.0` |
 
 The up-to-date commands can be found here: https://pytorch.org/get-started/locally/.
 

From 0f258774d34e756a654bc4d9ed49095cef1d1f67 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 5 Jul 2025 14:25:59 -0700
Subject: [PATCH 0786/1701] Minor README changes

---
 README.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index 04449e1d95..466e4edb29 100644
--- a/README.md
+++ b/README.md
@@ -183,9 +183,9 @@ python server.py
 
 Then browse to
 
-`http://localhost:7860/?__theme=dark`
+`http://127.0.0.1:7860`
 
-##### Manual install
+#### Manual install
 
 The `requirements*.txt` above contain various wheels precompiled through GitHub Actions. If you wish to compile things manually, or if you need to because no suitable wheels are available for your hardware, you can use `requirements_nowheels.txt` and then install your desired loaders manually.
 

From de4ccffff8bd0dff1fe4aca048115af2e398ea75 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 6 Jul 2025 16:24:57 -0700
Subject: [PATCH 0787/1701] Fix the duckduckgo search

---
 modules/web_search.py                         | 26 +++++++++++++------
 requirements/full/requirements.txt            |  1 -
 requirements/full/requirements_amd.txt        |  1 -
 requirements/full/requirements_amd_noavx2.txt |  1 -
 .../full/requirements_apple_intel.txt         |  1 -
 .../full/requirements_apple_silicon.txt       |  1 -
 requirements/full/requirements_cpu_only.txt   |  1 -
 .../full/requirements_cpu_only_noavx2.txt     |  1 -
 requirements/full/requirements_cuda128.txt    |  1 -
 .../full/requirements_cuda128_noavx2.txt      |  1 -
 requirements/full/requirements_noavx2.txt     |  1 -
 requirements/full/requirements_nowheels.txt   |  1 -
 requirements/portable/requirements.txt        |  1 -
 .../portable/requirements_apple_intel.txt     |  1 -
 .../portable/requirements_apple_silicon.txt   |  1 -
 .../portable/requirements_cpu_only.txt        |  1 -
 .../portable/requirements_cpu_only_noavx2.txt |  1 -
 requirements/portable/requirements_noavx2.txt |  1 -
 .../portable/requirements_nowheels.txt        |  1 -
 requirements/portable/requirements_vulkan.txt |  1 -
 .../portable/requirements_vulkan_noavx2.txt   |  1 -
 21 files changed, 18 insertions(+), 28 deletions(-)

diff --git a/modules/web_search.py b/modules/web_search.py
index 401a42bb36..3b1f6e1885 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -1,6 +1,9 @@
 import concurrent.futures
+import html
+import re
 from concurrent.futures import as_completed
 from datetime import datetime
+from urllib.parse import quote_plus
 
 import requests
 
@@ -44,19 +47,26 @@ def download_web_page(url, timeout=10):
         return ""
 
 
-def perform_web_search(query, num_pages=3, max_workers=5):
+def perform_web_search(query, num_pages=3, max_workers=5, timeout=10):
     """Perform web search and return results with content"""
-    from duckduckgo_search import DDGS
-
     try:
-        with DDGS() as ddgs:
-            results = list(ddgs.text(query, max_results=num_pages))
+        # Use DuckDuckGo HTML search endpoint
+        search_url = f"https://html.duckduckgo.com/html/?q={quote_plus(query)}"
+        headers = {'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36'}
+
+        response = requests.get(search_url, headers=headers, timeout=timeout)
+        response.raise_for_status()
+
+        # Extract results with regex
+        titles = re.findall(r']*class="[^"]*result__a[^"]*"[^>]*>(.*?)', response.text, re.DOTALL)
+        urls = re.findall(r']*class="[^"]*result__url[^"]*"[^>]*>(.*?)', response.text, re.DOTALL)
 
         # Prepare download tasks
         download_tasks = []
-        for i, result in enumerate(results):
-            url = result.get('href', '')
-            title = result.get('title', f'Search Result {i+1}')
+        for i in range(min(len(titles), len(urls), num_pages)):
+            url = f"https://{urls[i].strip()}"
+            title = re.sub(r'<[^>]+>', '', titles[i]).strip()
+            title = html.unescape(title)
             download_tasks.append((url, title, i))
 
         search_results = [None] * len(download_tasks)  # Pre-allocate to maintain order
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 1f7b8f9e82..9735fcee4c 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -2,7 +2,6 @@ accelerate==1.5.*
 bitsandbytes==0.45.*
 colorama
 datasets
-duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index ea2dbf6733..c7db011611 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -1,7 +1,6 @@
 accelerate==1.5.*
 colorama
 datasets
-duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 251d5e24e0..10c8933219 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -1,7 +1,6 @@
 accelerate==1.5.*
 colorama
 datasets
-duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 649eaaca11..e900f1bcc7 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -1,7 +1,6 @@
 accelerate==1.5.*
 colorama
 datasets
-duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 7f1895fc71..a853138940 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -1,7 +1,6 @@
 accelerate==1.5.*
 colorama
 datasets
-duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 6ec88d63de..3735a1a850 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -1,7 +1,6 @@
 accelerate==1.5.*
 colorama
 datasets
-duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 7e6c780d0c..f93a4d77b0 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -1,7 +1,6 @@
 accelerate==1.5.*
 colorama
 datasets
-duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index cbac577ad3..b5bee1e43e 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -2,7 +2,6 @@ accelerate==1.5.*
 bitsandbytes==0.45.*
 colorama
 datasets
-duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 3f7add1487..2cd912052a 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -2,7 +2,6 @@ accelerate==1.5.*
 bitsandbytes==0.45.*
 colorama
 datasets
-duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 9903bf7a14..35b683c878 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -2,7 +2,6 @@ accelerate==1.5.*
 bitsandbytes==0.45.*
 colorama
 datasets
-duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index b1c3c6eaf1..30020989eb 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -1,7 +1,6 @@
 accelerate==1.5.*
 colorama
 datasets
-duckduckgo_search==8.0.2
 einops
 fastapi==0.112.4
 gradio==4.37.*
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 785ce62364..d928ff08ce 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -1,4 +1,3 @@
-duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index d56d6a79f2..b25298fc3e 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -1,4 +1,3 @@
-duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 3e00a781b4..e52a7248c7 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -1,4 +1,3 @@
-duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 8006762e9f..7c2578d235 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -1,4 +1,3 @@
-duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 9ec080824a..c125c97e7e 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -1,4 +1,3 @@
-duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 8ca5441a2b..6daa06fc74 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -1,4 +1,3 @@
-duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index c22d6441e1..b7b73eff0e 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -1,4 +1,3 @@
-duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index a164c879fa..247d3a9cc7 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -1,4 +1,3 @@
-duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 3b6943b797..466a1b6e75 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -1,4 +1,3 @@
-duckduckgo_search==8.0.2
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15

From 8df1127ce291dc9f4c7f57da1faa879aa4f87ae7 Mon Sep 17 00:00:00 2001
From: Ali <17217248+abulhawa@users.noreply.github.com>
Date: Mon, 7 Jul 2025 01:42:34 +0200
Subject: [PATCH 0788/1701] Fix: use embedded Python in start_windows.bat to
 avoid system interpreter conflicts (#7120)

---
 start_windows.bat | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/start_windows.bat b/start_windows.bat
index 68c6194e76..08fedc6842 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -99,7 +99,7 @@ set "CUDA_HOME=%CUDA_PATH%"
 call "%CONDA_ROOT_PREFIX%\condabin\conda.bat" activate "%INSTALL_ENV_DIR%" || ( echo. && echo Miniforge hook not found. && goto end )
 
 @rem setup installer env
-call python one_click.py %*
+call "%INSTALL_ENV_DIR%\python.exe" one_click.py %*
 
 @rem below are functions for the script   next line skips these during normal execution
 goto end

From 959d4ddb917b93f0e2abe0551905725cada3ed1a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Philipp=20Cla=C3=9Fen?= 
Date: Mon, 7 Jul 2025 01:51:42 +0200
Subject: [PATCH 0789/1701] Fix for chat sidebars toggle buttons disappearing
 (#7106)

---
 js/main.js         | 4 +---
 modules/ui_chat.py | 2 +-
 2 files changed, 2 insertions(+), 4 deletions(-)

diff --git a/js/main.js b/js/main.js
index c805fc9a7a..17b853f4ec 100644
--- a/js/main.js
+++ b/js/main.js
@@ -27,9 +27,7 @@ document.querySelector(".header_bar").addEventListener("click", function(event)
     this.style.marginBottom = chatVisible ? "0px" : "19px";
 
     if (chatVisible && !showControlsChecked) {
-      document.querySelectorAll(
-        "#chat-tab > div > :nth-child(1), #chat-tab > div > :nth-child(3), #chat-tab > div > :nth-child(4), #extensions"
-      ).forEach(element => {
+      document.querySelectorAll("#extensions").forEach(element => {
         element.style.display = "none";
       });
     }
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index e2d8935a77..4dade1765e 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -21,7 +21,7 @@ def create_ui():
     shared.gradio['history'] = gr.State({'internal': [], 'visible': [], 'metadata': {}})
     shared.gradio['display'] = gr.JSON(value={}, visible=False)  # Hidden buffer
 
-    with gr.Tab('Chat', id='Chat', elem_id='chat-tab'):
+    with gr.Tab('Chat', elem_id='chat-tab'):
         with gr.Row(elem_id='past-chats-row', elem_classes=['pretty_scrollbar']):
             with gr.Column():
                 with gr.Row(elem_id='past-chats-buttons'):

From e6bc7742fb1e28fde10ee939f870eeb75545ec56 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 6 Jul 2025 17:29:29 -0700
Subject: [PATCH 0790/1701] Support installing user extensions in
 user_data/extensions/

---
 modules/extensions.py                         | 63 ++++++++++++-------
 modules/utils.py                              | 14 ++++-
 .../extensions/place-your-extensions-here.txt |  0
 3 files changed, 51 insertions(+), 26 deletions(-)
 create mode 100644 user_data/extensions/place-your-extensions-here.txt

diff --git a/modules/extensions.py b/modules/extensions.py
index 6729b996f4..be9bc38c7a 100644
--- a/modules/extensions.py
+++ b/modules/extensions.py
@@ -2,10 +2,10 @@
 import traceback
 from functools import partial
 from inspect import signature
+from pathlib import Path
 
 import gradio as gr
 
-import extensions
 import modules.shared as shared
 from modules.logging_colors import logger
 
@@ -28,36 +28,51 @@ def apply_settings(extension, name):
 def load_extensions():
     global state, setup_called
     state = {}
+
     for i, name in enumerate(shared.args.extensions):
-        if name in available_extensions:
-            if name != 'api':
-                logger.info(f'Loading the extension "{name}"')
-            try:
-                try:
-                    extension = importlib.import_module(f"extensions.{name}.script")
-                except ModuleNotFoundError:
-                    logger.error(f"Could not import the requirements for '{name}'. Make sure to install the requirements for the extension.\n\n* To install requirements for all available extensions, launch the\n  update_wizard script for your OS and choose the B option.\n\n* To install the requirements for this extension alone, launch the\n  cmd script for your OS and paste the following command in the\n  terminal window that appears:\n\nLinux / Mac:\n\npip install -r extensions/{name}/requirements.txt --upgrade\n\nWindows:\n\npip install -r extensions\\{name}\\requirements.txt --upgrade\n")
-                    raise
-
-                # Only run setup() and apply settings from settings.yaml once
-                if extension not in setup_called:
-                    apply_settings(extension, name)
-                    if hasattr(extension, "setup"):
-                        extension.setup()
-
-                    setup_called.add(extension)
-
-                state[name] = [True, i]
-            except:
-                logger.error(f'Failed to load the extension "{name}".')
-                traceback.print_exc()
+        if name not in available_extensions:
+            continue
+
+        if name != 'api':
+            logger.info(f'Loading the extension "{name}"')
+
+        try:
+            # Prefer user extension, fall back to system extension
+            user_script_path = Path(f'user_data/extensions/{name}/script.py')
+            if user_script_path.exists():
+                extension = importlib.import_module(f"user_data.extensions.{name}.script")
+            else:
+                extension = importlib.import_module(f"extensions.{name}.script")
+
+            if extension not in setup_called:
+                apply_settings(extension, name)
+                if hasattr(extension, "setup"):
+                    extension.setup()
+                setup_called.add(extension)
+
+            state[name] = [True, i, extension]  # Store extension object
+
+        except ModuleNotFoundError:
+            extension_location = Path('user_data/extensions') / name if user_script_path.exists() else Path('extensions') / name
+            logger.error(
+                f"Could not import the requirements for '{name}'. Make sure to install the requirements for the extension.\n\n"
+                f"* To install requirements for all available extensions, launch the\n  update_wizard script for your OS and choose the B option.\n\n"
+                f"* To install the requirements for this extension alone, launch the\n  cmd script for your OS and paste the following command in the\n  terminal window that appears:\n\n"
+                f"Linux / Mac:\n\npip install -r {extension_location}/requirements.txt --upgrade\n\n"
+                f"Windows:\n\npip install -r {extension_location}\\requirements.txt --upgrade\n"
+            )
+            raise
+
+        except Exception:
+            logger.error(f'Failed to load the extension "{name}".')
+            traceback.print_exc()
 
 
 # This iterator returns the extensions in the order specified in the command-line
 def iterator():
     for name in sorted(state, key=lambda x: state[x][1]):
         if state[name][0]:
-            yield getattr(extensions, name).script, name
+            yield state[name][2], name  # Use stored extension object
 
 
 # Extension functions that map string -> string
diff --git a/modules/utils.py b/modules/utils.py
index c285d40161..117ad590a4 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -183,8 +183,18 @@ def get_available_instruction_templates():
 
 
 def get_available_extensions():
-    extensions = sorted(set(map(lambda x: x.parts[1], Path('extensions').glob('*/script.py'))), key=natural_keys)
-    return extensions
+    # User extensions (higher priority)
+    user_extensions = []
+    user_ext_path = Path('user_data/extensions')
+    if user_ext_path.exists():
+        user_exts = map(lambda x: x.parts[2], user_ext_path.glob('*/script.py'))
+        user_extensions = sorted(set(user_exts), key=natural_keys)
+
+    # System extensions (excluding those overridden by user extensions)
+    system_exts = map(lambda x: x.parts[1], Path('extensions').glob('*/script.py'))
+    system_extensions = sorted(set(system_exts) - set(user_extensions), key=natural_keys)
+
+    return user_extensions + system_extensions
 
 
 def get_available_loras():
diff --git a/user_data/extensions/place-your-extensions-here.txt b/user_data/extensions/place-your-extensions-here.txt
new file mode 100644
index 0000000000..e69de29bb2

From 60123a67ac051d1780acb13f0f40fc9fb77860e8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 6 Jul 2025 17:44:41 -0700
Subject: [PATCH 0791/1701] Better log message when extension requirements are
 not found

---
 modules/extensions.py | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/modules/extensions.py b/modules/extensions.py
index be9bc38c7a..e00103124e 100644
--- a/modules/extensions.py
+++ b/modules/extensions.py
@@ -54,12 +54,17 @@ def load_extensions():
 
         except ModuleNotFoundError:
             extension_location = Path('user_data/extensions') / name if user_script_path.exists() else Path('extensions') / name
+            windows_path = str(extension_location).replace('/', '\\')
             logger.error(
                 f"Could not import the requirements for '{name}'. Make sure to install the requirements for the extension.\n\n"
-                f"* To install requirements for all available extensions, launch the\n  update_wizard script for your OS and choose the B option.\n\n"
-                f"* To install the requirements for this extension alone, launch the\n  cmd script for your OS and paste the following command in the\n  terminal window that appears:\n\n"
-                f"Linux / Mac:\n\npip install -r {extension_location}/requirements.txt --upgrade\n\n"
-                f"Windows:\n\npip install -r {extension_location}\\requirements.txt --upgrade\n"
+                f"* To install requirements automatically, launch the update_wizard script for your OS and:\n\n"
+                f"1. Choose option B (Install/update extensions requirements)\n"
+                f"2. Select '{name}' from the extension list\n\n"
+                f"* To install requirements manually, launch the cmd script for your OS and paste the following command:\n\n"
+                f"Linux / Mac:\n\n"
+                f"pip install -r {extension_location}/requirements.txt --upgrade\n\n"
+                f"Windows:\n\n"
+                f"pip install -r {windows_path}\\requirements.txt --upgrade\n"
             )
             raise
 

From e5767d4fc5871875b5c555257f94ba707dc1ad55 Mon Sep 17 00:00:00 2001
From: Alidr79 
Date: Mon, 7 Jul 2025 04:18:53 +0330
Subject: [PATCH 0792/1701] Update ui_model_menu.py blocking the --multi-user
 access in backend (#7098)

---
 modules/ui_model_menu.py |  4 ++++
 modules/ui_session.py    | 14 ++++++++------
 2 files changed, 12 insertions(+), 6 deletions(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index a18d7d10d0..e09e292e4b 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -130,6 +130,10 @@ def create_ui():
 
 
 def create_event_handlers():
+    mu = shared.args.multi_user
+    if mu:
+        return  
+
     shared.gradio['loader'].change(loaders.make_loader_params_visible, gradio('loader'), gradio(loaders.get_all_params()), show_progress=False)
 
     # In this event handler, the interface state is read and updated
diff --git a/modules/ui_session.py b/modules/ui_session.py
index a69e155bdd..f8d573ad76 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -27,9 +27,10 @@ def create_ui():
                         shared.gradio['bool_menu'] = gr.CheckboxGroup(choices=get_boolean_arguments(), value=get_boolean_arguments(active=True), label="Boolean command-line flags", elem_classes='checkboxgroup-table')
 
         shared.gradio['theme_state'] = gr.Textbox(visible=False, value='dark' if shared.settings['dark_theme'] else 'light')
-        shared.gradio['save_settings'].click(
-            ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-            handle_save_settings, gradio('interface_state', 'preset_menu', 'extensions_menu', 'show_controls', 'theme_state'), gradio('save_contents', 'save_filename', 'save_root', 'file_saver'), show_progress=False)
+        if not mu:
+            shared.gradio['save_settings'].click(
+                ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
+                handle_save_settings, gradio('interface_state', 'preset_menu', 'extensions_menu', 'show_controls', 'theme_state'), gradio('save_contents', 'save_filename', 'save_root', 'file_saver'), show_progress=False)
 
         shared.gradio['toggle_dark_mode'].click(
             lambda x: 'dark' if x == 'light' else 'light', gradio('theme_state'), gradio('theme_state')).then(
@@ -42,9 +43,10 @@ def create_ui():
         )
 
         # Reset interface event
-        shared.gradio['reset_interface'].click(
-            set_interface_arguments, gradio('extensions_menu', 'bool_menu'), None).then(
-            None, None, None, js='() => {document.body.innerHTML=\'
            Reloading...
            \'; setTimeout(function(){location.reload()},2500); return []}')
+        if not mu:
+            shared.gradio['reset_interface'].click(
+                set_interface_arguments, gradio('extensions_menu', 'bool_menu'), None).then(
+                None, None, None, js='() => {document.body.innerHTML=\'
            Reloading...
            \'; setTimeout(function(){location.reload()},2500); return []}')
 
 
 def handle_save_settings(state, preset, extensions, show_controls, theme):

From cbef2720ce83c05c51159be374ccc3f7ba46f07c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 6 Jul 2025 20:14:02 -0700
Subject: [PATCH 0793/1701] Revert "Fix: use embedded Python in
 start_windows.bat to avoid system interpreter conflicts (#7120)"

This reverts commit 8df1127ce291dc9f4c7f57da1faa879aa4f87ae7.
---
 start_windows.bat | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/start_windows.bat b/start_windows.bat
index 08fedc6842..68c6194e76 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -99,7 +99,7 @@ set "CUDA_HOME=%CUDA_PATH%"
 call "%CONDA_ROOT_PREFIX%\condabin\conda.bat" activate "%INSTALL_ENV_DIR%" || ( echo. && echo Miniforge hook not found. && goto end )
 
 @rem setup installer env
-call "%INSTALL_ENV_DIR%\python.exe" one_click.py %*
+call python one_click.py %*
 
 @rem below are functions for the script   next line skips these during normal execution
 goto end

From e52bc0acb2b2f24ced25326010629d5ed2ade3da Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 6 Jul 2025 20:28:35 -0700
Subject: [PATCH 0794/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_cuda128.txt             | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt      | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 9735fcee4c..34fd13bee8 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index c7db011611..9b18524a27 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -32,7 +32,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 10c8933219..6b9cc2a9bb 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -32,7 +32,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index e900f1bcc7..b7830355b6 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -32,7 +32,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index a853138940..e48d4dee6e 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -32,8 +32,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 3735a1a850..831ff51fe0 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -32,5 +32,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index f93a4d77b0..fb81f40b30 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -32,5 +32,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index b5bee1e43e..3d29c93fca 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 2cd912052a..7528aed0ef 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 35b683c878..e86e452492 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index d928ff08ce..e9fa61f163 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index b25298fc3e..4f9b0668cd 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index e52a7248c7..9f403a0bd6 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -18,6 +18,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 7c2578d235..8e9baa3245 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index c125c97e7e..ebd6359131 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 6daa06fc74..6802a3372b 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 247d3a9cc7..624c222db5 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 466a1b6e75..edb82a319a 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.23.0/llama_cpp_binaries-0.23.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 426e7a4cec7427927cb5c0da11b1330d26ab7c5e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 7 Jul 2025 08:43:01 -0700
Subject: [PATCH 0795/1701] Update the extensions documentation

---
 docs/07 - Extensions.md | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/docs/07 - Extensions.md b/docs/07 - Extensions.md
index 78497888e6..ebcd3c0ed9 100644
--- a/docs/07 - Extensions.md	
+++ b/docs/07 - Extensions.md	
@@ -1,8 +1,14 @@
 # Extensions
 
-Extensions are defined by files named `script.py` inside subfolders of `text-generation-webui/extensions`. They are loaded at startup if the folder name is specified after the `--extensions` flag.
+Extensions are defined by files named `script.py` inside subfolders of either:
+- `text-generation-webui/extensions`
+- `text-generation-webui/user_data/extensions`
 
-For instance, `extensions/silero_tts/script.py` gets loaded with `python server.py --extensions silero_tts`.
+They are loaded at startup if the folder name is specified after the `--extensions` flag.
+
+For instance, `extensions/silero_tts/script.py` or `user_data/extensions/silero_tts/script.py` gets loaded with `python server.py --extensions silero_tts`.
+
+**Note:** Extensions in `user_data/extensions/` take priority over those in `extensions/` when both exist with the same name.
 
 ## [text-generation-webui-extensions](https://github.com/oobabooga/text-generation-webui-extensions)
 

From 07e6f004c59e457ed5ccbd963a97214d73b1ed4b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 7 Jul 2025 11:28:47 -0700
Subject: [PATCH 0796/1701] Rename a button in the Session tab for clarity

---
 modules/ui_session.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_session.py b/modules/ui_session.py
index f8d573ad76..60b19f476c 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -17,7 +17,7 @@ def create_ui():
 
             with gr.Column():
                 gr.Markdown("## Extensions & flags")
-                shared.gradio['save_settings'] = gr.Button('Save settings to user_data/settings.yaml', elem_classes='refresh-button', interactive=not mu)
+                shared.gradio['save_settings'] = gr.Button('Save extensions settings to user_data/settings.yaml', elem_classes='refresh-button', interactive=not mu)
                 shared.gradio['reset_interface'] = gr.Button("Apply flags/extensions and restart", interactive=not mu)
                 with gr.Row():
                     with gr.Column():

From ca226a54c6d8c3f3cbb27159b92eb82860517a08 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 7 Jul 2025 11:29:37 -0700
Subject: [PATCH 0797/1701] Disable the message version navigation hover
 effects during streaming

---
 css/main.css | 14 ++++++--------
 1 file changed, 6 insertions(+), 8 deletions(-)

diff --git a/css/main.css b/css/main.css
index 036a2bb73e..240a94d5e4 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1291,18 +1291,16 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     opacity: 1;
 }
 
-/* Disable message action hover effects during generation */
-._generating .message:hover .message-actions,
-._generating .user-message:hover .message-actions,
-._generating .assistant-message:hover .message-actions {
+/* Disable message actions and version navigation hover effects during generation */
+._generating :is(.message, .user-message, .assistant-message):hover :is(.message-actions, .version-navigation) {
     opacity: 0 !important;
+    pointer-events: none;
 }
 
-/* Disable message action hover effects during scrolling */
-.scrolling .message:hover .message-actions,
-.scrolling .user-message:hover .message-actions,
-.scrolling .assistant-message:hover .message-actions {
+/* Disable message actions and version navigation hover effects during scrolling */
+.scrolling :is(.message, .user-message, .assistant-message):hover :is(.message-actions, .version-navigation) {
     opacity: 0 !important;
+    pointer-events: none;
 }
 
 .footer-button svg {

From 74d98186fc289087025759d16b7a8ecf5cf770da Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 7 Jul 2025 13:23:23 -0700
Subject: [PATCH 0798/1701] Slightly more robust autoscroll

---
 js/main.js | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/js/main.js b/js/main.js
index 17b853f4ec..e5b14d8ee6 100644
--- a/js/main.js
+++ b/js/main.js
@@ -187,8 +187,11 @@ const observer = new MutationObserver(function(mutations) {
 
   doSyntaxHighlighting();
 
-  if (!window.isScrolled && !isScrollingClassOnly && targetElement.scrollTop !== targetElement.scrollHeight) {
-    targetElement.scrollTop = targetElement.scrollHeight;
+  if (!window.isScrolled && !isScrollingClassOnly) {
+    const maxScroll = targetElement.scrollHeight - targetElement.clientHeight;
+    if (maxScroll > 0 && targetElement.scrollTop < maxScroll - 1) {
+      targetElement.scrollTop = maxScroll;
+    }
   }
 
   const chatElement = document.getElementById("chat");

From e8266b0356e577ab372aed0f580bd4da05ef3d19 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 7 Jul 2025 14:19:20 -0700
Subject: [PATCH 0799/1701] Use windows-2022 in workflows

---
 .github/workflows/build-everything-tgw.yml          | 6 +++---
 .github/workflows/build-portable-release-cuda.yml   | 2 +-
 .github/workflows/build-portable-release-vulkan.yml | 2 +-
 .github/workflows/build-portable-release.yml        | 2 +-
 4 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/.github/workflows/build-everything-tgw.yml b/.github/workflows/build-everything-tgw.yml
index b3a1e8dd49..23773010e9 100644
--- a/.github/workflows/build-everything-tgw.yml
+++ b/.github/workflows/build-everything-tgw.yml
@@ -18,7 +18,7 @@ jobs:
     uses: ./.github/workflows/build-portable-release-cuda.yml
     with:
       version: ${{ inputs.version }}
-      config: 'os:windows-2019'
+      config: 'os:windows-2022'
 
   build_release_cuda_linux:
     name: CUDA Linux
@@ -32,7 +32,7 @@ jobs:
     uses: ./.github/workflows/build-portable-release-vulkan.yml
     with:
       version: ${{ inputs.version }}
-      config: 'os:windows-2019'
+      config: 'os:windows-2022'
 
   build_release_vulkan_linux:
     name: Vulkan Linux
@@ -46,7 +46,7 @@ jobs:
     uses: ./.github/workflows/build-portable-release.yml
     with:
       version: ${{ inputs.version }}
-      config: 'os:windows-2019'
+      config: 'os:windows-2022'
 
   build_release_cpu_linux:
     name: CPU Linux
diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index 729ce695d7..82ca65362e 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -57,7 +57,7 @@ jobs:
         id: set-matrix
         run: |
           $matrix = @{
-              'os' = @('ubuntu-22.04', 'windows-2019')
+              'os' = @('ubuntu-22.04', 'windows-2022')
               'pyver' = @("3.11")
               'avx' = @("AVX2")
               'cuda' = @("11.7", "12.4")
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 98eb363522..62e8b66945 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -57,7 +57,7 @@ jobs:
         id: set-matrix
         run: |
           $matrix = @{
-              'os' = @('ubuntu-22.04', 'windows-2019')
+              'os' = @('ubuntu-22.04', 'windows-2022')
               'pyver' = @("3.11")
               'avx' = @("AVX2")
           }
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 8863d67367..3d6734afda 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -57,7 +57,7 @@ jobs:
         id: set-matrix
         run: |
           $matrix = @{
-              'os' = @('ubuntu-22.04', 'windows-2019', 'macos-13', 'macos-14')
+              'os' = @('ubuntu-22.04', 'windows-2022', 'macos-13', 'macos-14')
               'pyver' = @("3.11")
               'avx' = @("AVX2")
           }

From 86cb5e058711633bd4fa45faabcac699020386e5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 7 Jul 2025 21:02:19 -0700
Subject: [PATCH 0800/1701] Standardize margins and paddings across all chat
 styles

---
 css/chat_style-Dark.css            | 28 ++++++++++++++++++++++++++-
 css/chat_style-TheEncrypted777.css | 28 ++++++++++++++++++++++++++-
 css/chat_style-cai-chat.css        | 31 ++++++++++++++++++++++++++----
 css/chat_style-messenger.css       | 28 ++++++++++++++++++++++++++-
 css/chat_style-wpp.css             | 27 ++++++++++++++++++++++++--
 css/html_readable_style.css        | 28 ++++++++++++++++++++++++---
 6 files changed, 158 insertions(+), 12 deletions(-)

diff --git a/css/chat_style-Dark.css b/css/chat_style-Dark.css
index 6a4784ccd1..aae82f8b92 100644
--- a/css/chat_style-Dark.css
+++ b/css/chat_style-Dark.css
@@ -91,7 +91,6 @@
 }
 
 .message-body p {
-    margin-bottom: 0 !important;
     font-size: 16px !important;
     line-height: 1.5 !important;
     color: #e0e0e0 !important; /* Light color for text */
@@ -129,3 +128,30 @@
         font-size: 18px; /* Smaller username for mobile */
     }
 }
+
+/* Standard spacing */
+.chat .message-body :is(p, ul, ol) {
+    margin: 1.25em 0 !important;
+}
+
+.chat .message-body :is(p, ul, ol):first-child {
+    margin-top: 0 !important;
+}
+
+.chat .message-body :is(p, ul, ol):last-child {
+    margin-bottom: 0 !important;
+}
+
+.chat .message-body ul, .chat .message-body ol {
+    padding-inline-start: 2em;
+}
+
+.message-body li {
+    list-style-position: outside;
+    margin-top: 0.5em !important;
+    margin-bottom: 0.5em !important;
+}
+
+.message-body li > p {
+    display: inline !important;
+}
diff --git a/css/chat_style-TheEncrypted777.css b/css/chat_style-TheEncrypted777.css
index fbd47072fb..7f37377d82 100644
--- a/css/chat_style-TheEncrypted777.css
+++ b/css/chat_style-TheEncrypted777.css
@@ -87,7 +87,6 @@
 }
 
 .message-body p {
-    margin-bottom: 0 !important;
     font-size: 18px !important;
     line-height: 1.428571429 !important;
     color: rgb(243 244 246) !important;
@@ -135,3 +134,30 @@
         font-size: 20px;
     }
 }
+
+/* Standard spacing */
+.chat .message-body :is(p, ul, ol) {
+    margin: 1.25em 0 !important;
+}
+
+.chat .message-body :is(p, ul, ol):first-child {
+    margin-top: 0 !important;
+}
+
+.chat .message-body :is(p, ul, ol):last-child {
+    margin-bottom: 0 !important;
+}
+
+.chat .message-body ul, .chat .message-body ol {
+    padding-inline-start: 2em;
+}
+
+.message-body li {
+    list-style-position: outside;
+    margin-top: 0.5em !important;
+    margin-bottom: 0.5em !important;
+}
+
+.message-body li > p {
+    display: inline !important;
+}
diff --git a/css/chat_style-cai-chat.css b/css/chat_style-cai-chat.css
index b06b1269c7..d2e8f44aad 100644
--- a/css/chat_style-cai-chat.css
+++ b/css/chat_style-cai-chat.css
@@ -52,10 +52,6 @@
     font-weight: 500;
 }
 
-.message-body p, .chat .message-body ul, .chat .message-body ol {
-    margin-bottom: 10px !important;
-}
-
 .dark .message-body p em {
     color: rgb(138 138 138) !important;
 }
@@ -64,3 +60,30 @@
     color: rgb(110 110 110) !important;
     font-weight: 500;
 }
+
+/* Standard spacing */
+.chat .message-body :is(p, ul, ol) {
+    margin: 1.25em 0 !important;
+}
+
+.chat .message-body :is(p, ul, ol):first-child {
+    margin-top: 0 !important;
+}
+
+.chat .message-body :is(p, ul, ol):last-child {
+    margin-bottom: 0 !important;
+}
+
+.chat .message-body ul, .chat .message-body ol {
+    padding-inline-start: 2em;
+}
+
+.message-body li {
+    list-style-position: outside;
+    margin-top: 0.5em !important;
+    margin-bottom: 0.5em !important;
+}
+
+.message-body li > p {
+    display: inline !important;
+}
diff --git a/css/chat_style-messenger.css b/css/chat_style-messenger.css
index 65af5f7a5f..a6189a4aa6 100644
--- a/css/chat_style-messenger.css
+++ b/css/chat_style-messenger.css
@@ -86,7 +86,6 @@
 }
 
 .message-body p {
-    margin-bottom: 0 !important;
     font-size: 15px !important;
     line-height: 1.428571429 !important;
     font-weight: 500;
@@ -99,3 +98,30 @@
 .message-body p em {
     color: rgb(110 110 110) !important;
 }
+
+/* Standard spacing */
+.chat .message-body :is(p, ul, ol) {
+    margin: 1.25em 0 !important;
+}
+
+.chat .message-body :is(p, ul, ol):first-child {
+    margin-top: 0 !important;
+}
+
+.chat .message-body :is(p, ul, ol):last-child {
+    margin-bottom: 0 !important;
+}
+
+.chat .message-body ul, .chat .message-body ol {
+    padding-inline-start: 2em;
+}
+
+.message-body li {
+    list-style-position: outside;
+    margin-top: 0.5em !important;
+    margin-bottom: 0.5em !important;
+}
+
+.message-body li > p {
+    display: inline !important;
+}
diff --git a/css/chat_style-wpp.css b/css/chat_style-wpp.css
index b2ac4d2a38..c86e577da5 100644
--- a/css/chat_style-wpp.css
+++ b/css/chat_style-wpp.css
@@ -100,6 +100,29 @@
     margin-top: 8px;
 }
 
-.message-body p, .chat .message-body ul, .chat .message-body ol {
-    margin-bottom: 10px !important;
+/* Standard spacing */
+.chat .message-body :is(p, ul, ol) {
+    margin: 1.25em 0 !important;
+}
+
+.chat .message-body :is(p, ul, ol):first-child {
+    margin-top: 0 !important;
+}
+
+.chat .message-body :is(p, ul, ol):last-child {
+    margin-bottom: 0 !important;
+}
+
+.chat .message-body ul, .chat .message-body ol {
+    padding-inline-start: 2em;
+}
+
+.message-body li {
+    list-style-position: outside;
+    margin-top: 0.5em !important;
+    margin-bottom: 0.5em !important;
+}
+
+.message-body li > p {
+    display: inline !important;
 }
diff --git a/css/html_readable_style.css b/css/html_readable_style.css
index 760649d600..74a913e93e 100644
--- a/css/html_readable_style.css
+++ b/css/html_readable_style.css
@@ -11,9 +11,8 @@
 
 .readable-container p, .readable-container li {
     font-size: 16px !important;
-    color: #efefef !important;
-    margin-bottom: 22px;
     line-height: 1.4 !important;
+    color: #efefef !important;
 }
 
 .readable-container li > p {
@@ -30,4 +29,27 @@
 
 .readable-container .hoverable {
     font-size: 14px;
-}
\ No newline at end of file
+}
+
+/* Standard spacing */
+.readable-container :is(p, ul, ol) {
+    margin: 1.25em 0 !important;
+}
+
+.readable-container :is(p, ul, ol):first-child {
+    margin-top: 0 !important;
+}
+
+.readable-container :is(p, ul, ol):last-child {
+    margin-bottom: 0 !important;
+}
+
+.readable-container ul, .readable-container ol {
+    padding-inline-start: 2em;
+}
+
+.readable-container li {
+    list-style-position: outside;
+    margin-top: 0.5em !important;
+    margin-bottom: 0.5em !important;
+}

From c6c1b725e95594f03891cdabe0be51ca8cb33c0a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 7 Jul 2025 21:11:13 -0700
Subject: [PATCH 0801/1701] CSS simplifications

---
 css/chat_style-Dark.css            | 20 +-------------------
 css/chat_style-TheEncrypted777.css | 20 +-------------------
 css/chat_style-cai-chat.css        | 21 +--------------------
 css/chat_style-messenger.css       | 23 +----------------------
 css/chat_style-wpp.css             | 20 +-------------------
 css/html_readable_style.css        | 12 +-----------
 6 files changed, 6 insertions(+), 110 deletions(-)

diff --git a/css/chat_style-Dark.css b/css/chat_style-Dark.css
index aae82f8b92..11603ff55f 100644
--- a/css/chat_style-Dark.css
+++ b/css/chat_style-Dark.css
@@ -35,10 +35,6 @@
     color: #f0f0f0; /* Light text color for readability */
 }
 
-.text p {
-    margin-top: 2px;
-}
-
 .username {
     padding-left: 10px;
     font-size: 20px;
@@ -129,7 +125,7 @@
     }
 }
 
-/* Standard spacing */
+/* Standard spacing from instruct style */
 .chat .message-body :is(p, ul, ol) {
     margin: 1.25em 0 !important;
 }
@@ -141,17 +137,3 @@
 .chat .message-body :is(p, ul, ol):last-child {
     margin-bottom: 0 !important;
 }
-
-.chat .message-body ul, .chat .message-body ol {
-    padding-inline-start: 2em;
-}
-
-.message-body li {
-    list-style-position: outside;
-    margin-top: 0.5em !important;
-    margin-bottom: 0.5em !important;
-}
-
-.message-body li > p {
-    display: inline !important;
-}
diff --git a/css/chat_style-TheEncrypted777.css b/css/chat_style-TheEncrypted777.css
index 7f37377d82..e1f378397a 100644
--- a/css/chat_style-TheEncrypted777.css
+++ b/css/chat_style-TheEncrypted777.css
@@ -38,10 +38,6 @@
     text-shadow: 2px 2px 2px rgb(0 0 0 / 40%);
 }
 
-.text p {
-    margin-top: 2px;
-}
-
 .username {
     padding-left: 10px;
     font-size: 22px;
@@ -135,7 +131,7 @@
     }
 }
 
-/* Standard spacing */
+/* Standard spacing from instruct style */
 .chat .message-body :is(p, ul, ol) {
     margin: 1.25em 0 !important;
 }
@@ -147,17 +143,3 @@
 .chat .message-body :is(p, ul, ol):last-child {
     margin-bottom: 0 !important;
 }
-
-.chat .message-body ul, .chat .message-body ol {
-    padding-inline-start: 2em;
-}
-
-.message-body li {
-    list-style-position: outside;
-    margin-top: 0.5em !important;
-    margin-bottom: 0.5em !important;
-}
-
-.message-body li > p {
-    display: inline !important;
-}
diff --git a/css/chat_style-cai-chat.css b/css/chat_style-cai-chat.css
index d2e8f44aad..ef1f1f6231 100644
--- a/css/chat_style-cai-chat.css
+++ b/css/chat_style-cai-chat.css
@@ -9,11 +9,6 @@
     line-height: 22.5px !important;
 }
 
-.message-body {
-    margin-top: 3px;
-    font-size: 15px !important;
-}
-
 .circle-you {
     width: 50px;
     height: 50px;
@@ -61,7 +56,7 @@
     font-weight: 500;
 }
 
-/* Standard spacing */
+/* Standard spacing from instruct style */
 .chat .message-body :is(p, ul, ol) {
     margin: 1.25em 0 !important;
 }
@@ -73,17 +68,3 @@
 .chat .message-body :is(p, ul, ol):last-child {
     margin-bottom: 0 !important;
 }
-
-.chat .message-body ul, .chat .message-body ol {
-    padding-inline-start: 2em;
-}
-
-.message-body li {
-    list-style-position: outside;
-    margin-top: 0.5em !important;
-    margin-bottom: 0.5em !important;
-}
-
-.message-body li > p {
-    display: inline !important;
-}
diff --git a/css/chat_style-messenger.css b/css/chat_style-messenger.css
index a6189a4aa6..b6da2bb2bb 100644
--- a/css/chat_style-messenger.css
+++ b/css/chat_style-messenger.css
@@ -68,17 +68,10 @@
     max-width: 80%;
 }
 
-.text p {
-    margin-top: 5px;
-}
-
 .username {
     font-weight: bold;
 }
 
-.message-body {
-}
-
 .message-body img {
     max-width: 300px;
     max-height: 300px;
@@ -99,7 +92,7 @@
     color: rgb(110 110 110) !important;
 }
 
-/* Standard spacing */
+/* Standard spacing from instruct style */
 .chat .message-body :is(p, ul, ol) {
     margin: 1.25em 0 !important;
 }
@@ -111,17 +104,3 @@
 .chat .message-body :is(p, ul, ol):last-child {
     margin-bottom: 0 !important;
 }
-
-.chat .message-body ul, .chat .message-body ol {
-    padding-inline-start: 2em;
-}
-
-.message-body li {
-    list-style-position: outside;
-    margin-top: 0.5em !important;
-    margin-bottom: 0.5em !important;
-}
-
-.message-body li > p {
-    display: inline !important;
-}
diff --git a/css/chat_style-wpp.css b/css/chat_style-wpp.css
index c86e577da5..d11172ef56 100644
--- a/css/chat_style-wpp.css
+++ b/css/chat_style-wpp.css
@@ -83,10 +83,6 @@
     font-weight: 400;
 }
 
-.message-body p:first-child {
-    margin-top: 0 !important;
-}
-
 .dark .message-body p em {
     color: rgb(170 170 170) !important;
 }
@@ -100,7 +96,7 @@
     margin-top: 8px;
 }
 
-/* Standard spacing */
+/* Standard spacing from instruct style */
 .chat .message-body :is(p, ul, ol) {
     margin: 1.25em 0 !important;
 }
@@ -112,17 +108,3 @@
 .chat .message-body :is(p, ul, ol):last-child {
     margin-bottom: 0 !important;
 }
-
-.chat .message-body ul, .chat .message-body ol {
-    padding-inline-start: 2em;
-}
-
-.message-body li {
-    list-style-position: outside;
-    margin-top: 0.5em !important;
-    margin-bottom: 0.5em !important;
-}
-
-.message-body li > p {
-    display: inline !important;
-}
diff --git a/css/html_readable_style.css b/css/html_readable_style.css
index 74a913e93e..74f69b73ac 100644
--- a/css/html_readable_style.css
+++ b/css/html_readable_style.css
@@ -31,7 +31,7 @@
     font-size: 14px;
 }
 
-/* Standard spacing */
+/* Standard spacing from instruct style */
 .readable-container :is(p, ul, ol) {
     margin: 1.25em 0 !important;
 }
@@ -43,13 +43,3 @@
 .readable-container :is(p, ul, ol):last-child {
     margin-bottom: 0 !important;
 }
-
-.readable-container ul, .readable-container ol {
-    padding-inline-start: 2em;
-}
-
-.readable-container li {
-    list-style-position: outside;
-    margin-top: 0.5em !important;
-    margin-bottom: 0.5em !important;
-}

From 42191a36abe90b09b984bafbc552c62b951e7eef Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 8 Jul 2025 07:08:23 -0700
Subject: [PATCH 0802/1701] Keep navigation icons visible when switching
 versions

---
 js/main.js | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/js/main.js b/js/main.js
index e5b14d8ee6..138569c04d 100644
--- a/js/main.js
+++ b/js/main.js
@@ -147,11 +147,15 @@ window.isScrolled = false;
 let scrollTimeout;
 
 targetElement.addEventListener("scroll", function() {
+  let diff = targetElement.scrollHeight - targetElement.clientHeight;
+  let isAtBottomNow = Math.abs(targetElement.scrollTop - diff) <= 10 || diff == 0;
+
   // Add scrolling class to disable hover effects
-  targetElement.classList.add("scrolling");
+  if (window.isScrolled || !isAtBottomNow) {
+    targetElement.classList.add("scrolling");
+  }
 
-  let diff = targetElement.scrollHeight - targetElement.clientHeight;
-  if(Math.abs(targetElement.scrollTop - diff) <= 10 || diff == 0) {
+  if(isAtBottomNow) {
     window.isScrolled = false;
   } else {
     window.isScrolled = true;
@@ -163,7 +167,6 @@ targetElement.addEventListener("scroll", function() {
     targetElement.classList.remove("scrolling");
     doSyntaxHighlighting(); // Only run after scrolling stops
   }, 150);
-
 });
 
 // Create a MutationObserver instance

From 93e08c0d4a435734ada5d9f948371941d1a234be Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 8 Jul 2025 08:59:29 -0700
Subject: [PATCH 0803/1701] Update README

---
 README.md | 1 +
 1 file changed, 1 insertion(+)

diff --git a/README.md b/README.md
index 466e4edb29..c80dacde15 100644
--- a/README.md
+++ b/README.md
@@ -18,6 +18,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 - **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
 - **Web search**: Optionally search the internet with LLM-generated queries to add context to the conversation.
 - Aesthetic UI with dark and light themes.
+- Syntax highlighting for code blocks and LaTeX rendering for mathematical expressions.
 - `instruct` mode for instruction-following (like ChatGPT), and `chat-instruct`/`chat` modes for talking to custom characters.
 - Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
 - Edit messages, navigate between message versions, and branch conversations at any point.

From 2f544fe199be17cc60016abbd64d248e0c64c354 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 8 Jul 2025 09:02:42 -0700
Subject: [PATCH 0804/1701] Update the keyboard shortcuts documentation

---
 docs/13 - Keyboard Shortcuts.md | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/docs/13 - Keyboard Shortcuts.md b/docs/13 - Keyboard Shortcuts.md
index b48c7da7f4..3ba484234c 100644
--- a/docs/13 - Keyboard Shortcuts.md	
+++ b/docs/13 - Keyboard Shortcuts.md	
@@ -5,7 +5,6 @@
 | Shortcut                | Description                                      |
 |-------------------------|--------------------------------------------------|
 | Esc                     | Stop generation                                  |
-| Tab                     | Switch between current tab and Parameters tab    |
 
 #### Chat tab
 
@@ -15,6 +14,6 @@
 | Ctrl + Enter            | Regenerate                                       |
 | Alt + Enter             | Continue                                         |
 | Ctrl + Shift + Backspace| Remove last                                      |
-| Ctrl + Shift + K        | Copy last                                        |
-| Ctrl + Shift + L        | Replace last                                     |
 | Ctrl + Shift + M        | Impersonate                                      |
+| ← (Left Arrow)          | Navigate to previous version of last assistant message |
+| → (Right Arrow)         | Navigate to next version of last assistant message (or regenerate if at latest version) |

From 3e24a127c7b0193ef5a211750f1d5ffcd6a15de8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 8 Jul 2025 09:13:11 -0700
Subject: [PATCH 0805/1701] Remove more unnecessary files from portable builds

---
 .github/workflows/build-portable-release-cuda.yml   | 3 ++-
 .github/workflows/build-portable-release-vulkan.yml | 5 ++++-
 .github/workflows/build-portable-release.yml        | 7 +++++--
 3 files changed, 11 insertions(+), 4 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index 82ca65362e..2d2685dd7a 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -107,7 +107,7 @@ jobs:
             cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
             cd "text-generation-webui-${VERSION_CLEAN}"
 
-            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore
+            # Remove extensions that need additional requirements
             allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
@@ -163,6 +163,7 @@ jobs:
             if [[ "$CUDA_VERSION" == "11.7" ]]; then
                 rm requirements_cuda_temp.txt
             fi
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore requirements/ one_click.py
 
             # 6. Create ZIP file
             cd ..
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 62e8b66945..d275460b76 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -106,7 +106,7 @@ jobs:
             cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
             cd "text-generation-webui-${VERSION_CLEAN}"
 
-            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore
+            # Remove extensions that need additional requirements
             allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
@@ -150,6 +150,9 @@ jobs:
             echo "Installing Python packages from $REQ_FILE..."
             $PIP_PATH install --target="./$PACKAGES_PATH" -r "$REQ_FILE"
 
+            # 5. Clean up
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore requirements/ one_click.py
+
             # 6. Create ZIP file
             cd ..
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.zip"
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 3d6734afda..51567ec8bd 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -106,7 +106,7 @@ jobs:
             cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
             cd "text-generation-webui-${VERSION_CLEAN}"
 
-            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore
+            # Remove extensions that need additional requirements
             allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
@@ -176,7 +176,10 @@ jobs:
             echo "Installing Python packages from $REQ_FILE..."
             $PIP_PATH install --target="./$PACKAGES_PATH" -r "$REQ_FILE"
 
-            # 5. Create ZIP file
+            # 5. Clean up
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore requirements/ one_click.py
+
+            # 6. Create ZIP file
             cd ..
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.zip"
             echo "Creating archive: $ZIP_NAME"

From cd5d867b62fac8a20ada2c0f879397436f5d148f Mon Sep 17 00:00:00 2001
From: Cats <42129397+Cats1337@users.noreply.github.com>
Date: Tue, 8 Jul 2025 13:54:38 -0700
Subject: [PATCH 0806/1701] docs: Add Mirostat Explanation (#7128)

---
 docs/03 - Parameters Tab.md | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/docs/03 - Parameters Tab.md b/docs/03 - Parameters Tab.md
index 06eebe501b..8c758ce678 100644
--- a/docs/03 - Parameters Tab.md	
+++ b/docs/03 - Parameters Tab.md	
@@ -51,10 +51,12 @@ For more information about the parameters, the [transformers documentation](http
 * **guidance_scale**: The main parameter for Classifier-Free Guidance (CFG). [The paper](https://arxiv.org/pdf/2306.17806.pdf) suggests that 1.5 is a good value. It can be used in conjunction with a negative prompt or not.
 * **Negative prompt**: Only used when `guidance_scale != 1`. It is most useful for instruct models and custom system messages. You place your full prompt in this field with the system message replaced with the default one for the model (like "You are Llama, a helpful assistant...") to make the model pay more attention to your custom system message.
 * **penalty_alpha**: Contrastive Search is enabled by setting this to greater than zero and unchecking "do_sample". It should be used with a low value of top_k, for instance, top_k = 4.
-* **mirostat_mode**: Activates the Mirostat sampling technique. It aims to control perplexity during sampling. See the [paper](https://arxiv.org/abs/2007.14966).
-* **mirostat_tau**: No idea, see the paper for details. According to the Preset Arena, 8 is a good value. 
-* **mirostat_eta**: No idea, see the paper for details. According to the Preset Arena, 0.1 is a good value.
-* **dynamic_temperature**: Activates Dynamic Temperature. This modifies temperature to range between "dynatemp_low" (minimum) and "dynatemp_high" (maximum), with an entropy-based scaling. The steepness of the curve is controlled by "dynatemp_exponent".
+* **mirostat_mode**: Activates Mirostat sampling, an adaptive decoding method that dynamically controls output perplexity for higher-quality text generation. 0 is disabled. 1 is the classic Mirostat algorithm described in [the paper](https://arxiv.org/abs/2007.14966), but can be less stable, or “wobbly,” and produce less coherent text. 2 is the improved version that is more stable and has lower perplexity, recommended for most use cases. 
+*Note: Use either mirostat or dynamic_temperature, not both at the same time.*
+* **mirostat_tau**: Target perplexity for Mirostat sampling. Controls how “surprising” the text is. Higher values = more diverse, lower = more predictable. Preset Arena suggests 8 as a good value.
+* **mirostat_eta**: Learning rate for Mirostat’s perplexity adjustment. Higher values = adapts faster but less stable, lower values = slower but more stable. Preset Arena suggests 0.1 as a good value.
+* **dynamic_temperature**: Activates Dynamic Temperature. This modifies temperature to range between "dynatemp_low" (minimum) and "dynatemp_high" (maximum), with an entropy-based scaling. The steepness of the curve is controlled by "dynatemp_exponent". 
+*Note: Use either dynamic_temperature or mirostat, not both at the same time.*
 * **smoothing_factor**: Activates Quadratic Sampling. When `0 < smoothing_factor < 1`, the logits distribution becomes flatter. When `smoothing_factor > 1`, it becomes more peaked.
 * **temperature_last**: Makes temperature the last sampler instead of the first. With this, you can remove low probability tokens with a sampler like min_p and then use a high temperature to make the model creative without losing coherency. Note: this parameter takes precedence over "Sampler priority". That means that `temperature`/`dynamic_temperature`/`quadratic_sampling` will be removed from wherever they are and moved to the end of the stack.
 * **do_sample**: When unchecked, sampling is entirely disabled, and greedy decoding is used instead (the most likely token is always picked).

From d1e9301a43795abb56a46c63f6a5108ccbb80588 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 8 Jul 2025 19:57:25 -0700
Subject: [PATCH 0807/1701] Remove fragile js from
 9a58964834833e37fc830f06f4d4a87a142d4179

---
 js/main.js | 39 ---------------------------------------
 1 file changed, 39 deletions(-)

diff --git a/js/main.js b/js/main.js
index 138569c04d..e0f9314dad 100644
--- a/js/main.js
+++ b/js/main.js
@@ -1052,45 +1052,6 @@ new MutationObserver(() => addMiniDeletes()).observe(
 );
 addMiniDeletes();
 
-//------------------------------------------------
-// Maintain distance from bottom when input height changes
-//------------------------------------------------
-let wasAtBottom = false;
-let preservedDistance = 0;
-
-function checkIfAtBottom() {
-  const distanceFromBottom = targetElement.scrollHeight - targetElement.scrollTop - targetElement.clientHeight;
-  wasAtBottom = distanceFromBottom <= 1; // Allow for rounding errors
-}
-
-function preserveScrollPosition() {
-  preservedDistance = targetElement.scrollHeight - targetElement.scrollTop - targetElement.clientHeight;
-}
-
-function restoreScrollPosition() {
-  if (wasAtBottom) {
-    // Force to bottom
-    targetElement.scrollTop = targetElement.scrollHeight - targetElement.clientHeight;
-  } else {
-    // Restore original distance
-    targetElement.scrollTop = targetElement.scrollHeight - targetElement.clientHeight - preservedDistance;
-  }
-}
-
-// Check position before input
-chatInput.addEventListener("beforeinput", () => {
-  checkIfAtBottom();
-  preserveScrollPosition();
-});
-
-// Restore after input
-chatInput.addEventListener("input", () => {
-  requestAnimationFrame(() => restoreScrollPosition());
-});
-
-// Update wasAtBottom when user scrolls
-targetElement.addEventListener("scroll", checkIfAtBottom);
-
 //------------------------------------------------
 // Fix autoscroll after fonts load
 //------------------------------------------------

From 6c2bdda0f02f497ee91f8c13afab7de149823ff0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Jul 2025 18:38:45 -0700
Subject: [PATCH 0808/1701] Transformers loader: replace
 `use_flash_attention_2`/`use_eager_attention` with a unified
 `attn_implementation`

Closes #7107
---
 modules/loaders.py             | 3 +--
 modules/models_settings.py     | 6 ------
 modules/shared.py              | 3 +--
 modules/transformers_loader.py | 7 +------
 modules/ui.py                  | 3 +--
 modules/ui_model_menu.py       | 3 +--
 6 files changed, 5 insertions(+), 20 deletions(-)

diff --git a/modules/loaders.py b/modules/loaders.py
index 6fbd2198c3..2b2c6b784e 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -40,11 +40,10 @@
         'load_in_8bit',
         'load_in_4bit',
         'torch_compile',
-        'use_flash_attention_2',
+        'attn_implementation',
         'cpu',
         'disk',
         'use_double_quant',
-        'use_eager_attention',
         'bf16',
         'trust_remote_code',
         'no_use_fast',
diff --git a/modules/models_settings.py b/modules/models_settings.py
index bea5b4d64c..a06e594e90 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -15,7 +15,6 @@
 def get_fallback_settings():
     return {
         'bf16': False,
-        'use_eager_attention': False,
         'ctx_size': 2048,
         'rope_freq_base': 0,
         'compress_pos_emb': 1,
@@ -118,14 +117,9 @@ def get_model_metadata(model):
                 if metadata['rope_scaling']['type'] == 'linear':
                     model_settings['compress_pos_emb'] = metadata['rope_scaling']['factor']
 
-            # For Gemma-2
             if 'torch_dtype' in metadata and metadata['torch_dtype'] == 'bfloat16':
                 model_settings['bf16'] = True
 
-            # For Gemma-2
-            if 'architectures' in metadata and isinstance(metadata['architectures'], list) and 'Gemma2ForCausalLM' in metadata['architectures']:
-                model_settings['use_eager_attention'] = True
-
     # Try to find the Jinja instruct template
     path = Path(f'{shared.args.model_dir}/{model}') / 'tokenizer_config.json'
     if path.exists():
diff --git a/modules/shared.py b/modules/shared.py
index 5333ec4f4d..5764911491 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -61,8 +61,7 @@
 group.add_argument('--trust-remote-code', action='store_true', help='Set trust_remote_code=True while loading the model. Necessary for some models.')
 group.add_argument('--force-safetensors', action='store_true', help='Set use_safetensors=True while loading the model. This prevents arbitrary code execution.')
 group.add_argument('--no_use_fast', action='store_true', help='Set use_fast=False while loading the tokenizer (it\'s True by default). Use this if you have any problems related to use_fast.')
-group.add_argument('--use_flash_attention_2', action='store_true', help='Set use_flash_attention_2=True while loading the model.')
-group.add_argument('--use_eager_attention', action='store_true', help='Set attn_implementation= eager while loading the model.')
+group.add_argument('--attn-implementation', type=str, default='sdpa', help='Attention implementation. Valid options: sdpa, eager, flash_attention_2.')
 group.add_argument('--torch-compile', action='store_true', help='Compile the model with torch.compile for improved performance.')
 
 # bitsandbytes 4-bit
diff --git a/modules/transformers_loader.py b/modules/transformers_loader.py
index 905f5c4728..ef524b5784 100644
--- a/modules/transformers_loader.py
+++ b/modules/transformers_loader.py
@@ -135,20 +135,15 @@ def load_model_HF(model_name):
     params = {
         'low_cpu_mem_usage': True,
         'torch_dtype': torch.bfloat16 if shared.args.bf16 else torch.float16,
+        'attn_implementation': shared.args.attn_implementation,
     }
 
     if shared.args.trust_remote_code:
         params['trust_remote_code'] = True
 
-    if shared.args.use_flash_attention_2:
-        params['use_flash_attention_2'] = True
-
     if shared.args.force_safetensors:
         params['force_safetensors'] = True
 
-    if shared.args.use_eager_attention:
-        params['attn_implementation'] = 'eager'
-
     config = AutoConfig.from_pretrained(path_to_model, trust_remote_code=shared.args.trust_remote_code)
 
     if 'chatglm' in model_name.lower():
diff --git a/modules/ui.py b/modules/ui.py
index 0030bb0231..9d6777f0a1 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -144,7 +144,7 @@ def list_model_elements():
         'load_in_4bit',
         'torch_compile',
         'flash_attn',
-        'use_flash_attention_2',
+        'attn_implementation',
         'cpu',
         'disk',
         'row_split',
@@ -153,7 +153,6 @@ def list_model_elements():
         'mlock',
         'numa',
         'use_double_quant',
-        'use_eager_attention',
         'bf16',
         'autosplit',
         'enable_tp',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index e09e292e4b..2018a94359 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -44,6 +44,7 @@ def create_ui():
                             shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. Common values: 4096, 8192, 16384, 32768, 65536, 131072. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
+                            shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
                         with gr.Column():
                             shared.gradio['vram_info'] = gr.HTML(value=get_initial_vram_info())
@@ -52,7 +53,6 @@ def create_ui():
                             shared.gradio['load_in_8bit'] = gr.Checkbox(label="load-in-8bit", value=shared.args.load_in_8bit)
                             shared.gradio['load_in_4bit'] = gr.Checkbox(label="load-in-4bit", value=shared.args.load_in_4bit)
                             shared.gradio['torch_compile'] = gr.Checkbox(label="torch-compile", value=shared.args.torch_compile, info='Compile the model with torch.compile for improved performance.')
-                            shared.gradio['use_flash_attention_2'] = gr.Checkbox(label="use_flash_attention_2", value=shared.args.use_flash_attention_2, info='Set use_flash_attention_2=True while loading the model.')
                             shared.gradio['use_double_quant'] = gr.Checkbox(label="use_double_quant", value=shared.args.use_double_quant, info='Used by load-in-4bit.')
                             shared.gradio['autosplit'] = gr.Checkbox(label="autosplit", value=shared.args.autosplit, info='Automatically split the model tensors across the available GPUs.')
                             shared.gradio['enable_tp'] = gr.Checkbox(label="enable_tp", value=shared.args.enable_tp, info='Enable Tensor Parallelism (TP).')
@@ -96,7 +96,6 @@ def create_ui():
                                 shared.gradio['no_mmap'] = gr.Checkbox(label="no-mmap", value=shared.args.no_mmap)
                                 shared.gradio['mlock'] = gr.Checkbox(label="mlock", value=shared.args.mlock)
                                 shared.gradio['numa'] = gr.Checkbox(label="numa", value=shared.args.numa, info='NUMA support can help on some systems with non-uniform memory access.')
-                                shared.gradio['use_eager_attention'] = gr.Checkbox(label="use_eager_attention", value=shared.args.use_eager_attention, info='Set attn_implementation= eager while loading the model.')
                                 shared.gradio['bf16'] = gr.Checkbox(label="bf16", value=shared.args.bf16)
                                 shared.gradio['no_flash_attn'] = gr.Checkbox(label="no_flash_attn", value=shared.args.no_flash_attn)
                                 shared.gradio['no_xformers'] = gr.Checkbox(label="no_xformers", value=shared.args.no_xformers)

From c357601c01b52dbfec0670b377b7d60132e85c27 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Jul 2025 18:48:04 -0700
Subject: [PATCH 0809/1701] Bump transformers to 4.53

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_cuda128.txt         | 2 +-
 requirements/full/requirements_cuda128_noavx2.txt  | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 11 files changed, 11 insertions(+), 11 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 34fd13bee8..6f3b816e52 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -23,7 +23,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.53.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 9b18524a27..7eba3671ff 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.53.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 6b9cc2a9bb..6d6e990d20 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.53.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index b7830355b6..400774661c 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.53.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index e48d4dee6e..c8744377cd 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.53.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 831ff51fe0..d77dfd1c35 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.53.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index fb81f40b30..b1bfc351b8 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.53.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 3d29c93fca..58a246aa54 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -23,7 +23,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.53.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 7528aed0ef..b3859a51ae 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -23,7 +23,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.53.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index e86e452492..9e1a46e69f 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -23,7 +23,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.53.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 30020989eb..f3d2b08db8 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.50.*
+transformers==4.53.*
 tqdm
 wandb
 

From f045b72826820167d475aa2e6cac226485b586ef Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Jul 2025 19:46:26 -0700
Subject: [PATCH 0810/1701] Bump accelerate to 1.8

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_cuda128.txt         | 2 +-
 requirements/full/requirements_cuda128_noavx2.txt  | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 11 files changed, 11 insertions(+), 11 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 6f3b816e52..34f1f8d109 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -1,4 +1,4 @@
-accelerate==1.5.*
+accelerate==1.8.*
 bitsandbytes==0.45.*
 colorama
 datasets
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 7eba3671ff..af117c5c0f 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -1,4 +1,4 @@
-accelerate==1.5.*
+accelerate==1.8.*
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 6d6e990d20..8095c38c54 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.5.*
+accelerate==1.8.*
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 400774661c..d63207b6b3 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -1,4 +1,4 @@
-accelerate==1.5.*
+accelerate==1.8.*
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index c8744377cd..a2add551f8 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -1,4 +1,4 @@
-accelerate==1.5.*
+accelerate==1.8.*
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index d77dfd1c35..ec30e05d54 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -1,4 +1,4 @@
-accelerate==1.5.*
+accelerate==1.8.*
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index b1bfc351b8..d7cf75db11 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.5.*
+accelerate==1.8.*
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 58a246aa54..1db5c92810 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -1,4 +1,4 @@
-accelerate==1.5.*
+accelerate==1.8.*
 bitsandbytes==0.45.*
 colorama
 datasets
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index b3859a51ae..9eef3412f5 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.5.*
+accelerate==1.8.*
 bitsandbytes==0.45.*
 colorama
 datasets
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 9e1a46e69f..52a3068a48 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.5.*
+accelerate==1.8.*
 bitsandbytes==0.45.*
 colorama
 datasets
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index f3d2b08db8..4f35ffcc4d 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -1,4 +1,4 @@
-accelerate==1.5.*
+accelerate==1.8.*
 colorama
 datasets
 einops

From 8b3c7aa7952fdf92ae20c4f66e12df85ed8bcf21 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Jul 2025 19:46:55 -0700
Subject: [PATCH 0811/1701] Bump bitsandbytes to 0.46

---
 requirements/full/requirements.txt                | 2 +-
 requirements/full/requirements_cuda128.txt        | 2 +-
 requirements/full/requirements_cuda128_noavx2.txt | 2 +-
 requirements/full/requirements_noavx2.txt         | 2 +-
 4 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 34f1f8d109..791a45ebd8 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -1,5 +1,5 @@
 accelerate==1.8.*
-bitsandbytes==0.45.*
+bitsandbytes==0.46.*
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 1db5c92810..dd5c08accd 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -1,5 +1,5 @@
 accelerate==1.8.*
-bitsandbytes==0.45.*
+bitsandbytes==0.46.*
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 9eef3412f5..17dd31fd17 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -1,5 +1,5 @@
 accelerate==1.8.*
-bitsandbytes==0.45.*
+bitsandbytes==0.46.*
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 52a3068a48..be91420092 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -1,5 +1,5 @@
 accelerate==1.8.*
-bitsandbytes==0.45.*
+bitsandbytes==0.46.*
 colorama
 datasets
 einops

From b69f435311e47949c6341e298f4b8db396674926 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Jul 2025 19:56:50 -0700
Subject: [PATCH 0812/1701] Fix latest transformers being super slow

---
 modules/transformers_loader.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/modules/transformers_loader.py b/modules/transformers_loader.py
index ef524b5784..e4163b6d48 100644
--- a/modules/transformers_loader.py
+++ b/modules/transformers_loader.py
@@ -131,6 +131,8 @@ def load_tokenizer(model_name, tokenizer_dir=None):
 
 
 def load_model_HF(model_name):
+    torch._dynamo.config.disable = True
+
     path_to_model = Path(f'{shared.args.model_dir}/{model_name}')
     params = {
         'low_cpu_mem_usage': True,

From bd4881c4dcead542cd8abf7fc2020b259b8f7af6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Jul 2025 19:57:37 -0700
Subject: [PATCH 0813/1701] Use eager attention by default instead of sdpa

---
 modules/shared.py        | 2 +-
 modules/ui_model_menu.py | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 5764911491..b13156b738 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -61,7 +61,7 @@
 group.add_argument('--trust-remote-code', action='store_true', help='Set trust_remote_code=True while loading the model. Necessary for some models.')
 group.add_argument('--force-safetensors', action='store_true', help='Set use_safetensors=True while loading the model. This prevents arbitrary code execution.')
 group.add_argument('--no_use_fast', action='store_true', help='Set use_fast=False while loading the tokenizer (it\'s True by default). Use this if you have any problems related to use_fast.')
-group.add_argument('--attn-implementation', type=str, default='sdpa', help='Attention implementation. Valid options: sdpa, eager, flash_attention_2.')
+group.add_argument('--attn-implementation', type=str, default='eager', help='Attention implementation. Valid options: eager, sdpa, flash_attention_2.')
 group.add_argument('--torch-compile', action='store_true', help='Compile the model with torch.compile for improved performance.')
 
 # bitsandbytes 4-bit
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 2018a94359..365ca7e7e7 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -44,7 +44,7 @@ def create_ui():
                             shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. Common values: 4096, 8192, 16384, 32768, 65536, 131072. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
-                            shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
+                            shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['eager', 'sdpa', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
                         with gr.Column():
                             shared.gradio['vram_info'] = gr.HTML(value=get_initial_vram_info())

From e015355e4a28e365fa027271c67bd366220bbe52 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Jul 2025 20:03:53 -0700
Subject: [PATCH 0814/1701] Update README

---
 README.md         | 5 ++---
 modules/shared.py | 2 +-
 2 files changed, 3 insertions(+), 4 deletions(-)

diff --git a/README.md b/README.md
index c80dacde15..2f7fe4f5a6 100644
--- a/README.md
+++ b/README.md
@@ -235,7 +235,7 @@ List of command-line flags
 ```txt
 usage: server.py [-h] [--multi-user] [--model MODEL] [--lora LORA [LORA ...]] [--model-dir MODEL_DIR] [--lora-dir LORA_DIR] [--model-menu] [--settings SETTINGS]
                  [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR]
-                 [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--use_flash_attention_2] [--use_eager_attention] [--torch-compile] [--load-in-4bit]
+                 [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] [--torch-compile] [--load-in-4bit]
                  [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn] [--threads THREADS] [--threads-batch THREADS_BATCH] [--batch-size BATCH_SIZE] [--no-mmap]
                  [--mlock] [--gpu-layers N] [--tensor-split TENSOR_SPLIT] [--numa] [--no-kv-offload] [--row-split] [--extra-flags EXTRA_FLAGS] [--streaming-llm] [--ctx-size N] [--cache-type N]
                  [--model-draft MODEL_DRAFT] [--draft-max DRAFT_MAX] [--gpu-layers-draft GPU_LAYERS_DRAFT] [--device-draft DEVICE_DRAFT] [--ctx-size-draft CTX_SIZE_DRAFT] [--gpu-split GPU_SPLIT]
@@ -278,8 +278,7 @@ Transformers/Accelerate:
   --trust-remote-code                       Set trust_remote_code=True while loading the model. Necessary for some models.
   --force-safetensors                       Set use_safetensors=True while loading the model. This prevents arbitrary code execution.
   --no_use_fast                             Set use_fast=False while loading the tokenizer (it's True by default). Use this if you have any problems related to use_fast.
-  --use_flash_attention_2                   Set use_flash_attention_2=True while loading the model.
-  --use_eager_attention                     Set attn_implementation= eager while loading the model.
+  --attn-implementation IMPLEMENTATION      Attention implementation. Valid options: eager, sdpa, flash_attention_2.
   --torch-compile                           Compile the model with torch.compile for improved performance.
 
 bitsandbytes 4-bit:
diff --git a/modules/shared.py b/modules/shared.py
index b13156b738..0348d41e74 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -61,7 +61,7 @@
 group.add_argument('--trust-remote-code', action='store_true', help='Set trust_remote_code=True while loading the model. Necessary for some models.')
 group.add_argument('--force-safetensors', action='store_true', help='Set use_safetensors=True while loading the model. This prevents arbitrary code execution.')
 group.add_argument('--no_use_fast', action='store_true', help='Set use_fast=False while loading the tokenizer (it\'s True by default). Use this if you have any problems related to use_fast.')
-group.add_argument('--attn-implementation', type=str, default='eager', help='Attention implementation. Valid options: eager, sdpa, flash_attention_2.')
+group.add_argument('--attn-implementation', type=str, default='eager', metavar="IMPLEMENTATION", help='Attention implementation. Valid options: eager, sdpa, flash_attention_2.')
 group.add_argument('--torch-compile', action='store_true', help='Compile the model with torch.compile for improved performance.')
 
 # bitsandbytes 4-bit

From d1f4622a9666ce9b89e6f7796e6d8a3d33e4209b Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Thu, 10 Jul 2025 00:15:50 -0300
Subject: [PATCH 0815/1701] Update peft requirement from ==0.15.* to ==0.16.*
 in /requirements/full (#7127)

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_cuda128.txt         | 2 +-
 requirements/full/requirements_cuda128_noavx2.txt  | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 11 files changed, 11 insertions(+), 11 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 791a45ebd8..b2d0f98acf 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.15.*
+peft==0.16.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index af117c5c0f..e33a831d0d 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -9,7 +9,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.15.*
+peft==0.16.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 8095c38c54..3302afe5f2 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -9,7 +9,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.15.*
+peft==0.16.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index d63207b6b3..319dbc5752 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -9,7 +9,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.15.*
+peft==0.16.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index a2add551f8..2653c682e1 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -9,7 +9,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.15.*
+peft==0.16.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index ec30e05d54..ad09c53fd5 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -9,7 +9,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.15.*
+peft==0.16.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index d7cf75db11..c2ad776520 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -9,7 +9,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.15.*
+peft==0.16.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index dd5c08accd..e528b0ef1f 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.15.*
+peft==0.16.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 17dd31fd17..28b16740f4 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.15.*
+peft==0.16.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index be91420092..7e0fdaac28 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.15.*
+peft==0.16.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 4f35ffcc4d..ee5346e6e7 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -9,7 +9,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.15.*
+peft==0.16.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2

From 21e0e9f32bf7ed7989538c8b8f0755ae3f4d57c9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 9 Jul 2025 21:05:17 -0700
Subject: [PATCH 0816/1701] Add the triton-windows requirement on Windows to
 make transformers functional

---
 requirements/full/requirements.txt                 | 1 +
 requirements/full/requirements_amd.txt             | 1 +
 requirements/full/requirements_amd_noavx2.txt      | 1 +
 requirements/full/requirements_apple_intel.txt     | 1 +
 requirements/full/requirements_apple_silicon.txt   | 1 +
 requirements/full/requirements_cpu_only.txt        | 1 +
 requirements/full/requirements_cpu_only_noavx2.txt | 1 +
 requirements/full/requirements_cuda128.txt         | 1 +
 requirements/full/requirements_cuda128_noavx2.txt  | 1 +
 requirements/full/requirements_noavx2.txt          | 1 +
 requirements/full/requirements_nowheels.txt        | 1 +
 11 files changed, 11 insertions(+)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 791a45ebd8..4900038320 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -24,6 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
+triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index af117c5c0f..b72b680636 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -23,6 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
+triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 8095c38c54..898f345186 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -23,6 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
+triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index d63207b6b3..926baaf0e1 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -23,6 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
+triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index a2add551f8..b1fbebd66d 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -23,6 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
+triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index ec30e05d54..7cb2b2dac9 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -23,6 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
+triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index d7cf75db11..1aa90fbdd4 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -23,6 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
+triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index dd5c08accd..2f52baf98a 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -24,6 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
+triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 17dd31fd17..d06a235e56 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -24,6 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
+triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index be91420092..a5231cde5c 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -24,6 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
+triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 4f35ffcc4d..e4458ff638 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -23,6 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
+triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 

From e523f25b9fef994a8f7dd29d6713aafb0e53248f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 10 Jul 2025 05:35:57 -0700
Subject: [PATCH 0817/1701] Downgrade triton-windows to 3.2.0.post19

https://github.com/oobabooga/text-generation-webui/issues/7107#issuecomment-3057250374
---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_cuda128.txt         | 2 +-
 requirements/full/requirements_cuda128_noavx2.txt  | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 11 files changed, 11 insertions(+), 11 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index ab1ffce6e2..6d90e01c36 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -24,7 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
-triton-windows==3.3.1.post19; platform_system == "Windows"
+triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index ab88dae977..67b08f3f4b 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
-triton-windows==3.3.1.post19; platform_system == "Windows"
+triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 553346ae9b..94e9cb7f6f 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
-triton-windows==3.3.1.post19; platform_system == "Windows"
+triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index a053374d7b..e64ca5c9da 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
-triton-windows==3.3.1.post19; platform_system == "Windows"
+triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 15dd1f90a5..14f9878226 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
-triton-windows==3.3.1.post19; platform_system == "Windows"
+triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index bb02e30d27..6ed6bb916c 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
-triton-windows==3.3.1.post19; platform_system == "Windows"
+triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 5da4a198ec..8ab631a0ab 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
-triton-windows==3.3.1.post19; platform_system == "Windows"
+triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index f8dca3522d..bf9ee6cb20 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -24,7 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
-triton-windows==3.3.1.post19; platform_system == "Windows"
+triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index a3860b6a52..691f2ca4a1 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -24,7 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
-triton-windows==3.3.1.post19; platform_system == "Windows"
+triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 2f95e6f746..f569b1d4f6 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -24,7 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
-triton-windows==3.3.1.post19; platform_system == "Windows"
+triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 2e293dcc67..69a8218499 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
-triton-windows==3.3.1.post19; platform_system == "Windows"
+triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 

From 0f3a88057ca2e45fe8cae9256aed417c3a0e8601 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 10 Jul 2025 05:39:04 -0700
Subject: [PATCH 0818/1701] Don't downgrade triton-windows on CUDA 12.8

---
 requirements/full/requirements_cuda128.txt        | 2 +-
 requirements/full/requirements_cuda128_noavx2.txt | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index bf9ee6cb20..f8dca3522d 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -24,7 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
-triton-windows==3.2.0.post19; platform_system == "Windows"
+triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 691f2ca4a1..a3860b6a52 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -24,7 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.53.*
-triton-windows==3.2.0.post19; platform_system == "Windows"
+triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 

From 635e6efd18cbf12acd9c32142763c1623205947c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 10 Jul 2025 07:04:47 -0700
Subject: [PATCH 0819/1701] Ignore add_bos_token in instruct prompts, let the
 jinja2 template decide

---
 modules/chat.py            |  4 ++++
 modules/text_generation.py | 11 +++--------
 modules/ui_parameters.py   |  2 +-
 3 files changed, 8 insertions(+), 9 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 9290dd6223..827b60503d 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -643,6 +643,10 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
     output = apply_extensions('history', output)
     state = apply_extensions('state', state)
 
+    # Let the jinja2 template handle the BOS token
+    if state['mode'] in ['instruct', 'chat-instruct']:
+        state['add_bos_token'] = False
+
     # Initialize metadata if not present
     if 'metadata' not in output:
         output['metadata'] = {}
diff --git a/modules/text_generation.py b/modules/text_generation.py
index a75141f198..8d1950b98d 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -134,7 +134,6 @@ def encode(prompt, add_special_tokens=True, add_bos_token=True, truncation_lengt
                 input_ids = np.array(input_ids).reshape(1, len(input_ids))
         else:
             input_ids = shared.tokenizer.encode(str(prompt), return_tensors='pt', add_special_tokens=add_special_tokens)
-
             if hasattr(shared.tokenizer, 'bos_token_id') and shared.tokenizer.bos_token_id is not None:
                 if add_bos_token:
                     # Add BOS token if missing
@@ -142,13 +141,9 @@ def encode(prompt, add_special_tokens=True, add_bos_token=True, truncation_lengt
                         bos_tensor = torch.tensor([[shared.tokenizer.bos_token_id]])
                         input_ids = torch.cat((bos_tensor, input_ids), 1)
 
-                    # Prevent double BOS tokens from jinja templates
-                    while len(input_ids[0]) > 1 and input_ids[0][0] == shared.tokenizer.bos_token_id and input_ids[0][1] == shared.tokenizer.bos_token_id:
-                        input_ids = input_ids[:, 1:]
-                else:
-                    # Remove BOS tokens when not wanted
-                    while len(input_ids[0]) > 0 and input_ids[0][0] == shared.tokenizer.bos_token_id:
-                        input_ids = input_ids[:, 1:]
+                # Always prevent double BOS tokens (regardless of add_bos_token setting)
+                while len(input_ids[0]) > 1 and input_ids[0][0] == shared.tokenizer.bos_token_id and input_ids[0][1] == shared.tokenizer.bos_token_id:
+                    input_ids = input_ids[:, 1:]
 
         if truncation_length is not None:
             input_ids = input_ids[:, -truncation_length:]
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 5c92f32edb..45bef5f9e3 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -84,7 +84,7 @@ def create_ui():
 
                             shared.gradio['auto_max_new_tokens'] = gr.Checkbox(value=shared.settings['auto_max_new_tokens'], label='auto_max_new_tokens', info='Expand max_new_tokens to the available context length.')
                             shared.gradio['ban_eos_token'] = gr.Checkbox(value=shared.settings['ban_eos_token'], label='Ban the eos_token', info='Forces the model to never end the generation prematurely.')
-                            shared.gradio['add_bos_token'] = gr.Checkbox(value=shared.settings['add_bos_token'], label='Add the bos_token to the beginning of prompts', info='Disabling this can make the replies more creative.')
+                            shared.gradio['add_bos_token'] = gr.Checkbox(value=shared.settings['add_bos_token'], label='Add the bos_token to the beginning of prompts', info='Only applies to text completion (notebook). In chat mode, templates control BOS tokens.')
                             shared.gradio['skip_special_tokens'] = gr.Checkbox(value=shared.settings['skip_special_tokens'], label='Skip special tokens', info='Some specific models need this unset.')
                             shared.gradio['stream'] = gr.Checkbox(value=shared.settings['stream'], label='Activate text streaming')
                             shared.gradio['static_cache'] = gr.Checkbox(value=shared.settings['static_cache'], label='Static KV cache', info='Use a static cache for improved performance.')

From 188c7c8f2b987d0af95e2b3c9df2004cda269941 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 10 Jul 2025 18:42:52 -0700
Subject: [PATCH 0820/1701] Revert "CSS simplifications"

This reverts commit c6c1b725e95594f03891cdabe0be51ca8cb33c0a.
---
 css/chat_style-Dark.css            | 20 +++++++++++++++++++-
 css/chat_style-TheEncrypted777.css | 20 +++++++++++++++++++-
 css/chat_style-cai-chat.css        | 21 ++++++++++++++++++++-
 css/chat_style-messenger.css       | 23 ++++++++++++++++++++++-
 css/chat_style-wpp.css             | 20 +++++++++++++++++++-
 css/html_readable_style.css        | 12 +++++++++++-
 6 files changed, 110 insertions(+), 6 deletions(-)

diff --git a/css/chat_style-Dark.css b/css/chat_style-Dark.css
index 11603ff55f..aae82f8b92 100644
--- a/css/chat_style-Dark.css
+++ b/css/chat_style-Dark.css
@@ -35,6 +35,10 @@
     color: #f0f0f0; /* Light text color for readability */
 }
 
+.text p {
+    margin-top: 2px;
+}
+
 .username {
     padding-left: 10px;
     font-size: 20px;
@@ -125,7 +129,7 @@
     }
 }
 
-/* Standard spacing from instruct style */
+/* Standard spacing */
 .chat .message-body :is(p, ul, ol) {
     margin: 1.25em 0 !important;
 }
@@ -137,3 +141,17 @@
 .chat .message-body :is(p, ul, ol):last-child {
     margin-bottom: 0 !important;
 }
+
+.chat .message-body ul, .chat .message-body ol {
+    padding-inline-start: 2em;
+}
+
+.message-body li {
+    list-style-position: outside;
+    margin-top: 0.5em !important;
+    margin-bottom: 0.5em !important;
+}
+
+.message-body li > p {
+    display: inline !important;
+}
diff --git a/css/chat_style-TheEncrypted777.css b/css/chat_style-TheEncrypted777.css
index e1f378397a..7f37377d82 100644
--- a/css/chat_style-TheEncrypted777.css
+++ b/css/chat_style-TheEncrypted777.css
@@ -38,6 +38,10 @@
     text-shadow: 2px 2px 2px rgb(0 0 0 / 40%);
 }
 
+.text p {
+    margin-top: 2px;
+}
+
 .username {
     padding-left: 10px;
     font-size: 22px;
@@ -131,7 +135,7 @@
     }
 }
 
-/* Standard spacing from instruct style */
+/* Standard spacing */
 .chat .message-body :is(p, ul, ol) {
     margin: 1.25em 0 !important;
 }
@@ -143,3 +147,17 @@
 .chat .message-body :is(p, ul, ol):last-child {
     margin-bottom: 0 !important;
 }
+
+.chat .message-body ul, .chat .message-body ol {
+    padding-inline-start: 2em;
+}
+
+.message-body li {
+    list-style-position: outside;
+    margin-top: 0.5em !important;
+    margin-bottom: 0.5em !important;
+}
+
+.message-body li > p {
+    display: inline !important;
+}
diff --git a/css/chat_style-cai-chat.css b/css/chat_style-cai-chat.css
index ef1f1f6231..d2e8f44aad 100644
--- a/css/chat_style-cai-chat.css
+++ b/css/chat_style-cai-chat.css
@@ -9,6 +9,11 @@
     line-height: 22.5px !important;
 }
 
+.message-body {
+    margin-top: 3px;
+    font-size: 15px !important;
+}
+
 .circle-you {
     width: 50px;
     height: 50px;
@@ -56,7 +61,7 @@
     font-weight: 500;
 }
 
-/* Standard spacing from instruct style */
+/* Standard spacing */
 .chat .message-body :is(p, ul, ol) {
     margin: 1.25em 0 !important;
 }
@@ -68,3 +73,17 @@
 .chat .message-body :is(p, ul, ol):last-child {
     margin-bottom: 0 !important;
 }
+
+.chat .message-body ul, .chat .message-body ol {
+    padding-inline-start: 2em;
+}
+
+.message-body li {
+    list-style-position: outside;
+    margin-top: 0.5em !important;
+    margin-bottom: 0.5em !important;
+}
+
+.message-body li > p {
+    display: inline !important;
+}
diff --git a/css/chat_style-messenger.css b/css/chat_style-messenger.css
index b6da2bb2bb..a6189a4aa6 100644
--- a/css/chat_style-messenger.css
+++ b/css/chat_style-messenger.css
@@ -68,10 +68,17 @@
     max-width: 80%;
 }
 
+.text p {
+    margin-top: 5px;
+}
+
 .username {
     font-weight: bold;
 }
 
+.message-body {
+}
+
 .message-body img {
     max-width: 300px;
     max-height: 300px;
@@ -92,7 +99,7 @@
     color: rgb(110 110 110) !important;
 }
 
-/* Standard spacing from instruct style */
+/* Standard spacing */
 .chat .message-body :is(p, ul, ol) {
     margin: 1.25em 0 !important;
 }
@@ -104,3 +111,17 @@
 .chat .message-body :is(p, ul, ol):last-child {
     margin-bottom: 0 !important;
 }
+
+.chat .message-body ul, .chat .message-body ol {
+    padding-inline-start: 2em;
+}
+
+.message-body li {
+    list-style-position: outside;
+    margin-top: 0.5em !important;
+    margin-bottom: 0.5em !important;
+}
+
+.message-body li > p {
+    display: inline !important;
+}
diff --git a/css/chat_style-wpp.css b/css/chat_style-wpp.css
index d11172ef56..c86e577da5 100644
--- a/css/chat_style-wpp.css
+++ b/css/chat_style-wpp.css
@@ -83,6 +83,10 @@
     font-weight: 400;
 }
 
+.message-body p:first-child {
+    margin-top: 0 !important;
+}
+
 .dark .message-body p em {
     color: rgb(170 170 170) !important;
 }
@@ -96,7 +100,7 @@
     margin-top: 8px;
 }
 
-/* Standard spacing from instruct style */
+/* Standard spacing */
 .chat .message-body :is(p, ul, ol) {
     margin: 1.25em 0 !important;
 }
@@ -108,3 +112,17 @@
 .chat .message-body :is(p, ul, ol):last-child {
     margin-bottom: 0 !important;
 }
+
+.chat .message-body ul, .chat .message-body ol {
+    padding-inline-start: 2em;
+}
+
+.message-body li {
+    list-style-position: outside;
+    margin-top: 0.5em !important;
+    margin-bottom: 0.5em !important;
+}
+
+.message-body li > p {
+    display: inline !important;
+}
diff --git a/css/html_readable_style.css b/css/html_readable_style.css
index 74f69b73ac..74a913e93e 100644
--- a/css/html_readable_style.css
+++ b/css/html_readable_style.css
@@ -31,7 +31,7 @@
     font-size: 14px;
 }
 
-/* Standard spacing from instruct style */
+/* Standard spacing */
 .readable-container :is(p, ul, ol) {
     margin: 1.25em 0 !important;
 }
@@ -43,3 +43,13 @@
 .readable-container :is(p, ul, ol):last-child {
     margin-bottom: 0 !important;
 }
+
+.readable-container ul, .readable-container ol {
+    padding-inline-start: 2em;
+}
+
+.readable-container li {
+    list-style-position: outside;
+    margin-top: 0.5em !important;
+    margin-bottom: 0.5em !important;
+}

From caf69d871a74b603f52fc5f0fe644c8ec80b07f3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 10 Jul 2025 18:43:01 -0700
Subject: [PATCH 0821/1701] Revert "Standardize margins and paddings across all
 chat styles"

This reverts commit 86cb5e058711633bd4fa45faabcac699020386e5.
---
 css/chat_style-Dark.css            | 28 +--------------------------
 css/chat_style-TheEncrypted777.css | 28 +--------------------------
 css/chat_style-cai-chat.css        | 31 ++++--------------------------
 css/chat_style-messenger.css       | 28 +--------------------------
 css/chat_style-wpp.css             | 27 ++------------------------
 css/html_readable_style.css        | 28 +++------------------------
 6 files changed, 12 insertions(+), 158 deletions(-)

diff --git a/css/chat_style-Dark.css b/css/chat_style-Dark.css
index aae82f8b92..6a4784ccd1 100644
--- a/css/chat_style-Dark.css
+++ b/css/chat_style-Dark.css
@@ -91,6 +91,7 @@
 }
 
 .message-body p {
+    margin-bottom: 0 !important;
     font-size: 16px !important;
     line-height: 1.5 !important;
     color: #e0e0e0 !important; /* Light color for text */
@@ -128,30 +129,3 @@
         font-size: 18px; /* Smaller username for mobile */
     }
 }
-
-/* Standard spacing */
-.chat .message-body :is(p, ul, ol) {
-    margin: 1.25em 0 !important;
-}
-
-.chat .message-body :is(p, ul, ol):first-child {
-    margin-top: 0 !important;
-}
-
-.chat .message-body :is(p, ul, ol):last-child {
-    margin-bottom: 0 !important;
-}
-
-.chat .message-body ul, .chat .message-body ol {
-    padding-inline-start: 2em;
-}
-
-.message-body li {
-    list-style-position: outside;
-    margin-top: 0.5em !important;
-    margin-bottom: 0.5em !important;
-}
-
-.message-body li > p {
-    display: inline !important;
-}
diff --git a/css/chat_style-TheEncrypted777.css b/css/chat_style-TheEncrypted777.css
index 7f37377d82..fbd47072fb 100644
--- a/css/chat_style-TheEncrypted777.css
+++ b/css/chat_style-TheEncrypted777.css
@@ -87,6 +87,7 @@
 }
 
 .message-body p {
+    margin-bottom: 0 !important;
     font-size: 18px !important;
     line-height: 1.428571429 !important;
     color: rgb(243 244 246) !important;
@@ -134,30 +135,3 @@
         font-size: 20px;
     }
 }
-
-/* Standard spacing */
-.chat .message-body :is(p, ul, ol) {
-    margin: 1.25em 0 !important;
-}
-
-.chat .message-body :is(p, ul, ol):first-child {
-    margin-top: 0 !important;
-}
-
-.chat .message-body :is(p, ul, ol):last-child {
-    margin-bottom: 0 !important;
-}
-
-.chat .message-body ul, .chat .message-body ol {
-    padding-inline-start: 2em;
-}
-
-.message-body li {
-    list-style-position: outside;
-    margin-top: 0.5em !important;
-    margin-bottom: 0.5em !important;
-}
-
-.message-body li > p {
-    display: inline !important;
-}
diff --git a/css/chat_style-cai-chat.css b/css/chat_style-cai-chat.css
index d2e8f44aad..b06b1269c7 100644
--- a/css/chat_style-cai-chat.css
+++ b/css/chat_style-cai-chat.css
@@ -52,6 +52,10 @@
     font-weight: 500;
 }
 
+.message-body p, .chat .message-body ul, .chat .message-body ol {
+    margin-bottom: 10px !important;
+}
+
 .dark .message-body p em {
     color: rgb(138 138 138) !important;
 }
@@ -60,30 +64,3 @@
     color: rgb(110 110 110) !important;
     font-weight: 500;
 }
-
-/* Standard spacing */
-.chat .message-body :is(p, ul, ol) {
-    margin: 1.25em 0 !important;
-}
-
-.chat .message-body :is(p, ul, ol):first-child {
-    margin-top: 0 !important;
-}
-
-.chat .message-body :is(p, ul, ol):last-child {
-    margin-bottom: 0 !important;
-}
-
-.chat .message-body ul, .chat .message-body ol {
-    padding-inline-start: 2em;
-}
-
-.message-body li {
-    list-style-position: outside;
-    margin-top: 0.5em !important;
-    margin-bottom: 0.5em !important;
-}
-
-.message-body li > p {
-    display: inline !important;
-}
diff --git a/css/chat_style-messenger.css b/css/chat_style-messenger.css
index a6189a4aa6..65af5f7a5f 100644
--- a/css/chat_style-messenger.css
+++ b/css/chat_style-messenger.css
@@ -86,6 +86,7 @@
 }
 
 .message-body p {
+    margin-bottom: 0 !important;
     font-size: 15px !important;
     line-height: 1.428571429 !important;
     font-weight: 500;
@@ -98,30 +99,3 @@
 .message-body p em {
     color: rgb(110 110 110) !important;
 }
-
-/* Standard spacing */
-.chat .message-body :is(p, ul, ol) {
-    margin: 1.25em 0 !important;
-}
-
-.chat .message-body :is(p, ul, ol):first-child {
-    margin-top: 0 !important;
-}
-
-.chat .message-body :is(p, ul, ol):last-child {
-    margin-bottom: 0 !important;
-}
-
-.chat .message-body ul, .chat .message-body ol {
-    padding-inline-start: 2em;
-}
-
-.message-body li {
-    list-style-position: outside;
-    margin-top: 0.5em !important;
-    margin-bottom: 0.5em !important;
-}
-
-.message-body li > p {
-    display: inline !important;
-}
diff --git a/css/chat_style-wpp.css b/css/chat_style-wpp.css
index c86e577da5..b2ac4d2a38 100644
--- a/css/chat_style-wpp.css
+++ b/css/chat_style-wpp.css
@@ -100,29 +100,6 @@
     margin-top: 8px;
 }
 
-/* Standard spacing */
-.chat .message-body :is(p, ul, ol) {
-    margin: 1.25em 0 !important;
-}
-
-.chat .message-body :is(p, ul, ol):first-child {
-    margin-top: 0 !important;
-}
-
-.chat .message-body :is(p, ul, ol):last-child {
-    margin-bottom: 0 !important;
-}
-
-.chat .message-body ul, .chat .message-body ol {
-    padding-inline-start: 2em;
-}
-
-.message-body li {
-    list-style-position: outside;
-    margin-top: 0.5em !important;
-    margin-bottom: 0.5em !important;
-}
-
-.message-body li > p {
-    display: inline !important;
+.message-body p, .chat .message-body ul, .chat .message-body ol {
+    margin-bottom: 10px !important;
 }
diff --git a/css/html_readable_style.css b/css/html_readable_style.css
index 74a913e93e..760649d600 100644
--- a/css/html_readable_style.css
+++ b/css/html_readable_style.css
@@ -11,8 +11,9 @@
 
 .readable-container p, .readable-container li {
     font-size: 16px !important;
-    line-height: 1.4 !important;
     color: #efefef !important;
+    margin-bottom: 22px;
+    line-height: 1.4 !important;
 }
 
 .readable-container li > p {
@@ -29,27 +30,4 @@
 
 .readable-container .hoverable {
     font-size: 14px;
-}
-
-/* Standard spacing */
-.readable-container :is(p, ul, ol) {
-    margin: 1.25em 0 !important;
-}
-
-.readable-container :is(p, ul, ol):first-child {
-    margin-top: 0 !important;
-}
-
-.readable-container :is(p, ul, ol):last-child {
-    margin-bottom: 0 !important;
-}
-
-.readable-container ul, .readable-container ol {
-    padding-inline-start: 2em;
-}
-
-.readable-container li {
-    list-style-position: outside;
-    margin-top: 0.5em !important;
-    margin-bottom: 0.5em !important;
-}
+}
\ No newline at end of file

From 273888f2181fb6377a0921fb0eb198e5a0c7ca4c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 10 Jul 2025 18:56:46 -0700
Subject: [PATCH 0822/1701] Revert "Use eager attention by default instead of
 sdpa"

This reverts commit bd4881c4dcead542cd8abf7fc2020b259b8f7af6.
---
 modules/shared.py        | 2 +-
 modules/ui_model_menu.py | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 0348d41e74..d368062b88 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -61,7 +61,7 @@
 group.add_argument('--trust-remote-code', action='store_true', help='Set trust_remote_code=True while loading the model. Necessary for some models.')
 group.add_argument('--force-safetensors', action='store_true', help='Set use_safetensors=True while loading the model. This prevents arbitrary code execution.')
 group.add_argument('--no_use_fast', action='store_true', help='Set use_fast=False while loading the tokenizer (it\'s True by default). Use this if you have any problems related to use_fast.')
-group.add_argument('--attn-implementation', type=str, default='eager', metavar="IMPLEMENTATION", help='Attention implementation. Valid options: eager, sdpa, flash_attention_2.')
+group.add_argument('--attn-implementation', type=str, default='sdpa', metavar="IMPLEMENTATION", help='Attention implementation. Valid options: sdpa, eager, flash_attention_2.')
 group.add_argument('--torch-compile', action='store_true', help='Compile the model with torch.compile for improved performance.')
 
 # bitsandbytes 4-bit
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 365ca7e7e7..2018a94359 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -44,7 +44,7 @@ def create_ui():
                             shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. Common values: 4096, 8192, 16384, 32768, 65536, 131072. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
-                            shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['eager', 'sdpa', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
+                            shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
                         with gr.Column():
                             shared.gradio['vram_info'] = gr.HTML(value=get_initial_vram_info())

From 5a8a9c22e855cb8886c3215deacf1786d5c94066 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 11 Jul 2025 09:20:27 -0700
Subject: [PATCH 0823/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_cuda128.txt             | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt      | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 6d90e01c36..69cf32ad46 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 67b08f3f4b..9bce8e5743 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 94e9cb7f6f..6edf1b2e31 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index e64ca5c9da..82419dd209 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 14f9878226..ac823e68c2 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 6ed6bb916c..0b1514d9fa 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 8ab631a0ab..8274217679 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index f8dca3522d..6c890c6b5e 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index a3860b6a52..05b90d3b43 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index f569b1d4f6..d3b3efbf8e 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index e9fa61f163..f3d97d5239 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 4f9b0668cd..3cf8fcd1a2 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 9f403a0bd6..52b7e72e7e 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -18,6 +18,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 8e9baa3245..b5b114cc73 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index ebd6359131..4386608281 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 6802a3372b..efb62cbb74 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 624c222db5..1f0f18f37a 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index edb82a319a..75a04a26e7 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.24.0/llama_cpp_binaries-0.24.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 1d1b20bd77196cf07243e444eb8ed105cf7654d0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 11 Jul 2025 10:51:23 -0700
Subject: [PATCH 0824/1701] Remove the --torch-compile option (it doesn't do
 anything currently)

---
 README.md                      | 9 ++++-----
 modules/loaders.py             | 1 -
 modules/shared.py              | 1 -
 modules/transformers_loader.py | 3 ---
 modules/ui.py                  | 1 -
 modules/ui_model_menu.py       | 3 +--
 6 files changed, 5 insertions(+), 13 deletions(-)

diff --git a/README.md b/README.md
index 2f7fe4f5a6..907d8c38fd 100644
--- a/README.md
+++ b/README.md
@@ -235,9 +235,9 @@ List of command-line flags
 ```txt
 usage: server.py [-h] [--multi-user] [--model MODEL] [--lora LORA [LORA ...]] [--model-dir MODEL_DIR] [--lora-dir LORA_DIR] [--model-menu] [--settings SETTINGS]
                  [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR]
-                 [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] [--torch-compile] [--load-in-4bit]
-                 [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn] [--threads THREADS] [--threads-batch THREADS_BATCH] [--batch-size BATCH_SIZE] [--no-mmap]
-                 [--mlock] [--gpu-layers N] [--tensor-split TENSOR_SPLIT] [--numa] [--no-kv-offload] [--row-split] [--extra-flags EXTRA_FLAGS] [--streaming-llm] [--ctx-size N] [--cache-type N]
+                 [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] [--load-in-4bit] [--use_double_quant]
+                 [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn] [--threads THREADS] [--threads-batch THREADS_BATCH] [--batch-size BATCH_SIZE] [--no-mmap] [--mlock]
+                 [--gpu-layers N] [--tensor-split TENSOR_SPLIT] [--numa] [--no-kv-offload] [--row-split] [--extra-flags EXTRA_FLAGS] [--streaming-llm] [--ctx-size N] [--cache-type N]
                  [--model-draft MODEL_DRAFT] [--draft-max DRAFT_MAX] [--gpu-layers-draft GPU_LAYERS_DRAFT] [--device-draft DEVICE_DRAFT] [--ctx-size-draft CTX_SIZE_DRAFT] [--gpu-split GPU_SPLIT]
                  [--autosplit] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa] [--num_experts_per_token N] [--enable_tp] [--cpp-runner] [--deepspeed] [--nvme-offload-dir NVME_OFFLOAD_DIR]
                  [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT]
@@ -278,8 +278,7 @@ Transformers/Accelerate:
   --trust-remote-code                       Set trust_remote_code=True while loading the model. Necessary for some models.
   --force-safetensors                       Set use_safetensors=True while loading the model. This prevents arbitrary code execution.
   --no_use_fast                             Set use_fast=False while loading the tokenizer (it's True by default). Use this if you have any problems related to use_fast.
-  --attn-implementation IMPLEMENTATION      Attention implementation. Valid options: eager, sdpa, flash_attention_2.
-  --torch-compile                           Compile the model with torch.compile for improved performance.
+  --attn-implementation IMPLEMENTATION      Attention implementation. Valid options: sdpa, eager, flash_attention_2.
 
 bitsandbytes 4-bit:
   --load-in-4bit                            Load the model with 4-bit precision (using bitsandbytes).
diff --git a/modules/loaders.py b/modules/loaders.py
index 2b2c6b784e..f515aeca6c 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -39,7 +39,6 @@
         'quant_type',
         'load_in_8bit',
         'load_in_4bit',
-        'torch_compile',
         'attn_implementation',
         'cpu',
         'disk',
diff --git a/modules/shared.py b/modules/shared.py
index d368062b88..5e3e11c059 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -62,7 +62,6 @@
 group.add_argument('--force-safetensors', action='store_true', help='Set use_safetensors=True while loading the model. This prevents arbitrary code execution.')
 group.add_argument('--no_use_fast', action='store_true', help='Set use_fast=False while loading the tokenizer (it\'s True by default). Use this if you have any problems related to use_fast.')
 group.add_argument('--attn-implementation', type=str, default='sdpa', metavar="IMPLEMENTATION", help='Attention implementation. Valid options: sdpa, eager, flash_attention_2.')
-group.add_argument('--torch-compile', action='store_true', help='Compile the model with torch.compile for improved performance.')
 
 # bitsandbytes 4-bit
 group = parser.add_argument_group('bitsandbytes 4-bit')
diff --git a/modules/transformers_loader.py b/modules/transformers_loader.py
index e4163b6d48..2f7367a4d0 100644
--- a/modules/transformers_loader.py
+++ b/modules/transformers_loader.py
@@ -258,9 +258,6 @@ def load_model_HF(model_name):
         print()
         model = LoaderClass.from_pretrained(path_to_model, **params)
 
-    if shared.args.torch_compile:
-        model = torch.compile(model)
-
     return model
 
 
diff --git a/modules/ui.py b/modules/ui.py
index 9d6777f0a1..98acc03861 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -142,7 +142,6 @@ def list_model_elements():
         'num_experts_per_token',
         'load_in_8bit',
         'load_in_4bit',
-        'torch_compile',
         'flash_attn',
         'attn_implementation',
         'cpu',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 2018a94359..ba7c529e1e 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -52,7 +52,6 @@ def create_ui():
                             shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming-llm", value=shared.args.streaming_llm, info='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
                             shared.gradio['load_in_8bit'] = gr.Checkbox(label="load-in-8bit", value=shared.args.load_in_8bit)
                             shared.gradio['load_in_4bit'] = gr.Checkbox(label="load-in-4bit", value=shared.args.load_in_4bit)
-                            shared.gradio['torch_compile'] = gr.Checkbox(label="torch-compile", value=shared.args.torch_compile, info='Compile the model with torch.compile for improved performance.')
                             shared.gradio['use_double_quant'] = gr.Checkbox(label="use_double_quant", value=shared.args.use_double_quant, info='Used by load-in-4bit.')
                             shared.gradio['autosplit'] = gr.Checkbox(label="autosplit", value=shared.args.autosplit, info='Automatically split the model tensors across the available GPUs.')
                             shared.gradio['enable_tp'] = gr.Checkbox(label="enable_tp", value=shared.args.enable_tp, info='Enable Tensor Parallelism (TP).')
@@ -131,7 +130,7 @@ def create_ui():
 def create_event_handlers():
     mu = shared.args.multi_user
     if mu:
-        return  
+        return
 
     shared.gradio['loader'].change(loaders.make_loader_params_visible, gradio('loader'), gradio(loaders.get_all_params()), show_progress=False)
 

From 845432b9b47d948adad872cc593e85de4c96c5b9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 14 Jul 2025 21:03:18 -0700
Subject: [PATCH 0825/1701] Remove the obsolete modules/relative_imports.py
 file

---
 modules/relative_imports.py | 13 -------------
 1 file changed, 13 deletions(-)
 delete mode 100644 modules/relative_imports.py

diff --git a/modules/relative_imports.py b/modules/relative_imports.py
deleted file mode 100644
index 3c0eb56b77..0000000000
--- a/modules/relative_imports.py
+++ /dev/null
@@ -1,13 +0,0 @@
-import sys
-from pathlib import Path
-
-
-class RelativeImport:
-    def __init__(self, path):
-        self.import_path = Path(path)
-
-    def __enter__(self):
-        sys.path.insert(0, str(self.import_path))
-
-    def __exit__(self, exc_type, exc_value, traceback):
-        sys.path.remove(str(self.import_path))

From 03fb85e49ab8d8d9019db09d67081ae3a540f3d9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 15 Jul 2025 07:37:13 -0700
Subject: [PATCH 0826/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_cuda128.txt             | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt      | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 69cf32ad46..a45a14465e 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 9bce8e5743..b67e97a13c 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 6edf1b2e31..bf02245d4e 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 82419dd209..5b2dac777c 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index ac823e68c2..9e2ef7baf3 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 0b1514d9fa..3495859571 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 8274217679..094077408e 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 6c890c6b5e..432018c5c6 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 05b90d3b43..a737c5e1b8 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index d3b3efbf8e..914feeae8c 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index f3d97d5239..ceec09aa6d 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 3cf8fcd1a2..a53cbed8df 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 52b7e72e7e..823ec02848 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -18,6 +18,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index b5b114cc73..ce6c215b0c 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 4386608281..1db63e55b9 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index efb62cbb74..b8929f1772 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 1f0f18f37a..612c9ef28c 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 75a04a26e7..bf504c9572 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.26.0/llama_cpp_binaries-0.26.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 9371867238c4a38cfe955067fce7bcadfaef4205 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 15 Jul 2025 07:38:03 -0700
Subject: [PATCH 0827/1701] Update exllamav2

---
 requirements/full/requirements.txt                | 6 +++---
 requirements/full/requirements_amd.txt            | 4 ++--
 requirements/full/requirements_amd_noavx2.txt     | 4 ++--
 requirements/full/requirements_apple_intel.txt    | 2 +-
 requirements/full/requirements_apple_silicon.txt  | 2 +-
 requirements/full/requirements_cuda128.txt        | 6 +++---
 requirements/full/requirements_cuda128_noavx2.txt | 6 +++---
 requirements/full/requirements_noavx2.txt         | 6 +++---
 8 files changed, 18 insertions(+), 18 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index a45a14465e..946b312e5d 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -38,8 +38,8 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index b67e97a13c..8047ed78fa 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -35,5 +35,5 @@ tiktoken
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index bf02245d4e..1736fe5874 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -35,5 +35,5 @@ tiktoken
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 5b2dac777c..964ec5bd18 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -36,4 +36,4 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 9e2ef7baf3..305474043e 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,4 +37,4 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 432018c5c6..3b60343cb7 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -38,8 +38,8 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index a737c5e1b8..73c2166950 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -38,8 +38,8 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 914feeae8c..eb123b9970 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -38,8 +38,8 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.1/exllamav2-0.3.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

From a00983b2ba1341c6e17f8f37aa9386252668dc6e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 19 Jul 2025 12:05:50 -0700
Subject: [PATCH 0828/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_cuda128.txt             | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt      | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 946b312e5d..b83cc9879d 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 8047ed78fa..8224d98700 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 1736fe5874..22141a8a5d 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 964ec5bd18..e9290bcb41 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 305474043e..449fde875a 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 3495859571..8a84e4030d 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 094077408e..9488f5e75d 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 3b60343cb7..c69cfb5dcb 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 73c2166950..4f24e459ac 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index eb123b9970..537b200411 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index ceec09aa6d..53479a80f9 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index a53cbed8df..d7336d2fcd 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 823ec02848..1edaa5156b 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -18,6 +18,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index ce6c215b0c..04c9b28343 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 1db63e55b9..3c3563d324 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index b8929f1772..cf0d7b11e5 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 612c9ef28c..9bd8a37ce5 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index bf504c9572..b8519553fc 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.27.0/llama_cpp_binaries-0.27.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From ccf5e3e3a7d20d04559f37fbc8ca3ed346affc58 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 19 Jul 2025 12:07:38 -0700
Subject: [PATCH 0829/1701] Update exllamav3

---
 requirements/full/requirements.txt                | 4 ++--
 requirements/full/requirements_apple_intel.txt    | 2 +-
 requirements/full/requirements_apple_silicon.txt  | 2 +-
 requirements/full/requirements_cuda128.txt        | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt | 4 ++--
 requirements/full/requirements_noavx2.txt         | 4 ++--
 6 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index b83cc9879d..687f1f5a82 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -36,8 +36,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index e9290bcb41..04325464c4 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -35,5 +35,5 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 449fde875a..9497575fc4 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -36,5 +36,5 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index c69cfb5dcb..a2af51086b 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -36,8 +36,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 4f24e459ac..948a275abd 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -36,8 +36,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 537b200411..8f7106e4f0 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -36,8 +36,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.4/exllamav3-0.0.4+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 0c667de7a75d97bf8066e83349f3b1157b0e35ba Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 19 Jul 2025 12:14:41 -0700
Subject: [PATCH 0830/1701] UI: Add a None option for the speculative decoding
 model (closes #7145)

---
 modules/ui_model_menu.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index ba7c529e1e..0ab67e7a5b 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -62,8 +62,8 @@ def create_ui():
                             # Speculative decoding
                             with gr.Accordion("Speculative decoding", open=False, elem_classes='tgw-accordion') as shared.gradio['speculative_decoding_accordion']:
                                 with gr.Row():
-                                    shared.gradio['model_draft'] = gr.Dropdown(label="model-draft", choices=utils.get_available_models(), value=lambda: shared.args.model_draft, elem_classes='slim-dropdown', info='Draft model. Speculative decoding only works with models sharing the same vocabulary (e.g., same model family).', interactive=not mu)
-                                    ui.create_refresh_button(shared.gradio['model_draft'], lambda: None, lambda: {'choices': utils.get_available_models()}, 'refresh-button', interactive=not mu)
+                                    shared.gradio['model_draft'] = gr.Dropdown(label="model-draft", choices=['None'] + utils.get_available_models(), value=lambda: shared.args.model_draft, elem_classes='slim-dropdown', info='Draft model. Speculative decoding only works with models sharing the same vocabulary (e.g., same model family).', interactive=not mu)
+                                    ui.create_refresh_button(shared.gradio['model_draft'], lambda: None, lambda: {'choices': ['None'] + utils.get_available_models()}, 'refresh-button', interactive=not mu)
 
                                 shared.gradio['gpu_layers_draft'] = gr.Slider(label="gpu-layers-draft", minimum=0, maximum=256, value=shared.args.gpu_layers_draft, info='Number of layers to offload to the GPU for the draft model.')
                                 shared.gradio['draft_max'] = gr.Number(label="draft-max", precision=0, step=1, value=shared.args.draft_max, info='Number of tokens to draft for speculative decoding. Recommended value: 4.')

From d746484521c527f64b66264bc9d3ecc22b7461c2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 23 Jul 2025 11:52:51 -0700
Subject: [PATCH 0831/1701] Handle both int and str types in grammar char
 processing

---
 modules/grammar/grammar_utils.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/grammar/grammar_utils.py b/modules/grammar/grammar_utils.py
index 7f09ff82fb..af78f6b9cd 100644
--- a/modules/grammar/grammar_utils.py
+++ b/modules/grammar/grammar_utils.py
@@ -463,7 +463,7 @@ def __init__(self, grammar_str, start_rule_name, tokenizer):
         super().__init__(grammar_str, start_rule_name, tokenizer)
 
     def accept_char(self, char, stacks):
-        byte = ord(char)
+        byte = char if isinstance(char, int) else ord(char)
         new_stacks = []
         for stack in stacks:
             # stack is empty
@@ -549,7 +549,7 @@ def filter_vocab(self, stacks, device):
     # For each sub-rule in the grammar, cache whether each byte is accepted.
     @lru_cache(maxsize=None)
     def pos_char_acceptance(self, pos, char):
-        byte = ord(char)
+        byte = char if isinstance(char, int) else ord(char)
         num_chars = self.grammar_encoding[pos]
         pos += 1
         for i in range(0, num_chars, 2):

From f08bb9a2012eeac213232c2fe087ba330b1801fb Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 24 Jul 2025 10:34:59 -0700
Subject: [PATCH 0832/1701] Handle edge case in chat history loading (closes
 #7155)

---
 modules/chat.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index 827b60503d..1a16a6891e 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1175,6 +1175,9 @@ def save_last_chat_state(character, mode, unique_id):
 def load_history(unique_id, character, mode):
     p = get_history_file_path(unique_id, character, mode)
 
+    if not p.exists():
+        return {'internal': [], 'visible': [], 'metadata': {}}
+
     f = json.loads(open(p, 'rb').read())
     if 'internal' in f and 'visible' in f:
         history = f

From 74230f559ab5e8536ff22352c4910191667ab12c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 1 Aug 2025 11:03:15 -0700
Subject: [PATCH 0833/1701] Bump transformers to 4.54

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_cuda128.txt         | 2 +-
 requirements/full/requirements_cuda128_noavx2.txt  | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 11 files changed, 11 insertions(+), 11 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 687f1f5a82..9810c65abe 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -23,7 +23,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.53.*
+transformers==4.54.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 8224d98700..314e7d4f3d 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.53.*
+transformers==4.54.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 22141a8a5d..c7a8ba9b92 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.53.*
+transformers==4.54.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 04325464c4..6ce6ae9b30 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.53.*
+transformers==4.54.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 9497575fc4..531282109a 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.53.*
+transformers==4.54.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 8a84e4030d..6ba2900818 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.53.*
+transformers==4.54.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 9488f5e75d..a1bd0ffcf8 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.53.*
+transformers==4.54.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index a2af51086b..f21a5208da 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -23,7 +23,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.53.*
+transformers==4.54.*
 triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 948a275abd..e76d66683c 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -23,7 +23,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.53.*
+transformers==4.54.*
 triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 8f7106e4f0..5fbe49e72d 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -23,7 +23,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.53.*
+transformers==4.54.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 69a8218499..2158834472 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.53.*
+transformers==4.54.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb

From 02a3420a506631f50bdb3dbd3d6c22ef4344c343 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 10:09:30 -0700
Subject: [PATCH 0834/1701] Bump transformers to 4.55 (adds gpt-oss support)

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_cuda128.txt         | 2 +-
 requirements/full/requirements_cuda128_noavx2.txt  | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 11 files changed, 11 insertions(+), 11 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 9810c65abe..3a30a6c7dd 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -23,7 +23,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.54.*
+transformers==4.55.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 314e7d4f3d..ae269bc909 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.54.*
+transformers==4.55.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index c7a8ba9b92..14871b4bde 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.54.*
+transformers==4.55.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 6ce6ae9b30..493579394a 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.54.*
+transformers==4.55.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 531282109a..de33cdb8df 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.54.*
+transformers==4.55.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 6ba2900818..58496d9d1a 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.54.*
+transformers==4.55.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index a1bd0ffcf8..c532207698 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.54.*
+transformers==4.55.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index f21a5208da..804ef93497 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -23,7 +23,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.54.*
+transformers==4.55.*
 triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index e76d66683c..06d93d65f4 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -23,7 +23,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.54.*
+transformers==4.55.*
 triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 5fbe49e72d..f9e5fb73d5 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -23,7 +23,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.54.*
+transformers==4.55.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 2158834472..cd85a744e7 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -22,7 +22,7 @@ safetensors==0.5.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.54.*
+transformers==4.55.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb

From 59890435376110f58656037b1871be553bafbb1f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 11:22:18 -0700
Subject: [PATCH 0835/1701] Transformers: Support standalone .jinja chat
 templates (for GPT-OSS)

---
 modules/models_settings.py | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index a06e594e90..c4dfb0ed31 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -122,13 +122,25 @@ def get_model_metadata(model):
 
     # Try to find the Jinja instruct template
     path = Path(f'{shared.args.model_dir}/{model}') / 'tokenizer_config.json'
+    template = None
+
+    # 1. Prioritize reading from chat_template.jinja if it exists
+    jinja_path = Path(f'{shared.args.model_dir}/{model}') / 'chat_template.jinja'
+    if jinja_path.exists():
+        with open(jinja_path, 'r', encoding='utf-8') as f:
+            template = f.read()
+
     if path.exists():
         metadata = json.loads(open(path, 'r', encoding='utf-8').read())
-        if 'chat_template' in metadata:
+
+        # 2. Only read from metadata if we haven't already loaded from .jinja
+        if template is None and 'chat_template' in metadata:
             template = metadata['chat_template']
             if isinstance(template, list):
                 template = template[0]['template']
 
+        # 3. If a template was found from either source, process it
+        if template:
             for k in ['eos_token', 'bos_token']:
                 if k in metadata:
                     value = metadata[k]

From 3039aeffeb8958724de56912d0d90267b87a7074 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 11:35:17 -0700
Subject: [PATCH 0836/1701] Fix parsing the gpt-oss-20b template

---
 modules/models_settings.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index c4dfb0ed31..8ed7f953e3 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -90,7 +90,7 @@ def get_model_metadata(model):
             template = template.replace('eos_token', "'{}'".format(eos_token))
             template = template.replace('bos_token', "'{}'".format(bos_token))
 
-            template = re.sub(r'raise_exception\([^)]*\)', "''", template)
+            template = re.sub(r"\{\{-?\s*raise_exception\(.*?\)\s*-?\}\}", "", template, flags=re.DOTALL)
             template = re.sub(r'{% if add_generation_prompt %}.*', '', template, flags=re.DOTALL)
             model_settings['instruction_template'] = 'Custom (obtained from model metadata)'
             model_settings['instruction_template_str'] = template
@@ -149,7 +149,7 @@ def get_model_metadata(model):
 
                     template = template.replace(k, "'{}'".format(value))
 
-            template = re.sub(r'raise_exception\([^)]*\)', "''", template)
+            template = re.sub(r"\{\{-?\s*raise_exception\(.*?\)\s*-?\}\}", "", template, flags=re.DOTALL)
             template = re.sub(r'{% if add_generation_prompt %}.*', '', template, flags=re.DOTALL)
             model_settings['instruction_template'] = 'Custom (obtained from model metadata)'
             model_settings['instruction_template_str'] = template

From 3b28dc182186308648af5cddb44c811f1608c70d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 11:35:53 -0700
Subject: [PATCH 0837/1701] Don't pass torch_dtype to transformers loader, let
 it be autodetected

---
 modules/transformers_loader.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/modules/transformers_loader.py b/modules/transformers_loader.py
index 2f7367a4d0..e407212561 100644
--- a/modules/transformers_loader.py
+++ b/modules/transformers_loader.py
@@ -136,7 +136,6 @@ def load_model_HF(model_name):
     path_to_model = Path(f'{shared.args.model_dir}/{model_name}')
     params = {
         'low_cpu_mem_usage': True,
-        'torch_dtype': torch.bfloat16 if shared.args.bf16 else torch.float16,
         'attn_implementation': shared.args.attn_implementation,
     }
 

From 9f28f53cfc7d14cf8e3c9ebd00834126f73675b4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 11:56:00 -0700
Subject: [PATCH 0838/1701] Better parsing of the gpt-oss template

---
 modules/chat.py | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 1a16a6891e..c10d91a71a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -108,7 +108,14 @@ def get_generation_prompt(renderer, impersonate=False, strip_trailing_spaces=Tru
 
     suffix_plus_prefix = prompt.split("<<|user-message-1|>>")[1].split("<<|user-message-2|>>")[0]
     suffix = prompt.split("<<|user-message-2|>>")[1]
-    prefix = suffix_plus_prefix[len(suffix):]
+
+    # Remove the message suffix. The first case handles the GPT-OSS model
+    # in a way that is likely to not interfere with previous models.
+    if '<|start|>user' in suffix_plus_prefix or '<|start|>assistant' in suffix_plus_prefix:
+        start_index = suffix_plus_prefix.rindex('<|start|>')
+        prefix = suffix_plus_prefix[start_index:]
+    else:
+        prefix = suffix_plus_prefix[len(suffix):]
 
     if strip_trailing_spaces:
         prefix = prefix.rstrip(' ')

From 178c3e75cca827657a018a64ae3d7945d9e25231 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 12:38:06 -0700
Subject: [PATCH 0839/1701] Handle templates with channels separately

---
 modules/chat.py | 184 +++++++++++++++++++++++++++++++++++++++++-------
 1 file changed, 157 insertions(+), 27 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index c10d91a71a..f929f65359 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -86,6 +86,134 @@ def str_presenter(dumper, data):
 yaml.representer.SafeRepresenter.add_representer(str, str_presenter)
 
 
+# Template Handler Classes
+class TemplateHandler:
+    """Base class for handling different template types"""
+
+    def __init__(self, template_str):
+        self.template_str = template_str
+
+    def get_generation_prefix_suffix(self, renderer, impersonate=False, strip_trailing_spaces=True):
+        """Get prefix/suffix for generation"""
+        return "", ""
+
+    def get_stopping_strings(self, renderer):
+        """Get stopping strings for this template type"""
+        return []
+
+    def modify_for_continue(self, prompt, renderer, impersonate=False):
+        """Modify prompt for continue mode"""
+        return prompt
+
+    def supports_impersonate(self):
+        """Whether impersonate mode is supported"""
+        return False
+
+
+class LinearTemplateHandler(TemplateHandler):
+    """Handles traditional linear templates"""
+
+    def get_generation_prefix_suffix(self, renderer, impersonate=False, strip_trailing_spaces=True):
+        # This is the original, complex logic for deriving prefix/suffix for old templates.
+        if impersonate:
+            messages = [
+                {"role": "user", "content": "<<|user-message-1|>>"},
+                {"role": "user", "content": "<<|user-message-2|>>"},
+            ]
+        else:
+            messages = [
+                {"role": "assistant", "content": "<<|user-message-1|>>"},
+                {"role": "assistant", "content": "<<|user-message-2|>>"},
+            ]
+
+        prompt = renderer(messages=messages)
+        suffix_plus_prefix = prompt.split("<<|user-message-1|>>")[1].split("<<|user-message-2|>>")[0]
+        suffix = prompt.split("<<|user-message-2|>>")[1]
+
+        if '<|start|>user' in suffix_plus_prefix or '<|start|>assistant' in suffix_plus_prefix:
+            start_index = suffix_plus_prefix.rindex('<|start|>')
+            prefix = suffix_plus_prefix[start_index:]
+        else:
+            prefix = suffix_plus_prefix[len(suffix):]
+
+        if strip_trailing_spaces:
+            prefix = prefix.rstrip(' ')
+
+        return prefix, suffix
+
+    def get_stopping_strings(self, renderer):
+        # This is the original, correct logic for dynamically creating stopping strings for linear templates.
+        prefix_bot, suffix_bot = self.get_generation_prefix_suffix(renderer, impersonate=False)
+        prefix_user, suffix_user = self.get_generation_prefix_suffix(renderer, impersonate=True)
+
+        stopping_strings = [
+            suffix_user + prefix_bot,
+            suffix_user + prefix_user,
+            suffix_bot + prefix_bot,
+            suffix_bot + prefix_user,
+        ]
+
+        # Attempt to find a single EOT token to use as a stop string
+        for item in stopping_strings:
+            item = item.strip()
+            if item.startswith("<") and ">" in item:
+                stopping_strings.append(item.split(">")[0] + ">")
+                break
+            elif item.startswith("[") and "]" in item:
+                stopping_strings.append(item.split("]")[0] + "]")
+                break
+
+        return stopping_strings
+
+    def modify_for_continue(self, prompt, renderer, impersonate=False):
+        suffix = self.get_generation_prefix_suffix(renderer, impersonate)[1]
+        if len(suffix) > 0:
+            return prompt[:-len(suffix)]
+        return prompt
+
+    def supports_impersonate(self):
+        return True
+
+
+class ChannelTemplateHandler(TemplateHandler):
+    """Handles channel-based templates"""
+
+    def get_generation_prefix_suffix(self, renderer, impersonate=False, strip_trailing_spaces=True):
+        """
+        Gets the string to add to the prompt to start a new generation.
+        """
+        dummy_message = [{'role': 'user', 'content': '...'}]
+        prompt_without_gen = renderer(messages=dummy_message, add_generation_prompt=False)
+        prompt_with_gen = renderer(messages=dummy_message, add_generation_prompt=True)
+        generation_prompt = prompt_with_gen[len(prompt_without_gen):]
+
+        if strip_trailing_spaces:
+            generation_prompt = generation_prompt.rstrip(' ')
+
+        return generation_prompt, ""
+
+    def get_stopping_strings(self, renderer):
+        return [
+            '<|return|>',
+            '<|start|>user',
+            '<|start|>developer',
+            '<|call|>'
+        ]
+
+    def modify_for_continue(self, prompt, renderer, impersonate=False):
+        return prompt
+
+    def supports_impersonate(self):
+        return False
+
+
+def create_template_handler(template_str):
+    """Factory function to create appropriate handler"""
+    if '<|channel|>' in template_str:
+        return ChannelTemplateHandler(template_str)
+    return LinearTemplateHandler(template_str)
+
+
 def get_generation_prompt(renderer, impersonate=False, strip_trailing_spaces=True):
     '''
     Given a Jinja template, reverse-engineers the prefix and the suffix for
@@ -270,6 +398,15 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
         messages.append({"role": "user", "content": user_input})
 
+    # Create template handler based on current template
+    template_str = state['instruction_template_str'] if state['mode'] == 'instruct' else chat_template_str
+    handler = create_template_handler(template_str)
+
+    # Check impersonate support early
+    if impersonate and not handler.supports_impersonate():
+        logger.warning("Impersonate not supported for channel-based templates")
+        return ""
+
     def make_prompt(messages):
         if state['mode'] == 'chat-instruct' and _continue:
             prompt = renderer(messages=messages[:-1])
@@ -287,10 +424,10 @@ def make_prompt(messages):
             command = replace_character_names(command, state['name1'], state['name2'])
 
             if _continue:
-                prefix = get_generation_prompt(renderer, impersonate=impersonate, strip_trailing_spaces=False)[0]
+                prefix = handler.get_generation_prefix_suffix(renderer, impersonate=impersonate, strip_trailing_spaces=False)[0]
                 prefix += messages[-1]["content"]
             else:
-                prefix = get_generation_prompt(renderer, impersonate=impersonate)[0]
+                prefix = handler.get_generation_prefix_suffix(renderer, impersonate=impersonate)[0]
                 if not impersonate:
                     prefix = apply_extensions('bot_prefix', prefix, state)
 
@@ -298,16 +435,14 @@ def make_prompt(messages):
             outer_messages.append({"role": "assistant", "content": prefix})
 
             prompt = instruct_renderer(messages=outer_messages)
-            suffix = get_generation_prompt(instruct_renderer, impersonate=False)[1]
+            suffix = handler.get_generation_prefix_suffix(instruct_renderer, impersonate=False)[1]
             if len(suffix) > 0:
                 prompt = prompt[:-len(suffix)]
         else:
             if _continue:
-                suffix = get_generation_prompt(renderer, impersonate=impersonate)[1]
-                if len(suffix) > 0:
-                    prompt = prompt[:-len(suffix)]
+                prompt = handler.modify_for_continue(prompt, renderer, impersonate)
             else:
-                prefix = get_generation_prompt(renderer, impersonate=impersonate)[0]
+                prefix = handler.get_generation_prefix_suffix(renderer, impersonate=impersonate)[0]
                 if state['mode'] == 'chat' and not impersonate:
                     prefix = apply_extensions('bot_prefix', prefix, state)
 
@@ -433,31 +568,16 @@ def get_stopping_strings(state):
     if state['mode'] in ['instruct', 'chat-instruct']:
         template = jinja_env.from_string(state['instruction_template_str'])
         renderer = partial(template.render, add_generation_prompt=False)
-        renderers.append(renderer)
+        renderers.append((renderer, state['instruction_template_str']))
 
     if state['mode'] in ['chat', 'chat-instruct']:
         template = jinja_env.from_string(state['chat_template_str'])
         renderer = partial(template.render, add_generation_prompt=False, name1=state['name1'], name2=state['name2'])
-        renderers.append(renderer)
+        renderers.append((renderer, state['chat_template_str']))
 
-    for renderer in renderers:
-        prefix_bot, suffix_bot = get_generation_prompt(renderer, impersonate=False)
-        prefix_user, suffix_user = get_generation_prompt(renderer, impersonate=True)
-
-        stopping_strings += [
-            suffix_user + prefix_bot,
-            suffix_user + prefix_user,
-            suffix_bot + prefix_bot,
-            suffix_bot + prefix_user,
-        ]
-
-    # Try to find the EOT token
-    for item in stopping_strings.copy():
-        item = item.strip()
-        if item.startswith("<") and ">" in item:
-            stopping_strings.append(item.split(">")[0] + ">")
-        elif item.startswith("[") and "]" in item:
-            stopping_strings.append(item.split("]")[0] + "]")
+    for renderer, template_str in renderers:
+        handler = create_template_handler(template_str)
+        stopping_strings += handler.get_stopping_strings(renderer)
 
     if 'stopping_strings' in state and isinstance(state['stopping_strings'], list):
         stopping_strings += state.pop('stopping_strings')
@@ -809,6 +929,16 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
 
 
 def impersonate_wrapper(textbox, state):
+    # Check template support first
+    template_str = state['chat_template_str']
+    handler = create_template_handler(template_str)
+
+    if not handler.supports_impersonate():
+        logger.warning("Impersonate not supported for channel-based templates")
+        static_output = chat_html_wrapper(state['history'], state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
+        yield textbox, static_output
+        return
+
     text = textbox['text']
     static_output = chat_html_wrapper(state['history'], state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
 

From ecd16d6bf9f680ba5b25eb837bf61569dde81886 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 12:57:49 -0700
Subject: [PATCH 0840/1701] Automatically set skip_special_tokens to False for
 channel-based templates

---
 modules/chat.py | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index f929f65359..46d24a6ff2 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -770,6 +770,16 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
     output = apply_extensions('history', output)
     state = apply_extensions('state', state)
 
+    # Automatically set skip_special_tokens to False for channel-based templates
+    if state['mode'] in ['instruct', 'chat-instruct']:
+        template_str = state['instruction_template_str']
+    else:  # chat mode
+        template_str = state['chat_template_str']
+
+    handler = create_template_handler(template_str)
+    if isinstance(handler, ChannelTemplateHandler):
+        state['skip_special_tokens'] = False
+
     # Let the jinja2 template handle the BOS token
     if state['mode'] in ['instruct', 'chat-instruct']:
         state['add_bos_token'] = False

From 5c5a4dfc140d3e6558c97fc84c430faa2444ef28 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 13:03:18 -0700
Subject: [PATCH 0841/1701] Fix impersonate

---
 modules/chat.py | 43 ++++++++++++++-----------------------------
 1 file changed, 14 insertions(+), 29 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 46d24a6ff2..043908c9c4 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -105,10 +105,6 @@ def modify_for_continue(self, prompt, renderer, impersonate=False):
         """Modify prompt for continue mode"""
         return prompt
 
-    def supports_impersonate(self):
-        """Whether impersonate mode is supported"""
-        return False
-
 
 class LinearTemplateHandler(TemplateHandler):
     """Handles traditional linear templates"""
@@ -171,41 +167,41 @@ def modify_for_continue(self, prompt, renderer, impersonate=False):
             return prompt[:-len(suffix)]
         return prompt
 
-    def supports_impersonate(self):
-        return True
-
 
 class ChannelTemplateHandler(TemplateHandler):
     """Handles channel-based templates"""
 
     def get_generation_prefix_suffix(self, renderer, impersonate=False, strip_trailing_spaces=True):
         """
-        Gets the string to add to the prompt to start a new generation.
+        Gets the string to add to the prompt to start a new turn.
         """
-        dummy_message = [{'role': 'user', 'content': '...'}]
-        prompt_without_gen = renderer(messages=dummy_message, add_generation_prompt=False)
-        prompt_with_gen = renderer(messages=dummy_message, add_generation_prompt=True)
-        generation_prompt = prompt_with_gen[len(prompt_without_gen):]
+        if impersonate:
+            # For impersonate mode, we need the prefix for a user's turn.
+            prefix = "<|start|>user<|message|>"
+        else:
+            # For a normal reply, we need the prefix for the assistant's turn.
+            prefix = "<|start|>assistant"
 
         if strip_trailing_spaces:
-            generation_prompt = generation_prompt.rstrip(' ')
+            prefix = prefix.rstrip(' ')
 
-        return generation_prompt, ""
+        # The suffix is not needed for this template type's generation logic.
+        return prefix, ""
 
     def get_stopping_strings(self, renderer):
+        # Use specific tokens that unambiguously signal the end of a turn
+        # or the start of a different character's turn.
         return [
             '<|return|>',
             '<|start|>user',
             '<|start|>developer',
-            '<|call|>'
+            '<|call|>',
         ]
 
     def modify_for_continue(self, prompt, renderer, impersonate=False):
+        # Channels don't need suffix stripping for the continue logic to work.
         return prompt
 
-    def supports_impersonate(self):
-        return False
-
 
 def create_template_handler(template_str):
     """Factory function to create appropriate handler"""
@@ -402,11 +398,6 @@ def generate_chat_prompt(user_input, state, **kwargs):
     template_str = state['instruction_template_str'] if state['mode'] == 'instruct' else chat_template_str
     handler = create_template_handler(template_str)
 
-    # Check impersonate support early
-    if impersonate and not handler.supports_impersonate():
-        logger.warning("Impersonate not supported for channel-based templates")
-        return ""
-
     def make_prompt(messages):
         if state['mode'] == 'chat-instruct' and _continue:
             prompt = renderer(messages=messages[:-1])
@@ -943,12 +934,6 @@ def impersonate_wrapper(textbox, state):
     template_str = state['chat_template_str']
     handler = create_template_handler(template_str)
 
-    if not handler.supports_impersonate():
-        logger.warning("Impersonate not supported for channel-based templates")
-        static_output = chat_html_wrapper(state['history'], state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
-        yield textbox, static_output
-        return
-
     text = textbox['text']
     static_output = chat_html_wrapper(state['history'], state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
 

From 42e3a7a5ae7011d12987f23056220c506af69af6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 14:56:12 -0700
Subject: [PATCH 0842/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_cuda128.txt             | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt      | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 3a30a6c7dd..dd1e8d35cc 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index ae269bc909..b65f0b09f2 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 14871b4bde..6e6986540b 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 493579394a..84abd394b6 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index de33cdb8df..2deefbc4b6 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 58496d9d1a..8c1baf04e1 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index c532207698..67a44432cc 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 804ef93497..9fe3c54b2a 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 06d93d65f4..50602d8d89 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index f9e5fb73d5..abdcfc16fc 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 53479a80f9..30d7d9e43e 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index d7336d2fcd..a7c7808aec 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 1edaa5156b..b1f66f566e 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -18,6 +18,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 04c9b28343..76530338b9 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 3c3563d324..26235b83b7 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index cf0d7b11e5..cfa7631095 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 9bd8a37ce5..2f8c401d35 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index b8519553fc..e065057521 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.31.0/llama_cpp_binaries-0.31.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 6bb8212731db0dddb00d10494e56223718401d4c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 15:06:22 -0700
Subject: [PATCH 0843/1701] Fix thinking block rendering for GPT-OSS

---
 modules/html_generator.py | 63 ++++++++++++++++++++++++++++-----------
 1 file changed, 45 insertions(+), 18 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 6844c244d2..8777acf739 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -116,29 +116,56 @@ def extract_thinking_block(string):
     THINK_START_TAG = "<think>"
     THINK_END_TAG = "</think>"
 
-    # Look for think tag
+    # Look for think tag first
     start_pos = string.find(THINK_START_TAG)
     end_pos = string.find(THINK_END_TAG)
 
-    # Return if neither tag is in string
-    if start_pos == -1 and end_pos == -1:
-        return None, string
+    # If think tags found, use existing logic
+    if start_pos != -1 or end_pos != -1:
+        # handle missing start or end tags
+        if start_pos == -1:
+            thought_start = 0
+        else:
+            thought_start = start_pos + len(THINK_START_TAG)
+        if end_pos == -1:
+            thought_end = len(string)
+            content_start = len(string)
+        else:
+            thought_end = end_pos
+            content_start = end_pos + len(THINK_END_TAG)
+        thinking_content = string[thought_start:thought_end]
+        remaining_content = string[content_start:]
+        return thinking_content, remaining_content
+
+    # If think tags not found, try alternative format
+    ALT_START = "<|channel|>analysis<|message|>"
+    ALT_END = "<|end|>"
+    ALT_CONTENT_START = "<|start|>assistant<|channel|>final<|message|>"
+
+    alt_start_pos = string.find(ALT_START)
+    alt_end_pos = string.find(ALT_END)
+    alt_content_pos = string.find(ALT_CONTENT_START)
+
+    # Check if start tag or end tag is found
+    if alt_start_pos != -1 or alt_end_pos != -1:
+        if alt_start_pos == -1:
+            thought_start = 0
+        else:
+            thought_start = alt_start_pos + len(ALT_START)
 
-    # handle missing start or end tags
-    if start_pos == -1:
-        thought_start = 0
-    else:
-        thought_start = start_pos + len(THINK_START_TAG)
-    if end_pos == -1:
-        thought_end = len(string)
-        content_start = len(string)
-    else:
-        thought_end = end_pos
-        content_start = end_pos + len(THINK_END_TAG)
+        if alt_end_pos == -1:
+            thought_end = len(string)
+            content_start = len(string)
+        else:
+            thought_end = alt_end_pos
+            content_start = alt_content_pos + len(ALT_CONTENT_START) if alt_content_pos != -1 else len(string)
+
+        thinking_content = string[thought_start:thought_end]
+        remaining_content = string[content_start:]
+        return thinking_content, remaining_content
 
-    thinking_content = string[thought_start:thought_end]
-    remaining_content = string[content_start:]
-    return thinking_content, remaining_content
+    # Return if neither format is found
+    return None, string
 
 
 @functools.lru_cache(maxsize=None)

From 498778b8ac85990158713c1925ca657d1fa135c8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 15:19:11 -0700
Subject: [PATCH 0844/1701] Add a new 'Reasoning effort' UI element

---
 extensions/openai/typing.py | 1 +
 modules/chat.py             | 3 ++-
 modules/loaders.py          | 5 +++++
 modules/shared.py           | 1 +
 modules/ui.py               | 2 ++
 modules/ui_chat.py          | 3 ++-
 6 files changed, 13 insertions(+), 2 deletions(-)

diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 6643ed16fb..6bd3749ff6 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -43,6 +43,7 @@ class GenerationOptions(BaseModel):
     ban_eos_token: bool = False
     add_bos_token: bool = True
     enable_thinking: bool = True
+    reasoning_effort: str = "medium"
     skip_special_tokens: bool = True
     static_cache: bool = False
     truncation_length: int = 0
diff --git a/modules/chat.py b/modules/chat.py
index 043908c9c4..dd923d670a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -306,7 +306,8 @@ def generate_chat_prompt(user_input, state, **kwargs):
         builtin_tools=None,
         tools=state['tools'] if 'tools' in state else None,
         tools_in_user_message=False,
-        add_generation_prompt=False
+        add_generation_prompt=False,
+        reasoning_effort=state.get('reasoning_effort', 'medium')
     )
 
     chat_renderer = partial(
diff --git a/modules/loaders.py b/modules/loaders.py
index f515aeca6c..7546bc5ba4 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -137,6 +137,7 @@ def transformers_samplers():
         'ban_eos_token',
         'add_bos_token',
         'enable_thinking',
+        'reasoning_effort',
         'skip_special_tokens',
         'static_cache',
         'seed',
@@ -189,6 +190,7 @@ def transformers_samplers():
         'ban_eos_token',
         'add_bos_token',
         'enable_thinking',
+        'reasoning_effort',
         'skip_special_tokens',
         'seed',
         'sampler_priority',
@@ -236,6 +238,7 @@ def transformers_samplers():
         'ban_eos_token',
         'add_bos_token',
         'enable_thinking',
+        'reasoning_effort',
         'skip_special_tokens',
         'seed',
         'sampler_priority',
@@ -275,6 +278,7 @@ def transformers_samplers():
         'ban_eos_token',
         'add_bos_token',
         'enable_thinking',
+        'reasoning_effort',
         'skip_special_tokens',
         'seed',
         'custom_token_bans',
@@ -308,6 +312,7 @@ def transformers_samplers():
         'ban_eos_token',
         'add_bos_token',
         'enable_thinking',
+        'reasoning_effort',
         'seed',
         'sampler_priority',
         'dry_sequence_breakers',
diff --git a/modules/shared.py b/modules/shared.py
index 5e3e11c059..ab5198d1f5 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -211,6 +211,7 @@
     'ban_eos_token': False,
     'add_bos_token': True,
     'enable_thinking': True,
+    'reasoning_effort': 'medium',
     'skip_special_tokens': True,
     'stream': True,
     'static_cache': False,
diff --git a/modules/ui.py b/modules/ui.py
index 98acc03861..e7805046cc 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -215,6 +215,7 @@ def list_interface_input_elements():
         'ban_eos_token',
         'add_bos_token',
         'enable_thinking',
+        'reasoning_effort',
         'skip_special_tokens',
         'stream',
         'static_cache',
@@ -482,6 +483,7 @@ def setup_auto_save():
         'ban_eos_token',
         'add_bos_token',
         'enable_thinking',
+        'reasoning_effort',
         'skip_special_tokens',
         'stream',
         'static_cache',
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 4dade1765e..1d85a39863 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -78,7 +78,8 @@ def create_ui():
                 with gr.Row():
                     shared.gradio['start_with'] = gr.Textbox(label='Start reply with', placeholder='Sure thing!', value=shared.settings['start_with'], elem_classes=['add_scrollbar'])
 
-                shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', info='Used by Qwen3 to toggle  mode.')
+                shared.gradio['reasoning_effort'] = gr.Dropdown(value=shared.settings['reasoning_effort'], choices=['low', 'medium', 'high'], label='Reasoning effort', info='Used by GPT-OSS.')
+                shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', info='Used by pre-2507 Qwen3.')
                 shared.gradio['enable_web_search'] = gr.Checkbox(value=shared.settings.get('enable_web_search', False), label='Activate web search', elem_id='web-search')
                 with gr.Row(visible=shared.settings.get('enable_web_search', False)) as shared.gradio['web_search_row']:
                     shared.gradio['web_search_pages'] = gr.Number(value=shared.settings.get('web_search_pages', 3), precision=0, label='Number of pages to download', minimum=1, maximum=10)

From 0e42575c57b374e2b652b15cc2e03daec3170bc6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 15:36:20 -0700
Subject: [PATCH 0845/1701] Fix thinking block parsing for GPT-OSS under
 llama.cpp

---
 modules/html_generator.py | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 8777acf739..79237f7faf 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -146,19 +146,23 @@ def extract_thinking_block(string):
     alt_end_pos = string.find(ALT_END)
     alt_content_pos = string.find(ALT_CONTENT_START)
 
-    # Check if start tag or end tag is found
     if alt_start_pos != -1 or alt_end_pos != -1:
         if alt_start_pos == -1:
             thought_start = 0
         else:
             thought_start = alt_start_pos + len(ALT_START)
 
+        # If no explicit end tag but content start exists, use content start as end
         if alt_end_pos == -1:
-            thought_end = len(string)
-            content_start = len(string)
+            if alt_content_pos != -1:
+                thought_end = alt_content_pos
+                content_start = alt_content_pos + len(ALT_CONTENT_START)
+            else:
+                thought_end = len(string)
+                content_start = len(string)
         else:
             thought_end = alt_end_pos
-            content_start = alt_content_pos + len(ALT_CONTENT_START) if alt_content_pos != -1 else len(string)
+            content_start = alt_content_pos + len(ALT_CONTENT_START) if alt_content_pos != -1 else alt_end_pos + len(ALT_END)
 
         thinking_content = string[thought_start:thought_end]
         remaining_content = string[content_start:]

From 7d98ca619558c9b77fa26130e3656b0bf8843341 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 15:43:44 -0700
Subject: [PATCH 0846/1701] Make web search functional with thinking models

---
 modules/chat.py | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index dd923d670a..dbc0e6f6a8 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -307,7 +307,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
         tools=state['tools'] if 'tools' in state else None,
         tools_in_user_message=False,
         add_generation_prompt=False,
-        reasoning_effort=state.get('reasoning_effort', 'medium')
+        reasoning_effort=state['reasoning_effort'])
     )
 
     chat_renderer = partial(
@@ -730,9 +730,9 @@ def generate_search_query(user_message, state):
 
     # Use a minimal state for search query generation but keep the full history
     search_state = state.copy()
-    search_state['max_new_tokens'] = 64
-    search_state['auto_max_new_tokens'] = False
+    search_state['auto_max_new_tokens'] = True
     search_state['enable_thinking'] = False
+    search_state['reasoning_effort'] = 'low'
     search_state['start_with'] = ""
 
     # Generate the full prompt using existing history + augmented message
@@ -742,6 +742,12 @@ def generate_search_query(user_message, state):
     for reply in generate_reply(formatted_prompt, search_state, stopping_strings=[], is_chat=True):
         query = reply
 
+    # Check for thinking block delimiters and extract content after them
+    if "" in query:
+        query = query.rsplit("            ", 1)[1]
+    elif "<|start|>assistant<|channel|>final<|message|>" in query:
+        query = query.rsplit("<|start|>assistant<|channel|>final<|message|>", 1)[1]
+
     # Strip and remove surrounding quotes if present
     query = query.strip()
     if len(query) >= 2 and query.startswith('"') and query.endswith('"'):

From 701048cf336946177ea216d3456e5f7cdd1cab85 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 15:51:24 -0700
Subject: [PATCH 0847/1701] Try to avoid breaking jinja2 parsing for older
 models

---
 modules/models_settings.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 8ed7f953e3..3c068df0b9 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -91,6 +91,7 @@ def get_model_metadata(model):
             template = template.replace('bos_token', "'{}'".format(bos_token))
 
             template = re.sub(r"\{\{-?\s*raise_exception\(.*?\)\s*-?\}\}", "", template, flags=re.DOTALL)
+            template = re.sub(r'raise_exception\([^)]*\)', "''", template)
             template = re.sub(r'{% if add_generation_prompt %}.*', '', template, flags=re.DOTALL)
             model_settings['instruction_template'] = 'Custom (obtained from model metadata)'
             model_settings['instruction_template_str'] = template
@@ -150,6 +151,7 @@ def get_model_metadata(model):
                     template = template.replace(k, "'{}'".format(value))
 
             template = re.sub(r"\{\{-?\s*raise_exception\(.*?\)\s*-?\}\}", "", template, flags=re.DOTALL)
+            template = re.sub(r'raise_exception\([^)]*\)', "''", template)
             template = re.sub(r'{% if add_generation_prompt %}.*', '', template, flags=re.DOTALL)
             model_settings['instruction_template'] = 'Custom (obtained from model metadata)'
             model_settings['instruction_template_str'] = template

From e5b8d4d072f74281071a9ad911bede662d61767e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 15:52:56 -0700
Subject: [PATCH 0848/1701] Fix a typo

---
 modules/chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index dbc0e6f6a8..1f4e2af000 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -307,7 +307,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
         tools=state['tools'] if 'tools' in state else None,
         tools_in_user_message=False,
         add_generation_prompt=False,
-        reasoning_effort=state['reasoning_effort'])
+        reasoning_effort=state['reasoning_effort']
     )
 
     chat_renderer = partial(

From 80f6abb07e44cb70d65a0d43fe9676a02880eb2c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 16:01:19 -0700
Subject: [PATCH 0849/1701] Begin fixing 'Continue' with GPT-OSS

---
 modules/chat.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 1f4e2af000..b23340aa46 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -165,6 +165,7 @@ def modify_for_continue(self, prompt, renderer, impersonate=False):
         suffix = self.get_generation_prefix_suffix(renderer, impersonate)[1]
         if len(suffix) > 0:
             return prompt[:-len(suffix)]
+
         return prompt
 
 
@@ -199,7 +200,10 @@ def get_stopping_strings(self, renderer):
         ]
 
     def modify_for_continue(self, prompt, renderer, impersonate=False):
-        # Channels don't need suffix stripping for the continue logic to work.
+        suffix = '<|return|>'
+        if prompt.endswith(suffix):
+            return prompt[:-len(suffix)]
+
         return prompt
 
 

From 20adc3c96737e35b96f6b1d557a63b1d2c75a825 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 16:58:45 -0700
Subject: [PATCH 0850/1701] Start over new template handling (to avoid
 overcomplicating)

---
 modules/chat.py | 192 +++++++-----------------------------------------
 1 file changed, 28 insertions(+), 164 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index b23340aa46..82760cc8cd 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -86,134 +86,6 @@ def str_presenter(dumper, data):
 yaml.representer.SafeRepresenter.add_representer(str, str_presenter)
 
 
-# Template Handler Classes
-class TemplateHandler:
-    """Base class for handling different template types"""
-
-    def __init__(self, template_str):
-        self.template_str = template_str
-
-    def get_generation_prefix_suffix(self, renderer, impersonate=False, strip_trailing_spaces=True):
-        """Get prefix/suffix for generation"""
-        return "", ""
-
-    def get_stopping_strings(self, renderer):
-        """Get stopping strings for this template type"""
-        return []
-
-    def modify_for_continue(self, prompt, renderer, impersonate=False):
-        """Modify prompt for continue mode"""
-        return prompt
-
-
-class LinearTemplateHandler(TemplateHandler):
-    """Handles traditional linear templates"""
-
-    def get_generation_prefix_suffix(self, renderer, impersonate=False, strip_trailing_spaces=True):
-        # This is the original, complex logic for deriving prefix/suffix for old templates.
-        if impersonate:
-            messages = [
-                {"role": "user", "content": "<<|user-message-1|>>"},
-                {"role": "user", "content": "<<|user-message-2|>>"},
-            ]
-        else:
-            messages = [
-                {"role": "assistant", "content": "<<|user-message-1|>>"},
-                {"role": "assistant", "content": "<<|user-message-2|>>"},
-            ]
-
-        prompt = renderer(messages=messages)
-        suffix_plus_prefix = prompt.split("<<|user-message-1|>>")[1].split("<<|user-message-2|>>")[0]
-        suffix = prompt.split("<<|user-message-2|>>")[1]
-
-        if '<|start|>user' in suffix_plus_prefix or '<|start|>assistant' in suffix_plus_prefix:
-            start_index = suffix_plus_prefix.rindex('<|start|>')
-            prefix = suffix_plus_prefix[start_index:]
-        else:
-            prefix = suffix_plus_prefix[len(suffix):]
-
-        if strip_trailing_spaces:
-            prefix = prefix.rstrip(' ')
-
-        return prefix, suffix
-
-    def get_stopping_strings(self, renderer):
-        # This is the original, correct logic for dynamically creating stopping strings for linear templates.
-        prefix_bot, suffix_bot = self.get_generation_prefix_suffix(renderer, impersonate=False)
-        prefix_user, suffix_user = self.get_generation_prefix_suffix(renderer, impersonate=True)
-
-        stopping_strings = [
-            suffix_user + prefix_bot,
-            suffix_user + prefix_user,
-            suffix_bot + prefix_bot,
-            suffix_bot + prefix_user,
-        ]
-
-        # Attempt to find a single EOT token to use as a stop string
-        for item in stopping_strings:
-            item = item.strip()
-            if item.startswith("<") and ">" in item:
-                stopping_strings.append(item.split(">")[0] + ">")
-                break
-            elif item.startswith("[") and "]" in item:
-                stopping_strings.append(item.split("]")[0] + "]")
-                break
-
-        return stopping_strings
-
-    def modify_for_continue(self, prompt, renderer, impersonate=False):
-        suffix = self.get_generation_prefix_suffix(renderer, impersonate)[1]
-        if len(suffix) > 0:
-            return prompt[:-len(suffix)]
-
-        return prompt
-
-
-class ChannelTemplateHandler(TemplateHandler):
-    """Handles channel-based templates"""
-
-    def get_generation_prefix_suffix(self, renderer, impersonate=False, strip_trailing_spaces=True):
-        """
-        Gets the string to add to the prompt to start a new turn.
-        """
-        if impersonate:
-            # For impersonate mode, we need the prefix for a user's turn.
-            prefix = "<|start|>user<|message|>"
-        else:
-            # For a normal reply, we need the prefix for the assistant's turn.
-            prefix = "<|start|>assistant"
-
-        if strip_trailing_spaces:
-            prefix = prefix.rstrip(' ')
-
-        # The suffix is not needed for this template type's generation logic.
-        return prefix, ""
-
-    def get_stopping_strings(self, renderer):
-        # Use specific tokens that unambiguously signal the end of a turn
-        # or the start of a different character's turn.
-        return [
-            '<|return|>',
-            '<|start|>user',
-            '<|start|>developer',
-            '<|call|>',
-        ]
-
-    def modify_for_continue(self, prompt, renderer, impersonate=False):
-        suffix = '<|return|>'
-        if prompt.endswith(suffix):
-            return prompt[:-len(suffix)]
-
-        return prompt
-
-
-def create_template_handler(template_str):
-    """Factory function to create appropriate handler"""
-    if '<|channel|>' in template_str:
-        return ChannelTemplateHandler(template_str)
-    return LinearTemplateHandler(template_str)
-
-
 def get_generation_prompt(renderer, impersonate=False, strip_trailing_spaces=True):
     '''
     Given a Jinja template, reverse-engineers the prefix and the suffix for
@@ -236,14 +108,7 @@ def get_generation_prompt(renderer, impersonate=False, strip_trailing_spaces=Tru
 
     suffix_plus_prefix = prompt.split("<<|user-message-1|>>")[1].split("<<|user-message-2|>>")[0]
     suffix = prompt.split("<<|user-message-2|>>")[1]
-
-    # Remove the message suffix. The first case handles the GPT-OSS model
-    # in a way that is likely to not interfere with previous models.
-    if '<|start|>user' in suffix_plus_prefix or '<|start|>assistant' in suffix_plus_prefix:
-        start_index = suffix_plus_prefix.rindex('<|start|>')
-        prefix = suffix_plus_prefix[start_index:]
-    else:
-        prefix = suffix_plus_prefix[len(suffix):]
+    prefix = suffix_plus_prefix[len(suffix):]
 
     if strip_trailing_spaces:
         prefix = prefix.rstrip(' ')
@@ -399,10 +264,6 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
         messages.append({"role": "user", "content": user_input})
 
-    # Create template handler based on current template
-    template_str = state['instruction_template_str'] if state['mode'] == 'instruct' else chat_template_str
-    handler = create_template_handler(template_str)
-
     def make_prompt(messages):
         if state['mode'] == 'chat-instruct' and _continue:
             prompt = renderer(messages=messages[:-1])
@@ -420,10 +281,10 @@ def make_prompt(messages):
             command = replace_character_names(command, state['name1'], state['name2'])
 
             if _continue:
-                prefix = handler.get_generation_prefix_suffix(renderer, impersonate=impersonate, strip_trailing_spaces=False)[0]
+                prefix = get_generation_prompt(renderer, impersonate=impersonate, strip_trailing_spaces=False)[0]
                 prefix += messages[-1]["content"]
             else:
-                prefix = handler.get_generation_prefix_suffix(renderer, impersonate=impersonate)[0]
+                prefix = get_generation_prompt(renderer, impersonate=impersonate)[0]
                 if not impersonate:
                     prefix = apply_extensions('bot_prefix', prefix, state)
 
@@ -431,14 +292,16 @@ def make_prompt(messages):
             outer_messages.append({"role": "assistant", "content": prefix})
 
             prompt = instruct_renderer(messages=outer_messages)
-            suffix = handler.get_generation_prefix_suffix(instruct_renderer, impersonate=False)[1]
+            suffix = get_generation_prompt(instruct_renderer, impersonate=False)[1]
             if len(suffix) > 0:
                 prompt = prompt[:-len(suffix)]
         else:
             if _continue:
-                prompt = handler.modify_for_continue(prompt, renderer, impersonate)
+                suffix = get_generation_prompt(renderer, impersonate=impersonate)[1]
+                if len(suffix) > 0:
+                    prompt = prompt[:-len(suffix)]
             else:
-                prefix = handler.get_generation_prefix_suffix(renderer, impersonate=impersonate)[0]
+                prefix = get_generation_prompt(renderer, impersonate=impersonate)[0]
                 if state['mode'] == 'chat' and not impersonate:
                     prefix = apply_extensions('bot_prefix', prefix, state)
 
@@ -564,16 +427,31 @@ def get_stopping_strings(state):
     if state['mode'] in ['instruct', 'chat-instruct']:
         template = jinja_env.from_string(state['instruction_template_str'])
         renderer = partial(template.render, add_generation_prompt=False)
-        renderers.append((renderer, state['instruction_template_str']))
+        renderers.append(renderer)
 
     if state['mode'] in ['chat', 'chat-instruct']:
         template = jinja_env.from_string(state['chat_template_str'])
         renderer = partial(template.render, add_generation_prompt=False, name1=state['name1'], name2=state['name2'])
-        renderers.append((renderer, state['chat_template_str']))
+        renderers.append(renderer)
+
+    for renderer in renderers:
+        prefix_bot, suffix_bot = get_generation_prompt(renderer, impersonate=False)
+        prefix_user, suffix_user = get_generation_prompt(renderer, impersonate=True)
+
+        stopping_strings += [
+            suffix_user + prefix_bot,
+            suffix_user + prefix_user,
+            suffix_bot + prefix_bot,
+            suffix_bot + prefix_user,
+        ]
 
-    for renderer, template_str in renderers:
-        handler = create_template_handler(template_str)
-        stopping_strings += handler.get_stopping_strings(renderer)
+    # Try to find the EOT token
+    for item in stopping_strings.copy():
+        item = item.strip()
+        if item.startswith("<") and ">" in item:
+            stopping_strings.append(item.split(">")[0] + ">")
+        elif item.startswith("[") and "]" in item:
+            stopping_strings.append(item.split("]")[0] + "]")
 
     if 'stopping_strings' in state and isinstance(state['stopping_strings'], list):
         stopping_strings += state.pop('stopping_strings')
@@ -772,16 +650,6 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
     output = apply_extensions('history', output)
     state = apply_extensions('state', state)
 
-    # Automatically set skip_special_tokens to False for channel-based templates
-    if state['mode'] in ['instruct', 'chat-instruct']:
-        template_str = state['instruction_template_str']
-    else:  # chat mode
-        template_str = state['chat_template_str']
-
-    handler = create_template_handler(template_str)
-    if isinstance(handler, ChannelTemplateHandler):
-        state['skip_special_tokens'] = False
-
     # Let the jinja2 template handle the BOS token
     if state['mode'] in ['instruct', 'chat-instruct']:
         state['add_bos_token'] = False
@@ -941,10 +809,6 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
 
 
 def impersonate_wrapper(textbox, state):
-    # Check template support first
-    template_str = state['chat_template_str']
-    handler = create_template_handler(template_str)
-
     text = textbox['text']
     static_output = chat_html_wrapper(state['history'], state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
 

From bfbbfc2361e26b03e5af9a26434391be9fd257f1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 17:33:01 -0700
Subject: [PATCH 0851/1701] Ignore add_generation_prompt in GPT-OSS

---
 modules/models_settings.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 3c068df0b9..e35e1c046e 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -93,6 +93,7 @@ def get_model_metadata(model):
             template = re.sub(r"\{\{-?\s*raise_exception\(.*?\)\s*-?\}\}", "", template, flags=re.DOTALL)
             template = re.sub(r'raise_exception\([^)]*\)', "''", template)
             template = re.sub(r'{% if add_generation_prompt %}.*', '', template, flags=re.DOTALL)
+            template = re.sub(r'elif loop\.last and not add_generation_prompt', 'elif False', template)  # Handle GPT-OSS
             model_settings['instruction_template'] = 'Custom (obtained from model metadata)'
             model_settings['instruction_template_str'] = template
 
@@ -153,6 +154,7 @@ def get_model_metadata(model):
             template = re.sub(r"\{\{-?\s*raise_exception\(.*?\)\s*-?\}\}", "", template, flags=re.DOTALL)
             template = re.sub(r'raise_exception\([^)]*\)', "''", template)
             template = re.sub(r'{% if add_generation_prompt %}.*', '', template, flags=re.DOTALL)
+            template = re.sub(r'elif loop\.last and not add_generation_prompt', 'elif False', template)  # Handle GPT-OSS
             model_settings['instruction_template'] = 'Custom (obtained from model metadata)'
             model_settings['instruction_template_str'] = template
 

From fbea21a1f13186740012c55aa7877a5aeda89c2f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 17:33:27 -0700
Subject: [PATCH 0852/1701] Only use enable_thinking if the template supports
 it

---
 modules/chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 82760cc8cd..e7fd86f999 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -307,7 +307,7 @@ def make_prompt(messages):
 
                 prompt += prefix
 
-        if state['mode'] == 'instruct' and not any((_continue, impersonate, state['enable_thinking'])):
+        if state['mode'] == 'instruct' and 'enable_thinking' in state['instruction_template_str'] and not any((_continue, impersonate, state['enable_thinking'])):
             prompt += get_thinking_suppression_string(instruction_template)
 
         return prompt

From 7c82d65a9d071342cc501246760f8a875e5097a7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 5 Aug 2025 18:05:09 -0700
Subject: [PATCH 0853/1701] Handle GPT-OSS as a special template case

---
 modules/chat.py | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index e7fd86f999..66f89c7019 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -302,6 +302,13 @@ def make_prompt(messages):
                     prompt = prompt[:-len(suffix)]
             else:
                 prefix = get_generation_prompt(renderer, impersonate=impersonate)[0]
+
+                # Handle GPT-OSS as a special case
+                if '<|channel|>final<|message|>' in state['instruction_template_str']:
+                    prefix = prefix.rstrip("<|channel|>final<|message|>")
+                    if impersonate:
+                        prefix += "<|message|>"
+
                 if state['mode'] == 'chat' and not impersonate:
                     prefix = apply_extensions('bot_prefix', prefix, state)
 
@@ -460,6 +467,12 @@ def get_stopping_strings(state):
     result = [item for item in stopping_strings if not any(item.startswith(other) and item != other for other in stopping_strings)]
     result = list(set(result))
 
+    # Handle GPT-OSS as a special case
+    if '<|channel|>final<|message|>' in state['instruction_template_str'] and "<|end|>" in result:
+        result.remove("<|end|>")
+        result.append("<|result|>")
+        result = list(set(result))
+
     if shared.args.verbose:
         logger.info("STOPPING_STRINGS=")
         pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(result)
@@ -650,6 +663,10 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
     output = apply_extensions('history', output)
     state = apply_extensions('state', state)
 
+    # Handle GPT-OSS as a special case
+    if '<|channel|>final<|message|>' in state['instruction_template_str']:
+        state['skip_special_tokens'] = False
+
     # Let the jinja2 template handle the BOS token
     if state['mode'] in ['instruct', 'chat-instruct']:
         state['add_bos_token'] = False

From 6ce4b353c49a1b9b86cf842a1d30ec2198f5d9b7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 6 Aug 2025 06:42:45 -0700
Subject: [PATCH 0854/1701] Fix the GPT-OSS template

---
 modules/chat.py | 38 ++++++++++++++++++++++++++++++++++++--
 1 file changed, 36 insertions(+), 2 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 66f89c7019..e07dfd1c65 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -211,7 +211,39 @@ def generate_chat_prompt(user_input, state, **kwargs):
             messages.insert(insert_pos, {"role": "tool", "content": tool_msg})
 
         if assistant_msg:
-            messages.insert(insert_pos, {"role": "assistant", "content": assistant_msg})
+            # Handle GPT-OSS as a special case
+            if '<|channel|>analysis<|message|>' in assistant_msg or '<|channel|>final<|message|>' in assistant_msg:
+
+                thinking_content = ""
+                final_content = ""
+
+                # Extract analysis content if present
+                if '<|channel|>analysis<|message|>' in assistant_msg:
+                    analysis_start = assistant_msg.find('<|channel|>analysis<|message|>') + len('<|channel|>analysis<|message|>')
+                    if '<|start|>assistant<|channel|>final<|message|>' in assistant_msg:
+                        analysis_end = assistant_msg.find('<|start|>assistant<|channel|>final<|message|>')
+                    else:
+                        analysis_end = len(assistant_msg)
+
+                    thinking_content = assistant_msg[analysis_start:analysis_end].strip()
+
+                # Extract final content if present
+                if '<|start|>assistant<|channel|>final<|message|>' in assistant_msg:
+                    final_start = assistant_msg.find('<|start|>assistant<|channel|>final<|message|>') + len('<|start|>assistant<|channel|>final<|message|>')
+                    final_content = assistant_msg[final_start:].strip()
+                elif '<|channel|>final<|message|>' in assistant_msg:
+                    final_start = assistant_msg.find('<|channel|>final<|message|>') + len('<|channel|>final<|message|>')
+                    final_content = assistant_msg[final_start:].strip()
+
+                # Insert as structured message
+                msg_dict = {"role": "assistant", "content": final_content}
+                if thinking_content:
+                    msg_dict["thinking"] = thinking_content
+
+                messages.insert(insert_pos, msg_dict)
+
+            else:
+                messages.insert(insert_pos, {"role": "assistant", "content": assistant_msg})
 
         if user_msg not in ['', '<|BEGIN-VISIBLE-CHAT|>']:
             # Check for user message attachments in metadata
@@ -305,7 +337,9 @@ def make_prompt(messages):
 
                 # Handle GPT-OSS as a special case
                 if '<|channel|>final<|message|>' in state['instruction_template_str']:
-                    prefix = prefix.rstrip("<|channel|>final<|message|>")
+                    if prefix.endswith("<|channel|>final<|message|>"):
+                        prefix = prefix[:-len("<|channel|>final<|message|>")]
+
                     if impersonate:
                         prefix += "<|message|>"
 

From 0c1403f2c72133e1ff63154d21f37954a2e1c343 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 6 Aug 2025 08:05:37 -0700
Subject: [PATCH 0855/1701] Handle GPT-OSS as a special case when continuing

---
 modules/chat.py | 47 ++++++++++++++++++++++++++++++++---------------
 1 file changed, 32 insertions(+), 15 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index e07dfd1c65..64588b9d38 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -237,7 +237,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
                 # Insert as structured message
                 msg_dict = {"role": "assistant", "content": final_content}
-                if thinking_content:
+                if '<|channel|>analysis<|message|>' in assistant_msg:
                     msg_dict["thinking"] = thinking_content
 
                 messages.insert(insert_pos, msg_dict)
@@ -328,25 +328,42 @@ def make_prompt(messages):
             if len(suffix) > 0:
                 prompt = prompt[:-len(suffix)]
         else:
-            if _continue:
-                suffix = get_generation_prompt(renderer, impersonate=impersonate)[1]
-                if len(suffix) > 0:
-                    prompt = prompt[:-len(suffix)]
+            # Handle GPT-OSS as a special case when continuing
+            if _continue and '<|channel|>final<|message|>' in state['instruction_template_str']:
+                # This prevents the template from stripping the analysis block of the message being continued.
+
+                last_message_to_continue = messages[-1]
+                prompt = renderer(messages=messages[:-1])
+
+                assistant_reply_so_far = ""
+                if 'thinking' in last_message_to_continue:
+                    assistant_reply_so_far += f"<|start|>assistant<|channel|>analysis<|message|>{last_message_to_continue['thinking']}<|end|>"
+
+                assistant_reply_so_far += f"<|start|>assistant<|channel|>final<|message|>{last_message_to_continue.get('content', '')}"
+
+                prompt += assistant_reply_so_far
+
             else:
-                prefix = get_generation_prompt(renderer, impersonate=impersonate)[0]
+                prompt = renderer(messages=messages)
+                if _continue:
+                    suffix = get_generation_prompt(renderer, impersonate=impersonate)[1]
+                    if len(suffix) > 0:
+                        prompt = prompt[:-len(suffix)]
+                else:
+                    prefix = get_generation_prompt(renderer, impersonate=impersonate)[0]
 
-                # Handle GPT-OSS as a special case
-                if '<|channel|>final<|message|>' in state['instruction_template_str']:
-                    if prefix.endswith("<|channel|>final<|message|>"):
-                        prefix = prefix[:-len("<|channel|>final<|message|>")]
+                    # Handle GPT-OSS as a special case when not continuing
+                    if '<|channel|>final<|message|>' in state['instruction_template_str']:
+                        if prefix.endswith("<|channel|>final<|message|>"):
+                            prefix = prefix[:-len("<|channel|>final<|message|>")]
 
-                    if impersonate:
-                        prefix += "<|message|>"
+                        if impersonate:
+                            prefix += "<|message|>"
 
-                if state['mode'] == 'chat' and not impersonate:
-                    prefix = apply_extensions('bot_prefix', prefix, state)
+                    if state['mode'] == 'chat' and not impersonate:
+                        prefix = apply_extensions('bot_prefix', prefix, state)
 
-                prompt += prefix
+                    prompt += prefix
 
         if state['mode'] == 'instruct' and 'enable_thinking' in state['instruction_template_str'] and not any((_continue, impersonate, state['enable_thinking'])):
             prompt += get_thinking_suppression_string(instruction_template)

From 3e24f455c8cca90310d5a3f9db31ed2007520db3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 6 Aug 2025 10:18:42 -0700
Subject: [PATCH 0856/1701] Fix continue for GPT-OSS (hopefully the final fix)

---
 modules/chat.py | 54 ++++++++++++++++++++++++++++++++-----------------
 1 file changed, 36 insertions(+), 18 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 64588b9d38..1ab91b5e7a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -219,21 +219,39 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
                 # Extract analysis content if present
                 if '<|channel|>analysis<|message|>' in assistant_msg:
-                    analysis_start = assistant_msg.find('<|channel|>analysis<|message|>') + len('<|channel|>analysis<|message|>')
-                    if '<|start|>assistant<|channel|>final<|message|>' in assistant_msg:
-                        analysis_end = assistant_msg.find('<|start|>assistant<|channel|>final<|message|>')
-                    else:
-                        analysis_end = len(assistant_msg)
-
-                    thinking_content = assistant_msg[analysis_start:analysis_end].strip()
+                    # Split the message by the analysis tag to isolate the content that follows
+                    parts = assistant_msg.split('<|channel|>analysis<|message|>', 1)
+                    if len(parts) > 1:
+                        # The content is everything after the tag
+                        potential_content = parts[1]
+
+                        # Now, find the end of this content block
+                        analysis_end_tag = '<|end|>'
+                        if analysis_end_tag in potential_content:
+                            thinking_content = potential_content.split(analysis_end_tag, 1)[0].strip()
+                        else:
+                            # Fallback: if no <|end|> tag, stop at the start of the final channel if it exists
+                            final_channel_tag = '<|channel|>final<|message|>'
+                            if final_channel_tag in potential_content:
+                                thinking_content = potential_content.split(final_channel_tag, 1)[0].strip()
+                            else:
+                                thinking_content = potential_content.strip()
 
                 # Extract final content if present
-                if '<|start|>assistant<|channel|>final<|message|>' in assistant_msg:
-                    final_start = assistant_msg.find('<|start|>assistant<|channel|>final<|message|>') + len('<|start|>assistant<|channel|>final<|message|>')
-                    final_content = assistant_msg[final_start:].strip()
-                elif '<|channel|>final<|message|>' in assistant_msg:
-                    final_start = assistant_msg.find('<|channel|>final<|message|>') + len('<|channel|>final<|message|>')
-                    final_content = assistant_msg[final_start:].strip()
+                final_tag_to_find = '<|channel|>final<|message|>'
+                if final_tag_to_find in assistant_msg:
+                    # Split the message by the final tag to isolate the content that follows
+                    parts = assistant_msg.split(final_tag_to_find, 1)
+                    if len(parts) > 1:
+                        # The content is everything after the tag
+                        potential_content = parts[1]
+
+                        # Now, find the end of this content block
+                        final_end_tag = '<|end|>'
+                        if final_end_tag in potential_content:
+                            final_content = potential_content.split(final_end_tag, 1)[0].strip()
+                        else:
+                            final_content = potential_content.strip()
 
                 # Insert as structured message
                 msg_dict = {"role": "assistant", "content": final_content}
@@ -330,16 +348,16 @@ def make_prompt(messages):
         else:
             # Handle GPT-OSS as a special case when continuing
             if _continue and '<|channel|>final<|message|>' in state['instruction_template_str']:
-                # This prevents the template from stripping the analysis block of the message being continued.
-
                 last_message_to_continue = messages[-1]
                 prompt = renderer(messages=messages[:-1])
 
-                assistant_reply_so_far = ""
+                # Start the assistant turn wrapper
+                assistant_reply_so_far = "<|start|>assistant"
+
                 if 'thinking' in last_message_to_continue:
-                    assistant_reply_so_far += f"<|start|>assistant<|channel|>analysis<|message|>{last_message_to_continue['thinking']}<|end|>"
+                    assistant_reply_so_far += f"<|channel|>analysis<|message|>{last_message_to_continue['thinking']}<|end|>"
 
-                assistant_reply_so_far += f"<|start|>assistant<|channel|>final<|message|>{last_message_to_continue.get('content', '')}"
+                assistant_reply_so_far += f"<|channel|>final<|message|>{last_message_to_continue.get('content', '')}"
 
                 prompt += assistant_reply_so_far
 

From f1147c992618ee17a7f5a37331d99d00ad02fd79 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 6 Aug 2025 19:32:36 -0700
Subject: [PATCH 0857/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_cuda128.txt             | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt      | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index dd1e8d35cc..f17cae8a2c 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index b65f0b09f2..51f4571fa3 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 6e6986540b..37021c7727 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 84abd394b6..f54ae19185 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 2deefbc4b6..e495455b11 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 8c1baf04e1..72847534ca 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 67a44432cc..ed641a2408 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 9fe3c54b2a..d7fe735b55 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 50602d8d89..cb71f74b5b 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index abdcfc16fc..d6bed57630 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 30d7d9e43e..1f17dc5028 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index a7c7808aec..82254842d6 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index b1f66f566e..986a3d493c 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -18,6 +18,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 76530338b9..833e923bec 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 26235b83b7..6a894d49ee 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index cfa7631095..0afb19c26d 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 2f8c401d35..a404f50c57 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index e065057521..75176656f7 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.32.0/llama_cpp_binaries-0.32.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From b391ac8eb1ba63e449f0ef021db56d6513dce646 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 8 Aug 2025 17:51:24 -0700
Subject: [PATCH 0858/1701] Fix getting the ctx-size for EXL3/EXL2/Transformers
 models

---
 modules/models_settings.py | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index e35e1c046e..4e53dc813e 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -106,9 +106,16 @@ def get_model_metadata(model):
 
             for k in ['max_position_embeddings', 'model_max_length', 'max_seq_len']:
                 if k in metadata:
-                    model_settings['truncation_length'] = metadata[k]
-                    model_settings['truncation_length_info'] = metadata[k]
-                    model_settings['ctx_size'] = min(metadata[k], 8192)
+                    value = metadata[k]
+                elif k in metadata.get('text_config', {}):
+                    value = metadata['text_config'][k]
+                else:
+                    continue
+
+                model_settings['truncation_length'] = value
+                model_settings['truncation_length_info'] = value
+                model_settings['ctx_size'] = min(value, 8192)
+                break
 
             if 'rope_theta' in metadata:
                 model_settings['rope_freq_base'] = metadata['rope_theta']

From 88127f46c124723554b5e87cad9c868348ed4c53 Mon Sep 17 00:00:00 2001
From: Katehuuh <133996730+Katehuuh@users.noreply.github.com>
Date: Sat, 9 Aug 2025 04:31:16 +0200
Subject: [PATCH 0859/1701] Add multimodal support (ExLlamaV3) (#7174)

---
 css/main.css                     |  13 ++
 docs/12 - OpenAI API.md          |  18 ++
 extensions/openai/completions.py | 102 +++++++++-
 extensions/openai/image_utils.py |  97 ++++++++++
 extensions/openai/typing.py      |  16 +-
 modules/chat.py                  | 118 +++++++++---
 modules/exllamav3.py             | 313 +++++++++++++++++++++++++++++++
 modules/html_generator.py        |  31 ++-
 modules/loaders.py               |  40 ++++
 modules/models.py                |  13 +-
 modules/shared.py                |   2 +
 modules/text_generation.py       |  10 +-
 modules/ui_chat.py               |   2 +-
 13 files changed, 723 insertions(+), 52 deletions(-)
 create mode 100644 extensions/openai/image_utils.py
 create mode 100644 modules/exllamav3.py

diff --git a/css/main.css b/css/main.css
index 240a94d5e4..de16d81d2c 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1577,6 +1577,19 @@ strong {
     margin-top: 4px;
 }
 
+.image-attachment {
+    flex-direction: column;
+}
+
+.image-preview {
+    border-radius: 16px;
+    margin-bottom: 5px;
+    object-fit: cover;
+    object-position: center;
+    border: 2px solid var(--border-color-primary);
+    aspect-ratio: 1 / 1;
+}
+
 button:focus {
     outline: none;
 }
diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index ec99939764..b7b5fbc18d 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -77,6 +77,24 @@ curl http://127.0.0.1:5000/v1/chat/completions \
   }'
 ```
 
+#### Multimodal support (ExLlamaV3)
+
+```shell
+curl http://127.0.0.1:5000/v1/chat/completions \
+  -H "Content-Type: application/json" \
+  -d '{
+    "messages": [
+      {
+        "role": "user",
+        "content": [
+          {"type": "text", "text": "What color is this image?"},
+          {"type": "image_url", "image_url": {"url": "https://github.com/turboderp-org/exllamav3/blob/master/examples/media/cat.png?raw=true"}}
+        ]
+      }
+    ]
+  }'
+```
+
 #### SSE streaming
 
 ```shell
diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 5181b18bff..3d389f0b27 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -7,6 +7,7 @@
 from pydantic import ValidationError
 
 from extensions.openai.errors import InvalidRequestError
+from extensions.openai.image_utils import convert_openai_messages_to_images
 from extensions.openai.typing import ToolDefinition
 from extensions.openai.utils import debug_msg, getToolCallId, parseToolCall
 from modules import shared
@@ -16,6 +17,7 @@
     load_character_memoized,
     load_instruction_template_memoized
 )
+from modules.logging_colors import logger
 from modules.presets import load_preset_memoized
 from modules.text_generation import decode, encode, generate_reply
 
@@ -82,6 +84,21 @@ def process_parameters(body, is_legacy=False):
     return generate_params
 
 
+def process_multimodal_content(content):
+    """Extract text from OpenAI multimodal format for non-multimodal models"""
+    if isinstance(content, str):
+        return content
+
+    if isinstance(content, list):
+        text_parts = []
+        for item in content:
+            if isinstance(item, dict) and item.get('type') == 'text':
+                text_parts.append(item.get('text', ''))
+        return ' '.join(text_parts) if text_parts else str(content)
+
+    return str(content)
+
+
 def convert_history(history):
     '''
     Chat histories in this program are in the format [message, reply].
@@ -99,8 +116,11 @@ def convert_history(history):
         role = entry["role"]
 
         if role == "user":
+            # Extract text content (images handled by model-specific code)
+            content = process_multimodal_content(content)
             user_input = content
             user_input_last = True
+
             if current_message:
                 chat_dialogue.append([current_message, '', ''])
                 current_message = ""
@@ -126,7 +146,11 @@ def convert_history(history):
     if not user_input_last:
         user_input = ""
 
-    return user_input, system_message, {'internal': chat_dialogue, 'visible': copy.deepcopy(chat_dialogue)}
+    return user_input, system_message, {
+        'internal': chat_dialogue,
+        'visible': copy.deepcopy(chat_dialogue),
+        'messages': history  # Store original messages for multimodal models
+    }
 
 
 def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, prompt_only=False) -> dict:
@@ -150,9 +174,23 @@ def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, p
         elif m['role'] == 'function':
             raise InvalidRequestError(message="role: function is not supported.", param='messages')
 
-        if 'content' not in m and "image_url" not in m:
+        # Handle multimodal content validation
+        content = m.get('content')
+        if content is None:
             raise InvalidRequestError(message="messages: missing content", param='messages')
 
+        # Validate multimodal content structure
+        if isinstance(content, list):
+            for item in content:
+                if not isinstance(item, dict) or 'type' not in item:
+                    raise InvalidRequestError(message="messages: invalid content item format", param='messages')
+                if item['type'] not in ['text', 'image_url']:
+                    raise InvalidRequestError(message="messages: unsupported content type", param='messages')
+                if item['type'] == 'text' and 'text' not in item:
+                    raise InvalidRequestError(message="messages: missing text in content item", param='messages')
+                if item['type'] == 'image_url' and ('image_url' not in item or 'url' not in item['image_url']):
+                    raise InvalidRequestError(message="messages: missing image_url in content item", param='messages')
+
     # Chat Completions
     object_type = 'chat.completion' if not stream else 'chat.completion.chunk'
     created_time = int(time.time())
@@ -336,9 +374,26 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False):
 
     prompt_str = 'context' if is_legacy else 'prompt'
 
-    # ... encoded as a string, array of strings, array of tokens, or array of token arrays.
-    if prompt_str not in body:
-        raise InvalidRequestError("Missing required input", param=prompt_str)
+    # Handle both prompt and messages format for unified multimodal support
+    if prompt_str not in body or body[prompt_str] is None:
+        if 'messages' in body:
+            # Convert messages format to prompt for completions endpoint
+            prompt_text = ""
+            for message in body.get('messages', []):
+                if isinstance(message, dict) and 'content' in message:
+                    # Extract text content from multimodal messages
+                    content = message['content']
+                    if isinstance(content, str):
+                        prompt_text += content
+                    elif isinstance(content, list):
+                        for item in content:
+                            if isinstance(item, dict) and item.get('type') == 'text':
+                                prompt_text += item.get('text', '')
+
+            # Allow empty prompts for image-only requests
+            body[prompt_str] = prompt_text
+        else:
+            raise InvalidRequestError("Missing required input", param=prompt_str)
 
     # common params
     generate_params = process_parameters(body, is_legacy=is_legacy)
@@ -349,9 +404,18 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False):
     suffix = body['suffix'] if body['suffix'] else ''
     echo = body['echo']
 
+    # Add messages to generate_params if present for multimodal processing
+    if 'messages' in body:
+        generate_params['messages'] = body['messages']
+
     if not stream:
         prompt_arg = body[prompt_str]
-        if isinstance(prompt_arg, str) or (isinstance(prompt_arg, list) and isinstance(prompt_arg[0], int)):
+
+        # Handle empty/None prompts (e.g., image-only requests)
+        if prompt_arg is None:
+            prompt_arg = ""
+
+        if isinstance(prompt_arg, str) or (isinstance(prompt_arg, list) and len(prompt_arg) > 0 and isinstance(prompt_arg[0], int)):
             prompt_arg = [prompt_arg]
 
         resp_list_data = []
@@ -374,7 +438,19 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False):
 
             # generate reply #######################################
             debug_msg({'prompt': prompt, 'generate_params': generate_params})
-            generator = generate_reply(prompt, generate_params, is_chat=False)
+
+            # Use multimodal generation if images are present
+            if 'messages' in generate_params:
+                raw_images = convert_openai_messages_to_images(generate_params['messages'])
+                if raw_images:
+                    logger.info(f"Using multimodal generation for {len(raw_images)} images")
+                    generate_params['raw_images'] = raw_images
+                    generator = shared.model.generate_with_streaming(prompt, generate_params)
+                else:
+                    generator = generate_reply(prompt, generate_params, is_chat=False)
+            else:
+                generator = generate_reply(prompt, generate_params, is_chat=False)
+
             answer = ''
 
             for a in generator:
@@ -447,7 +523,17 @@ def text_streaming_chunk(content):
 
         # generate reply #######################################
         debug_msg({'prompt': prompt, 'generate_params': generate_params})
-        generator = generate_reply(prompt, generate_params, is_chat=False)
+        # Use multimodal generation if images are present
+        if 'messages' in generate_params:
+            raw_images = convert_openai_messages_to_images(generate_params['messages'])
+            if raw_images:
+                logger.info(f"Using multimodal generation for {len(raw_images)} images")
+                generate_params['raw_images'] = raw_images
+                generator = shared.model.generate_with_streaming(prompt, generate_params)
+            else:
+                generator = generate_reply(prompt, generate_params, is_chat=False)
+        else:
+            generator = generate_reply(prompt, generate_params, is_chat=False)
 
         answer = ''
         seen_content = ''
diff --git a/extensions/openai/image_utils.py b/extensions/openai/image_utils.py
new file mode 100644
index 0000000000..c54f053294
--- /dev/null
+++ b/extensions/openai/image_utils.py
@@ -0,0 +1,97 @@
+"""
+Shared image processing utilities for multimodal support.
+Used by both ExLlamaV3 and llama.cpp implementations.
+"""
+import base64
+import io
+from typing import Any, List, Tuple
+
+from PIL import Image
+
+from modules.logging_colors import logger
+
+
+def decode_base64_image(base64_string: str) -> Image.Image:
+    """Decodes a base64 string to a PIL Image."""
+    try:
+        if base64_string.startswith('data:image/'):
+            base64_string = base64_string.split(',', 1)[1]
+
+        image_data = base64.b64decode(base64_string)
+        image = Image.open(io.BytesIO(image_data))
+        return image
+    except Exception as e:
+        logger.error(f"Failed to decode base64 image: {e}")
+        raise ValueError(f"Invalid base64 image data: {e}")
+
+
+def process_message_content(content: Any) -> Tuple[str, List[Image.Image]]:
+    """
+    Processes message content that may contain text and images.
+    Returns: A tuple of (text_content, list_of_pil_images).
+    """
+    if isinstance(content, str):
+        return content, []
+
+    if isinstance(content, list):
+        text_parts = []
+        images = []
+        for item in content:
+            if not isinstance(item, dict):
+                continue
+
+            item_type = item.get('type', '')
+            if item_type == 'text':
+                text_parts.append(item.get('text', ''))
+            elif item_type == 'image_url':
+                image_url_data = item.get('image_url', {})
+                image_url = image_url_data.get('url', '')
+
+                if image_url.startswith('data:image/'):
+                    try:
+                        images.append(decode_base64_image(image_url))
+                    except Exception as e:
+                        logger.warning(f"Failed to process a base64 image: {e}")
+                elif image_url.startswith('http'):
+                    # Support external URLs
+                    try:
+                        import requests
+                        response = requests.get(image_url, timeout=10)
+                        response.raise_for_status()
+                        image_data = response.content
+                        image = Image.open(io.BytesIO(image_data))
+                        images.append(image)
+                        logger.info("Successfully loaded external image from URL")
+                    except Exception as e:
+                        logger.warning(f"Failed to fetch external image: {e}")
+                else:
+                    logger.warning(f"Unsupported image URL format: {image_url[:70]}...")
+
+        return ' '.join(text_parts), images
+
+    return str(content), []
+
+
+def convert_image_attachments_to_pil(image_attachments: List[dict]) -> List[Image.Image]:
+    """Convert webui image_attachments format to PIL Images."""
+    pil_images = []
+    for attachment in image_attachments:
+        if attachment.get('type') == 'image' and 'image_data' in attachment:
+            try:
+                image = decode_base64_image(attachment['image_data'])
+                if image.mode != 'RGB':
+                    image = image.convert('RGB')
+                pil_images.append(image)
+            except Exception as e:
+                logger.warning(f"Failed to process image attachment: {e}")
+    return pil_images
+
+
+def convert_openai_messages_to_images(messages: List[dict]) -> List[Image.Image]:
+    """Convert OpenAI messages format to PIL Images."""
+    all_images = []
+    for message in messages:
+        if isinstance(message, dict) and 'content' in message:
+            _, images = process_message_content(message['content'])
+            all_images.extend(images)
+    return all_images
diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 6bd3749ff6..e9f92da561 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -2,7 +2,7 @@
 import time
 from typing import Dict, List, Optional
 
-from pydantic import BaseModel, Field, validator
+from pydantic import BaseModel, Field, field_validator, validator
 
 
 class GenerationOptions(BaseModel):
@@ -99,7 +99,8 @@ class ToolCall(BaseModel):
 
 class CompletionRequestParams(BaseModel):
     model: str | None = Field(default=None, description="Unused parameter. To change the model, use the /v1/internal/model/load endpoint.")
-    prompt: str | List[str]
+    prompt: str | List[str] | None = Field(default=None, description="Text prompt for completion. Can also use 'messages' format for multimodal.")
+    messages: List[dict] | None = Field(default=None, description="OpenAI messages format for multimodal support. Alternative to 'prompt'.")
     best_of: int | None = Field(default=1, description="Unused parameter.")
     echo: bool | None = False
     frequency_penalty: float | None = 0
@@ -115,6 +116,17 @@ class CompletionRequestParams(BaseModel):
     top_p: float | None = 1
     user: str | None = Field(default=None, description="Unused parameter.")
 
+    @field_validator('prompt', 'messages')
+    @classmethod
+    def validate_prompt_or_messages(cls, v, info):
+        """Ensure either 'prompt' or 'messages' is provided for completions."""
+        if info.field_name == 'prompt':  # If we're validating 'prompt', check if neither prompt nor messages will be set
+            messages = info.data.get('messages')
+            if v is None and messages is None:
+                raise ValueError("Either 'prompt' or 'messages' must be provided")
+
+        return v
+
 
 class CompletionRequest(GenerationOptions, CompletionRequestParams):
     pass
diff --git a/modules/chat.py b/modules/chat.py
index 1ab91b5e7a..354ae46b30 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -271,16 +271,27 @@ def generate_chat_prompt(user_input, state, **kwargs):
             # Add attachment content if present AND if past attachments are enabled
             if (state.get('include_past_attachments', True) and user_key in metadata and "attachments" in metadata[user_key]):
                 attachments_text = ""
+                image_refs = ""
+
                 for attachment in metadata[user_key]["attachments"]:
-                    filename = attachment.get("name", "file")
-                    content = attachment.get("content", "")
-                    if attachment.get("type") == "text/html" and attachment.get("url"):
-                        attachments_text += f"\nName: {filename}\nURL: {attachment['url']}\nContents:\n\n=====\n{content}\n=====\n\n"
+                    if attachment.get("type") == "image":
+                        # Add image reference for multimodal models
+                        image_refs += "<__media__>"
                     else:
-                        attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
+                        # Handle text/PDF attachments
+                        filename = attachment.get("name", "file")
+                        content = attachment.get("content", "")
+                        if attachment.get("type") == "text/html" and attachment.get("url"):
+                            attachments_text += f"\nName: {filename}\nURL: {attachment['url']}\nContents:\n\n=====\n{content}\n=====\n\n"
+                        else:
+                            attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
 
-                if attachments_text:
-                    enhanced_user_msg = f"{user_msg}\n\nATTACHMENTS:\n{attachments_text}"
+                if image_refs or attachments_text:
+                    enhanced_user_msg = user_msg
+                    if image_refs:
+                        enhanced_user_msg += f" {image_refs}"
+                    if attachments_text:
+                        enhanced_user_msg += f"\n\nATTACHMENTS:\n{attachments_text}"
 
             messages.insert(insert_pos, {"role": "user", "content": enhanced_user_msg})
 
@@ -301,16 +312,23 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
             if user_key in metadata and "attachments" in metadata[user_key]:
                 attachments_text = ""
+                image_refs = ""
+
                 for attachment in metadata[user_key]["attachments"]:
-                    filename = attachment.get("name", "file")
-                    content = attachment.get("content", "")
-                    if attachment.get("type") == "text/html" and attachment.get("url"):
-                        attachments_text += f"\nName: {filename}\nURL: {attachment['url']}\nContents:\n\n=====\n{content}\n=====\n\n"
+                    if attachment.get("type") == "image":
+                        image_refs += "<__media__>"
                     else:
-                        attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
+                        filename = attachment.get("name", "file")
+                        content = attachment.get("content", "")
+                        if attachment.get("type") == "text/html" and attachment.get("url"):
+                            attachments_text += f"\nName: {filename}\nURL: {attachment['url']}\nContents:\n\n=====\n{content}\n=====\n\n"
+                        else:
+                            attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
 
-                if attachments_text:
-                    user_input = f"{user_input}\n\nATTACHMENTS:\n{attachments_text}"
+                if image_refs or attachments_text:
+                    user_input = f"{user_input} {image_refs}"
+                    if attachments_text:
+                        user_input += f"\n\nATTACHMENTS:\n{attachments_text}"
 
         messages.append({"role": "user", "content": user_input})
 
@@ -594,29 +612,64 @@ def add_message_attachment(history, row_idx, file_path, is_user=True):
     file_extension = path.suffix.lower()
 
     try:
-        # Handle different file types
-        if file_extension == '.pdf':
+        # Handle image files
+        if file_extension in ['.jpg', '.jpeg', '.png', '.webp', '.bmp', '.gif']:
+            # Convert image to base64
+            with open(path, 'rb') as f:
+                image_data = base64.b64encode(f.read()).decode('utf-8')
+
+            # Determine MIME type from extension
+            mime_type_map = {
+                '.jpg': 'image/jpeg',
+                '.jpeg': 'image/jpeg',
+                '.png': 'image/png',
+                '.webp': 'image/webp',
+                '.bmp': 'image/bmp',
+                '.gif': 'image/gif'
+            }
+            mime_type = mime_type_map.get(file_extension, 'image/jpeg')
+
+            # Format as data URL
+            data_url = f"data:{mime_type};base64,{image_data}"
+
+            # Generate unique image ID
+            image_id = len([att for att in history['metadata'][key]["attachments"] if att.get("type") == "image"]) + 1
+
+            attachment = {
+                "name": filename,
+                "type": "image",
+                "image_data": data_url,
+                "image_id": image_id,
+                "file_path": str(path)  # For UI preview
+            }
+        elif file_extension == '.pdf':
             # Process PDF file
             content = extract_pdf_text(path)
-            file_type = "application/pdf"
+            attachment = {
+                "name": filename,
+                "type": "application/pdf",
+                "content": content,
+            }
         elif file_extension == '.docx':
             content = extract_docx_text(path)
-            file_type = "application/docx"
+            attachment = {
+                "name": filename,
+                "type": "application/docx",
+                "content": content,
+            }
         else:
             # Default handling for text files
             with open(path, 'r', encoding='utf-8') as f:
                 content = f.read()
-            file_type = "text/plain"
 
-        # Add attachment
-        attachment = {
-            "name": filename,
-            "type": file_type,
-            "content": content,
-        }
+            attachment = {
+                "name": filename,
+                "type": "text/plain",
+                "content": content,
+            }
 
         history['metadata'][key]["attachments"].append(attachment)
-        return content  # Return the content for reuse
+        return attachment  # Return the attachment for reuse
     except Exception as e:
         logger.error(f"Error processing attachment {filename}: {e}")
         return None
@@ -759,6 +812,19 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
         for file_path in files:
             add_message_attachment(output, row_idx, file_path, is_user=True)
 
+        # Collect image attachments for ExLlamaV3
+        image_attachments = []
+        if 'metadata' in output:
+            user_key = f"user_{row_idx}"
+            if user_key in output['metadata'] and "attachments" in output['metadata'][user_key]:
+                for attachment in output['metadata'][user_key]["attachments"]:
+                    if attachment.get("type") == "image":
+                        image_attachments.append(attachment)
+
+        # Add image attachments to state for the generation
+        if image_attachments:
+            state['image_attachments'] = image_attachments
+
         # Add web search results as attachments if enabled
         if state.get('enable_web_search', False):
             search_query = generate_search_query(text, state)
diff --git a/modules/exllamav3.py b/modules/exllamav3.py
new file mode 100644
index 0000000000..c2532ec301
--- /dev/null
+++ b/modules/exllamav3.py
@@ -0,0 +1,313 @@
+import traceback
+from pathlib import Path
+from typing import Any, List, Tuple
+
+from exllamav3 import Cache, Config, Generator, Model, Tokenizer
+from exllamav3.cache import CacheLayer_fp16, CacheLayer_quant
+
+from extensions.openai.image_utils import (
+    convert_image_attachments_to_pil,
+    convert_openai_messages_to_images
+)
+from modules import shared
+from modules.logging_colors import logger
+
+try:
+    import flash_attn
+except Exception:
+    logger.warning('Failed to load flash-attention due to the following error:\n')
+    traceback.print_exc()
+
+
+class Exllamav3Model:
+    def __init__(self):
+        pass
+
+    @classmethod
+    def from_pretrained(cls, path_to_model):
+        path_to_model = Path(f'{shared.args.model_dir}') / Path(path_to_model)
+
+        # Reset global MMTokenAllocator to prevent token ID corruption when switching models
+        from exllamav3.tokenizer.mm_embedding import (
+            FIRST_MM_EMBEDDING_INDEX,
+            global_allocator
+        )
+        global_allocator.next_token_index = FIRST_MM_EMBEDDING_INDEX
+        logger.info("Reset MMTokenAllocator for clean multimodal token allocation")
+
+        config = Config.from_directory(str(path_to_model))
+        model = Model.from_config(config)
+
+        # Calculate the closest multiple of 256 at or above the chosen value
+        max_tokens = shared.args.ctx_size
+        if max_tokens % 256 != 0:
+            adjusted_tokens = ((max_tokens // 256) + 1) * 256
+            logger.warning(f"max_num_tokens must be a multiple of 256. Adjusting from {max_tokens} to {adjusted_tokens}")
+            max_tokens = adjusted_tokens
+
+        # Parse cache type (ExLlamaV2 pattern)
+        cache_type = shared.args.cache_type.lower()
+        cache_kwargs = {}
+        if cache_type == 'fp16':
+            layer_type = CacheLayer_fp16
+        elif cache_type.startswith('q'):
+            layer_type = CacheLayer_quant
+            if '_' in cache_type:
+                # Different bits for k and v (e.g., q4_q8)
+                k_part, v_part = cache_type.split('_')
+                k_bits = int(k_part[1:])
+                v_bits = int(v_part[1:])
+            else:
+                # Same bits for k and v (e.g., q4)
+                k_bits = v_bits = int(cache_type[1:])
+
+            # Validate bit ranges
+            if not (2 <= k_bits <= 8 and 2 <= v_bits <= 8):
+                logger.warning(f"Invalid quantization bits: k_bits={k_bits}, v_bits={v_bits}. Must be between 2 and 8. Falling back to fp16.")
+                layer_type = CacheLayer_fp16
+            else:
+                cache_kwargs = {'k_bits': k_bits, 'v_bits': v_bits}
+        else:
+            logger.warning(f"Unrecognized cache type: {cache_type}. Falling back to fp16.")
+            layer_type = CacheLayer_fp16
+
+        cache = Cache(model, max_num_tokens=max_tokens, layer_type=layer_type, **cache_kwargs)
+
+        load_params = {'progressbar': True}
+        if shared.args.gpu_split:
+            split = [float(alloc) for alloc in shared.args.gpu_split.split(",")]
+            load_params['use_per_device'] = split
+
+        model.load(**load_params)
+
+        tokenizer = Tokenizer.from_config(config)
+
+        # Load vision model component (ExLlamaV3 native)
+        vision_model = None
+        try:
+            logger.info("Loading vision model component...")
+            vision_model = Model.from_config(config, component="vision")
+            vision_model.load(progressbar=True)
+            logger.info("Vision model loaded successfully")
+        except Exception as e:
+            logger.warning(f"Vision model loading failed (multimodal disabled): {e}")
+
+        generator = Generator(
+            model=model,
+            cache=cache,
+            tokenizer=tokenizer,
+        )
+
+        result = cls()
+        result.model = model
+        result.cache = cache
+        result.tokenizer = tokenizer
+        result.generator = generator
+        result.config = config
+        result.max_tokens = max_tokens
+        result.vision_model = vision_model
+
+        return result
+
+    def is_multimodal(self) -> bool:
+        """Check if this model supports multimodal input."""
+        return hasattr(self, 'vision_model') and self.vision_model is not None
+
+    def _process_images_for_generation(self, prompt: str, state: dict) -> Tuple[str, List[Any]]:
+        """
+        Process all possible image inputs and return modified prompt + embeddings.
+        Returns: (processed_prompt, image_embeddings)
+        """
+        if not self.is_multimodal():
+            return prompt, []
+
+        # Collect images from various sources using shared utilities
+        pil_images = []
+
+        # From webui image_attachments (preferred format)
+        if 'image_attachments' in state and state['image_attachments']:
+            pil_images.extend(convert_image_attachments_to_pil(state['image_attachments']))
+
+        # From OpenAI API raw_images
+        elif 'raw_images' in state and state['raw_images']:
+            pil_images.extend(state['raw_images'])
+
+        # From OpenAI API messages format
+        elif 'messages' in state and state['messages']:
+            pil_images.extend(convert_openai_messages_to_images(state['messages']))
+
+        if not pil_images:
+            return prompt, []
+
+        # ExLlamaV3-specific: Generate embeddings
+        try:
+            # Use pre-computed embeddings if available (proper MMEmbedding lifetime)
+            if 'image_embeddings' in state and state['image_embeddings']:
+                # Use existing embeddings - this preserves MMEmbedding lifetime
+                image_embeddings = state['image_embeddings']
+            else:
+                # Do not reset the cache/allocator index; it causes token ID conflicts during generation.
+
+                logger.info(f"Processing {len(pil_images)} image(s) with ExLlamaV3 vision model")
+                image_embeddings = [
+                    self.vision_model.get_image_embeddings(tokenizer=self.tokenizer, image=img)
+                    for img in pil_images
+                ]
+
+            # ExLlamaV3-specific: Handle prompt processing with placeholders
+            placeholders = [ie.text_alias for ie in image_embeddings]
+
+            if '<__media__>' in prompt:
+                # Web chat: Replace <__media__> placeholders
+                for alias in placeholders:
+                    prompt = prompt.replace('<__media__>', alias, 1)
+                logger.info(f"Replaced {len(placeholders)} <__media__> placeholder(s)")
+            else:
+                # API: Prepend embedding aliases
+                combined_placeholders = "\n".join(placeholders)
+                prompt = combined_placeholders + "\n" + prompt
+                logger.info(f"Prepended {len(placeholders)} embedding(s) to prompt")
+
+            return prompt, image_embeddings
+
+        except Exception as e:
+            logger.error(f"Failed to process images: {e}")
+            return prompt, []
+
+    def generate_with_streaming(self, prompt, state):
+        """
+        Generate text with streaming using native ExLlamaV3 API
+        """
+        from exllamav3 import Job
+        from exllamav3.generator.sampler.presets import ComboSampler
+
+        # Process images and modify prompt (ExLlamaV3-specific)
+        prompt, image_embeddings = self._process_images_for_generation(prompt, state)
+
+        sampler = ComboSampler(
+            rep_p=state.get('repetition_penalty', 1.0),
+            freq_p=state.get('frequency_penalty', 0.0),
+            pres_p=state.get('presence_penalty', 0.0),
+            temperature=state.get('temperature', 0.7),
+            min_p=state.get('min_p', 0.0),
+            top_k=state.get('top_k', 0),
+            top_p=state.get('top_p', 1.0),
+        )
+
+        # Encode prompt with embeddings (ExLlamaV3-specific)
+        if image_embeddings:
+            input_ids = self.tokenizer.encode(
+                prompt,
+                encode_special_tokens=True,
+                embeddings=image_embeddings,
+            )
+        else:
+            input_ids = self.tokenizer.encode(prompt, encode_special_tokens=True)
+
+        # Get stop conditions from state (webui format) - keep as strings like ExLlamaV3 examples
+        stop_conditions = []
+        if 'stopping_strings' in state and state['stopping_strings']:
+            # Use strings directly (ExLlamaV3 handles the conversion internally)
+            stop_conditions.extend(state['stopping_strings'])
+
+        # Add EOS token ID as ExLlamaV3 examples do
+        if hasattr(self.tokenizer, 'eos_token_id') and self.tokenizer.eos_token_id is not None:
+            stop_conditions.append(self.tokenizer.eos_token_id)
+
+        job = Job(
+            input_ids=input_ids,
+            max_new_tokens=state.get('max_new_tokens', 500),
+            decode_special_tokens=True,
+            embeddings=image_embeddings if image_embeddings else None,
+            sampler=sampler,
+            stop_conditions=stop_conditions if stop_conditions else None,
+        )
+
+        # Stream generation
+        self.generator.enqueue(job)
+
+        response_text = ""
+        try:
+            while self.generator.num_remaining_jobs():
+                results = self.generator.iterate()
+                for result in results:
+                    if "eos" in result and result["eos"]:
+                        break
+
+                    chunk = result.get("text", "")
+                    if chunk:
+                        response_text += chunk
+                        yield response_text
+        finally:
+            # No cleanup needed. MMEmbedding lifetime is managed by Python.
+            # Cache and page table resets are unnecessary and can cause token ID conflicts.
+            pass
+
+    def generate(self, prompt, state):
+        """
+        Generate text using native ExLlamaV3 API (non-streaming)
+        """
+        output = self.generator.generate(
+            prompt=prompt,
+            max_new_tokens=state.get('max_new_tokens', 500),
+            temperature=state.get('temperature', 0.7),
+            top_p=state.get('top_p', 1.0),
+            top_k=state.get('top_k', 0),
+            repetition_penalty=state.get('repetition_penalty', 1.0),
+            frequency_penalty=state.get('frequency_penalty', 0.0),
+            presence_penalty=state.get('presence_penalty', 0.0),
+            min_p=state.get('min_p', 0.0),
+        )
+
+        return output
+
+    def encode(self, string, **kwargs):
+        return self.tokenizer.encode(string, **kwargs)
+
+    def decode(self, ids, **kwargs):
+        return self.tokenizer.decode(ids, **kwargs)
+
+    @property
+    def last_prompt_token_count(self):
+        # This would need to be tracked during generation
+        return 0
+
+    def unload(self):
+        logger.info("Unloading ExLlamaV3 model components...")
+
+        if hasattr(self, 'vision_model') and self.vision_model is not None:
+            try:
+                del self.vision_model
+            except Exception as e:
+                logger.warning(f"Error unloading vision model: {e}")
+            self.vision_model = None
+
+        if hasattr(self, 'model') and self.model is not None:
+            try:
+                self.model.unload()
+                del self.model
+            except Exception as e:
+                logger.warning(f"Error unloading main model: {e}")
+            self.model = None
+
+        if hasattr(self, 'cache') and self.cache is not None:
+            self.cache = None
+
+        if hasattr(self, 'generator') and self.generator is not None:
+            self.generator = None
+
+        if hasattr(self, 'tokenizer') and self.tokenizer is not None:
+            self.tokenizer = None
+
+        # Force GPU memory cleanup
+        import gc
+
+        import torch
+        gc.collect()
+
+        if torch.cuda.is_available():
+            torch.cuda.empty_cache()
+            torch.cuda.synchronize()
+            torch.cuda.empty_cache()
+
+        logger.info("ExLlamaV3 model fully unloaded")
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 79237f7faf..63a0cdd0c9 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -406,16 +406,27 @@ def format_message_attachments(history, role, index):
         for attachment in attachments:
             name = html.escape(attachment["name"])
 
-            # Make clickable if URL exists
-            if "url" in attachment:
-                name = f'{name}'
-
-            attachments_html += (
-                f'
            '
-                f'
            {attachment_svg}
            '
-                f'
            {name}
            '
-                f'
            '
-            )
+            if attachment.get("type") == "image":
+                # Show image preview
+                file_path = attachment.get("file_path", "")
+                attachments_html += (
+                    f'
            '
+                    f'{name}'
+                    f'
            {name}
            '
+                    f'
            '
+                )
+            else:
+                # Make clickable if URL exists (web search)
+                if "url" in attachment:
+                    name = f'{name}'
+
+                attachments_html += (
+                    f'
            '
+                    f'
            {attachment_svg}
            '
+                    f'
            {name}
            '
+                    f'
            '
+                )
+
         attachments_html += '
            '
         return attachments_html
 
diff --git a/modules/loaders.py b/modules/loaders.py
index 7546bc5ba4..e9437c164c 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -55,6 +55,11 @@
         'trust_remote_code',
         'no_use_fast',
     ],
+    'ExLlamav3': [
+        'ctx_size',
+        'cache_type',
+        'gpu_split',
+    ],
     'ExLlamav2_HF': [
         'ctx_size',
         'cache_type',
@@ -248,6 +253,41 @@ def transformers_samplers():
         'grammar_string',
         'grammar_file_row',
     },
+    'ExLlamav3': {
+        'temperature',
+        'dynatemp_low',
+        'dynatemp_high',
+        'dynatemp_exponent',
+        'smoothing_factor',
+        'min_p',
+        'top_p',
+        'top_k',
+        'typical_p',
+        'xtc_threshold',
+        'xtc_probability',
+        'tfs',
+        'top_a',
+        'dry_multiplier',
+        'dry_allowed_length',
+        'dry_base',
+        'repetition_penalty',
+        'frequency_penalty',
+        'presence_penalty',
+        'repetition_penalty_range',
+        'mirostat_mode',
+        'mirostat_tau',
+        'mirostat_eta',
+        'dynamic_temperature',
+        'temperature_last',
+        'auto_max_new_tokens',
+        'ban_eos_token',
+        'add_bos_token',
+        'enable_thinking',
+        'skip_special_tokens',
+        'seed',
+        'custom_token_bans',
+        'dry_sequence_breakers',
+    },
     'ExLlamav2': {
         'temperature',
         'dynatemp_low',
diff --git a/modules/models.py b/modules/models.py
index c1e7fb5695..cc500a40b7 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -19,6 +19,7 @@ def load_model(model_name, loader=None):
         'llama.cpp': llama_cpp_server_loader,
         'Transformers': transformers_loader,
         'ExLlamav3_HF': ExLlamav3_HF_loader,
+        'ExLlamav3': ExLlamav3_loader,
         'ExLlamav2_HF': ExLlamav2_HF_loader,
         'ExLlamav2': ExLlamav2_loader,
         'TensorRT-LLM': TensorRT_LLM_loader,
@@ -88,6 +89,14 @@ def ExLlamav3_HF_loader(model_name):
     return Exllamav3HF.from_pretrained(model_name)
 
 
+def ExLlamav3_loader(model_name):
+    from modules.exllamav3 import Exllamav3Model
+
+    model = Exllamav3Model.from_pretrained(model_name)
+    tokenizer = model.tokenizer
+    return model, tokenizer
+
+
 def ExLlamav2_HF_loader(model_name):
     from modules.exllamav2_hf import Exllamav2HF
 
@@ -116,7 +125,9 @@ def unload_model(keep_model_name=False):
         return
 
     is_llamacpp = (shared.model.__class__.__name__ == 'LlamaServer')
-    if shared.model.__class__.__name__ == 'Exllamav3HF':
+    if shared.args.loader in ['ExLlamav3_HF', 'ExLlamav3']:
+        shared.model.unload()
+    elif shared.args.loader in ['ExLlamav2_HF', 'ExLlamav2'] and hasattr(shared.model, 'unload'):
         shared.model.unload()
 
     shared.model = shared.tokenizer = None
diff --git a/modules/shared.py b/modules/shared.py
index ab5198d1f5..1de4306b37 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -318,6 +318,8 @@ def fix_loader_name(name):
         return 'ExLlamav2_HF'
     elif name in ['exllamav3-hf', 'exllamav3_hf', 'exllama-v3-hf', 'exllama_v3_hf', 'exllama-v3_hf', 'exllama3-hf', 'exllama3_hf', 'exllama-3-hf', 'exllama_3_hf', 'exllama-3_hf']:
         return 'ExLlamav3_HF'
+    elif name in ['exllamav3']:
+        return 'ExLlamav3'
     elif name in ['tensorrt', 'tensorrtllm', 'tensorrt_llm', 'tensorrt-llm', 'tensort', 'tensortllm']:
         return 'TensorRT-LLM'
 
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 8d1950b98d..d6a87ce8b1 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -40,7 +40,7 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
             yield ''
             return
 
-        if shared.model.__class__.__name__ in ['LlamaServer', 'Exllamav2Model', 'TensorRTLLMModel']:
+        if shared.model.__class__.__name__ in ['LlamaServer', 'Exllamav2Model', 'Exllamav3Model', 'TensorRTLLMModel']:
             generate_func = generate_reply_custom
         else:
             generate_func = generate_reply_HF
@@ -128,9 +128,9 @@ def encode(prompt, add_special_tokens=True, add_bos_token=True, truncation_lengt
 
         from modules.torch_utils import get_device
 
-        if shared.model.__class__.__name__ in ['Exllamav2Model', 'TensorRTLLMModel']:
+        if shared.model.__class__.__name__ in ['Exllamav2Model', 'Exllamav3Model', 'TensorRTLLMModel']:
             input_ids = shared.tokenizer.encode(str(prompt))
-            if shared.model.__class__.__name__ != 'Exllamav2Model':
+            if shared.model.__class__.__name__ not in ['Exllamav2Model', 'Exllamav3Model']:
                 input_ids = np.array(input_ids).reshape(1, len(input_ids))
         else:
             input_ids = shared.tokenizer.encode(str(prompt), return_tensors='pt', add_special_tokens=add_special_tokens)
@@ -148,7 +148,7 @@ def encode(prompt, add_special_tokens=True, add_bos_token=True, truncation_lengt
         if truncation_length is not None:
             input_ids = input_ids[:, -truncation_length:]
 
-        if shared.model.__class__.__name__ in ['Exllamav2Model', 'TensorRTLLMModel'] or shared.args.cpu:
+        if shared.model.__class__.__name__ in ['Exllamav2Model', 'Exllamav3Model', 'TensorRTLLMModel'] or shared.args.cpu:
             return input_ids
         else:
             device = get_device()
@@ -295,6 +295,8 @@ def generate_reply_HF(question, original_question, state, stopping_strings=None,
         _StopEverythingStoppingCriteria
     )
 
+    # Native ExLlamav3Model handles multimodal internally - no special routing needed
+
     if shared.args.loader == 'Transformers':
         clear_torch_cache()
 
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 1d85a39863..3b922fb454 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -54,7 +54,7 @@ def create_ui():
                         gr.HTML(value='
            ☰
            ', elem_id='gr-hover')
 
                     with gr.Column(scale=10, elem_id='chat-input-container'):
-                        shared.gradio['textbox'] = gr.MultimodalTextbox(label='', placeholder='Send a message', file_types=['text', '.pdf'], file_count="multiple", elem_id='chat-input', elem_classes=['add_scrollbar'])
+                        shared.gradio['textbox'] = gr.MultimodalTextbox(label='', placeholder='Send a message', file_types=['text', '.pdf', 'image'], file_count="multiple", elem_id='chat-input', elem_classes=['add_scrollbar'])
                         shared.gradio['typing-dots'] = gr.HTML(value='
            ', label='typing', elem_id='typing-container')
 
                     with gr.Column(scale=1, elem_id='generate-stop-container'):

From 6e9de75727ace45b3bf71ea3a98ef350b6d7414d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 8 Aug 2025 19:35:09 -0700
Subject: [PATCH 0860/1701] Support loading chat templates from
 chat_template.json files

---
 modules/models_settings.py | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 4e53dc813e..729d5dd108 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -139,16 +139,26 @@ def get_model_metadata(model):
         with open(jinja_path, 'r', encoding='utf-8') as f:
             template = f.read()
 
+    # 2. If no .jinja file, try chat_template.json
+    if template is None:
+        json_template_path = Path(f'{shared.args.model_dir}/{model}') / 'chat_template.json'
+        if json_template_path.exists():
+            with open(json_template_path, 'r', encoding='utf-8') as f:
+                json_data = json.load(f)
+                if 'chat_template' in json_data:
+                    template = json_data['chat_template']
+
+    # 3. Fall back to tokenizer_config.json metadata
     if path.exists():
         metadata = json.loads(open(path, 'r', encoding='utf-8').read())
 
-        # 2. Only read from metadata if we haven't already loaded from .jinja
+        # Only read from metadata if we haven't already loaded from .jinja or .json
         if template is None and 'chat_template' in metadata:
             template = metadata['chat_template']
             if isinstance(template, list):
                 template = template[0]['template']
 
-        # 3. If a template was found from either source, process it
+        # 4. If a template was found from any source, process it
         if template:
             for k in ['eos_token', 'bos_token']:
                 if k in metadata:

From 8fcadff8d3120d1f3e844cd030d59a8c2b0b2dfd Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 8 Aug 2025 20:13:54 -0700
Subject: [PATCH 0861/1701] mtmd: Use the base64 attachment for the UI preview
 instead of the file

---
 modules/chat.py           | 1 -
 modules/html_generator.py | 5 ++---
 2 files changed, 2 insertions(+), 4 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 354ae46b30..988002394d 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -640,7 +640,6 @@ def add_message_attachment(history, row_idx, file_path, is_user=True):
                 "type": "image",
                 "image_data": data_url,
                 "image_id": image_id,
-                "file_path": str(path)  # For UI preview
             }
         elif file_extension == '.pdf':
             # Process PDF file
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 63a0cdd0c9..cb14a72248 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -407,11 +407,10 @@ def format_message_attachments(history, role, index):
             name = html.escape(attachment["name"])
 
             if attachment.get("type") == "image":
-                # Show image preview
-                file_path = attachment.get("file_path", "")
+                image_data = attachment.get("image_data", "")
                 attachments_html += (
                     f'
            '
-                    f'{name}'
+                    f'{name}'
                     f'
            {name}
            '
                     f'
            '
                 )

From 544c3a7c9f305b6a2141c3d02770250058d43322 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 8 Aug 2025 21:15:53 -0700
Subject: [PATCH 0862/1701] Polish the new exllamav3 loader

---
 modules/exllamav3.py | 152 +++++++++++++++++++++++++++++--------------
 modules/loaders.py   |  21 +-----
 2 files changed, 104 insertions(+), 69 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index c2532ec301..295c273712 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -2,8 +2,22 @@
 from pathlib import Path
 from typing import Any, List, Tuple
 
+import torch
 from exllamav3 import Cache, Config, Generator, Model, Tokenizer
 from exllamav3.cache import CacheLayer_fp16, CacheLayer_quant
+from exllamav3.generator import Job
+# Import the base sampler components directly from exllamav3
+from exllamav3.generator.sampler import (
+    CustomSampler,
+    SS_Argmax,
+    SS_MinP,
+    SS_PresFreqP,
+    SS_RepP,
+    SS_Sample,
+    SS_Temperature,
+    SS_TopK,
+    SS_TopP
+)
 
 from extensions.openai.image_utils import (
     convert_image_attachments_to_pil,
@@ -11,6 +25,7 @@
 )
 from modules import shared
 from modules.logging_colors import logger
+from modules.text_generation import get_max_prompt_length
 
 try:
     import flash_attn
@@ -79,7 +94,6 @@ def from_pretrained(cls, path_to_model):
             load_params['use_per_device'] = split
 
         model.load(**load_params)
-
         tokenizer = Tokenizer.from_config(config)
 
         # Load vision model component (ExLlamaV3 native)
@@ -127,11 +141,9 @@ def _process_images_for_generation(self, prompt: str, state: dict) -> Tuple[str,
         # From webui image_attachments (preferred format)
         if 'image_attachments' in state and state['image_attachments']:
             pil_images.extend(convert_image_attachments_to_pil(state['image_attachments']))
-
         # From OpenAI API raw_images
         elif 'raw_images' in state and state['raw_images']:
             pil_images.extend(state['raw_images'])
-
         # From OpenAI API messages format
         elif 'messages' in state and state['messages']:
             pil_images.extend(convert_openai_messages_to_images(state['messages']))
@@ -147,7 +159,6 @@ def _process_images_for_generation(self, prompt: str, state: dict) -> Tuple[str,
                 image_embeddings = state['image_embeddings']
             else:
                 # Do not reset the cache/allocator index; it causes token ID conflicts during generation.
-
                 logger.info(f"Processing {len(pil_images)} image(s) with ExLlamaV3 vision model")
                 image_embeddings = [
                     self.vision_model.get_image_embeddings(tokenizer=self.tokenizer, image=img)
@@ -178,46 +189,98 @@ def generate_with_streaming(self, prompt, state):
         """
         Generate text with streaming using native ExLlamaV3 API
         """
-        from exllamav3 import Job
-        from exllamav3.generator.sampler.presets import ComboSampler
-
         # Process images and modify prompt (ExLlamaV3-specific)
         prompt, image_embeddings = self._process_images_for_generation(prompt, state)
 
-        sampler = ComboSampler(
-            rep_p=state.get('repetition_penalty', 1.0),
-            freq_p=state.get('frequency_penalty', 0.0),
-            pres_p=state.get('presence_penalty', 0.0),
-            temperature=state.get('temperature', 0.7),
-            min_p=state.get('min_p', 0.0),
-            top_k=state.get('top_k', 0),
-            top_p=state.get('top_p', 1.0),
-        )
+        # -- Manually build and sort the sampler stack --
+        # Greedy decoding is a special case
+        if state['temperature'] == 0:
+            sampler = CustomSampler([SS_Argmax()])
+        else:
+            # 1. Create a list of all active, unordered samplers
+            unordered_samplers = []
+
+            # Penalties
+            penalty_range = state['repetition_penalty_range']
+            if penalty_range <= 0:
+                penalty_range = -1 # ExllamaV3 uses -1 for whole context
+            rep_decay = 0 # Not a configurable parameter
+
+            # Add penalty samplers if they are active
+            if state['repetition_penalty'] != 1.0:
+                 unordered_samplers.append(SS_RepP(state['repetition_penalty'], penalty_range, rep_decay))
+            if state['presence_penalty'] != 0.0 or state['frequency_penalty'] != 0.0:
+                 unordered_samplers.append(SS_PresFreqP(state['presence_penalty'], state['frequency_penalty'], penalty_range, rep_decay))
+
+            # Standard samplers
+            if state['top_k'] > 0:
+                unordered_samplers.append(SS_TopK(state['top_k']))
+            if state['top_p'] < 1.0:
+                unordered_samplers.append(SS_TopP(state['top_p']))
+            if state['min_p'] > 0.0:
+                unordered_samplers.append(SS_MinP(state['min_p']))
+
+            # Temperature
+            unordered_samplers.append(SS_Temperature(state['temperature']))
+
+            # 2. Define the mapping from class names to the priority list keys
+            class_name_to_nickname = {
+                'SS_RepP': 'repetition_penalty',
+                'SS_PresFreqP': 'presence_frequency_penalty',
+                'SS_TopK': 'top_k',
+                'SS_TopP': 'top_p',
+                'SS_MinP': 'min_p',
+                'SS_Temperature': 'temperature',
+            }
+
+            # 3. Get the priority list and handle temperature_last
+            default_priority = ['repetition_penalty', 'presence_frequency_penalty', 'top_k', 'top_p', 'min_p', 'temperature']
+            sampler_priority = state.get('sampler_priority', default_priority)
+
+            if state['temperature_last'] and 'temperature' in sampler_priority:
+                sampler_priority.append(sampler_priority.pop(sampler_priority.index('temperature')))
+
+            # 4. Sort the unordered list based on the priority list
+            def custom_sort_key(sampler_obj):
+                class_name = sampler_obj.__class__.__name__
+                nickname = class_name_to_nickname.get(class_name)
+                if nickname in sampler_priority:
+                    return sampler_priority.index(nickname)
+                return -1
+
+            ordered_samplers = sorted(unordered_samplers, key=custom_sort_key)
+
+            # 5. Add the final sampling stage and build the sampler
+            ordered_samplers.append(SS_Sample())
+            sampler = CustomSampler(ordered_samplers)
+        # -- End of sampler building --
 
         # Encode prompt with embeddings (ExLlamaV3-specific)
-        if image_embeddings:
-            input_ids = self.tokenizer.encode(
-                prompt,
-                encode_special_tokens=True,
-                embeddings=image_embeddings,
-            )
+        input_ids = self.tokenizer.encode(
+            prompt,
+            add_bos=state['add_bos_token'],
+            encode_special_tokens=True,
+            embeddings=image_embeddings,
+        )
+
+        input_ids = input_ids[:, -get_max_prompt_length(state):]
+
+        # Determine max_new_tokens
+        if state['auto_max_new_tokens']:
+            max_new_tokens = state['truncation_length'] - input_ids.shape[-1]
         else:
-            input_ids = self.tokenizer.encode(prompt, encode_special_tokens=True)
+            max_new_tokens = state['max_new_tokens']
 
-        # Get stop conditions from state (webui format) - keep as strings like ExLlamaV3 examples
+        # Get stop conditions
         stop_conditions = []
-        if 'stopping_strings' in state and state['stopping_strings']:
-            # Use strings directly (ExLlamaV3 handles the conversion internally)
-            stop_conditions.extend(state['stopping_strings'])
-
-        # Add EOS token ID as ExLlamaV3 examples do
-        if hasattr(self.tokenizer, 'eos_token_id') and self.tokenizer.eos_token_id is not None:
-            stop_conditions.append(self.tokenizer.eos_token_id)
+        if not state['ban_eos_token']:
+            if hasattr(self.tokenizer, 'eos_token_id') and self.tokenizer.eos_token_id is not None:
+                stop_conditions.append(self.tokenizer.eos_token_id)
 
         job = Job(
             input_ids=input_ids,
-            max_new_tokens=state.get('max_new_tokens', 500),
-            decode_special_tokens=True,
+            max_new_tokens=max_new_tokens,
+            decode_special_tokens=not state['skip_special_tokens'],
             embeddings=image_embeddings if image_embeddings else None,
             sampler=sampler,
             stop_conditions=stop_conditions if stop_conditions else None,
@@ -244,25 +307,16 @@ def generate_with_streaming(self, prompt, state):
             pass
 
     def generate(self, prompt, state):
-        """
-        Generate text using native ExLlamaV3 API (non-streaming)
-        """
-        output = self.generator.generate(
-            prompt=prompt,
-            max_new_tokens=state.get('max_new_tokens', 500),
-            temperature=state.get('temperature', 0.7),
-            top_p=state.get('top_p', 1.0),
-            top_k=state.get('top_k', 0),
-            repetition_penalty=state.get('repetition_penalty', 1.0),
-            frequency_penalty=state.get('frequency_penalty', 0.0),
-            presence_penalty=state.get('presence_penalty', 0.0),
-            min_p=state.get('min_p', 0.0),
-        )
+        output = ""
+        for chunk in self.generate_with_streaming(prompt, state):
+            output = chunk
 
         return output
 
     def encode(self, string, **kwargs):
-        return self.tokenizer.encode(string, **kwargs)
+        # Default add_bos to True for consistency with exllamav2 behavior
+        add_bos = kwargs.pop('add_bos', True)
+        return self.tokenizer.encode(string, add_bos=add_bos, **kwargs)
 
     def decode(self, ids, **kwargs):
         return self.tokenizer.decode(ids, **kwargs)
@@ -301,8 +355,6 @@ def unload(self):
 
         # Force GPU memory cleanup
         import gc
-
-        import torch
         gc.collect()
 
         if torch.cuda.is_available():
diff --git a/modules/loaders.py b/modules/loaders.py
index e9437c164c..151de990e5 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -255,38 +255,21 @@ def transformers_samplers():
     },
     'ExLlamav3': {
         'temperature',
-        'dynatemp_low',
-        'dynatemp_high',
-        'dynatemp_exponent',
-        'smoothing_factor',
         'min_p',
         'top_p',
         'top_k',
-        'typical_p',
-        'xtc_threshold',
-        'xtc_probability',
-        'tfs',
-        'top_a',
-        'dry_multiplier',
-        'dry_allowed_length',
-        'dry_base',
         'repetition_penalty',
         'frequency_penalty',
         'presence_penalty',
         'repetition_penalty_range',
-        'mirostat_mode',
-        'mirostat_tau',
-        'mirostat_eta',
-        'dynamic_temperature',
         'temperature_last',
+        'sampler_priority',
         'auto_max_new_tokens',
         'ban_eos_token',
         'add_bos_token',
         'enable_thinking',
-        'skip_special_tokens',
         'seed',
-        'custom_token_bans',
-        'dry_sequence_breakers',
+        'skip_special_tokens',
     },
     'ExLlamav2': {
         'temperature',

From 9e260332cc9da24a407bb59aadf0cf6a9cf0d88c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 8 Aug 2025 21:22:47 -0700
Subject: [PATCH 0863/1701] Remove some unnecessary code

---
 modules/exllamav3.py | 13 ++-----------
 1 file changed, 2 insertions(+), 11 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 295c273712..d616d2f5be 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -26,6 +26,7 @@
 from modules import shared
 from modules.logging_colors import logger
 from modules.text_generation import get_max_prompt_length
+from modules.torch_utils import clear_torch_cache
 
 try:
     import flash_attn
@@ -342,6 +343,7 @@ def unload(self):
                 del self.model
             except Exception as e:
                 logger.warning(f"Error unloading main model: {e}")
+
             self.model = None
 
         if hasattr(self, 'cache') and self.cache is not None:
@@ -352,14 +354,3 @@ def unload(self):
 
         if hasattr(self, 'tokenizer') and self.tokenizer is not None:
             self.tokenizer = None
-
-        # Force GPU memory cleanup
-        import gc
-        gc.collect()
-
-        if torch.cuda.is_available():
-            torch.cuda.empty_cache()
-            torch.cuda.synchronize()
-            torch.cuda.empty_cache()
-
-        logger.info("ExLlamaV3 model fully unloaded")

From 1168004067dbe37791e6911fc3ed3386d8131ce3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 9 Aug 2025 07:01:55 -0700
Subject: [PATCH 0864/1701] Minor change

---
 modules/exllamav3.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index d616d2f5be..f6c56cb064 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -6,7 +6,6 @@
 from exllamav3 import Cache, Config, Generator, Model, Tokenizer
 from exllamav3.cache import CacheLayer_fp16, CacheLayer_quant
 from exllamav3.generator import Job
-# Import the base sampler components directly from exllamav3
 from exllamav3.generator.sampler import (
     CustomSampler,
     SS_Argmax,

From 3f5ec9644f5aec2045126cdc5d962ee6f0b44c14 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 9 Aug 2025 07:06:07 -0700
Subject: [PATCH 0865/1701] mtmd: Place the image <__media__> at the top of the
 prompt

---
 modules/chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 988002394d..0a03a0840f 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -289,7 +289,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
                 if image_refs or attachments_text:
                     enhanced_user_msg = user_msg
                     if image_refs:
-                        enhanced_user_msg += f" {image_refs}"
+                        enhanced_user_msg = f"{image_refs}\n\n{enhanced_user_msg}"
                     if attachments_text:
                         enhanced_user_msg += f"\n\nATTACHMENTS:\n{attachments_text}"
 

From d9db8f63a719f799bac8f05ed567a1ba38041a72 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 9 Aug 2025 07:25:42 -0700
Subject: [PATCH 0866/1701] mtmd: Simplifications

---
 extensions/openai/completions.py | 33 +++++++-------------------------
 1 file changed, 7 insertions(+), 26 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 3d389f0b27..ff64527a3b 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -407,6 +407,10 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False):
     # Add messages to generate_params if present for multimodal processing
     if 'messages' in body:
         generate_params['messages'] = body['messages']
+        raw_images = convert_openai_messages_to_images(generate_params['messages'])
+        if raw_images:
+            logger.info(f"Found {len(raw_images)} image(s) in request.")
+            generate_params['raw_images'] = raw_images
 
     if not stream:
         prompt_arg = body[prompt_str]
@@ -423,7 +427,7 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False):
         total_prompt_token_count = 0
 
         for idx, prompt in enumerate(prompt_arg, start=0):
-            if isinstance(prompt[0], int):
+            if isinstance(prompt, list) and len(prompt) > 0 and isinstance(prompt[0], int):
                 # token lists
                 if requested_model == shared.model_name:
                     prompt = decode(prompt)[0]
@@ -438,19 +442,7 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False):
 
             # generate reply #######################################
             debug_msg({'prompt': prompt, 'generate_params': generate_params})
-
-            # Use multimodal generation if images are present
-            if 'messages' in generate_params:
-                raw_images = convert_openai_messages_to_images(generate_params['messages'])
-                if raw_images:
-                    logger.info(f"Using multimodal generation for {len(raw_images)} images")
-                    generate_params['raw_images'] = raw_images
-                    generator = shared.model.generate_with_streaming(prompt, generate_params)
-                else:
-                    generator = generate_reply(prompt, generate_params, is_chat=False)
-            else:
-                generator = generate_reply(prompt, generate_params, is_chat=False)
-
+            generator = generate_reply(prompt, generate_params, is_chat=False)
             answer = ''
 
             for a in generator:
@@ -523,18 +515,7 @@ def text_streaming_chunk(content):
 
         # generate reply #######################################
         debug_msg({'prompt': prompt, 'generate_params': generate_params})
-        # Use multimodal generation if images are present
-        if 'messages' in generate_params:
-            raw_images = convert_openai_messages_to_images(generate_params['messages'])
-            if raw_images:
-                logger.info(f"Using multimodal generation for {len(raw_images)} images")
-                generate_params['raw_images'] = raw_images
-                generator = shared.model.generate_with_streaming(prompt, generate_params)
-            else:
-                generator = generate_reply(prompt, generate_params, is_chat=False)
-        else:
-            generator = generate_reply(prompt, generate_params, is_chat=False)
-
+        generator = generate_reply(prompt, generate_params, is_chat=False)
         answer = ''
         seen_content = ''
         completion_token_count = 0

From fa9be444fa0b3e18763b1cd3d0dd07ad565ac1bb Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 9 Aug 2025 07:26:59 -0700
Subject: [PATCH 0867/1701] Use ExLlamav3 instead of ExLlamav3_HF by default
 for EXL3 models

---
 modules/models_settings.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 729d5dd108..d3bf4a36ec 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -211,11 +211,11 @@ def infer_loader(model_name, model_settings, hf_quant_method=None):
     elif re.match(r'.*\.gguf', model_name.lower()):
         loader = 'llama.cpp'
     elif hf_quant_method == 'exl3':
-        loader = 'ExLlamav3_HF'
+        loader = 'ExLlamav3'
     elif hf_quant_method in ['exl2', 'gptq']:
         loader = 'ExLlamav2_HF'
     elif re.match(r'.*exl3', model_name.lower()):
-        loader = 'ExLlamav3_HF'
+        loader = 'ExLlamav3'
     elif re.match(r'.*exl2', model_name.lower()):
         loader = 'ExLlamav2_HF'
     else:

From f396b82a4f92f5823ed2a9bd1ff32d915da4cf9a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 9 Aug 2025 07:31:36 -0700
Subject: [PATCH 0868/1701] mtmd: Better way to detect if an EXL3 model is
 multimodal

---
 modules/exllamav3.py | 17 ++++++++++-------
 1 file changed, 10 insertions(+), 7 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index f6c56cb064..70f6c2f1c6 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -98,13 +98,16 @@ def from_pretrained(cls, path_to_model):
 
         # Load vision model component (ExLlamaV3 native)
         vision_model = None
-        try:
-            logger.info("Loading vision model component...")
-            vision_model = Model.from_config(config, component="vision")
-            vision_model.load(progressbar=True)
-            logger.info("Vision model loaded successfully")
-        except Exception as e:
-            logger.warning(f"Vision model loading failed (multimodal disabled): {e}")
+        if "vision_config" in config.config_dict:
+            logger.info("Vision component detected in model config. Attempting to load...")
+            try:
+                vision_model = Model.from_config(config, component="vision")
+                vision_model.load(progressbar=True)
+                logger.info("Vision model loaded successfully.")
+            except Exception as e:
+                logger.warning(f"Vision model loading failed (multimodal disabled): {e}")
+        else:
+            logger.info("No vision component in model config. Skipping multimodal setup.")
 
         generator = Generator(
             model=model,

From 59c6138e989861020816041821369fd9cd6b0ffa Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 9 Aug 2025 07:32:15 -0700
Subject: [PATCH 0869/1701] Remove a log message

---
 modules/exllamav3.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 70f6c2f1c6..bdb0c76080 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -48,7 +48,6 @@ def from_pretrained(cls, path_to_model):
             global_allocator
         )
         global_allocator.next_token_index = FIRST_MM_EMBEDDING_INDEX
-        logger.info("Reset MMTokenAllocator for clean multimodal token allocation")
 
         config = Config.from_directory(str(path_to_model))
         model = Model.from_config(config)

From 2fe79a93ccf21121db3dd076050df93f08c5bdb9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 9 Aug 2025 07:50:24 -0700
Subject: [PATCH 0870/1701] mtmd: Handle another case after
 3f5ec9644f5aec2045126cdc5d962ee6f0b44c14

---
 modules/chat.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 0a03a0840f..b127b4895a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -326,7 +326,9 @@ def generate_chat_prompt(user_input, state, **kwargs):
                             attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
 
                 if image_refs or attachments_text:
-                    user_input = f"{user_input} {image_refs}"
+                    user_input = user_input
+                    if image_refs:
+                        user_input = f"{image_refs}\n\n{user_input}"
                     if attachments_text:
                         user_input += f"\n\nATTACHMENTS:\n{attachments_text}"
 

From a6d6bee88cf52dc4bbfaeac91df5a951810ab0dd Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 9 Aug 2025 07:51:03 -0700
Subject: [PATCH 0871/1701] Change a comment

---
 modules/chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index b127b4895a..639feebf62 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -813,7 +813,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
         for file_path in files:
             add_message_attachment(output, row_idx, file_path, is_user=True)
 
-        # Collect image attachments for ExLlamaV3
+        # Collect image attachments for multimodal generation
         image_attachments = []
         if 'metadata' in output:
             user_key = f"user_{row_idx}"

From d489eb589a698e37540861fe2951ef66efdb772d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 9 Aug 2025 14:10:41 -0700
Subject: [PATCH 0872/1701] Attempt at fixing new exllamav3 loader undefined
 behavior when switching conversations

---
 modules/exllamav3.py | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index bdb0c76080..e3a2d95a7d 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -304,9 +304,7 @@ def custom_sort_key(sampler_obj):
                         response_text += chunk
                         yield response_text
         finally:
-            # No cleanup needed. MMEmbedding lifetime is managed by Python.
-            # Cache and page table resets are unnecessary and can cause token ID conflicts.
-            pass
+            self.generator.clear_queue()
 
     def generate(self, prompt, state):
         output = ""

From a289a92b9408e2542632ffa600ef57c373200aec Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 9 Aug 2025 17:10:58 -0700
Subject: [PATCH 0873/1701] Fix exllamav3 token count

---
 modules/exllamav3.py | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index e3a2d95a7d..268a64ecc8 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -267,9 +267,11 @@ def custom_sort_key(sampler_obj):
 
         input_ids = input_ids[:, -get_max_prompt_length(state):]
 
+        self._last_prompt_token_count = input_ids.shape[-1]
+
         # Determine max_new_tokens
         if state['auto_max_new_tokens']:
-            max_new_tokens = state['truncation_length'] - input_ids.shape[-1]
+            max_new_tokens = state['truncation_length'] - self._last_prompt_token_count
         else:
             max_new_tokens = state['max_new_tokens']
 
@@ -323,8 +325,7 @@ def decode(self, ids, **kwargs):
 
     @property
     def last_prompt_token_count(self):
-        # This would need to be tracked during generation
-        return 0
+        return getattr(self, '_last_prompt_token_count', 0)
 
     def unload(self):
         logger.info("Unloading ExLlamaV3 model components...")

From eb16f6401794340c751e8105bc3837967b9e054e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 9 Aug 2025 17:12:16 -0700
Subject: [PATCH 0874/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_cuda128.txt             | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt      | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index f17cae8a2c..323ef0f92e 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 51f4571fa3..2a7c936156 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 37021c7727..0106fbea10 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index f54ae19185..d5db4a1cff 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index e495455b11..694f1ff895 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 72847534ca..392637e2dc 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index ed641a2408..88eaa93042 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index d7fe735b55..6accc2f049 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index cb71f74b5b..3025f092ae 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index d6bed57630..7394bdcfc2 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 1f17dc5028..a095a4c709 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 82254842d6..ea43e56eb8 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 986a3d493c..79737728ca 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -18,6 +18,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 833e923bec..d39786bd82 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 6a894d49ee..0b373fa972 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 0afb19c26d..fe9dccac76 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index a404f50c57..b3cfd525f1 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 75176656f7..02aa03e3b0 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.33.0/llama_cpp_binaries-0.33.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From d86b0ec0103f24f27329a600d1cbaf3a5ea4c517 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Sun, 10 Aug 2025 01:27:25 -0300
Subject: [PATCH 0875/1701] Add multimodal support (llama.cpp) (#7027)

---
 extensions/openai/image_utils.py            |  9 ++++
 modules/chat.py                             | 26 ++++++------
 modules/llama_cpp_server.py                 | 46 ++++++++++++++++++---
 modules/loaders.py                          |  2 +
 modules/shared.py                           |  1 +
 modules/ui.py                               |  1 +
 modules/ui_model_menu.py                    |  6 +++
 modules/utils.py                            | 13 ++++++
 user_data/mmproj/place-your-mmproj-here.txt |  0
 9 files changed, 86 insertions(+), 18 deletions(-)
 create mode 100644 user_data/mmproj/place-your-mmproj-here.txt

diff --git a/extensions/openai/image_utils.py b/extensions/openai/image_utils.py
index c54f053294..658f00d74d 100644
--- a/extensions/openai/image_utils.py
+++ b/extensions/openai/image_utils.py
@@ -11,6 +11,15 @@
 from modules.logging_colors import logger
 
 
+def convert_pil_to_base64(image: Image.Image) -> str:
+    """Converts a PIL Image to a base64 encoded string."""
+    buffered = io.BytesIO()
+    # Save image to an in-memory bytes buffer in PNG format
+    image.save(buffered, format="PNG")
+    # Encode the bytes to a base64 string
+    return base64.b64encode(buffered.getvalue()).decode('utf-8')
+
+
 def decode_base64_image(base64_string: str) -> Image.Image:
     """Decodes a base64 string to a PIL Image."""
     try:
diff --git a/modules/chat.py b/modules/chat.py
index 639feebf62..696fa350d4 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -813,19 +813,6 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
         for file_path in files:
             add_message_attachment(output, row_idx, file_path, is_user=True)
 
-        # Collect image attachments for multimodal generation
-        image_attachments = []
-        if 'metadata' in output:
-            user_key = f"user_{row_idx}"
-            if user_key in output['metadata'] and "attachments" in output['metadata'][user_key]:
-                for attachment in output['metadata'][user_key]["attachments"]:
-                    if attachment.get("type") == "image":
-                        image_attachments.append(attachment)
-
-        # Add image attachments to state for the generation
-        if image_attachments:
-            state['image_attachments'] = image_attachments
-
         # Add web search results as attachments if enabled
         if state.get('enable_web_search', False):
             search_query = generate_search_query(text, state)
@@ -881,6 +868,19 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
                     'metadata': output['metadata']
                 }
 
+    # Collect image attachments for multimodal generation
+    image_attachments = []
+    if 'metadata' in output:
+        user_key = f"user_{row_idx}"
+        if user_key in output['metadata'] and "attachments" in output['metadata'][user_key]:
+            for attachment in output['metadata'][user_key]["attachments"]:
+                if attachment.get("type") == "image":
+                    image_attachments.append(attachment)
+
+    # Add image attachments to state for the generation
+    if image_attachments:
+        state['image_attachments'] = image_attachments
+
     # Generate the prompt
     kwargs = {
         '_continue': _continue,
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index e64f16944a..072ff83bf5 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -12,6 +12,10 @@
 import llama_cpp_binaries
 import requests
 
+from extensions.openai.image_utils import (
+    convert_image_attachments_to_pil,
+    convert_pil_to_base64
+)
 from modules import shared
 from modules.logging_colors import logger
 
@@ -128,15 +132,40 @@ def generate_with_streaming(self, prompt, state):
         url = f"http://127.0.0.1:{self.port}/completion"
         payload = self.prepare_payload(state)
 
-        token_ids = self.encode(prompt, add_bos_token=state["add_bos_token"])
-        self.last_prompt_token_count = len(token_ids)
+        pil_images = []
+        # Check for images from the Web UI (image_attachments)
+        if 'image_attachments' in state and state['image_attachments']:
+            pil_images.extend(convert_image_attachments_to_pil(state['image_attachments']))
+        # Else, check for images from the API (raw_images)
+        elif 'raw_images' in state and state['raw_images']:
+            pil_images.extend(state.get('raw_images', []))
+
+        if pil_images:
+            # Multimodal case
+            IMAGE_TOKEN_COST_ESTIMATE = 600  # A safe, conservative estimate per image
+
+            base64_images = [convert_pil_to_base64(img) for img in pil_images]
+            multimodal_prompt_object = {
+                "prompt": prompt,
+                "multimodal_data": base64_images
+            }
+            payload["prompt"] = multimodal_prompt_object
+
+            # Calculate an estimated token count
+            text_tokens = self.encode(prompt, add_bos_token=state["add_bos_token"])
+            self.last_prompt_token_count = len(text_tokens) + (len(pil_images) * IMAGE_TOKEN_COST_ESTIMATE)
+        else:
+            # Text only case
+            token_ids = self.encode(prompt, add_bos_token=state["add_bos_token"])
+            self.last_prompt_token_count = len(token_ids)
+            payload["prompt"] = token_ids
+
         if state['auto_max_new_tokens']:
-            max_new_tokens = state['truncation_length'] - len(token_ids)
+            max_new_tokens = state['truncation_length'] - self.last_prompt_token_count
         else:
             max_new_tokens = state['max_new_tokens']
 
         payload.update({
-            "prompt": token_ids,
             "n_predict": max_new_tokens,
             "stream": True,
             "cache_prompt": True
@@ -144,7 +173,7 @@ def generate_with_streaming(self, prompt, state):
 
         if shared.args.verbose:
             logger.info("GENERATE_PARAMS=")
-            printable_payload = {k: v for k, v in payload.items() if k != "prompt"}
+            printable_payload = {k: (v if k != "prompt" else "[multimodal object]" if pil_images else v) for k, v in payload.items()}
             pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
             print()
 
@@ -295,6 +324,13 @@ def _start_server(self):
             cmd += ["--rope-freq-scale", str(1.0 / shared.args.compress_pos_emb)]
         if shared.args.rope_freq_base > 0:
             cmd += ["--rope-freq-base", str(shared.args.rope_freq_base)]
+        if shared.args.mmproj not in [None, 'None']:
+            path = Path(shared.args.mmproj)
+            if not path.exists():
+                path = Path('user_data/mmproj') / shared.args.mmproj
+
+            if path.exists():
+                cmd += ["--mmproj", str(path)]
         if shared.args.model_draft not in [None, 'None']:
             path = Path(shared.args.model_draft)
             if not path.exists():
diff --git a/modules/loaders.py b/modules/loaders.py
index 151de990e5..feca9985a6 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -28,6 +28,8 @@
         'device_draft',
         'ctx_size_draft',
         'speculative_decoding_accordion',
+        'mmproj',
+        'mmproj_accordion',
         'vram_info',
     ],
     'Transformers': [
diff --git a/modules/shared.py b/modules/shared.py
index 1de4306b37..e9d8a62fdb 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -85,6 +85,7 @@
 group.add_argument('--row-split', action='store_true', help='Split the model by rows across GPUs. This may improve multi-gpu performance.')
 group.add_argument('--extra-flags', type=str, default=None, help='Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"')
 group.add_argument('--streaming-llm', action='store_true', help='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
+group.add_argument('--mmproj', type=str, default=None, help='Path to the mmproj file for vision models.')
 
 # Cache
 group = parser.add_argument_group('Context and cache')
diff --git a/modules/ui.py b/modules/ui.py
index e7805046cc..1171cd48eb 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -167,6 +167,7 @@ def list_model_elements():
         'gpu_layers_draft',
         'device_draft',
         'ctx_size_draft',
+        'mmproj',
     ]
 
     return elements
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 0ab67e7a5b..9fa8a4f40c 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -59,6 +59,12 @@ def create_ui():
                             shared.gradio['trust_remote_code'] = gr.Checkbox(label="trust-remote-code", value=shared.args.trust_remote_code, info='Set trust_remote_code=True while loading the tokenizer/model. To enable this option, start the web UI with the --trust-remote-code flag.', interactive=shared.args.trust_remote_code)
                             shared.gradio['tensorrt_llm_info'] = gr.Markdown('* TensorRT-LLM has to be installed manually in a separate Python 3.10 environment at the moment. For a guide, consult the description of [this PR](https://github.com/oobabooga/text-generation-webui/pull/5715). \n\n* `ctx_size` is only used when `cpp-runner` is checked.\n\n* `cpp_runner` does not support streaming at the moment.')
 
+                            # Multimodal
+                            with gr.Accordion("Multimodal (vision)", open=False, elem_classes='tgw-accordion') as shared.gradio['mmproj_accordion']:
+                                with gr.Row():
+                                    shared.gradio['mmproj'] = gr.Dropdown(label="mmproj file", choices=utils.get_available_mmproj(), value=lambda: shared.args.mmproj or 'None', elem_classes='slim-dropdown', info='Select a file that matches your model. Must be placed in user_data/mmproj/', interactive=not mu)
+                                    ui.create_refresh_button(shared.gradio['mmproj'], lambda: None, lambda: {'choices': utils.get_available_mmproj()}, 'refresh-button', interactive=not mu)
+
                             # Speculative decoding
                             with gr.Accordion("Speculative decoding", open=False, elem_classes='tgw-accordion') as shared.gradio['speculative_decoding_accordion']:
                                 with gr.Row():
diff --git a/modules/utils.py b/modules/utils.py
index 117ad590a4..4927ef04d4 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -154,6 +154,19 @@ def get_available_ggufs():
     return sorted(model_list, key=natural_keys)
 
 
+def get_available_mmproj():
+    mmproj_dir = Path('user_data/mmproj')
+    if not mmproj_dir.exists():
+        return ['None']
+
+    mmproj_files = []
+    for item in mmproj_dir.iterdir():
+        if item.is_file() and item.suffix.lower() in ('.gguf', '.bin'):
+            mmproj_files.append(item.name)
+
+    return ['None'] + sorted(mmproj_files, key=natural_keys)
+
+
 def get_available_presets():
     return sorted(set((k.stem for k in Path('user_data/presets').glob('*.yaml'))), key=natural_keys)
 
diff --git a/user_data/mmproj/place-your-mmproj-here.txt b/user_data/mmproj/place-your-mmproj-here.txt
new file mode 100644
index 0000000000..e69de29bb2

From c6b4d1e87f67dd990c66eaecb35c9cc70d0ae4e3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 9 Aug 2025 21:33:12 -0700
Subject: [PATCH 0876/1701] Fix the exllamav2 loader ignoring add_bos

---
 modules/exllamav2.py | 3 ++-
 modules/exllamav3.py | 1 -
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/exllamav2.py b/modules/exllamav2.py
index 6bb422eaec..5d5c5b567c 100644
--- a/modules/exllamav2.py
+++ b/modules/exllamav2.py
@@ -135,7 +135,8 @@ def from_pretrained(self, path_to_model):
         return result, result
 
     def encode(self, string, **kwargs):
-        return self.tokenizer.encode(string, add_bos=True, encode_special_tokens=True)
+        add_bos = kwargs.pop('add_bos', True)
+        return self.tokenizer.encode(string, add_bos=add_bos, encode_special_tokens=True, **kwargs)
 
     def decode(self, ids, **kwargs):
         if isinstance(ids, list):
diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 268a64ecc8..9201801c1b 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -316,7 +316,6 @@ def generate(self, prompt, state):
         return output
 
     def encode(self, string, **kwargs):
-        # Default add_bos to True for consistency with exllamav2 behavior
         add_bos = kwargs.pop('add_bos', True)
         return self.tokenizer.encode(string, add_bos=add_bos, **kwargs)
 

From 2f90ac98807a4ffa6a761bbcef5cf81a9de568b8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 9 Aug 2025 21:41:38 -0700
Subject: [PATCH 0877/1701] Move the new image_utils.py file to modules/

---
 extensions/openai/completions.py              | 2 +-
 modules/exllamav3.py                          | 4 ++--
 {extensions/openai => modules}/image_utils.py | 0
 modules/llama_cpp_server.py                   | 4 ++--
 4 files changed, 5 insertions(+), 5 deletions(-)
 rename {extensions/openai => modules}/image_utils.py (100%)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index ff64527a3b..f49440603c 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -7,7 +7,6 @@
 from pydantic import ValidationError
 
 from extensions.openai.errors import InvalidRequestError
-from extensions.openai.image_utils import convert_openai_messages_to_images
 from extensions.openai.typing import ToolDefinition
 from extensions.openai.utils import debug_msg, getToolCallId, parseToolCall
 from modules import shared
@@ -17,6 +16,7 @@
     load_character_memoized,
     load_instruction_template_memoized
 )
+from modules.image_utils import convert_openai_messages_to_images
 from modules.logging_colors import logger
 from modules.presets import load_preset_memoized
 from modules.text_generation import decode, encode, generate_reply
diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 9201801c1b..9d597ce780 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -18,11 +18,11 @@
     SS_TopP
 )
 
-from extensions.openai.image_utils import (
+from modules import shared
+from modules.image_utils import (
     convert_image_attachments_to_pil,
     convert_openai_messages_to_images
 )
-from modules import shared
 from modules.logging_colors import logger
 from modules.text_generation import get_max_prompt_length
 from modules.torch_utils import clear_torch_cache
diff --git a/extensions/openai/image_utils.py b/modules/image_utils.py
similarity index 100%
rename from extensions/openai/image_utils.py
rename to modules/image_utils.py
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 072ff83bf5..3e8127ab18 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -12,11 +12,11 @@
 import llama_cpp_binaries
 import requests
 
-from extensions.openai.image_utils import (
+from modules import shared
+from modules.image_utils import (
     convert_image_attachments_to_pil,
     convert_pil_to_base64
 )
-from modules import shared
 from modules.logging_colors import logger
 
 llamacpp_valid_cache_types = {"fp16", "q8_0", "q4_0"}

From 4663b1a56e99b8d637f9ac67c8b8d0e09d496ec7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 9 Aug 2025 21:45:50 -0700
Subject: [PATCH 0878/1701] Update docs

---
 docs/12 - OpenAI API.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index b7b5fbc18d..fc76cd8b50 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -77,7 +77,7 @@ curl http://127.0.0.1:5000/v1/chat/completions \
   }'
 ```
 
-#### Multimodal support (ExLlamaV3)
+#### Multimodal support (llama.cpp and ExLlamaV3)
 
 ```shell
 curl http://127.0.0.1:5000/v1/chat/completions \

From 0ea62d88f60689b44dd4ee42ae9ba0ff871a29c2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 9 Aug 2025 21:47:02 -0700
Subject: [PATCH 0879/1701] mtmd: Fix "continue" when an image is present

---
 modules/chat.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 696fa350d4..42bb58a5b1 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -868,6 +868,8 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
                     'metadata': output['metadata']
                 }
 
+    row_idx = len(output['internal']) - 1
+
     # Collect image attachments for multimodal generation
     image_attachments = []
     if 'metadata' in output:
@@ -895,7 +897,6 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
         prompt = generate_chat_prompt(text, state, **kwargs)
 
     # Add timestamp for assistant's response at the start of generation
-    row_idx = len(output['internal']) - 1
     update_message_metadata(output['metadata'], "assistant", row_idx, timestamp=get_current_timestamp(), model_name=shared.model_name)
 
     # Generate

From 1fb580785937ad42e8657a5fd894dfcd5a1fdeb3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 10 Aug 2025 06:54:44 -0700
Subject: [PATCH 0880/1701] mtmd: Fix API text completion when no images are
 sent

---
 extensions/openai/completions.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index f49440603c..6f4dfc29f1 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -405,7 +405,7 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False):
     echo = body['echo']
 
     # Add messages to generate_params if present for multimodal processing
-    if 'messages' in body:
+    if body.get('messages'):
         generate_params['messages'] = body['messages']
         raw_images = convert_openai_messages_to_images(generate_params['messages'])
         if raw_images:

From 6fbf162d712cef876b128651fdebeb08a4f32538 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 10 Aug 2025 07:21:55 -0700
Subject: [PATCH 0881/1701] Default max_tokens to 512 in the API instead of 16

---
 extensions/openai/typing.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index e9f92da561..9036627018 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -106,7 +106,7 @@ class CompletionRequestParams(BaseModel):
     frequency_penalty: float | None = 0
     logit_bias: dict | None = None
     logprobs: int | None = None
-    max_tokens: int | None = 16
+    max_tokens: int | None = 512
     n: int | None = Field(default=1, description="Unused parameter.")
     presence_penalty: float | None = 0
     stop: str | List[str] | None = None
@@ -232,7 +232,7 @@ class LogitsRequestParams(BaseModel):
     use_samplers: bool = False
     top_logits: int | None = 50
     frequency_penalty: float | None = 0
-    max_tokens: int | None = 16
+    max_tokens: int | None = 512
     presence_penalty: float | None = 0
     temperature: float | None = 1
     top_p: float | None = 1

From cc964ee579463d4c3acb35c188ee8eb38e23ce1a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 10 Aug 2025 07:44:38 -0700
Subject: [PATCH 0882/1701] mtmd: Increase the size of the UI image preview

---
 css/main.css | 1 +
 1 file changed, 1 insertion(+)

diff --git a/css/main.css b/css/main.css
index de16d81d2c..062d3eb215 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1579,6 +1579,7 @@ strong {
 
 .image-attachment {
     flex-direction: column;
+    max-width: 314px;
 }
 
 .image-preview {

From 9ec310d858824d5f9186bca277a3ab77ac556b75 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 10 Aug 2025 07:54:21 -0700
Subject: [PATCH 0883/1701] UI: Fix the color of italic text

---
 css/html_instruct_style.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index 9831ee8fb9..3e5ebe67c8 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -13,7 +13,7 @@
     line-height: 28px !important;
 }
 
-.dark .chat .message-body :is(p, li, q, h1, h2, h3, h4, h5, h6) {
+.dark .chat .message-body :is(p, li, q, em, h1, h2, h3, h4, h5, h6) {
     color: #d1d5db !important;
 }
 

From c5340533c0b3a9edaea6c253f99250f09f2c26a5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 10 Aug 2025 20:39:04 -0700
Subject: [PATCH 0884/1701] mtmd: Add another API example

---
 docs/12 - OpenAI API.md | 38 ++++++++++++++++++++++++++++++++++++--
 1 file changed, 36 insertions(+), 2 deletions(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index fc76cd8b50..5dc98a5124 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -77,7 +77,9 @@ curl http://127.0.0.1:5000/v1/chat/completions \
   }'
 ```
 
-#### Multimodal support (llama.cpp and ExLlamaV3)
+#### Multimodal/vision (llama.cpp and ExLlamaV3)
+
+##### /v1/chat/completions (recommended!)
 
 ```shell
 curl http://127.0.0.1:5000/v1/chat/completions \
@@ -87,7 +89,7 @@ curl http://127.0.0.1:5000/v1/chat/completions \
       {
         "role": "user",
         "content": [
-          {"type": "text", "text": "What color is this image?"},
+          {"type": "text", "text": "Please describe what you see in this image."},
           {"type": "image_url", "image_url": {"url": "https://github.com/turboderp-org/exllamav3/blob/master/examples/media/cat.png?raw=true"}}
         ]
       }
@@ -95,6 +97,38 @@ curl http://127.0.0.1:5000/v1/chat/completions \
   }'
 ```
 
+##### /v1/completions
+
+```shell
+curl http://127.0.0.1:5000/v1/completions \
+  -H "Content-Type: application/json" \
+  -d '{
+    "messages": [
+      {
+        "role": "user",
+        "content": [
+          {
+            "type": "text",
+            "text": "About image <__media__> and image <__media__>, what I can say is that the first one"
+          },
+          {
+            "type": "image_url",
+            "image_url": {
+              "url": "https://github.com/turboderp-org/exllamav3/blob/master/examples/media/cat.png?raw=true"
+            }
+          },
+          {
+            "type": "image_url",
+            "image_url": {
+              "url": "https://github.com/turboderp-org/exllamav3/blob/master/examples/media/strawberry.png?raw=true"
+            }
+          }
+        ]
+      }
+    ]
+  }'
+```
+
 #### SSE streaming
 
 ```shell

From 4d8dbbab648d14680741324d187bee23a8bea486 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 07:26:11 -0700
Subject: [PATCH 0885/1701] API: Fix sampler_priority usage for ExLlamaV3

---
 modules/exllamav3.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 9d597ce780..8f68666942 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -237,7 +237,7 @@ def generate_with_streaming(self, prompt, state):
 
             # 3. Get the priority list and handle temperature_last
             default_priority = ['repetition_penalty', 'presence_frequency_penalty', 'top_k', 'top_p', 'min_p', 'temperature']
-            sampler_priority = state.get('sampler_priority', default_priority)
+            sampler_priority = state.get('sampler_priority') or default_priority
 
             if state['temperature_last'] and 'temperature' in sampler_priority:
                 sampler_priority.append(sampler_priority.pop(sampler_priority.index('temperature')))

From 4809ddfeb85e8b8d28bb617366c86fd8037815ee Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 07:35:22 -0700
Subject: [PATCH 0886/1701] Exllamav3: small sampler fixes

---
 modules/exllamav3.py | 14 ++++++--------
 1 file changed, 6 insertions(+), 8 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 8f68666942..5c142ec25f 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -6,7 +6,9 @@
 from exllamav3 import Cache, Config, Generator, Model, Tokenizer
 from exllamav3.cache import CacheLayer_fp16, CacheLayer_quant
 from exllamav3.generator import Job
-from exllamav3.generator.sampler import (
+
+from modules import shared
+from modules.exllamav3_custom_sampler import (
     CustomSampler,
     SS_Argmax,
     SS_MinP,
@@ -17,8 +19,6 @@
     SS_TopK,
     SS_TopP
 )
-
-from modules import shared
 from modules.image_utils import (
     convert_image_attachments_to_pil,
     convert_openai_messages_to_images
@@ -194,7 +194,6 @@ def generate_with_streaming(self, prompt, state):
         # Process images and modify prompt (ExLlamaV3-specific)
         prompt, image_embeddings = self._process_images_for_generation(prompt, state)
 
-        # -- Manually build and sort the sampler stack --
         # Greedy decoding is a special case
         if state['temperature'] == 0:
             sampler = CustomSampler([SS_Argmax()])
@@ -205,7 +204,7 @@ def generate_with_streaming(self, prompt, state):
             # Penalties
             penalty_range = state['repetition_penalty_range']
             if penalty_range <= 0:
-                penalty_range = -1 # ExllamaV3 uses -1 for whole context
+                penalty_range = int(10e7)  # Use large number for "full context"
             rep_decay = 0 # Not a configurable parameter
 
             # Add penalty samplers if they are active
@@ -222,7 +221,7 @@ def generate_with_streaming(self, prompt, state):
             if state['min_p'] > 0.0:
                 unordered_samplers.append(SS_MinP(state['min_p']))
 
-            # Temperature
+            # Temperature (SS_NoOp is returned if temp is 1.0)
             unordered_samplers.append(SS_Temperature(state['temperature']))
 
             # 2. Define the mapping from class names to the priority list keys
@@ -246,7 +245,7 @@ def generate_with_streaming(self, prompt, state):
             def custom_sort_key(sampler_obj):
                 class_name = sampler_obj.__class__.__name__
                 nickname = class_name_to_nickname.get(class_name)
-                if nickname in sampler_priority:
+                if nickname and nickname in sampler_priority:
                     return sampler_priority.index(nickname)
                 return -1
 
@@ -255,7 +254,6 @@ def custom_sort_key(sampler_obj):
             # 5. Add the final sampling stage and build the sampler
             ordered_samplers.append(SS_Sample())
             sampler = CustomSampler(ordered_samplers)
-        # -- End of sampler building --
 
         # Encode prompt with embeddings (ExLlamaV3-specific)
         input_ids = self.tokenizer.encode(

From 1cb800d3923093e102470c6dde4e4a8b451e0a33 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 07:37:10 -0700
Subject: [PATCH 0887/1701] Docs: small change

---
 docs/12 - OpenAI API.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index 5dc98a5124..fd3309c7ad 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -79,7 +79,7 @@ curl http://127.0.0.1:5000/v1/chat/completions \
 
 #### Multimodal/vision (llama.cpp and ExLlamaV3)
 
-##### /v1/chat/completions (recommended!)
+##### With /v1/chat/completions (recommended!)
 
 ```shell
 curl http://127.0.0.1:5000/v1/chat/completions \
@@ -97,7 +97,7 @@ curl http://127.0.0.1:5000/v1/chat/completions \
   }'
 ```
 
-##### /v1/completions
+##### With /v1/completions
 
 ```shell
 curl http://127.0.0.1:5000/v1/completions \

From 52d1cbbbe95bed853241ab422b0558ab029d7d08 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 07:38:39 -0700
Subject: [PATCH 0888/1701] Fix an import

---
 modules/exllamav3.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 5c142ec25f..3fabdb6b48 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -8,7 +8,7 @@
 from exllamav3.generator import Job
 
 from modules import shared
-from modules.exllamav3_custom_sampler import (
+from exllamav3.generator.sampler import (
     CustomSampler,
     SS_Argmax,
     SS_MinP,

From 38c0b4a1adc613e9e3a237835faa1d88632733ef Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 07:39:53 -0700
Subject: [PATCH 0889/1701] Default ctx-size to 8192 when not found in the
 metadata

---
 modules/models_settings.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index d3bf4a36ec..bf7b1cf916 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -15,7 +15,7 @@
 def get_fallback_settings():
     return {
         'bf16': False,
-        'ctx_size': 2048,
+        'ctx_size': 8192,
         'rope_freq_base': 0,
         'compress_pos_emb': 1,
         'alpha_value': 1,

From b62c8845f34f3faac2481e368e6a4c67fd33fa59 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 08:22:17 -0700
Subject: [PATCH 0890/1701] mtmd: Fix /chat/completions for llama.cpp

---
 extensions/openai/completions.py | 18 +++++++++++++++---
 modules/chat.py                  | 23 ++++++++++++-----------
 modules/llama_cpp_server.py      |  8 ++++++--
 3 files changed, 33 insertions(+), 16 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 6f4dfc29f1..c3037d0c24 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -85,16 +85,28 @@ def process_parameters(body, is_legacy=False):
 
 
 def process_multimodal_content(content):
-    """Extract text from OpenAI multimodal format for non-multimodal models"""
+    """Extract text and add image placeholders from OpenAI multimodal format"""
     if isinstance(content, str):
         return content
 
     if isinstance(content, list):
         text_parts = []
+        image_placeholders = ""
         for item in content:
-            if isinstance(item, dict) and item.get('type') == 'text':
+            if not isinstance(item, dict):
+                continue
+
+            item_type = item.get('type', '')
+            if item_type == 'text':
                 text_parts.append(item.get('text', ''))
-        return ' '.join(text_parts) if text_parts else str(content)
+            elif item_type == 'image_url':
+                image_placeholders += "<__media__>"
+
+        final_text = ' '.join(text_parts)
+        if image_placeholders:
+            return f"{image_placeholders}\n\n{final_text}"
+        else:
+            return final_text
 
     return str(content)
 
diff --git a/modules/chat.py b/modules/chat.py
index 42bb58a5b1..7b1629dd5b 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -870,18 +870,19 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
 
     row_idx = len(output['internal']) - 1
 
-    # Collect image attachments for multimodal generation
-    image_attachments = []
+    # Collect image attachments for multimodal generation from the entire history
+    all_image_attachments = []
     if 'metadata' in output:
-        user_key = f"user_{row_idx}"
-        if user_key in output['metadata'] and "attachments" in output['metadata'][user_key]:
-            for attachment in output['metadata'][user_key]["attachments"]:
-                if attachment.get("type") == "image":
-                    image_attachments.append(attachment)
-
-    # Add image attachments to state for the generation
-    if image_attachments:
-        state['image_attachments'] = image_attachments
+        for i in range(len(output['internal'])):
+            user_key = f"user_{i}"
+            if user_key in output['metadata'] and "attachments" in output['metadata'][user_key]:
+                for attachment in output['metadata'][user_key]["attachments"]:
+                    if attachment.get("type") == "image":
+                        all_image_attachments.append(attachment)
+
+    # Add all collected image attachments to state for the generation
+    if all_image_attachments:
+        state['image_attachments'] = all_image_attachments
 
     # Generate the prompt
     kwargs = {
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 3e8127ab18..63c8eda087 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -15,6 +15,7 @@
 from modules import shared
 from modules.image_utils import (
     convert_image_attachments_to_pil,
+    convert_openai_messages_to_images,
     convert_pil_to_base64
 )
 from modules.logging_colors import logger
@@ -133,10 +134,13 @@ def generate_with_streaming(self, prompt, state):
         payload = self.prepare_payload(state)
 
         pil_images = []
-        # Check for images from the Web UI (image_attachments)
+        # Source 1: Web UI (from chatbot_wrapper)
         if 'image_attachments' in state and state['image_attachments']:
             pil_images.extend(convert_image_attachments_to_pil(state['image_attachments']))
-        # Else, check for images from the API (raw_images)
+        # Source 2: Chat Completions API (/v1/chat/completions)
+        elif 'history' in state and state.get('history', {}).get('messages'):
+            pil_images.extend(convert_openai_messages_to_images(state['history']['messages']))
+        # Source 3: Legacy Completions API (/v1/completions)
         elif 'raw_images' in state and state['raw_images']:
             pil_images.extend(state.get('raw_images', []))
 

From b10d525bf7618c415c88937e26c1a2240c3b2fcf Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 12:05:22 -0700
Subject: [PATCH 0891/1701] UI: Update a tooltip

---
 js/main.js | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index e0f9314dad..66a344b318 100644
--- a/js/main.js
+++ b/js/main.js
@@ -977,7 +977,7 @@ if (document.readyState === "loading") {
 //------------------------------------------------
 
 // File upload button
-document.querySelector("#chat-input .upload-button").title = "Upload text files, PDFs, and DOCX documents";
+document.querySelector("#chat-input .upload-button").title = "Upload text files, PDFs, DOCX documents, and images";
 
 // Activate web search
 document.getElementById("web-search").title = "Search the internet with DuckDuckGo";

From 1ba1211ca027b887f160f6894a004b7d96ea0eee Mon Sep 17 00:00:00 2001
From: Mykeehu 
Date: Mon, 11 Aug 2025 21:13:56 +0200
Subject: [PATCH 0892/1701] Fix edit window and buttons in Messenger theme
 (#7100)

---
 css/chat_style-messenger.css | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/css/chat_style-messenger.css b/css/chat_style-messenger.css
index 65af5f7a5f..583703c0db 100644
--- a/css/chat_style-messenger.css
+++ b/css/chat_style-messenger.css
@@ -99,3 +99,9 @@
 .message-body p em {
     color: rgb(110 110 110) !important;
 }
+.editing-textarea {
+    width: max(30rem) !important;
+}
+.circle-you + .text .edit-control-button, .circle-you + .text .editing-textarea {
+    color: #000 !important;
+}

From 999471256c0626bb29e9caa65bbf96b8d2cb52d6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 12:32:17 -0700
Subject: [PATCH 0893/1701] Lint

---
 modules/exllamav2.py |  2 +-
 modules/exllamav3.py | 11 ++++-------
 2 files changed, 5 insertions(+), 8 deletions(-)

diff --git a/modules/exllamav2.py b/modules/exllamav2.py
index 5d5c5b567c..3b3233d25b 100644
--- a/modules/exllamav2.py
+++ b/modules/exllamav2.py
@@ -3,6 +3,7 @@
 from pathlib import Path
 
 import torch
+
 from exllamav2 import (
     ExLlamaV2,
     ExLlamaV2Cache,
@@ -15,7 +16,6 @@
     ExLlamaV2Tokenizer
 )
 from exllamav2.generator import ExLlamaV2Sampler, ExLlamaV2StreamingGenerator
-
 from modules import shared
 from modules.logging_colors import logger
 from modules.text_generation import get_max_prompt_length
diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 3fabdb6b48..980230f814 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -2,12 +2,9 @@
 from pathlib import Path
 from typing import Any, List, Tuple
 
-import torch
 from exllamav3 import Cache, Config, Generator, Model, Tokenizer
 from exllamav3.cache import CacheLayer_fp16, CacheLayer_quant
 from exllamav3.generator import Job
-
-from modules import shared
 from exllamav3.generator.sampler import (
     CustomSampler,
     SS_Argmax,
@@ -19,13 +16,13 @@
     SS_TopK,
     SS_TopP
 )
+from modules import shared
 from modules.image_utils import (
     convert_image_attachments_to_pil,
     convert_openai_messages_to_images
 )
 from modules.logging_colors import logger
 from modules.text_generation import get_max_prompt_length
-from modules.torch_utils import clear_torch_cache
 
 try:
     import flash_attn
@@ -205,13 +202,13 @@ def generate_with_streaming(self, prompt, state):
             penalty_range = state['repetition_penalty_range']
             if penalty_range <= 0:
                 penalty_range = int(10e7)  # Use large number for "full context"
-            rep_decay = 0 # Not a configurable parameter
+            rep_decay = 0  # Not a configurable parameter
 
             # Add penalty samplers if they are active
             if state['repetition_penalty'] != 1.0:
-                 unordered_samplers.append(SS_RepP(state['repetition_penalty'], penalty_range, rep_decay))
+                unordered_samplers.append(SS_RepP(state['repetition_penalty'], penalty_range, rep_decay))
             if state['presence_penalty'] != 0.0 or state['frequency_penalty'] != 0.0:
-                 unordered_samplers.append(SS_PresFreqP(state['presence_penalty'], state['frequency_penalty'], penalty_range, rep_decay))
+                unordered_samplers.append(SS_PresFreqP(state['presence_penalty'], state['frequency_penalty'], penalty_range, rep_decay))
 
             # Standard samplers
             if state['top_k'] > 0:

From a78ca6ffcdf0c53efdce8bfa6b37825590f5ae6e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 12:33:38 -0700
Subject: [PATCH 0894/1701] Remove a comment

---
 modules/text_generation.py | 2 --
 1 file changed, 2 deletions(-)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index d6a87ce8b1..27c5de7dff 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -295,8 +295,6 @@ def generate_reply_HF(question, original_question, state, stopping_strings=None,
         _StopEverythingStoppingCriteria
     )
 
-    # Native ExLlamav3Model handles multimodal internally - no special routing needed
-
     if shared.args.loader == 'Transformers':
         clear_torch_cache()
 

From 765af1ba1736b209427232d5bec1b2e55b099e1b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 12:39:18 -0700
Subject: [PATCH 0895/1701] API: Improve a validation

---
 extensions/openai/typing.py | 17 ++++++-----------
 1 file changed, 6 insertions(+), 11 deletions(-)

diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 9036627018..56d915822a 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -2,7 +2,7 @@
 import time
 from typing import Dict, List, Optional
 
-from pydantic import BaseModel, Field, field_validator, validator
+from pydantic import BaseModel, Field, model_validator, validator
 
 
 class GenerationOptions(BaseModel):
@@ -116,16 +116,11 @@ class CompletionRequestParams(BaseModel):
     top_p: float | None = 1
     user: str | None = Field(default=None, description="Unused parameter.")
 
-    @field_validator('prompt', 'messages')
-    @classmethod
-    def validate_prompt_or_messages(cls, v, info):
-        """Ensure either 'prompt' or 'messages' is provided for completions."""
-        if info.field_name == 'prompt':  # If we're validating 'prompt', check if neither prompt nor messages will be set
-            messages = info.data.get('messages')
-            if v is None and messages is None:
-                raise ValueError("Either 'prompt' or 'messages' must be provided")
-
-        return v
+    @model_validator(mode='after')
+    def validate_prompt_or_messages(self):
+        if self.prompt is None and self.messages is None:
+            raise ValueError("Either 'prompt' or 'messages' must be provided")
+        return self
 
 
 class CompletionRequest(GenerationOptions, CompletionRequestParams):

From 1e3c4e8bdbc3e8d313bfab016bc6f1853c4ad4b7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 14:40:59 -0700
Subject: [PATCH 0896/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_cuda128.txt             | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt      | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 323ef0f92e..789539fc21 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 2a7c936156..d79224784a 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 0106fbea10..2a3337a358 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index d5db4a1cff..7287497da0 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 694f1ff895..48ebe38167 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 392637e2dc..ccf80d068a 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 88eaa93042..e819dd042f 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 6accc2f049..8b9c882c50 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 3025f092ae..ce81c5ff16 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 7394bdcfc2..6233b84a61 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index a095a4c709..e3a863ec8e 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index ea43e56eb8..26f813d257 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 79737728ca..4de1159dbc 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -18,6 +18,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index d39786bd82..fded989820 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 0b373fa972..013364ff3a 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index fe9dccac76..85e95eb359 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index b3cfd525f1..945dcf4929 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 02aa03e3b0..bf1eff03c6 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.34.0/llama_cpp_binaries-0.34.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 0e88a621fd96bc75b908d078972ab8117e957f55 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 15:16:03 -0700
Subject: [PATCH 0897/1701] UI: Better organize the right sidebar

---
 modules/ui_chat.py | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 3b922fb454..94c980bb3f 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -78,12 +78,19 @@ def create_ui():
                 with gr.Row():
                     shared.gradio['start_with'] = gr.Textbox(label='Start reply with', placeholder='Sure thing!', value=shared.settings['start_with'], elem_classes=['add_scrollbar'])
 
+                gr.HTML("
            ")
+
                 shared.gradio['reasoning_effort'] = gr.Dropdown(value=shared.settings['reasoning_effort'], choices=['low', 'medium', 'high'], label='Reasoning effort', info='Used by GPT-OSS.')
                 shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', info='Used by pre-2507 Qwen3.')
+
+                gr.HTML("
            ")
+
                 shared.gradio['enable_web_search'] = gr.Checkbox(value=shared.settings.get('enable_web_search', False), label='Activate web search', elem_id='web-search')
                 with gr.Row(visible=shared.settings.get('enable_web_search', False)) as shared.gradio['web_search_row']:
                     shared.gradio['web_search_pages'] = gr.Number(value=shared.settings.get('web_search_pages', 3), precision=0, label='Number of pages to download', minimum=1, maximum=10)
 
+                gr.HTML("
            ")
+
                 with gr.Row():
                     shared.gradio['mode'] = gr.Radio(choices=['instruct', 'chat-instruct', 'chat'], value=None, label='Mode', info='Defines how the chat prompt is generated. In instruct and chat-instruct modes, the instruction template Parameters > Instruction template is used.', elem_id='chat-mode')
 
@@ -93,6 +100,8 @@ def create_ui():
                 with gr.Row():
                     shared.gradio['chat-instruct_command'] = gr.Textbox(value=shared.settings['chat-instruct_command'], lines=12, label='Command for chat-instruct mode', info='<|character|> and <|prompt|> get replaced with the bot name and the regular chat prompt respectively.', visible=shared.settings['mode'] == 'chat-instruct', elem_classes=['add_scrollbar'])
 
+                gr.HTML("
            ")
+
                 with gr.Row():
                     shared.gradio['count_tokens'] = gr.Button('Count tokens', size='sm')
 

From 0e3def449a8bf71ab40c052e4206f612aeba0a60 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 15:17:25 -0700
Subject: [PATCH 0898/1701] llama.cpp: --swa-full to llama-server when
 streaming-llm is checked

---
 modules/llama_cpp_server.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 63c8eda087..58534f2634 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -356,6 +356,7 @@ def _start_server(self):
                 cmd += ["--ctx-size-draft", str(shared.args.ctx_size_draft)]
         if shared.args.streaming_llm:
             cmd += ["--cache-reuse", "1"]
+            cmd += ["--swa-full"]
         if shared.args.extra_flags:
             # Clean up the input
             extra_flags = shared.args.extra_flags.strip()

From c47e6deda279f27c7bff1a31351e72c0d5025052 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 16:20:20 -0700
Subject: [PATCH 0899/1701] Update README

---
 README.md | 1 +
 1 file changed, 1 insertion(+)

diff --git a/README.md b/README.md
index 907d8c38fd..6e59f7daa2 100644
--- a/README.md
+++ b/README.md
@@ -16,6 +16,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 - Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory.
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
+- **Vision (multimodal models)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal%E2Tutorial)).
 - **Web search**: Optionally search the internet with LLM-generated queries to add context to the conversation.
 - Aesthetic UI with dark and light themes.
 - Syntax highlighting for code blocks and LaTeX rendering for mathematical expressions.

From e6447cd24acbde845dbb4aa27acfd4c17b5c849c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 17:42:35 -0700
Subject: [PATCH 0900/1701] mtmd: Update the llama-server request

---
 modules/llama_cpp_server.py | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 58534f2634..e82edb90ba 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -149,11 +149,10 @@ def generate_with_streaming(self, prompt, state):
             IMAGE_TOKEN_COST_ESTIMATE = 600  # A safe, conservative estimate per image
 
             base64_images = [convert_pil_to_base64(img) for img in pil_images]
-            multimodal_prompt_object = {
-                "prompt": prompt,
+            payload["prompt"] = {
+                "prompt_string": prompt,
                 "multimodal_data": base64_images
             }
-            payload["prompt"] = multimodal_prompt_object
 
             # Calculate an estimated token count
             text_tokens = self.encode(prompt, add_bos_token=state["add_bos_token"])

From d8fcc71616307a8ecacea93b7bdfa1117a23e1fe Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 11 Aug 2025 18:02:33 -0700
Subject: [PATCH 0901/1701] mtmd: Fail early if images are provided but the
 model doesn't support them (llama.cpp)

---
 modules/llama_cpp_server.py | 15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index e82edb90ba..51dacb842c 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -34,6 +34,7 @@ def __init__(self, model_path, server_path=None):
         self.process = None
         self.session = requests.Session()
         self.vocabulary_size = None
+        self.has_multimodal = False
         self.bos_token = ""
         self.last_prompt_token_count = 0
 
@@ -144,6 +145,10 @@ def generate_with_streaming(self, prompt, state):
         elif 'raw_images' in state and state['raw_images']:
             pil_images.extend(state.get('raw_images', []))
 
+        # Fail early if images are provided but the model doesn't support them
+        if pil_images and not self.has_multimodal:
+            raise RuntimeError("The loaded llama.cpp model does not support multimodal requests. You must load a vision model and provide an mmproj file.")
+
         if pil_images:
             # Multimodal case
             IMAGE_TOKEN_COST_ESTIMATE = 600  # A safe, conservative estimate per image
@@ -261,8 +266,8 @@ def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
         else:
             raise Exception(f"Unexpected response format: 'completion_probabilities' not found in {result}")
 
-    def _get_vocabulary_size(self):
-        """Get and store the model's maximum context length."""
+    def _get_model_properties(self):
+        """Get and store the model's properties, including vocab size and multimodal capability."""
         url = f"http://127.0.0.1:{self.port}/v1/models"
         response = self.session.get(url).json()
 
@@ -271,6 +276,10 @@ def _get_vocabulary_size(self):
             if "meta" in model_info and "n_vocab" in model_info["meta"]:
                 self.vocabulary_size = model_info["meta"]["n_vocab"]
 
+            # Check for multimodal capability
+            if "capabilities" in model_info and "multimodal" in model_info["capabilities"]:
+                self.has_multimodal = True
+
     def _get_bos_token(self):
         """Get and store the model's BOS token."""
         url = f"http://127.0.0.1:{self.port}/props"
@@ -421,7 +430,7 @@ def _start_server(self):
             time.sleep(1)
 
         # Server is now healthy, get model info
-        self._get_vocabulary_size()
+        self._get_model_properties()
         self._get_bos_token()
         return self.port
 

From 0882970a9445badcd953f27e4e10ecf869c103a5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 12 Aug 2025 07:00:24 -0700
Subject: [PATCH 0902/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_cuda128.txt             | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt      | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 789539fc21..eb7742b1fc 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index d79224784a..47bcb60a1b 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 2a3337a358..6958ce3794 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 7287497da0..0890b2a5fa 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 48ebe38167..da3010c690 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index ccf80d068a..3a9a953bdf 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index e819dd042f..a3e176d3dd 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 8b9c882c50..807d0a2194 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index ce81c5ff16..41e965742e 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 6233b84a61..72ba710323 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index e3a863ec8e..0c7f1d29ba 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 26f813d257..09f1c5029f 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 4de1159dbc..75296cb47c 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -18,6 +18,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index fded989820..ff3d7cb1ca 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 013364ff3a..97414bde08 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 85e95eb359..7f54320562 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 945dcf4929..c1764eadbc 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index bf1eff03c6..142b67ec40 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.35.0/llama_cpp_binaries-0.35.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 2238302b496a4145ee98e0eab0bf3d9f19a9c83b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 12 Aug 2025 08:50:45 -0700
Subject: [PATCH 0903/1701] ExLlamaV3: Add speculative decoding

---
 modules/exllamav3.py | 58 ++++++++++++++++++++++++++++++++++++++++++++
 modules/loaders.py   |  4 +++
 2 files changed, 62 insertions(+)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 980230f814..7fc6c5b1a0 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -85,6 +85,7 @@ def from_pretrained(cls, path_to_model):
         cache = Cache(model, max_num_tokens=max_tokens, layer_type=layer_type, **cache_kwargs)
 
         load_params = {'progressbar': True}
+        split = None
         if shared.args.gpu_split:
             split = [float(alloc) for alloc in shared.args.gpu_split.split(",")]
             load_params['use_per_device'] = split
@@ -92,6 +93,45 @@ def from_pretrained(cls, path_to_model):
         model.load(**load_params)
         tokenizer = Tokenizer.from_config(config)
 
+        # Initialize draft model for speculative decoding
+        draft_model = None
+        draft_cache = None
+        if shared.args.model_draft and shared.args.model_draft.lower() not in ["", "none"]:
+            logger.info(f"Loading draft model for speculative decoding: {shared.args.model_draft}")
+
+            draft_path = Path(shared.args.model_draft)
+            if not draft_path.is_dir():
+                draft_path = Path(f'{shared.args.model_dir}') / Path(shared.args.model_draft)
+
+            if not draft_path.is_dir():
+                logger.warning(f"Draft model not found at {draft_path}, speculative decoding disabled.")
+            else:
+                draft_config = Config.from_directory(str(draft_path))
+
+                # Set context size for draft model with 256-multiple validation
+                if shared.args.ctx_size_draft > 0:
+                    draft_max_tokens = shared.args.ctx_size_draft
+                else:
+                    draft_max_tokens = shared.args.ctx_size
+
+                # Validate draft model context size is a multiple of 256
+                if draft_max_tokens % 256 != 0:
+                    adjusted_draft_tokens = ((draft_max_tokens // 256) + 1) * 256
+                    logger.warning(f"Draft model max_num_tokens must be a multiple of 256. Adjusting from {draft_max_tokens} to {adjusted_draft_tokens}")
+                    draft_max_tokens = adjusted_draft_tokens
+
+                draft_config.max_seq_len = draft_max_tokens
+
+                draft_model = Model.from_config(draft_config)
+                draft_cache = Cache(draft_model, max_num_tokens=draft_max_tokens, layer_type=layer_type, **cache_kwargs)
+
+                draft_load_params = {'progressbar': True}
+                if split:
+                    draft_load_params['use_per_device'] = split
+
+                draft_model.load(**draft_load_params)
+                logger.info(f"Draft model loaded successfully. Max speculative tokens: {shared.args.draft_max}")
+
         # Load vision model component (ExLlamaV3 native)
         vision_model = None
         if "vision_config" in config.config_dict:
@@ -109,6 +149,9 @@ def from_pretrained(cls, path_to_model):
             model=model,
             cache=cache,
             tokenizer=tokenizer,
+            draft_model=draft_model,
+            draft_cache=draft_cache,
+            num_speculative_tokens=shared.args.draft_max if draft_model is not None else 0,
         )
 
         result = cls()
@@ -119,6 +162,8 @@ def from_pretrained(cls, path_to_model):
         result.config = config
         result.max_tokens = max_tokens
         result.vision_model = vision_model
+        result.draft_model = draft_model
+        result.draft_cache = draft_cache
 
         return result
 
@@ -289,6 +334,7 @@ def custom_sort_key(sampler_obj):
         self.generator.enqueue(job)
 
         response_text = ""
+
         try:
             while self.generator.num_remaining_jobs():
                 results = self.generator.iterate()
@@ -300,6 +346,7 @@ def custom_sort_key(sampler_obj):
                     if chunk:
                         response_text += chunk
                         yield response_text
+
         finally:
             self.generator.clear_queue()
 
@@ -331,6 +378,17 @@ def unload(self):
                 logger.warning(f"Error unloading vision model: {e}")
             self.vision_model = None
 
+        if hasattr(self, 'draft_model') and self.draft_model is not None:
+            try:
+                self.draft_model.unload()
+                del self.draft_model
+            except Exception as e:
+                logger.warning(f"Error unloading draft model: {e}")
+            self.draft_model = None
+
+        if hasattr(self, 'draft_cache') and self.draft_cache is not None:
+            self.draft_cache = None
+
         if hasattr(self, 'model') and self.model is not None:
             try:
                 self.model.unload()
diff --git a/modules/loaders.py b/modules/loaders.py
index feca9985a6..8b7e6ccee4 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -61,6 +61,10 @@
         'ctx_size',
         'cache_type',
         'gpu_split',
+        'model_draft',
+        'draft_max',
+        'ctx_size_draft',
+        'speculative_decoding_accordion',
     ],
     'ExLlamav2_HF': [
         'ctx_size',

From 2f6a629393afdb33e7fd355be10f6c72185412af Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 12 Aug 2025 08:51:01 -0700
Subject: [PATCH 0904/1701] UI: Minor improvement after
 0e88a621fd96bc75b908d078972ab8117e957f55

---
 js/main.js | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index 66a344b318..4b4b14c2cc 100644
--- a/js/main.js
+++ b/js/main.js
@@ -583,7 +583,7 @@ function moveToChatTab() {
 
   const chatControlsFirstChild = document.querySelector("#chat-controls").firstElementChild;
   const newParent = chatControlsFirstChild;
-  let newPosition = newParent.children.length - 2;
+  let newPosition = newParent.children.length - 3;
 
   newParent.insertBefore(grandParent, newParent.children[newPosition]);
   document.getElementById("save-character").style.display = "none";

From 8d7b88106a34102863a491a9c8848871c5118a85 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 12 Aug 2025 13:20:16 -0700
Subject: [PATCH 0905/1701] Revert "mtmd: Fail early if images are provided but
 the model doesn't support them (llama.cpp)"

This reverts commit d8fcc71616307a8ecacea93b7bdfa1117a23e1fe.
---
 modules/llama_cpp_server.py | 15 +++------------
 1 file changed, 3 insertions(+), 12 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 51dacb842c..e82edb90ba 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -34,7 +34,6 @@ def __init__(self, model_path, server_path=None):
         self.process = None
         self.session = requests.Session()
         self.vocabulary_size = None
-        self.has_multimodal = False
         self.bos_token = ""
         self.last_prompt_token_count = 0
 
@@ -145,10 +144,6 @@ def generate_with_streaming(self, prompt, state):
         elif 'raw_images' in state and state['raw_images']:
             pil_images.extend(state.get('raw_images', []))
 
-        # Fail early if images are provided but the model doesn't support them
-        if pil_images and not self.has_multimodal:
-            raise RuntimeError("The loaded llama.cpp model does not support multimodal requests. You must load a vision model and provide an mmproj file.")
-
         if pil_images:
             # Multimodal case
             IMAGE_TOKEN_COST_ESTIMATE = 600  # A safe, conservative estimate per image
@@ -266,8 +261,8 @@ def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
         else:
             raise Exception(f"Unexpected response format: 'completion_probabilities' not found in {result}")
 
-    def _get_model_properties(self):
-        """Get and store the model's properties, including vocab size and multimodal capability."""
+    def _get_vocabulary_size(self):
+        """Get and store the model's maximum context length."""
         url = f"http://127.0.0.1:{self.port}/v1/models"
         response = self.session.get(url).json()
 
@@ -276,10 +271,6 @@ def _get_model_properties(self):
             if "meta" in model_info and "n_vocab" in model_info["meta"]:
                 self.vocabulary_size = model_info["meta"]["n_vocab"]
 
-            # Check for multimodal capability
-            if "capabilities" in model_info and "multimodal" in model_info["capabilities"]:
-                self.has_multimodal = True
-
     def _get_bos_token(self):
         """Get and store the model's BOS token."""
         url = f"http://127.0.0.1:{self.port}/props"
@@ -430,7 +421,7 @@ def _start_server(self):
             time.sleep(1)
 
         # Server is now healthy, get model info
-        self._get_model_properties()
+        self._get_vocabulary_size()
         self._get_bos_token()
         return self.port
 

From 7301452b4183efab97de71dae27486874a3d73f6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 12 Aug 2025 13:23:24 -0700
Subject: [PATCH 0906/1701] UI: Minor info message change

---
 modules/ui_model_menu.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 9fa8a4f40c..6972a17eb0 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -42,7 +42,7 @@ def create_ui():
                     with gr.Row():
                         with gr.Column():
                             shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
-                            shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. Common values: 4096, 8192, 16384, 32768, 65536, 131072. ⚠️ Lower this value if you can\'t load the model.')
+                            shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. Common values: 4096, 8192, 16384, 32768, 65536, 131072.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')

From 2f979ce2942efc82ad90dfc28c7407c473da5169 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 12 Aug 2025 13:33:49 -0700
Subject: [PATCH 0907/1701] docs: Add a multimodal tutorial

---
 docs/Multimodal Tutorial.md | 66 +++++++++++++++++++++++++++++++++++++
 1 file changed, 66 insertions(+)
 create mode 100644 docs/Multimodal Tutorial.md

diff --git a/docs/Multimodal Tutorial.md b/docs/Multimodal Tutorial.md
new file mode 100644
index 0000000000..a30889f734
--- /dev/null
+++ b/docs/Multimodal Tutorial.md	
@@ -0,0 +1,66 @@
+## Getting started
+
+### 1. Find a multimodal model
+
+GGUF models with vision capabilities are uploaded along a `mmproj` file to Hugging Face.
+
+For instance, [unsloth/gemma-3-4b-it-GGUF](https://huggingface.co/unsloth/gemma-3-4b-it-GGUF/tree/main) has this:
+
+print1
+
+### 2. Download the model to `user_data/models`
+
+As an example, download
+
+https://huggingface.co/unsloth/gemma-3-4b-it-GGUF/resolve/main/gemma-3-4b-it-Q4_K_S.gguf?download=true
+
+to your `text-generation-webui/user_data/models` folder.
+
+### 3. Download the associated mmproj file to `user_data/mmproj`
+
+Then download
+
+https://huggingface.co/unsloth/gemma-3-4b-it-GGUF/resolve/main/mmproj-F16.gguf?download=true
+
+to your `text-generation-webui/user_data/mmproj` folder. Name it `mmproj-gemma-3-4b-it-F16.gguf` to give it a recognizable name.
+
+### 4. Load the model
+
+1. Launch the web UI
+2. Navigate to the Model tab
+3. Select the GGUF model in the Model dropdown:
+
+print2
+
+4. Select the mmproj file in the Multimodal (vision) menu:
+
+print3
+
+5. Click "Load"
+
+### 5. Send a message with an image
+
+Select your image by clicking on the 📎 icon and send your message:
+
+print5
+
+The model will reply with great understanding of the image contents:
+
+print6
+
+## Multimodal with ExLlamaV3
+
+Multimodal also works with the ExLlamaV3 loader (the non-HF one).
+
+No additional files are necessary, just load a multimodal EXL3 model and send an image.
+
+Examples of models that you can use:
+
+- https://huggingface.co/turboderp/gemma-3-27b-it-exl3
+- https://huggingface.co/turboderp/Mistral-Small-3.1-24B-Instruct-2503-exl3
+
+## Multimodal API examples
+
+In the page below you can find some ready-to-use examples:
+
+[Multimodal/vision (llama.cpp and ExLlamaV3)](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#multimodalvision-llamacpp-and-exllamav3)

From 41b95e9ec3dada8a931abb1a1ca974529d12d177 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 12 Aug 2025 13:37:37 -0700
Subject: [PATCH 0908/1701] Lint

---
 modules/exllamav2.py | 2 +-
 modules/exllamav3.py | 1 +
 2 files changed, 2 insertions(+), 1 deletion(-)

diff --git a/modules/exllamav2.py b/modules/exllamav2.py
index 3b3233d25b..5d5c5b567c 100644
--- a/modules/exllamav2.py
+++ b/modules/exllamav2.py
@@ -3,7 +3,6 @@
 from pathlib import Path
 
 import torch
-
 from exllamav2 import (
     ExLlamaV2,
     ExLlamaV2Cache,
@@ -16,6 +15,7 @@
     ExLlamaV2Tokenizer
 )
 from exllamav2.generator import ExLlamaV2Sampler, ExLlamaV2StreamingGenerator
+
 from modules import shared
 from modules.logging_colors import logger
 from modules.text_generation import get_max_prompt_length
diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 7fc6c5b1a0..66e2569333 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -16,6 +16,7 @@
     SS_TopK,
     SS_TopP
 )
+
 from modules import shared
 from modules.image_utils import (
     convert_image_attachments_to_pil,

From bd05fb899e7bb2889d616970e39f8cca41541e79 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 12 Aug 2025 14:19:18 -0700
Subject: [PATCH 0909/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 6e59f7daa2..5e7f37def4 100644
--- a/README.md
+++ b/README.md
@@ -16,7 +16,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 - Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory.
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
-- **Vision (multimodal models)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal%E2Tutorial)).
+- **Vision (multimodal models)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
 - **Web search**: Optionally search the internet with LLM-generated queries to add context to the conversation.
 - Aesthetic UI with dark and light themes.
 - Syntax highlighting for code blocks and LaTeX rendering for mathematical expressions.

From 331eab81f785b08eabc41e320f59c45c42a7d73f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 13 Aug 2025 06:44:34 -0700
Subject: [PATCH 0910/1701] mtmd: Explain base64 inputs in the API docs

---
 docs/12 - OpenAI API.md | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index fd3309c7ad..f85991f4f7 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -97,6 +97,8 @@ curl http://127.0.0.1:5000/v1/chat/completions \
   }'
 ```
 
+For base64-encoded images, just replace the inner "url" value with the base64 string, formatted as `data:image/FORMAT;base64,BASE64_STRING` where FORMAT is the image type (image/png, image/jpeg, image/gif, etc.) and BASE64_STRING is your encoded image.
+
 ##### With /v1/completions
 
 ```shell
@@ -129,6 +131,8 @@ curl http://127.0.0.1:5000/v1/completions \
   }'
 ```
 
+For base64-encoded images, just replace the inner "url" values with the base64 strings, formatted as `data:image/FORMAT;base64,BASE64_STRING` where FORMAT is the image type (image/png, image/jpeg, image/gif, etc.) and BASE64_STRING is your encoded image.
+
 #### SSE streaming
 
 ```shell

From 725a8bcf60dd769aa2d62760750c74b772f9f504 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 13 Aug 2025 06:49:28 -0700
Subject: [PATCH 0911/1701] Small docs change

---
 docs/12 - OpenAI API.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index f85991f4f7..83bba41f97 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -97,7 +97,7 @@ curl http://127.0.0.1:5000/v1/chat/completions \
   }'
 ```
 
-For base64-encoded images, just replace the inner "url" value with the base64 string, formatted as `data:image/FORMAT;base64,BASE64_STRING` where FORMAT is the image type (image/png, image/jpeg, image/gif, etc.) and BASE64_STRING is your encoded image.
+For base64-encoded images, just replace the inner "url" value with this format: `data:image/FORMAT;base64,BASE64_STRING` where FORMAT is the file type (png, jpeg, gif, etc.) and BASE64_STRING is your base64-encoded image data.
 
 ##### With /v1/completions
 
@@ -131,7 +131,7 @@ curl http://127.0.0.1:5000/v1/completions \
   }'
 ```
 
-For base64-encoded images, just replace the inner "url" values with the base64 strings, formatted as `data:image/FORMAT;base64,BASE64_STRING` where FORMAT is the image type (image/png, image/jpeg, image/gif, etc.) and BASE64_STRING is your encoded image.
+For base64-encoded images, just replace the inner "url" values with this format: `data:image/FORMAT;base64,BASE64_STRING` where FORMAT is the file type (png, jpeg, gif, etc.) and BASE64_STRING is your base64-encoded image data.
 
 #### SSE streaming
 

From 57f6e9af5a8defd67959672e8dc92040be91a0a5 Mon Sep 17 00:00:00 2001
From: altoiddealer 
Date: Wed, 13 Aug 2025 15:47:27 -0400
Subject: [PATCH 0912/1701] Set multimodal status during Model Loading (#7199)

---
 modules/exllamav3.py        | 10 +++++-----
 modules/llama_cpp_server.py | 24 ++++++++++++++++++++----
 modules/models.py           |  4 ++++
 modules/shared.py           |  1 +
 4 files changed, 30 insertions(+), 9 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 66e2569333..e580bbda64 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -177,9 +177,6 @@ def _process_images_for_generation(self, prompt: str, state: dict) -> Tuple[str,
         Process all possible image inputs and return modified prompt + embeddings.
         Returns: (processed_prompt, image_embeddings)
         """
-        if not self.is_multimodal():
-            return prompt, []
-
         # Collect images from various sources using shared utilities
         pil_images = []
 
@@ -234,8 +231,11 @@ def generate_with_streaming(self, prompt, state):
         """
         Generate text with streaming using native ExLlamaV3 API
         """
-        # Process images and modify prompt (ExLlamaV3-specific)
-        prompt, image_embeddings = self._process_images_for_generation(prompt, state)
+        image_embeddings = []
+
+        if shared.is_multimodal:
+            # Process images and modify prompt (ExLlamaV3-specific)
+            prompt, image_embeddings = self._process_images_for_generation(prompt, state)
 
         # Greedy decoding is a special case
         if state['temperature'] == 0:
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index e82edb90ba..5953803a7f 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -8,6 +8,7 @@
 import threading
 import time
 from pathlib import Path
+from typing import Any, List
 
 import llama_cpp_binaries
 import requests
@@ -129,10 +130,10 @@ def prepare_payload(self, state):
 
         return payload
 
-    def generate_with_streaming(self, prompt, state):
-        url = f"http://127.0.0.1:{self.port}/completion"
-        payload = self.prepare_payload(state)
-
+    def _process_images_for_generation(self, state: dict) -> List[Any]:
+        """
+        Process all possible image inputs and return PIL images
+        """
         pil_images = []
         # Source 1: Web UI (from chatbot_wrapper)
         if 'image_attachments' in state and state['image_attachments']:
@@ -144,6 +145,21 @@ def generate_with_streaming(self, prompt, state):
         elif 'raw_images' in state and state['raw_images']:
             pil_images.extend(state.get('raw_images', []))
 
+        return pil_images
+
+    def is_multimodal(self) -> bool:
+        """Check if this model supports multimodal input."""
+        return shared.args.mmproj not in [None, 'None']
+
+    def generate_with_streaming(self, prompt, state):
+        url = f"http://127.0.0.1:{self.port}/completion"
+        payload = self.prepare_payload(state)
+
+        pil_images = []
+
+        if shared.is_multimodal:
+            pil_images = self._process_images_for_generation(state)
+
         if pil_images:
             # Multimodal case
             IMAGE_TOKEN_COST_ESTIMATE = 600  # A safe, conservative estimate per image
diff --git a/modules/models.py b/modules/models.py
index cc500a40b7..938eed3dc2 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -55,6 +55,10 @@ def load_model(model_name, loader=None):
     if loader.lower().startswith('exllama') or loader.lower().startswith('tensorrt') or loader == 'llama.cpp':
         shared.settings['truncation_length'] = shared.args.ctx_size
 
+    shared.is_multimodal = False    
+    if loader.lower() in ('exllamav3', 'llama.cpp'):
+        shared.is_multimodal = model.is_multimodal()
+
     logger.info(f"Loaded \"{model_name}\" in {(time.time()-t0):.2f} seconds.")
     logger.info(f"LOADER: \"{loader}\"")
     logger.info(f"TRUNCATION LENGTH: {shared.settings['truncation_length']}")
diff --git a/modules/shared.py b/modules/shared.py
index e9d8a62fdb..a1f4571e33 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -16,6 +16,7 @@
 tokenizer = None
 model_name = 'None'
 is_seq2seq = False
+is_multimodal = False
 model_dirty_from_training = False
 lora_names = []
 

From 73a8a737b23fc195c52ef1d9021993fd13e28e33 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 13 Aug 2025 18:23:18 -0700
Subject: [PATCH 0913/1701] docs: Improve the multimodal examples slightly

---
 docs/12 - OpenAI API.md | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index 83bba41f97..227541a303 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -93,7 +93,10 @@ curl http://127.0.0.1:5000/v1/chat/completions \
           {"type": "image_url", "image_url": {"url": "https://github.com/turboderp-org/exllamav3/blob/master/examples/media/cat.png?raw=true"}}
         ]
       }
-    ]
+    ],
+    "temperature": 0.6,
+    "top_p": 0.95,
+    "top_k": 20
   }'
 ```
 
@@ -127,7 +130,10 @@ curl http://127.0.0.1:5000/v1/completions \
           }
         ]
       }
-    ]
+    ],
+    "temperature": 0.6,
+    "top_p": 0.95,
+    "top_k": 20
   }'
 ```
 

From d771ca4a13b9837e169cd44815bb3a86bc6c8a4b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 14 Aug 2025 12:02:30 -0700
Subject: [PATCH 0914/1701] Fix web search (attempt)

---
 modules/web_search.py | 19 +++++++++++++------
 1 file changed, 13 insertions(+), 6 deletions(-)

diff --git a/modules/web_search.py b/modules/web_search.py
index 3b1f6e1885..597af4b289 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -1,6 +1,8 @@
 import concurrent.futures
 import html
+import random
 import re
+import urllib.request
 from concurrent.futures import as_completed
 from datetime import datetime
 from urllib.parse import quote_plus
@@ -50,16 +52,21 @@ def download_web_page(url, timeout=10):
 def perform_web_search(query, num_pages=3, max_workers=5, timeout=10):
     """Perform web search and return results with content"""
     try:
-        # Use DuckDuckGo HTML search endpoint
         search_url = f"https://html.duckduckgo.com/html/?q={quote_plus(query)}"
-        headers = {'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36'}
 
-        response = requests.get(search_url, headers=headers, timeout=timeout)
-        response.raise_for_status()
+        agents = [
+            "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36",
+            "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36"
+        ]
+
+        response_text = ""
+        req = urllib.request.Request(search_url, headers={'User-Agent': random.choice(agents)})
+        with urllib.request.urlopen(req, timeout=timeout) as response:
+            response_text = response.read().decode('utf-8')
 
         # Extract results with regex
-        titles = re.findall(r']*class="[^"]*result__a[^"]*"[^>]*>(.*?)', response.text, re.DOTALL)
-        urls = re.findall(r']*class="[^"]*result__url[^"]*"[^>]*>(.*?)', response.text, re.DOTALL)
+        titles = re.findall(r']*class="[^"]*result__a[^"]*"[^>]*>(.*?)', response_text, re.DOTALL)
+        urls = re.findall(r']*class="[^"]*result__url[^"]*"[^>]*>(.*?)', response_text, re.DOTALL)
 
         # Prepare download tasks
         download_tasks = []

From dbabe67e776d46bb0a84987a9c484a59bd75d8db Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 Aug 2025 13:19:11 -0700
Subject: [PATCH 0915/1701] ExLlamaV3: Enable the --enable-tp option, add a
 --tp-backend option

---
 modules/exllamav3.py     | 5 +++++
 modules/loaders.py       | 2 ++
 modules/shared.py        | 6 +++++-
 modules/ui.py            | 1 +
 modules/ui_model_menu.py | 4 +++-
 5 files changed, 16 insertions(+), 2 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index e580bbda64..739629772a 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -91,6 +91,11 @@ def from_pretrained(cls, path_to_model):
             split = [float(alloc) for alloc in shared.args.gpu_split.split(",")]
             load_params['use_per_device'] = split
 
+        # Tensor-parallelism
+        if shared.args.enable_tp:
+            load_params['tensor_p'] = True
+            load_params['tp_backend'] = shared.args.tp_backend
+
         model.load(**load_params)
         tokenizer = Tokenizer.from_config(config)
 
diff --git a/modules/loaders.py b/modules/loaders.py
index 8b7e6ccee4..295db1e750 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -65,6 +65,8 @@
         'draft_max',
         'ctx_size_draft',
         'speculative_decoding_accordion',
+        'enable_tp',
+        'tp_backend',
     ],
     'ExLlamav2_HF': [
         'ctx_size',
diff --git a/modules/shared.py b/modules/shared.py
index a1f4571e33..644261a0bf 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -101,6 +101,11 @@
 group.add_argument('--device-draft', type=str, default=None, help='Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1')
 group.add_argument('--ctx-size-draft', type=int, default=0, help='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
 
+# ExLlamaV3
+group = parser.add_argument_group('ExLlamaV3')
+group.add_argument('--enable-tp', '--enable_tp', action='store_true', help='Enable Tensor Parallelism (TP) to split the model across GPUs.')
+group.add_argument('--tp-backend', type=str, default='native', help='The backend for tensor parallelism. Valid options: native, nccl. Default: native.')
+
 # ExLlamaV2
 group = parser.add_argument_group('ExLlamaV2')
 group.add_argument('--gpu-split', type=str, help='Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.')
@@ -110,7 +115,6 @@
 group.add_argument('--no_xformers', action='store_true', help='Force xformers to not be used.')
 group.add_argument('--no_sdpa', action='store_true', help='Force Torch SDPA to not be used.')
 group.add_argument('--num_experts_per_token', type=int, default=2, metavar='N', help='Number of experts to use for generation. Applies to MoE models like Mixtral.')
-group.add_argument('--enable_tp', action='store_true', help='Enable Tensor Parallelism (TP) in ExLlamaV2.')
 
 # TensorRT-LLM
 group = parser.add_argument_group('TensorRT-LLM')
diff --git a/modules/ui.py b/modules/ui.py
index 1171cd48eb..502005e747 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -155,6 +155,7 @@ def list_model_elements():
         'bf16',
         'autosplit',
         'enable_tp',
+        'tp_backend',
         'no_flash_attn',
         'no_xformers',
         'no_sdpa',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 6972a17eb0..dd2406274d 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -46,6 +46,8 @@ def create_ui():
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
+                            shared.gradio['tp_backend'] = gr.Dropdown(label="tp-backend", choices=['native', 'nccl'], value=shared.args.tp_backend, info='The backend for tensor parallelism.')
+
                         with gr.Column():
                             shared.gradio['vram_info'] = gr.HTML(value=get_initial_vram_info())
                             shared.gradio['flash_attn'] = gr.Checkbox(label="flash-attn", value=shared.args.flash_attn, info='Use flash-attention.')
@@ -54,7 +56,7 @@ def create_ui():
                             shared.gradio['load_in_4bit'] = gr.Checkbox(label="load-in-4bit", value=shared.args.load_in_4bit)
                             shared.gradio['use_double_quant'] = gr.Checkbox(label="use_double_quant", value=shared.args.use_double_quant, info='Used by load-in-4bit.')
                             shared.gradio['autosplit'] = gr.Checkbox(label="autosplit", value=shared.args.autosplit, info='Automatically split the model tensors across the available GPUs.')
-                            shared.gradio['enable_tp'] = gr.Checkbox(label="enable_tp", value=shared.args.enable_tp, info='Enable Tensor Parallelism (TP).')
+                            shared.gradio['enable_tp'] = gr.Checkbox(label="enable_tp", value=shared.args.enable_tp, info='Enable tensor parallelism (TP).')
                             shared.gradio['cpp_runner'] = gr.Checkbox(label="cpp-runner", value=shared.args.cpp_runner, info='Enable inference with ModelRunnerCpp, which is faster than the default ModelRunner.')
                             shared.gradio['trust_remote_code'] = gr.Checkbox(label="trust-remote-code", value=shared.args.trust_remote_code, info='Set trust_remote_code=True while loading the tokenizer/model. To enable this option, start the web UI with the --trust-remote-code flag.', interactive=shared.args.trust_remote_code)
                             shared.gradio['tensorrt_llm_info'] = gr.Markdown('* TensorRT-LLM has to be installed manually in a separate Python 3.10 environment at the moment. For a guide, consult the description of [this PR](https://github.com/oobabooga/text-generation-webui/pull/5715). \n\n* `ctx_size` is only used when `cpp-runner` is checked.\n\n* `cpp_runner` does not support streaming at the moment.')

From a633793a0007d435ed4f2cf08f0fbb4b77651b91 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 Aug 2025 13:19:42 -0700
Subject: [PATCH 0916/1701] Bump exllamav3 to 0.0.6

---
 requirements/full/requirements.txt                | 4 ++--
 requirements/full/requirements_apple_intel.txt    | 2 +-
 requirements/full/requirements_apple_silicon.txt  | 2 +-
 requirements/full/requirements_cuda128.txt        | 4 ++--
 requirements/full/requirements_cuda128_noavx2.txt | 4 ++--
 requirements/full/requirements_noavx2.txt         | 4 ++--
 6 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index eb7742b1fc..d0282ee920 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -36,8 +36,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 0890b2a5fa..13361a7807 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -35,5 +35,5 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index da3010c690..e4a30168b5 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -36,5 +36,5 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5-py3-none-any.whl
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
index 807d0a2194..eeee9ff35d 100644
--- a/requirements/full/requirements_cuda128.txt
+++ b/requirements/full/requirements_cuda128.txt
@@ -36,8 +36,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
index 41e965742e..d08f23caea 100644
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ b/requirements/full/requirements_cuda128_noavx2.txt
@@ -36,8 +36,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 72ba710323..71fd227d2a 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -36,8 +36,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.5/exllamav3-0.0.5+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 9651b5c873649e5c967142f9f78e7ad6cf59aaf5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 Aug 2025 13:22:55 -0700
Subject: [PATCH 0917/1701] Make CUDA 12.8 the default CUDA option, remove the
 CUDA 12.4 option

Exllamav3 doesn't compile with torch 2.6 anymore, and torch 2.7
requires newer CUDA.
---
 one_click.py                                  | 20 +++++++--
 requirements/full/requirements.txt            | 14 +++---
 requirements/full/requirements_cuda128.txt    | 45 -------------------
 .../full/requirements_cuda128_noavx2.txt      | 45 -------------------
 requirements/full/requirements_noavx2.txt     | 14 +++---
 5 files changed, 30 insertions(+), 108 deletions(-)
 delete mode 100644 requirements/full/requirements_cuda128.txt
 delete mode 100644 requirements/full/requirements_cuda128_noavx2.txt

diff --git a/one_click.py b/one_click.py
index 050da76b10..486e893e16 100644
--- a/one_click.py
+++ b/one_click.py
@@ -16,7 +16,7 @@
 # os.environ["HCC_AMDGPU_TARGET"] = 'gfx1030'
 
 # Define the required versions
-TORCH_VERSION = "2.6.0"
+TORCH_VERSION = "2.7.0"
 PYTHON_VERSION = "3.11"
 LIBSTDCXX_VERSION_LINUX = "12.1.0"
 
@@ -113,17 +113,16 @@ def get_gpu_choice():
             choice = get_user_choice(
                 "What is your GPU?",
                 {
-                    'A': 'NVIDIA - CUDA 12.4',
+                    'A': 'NVIDIA',
                     'B': 'AMD - Linux/macOS only, requires ROCm 6.2.4',
                     'C': 'Apple M Series',
                     'D': 'Intel Arc (beta)',
-                    'E': 'NVIDIA - CUDA 12.8',
                     'N': 'CPU mode'
                 },
             )
 
         # Convert choice to GPU name
-        gpu_choice = {"A": "NVIDIA", "B": "AMD", "C": "APPLE", "D": "INTEL", "E": "NVIDIA_CUDA128", "N": "NONE"}[choice]
+        gpu_choice = {"A": "NVIDIA_CUDA128", "B": "AMD", "C": "APPLE", "D": "INTEL", "N": "NONE"}[choice]
 
         # Save choice to state
         state['gpu_choice'] = gpu_choice
@@ -368,6 +367,19 @@ def update_requirements(initial_installation=False, pull=True):
             assert_success=True
         )
 
+    # Check for outdated CUDA 12.4 installs and refuse to update
+    state = load_state()
+    if state.get('gpu_choice') == 'NVIDIA':
+        print_big_message(
+            "Your current installation uses CUDA 12.4, which has been removed.\n"
+            "To update to the new default (CUDA 12.8), a clean installation is required.\n\n"
+            "INSTRUCTIONS:\n"
+            "1. Delete the 'installer_files' folder in your text-generation-webui directory.\n"
+            "2. Run the start script again (e.g., start_windows.bat).\n\n"
+            "This will create a fresh environment with the latest software."
+        )
+        sys.exit(0)
+
     current_commit = get_current_commit()
     wheels_changed = not os.path.exists(state_file)
     if not wheels_changed:
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index d0282ee920..eeee9ff35d 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -24,7 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.55.*
-triton-windows==3.2.0.post19; platform_system == "Windows"
+triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 
@@ -36,10 +36,10 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128.txt b/requirements/full/requirements_cuda128.txt
deleted file mode 100644
index eeee9ff35d..0000000000
--- a/requirements/full/requirements_cuda128.txt
+++ /dev/null
@@ -1,45 +0,0 @@
-accelerate==1.8.*
-bitsandbytes==0.46.*
-colorama
-datasets
-einops
-fastapi==0.112.4
-gradio==4.37.*
-html2text==2025.4.15
-jinja2==3.1.6
-markdown
-numpy==2.2.*
-pandas
-peft==0.16.*
-Pillow>=9.5.0
-psutil
-pydantic==2.8.2
-PyPDF2==3.0.1
-python-docx==1.1.2
-pyyaml
-requests
-rich
-safetensors==0.5.*
-scipy
-sentencepiece
-tensorboard
-transformers==4.55.*
-triton-windows==3.3.1.post19; platform_system == "Windows"
-tqdm
-wandb
-
-# API
-flask_cloudflared==0.0.14
-sse-starlette==1.6.5
-tiktoken
-
-# CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/full/requirements_cuda128_noavx2.txt b/requirements/full/requirements_cuda128_noavx2.txt
deleted file mode 100644
index d08f23caea..0000000000
--- a/requirements/full/requirements_cuda128_noavx2.txt
+++ /dev/null
@@ -1,45 +0,0 @@
-accelerate==1.8.*
-bitsandbytes==0.46.*
-colorama
-datasets
-einops
-fastapi==0.112.4
-gradio==4.37.*
-html2text==2025.4.15
-jinja2==3.1.6
-markdown
-numpy==2.2.*
-pandas
-peft==0.16.*
-Pillow>=9.5.0
-psutil
-pydantic==2.8.2
-PyPDF2==3.0.1
-python-docx==1.1.2
-pyyaml
-requests
-rich
-safetensors==0.5.*
-scipy
-sentencepiece
-tensorboard
-transformers==4.55.*
-triton-windows==3.3.1.post19; platform_system == "Windows"
-tqdm
-wandb
-
-# API
-flask_cloudflared==0.0.14
-sse-starlette==1.6.5
-tiktoken
-
-# CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 71fd227d2a..d08f23caea 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -24,7 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.55.*
-triton-windows==3.2.0.post19; platform_system == "Windows"
+triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 
@@ -36,10 +36,10 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu124.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

From 3a91ca2dd191716be9e9f3f20627c1e1a80d13f1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 Aug 2025 13:57:23 -0700
Subject: [PATCH 0918/1701] Update flash attention

---
 requirements/full/requirements.txt        | 4 ++--
 requirements/full/requirements_noavx2.txt | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index eeee9ff35d..d57a457c83 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -41,5 +41,5 @@ https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/Dao-AILab/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu12torch2.7cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index d08f23caea..b073a3a903 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -41,5 +41,5 @@ https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/Dao-AILab/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu12torch2.7cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

From 64eba9576cb806d2213b7efbb82469aa70a9fd71 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 Aug 2025 14:08:40 -0700
Subject: [PATCH 0919/1701] mtmd: Fix a bug when "include past attachments" is
 unchecked

---
 modules/chat.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 7b1629dd5b..ab6b43c078 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -269,7 +269,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
             enhanced_user_msg = user_msg
 
             # Add attachment content if present AND if past attachments are enabled
-            if (state.get('include_past_attachments', True) and user_key in metadata and "attachments" in metadata[user_key]):
+            if user_key in metadata and "attachments" in metadata[user_key]:
                 attachments_text = ""
                 image_refs = ""
 
@@ -277,7 +277,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
                     if attachment.get("type") == "image":
                         # Add image reference for multimodal models
                         image_refs += "<__media__>"
-                    else:
+                    elif state.get('include_past_attachments', True):
                         # Handle text/PDF attachments
                         filename = attachment.get("name", "file")
                         content = attachment.get("content", "")

From 58797a9eb5f386cc8262f6d8f1a152494249c28d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 Aug 2025 14:17:20 -0700
Subject: [PATCH 0920/1701] Minor change after
 9651b5c873649e5c967142f9f78e7ad6cf59aaf5

---
 one_click.py | 12 ++----------
 1 file changed, 2 insertions(+), 10 deletions(-)

diff --git a/one_click.py b/one_click.py
index 486e893e16..67be9e4b12 100644
--- a/one_click.py
+++ b/one_click.py
@@ -135,9 +135,7 @@ def get_pytorch_install_command(gpu_choice):
     """Get PyTorch installation command based on GPU choice"""
     base_cmd = f"python -m pip install torch=={TORCH_VERSION} "
 
-    if gpu_choice == "NVIDIA":
-        return base_cmd + "--index-url https://download.pytorch.org/whl/cu124"
-    elif gpu_choice == "NVIDIA_CUDA128":
+    if gpu_choice == "NVIDIA_CUDA128":
         return "python -m pip install torch==2.7.1 --index-url https://download.pytorch.org/whl/cu128"
     elif gpu_choice == "AMD":
         return base_cmd + "--index-url https://download.pytorch.org/whl/rocm6.2.4"
@@ -156,9 +154,7 @@ def get_pytorch_update_command(gpu_choice):
     """Get PyTorch update command based on GPU choice"""
     base_cmd = f"python -m pip install --upgrade torch=={TORCH_VERSION} "
 
-    if gpu_choice == "NVIDIA":
-        return f"{base_cmd} --index-url https://download.pytorch.org/whl/cu124"
-    elif gpu_choice == "NVIDIA_CUDA128":
+    if gpu_choice == "NVIDIA_CUDA128":
         return "python -m pip install --upgrade torch==2.7.1 --index-url https://download.pytorch.org/whl/cu128"
     elif gpu_choice == "AMD":
         return f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.2.4"
@@ -181,8 +177,6 @@ def get_requirements_file(gpu_choice):
         file_name = f"requirements_apple_{'intel' if is_x86_64() else 'silicon'}.txt"
     elif gpu_choice in ["INTEL", "NONE"]:
         file_name = f"requirements_cpu_only{'_noavx2' if not cpu_has_avx2() else ''}.txt"
-    elif gpu_choice == "NVIDIA":
-        file_name = f"requirements{'_noavx2' if not cpu_has_avx2() else ''}.txt"
     elif gpu_choice == "NVIDIA_CUDA128":
         file_name = f"requirements_cuda128{'_noavx2' if not cpu_has_avx2() else ''}.txt"
     else:
@@ -330,8 +324,6 @@ def install_webui():
                 cmd_flags_file.write("\n--cpu\n")
 
     # Handle CUDA version display
-    elif any((is_windows(), is_linux())) and gpu_choice == "NVIDIA":
-        print("CUDA: 12.4")
     elif any((is_windows(), is_linux())) and gpu_choice == "NVIDIA_CUDA128":
         print("CUDA: 12.8")
 

From 35707c2dd89a9983f2038ded9bc67d13aa7bc213 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 Aug 2025 21:39:57 -0700
Subject: [PATCH 0921/1701] Update README

---
 README.md | 38 +++++++++++++++++++-------------------
 1 file changed, 19 insertions(+), 19 deletions(-)

diff --git a/README.md b/README.md
index 5e7f37def4..93d3113169 100644
--- a/README.md
+++ b/README.md
@@ -32,13 +32,13 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 
 ## How to install
 
-#### Option 1: Portable builds (get started in 1 minute)
+#### ✅ Option 1: Portable builds (get started in 1 minute)
 
 No installation needed – just download, unzip and run. All dependencies included.
 
 Compatible with GGUF (llama.cpp) models on Windows, Linux, and macOS.
 
-Download from here: https://github.com/oobabooga/text-generation-webui/releases
+Download from here: **https://github.com/oobabooga/text-generation-webui/releases**
 
 #### Option 2: One-click installer
 
@@ -57,23 +57,6 @@ You can pass command-line flags directly (e.g., `./start_linux.sh --help`), or a
 
 To update, run the update script for your OS: `update_wizard_windows.bat`, `update_wizard_linux.sh`, or `update_wizard_macos.sh`.
 
-
            
-
-One-click installer details
-
-
-### One-click-installer
-
-The script uses Miniforge to set up a Conda environment in the `installer_files` folder.
-
-If you ever need to install something manually in the `installer_files` environment, you can launch an interactive shell using the cmd script: `cmd_linux.sh`, `cmd_windows.bat`, or `cmd_macos.sh`.
-
-* There is no need to run any of those scripts (`start_`, `update_wizard_`, or `cmd_`) as admin/root.
-* To install requirements for extensions, it is recommended to use the update wizard script with the "Install/update extensions requirements" option. At the end, this script will install the main requirements for the project to make sure that they take precedence in case of version conflicts.
-* For automated installation, you can use the `GPU_CHOICE`, `LAUNCH_AFTER_INSTALL`, and `INSTALL_EXTENSIONS` environment variables. For instance: `GPU_CHOICE=A LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh`.
-
-
            
-
 
            
 
 Manual portable installation with venv
@@ -108,6 +91,23 @@ deactivate
 ```
 
            
 
+
            
+
+One-click installer details
+
+
+### One-click-installer
+
+The script uses Miniforge to set up a Conda environment in the `installer_files` folder.
+
+If you ever need to install something manually in the `installer_files` environment, you can launch an interactive shell using the cmd script: `cmd_linux.sh`, `cmd_windows.bat`, or `cmd_macos.sh`.
+
+* There is no need to run any of those scripts (`start_`, `update_wizard_`, or `cmd_`) as admin/root.
+* To install requirements for extensions, it is recommended to use the update wizard script with the "Install/update extensions requirements" option. At the end, this script will install the main requirements for the project to make sure that they take precedence in case of version conflicts.
+* For automated installation, you can use the `GPU_CHOICE`, `LAUNCH_AFTER_INSTALL`, and `INSTALL_EXTENSIONS` environment variables. For instance: `GPU_CHOICE=A LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh`.
+
+
            
+
 
            
 
 Manual full installation with conda or docker

From 3dec47eaf8f94ba085e7d4d06522ce398d04bdbe Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 Aug 2025 21:43:46 -0700
Subject: [PATCH 0922/1701] Small one-click installer changes

---
 one_click.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/one_click.py b/one_click.py
index 67be9e4b12..1ea5e59f6e 100644
--- a/one_click.py
+++ b/one_click.py
@@ -16,7 +16,7 @@
 # os.environ["HCC_AMDGPU_TARGET"] = 'gfx1030'
 
 # Define the required versions
-TORCH_VERSION = "2.7.0"
+TORCH_VERSION = "2.7.1"
 PYTHON_VERSION = "3.11"
 LIBSTDCXX_VERSION_LINUX = "12.1.0"
 
@@ -136,7 +136,7 @@ def get_pytorch_install_command(gpu_choice):
     base_cmd = f"python -m pip install torch=={TORCH_VERSION} "
 
     if gpu_choice == "NVIDIA_CUDA128":
-        return "python -m pip install torch==2.7.1 --index-url https://download.pytorch.org/whl/cu128"
+        return base_cmd + "--index-url https://download.pytorch.org/whl/cu128"
     elif gpu_choice == "AMD":
         return base_cmd + "--index-url https://download.pytorch.org/whl/rocm6.2.4"
     elif gpu_choice in ["APPLE", "NONE"]:

From 320f7339cdd3bf52ff705e0679d55fd738e61218 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 Aug 2025 21:56:35 -0700
Subject: [PATCH 0923/1701] Update README

---
 README.md | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/README.md b/README.md
index 93d3113169..3832030c5e 100644
--- a/README.md
+++ b/README.md
@@ -139,19 +139,19 @@ conda activate textgen
 
 | System | GPU | Command |
 |--------|---------|---------|
-| Linux/WSL | NVIDIA | `pip3 install torch==2.6.0 --index-url https://download.pytorch.org/whl/cu124` |
-| Linux/WSL | CPU only | `pip3 install torch==2.6.0 --index-url https://download.pytorch.org/whl/cpu` |
-| Linux | AMD | `pip3 install torch==2.6.0 --index-url https://download.pytorch.org/whl/rocm6.2.4` |
-| MacOS + MPS | Any | `pip3 install torch==2.6.0` |
-| Windows | NVIDIA | `pip3 install torch==2.6.0 --index-url https://download.pytorch.org/whl/cu124` |
-| Windows | CPU only | `pip3 install torch==2.6.0` |
+| Linux/WSL | NVIDIA | `pip3 install torch==2.7.1 --index-url https://download.pytorch.org/whl/cu128` |
+| Linux/WSL | CPU only | `pip3 install torch==2.7.1 --index-url https://download.pytorch.org/whl/cpu` |
+| Linux | AMD | `pip3 install torch==2.7.1 --index-url https://download.pytorch.org/whl/rocm6.2.4` |
+| MacOS + MPS | Any | `pip3 install torch==2.7.1` |
+| Windows | NVIDIA | `pip3 install torch==2.7.1 --index-url https://download.pytorch.org/whl/cu128` |
+| Windows | CPU only | `pip3 install torch==2.7.1` |
 
 The up-to-date commands can be found here: https://pytorch.org/get-started/locally/.
 
 If you need `nvcc` to compile some library manually, you will additionally need to install this:
 
 ```
-conda install -y -c "nvidia/label/cuda-12.4.1" cuda
+conda install -y -c "nvidia/label/cuda-12.8.1" cuda
 ```
 
 #### 3. Install the web UI

From 6bf31479d92ab9da0c643ca907a90b300a230f25 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 Aug 2025 22:00:21 -0700
Subject: [PATCH 0924/1701] Update README

---
 README.md | 48 +++++++++++++++++++++++-------------------------
 1 file changed, 23 insertions(+), 25 deletions(-)

diff --git a/README.md b/README.md
index 3832030c5e..ccde678ade 100644
--- a/README.md
+++ b/README.md
@@ -40,29 +40,7 @@ Compatible with GGUF (llama.cpp) models on Windows, Linux, and macOS.
 
 Download from here: **https://github.com/oobabooga/text-generation-webui/releases**
 
-#### Option 2: One-click installer
-
-For users who need additional backends (ExLlamaV3, Transformers) or extensions (TTS, voice input, translation, etc). Requires ~10GB disk space and downloads PyTorch.
-
-1. Clone the repository, or [download its source code](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip) and extract it.
-2. Run the startup script for your OS: `start_windows.bat`, `start_linux.sh`, or `start_macos.sh`.
-3. When prompted, select your GPU vendor.
-4. After installation, open `http://127.0.0.1:7860` in your browser.
-
-To restart the web UI later, run the same `start_` script.
-
-To reinstall with a fresh Python environment, delete the `installer_files` folder and run the `start_` script again.
-
-You can pass command-line flags directly (e.g., `./start_linux.sh --help`), or add them to `user_data/CMD_FLAGS.txt` (e.g., `--api` to enable the API).
-
-To update, run the update script for your OS: `update_wizard_windows.bat`, `update_wizard_linux.sh`, or `update_wizard_macos.sh`.
-
-
            
-
-Manual portable installation with venv
-
-
-### Manual portable installation with venv
+#### Option 2: Manual portable install with venv
 
 Very fast setup that should work on any Python 3.9+:
 
@@ -81,7 +59,7 @@ venv\Scripts\activate
 source venv/bin/activate
 
 # Install dependencies (choose appropriate file under requirements/portable for your hardware)
-pip install -r requirements/portable/requirements.txt
+pip install -r requirements/portable/requirements.txt --upgrade
 
 # Launch server (basic command)
 python server.py --portable --api --auto-launch
@@ -89,9 +67,29 @@ python server.py --portable --api --auto-launch
 # When done working, deactivate
 deactivate
 ```
-
            
+
+#### Option 3: One-click installer
+
+For users who need additional backends (ExLlamaV3, Transformers) or extensions (TTS, voice input, translation, etc). Requires ~10GB disk space and downloads PyTorch.
+
+1. Clone the repository, or [download its source code](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip) and extract it.
+2. Run the startup script for your OS: `start_windows.bat`, `start_linux.sh`, or `start_macos.sh`.
+3. When prompted, select your GPU vendor.
+4. After installation, open `http://127.0.0.1:7860` in your browser.
+
+To restart the web UI later, run the same `start_` script.
+
+To reinstall with a fresh Python environment, delete the `installer_files` folder and run the `start_` script again.
+
+You can pass command-line flags directly (e.g., `./start_linux.sh --help`), or add them to `user_data/CMD_FLAGS.txt` (e.g., `--api` to enable the API).
+
+To update, run the update script for your OS: `update_wizard_windows.bat`, `update_wizard_linux.sh`, or `update_wizard_macos.sh`.
 
 
            
+
+Manual portable installation with venv
+
+
 
 One-click installer details
 

From 8cdb911a6e637c355dc9eac2ab43f94eab7b3281 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 Aug 2025 22:06:12 -0700
Subject: [PATCH 0925/1701] Update README

---
 README.md | 29 +++++++++++++----------------
 1 file changed, 13 insertions(+), 16 deletions(-)

diff --git a/README.md b/README.md
index ccde678ade..f213f7a9dd 100644
--- a/README.md
+++ b/README.md
@@ -2,8 +2,6 @@
 
 A Gradio web UI for Large Language Models.
 
-Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.com/AUTOMATIC1111/stable-diffusion-webui) of text generation.
-
 [Try the Deep Reason extension](https://oobabooga.gumroad.com/l/deep_reason)
 
 |![Image1](https://github.com/oobabooga/screenshots/raw/main/INSTRUCT-3.5.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/CHAT-3.5.png) |
@@ -79,16 +77,11 @@ For users who need additional backends (ExLlamaV3, Transformers) or extensions (
 
 To restart the web UI later, run the same `start_` script.
 
-To reinstall with a fresh Python environment, delete the `installer_files` folder and run the `start_` script again.
-
 You can pass command-line flags directly (e.g., `./start_linux.sh --help`), or add them to `user_data/CMD_FLAGS.txt` (e.g., `--api` to enable the API).
 
 To update, run the update script for your OS: `update_wizard_windows.bat`, `update_wizard_linux.sh`, or `update_wizard_macos.sh`.
 
-
            
-
-Manual portable installation with venv
-
+To reinstall with a fresh Python environment, delete the `installer_files` folder and run the `start_` script again.
 
 
 One-click installer details
@@ -236,13 +229,13 @@ usage: server.py [-h] [--multi-user] [--model MODEL] [--lora LORA [LORA ...]] [-
                  [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR]
                  [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] [--load-in-4bit] [--use_double_quant]
                  [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn] [--threads THREADS] [--threads-batch THREADS_BATCH] [--batch-size BATCH_SIZE] [--no-mmap] [--mlock]
-                 [--gpu-layers N] [--tensor-split TENSOR_SPLIT] [--numa] [--no-kv-offload] [--row-split] [--extra-flags EXTRA_FLAGS] [--streaming-llm] [--ctx-size N] [--cache-type N]
-                 [--model-draft MODEL_DRAFT] [--draft-max DRAFT_MAX] [--gpu-layers-draft GPU_LAYERS_DRAFT] [--device-draft DEVICE_DRAFT] [--ctx-size-draft CTX_SIZE_DRAFT] [--gpu-split GPU_SPLIT]
-                 [--autosplit] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa] [--num_experts_per_token N] [--enable_tp] [--cpp-runner] [--deepspeed] [--nvme-offload-dir NVME_OFFLOAD_DIR]
-                 [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT]
-                 [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE]
-                 [--subpath SUBPATH] [--old-colors] [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY]
-                 [--api-enable-ipv6] [--api-disable-ipv4] [--nowebui]
+                 [--gpu-layers N] [--tensor-split TENSOR_SPLIT] [--numa] [--no-kv-offload] [--row-split] [--extra-flags EXTRA_FLAGS] [--streaming-llm] [--mmproj MMPROJ] [--ctx-size N] [--cache-type N]
+                 [--model-draft MODEL_DRAFT] [--draft-max DRAFT_MAX] [--gpu-layers-draft GPU_LAYERS_DRAFT] [--device-draft DEVICE_DRAFT] [--ctx-size-draft CTX_SIZE_DRAFT] [--enable-tp]
+                 [--tp-backend TP_BACKEND] [--gpu-split GPU_SPLIT] [--autosplit] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa] [--num_experts_per_token N] [--cpp-runner] [--deepspeed]
+                 [--nvme-offload-dir NVME_OFFLOAD_DIR] [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB] [--listen]
+                 [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE]
+                 [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors] [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY]
+                 [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4] [--nowebui]
 
 Text generation web UI
 
@@ -299,6 +292,7 @@ llama.cpp:
   --row-split                               Split the model by rows across GPUs. This may improve multi-gpu performance.
   --extra-flags EXTRA_FLAGS                 Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"
   --streaming-llm                           Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.
+  --mmproj MMPROJ                           Path to the mmproj file for vision models.
 
 Context and cache:
   --ctx-size N, --n_ctx N, --max_seq_len N  Context size in tokens.
@@ -312,6 +306,10 @@ Speculative decoding:
   --device-draft DEVICE_DRAFT               Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1
   --ctx-size-draft CTX_SIZE_DRAFT           Size of the prompt context for the draft model. If 0, uses the same as the main model.
 
+ExLlamaV3:
+  --enable-tp, --enable_tp                  Enable Tensor Parallelism (TP) to split the model across GPUs.
+  --tp-backend TP_BACKEND                   The backend for tensor parallelism. Valid options: native, nccl. Default: native.
+
 ExLlamaV2:
   --gpu-split GPU_SPLIT                     Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.
   --autosplit                               Autosplit the model tensors across the available GPUs. This causes --gpu-split to be ignored.
@@ -320,7 +318,6 @@ ExLlamaV2:
   --no_xformers                             Force xformers to not be used.
   --no_sdpa                                 Force Torch SDPA to not be used.
   --num_experts_per_token N                 Number of experts to use for generation. Applies to MoE models like Mixtral.
-  --enable_tp                               Enable Tensor Parallelism (TP) in ExLlamaV2.
 
 TensorRT-LLM:
   --cpp-runner                              Use the ModelRunnerCpp runner, which is faster than the default ModelRunner but doesn't support streaming yet.

From 8a14aa62ff369615de895d155b0a105a3b4f7cb8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 Aug 2025 22:06:59 -0700
Subject: [PATCH 0926/1701] Update README

---
 README.md | 1 +
 1 file changed, 1 insertion(+)

diff --git a/README.md b/README.md
index f213f7a9dd..6e5cb5d8dc 100644
--- a/README.md
+++ b/README.md
@@ -83,6 +83,7 @@ To update, run the update script for your OS: `update_wizard_windows.bat`, `upda
 
 To reinstall with a fresh Python environment, delete the `installer_files` folder and run the `start_` script again.
 
+
            
 
 One-click installer details
 

From 6b1b2e2373df2a17ac48eaf5c53494aa8f4b8a57 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 Aug 2025 22:19:20 -0700
Subject: [PATCH 0927/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 6e5cb5d8dc..6b49cee06a 100644
--- a/README.md
+++ b/README.md
@@ -378,7 +378,7 @@ text-generation-webui
         └── llama-2-13b-chat.Q4_K_M.gguf
 ```
 
-* The remaining model types (like 16-bit Transformers models and EXL2 models) are made of several files and must be placed in a subfolder. Example:
+* The remaining model types (like 16-bit Transformers models and EXL3 models) are made of several files and must be placed in a subfolder. Example:
 
 ```
 text-generation-webui

From 15f99b1b710aced1ce8db70748a2a82602457661 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 18 Aug 2025 05:51:04 -0700
Subject: [PATCH 0928/1701] Installer: Fix a requirement file

---
 one_click.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/one_click.py b/one_click.py
index 1ea5e59f6e..2c7c2c0c9f 100644
--- a/one_click.py
+++ b/one_click.py
@@ -171,14 +171,14 @@ def get_requirements_file(gpu_choice):
     """Get requirements file path based on GPU choice"""
     requirements_base = os.path.join("requirements", "full")
 
-    if gpu_choice == "AMD":
+    if gpu_choice == "NVIDIA_CUDA128":
+        file_name = f"requirements{'_noavx2' if not cpu_has_avx2() else ''}.txt"
+    elif gpu_choice == "AMD":
         file_name = f"requirements_amd{'_noavx2' if not cpu_has_avx2() else ''}.txt"
     elif gpu_choice == "APPLE":
         file_name = f"requirements_apple_{'intel' if is_x86_64() else 'silicon'}.txt"
     elif gpu_choice in ["INTEL", "NONE"]:
         file_name = f"requirements_cpu_only{'_noavx2' if not cpu_has_avx2() else ''}.txt"
-    elif gpu_choice == "NVIDIA_CUDA128":
-        file_name = f"requirements_cuda128{'_noavx2' if not cpu_has_avx2() else ''}.txt"
     else:
         raise ValueError(f"Unknown GPU choice: {gpu_choice}")
 

From 08594e52636d6a6d583a87ecb6fd10e49821c500 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 18 Aug 2025 05:59:46 -0700
Subject: [PATCH 0929/1701] Installer: Slight improvement

---
 one_click.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/one_click.py b/one_click.py
index 2c7c2c0c9f..58c0e8ff18 100644
--- a/one_click.py
+++ b/one_click.py
@@ -408,7 +408,7 @@ def update_requirements(initial_installation=False, pull=True):
             with open(requirements_file, 'r') as f:
                 after_pull_whl_lines = [line for line in f if '.whl' in line]
 
-        wheels_changed = wheels_changed or (before_pull_whl_lines != after_pull_whl_lines)
+            wheels_changed = wheels_changed or (before_pull_whl_lines != after_pull_whl_lines)
 
         # Check for changes to installer files
         for file in files_to_check:

From 7d23a55901a43c323d2afe6d8c4585e7c9c3bca2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 18 Aug 2025 09:05:47 -0700
Subject: [PATCH 0930/1701] Fix model unloading when switching loaders (closes
 #7203)

---
 modules/models.py | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/modules/models.py b/modules/models.py
index 938eed3dc2..e620957b65 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -128,10 +128,12 @@ def unload_model(keep_model_name=False):
     if shared.model is None:
         return
 
-    is_llamacpp = (shared.model.__class__.__name__ == 'LlamaServer')
-    if shared.args.loader in ['ExLlamav3_HF', 'ExLlamav3']:
+    model_class_name = shared.model.__class__.__name__
+    is_llamacpp = (model_class_name == 'LlamaServer')
+
+    if model_class_name in ['Exllamav3Model', 'Exllamav3HF']:
         shared.model.unload()
-    elif shared.args.loader in ['ExLlamav2_HF', 'ExLlamav2'] and hasattr(shared.model, 'unload'):
+    elif model_class_name in ['Exllamav2Model', 'Exllamav2HF'] and hasattr(shared.model, 'unload'):
         shared.model.unload()
 
     shared.model = shared.tokenizer = None

From 8805a50d24066dd5645b2dbb85595bc07d75c34c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 18 Aug 2025 15:31:01 -0700
Subject: [PATCH 0931/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index d57a457c83..9f906b2620 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 47bcb60a1b..70e031b8a4 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 6958ce3794..81556326d3 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 13361a7807..7b9d365076 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index e4a30168b5..0fc9162ff5 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 3a9a953bdf..3565a994d5 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index a3e176d3dd..64c174163a 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index b073a3a903..2b16230833 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 0c7f1d29ba..943ea600e7 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 09f1c5029f..394b89b6d3 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 75296cb47c..cffe3aea35 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -18,6 +18,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index ff3d7cb1ca..d274e2c8c1 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 97414bde08..47ec086e95 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 7f54320562..9a0a3694b0 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index c1764eadbc..45e96da915 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 142b67ec40..9183562e63 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.36.0/llama_cpp_binaries-0.36.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From cbba58bef9f70a366413cce145a907658a24a982 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 18 Aug 2025 15:50:09 -0700
Subject: [PATCH 0932/1701] UI: Fix code blocks having an extra empty line

---
 modules/html_generator.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index cb14a72248..279f9ba6fe 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -306,6 +306,9 @@ def process_markdown_content(string):
         # Convert to HTML using markdown
         html_output = markdown.markdown(result, extensions=['fenced_code', 'tables', SaneListExtension()])
 
+    # Remove extra newlines before 
+    html_output = re.sub(r'\s*', '', html_output)
+
     # Unescape code blocks
     pattern = re.compile(r']*>(.*?)', re.DOTALL)
     html_output = pattern.sub(lambda x: html.unescape(x.group()), html_output)

From 5b06284a8af7d5bf068210124797fa7e4b31ade4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 19 Aug 2025 06:23:21 -0700
Subject: [PATCH 0933/1701] UI: Keep ExLlamav3_HF selected if already selected
 for EXL3 models

---
 modules/models_settings.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index bf7b1cf916..c325fa0c4e 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -251,7 +251,7 @@ def apply_model_settings_to_state(model, state):
     model_settings = get_model_metadata(model)
     if 'loader' in model_settings:
         loader = model_settings.pop('loader')
-        if not (loader == 'ExLlamav2_HF' and state['loader'] in ['ExLlamav2']):
+        if not ((loader == 'ExLlamav2_HF' and state['loader'] == 'ExLlamav2') or (loader == 'ExLlamav3_HF' and state['loader'] == 'ExLlamav3')):
             state['loader'] = loader
 
     for k in model_settings:

From e0f5905a97bd40a343003b4626e08b3fec9416de Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 19 Aug 2025 06:34:05 -0700
Subject: [PATCH 0934/1701] Code formatting

---
 modules/exllamav3.py | 3 ++-
 one_click.py         | 2 +-
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 739629772a..fd676a005d 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -236,11 +236,12 @@ def generate_with_streaming(self, prompt, state):
         """
         Generate text with streaming using native ExLlamaV3 API
         """
-        image_embeddings = []
 
         if shared.is_multimodal:
             # Process images and modify prompt (ExLlamaV3-specific)
             prompt, image_embeddings = self._process_images_for_generation(prompt, state)
+        else:
+            image_embeddings = []
 
         # Greedy decoding is a special case
         if state['temperature'] == 0:
diff --git a/one_click.py b/one_click.py
index 58c0e8ff18..881d7489b9 100644
--- a/one_click.py
+++ b/one_click.py
@@ -155,7 +155,7 @@ def get_pytorch_update_command(gpu_choice):
     base_cmd = f"python -m pip install --upgrade torch=={TORCH_VERSION} "
 
     if gpu_choice == "NVIDIA_CUDA128":
-        return "python -m pip install --upgrade torch==2.7.1 --index-url https://download.pytorch.org/whl/cu128"
+        return f"{base_cmd} --index-url https://download.pytorch.org/whl/cu128"
     elif gpu_choice == "AMD":
         return f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.2.4"
     elif gpu_choice in ["APPLE", "NONE"]:

From 1972479610f4b1482912ff012469e8ab9cbaa908 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 19 Aug 2025 06:48:22 -0700
Subject: [PATCH 0935/1701] Add the TP option to exllamav3_HF

---
 modules/exllamav3_hf.py | 5 +++++
 modules/loaders.py      | 2 ++
 2 files changed, 7 insertions(+)

diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index 1254ff5dad..d9f4ed576a 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -74,6 +74,11 @@ def __init__(self, model_dir):
             split = [float(alloc) for alloc in shared.args.gpu_split.split(",")]
             load_params['use_per_device'] = split
 
+        # Tensor-parallelism
+        if shared.args.enable_tp:
+            load_params['tensor_p'] = True
+            load_params['tp_backend'] = shared.args.tp_backend
+
         self.ex_model.load(**load_params)
         self.past_seq = None
         self.max_tokens = max_tokens
diff --git a/modules/loaders.py b/modules/loaders.py
index 295db1e750..f88e976dd1 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -56,6 +56,8 @@
         'cfg_cache',
         'trust_remote_code',
         'no_use_fast',
+        'enable_tp',
+        'tp_backend',
     ],
     'ExLlamav3': [
         'ctx_size',

From 9e7b326e3402de37adadc8509764738e98113763 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 19 Aug 2025 06:50:40 -0700
Subject: [PATCH 0936/1701] Lint

---
 modules/models.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/models.py b/modules/models.py
index e620957b65..ca3d184f71 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -55,7 +55,7 @@ def load_model(model_name, loader=None):
     if loader.lower().startswith('exllama') or loader.lower().startswith('tensorrt') or loader == 'llama.cpp':
         shared.settings['truncation_length'] = shared.args.ctx_size
 
-    shared.is_multimodal = False    
+    shared.is_multimodal = False
     if loader.lower() in ('exllamav3', 'llama.cpp'):
         shared.is_multimodal = model.is_multimodal()
 

From fd41f2fafcc6e286b69ba7efe2f5214d89f834ca Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 22 Aug 2025 11:18:56 -0700
Subject: [PATCH 0937/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 9f906b2620..f1d40000ba 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 70e031b8a4..437a10d957 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 81556326d3..0170d951ac 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 7b9d365076..7b369c4028 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 0fc9162ff5..1d1f44e0cf 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 3565a994d5..e63e97052c 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 64c174163a..c03a718a95 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 2b16230833..70b73e83e0 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 943ea600e7..ab91a76325 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 394b89b6d3..0faa6502d3 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index cffe3aea35..e1024942bc 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -18,6 +18,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index d274e2c8c1..3d2b6338da 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 47ec086e95..a95b30b361 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 9a0a3694b0..2eb7f59738 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 45e96da915..3244c9d434 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 9183562e63..685c7d1c06 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.37.0/llama_cpp_binaries-0.37.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From f247c2ae62fa246414bededf901df058665f819b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 22 Aug 2025 11:46:02 -0700
Subject: [PATCH 0938/1701] Make --model work with absolute paths, eg --model
 /tmp/gemma-3-270m-it-IQ4_NL.gguf

---
 modules/models.py          | 12 ++++++++++--
 modules/models_settings.py | 16 +++++++++-------
 modules/utils.py           | 13 +++++++++++++
 server.py                  | 13 +++----------
 4 files changed, 35 insertions(+), 19 deletions(-)

diff --git a/modules/models.py b/modules/models.py
index ca3d184f71..cae88ac598 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -5,6 +5,7 @@
 import modules.shared as shared
 from modules.logging_colors import logger
 from modules.models_settings import get_model_metadata
+from modules.utils import resolve_model_path
 
 last_generation_time = time.time()
 
@@ -69,17 +70,24 @@ def load_model(model_name, loader=None):
 def llama_cpp_server_loader(model_name):
     from modules.llama_cpp_server import LlamaServer
 
-    path = Path(f'{shared.args.model_dir}/{model_name}')
+    path = resolve_model_path(model_name)
+
     if path.is_file():
         model_file = path
     else:
-        model_file = sorted(Path(f'{shared.args.model_dir}/{model_name}').glob('*.gguf'))[0]
+        gguf_files = sorted(path.glob('*.gguf'))
+        if not gguf_files:
+            logger.error(f"No .gguf models found in the directory: {path}")
+            return None, None
+
+        model_file = gguf_files[0]
 
     try:
         model = LlamaServer(model_file)
         return model, model
     except Exception as e:
         logger.error(f"Error loading the model with llama.cpp: {str(e)}")
+        return None, None
 
 
 def transformers_loader(model_name):
diff --git a/modules/models_settings.py b/modules/models_settings.py
index c325fa0c4e..aa16fdb99d 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -10,6 +10,7 @@
 
 from modules import chat, loaders, metadata_gguf, shared, ui
 from modules.logging_colors import logger
+from modules.utils import resolve_model_path
 
 
 def get_fallback_settings():
@@ -26,6 +27,7 @@ def get_fallback_settings():
 
 
 def get_model_metadata(model):
+    model_path = resolve_model_path(model)
     model_settings = {}
 
     # Get settings from user_data/models/config.yaml and user_data/models/config-user.yaml
@@ -35,7 +37,7 @@ def get_model_metadata(model):
             for k in settings[pat]:
                 model_settings[k] = settings[pat][k]
 
-    path = Path(f'{shared.args.model_dir}/{model}/config.json')
+    path = model_path / 'config.json'
     if path.exists():
         hf_metadata = json.loads(open(path, 'r', encoding='utf-8').read())
     else:
@@ -51,7 +53,7 @@ def get_model_metadata(model):
 
     # GGUF metadata
     if model_settings['loader'] == 'llama.cpp':
-        path = Path(f'{shared.args.model_dir}/{model}')
+        path = model_path
         if path.is_file():
             model_file = path
         else:
@@ -130,18 +132,18 @@ def get_model_metadata(model):
                 model_settings['bf16'] = True
 
     # Try to find the Jinja instruct template
-    path = Path(f'{shared.args.model_dir}/{model}') / 'tokenizer_config.json'
+    path = model_path / 'tokenizer_config.json'
     template = None
 
     # 1. Prioritize reading from chat_template.jinja if it exists
-    jinja_path = Path(f'{shared.args.model_dir}/{model}') / 'chat_template.jinja'
+    jinja_path = model_path / 'chat_template.jinja'
     if jinja_path.exists():
         with open(jinja_path, 'r', encoding='utf-8') as f:
             template = f.read()
 
     # 2. If no .jinja file, try chat_template.json
     if template is None:
-        json_template_path = Path(f'{shared.args.model_dir}/{model}') / 'chat_template.json'
+        json_template_path = model_path / 'chat_template.json'
         if json_template_path.exists():
             with open(json_template_path, 'r', encoding='utf-8') as f:
                 json_data = json.load(f)
@@ -201,7 +203,7 @@ def get_model_metadata(model):
 
 
 def infer_loader(model_name, model_settings, hf_quant_method=None):
-    path_to_model = Path(f'{shared.args.model_dir}/{model_name}')
+    path_to_model = resolve_model_path(model_name)
     if not path_to_model.exists():
         loader = None
     elif shared.args.portable:
@@ -357,7 +359,7 @@ def get_model_size_mb(model_file: Path) -> float:
 
 
 def estimate_vram(gguf_file, gpu_layers, ctx_size, cache_type):
-    model_file = Path(f'{shared.args.model_dir}/{gguf_file}')
+    model_file = resolve_model_path(gguf_file)
     metadata = load_gguf_metadata_with_cache(model_file)
     size_in_mb = get_model_size_mb(model_file)
 
diff --git a/modules/utils.py b/modules/utils.py
index 4927ef04d4..e8d23a0284 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -86,6 +86,19 @@ def check_model_loaded():
     return True, None
 
 
+def resolve_model_path(model_name_or_path):
+    """
+    Resolves a model path, checking for a direct path
+    before the default models directory.
+    """
+
+    path_candidate = Path(model_name_or_path)
+    if path_candidate.exists():
+        return path_candidate
+    else:
+        return Path(f'{shared.args.model_dir}/{model_name_or_path}')
+
+
 def get_available_models():
     # Get all GGUF files
     gguf_files = get_available_ggufs()
diff --git a/server.py b/server.py
index 7ce3c208f9..e6687a3cb6 100644
--- a/server.py
+++ b/server.py
@@ -283,21 +283,14 @@ def create_interface():
 
     # If any model has been selected, load it
     if shared.model_name != 'None':
-        p = Path(shared.model_name)
-        if p.exists():
-            model_name = p.parts[-1]
-            shared.model_name = model_name
-        else:
-            model_name = shared.model_name
-
-        model_settings = get_model_metadata(model_name)
+        model_settings = get_model_metadata(shared.model_name)
         update_model_parameters(model_settings, initial=True)  # hijack the command-line arguments
 
         # Auto-adjust GPU layers if not provided by user and it's a llama.cpp model
         if 'gpu_layers' not in shared.provided_arguments and shared.args.loader == 'llama.cpp' and 'gpu_layers' in model_settings:
             vram_usage, adjusted_layers = update_gpu_layers_and_vram(
                 shared.args.loader,
-                model_name,
+                shared.model_name,
                 model_settings['gpu_layers'],
                 shared.args.ctx_size,
                 shared.args.cache_type,
@@ -308,7 +301,7 @@ def create_interface():
             shared.args.gpu_layers = adjusted_layers
 
         # Load the model
-        shared.model, shared.tokenizer = load_model(model_name)
+        shared.model, shared.tokenizer = load_model(shared.model_name)
         if shared.args.lora:
             add_lora_to_model(shared.args.lora)
 

From 7fe8da89448f795838a34adc0b2246855127e4a2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 22 Aug 2025 14:42:56 -0700
Subject: [PATCH 0939/1701] Minor simplification after
 f247c2ae62fa246414bededf901df058665f819b

---
 modules/llama_cpp_server.py | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 5953803a7f..8f1924cb07 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -20,6 +20,7 @@
     convert_pil_to_base64
 )
 from modules.logging_colors import logger
+from modules.utils import resolve_model_path
 
 llamacpp_valid_cache_types = {"fp16", "q8_0", "q4_0"}
 
@@ -351,14 +352,12 @@ def _start_server(self):
             if path.exists():
                 cmd += ["--mmproj", str(path)]
         if shared.args.model_draft not in [None, 'None']:
-            path = Path(shared.args.model_draft)
-            if not path.exists():
-                path = Path(f'{shared.args.model_dir}/{shared.args.model_draft}')
+            path = resolve_model_path(shared.args.model_draft)
 
             if path.is_file():
                 model_file = path
             else:
-                model_file = sorted(Path(f'{shared.args.model_dir}/{shared.args.model_draft}').glob('*.gguf'))[0]
+                model_file = sorted(path.glob('*.gguf'))[0]
 
             cmd += ["--model-draft", model_file]
             if shared.args.draft_max > 0:

From 8be798e15f48bd1f498d2c609ddf2f31cf22524b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 24 Aug 2025 12:19:19 -0700
Subject: [PATCH 0940/1701] llama.cpp: Fix stderr deadlock while loading some
 multimodal models

---
 modules/llama_cpp_server.py | 76 +++++++++++++++++++++++--------------
 1 file changed, 48 insertions(+), 28 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 8f1924cb07..e3dd43b429 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -410,8 +410,7 @@ def _start_server(self):
         self.process = subprocess.Popen(
             cmd,
             stderr=subprocess.PIPE,
-            text=True,
-            bufsize=1,
+            bufsize=0,
             env=env
         )
 
@@ -473,34 +472,55 @@ def filter_stderr_with_progress(process_stderr):
     last_was_progress = False
 
     try:
-        for raw in iter(process_stderr.readline, ''):
-            line = raw.rstrip('\r\n')
-            match = progress_re.search(line)
-
-            if match:
-                progress = float(match.group(1))
-
-                # Extract just the part from "prompt processing" onwards
-                prompt_processing_idx = line.find('prompt processing')
-                if prompt_processing_idx != -1:
-                    display_line = line[prompt_processing_idx:]
-                else:
-                    display_line = line  # fallback to full line
-
-                # choose carriage return for in-progress or newline at completion
-                end_char = '\r' if progress < 1.0 else '\n'
-                print(display_line, end=end_char, file=sys.stderr, flush=True)
-                last_was_progress = (progress < 1.0)
+        # Read in binary mode and decode manually
+        buffer = b""
+        while True:
+            # Read chunks aggressively to prevent buffer overflow
+            chunk = process_stderr.read(4096)
+            if not chunk:
+                break
 
-            # skip noise lines
-            elif not (line.startswith(('srv ', 'slot ')) or 'log_server_r: request: GET /health' in line):
-                # if we were in progress, finish that line first
-                if last_was_progress:
-                    print(file=sys.stderr)
+            buffer += chunk
 
-                print(line, file=sys.stderr, flush=True)
-                last_was_progress = False
+            # Process complete lines
+            while b'\n' in buffer:
+                line_bytes, buffer = buffer.split(b'\n', 1)
+                try:
+                    line = line_bytes.decode('utf-8', errors='replace').strip('\r\n')
+                    if line:  # Process non-empty lines
+                        match = progress_re.search(line)
+
+                        if match:
+                            progress = float(match.group(1))
+
+                            # Extract just the part from "prompt processing" onwards
+                            prompt_processing_idx = line.find('prompt processing')
+                            if prompt_processing_idx != -1:
+                                display_line = line[prompt_processing_idx:]
+                            else:
+                                display_line = line  # fallback to full line
+
+                            # choose carriage return for in-progress or newline at completion
+                            end_char = '\r' if progress < 1.0 else '\n'
+                            print(display_line, end=end_char, file=sys.stderr, flush=True)
+                            last_was_progress = (progress < 1.0)
+
+                        # skip noise lines
+                        elif not (line.startswith(('srv ', 'slot ')) or 'log_server_r: request: GET /health' in line):
+                            # if we were in progress, finish that line first
+                            if last_was_progress:
+                                print(file=sys.stderr)
+
+                            print(line, file=sys.stderr, flush=True)
+                            last_was_progress = False
+
+                except Exception:
+                    continue
 
     except (ValueError, IOError):
-        # silently ignore broken output or IO errors
         pass
+    finally:
+        try:
+            process_stderr.close()
+        except:
+            pass

From 2478294c06dac4ec749f9d37532e1e258a322ee6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 24 Aug 2025 12:37:41 -0700
Subject: [PATCH 0941/1701] UI: Preload the instruct and chat fonts

---
 modules/block_requests.py | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/modules/block_requests.py b/modules/block_requests.py
index dc1ee4671e..618b4bd682 100644
--- a/modules/block_requests.py
+++ b/modules/block_requests.py
@@ -38,7 +38,6 @@ def my_get(url, **kwargs):
     return requests.api.request('get', 'http://127.0.0.1/', **kwargs)
 
 
-# Kindly provided by our friend WizardLM-30B
 def my_open(*args, **kwargs):
     filename = str(args[0])
     if filename.endswith(('index.html', 'share.html')):
@@ -52,6 +51,10 @@ def my_open(*args, **kwargs):
         file_contents = file_contents.replace('cdnjs.cloudflare.com', '127.0.0.1')
         file_contents = file_contents.replace(
             '',
+            '\n    '
+            '\n    '
+            '\n    '
+            '\n    '
             '\n    '
             '\n    '
             '\n    '

From 1f77427088a1487fd9afc8d43282aba6d51557b6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 24 Aug 2025 19:56:22 -0700
Subject: [PATCH 0942/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index f1d40000ba..c53e772229 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 437a10d957..776dbc7bd4 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 0170d951ac..7205a72e7e 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 7b369c4028..f4e00c5b3a 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -33,7 +33,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 1d1f44e0cf..0b28ad216d 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -33,8 +33,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index e63e97052c..fcbb48f0be 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index c03a718a95..1705791ebd 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -33,5 +33,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 70b73e83e0..7e95121968 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index ab91a76325..9eb8b33c0c 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 0faa6502d3..0a27f61b4f 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index e1024942bc..79674e5a92 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -18,6 +18,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 3d2b6338da..3ebb7d8bd6 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index a95b30b361..81b78fe6e4 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 2eb7f59738..58dc529a6a 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 3244c9d434..5ad8ede10c 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 685c7d1c06..0adf6e48df 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -18,5 +18,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.38.0/llama_cpp_binaries-0.38.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 3bc48014a5bfdf633b814c23bbb5b42212293b06 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 16:48:21 -0700
Subject: [PATCH 0943/1701] chat.py code simplifications

---
 modules/chat.py | 29 ++++++++++++-----------------
 1 file changed, 12 insertions(+), 17 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index ab6b43c078..022ab8c930 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -213,13 +213,11 @@ def generate_chat_prompt(user_input, state, **kwargs):
         if assistant_msg:
             # Handle GPT-OSS as a special case
             if '<|channel|>analysis<|message|>' in assistant_msg or '<|channel|>final<|message|>' in assistant_msg:
-
                 thinking_content = ""
                 final_content = ""
 
                 # Extract analysis content if present
                 if '<|channel|>analysis<|message|>' in assistant_msg:
-                    # Split the message by the analysis tag to isolate the content that follows
                     parts = assistant_msg.split('<|channel|>analysis<|message|>', 1)
                     if len(parts) > 1:
                         # The content is everything after the tag
@@ -240,7 +238,6 @@ def generate_chat_prompt(user_input, state, **kwargs):
                 # Extract final content if present
                 final_tag_to_find = '<|channel|>final<|message|>'
                 if final_tag_to_find in assistant_msg:
-                    # Split the message by the final tag to isolate the content that follows
                     parts = assistant_msg.split(final_tag_to_find, 1)
                     if len(parts) > 1:
                         # The content is everything after the tag
@@ -261,6 +258,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
                 messages.insert(insert_pos, msg_dict)
 
             else:
+                # Default case (used by all other models)
                 messages.insert(insert_pos, {"role": "assistant", "content": assistant_msg})
 
         if user_msg not in ['', '<|BEGIN-VISIBLE-CHAT|>']:
@@ -286,18 +284,17 @@ def generate_chat_prompt(user_input, state, **kwargs):
                         else:
                             attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
 
-                if image_refs or attachments_text:
-                    enhanced_user_msg = user_msg
-                    if image_refs:
-                        enhanced_user_msg = f"{image_refs}\n\n{enhanced_user_msg}"
-                    if attachments_text:
-                        enhanced_user_msg += f"\n\nATTACHMENTS:\n{attachments_text}"
+                if image_refs:
+                    enhanced_user_msg = f"{image_refs}\n\n{enhanced_user_msg}"
+                if attachments_text:
+                    enhanced_user_msg += f"\n\nATTACHMENTS:\n{attachments_text}"
 
             messages.insert(insert_pos, {"role": "user", "content": enhanced_user_msg})
 
+    # Handle the current user input
     user_input = user_input.strip()
 
-    # Check if we have attachments even with empty input
+    # Check if we have attachments
     has_attachments = False
     if not impersonate and not _continue and len(history_data.get('metadata', {})) > 0:
         current_row_idx = len(history)
@@ -306,7 +303,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
     if (user_input or has_attachments) and not impersonate and not _continue:
         # For the current user input being processed, check if we need to add attachments
-        if not impersonate and not _continue and len(history_data.get('metadata', {})) > 0:
+        if len(history_data.get('metadata', {})) > 0:
             current_row_idx = len(history)
             user_key = f"user_{current_row_idx}"
 
@@ -325,12 +322,10 @@ def generate_chat_prompt(user_input, state, **kwargs):
                         else:
                             attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
 
-                if image_refs or attachments_text:
-                    user_input = user_input
-                    if image_refs:
-                        user_input = f"{image_refs}\n\n{user_input}"
-                    if attachments_text:
-                        user_input += f"\n\nATTACHMENTS:\n{attachments_text}"
+                if image_refs:
+                    user_input = f"{image_refs}\n\n{user_input}"
+                if attachments_text:
+                    user_input += f"\n\nATTACHMENTS:\n{attachments_text}"
 
         messages.append({"role": "user", "content": user_input})
 

From d08800c359bbc90172294a78f569cf284148d4b4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 17:03:37 -0700
Subject: [PATCH 0944/1701] chat.py improvements

---
 modules/chat.py | 16 +++++++---------
 1 file changed, 7 insertions(+), 9 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 022ab8c930..cd82b8134b 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -336,10 +336,6 @@ def make_prompt(messages):
             prompt = renderer(messages=messages)
 
         if state['mode'] == 'chat-instruct':
-            outer_messages = []
-            if state['custom_system_message'].strip() != '':
-                outer_messages.append({"role": "system", "content": state['custom_system_message']})
-
             command = state['chat-instruct_command']
             command = command.replace('<|character|>', state['name2'] if not impersonate else state['name1'])
             command = command.replace('<|prompt|>', prompt)
@@ -353,29 +349,31 @@ def make_prompt(messages):
                 if not impersonate:
                     prefix = apply_extensions('bot_prefix', prefix, state)
 
+            suffix = get_generation_prompt(instruct_renderer, impersonate=False)[1]
+
+            outer_messages = []
+            if state['custom_system_message'].strip() != '':
+                outer_messages.append({"role": "system", "content": state['custom_system_message']})
+
             outer_messages.append({"role": "user", "content": command})
             outer_messages.append({"role": "assistant", "content": prefix})
 
             prompt = instruct_renderer(messages=outer_messages)
-            suffix = get_generation_prompt(instruct_renderer, impersonate=False)[1]
             if len(suffix) > 0:
                 prompt = prompt[:-len(suffix)]
         else:
             # Handle GPT-OSS as a special case when continuing
+            # (otherwise the thinking block gets removed...)
             if _continue and '<|channel|>final<|message|>' in state['instruction_template_str']:
                 last_message_to_continue = messages[-1]
                 prompt = renderer(messages=messages[:-1])
 
-                # Start the assistant turn wrapper
                 assistant_reply_so_far = "<|start|>assistant"
-
                 if 'thinking' in last_message_to_continue:
                     assistant_reply_so_far += f"<|channel|>analysis<|message|>{last_message_to_continue['thinking']}<|end|>"
 
                 assistant_reply_so_far += f"<|channel|>final<|message|>{last_message_to_continue.get('content', '')}"
-
                 prompt += assistant_reply_so_far
-
             else:
                 prompt = renderer(messages=messages)
                 if _continue:

From f919cdf881ee45641d588fc664d2c4fe1cc71c4a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 17:20:51 -0700
Subject: [PATCH 0945/1701] chat.py code simplifications

---
 modules/chat.py | 128 +++++++++++++++---------------------------------
 1 file changed, 40 insertions(+), 88 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index cd82b8134b..023f5a3eea 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -86,36 +86,6 @@ def str_presenter(dumper, data):
 yaml.representer.SafeRepresenter.add_representer(str, str_presenter)
 
 
-def get_generation_prompt(renderer, impersonate=False, strip_trailing_spaces=True):
-    '''
-    Given a Jinja template, reverse-engineers the prefix and the suffix for
-    an assistant message (if impersonate=False) or an user message
-    (if impersonate=True)
-    '''
-
-    if impersonate:
-        messages = [
-            {"role": "user", "content": "<<|user-message-1|>>"},
-            {"role": "user", "content": "<<|user-message-2|>>"},
-        ]
-    else:
-        messages = [
-            {"role": "assistant", "content": "<<|user-message-1|>>"},
-            {"role": "assistant", "content": "<<|user-message-2|>>"},
-        ]
-
-    prompt = renderer(messages=messages)
-
-    suffix_plus_prefix = prompt.split("<<|user-message-1|>>")[1].split("<<|user-message-2|>>")[0]
-    suffix = prompt.split("<<|user-message-2|>>")[1]
-    prefix = suffix_plus_prefix[len(suffix):]
-
-    if strip_trailing_spaces:
-        prefix = prefix.rstrip(' ')
-
-    return prefix, suffix
-
-
 def get_thinking_suppression_string(template):
     """
     Determines what string needs to be added to suppress thinking mode
@@ -341,26 +311,16 @@ def make_prompt(messages):
             command = command.replace('<|prompt|>', prompt)
             command = replace_character_names(command, state['name1'], state['name2'])
 
-            if _continue:
-                prefix = get_generation_prompt(renderer, impersonate=impersonate, strip_trailing_spaces=False)[0]
-                prefix += messages[-1]["content"]
-            else:
-                prefix = get_generation_prompt(renderer, impersonate=impersonate)[0]
-                if not impersonate:
-                    prefix = apply_extensions('bot_prefix', prefix, state)
-
-            suffix = get_generation_prompt(instruct_renderer, impersonate=False)[1]
-
             outer_messages = []
             if state['custom_system_message'].strip() != '':
                 outer_messages.append({"role": "system", "content": state['custom_system_message']})
 
             outer_messages.append({"role": "user", "content": command})
-            outer_messages.append({"role": "assistant", "content": prefix})
 
-            prompt = instruct_renderer(messages=outer_messages)
-            if len(suffix) > 0:
-                prompt = prompt[:-len(suffix)]
+            prompt = instruct_renderer(
+                messages=outer_messages,
+                add_generation_prompt=True
+            )
         else:
             # Handle GPT-OSS as a special case when continuing
             # (otherwise the thinking block gets removed...)
@@ -375,29 +335,10 @@ def make_prompt(messages):
                 assistant_reply_so_far += f"<|channel|>final<|message|>{last_message_to_continue.get('content', '')}"
                 prompt += assistant_reply_so_far
             else:
-                prompt = renderer(messages=messages)
-                if _continue:
-                    suffix = get_generation_prompt(renderer, impersonate=impersonate)[1]
-                    if len(suffix) > 0:
-                        prompt = prompt[:-len(suffix)]
-                else:
-                    prefix = get_generation_prompt(renderer, impersonate=impersonate)[0]
-
-                    # Handle GPT-OSS as a special case when not continuing
-                    if '<|channel|>final<|message|>' in state['instruction_template_str']:
-                        if prefix.endswith("<|channel|>final<|message|>"):
-                            prefix = prefix[:-len("<|channel|>final<|message|>")]
-
-                        if impersonate:
-                            prefix += "<|message|>"
-
-                    if state['mode'] == 'chat' and not impersonate:
-                        prefix = apply_extensions('bot_prefix', prefix, state)
-
-                    prompt += prefix
-
-        if state['mode'] == 'instruct' and 'enable_thinking' in state['instruction_template_str'] and not any((_continue, impersonate, state['enable_thinking'])):
-            prompt += get_thinking_suppression_string(instruction_template)
+                prompt = renderer(
+                    messages=messages,
+                    add_generation_prompt=True
+                )
 
         return prompt
 
@@ -523,24 +464,41 @@ def get_stopping_strings(state):
         renderer = partial(template.render, add_generation_prompt=False, name1=state['name1'], name2=state['name2'])
         renderers.append(renderer)
 
+    fake_messages = [
+        {"role": "user", "content": "first user message"},
+        {"role": "assistant", "content": "first assistant message"},
+        {"role": "user", "content": "second user message"},
+        {"role": "assistant", "content": "second assistant message"},
+    ]
+
+    stopping_strings = []
     for renderer in renderers:
-        prefix_bot, suffix_bot = get_generation_prompt(renderer, impersonate=False)
-        prefix_user, suffix_user = get_generation_prompt(renderer, impersonate=True)
-
-        stopping_strings += [
-            suffix_user + prefix_bot,
-            suffix_user + prefix_user,
-            suffix_bot + prefix_bot,
-            suffix_bot + prefix_user,
+        prompt = renderer(messages=fake_messages)
+
+        # Find positions of each message content
+        first_user_end = prompt.find("first user message") + len("first user message")
+        first_assistant_start = prompt.find("first assistant message")
+        first_assistant_end = prompt.find("first assistant message") + len("first assistant message")
+        second_user_start = prompt.find("second user message")
+        second_assistant_end = prompt.find("second assistant message") + len("second assistant message")
+
+        # Extract pieces of text potentially containing unique stopping strings
+        texts = [
+            prompt[first_user_end:first_assistant_start],
+            prompt[first_assistant_end:second_user_start],
+            prompt[second_assistant_end:]
         ]
 
-    # Try to find the EOT token
-    for item in stopping_strings.copy():
-        item = item.strip()
-        if item.startswith("<") and ">" in item:
-            stopping_strings.append(item.split(">")[0] + ">")
-        elif item.startswith("[") and "]" in item:
-            stopping_strings.append(item.split("]")[0] + "]")
+        for text in texts:
+            text = text.strip()
+            if text.startswith("<") and ">" in text:
+                stopping_strings.append(text.split(">")[0] + ">")
+            elif text.startswith("[") and "]" in text:
+                stopping_strings.append(text.split("]")[0] + "]")
+            elif text.startswith("(") and ")" in text:
+                stopping_strings.append(text.split(")")[0] + ")")
+            elif text.startswith("{") and "}" in text:
+                stopping_strings.append(text.split("}")[0] + "}")
 
     if 'stopping_strings' in state and isinstance(state['stopping_strings'], list):
         stopping_strings += state.pop('stopping_strings')
@@ -549,12 +507,6 @@ def get_stopping_strings(state):
     result = [item for item in stopping_strings if not any(item.startswith(other) and item != other for other in stopping_strings)]
     result = list(set(result))
 
-    # Handle GPT-OSS as a special case
-    if '<|channel|>final<|message|>' in state['instruction_template_str'] and "<|end|>" in result:
-        result.remove("<|end|>")
-        result.append("<|result|>")
-        result = list(set(result))
-
     if shared.args.verbose:
         logger.info("STOPPING_STRINGS=")
         pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(result)

From aad0104c1b536d62c19e59f4afc5a90c703f169f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 17:33:13 -0700
Subject: [PATCH 0946/1701] Remove a function

---
 modules/chat.py | 38 --------------------------------------
 1 file changed, 38 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 023f5a3eea..05ed02bf91 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -86,44 +86,6 @@ def str_presenter(dumper, data):
 yaml.representer.SafeRepresenter.add_representer(str, str_presenter)
 
 
-def get_thinking_suppression_string(template):
-    """
-    Determines what string needs to be added to suppress thinking mode
-    by comparing template renderings with thinking enabled vs disabled.
-    """
-
-    # Render with thinking enabled
-    with_thinking = template.render(
-        messages=[{'role': 'user', 'content': ''}],
-        builtin_tools=None,
-        tools=None,
-        tools_in_user_message=False,
-        add_generation_prompt=True,
-        enable_thinking=True
-    )
-
-    # Render with thinking disabled
-    without_thinking = template.render(
-        messages=[{'role': 'user', 'content': ''}],
-        builtin_tools=None,
-        tools=None,
-        tools_in_user_message=False,
-        add_generation_prompt=True,
-        enable_thinking=False
-    )
-
-    # Find the difference (what gets added to suppress thinking)
-    i = 0
-    while i < min(len(with_thinking), len(without_thinking)) and with_thinking[i] == without_thinking[i]:
-        i += 1
-
-    j = 0
-    while j < min(len(with_thinking), len(without_thinking)) - i and with_thinking[-1 - j] == without_thinking[-1 - j]:
-        j += 1
-
-    return without_thinking[i:len(without_thinking) - j if j else None]
-
-
 def generate_chat_prompt(user_input, state, **kwargs):
     impersonate = kwargs.get('impersonate', False)
     _continue = kwargs.get('_continue', False)

From adeca8a65888f97b94dfdaff6b2492c031ec1ccd Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 17:36:01 -0700
Subject: [PATCH 0947/1701] Remove changes to the jinja2 templates

---
 modules/models_settings.py | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index aa16fdb99d..7645880ff8 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -94,8 +94,6 @@ def get_model_metadata(model):
 
             template = re.sub(r"\{\{-?\s*raise_exception\(.*?\)\s*-?\}\}", "", template, flags=re.DOTALL)
             template = re.sub(r'raise_exception\([^)]*\)', "''", template)
-            template = re.sub(r'{% if add_generation_prompt %}.*', '', template, flags=re.DOTALL)
-            template = re.sub(r'elif loop\.last and not add_generation_prompt', 'elif False', template)  # Handle GPT-OSS
             model_settings['instruction_template'] = 'Custom (obtained from model metadata)'
             model_settings['instruction_template_str'] = template
 
@@ -172,8 +170,6 @@ def get_model_metadata(model):
 
             template = re.sub(r"\{\{-?\s*raise_exception\(.*?\)\s*-?\}\}", "", template, flags=re.DOTALL)
             template = re.sub(r'raise_exception\([^)]*\)', "''", template)
-            template = re.sub(r'{% if add_generation_prompt %}.*', '', template, flags=re.DOTALL)
-            template = re.sub(r'elif loop\.last and not add_generation_prompt', 'elif False', template)  # Handle GPT-OSS
             model_settings['instruction_template'] = 'Custom (obtained from model metadata)'
             model_settings['instruction_template_str'] = template
 

From 3ad59703748dcd5685dcbb7368df45914661e8da Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 17:43:21 -0700
Subject: [PATCH 0948/1701] Make the llama.cpp --verbose output less verbose

---
 modules/llama_cpp_server.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index e3dd43b429..8579f84397 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -193,7 +193,7 @@ def generate_with_streaming(self, prompt, state):
 
         if shared.args.verbose:
             logger.info("GENERATE_PARAMS=")
-            printable_payload = {k: (v if k != "prompt" else "[multimodal object]" if pil_images else v) for k, v in payload.items()}
+            printable_payload = {k: v for k, v in payload.items() if k != "prompt"}
             pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
             print()
 

From b330ec35174f6b1b7e26922bdeef16069441bd8a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 17:54:15 -0700
Subject: [PATCH 0949/1701] Simplifications

---
 modules/chat.py | 26 +++++++++++---------------
 1 file changed, 11 insertions(+), 15 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 05ed02bf91..530e3a0a87 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -108,6 +108,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
         tools=state['tools'] if 'tools' in state else None,
         tools_in_user_message=False,
         add_generation_prompt=False,
+        enable_thinking=state['enable_thinking'],
         reasoning_effort=state['reasoning_effort']
     )
 
@@ -262,10 +263,10 @@ def generate_chat_prompt(user_input, state, **kwargs):
         messages.append({"role": "user", "content": user_input})
 
     def make_prompt(messages):
-        if state['mode'] == 'chat-instruct' and _continue:
-            prompt = renderer(messages=messages[:-1])
-        else:
-            prompt = renderer(messages=messages)
+        prompt = renderer(
+            messages=messages[:-1] if _continue else messages,
+            add_generation_prompt=(state['mode'] != 'chat-instruct')
+        )
 
         if state['mode'] == 'chat-instruct':
             command = state['chat-instruct_command']
@@ -287,20 +288,15 @@ def make_prompt(messages):
             # Handle GPT-OSS as a special case when continuing
             # (otherwise the thinking block gets removed...)
             if _continue and '<|channel|>final<|message|>' in state['instruction_template_str']:
-                last_message_to_continue = messages[-1]
-                prompt = renderer(messages=messages[:-1])
-
                 assistant_reply_so_far = "<|start|>assistant"
-                if 'thinking' in last_message_to_continue:
-                    assistant_reply_so_far += f"<|channel|>analysis<|message|>{last_message_to_continue['thinking']}<|end|>"
+                if 'thinking' in messages[-1]:
+                    assistant_reply_so_far += f"<|channel|>analysis<|message|>{messages[-1]['thinking']}<|end|>"
 
-                assistant_reply_so_far += f"<|channel|>final<|message|>{last_message_to_continue.get('content', '')}"
+                assistant_reply_so_far += f"<|channel|>final<|message|>"
                 prompt += assistant_reply_so_far
-            else:
-                prompt = renderer(
-                    messages=messages,
-                    add_generation_prompt=True
-                )
+
+        if _continue:
+            prompt += messages[-1].get('content', '')
 
         return prompt
 

From c1aa4590ea3d69ba9ae8edd3bf222af27a3cd13b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 18:05:40 -0700
Subject: [PATCH 0950/1701] Code simplifications, fix impersonate

---
 modules/chat.py | 60 ++++++++++++++++++++++++++++---------------------
 1 file changed, 34 insertions(+), 26 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 530e3a0a87..7c2ab4a397 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -228,44 +228,48 @@ def generate_chat_prompt(user_input, state, **kwargs):
     user_input = user_input.strip()
 
     # Check if we have attachments
-    has_attachments = False
-    if not impersonate and not _continue and len(history_data.get('metadata', {})) > 0:
-        current_row_idx = len(history)
-        user_key = f"user_{current_row_idx}"
-        has_attachments = user_key in metadata and "attachments" in metadata[user_key]
-
-    if (user_input or has_attachments) and not impersonate and not _continue:
-        # For the current user input being processed, check if we need to add attachments
+    if not (impersonate or _continue):
+        has_attachments = False
         if len(history_data.get('metadata', {})) > 0:
             current_row_idx = len(history)
             user_key = f"user_{current_row_idx}"
+            has_attachments = user_key in metadata and "attachments" in metadata[user_key]
 
-            if user_key in metadata and "attachments" in metadata[user_key]:
-                attachments_text = ""
-                image_refs = ""
+        if user_input or has_attachments:
+            # For the current user input being processed, check if we need to add attachments
+            if len(history_data.get('metadata', {})) > 0:
+                current_row_idx = len(history)
+                user_key = f"user_{current_row_idx}"
 
-                for attachment in metadata[user_key]["attachments"]:
-                    if attachment.get("type") == "image":
-                        image_refs += "<__media__>"
-                    else:
-                        filename = attachment.get("name", "file")
-                        content = attachment.get("content", "")
-                        if attachment.get("type") == "text/html" and attachment.get("url"):
-                            attachments_text += f"\nName: {filename}\nURL: {attachment['url']}\nContents:\n\n=====\n{content}\n=====\n\n"
+                if user_key in metadata and "attachments" in metadata[user_key]:
+                    attachments_text = ""
+                    image_refs = ""
+
+                    for attachment in metadata[user_key]["attachments"]:
+                        if attachment.get("type") == "image":
+                            image_refs += "<__media__>"
                         else:
-                            attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
+                            filename = attachment.get("name", "file")
+                            content = attachment.get("content", "")
+                            if attachment.get("type") == "text/html" and attachment.get("url"):
+                                attachments_text += f"\nName: {filename}\nURL: {attachment['url']}\nContents:\n\n=====\n{content}\n=====\n\n"
+                            else:
+                                attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
 
-                if image_refs:
-                    user_input = f"{image_refs}\n\n{user_input}"
-                if attachments_text:
-                    user_input += f"\n\nATTACHMENTS:\n{attachments_text}"
+                    if image_refs:
+                        user_input = f"{image_refs}\n\n{user_input}"
+                    if attachments_text:
+                        user_input += f"\n\nATTACHMENTS:\n{attachments_text}"
 
-        messages.append({"role": "user", "content": user_input})
+            messages.append({"role": "user", "content": user_input})
+
+    if impersonate:
+        messages.append({"role": "user", "content": "fake user message replace me"})
 
     def make_prompt(messages):
         prompt = renderer(
             messages=messages[:-1] if _continue else messages,
-            add_generation_prompt=(state['mode'] != 'chat-instruct')
+            add_generation_prompt=(state['mode'] != 'chat-instruct' and not impersonate)
         )
 
         if state['mode'] == 'chat-instruct':
@@ -298,6 +302,10 @@ def make_prompt(messages):
         if _continue:
             prompt += messages[-1].get('content', '')
 
+        if impersonate:
+            prompt = prompt.split("fake user message replace me", 1)[0]
+            prompt += user_input
+
         return prompt
 
     prompt = make_prompt(messages)

From ded6c41cf8b5a95441516515f669766407d9692d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 18:16:17 -0700
Subject: [PATCH 0951/1701] Fix impersonate for chat-instruct

---
 modules/chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 7c2ab4a397..d2513e07a9 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -263,7 +263,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
             messages.append({"role": "user", "content": user_input})
 
-    if impersonate:
+    if impersonate and state['mode'] != 'chat-instruct':
         messages.append({"role": "user", "content": "fake user message replace me"})
 
     def make_prompt(messages):

From b657be73814329d9d8d81f1cec49fe7c738dc3ee Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 18:22:08 -0700
Subject: [PATCH 0952/1701] Obtain stopping strings in chat mode

---
 modules/chat.py | 20 +++++++++++---------
 1 file changed, 11 insertions(+), 9 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index d2513e07a9..8a9a5a1b22 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -456,15 +456,17 @@ def get_stopping_strings(state):
         ]
 
         for text in texts:
-            text = text.strip()
-            if text.startswith("<") and ">" in text:
-                stopping_strings.append(text.split(">")[0] + ">")
-            elif text.startswith("[") and "]" in text:
-                stopping_strings.append(text.split("]")[0] + "]")
-            elif text.startswith("(") and ")" in text:
-                stopping_strings.append(text.split(")")[0] + ")")
-            elif text.startswith("{") and "}" in text:
-                stopping_strings.append(text.split("}")[0] + "}")
+            stripped_text = text.strip()
+            if stripped_text.startswith("<") and ">" in stripped_text:
+                stopping_strings.append(stripped_text.split(">")[0] + ">")
+            elif stripped_text.startswith("[") and "]" in stripped_text:
+                stopping_strings.append(stripped_text.split("]")[0] + "]")
+            elif stripped_text.startswith("(") and ")" in stripped_text:
+                stopping_strings.append(stripped_text.split(")")[0] + ")")
+            elif stripped_text.startswith("{") and "}" in stripped_text:
+                stopping_strings.append(stripped_text.split("}")[0] + "}")
+            elif ":" in text:
+                stopping_strings.append(text.split(":")[0] + ":")
 
     if 'stopping_strings' in state and isinstance(state['stopping_strings'], list):
         stopping_strings += state.pop('stopping_strings')

From 6c165d2e55f41f6e0259e2175b9c7314b28a221f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 18:28:43 -0700
Subject: [PATCH 0953/1701] Fix the chat template

---
 modules/shared.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/shared.py b/modules/shared.py
index 644261a0bf..c3d96b70a3 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -287,7 +287,7 @@
     'greeting': 'How can I help you today?',
     'custom_system_message': '',
     'instruction_template_str': "{%- set ns = namespace(found=false) -%}\n{%- for message in messages -%}\n    {%- if message['role'] == 'system' -%}\n        {%- set ns.found = true -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if not ns.found -%}\n    {{- '' + 'Below is an instruction that describes a task. Write a response that appropriately completes the request.' + '\\n\\n' -}}\n{%- endif %}\n{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {{- '' + message['content'] + '\\n\\n' -}}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{-'### Instruction:\\n' + message['content'] + '\\n\\n'-}}\n        {%- else -%}\n            {{-'### Response:\\n' + message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{-'### Response:\\n'-}}\n{%- endif -%}",
-    'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{- name1 + ': ' + message['content'] + '\\n'-}}\n        {%- else -%}\n            {{- name2 + ': ' + message['content'] + '\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}",
+    'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{- name1 + ': ' + message['content'] + '\\n'-}}\n        {%- else -%}\n            {{- name2 + ': ' + message['content'] + '\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt %}\n    {{- name2 + ': ' -}}\n{%- endif %}",
 
     # Extensions
     'default_extensions': [],

From a531328f7eef0d7e6f4ac85186409ee2320586ee Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 18:41:58 -0700
Subject: [PATCH 0954/1701] Fix the GPT-OSS stopping string

---
 modules/chat.py | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index 8a9a5a1b22..96d36ba545 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -475,6 +475,12 @@ def get_stopping_strings(state):
     result = [item for item in stopping_strings if not any(item.startswith(other) and item != other for other in stopping_strings)]
     result = list(set(result))
 
+    # Handle GPT-OSS as a special case
+    if '<|channel|>final<|message|>' in state['instruction_template_str'] and "<|end|>" in result:
+        result.remove("<|end|>")
+        result.append("<|result|>")
+        result = list(set(result))
+
     if shared.args.verbose:
         logger.info("STOPPING_STRINGS=")
         pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(result)

From 8f660aefe361d396847ebce03d86f0e501561c17 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 18:50:16 -0700
Subject: [PATCH 0955/1701] Fix chat-instruct replies leaking the bot name
 sometimes

---
 modules/chat.py | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index 96d36ba545..a24a5be124 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -825,6 +825,12 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
 
         # Extract the reply
         if state['mode'] in ['chat', 'chat-instruct']:
+            reply = reply.lstrip()
+            if reply.startswith(state['name2'] + ':'):
+                reply = reply[len(state['name2'] + ':'):]
+            elif reply.startswith(state['name1'] + ':'):
+                reply = reply[len(state['name1'] + ':'):]
+
             visible_reply = re.sub("(||{{user}})", state['name1'], reply)
         else:
             visible_reply = reply

From 8fcb4b310242a37b58bd006c28b7bb29a688e767 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 19:10:46 -0700
Subject: [PATCH 0956/1701] Make bot_prefix extensions functional again

---
 modules/chat.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index a24a5be124..5eb9f30182 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -306,6 +306,9 @@ def make_prompt(messages):
             prompt = prompt.split("fake user message replace me", 1)[0]
             prompt += user_input
 
+        if state['mode'] in ['chat', 'chat-instruct'] and not impersonate and not _continue:
+            prompt += apply_extensions('bot_prefix', "", state)
+
         return prompt
 
     prompt = make_prompt(messages)

From 6a7166fffaa4361a923cf9d6a15a2f6b96a8be6d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 19:46:48 -0700
Subject: [PATCH 0957/1701] Add support for the Seed-OSS template

---
 modules/chat.py           | 29 ++++++++++++++++++++++++++++-
 modules/html_generator.py | 28 ++++++++++++++++++++++++++--
 2 files changed, 54 insertions(+), 3 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 5eb9f30182..818d1014aa 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -109,7 +109,8 @@ def generate_chat_prompt(user_input, state, **kwargs):
         tools_in_user_message=False,
         add_generation_prompt=False,
         enable_thinking=state['enable_thinking'],
-        reasoning_effort=state['reasoning_effort']
+        reasoning_effort=state['reasoning_effort'],
+        thinking_budget=-1 if state.get('enable_thinking', True) else 0
     )
 
     chat_renderer = partial(
@@ -190,6 +191,30 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
                 messages.insert(insert_pos, msg_dict)
 
+            # Handle Seed-OSS
+            elif '' in assistant_msg:
+                thinking_content = ""
+                final_content = assistant_msg
+
+                # Extract thinking content if present
+                if '' in assistant_msg:
+                    parts = assistant_msg.split('', 1)
+                    if len(parts) > 1:
+                        potential_content = parts[1]
+                        if '' in potential_content:
+                            thinking_content = potential_content.split('', 1)[0].strip()
+                            final_content = parts[0] + potential_content.split('', 1)[1]
+                        else:
+                            thinking_content = potential_content.strip()
+                            final_content = parts[0]
+
+                # Insert as structured message
+                msg_dict = {"role": "assistant", "content": final_content.strip()}
+                if thinking_content:
+                    msg_dict["reasoning_content"] = thinking_content
+
+                messages.insert(insert_pos, msg_dict)
+
             else:
                 # Default case (used by all other models)
                 messages.insert(insert_pos, {"role": "assistant", "content": assistant_msg})
@@ -687,6 +712,8 @@ def generate_search_query(user_message, state):
         query = query.rsplit("", 1)[1]
     elif "<|start|>assistant<|channel|>final<|message|>" in query:
         query = query.rsplit("<|start|>assistant<|channel|>final<|message|>", 1)[1]
+    elif "" in query:
+        query = query.rsplit("", 1)[1]
 
     # Strip and remove surrounding quotes if present
     query = query.strip()
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 279f9ba6fe..63844f357d 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -137,7 +137,7 @@ def extract_thinking_block(string):
         remaining_content = string[content_start:]
         return thinking_content, remaining_content
 
-    # If think tags not found, try alternative format
+    # If think tags not found, try GPT-OSS alternative format
     ALT_START = "<|channel|>analysis<|message|>"
     ALT_END = "<|end|>"
     ALT_CONTENT_START = "<|start|>assistant<|channel|>final<|message|>"
@@ -168,7 +168,31 @@ def extract_thinking_block(string):
         remaining_content = string[content_start:]
         return thinking_content, remaining_content
 
-    # Return if neither format is found
+    # Try seed:think format
+    SEED_START = "<seed:think>"
+    SEED_END = "</seed:think>"
+
+    seed_start_pos = string.find(SEED_START)
+    seed_end_pos = string.find(SEED_END)
+
+    if seed_start_pos != -1 or seed_end_pos != -1:
+        if seed_start_pos == -1:
+            thought_start = 0
+        else:
+            thought_start = seed_start_pos + len(SEED_START)
+
+        if seed_end_pos == -1:
+            thought_end = len(string)
+            content_start = len(string)
+        else:
+            thought_end = seed_end_pos
+            content_start = seed_end_pos + len(SEED_END)
+
+        thinking_content = string[thought_start:thought_end]
+        remaining_content = string[content_start:]
+        return thinking_content, remaining_content
+
+    # Return if no format is found
     return None, string
 
 

From 02ca96fa44e0a29eb52aad46fafd0995e1d91d42 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 25 Aug 2025 22:17:22 -0700
Subject: [PATCH 0958/1701] Multiple fixes

---
 modules/chat.py   | 38 +++++++++++++++++++++++---------------
 modules/shared.py |  4 ++--
 2 files changed, 25 insertions(+), 17 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 818d1014aa..3c61a0dd28 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -292,9 +292,22 @@ def generate_chat_prompt(user_input, state, **kwargs):
         messages.append({"role": "user", "content": "fake user message replace me"})
 
     def make_prompt(messages):
+        last_message = messages[-1].copy()
+        if _continue:
+            if state['mode'] == 'chat-instruct':
+                messages = messages[:-1]
+            else:
+                messages[-1]["content"] = "fake assistant message replace me"
+                messages.append({"role": "assistant", "content": "this will get deleted"})
+
+        if state['mode'] != 'chat-instruct':
+            add_generation_prompt = (not _continue and not impersonate)
+        else:
+            add_generation_prompt = False
+
         prompt = renderer(
-            messages=messages[:-1] if _continue else messages,
-            add_generation_prompt=(state['mode'] != 'chat-instruct' and not impersonate)
+            messages=messages,
+            add_generation_prompt=add_generation_prompt
         )
 
         if state['mode'] == 'chat-instruct':
@@ -308,24 +321,19 @@ def make_prompt(messages):
                 outer_messages.append({"role": "system", "content": state['custom_system_message']})
 
             outer_messages.append({"role": "user", "content": command})
+            if _continue:
+                outer_messages.append(last_message.copy())
+                outer_messages[-1]["content"] = "fake assistant message replace me"
+                outer_messages.append({"role": "assistant", "content": "this will get deleted"})
 
             prompt = instruct_renderer(
                 messages=outer_messages,
-                add_generation_prompt=True
+                add_generation_prompt=not _continue
             )
-        else:
-            # Handle GPT-OSS as a special case when continuing
-            # (otherwise the thinking block gets removed...)
-            if _continue and '<|channel|>final<|message|>' in state['instruction_template_str']:
-                assistant_reply_so_far = "<|start|>assistant"
-                if 'thinking' in messages[-1]:
-                    assistant_reply_so_far += f"<|channel|>analysis<|message|>{messages[-1]['thinking']}<|end|>"
-
-                assistant_reply_so_far += f"<|channel|>final<|message|>"
-                prompt += assistant_reply_so_far
 
         if _continue:
-            prompt += messages[-1].get('content', '')
+            prompt = prompt.split("fake assistant message replace me", 1)[0]
+            prompt += last_message.get("content", "")
 
         if impersonate:
             prompt = prompt.split("fake user message replace me", 1)[0]
@@ -453,7 +461,7 @@ def get_stopping_strings(state):
         renderer = partial(template.render, add_generation_prompt=False)
         renderers.append(renderer)
 
-    if state['mode'] in ['chat', 'chat-instruct']:
+    if state['mode'] in ['chat']:
         template = jinja_env.from_string(state['chat_template_str'])
         renderer = partial(template.render, add_generation_prompt=False, name1=state['name1'], name2=state['name2'])
         renderers.append(renderer)
diff --git a/modules/shared.py b/modules/shared.py
index c3d96b70a3..3e72acca39 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -203,7 +203,7 @@
     'start_with': '',
     'mode': 'instruct',
     'chat_style': 'cai-chat',
-    'chat-instruct_command': 'Continue the chat dialogue below. Write a single reply for the character "<|character|>".\n\n<|prompt|>',
+    'chat-instruct_command': 'Continue the chat dialogue below. Write a single reply for the character "<|character|>". Reply directly, without starting the reply with the character name.\n\n<|prompt|>',
     'enable_web_search': False,
     'web_search_pages': 3,
     'prompt-notebook': '',
@@ -287,7 +287,7 @@
     'greeting': 'How can I help you today?',
     'custom_system_message': '',
     'instruction_template_str': "{%- set ns = namespace(found=false) -%}\n{%- for message in messages -%}\n    {%- if message['role'] == 'system' -%}\n        {%- set ns.found = true -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if not ns.found -%}\n    {{- '' + 'Below is an instruction that describes a task. Write a response that appropriately completes the request.' + '\\n\\n' -}}\n{%- endif %}\n{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {{- '' + message['content'] + '\\n\\n' -}}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{-'### Instruction:\\n' + message['content'] + '\\n\\n'-}}\n        {%- else -%}\n            {{-'### Response:\\n' + message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{-'### Response:\\n'-}}\n{%- endif -%}",
-    'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{- name1 + ': ' + message['content'] + '\\n'-}}\n        {%- else -%}\n            {{- name2 + ': ' + message['content'] + '\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt %}\n    {{- name2 + ': ' -}}\n{%- endif %}",
+    'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{- name1 + ': ' + message['content'] + '\\n'-}}\n        {%- else -%}\n            {{- name2 + ': ' + message['content'] + '\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt %}\n    {{- name2 + ':' -}}\n{%- endif %}",
 
     # Extensions
     'default_extensions': [],

From 750adf793dcf1bc4c5140c84f76b932dc454c194 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 26 Aug 2025 11:58:49 -0700
Subject: [PATCH 0959/1701] UI: Preserve chat scroll position on textarea
 resize

---
 js/main.js | 54 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 54 insertions(+)

diff --git a/js/main.js b/js/main.js
index 4b4b14c2cc..4ada64f62f 100644
--- a/js/main.js
+++ b/js/main.js
@@ -1065,3 +1065,57 @@ document.fonts.addEventListener("loadingdone", (event) => {
     }
   }, 50);
 });
+
+//------------------------------------------------
+// Preserve chat scroll position on textarea resize
+//------------------------------------------------
+(function() {
+  let chatParent = null;
+  let initialState = null;
+  let debounceTimeout = null;
+
+  function getChatParent() {
+    if (!chatParent) chatParent = document.querySelector(".chat-parent");
+    return chatParent;
+  }
+
+  function getTextarea() {
+    return document.querySelector("#chat-input textarea");
+  }
+
+  document.addEventListener("input", function(e) {
+    if (e.target.matches("#chat-input textarea")) {
+      const chat = getChatParent();
+      const textarea = getTextarea();
+
+      if (chat && textarea) {
+        // Capture initial state only on first input of a typing sequence
+        if (!initialState) {
+          initialState = {
+            scrollTop: chat.scrollTop,
+            textareaHeight: textarea.offsetHeight
+          };
+        }
+
+        // Clear existing timeout
+        clearTimeout(debounceTimeout);
+
+        // Wait for typing to stop (50ms delay)
+        debounceTimeout = setTimeout(() => {
+          const finalTextareaHeight = textarea.offsetHeight;
+          const totalGrowth = finalTextareaHeight - initialState.textareaHeight;
+          const targetScroll = initialState.scrollTop + totalGrowth;
+
+          const restore = () => { chat.scrollTop = targetScroll; };
+
+          restore();
+          requestAnimationFrame(restore);
+          setTimeout(restore, 0);
+          setTimeout(restore, 10);
+
+          initialState = null;
+        }, 50);
+      }
+    }
+  }, true);
+})();

From ccc8a2229dd82ae8d77274341785e043f6f3a343 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 26 Aug 2025 13:59:54 -0700
Subject: [PATCH 0960/1701] Revert "UI: Preserve chat scroll position on
 textarea resize"

This reverts commit 750adf793dcf1bc4c5140c84f76b932dc454c194.
---
 js/main.js | 54 ------------------------------------------------------
 1 file changed, 54 deletions(-)

diff --git a/js/main.js b/js/main.js
index 4ada64f62f..4b4b14c2cc 100644
--- a/js/main.js
+++ b/js/main.js
@@ -1065,57 +1065,3 @@ document.fonts.addEventListener("loadingdone", (event) => {
     }
   }, 50);
 });
-
-//------------------------------------------------
-// Preserve chat scroll position on textarea resize
-//------------------------------------------------
-(function() {
-  let chatParent = null;
-  let initialState = null;
-  let debounceTimeout = null;
-
-  function getChatParent() {
-    if (!chatParent) chatParent = document.querySelector(".chat-parent");
-    return chatParent;
-  }
-
-  function getTextarea() {
-    return document.querySelector("#chat-input textarea");
-  }
-
-  document.addEventListener("input", function(e) {
-    if (e.target.matches("#chat-input textarea")) {
-      const chat = getChatParent();
-      const textarea = getTextarea();
-
-      if (chat && textarea) {
-        // Capture initial state only on first input of a typing sequence
-        if (!initialState) {
-          initialState = {
-            scrollTop: chat.scrollTop,
-            textareaHeight: textarea.offsetHeight
-          };
-        }
-
-        // Clear existing timeout
-        clearTimeout(debounceTimeout);
-
-        // Wait for typing to stop (50ms delay)
-        debounceTimeout = setTimeout(() => {
-          const finalTextareaHeight = textarea.offsetHeight;
-          const totalGrowth = finalTextareaHeight - initialState.textareaHeight;
-          const targetScroll = initialState.scrollTop + totalGrowth;
-
-          const restore = () => { chat.scrollTop = targetScroll; };
-
-          restore();
-          requestAnimationFrame(restore);
-          setTimeout(restore, 0);
-          setTimeout(restore, 10);
-
-          initialState = null;
-        }, 50);
-      }
-    }
-  }, true);
-})();

From 8042f76399f9aa84c4d16dd10fb9ddf3d01238a4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 27 Aug 2025 05:37:01 -0700
Subject: [PATCH 0961/1701] Make portable installs functional with Python 3.13

---
 requirements/full/requirements.txt                     | 1 +
 requirements/full/requirements_amd.txt                 | 1 +
 requirements/full/requirements_amd_noavx2.txt          | 1 +
 requirements/full/requirements_apple_intel.txt         | 1 +
 requirements/full/requirements_apple_silicon.txt       | 1 +
 requirements/full/requirements_cpu_only.txt            | 1 +
 requirements/full/requirements_cpu_only_noavx2.txt     | 1 +
 requirements/full/requirements_noavx2.txt              | 1 +
 requirements/full/requirements_nowheels.txt            | 1 +
 requirements/portable/requirements.txt                 | 1 +
 requirements/portable/requirements_apple_intel.txt     | 1 +
 requirements/portable/requirements_apple_silicon.txt   | 1 +
 requirements/portable/requirements_cpu_only.txt        | 1 +
 requirements/portable/requirements_cpu_only_noavx2.txt | 1 +
 requirements/portable/requirements_noavx2.txt          | 1 +
 requirements/portable/requirements_nowheels.txt        | 1 +
 requirements/portable/requirements_vulkan.txt          | 1 +
 requirements/portable/requirements_vulkan_noavx2.txt   | 1 +
 18 files changed, 18 insertions(+)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index c53e772229..77ddc8fb61 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -1,4 +1,5 @@
 accelerate==1.8.*
+audioop-lts<1.0; python_version >= "3.13"
 bitsandbytes==0.46.*
 colorama
 datasets
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 776dbc7bd4..802f6724c8 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -1,4 +1,5 @@
 accelerate==1.8.*
+audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 7205a72e7e..bbb4fa59d0 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -1,4 +1,5 @@
 accelerate==1.8.*
+audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index f4e00c5b3a..b721bcce63 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -1,4 +1,5 @@
 accelerate==1.8.*
+audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 0b28ad216d..80b168d2da 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -1,4 +1,5 @@
 accelerate==1.8.*
+audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index fcbb48f0be..5bfcdea6d4 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -1,4 +1,5 @@
 accelerate==1.8.*
+audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 1705791ebd..31743a21c4 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -1,4 +1,5 @@
 accelerate==1.8.*
+audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 7e95121968..0d04d22989 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -1,4 +1,5 @@
 accelerate==1.8.*
+audioop-lts<1.0; python_version >= "3.13"
 bitsandbytes==0.46.*
 colorama
 datasets
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index cd85a744e7..74d8604708 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -1,4 +1,5 @@
 accelerate==1.8.*
+audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
 einops
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 9eb8b33c0c..ca0c4017c8 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -1,3 +1,4 @@
+audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 0a27f61b4f..b5a853bad1 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -1,3 +1,4 @@
+audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 79674e5a92..995f5f268c 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -1,3 +1,4 @@
+audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 3ebb7d8bd6..3b5d94424f 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -1,3 +1,4 @@
+audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 81b78fe6e4..4bc705ec98 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -1,3 +1,4 @@
+audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 58dc529a6a..a4dc4de97b 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -1,3 +1,4 @@
+audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index b7b73eff0e..be624bb1e5 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -1,3 +1,4 @@
+audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 5ad8ede10c..4367e18018 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -1,3 +1,4 @@
+audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 0adf6e48df..2130efcceb 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -1,3 +1,4 @@
+audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 gradio==4.37.*
 html2text==2025.4.15

From 73442a2b6d0f2de333c26cbdde862f3f7b84d8a8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 27 Aug 2025 05:43:13 -0700
Subject: [PATCH 0962/1701] UI: Better handle the chat input position with CSS

This also solves scrolling issues with the main chat content
when the height of the textarea increases.
---
 css/chat_style-messenger.css |  2 ++
 css/main.css                 |  4 ++++
 js/main.js                   | 27 +++++++++++++++++++++++++++
 3 files changed, 33 insertions(+)

diff --git a/css/chat_style-messenger.css b/css/chat_style-messenger.css
index 583703c0db..70fd6d4a28 100644
--- a/css/chat_style-messenger.css
+++ b/css/chat_style-messenger.css
@@ -99,9 +99,11 @@
 .message-body p em {
     color: rgb(110 110 110) !important;
 }
+
 .editing-textarea {
     width: max(30rem) !important;
 }
+
 .circle-you + .text .edit-control-button, .circle-you + .text .editing-textarea {
     color: #000 !important;
 }
diff --git a/css/main.css b/css/main.css
index 062d3eb215..b799f59587 100644
--- a/css/main.css
+++ b/css/main.css
@@ -404,6 +404,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     flex: 1;
     overflow: auto !important;
     border-radius: 0 !important;
+    margin-bottom: 75px;
 }
 
 .chat-parent .prose {
@@ -626,6 +627,9 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     max-width: 54rem;
     left: 50%;
     transform: translateX(-50%);
+    position: absolute;
+    bottom: 0;
+    background: var(--body-background-fill);
 }
 
 @media print {
diff --git a/js/main.js b/js/main.js
index 4b4b14c2cc..9b9d685a68 100644
--- a/js/main.js
+++ b/js/main.js
@@ -1065,3 +1065,30 @@ document.fonts.addEventListener("loadingdone", (event) => {
     }
   }, 50);
 });
+
+(function() {
+  const chatParent = document.querySelector(".chat-parent");
+  const chatInputRow = document.querySelector("#chat-input-row");
+  const originalMarginBottom = 75;
+  let originalHeight = chatInputRow.offsetHeight;
+
+  function updateMargin() {
+    const currentHeight = chatInputRow.offsetHeight;
+    const heightDifference = currentHeight - originalHeight;
+    chatParent.style.marginBottom = `${originalMarginBottom + heightDifference}px`;
+  }
+
+  // Watch for changes that might affect height
+  const observer = new MutationObserver(updateMargin);
+  observer.observe(chatInputRow, {
+    childList: true,
+    subtree: true,
+    attributes: true
+  });
+
+  // Also listen for window resize
+  window.addEventListener("resize", updateMargin);
+
+  // Initial call to set the margin based on current state
+  updateMargin();
+})();

From 0b4518e61cfe7993017c54d09328dd364301128f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 27 Aug 2025 05:53:09 -0700
Subject: [PATCH 0963/1701] "Text generation web UI" -> "Text Generation Web
 UI"

---
 README.md         | 4 ++--
 modules/shared.py | 2 +-
 server.py         | 6 +++---
 3 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/README.md b/README.md
index 6b49cee06a..d42697dd54 100644
--- a/README.md
+++ b/README.md
@@ -1,4 +1,4 @@
-# Text generation web UI
+# Text Generation Web UI
 
 A Gradio web UI for Large Language Models.
 
@@ -238,7 +238,7 @@ usage: server.py [-h] [--multi-user] [--model MODEL] [--lora LORA [LORA ...]] [-
                  [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors] [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY]
                  [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4] [--nowebui]
 
-Text generation web UI
+Text Generation Web UI
 
 options:
   -h, --help                                show this help message and exit
diff --git a/modules/shared.py b/modules/shared.py
index 3e72acca39..a308523957 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -31,7 +31,7 @@
 need_restart = False
 
 # Parser copied from https://github.com/vladmandic/automatic
-parser = argparse.ArgumentParser(description="Text generation web UI", conflict_handler='resolve', add_help=True, formatter_class=lambda prog: argparse.HelpFormatter(prog, max_help_position=55, indent_increment=2, width=200))
+parser = argparse.ArgumentParser(description="Text Generation Web UI", conflict_handler='resolve', add_help=True, formatter_class=lambda prog: argparse.HelpFormatter(prog, max_help_position=55, indent_increment=2, width=200))
 
 # Basic settings
 group = parser.add_argument_group('Basic settings')
diff --git a/server.py b/server.py
index e6687a3cb6..52463a3c39 100644
--- a/server.py
+++ b/server.py
@@ -70,7 +70,7 @@
 
 
 def signal_handler(sig, frame):
-    logger.info("Received Ctrl+C. Shutting down Text generation web UI gracefully.")
+    logger.info("Received Ctrl+C. Shutting down Text Generation Web UI gracefully.")
 
     # Explicitly stop LlamaServer to avoid __del__ cleanup issues during shutdown
     if shared.model and shared.model.__class__.__name__ == 'LlamaServer':
@@ -87,7 +87,7 @@ def signal_handler(sig, frame):
 
 def create_interface():
 
-    title = 'Text generation web UI'
+    title = 'Text Generation Web UI'
 
     # Password authentication
     auth = []
@@ -230,7 +230,7 @@ def create_interface():
 
 if __name__ == "__main__":
 
-    logger.info("Starting Text generation web UI")
+    logger.info("Starting Text Generation Web UI")
     do_cmd_flags_warnings()
 
     # Load custom settings

From 030ba7bfeb0e7aed7c8a176e13cc64cd75489d23 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 27 Aug 2025 07:44:35 -0700
Subject: [PATCH 0964/1701] UI: Mention that Seed-OSS uses enable_thinking

---
 modules/ui_chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 94c980bb3f..1dbac13b79 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -81,7 +81,7 @@ def create_ui():
                 gr.HTML("
            ")
 
                 shared.gradio['reasoning_effort'] = gr.Dropdown(value=shared.settings['reasoning_effort'], choices=['low', 'medium', 'high'], label='Reasoning effort', info='Used by GPT-OSS.')
-                shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', info='Used by pre-2507 Qwen3.')
+                shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', info='Used by Seed-OSS, pre-2507 Qwen3.')
 
                 gr.HTML("
            ")
 

From a92758a1444626167468f0b0552a642b1e9245a2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 27 Aug 2025 16:15:20 -0700
Subject: [PATCH 0965/1701] llama.cpp: Fix obtaining the maximum sequence
 length for GPT-OSS

---
 modules/models_settings.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 7645880ff8..6dc000b424 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -68,7 +68,7 @@ def get_model_metadata(model):
         metadata = load_gguf_metadata_with_cache(model_file)
 
         for k in metadata:
-            if k.endswith('context_length'):
+            if k.endswith('.context_length'):
                 model_settings['ctx_size'] = min(metadata[k], 8192)
                 model_settings['truncation_length_info'] = metadata[k]
             elif k.endswith('rope.freq_base'):

From ba6041251d200dfffaf6ea46dd492554a254b241 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 28 Aug 2025 06:20:00 -0700
Subject: [PATCH 0966/1701] UI: Minor change

---
 modules/ui_chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 1dbac13b79..31a7a4fcc0 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -81,7 +81,7 @@ def create_ui():
                 gr.HTML("
            ")
 
                 shared.gradio['reasoning_effort'] = gr.Dropdown(value=shared.settings['reasoning_effort'], choices=['low', 'medium', 'high'], label='Reasoning effort', info='Used by GPT-OSS.')
-                shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', info='Used by Seed-OSS, pre-2507 Qwen3.')
+                shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', info='Used by Seed-OSS and pre-2507 Qwen3.')
 
                 gr.HTML("
            ")
 

From a336a8bbeb53136c40040be8d7e18e79eec034df Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 28 Aug 2025 08:26:40 -0700
Subject: [PATCH 0967/1701] UI: Fix italic and quote color in headings

---
 css/html_instruct_style.css | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index 3e5ebe67c8..22901c4d11 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -13,7 +13,9 @@
     line-height: 28px !important;
 }
 
-.dark .chat .message-body :is(p, li, q, em, h1, h2, h3, h4, h5, h6) {
+.dark .chat .message-body :is(p,li,h1,h2,h3,h4,h5,h6),
+.dark .chat .message-body em:not(:is(h1,h2,h3,h4,h5,h6) em),
+.dark .chat .message-body q:not(:is(h1,h2,h3,h4,h5,h6) q) {
     color: #d1d5db !important;
 }
 

From cfc83745ec96ad963282620524f94b08776de5b6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 28 Aug 2025 08:34:48 -0700
Subject: [PATCH 0968/1701] UI: Improve right sidebar borders in light mode

---
 css/main.css       | 9 +++++++++
 modules/ui_chat.py | 8 ++++----
 2 files changed, 13 insertions(+), 4 deletions(-)

diff --git a/css/main.css b/css/main.css
index b799f59587..a7ed253480 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1697,3 +1697,12 @@ button:focus {
 #chat-input span {
     display: none;
 }
+
+.sidebar-vertical-separator {
+    margin: 0;
+    border-bottom: var(--input-border-width) solid var(--input-border-color);
+}
+
+.dark .sidebar-vertical-separator {
+    border-bottom: 1px solid rgba(255,255,255,0.1);
+}
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 31a7a4fcc0..1e8218a950 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -78,18 +78,18 @@ def create_ui():
                 with gr.Row():
                     shared.gradio['start_with'] = gr.Textbox(label='Start reply with', placeholder='Sure thing!', value=shared.settings['start_with'], elem_classes=['add_scrollbar'])
 
-                gr.HTML("
            ")
+                gr.HTML("
            ")
 
                 shared.gradio['reasoning_effort'] = gr.Dropdown(value=shared.settings['reasoning_effort'], choices=['low', 'medium', 'high'], label='Reasoning effort', info='Used by GPT-OSS.')
                 shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', info='Used by Seed-OSS and pre-2507 Qwen3.')
 
-                gr.HTML("
            ")
+                gr.HTML("
            ")
 
                 shared.gradio['enable_web_search'] = gr.Checkbox(value=shared.settings.get('enable_web_search', False), label='Activate web search', elem_id='web-search')
                 with gr.Row(visible=shared.settings.get('enable_web_search', False)) as shared.gradio['web_search_row']:
                     shared.gradio['web_search_pages'] = gr.Number(value=shared.settings.get('web_search_pages', 3), precision=0, label='Number of pages to download', minimum=1, maximum=10)
 
-                gr.HTML("
            ")
+                gr.HTML("
            ")
 
                 with gr.Row():
                     shared.gradio['mode'] = gr.Radio(choices=['instruct', 'chat-instruct', 'chat'], value=None, label='Mode', info='Defines how the chat prompt is generated. In instruct and chat-instruct modes, the instruction template Parameters > Instruction template is used.', elem_id='chat-mode')
@@ -100,7 +100,7 @@ def create_ui():
                 with gr.Row():
                     shared.gradio['chat-instruct_command'] = gr.Textbox(value=shared.settings['chat-instruct_command'], lines=12, label='Command for chat-instruct mode', info='<|character|> and <|prompt|> get replaced with the bot name and the regular chat prompt respectively.', visible=shared.settings['mode'] == 'chat-instruct', elem_classes=['add_scrollbar'])
 
-                gr.HTML("
            ")
+                gr.HTML("
            ")
 
                 with gr.Row():
                     shared.gradio['count_tokens'] = gr.Button('Count tokens', size='sm')

From cb8780a4ce617b9a53d0ffb535f6b18b82b5f3bf Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 28 Aug 2025 11:13:19 -0700
Subject: [PATCH 0969/1701] Safer check for is_multimodal when loading models

Avoids unrelated multimodal error when a model fails to load due
to lack of memory.
---
 modules/models.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/models.py b/modules/models.py
index cae88ac598..133131d7f1 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -57,7 +57,7 @@ def load_model(model_name, loader=None):
         shared.settings['truncation_length'] = shared.args.ctx_size
 
     shared.is_multimodal = False
-    if loader.lower() in ('exllamav3', 'llama.cpp'):
+    if loader.lower() in ('exllamav3', 'llama.cpp') and hasattr(model, 'is_multimodal'):
         shared.is_multimodal = model.is_multimodal()
 
     logger.info(f"Loaded \"{model_name}\" in {(time.time()-t0):.2f} seconds.")

From d9eec31886246d5501b6502457c917fb46e9e748 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 28 Aug 2025 17:46:29 -0700
Subject: [PATCH 0970/1701] UI: Suppress "Attempted to select a non-interactive
 or hidden tab" warnings

---
 js/global_scope_js.js | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index ebed1f3dcd..89b51d67d2 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -372,3 +372,18 @@ observer.observe(document.documentElement, {
   subtree: true,
   attributeFilter: ["style"]
 });
+
+//------------------------------------------------
+// Suppress "Attempted to select a non-interactive or hidden tab" warning
+//------------------------------------------------
+(function() {
+  const originalWarn = console.warn;
+
+  console.warn = function(...args) {
+    if (args[0] && args[0].includes("Attempted to select a non-interactive or hidden tab")) {
+      return;
+    }
+
+    originalWarn.apply(console, args);
+  };
+})();

From 272095547845a69a725c4564d63f40e50f47e563 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 28 Aug 2025 19:48:16 -0700
Subject: [PATCH 0971/1701] Fix a bug after
 d9eec31886246d5501b6502457c917fb46e9e748

---
 js/global_scope_js.js | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 89b51d67d2..d8de2b58ba 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -380,7 +380,7 @@ observer.observe(document.documentElement, {
   const originalWarn = console.warn;
 
   console.warn = function(...args) {
-    if (args[0] && args[0].includes("Attempted to select a non-interactive or hidden tab")) {
+    if (args[0] && typeof args[0] === 'string' && args[0].includes("Attempted to select a non-interactive or hidden tab")) {
       return;
     }
 

From fc2eb48664bc8e29034904cac43e6c0bc89aa727 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 28 Aug 2025 20:19:03 -0700
Subject: [PATCH 0972/1701] Style fixes after
 73442a2b6d0f2de333c26cbdde862f3f7b84d8a8

---
 css/main.css        | 12 ------------
 js/show_controls.js | 12 ------------
 2 files changed, 24 deletions(-)

diff --git a/css/main.css b/css/main.css
index a7ed253480..cde01aa4aa 100644
--- a/css/main.css
+++ b/css/main.css
@@ -429,10 +429,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     margin-left: 5px;
 }
 
-.chat-parent.bigchat {
-    flex: 1;
-}
-
 .chat > .messages {
     display: flex;
     flex-direction: column;
@@ -832,10 +828,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     padding: 1rem;
 }
 
-#chat-input-row.bigchat {
-    padding-bottom: 1px !important;
-}
-
 #chat-col {
     height: 100dvh;
     display: flex;
@@ -851,10 +843,6 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     }
 }
 
-#chat-col.bigchat {
-    padding-bottom: 15px !important;
-}
-
 .message-body ol, .message-body ul {
     margin-top: 0 !important;
     margin-bottom: 1.25em !important;
diff --git a/js/show_controls.js b/js/show_controls.js
index f974d412fa..ff513395ee 100644
--- a/js/show_controls.js
+++ b/js/show_controls.js
@@ -20,12 +20,6 @@ function toggle_controls(value) {
       extensions.style.display = "inherit";
     }
 
-    // Remove bigchat classes
-    chatParent.classList.remove("bigchat");
-    document.getElementById("chat-input-row").classList.remove("bigchat");
-    document.getElementById("chat-col").classList.remove("bigchat");
-    document.getElementById("chat-tab").style.paddingBottom = "";
-
     let gallery_element = document.getElementById("gallery-extension");
     if (gallery_element) {
       gallery_element.style.display = "block";
@@ -47,11 +41,5 @@ function toggle_controls(value) {
     if (extensions) {
       extensions.style.display = "none";
     }
-
-    // Add bigchat classes
-    chatParent.classList.add("bigchat");
-    document.getElementById("chat-input-row").classList.add("bigchat");
-    document.getElementById("chat-col").classList.add("bigchat");
-    document.getElementById("chat-tab").style.paddingBottom = "0px";
   }
 }

From d78b7d0fad31c6b9bf89a89f748e1b00c27c5946 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 28 Aug 2025 20:22:07 -0700
Subject: [PATCH 0973/1701] Lint

---
 js/global_scope_js.js | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index d8de2b58ba..4d8c11215c 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -380,7 +380,7 @@ observer.observe(document.documentElement, {
   const originalWarn = console.warn;
 
   console.warn = function(...args) {
-    if (args[0] && typeof args[0] === 'string' && args[0].includes("Attempted to select a non-interactive or hidden tab")) {
+    if (args[0] && typeof args[0] === "string" && args[0].includes("Attempted to select a non-interactive or hidden tab")) {
       return;
     }
 

From 084675cf75d08ce2d82fe440abbaf52e429eed3a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 29 Aug 2025 09:11:10 -0700
Subject: [PATCH 0974/1701] UI: Improve thinking blocks in chat-instruct mode

---
 css/main.css | 1 +
 1 file changed, 1 insertion(+)

diff --git a/css/main.css b/css/main.css
index cde01aa4aa..f279337291 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1354,6 +1354,7 @@ div.svelte-362y77>*, div.svelte-362y77>.form>* {
     cursor: pointer;
     user-select: none;
     font-size: 14px;
+    line-height: var(--line-sm);
     color: rgb(0 0 0 / 70%);
     transition: background-color 0.2s;
 }

From a2b37adb265847c878d71107fca988851090b46f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 29 Aug 2025 09:25:44 -0700
Subject: [PATCH 0975/1701] UI: Preload the correct fonts for chat mode

---
 modules/block_requests.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/modules/block_requests.py b/modules/block_requests.py
index 618b4bd682..911e41d9a0 100644
--- a/modules/block_requests.py
+++ b/modules/block_requests.py
@@ -53,8 +53,9 @@ def my_open(*args, **kwargs):
             '',
             '\n    '
             '\n    '
-            '\n    '
-            '\n    '
+            '\n    '
+            '\n    '
+            '\n    '
             '\n    '
             '\n    '
             '\n    '

From 07a2e226c165fd5917c778f18c0f0fd4bcef38b7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 29 Aug 2025 14:08:38 -0700
Subject: [PATCH 0976/1701] UI: Minor font color fixes in instruct mode

---
 css/html_instruct_style.css | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index 22901c4d11..6dee0a8980 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -14,8 +14,8 @@
 }
 
 .dark .chat .message-body :is(p,li,h1,h2,h3,h4,h5,h6),
-.dark .chat .message-body em:not(:is(h1,h2,h3,h4,h5,h6) em),
-.dark .chat .message-body q:not(:is(h1,h2,h3,h4,h5,h6) q) {
+.dark .chat .message-body em:not(:is(h1,h2,h3,h4,h5,h6,b,strong) em),
+.dark .chat .message-body q:not(:is(h1,h2,h3,h4,h5,h6,b,strong) q) {
     color: #d1d5db !important;
 }
 

From 08f90f4b64565424f812e9a0447338c022d883f2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 29 Aug 2025 14:09:04 -0700
Subject: [PATCH 0977/1701] Lint

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index f279337291..c7ee57da0c 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1693,5 +1693,5 @@ button:focus {
 }
 
 .dark .sidebar-vertical-separator {
-    border-bottom: 1px solid rgba(255,255,255,0.1);
+    border-bottom: 1px solid rgb(255 255 255 / 10%);
 }

From a3eb67e466f9cc23a6c3607842375e8bf50f2dd0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 30 Aug 2025 08:42:26 -0700
Subject: [PATCH 0978/1701] Fix the UI failing to launch if the Notebook prompt
 is too long

---
 modules/prompts.py     | 3 +--
 modules/ui_default.py  | 3 +--
 modules/ui_notebook.py | 3 +--
 server.py              | 8 ++++++++
 4 files changed, 11 insertions(+), 6 deletions(-)

diff --git a/modules/prompts.py b/modules/prompts.py
index 79d9b56e46..b800af91a3 100644
--- a/modules/prompts.py
+++ b/modules/prompts.py
@@ -22,8 +22,7 @@ def load_prompt(fname):
     if file_path.exists():
         with open(file_path, 'r', encoding='utf-8') as f:
             text = f.read()
-            if len(text) > 0 and text[-1] == '\n':
-                text = text[:-1]
+            text = text.rstrip()
 
             return text
     else:
diff --git a/modules/ui_default.py b/modules/ui_default.py
index 44af48a306..c0feae1947 100644
--- a/modules/ui_default.py
+++ b/modules/ui_default.py
@@ -22,8 +22,7 @@ def create_ui():
         with gr.Row():
             with gr.Column():
                 with gr.Row():
-                    initial_text = load_prompt(shared.settings['prompt-notebook'])
-                    shared.gradio['textbox-default'] = gr.Textbox(value=initial_text, lines=27, label='Input', elem_classes=['textbox_default', 'add_scrollbar'])
+                    shared.gradio['textbox-default'] = gr.Textbox(value="", lines=27, label='Input', elem_classes=['textbox_default', 'add_scrollbar'])
                     shared.gradio['token-counter-default'] = gr.HTML(value="0", elem_id="default-token-counter")
 
                 with gr.Row():
diff --git a/modules/ui_notebook.py b/modules/ui_notebook.py
index 939d81f70b..9fab879b82 100644
--- a/modules/ui_notebook.py
+++ b/modules/ui_notebook.py
@@ -30,8 +30,7 @@ def create_ui():
             with gr.Column(scale=4):
                 with gr.Tab('Raw'):
                     with gr.Row():
-                        initial_text = load_prompt(shared.settings['prompt-notebook'])
-                        shared.gradio['textbox-notebook'] = gr.Textbox(label="", value=initial_text, lines=27, elem_id='textbox-notebook', elem_classes=['textbox', 'add_scrollbar'])
+                        shared.gradio['textbox-notebook'] = gr.Textbox(label="", value="", lines=27, elem_id='textbox-notebook', elem_classes=['textbox', 'add_scrollbar'])
                         shared.gradio['token-counter-notebook'] = gr.HTML(value="0", elem_id="notebook-token-counter")
 
                 with gr.Tab('Markdown'):
diff --git a/server.py b/server.py
index 52463a3c39..c804c34232 100644
--- a/server.py
+++ b/server.py
@@ -6,6 +6,7 @@
 from modules import shared
 from modules.block_requests import OpenMonkeyPatch, RequestBlocker
 from modules.logging_colors import logger
+from modules.prompts import load_prompt
 
 # Set up Gradio temp directory path
 gradio_temp_path = Path('user_data') / 'cache' / 'gradio'
@@ -109,6 +110,13 @@ def create_interface():
         'filter_by_loader': (shared.args.loader or 'All') if not shared.args.portable else 'llama.cpp'
     })
 
+    if shared.settings['prompt-notebook']:
+        prompt = load_prompt(shared.settings['prompt-notebook'])
+        shared.persistent_interface_state.update({
+            'textbox-default': prompt,
+            'textbox-notebook': prompt
+        })
+
     # Clear existing cache files
     for cache_file in ['pfp_character.png', 'pfp_character_thumb.png']:
         cache_path = Path(f"user_data/cache/{cache_file}")

From 96136ea76008cf1fb440050936b6b5bca5bd3d85 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 30 Aug 2025 10:13:32 -0700
Subject: [PATCH 0979/1701] Fix LaTeX rendering for equations with asterisks

---
 modules/html_generator.py | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 63844f357d..9f8c28e53a 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -243,6 +243,27 @@ def process_markdown_content(string):
     if not string:
         return ""
 
+    # Define a unique placeholder for LaTeX asterisks
+    LATEX_ASTERISK_PLACEHOLDER = "LATEXASTERISKPLACEHOLDER"
+
+    def protect_asterisks_in_latex(match):
+        """A replacer function for re.sub to protect asterisks in multiple LaTeX formats."""
+        # Check which delimiter group was captured
+        if match.group(1) is not None:  # Content from $$...$$
+            content = match.group(1)
+            modified_content = content.replace('*', LATEX_ASTERISK_PLACEHOLDER)
+            return f'$${modified_content}$$'
+        elif match.group(2) is not None:  # Content from \[...\]
+            content = match.group(2)
+            modified_content = content.replace('*', LATEX_ASTERISK_PLACEHOLDER)
+            return f'\\[{modified_content}\\]'
+        elif match.group(3) is not None:  # Content from \(...\)
+            content = match.group(3)
+            modified_content = content.replace('*', LATEX_ASTERISK_PLACEHOLDER)
+            return f'\\({modified_content}\\)'
+
+        return match.group(0) # Fallback
+
     # Make \[ \]  LaTeX equations inline
     pattern = r'^\s*\\\[\s*\n([\s\S]*?)\n\s*\\\]\s*$'
     replacement = r'\\[ \1 \\]'
@@ -272,6 +293,10 @@ def process_markdown_content(string):
     string = string.replace('\\end{equation*}', '$$')
     string = re.sub(r"(.)```", r"\1\n```", string)
 
+    # Protect asterisks within all LaTeX blocks before markdown conversion
+    latex_pattern = re.compile(r'\$\$(.*?)\$\$|\\\[(.*?)\\\]|\\\((.*?)\\\)', re.DOTALL)
+    string = latex_pattern.sub(protect_asterisks_in_latex, string)
+
     result = ''
     is_code = False
     is_latex = False
@@ -330,6 +355,9 @@ def process_markdown_content(string):
         # Convert to HTML using markdown
         html_output = markdown.markdown(result, extensions=['fenced_code', 'tables', SaneListExtension()])
 
+    # Restore the LaTeX asterisks after markdown conversion
+    html_output = html_output.replace(LATEX_ASTERISK_PLACEHOLDER, '*')
+
     # Remove extra newlines before 
     html_output = re.sub(r'\s*', '', html_output)
 

From cf1aad2a687622358e121497c30508e960267662 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 30 Aug 2025 12:16:45 -0700
Subject: [PATCH 0980/1701] Fix "continue" for Byte-OSS for partial thinking
 blocks

---
 modules/chat.py | 18 +++++++++++++++++-
 1 file changed, 17 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 3c61a0dd28..6d85bc6ee1 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -333,7 +333,23 @@ def make_prompt(messages):
 
         if _continue:
             prompt = prompt.split("fake assistant message replace me", 1)[0]
-            prompt += last_message.get("content", "")
+            content = last_message.get("content", "")
+            thinking = last_message.get("thinking", "")
+            reasoning = last_message.get("reasoning_content", "")
+
+            partial_thought = thinking or reasoning
+            # Handle partial thinking blocks (GPT-OSS and Seed-OSS)
+            if partial_thought and partial_thought.strip():
+                search_string = partial_thought.strip()
+                index = prompt.rfind(search_string)
+                if index != -1:
+                    prompt = prompt[:index] + partial_thought
+                else:
+                    # Fallback
+                    prompt += content
+            else:
+                # All other cases
+                prompt += content
 
         if impersonate:
             prompt = prompt.split("fake user message replace me", 1)[0]

From 3a3e247f3cb21ec7c13bcfa9f21757e216d5c7ec Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 30 Aug 2025 12:36:35 -0700
Subject: [PATCH 0981/1701] Even better way to handle continue for thinking
 blocks

---
 modules/chat.py | 11 +++++------
 1 file changed, 5 insertions(+), 6 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 6d85bc6ee1..ad2f4001e1 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -333,20 +333,19 @@ def make_prompt(messages):
 
         if _continue:
             prompt = prompt.split("fake assistant message replace me", 1)[0]
+
             content = last_message.get("content", "")
-            thinking = last_message.get("thinking", "")
-            reasoning = last_message.get("reasoning_content", "")
+            partial_thought = last_message.get("thinking", "") or last_message.get("reasoning_content", "")
 
-            partial_thought = thinking or reasoning
             # Handle partial thinking blocks (GPT-OSS and Seed-OSS)
-            if partial_thought and partial_thought.strip():
+            if not content and partial_thought and partial_thought.strip():
                 search_string = partial_thought.strip()
                 index = prompt.rfind(search_string)
                 if index != -1:
                     prompt = prompt[:index] + partial_thought
                 else:
-                    # Fallback
-                    prompt += content
+                    # Fallback if search fails: just append the thought
+                    prompt += partial_thought
             else:
                 # All other cases
                 prompt += content

From 21d790f87ea21fc1f7f9f938621d805b21564493 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 30 Aug 2025 14:48:07 -0700
Subject: [PATCH 0982/1701] Optimize LaTeX rendering during streaming for long
 replies

---
 js/main.js | 20 +++++++++++++-------
 1 file changed, 13 insertions(+), 7 deletions(-)

diff --git a/js/main.js b/js/main.js
index 9b9d685a68..c9ee628465 100644
--- a/js/main.js
+++ b/js/main.js
@@ -260,13 +260,19 @@ function doSyntaxHighlighting() {
           codeBlock.classList.add("pretty_scrollbar");
         });
 
-        renderMathInElement(messageBody, {
-          delimiters: [
-            { left: "$$", right: "$$", display: true },
-            { left: "$", right: "$", display: false },
-            { left: "\\(", right: "\\)", display: false },
-            { left: "\\[", right: "\\]", display: true },
-          ],
+        // Only render math in visible elements
+        const mathContainers = messageBody.querySelectorAll("p, div, span, li, td, th, h1, h2, h3, h4, h5, h6, blockquote, figcaption, caption, dd, dt");
+        mathContainers.forEach(container => {
+          if (isElementVisibleOnScreen(container)) {
+            renderMathInElement(container, {
+              delimiters: [
+                { left: "$$", right: "$$", display: true },
+                { left: "$", right: "$", display: false },
+                { left: "\\(", right: "\\)", display: false },
+                { left: "\\[", right: "\\]", display: true },
+              ],
+            });
+          }
         });
       } else if (hasSeenVisible) {
         // We've seen visible messages but this one is not visible

From 5920ad8834dd8d3077376636c8347e7547bb6a04 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 30 Aug 2025 15:22:50 -0700
Subject: [PATCH 0983/1701] UI: Give streaming instruct messages more vertical
 space

---
 js/main.js | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index c9ee628465..7246ca874b 100644
--- a/js/main.js
+++ b/js/main.js
@@ -206,7 +206,7 @@ const observer = new MutationObserver(function(mutations) {
       // Add padding to the messages container to create room for the last message.
       // The purpose of this is to avoid constant scrolling during streaming in
       // instruct mode.
-      const bufferHeight = Math.max(0, Math.max(0.7 * window.innerHeight, window.innerHeight - prevSibling.offsetHeight - 84) - lastChild.offsetHeight);
+      const bufferHeight = Math.max(0, Math.max(window.innerHeight - 128 - 84, window.innerHeight - prevSibling.offsetHeight - 84) - lastChild.offsetHeight);
       messagesContainer.style.paddingBottom = `${bufferHeight}px`;
     }
   }

From 5631d4e3d69a7d3e77899efcd71b8e4860cfa346 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 30 Aug 2025 15:34:49 -0700
Subject: [PATCH 0984/1701] Minor change after
 21d790f87ea21fc1f7f9f938621d805b21564493

---
 js/main.js | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index 7246ca874b..c08dffcfb3 100644
--- a/js/main.js
+++ b/js/main.js
@@ -261,7 +261,7 @@ function doSyntaxHighlighting() {
         });
 
         // Only render math in visible elements
-        const mathContainers = messageBody.querySelectorAll("p, div, span, li, td, th, h1, h2, h3, h4, h5, h6, blockquote, figcaption, caption, dd, dt");
+        const mathContainers = messageBody.querySelectorAll("p, span, li, td, th, h1, h2, h3, h4, h5, h6, blockquote, figcaption, caption, dd, dt");
         mathContainers.forEach(container => {
           if (isElementVisibleOnScreen(container)) {
             renderMathInElement(container, {

From 7b80e9a2ad0cb12ac732bd2bcf7bc0bd1cb3a0e6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 30 Aug 2025 20:22:11 -0700
Subject: [PATCH 0985/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 77ddc8fb61..b9b7dcbe80 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 802f6724c8..fb0ee8f88b 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index bbb4fa59d0..080615e318 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index b721bcce63..6b8181a78a 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 80b168d2da..5f44da7507 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 5bfcdea6d4..aa82f50c84 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 31743a21c4..452ad80186 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 0d04d22989..2bd992fcf0 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index ca0c4017c8..c170d1e2cd 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index b5a853bad1..2d71b660f5 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 995f5f268c..3a9b79e156 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 3b5d94424f..202b726d3e 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 4bc705ec98..3c1a14bd19 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index a4dc4de97b..6ba0abcc58 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 4367e18018..cd94fb8cbd 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 2130efcceb..51727f2d12 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.39.0/llama_cpp_binaries-0.39.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 13876a1ee8b0d40ab54f4683b02c4534543f8aa8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 30 Aug 2025 20:27:32 -0700
Subject: [PATCH 0986/1701] llama.cpp: Remove the --flash-attn flag (it's
 always on now)

---
 modules/llama_cpp_server.py | 2 --
 modules/loaders.py          | 1 -
 modules/shared.py           | 4 ----
 modules/ui.py               | 1 -
 modules/ui_model_menu.py    | 1 -
 5 files changed, 9 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 8579f84397..6a094c9d9f 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -318,8 +318,6 @@ def _start_server(self):
             "--no-webui",
         ]
 
-        if shared.args.flash_attn:
-            cmd.append("--flash-attn")
         if shared.args.threads > 0:
             cmd += ["--threads", str(shared.args.threads)]
         if shared.args.threads_batch > 0:
diff --git a/modules/loaders.py b/modules/loaders.py
index f88e976dd1..fe982ab55f 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -16,7 +16,6 @@
         'streaming_llm',
         'rope_freq_base',
         'compress_pos_emb',
-        'flash_attn',
         'row_split',
         'no_kv_offload',
         'no_mmap',
diff --git a/modules/shared.py b/modules/shared.py
index a308523957..4daf43c9aa 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -73,7 +73,6 @@
 
 # llama.cpp
 group = parser.add_argument_group('llama.cpp')
-group.add_argument('--flash-attn', action='store_true', help='Use flash-attention.')
 group.add_argument('--threads', type=int, default=0, help='Number of threads to use.')
 group.add_argument('--threads-batch', type=int, default=0, help='Number of threads to use for batches/prompt processing.')
 group.add_argument('--batch-size', type=int, default=256, help='Maximum number of prompt tokens to batch together when calling llama_eval.')
@@ -159,9 +158,6 @@
 group.add_argument('--api-disable-ipv4', action='store_true', help='Disable IPv4 for the API')
 group.add_argument('--nowebui', action='store_true', help='Do not launch the Gradio UI. Useful for launching the API in standalone mode.')
 
-# Deprecated parameters
-group = parser.add_argument_group('Deprecated')
-
 # Handle CMD_FLAGS.txt
 cmd_flags_path = Path(__file__).parent.parent / "user_data" / "CMD_FLAGS.txt"
 if cmd_flags_path.exists():
diff --git a/modules/ui.py b/modules/ui.py
index 502005e747..12f43768aa 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -142,7 +142,6 @@ def list_model_elements():
         'num_experts_per_token',
         'load_in_8bit',
         'load_in_4bit',
-        'flash_attn',
         'attn_implementation',
         'cpu',
         'disk',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index dd2406274d..729700d473 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -50,7 +50,6 @@ def create_ui():
 
                         with gr.Column():
                             shared.gradio['vram_info'] = gr.HTML(value=get_initial_vram_info())
-                            shared.gradio['flash_attn'] = gr.Checkbox(label="flash-attn", value=shared.args.flash_attn, info='Use flash-attention.')
                             shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming-llm", value=shared.args.streaming_llm, info='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
                             shared.gradio['load_in_8bit'] = gr.Checkbox(label="load-in-8bit", value=shared.args.load_in_8bit)
                             shared.gradio['load_in_4bit'] = gr.Checkbox(label="load-in-4bit", value=shared.args.load_in_4bit)

From 8028d8854122887616a5a5322704904fffa98a93 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 30 Aug 2025 21:29:20 -0700
Subject: [PATCH 0987/1701] Lint

---
 modules/models.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/modules/models.py b/modules/models.py
index 133131d7f1..d2b9cc987c 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -1,6 +1,5 @@
 import sys
 import time
-from pathlib import Path
 
 import modules.shared as shared
 from modules.logging_colors import logger

From 387e249decfb8ca7e119e5971da11d3605e7e3e3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 30 Aug 2025 21:31:27 -0700
Subject: [PATCH 0988/1701] Change an info message

---
 modules/ui_chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 1e8218a950..7c3886072f 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -92,7 +92,7 @@ def create_ui():
                 gr.HTML("
            ")
 
                 with gr.Row():
-                    shared.gradio['mode'] = gr.Radio(choices=['instruct', 'chat-instruct', 'chat'], value=None, label='Mode', info='Defines how the chat prompt is generated. In instruct and chat-instruct modes, the instruction template Parameters > Instruction template is used.', elem_id='chat-mode')
+                    shared.gradio['mode'] = gr.Radio(choices=['instruct', 'chat-instruct', 'chat'], value=None, label='Mode', info='In instruct and chat-instruct modes, the template under Parameters > Instruction template is used.', elem_id='chat-mode')
 
                 with gr.Row():
                     shared.gradio['chat_style'] = gr.Dropdown(choices=utils.get_available_chat_styles(), label='Chat style', value=shared.settings['chat_style'], visible=shared.settings['mode'] != 'instruct')

From 00ebb295d32cc89da239f62b68a15bb1ae4bc636 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 31 Aug 2025 16:27:23 -0700
Subject: [PATCH 0989/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index b9b7dcbe80..9cf069c78d 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index fb0ee8f88b..81434cc83d 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 080615e318..bf547be76b 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 6b8181a78a..64dfcab739 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 5f44da7507..cb29ad4eae 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index aa82f50c84..e0a1078230 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 452ad80186..5b6a4bf4bc 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 2bd992fcf0..39c6f768e0 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index c170d1e2cd..8407fa2964 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 2d71b660f5..3d2be6de44 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 3a9b79e156..2bf635b3dd 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 202b726d3e..3b9fc16f50 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 3c1a14bd19..e4d2900d27 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 6ba0abcc58..5b492b4266 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index cd94fb8cbd..90e7d38b7d 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 51727f2d12..fe21a1c769 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.40.0/llama_cpp_binaries-0.40.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From d843afcf66afeea23c941d6418d63233d0702d2e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 2 Sep 2025 05:43:33 -0700
Subject: [PATCH 0990/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 9cf069c78d..3a3b899cda 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 81434cc83d..388da65c77 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index bf547be76b..d163577929 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 64dfcab739..dde8d4a1fc 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index cb29ad4eae..9b1776ca24 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index e0a1078230..17d907bc52 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 5b6a4bf4bc..8c0954283e 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 39c6f768e0..553e8cfb14 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 8407fa2964..e77ce7b196 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 3d2be6de44..dc45ef3758 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 2bf635b3dd..541f96d405 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 3b9fc16f50..2af3b4b927 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index e4d2900d27..6a5f574020 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 5b492b4266..a7f2405b22 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 90e7d38b7d..bb2b0f284f 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index fe21a1c769..404f126766 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.41.0/llama_cpp_binaries-0.41.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 00ed878b054df15311be871381b09a0a8ecd1135 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 2 Sep 2025 10:16:26 -0700
Subject: [PATCH 0991/1701] Slightly more robust model loading

---
 modules/models.py | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/modules/models.py b/modules/models.py
index d2b9cc987c..9535ea82a8 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -45,12 +45,13 @@ def load_model(model_name, loader=None):
         model, tokenizer = output
     else:
         model = output
-        if model is None:
-            return None, None
-        else:
+        if model is not None:
             from modules.transformers_loader import load_tokenizer
             tokenizer = load_tokenizer(model_name)
 
+    if model is None:
+        return None, None
+
     shared.settings.update({k: v for k, v in metadata.items() if k in shared.settings})
     if loader.lower().startswith('exllama') or loader.lower().startswith('tensorrt') or loader == 'llama.cpp':
         shared.settings['truncation_length'] = shared.args.ctx_size

From c6ea67bbdbb4b1f7de7d6f0a8d6909c54c62c348 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 2 Sep 2025 10:22:03 -0700
Subject: [PATCH 0992/1701] Lint

---
 modules/html_generator.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 9f8c28e53a..492b52bd31 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -262,7 +262,7 @@ def protect_asterisks_in_latex(match):
             modified_content = content.replace('*', LATEX_ASTERISK_PLACEHOLDER)
             return f'\\({modified_content}\\)'
 
-        return match.group(0) # Fallback
+        return match.group(0)  # Fallback
 
     # Make \[ \]  LaTeX equations inline
     pattern = r'^\s*\\\[\s*\n([\s\S]*?)\n\s*\\\]\s*$'

From 2395c647d45769fe8c440d75219fb838a74869e3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 2 Sep 2025 12:11:15 -0700
Subject: [PATCH 0993/1701] Fix the instruct message height on mobile

---
 js/main.js | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index c08dffcfb3..c31621f63a 100644
--- a/js/main.js
+++ b/js/main.js
@@ -206,7 +206,13 @@ const observer = new MutationObserver(function(mutations) {
       // Add padding to the messages container to create room for the last message.
       // The purpose of this is to avoid constant scrolling during streaming in
       // instruct mode.
-      const bufferHeight = Math.max(0, Math.max(window.innerHeight - 128 - 84, window.innerHeight - prevSibling.offsetHeight - 84) - lastChild.offsetHeight);
+      let bufferHeight = Math.max(0, Math.max(window.innerHeight - 128 - 84, window.innerHeight - prevSibling.offsetHeight - 84) - lastChild.offsetHeight);
+
+      // Subtract header height when screen width is <= 924px
+      if (window.innerWidth <= 924) {
+        bufferHeight = Math.max(0, bufferHeight - 32);
+      }
+
       messagesContainer.style.paddingBottom = `${bufferHeight}px`;
     }
   }

From f3829b268a870c8113dc4146a13e5d9e07fd1aea Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 2 Sep 2025 12:12:17 -0700
Subject: [PATCH 0994/1701] llama.cpp: Always pass --flash-attn on

---
 modules/llama_cpp_server.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 6a094c9d9f..38589cf290 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -316,6 +316,7 @@ def _start_server(self):
             "--batch-size", str(shared.args.batch_size),
             "--port", str(self.port),
             "--no-webui",
+            "--flash-attn", "on",
         ]
 
         if shared.args.threads > 0:

From ba62783b72c41fc266adca026a2bc0d4fd438458 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 2 Sep 2025 14:22:22 -0700
Subject: [PATCH 0995/1701] UI: Don't use `$ $` for LaTeX, only `$$ $$`

---
 js/main.js | 1 -
 1 file changed, 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index c31621f63a..a4378c1b79 100644
--- a/js/main.js
+++ b/js/main.js
@@ -273,7 +273,6 @@ function doSyntaxHighlighting() {
             renderMathInElement(container, {
               delimiters: [
                 { left: "$$", right: "$$", display: true },
-                { left: "$", right: "$", display: false },
                 { left: "\\(", right: "\\)", display: false },
                 { left: "\\[", right: "\\]", display: true },
               ],

From 557b78d31eadd5bab9c8495dcb01c44c7608a7b3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 3 Sep 2025 16:50:03 -0700
Subject: [PATCH 0996/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 3a3b899cda..843d24277c 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 388da65c77..fc45ab1d8c 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index d163577929..9bbc437042 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index dde8d4a1fc..97b50b8549 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 9b1776ca24..f67b0cf3bf 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 17d907bc52..bb29982ec4 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 8c0954283e..e78224ff72 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 553e8cfb14..5ebd291990 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index e77ce7b196..b42832821d 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index dc45ef3758..c8953afdd2 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 541f96d405..d6bd0cbba4 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 2af3b4b927..7c99a4ab2a 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 6a5f574020..8756eda4b4 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index a7f2405b22..2dcccd8fc4 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index bb2b0f284f..df0f014839 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 404f126766..8fc857c6c1 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.42.0/llama_cpp_binaries-0.42.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 483927a5be10228f905158a26398109d8b2286b8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 17 Sep 2025 05:09:12 -0700
Subject: [PATCH 0997/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 843d24277c..baf3d9ebe7 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index fc45ab1d8c..b56577f006 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 9bbc437042..c80f21bd7a 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 97b50b8549..263496c072 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index f67b0cf3bf..cc3f2a9972 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index bb29982ec4..2e35f1a58f 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index e78224ff72..461063095c 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 5ebd291990..84f8b08979 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index b42832821d..124efa7c98 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index c8953afdd2..ee44da1ea7 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index d6bd0cbba4..b8ea820eb5 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 7c99a4ab2a..a2e1f8f97c 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 8756eda4b4..b0d8a05ed0 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 2dcccd8fc4..41110ab545 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index df0f014839..e393f5a2bb 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 8fc857c6c1..fdf1435d42 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.43.0/llama_cpp_binaries-0.43.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 8f731a566c88b37e0b93d47cf9038757e196baab Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Wed, 17 Sep 2025 12:17:16 -0300
Subject: [PATCH 0998/1701] Update peft requirement from ==0.16.* to ==0.17.*
 in /requirements/full (#7172)

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index baf3d9ebe7..68a642bdda 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -11,7 +11,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.16.*
+peft==0.17.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index b56577f006..c06ee19598 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.16.*
+peft==0.17.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index c80f21bd7a..f95e745608 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.16.*
+peft==0.17.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 263496c072..7ecc79311d 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.16.*
+peft==0.17.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index cc3f2a9972..bf76d86215 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.16.*
+peft==0.17.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 2e35f1a58f..ab5efdf901 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.16.*
+peft==0.17.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 461063095c..fd2db9481a 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.16.*
+peft==0.17.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 84f8b08979..b198287e02 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -11,7 +11,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.16.*
+peft==0.17.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 74d8604708..22f63fb418 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.16.*
+peft==0.17.*
 Pillow>=9.5.0
 psutil
 pydantic==2.8.2

From fe15b67160a195e6e0129a79942b5803524a4e5d Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Wed, 17 Sep 2025 12:17:58 -0300
Subject: [PATCH 0999/1701] Update bitsandbytes requirement in
 /requirements/full (#7193)

---
 requirements/full/requirements.txt        | 2 +-
 requirements/full/requirements_noavx2.txt | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 68a642bdda..80a27bce46 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -1,6 +1,6 @@
 accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
-bitsandbytes==0.46.*
+bitsandbytes==0.47.*
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index b198287e02..1c92b6b22d 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -1,6 +1,6 @@
 accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
-bitsandbytes==0.46.*
+bitsandbytes==0.47.*
 colorama
 datasets
 einops

From 7131a478b99d574306f83a6db994648384b23e2d Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Wed, 17 Sep 2025 12:18:13 -0300
Subject: [PATCH 1000/1701] Update safetensors requirement in
 /requirements/full (#7192)

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 80a27bce46..6bef240ae5 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -20,7 +20,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.5.*
+safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index c06ee19598..acee1defdd 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -19,7 +19,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.5.*
+safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index f95e745608..ca4d7d45e3 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -19,7 +19,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.5.*
+safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 7ecc79311d..351035d0f0 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -19,7 +19,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.5.*
+safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index bf76d86215..92a014b3b9 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -19,7 +19,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.5.*
+safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index ab5efdf901..51083d21fe 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -19,7 +19,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.5.*
+safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index fd2db9481a..3c231ace9c 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -19,7 +19,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.5.*
+safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 1c92b6b22d..1bed9c4ffe 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -20,7 +20,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.5.*
+safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 22f63fb418..7e341ed98a 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -19,7 +19,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.5.*
+safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard

From 8087a57fd84ec665c6ad0549738892d02b06003f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 17 Sep 2025 08:19:18 -0700
Subject: [PATCH 1001/1701] Bump transformers to 4.56

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 6bef240ae5..38d07c4b39 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -24,7 +24,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.55.*
+transformers==4.56.*
 triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index acee1defdd..15b8aaed45 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -23,7 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.55.*
+transformers==4.56.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index ca4d7d45e3..a9c29a3cdd 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -23,7 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.55.*
+transformers==4.56.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 351035d0f0..dc3bb6a1c8 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -23,7 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.55.*
+transformers==4.56.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 92a014b3b9..25f1e8bdd0 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -23,7 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.55.*
+transformers==4.56.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 51083d21fe..611d3f0f7f 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -23,7 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.55.*
+transformers==4.56.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 3c231ace9c..2696788612 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -23,7 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.55.*
+transformers==4.56.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 1bed9c4ffe..ccf7c5ea8c 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -24,7 +24,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.55.*
+transformers==4.56.*
 triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 7e341ed98a..1f63e30490 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -23,7 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.55.*
+transformers==4.56.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb

From 9c0a833a0a4bea303b9a52cd007cc4df6f8b91d8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 17 Sep 2025 11:58:54 -0700
Subject: [PATCH 1002/1701] Revert "Update bitsandbytes requirement in
 /requirements/full (#7193)"

This reverts commit fe15b67160a195e6e0129a79942b5803524a4e5d.
---
 requirements/full/requirements.txt        | 2 +-
 requirements/full/requirements_noavx2.txt | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 38d07c4b39..008799cd9c 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -1,6 +1,6 @@
 accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
-bitsandbytes==0.47.*
+bitsandbytes==0.46.*
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index ccf7c5ea8c..1ed1008dba 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -1,6 +1,6 @@
 accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
-bitsandbytes==0.47.*
+bitsandbytes==0.46.*
 colorama
 datasets
 einops

From 9e9ab3989265a084d0f63ec1c07d39aab9141ab5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 17 Sep 2025 12:29:22 -0700
Subject: [PATCH 1003/1701] Make exllamav3_hf and exllamav2_hf functional again

---
 modules/exllamav2_hf.py |  4 +++-
 modules/exllamav3_hf.py | 10 ++++++----
 2 files changed, 9 insertions(+), 5 deletions(-)

diff --git a/modules/exllamav2_hf.py b/modules/exllamav2_hf.py
index 4aa46375b6..4ba185902c 100644
--- a/modules/exllamav2_hf.py
+++ b/modules/exllamav2_hf.py
@@ -35,7 +35,9 @@
 
 class Exllamav2HF(PreTrainedModel, GenerationMixin):
     def __init__(self, config: ExLlamaV2Config):
-        super().__init__(PretrainedConfig())
+        hf_config = PretrainedConfig.from_pretrained(config.model_dir)
+        super().__init__(hf_config)
+
         self.ex_config = config
         self.loras = None
         self.generation_config = GenerationConfig()
diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index d9f4ed576a..05b473b71f 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -27,11 +27,13 @@
 
 class Exllamav3HF(PreTrainedModel, GenerationMixin):
     def __init__(self, model_dir):
-        super().__init__(PretrainedConfig())
-        self.generation_config = GenerationConfig()
+        hf_config = PretrainedConfig.from_pretrained(model_dir)
+        super().__init__(hf_config)
+
+        exl3_config = Config.from_directory(model_dir)
 
-        config = Config.from_directory(model_dir)
-        self.ex_model = Model.from_config(config)
+        self.generation_config = GenerationConfig()
+        self.ex_model = Model.from_config(exl3_config)
 
         # Calculate the closest multiple of 256 at or above the chosen value
         max_tokens = shared.args.ctx_size

From dd6d2223a56366a0097df254bff1f6a658c6307a Mon Sep 17 00:00:00 2001
From: stevenxdavis <31548921+stevenxdavis@users.noreply.github.com>
Date: Wed, 17 Sep 2025 14:39:04 -0500
Subject: [PATCH 1004/1701] Changing transformers_loader.py to Match User
 Expectations for --bf16 and Flash Attention 2 (#7217)

---
 modules/transformers_loader.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/transformers_loader.py b/modules/transformers_loader.py
index e407212561..7866f44888 100644
--- a/modules/transformers_loader.py
+++ b/modules/transformers_loader.py
@@ -137,6 +137,7 @@ def load_model_HF(model_name):
     params = {
         'low_cpu_mem_usage': True,
         'attn_implementation': shared.args.attn_implementation,
+        'torch_dtype': torch.bfloat16 if shared.args.bf16 else torch.float16,
     }
 
     if shared.args.trust_remote_code:

From e4412f063486d673f36532a770fb7de72e1ba826 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 17 Sep 2025 21:57:17 -0700
Subject: [PATCH 1005/1701] Slightly more robust syntax highlighting

---
 js/main.js | 72 ++++++++++++++++++++++++++++--------------------------
 1 file changed, 38 insertions(+), 34 deletions(-)

diff --git a/js/main.js b/js/main.js
index a4378c1b79..29b53cc933 100644
--- a/js/main.js
+++ b/js/main.js
@@ -249,44 +249,48 @@ function doSyntaxHighlighting() {
   if (messageBodies.length > 0) {
     observer.disconnect();
 
-    let hasSeenVisible = false;
-
-    // Go from last message to first
-    for (let i = messageBodies.length - 1; i >= 0; i--) {
-      const messageBody = messageBodies[i];
-
-      if (isElementVisibleOnScreen(messageBody)) {
-        hasSeenVisible = true;
-
-        // Handle both code and math in a single pass through each message
-        const codeBlocks = messageBody.querySelectorAll("pre code:not([data-highlighted])");
-        codeBlocks.forEach((codeBlock) => {
-          hljs.highlightElement(codeBlock);
-          codeBlock.setAttribute("data-highlighted", "true");
-          codeBlock.classList.add("pretty_scrollbar");
-        });
-
-        // Only render math in visible elements
-        const mathContainers = messageBody.querySelectorAll("p, span, li, td, th, h1, h2, h3, h4, h5, h6, blockquote, figcaption, caption, dd, dt");
-        mathContainers.forEach(container => {
-          if (isElementVisibleOnScreen(container)) {
-            renderMathInElement(container, {
-              delimiters: [
-                { left: "$$", right: "$$", display: true },
-                { left: "\\(", right: "\\)", display: false },
-                { left: "\\[", right: "\\]", display: true },
-              ],
-            });
-          }
-        });
-      } else if (hasSeenVisible) {
+    try {
+      let hasSeenVisible = false;
+
+      // Go from last message to first
+      for (let i = messageBodies.length - 1; i >= 0; i--) {
+        const messageBody = messageBodies[i];
+
+        if (isElementVisibleOnScreen(messageBody)) {
+          hasSeenVisible = true;
+
+          // Handle both code and math in a single pass through each message
+          const codeBlocks = messageBody.querySelectorAll("pre code:not([data-highlighted])");
+          codeBlocks.forEach((codeBlock) => {
+            hljs.highlightElement(codeBlock);
+            codeBlock.setAttribute("data-highlighted", "true");
+            codeBlock.classList.add("pretty_scrollbar");
+          });
+
+          // Only render math in visible elements
+          const mathContainers = messageBody.querySelectorAll("p, span, li, td, th, h1, h2, h3, h4, h5, h6, blockquote, figcaption, caption, dd, dt");
+          mathContainers.forEach(container => {
+            if (isElementVisibleOnScreen(container)) {
+              renderMathInElement(container, {
+                delimiters: [
+                  { left: "$$", right: "$$", display: true },
+                  { left: "\\(", right: "\\)", display: false },
+                  { left: "\\[", right: "\\]", display: true },
+                ],
+              });
+            }
+          });
+        } else if (hasSeenVisible) {
         // We've seen visible messages but this one is not visible
         // Since we're going from last to first, we can break
-        break;
+          break;
+        }
       }
-    }
 
-    observer.observe(targetElement, config);
+    } finally {
+
+      observer.observe(targetElement, config);
+    }
   }
 }
 

From 005fcf3f984758e3034afe218d21d78600a53756 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 17 Sep 2025 21:58:37 -0700
Subject: [PATCH 1006/1701] Formatting

---
 js/main.js | 2 --
 1 file changed, 2 deletions(-)

diff --git a/js/main.js b/js/main.js
index 29b53cc933..67f602793c 100644
--- a/js/main.js
+++ b/js/main.js
@@ -286,9 +286,7 @@ function doSyntaxHighlighting() {
           break;
         }
       }
-
     } finally {
-
       observer.observe(targetElement, config);
     }
   }

From 1e863a71137c43bde82a9c623765f809f084dfa1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 19 Sep 2025 16:12:50 -0700
Subject: [PATCH 1007/1701] Fix exllamav3 ignoring the stop button

---
 modules/exllamav3.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index fd676a005d..f7078028bc 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -344,6 +344,9 @@ def custom_sort_key(sampler_obj):
 
         try:
             while self.generator.num_remaining_jobs():
+                if shared.stop_everything:
+                    break
+
                 results = self.generator.iterate()
                 for result in results:
                     if "eos" in result and result["eos"]:

From 8c9df34696c065ce9bbb5a2bae0e9aa1a617b370 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 20 Sep 2025 20:57:15 -0700
Subject: [PATCH 1008/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 008799cd9c..85119c65d1 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 15b8aaed45..ffd496f397 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index a9c29a3cdd..7a35b5537c 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index dc3bb6a1c8..ebf132429e 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 25f1e8bdd0..00303ff9f1 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 611d3f0f7f..9a57850103 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 2696788612..d777a01367 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 1ed1008dba..f35dd111cc 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 124efa7c98..2162fddf8f 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index ee44da1ea7..91150ed1e9 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index b8ea820eb5..222403862e 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index a2e1f8f97c..847e445075 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index b0d8a05ed0..f34e18475d 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 41110ab545..771d036291 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index e393f5a2bb..bb3a5ab898 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index fdf1435d42..fbc5228275 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.45.0/llama_cpp_binaries-0.45.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From b7effb22e06d6cf05b3c2da2ed6c7fced339a2fc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 5 Oct 2025 07:29:38 -0700
Subject: [PATCH 1009/1701] Update exllamav3

---
 requirements/full/requirements.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_noavx2.txt        | 4 ++--
 4 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 85119c65d1..84192fc41f 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -37,8 +37,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index ebf132429e..6798cd45c6 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -36,5 +36,5 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 00303ff9f1..7a83a7c609 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,5 +37,5 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6-py3-none-any.whl
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index f35dd111cc..a08b75a621 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -37,8 +37,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.6/exllamav3-0.0.6+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 0eb8543d746910fa519b4c9e6a1a93f39536d81c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 5 Oct 2025 07:30:33 -0700
Subject: [PATCH 1010/1701] Update transformers

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 84192fc41f..6e214fcaed 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -24,7 +24,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.56.*
+transformers==4.57.*
 triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index ffd496f397..422965c157 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -23,7 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.56.*
+transformers==4.57.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 7a35b5537c..a655600da9 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -23,7 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.56.*
+transformers==4.57.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 6798cd45c6..ab0696db75 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -23,7 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.56.*
+transformers==4.57.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 7a83a7c609..0050a71726 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -23,7 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.56.*
+transformers==4.57.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 9a57850103..92a3659eea 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -23,7 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.56.*
+transformers==4.57.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index d777a01367..fe987cbd8a 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -23,7 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.56.*
+transformers==4.57.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index a08b75a621..c86b90395f 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -24,7 +24,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.56.*
+transformers==4.57.*
 triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 1f63e30490..3bd20dd9c8 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -23,7 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
-transformers==4.56.*
+transformers==4.57.*
 triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb

From 64829071e0280e7795a45d11a057c7a7c8e72ea8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 5 Oct 2025 07:32:41 -0700
Subject: [PATCH 1011/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 6e214fcaed..29526c7950 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 422965c157..92eb51e05e 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index a655600da9..11d19547bb 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index ab0696db75..fe1fdc9fe1 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 0050a71726..b05eab13c6 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 92a3659eea..590e7e88b0 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index fe987cbd8a..56abaaa15d 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index c86b90395f..1202fff3e4 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 2162fddf8f..577183e9f3 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 91150ed1e9..347b7e1ed6 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 222403862e..810ae2323e 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 847e445075..6d16845e5d 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index f34e18475d..e0e5edece0 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 771d036291..aa0c8c37e6 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index bb3a5ab898..29a63ec05d 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index fbc5228275..e1a90292e7 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.46.0/llama_cpp_binaries-0.46.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 38a7fd685d4ed669fde78e54c9423a38588202b6 Mon Sep 17 00:00:00 2001
From: Remowylliams 
Date: Sun, 5 Oct 2025 09:34:47 -0500
Subject: [PATCH 1012/1701] chat.py fixes Instruct mode History

---
 modules/chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index ad2f4001e1..70c55b740d 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1194,7 +1194,7 @@ def find_all_histories_with_first_prompts(state):
         if re.match(r'^[0-9]{8}-[0-9]{2}-[0-9]{2}-[0-9]{2}$', filename):
             first_prompt = ""
             if data and 'visible' in data and len(data['visible']) > 0:
-                if data['internal'][0][0] == '<|BEGIN-VISIBLE-CHAT|>':
+                if len(data['internal']) > 1 and data['internal'][0][0] == '<|BEGIN-VISIBLE-CHAT|>':
                     if len(data['visible']) > 1:
                         first_prompt = html.unescape(data['visible'][1][0])
                     elif i == 0:

From 898a3ed2feb9750f8998f371ef3af52f36a7cb6e Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Tue, 7 Oct 2025 18:33:28 -0300
Subject: [PATCH 1013/1701] Add sponsor (Warp) to README <3

---
 README.md | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/README.md b/README.md
index d42697dd54..c3f7d8a65b 100644
--- a/README.md
+++ b/README.md
@@ -1,3 +1,16 @@
+
            
+   Special thanks to:
+   
            
+   
            
+   
+      Warp sponsorship
+   
+
+### [Warp, built for coding with multiple AI agents](https://go.warp.dev/text-generation-webui)
+[Available for macOS, Linux, & Windows](https://go.warp.dev/text-generation-webui)
            
+
            
+
            
+
 # Text Generation Web UI
 
 A Gradio web UI for Large Language Models.

From 292c91abbb08aef017c8b21aeadd807e5b702c12 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 8 Oct 2025 08:31:34 -0700
Subject: [PATCH 1014/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 29526c7950..eeb9728479 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 92eb51e05e..968bf635ac 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 11d19547bb..40adf20e63 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index fe1fdc9fe1..2be03386ba 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index b05eab13c6..b715d17ba6 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 590e7e88b0..2c46247b5e 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 56abaaa15d..5c607becb5 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 1202fff3e4..bfcdbb1da2 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 577183e9f3..50c0a89104 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 347b7e1ed6..a6ab4179cc 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 810ae2323e..d6fd065e13 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 6d16845e5d..efa587ea9d 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index e0e5edece0..10eb57a561 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index aa0c8c37e6..45be85e45a 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 29a63ec05d..1a72b16bbc 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index e1a90292e7..0d684da823 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.47.0/llama_cpp_binaries-0.47.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From d229dfe991e7fb70b825daae47d99c5d15fddf0e Mon Sep 17 00:00:00 2001
From: Ionoclast Laboratories 
Date: Wed, 8 Oct 2025 08:40:53 -0700
Subject: [PATCH 1015/1701] Fix portable apple intel requirement for llama
 binaries (issue #7238) (#7239)

---
 requirements/portable/requirements_apple_intel.txt | 1 +
 1 file changed, 1 insertion(+)

diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index a6ab4179cc..e844165e69 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -21,3 +21,4 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
\ No newline at end of file

From 93aa7b3ed3f770c25e21594bafe41bb2dded2caa Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Oct 2025 08:49:27 -0700
Subject: [PATCH 1016/1701] Better handle multigpu setups with transformers +
 bitsandbytes

---
 modules/torch_utils.py | 15 +--------------
 1 file changed, 1 insertion(+), 14 deletions(-)

diff --git a/modules/torch_utils.py b/modules/torch_utils.py
index ad9b26ade7..1bc82f0387 100644
--- a/modules/torch_utils.py
+++ b/modules/torch_utils.py
@@ -2,25 +2,12 @@
 
 import torch
 from accelerate.utils import is_npu_available, is_xpu_available
-from transformers import is_torch_npu_available, is_torch_xpu_available
 
 from modules import shared
 
 
 def get_device():
-    if torch.cuda.is_available():
-        return torch.device('cuda')
-    elif shared.args.deepspeed:
-        import deepspeed
-        return deepspeed.get_accelerator().current_device_name()
-    elif torch.backends.mps.is_available():
-        return torch.device('mps')
-    elif is_torch_xpu_available():
-        return torch.device('xpu:0')
-    elif is_torch_npu_available():
-        return torch.device('npu:0')
-    else:
-        return None
+    return getattr(shared.model, 'device', None)
 
 
 def clear_torch_cache():

From 282aa1918907fceec7f903d3dc2bc8492ce8e885 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Oct 2025 09:26:35 -0700
Subject: [PATCH 1017/1701] Safer profile picture uploading

---
 modules/chat.py    | 55 ++++++++++++++++++++++++++++++++++++----------
 modules/ui_chat.py |  8 +++----
 2 files changed, 48 insertions(+), 15 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 70c55b740d..a52d96af3a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -3,8 +3,10 @@
 import functools
 import html
 import json
+import os
 import pprint
 import re
+import shutil
 import time
 from datetime import datetime
 from functools import partial
@@ -1385,12 +1387,17 @@ def generate_pfp_cache(character):
     for path in [Path(f"user_data/characters/{character}.{extension}") for extension in ['png', 'jpg', 'jpeg']]:
         if path.exists():
             original_img = Image.open(path)
-            original_img.save(Path(f'{cache_folder}/pfp_character.png'), format='PNG')
+            # Define file paths
+            pfp_path = Path(f'{cache_folder}/pfp_character.png')
+            thumb_path = Path(f'{cache_folder}/pfp_character_thumb.png')
 
+            # Save main picture and thumbnail
+            original_img.save(pfp_path, format='PNG')
             thumb = make_thumbnail(original_img)
-            thumb.save(Path(f'{cache_folder}/pfp_character_thumb.png'), format='PNG')
+            thumb.save(thumb_path, format='PNG')
 
-            return thumb
+            # Return the path to the thumbnail, not the in-memory PIL Image object.
+            return str(thumb_path)
 
     return None
 
@@ -1507,7 +1514,22 @@ def load_instruction_template_memoized(template):
     return load_instruction_template(template)
 
 
-def upload_character(file, img, tavern=False):
+def open_image_safely(path):
+    if path is None or not isinstance(path, str) or not Path(path).exists():
+        return None
+
+    if os.path.islink(path):
+        return None
+
+    try:
+        return Image.open(path)
+    except Exception as e:
+        logger.error(f"Failed to open image file: {path}. Reason: {e}")
+        return None
+
+
+def upload_character(file, img_path, tavern=False):
+    img = open_image_safely(img_path)
     decoded_file = file if isinstance(file, str) else file.decode('utf-8')
     try:
         data = json.loads(decoded_file)
@@ -1554,12 +1576,17 @@ def build_pygmalion_style_context(data):
     return context
 
 
-def upload_tavern_character(img, _json):
+def upload_tavern_character(img_path, _json):
     _json = {'char_name': _json['name'], 'char_persona': _json['description'], 'char_greeting': _json['first_mes'], 'example_dialogue': _json['mes_example'], 'world_scenario': _json['scenario']}
-    return upload_character(json.dumps(_json), img, tavern=True)
+    return upload_character(json.dumps(_json), img_path, tavern=True)
+
 
+def check_tavern_character(img_path):
+    img = open_image_safely(img_path)
+
+    if img is None:
+        return "Invalid or disallowed image file.", None, None, gr.update(interactive=False)
 
-def check_tavern_character(img):
     if "chara" not in img.info:
         return "Not a TavernAI card", None, None, gr.update(interactive=False)
 
@@ -1571,7 +1598,8 @@ def check_tavern_character(img):
     return _json['name'], _json['description'], _json, gr.update(interactive=True)
 
 
-def upload_your_profile_picture(img):
+def upload_your_profile_picture(img_path):
+    img = open_image_safely(img_path)
     cache_folder = Path(shared.args.disk_cache_dir)
     if not cache_folder.exists():
         cache_folder.mkdir()
@@ -1614,15 +1642,19 @@ def save_character(name, greeting, context, picture, filename):
     save_file(filepath, data)
     path_to_img = Path(f'user_data/characters/{filename}.png')
     if picture is not None:
-        picture.save(path_to_img)
+        # Copy the image file from its source path to the character folder
+        shutil.copy(picture, path_to_img)
         logger.info(f'Saved {path_to_img}.')
 
 
 def delete_character(name, instruct=False):
+    # Check for character data files
     for extension in ["yml", "yaml", "json"]:
         delete_file(Path(f'user_data/characters/{name}.{extension}'))
 
-    delete_file(Path(f'user_data/characters/{name}.png'))
+    # Check for character image files
+    for extension in ["png", "jpg", "jpeg"]:
+        delete_file(Path(f'user_data/characters/{name}.{extension}'))
 
 
 def jinja_template_from_old_format(params, verbose=False):
@@ -1974,8 +2006,9 @@ def handle_character_menu_change(state):
     ]
 
 
-def handle_character_picture_change(picture):
+def handle_character_picture_change(picture_path):
     """Update or clear cache when character picture changes"""
+    picture = open_image_safely(picture_path)
     cache_folder = Path(shared.args.disk_cache_dir)
     if not cache_folder.exists():
         cache_folder.mkdir()
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 7c3886072f..c342ce5b0c 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -152,14 +152,14 @@ def create_character_settings_ui():
                     with gr.Tab('YAML or JSON'):
                         with gr.Row():
                             shared.gradio['upload_json'] = gr.File(type='binary', file_types=['.json', '.yaml'], label='JSON or YAML File', interactive=not mu)
-                            shared.gradio['upload_img_bot'] = gr.Image(type='pil', label='Profile Picture (optional)', interactive=not mu)
+                            shared.gradio['upload_img_bot'] = gr.Image(type='filepath', label='Profile Picture (optional)', interactive=not mu)
 
                         shared.gradio['Submit character'] = gr.Button(value='Submit', interactive=False)
 
                     with gr.Tab('TavernAI PNG'):
                         with gr.Row():
                             with gr.Column():
-                                shared.gradio['upload_img_tavern'] = gr.Image(type='pil', label='TavernAI PNG File', elem_id='upload_img_tavern', interactive=not mu)
+                                shared.gradio['upload_img_tavern'] = gr.Image(type='filepath', label='TavernAI PNG File', elem_id='upload_img_tavern', interactive=not mu)
                                 shared.gradio['tavern_json'] = gr.State()
                             with gr.Column():
                                 shared.gradio['tavern_name'] = gr.Textbox(value='', lines=1, label='Name', interactive=False)
@@ -168,8 +168,8 @@ def create_character_settings_ui():
                         shared.gradio['Submit tavern character'] = gr.Button(value='Submit', interactive=False)
 
             with gr.Column(scale=1):
-                shared.gradio['character_picture'] = gr.Image(label='Character picture', type='pil', interactive=not mu)
-                shared.gradio['your_picture'] = gr.Image(label='Your picture', type='pil', value=Image.open(Path('user_data/cache/pfp_me.png')) if Path('user_data/cache/pfp_me.png').exists() else None, interactive=not mu)
+                shared.gradio['character_picture'] = gr.Image(label='Character picture', type='filepath', interactive=not mu)
+                shared.gradio['your_picture'] = gr.Image(label='Your picture', type='filepath', value=Image.open(Path('user_data/cache/pfp_me.png')) if Path('user_data/cache/pfp_me.png').exists() else None, interactive=not mu)
 
 
 def create_chat_settings_ui():

From 0f3793d6086ec787a9b8f86e8d42b3ea63fb038e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Oct 2025 09:38:22 -0700
Subject: [PATCH 1018/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 6 +++---
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 35 insertions(+), 35 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index eeb9728479..d08209f67b 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 968bf635ac..b3b0005e09 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 40adf20e63..5e0d375e85 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 2be03386ba..425440309e 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -34,7 +34,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index b715d17ba6..dda91eac11 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -34,8 +34,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7-py3-none-any.whl
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 2c46247b5e..f68fdd9da1 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 5c607becb5..b40f1af216 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -34,5 +34,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index bfcdbb1da2..9143542cee 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -35,8 +35,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 50c0a89104..7a38b1e6eb 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index e844165e69..047d1c5476 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
\ No newline at end of file
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
\ No newline at end of file
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index d6fd065e13..5c8ae4df94 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -19,6 +19,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index efa587ea9d..f41efd5866 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 10eb57a561..6915805096 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 45be85e45a..ca66098c37 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 1a72b16bbc..36aff3617a 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 0d684da823..be7170e3d5 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -19,5 +19,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.48.0/llama_cpp_binaries-0.48.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 1aa2b924d2b2fa43f1c7a46f28599c8025e779e1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Oct 2025 10:52:31 -0700
Subject: [PATCH 1019/1701] Downloader: Gracefully handle '416 Range Not
 Satisfiable' when continuing downloads

---
 download-model.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/download-model.py b/download-model.py
index c0a3aa36b5..f9bca97b3f 100644
--- a/download-model.py
+++ b/download-model.py
@@ -255,6 +255,11 @@ def get_single_file(self, url, output_folder, start_from_scratch=False):
                         mode = 'ab'
 
                     with session.get(url, stream=True, headers=headers, timeout=30) as r:
+                        # Assume error 416 (range unsatisfiable) means the download is complete.
+                        # Small size differences can happen in Git LFS
+                        if mode == 'ab' and r.status_code == 416:
+                            break
+
                         r.raise_for_status()
                         total_size_from_stream = int(r.headers.get('content-length', 0))
                         if mode == 'ab':

From 218dc01b5189b2dcf08045bf0a8f17e4fad176a2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Oct 2025 10:59:34 -0700
Subject: [PATCH 1020/1701] Add fallbacks after
 93aa7b3ed3f770c25e21594bafe41bb2dded2caa

---
 modules/torch_utils.py | 17 ++++++++++++++++-
 1 file changed, 16 insertions(+), 1 deletion(-)

diff --git a/modules/torch_utils.py b/modules/torch_utils.py
index 1bc82f0387..418520a87d 100644
--- a/modules/torch_utils.py
+++ b/modules/torch_utils.py
@@ -2,12 +2,27 @@
 
 import torch
 from accelerate.utils import is_npu_available, is_xpu_available
+from transformers import is_torch_npu_available, is_torch_xpu_available
 
 from modules import shared
 
 
 def get_device():
-    return getattr(shared.model, 'device', None)
+    if hasattr(shared.model, 'device'):
+        return shared.model.device
+    elif torch.cuda.is_available():
+        return torch.device('cuda')
+    elif shared.args.deepspeed:
+        import deepspeed
+        return deepspeed.get_accelerator().current_device_name()
+    elif torch.backends.mps.is_available():
+        return torch.device('mps')
+    elif is_torch_xpu_available():
+        return torch.device('xpu:0')
+    elif is_torch_npu_available():
+        return torch.device('npu:0')
+    else:
+        return None
 
 
 def clear_torch_cache():

From 7f06aec3a1fc2e6924d87035483ce10ce65af058 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Oct 2025 11:24:25 -0700
Subject: [PATCH 1021/1701] exllamav3: Implement the logits function for
 /v1/internal/logits

---
 modules/exllamav3.py | 45 ++++++++++++++++++++++++++++++++++++++++++--
 modules/logits.py    |  3 ++-
 modules/models.py    |  3 +--
 3 files changed, 46 insertions(+), 5 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index f7078028bc..d884bbf7e2 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -2,6 +2,8 @@
 from pathlib import Path
 from typing import Any, List, Tuple
 
+import torch
+
 from exllamav3 import Cache, Config, Generator, Model, Tokenizer
 from exllamav3.cache import CacheLayer_fp16, CacheLayer_quant
 from exllamav3.generator import Job
@@ -16,7 +18,6 @@
     SS_TopK,
     SS_TopP
 )
-
 from modules import shared
 from modules.image_utils import (
     convert_image_attachments_to_pil,
@@ -171,7 +172,7 @@ def from_pretrained(cls, path_to_model):
         result.draft_model = draft_model
         result.draft_cache = draft_cache
 
-        return result
+        return result, result
 
     def is_multimodal(self) -> bool:
         """Check if this model supports multimodal input."""
@@ -367,11 +368,51 @@ def generate(self, prompt, state):
 
         return output
 
+    def get_logits(self, token_ids, **kwargs):
+        """
+        Process a batch of token_ids and return the logits for the last token.
+        This will reset and overwrite the model's cache.
+        """
+        # Initialize a single params dictionary that will be updated in-place
+        params = {
+            "cache": self.cache,
+            "reconstruct": False,
+            "attn_mode": "flash_attn",
+            "batch_shape": (1, self.max_tokens),
+            "past_len": 0
+        }
+        params.update(kwargs)
+
+        # Process prefix tokens to fill the cache and generate recurrent state
+        if token_ids.shape[-1] > 1:
+            prefix_ids = token_ids[:, :-1]
+
+            # This forward call updates the 'params' dict with the recurrent state
+            self.model.forward(
+                input_ids=prefix_ids,
+                params=params
+            )
+
+            # Update past_len for the next call
+            params["past_len"] = prefix_ids.shape[-1]
+
+        # Process the last token, now using the state-filled 'params' dict
+        last_token_ids = token_ids[:, -1:]
+        logits = self.model.forward(
+            input_ids=last_token_ids,
+            params=params
+        )
+
+        return logits.float().cpu()
+
     def encode(self, string, **kwargs):
         add_bos = kwargs.pop('add_bos', True)
         return self.tokenizer.encode(string, add_bos=add_bos, **kwargs)
 
     def decode(self, ids, **kwargs):
+        if isinstance(ids, torch.Tensor) and ids.dim() == 0:
+            ids = ids.view(1)
+
         return self.tokenizer.decode(ids, **kwargs)
 
     @property
diff --git a/modules/logits.py b/modules/logits.py
index 56a2057291..d668e44e4b 100644
--- a/modules/logits.py
+++ b/modules/logits.py
@@ -71,6 +71,7 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
         from modules.torch_utils import get_device
 
         is_non_hf_exllamav2 = shared.model.__class__.__name__ == 'Exllamav2Model'
+        is_non_hf_exllamav3 = shared.model.__class__.__name__ == 'Exllamav3Model'
 
         if not use_samplers:
             state = {'stream': True}
@@ -88,7 +89,7 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
 
             scores = sampler_hijack.global_scores[-1]
         else:
-            if is_non_hf_exllamav2:
+            if is_non_hf_exllamav2 or is_non_hf_exllamav3:
                 device = get_device()
                 tokens = shared.tokenizer.encode(prompt)
                 if device:
diff --git a/modules/models.py b/modules/models.py
index 9535ea82a8..8c0f1c3700 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -104,8 +104,7 @@ def ExLlamav3_HF_loader(model_name):
 def ExLlamav3_loader(model_name):
     from modules.exllamav3 import Exllamav3Model
 
-    model = Exllamav3Model.from_pretrained(model_name)
-    tokenizer = model.tokenizer
+    model, tokenizer = Exllamav3Model.from_pretrained(model_name)
     return model, tokenizer
 
 

From deb37b821bb13f241dcc3e6aff7cbf41bbae143f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Oct 2025 12:05:45 -0700
Subject: [PATCH 1022/1701] Same as 7f06aec3a1fc2e6924d87035483ce10ce65af058
 but for exllamav3_hf

---
 modules/exllamav3_hf.py | 137 ++++++++++++++--------------------------
 1 file changed, 48 insertions(+), 89 deletions(-)

diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index 05b473b71f..c606912bb4 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -103,6 +103,12 @@ def __call__(self, *args, **kwargs):
         labels = kwargs.get('labels', None)
         past_key_values = kwargs.get('past_key_values', None)
 
+        # Reset the internal sequence state for standalone calls (logit viewer)
+        # or the very first step of a new generation.
+        if past_key_values is None:
+            self.past_seq = None
+            self.past_seq_negative = None
+
         if len(args) > 0:
             if not shared.args.cfg_cache:
                 logger.error("Please enable the cfg-cache option to use CFG with ExLlamav3_HF.")
@@ -119,8 +125,8 @@ def __call__(self, *args, **kwargs):
             ex_cache = self.ex_cache
 
         seq = input_ids[0].tolist()
-        if is_negative and past_key_values is not None:
-            seq = past_key_values + seq
+        if is_negative and past_key_values is not None and isinstance(past_key_values, list):
+             seq = past_key_values + seq
 
         seq_tensor = torch.tensor(seq)
         reset = True
@@ -128,97 +134,50 @@ def __call__(self, *args, **kwargs):
         # Maximum number of tokens to process in a single forward pass
         max_chunk_size = 256
 
+        if past_seq is not None:
+            min_length = min(past_seq.shape[0], seq_tensor.shape[0])
+            indices = torch.nonzero(~torch.eq(past_seq[:min_length], seq_tensor[:min_length]))
+            if len(indices) == 0 and seq_tensor.shape[0] > past_seq.shape[0]:
+                reset = False
+
+        # Create a single `params` dictionary that will be used and modified
+        # in-place across all `forward` calls within this function.
+        params = {
+            "attn_mode": "flash_attn",
+            "cache": ex_cache,
+            "batch_shape": (1, self.max_tokens),
+            "reconstruct": False,
+            "past_len": 0
+        }
+
         # Make the forward call
         if labels is None:
-            if past_seq is not None:
-                min_length = min(past_seq.shape[0], seq_tensor.shape[0])
-                indices = torch.nonzero(~torch.eq(past_seq[:min_length], seq_tensor[:min_length]))
-                if len(indices) > 0:
-                    longest_prefix = indices[0].item()
-                else:
-                    longest_prefix = min_length
-
-                if longest_prefix > 0:
-                    reset = False
-                    current_len = longest_prefix
-                    remaining_tokens = len(seq_tensor) - longest_prefix - 1
-
-                    if remaining_tokens > 0:
-                        # Process tokens from longest_prefix to second-to-last token
-                        tokens_to_process = seq_tensor[longest_prefix:-1]
-
-                        # Process in chunks if the number of tokens is large
-                        for i in range(0, tokens_to_process.shape[0], max_chunk_size):
-                            chunk = tokens_to_process[i:i + max_chunk_size]
-                            self.ex_model.forward(
-                                input_ids=chunk.view(1, -1),
-                                params={
-                                    "attn_mode": "flash_attn",
-                                    "cache": ex_cache,
-                                    "past_len": longest_prefix + i,
-                                    "batch_shape": (1, self.max_tokens),
-                                    "reconstruct": False  # Force memory-efficient path
-                                }
-                            )
-
-                        current_len = longest_prefix + remaining_tokens
-
-            if reset:
-                if len(seq_tensor) > 1:
-                    # Process all tokens except the last one
-                    tokens_to_process = seq_tensor[:-1]
-
-                    # Process in chunks if the number of tokens is large
-                    current_len = 0
-                    for i in range(0, tokens_to_process.shape[0], max_chunk_size):
-                        chunk = tokens_to_process[i:i + max_chunk_size]
-                        self.ex_model.forward(
-                            input_ids=chunk.view(1, -1),
-                            params={
-                                "attn_mode": "flash_attn",
-                                "cache": ex_cache,
-                                "past_len": current_len,
-                                "batch_shape": (1, self.max_tokens),
-                                "reconstruct": False  # Force memory-efficient path
-                            }
-                        )
-                        current_len += chunk.shape[0]
-                else:
-                    current_len = 0
-
-            # Process the last token and get logits
-            logits = self.ex_model.forward(
-                input_ids=seq_tensor[-1:].view(1, -1),
-                params={
-                    "attn_mode": "flash_attn",
-                    "cache": ex_cache,
-                    "past_len": current_len,
-                    "batch_shape": (1, self.max_tokens),
-                    "reconstruct": False  # Force memory-efficient path
-                }
-            ).to(input_ids.device).float()
+            # If it's an efficient continuation, process only the new tokens
+            if not reset:
+                params["past_len"] = past_seq.shape[0]
+                tokens_to_process = seq_tensor[past_seq.shape[0]:]
+            # Otherwise, process the whole sequence from scratch
+            else:
+                tokens_to_process = seq_tensor
+
+            # Process all but the last token of the sequence/sub-sequence
+            if tokens_to_process.shape[0] > 1:
+                prefix_to_process = tokens_to_process[:-1]
+
+                # Process in chunks if the number of tokens is large
+                for i in range(0, prefix_to_process.shape[0], max_chunk_size):
+                    chunk = prefix_to_process[i:i + max_chunk_size]
+                    self.ex_model.forward(input_ids=chunk.view(1, -1), params=params)
+                    params["past_len"] += chunk.shape[0]
+
+            # Process the last token to get logits
+            last_token = tokens_to_process[-1:].view(1, -1)
+            logits = self.ex_model.forward(input_ids=last_token, params=params).to(input_ids.device).float()
         else:
             # When processing with labels, handle as a complete sequence
-            # Process in chunks if the number of tokens is large
-            tokens_to_process = seq_tensor
-            all_logits = None
-
-            for i in range(0, tokens_to_process.shape[0], max_chunk_size):
-                chunk = tokens_to_process[i:i + max_chunk_size]
-                chunk_logits = self.ex_model.forward(
-                    input_ids=chunk.view(1, -1),
-                    params={
-                        "attn_mode": "flash_attn_nc",  # No caching for training
-                        "reconstruct": False  # Force memory-efficient path
-                    }
-                ).float()
-
-                if all_logits is None:
-                    all_logits = chunk_logits
-                else:
-                    all_logits = torch.cat([all_logits, chunk_logits], dim=1)
-
-            logits = all_logits
+            params["attn_mode"] = "flash_attn_nc"
+            logits = self.ex_model.forward(input_ids=seq_tensor.view(1,-1), params=params).float()
+
 
         if is_negative:
             self.past_seq_negative = seq_tensor

From e0f0fae59dc4c5fcb789744182d20f0f73061129 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Oct 2025 13:03:48 -0700
Subject: [PATCH 1023/1701] Exllamav3: Add fla to requirements for qwen3-next

---
 requirements/full/requirements.txt               | 1 +
 requirements/full/requirements_apple_intel.txt   | 2 --
 requirements/full/requirements_apple_silicon.txt | 2 --
 requirements/full/requirements_noavx2.txt        | 1 +
 4 files changed, 2 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index d08209f67b..09527a66dd 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -5,6 +5,7 @@ colorama
 datasets
 einops
 fastapi==0.112.4
+flash-linear-attention==0.3.2
 gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 425440309e..0bb837ba93 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -36,5 +36,3 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7-py3-none-any.whl
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index dda91eac11..514c066200 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,5 +37,3 @@ tiktoken
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7-py3-none-any.whl
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 9143542cee..364c4964da 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -5,6 +5,7 @@ colorama
 datasets
 einops
 fastapi==0.112.4
+flash-linear-attention==0.3.2
 gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6

From 977ffbaa04425cda291ad317aec99fd963982558 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Oct 2025 15:53:14 -0700
Subject: [PATCH 1024/1701] Update exllamav3 to 0.0.8

---
 requirements/full/requirements.txt        | 4 ++--
 requirements/full/requirements_noavx2.txt | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 09527a66dd..603283735c 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -38,8 +38,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 364c4964da..9d4956c7ff 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -38,8 +38,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From af3c70651c9f9dc543339ada09d90d19357fc7d3 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Thu, 9 Oct 2025 19:53:34 -0300
Subject: [PATCH 1025/1701] Update bitsandbytes requirement in
 /requirements/full (#7255)

---
 requirements/full/requirements.txt        | 2 +-
 requirements/full/requirements_noavx2.txt | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 603283735c..2fd0b14388 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -1,6 +1,6 @@
 accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
-bitsandbytes==0.46.*
+bitsandbytes==0.48.*
 colorama
 datasets
 einops
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 9d4956c7ff..f50f89a105 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -1,6 +1,6 @@
 accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
-bitsandbytes==0.46.*
+bitsandbytes==0.48.*
 colorama
 datasets
 einops

From 748007f6eef39e432193fe2908a0273b25c791ba Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Oct 2025 16:50:00 -0700
Subject: [PATCH 1026/1701] Revert "Update exllamav3 to 0.0.8"

This reverts commit 977ffbaa04425cda291ad317aec99fd963982558.
---
 requirements/full/requirements.txt        | 4 ++--
 requirements/full/requirements_noavx2.txt | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 2fd0b14388..97eac769ae 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -38,8 +38,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index f50f89a105..9de9e65aaa 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -38,8 +38,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 0d03813e98e0093fe78b60f37126e1aa21ac12f1 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Thu, 9 Oct 2025 21:01:13 -0300
Subject: [PATCH 1027/1701] Update modules/chat.py

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
---
 modules/chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index a52d96af3a..9857479a82 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1196,7 +1196,7 @@ def find_all_histories_with_first_prompts(state):
         if re.match(r'^[0-9]{8}-[0-9]{2}-[0-9]{2}-[0-9]{2}$', filename):
             first_prompt = ""
             if data and 'visible' in data and len(data['visible']) > 0:
-                if len(data['internal']) > 1 and data['internal'][0][0] == '<|BEGIN-VISIBLE-CHAT|>':
+                if len(data['internal']) > 0 and data['internal'][0][0] == '<|BEGIN-VISIBLE-CHAT|>':
                     if len(data['visible']) > 1:
                         first_prompt = html.unescape(data['visible'][1][0])
                     elif i == 0:

From bf5d85c9227d115f02fbc083dfe4e5cf582601ea Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 9 Oct 2025 17:22:41 -0700
Subject: [PATCH 1028/1701] Revert "Downloader: Gracefully handle '416 Range
 Not Satisfiable' when continuing downloads"

This reverts commit 1aa2b924d2b2fa43f1c7a46f28599c8025e779e1.
---
 download-model.py | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/download-model.py b/download-model.py
index f9bca97b3f..c0a3aa36b5 100644
--- a/download-model.py
+++ b/download-model.py
@@ -255,11 +255,6 @@ def get_single_file(self, url, output_folder, start_from_scratch=False):
                         mode = 'ab'
 
                     with session.get(url, stream=True, headers=headers, timeout=30) as r:
-                        # Assume error 416 (range unsatisfiable) means the download is complete.
-                        # Small size differences can happen in Git LFS
-                        if mode == 'ab' and r.status_code == 416:
-                            break
-
                         r.raise_for_status()
                         total_size_from_stream = int(r.headers.get('content-length', 0))
                         if mode == 'ab':

From 25360387ec6bef6e9e71267b38ef1ca452f6976a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 10 Oct 2025 08:24:35 -0700
Subject: [PATCH 1029/1701] Downloader: Fix resuming downloads after HF moved
 to Xet

---
 download-model.py | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)

diff --git a/download-model.py b/download-model.py
index c0a3aa36b5..f17e1285ba 100644
--- a/download-model.py
+++ b/download-model.py
@@ -242,9 +242,19 @@ def get_single_file(self, url, output_folder, start_from_scratch=False):
                 try:
                     if output_path.exists() and not start_from_scratch:
                         current_file_size_on_disk = output_path.stat().st_size
-                        r_head = session.head(url, timeout=20)
-                        r_head.raise_for_status()
-                        total_size = int(r_head.headers.get('content-length', 0))
+
+                        # Make a HEAD request without following redirects to get metadata first
+                        r_head = session.head(url, timeout=20, allow_redirects=True)
+                        r_head.raise_for_status() # Will raise an error for 4xx or 5xx status codes
+
+                        # Check for the new 'x-linked-size' header from Hugging Face
+                        if 'x-linked-size' in r_head.headers:
+                            total_size = int(r_head.headers['x-linked-size'])
+                        # Fallback to the old 'content-length' just in case
+                        elif 'content-length' in r_head.headers:
+                            total_size = int(r_head.headers.get('content-length', 0))
+                        else:
+                            total_size = 0
 
                         if current_file_size_on_disk >= total_size and total_size > 0:
                             if self.progress_queue is not None and total_size > 0:

From 5d734cc7cae7725345d03423505e0238a98ea843 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 10 Oct 2025 12:54:54 -0700
Subject: [PATCH 1030/1701] Remove unused CSS

---
 css/main.css | 23 +----------------------
 1 file changed, 1 insertion(+), 22 deletions(-)

diff --git a/css/main.css b/css/main.css
index c7ee57da0c..fd79d24c3e 100644
--- a/css/main.css
+++ b/css/main.css
@@ -22,10 +22,6 @@
     font-style: italic;
 }
 
-.tabs.svelte-710i53 {
-    margin-top: 0
-}
-
 .padded.svelte-12cmxck {
     padding: 3px 0;
 }
@@ -126,10 +122,6 @@ span.math.inline {
     vertical-align: baseline !important;
 }
 
-div.svelte-15lo0d8 > *, div.svelte-15lo0d8 > .form > * {
-    flex-wrap: nowrap;
-}
-
 gradio-app > :first-child {
     padding: 0 !important;
 }
@@ -325,7 +317,7 @@ audio {
 /* ----------------------------------------------
   Chat tab
 ---------------------------------------------- */
-.h-\[40dvh\], .wrap.svelte-byatnx.svelte-byatnx.svelte-byatnx {
+.h-\[40dvh\] {
     height: 66.67dvh
 }
 
@@ -338,23 +330,10 @@ audio {
     width: unset
 }
 
-div.svelte-362y77>*, div.svelte-362y77>.form>* {
-    flex-wrap: nowrap
-}
-
 .pending.svelte-1ed2p3z {
     opacity: 1;
 }
 
-.wrap.svelte-6roggh.svelte-6roggh {
-    max-height: 92.5%;
-}
-
-/* This is for the microphone button in the whisper extension */
-.sm.svelte-1ipelgc {
-    width: 100%;
-}
-
 #chat-tab {
     padding: 0;
 }

From 78ff21d512565b437223befaecdbe72b30b0f6ba Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 10 Oct 2025 15:19:27 -0700
Subject: [PATCH 1031/1701] Organize the --help message

---
 modules/shared.py | 58 +++++++++++++++++++++++------------------------
 1 file changed, 29 insertions(+), 29 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 4daf43c9aa..af26f185c7 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -50,6 +50,35 @@
 group = parser.add_argument_group('Model loader')
 group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF, ExLlamav2, TensorRT-LLM.')
 
+# Cache
+group = parser.add_argument_group('Context and cache')
+group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=8192, metavar='N', help='Context size in tokens.')
+group.add_argument('--cache-type', '--cache_type', type=str, default='fp16', metavar='N', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).')
+
+# Speculative decoding
+group = parser.add_argument_group('Speculative decoding')
+group.add_argument('--model-draft', type=str, default=None, help='Path to the draft model for speculative decoding.')
+group.add_argument('--draft-max', type=int, default=4, help='Number of tokens to draft for speculative decoding.')
+group.add_argument('--gpu-layers-draft', type=int, default=256, help='Number of layers to offload to the GPU for the draft model.')
+group.add_argument('--device-draft', type=str, default=None, help='Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1')
+group.add_argument('--ctx-size-draft', type=int, default=0, help='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
+
+# llama.cpp
+group = parser.add_argument_group('llama.cpp')
+group.add_argument('--gpu-layers', '--n-gpu-layers', type=int, default=256, metavar='N', help='Number of layers to offload to the GPU.')
+group.add_argument('--mmproj', type=str, default=None, help='Path to the mmproj file for vision models.')
+group.add_argument('--streaming-llm', action='store_true', help='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
+group.add_argument('--tensor-split', type=str, default=None, help='Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.')
+group.add_argument('--row-split', action='store_true', help='Split the model by rows across GPUs. This may improve multi-gpu performance.')
+group.add_argument('--no-mmap', action='store_true', help='Prevent mmap from being used.')
+group.add_argument('--mlock', action='store_true', help='Force the system to keep the model in RAM.')
+group.add_argument('--no-kv-offload', action='store_true', help='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
+group.add_argument('--batch-size', type=int, default=256, help='Maximum number of prompt tokens to batch together when calling llama_eval.')
+group.add_argument('--threads', type=int, default=0, help='Number of threads to use.')
+group.add_argument('--threads-batch', type=int, default=0, help='Number of threads to use for batches/prompt processing.')
+group.add_argument('--numa', action='store_true', help='Activate NUMA task allocation for llama.cpp.')
+group.add_argument('--extra-flags', type=str, default=None, help='Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"')
+
 # Transformers/Accelerate
 group = parser.add_argument_group('Transformers/Accelerate')
 group.add_argument('--cpu', action='store_true', help='Use the CPU to generate text. Warning: Training on CPU is extremely slow.')
@@ -71,35 +100,6 @@
 group.add_argument('--compute_dtype', type=str, default='float16', help='compute dtype for 4-bit. Valid options: bfloat16, float16, float32.')
 group.add_argument('--quant_type', type=str, default='nf4', help='quant_type for 4-bit. Valid options: nf4, fp4.')
 
-# llama.cpp
-group = parser.add_argument_group('llama.cpp')
-group.add_argument('--threads', type=int, default=0, help='Number of threads to use.')
-group.add_argument('--threads-batch', type=int, default=0, help='Number of threads to use for batches/prompt processing.')
-group.add_argument('--batch-size', type=int, default=256, help='Maximum number of prompt tokens to batch together when calling llama_eval.')
-group.add_argument('--no-mmap', action='store_true', help='Prevent mmap from being used.')
-group.add_argument('--mlock', action='store_true', help='Force the system to keep the model in RAM.')
-group.add_argument('--gpu-layers', '--n-gpu-layers', type=int, default=256, metavar='N', help='Number of layers to offload to the GPU.')
-group.add_argument('--tensor-split', type=str, default=None, help='Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.')
-group.add_argument('--numa', action='store_true', help='Activate NUMA task allocation for llama.cpp.')
-group.add_argument('--no-kv-offload', action='store_true', help='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
-group.add_argument('--row-split', action='store_true', help='Split the model by rows across GPUs. This may improve multi-gpu performance.')
-group.add_argument('--extra-flags', type=str, default=None, help='Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"')
-group.add_argument('--streaming-llm', action='store_true', help='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
-group.add_argument('--mmproj', type=str, default=None, help='Path to the mmproj file for vision models.')
-
-# Cache
-group = parser.add_argument_group('Context and cache')
-group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=8192, metavar='N', help='Context size in tokens.')
-group.add_argument('--cache-type', '--cache_type', type=str, default='fp16', metavar='N', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).')
-
-# Speculative decoding
-group = parser.add_argument_group('Speculative decoding')
-group.add_argument('--model-draft', type=str, default=None, help='Path to the draft model for speculative decoding.')
-group.add_argument('--draft-max', type=int, default=4, help='Number of tokens to draft for speculative decoding.')
-group.add_argument('--gpu-layers-draft', type=int, default=256, help='Number of layers to offload to the GPU for the draft model.')
-group.add_argument('--device-draft', type=str, default=None, help='Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1')
-group.add_argument('--ctx-size-draft', type=int, default=0, help='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
-
 # ExLlamaV3
 group = parser.add_argument_group('ExLlamaV3')
 group.add_argument('--enable-tp', '--enable_tp', action='store_true', help='Enable Tensor Parallelism (TP) to split the model across GPUs.')

From a74596374d3c924e62bdaf9bcd518d5cbc9f489f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 10 Oct 2025 17:51:31 -0700
Subject: [PATCH 1032/1701] Reapply "Update exllamav3 to 0.0.8"

This reverts commit 748007f6eef39e432193fe2908a0273b25c791ba.
---
 requirements/full/requirements.txt        | 4 ++--
 requirements/full/requirements_noavx2.txt | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 97eac769ae..2fd0b14388 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -38,8 +38,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 9de9e65aaa..f50f89a105 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -38,8 +38,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.7/exllamav3-0.0.7+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From dd0b0034930b4dbda242b9f19bcc5693a2e600f1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 10 Oct 2025 17:52:16 -0700
Subject: [PATCH 1033/1701] Bump pydantic to 2.11.0

---
 requirements/full/requirements.txt                     | 2 +-
 requirements/full/requirements_amd.txt                 | 2 +-
 requirements/full/requirements_amd_noavx2.txt          | 2 +-
 requirements/full/requirements_apple_intel.txt         | 2 +-
 requirements/full/requirements_apple_silicon.txt       | 2 +-
 requirements/full/requirements_cpu_only.txt            | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt     | 2 +-
 requirements/full/requirements_noavx2.txt              | 2 +-
 requirements/full/requirements_nowheels.txt            | 2 +-
 requirements/portable/requirements.txt                 | 2 +-
 requirements/portable/requirements_apple_intel.txt     | 2 +-
 requirements/portable/requirements_apple_silicon.txt   | 2 +-
 requirements/portable/requirements_cpu_only.txt        | 2 +-
 requirements/portable/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/portable/requirements_noavx2.txt          | 2 +-
 requirements/portable/requirements_nowheels.txt        | 2 +-
 requirements/portable/requirements_vulkan.txt          | 2 +-
 requirements/portable/requirements_vulkan_noavx2.txt   | 2 +-
 18 files changed, 18 insertions(+), 18 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 2fd0b14388..1a6a1953bd 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -15,7 +15,7 @@ pandas
 peft==0.17.*
 Pillow>=9.5.0
 psutil
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index b3b0005e09..8a5c431b1b 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -13,7 +13,7 @@ pandas
 peft==0.17.*
 Pillow>=9.5.0
 psutil
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 5e0d375e85..7338468b21 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -13,7 +13,7 @@ pandas
 peft==0.17.*
 Pillow>=9.5.0
 psutil
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 0bb837ba93..f65aa84523 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -13,7 +13,7 @@ pandas
 peft==0.17.*
 Pillow>=9.5.0
 psutil
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 514c066200..c74fbf5b95 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -13,7 +13,7 @@ pandas
 peft==0.17.*
 Pillow>=9.5.0
 psutil
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index f68fdd9da1..4b25bcee39 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -13,7 +13,7 @@ pandas
 peft==0.17.*
 Pillow>=9.5.0
 psutil
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index b40f1af216..de15a37488 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -13,7 +13,7 @@ pandas
 peft==0.17.*
 Pillow>=9.5.0
 psutil
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index f50f89a105..aeb15d828c 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -15,7 +15,7 @@ pandas
 peft==0.17.*
 Pillow>=9.5.0
 psutil
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 3bd20dd9c8..bca8e478bb 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -13,7 +13,7 @@ pandas
 peft==0.17.*
 Pillow>=9.5.0
 psutil
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 7a38b1e6eb..5da97792cb 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -5,7 +5,7 @@ html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 047d1c5476..1f43809b86 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -5,7 +5,7 @@ html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 5c8ae4df94..5f26a73d4a 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -5,7 +5,7 @@ html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index f41efd5866..db23729c56 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -5,7 +5,7 @@ html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 6915805096..7b57af587f 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -5,7 +5,7 @@ html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index ca66098c37..35768ac44f 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -5,7 +5,7 @@ html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index be624bb1e5..d2ac54636d 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -5,7 +5,7 @@ html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 36aff3617a..7317784e01 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -5,7 +5,7 @@ html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index be7170e3d5..dfc0d2fc72 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -5,7 +5,7 @@ html2text==2025.4.15
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-pydantic==2.8.2
+pydantic==2.11.0
 PyPDF2==3.0.1
 python-docx==1.1.2
 pyyaml

From 1831b3fb51ee055582d6db9018d51f921c46adde Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 10 Oct 2025 17:53:34 -0700
Subject: [PATCH 1034/1701] Use my custom gradio_client build (small changes to
 work with pydantic 2.11)

---
 requirements/full/requirements.txt                     | 5 ++++-
 requirements/full/requirements_amd.txt                 | 5 ++++-
 requirements/full/requirements_amd_noavx2.txt          | 5 ++++-
 requirements/full/requirements_apple_intel.txt         | 5 ++++-
 requirements/full/requirements_apple_silicon.txt       | 5 ++++-
 requirements/full/requirements_cpu_only.txt            | 5 ++++-
 requirements/full/requirements_cpu_only_noavx2.txt     | 5 ++++-
 requirements/full/requirements_noavx2.txt              | 5 ++++-
 requirements/full/requirements_nowheels.txt            | 5 ++++-
 requirements/portable/requirements.txt                 | 5 ++++-
 requirements/portable/requirements_apple_intel.txt     | 7 +++++--
 requirements/portable/requirements_apple_silicon.txt   | 5 ++++-
 requirements/portable/requirements_cpu_only.txt        | 5 ++++-
 requirements/portable/requirements_cpu_only_noavx2.txt | 5 ++++-
 requirements/portable/requirements_noavx2.txt          | 5 ++++-
 requirements/portable/requirements_nowheels.txt        | 5 ++++-
 requirements/portable/requirements_vulkan.txt          | 5 ++++-
 requirements/portable/requirements_vulkan_noavx2.txt   | 5 ++++-
 18 files changed, 73 insertions(+), 19 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 1a6a1953bd..b2e5af6668 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -6,7 +6,6 @@ datasets
 einops
 fastapi==0.112.4
 flash-linear-attention==0.3.2
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -30,6 +29,10 @@ triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 8a5c431b1b..d32a493023 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -4,7 +4,6 @@ colorama
 datasets
 einops
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -28,6 +27,10 @@ triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 7338468b21..8a89f910ad 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -4,7 +4,6 @@ colorama
 datasets
 einops
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -28,6 +27,10 @@ triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index f65aa84523..28078f532e 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -4,7 +4,6 @@ colorama
 datasets
 einops
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -28,6 +27,10 @@ triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index c74fbf5b95..52113495b8 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -4,7 +4,6 @@ colorama
 datasets
 einops
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -28,6 +27,10 @@ triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 4b25bcee39..ec36f227ec 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -4,7 +4,6 @@ colorama
 datasets
 einops
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -28,6 +27,10 @@ triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index de15a37488..68041f71ce 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -4,7 +4,6 @@ colorama
 datasets
 einops
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -28,6 +27,10 @@ triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index aeb15d828c..f8b12c70a2 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -6,7 +6,6 @@ datasets
 einops
 fastapi==0.112.4
 flash-linear-attention==0.3.2
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -30,6 +29,10 @@ triton-windows==3.3.1.post19; platform_system == "Windows"
 tqdm
 wandb
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index bca8e478bb..47b0dcbd4d 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -4,7 +4,6 @@ colorama
 datasets
 einops
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -28,6 +27,10 @@ triton-windows==3.2.0.post19; platform_system == "Windows"
 tqdm
 wandb
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 5da97792cb..94bd3d5332 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -13,6 +12,10 @@ requests
 rich
 tqdm
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 1f43809b86..8cdeacb8c5 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -13,6 +12,10 @@ requests
 rich
 tqdm
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
@@ -21,4 +24,4 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
\ No newline at end of file
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 5f26a73d4a..18bdde4035 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -13,6 +12,10 @@ requests
 rich
 tqdm
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index db23729c56..88a11b0fb6 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -13,6 +12,10 @@ requests
 rich
 tqdm
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 7b57af587f..786cb20885 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -13,6 +12,10 @@ requests
 rich
 tqdm
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 35768ac44f..66a96e4090 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -13,6 +12,10 @@ requests
 rich
 tqdm
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index d2ac54636d..1a7b568393 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -13,6 +12,10 @@ requests
 rich
 tqdm
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 7317784e01..f486310b1b 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -13,6 +12,10 @@ requests
 rich
 tqdm
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index dfc0d2fc72..6ed9599ac8 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-gradio==4.37.*
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
@@ -13,6 +12,10 @@ requests
 rich
 tqdm
 
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5

From c7dd920dc87472c1d4545c0e10bd6270c6a83fb8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 11 Oct 2025 14:12:05 -0700
Subject: [PATCH 1035/1701] Fix metadata leaking into branched chats

---
 modules/chat.py | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index 9857479a82..8db7953a76 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1831,6 +1831,10 @@ def handle_branch_chat_click(state):
         history = state['history']
         history['visible'] = history['visible'][:branch_from_index + 1]
         history['internal'] = history['internal'][:branch_from_index + 1]
+        # Prune the metadata dictionary to remove entries beyond the branch point
+        if 'metadata' in history:
+            history['metadata'] = {k: v for k, v in history['metadata'].items() if int(k.split('_')[-1]) <= branch_from_index}
+
     new_unique_id = datetime.now().strftime('%Y%m%d-%H-%M-%S')
     save_history(history, new_unique_id, state['character_menu'], state['mode'])
 

From 655c3e86e310414f806154f93098a1cd68382981 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 11 Oct 2025 17:00:25 -0700
Subject: [PATCH 1036/1701] Fix "continue" missing an initial space in
 chat-instruct/chat modes

---
 modules/chat.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 8db7953a76..55984d7a40 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -880,7 +880,9 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
 
         # Extract the reply
         if state['mode'] in ['chat', 'chat-instruct']:
-            reply = reply.lstrip()
+            if not _continue:
+                reply = reply.lstrip()
+
             if reply.startswith(state['name2'] + ':'):
                 reply = reply[len(state['name2'] + ':'):]
             elif reply.startswith(state['name1'] + ':'):

From 968c79db062bf3dcc2e435082702c324be7f7c2e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 11 Oct 2025 17:20:10 -0700
Subject: [PATCH 1037/1701] Minor README fix (closes #7251)

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index c3f7d8a65b..c4729d5590 100644
--- a/README.md
+++ b/README.md
@@ -164,7 +164,7 @@ conda install -y -c "nvidia/label/cuda-12.8.1" cuda
 ```
 git clone https://github.com/oobabooga/text-generation-webui
 cd text-generation-webui
-pip install -r 
+pip install -r requirements/full/
 ```
 
 Requirements file to use:

From 611399e089f0c6b5f4b11760f7d87dbb0d8d6378 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 11 Oct 2025 17:22:48 -0700
Subject: [PATCH 1038/1701] Update README

---
 README.md | 75 +++++++++++++++++++++++++++----------------------------
 1 file changed, 37 insertions(+), 38 deletions(-)

diff --git a/README.md b/README.md
index c4729d5590..d350d9596c 100644
--- a/README.md
+++ b/README.md
@@ -240,16 +240,16 @@ List of command-line flags
 
 ```txt
 usage: server.py [-h] [--multi-user] [--model MODEL] [--lora LORA [LORA ...]] [--model-dir MODEL_DIR] [--lora-dir LORA_DIR] [--model-menu] [--settings SETTINGS]
-                 [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR]
-                 [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] [--load-in-4bit] [--use_double_quant]
-                 [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--flash-attn] [--threads THREADS] [--threads-batch THREADS_BATCH] [--batch-size BATCH_SIZE] [--no-mmap] [--mlock]
-                 [--gpu-layers N] [--tensor-split TENSOR_SPLIT] [--numa] [--no-kv-offload] [--row-split] [--extra-flags EXTRA_FLAGS] [--streaming-llm] [--mmproj MMPROJ] [--ctx-size N] [--cache-type N]
-                 [--model-draft MODEL_DRAFT] [--draft-max DRAFT_MAX] [--gpu-layers-draft GPU_LAYERS_DRAFT] [--device-draft DEVICE_DRAFT] [--ctx-size-draft CTX_SIZE_DRAFT] [--enable-tp]
-                 [--tp-backend TP_BACKEND] [--gpu-split GPU_SPLIT] [--autosplit] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa] [--num_experts_per_token N] [--cpp-runner] [--deepspeed]
-                 [--nvme-offload-dir NVME_OFFLOAD_DIR] [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB] [--listen]
-                 [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE]
-                 [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors] [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY]
-                 [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4] [--nowebui]
+                 [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--ctx-size N] [--cache-type N] [--model-draft MODEL_DRAFT]
+                 [--draft-max DRAFT_MAX] [--gpu-layers-draft GPU_LAYERS_DRAFT] [--device-draft DEVICE_DRAFT] [--ctx-size-draft CTX_SIZE_DRAFT] [--gpu-layers N] [--mmproj MMPROJ] [--streaming-llm]
+                 [--tensor-split TENSOR_SPLIT] [--row-split] [--no-mmap] [--mlock] [--no-kv-offload] [--batch-size BATCH_SIZE] [--threads THREADS] [--threads-batch THREADS_BATCH] [--numa]
+                 [--extra-flags EXTRA_FLAGS] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code]
+                 [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE]
+                 [--enable-tp] [--tp-backend TP_BACKEND] [--gpu-split GPU_SPLIT] [--autosplit] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa] [--num_experts_per_token N] [--cpp-runner]
+                 [--deepspeed] [--nvme-offload-dir NVME_OFFLOAD_DIR] [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB]
+                 [--listen] [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH]
+                 [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors] [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT]
+                 [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4] [--nowebui]
 
 Text Generation Web UI
 
@@ -273,6 +273,33 @@ Model loader:
   --loader LOADER                           Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF, ExLlamav2,
                                             TensorRT-LLM.
 
+Context and cache:
+  --ctx-size N, --n_ctx N, --max_seq_len N  Context size in tokens.
+  --cache-type N, --cache_type N            KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits
+                                            separately, e.g. q4_q8).
+
+Speculative decoding:
+  --model-draft MODEL_DRAFT                 Path to the draft model for speculative decoding.
+  --draft-max DRAFT_MAX                     Number of tokens to draft for speculative decoding.
+  --gpu-layers-draft GPU_LAYERS_DRAFT       Number of layers to offload to the GPU for the draft model.
+  --device-draft DEVICE_DRAFT               Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1
+  --ctx-size-draft CTX_SIZE_DRAFT           Size of the prompt context for the draft model. If 0, uses the same as the main model.
+
+llama.cpp:
+  --gpu-layers N, --n-gpu-layers N          Number of layers to offload to the GPU.
+  --mmproj MMPROJ                           Path to the mmproj file for vision models.
+  --streaming-llm                           Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.
+  --tensor-split TENSOR_SPLIT               Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.
+  --row-split                               Split the model by rows across GPUs. This may improve multi-gpu performance.
+  --no-mmap                                 Prevent mmap from being used.
+  --mlock                                   Force the system to keep the model in RAM.
+  --no-kv-offload                           Do not offload the K, Q, V to the GPU. This saves VRAM but reduces the performance.
+  --batch-size BATCH_SIZE                   Maximum number of prompt tokens to batch together when calling llama_eval.
+  --threads THREADS                         Number of threads to use.
+  --threads-batch THREADS_BATCH             Number of threads to use for batches/prompt processing.
+  --numa                                    Activate NUMA task allocation for llama.cpp.
+  --extra-flags EXTRA_FLAGS                 Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"
+
 Transformers/Accelerate:
   --cpu                                     Use the CPU to generate text. Warning: Training on CPU is extremely slow.
   --cpu-memory CPU_MEMORY                   Maximum CPU memory in GiB. Use this for CPU offloading.
@@ -292,34 +319,6 @@ bitsandbytes 4-bit:
   --compute_dtype COMPUTE_DTYPE             compute dtype for 4-bit. Valid options: bfloat16, float16, float32.
   --quant_type QUANT_TYPE                   quant_type for 4-bit. Valid options: nf4, fp4.
 
-llama.cpp:
-  --flash-attn                              Use flash-attention.
-  --threads THREADS                         Number of threads to use.
-  --threads-batch THREADS_BATCH             Number of threads to use for batches/prompt processing.
-  --batch-size BATCH_SIZE                   Maximum number of prompt tokens to batch together when calling llama_eval.
-  --no-mmap                                 Prevent mmap from being used.
-  --mlock                                   Force the system to keep the model in RAM.
-  --gpu-layers N, --n-gpu-layers N          Number of layers to offload to the GPU.
-  --tensor-split TENSOR_SPLIT               Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.
-  --numa                                    Activate NUMA task allocation for llama.cpp.
-  --no-kv-offload                           Do not offload the K, Q, V to the GPU. This saves VRAM but reduces the performance.
-  --row-split                               Split the model by rows across GPUs. This may improve multi-gpu performance.
-  --extra-flags EXTRA_FLAGS                 Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"
-  --streaming-llm                           Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.
-  --mmproj MMPROJ                           Path to the mmproj file for vision models.
-
-Context and cache:
-  --ctx-size N, --n_ctx N, --max_seq_len N  Context size in tokens.
-  --cache-type N, --cache_type N            KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits
-                                            separately, e.g. q4_q8).
-
-Speculative decoding:
-  --model-draft MODEL_DRAFT                 Path to the draft model for speculative decoding.
-  --draft-max DRAFT_MAX                     Number of tokens to draft for speculative decoding.
-  --gpu-layers-draft GPU_LAYERS_DRAFT       Number of layers to offload to the GPU for the draft model.
-  --device-draft DEVICE_DRAFT               Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1
-  --ctx-size-draft CTX_SIZE_DRAFT           Size of the prompt context for the draft model. If 0, uses the same as the main model.
-
 ExLlamaV3:
   --enable-tp, --enable_tp                  Enable Tensor Parallelism (TP) to split the model across GPUs.
   --tp-backend TP_BACKEND                   The backend for tensor parallelism. Valid options: native, nccl. Default: native.

From 308e726e11887c504569e48ac93f43bfdd6b8368 Mon Sep 17 00:00:00 2001
From: mamei16 
Date: Mon, 13 Oct 2025 04:00:11 +0200
Subject: [PATCH 1039/1701] log error when llama-server request exceeds context
 size (#7263)

---
 modules/llama_cpp_server.py | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 38589cf290..27890d8c90 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -200,7 +200,10 @@ def generate_with_streaming(self, prompt, state):
         # Make the generation request
         response = self.session.post(url, json=payload, stream=True)
         try:
-            response.raise_for_status()  # Raise an exception for HTTP errors
+            if response.status_code == 400 and response.json()["error"]["type"] == "exceed_context_size_error":
+                logger.error("The request exceeds the available context size, try increasing it")
+            else:
+                response.raise_for_status()  # Raise an exception for HTTP errors
 
             full_text = ""
 

From 047855c591bc7293d85daee2ebd2adae40212c03 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 13 Oct 2025 15:32:03 -0700
Subject: [PATCH 1040/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 6 +++---
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 35 insertions(+), 35 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index b2e5af6668..0031696f33 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -39,8 +39,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index d32a493023..ec1bae230d 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,7 +37,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 8a89f910ad..748d1241fe 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -37,7 +37,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 28078f532e..96dda146ea 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 52113495b8..d79f326aba 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,6 +37,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index ec36f227ec..ae798bcd20 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 68041f71ce..0dde3cdac3 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index f8b12c70a2..ae671116f5 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -39,8 +39,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 94bd3d5332..a9b900e4ab 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 8cdeacb8c5..b6314e9c5e 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -22,6 +22,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 18bdde4035..f1adf3b162 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -22,6 +22,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 88a11b0fb6..402fc6592e 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 786cb20885..b2cc4866c6 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 66a96e4090..5a922619e1 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index f486310b1b..2fb2d60fe7 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 6ed9599ac8..b60c5807b3 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.49.0/llama_cpp_binaries-0.49.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From efaf2aef3daa84101e36fa6d4b4084acebb3e1ed Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 13 Oct 2025 15:32:25 -0700
Subject: [PATCH 1041/1701] Update exllamav3 to 0.0.9

---
 requirements/full/requirements.txt        | 4 ++--
 requirements/full/requirements_noavx2.txt | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 0031696f33..a38457fb68 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -41,8 +41,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.9/exllamav3-0.0.9+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.9/exllamav3-0.0.9+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index ae671116f5..ca8a242752 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -41,8 +41,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.8/exllamav3-0.0.8+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.9/exllamav3-0.0.9+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.9/exllamav3-0.0.9+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From b5a6904c4ac4049823396090360b6f566f4e4603 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Oct 2025 20:42:37 -0700
Subject: [PATCH 1042/1701] Make --trust-remote-code immutable from the UI/API

---
 modules/loaders.py             | 3 ---
 modules/shared.py              | 9 ++++++++-
 modules/transformers_loader.py | 6 +++---
 modules/ui.py                  | 1 -
 modules/ui_model_menu.py       | 1 -
 5 files changed, 11 insertions(+), 9 deletions(-)

diff --git a/modules/loaders.py b/modules/loaders.py
index fe982ab55f..609a54c6df 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -45,7 +45,6 @@
         'disk',
         'use_double_quant',
         'bf16',
-        'trust_remote_code',
         'no_use_fast',
     ],
     'ExLlamav3_HF': [
@@ -53,7 +52,6 @@
         'cache_type',
         'gpu_split',
         'cfg_cache',
-        'trust_remote_code',
         'no_use_fast',
         'enable_tp',
         'tp_backend',
@@ -82,7 +80,6 @@
         'no_xformers',
         'no_sdpa',
         'cfg_cache',
-        'trust_remote_code',
         'no_use_fast',
     ],
     'ExLlamav2': [
diff --git a/modules/shared.py b/modules/shared.py
index af26f185c7..e54ba654a2 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -174,6 +174,7 @@
 
 
 args = parser.parse_args()
+original_args = copy.deepcopy(args)
 args_defaults = parser.parse_args([])
 
 # Create a mapping of all argument aliases to their canonical names
@@ -295,7 +296,13 @@
 def do_cmd_flags_warnings():
     # Security warnings
     if args.trust_remote_code:
-        logger.warning('trust_remote_code is enabled. This is dangerous.')
+        logger.warning(
+            "The `--trust-remote-code` flag is enabled.\n"
+            "This allows models to execute arbitrary code on your machine.\n\n"
+            "1. Only use with models from sources you fully trust.\n"
+            "2. Set an access password with `--gradio-auth`."
+        )
+
     if 'COLAB_GPU' not in os.environ and not args.nowebui:
         if args.share:
             logger.warning("The gradio \"share link\" feature uses a proprietary executable to create a reverse tunnel. Use it with care.")
diff --git a/modules/transformers_loader.py b/modules/transformers_loader.py
index 7866f44888..f1af1299fa 100644
--- a/modules/transformers_loader.py
+++ b/modules/transformers_loader.py
@@ -123,7 +123,7 @@ def load_tokenizer(model_name, tokenizer_dir=None):
 
         tokenizer = AutoTokenizer.from_pretrained(
             path_to_model,
-            trust_remote_code=shared.args.trust_remote_code,
+            trust_remote_code=shared.original_args.trust_remote_code,
             use_fast=not shared.args.no_use_fast
         )
 
@@ -140,13 +140,13 @@ def load_model_HF(model_name):
         'torch_dtype': torch.bfloat16 if shared.args.bf16 else torch.float16,
     }
 
-    if shared.args.trust_remote_code:
+    if shared.original_args.trust_remote_code:
         params['trust_remote_code'] = True
 
     if shared.args.force_safetensors:
         params['force_safetensors'] = True
 
-    config = AutoConfig.from_pretrained(path_to_model, trust_remote_code=shared.args.trust_remote_code)
+    config = AutoConfig.from_pretrained(path_to_model, trust_remote_code=shared.original_args.trust_remote_code)
 
     if 'chatglm' in model_name.lower():
         LoaderClass = AutoModel
diff --git a/modules/ui.py b/modules/ui.py
index 12f43768aa..76533767a6 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -160,7 +160,6 @@ def list_model_elements():
         'no_sdpa',
         'cfg_cache',
         'cpp_runner',
-        'trust_remote_code',
         'no_use_fast',
         'model_draft',
         'draft_max',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 729700d473..50ada9f90c 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -57,7 +57,6 @@ def create_ui():
                             shared.gradio['autosplit'] = gr.Checkbox(label="autosplit", value=shared.args.autosplit, info='Automatically split the model tensors across the available GPUs.')
                             shared.gradio['enable_tp'] = gr.Checkbox(label="enable_tp", value=shared.args.enable_tp, info='Enable tensor parallelism (TP).')
                             shared.gradio['cpp_runner'] = gr.Checkbox(label="cpp-runner", value=shared.args.cpp_runner, info='Enable inference with ModelRunnerCpp, which is faster than the default ModelRunner.')
-                            shared.gradio['trust_remote_code'] = gr.Checkbox(label="trust-remote-code", value=shared.args.trust_remote_code, info='Set trust_remote_code=True while loading the tokenizer/model. To enable this option, start the web UI with the --trust-remote-code flag.', interactive=shared.args.trust_remote_code)
                             shared.gradio['tensorrt_llm_info'] = gr.Markdown('* TensorRT-LLM has to be installed manually in a separate Python 3.10 environment at the moment. For a guide, consult the description of [this PR](https://github.com/oobabooga/text-generation-webui/pull/5715). \n\n* `ctx_size` is only used when `cpp-runner` is checked.\n\n* `cpp_runner` does not support streaming at the moment.')
 
                             # Multimodal

From c93d567f9728151ceb12b11f3eca297afb4645e6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 15 Oct 2025 06:40:48 -0700
Subject: [PATCH 1043/1701] Update exllamav3 to 0.0.10

---
 requirements/full/requirements.txt        | 4 ++--
 requirements/full/requirements_noavx2.txt | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index a38457fb68..932ae17f3b 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -41,8 +41,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.9/exllamav3-0.0.9+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.9/exllamav3-0.0.9+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index ca8a242752..1c94183026 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -41,8 +41,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.9/exllamav3-0.0.9+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.9/exllamav3-0.0.9+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 163d86344316dc487aa064319d361b9b2e0aaf88 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 15 Oct 2025 11:23:10 -0700
Subject: [PATCH 1044/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 6 +++---
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 35 insertions(+), 35 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 932ae17f3b..0609f69bbe 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -39,8 +39,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index ec1bae230d..73d3518504 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,7 +37,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 748d1241fe..dad354e6ac 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -37,7 +37,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 96dda146ea..ab48c83990 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index d79f326aba..3a341b8a1b 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,6 +37,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index ae798bcd20..3765d329bf 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 0dde3cdac3..a3e4e51631 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 1c94183026..abdd262e64 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -39,8 +39,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index a9b900e4ab..d43007d06c 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index b6314e9c5e..77a66dd8ac 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -22,6 +22,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index f1adf3b162..a2479e9823 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -22,6 +22,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 402fc6592e..8d8e5233eb 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index b2cc4866c6..08b7240c6b 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 5a922619e1..857c5e6672 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 2fb2d60fe7..d2e66763e2 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index b60c5807b3..349272949b 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.50.0/llama_cpp_binaries-0.50.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From c871d9cdbd1b2021b8978f00c1a3b3a1f66d6448 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 15 Oct 2025 13:05:41 -0700
Subject: [PATCH 1045/1701] Revert "Same as
 7f06aec3a1fc2e6924d87035483ce10ce65af058 but for exllamav3_hf"

This reverts commit deb37b821bb13f241dcc3e6aff7cbf41bbae143f.
---
 modules/exllamav3_hf.py | 137 ++++++++++++++++++++++++++--------------
 1 file changed, 89 insertions(+), 48 deletions(-)

diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index c606912bb4..05b473b71f 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -103,12 +103,6 @@ def __call__(self, *args, **kwargs):
         labels = kwargs.get('labels', None)
         past_key_values = kwargs.get('past_key_values', None)
 
-        # Reset the internal sequence state for standalone calls (logit viewer)
-        # or the very first step of a new generation.
-        if past_key_values is None:
-            self.past_seq = None
-            self.past_seq_negative = None
-
         if len(args) > 0:
             if not shared.args.cfg_cache:
                 logger.error("Please enable the cfg-cache option to use CFG with ExLlamav3_HF.")
@@ -125,8 +119,8 @@ def __call__(self, *args, **kwargs):
             ex_cache = self.ex_cache
 
         seq = input_ids[0].tolist()
-        if is_negative and past_key_values is not None and isinstance(past_key_values, list):
-             seq = past_key_values + seq
+        if is_negative and past_key_values is not None:
+            seq = past_key_values + seq
 
         seq_tensor = torch.tensor(seq)
         reset = True
@@ -134,50 +128,97 @@ def __call__(self, *args, **kwargs):
         # Maximum number of tokens to process in a single forward pass
         max_chunk_size = 256
 
-        if past_seq is not None:
-            min_length = min(past_seq.shape[0], seq_tensor.shape[0])
-            indices = torch.nonzero(~torch.eq(past_seq[:min_length], seq_tensor[:min_length]))
-            if len(indices) == 0 and seq_tensor.shape[0] > past_seq.shape[0]:
-                reset = False
-
-        # Create a single `params` dictionary that will be used and modified
-        # in-place across all `forward` calls within this function.
-        params = {
-            "attn_mode": "flash_attn",
-            "cache": ex_cache,
-            "batch_shape": (1, self.max_tokens),
-            "reconstruct": False,
-            "past_len": 0
-        }
-
         # Make the forward call
         if labels is None:
-            # If it's an efficient continuation, process only the new tokens
-            if not reset:
-                params["past_len"] = past_seq.shape[0]
-                tokens_to_process = seq_tensor[past_seq.shape[0]:]
-            # Otherwise, process the whole sequence from scratch
-            else:
-                tokens_to_process = seq_tensor
-
-            # Process all but the last token of the sequence/sub-sequence
-            if tokens_to_process.shape[0] > 1:
-                prefix_to_process = tokens_to_process[:-1]
-
-                # Process in chunks if the number of tokens is large
-                for i in range(0, prefix_to_process.shape[0], max_chunk_size):
-                    chunk = prefix_to_process[i:i + max_chunk_size]
-                    self.ex_model.forward(input_ids=chunk.view(1, -1), params=params)
-                    params["past_len"] += chunk.shape[0]
-
-            # Process the last token to get logits
-            last_token = tokens_to_process[-1:].view(1, -1)
-            logits = self.ex_model.forward(input_ids=last_token, params=params).to(input_ids.device).float()
+            if past_seq is not None:
+                min_length = min(past_seq.shape[0], seq_tensor.shape[0])
+                indices = torch.nonzero(~torch.eq(past_seq[:min_length], seq_tensor[:min_length]))
+                if len(indices) > 0:
+                    longest_prefix = indices[0].item()
+                else:
+                    longest_prefix = min_length
+
+                if longest_prefix > 0:
+                    reset = False
+                    current_len = longest_prefix
+                    remaining_tokens = len(seq_tensor) - longest_prefix - 1
+
+                    if remaining_tokens > 0:
+                        # Process tokens from longest_prefix to second-to-last token
+                        tokens_to_process = seq_tensor[longest_prefix:-1]
+
+                        # Process in chunks if the number of tokens is large
+                        for i in range(0, tokens_to_process.shape[0], max_chunk_size):
+                            chunk = tokens_to_process[i:i + max_chunk_size]
+                            self.ex_model.forward(
+                                input_ids=chunk.view(1, -1),
+                                params={
+                                    "attn_mode": "flash_attn",
+                                    "cache": ex_cache,
+                                    "past_len": longest_prefix + i,
+                                    "batch_shape": (1, self.max_tokens),
+                                    "reconstruct": False  # Force memory-efficient path
+                                }
+                            )
+
+                        current_len = longest_prefix + remaining_tokens
+
+            if reset:
+                if len(seq_tensor) > 1:
+                    # Process all tokens except the last one
+                    tokens_to_process = seq_tensor[:-1]
+
+                    # Process in chunks if the number of tokens is large
+                    current_len = 0
+                    for i in range(0, tokens_to_process.shape[0], max_chunk_size):
+                        chunk = tokens_to_process[i:i + max_chunk_size]
+                        self.ex_model.forward(
+                            input_ids=chunk.view(1, -1),
+                            params={
+                                "attn_mode": "flash_attn",
+                                "cache": ex_cache,
+                                "past_len": current_len,
+                                "batch_shape": (1, self.max_tokens),
+                                "reconstruct": False  # Force memory-efficient path
+                            }
+                        )
+                        current_len += chunk.shape[0]
+                else:
+                    current_len = 0
+
+            # Process the last token and get logits
+            logits = self.ex_model.forward(
+                input_ids=seq_tensor[-1:].view(1, -1),
+                params={
+                    "attn_mode": "flash_attn",
+                    "cache": ex_cache,
+                    "past_len": current_len,
+                    "batch_shape": (1, self.max_tokens),
+                    "reconstruct": False  # Force memory-efficient path
+                }
+            ).to(input_ids.device).float()
         else:
             # When processing with labels, handle as a complete sequence
-            params["attn_mode"] = "flash_attn_nc"
-            logits = self.ex_model.forward(input_ids=seq_tensor.view(1,-1), params=params).float()
-
+            # Process in chunks if the number of tokens is large
+            tokens_to_process = seq_tensor
+            all_logits = None
+
+            for i in range(0, tokens_to_process.shape[0], max_chunk_size):
+                chunk = tokens_to_process[i:i + max_chunk_size]
+                chunk_logits = self.ex_model.forward(
+                    input_ids=chunk.view(1, -1),
+                    params={
+                        "attn_mode": "flash_attn_nc",  # No caching for training
+                        "reconstruct": False  # Force memory-efficient path
+                    }
+                ).float()
+
+                if all_logits is None:
+                    all_logits = chunk_logits
+                else:
+                    all_logits = torch.cat([all_logits, chunk_logits], dim=1)
+
+            logits = all_logits
 
         if is_negative:
             self.past_seq_negative = seq_tensor

From a156ebbf76f9e56a8fd17c7355d337c6082c33eb Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 15 Oct 2025 13:15:01 -0700
Subject: [PATCH 1046/1701] Lint

---
 download-model.py       | 2 +-
 modules/exllamav3_hf.py | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/download-model.py b/download-model.py
index f17e1285ba..756d529f73 100644
--- a/download-model.py
+++ b/download-model.py
@@ -245,7 +245,7 @@ def get_single_file(self, url, output_folder, start_from_scratch=False):
 
                         # Make a HEAD request without following redirects to get metadata first
                         r_head = session.head(url, timeout=20, allow_redirects=True)
-                        r_head.raise_for_status() # Will raise an error for 4xx or 5xx status codes
+                        r_head.raise_for_status()  # Will raise an error for 4xx or 5xx status codes
 
                         # Check for the new 'x-linked-size' header from Hugging Face
                         if 'x-linked-size' in r_head.headers:
diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index 05b473b71f..e05f8d7dc2 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -4,8 +4,6 @@
 from typing import Any, Dict, Optional, Union
 
 import torch
-from exllamav3 import Cache, Config, Model
-from exllamav3.cache import CacheLayer_fp16, CacheLayer_quant
 from torch.nn import CrossEntropyLoss
 from transformers import (
     GenerationConfig,
@@ -15,6 +13,8 @@
 )
 from transformers.modeling_outputs import CausalLMOutputWithPast
 
+from exllamav3 import Cache, Config, Model
+from exllamav3.cache import CacheLayer_fp16, CacheLayer_quant
 from modules import shared
 from modules.logging_colors import logger
 

From 0d857442053218f89ec29d511c761e3659bbd870 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Mon, 20 Oct 2025 20:36:55 -0300
Subject: [PATCH 1047/1701] Bump triton-windows in /requirements/full (#7274)

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 0609f69bbe..1cd1f86e91 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -25,7 +25,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.3.1.post19; platform_system == "Windows"
+triton-windows==3.5.0.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 73d3518504..7bf075b82d 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.2.0.post19; platform_system == "Windows"
+triton-windows==3.5.0.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index dad354e6ac..b84d5993dc 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.2.0.post19; platform_system == "Windows"
+triton-windows==3.5.0.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index ab48c83990..f98600219f 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.2.0.post19; platform_system == "Windows"
+triton-windows==3.5.0.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 3a341b8a1b..14016bcb39 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.2.0.post19; platform_system == "Windows"
+triton-windows==3.5.0.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 3765d329bf..5ed55a605e 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.2.0.post19; platform_system == "Windows"
+triton-windows==3.5.0.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index a3e4e51631..5e7ac929db 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.2.0.post19; platform_system == "Windows"
+triton-windows==3.5.0.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index abdd262e64..787e905844 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -25,7 +25,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.3.1.post19; platform_system == "Windows"
+triton-windows==3.5.0.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 47b0dcbd4d..2bb65a1d04 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.2.0.post19; platform_system == "Windows"
+triton-windows==3.5.0.post21; platform_system == "Windows"
 tqdm
 wandb
 

From 9476123ee69999249ddebe5feb57e9838d7e9f10 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 20 Oct 2025 19:43:26 -0700
Subject: [PATCH 1048/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 6 +++---
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 35 insertions(+), 35 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 0609f69bbe..553f9c9741 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -39,8 +39,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 73d3518504..85e2020996 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,7 +37,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index dad354e6ac..4922d692d8 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -37,7 +37,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index ab48c83990..b7c6ff14ad 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 3a341b8a1b..ccedec864b 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,6 +37,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 3765d329bf..da477a2172 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index a3e4e51631..0ecbf6c123 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index abdd262e64..11ef84eb6c 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -39,8 +39,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index d43007d06c..97fa21b86a 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 77a66dd8ac..318a7dfa92 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -22,6 +22,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index a2479e9823..d26a85c33b 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -22,6 +22,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 8d8e5233eb..af73d19f6d 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 08b7240c6b..8330aa3448 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 857c5e6672..85386971e8 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index d2e66763e2..72387a15f0 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 349272949b..86d3caaf87 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.53.0/llama_cpp_binaries-0.53.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 24fd2b4dece947ad82770044fdc595dfaa20c0f0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 21 Oct 2025 07:26:38 -0700
Subject: [PATCH 1049/1701] Update exllamav3 to 0.0.11

---
 requirements/full/requirements.txt        | 4 ++--
 requirements/full/requirements_noavx2.txt | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 95f2e18cfe..bd5920244f 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -41,8 +41,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.11/exllamav3-0.0.11+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.11/exllamav3-0.0.11+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 6dd7383bc3..c05072486a 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -41,8 +41,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.10/exllamav3-0.0.10+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.11/exllamav3-0.0.11+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.11/exllamav3-0.0.11+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 138cc654c4330a01bd35fdfcaa6b98d18943cb28 Mon Sep 17 00:00:00 2001
From: reksarka 
Date: Wed, 22 Oct 2025 20:55:17 +0300
Subject: [PATCH 1050/1701] Make it possible to run a portable Web UI build via
 a symlink (#7277)

---
 start_linux.sh | 2 +-
 start_macos.sh | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/start_linux.sh b/start_linux.sh
index a991e1bc9f..fac8df3b0f 100755
--- a/start_linux.sh
+++ b/start_linux.sh
@@ -5,7 +5,7 @@ export PYTHONNOUSERSITE=1
 unset PYTHONPATH
 unset PYTHONHOME
 
-cd "$(dirname "${BASH_SOURCE[0]}")"
+cd "$(dirname "$(readlink -f "${BASH_SOURCE[0]}")")"
 
 # Portable install case
 if [ -d "portable_env" ]; then
diff --git a/start_macos.sh b/start_macos.sh
index 6426229abd..6329cc0c6b 100755
--- a/start_macos.sh
+++ b/start_macos.sh
@@ -5,7 +5,7 @@ export PYTHONNOUSERSITE=1
 unset PYTHONPATH
 unset PYTHONHOME
 
-cd "$(dirname "${BASH_SOURCE[0]}")"
+cd "$(dirname "$(readlink -f "${BASH_SOURCE[0]}")")"
 
 # Portable install case
 if [ -d "portable_env" ]; then

From 9a84a828fc4b0e5cc632c98736f2b31adedf2106 Mon Sep 17 00:00:00 2001
From: Immanuel 
Date: Wed, 22 Oct 2025 19:59:27 +0200
Subject: [PATCH 1051/1701] Fixed python requirements for apple devices with
 macos tahoe (#7273)

---
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 4 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index efad6b255a..05ac2e4624 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
\ No newline at end of file
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index c3faff2d96..3919f0ccc5 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,6 +37,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 318a7dfa92..7933188a80 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -22,6 +22,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index d26a85c33b..538fbc14a9 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -22,6 +22,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and platform_release < "25.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"

From f4c9e671558bcdddf566f47bdad5f9e38e4018ae Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 23 Oct 2025 08:19:32 -0700
Subject: [PATCH 1052/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 6 +++---
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 35 insertions(+), 35 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index bd5920244f..6b579bc435 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -39,8 +39,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.11/exllamav3-0.0.11+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.11/exllamav3-0.0.11+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index a694fd8b96..277beb86e5 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,7 +37,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 73a54f3b6d..4637702eb9 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -37,7 +37,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 05ac2e4624..7784181d9a 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
\ No newline at end of file
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
\ No newline at end of file
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 3919f0ccc5..88b4effb14 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,6 +37,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 0b0d0dca88..aca3e51f5c 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index f46ecda62c..e75df34874 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index c05072486a..16dfea5a01 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -39,8 +39,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.11/exllamav3-0.0.11+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.11/exllamav3-0.0.11+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 97fa21b86a..c5885c3eba 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 7933188a80..a50a3e84fb 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -22,6 +22,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 538fbc14a9..df0beb224e 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -22,6 +22,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index af73d19f6d..75e60b8cf1 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 8330aa3448..b9835fa1c8 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 85386971e8..7cf4755df7 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 72387a15f0..035c8d4e27 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 86d3caaf87..bb3bd518c7 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -22,5 +22,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.54.0/llama_cpp_binaries-0.54.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From c8cd840b24910889151a9fc2da590f2b65f41b04 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Tue, 28 Oct 2025 10:07:03 -0300
Subject: [PATCH 1053/1701] Bump flash-linear-attention from 0.3.2 to 0.4.0 in
 /requirements/full (#7285)

Bumps [flash-linear-attention](https://github.com/fla-org/flash-linear-attention) from 0.3.2 to 0.4.0.
- [Release notes](https://github.com/fla-org/flash-linear-attention/releases)
- [Commits](https://github.com/fla-org/flash-linear-attention/compare/v0.3.2...v0.4.0)

---
updated-dependencies:
- dependency-name: flash-linear-attention
  dependency-version: 0.4.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] 
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
---
 requirements/full/requirements.txt        | 2 +-
 requirements/full/requirements_noavx2.txt | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 6b579bc435..798520fb65 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -5,7 +5,7 @@ colorama
 datasets
 einops
 fastapi==0.112.4
-flash-linear-attention==0.3.2
+flash-linear-attention==0.4.0
 html2text==2025.4.15
 jinja2==3.1.6
 markdown
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 16dfea5a01..11ec4d37e5 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -5,7 +5,7 @@ colorama
 datasets
 einops
 fastapi==0.112.4
-flash-linear-attention==0.3.2
+flash-linear-attention==0.4.0
 html2text==2025.4.15
 jinja2==3.1.6
 markdown

From 338ae36f734a3653aad54f3e01e1e8da8298735a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 28 Oct 2025 12:43:16 -0700
Subject: [PATCH 1054/1701] Add weights_only=True to torch.load in Training_PRO

---
 extensions/Training_PRO/script.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/extensions/Training_PRO/script.py b/extensions/Training_PRO/script.py
index cb11a8df0a..e2f90f17e2 100644
--- a/extensions/Training_PRO/script.py
+++ b/extensions/Training_PRO/script.py
@@ -823,7 +823,7 @@ def generate_and_tokenize_prompt(data_point):
         lora_model = get_peft_model(shared.model, config)
         if not always_override and Path(f"{lora_file_path}/adapter_model.bin").is_file():
             logger.info("Loading existing LoRA data...")
-            state_dict_peft = torch.load(f"{lora_file_path}/adapter_model.bin")
+            state_dict_peft = torch.load(f"{lora_file_path}/adapter_model.bin", weights_only=True)
             set_peft_model_state_dict(lora_model, state_dict_peft)
 
             print(f" + Continue Training on {RED}{lora_file_path}/adapter_model.bin{RESET}")

From 6871484398292bb602b89ac202a3389d082d15f1 Mon Sep 17 00:00:00 2001
From: Trenten Miller <62214409+inyourface34456@users.noreply.github.com>
Date: Tue, 28 Oct 2025 15:48:04 -0400
Subject: [PATCH 1055/1701] fix: Rename 'evaluation_strategy' to
 'eval_strategy' in training

---
 modules/training.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/training.py b/modules/training.py
index 2354c39df9..1de0b7026e 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -611,7 +611,7 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
             bf16=shared.args.bf16,
             optim=optimizer,
             logging_steps=2 if stop_at_loss > 0 else 5,
-            evaluation_strategy="steps" if eval_data is not None else "no",
+            eval_strategy="steps" if eval_data is not None else "no",
             eval_steps=math.ceil(eval_steps / gradient_accumulation_steps) if eval_data is not None else None,
             save_strategy="steps" if eval_data is not None else "no",
             output_dir=lora_file_path,
@@ -620,7 +620,7 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
             # TODO: Enable multi-device support
             ddp_find_unused_parameters=None,
             no_cuda=shared.args.cpu,
-            use_ipex=True if is_torch_xpu_available() and not shared.args.cpu else False
+            # use_ipex=True if is_torch_xpu_available() and not shared.args.cpu else False
         ),
         data_collator=transformers.DataCollatorForLanguageModeling(shared.tokenizer, mlm=False),
         callbacks=list([Callbacks()])

From cd645f80f8fd2488b7be96c4d1f2c8b301f85655 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 1 Nov 2025 19:58:18 -0700
Subject: [PATCH 1056/1701] Update exllamav3 to 0.0.12

---
 requirements/full/requirements.txt        | 4 ++--
 requirements/full/requirements_noavx2.txt | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 798520fb65..5677e20e55 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -41,8 +41,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.11/exllamav3-0.0.11+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.11/exllamav3-0.0.11+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 11ec4d37e5..ac0a175408 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -41,8 +41,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.11/exllamav3-0.0.11+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.11/exllamav3-0.0.11+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 67f92888911e4b1912743d3bfe5bda1610204ec2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 2 Nov 2025 14:01:00 -0800
Subject: [PATCH 1057/1701] Pin huggingface-hub to 0.36.0 (solves #7284 and
 #7289)

---
 requirements/full/requirements.txt                     | 1 +
 requirements/full/requirements_amd.txt                 | 1 +
 requirements/full/requirements_amd_noavx2.txt          | 1 +
 requirements/full/requirements_apple_intel.txt         | 3 ++-
 requirements/full/requirements_apple_silicon.txt       | 1 +
 requirements/full/requirements_cpu_only.txt            | 1 +
 requirements/full/requirements_cpu_only_noavx2.txt     | 1 +
 requirements/full/requirements_noavx2.txt              | 1 +
 requirements/full/requirements_nowheels.txt            | 1 +
 requirements/portable/requirements.txt                 | 1 +
 requirements/portable/requirements_apple_intel.txt     | 1 +
 requirements/portable/requirements_apple_silicon.txt   | 1 +
 requirements/portable/requirements_cpu_only.txt        | 1 +
 requirements/portable/requirements_cpu_only_noavx2.txt | 1 +
 requirements/portable/requirements_noavx2.txt          | 1 +
 requirements/portable/requirements_nowheels.txt        | 1 +
 requirements/portable/requirements_vulkan.txt          | 1 +
 requirements/portable/requirements_vulkan_noavx2.txt   | 1 +
 18 files changed, 19 insertions(+), 1 deletion(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 5677e20e55..bc206a02b8 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -7,6 +7,7 @@ einops
 fastapi==0.112.4
 flash-linear-attention==0.4.0
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 277beb86e5..8dc077fe2d 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -5,6 +5,7 @@ datasets
 einops
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 4637702eb9..277656964b 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -5,6 +5,7 @@ datasets
 einops
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 7784181d9a..a8f4ca2fed 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -5,6 +5,7 @@ datasets
 einops
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
@@ -38,4 +39,4 @@ tiktoken
 
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
\ No newline at end of file
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 88b4effb14..f2d1495f41 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -5,6 +5,7 @@ datasets
 einops
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index aca3e51f5c..6032cfd67a 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -5,6 +5,7 @@ datasets
 einops
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index e75df34874..f20a14227f 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -5,6 +5,7 @@ datasets
 einops
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index ac0a175408..85c26ef464 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -7,6 +7,7 @@ einops
 fastapi==0.112.4
 flash-linear-attention==0.4.0
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 2bb65a1d04..09a12418ca 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -5,6 +5,7 @@ datasets
 einops
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index c5885c3eba..4b104636c7 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -1,6 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index a50a3e84fb..a5f398550c 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -1,6 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index df0beb224e..92016c0ace 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -1,6 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 75e60b8cf1..c38c31e25f 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -1,6 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index b9835fa1c8..94e3d8abfe 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -1,6 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 7cf4755df7..d0d6f15ee9 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -1,6 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 1a7b568393..c9831985e5 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -1,6 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 035c8d4e27..a90c0d1a33 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -1,6 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index bb3bd518c7..9c6792664f 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -1,6 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
+huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*

From 92d9cd36a66c43e71ea2dbf52d7fcb051c109c0a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 5 Nov 2025 05:43:34 -0800
Subject: [PATCH 1058/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 6 +++---
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 35 insertions(+), 35 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index bc206a02b8..048bac8f72 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 8dc077fe2d..e0d8dad529 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -38,7 +38,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 277656964b..da43bedd02 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -38,7 +38,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index a8f4ca2fed..93f4b2594d 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index f2d1495f41..f05cd09d49 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -38,6 +38,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 6032cfd67a..d59629ced1 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index f20a14227f..bb0da1387f 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 85c26ef464..e6a2e6e3ff 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 4b104636c7..1821e77393 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index a5f398550c..f2f35d631f 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,6 +23,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 92016c0ace..90bf681d56 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,6 +23,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index c38c31e25f..434430f47d 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 94e3d8abfe..247226fa0b 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index d0d6f15ee9..816c849815 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index a90c0d1a33..1cfc5682b7 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 9c6792664f..c5c417ca03 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.56.0/llama_cpp_binaries-0.56.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 6be1bfcc8754c005149adfa9cc067a7bb59c3c56 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 5 Nov 2025 05:45:10 -0800
Subject: [PATCH 1059/1701] Remove the CUDA 11.7 portable builds

---
 .github/workflows/build-portable-release-cuda.yml | 13 ++-----------
 1 file changed, 2 insertions(+), 11 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index 2d2685dd7a..3b1ab8a862 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -60,7 +60,7 @@ jobs:
               'os' = @('ubuntu-22.04', 'windows-2022')
               'pyver' = @("3.11")
               'avx' = @("AVX2")
-              'cuda' = @("11.7", "12.4")
+              'cuda' = @("12.4")
           }
 
           if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})}
@@ -147,22 +147,13 @@ jobs:
 
             # Create CUDA-specific requirements file if needed
             cd "text-generation-webui-${VERSION_CLEAN}"
-            if [[ "$CUDA_VERSION" == "11.7" ]]; then
-                echo "Creating CUDA 11.7 specific requirements file"
-                sed 's/cu124/cu117/g' "$BASE_REQ_FILE" > requirements_cuda_temp.txt
-                REQ_FILE="requirements_cuda_temp.txt"
-            else
-                REQ_FILE="$BASE_REQ_FILE"
-            fi
+            REQ_FILE="$BASE_REQ_FILE"
 
             # 4. Install packages
             echo "Installing Python packages from $REQ_FILE..."
             $PIP_PATH install --target="./$PACKAGES_PATH" -r "$REQ_FILE"
 
             # 5. Clean up
-            if [[ "$CUDA_VERSION" == "11.7" ]]; then
-                rm requirements_cuda_temp.txt
-            fi
             rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore requirements/ one_click.py
 
             # 6. Create ZIP file

From e7534a90d8aad983f4b3b0c3fe67e9c29017fcaf Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 5 Nov 2025 18:46:01 -0800
Subject: [PATCH 1060/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 6 +++---
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 35 insertions(+), 35 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 048bac8f72..aa66653244 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index e0d8dad529..3bdb077463 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -38,7 +38,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index da43bedd02..0ad1124238 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -38,7 +38,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 93f4b2594d..6212441efc 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index f05cd09d49..a24cdd7ab0 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -38,6 +38,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index d59629ced1..62dc99e0ec 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index bb0da1387f..7f889de2c1 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index e6a2e6e3ff..74758177aa 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 1821e77393..f5c10dc4fb 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index f2f35d631f..2dc56261b8 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,6 +23,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 90bf681d56..0256fe98c3 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,6 +23,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 434430f47d..9f68d17c10 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 247226fa0b..7a444f81ec 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 816c849815..68cb06b792 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 1cfc5682b7..350ddb0917 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index c5c417ca03..4a88967dbf 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.58.0/llama_cpp_binaries-0.58.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 6a3bf1de927d1cd99442be258eb03529b8dad616 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 9 Nov 2025 19:43:53 -0800
Subject: [PATCH 1061/1701] Update exllamav3 to 0.0.14

---
 requirements/full/requirements.txt        | 4 ++--
 requirements/full/requirements_noavx2.txt | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index aa66653244..eea0e5faad 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,8 +42,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.14/exllamav3-0.0.14+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.14/exllamav3-0.0.14+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 74758177aa..ed2e395333 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -42,8 +42,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.12/exllamav3-0.0.12+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.14/exllamav3-0.0.14+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.14/exllamav3-0.0.14+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From a26e28bdeaeb726e9f399158023258342f22b153 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 18 Nov 2025 11:24:16 -0800
Subject: [PATCH 1062/1701] Update exllamav3 to 0.0.15

---
 requirements/full/requirements.txt        | 4 ++--
 requirements/full/requirements_noavx2.txt | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index eea0e5faad..ddbbcd868e 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,8 +42,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.14/exllamav3-0.0.14+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.14/exllamav3-0.0.14+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index ed2e395333..ec00c40790 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -42,8 +42,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.14/exllamav3-0.0.14+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.14/exllamav3-0.0.14+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 3d7e9856a2c3608974ef1f3d791a99bb38c3988f Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Tue, 18 Nov 2025 18:51:15 -0300
Subject: [PATCH 1063/1701] Update peft requirement from ==0.17.* to ==0.18.*
 in /requirements/full (#7310)

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index ddbbcd868e..a4a0e35cfd 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -12,7 +12,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.17.*
+peft==0.18.*
 Pillow>=9.5.0
 psutil
 pydantic==2.11.0
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 3bdb077463..7f0391f48f 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.17.*
+peft==0.18.*
 Pillow>=9.5.0
 psutil
 pydantic==2.11.0
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 0ad1124238..685e1cb19a 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.17.*
+peft==0.18.*
 Pillow>=9.5.0
 psutil
 pydantic==2.11.0
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 6212441efc..15fb2459c3 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.17.*
+peft==0.18.*
 Pillow>=9.5.0
 psutil
 pydantic==2.11.0
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index a24cdd7ab0..f786f4e121 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.17.*
+peft==0.18.*
 Pillow>=9.5.0
 psutil
 pydantic==2.11.0
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 62dc99e0ec..e743bda0e9 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.17.*
+peft==0.18.*
 Pillow>=9.5.0
 psutil
 pydantic==2.11.0
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 7f889de2c1..28793d430c 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.17.*
+peft==0.18.*
 Pillow>=9.5.0
 psutil
 pydantic==2.11.0
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index ec00c40790..c6afa75a65 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -12,7 +12,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.17.*
+peft==0.18.*
 Pillow>=9.5.0
 psutil
 pydantic==2.11.0
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 09a12418ca..2d7debc25a 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -10,7 +10,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pandas
-peft==0.17.*
+peft==0.18.*
 Pillow>=9.5.0
 psutil
 pydantic==2.11.0

From 4a36b7be5b90db498834d5688f5f511d3f195db9 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Tue, 18 Nov 2025 18:51:26 -0300
Subject: [PATCH 1064/1701] Bump triton-windows in /requirements/full (#7311)

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index a4a0e35cfd..d9fde6d644 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -26,7 +26,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.5.0.post21; platform_system == "Windows"
+triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 7f0391f48f..2b9738cc36 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -24,7 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.5.0.post21; platform_system == "Windows"
+triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 685e1cb19a..e07e5f2860 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -24,7 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.5.0.post21; platform_system == "Windows"
+triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 15fb2459c3..b0330a45c6 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -24,7 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.5.0.post21; platform_system == "Windows"
+triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index f786f4e121..684c38c9c9 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -24,7 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.5.0.post21; platform_system == "Windows"
+triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index e743bda0e9..383d458f8c 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -24,7 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.5.0.post21; platform_system == "Windows"
+triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 28793d430c..375839bb63 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -24,7 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.5.0.post21; platform_system == "Windows"
+triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index c6afa75a65..9862519a09 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -26,7 +26,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.5.0.post21; platform_system == "Windows"
+triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 2d7debc25a..818d22447e 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -24,7 +24,7 @@ scipy
 sentencepiece
 tensorboard
 transformers==4.57.*
-triton-windows==3.5.0.post21; platform_system == "Windows"
+triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
 wandb
 

From d85b95bb15cd9cff7a9593be4c6127ff718976fb Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 18 Nov 2025 14:06:04 -0800
Subject: [PATCH 1065/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 6 +++---
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 6 +++---
 requirements/portable/requirements_apple_silicon.txt   | 6 +++---
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 16 files changed, 35 insertions(+), 35 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index ddbbcd868e..16aa0ad7fe 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 3bdb077463..b339f5490d 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -38,7 +38,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 0ad1124238..7acdabb84a 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -38,7 +38,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 6212441efc..b658dc9600 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index a24cdd7ab0..34a9fd498d 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -38,6 +38,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 62dc99e0ec..c10e85ee41 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 7f889de2c1..76585ec8bb 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index ec00c40790..1ca034f715 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index f5c10dc4fb..a876e0b858 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 2dc56261b8..1e151785e7 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,6 +23,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 0256fe98c3..a8c775beba 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,6 +23,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 9f68d17c10..008c20af0a 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 7a444f81ec..d43ab75813 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 68cb06b792..c7f4703ddb 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 350ddb0917..4987494246 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 4a88967dbf..1b4e3bd7c6 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.60.0/llama_cpp_binaries-0.60.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From c45f35ccc27b02c3db5089ff84837c82f5c086a8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 18 Nov 2025 14:06:42 -0800
Subject: [PATCH 1066/1701] Remove the macos 13 wheels (deprecated by GitHub)

---
 requirements/full/requirements_apple_silicon.txt     | 1 -
 requirements/portable/requirements_apple_intel.txt   | 1 -
 requirements/portable/requirements_apple_silicon.txt | 1 -
 3 files changed, 3 deletions(-)

diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 34a9fd498d..83161b2721 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -40,4 +40,3 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 1e151785e7..f7e4e6e42f 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -25,4 +25,3 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index a8c775beba..99f55d3337 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -25,4 +25,3 @@ tiktoken
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0"

From 327a234d236064049b252fed32de10d4436cea8a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 18 Nov 2025 16:24:27 -0800
Subject: [PATCH 1067/1701] Add ROCm requirements.txt files

---
 requirements/full/requirements.txt            |  4 +--
 requirements/full/requirements_amd.txt        |  2 +-
 .../full/requirements_apple_intel.txt         |  4 +--
 .../full/requirements_apple_silicon.txt       |  4 +--
 requirements/full/requirements_cpu_only.txt   |  4 +--
 .../full/requirements_cpu_only_noavx2.txt     |  4 +--
 requirements/full/requirements_noavx2.txt     |  4 +--
 requirements/portable/requirements_amd.txt    | 27 +++++++++++++++++++
 .../portable/requirements_amd_noavx2.txt      | 27 +++++++++++++++++++
 requirements/portable/requirements_vulkan.txt |  2 +-
 10 files changed, 68 insertions(+), 14 deletions(-)
 create mode 100644 requirements/portable/requirements_amd.txt
 create mode 100644 requirements/portable/requirements_amd_noavx2.txt

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 0f4eac5275..7075f955ab 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index f8d78dbc44..39e4abd618 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -39,6 +39,6 @@ tiktoken
 
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 5cc516d292..dc1d182003 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 9639c6b533..0ccf95a5a3 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 914de4e8be..357daa9f32 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index d25879ba51..4e8db9e94a 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 731839a7a4..2204a40e76 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
new file mode 100644
index 0000000000..3dfa6d3f9f
--- /dev/null
+++ b/requirements/portable/requirements_amd.txt
@@ -0,0 +1,27 @@
+audioop-lts<1.0; python_version >= "3.13"
+fastapi==0.112.4
+html2text==2025.4.15
+huggingface-hub==0.36.0
+jinja2==3.1.6
+markdown
+numpy==2.2.*
+pydantic==2.11.0
+PyPDF2==3.0.1
+python-docx==1.1.2
+pyyaml
+requests
+rich
+tqdm
+
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
+
+# AMD wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
new file mode 100644
index 0000000000..57c6d55228
--- /dev/null
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -0,0 +1,27 @@
+audioop-lts<1.0; python_version >= "3.13"
+fastapi==0.112.4
+html2text==2025.4.15
+huggingface-hub==0.36.0
+jinja2==3.1.6
+markdown
+numpy==2.2.*
+pydantic==2.11.0
+PyPDF2==3.0.1
+python-docx==1.1.2
+pyyaml
+requests
+rich
+tqdm
+
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
+
+# AMD wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 4987494246..d339f97f2d 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -22,6 +22,6 @@ flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
-# CUDA wheels
+# Vulkan wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From d6f39e1fef16c5e92fe8eb490d51e76d0e230d99 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 18 Nov 2025 16:32:20 -0800
Subject: [PATCH 1068/1701] Add ROCm portable builds

---
 .github/workflows/build-everything-tgw.yml    |   7 +
 .../workflows/build-portable-release-rocm.yml | 165 ++++++++++++++++++
 2 files changed, 172 insertions(+)
 create mode 100644 .github/workflows/build-portable-release-rocm.yml

diff --git a/.github/workflows/build-everything-tgw.yml b/.github/workflows/build-everything-tgw.yml
index 23773010e9..78fd2d6bc1 100644
--- a/.github/workflows/build-everything-tgw.yml
+++ b/.github/workflows/build-everything-tgw.yml
@@ -41,6 +41,13 @@ jobs:
       version: ${{ inputs.version }}
       config: 'os:ubuntu-22.04'
 
+  build_release_rocm_linux:
+    name: ROCm Linux
+    uses: ./.github/workflows/build-portable-release-rocm.yml
+    with:
+      version: ${{ inputs.version }}
+      config: 'os:ubuntu-22.04'
+
   build_release_cpu_windows:
     name: CPU Windows
     uses: ./.github/workflows/build-portable-release.yml
diff --git a/.github/workflows/build-portable-release-rocm.yml b/.github/workflows/build-portable-release-rocm.yml
new file mode 100644
index 0000000000..ed7f3a87f6
--- /dev/null
+++ b/.github/workflows/build-portable-release-rocm.yml
@@ -0,0 +1,165 @@
+name: Build ROCm
+
+on:
+  workflow_dispatch:
+    inputs:
+      version:
+        description: 'Version tag of text-generation-webui to build: v3.0'
+        default: 'v3.0'
+        required: true
+        type: string
+      config:
+        description: 'Override configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2'
+        default: 'Default'
+        required: false
+        type: string
+      exclude:
+        description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
+        default: 'None'
+        required: false
+        type: string
+  workflow_call:
+    inputs:
+      version:
+        description: 'Version tag of text-generation-webui to build: v3.0'
+        default: 'v3.0'
+        required: true
+        type: string
+      config:
+        description: 'Configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2'
+        default: 'Default'
+        required: false
+        type: string
+      exclude:
+        description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
+        default: 'None'
+        required: false
+        type: string
+
+permissions:
+  contents: write
+
+jobs:
+  define_matrix:
+    name: Define Build Matrix
+    runs-on: ubuntu-latest
+    outputs:
+      matrix: ${{ steps.set-matrix.outputs.matrix }}
+    defaults:
+      run:
+        shell: pwsh
+    env:
+      CONFIGIN: ${{ inputs.config }}
+      EXCLUDEIN: ${{ inputs.exclude }}
+
+    steps:
+      - name: Define Job Output
+        id: set-matrix
+        run: |
+          $matrix = @{
+              'os' = @('ubuntu-22.04')
+              'pyver' = @("3.11")
+              'avx' = @("AVX2")
+          }
+
+          if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})}
+
+          if ($env:EXCLUDEIN -ne 'None') {
+              $exclusions = @()
+              $exclusions += $env:EXCLUDEIN.split(';').replace(':','=').replace(',',"`n") | ConvertFrom-StringData
+              $matrix['exclude'] = $exclusions
+          }
+
+          $matrixOut = ConvertTo-Json $matrix -Compress
+          Write-Output ('matrix=' + $matrixOut) >> $env:GITHUB_OUTPUT
+
+  build_wheels:
+    name: ${{ matrix.os }} ${{ matrix.pyver }} CPU ${{ matrix.avx }}
+    needs: define_matrix
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix: ${{ fromJSON(needs.define_matrix.outputs.matrix) }}
+    defaults:
+      run:
+        shell: pwsh
+    env:
+      AVXVER: ${{ matrix.avx }}
+      PCKGVER: ${{ inputs.version }}
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          repository: 'oobabooga/text-generation-webui'
+          ref: ${{ inputs.version }}
+          submodules: 'recursive'
+
+      - uses: actions/setup-python@v4
+        with:
+          python-version: ${{ matrix.pyver }}
+
+      - name: Build Package
+        shell: bash
+        run: |
+            VERSION_CLEAN="${{ inputs.version }}"
+            VERSION_CLEAN="${VERSION_CLEAN#v}"
+            cd ..
+            cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
+            cd "text-generation-webui-${VERSION_CLEAN}"
+
+            # Remove extensions that need additional requirements
+            allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
+            find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
+
+            # Define common variables
+            AVX_SUPPORT="${{ matrix.avx }}"
+            VERSION="${{ inputs.version }}"
+
+            # 1. Set platform-specific variables (Linux only for ROCm)
+            PLATFORM="linux"
+            PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-unknown-linux-gnu-install_only.tar.gz"
+            PIP_PATH="portable_env/bin/python -m pip"
+            PACKAGES_PATH="portable_env/lib/python3.11/site-packages"
+            rm start_macos.sh start_windows.bat
+
+            # 2. Download and extract Python
+            cd ..
+            echo "Downloading Python for $PLATFORM..."
+            curl -L -o python-build.tar.gz "$PYTHON_URL"
+            tar -xzf python-build.tar.gz
+            mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
+
+            # 3. Prepare requirements file based on AVX
+            if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
+                BASE_REQ_FILE="requirements/portable/requirements_amd.txt"
+            else
+                BASE_REQ_FILE="requirements/portable/requirements_amd_noavx2.txt"
+            fi
+            REQ_FILE="$BASE_REQ_FILE"
+
+            cd "text-generation-webui-${VERSION_CLEAN}"
+
+            # 4. Install packages
+            echo "Installing Python packages from $REQ_FILE..."
+            $PIP_PATH install --target="./$PACKAGES_PATH" -r "$REQ_FILE"
+
+            # 5. Clean up
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore requirements/ one_click.py
+
+            # 6. Create ZIP file
+            cd ..
+            ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-rocm.zip"
+            echo "Creating archive: $ZIP_NAME"
+
+            zip -r "$ZIP_NAME" "text-generation-webui-${VERSION_CLEAN}"
+
+      - name: Upload files to a GitHub release
+        id: upload-release
+        uses: svenstaro/upload-release-action@2.7.0
+        continue-on-error: true
+        with:
+          repo_token: ${{ secrets.GITHUB_TOKEN }}
+          file: ../textgen-portable-*.zip
+          tag: ${{ inputs.version }}
+          file_glob: true
+          make_latest: false
+          overwrite: true

From 0d4eff284ca43c84bc37e49ea24d95106dacf098 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 19 Nov 2025 05:23:43 -0800
Subject: [PATCH 1069/1701] Add a --cpu-moe model for llama.cpp

---
 modules/llama_cpp_server.py | 2 ++
 modules/loaders.py          | 1 +
 modules/shared.py           | 1 +
 modules/ui.py               | 1 +
 modules/ui_model_menu.py    | 3 ++-
 5 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 27890d8c90..47d9d27c10 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -326,6 +326,8 @@ def _start_server(self):
             cmd += ["--threads", str(shared.args.threads)]
         if shared.args.threads_batch > 0:
             cmd += ["--threads-batch", str(shared.args.threads_batch)]
+        if shared.args.cpu_moe:
+            cmd.append("--cpu-moe")
         if shared.args.no_mmap:
             cmd.append("--no-mmap")
         if shared.args.mlock:
diff --git a/modules/loaders.py b/modules/loaders.py
index 609a54c6df..0f0f6d1e96 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -6,6 +6,7 @@
 loaders_and_params = OrderedDict({
     'llama.cpp': [
         'gpu_layers',
+        'cpu_moe',
         'threads',
         'threads_batch',
         'batch_size',
diff --git a/modules/shared.py b/modules/shared.py
index e54ba654a2..1cca123329 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -66,6 +66,7 @@
 # llama.cpp
 group = parser.add_argument_group('llama.cpp')
 group.add_argument('--gpu-layers', '--n-gpu-layers', type=int, default=256, metavar='N', help='Number of layers to offload to the GPU.')
+group.add_argument('--cpu-moe', action='store_true', help='Move the experts to the CPU (for MoE models).')
 group.add_argument('--mmproj', type=str, default=None, help='Path to the mmproj file for vision models.')
 group.add_argument('--streaming-llm', action='store_true', help='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
 group.add_argument('--tensor-split', type=str, default=None, help='Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.')
diff --git a/modules/ui.py b/modules/ui.py
index 76533767a6..d8dcedfb45 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -125,6 +125,7 @@ def list_model_elements():
         'loader',
         'cpu_memory',
         'gpu_layers',
+        'cpu_moe',
         'threads',
         'threads_batch',
         'batch_size',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 50ada9f90c..31ab929f6e 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -50,6 +50,7 @@ def create_ui():
 
                         with gr.Column():
                             shared.gradio['vram_info'] = gr.HTML(value=get_initial_vram_info())
+                            shared.gradio['cpu_moe'] = gr.Checkbox(label="cpu-moe", value=shared.args.cpu_moe, info='Move the experts to the CPU. Saves VRAM on MoE models.')
                             shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming-llm", value=shared.args.streaming_llm, info='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
                             shared.gradio['load_in_8bit'] = gr.Checkbox(label="load-in-8bit", value=shared.args.load_in_8bit)
                             shared.gradio['load_in_4bit'] = gr.Checkbox(label="load-in-4bit", value=shared.args.load_in_4bit)
@@ -94,7 +95,7 @@ def create_ui():
                                 shared.gradio['num_experts_per_token'] = gr.Number(label="Number of experts per token", value=shared.args.num_experts_per_token, info='Only applies to MoE models like Mixtral.')
 
                             with gr.Column():
-                                shared.gradio['cpu'] = gr.Checkbox(label="cpu", value=shared.args.cpu, info='llama.cpp: Use llama-cpp-python compiled without GPU acceleration. Transformers: use PyTorch in CPU mode.')
+                                shared.gradio['cpu'] = gr.Checkbox(label="cpu", value=shared.args.cpu, info='Use PyTorch in CPU mode.')
                                 shared.gradio['disk'] = gr.Checkbox(label="disk", value=shared.args.disk)
                                 shared.gradio['row_split'] = gr.Checkbox(label="row_split", value=shared.args.row_split, info='Split the model by rows across GPUs. This may improve multi-gpu performance.')
                                 shared.gradio['no_kv_offload'] = gr.Checkbox(label="no_kv_offload", value=shared.args.no_kv_offload, info='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')

From b0baf7518bfd76d761b04ec7168e184089e18cab Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 19 Nov 2025 06:07:15 -0800
Subject: [PATCH 1070/1701] Remove macos x86-64 portable builds (macos-13
 runner deprecated by GitHub)

---
 .github/workflows/build-portable-release.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 51567ec8bd..a5b5b7c534 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -57,7 +57,7 @@ jobs:
         id: set-matrix
         run: |
           $matrix = @{
-              'os' = @('ubuntu-22.04', 'windows-2022', 'macos-13', 'macos-14')
+              'os' = @('ubuntu-22.04', 'windows-2022', 'macos-14')
               'pyver' = @("3.11")
               'avx' = @("AVX2")
           }

From 8f0048663d815bd9f4e253f5e87f9ea4d309caf9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 21 Nov 2025 07:09:16 -0800
Subject: [PATCH 1071/1701] More modular HTML generator

---
 modules/html_generator.py | 107 ++++++++++++++++++++++++--------------
 1 file changed, 68 insertions(+), 39 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 492b52bd31..ccc232a049 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -196,50 +196,45 @@ def extract_thinking_block(string):
     return None, string
 
 
-@functools.lru_cache(maxsize=None)
-def convert_to_markdown(string, message_id=None):
-    if not string:
-        return ""
-
-    # Use a default message ID if none provided
-    if message_id is None:
-        message_id = "unknown"
-
-    # Extract thinking block if present
-    thinking_content, remaining_content = extract_thinking_block(string)
-
-    # Process the main content
-    html_output = process_markdown_content(remaining_content)
-
-    # If thinking content was found, process it using the same function
-    if thinking_content is not None:
-        thinking_html = process_markdown_content(thinking_content)
-
-        # Generate unique ID for the thinking block
-        block_id = f"thinking-{message_id}-0"
-
-        # Check if thinking is complete or still in progress
-        is_streaming = not remaining_content
-        title_text = "Thinking..." if is_streaming else "Thought"
+def build_thinking_block(thinking_content, message_id, has_remaining_content):
+    """Build HTML for a thinking block."""
+    if thinking_content is None:
+        return None
+
+    # Process the thinking content through markdown
+    thinking_html = process_markdown_content(thinking_content)
+
+    # Generate unique ID for the thinking block
+    block_id = f"thinking-{message_id}-0"
+
+    # Check if thinking is complete or still in progress
+    is_streaming = not has_remaining_content
+    title_text = "Thinking..." if is_streaming else "Thought"
+
+    return f'''
+    
            
+        
+            {info_svg_small}
+            {title_text}
+        
+        
            {thinking_html}
            
+    
            
+    '''
 
-        thinking_block = f'''
-        
            
-            
-                {info_svg_small}
-                {title_text}
-            
-            
            {thinking_html}
            
-        
            
-        '''
 
-        # Prepend the thinking block to the message HTML
-        html_output = thinking_block + html_output
+def build_main_content_block(content):
+    """Build HTML for the main content block."""
+    if not content:
+        return ""
 
-    return html_output
+    return process_markdown_content(content)
 
 
 def process_markdown_content(string):
-    """Process a string through the markdown conversion pipeline."""
+    """
+    Process a string through the markdown conversion pipeline.
+    Uses robust manual parsing to ensure correct LaTeX and Code Block rendering.
+    """
     if not string:
         return ""
 
@@ -280,7 +275,7 @@ def protect_asterisks_in_latex(match):
     pattern = re.compile(r'\\begin{blockquote}(.*?)\\end{blockquote}', re.DOTALL)
     string = pattern.sub(replace_blockquote, string)
 
-    # Code
+    # Code block standardization
     string = string.replace('\\begin{code}', '```')
     string = string.replace('\\end{code}', '```')
     string = string.replace('\\begin{align*}', '$$')
@@ -301,6 +296,7 @@ def protect_asterisks_in_latex(match):
     is_code = False
     is_latex = False
 
+    # Manual line iteration for robust structure parsing
     for line in string.split('\n'):
         stripped_line = line.strip()
 
@@ -371,6 +367,39 @@ def protect_asterisks_in_latex(match):
     return html_output
 
 
+@functools.lru_cache(maxsize=None)
+def convert_to_markdown(string, message_id=None):
+    """
+    Convert a string to markdown HTML with support for multiple block types.
+    Blocks are assembled in order: thinking, main content, etc.
+    """
+    if not string:
+        return ""
+
+    # Use a default message ID if none provided
+    if message_id is None:
+        message_id = "unknown"
+
+    # Extract different components from the string
+    thinking_content, remaining_content = extract_thinking_block(string)
+
+    # Build individual HTML blocks
+    blocks = []
+
+    # Add thinking block if present
+    thinking_html = build_thinking_block(thinking_content, message_id, bool(remaining_content))
+    if thinking_html:
+        blocks.append(thinking_html)
+
+    # Add main content block
+    main_html = build_main_content_block(remaining_content)
+    if main_html:
+        blocks.append(main_html)
+
+    # Assemble all blocks into final HTML
+    return ''.join(blocks)
+
+
 def convert_to_markdown_wrapped(string, message_id=None, use_cache=True):
     '''
     Used to avoid caching convert_to_markdown calls during streaming.

From 400bb0694b148a8c099636e2d546c52df3614cfd Mon Sep 17 00:00:00 2001
From: GodEmperor785 <32663080+GodEmperor785@users.noreply.github.com>
Date: Fri, 21 Nov 2025 20:56:02 +0100
Subject: [PATCH 1072/1701] Add slider for --ubatch-size for llama.cpp loader,
 change defaults for better MoE performance (#7316)

---
 modules/llama_cpp_server.py | 1 +
 modules/loaders.py          | 1 +
 modules/shared.py           | 3 ++-
 modules/ui.py               | 1 +
 modules/ui_model_menu.py    | 1 +
 5 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 47d9d27c10..f83ed663d7 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -317,6 +317,7 @@ def _start_server(self):
             "--ctx-size", str(shared.args.ctx_size),
             "--gpu-layers", str(shared.args.gpu_layers),
             "--batch-size", str(shared.args.batch_size),
+            "--ubatch-size", str(shared.args.ubatch_size),
             "--port", str(self.port),
             "--no-webui",
             "--flash-attn", "on",
diff --git a/modules/loaders.py b/modules/loaders.py
index 0f0f6d1e96..0bf3781b97 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -10,6 +10,7 @@
         'threads',
         'threads_batch',
         'batch_size',
+        'ubatch_size',
         'ctx_size',
         'cache_type',
         'tensor_split',
diff --git a/modules/shared.py b/modules/shared.py
index 1cca123329..289fda54dd 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -74,7 +74,8 @@
 group.add_argument('--no-mmap', action='store_true', help='Prevent mmap from being used.')
 group.add_argument('--mlock', action='store_true', help='Force the system to keep the model in RAM.')
 group.add_argument('--no-kv-offload', action='store_true', help='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
-group.add_argument('--batch-size', type=int, default=256, help='Maximum number of prompt tokens to batch together when calling llama_eval.')
+group.add_argument('--batch-size', type=int, default=1024, help='Maximum number of prompt tokens to batch together when calling llama-server. This is the application level batch size')
+group.add_argument('--ubatch-size', type=int, default=1024, help='Maximum number of prompt tokens to batch together when calling llama-server. This is the max physical batch size for computation (device level).')
 group.add_argument('--threads', type=int, default=0, help='Number of threads to use.')
 group.add_argument('--threads-batch', type=int, default=0, help='Number of threads to use for batches/prompt processing.')
 group.add_argument('--numa', action='store_true', help='Activate NUMA task allocation for llama.cpp.')
diff --git a/modules/ui.py b/modules/ui.py
index d8dcedfb45..f99e8b6a7f 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -129,6 +129,7 @@ def list_model_elements():
         'threads',
         'threads_batch',
         'batch_size',
+        'ubatch_size',
         'ctx_size',
         'cache_type',
         'tensor_split',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 31ab929f6e..86adc2297e 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -84,6 +84,7 @@ def create_ui():
                                 shared.gradio['threads'] = gr.Slider(label="threads", minimum=0, step=1, maximum=256, value=shared.args.threads)
                                 shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
                                 shared.gradio['batch_size'] = gr.Slider(label="batch_size", minimum=1, maximum=4096, step=1, value=shared.args.batch_size)
+                                shared.gradio['ubatch_size'] = gr.Slider(label="ubatch_size", minimum=1, maximum=4096, step=1, value=shared.args.ubatch_size)
                                 shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
                                 shared.gradio['extra_flags'] = gr.Textbox(label='extra-flags', info='Additional flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"', value=shared.args.extra_flags)
                                 shared.gradio['cpu_memory'] = gr.Number(label="Maximum CPU memory in GiB. Use this for CPU offloading.", value=shared.args.cpu_memory)

From 3cad0cd4c189e98bea4c781b240a7cf9f4106dfc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 28 Nov 2025 03:52:37 -0800
Subject: [PATCH 1073/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 4 ++--
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 4 ++--
 requirements/portable/requirements_amd_noavx2.txt      | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 4 ++--
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 36 insertions(+), 36 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 7075f955ab..676283962f 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 39e4abd618..d9bf64cb67 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -38,7 +38,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 00e2172ca7..2c06c22f12 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -38,7 +38,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index dc1d182003..a4929d4006 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 0ccf95a5a3..6a7ce8a62b 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 357daa9f32..05957700fc 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 4e8db9e94a..9e43c409c7 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 2204a40e76..bf3c073480 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index a876e0b858..5f9b69e894 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 3dfa6d3f9f..10a6b5af8b 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index 57c6d55228..4686b9de71 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index f7e4e6e42f..890568f4d4 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 99f55d3337..e480db8f41 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 008c20af0a..c2a7d0403a 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index d43ab75813..e243d4bea4 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index c7f4703ddb..3db163b01b 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index d339f97f2d..7758131cf5 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 1b4e3bd7c6..dcee8d05e7 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.61.0/llama_cpp_binaries-0.61.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 78b315344abca1863366219a1bbe792e91c817e1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 28 Nov 2025 06:45:05 -0800
Subject: [PATCH 1074/1701] Update exllamav3

---
 requirements/full/requirements.txt        | 4 ++--
 requirements/full/requirements_noavx2.txt | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 676283962f..4be87c56e1 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,8 +42,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index bf3c073480..9931e1dcd2 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -42,8 +42,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.15/exllamav3-0.0.15+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 5327bc939705c0f65e77f8925ffd248fd3e66b88 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Fri, 28 Nov 2025 22:48:05 -0300
Subject: [PATCH 1075/1701] Update modules/shared.py

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
---
 modules/shared.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/shared.py b/modules/shared.py
index 289fda54dd..134c0caccc 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -74,7 +74,7 @@
 group.add_argument('--no-mmap', action='store_true', help='Prevent mmap from being used.')
 group.add_argument('--mlock', action='store_true', help='Force the system to keep the model in RAM.')
 group.add_argument('--no-kv-offload', action='store_true', help='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
-group.add_argument('--batch-size', type=int, default=1024, help='Maximum number of prompt tokens to batch together when calling llama-server. This is the application level batch size')
+group.add_argument('--batch-size', type=int, default=1024, help='Maximum number of prompt tokens to batch together when calling llama-server. This is the application level batch size.')
 group.add_argument('--ubatch-size', type=int, default=1024, help='Maximum number of prompt tokens to batch together when calling llama-server. This is the max physical batch size for computation (device level).')
 group.add_argument('--threads', type=int, default=0, help='Number of threads to use.')
 group.add_argument('--threads-batch', type=int, default=0, help='Number of threads to use for batches/prompt processing.')

From 661e42d2b7458a82ea0e98b1c732ba6a91dc4100 Mon Sep 17 00:00:00 2001
From: aidevtime 
Date: Sat, 29 Nov 2025 09:59:36 +0800
Subject: [PATCH 1076/1701] fix(deps): upgrade coqui-tts to >=0.27.0 for
 transformers 4.55 compatibility (#7329)

---
 extensions/coqui_tts/requirements.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/extensions/coqui_tts/requirements.txt b/extensions/coqui_tts/requirements.txt
index b0b691e8fe..e7eb369171 100644
--- a/extensions/coqui_tts/requirements.txt
+++ b/extensions/coqui_tts/requirements.txt
@@ -1 +1 @@
-coqui-tts==0.25.1
+coqui-tts>=0.27.0

From e24ba92ef231963387f5719125943acd83b46c46 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 1 Dec 2025 08:05:21 -0800
Subject: [PATCH 1077/1701] UI: Optimize typing in all textareas

---
 js/main.js | 61 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 61 insertions(+)

diff --git a/js/main.js b/js/main.js
index 67f602793c..337131c2ff 100644
--- a/js/main.js
+++ b/js/main.js
@@ -1105,3 +1105,64 @@ document.fonts.addEventListener("loadingdone", (event) => {
   // Initial call to set the margin based on current state
   updateMargin();
 })();
+
+//------------------------------------------------
+// Optimize typing in all textareas
+//------------------------------------------------
+
+(function() {
+  document.querySelectorAll("textarea").forEach(textarea => {
+    const computedStyle = getComputedStyle(textarea);
+    const MIN_HEIGHT = parseInt(computedStyle.minHeight) || textarea.offsetHeight || 42;
+    const configuredMax = parseInt(computedStyle.maxHeight) || 400;
+
+    let rafId = null;
+    let isOurResize = false;
+
+    function doResize() {
+      rafId = null;
+      isOurResize = true;
+
+      // Recalculate max height each time
+      const maxHeight = Math.min(configuredMax, window.innerHeight * 0.5);
+
+      textarea.style.height = "auto";
+      const contentHeight = textarea.scrollHeight;
+      const clampedHeight = Math.min(maxHeight, Math.max(MIN_HEIGHT, contentHeight));
+
+      textarea.style.height = clampedHeight + "px";
+      textarea.style.overflowY = contentHeight > maxHeight ? "auto" : "hidden";
+
+      isOurResize = false;
+    }
+
+    function scheduleResize() {
+      if (rafId === null) {
+        rafId = requestAnimationFrame(doResize);
+      }
+    }
+
+    const desc = Object.getOwnPropertyDescriptor(CSSStyleDeclaration.prototype, "height");
+    const originalSet = desc?.set || function(v) { this.setProperty("height", v); };
+    const originalGet = desc?.get || function() { return this.getPropertyValue("height"); };
+
+    Object.defineProperty(textarea.style, "height", {
+      get() { return originalGet.call(this); },
+      set(value) {
+        if (isOurResize) originalSet.call(this, value);
+        else scheduleResize();
+      },
+      configurable: true
+    });
+
+    textarea.addEventListener("input", scheduleResize, { passive: true });
+    doResize();
+  });
+
+  // Trigger resize on all textareas when window resizes
+  window.addEventListener("resize", () => {
+    document.querySelectorAll("textarea").forEach(ta => {
+      ta.dispatchEvent(new Event("input"));
+    });
+  }, { passive: true });
+})();

From a83821e941f8f6d33556d101d7aa241b41db562e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 1 Dec 2025 10:34:23 -0800
Subject: [PATCH 1078/1701] Revert "UI: Optimize typing in all textareas"

This reverts commit e24ba92ef231963387f5719125943acd83b46c46.
---
 js/main.js | 61 ------------------------------------------------------
 1 file changed, 61 deletions(-)

diff --git a/js/main.js b/js/main.js
index 337131c2ff..67f602793c 100644
--- a/js/main.js
+++ b/js/main.js
@@ -1105,64 +1105,3 @@ document.fonts.addEventListener("loadingdone", (event) => {
   // Initial call to set the margin based on current state
   updateMargin();
 })();
-
-//------------------------------------------------
-// Optimize typing in all textareas
-//------------------------------------------------
-
-(function() {
-  document.querySelectorAll("textarea").forEach(textarea => {
-    const computedStyle = getComputedStyle(textarea);
-    const MIN_HEIGHT = parseInt(computedStyle.minHeight) || textarea.offsetHeight || 42;
-    const configuredMax = parseInt(computedStyle.maxHeight) || 400;
-
-    let rafId = null;
-    let isOurResize = false;
-
-    function doResize() {
-      rafId = null;
-      isOurResize = true;
-
-      // Recalculate max height each time
-      const maxHeight = Math.min(configuredMax, window.innerHeight * 0.5);
-
-      textarea.style.height = "auto";
-      const contentHeight = textarea.scrollHeight;
-      const clampedHeight = Math.min(maxHeight, Math.max(MIN_HEIGHT, contentHeight));
-
-      textarea.style.height = clampedHeight + "px";
-      textarea.style.overflowY = contentHeight > maxHeight ? "auto" : "hidden";
-
-      isOurResize = false;
-    }
-
-    function scheduleResize() {
-      if (rafId === null) {
-        rafId = requestAnimationFrame(doResize);
-      }
-    }
-
-    const desc = Object.getOwnPropertyDescriptor(CSSStyleDeclaration.prototype, "height");
-    const originalSet = desc?.set || function(v) { this.setProperty("height", v); };
-    const originalGet = desc?.get || function() { return this.getPropertyValue("height"); };
-
-    Object.defineProperty(textarea.style, "height", {
-      get() { return originalGet.call(this); },
-      set(value) {
-        if (isOurResize) originalSet.call(this, value);
-        else scheduleResize();
-      },
-      configurable: true
-    });
-
-    textarea.addEventListener("input", scheduleResize, { passive: true });
-    doResize();
-  });
-
-  // Trigger resize on all textareas when window resizes
-  window.addEventListener("resize", () => {
-    document.querySelectorAll("textarea").forEach(ta => {
-      ta.dispatchEvent(new Event("input"));
-    });
-  }, { passive: true });
-})();

From b3666e140de349a651aea22a6f418572925a5c62 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Tue, 2 Dec 2025 14:55:38 -0300
Subject: [PATCH 1079/1701] Add image generation support (#7328)

---
 README.md                                     |   7 +-
 css/main.css                                  |  96 +-
 docs/Image Generation Tutorial.md             |  20 +
 modules/image_models.py                       | 183 ++++
 modules/shared.py                             |  53 +-
 modules/ui.py                                 |  40 +-
 modules/ui_image_generation.py                | 847 ++++++++++++++++++
 modules/utils.py                              |  22 +-
 requirements/full/requirements.txt            |   4 +
 requirements/full/requirements_amd.txt        |   4 +
 requirements/full/requirements_amd_noavx2.txt |   4 +
 .../full/requirements_apple_intel.txt         |   4 +
 .../full/requirements_apple_silicon.txt       |   4 +
 requirements/full/requirements_cpu_only.txt   |   4 +
 .../full/requirements_cpu_only_noavx2.txt     |   4 +
 requirements/full/requirements_noavx2.txt     |   4 +
 requirements/full/requirements_nowheels.txt   |   4 +
 server.py                                     |  24 +
 18 files changed, 1314 insertions(+), 14 deletions(-)
 create mode 100644 docs/Image Generation Tutorial.md
 create mode 100644 modules/image_models.py
 create mode 100644 modules/ui_image_generation.py

diff --git a/README.md b/README.md
index d350d9596c..174fc2ac37 100644
--- a/README.md
+++ b/README.md
@@ -28,6 +28,8 @@ A Gradio web UI for Large Language Models.
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
 - **Vision (multimodal models)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
+Image generation: A dedicated tab for diffusers models like Z-Image-Turbo and Qwen-Image. Features 4-bit/8-bit quantization and a persistent gallery with metadata (tutorial).
+- **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo** and **Qwen-Image**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
 - **Web search**: Optionally search the internet with LLM-generated queries to add context to the conversation.
 - Aesthetic UI with dark and light themes.
 - Syntax highlighting for code blocks and LaTeX rendering for mathematical expressions.
@@ -432,6 +434,7 @@ https://colab.research.google.com/github/oobabooga/text-generation-webui/blob/ma
 
 https://www.reddit.com/r/Oobabooga/
 
-## Acknowledgment
+## Acknowledgments
 
-In August 2023, [Andreessen Horowitz](https://a16z.com/) (a16z) provided a generous grant to encourage and support my independent work on this project. I am **extremely** grateful for their trust and recognition.
+- In August 2023, [Andreessen Horowitz](https://a16z.com/) (a16z) provided a generous grant to encourage and support my independent work on this project. I am **extremely** grateful for their trust and recognition.
+- This project was inspired by [AUTOMATIC1111/stable-diffusion-webui](https://github.com/AUTOMATIC1111/stable-diffusion-webui) and wouldn't exist without it.
diff --git a/css/main.css b/css/main.css
index fd79d24c3e..5c1c356dbe 100644
--- a/css/main.css
+++ b/css/main.css
@@ -93,11 +93,11 @@ ol li p, ul li p {
     display: inline-block;
 }
 
-#notebook-parent-tab, #chat-tab, #parameters, #chat-settings, #lora, #training-tab, #model-tab, #session-tab, #character-tab {
+#notebook-parent-tab, #chat-tab, #parameters, #chat-settings, #lora, #training-tab, #model-tab, #session-tab, #character-tab, #image-ai-tab {
     border: 0;
 }
 
-#notebook-parent-tab, #parameters, #chat-settings, #lora, #training-tab, #model-tab, #session-tab, #character-tab {
+#notebook-parent-tab, #parameters, #chat-settings, #lora, #training-tab, #model-tab, #session-tab, #character-tab, #image-ai-tab {
     padding: 1rem;
 }
 
@@ -244,37 +244,46 @@ button {
     font-size: 100% !important;
 }
 
-.pretty_scrollbar::-webkit-scrollbar {
+.pretty_scrollbar::-webkit-scrollbar,
+#image-history-gallery > :nth-child(2)::-webkit-scrollbar {
     width: 8px;
     height: 8px;
 }
 
-.pretty_scrollbar::-webkit-scrollbar-track {
+.pretty_scrollbar::-webkit-scrollbar-track,
+#image-history-gallery > :nth-child(2)::-webkit-scrollbar-track {
     background: transparent;
 }
 
 .pretty_scrollbar::-webkit-scrollbar-thumb,
-.pretty_scrollbar::-webkit-scrollbar-thumb:hover {
+.pretty_scrollbar::-webkit-scrollbar-thumb:hover,
+#image-history-gallery > :nth-child(2)::-webkit-scrollbar-thumb,
+#image-history-gallery > :nth-child(2)::-webkit-scrollbar-thumb:hover {
     background: var(--neutral-300);
     border-radius: 30px;
 }
 
 .dark .pretty_scrollbar::-webkit-scrollbar-thumb,
-.dark .pretty_scrollbar::-webkit-scrollbar-thumb:hover {
+.dark .pretty_scrollbar::-webkit-scrollbar-thumb:hover,
+.dark #image-history-gallery > :nth-child(2)::-webkit-scrollbar-thumb,
+.dark #image-history-gallery > :nth-child(2)::-webkit-scrollbar-thumb:hover {
     background: rgb(255 255 255 / 6.25%);
     border-radius: 10px;
 }
 
-.pretty_scrollbar::-webkit-resizer {
+.pretty_scrollbar::-webkit-resizer,
+#image-history-gallery > :nth-child(2)::-webkit-resizer {
     background: #c5c5d2;
 }
 
-.dark .pretty_scrollbar::-webkit-resizer {
+.dark .pretty_scrollbar::-webkit-resizer,
+.dark #image-history-gallery > :nth-child(2)::-webkit-resizer {
     background: #ccc;
     border-radius: 10px;
 }
 
-.pretty_scrollbar::-webkit-scrollbar-corner {
+.pretty_scrollbar::-webkit-scrollbar-corner,
+#image-history-gallery > :nth-child(2)::-webkit-scrollbar-corner {
     background: transparent;
 }
 
@@ -1674,3 +1683,72 @@ button:focus {
 .dark .sidebar-vertical-separator {
     border-bottom: 1px solid rgb(255 255 255 / 10%);
 }
+
+button#swap-height-width {
+    position: absolute;
+    top: -50px;
+    right: 0;
+    border: 0;
+}
+
+#image-output-gallery, #image-output-gallery > :nth-child(2) {
+    height: calc(100vh - 83px);
+    max-height: calc(100vh - 83px);
+}
+
+#image-history-gallery, #image-history-gallery > :nth-child(2) {
+    height: calc(100vh - 174px);
+    max-height: calc(100vh - 174px);
+}
+
+/* Additional CSS for the paginated image gallery */
+
+/* Page info styling */
+#image-page-info {
+    display: flex;
+    align-items: center;
+    justify-content: center;
+    min-width: 200px;
+    font-size: 0.9em;
+    color: var(--body-text-color-subdued);
+}
+
+/* Settings display panel */
+#image-ai-tab .settings-display-panel {
+    background: var(--background-fill-secondary);
+    padding: 12px;
+    border-radius: 8px;
+    font-size: 0.9em;
+    max-height: 300px;
+    overflow-y: auto;
+    margin-top: 8px;
+}
+
+/* Gallery status message */
+#image-ai-tab .gallery-status {
+    color: var(--color-accent);
+    font-size: 0.85em;
+    margin-top: 4px;
+}
+
+/* Pagination button row alignment */
+#image-ai-tab .pagination-controls {
+    display: flex;
+    align-items: center;
+    gap: 8px;
+    flex-wrap: wrap;
+}
+
+/* Selected image preview container */
+#image-ai-tab .selected-preview-container {
+    border: 1px solid var(--border-color-primary);
+    border-radius: 8px;
+    padding: 8px;
+    background: var(--background-fill-secondary);
+}
+
+/* Fix a gr.Markdown UI glitch when clicking Next in the
+ * Image AI > Gallery tab */
+.min.svelte-1yrv54 {
+    min-height: 0;
+}
diff --git a/docs/Image Generation Tutorial.md b/docs/Image Generation Tutorial.md
new file mode 100644
index 0000000000..e7022c34f1
--- /dev/null
+++ b/docs/Image Generation Tutorial.md	
@@ -0,0 +1,20 @@
+# Image Generation Tutorial
+
+This feature allows you to generate images using high-speed models like Z-Image-Turbo directly within the web UI.
+
+## How to use
+
+1. Click on the **Image AI** tab at the top of the interface.
+2. Select the **Model** sub-tab.
+3. Copy and paste the following link into the **Download model** box:
+
+```
+https://huggingface.co/Tongyi-MAI/Z-Image-Turbo
+```
+
+4. Click the **Download** button and wait for the confirmation message.
+5. In the **Model** dropdown menu, select the model you just downloaded (if you don't see it, click the 🔄 refresh button).
+6. Click **Load**.
+7. Go to the **Generate** sub-tab, type a prompt, and click **GENERATE**.
+
+> **Note for Z-Image-Turbo:** For the best results with this specific model, keep the **CFG Scale** slider at **0**.
diff --git a/modules/image_models.py b/modules/image_models.py
new file mode 100644
index 0000000000..e6f9a17273
--- /dev/null
+++ b/modules/image_models.py
@@ -0,0 +1,183 @@
+import time
+
+import modules.shared as shared
+from modules.logging_colors import logger
+from modules.torch_utils import get_device
+from modules.utils import resolve_model_path
+
+
+def get_quantization_config(quant_method):
+    """
+    Get the appropriate quantization config based on the selected method.
+
+    Args:
+        quant_method: One of 'none', 'bnb-8bit', 'bnb-4bit', 'quanto-8bit', 'quanto-4bit', 'quanto-2bit'
+
+    Returns:
+        PipelineQuantizationConfig or None
+    """
+    import torch
+    from diffusers import BitsAndBytesConfig, QuantoConfig
+    from diffusers.quantizers import PipelineQuantizationConfig
+
+    if quant_method == 'none' or not quant_method:
+        return None
+
+    # Bitsandbytes 8-bit quantization
+    elif quant_method == 'bnb-8bit':
+        return PipelineQuantizationConfig(
+            quant_mapping={
+                "transformer": BitsAndBytesConfig(
+                    load_in_8bit=True
+                )
+            }
+        )
+
+    # Bitsandbytes 4-bit quantization
+    elif quant_method == 'bnb-4bit':
+        return PipelineQuantizationConfig(
+            quant_mapping={
+                "transformer": BitsAndBytesConfig(
+                    load_in_4bit=True,
+                    bnb_4bit_quant_type="nf4",
+                    bnb_4bit_compute_dtype=torch.bfloat16,
+                    bnb_4bit_use_double_quant=True
+                )
+            }
+        )
+
+    # Quanto 8-bit quantization
+    elif quant_method == 'quanto-8bit':
+        return PipelineQuantizationConfig(
+            quant_mapping={
+                "transformer": QuantoConfig(weights_dtype="int8")
+            }
+        )
+
+    # Quanto 4-bit quantization
+    elif quant_method == 'quanto-4bit':
+        return PipelineQuantizationConfig(
+            quant_mapping={
+                "transformer": QuantoConfig(weights_dtype="int4")
+            }
+        )
+
+    # Quanto 2-bit quantization
+    elif quant_method == 'quanto-2bit':
+        return PipelineQuantizationConfig(
+            quant_mapping={
+                "transformer": QuantoConfig(weights_dtype="int2")
+            }
+        )
+
+    else:
+        logger.warning(f"Unknown quantization method: {quant_method}. Loading without quantization.")
+        return None
+
+
+def get_pipeline_type(pipe):
+    """
+    Detect the pipeline type based on the loaded pipeline class.
+
+    Returns:
+        str: 'zimage', 'qwenimage', or 'unknown'
+    """
+    class_name = pipe.__class__.__name__
+    if class_name == 'ZImagePipeline':
+        return 'zimage'
+    elif class_name == 'QwenImagePipeline':
+        return 'qwenimage'
+    else:
+        return 'unknown'
+
+
+def load_image_model(model_name, dtype='bfloat16', attn_backend='sdpa', cpu_offload=False, compile_model=False, quant_method='none'):
+    """
+    Load a diffusers image generation model.
+
+    Args:
+        model_name: Name of the model directory
+        dtype: 'bfloat16' or 'float16'
+        attn_backend: 'sdpa', 'flash_attention_2', or 'flash_attention_3'
+        cpu_offload: Enable CPU offloading for low VRAM
+        compile_model: Compile the model for faster inference (slow first run)
+        quant_method: Quantization method - 'none', 'bnb-8bit', 'bnb-4bit', 'quanto-8bit', 'quanto-4bit', 'quanto-2bit'
+    """
+    import torch
+    from diffusers import DiffusionPipeline
+
+    logger.info(f"Loading image model \"{model_name}\" with quantization: {quant_method}")
+    t0 = time.time()
+
+    dtype_map = {"bfloat16": torch.bfloat16, "float16": torch.float16}
+    target_dtype = dtype_map.get(dtype, torch.bfloat16)
+
+    model_path = resolve_model_path(model_name, image_model=True)
+
+    try:
+        # Get quantization config based on selected method
+        pipeline_quant_config = get_quantization_config(quant_method)
+
+        # Load the pipeline
+        load_kwargs = {
+            "torch_dtype": target_dtype,
+            "low_cpu_mem_usage": True,
+        }
+
+        if pipeline_quant_config is not None:
+            load_kwargs["quantization_config"] = pipeline_quant_config
+
+        # Use DiffusionPipeline for automatic pipeline detection
+        # This handles both ZImagePipeline and QwenImagePipeline
+        pipe = DiffusionPipeline.from_pretrained(
+            str(model_path),
+            **load_kwargs
+        )
+
+        pipeline_type = get_pipeline_type(pipe)
+
+        if not cpu_offload:
+            pipe.to(get_device())
+
+        # Set attention backend (if supported by the pipeline)
+        if hasattr(pipe, 'transformer') and hasattr(pipe.transformer, 'set_attention_backend'):
+            if attn_backend == 'flash_attention_2':
+                pipe.transformer.set_attention_backend("flash")
+            elif attn_backend == 'flash_attention_3':
+                pipe.transformer.set_attention_backend("_flash_3")
+            # sdpa is the default, no action needed
+
+        if compile_model:
+            if hasattr(pipe, 'transformer') and hasattr(pipe.transformer, 'compile'):
+                logger.info("Compiling model (first run will be slow)...")
+                pipe.transformer.compile()
+
+        if cpu_offload:
+            pipe.enable_model_cpu_offload()
+
+        shared.image_model = pipe
+        shared.image_model_name = model_name
+        shared.image_pipeline_type = pipeline_type
+
+        logger.info(f"Loaded image model \"{model_name}\" in {(time.time() - t0):.2f} seconds.")
+        return pipe
+
+    except Exception as e:
+        logger.error(f"Failed to load image model: {str(e)}")
+        return None
+
+
+def unload_image_model():
+    """Unload the current image model and free VRAM."""
+    if shared.image_model is None:
+        return
+
+    del shared.image_model
+    shared.image_model = None
+    shared.image_model_name = 'None'
+    shared.image_pipeline_type = None
+
+    from modules.torch_utils import clear_torch_cache
+    clear_torch_cache()
+
+    logger.info("Image model unloaded.")
diff --git a/modules/shared.py b/modules/shared.py
index 134c0caccc..316f7729b7 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -11,7 +11,7 @@
 from modules.logging_colors import logger
 from modules.presets import default_preset
 
-# Model variables
+# Text model variables
 model = None
 tokenizer = None
 model_name = 'None'
@@ -20,6 +20,11 @@
 model_dirty_from_training = False
 lora_names = []
 
+# Image model variables
+image_model = None
+image_model_name = 'None'
+image_pipeline_type = None
+
 # Generation variables
 stop_everything = False
 generation_lock = None
@@ -46,6 +51,18 @@
 group.add_argument('--verbose', action='store_true', help='Print the prompts to the terminal.')
 group.add_argument('--idle-timeout', type=int, default=0, help='Unload model after this many minutes of inactivity. It will be automatically reloaded when you try to use it again.')
 
+# Image generation
+group = parser.add_argument_group('Image model')
+group.add_argument('--image-model', type=str, help='Name of the image model to select on startup (overrides saved setting).')
+group.add_argument('--image-model-dir', type=str, default='user_data/image_models', help='Path to directory with all the image models.')
+group.add_argument('--image-dtype', type=str, default=None, choices=['bfloat16', 'float16'], help='Data type for image model.')
+group.add_argument('--image-attn-backend', type=str, default=None, choices=['sdpa', 'flash_attention_2', 'flash_attention_3'], help='Attention backend for image model.')
+group.add_argument('--image-cpu-offload', action='store_true', help='Enable CPU offloading for image model.')
+group.add_argument('--image-compile', action='store_true', help='Compile the image model for faster inference.')
+group.add_argument('--image-quant', type=str, default=None,
+                   choices=['none', 'bnb-8bit', 'bnb-4bit', 'quanto-8bit', 'quanto-4bit', 'quanto-2bit'],
+                   help='Quantization method for image model.')
+
 # Model loader
 group = parser.add_argument_group('Model loader')
 group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF, ExLlamav2, TensorRT-LLM.')
@@ -290,6 +307,24 @@
 
     # Extensions
     'default_extensions': [],
+
+    # Image generation settings
+    'image_prompt': '',
+    'image_neg_prompt': '',
+    'image_width': 1024,
+    'image_height': 1024,
+    'image_aspect_ratio': '1:1 Square',
+    'image_steps': 9,
+    'image_cfg_scale': 0.0,
+    'image_seed': -1,
+    'image_batch_size': 1,
+    'image_batch_count': 1,
+    'image_model_menu': 'None',
+    'image_dtype': 'bfloat16',
+    'image_attn_backend': 'sdpa',
+    'image_cpu_offload': False,
+    'image_compile': False,
+    'image_quant': 'none',
 }
 
 default_settings = copy.deepcopy(settings)
@@ -314,6 +349,22 @@ def do_cmd_flags_warnings():
                 logger.warning('\nThe multi-user mode is highly experimental and should not be shared publicly.')
 
 
+def apply_image_model_cli_overrides():
+    """Apply command-line overrides for image model settings."""
+    if args.image_model is not None:
+        settings['image_model_menu'] = args.image_model
+    if args.image_dtype is not None:
+        settings['image_dtype'] = args.image_dtype
+    if args.image_attn_backend is not None:
+        settings['image_attn_backend'] = args.image_attn_backend
+    if args.image_cpu_offload:
+        settings['image_cpu_offload'] = True
+    if args.image_compile:
+        settings['image_compile'] = True
+    if args.image_quant is not None:
+        settings['image_quant'] = args.image_quant
+
+
 def fix_loader_name(name):
     if not name:
         return name
diff --git a/modules/ui.py b/modules/ui.py
index f99e8b6a7f..9700d29711 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -280,6 +280,26 @@ def list_interface_input_elements():
         'include_past_attachments',
     ]
 
+    # Image generation elements
+    elements += [
+        'image_prompt',
+        'image_neg_prompt',
+        'image_width',
+        'image_height',
+        'image_aspect_ratio',
+        'image_steps',
+        'image_cfg_scale',
+        'image_seed',
+        'image_batch_size',
+        'image_batch_count',
+        'image_model_menu',
+        'image_dtype',
+        'image_attn_backend',
+        'image_compile',
+        'image_cpu_offload',
+        'image_quant',
+    ]
+
     return elements
 
 
@@ -509,7 +529,25 @@ def setup_auto_save():
         'theme_state',
         'show_two_notebook_columns',
         'paste_to_attachment',
-        'include_past_attachments'
+        'include_past_attachments',
+
+        # Image generation tab (ui_image_generation.py)
+        'image_prompt',
+        'image_neg_prompt',
+        'image_width',
+        'image_height',
+        'image_aspect_ratio',
+        'image_steps',
+        'image_cfg_scale',
+        'image_seed',
+        'image_batch_size',
+        'image_batch_count',
+        'image_model_menu',
+        'image_dtype',
+        'image_attn_backend',
+        'image_compile',
+        'image_cpu_offload',
+        'image_quant',
     ]
 
     for element_name in change_elements:
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
new file mode 100644
index 0000000000..92c447c88e
--- /dev/null
+++ b/modules/ui_image_generation.py
@@ -0,0 +1,847 @@
+import json
+import os
+import time
+import traceback
+from datetime import datetime
+from pathlib import Path
+
+import gradio as gr
+import numpy as np
+from PIL import Image
+from PIL.PngImagePlugin import PngInfo
+
+from modules import shared, ui, utils
+from modules.image_models import (
+    get_pipeline_type,
+    load_image_model,
+    unload_image_model
+)
+from modules.logging_colors import logger
+from modules.text_generation import stop_everything_event
+from modules.torch_utils import get_device
+from modules.utils import gradio
+
+ASPECT_RATIOS = {
+    "1:1 Square": (1, 1),
+    "16:9 Cinema": (16, 9),
+    "9:16 Mobile": (9, 16),
+    "4:3 Photo": (4, 3),
+    "Custom": None,
+}
+
+STEP = 16
+IMAGES_PER_PAGE = 64
+
+# Settings keys to save in PNG metadata (Generate tab only)
+METADATA_SETTINGS_KEYS = [
+    'image_prompt',
+    'image_neg_prompt',
+    'image_width',
+    'image_height',
+    'image_aspect_ratio',
+    'image_steps',
+    'image_seed',
+    'image_batch_size',
+    'image_batch_count',
+    'image_cfg_scale',
+]
+
+# Cache for all image paths
+_image_cache = []
+_cache_timestamp = 0
+
+
+def round_to_step(value, step=STEP):
+    return round(value / step) * step
+
+
+def clamp(value, min_val, max_val):
+    return max(min_val, min(max_val, value))
+
+
+def apply_aspect_ratio(aspect_ratio, current_width, current_height):
+    if aspect_ratio == "Custom" or aspect_ratio not in ASPECT_RATIOS:
+        return current_width, current_height
+
+    w_ratio, h_ratio = ASPECT_RATIOS[aspect_ratio]
+
+    if w_ratio == h_ratio:
+        base = min(current_width, current_height)
+        new_width = base
+        new_height = base
+    elif w_ratio < h_ratio:
+        new_width = current_width
+        new_height = round_to_step(current_width * h_ratio / w_ratio)
+    else:
+        new_height = current_height
+        new_width = round_to_step(current_height * w_ratio / h_ratio)
+
+    new_width = clamp(new_width, 256, 2048)
+    new_height = clamp(new_height, 256, 2048)
+
+    return int(new_width), int(new_height)
+
+
+def update_height_from_width(width, aspect_ratio):
+    if aspect_ratio == "Custom" or aspect_ratio not in ASPECT_RATIOS:
+        return gr.update()
+
+    w_ratio, h_ratio = ASPECT_RATIOS[aspect_ratio]
+    new_height = round_to_step(width * h_ratio / w_ratio)
+    new_height = clamp(new_height, 256, 2048)
+
+    return int(new_height)
+
+
+def update_width_from_height(height, aspect_ratio):
+    if aspect_ratio == "Custom" or aspect_ratio not in ASPECT_RATIOS:
+        return gr.update()
+
+    w_ratio, h_ratio = ASPECT_RATIOS[aspect_ratio]
+    new_width = round_to_step(height * w_ratio / h_ratio)
+    new_width = clamp(new_width, 256, 2048)
+
+    return int(new_width)
+
+
+def swap_dimensions_and_update_ratio(width, height, aspect_ratio):
+    new_width, new_height = height, width
+
+    new_ratio = "Custom"
+    for name, ratios in ASPECT_RATIOS.items():
+        if ratios is None:
+            continue
+        w_r, h_r = ratios
+        expected_height = new_width * h_r / w_r
+        if abs(expected_height - new_height) < STEP:
+            new_ratio = name
+            break
+
+    return new_width, new_height, new_ratio
+
+
+def build_generation_metadata(state, actual_seed):
+    """Build metadata dict from generation settings."""
+    metadata = {}
+    for key in METADATA_SETTINGS_KEYS:
+        if key in state:
+            metadata[key] = state[key]
+
+    # Store the actual seed used (not -1)
+    metadata['image_seed'] = actual_seed
+    metadata['generated_at'] = datetime.now().isoformat()
+    metadata['model'] = shared.image_model_name
+
+    return metadata
+
+
+def save_generated_images(images, state, actual_seed):
+    """Save images with generation metadata embedded in PNG."""
+    date_str = datetime.now().strftime("%Y-%m-%d")
+    folder_path = os.path.join("user_data", "image_outputs", date_str)
+    os.makedirs(folder_path, exist_ok=True)
+
+    metadata = build_generation_metadata(state, actual_seed)
+    metadata_json = json.dumps(metadata, ensure_ascii=False)
+
+    for idx, img in enumerate(images):
+        timestamp = datetime.now().strftime("%H-%M-%S")
+        filename = f"{timestamp}_{actual_seed:010d}_{idx:03d}.png"
+        filepath = os.path.join(folder_path, filename)
+
+        # Create PNG metadata
+        png_info = PngInfo()
+        png_info.add_text("image_gen_settings", metadata_json)
+
+        # Save with metadata
+        img.save(filepath, pnginfo=png_info)
+
+
+def read_image_metadata(image_path):
+    """Read generation metadata from PNG file."""
+    try:
+        with Image.open(image_path) as img:
+            if hasattr(img, 'text') and 'image_gen_settings' in img.text:
+                return json.loads(img.text['image_gen_settings'])
+    except Exception as e:
+        logger.debug(f"Could not read metadata from {image_path}: {e}")
+    return None
+
+
+def format_metadata_for_display(metadata):
+    """Format metadata as readable text."""
+    if not metadata:
+        return "No generation settings found in this image."
+
+    lines = ["**Generation Settings**", ""]
+
+    # Display in a nice order
+    display_order = [
+        ('image_prompt', 'Prompt'),
+        ('image_neg_prompt', 'Negative Prompt'),
+        ('image_width', 'Width'),
+        ('image_height', 'Height'),
+        ('image_aspect_ratio', 'Aspect Ratio'),
+        ('image_steps', 'Steps'),
+        ('image_cfg_scale', 'CFG Scale'),
+        ('image_seed', 'Seed'),
+        ('image_batch_size', 'Batch Size'),
+        ('image_batch_count', 'Batch Count'),
+        ('model', 'Model'),
+        ('generated_at', 'Generated At'),
+    ]
+
+    for key, label in display_order:
+        if key in metadata:
+            value = metadata[key]
+            if key in ['image_prompt', 'image_neg_prompt'] and value:
+                # Truncate long prompts for display
+                if len(str(value)) > 200:
+                    value = str(value)[:200] + "..."
+            lines.append(f"**{label}:** {value}")
+
+    return "\n\n".join(lines)
+
+
+def get_all_history_images(force_refresh=False):
+    """Get all history images sorted by modification time (newest first). Uses caching."""
+    global _image_cache, _cache_timestamp
+
+    output_dir = os.path.join("user_data", "image_outputs")
+    if not os.path.exists(output_dir):
+        return []
+
+    # Check if we need to refresh cache
+    current_time = time.time()
+    if not force_refresh and _image_cache and (current_time - _cache_timestamp) < 2:
+        return _image_cache
+
+    image_files = []
+    for root, _, files in os.walk(output_dir):
+        for file in files:
+            if file.endswith((".png", ".jpg", ".jpeg")):
+                full_path = os.path.join(root, file)
+                image_files.append((full_path, os.path.getmtime(full_path)))
+
+    image_files.sort(key=lambda x: x[1], reverse=True)
+    _image_cache = [x[0] for x in image_files]
+    _cache_timestamp = current_time
+
+    return _image_cache
+
+
+def get_paginated_images(page=0, force_refresh=False):
+    """Get images for a specific page."""
+    all_images = get_all_history_images(force_refresh)
+    total_images = len(all_images)
+    total_pages = max(1, (total_images + IMAGES_PER_PAGE - 1) // IMAGES_PER_PAGE)
+
+    # Clamp page to valid range
+    page = max(0, min(page, total_pages - 1))
+
+    start_idx = page * IMAGES_PER_PAGE
+    end_idx = min(start_idx + IMAGES_PER_PAGE, total_images)
+
+    page_images = all_images[start_idx:end_idx]
+
+    return page_images, page, total_pages, total_images
+
+
+def get_initial_page_info():
+    """Get page info string for initial load."""
+    _, page, total_pages, total_images = get_paginated_images(0)
+    return f"Page {page + 1} of {total_pages} ({total_images} total images)"
+
+
+def refresh_gallery(current_page=0):
+    """Refresh gallery with current page."""
+    images, page, total_pages, total_images = get_paginated_images(current_page, force_refresh=True)
+    page_info = f"Page {page + 1} of {total_pages} ({total_images} total images)"
+    return images, page, page_info
+
+
+def go_to_page(page_num, current_page):
+    """Go to a specific page (1-indexed input)."""
+    try:
+        page = int(page_num) - 1  # Convert to 0-indexed
+    except (ValueError, TypeError):
+        page = current_page
+
+    images, page, total_pages, total_images = get_paginated_images(page)
+    page_info = f"Page {page + 1} of {total_pages} ({total_images} total images)"
+    return images, page, page_info
+
+
+def next_page(current_page):
+    """Go to next page."""
+    images, page, total_pages, total_images = get_paginated_images(current_page + 1)
+    page_info = f"Page {page + 1} of {total_pages} ({total_images} total images)"
+    return images, page, page_info
+
+
+def prev_page(current_page):
+    """Go to previous page."""
+    images, page, total_pages, total_images = get_paginated_images(current_page - 1)
+    page_info = f"Page {page + 1} of {total_pages} ({total_images} total images)"
+    return images, page, page_info
+
+
+def on_gallery_select(evt: gr.SelectData, current_page):
+    """Handle image selection from gallery."""
+    if evt.index is None:
+        return "", "Select an image to view its settings"
+
+    # Get the current page's images to find the actual file path
+    all_images = get_all_history_images()
+    total_images = len(all_images)
+
+    # Calculate the actual index in the full list
+    start_idx = current_page * IMAGES_PER_PAGE
+    actual_idx = start_idx + evt.index
+
+    if actual_idx >= total_images:
+        return "", "Image not found"
+
+    image_path = all_images[actual_idx]
+    metadata = read_image_metadata(image_path)
+    metadata_display = format_metadata_for_display(metadata)
+
+    return image_path, metadata_display
+
+
+def send_to_generate(selected_image_path):
+    """Load settings from selected image and return updates for all Generate tab inputs."""
+    if not selected_image_path or not os.path.exists(selected_image_path):
+        return [gr.update()] * 10 + ["No image selected"]
+
+    metadata = read_image_metadata(selected_image_path)
+    if not metadata:
+        return [gr.update()] * 10 + ["No settings found in this image"]
+
+    # Return updates for each input element in order
+    updates = [
+        gr.update(value=metadata.get('image_prompt', '')),
+        gr.update(value=metadata.get('image_neg_prompt', '')),
+        gr.update(value=metadata.get('image_width', 1024)),
+        gr.update(value=metadata.get('image_height', 1024)),
+        gr.update(value=metadata.get('image_aspect_ratio', '1:1 Square')),
+        gr.update(value=metadata.get('image_steps', 9)),
+        gr.update(value=metadata.get('image_seed', -1)),
+        gr.update(value=metadata.get('image_batch_size', 1)),
+        gr.update(value=metadata.get('image_batch_count', 1)),
+        gr.update(value=metadata.get('image_cfg_scale', 0.0)),
+    ]
+
+    status = f"✓ Settings loaded from image (seed: {metadata.get('image_seed', 'unknown')})"
+    return updates + [status]
+
+
+def read_dropped_image_metadata(image_path):
+    """Read metadata from a dropped/uploaded image."""
+    if not image_path:
+        return "Drop an image to view its generation settings."
+
+    metadata = read_image_metadata(image_path)
+    return format_metadata_for_display(metadata)
+
+
+def create_ui():
+    if shared.settings['image_model_menu'] != 'None':
+        shared.image_model_name = shared.settings['image_model_menu']
+
+    with gr.Tab("Image AI", elem_id="image-ai-tab"):
+        with gr.Tabs():
+            # TAB 1: GENERATE
+            with gr.TabItem("Generate"):
+                with gr.Row():
+                    with gr.Column(scale=4, min_width=350):
+                        shared.gradio['image_prompt'] = gr.Textbox(
+                            label="Prompt",
+                            placeholder="Describe your imagination...",
+                            lines=3,
+                            autofocus=True,
+                            value=shared.settings['image_prompt']
+                        )
+                        shared.gradio['image_neg_prompt'] = gr.Textbox(
+                            label="Negative Prompt",
+                            placeholder="Low quality...",
+                            lines=3,
+                            value=shared.settings['image_neg_prompt']
+                        )
+
+                        shared.gradio['image_generate_btn'] = gr.Button("Generate", variant="primary", size="lg")
+                        shared.gradio['image_stop_btn'] = gr.Button("Stop", size="lg", visible=False)
+                        gr.HTML("
            ")
+
+                        gr.Markdown("### Dimensions")
+                        with gr.Row():
+                            with gr.Column():
+                                shared.gradio['image_width'] = gr.Slider(256, 2048, value=shared.settings['image_width'], step=STEP, label="Width")
+                            with gr.Column():
+                                shared.gradio['image_height'] = gr.Slider(256, 2048, value=shared.settings['image_height'], step=STEP, label="Height")
+                            shared.gradio['image_swap_btn'] = gr.Button("⇄ Swap", elem_classes='refresh-button', scale=0, min_width=80, elem_id="swap-height-width")
+
+                        with gr.Row():
+                            shared.gradio['image_aspect_ratio'] = gr.Radio(
+                                choices=["1:1 Square", "16:9 Cinema", "9:16 Mobile", "4:3 Photo", "Custom"],
+                                value=shared.settings['image_aspect_ratio'],
+                                label="Aspect Ratio",
+                                interactive=True
+                            )
+
+                        gr.Markdown("### Config")
+                        with gr.Row():
+                            with gr.Column():
+                                shared.gradio['image_steps'] = gr.Slider(1, 100, value=shared.settings['image_steps'], step=1, label="Steps")
+                                shared.gradio['image_cfg_scale'] = gr.Slider(
+                                    0.0, 10.0,
+                                    value=shared.settings['image_cfg_scale'],
+                                    step=0.1,
+                                    label="CFG Scale",
+                                    info="Z-Image Turbo: 0.0 | Qwen: 4.0"
+                                )
+                                shared.gradio['image_seed'] = gr.Number(label="Seed", value=shared.settings['image_seed'], precision=0, info="-1 = Random")
+                            with gr.Column():
+                                shared.gradio['image_batch_size'] = gr.Slider(1, 32, value=shared.settings['image_batch_size'], step=1, label="Batch Size (VRAM Heavy)", info="Generates N images at once.")
+                                shared.gradio['image_batch_count'] = gr.Slider(1, 128, value=shared.settings['image_batch_count'], step=1, label="Sequential Count (Loop)", info="Repeats the generation N times.")
+
+                    with gr.Column(scale=6, min_width=500):
+                        with gr.Column(elem_classes=["viewport-container"]):
+                            shared.gradio['image_output_gallery'] = gr.Gallery(label="Output", show_label=False, columns=2, rows=2, height="80vh", object_fit="contain", preview=True, elem_id="image-output-gallery")
+
+            # TAB 2: GALLERY (with pagination)
+            with gr.TabItem("Gallery"):
+                with gr.Row():
+                    with gr.Column(scale=3):
+                        # Pagination controls
+                        with gr.Row():
+                            shared.gradio['image_refresh_history'] = gr.Button("🔄 Refresh", elem_classes="refresh-button")
+                            shared.gradio['image_prev_page'] = gr.Button("◀ Prev", elem_classes="refresh-button")
+                            shared.gradio['image_page_info'] = gr.Markdown(value=get_initial_page_info, elem_id="image-page-info")
+                            shared.gradio['image_next_page'] = gr.Button("Next ▶", elem_classes="refresh-button")
+                            shared.gradio['image_page_input'] = gr.Number(value=1, label="Page", precision=0, minimum=1, scale=0, min_width=80)
+                            shared.gradio['image_go_to_page'] = gr.Button("Go", elem_classes="refresh-button", scale=0, min_width=50)
+
+                        # State for current page and selected image path
+                        shared.gradio['image_current_page'] = gr.State(value=0)
+                        shared.gradio['image_selected_path'] = gr.State(value="")
+
+                        # Paginated gallery using gr.Gallery
+                        shared.gradio['image_history_gallery'] = gr.Gallery(
+                            value=lambda: get_paginated_images(0)[0],
+                            label="Image History",
+                            show_label=False,
+                            columns=6,
+                            object_fit="cover",
+                            height="auto",
+                            allow_preview=True,
+                            elem_id="image-history-gallery"
+                        )
+
+                    with gr.Column(scale=1):
+                        gr.Markdown("### Selected Image")
+                        shared.gradio['image_settings_display'] = gr.Markdown("Select an image to view its settings")
+                        shared.gradio['image_send_to_generate'] = gr.Button("Send to Generate", variant="primary")
+                        shared.gradio['image_gallery_status'] = gr.Markdown("")
+
+                        gr.Markdown("### Import Image")
+                        shared.gradio['image_drop_upload'] = gr.Image(
+                            label="Drop image here to view settings",
+                            type="filepath",
+                            height=150
+                        )
+
+            # TAB 3: MODEL
+            with gr.TabItem("Model"):
+                with gr.Row():
+                    with gr.Column():
+                        with gr.Row():
+                            shared.gradio['image_model_menu'] = gr.Dropdown(
+                                choices=utils.get_available_image_models(),
+                                value=shared.settings['image_model_menu'],
+                                label='Model',
+                                elem_classes='slim-dropdown'
+                            )
+                            shared.gradio['image_refresh_models'] = gr.Button("🔄", elem_classes='refresh-button', scale=0, min_width=40)
+                            shared.gradio['image_load_model'] = gr.Button("Load", variant='primary', elem_classes='refresh-button')
+                            shared.gradio['image_unload_model'] = gr.Button("Unload", elem_classes='refresh-button')
+
+                        gr.Markdown("## Settings")
+                        with gr.Row():
+                            with gr.Column():
+                                shared.gradio['image_quant'] = gr.Dropdown(
+                                    label='Quantization',
+                                    choices=['none', 'bnb-8bit', 'bnb-4bit', 'quanto-8bit', 'quanto-4bit', 'quanto-2bit'],
+                                    value=shared.settings['image_quant'],
+                                    info='Quantization method for reduced VRAM usage. Quanto supports lower precisions (2-bit, 4-bit, 8-bit).'
+                                )
+
+                                shared.gradio['image_dtype'] = gr.Dropdown(
+                                    choices=['bfloat16', 'float16'],
+                                    value=shared.settings['image_dtype'],
+                                    label='Data Type',
+                                    info='bfloat16 recommended for modern GPUs'
+                                )
+                                shared.gradio['image_attn_backend'] = gr.Dropdown(
+                                    choices=['sdpa', 'flash_attention_2', 'flash_attention_3'],
+                                    value=shared.settings['image_attn_backend'],
+                                    label='Attention Backend',
+                                    info='SDPA is default. Flash Attention requires compatible GPU.'
+                                )
+                            with gr.Column():
+                                shared.gradio['image_compile'] = gr.Checkbox(
+                                    value=shared.settings['image_compile'],
+                                    label='Compile Model',
+                                    info='Faster inference after first run. First run will be slow.'
+                                )
+                                shared.gradio['image_cpu_offload'] = gr.Checkbox(
+                                    value=shared.settings['image_cpu_offload'],
+                                    label='CPU Offload',
+                                    info='Enable for low VRAM GPUs. Slower but uses less memory.'
+                                )
+
+                    with gr.Column():
+                        shared.gradio['image_download_path'] = gr.Textbox(
+                            label="Download model",
+                            placeholder="Tongyi-MAI/Z-Image-Turbo",
+                            info="Enter HuggingFace path. Use : for branch, e.g. user/model:main"
+                        )
+                        shared.gradio['image_download_btn'] = gr.Button("Download", variant='primary')
+                        shared.gradio['image_model_status'] = gr.Markdown(
+                            value=f"Model: **{shared.settings['image_model_menu']}** (not loaded)" if shared.settings['image_model_menu'] != 'None' else "No model selected"
+                        )
+
+
+def create_event_handlers():
+    # Dimension controls
+    shared.gradio['image_aspect_ratio'].change(
+        apply_aspect_ratio,
+        gradio('image_aspect_ratio', 'image_width', 'image_height'),
+        gradio('image_width', 'image_height'),
+        show_progress=False
+    )
+
+    shared.gradio['image_width'].release(
+        update_height_from_width,
+        gradio('image_width', 'image_aspect_ratio'),
+        gradio('image_height'),
+        show_progress=False
+    )
+
+    shared.gradio['image_height'].release(
+        update_width_from_height,
+        gradio('image_height', 'image_aspect_ratio'),
+        gradio('image_width'),
+        show_progress=False
+    )
+
+    shared.gradio['image_swap_btn'].click(
+        swap_dimensions_and_update_ratio,
+        gradio('image_width', 'image_height', 'image_aspect_ratio'),
+        gradio('image_width', 'image_height', 'image_aspect_ratio'),
+        show_progress=False
+    )
+
+    # Generation
+    shared.gradio['image_generate_btn'].click(
+        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
+        lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('image_stop_btn', 'image_generate_btn')).then(
+        generate, gradio('interface_state'), gradio('image_output_gallery'), show_progress=False).then(
+        lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('image_stop_btn', 'image_generate_btn'))
+
+    shared.gradio['image_prompt'].submit(
+        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
+        lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('image_stop_btn', 'image_generate_btn')).then(
+        generate, gradio('interface_state'), gradio('image_output_gallery'), show_progress=False).then(
+        lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('image_stop_btn', 'image_generate_btn'))
+
+    shared.gradio['image_neg_prompt'].submit(
+        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
+        lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('image_stop_btn', 'image_generate_btn')).then(
+        generate, gradio('interface_state'), gradio('image_output_gallery'), show_progress=False).then(
+        lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('image_stop_btn', 'image_generate_btn'))
+
+    # Stop button
+    shared.gradio['image_stop_btn'].click(
+        stop_everything_event, None, None, show_progress=False
+    )
+
+    # Model management
+    shared.gradio['image_refresh_models'].click(
+        lambda: gr.update(choices=utils.get_available_image_models()),
+        None,
+        gradio('image_model_menu'),
+        show_progress=False
+    )
+
+    shared.gradio['image_load_model'].click(
+        load_image_model_wrapper,
+        gradio('image_model_menu', 'image_dtype', 'image_attn_backend', 'image_cpu_offload', 'image_compile', 'image_quant'),
+        gradio('image_model_status'),
+        show_progress=True
+    )
+
+    shared.gradio['image_unload_model'].click(
+        unload_image_model_wrapper,
+        None,
+        gradio('image_model_status'),
+        show_progress=False
+    )
+
+    shared.gradio['image_download_btn'].click(
+        download_image_model_wrapper,
+        gradio('image_download_path'),
+        gradio('image_model_status', 'image_model_menu'),
+        show_progress=True
+    )
+
+    # Gallery pagination handlers
+    shared.gradio['image_refresh_history'].click(
+        refresh_gallery,
+        gradio('image_current_page'),
+        gradio('image_history_gallery', 'image_current_page', 'image_page_info'),
+        show_progress=False
+    )
+
+    shared.gradio['image_next_page'].click(
+        next_page,
+        gradio('image_current_page'),
+        gradio('image_history_gallery', 'image_current_page', 'image_page_info'),
+        show_progress=False
+    )
+
+    shared.gradio['image_prev_page'].click(
+        prev_page,
+        gradio('image_current_page'),
+        gradio('image_history_gallery', 'image_current_page', 'image_page_info'),
+        show_progress=False
+    )
+
+    shared.gradio['image_go_to_page'].click(
+        go_to_page,
+        gradio('image_page_input', 'image_current_page'),
+        gradio('image_history_gallery', 'image_current_page', 'image_page_info'),
+        show_progress=False
+    )
+
+    # Image selection from gallery
+    shared.gradio['image_history_gallery'].select(
+        on_gallery_select,
+        gradio('image_current_page'),
+        gradio('image_selected_path', 'image_settings_display'),
+        show_progress=False
+    )
+
+    # Send to Generate
+    shared.gradio['image_send_to_generate'].click(
+        send_to_generate,
+        gradio('image_selected_path'),
+        gradio(
+            'image_prompt',
+            'image_neg_prompt',
+            'image_width',
+            'image_height',
+            'image_aspect_ratio',
+            'image_steps',
+            'image_seed',
+            'image_batch_size',
+            'image_batch_count',
+            'image_cfg_scale',
+            'image_gallery_status'
+        ),
+        show_progress=False
+    )
+
+    shared.gradio['image_drop_upload'].change(
+        read_dropped_image_metadata,
+        gradio('image_drop_upload'),
+        gradio('image_settings_display'),
+        show_progress=False
+    )
+
+
+def generate(state):
+    """
+    Generate images using the loaded model.
+    Automatically adjusts parameters based on pipeline type.
+    """
+    import torch
+
+    from modules.torch_utils import clear_torch_cache
+
+    clear_torch_cache()
+
+    try:
+        model_name = state['image_model_menu']
+
+        if not model_name or model_name == 'None':
+            logger.error("No image model selected. Go to the Model tab and select a model.")
+            return []
+
+        if shared.image_model is None:
+            result = load_image_model(
+                model_name,
+                dtype=state['image_dtype'],
+                attn_backend=state['image_attn_backend'],
+                cpu_offload=state['image_cpu_offload'],
+                compile_model=state['image_compile'],
+                quant_method=state['image_quant']
+            )
+            if result is None:
+                logger.error(f"Failed to load model `{model_name}`.")
+                return []
+
+            shared.image_model_name = model_name
+
+        seed = state['image_seed']
+        if seed == -1:
+            seed = np.random.randint(0, 2**32 - 1)
+
+        device = get_device()
+        if device is None:
+            device = "cpu"
+        generator = torch.Generator(device).manual_seed(int(seed))
+
+        all_images = []
+
+        # Get pipeline type for parameter adjustment
+        pipeline_type = getattr(shared, 'image_pipeline_type', None)
+        if pipeline_type is None:
+            pipeline_type = get_pipeline_type(shared.image_model)
+
+        # Process Prompt
+        prompt = state['image_prompt']
+
+        # Apply "Positive Magic" for Qwen models only
+        if pipeline_type == 'qwenimage':
+            magic_suffix = ", Ultra HD, 4K, cinematic composition"
+            # Avoid duplication if user already added it
+            if magic_suffix.strip(", ") not in prompt:
+                prompt += magic_suffix
+
+        # Reset stop flag at start
+        shared.stop_everything = False
+
+        # Callback to check for interruption during diffusion steps
+        def interrupt_callback(pipe, step_index, timestep, callback_kwargs):
+            if shared.stop_everything:
+                pipe._interrupt = True
+
+            return callback_kwargs
+
+        # Build generation kwargs
+        gen_kwargs = {
+            "prompt": prompt,
+            "negative_prompt": state['image_neg_prompt'],
+            "height": int(state['image_height']),
+            "width": int(state['image_width']),
+            "num_inference_steps": int(state['image_steps']),
+            "num_images_per_prompt": int(state['image_batch_size']),
+            "generator": generator,
+            "callback_on_step_end": interrupt_callback,
+        }
+
+        # Add pipeline-specific parameters for CFG
+        cfg_val = state.get('image_cfg_scale', 0.0)
+
+        if pipeline_type == 'qwenimage':
+            # Qwen-Image uses true_cfg_scale (typically 4.0)
+            gen_kwargs["true_cfg_scale"] = cfg_val
+        else:
+            # Z-Image and others use guidance_scale (typically 0.0 for Turbo)
+            gen_kwargs["guidance_scale"] = cfg_val
+
+        t0 = time.time()
+        for i in range(int(state['image_batch_count'])):
+            if shared.stop_everything:
+                break
+
+            generator.manual_seed(int(seed + i))
+            batch_results = shared.image_model(**gen_kwargs).images
+            all_images.extend(batch_results)
+
+        t1 = time.time()
+        save_generated_images(all_images, state, seed)
+
+        total_images = int(state['image_batch_count']) * int(state['image_batch_size'])
+        total_steps = state["image_steps"] * int(state['image_batch_count'])
+        logger.info(f'Generated {total_images} {"image" if total_images == 1 else "images"} in {(t1 - t0):.2f} seconds ({total_steps / (t1 - t0):.2f} steps/s, seed {seed})')
+
+        return all_images
+
+    except Exception as e:
+        logger.error(f"Image generation failed: {e}")
+        traceback.print_exc()
+        return []
+
+
+def load_image_model_wrapper(model_name, dtype, attn_backend, cpu_offload, compile_model, quant_method):
+    if not model_name or model_name == 'None':
+        yield "No model selected"
+        return
+
+    try:
+        yield f"Loading `{model_name}`..."
+        unload_image_model()
+
+        result = load_image_model(
+            model_name,
+            dtype=dtype,
+            attn_backend=attn_backend,
+            cpu_offload=cpu_offload,
+            compile_model=compile_model,
+            quant_method=quant_method
+        )
+
+        if result is not None:
+            shared.image_model_name = model_name
+            yield f"✓ Loaded **{model_name}** (quantization: {quant_method})"
+        else:
+            yield f"✗ Failed to load `{model_name}`"
+    except Exception:
+        yield f"Error:\n```\n{traceback.format_exc()}\n```"
+
+
+def unload_image_model_wrapper():
+    previous_name = shared.image_model_name
+    unload_image_model()
+    if previous_name != 'None':
+        return f"Model: **{previous_name}** (unloaded)"
+    return "No model loaded"
+
+
+def download_image_model_wrapper(model_path):
+    from huggingface_hub import snapshot_download
+
+    if not model_path:
+        yield "No model specified", gr.update()
+        return
+
+    try:
+        model_path = model_path.strip()
+        if model_path.startswith('https://huggingface.co/'):
+            model_path = model_path[len('https://huggingface.co/'):]
+        elif model_path.startswith('huggingface.co/'):
+            model_path = model_path[len('huggingface.co/'):]
+
+        if ':' in model_path:
+            model_id, branch = model_path.rsplit(':', 1)
+        else:
+            model_id, branch = model_path, 'main'
+
+        folder_name = model_id.replace('/', '_')
+        output_folder = Path(shared.args.image_model_dir) / folder_name
+
+        yield f"Downloading `{model_id}` (branch: {branch})...", gr.update()
+
+        snapshot_download(
+            repo_id=model_id,
+            revision=branch,
+            local_dir=output_folder,
+            local_dir_use_symlinks=False,
+        )
+
+        new_choices = utils.get_available_image_models()
+        yield f"✓ Downloaded to `{output_folder}`", gr.update(choices=new_choices, value=folder_name)
+    except Exception:
+        yield f"Error:\n```\n{traceback.format_exc()}\n```", gr.update()
diff --git a/modules/utils.py b/modules/utils.py
index e8d23a0284..b478f0664e 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -86,7 +86,7 @@ def check_model_loaded():
     return True, None
 
 
-def resolve_model_path(model_name_or_path):
+def resolve_model_path(model_name_or_path, image_model=False):
     """
     Resolves a model path, checking for a direct path
     before the default models directory.
@@ -95,6 +95,8 @@ def resolve_model_path(model_name_or_path):
     path_candidate = Path(model_name_or_path)
     if path_candidate.exists():
         return path_candidate
+    elif image_model:
+        return Path(f'{shared.args.image_model_dir}/{model_name_or_path}')
     else:
         return Path(f'{shared.args.model_dir}/{model_name_or_path}')
 
@@ -153,6 +155,24 @@ def get_available_models():
     return filtered_gguf_files + model_dirs
 
 
+def get_available_image_models():
+    model_dir = Path(shared.args.image_model_dir)
+    model_dir.mkdir(parents=True, exist_ok=True)
+
+    # Find valid model directories
+    model_dirs = []
+    for item in os.listdir(model_dir):
+        item_path = model_dir / item
+        if not item_path.is_dir():
+            continue
+
+        model_dirs.append(item)
+
+    model_dirs = sorted(model_dirs, key=natural_keys)
+
+    return model_dirs
+
+
 def get_available_ggufs():
     model_list = []
     model_dir = Path(shared.args.model_dir)
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 4be87c56e1..7076df4fbd 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -11,6 +11,7 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
+optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
@@ -34,6 +35,9 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
+# Diffusers
+diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index d9bf64cb67..d626408928 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -9,6 +9,7 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
+optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
@@ -32,6 +33,9 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
+# Diffusers
+diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 2c06c22f12..985511fac6 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -9,6 +9,7 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
+optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
@@ -32,6 +33,9 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
+# Diffusers
+diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index a4929d4006..1c75806d51 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -9,6 +9,7 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
+optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
@@ -32,6 +33,9 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
+# Diffusers
+diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 6a7ce8a62b..0f536b6a7c 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -9,6 +9,7 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
+optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
@@ -32,6 +33,9 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
+# Diffusers
+diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 05957700fc..0f8970d0d5 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -9,6 +9,7 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
+optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
@@ -32,6 +33,9 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
+# Diffusers
+diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 9e43c409c7..e862c1a194 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -9,6 +9,7 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
+optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
@@ -32,6 +33,9 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
+# Diffusers
+diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 9931e1dcd2..a734ce820b 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -11,6 +11,7 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
+optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
@@ -34,6 +35,9 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
+# Diffusers
+diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 818d22447e..8561462c27 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -9,6 +9,7 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
+optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
@@ -32,6 +33,9 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
+# Diffusers
+diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
+
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/server.py b/server.py
index c804c34232..b02c50a25e 100644
--- a/server.py
+++ b/server.py
@@ -5,6 +5,7 @@
 
 from modules import shared
 from modules.block_requests import OpenMonkeyPatch, RequestBlocker
+from modules.image_models import load_image_model
 from modules.logging_colors import logger
 from modules.prompts import load_prompt
 
@@ -50,6 +51,7 @@
     ui_chat,
     ui_default,
     ui_file_saving,
+    ui_image_generation,
     ui_model_menu,
     ui_notebook,
     ui_parameters,
@@ -163,6 +165,7 @@ def create_interface():
         ui_chat.create_character_settings_ui()  # Character tab
         ui_model_menu.create_ui()  # Model tab
         if not shared.args.portable:
+            ui_image_generation.create_ui()  # Image generation tab
             training.create_ui()  # Training tab
         ui_session.create_ui()  # Session tab
 
@@ -170,6 +173,8 @@ def create_interface():
         ui_chat.create_event_handlers()
         ui_default.create_event_handlers()
         ui_notebook.create_event_handlers()
+        if not shared.args.portable:
+            ui_image_generation.create_event_handlers()
 
         # Other events
         ui_file_saving.create_event_handlers()
@@ -256,6 +261,9 @@ def create_interface():
         if new_settings:
             shared.settings.update(new_settings)
 
+    # Apply CLI overrides for image model settings (CLI flags take precedence over saved settings)
+    shared.apply_image_model_cli_overrides()
+
     # Fallback settings for models
     shared.model_config['.*'] = get_fallback_settings()
     shared.model_config.move_to_end('.*', last=False)  # Move to the beginning
@@ -313,6 +321,22 @@ def create_interface():
         if shared.args.lora:
             add_lora_to_model(shared.args.lora)
 
+    # Load image model if specified via CLI
+    if shared.args.image_model:
+        logger.info(f"Loading image model: {shared.args.image_model}")
+        result = load_image_model(
+            shared.args.image_model,
+            dtype=shared.settings.get('image_dtype', 'bfloat16'),
+            attn_backend=shared.settings.get('image_attn_backend', 'sdpa'),
+            cpu_offload=shared.settings.get('image_cpu_offload', False),
+            compile_model=shared.settings.get('image_compile', False),
+            quant_method=shared.settings.get('image_quant', 'none')
+        )
+        if result is not None:
+            shared.image_model_name = shared.args.image_model
+        else:
+            logger.error(f"Failed to load image model: {shared.args.image_model}")
+
     shared.generation_lock = Lock()
 
     if shared.args.idle_timeout > 0:

From 6291e72129aa807768aa86390498613ee1408419 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 2 Dec 2025 09:57:04 -0800
Subject: [PATCH 1080/1701] Remove quanto for now (requires messy compilation)

---
 modules/image_models.py                       | 28 ++-----------------
 modules/shared.py                             |  2 +-
 modules/ui_image_generation.py                |  2 +-
 requirements/full/requirements.txt            |  1 -
 requirements/full/requirements_amd.txt        |  1 -
 requirements/full/requirements_amd_noavx2.txt |  1 -
 .../full/requirements_apple_intel.txt         |  1 -
 .../full/requirements_apple_silicon.txt       |  1 -
 requirements/full/requirements_cpu_only.txt   |  1 -
 .../full/requirements_cpu_only_noavx2.txt     |  1 -
 requirements/full/requirements_noavx2.txt     |  1 -
 requirements/full/requirements_nowheels.txt   |  1 -
 12 files changed, 4 insertions(+), 37 deletions(-)

diff --git a/modules/image_models.py b/modules/image_models.py
index e6f9a17273..4d7cc5f9fb 100644
--- a/modules/image_models.py
+++ b/modules/image_models.py
@@ -11,7 +11,7 @@ def get_quantization_config(quant_method):
     Get the appropriate quantization config based on the selected method.
 
     Args:
-        quant_method: One of 'none', 'bnb-8bit', 'bnb-4bit', 'quanto-8bit', 'quanto-4bit', 'quanto-2bit'
+        quant_method: One of 'none', 'bnb-8bit', 'bnb-4bit'
 
     Returns:
         PipelineQuantizationConfig or None
@@ -46,30 +46,6 @@ def get_quantization_config(quant_method):
             }
         )
 
-    # Quanto 8-bit quantization
-    elif quant_method == 'quanto-8bit':
-        return PipelineQuantizationConfig(
-            quant_mapping={
-                "transformer": QuantoConfig(weights_dtype="int8")
-            }
-        )
-
-    # Quanto 4-bit quantization
-    elif quant_method == 'quanto-4bit':
-        return PipelineQuantizationConfig(
-            quant_mapping={
-                "transformer": QuantoConfig(weights_dtype="int4")
-            }
-        )
-
-    # Quanto 2-bit quantization
-    elif quant_method == 'quanto-2bit':
-        return PipelineQuantizationConfig(
-            quant_mapping={
-                "transformer": QuantoConfig(weights_dtype="int2")
-            }
-        )
-
     else:
         logger.warning(f"Unknown quantization method: {quant_method}. Loading without quantization.")
         return None
@@ -101,7 +77,7 @@ def load_image_model(model_name, dtype='bfloat16', attn_backend='sdpa', cpu_offl
         attn_backend: 'sdpa', 'flash_attention_2', or 'flash_attention_3'
         cpu_offload: Enable CPU offloading for low VRAM
         compile_model: Compile the model for faster inference (slow first run)
-        quant_method: Quantization method - 'none', 'bnb-8bit', 'bnb-4bit', 'quanto-8bit', 'quanto-4bit', 'quanto-2bit'
+        quant_method: Quantization method - 'none', 'bnb-8bit', 'bnb-4bit'
     """
     import torch
     from diffusers import DiffusionPipeline
diff --git a/modules/shared.py b/modules/shared.py
index 316f7729b7..a96cd70ce5 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -60,7 +60,7 @@
 group.add_argument('--image-cpu-offload', action='store_true', help='Enable CPU offloading for image model.')
 group.add_argument('--image-compile', action='store_true', help='Compile the image model for faster inference.')
 group.add_argument('--image-quant', type=str, default=None,
-                   choices=['none', 'bnb-8bit', 'bnb-4bit', 'quanto-8bit', 'quanto-4bit', 'quanto-2bit'],
+                   choices=['none', 'bnb-8bit', 'bnb-4bit'],
                    help='Quantization method for image model.')
 
 # Model loader
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 92c447c88e..5ef66820e4 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -471,7 +471,7 @@ def create_ui():
                             with gr.Column():
                                 shared.gradio['image_quant'] = gr.Dropdown(
                                     label='Quantization',
-                                    choices=['none', 'bnb-8bit', 'bnb-4bit', 'quanto-8bit', 'quanto-4bit', 'quanto-2bit'],
+                                    choices=['none', 'bnb-8bit', 'bnb-4bit'],
                                     value=shared.settings['image_quant'],
                                     info='Quantization method for reduced VRAM usage. Quanto supports lower precisions (2-bit, 4-bit, 8-bit).'
                                 )
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 7076df4fbd..a031bbabdc 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -11,7 +11,6 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index d626408928..ee8d67acb2 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -9,7 +9,6 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 985511fac6..7b36b15151 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -9,7 +9,6 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 1c75806d51..4f72d5ac7a 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -9,7 +9,6 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 0f536b6a7c..7942b9f01e 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -9,7 +9,6 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 0f8970d0d5..96013c358f 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -9,7 +9,6 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index e862c1a194..0813f06c77 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -9,7 +9,6 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index a734ce820b..ea7edf6c86 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -11,7 +11,6 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 8561462c27..15247d72a3 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -9,7 +9,6 @@ huggingface-hub==0.36.0
 jinja2==3.1.6
 markdown
 numpy==2.2.*
-optimum-quanto==0.2.7
 pandas
 peft==0.18.*
 Pillow>=9.5.0

From 9d07d3a229e3e949cfb32e9ef1fe32b328759985 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 2 Dec 2025 10:06:57 -0800
Subject: [PATCH 1081/1701] Make portable builds functional again after
 b3666e140de349a651aea22a6f418572925a5c62

---
 modules/image_models.py        |  3 +-
 modules/ui.py                  | 77 ++++++++++++++++++----------------
 modules/ui_image_generation.py |  3 +-
 3 files changed, 44 insertions(+), 39 deletions(-)

diff --git a/modules/image_models.py b/modules/image_models.py
index 4d7cc5f9fb..f2e0276ab8 100644
--- a/modules/image_models.py
+++ b/modules/image_models.py
@@ -2,7 +2,6 @@
 
 import modules.shared as shared
 from modules.logging_colors import logger
-from modules.torch_utils import get_device
 from modules.utils import resolve_model_path
 
 
@@ -82,6 +81,8 @@ def load_image_model(model_name, dtype='bfloat16', attn_backend='sdpa', cpu_offl
     import torch
     from diffusers import DiffusionPipeline
 
+    from modules.torch_utils import get_device
+
     logger.info(f"Loading image model \"{model_name}\" with quantization: {quant_method}")
     t0 = time.time()
 
diff --git a/modules/ui.py b/modules/ui.py
index 9700d29711..ff5686e83d 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -280,25 +280,26 @@ def list_interface_input_elements():
         'include_past_attachments',
     ]
 
-    # Image generation elements
-    elements += [
-        'image_prompt',
-        'image_neg_prompt',
-        'image_width',
-        'image_height',
-        'image_aspect_ratio',
-        'image_steps',
-        'image_cfg_scale',
-        'image_seed',
-        'image_batch_size',
-        'image_batch_count',
-        'image_model_menu',
-        'image_dtype',
-        'image_attn_backend',
-        'image_compile',
-        'image_cpu_offload',
-        'image_quant',
-    ]
+    if not shared.args.portable:
+        # Image generation elements
+        elements += [
+            'image_prompt',
+            'image_neg_prompt',
+            'image_width',
+            'image_height',
+            'image_aspect_ratio',
+            'image_steps',
+            'image_cfg_scale',
+            'image_seed',
+            'image_batch_size',
+            'image_batch_count',
+            'image_model_menu',
+            'image_dtype',
+            'image_attn_backend',
+            'image_compile',
+            'image_cpu_offload',
+            'image_quant',
+        ]
 
     return elements
 
@@ -531,25 +532,29 @@ def setup_auto_save():
         'paste_to_attachment',
         'include_past_attachments',
 
-        # Image generation tab (ui_image_generation.py)
-        'image_prompt',
-        'image_neg_prompt',
-        'image_width',
-        'image_height',
-        'image_aspect_ratio',
-        'image_steps',
-        'image_cfg_scale',
-        'image_seed',
-        'image_batch_size',
-        'image_batch_count',
-        'image_model_menu',
-        'image_dtype',
-        'image_attn_backend',
-        'image_compile',
-        'image_cpu_offload',
-        'image_quant',
     ]
 
+    if not shared.args.portable:
+        # Image generation tab (ui_image_generation.py)
+        change_elements += [
+            'image_prompt',
+            'image_neg_prompt',
+            'image_width',
+            'image_height',
+            'image_aspect_ratio',
+            'image_steps',
+            'image_cfg_scale',
+            'image_seed',
+            'image_batch_size',
+            'image_batch_count',
+            'image_model_menu',
+            'image_dtype',
+            'image_attn_backend',
+            'image_compile',
+            'image_cpu_offload',
+            'image_quant',
+        ]
+
     for element_name in change_elements:
         if element_name in shared.gradio:
             shared.gradio[element_name].change(
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 5ef66820e4..4f90d35217 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -18,7 +18,6 @@
 )
 from modules.logging_colors import logger
 from modules.text_generation import stop_everything_event
-from modules.torch_utils import get_device
 from modules.utils import gradio
 
 ASPECT_RATIOS = {
@@ -667,7 +666,7 @@ def generate(state):
     """
     import torch
 
-    from modules.torch_utils import clear_torch_cache
+    from modules.torch_utils import clear_torch_cache, get_device
 
     clear_torch_cache()
 

From 97281ff83120c20180840ce8db3be80a255c861e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 2 Dec 2025 11:20:52 -0800
Subject: [PATCH 1082/1701] UI: Fix an index error in the new image gallery

---
 modules/ui_image_generation.py | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 4f90d35217..8ec879e9ae 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -290,8 +290,11 @@ def on_gallery_select(evt: gr.SelectData, current_page):
     if evt.index is None:
         return "", "Select an image to view its settings"
 
+    if not _image_cache:
+        get_all_history_images()
+
     # Get the current page's images to find the actual file path
-    all_images = get_all_history_images()
+    all_images = _image_cache
     total_images = len(all_images)
 
     # Calculate the actual index in the full list
@@ -506,9 +509,7 @@ def create_ui():
                             info="Enter HuggingFace path. Use : for branch, e.g. user/model:main"
                         )
                         shared.gradio['image_download_btn'] = gr.Button("Download", variant='primary')
-                        shared.gradio['image_model_status'] = gr.Markdown(
-                            value=f"Model: **{shared.settings['image_model_menu']}** (not loaded)" if shared.settings['image_model_menu'] != 'None' else "No model selected"
-                        )
+                        shared.gradio['image_model_status'] = gr.Markdown(value="")
 
 
 def create_event_handlers():

From 9448bf1caa8ad5bd704ae2610bc0b2738c9ca51f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 2 Dec 2025 14:22:51 -0800
Subject: [PATCH 1083/1701] Image generation: add torchao quantization
 (supports torch.compile)

---
 modules/image_models.py                       | 31 +++++++++++++++++--
 modules/shared.py                             |  2 +-
 modules/ui_image_generation.py                |  4 +--
 requirements/full/requirements.txt            |  1 +
 requirements/full/requirements_amd.txt        |  1 +
 requirements/full/requirements_amd_noavx2.txt |  1 +
 .../full/requirements_apple_intel.txt         |  1 +
 .../full/requirements_apple_silicon.txt       |  1 +
 requirements/full/requirements_cpu_only.txt   |  1 +
 .../full/requirements_cpu_only_noavx2.txt     |  1 +
 requirements/full/requirements_noavx2.txt     |  1 +
 requirements/full/requirements_nowheels.txt   |  1 +
 12 files changed, 40 insertions(+), 6 deletions(-)

diff --git a/modules/image_models.py b/modules/image_models.py
index f2e0276ab8..2ef1e7309b 100644
--- a/modules/image_models.py
+++ b/modules/image_models.py
@@ -10,13 +10,14 @@ def get_quantization_config(quant_method):
     Get the appropriate quantization config based on the selected method.
 
     Args:
-        quant_method: One of 'none', 'bnb-8bit', 'bnb-4bit'
+        quant_method: One of 'none', 'bnb-8bit', 'bnb-4bit',
+                      'torchao-int8wo', 'torchao-fp4', 'torchao-float8wo'
 
     Returns:
         PipelineQuantizationConfig or None
     """
     import torch
-    from diffusers import BitsAndBytesConfig, QuantoConfig
+    from diffusers import BitsAndBytesConfig, TorchAoConfig
     from diffusers.quantizers import PipelineQuantizationConfig
 
     if quant_method == 'none' or not quant_method:
@@ -45,6 +46,30 @@ def get_quantization_config(quant_method):
             }
         )
 
+    # torchao int8 weight-only
+    elif quant_method == 'torchao-int8wo':
+        return PipelineQuantizationConfig(
+            quant_mapping={
+                "transformer": TorchAoConfig("int8wo")
+            }
+        )
+
+    # torchao fp4 (e2m1)
+    elif quant_method == 'torchao-fp4':
+        return PipelineQuantizationConfig(
+            quant_mapping={
+                "transformer": TorchAoConfig("fp4_e2m1")
+            }
+        )
+
+    # torchao float8 weight-only
+    elif quant_method == 'torchao-float8wo':
+        return PipelineQuantizationConfig(
+            quant_mapping={
+                "transformer": TorchAoConfig("float8wo")
+            }
+        )
+
     else:
         logger.warning(f"Unknown quantization method: {quant_method}. Loading without quantization.")
         return None
@@ -76,7 +101,7 @@ def load_image_model(model_name, dtype='bfloat16', attn_backend='sdpa', cpu_offl
         attn_backend: 'sdpa', 'flash_attention_2', or 'flash_attention_3'
         cpu_offload: Enable CPU offloading for low VRAM
         compile_model: Compile the model for faster inference (slow first run)
-        quant_method: Quantization method - 'none', 'bnb-8bit', 'bnb-4bit'
+        quant_method: 'none', 'bnb-8bit', 'bnb-4bit', or torchao options (int8wo, fp4, float8wo)
     """
     import torch
     from diffusers import DiffusionPipeline
diff --git a/modules/shared.py b/modules/shared.py
index a96cd70ce5..4a0fd986a2 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -60,7 +60,7 @@
 group.add_argument('--image-cpu-offload', action='store_true', help='Enable CPU offloading for image model.')
 group.add_argument('--image-compile', action='store_true', help='Compile the image model for faster inference.')
 group.add_argument('--image-quant', type=str, default=None,
-                   choices=['none', 'bnb-8bit', 'bnb-4bit'],
+                   choices=['none', 'bnb-8bit', 'bnb-4bit', 'torchao-int8wo', 'torchao-fp4', 'torchao-float8wo'],
                    help='Quantization method for image model.')
 
 # Model loader
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 8ec879e9ae..1b0e58f361 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -473,9 +473,9 @@ def create_ui():
                             with gr.Column():
                                 shared.gradio['image_quant'] = gr.Dropdown(
                                     label='Quantization',
-                                    choices=['none', 'bnb-8bit', 'bnb-4bit'],
+                                    choices=['none', 'bnb-8bit', 'bnb-4bit', 'torchao-int8wo', 'torchao-fp4', 'torchao-float8wo'],
                                     value=shared.settings['image_quant'],
-                                    info='Quantization method for reduced VRAM usage. Quanto supports lower precisions (2-bit, 4-bit, 8-bit).'
+                                    info='BnB: bitsandbytes quantization. torchao: int8wo, fp4, float8wo.'
                                 )
 
                                 shared.gradio['image_dtype'] = gr.Dropdown(
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index a031bbabdc..4e02d76f0b 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -25,6 +25,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
+torchao==0.14.*
 transformers==4.57.*
 triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index ee8d67acb2..c0fbd9ab85 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -23,6 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
+torchao==0.14.*
 transformers==4.57.*
 triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 7b36b15151..b330646a90 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -23,6 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
+torchao==0.14.*
 transformers==4.57.*
 triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 4f72d5ac7a..c2c643375f 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -23,6 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
+torchao==0.14.*
 transformers==4.57.*
 triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 7942b9f01e..10889215a5 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -23,6 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
+torchao==0.14.*
 transformers==4.57.*
 triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 96013c358f..ffcd64736b 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -23,6 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
+torchao==0.14.*
 transformers==4.57.*
 triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 0813f06c77..daa2444ffe 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -23,6 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
+torchao==0.14.*
 transformers==4.57.*
 triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index ea7edf6c86..a43a77244a 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -25,6 +25,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
+torchao==0.14.*
 transformers==4.57.*
 triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 15247d72a3..ac24f0cbc9 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -23,6 +23,7 @@ safetensors==0.6.*
 scipy
 sentencepiece
 tensorboard
+torchao==0.14.*
 transformers==4.57.*
 triton-windows==3.5.1.post21; platform_system == "Windows"
 tqdm

From 5433ef33331b3b24da9d58b73287398457459de1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 3 Dec 2025 11:50:35 -0800
Subject: [PATCH 1084/1701] Add an API endpoint for generating images

---
 extensions/openai/images.py | 188 ++++++++++++++++++++++++++----------
 extensions/openai/script.py |  52 +++++-----
 extensions/openai/typing.py |  40 ++++++++
 3 files changed, 204 insertions(+), 76 deletions(-)

diff --git a/extensions/openai/images.py b/extensions/openai/images.py
index 92bd85f08b..3a5288e626 100644
--- a/extensions/openai/images.py
+++ b/extensions/openai/images.py
@@ -1,70 +1,154 @@
+"""
+OpenAI-compatible image generation using local diffusion models.
+"""
+
+import base64
+import io
+import json
 import os
 import time
+from datetime import datetime
 
-import requests
-
+import numpy as np
 from extensions.openai.errors import ServiceUnavailableError
+from modules import shared
+from modules.logging_colors import logger
+from PIL.PngImagePlugin import PngInfo
 
 
-def generations(prompt: str, size: str, response_format: str, n: int):
-    # Stable Diffusion callout wrapper for txt2img
-    # Low effort implementation for compatibility. With only "prompt" being passed and assuming DALL-E
-    # the results will be limited and likely poor. SD has hundreds of models and dozens of settings.
-    # If you want high quality tailored results you should just use the Stable Diffusion API directly.
-    # it's too general an API to try and shape the result with specific tags like negative prompts
-    # or "masterpiece", etc. SD configuration is beyond the scope of this API.
-    # At this point I will not add the edits and variations endpoints (ie. img2img) because they
-    # require changing the form data handling to accept multipart form data, also to properly support
-    # url return types will require file management and a web serving files... Perhaps later!
-    base_model_size = 512 if 'SD_BASE_MODEL_SIZE' not in os.environ else int(os.environ.get('SD_BASE_MODEL_SIZE', 512))
-    sd_defaults = {
-        'sampler_name': 'DPM++ 2M Karras',  # vast improvement
-        'steps': 30,
-    }
+def generations(prompt: str, size: str, response_format: str, n: int,
+                negative_prompt: str = "", steps: int = 9, seed: int = -1,
+                cfg_scale: float = 0.0, batch_count: int = 1):
+    """
+    Generate images using the loaded diffusion model.
+
+    Args:
+        prompt: Text description of the desired image
+        size: Image dimensions as "WIDTHxHEIGHT"
+        response_format: 'url' or 'b64_json'
+        n: Number of images per batch
+        negative_prompt: What to avoid in the image
+        steps: Number of inference steps
+        seed: Random seed (-1 for random)
+        cfg_scale: Classifier-free guidance scale
+        batch_count: Number of sequential batches
+
+    Returns:
+        dict with 'created' timestamp and 'data' list of images
+    """
+    import torch
+    from modules.image_models import get_pipeline_type
+    from modules.torch_utils import clear_torch_cache, get_device
+
+    if shared.image_model is None:
+        raise ServiceUnavailableError("No image model loaded. Load a model via the UI first.")
+
+    clear_torch_cache()
 
-    width, height = [int(x) for x in size.split('x')]  # ignore the restrictions on size
+    # Parse dimensions
+    try:
+        width, height = [int(x) for x in size.split('x')]
+    except (ValueError, IndexError):
+        width, height = 1024, 1024
 
-    # to hack on better generation, edit default payload.
-    payload = {
-        'prompt': prompt,  # ignore prompt limit of 1000 characters
-        'width': width,
-        'height': height,
-        'batch_size': n,
+    # Handle seed
+    if seed == -1:
+        seed = np.random.randint(0, 2**32 - 1)
+
+    device = get_device() or "cpu"
+    generator = torch.Generator(device).manual_seed(int(seed))
+
+    # Get pipeline type for CFG parameter name
+    pipeline_type = getattr(shared, 'image_pipeline_type', None) or get_pipeline_type(shared.image_model)
+
+    # Build generation kwargs
+    gen_kwargs = {
+        "prompt": prompt,
+        "negative_prompt": negative_prompt,
+        "height": height,
+        "width": width,
+        "num_inference_steps": steps,
+        "num_images_per_prompt": n,
+        "generator": generator,
     }
-    payload.update(sd_defaults)
-
-    scale = min(width, height) / base_model_size
-    if scale >= 1.2:
-        # for better performance with the default size (1024), and larger res.
-        scaler = {
-            'width': width // scale,
-            'height': height // scale,
-            'hr_scale': scale,
-            'enable_hr': True,
-            'hr_upscaler': 'Latent',
-            'denoising_strength': 0.68,
-        }
-        payload.update(scaler)
 
+    # Pipeline-specific CFG parameter
+    if pipeline_type == 'qwenimage':
+        gen_kwargs["true_cfg_scale"] = cfg_scale
+    else:
+        gen_kwargs["guidance_scale"] = cfg_scale
+
+    # Generate
+    all_images = []
+    t0 = time.time()
+
+    shared.stop_everything = False
+
+    def interrupt_callback(pipe, step_index, timestep, callback_kwargs):
+        if shared.stop_everything:
+            pipe._interrupt = True
+        return callback_kwargs
+
+    gen_kwargs["callback_on_step_end"] = interrupt_callback
+
+    for i in range(batch_count):
+        if shared.stop_everything:
+            break
+        generator.manual_seed(int(seed + i))
+        batch_results = shared.image_model(**gen_kwargs).images
+        all_images.extend(batch_results)
+
+    t1 = time.time()
+    total_images = len(all_images)
+    total_steps = steps * batch_count
+    logger.info(f'Generated {total_images} {"image" if total_images == 1 else "images"} in {(t1 - t0):.2f} seconds ({total_steps / (t1 - t0):.2f} steps/s, seed {seed})')
+
+    # Save images
+    _save_images(all_images, prompt, negative_prompt, width, height, steps, seed, cfg_scale)
+
+    # Build response
     resp = {
         'created': int(time.time()),
         'data': []
     }
-    from extensions.openai.script import params
-
-    # TODO: support SD_WEBUI_AUTH username:password pair.
-    sd_url = f"{os.environ.get('SD_WEBUI_URL', params.get('sd_webui_url', ''))}/sdapi/v1/txt2img"
-
-    response = requests.post(url=sd_url, json=payload)
-    r = response.json()
-    if response.status_code != 200 or 'images' not in r:
-        print(r)
-        raise ServiceUnavailableError(r.get('error', 'Unknown error calling Stable Diffusion'), code=response.status_code, internal_message=r.get('errors', None))
-    # r['parameters']...
-    for b64_json in r['images']:
+
+    for img in all_images:
+        b64 = _image_to_base64(img)
         if response_format == 'b64_json':
-            resp['data'].extend([{'b64_json': b64_json}])
+            resp['data'].append({'b64_json': b64})
         else:
-            resp['data'].extend([{'url': f'data:image/png;base64,{b64_json}'}])  # yeah it's lazy. requests.get() will not work with this
+            resp['data'].append({'url': f'data:image/png;base64,{b64}'})
 
     return resp
+
+
+def _image_to_base64(image) -> str:
+    buffered = io.BytesIO()
+    image.save(buffered, format="PNG")
+    return base64.b64encode(buffered.getvalue()).decode('utf-8')
+
+
+def _save_images(images, prompt, negative_prompt, width, height, steps, seed, cfg_scale):
+    """Save images with metadata."""
+    date_str = datetime.now().strftime("%Y-%m-%d")
+    folder = os.path.join("user_data", "image_outputs", date_str)
+    os.makedirs(folder, exist_ok=True)
+
+    metadata = {
+        'image_prompt': prompt,
+        'image_neg_prompt': negative_prompt,
+        'image_width': width,
+        'image_height': height,
+        'image_steps': steps,
+        'image_seed': seed,
+        'image_cfg_scale': cfg_scale,
+        'model': getattr(shared, 'image_model_name', 'unknown'),
+    }
+
+    for idx, img in enumerate(images):
+        ts = datetime.now().strftime("%H-%M-%S")
+        filepath = os.path.join(folder, f"{ts}_{seed:010d}_{idx:03d}.png")
+
+        png_info = PngInfo()
+        png_info.add_text("image_gen_settings", json.dumps(metadata))
+        img.save(filepath, pnginfo=png_info)
diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 9440cb1ebc..1e98273183 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -7,26 +7,23 @@
 from collections import deque
 from threading import Thread
 
-import uvicorn
-from fastapi import Depends, FastAPI, Header, HTTPException
-from fastapi.middleware.cors import CORSMiddleware
-from fastapi.requests import Request
-from fastapi.responses import JSONResponse
-from pydub import AudioSegment
-from sse_starlette import EventSourceResponse
-from starlette.concurrency import iterate_in_threadpool
-
 import extensions.openai.completions as OAIcompletions
-import extensions.openai.images as OAIimages
 import extensions.openai.logits as OAIlogits
 import extensions.openai.models as OAImodels
-from extensions.openai.errors import ServiceUnavailableError
+import uvicorn
 from extensions.openai.tokens import token_count, token_decode, token_encode
 from extensions.openai.utils import _start_cloudflared
+from fastapi import Depends, FastAPI, Header, HTTPException
+from fastapi.middleware.cors import CORSMiddleware
+from fastapi.requests import Request
+from fastapi.responses import JSONResponse
 from modules import shared
 from modules.logging_colors import logger
 from modules.models import unload_model
 from modules.text_generation import stop_everything_event
+from pydub import AudioSegment
+from sse_starlette import EventSourceResponse
+from starlette.concurrency import iterate_in_threadpool
 
 from .typing import (
     ChatCompletionRequest,
@@ -40,6 +37,8 @@
     EmbeddingsResponse,
     EncodeRequest,
     EncodeResponse,
+    ImageGenerationRequest,
+    ImageGenerationResponse,
     LoadLorasRequest,
     LoadModelRequest,
     LogitsRequest,
@@ -228,19 +227,24 @@ async def handle_audio_transcription(request: Request):
     return JSONResponse(content=transcription)
 
 
-@app.post('/v1/images/generations', dependencies=check_key)
-async def handle_image_generation(request: Request):
-
-    if not os.environ.get('SD_WEBUI_URL', params.get('sd_webui_url', '')):
-        raise ServiceUnavailableError("Stable Diffusion not available. SD_WEBUI_URL not set.")
-
-    body = await request.json()
-    prompt = body['prompt']
-    size = body.get('size', '1024x1024')
-    response_format = body.get('response_format', 'url')  # or b64_json
-    n = body.get('n', 1)  # ignore the batch limits of max 10
-
-    response = await OAIimages.generations(prompt=prompt, size=size, response_format=response_format, n=n)
+@app.post('/v1/images/generations', response_model=ImageGenerationResponse, dependencies=check_key)
+async def handle_image_generation(request_data: ImageGenerationRequest):
+    import extensions.openai.images as OAIimages
+
+    width, height = request_data.get_width_height()
+
+    response = await asyncio.to_thread(
+        OAIimages.generations,
+        prompt=request_data.prompt,
+        size=f"{width}x{height}",
+        response_format=request_data.response_format,
+        n=request_data.batch_size,  # <-- use resolved batch_size
+        negative_prompt=request_data.negative_prompt,
+        steps=request_data.steps,
+        seed=request_data.seed,
+        cfg_scale=request_data.cfg_scale,
+        batch_count=request_data.batch_count,
+    )
     return JSONResponse(response)
 
 
diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 56d915822a..a24b844be3 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -264,6 +264,46 @@ class LoadLorasRequest(BaseModel):
     lora_names: List[str]
 
 
+class ImageGenerationRequest(BaseModel):
+    """OpenAI-compatible image generation request with extended parameters."""
+    # Required
+    prompt: str
+
+    # Generation parameters
+    negative_prompt: str = ""
+    size: str = Field(default="1024x1024", description="'WIDTHxHEIGHT'")
+    steps: int = Field(default=9, ge=1)
+    cfg_scale: float = Field(default=0.0, ge=0.0)
+    seed: int = Field(default=-1, description="-1 for random")
+    batch_size: int | None = Field(default=None, ge=1, description="Parallel batch size (VRAM heavy)")
+    n: int = Field(default=1, ge=1, description="Alias for batch_size (OpenAI compatibility)")
+    batch_count: int = Field(default=1, ge=1, description="Sequential batch count")
+
+    # OpenAI compatibility (unused)
+    model: str | None = None
+    response_format: str = "b64_json"
+    user: str | None = None
+
+    @model_validator(mode='after')
+    def resolve_batch_size(self):
+        """Use batch_size if provided, otherwise fall back to n."""
+        if self.batch_size is None:
+            self.batch_size = self.n
+        return self
+
+    def get_width_height(self) -> tuple[int, int]:
+        try:
+            parts = self.size.lower().split('x')
+            return int(parts[0]), int(parts[1])
+        except (ValueError, IndexError):
+            return 1024, 1024
+
+
+class ImageGenerationResponse(BaseModel):
+    created: int = int(time.time())
+    data: List[dict]
+
+
 def to_json(obj):
     return json.dumps(obj.__dict__, indent=4)
 

From 5ad174fad2afbbe2c06e841c87e06515a33093ed Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 3 Dec 2025 11:58:54 -0800
Subject: [PATCH 1085/1701] docs: Add an image generation API example

---
 docs/12 - OpenAI API.md | 31 +++++++++++++++++++++++++++++--
 1 file changed, 29 insertions(+), 2 deletions(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index 227541a303..94a95b103e 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -139,6 +139,35 @@ curl http://127.0.0.1:5000/v1/completions \
 
 For base64-encoded images, just replace the inner "url" values with this format: `data:image/FORMAT;base64,BASE64_STRING` where FORMAT is the file type (png, jpeg, gif, etc.) and BASE64_STRING is your base64-encoded image data.
 
+#### Image generation
+
+```shell
+curl http://127.0.0.1:5000/v1/images/generations \
+  -H "Content-Type: application/json" \
+  -d '{
+    "prompt": "an orange tree",
+    "steps": 9,
+    "cfg_scale": 0,
+    "batch_size": 1,
+    "batch_count": 1
+  }'
+```
+
+You need to load an image model first. You can do this via the UI, or by adding `--image-model your_model_name` when launching the server.
+
+The output is a JSON object containing a `data` array. Each element has a `b64_json` field with the base64-encoded PNG image:
+
+```json
+{
+  "created": 1764791227,
+  "data": [
+    {
+      "b64_json": "iVBORw0KGgo..."
+    }
+  ]
+}
+```
+
 #### SSE streaming
 
 ```shell
@@ -419,7 +448,6 @@ The following environment variables can be used (they take precedence over every
 | `OPENEDAI_CERT_PATH`      | SSL certificate file path         |            cert.pem                |
 | `OPENEDAI_KEY_PATH`       | SSL key file path                    |             key.pem               |
 | `OPENEDAI_DEBUG`          | Enable debugging (set to 1)    | 1                          |
-| `SD_WEBUI_URL`           | WebUI URL (used by endpoint) | http://127.0.0.1:7861 |
 | `OPENEDAI_EMBEDDING_MODEL` | Embedding model (if applicable) |          sentence-transformers/all-mpnet-base-v2                  |
 | `OPENEDAI_EMBEDDING_DEVICE` | Embedding device (if applicable) |           cuda                 |
 
@@ -430,7 +458,6 @@ You can also set the following variables in your `settings.yaml` file:
 ```
 openai-embedding_device: cuda
 openai-embedding_model: "sentence-transformers/all-mpnet-base-v2"
-openai-sd_webui_url: http://127.0.0.1:7861
 openai-debug: 1
 ```
 

From 4468c49439685dc8bc68e9d7a6109694a2eab72b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 3 Dec 2025 12:02:47 -0800
Subject: [PATCH 1086/1701] Add semaphore to image generation API endpoint

---
 extensions/openai/script.py | 33 +++++++++++++++++----------------
 1 file changed, 17 insertions(+), 16 deletions(-)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 1e98273183..65805629da 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -53,12 +53,12 @@
 params = {
     'embedding_device': 'cpu',
     'embedding_model': 'sentence-transformers/all-mpnet-base-v2',
-    'sd_webui_url': '',
     'debug': 0
 }
 
 
 streaming_semaphore = asyncio.Semaphore(1)
+image_generation_semaphore = asyncio.Semaphore(1)
 
 
 def verify_api_key(authorization: str = Header(None)) -> None:
@@ -231,21 +231,22 @@ async def handle_audio_transcription(request: Request):
 async def handle_image_generation(request_data: ImageGenerationRequest):
     import extensions.openai.images as OAIimages
 
-    width, height = request_data.get_width_height()
-
-    response = await asyncio.to_thread(
-        OAIimages.generations,
-        prompt=request_data.prompt,
-        size=f"{width}x{height}",
-        response_format=request_data.response_format,
-        n=request_data.batch_size,  # <-- use resolved batch_size
-        negative_prompt=request_data.negative_prompt,
-        steps=request_data.steps,
-        seed=request_data.seed,
-        cfg_scale=request_data.cfg_scale,
-        batch_count=request_data.batch_count,
-    )
-    return JSONResponse(response)
+    async with image_generation_semaphore:
+        width, height = request_data.get_width_height()
+
+        response = await asyncio.to_thread(
+            OAIimages.generations,
+            prompt=request_data.prompt,
+            size=f"{width}x{height}",
+            response_format=request_data.response_format,
+            n=request_data.batch_size,  # <-- use resolved batch_size
+            negative_prompt=request_data.negative_prompt,
+            steps=request_data.steps,
+            seed=request_data.seed,
+            cfg_scale=request_data.cfg_scale,
+            batch_count=request_data.batch_count,
+        )
+        return JSONResponse(response)
 
 
 @app.post("/v1/embeddings", response_model=EmbeddingsResponse, dependencies=check_key)

From 906dc549697478be5b8816e12b7dc94fd34133b0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 3 Dec 2025 12:15:38 -0800
Subject: [PATCH 1087/1701] Load `--image-model` before `--model`

---
 server.py | 32 ++++++++++++++++----------------
 1 file changed, 16 insertions(+), 16 deletions(-)

diff --git a/server.py b/server.py
index b02c50a25e..0c5d14ce91 100644
--- a/server.py
+++ b/server.py
@@ -275,6 +275,22 @@ def create_interface():
         if extension not in shared.args.extensions:
             shared.args.extensions.append(extension)
 
+    # Load image model if specified via CLI
+    if shared.args.image_model:
+        logger.info(f"Loading image model: {shared.args.image_model}")
+        result = load_image_model(
+            shared.args.image_model,
+            dtype=shared.settings.get('image_dtype', 'bfloat16'),
+            attn_backend=shared.settings.get('image_attn_backend', 'sdpa'),
+            cpu_offload=shared.settings.get('image_cpu_offload', False),
+            compile_model=shared.settings.get('image_compile', False),
+            quant_method=shared.settings.get('image_quant', 'none')
+        )
+        if result is not None:
+            shared.image_model_name = shared.args.image_model
+        else:
+            logger.error(f"Failed to load image model: {shared.args.image_model}")
+
     available_models = utils.get_available_models()
 
     # Model defined through --model
@@ -321,22 +337,6 @@ def create_interface():
         if shared.args.lora:
             add_lora_to_model(shared.args.lora)
 
-    # Load image model if specified via CLI
-    if shared.args.image_model:
-        logger.info(f"Loading image model: {shared.args.image_model}")
-        result = load_image_model(
-            shared.args.image_model,
-            dtype=shared.settings.get('image_dtype', 'bfloat16'),
-            attn_backend=shared.settings.get('image_attn_backend', 'sdpa'),
-            cpu_offload=shared.settings.get('image_cpu_offload', False),
-            compile_model=shared.settings.get('image_compile', False),
-            quant_method=shared.settings.get('image_quant', 'none')
-        )
-        if result is not None:
-            shared.image_model_name = shared.args.image_model
-        else:
-            logger.error(f"Failed to load image model: {shared.args.image_model}")
-
     shared.generation_lock = Lock()
 
     if shared.args.idle_timeout > 0:

From 373baa5c9cb46b629b5f7362986ac3ceda97bc08 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 3 Dec 2025 14:43:56 -0800
Subject: [PATCH 1088/1701] UI: Minor image gallery improvements

---
 modules/ui_image_generation.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 1b0e58f361..2ea34551f8 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -29,7 +29,7 @@
 }
 
 STEP = 16
-IMAGES_PER_PAGE = 64
+IMAGES_PER_PAGE = 32
 
 # Settings keys to save in PNG metadata (Generate tab only)
 METADATA_SETTINGS_KEYS = [
@@ -172,7 +172,7 @@ def format_metadata_for_display(metadata):
     if not metadata:
         return "No generation settings found in this image."
 
-    lines = ["**Generation Settings**", ""]
+    lines = []
 
     # Display in a nice order
     display_order = [
@@ -441,7 +441,7 @@ def create_ui():
                         )
 
                     with gr.Column(scale=1):
-                        gr.Markdown("### Selected Image")
+                        gr.Markdown("### Generation Settings")
                         shared.gradio['image_settings_display'] = gr.Markdown("Select an image to view its settings")
                         shared.gradio['image_send_to_generate'] = gr.Button("Send to Generate", variant="primary")
                         shared.gradio['image_gallery_status'] = gr.Markdown("")

From 59285d501d505e0e1939861baea161870c4ed4c6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 3 Dec 2025 16:03:31 -0800
Subject: [PATCH 1089/1701] Image generation: Small UI improvements

---
 js/switch_tabs.js              | 14 ++++++++++++++
 modules/ui_image_generation.py |  5 +++--
 2 files changed, 17 insertions(+), 2 deletions(-)

diff --git a/js/switch_tabs.js b/js/switch_tabs.js
index 7fb78aea85..36e5736bad 100644
--- a/js/switch_tabs.js
+++ b/js/switch_tabs.js
@@ -36,3 +36,17 @@ function switch_to_character() {
   document.getElementById("character-tab-button").click();
   scrollToTop();
 }
+
+function switch_to_image_ai_generate() {
+  const container = document.querySelector("#image-ai-tab");
+  const buttons = container.getElementsByTagName("button");
+
+  for (let i = 0; i < buttons.length; i++) {
+    if (buttons[i].textContent.trim() === "Generate") {
+      buttons[i].click();
+      break;
+    }
+  }
+
+  scrollToTop();
+}
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 2ea34551f8..3a64bc8b70 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -418,9 +418,9 @@ def create_ui():
                         # Pagination controls
                         with gr.Row():
                             shared.gradio['image_refresh_history'] = gr.Button("🔄 Refresh", elem_classes="refresh-button")
-                            shared.gradio['image_prev_page'] = gr.Button("◀ Prev", elem_classes="refresh-button")
+                            shared.gradio['image_prev_page'] = gr.Button("◀ Prev Page", elem_classes="refresh-button")
                             shared.gradio['image_page_info'] = gr.Markdown(value=get_initial_page_info, elem_id="image-page-info")
-                            shared.gradio['image_next_page'] = gr.Button("Next ▶", elem_classes="refresh-button")
+                            shared.gradio['image_next_page'] = gr.Button("Next Page ▶", elem_classes="refresh-button")
                             shared.gradio['image_page_input'] = gr.Number(value=1, label="Page", precision=0, minimum=1, scale=0, min_width=80)
                             shared.gradio['image_go_to_page'] = gr.Button("Go", elem_classes="refresh-button", scale=0, min_width=50)
 
@@ -649,6 +649,7 @@ def create_event_handlers():
             'image_cfg_scale',
             'image_gallery_status'
         ),
+        js=f'() => {{{ui.switch_tabs_js}; switch_to_image_ai_generate()}}',
         show_progress=False
     )
 

From 49c60882bf8f4806d0f47e890d7f434007941796 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 3 Dec 2025 16:07:51 -0800
Subject: [PATCH 1090/1701] Image generation: Safer image uploading

---
 modules/chat.py                | 16 +---------------
 modules/image_utils.py         | 20 ++++++++++++++++----
 modules/ui_image_generation.py |  4 ++--
 3 files changed, 19 insertions(+), 21 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 55984d7a40..acfc2f6663 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -3,7 +3,6 @@
 import functools
 import html
 import json
-import os
 import pprint
 import re
 import shutil
@@ -26,6 +25,7 @@
     convert_to_markdown,
     make_thumbnail
 )
+from modules.image_utils import open_image_safely
 from modules.logging_colors import logger
 from modules.text_generation import (
     generate_reply,
@@ -1516,20 +1516,6 @@ def load_instruction_template_memoized(template):
     return load_instruction_template(template)
 
 
-def open_image_safely(path):
-    if path is None or not isinstance(path, str) or not Path(path).exists():
-        return None
-
-    if os.path.islink(path):
-        return None
-
-    try:
-        return Image.open(path)
-    except Exception as e:
-        logger.error(f"Failed to open image file: {path}. Reason: {e}")
-        return None
-
-
 def upload_character(file, img_path, tavern=False):
     img = open_image_safely(img_path)
     decoded_file = file if isinstance(file, str) else file.decode('utf-8')
diff --git a/modules/image_utils.py b/modules/image_utils.py
index 658f00d74d..d2809fef36 100644
--- a/modules/image_utils.py
+++ b/modules/image_utils.py
@@ -1,9 +1,7 @@
-"""
-Shared image processing utilities for multimodal support.
-Used by both ExLlamaV3 and llama.cpp implementations.
-"""
 import base64
 import io
+import os
+from pathlib import Path
 from typing import Any, List, Tuple
 
 from PIL import Image
@@ -11,6 +9,20 @@
 from modules.logging_colors import logger
 
 
+def open_image_safely(path):
+    if path is None or not isinstance(path, str) or not Path(path).exists():
+        return None
+
+    if os.path.islink(path):
+        return None
+
+    try:
+        return Image.open(path)
+    except Exception as e:
+        logger.error(f"Failed to open image file: {path}. Reason: {e}")
+        return None
+
+
 def convert_pil_to_base64(image: Image.Image) -> str:
     """Converts a PIL Image to a base64 encoded string."""
     buffered = io.BytesIO()
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 3a64bc8b70..06ef3e826a 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -7,7 +7,6 @@
 
 import gradio as gr
 import numpy as np
-from PIL import Image
 from PIL.PngImagePlugin import PngInfo
 
 from modules import shared, ui, utils
@@ -16,6 +15,7 @@
     load_image_model,
     unload_image_model
 )
+from modules.image_utils import open_image_safely
 from modules.logging_colors import logger
 from modules.text_generation import stop_everything_event
 from modules.utils import gradio
@@ -159,7 +159,7 @@ def save_generated_images(images, state, actual_seed):
 def read_image_metadata(image_path):
     """Read generation metadata from PNG file."""
     try:
-        with Image.open(image_path) as img:
+        with open_image_safely(image_path) as img:
             if hasattr(img, 'text') and 'image_gen_settings' in img.text:
                 return json.loads(img.text['image_gen_settings'])
     except Exception as e:

From fbca54957eec4ca49c73eac56159fa92603724b7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 3 Dec 2025 16:13:07 -0800
Subject: [PATCH 1091/1701] Image generation: Yield partial results for batch
 count > 1

---
 modules/ui_image_generation.py | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 06ef3e826a..08cf3f64c5 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -677,7 +677,8 @@ def generate(state):
 
         if not model_name or model_name == 'None':
             logger.error("No image model selected. Go to the Model tab and select a model.")
-            return []
+            yield []
+            return
 
         if shared.image_model is None:
             result = load_image_model(
@@ -690,7 +691,8 @@ def generate(state):
             )
             if result is None:
                 logger.error(f"Failed to load model `{model_name}`.")
-                return []
+                yield []
+                return
 
             shared.image_model_name = model_name
 
@@ -760,6 +762,7 @@ def interrupt_callback(pipe, step_index, timestep, callback_kwargs):
             generator.manual_seed(int(seed + i))
             batch_results = shared.image_model(**gen_kwargs).images
             all_images.extend(batch_results)
+            yield all_images
 
         t1 = time.time()
         save_generated_images(all_images, state, seed)
@@ -768,12 +771,12 @@ def interrupt_callback(pipe, step_index, timestep, callback_kwargs):
         total_steps = state["image_steps"] * int(state['image_batch_count'])
         logger.info(f'Generated {total_images} {"image" if total_images == 1 else "images"} in {(t1 - t0):.2f} seconds ({total_steps / (t1 - t0):.2f} steps/s, seed {seed})')
 
-        return all_images
+        yield all_images
 
     except Exception as e:
         logger.error(f"Image generation failed: {e}")
         traceback.print_exc()
-        return []
+        yield []
 
 
 def load_image_model_wrapper(model_name, dtype, attn_backend, cpu_offload, compile_model, quant_method):

From c93d27add3b15c0494d436f1a3361c0b4862c626 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 3 Dec 2025 18:29:43 -0800
Subject: [PATCH 1092/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 4 ++--
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 4 ++--
 requirements/portable/requirements_amd_noavx2.txt      | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 4 ++--
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 36 insertions(+), 36 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 4e02d76f0b..c1f87988e1 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -44,8 +44,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index c0fbd9ab85..512231e037 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -42,7 +42,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index b330646a90..1ccc507c6f 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -42,7 +42,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index c2c643375f..4d8aa7714d 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -42,5 +42,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 10889215a5..5290aa714d 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -42,5 +42,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index ffcd64736b..d9c76a3144 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -42,5 +42,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index daa2444ffe..1a796c2199 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -42,5 +42,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index a43a77244a..8105abaad6 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -44,8 +44,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 5f9b69e894..0b944d48b6 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 10a6b5af8b..9b43c901e9 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index 4686b9de71..825e6dec50 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 890568f4d4..864b67756d 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index e480db8f41..03e090cf02 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index c2a7d0403a..7dd851e876 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index e243d4bea4..63949d9f94 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 3db163b01b..34652264a0 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 7758131cf5..28849ff7fa 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index dcee8d05e7..365a798e5b 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.62.0/llama_cpp_binaries-0.62.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From c357eed4c73099343418f5340dc1c28013cb7486 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 3 Dec 2025 18:40:34 -0800
Subject: [PATCH 1093/1701] Image: Remove the flash_attention_3 option (no idea
 how to get it working)

---
 modules/image_models.py        | 4 +---
 modules/shared.py              | 2 +-
 modules/ui_image_generation.py | 2 +-
 3 files changed, 3 insertions(+), 5 deletions(-)

diff --git a/modules/image_models.py b/modules/image_models.py
index 2ef1e7309b..8363533b28 100644
--- a/modules/image_models.py
+++ b/modules/image_models.py
@@ -98,7 +98,7 @@ def load_image_model(model_name, dtype='bfloat16', attn_backend='sdpa', cpu_offl
     Args:
         model_name: Name of the model directory
         dtype: 'bfloat16' or 'float16'
-        attn_backend: 'sdpa', 'flash_attention_2', or 'flash_attention_3'
+        attn_backend: 'sdpa' or 'flash_attention_2'
         cpu_offload: Enable CPU offloading for low VRAM
         compile_model: Compile the model for faster inference (slow first run)
         quant_method: 'none', 'bnb-8bit', 'bnb-4bit', or torchao options (int8wo, fp4, float8wo)
@@ -145,8 +145,6 @@ def load_image_model(model_name, dtype='bfloat16', attn_backend='sdpa', cpu_offl
         if hasattr(pipe, 'transformer') and hasattr(pipe.transformer, 'set_attention_backend'):
             if attn_backend == 'flash_attention_2':
                 pipe.transformer.set_attention_backend("flash")
-            elif attn_backend == 'flash_attention_3':
-                pipe.transformer.set_attention_backend("_flash_3")
             # sdpa is the default, no action needed
 
         if compile_model:
diff --git a/modules/shared.py b/modules/shared.py
index 4a0fd986a2..4e17497b46 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -56,7 +56,7 @@
 group.add_argument('--image-model', type=str, help='Name of the image model to select on startup (overrides saved setting).')
 group.add_argument('--image-model-dir', type=str, default='user_data/image_models', help='Path to directory with all the image models.')
 group.add_argument('--image-dtype', type=str, default=None, choices=['bfloat16', 'float16'], help='Data type for image model.')
-group.add_argument('--image-attn-backend', type=str, default=None, choices=['sdpa', 'flash_attention_2', 'flash_attention_3'], help='Attention backend for image model.')
+group.add_argument('--image-attn-backend', type=str, default=None, choices=['sdpa', 'flash_attention_2'], help='Attention backend for image model.')
 group.add_argument('--image-cpu-offload', action='store_true', help='Enable CPU offloading for image model.')
 group.add_argument('--image-compile', action='store_true', help='Compile the image model for faster inference.')
 group.add_argument('--image-quant', type=str, default=None,
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 08cf3f64c5..fdf1af864e 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -485,7 +485,7 @@ def create_ui():
                                     info='bfloat16 recommended for modern GPUs'
                                 )
                                 shared.gradio['image_attn_backend'] = gr.Dropdown(
-                                    choices=['sdpa', 'flash_attention_2', 'flash_attention_3'],
+                                    choices=['sdpa', 'flash_attention_2'],
                                     value=shared.settings['image_attn_backend'],
                                     label='Attention Backend',
                                     info='SDPA is default. Flash Attention requires compatible GPU.'

From 235b94f097d7efa37f3f33de6b1787ce5673b742 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 3 Dec 2025 18:43:30 -0800
Subject: [PATCH 1094/1701] Image: Add placeholder file for
 user_data/image_models

---
 user_data/image_models/place-your-models-here.txt | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 create mode 100644 user_data/image_models/place-your-models-here.txt

diff --git a/user_data/image_models/place-your-models-here.txt b/user_data/image_models/place-your-models-here.txt
new file mode 100644
index 0000000000..e69de29bb2

From 14dbc3488e5780be50a7b6211742ec2137cdc503 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 05:32:58 -0800
Subject: [PATCH 1095/1701] Image: Clear the torch cache after generation, not
 before

---
 modules/ui_image_generation.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index fdf1af864e..7c393f137d 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -670,8 +670,6 @@ def generate(state):
 
     from modules.torch_utils import clear_torch_cache, get_device
 
-    clear_torch_cache()
-
     try:
         model_name = state['image_model_menu']
 
@@ -772,11 +770,13 @@ def interrupt_callback(pipe, step_index, timestep, callback_kwargs):
         logger.info(f'Generated {total_images} {"image" if total_images == 1 else "images"} in {(t1 - t0):.2f} seconds ({total_steps / (t1 - t0):.2f} steps/s, seed {seed})')
 
         yield all_images
+        clear_torch_cache()
 
     except Exception as e:
         logger.error(f"Image generation failed: {e}")
         traceback.print_exc()
         yield []
+        clear_torch_cache()
 
 
 def load_image_model_wrapper(model_name, dtype, attn_backend, cpu_offload, compile_model, quant_method):

From a838223d185597071ba4d346190f24708fa51062 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 05:49:57 -0800
Subject: [PATCH 1096/1701] Image: Add a progress bar during generation

---
 modules/ui_image_generation.py | 98 +++++++++++++++++++++++++---------
 1 file changed, 72 insertions(+), 26 deletions(-)

diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 7c393f137d..62760a8a35 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -373,7 +373,10 @@ def create_ui():
 
                         shared.gradio['image_generate_btn'] = gr.Button("Generate", variant="primary", size="lg")
                         shared.gradio['image_stop_btn'] = gr.Button("Stop", size="lg", visible=False)
-                        gr.HTML("
            ")
+                        shared.gradio['image_progress'] = gr.HTML(
+                            value=progress_bar_html(),
+                            elem_id="image-progress"
+                        )
 
                         gr.Markdown("### Dimensions")
                         with gr.Row():
@@ -546,19 +549,19 @@ def create_event_handlers():
     shared.gradio['image_generate_btn'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('image_stop_btn', 'image_generate_btn')).then(
-        generate, gradio('interface_state'), gradio('image_output_gallery'), show_progress=False).then(
+        generate, gradio('interface_state'), gradio('image_output_gallery', 'image_progress'), show_progress=False).then(
         lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('image_stop_btn', 'image_generate_btn'))
 
     shared.gradio['image_prompt'].submit(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('image_stop_btn', 'image_generate_btn')).then(
-        generate, gradio('interface_state'), gradio('image_output_gallery'), show_progress=False).then(
+        generate, gradio('interface_state'), gradio('image_output_gallery', 'image_progress'), show_progress=False).then(
         lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('image_stop_btn', 'image_generate_btn'))
 
     shared.gradio['image_neg_prompt'].submit(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         lambda: [gr.update(visible=True), gr.update(visible=False)], None, gradio('image_stop_btn', 'image_generate_btn')).then(
-        generate, gradio('interface_state'), gradio('image_output_gallery'), show_progress=False).then(
+        generate, gradio('interface_state'), gradio('image_output_gallery', 'image_progress'), show_progress=False).then(
         lambda: [gr.update(visible=False), gr.update(visible=True)], None, gradio('image_stop_btn', 'image_generate_btn'))
 
     # Stop button
@@ -661,11 +664,27 @@ def create_event_handlers():
     )
 
 
+def progress_bar_html(progress=0, text=""):
+    """Generate HTML for progress bar. Empty div when progress <= 0."""
+    if progress <= 0:
+        return '
            '
+
+    return f'''
            
+        
            
+            
            
+        
            
+        
            {text}
            
+    
            '''
+
+
 def generate(state):
     """
     Generate images using the loaded model.
     Automatically adjusts parameters based on pipeline type.
     """
+    import queue
+    import threading
+
     import torch
 
     from modules.torch_utils import clear_torch_cache, get_device
@@ -675,7 +694,7 @@ def generate(state):
 
         if not model_name or model_name == 'None':
             logger.error("No image model selected. Go to the Model tab and select a model.")
-            yield []
+            yield [], progress_bar_html()
             return
 
         if shared.image_model is None:
@@ -689,7 +708,7 @@ def generate(state):
             )
             if result is None:
                 logger.error(f"Failed to load model `{model_name}`.")
-                yield []
+                yield [], progress_bar_html()
                 return
 
             shared.image_model_name = model_name
@@ -713,69 +732,96 @@ def generate(state):
         # Process Prompt
         prompt = state['image_prompt']
 
-        # Apply "Positive Magic" for Qwen models only
         if pipeline_type == 'qwenimage':
             magic_suffix = ", Ultra HD, 4K, cinematic composition"
-            # Avoid duplication if user already added it
             if magic_suffix.strip(", ") not in prompt:
                 prompt += magic_suffix
 
-        # Reset stop flag at start
         shared.stop_everything = False
 
-        # Callback to check for interruption during diffusion steps
+        batch_count = int(state['image_batch_count'])
+        steps_per_batch = int(state['image_steps'])
+        total_steps = steps_per_batch * batch_count
+
+        # Queue for progress updates from callback
+        progress_queue = queue.Queue()
+
         def interrupt_callback(pipe, step_index, timestep, callback_kwargs):
             if shared.stop_everything:
                 pipe._interrupt = True
-
+            progress_queue.put(step_index + 1)
             return callback_kwargs
 
-        # Build generation kwargs
         gen_kwargs = {
             "prompt": prompt,
             "negative_prompt": state['image_neg_prompt'],
             "height": int(state['image_height']),
             "width": int(state['image_width']),
-            "num_inference_steps": int(state['image_steps']),
+            "num_inference_steps": steps_per_batch,
             "num_images_per_prompt": int(state['image_batch_size']),
             "generator": generator,
             "callback_on_step_end": interrupt_callback,
         }
 
-        # Add pipeline-specific parameters for CFG
         cfg_val = state.get('image_cfg_scale', 0.0)
-
         if pipeline_type == 'qwenimage':
-            # Qwen-Image uses true_cfg_scale (typically 4.0)
             gen_kwargs["true_cfg_scale"] = cfg_val
         else:
-            # Z-Image and others use guidance_scale (typically 0.0 for Turbo)
             gen_kwargs["guidance_scale"] = cfg_val
 
         t0 = time.time()
-        for i in range(int(state['image_batch_count'])):
+
+        for batch_idx in range(batch_count):
             if shared.stop_everything:
                 break
 
-            generator.manual_seed(int(seed + i))
-            batch_results = shared.image_model(**gen_kwargs).images
-            all_images.extend(batch_results)
-            yield all_images
+            generator.manual_seed(int(seed + batch_idx))
+
+            # Run generation in thread so we can yield progress
+            result_holder = []
+            error_holder = []
+
+            def run_batch():
+                try:
+                    result_holder.extend(shared.image_model(**gen_kwargs).images)
+                except Exception as e:
+                    error_holder.append(e)
+
+            thread = threading.Thread(target=run_batch)
+            thread.start()
+
+            # Yield progress updates while generation runs
+            while thread.is_alive():
+                try:
+                    step = progress_queue.get(timeout=0.1)
+                    absolute_step = batch_idx * steps_per_batch + step
+                    pct = absolute_step / total_steps
+                    text = f"Batch {batch_idx + 1}/{batch_count} — Step {step}/{steps_per_batch}"
+                    yield all_images, progress_bar_html(pct, text)
+                except queue.Empty:
+                    pass
+
+            thread.join()
+
+            if error_holder:
+                raise error_holder[0]
+
+            all_images.extend(result_holder)
+            yield all_images, progress_bar_html((batch_idx + 1) / batch_count, f"Batch {batch_idx + 1}/{batch_count} complete")
 
         t1 = time.time()
         save_generated_images(all_images, state, seed)
 
-        total_images = int(state['image_batch_count']) * int(state['image_batch_size'])
-        total_steps = state["image_steps"] * int(state['image_batch_count'])
+        total_images = batch_count * int(state['image_batch_size'])
         logger.info(f'Generated {total_images} {"image" if total_images == 1 else "images"} in {(t1 - t0):.2f} seconds ({total_steps / (t1 - t0):.2f} steps/s, seed {seed})')
 
-        yield all_images
+        yield all_images, progress_bar_html()
         clear_torch_cache()
 
     except Exception as e:
         logger.error(f"Image generation failed: {e}")
         traceback.print_exc()
-        yield []
+        yield [], progress_bar_html()
         clear_torch_cache()
 
 

From 7fb9f19bd8866ceb47765cfc428e6a21d105d498 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 06:17:40 -0800
Subject: [PATCH 1097/1701] Progress bar style improvements

---
 css/main.css                   | 39 ++++++++++++++++++++++++++++++++++
 modules/ui_image_generation.py | 10 ++++-----
 2 files changed, 44 insertions(+), 5 deletions(-)

diff --git a/css/main.css b/css/main.css
index 5c1c356dbe..4dea6b9e71 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1752,3 +1752,42 @@ button#swap-height-width {
 .min.svelte-1yrv54 {
     min-height: 0;
 }
+
+/* Image Generation Progress Bar */
+#image-progress .image-ai-separator {
+    height: 24px;
+    margin: 20px 0;
+    border-top: 1px solid var(--input-border-color);
+}
+
+#image-progress .image-ai-progress-wrapper {
+    height: 24px;
+    margin: 20px 0;
+}
+
+#image-progress .image-ai-progress-track {
+    background: #e5e7eb;
+    border-radius: 4px;
+    overflow: hidden;
+    height: 8px;
+}
+
+.dark #image-progress .image-ai-progress-track {
+    background: #333;
+}
+
+#image-progress .image-ai-progress-fill {
+    background: #4a9eff;
+    height: 100%;
+}
+
+#image-progress .image-ai-progress-text {
+    text-align: center;
+    font-size: 12px;
+    color: #666;
+    margin-top: 4px;
+}
+
+.dark #image-progress .image-ai-progress-text {
+    color: #888;
+}
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 62760a8a35..e85f15207e 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -667,13 +667,13 @@ def create_event_handlers():
 def progress_bar_html(progress=0, text=""):
     """Generate HTML for progress bar. Empty div when progress <= 0."""
     if progress <= 0:
-        return '
            '
+        return '
            '
 
-    return f'''
            
-        
            
-            
            
+    return f'''
            
+        
            
+            
            
         
            
-        
            {text}
            
+        
            {text}
            
     
            '''
 
 

From 27931537176fef1bc1335815097ebc780cbf1dbf Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 07:57:23 -0800
Subject: [PATCH 1098/1701] Image: Add LLM-generated prompt variations

---
 modules/shared.py              |  1 +
 modules/ui.py                  |  2 ++
 modules/ui_image_generation.py | 58 ++++++++++++++++++++++++++++++++++
 3 files changed, 61 insertions(+)

diff --git a/modules/shared.py b/modules/shared.py
index 4e17497b46..1ecc0d28c4 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -319,6 +319,7 @@
     'image_seed': -1,
     'image_batch_size': 1,
     'image_batch_count': 1,
+    'image_llm_variations': False,
     'image_model_menu': 'None',
     'image_dtype': 'bfloat16',
     'image_attn_backend': 'sdpa',
diff --git a/modules/ui.py b/modules/ui.py
index ff5686e83d..d95f7bc60a 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -293,6 +293,7 @@ def list_interface_input_elements():
             'image_seed',
             'image_batch_size',
             'image_batch_count',
+            'image_llm_variations',
             'image_model_menu',
             'image_dtype',
             'image_attn_backend',
@@ -547,6 +548,7 @@ def setup_auto_save():
             'image_seed',
             'image_batch_size',
             'image_batch_count',
+            'image_llm_variations',
             'image_model_menu',
             'image_dtype',
             'image_attn_backend',
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index e85f15207e..ceb470ffb0 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -10,6 +10,7 @@
 from PIL.PngImagePlugin import PngInfo
 
 from modules import shared, ui, utils
+from modules.utils import check_model_loaded
 from modules.image_models import (
     get_pipeline_type,
     load_image_model,
@@ -409,6 +410,11 @@ def create_ui():
                             with gr.Column():
                                 shared.gradio['image_batch_size'] = gr.Slider(1, 32, value=shared.settings['image_batch_size'], step=1, label="Batch Size (VRAM Heavy)", info="Generates N images at once.")
                                 shared.gradio['image_batch_count'] = gr.Slider(1, 128, value=shared.settings['image_batch_count'], step=1, label="Sequential Count (Loop)", info="Repeats the generation N times.")
+                                shared.gradio['image_llm_variations'] = gr.Checkbox(
+                                    value=shared.settings['image_llm_variations'],
+                                    label='LLM Prompt Variations',
+                                    info='Use the loaded LLM to generate creative prompt variations for each sequential batch.'
+                                )
 
                     with gr.Column(scale=6, min_width=500):
                         with gr.Column(elem_classes=["viewport-container"]):
@@ -664,6 +670,54 @@ def create_event_handlers():
     )
 
 
+def generate_prompt_variation(state):
+    """Generate a creative variation of the image prompt using the LLM."""
+    from modules.chat import generate_chat_prompt
+    from modules.text_generation import generate_reply
+
+    prompt = state['image_prompt']
+
+    # Check if LLM is loaded
+    model_loaded, _ = check_model_loaded()
+    if not model_loaded:
+        logger.warning("No LLM loaded for prompt variation. Using original prompt.")
+        return prompt
+
+    augmented_message = f"{prompt}\n\n=====\n\nPlease create a creative variation of the image generation prompt above. Keep the same general subject and style, but vary the details, composition, lighting, or mood. Respond with only the new prompt, nothing else."
+
+    # Use minimal state for generation
+    var_state = state.copy()
+    var_state['history'] = {'internal': [], 'visible': [], 'metadata': {}}
+    var_state['auto_max_new_tokens'] = True
+    var_state['enable_thinking'] = False
+    var_state['reasoning_effort'] = 'low'
+    var_state['start_with'] = ""
+
+    formatted_prompt = generate_chat_prompt(augmented_message, var_state)
+
+    variation = ""
+    for reply in generate_reply(formatted_prompt, var_state, stopping_strings=[], is_chat=True):
+        variation = reply
+
+    # Strip thinking blocks if present
+    if "" in variation:
+        variation = variation.rsplit("", 1)[1]
+    elif "<|start|>assistant<|channel|>final<|message|>" in variation:
+        variation = variation.rsplit("<|start|>assistant<|channel|>final<|message|>", 1)[1]
+    elif "" in variation:
+        variation = variation.rsplit("", 1)[1]
+
+    variation = variation.strip()
+    if len(variation) >= 2 and variation.startswith('"') and variation.endswith('"'):
+        variation = variation[1:-1]
+
+    if variation:
+        logger.info(f"Prompt variation: {variation}...")
+        return variation
+
+    return prompt
+
+
 def progress_bar_html(progress=0, text=""):
     """Generate HTML for progress bar. Empty div when progress <= 0."""
     if progress <= 0:
@@ -777,6 +831,10 @@ def interrupt_callback(pipe, step_index, timestep, callback_kwargs):
 
             generator.manual_seed(int(seed + batch_idx))
 
+            # Generate prompt variation if enabled
+            if state['image_llm_variations']:
+                gen_kwargs["prompt"] = generate_prompt_variation(state)
+
             # Run generation in thread so we can yield progress
             result_holder = []
             error_holder = []

From 5763947c375eef6cb1df55e6788538ab30ac4e99 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 10:23:00 -0800
Subject: [PATCH 1099/1701] Image: Simplify the API code, add the
 llm_variations option

---
 extensions/openai/images.py    | 147 ++++++---------------------------
 extensions/openai/script.py    |  32 +++----
 extensions/openai/typing.py    |  15 ++--
 modules/ui_image_generation.py |  10 +--
 4 files changed, 50 insertions(+), 154 deletions(-)

diff --git a/extensions/openai/images.py b/extensions/openai/images.py
index 3a5288e626..1ecb1e6399 100644
--- a/extensions/openai/images.py
+++ b/extensions/openai/images.py
@@ -4,117 +4,50 @@
 
 import base64
 import io
-import json
-import os
 import time
-from datetime import datetime
 
-import numpy as np
 from extensions.openai.errors import ServiceUnavailableError
 from modules import shared
-from modules.logging_colors import logger
-from PIL.PngImagePlugin import PngInfo
 
 
-def generations(prompt: str, size: str, response_format: str, n: int,
-                negative_prompt: str = "", steps: int = 9, seed: int = -1,
-                cfg_scale: float = 0.0, batch_count: int = 1):
+def generations(request):
     """
     Generate images using the loaded diffusion model.
-
-    Args:
-        prompt: Text description of the desired image
-        size: Image dimensions as "WIDTHxHEIGHT"
-        response_format: 'url' or 'b64_json'
-        n: Number of images per batch
-        negative_prompt: What to avoid in the image
-        steps: Number of inference steps
-        seed: Random seed (-1 for random)
-        cfg_scale: Classifier-free guidance scale
-        batch_count: Number of sequential batches
-
-    Returns:
-        dict with 'created' timestamp and 'data' list of images
+    Returns dict with 'created' timestamp and 'data' list of images.
     """
-    import torch
-    from modules.image_models import get_pipeline_type
-    from modules.torch_utils import clear_torch_cache, get_device
+    from modules.ui_image_generation import generate
 
     if shared.image_model is None:
         raise ServiceUnavailableError("No image model loaded. Load a model via the UI first.")
 
-    clear_torch_cache()
-
-    # Parse dimensions
-    try:
-        width, height = [int(x) for x in size.split('x')]
-    except (ValueError, IndexError):
-        width, height = 1024, 1024
-
-    # Handle seed
-    if seed == -1:
-        seed = np.random.randint(0, 2**32 - 1)
-
-    device = get_device() or "cpu"
-    generator = torch.Generator(device).manual_seed(int(seed))
-
-    # Get pipeline type for CFG parameter name
-    pipeline_type = getattr(shared, 'image_pipeline_type', None) or get_pipeline_type(shared.image_model)
-
-    # Build generation kwargs
-    gen_kwargs = {
-        "prompt": prompt,
-        "negative_prompt": negative_prompt,
-        "height": height,
-        "width": width,
-        "num_inference_steps": steps,
-        "num_images_per_prompt": n,
-        "generator": generator,
-    }
-
-    # Pipeline-specific CFG parameter
-    if pipeline_type == 'qwenimage':
-        gen_kwargs["true_cfg_scale"] = cfg_scale
-    else:
-        gen_kwargs["guidance_scale"] = cfg_scale
-
-    # Generate
-    all_images = []
-    t0 = time.time()
-
-    shared.stop_everything = False
-
-    def interrupt_callback(pipe, step_index, timestep, callback_kwargs):
-        if shared.stop_everything:
-            pipe._interrupt = True
-        return callback_kwargs
-
-    gen_kwargs["callback_on_step_end"] = interrupt_callback
+    width, height = request.get_width_height()
 
-    for i in range(batch_count):
-        if shared.stop_everything:
-            break
-        generator.manual_seed(int(seed + i))
-        batch_results = shared.image_model(**gen_kwargs).images
-        all_images.extend(batch_results)
-
-    t1 = time.time()
-    total_images = len(all_images)
-    total_steps = steps * batch_count
-    logger.info(f'Generated {total_images} {"image" if total_images == 1 else "images"} in {(t1 - t0):.2f} seconds ({total_steps / (t1 - t0):.2f} steps/s, seed {seed})')
-
-    # Save images
-    _save_images(all_images, prompt, negative_prompt, width, height, steps, seed, cfg_scale)
+    # Build state dict: GenerationOptions fields + image-specific keys
+    state = request.model_dump()
+    state.update({
+        'image_model_menu': shared.image_model_name,
+        'image_prompt': request.prompt,
+        'image_neg_prompt': request.negative_prompt,
+        'image_width': width,
+        'image_height': height,
+        'image_steps': request.steps,
+        'image_seed': request.image_seed,
+        'image_batch_size': request.batch_size,
+        'image_batch_count': request.batch_count,
+        'image_cfg_scale': request.cfg_scale,
+        'image_llm_variations': request.llm_variations,
+    })
+
+    # Exhaust generator, keep final result
+    images = []
+    for images, _ in generate(state, save_images=False):
+        pass
 
     # Build response
-    resp = {
-        'created': int(time.time()),
-        'data': []
-    }
-
-    for img in all_images:
+    resp = {'created': int(time.time()), 'data': []}
+    for img in images:
         b64 = _image_to_base64(img)
-        if response_format == 'b64_json':
+        if request.response_format == 'b64_json':
             resp['data'].append({'b64_json': b64})
         else:
             resp['data'].append({'url': f'data:image/png;base64,{b64}'})
@@ -126,29 +59,3 @@ def _image_to_base64(image) -> str:
     buffered = io.BytesIO()
     image.save(buffered, format="PNG")
     return base64.b64encode(buffered.getvalue()).decode('utf-8')
-
-
-def _save_images(images, prompt, negative_prompt, width, height, steps, seed, cfg_scale):
-    """Save images with metadata."""
-    date_str = datetime.now().strftime("%Y-%m-%d")
-    folder = os.path.join("user_data", "image_outputs", date_str)
-    os.makedirs(folder, exist_ok=True)
-
-    metadata = {
-        'image_prompt': prompt,
-        'image_neg_prompt': negative_prompt,
-        'image_width': width,
-        'image_height': height,
-        'image_steps': steps,
-        'image_seed': seed,
-        'image_cfg_scale': cfg_scale,
-        'model': getattr(shared, 'image_model_name', 'unknown'),
-    }
-
-    for idx, img in enumerate(images):
-        ts = datetime.now().strftime("%H-%M-%S")
-        filepath = os.path.join(folder, f"{ts}_{seed:010d}_{idx:03d}.png")
-
-        png_info = PngInfo()
-        png_info.add_text("image_gen_settings", json.dumps(metadata))
-        img.save(filepath, pnginfo=png_info)
diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 65805629da..12f99ba4b5 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -7,23 +7,24 @@
 from collections import deque
 from threading import Thread
 
-import extensions.openai.completions as OAIcompletions
-import extensions.openai.logits as OAIlogits
-import extensions.openai.models as OAImodels
 import uvicorn
-from extensions.openai.tokens import token_count, token_decode, token_encode
-from extensions.openai.utils import _start_cloudflared
 from fastapi import Depends, FastAPI, Header, HTTPException
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.requests import Request
 from fastapi.responses import JSONResponse
+from pydub import AudioSegment
+from sse_starlette import EventSourceResponse
+from starlette.concurrency import iterate_in_threadpool
+
+import extensions.openai.completions as OAIcompletions
+import extensions.openai.logits as OAIlogits
+import extensions.openai.models as OAImodels
+from extensions.openai.tokens import token_count, token_decode, token_encode
+from extensions.openai.utils import _start_cloudflared
 from modules import shared
 from modules.logging_colors import logger
 from modules.models import unload_model
 from modules.text_generation import stop_everything_event
-from pydub import AudioSegment
-from sse_starlette import EventSourceResponse
-from starlette.concurrency import iterate_in_threadpool
 
 from .typing import (
     ChatCompletionRequest,
@@ -232,20 +233,7 @@ async def handle_image_generation(request_data: ImageGenerationRequest):
     import extensions.openai.images as OAIimages
 
     async with image_generation_semaphore:
-        width, height = request_data.get_width_height()
-
-        response = await asyncio.to_thread(
-            OAIimages.generations,
-            prompt=request_data.prompt,
-            size=f"{width}x{height}",
-            response_format=request_data.response_format,
-            n=request_data.batch_size,  # <-- use resolved batch_size
-            negative_prompt=request_data.negative_prompt,
-            steps=request_data.steps,
-            seed=request_data.seed,
-            cfg_scale=request_data.cfg_scale,
-            batch_count=request_data.batch_count,
-        )
+        response = await asyncio.to_thread(OAIimages.generations, request_data)
         return JSONResponse(response)
 
 
diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index a24b844be3..dfdb9a7e4c 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -264,20 +264,18 @@ class LoadLorasRequest(BaseModel):
     lora_names: List[str]
 
 
-class ImageGenerationRequest(BaseModel):
-    """OpenAI-compatible image generation request with extended parameters."""
-    # Required
+class ImageGenerationRequestParams(BaseModel):
+    """Image-specific parameters for generation."""
     prompt: str
-
-    # Generation parameters
     negative_prompt: str = ""
     size: str = Field(default="1024x1024", description="'WIDTHxHEIGHT'")
     steps: int = Field(default=9, ge=1)
     cfg_scale: float = Field(default=0.0, ge=0.0)
-    seed: int = Field(default=-1, description="-1 for random")
+    image_seed: int = Field(default=-1, description="-1 for random")
     batch_size: int | None = Field(default=None, ge=1, description="Parallel batch size (VRAM heavy)")
     n: int = Field(default=1, ge=1, description="Alias for batch_size (OpenAI compatibility)")
     batch_count: int = Field(default=1, ge=1, description="Sequential batch count")
+    llm_variations: bool = False
 
     # OpenAI compatibility (unused)
     model: str | None = None
@@ -286,7 +284,6 @@ class ImageGenerationRequest(BaseModel):
 
     @model_validator(mode='after')
     def resolve_batch_size(self):
-        """Use batch_size if provided, otherwise fall back to n."""
         if self.batch_size is None:
             self.batch_size = self.n
         return self
@@ -299,6 +296,10 @@ def get_width_height(self) -> tuple[int, int]:
             return 1024, 1024
 
 
+class ImageGenerationRequest(GenerationOptions, ImageGenerationRequestParams):
+    pass
+
+
 class ImageGenerationResponse(BaseModel):
     created: int = int(time.time())
     data: List[dict]
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index ceb470ffb0..6ac0bc24c9 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -10,7 +10,6 @@
 from PIL.PngImagePlugin import PngInfo
 
 from modules import shared, ui, utils
-from modules.utils import check_model_loaded
 from modules.image_models import (
     get_pipeline_type,
     load_image_model,
@@ -19,7 +18,7 @@
 from modules.image_utils import open_image_safely
 from modules.logging_colors import logger
 from modules.text_generation import stop_everything_event
-from modules.utils import gradio
+from modules.utils import check_model_loaded, gradio
 
 ASPECT_RATIOS = {
     "1:1 Square": (1, 1),
@@ -725,13 +724,13 @@ def progress_bar_html(progress=0, text=""):
 
     return f'''
            
         
            
-            
            
+            
            
         
            
         
            {text}
            
     
            '''
 
 
-def generate(state):
+def generate(state, save_images=True):
     """
     Generate images using the loaded model.
     Automatically adjusts parameters based on pipeline type.
@@ -868,7 +867,8 @@ def run_batch():
             yield all_images, progress_bar_html((batch_idx + 1) / batch_count, f"Batch {batch_idx + 1}/{batch_count} complete")
 
         t1 = time.time()
-        save_generated_images(all_images, state, seed)
+        if save_images:
+            save_generated_images(all_images, state, seed)
 
         total_images = batch_count * int(state['image_batch_size'])
         logger.info(f'Generated {total_images} {"image" if total_images == 1 else "images"} in {(t1 - t0):.2f} seconds ({total_steps / (t1 - t0):.2f} steps/s, seed {seed})')

From ffef3c7b1dc5f3678a03abfa98fd54547de23796 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 10:44:35 -0800
Subject: [PATCH 1100/1701] Image: Make the LLM Variations prompt configurable

---
 extensions/openai/images.py    |  1 +
 extensions/openai/typing.py    |  1 +
 modules/shared.py              |  1 +
 modules/ui.py                  |  2 ++
 modules/ui_image_generation.py | 36 ++++++++++++++++++++++++++++------
 5 files changed, 35 insertions(+), 6 deletions(-)

diff --git a/extensions/openai/images.py b/extensions/openai/images.py
index 1ecb1e6399..0bb91a1ea7 100644
--- a/extensions/openai/images.py
+++ b/extensions/openai/images.py
@@ -36,6 +36,7 @@ def generations(request):
         'image_batch_count': request.batch_count,
         'image_cfg_scale': request.cfg_scale,
         'image_llm_variations': request.llm_variations,
+        'image_llm_variations_prompt': request.llm_variations_prompt or shared.settings.get('image_llm_variations_prompt', ''),
     })
 
     # Exhaust generator, keep final result
diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index dfdb9a7e4c..31a5dc6dc1 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -276,6 +276,7 @@ class ImageGenerationRequestParams(BaseModel):
     n: int = Field(default=1, ge=1, description="Alias for batch_size (OpenAI compatibility)")
     batch_count: int = Field(default=1, ge=1, description="Sequential batch count")
     llm_variations: bool = False
+    llm_variations_prompt: str | None = None
 
     # OpenAI compatibility (unused)
     model: str | None = None
diff --git a/modules/shared.py b/modules/shared.py
index 1ecc0d28c4..7fe9dbcf5c 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -320,6 +320,7 @@
     'image_batch_size': 1,
     'image_batch_count': 1,
     'image_llm_variations': False,
+    'image_llm_variations_prompt': 'Please create a creative variation of the image generation prompt above. Keep the same general subject and style, but vary the details, composition, lighting, or mood. Respond with only the new prompt, nothing else.',
     'image_model_menu': 'None',
     'image_dtype': 'bfloat16',
     'image_attn_backend': 'sdpa',
diff --git a/modules/ui.py b/modules/ui.py
index d95f7bc60a..919a5740b5 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -294,6 +294,7 @@ def list_interface_input_elements():
             'image_batch_size',
             'image_batch_count',
             'image_llm_variations',
+            'image_llm_variations_prompt',
             'image_model_menu',
             'image_dtype',
             'image_attn_backend',
@@ -549,6 +550,7 @@ def setup_auto_save():
             'image_batch_size',
             'image_batch_count',
             'image_llm_variations',
+            'image_llm_variations_prompt',
             'image_model_menu',
             'image_dtype',
             'image_attn_backend',
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 6ac0bc24c9..6eeb3d51f8 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -406,14 +406,25 @@ def create_ui():
                                     info="Z-Image Turbo: 0.0 | Qwen: 4.0"
                                 )
                                 shared.gradio['image_seed'] = gr.Number(label="Seed", value=shared.settings['image_seed'], precision=0, info="-1 = Random")
+
                             with gr.Column():
                                 shared.gradio['image_batch_size'] = gr.Slider(1, 32, value=shared.settings['image_batch_size'], step=1, label="Batch Size (VRAM Heavy)", info="Generates N images at once.")
                                 shared.gradio['image_batch_count'] = gr.Slider(1, 128, value=shared.settings['image_batch_count'], step=1, label="Sequential Count (Loop)", info="Repeats the generation N times.")
-                                shared.gradio['image_llm_variations'] = gr.Checkbox(
-                                    value=shared.settings['image_llm_variations'],
-                                    label='LLM Prompt Variations',
-                                    info='Use the loaded LLM to generate creative prompt variations for each sequential batch.'
-                                )
+
+                        gr.Markdown("### LLM Variations")
+                        shared.gradio['image_llm_variations'] = gr.Checkbox(
+                            value=shared.settings['image_llm_variations'],
+                            label='Activate',
+                            info='Use the loaded LLM to generate creative prompt variations for each sequential batch.'
+                        )
+                        shared.gradio['image_llm_variations_prompt'] = gr.Textbox(
+                            value=shared.settings['image_llm_variations_prompt'],
+                            label='Variation Prompt',
+                            lines=3,
+                            placeholder='Instructions for generating prompt variations...',
+                            visible=shared.settings['image_llm_variations'],
+                            info='The instruction given to the LLM for generating variations.'
+                        )
 
                     with gr.Column(scale=6, min_width=500):
                         with gr.Column(elem_classes=["viewport-container"]):
@@ -668,6 +679,14 @@ def create_event_handlers():
         show_progress=False
     )
 
+    # LLM Variations visibility toggle
+    shared.gradio['image_llm_variations'].change(
+        lambda x: gr.update(visible=x),
+        gradio('image_llm_variations'),
+        gradio('image_llm_variations_prompt'),
+        show_progress=False
+    )
+
 
 def generate_prompt_variation(state):
     """Generate a creative variation of the image prompt using the LLM."""
@@ -682,7 +701,12 @@ def generate_prompt_variation(state):
         logger.warning("No LLM loaded for prompt variation. Using original prompt.")
         return prompt
 
-    augmented_message = f"{prompt}\n\n=====\n\nPlease create a creative variation of the image generation prompt above. Keep the same general subject and style, but vary the details, composition, lighting, or mood. Respond with only the new prompt, nothing else."
+    # Get the custom variation prompt or use default
+    variation_instruction = state.get('image_llm_variations_prompt', '')
+    if not variation_instruction:
+        variation_instruction = 'Please create a creative variation of the image generation prompt above. Keep the same general subject and style, but vary the details, composition, lighting, or mood. Respond with only the new prompt, nothing else.'
+
+    augmented_message = f"{prompt}\n\n=====\n\n{variation_instruction}"
 
     # Use minimal state for generation
     var_state = state.copy()

From a90739f498e333ff3ffa0cba418139b1d1e00822 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 10:50:40 -0800
Subject: [PATCH 1101/1701] Image: Better LLM variation default prompt

---
 modules/shared.py              | 2 +-
 modules/ui_image_generation.py | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 7fe9dbcf5c..2f39e4959a 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -320,7 +320,7 @@
     'image_batch_size': 1,
     'image_batch_count': 1,
     'image_llm_variations': False,
-    'image_llm_variations_prompt': 'Please create a creative variation of the image generation prompt above. Keep the same general subject and style, but vary the details, composition, lighting, or mood. Respond with only the new prompt, nothing else.',
+    'image_llm_variations_prompt': 'Your task is to create a creative variation of the image generation prompt above. Keep the main subject but feel free to add an interesting setting, scenario, pose, atmosphere, or stylistic twist. Be specific and vivid. Respond with only the new prompt, nothing else.',
     'image_model_menu': 'None',
     'image_dtype': 'bfloat16',
     'image_attn_backend': 'sdpa',
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 6eeb3d51f8..1cf1b9550e 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -704,7 +704,7 @@ def generate_prompt_variation(state):
     # Get the custom variation prompt or use default
     variation_instruction = state.get('image_llm_variations_prompt', '')
     if not variation_instruction:
-        variation_instruction = 'Please create a creative variation of the image generation prompt above. Keep the same general subject and style, but vary the details, composition, lighting, or mood. Respond with only the new prompt, nothing else.'
+        variation_instruction = 'Your task is to create a creative variation of the image generation prompt above. Keep the main subject but feel free to add an interesting setting, scenario, pose, atmosphere, or stylistic twist. Be specific and vivid. Respond with only the new prompt, nothing else.'
 
     augmented_message = f"{prompt}\n\n=====\n\n{variation_instruction}"
 

From ac31a7c0082c08ae126c7140434c7fa053e53ea9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 15:45:04 -0800
Subject: [PATCH 1102/1701] Image: Organize the UI

---
 css/main.css                   |  6 ++++++
 modules/ui_image_generation.py | 30 ++++++++++++++----------------
 2 files changed, 20 insertions(+), 16 deletions(-)

diff --git a/css/main.css b/css/main.css
index 4dea6b9e71..e758696090 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1791,3 +1791,9 @@ button#swap-height-width {
 .dark #image-progress .image-ai-progress-text {
     color: #888;
 }
+
+#llm-prompt-variations {
+    position: absolute;
+    top: 0;
+    left: calc(100% - 174px);
+}
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 1cf1b9550e..5a1a8f797f 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -370,6 +370,19 @@ def create_ui():
                             lines=3,
                             value=shared.settings['image_neg_prompt']
                         )
+                        shared.gradio['image_llm_variations'] = gr.Checkbox(
+                            value=shared.settings['image_llm_variations'],
+                            label='LLM Prompt Variations',
+                            elem_id="llm-prompt-variations",
+                        )
+                        shared.gradio['image_llm_variations_prompt'] = gr.Textbox(
+                            value=shared.settings['image_llm_variations_prompt'],
+                            label='Variation Prompt',
+                            lines=3,
+                            placeholder='Instructions for generating prompt variations...',
+                            visible=shared.settings['image_llm_variations'],
+                            info='Use the loaded LLM to generate creative prompt variations for each sequential batch.'
+                        )
 
                         shared.gradio['image_generate_btn'] = gr.Button("Generate", variant="primary", size="lg")
                         shared.gradio['image_stop_btn'] = gr.Button("Stop", size="lg", visible=False)
@@ -411,21 +424,6 @@ def create_ui():
                                 shared.gradio['image_batch_size'] = gr.Slider(1, 32, value=shared.settings['image_batch_size'], step=1, label="Batch Size (VRAM Heavy)", info="Generates N images at once.")
                                 shared.gradio['image_batch_count'] = gr.Slider(1, 128, value=shared.settings['image_batch_count'], step=1, label="Sequential Count (Loop)", info="Repeats the generation N times.")
 
-                        gr.Markdown("### LLM Variations")
-                        shared.gradio['image_llm_variations'] = gr.Checkbox(
-                            value=shared.settings['image_llm_variations'],
-                            label='Activate',
-                            info='Use the loaded LLM to generate creative prompt variations for each sequential batch.'
-                        )
-                        shared.gradio['image_llm_variations_prompt'] = gr.Textbox(
-                            value=shared.settings['image_llm_variations_prompt'],
-                            label='Variation Prompt',
-                            lines=3,
-                            placeholder='Instructions for generating prompt variations...',
-                            visible=shared.settings['image_llm_variations'],
-                            info='The instruction given to the LLM for generating variations.'
-                        )
-
                     with gr.Column(scale=6, min_width=500):
                         with gr.Column(elem_classes=["viewport-container"]):
                             shared.gradio['image_output_gallery'] = gr.Gallery(label="Output", show_label=False, columns=2, rows=2, height="80vh", object_fit="contain", preview=True, elem_id="image-output-gallery")
@@ -735,7 +733,7 @@ def generate_prompt_variation(state):
         variation = variation[1:-1]
 
     if variation:
-        logger.info(f"Prompt variation: {variation}...")
+        logger.info(f"Prompt variation: {variation}")
         return variation
 
     return prompt

From 47a0fcd614b89a5705b65d93eaf89de544ab8880 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 16:11:35 -0800
Subject: [PATCH 1103/1701] Image: PNG metadata improvements

---
 modules/ui_image_generation.py | 38 +++++++++++++++++-----------------
 1 file changed, 19 insertions(+), 19 deletions(-)

diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 5a1a8f797f..a377f329f2 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -40,8 +40,6 @@
     'image_aspect_ratio',
     'image_steps',
     'image_seed',
-    'image_batch_size',
-    'image_batch_count',
     'image_cfg_scale',
 ]
 
@@ -184,8 +182,6 @@ def format_metadata_for_display(metadata):
         ('image_steps', 'Steps'),
         ('image_cfg_scale', 'CFG Scale'),
         ('image_seed', 'Seed'),
-        ('image_batch_size', 'Batch Size'),
-        ('image_batch_count', 'Batch Count'),
         ('model', 'Model'),
         ('generated_at', 'Generated At'),
     ]
@@ -314,11 +310,11 @@ def on_gallery_select(evt: gr.SelectData, current_page):
 def send_to_generate(selected_image_path):
     """Load settings from selected image and return updates for all Generate tab inputs."""
     if not selected_image_path or not os.path.exists(selected_image_path):
-        return [gr.update()] * 10 + ["No image selected"]
+        return [gr.update()] * 8 + ["No image selected"]
 
     metadata = read_image_metadata(selected_image_path)
     if not metadata:
-        return [gr.update()] * 10 + ["No settings found in this image"]
+        return [gr.update()] * 8 + ["No settings found in this image"]
 
     # Return updates for each input element in order
     updates = [
@@ -329,8 +325,6 @@ def send_to_generate(selected_image_path):
         gr.update(value=metadata.get('image_aspect_ratio', '1:1 Square')),
         gr.update(value=metadata.get('image_steps', 9)),
         gr.update(value=metadata.get('image_seed', -1)),
-        gr.update(value=metadata.get('image_batch_size', 1)),
-        gr.update(value=metadata.get('image_batch_count', 1)),
         gr.update(value=metadata.get('image_cfg_scale', 0.0)),
     ]
 
@@ -661,8 +655,6 @@ def create_event_handlers():
             'image_aspect_ratio',
             'image_steps',
             'image_seed',
-            'image_batch_size',
-            'image_batch_count',
             'image_cfg_scale',
             'image_gallery_status'
         ),
@@ -795,7 +787,7 @@ def generate(state, save_images=True):
         device = get_device()
         if device is None:
             device = "cpu"
-        generator = torch.Generator(device).manual_seed(int(seed))
+        generator = torch.Generator(device)
 
         all_images = []
 
@@ -804,14 +796,8 @@ def generate(state, save_images=True):
         if pipeline_type is None:
             pipeline_type = get_pipeline_type(shared.image_model)
 
-        # Process Prompt
         prompt = state['image_prompt']
 
-        if pipeline_type == 'qwenimage':
-            magic_suffix = ", Ultra HD, 4K, cinematic composition"
-            if magic_suffix.strip(", ") not in prompt:
-                prompt += magic_suffix
-
         shared.stop_everything = False
 
         batch_count = int(state['image_batch_count'])
@@ -862,7 +848,15 @@ def interrupt_callback(pipe, step_index, timestep, callback_kwargs):
 
             def run_batch():
                 try:
+                    # Apply magic suffix only at generation time for qwenimage
+                    clean_prompt = gen_kwargs["prompt"]
+                    if pipeline_type == 'qwenimage':
+                        magic_suffix = ", Ultra HD, 4K, cinematic composition"
+                        if magic_suffix.strip(", ") not in clean_prompt:
+                            gen_kwargs["prompt"] = clean_prompt + magic_suffix
+
                     result_holder.extend(shared.image_model(**gen_kwargs).images)
+                    gen_kwargs["prompt"] = clean_prompt  # restore
                 except Exception as e:
                     error_holder.append(e)
 
@@ -885,12 +879,18 @@ def run_batch():
             if error_holder:
                 raise error_holder[0]
 
+            # Save this batch's images with the actual prompt and seed used
+            if save_images:
+                batch_seed = seed + batch_idx
+                original_prompt = state['image_prompt']
+                state['image_prompt'] = gen_kwargs["prompt"]
+                save_generated_images(result_holder, state, batch_seed)
+                state['image_prompt'] = original_prompt
+
             all_images.extend(result_holder)
             yield all_images, progress_bar_html((batch_idx + 1) / batch_count, f"Batch {batch_idx + 1}/{batch_count} complete")
 
         t1 = time.time()
-        if save_images:
-            save_generated_images(all_images, state, seed)
 
         total_images = batch_count * int(state['image_batch_size'])
         logger.info(f'Generated {total_images} {"image" if total_images == 1 else "images"} in {(t1 - t0):.2f} seconds ({total_steps / (t1 - t0):.2f} steps/s, seed {seed})')

From b451bac082fadf00eca81e1e569d59a4df3ec8ac Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 16:33:46 -0800
Subject: [PATCH 1104/1701] Image: Improve a log message

---
 modules/ui_image_generation.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index a377f329f2..2cad5dc41e 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -725,7 +725,8 @@ def generate_prompt_variation(state):
         variation = variation[1:-1]
 
     if variation:
-        logger.info(f"Prompt variation: {variation}")
+        logger.info("Prompt variation:")
+        print(variation)
         return variation
 
     return prompt

From c7ad28a4cd076a025addc0656e42a74d91555dd8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 17:20:50 -0800
Subject: [PATCH 1105/1701] Image: Add the LLM-generated prompt to the API
 result

---
 extensions/openai/images.py    | 10 ++++++++--
 modules/ui_image_generation.py |  8 +++++++-
 2 files changed, 15 insertions(+), 3 deletions(-)

diff --git a/extensions/openai/images.py b/extensions/openai/images.py
index 0bb91a1ea7..e60470c3d1 100644
--- a/extensions/openai/images.py
+++ b/extensions/openai/images.py
@@ -48,10 +48,16 @@ def generations(request):
     resp = {'created': int(time.time()), 'data': []}
     for img in images:
         b64 = _image_to_base64(img)
+        image_obj = {
+            'revised_prompt': img.info.get('revised_prompt', request.prompt)
+        }
+
         if request.response_format == 'b64_json':
-            resp['data'].append({'b64_json': b64})
+            image_obj['b64_json'] = b64
         else:
-            resp['data'].append({'url': f'data:image/png;base64,{b64}'})
+            image_obj['url'] = f'data:image/png;base64,{b64}'
+
+        resp['data'].append(image_obj)
 
     return resp
 
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 2cad5dc41e..424589b6ea 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -856,7 +856,13 @@ def run_batch():
                         if magic_suffix.strip(", ") not in clean_prompt:
                             gen_kwargs["prompt"] = clean_prompt + magic_suffix
 
-                    result_holder.extend(shared.image_model(**gen_kwargs).images)
+                    batch_results = shared.image_model(**gen_kwargs).images
+
+                    # Store the modified prompt in the metadata
+                    for img in batch_results:
+                        img.info["revised_prompt"] = clean_prompt
+
+                    result_holder.extend(batch_results)
                     gen_kwargs["prompt"] = clean_prompt  # restore
                 except Exception as e:
                     error_holder.append(e)

From 3ef428efaa0e447d8d553e9387990b890aac5c6b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 17:34:17 -0800
Subject: [PATCH 1106/1701] Image: Remove llm_variations from the API

---
 extensions/openai/images.py | 3 +--
 extensions/openai/typing.py | 8 +-------
 2 files changed, 2 insertions(+), 9 deletions(-)

diff --git a/extensions/openai/images.py b/extensions/openai/images.py
index e60470c3d1..f46d549dbb 100644
--- a/extensions/openai/images.py
+++ b/extensions/openai/images.py
@@ -35,8 +35,7 @@ def generations(request):
         'image_batch_size': request.batch_size,
         'image_batch_count': request.batch_count,
         'image_cfg_scale': request.cfg_scale,
-        'image_llm_variations': request.llm_variations,
-        'image_llm_variations_prompt': request.llm_variations_prompt or shared.settings.get('image_llm_variations_prompt', ''),
+        'image_llm_variations': False,
     })
 
     # Exhaust generator, keep final result
diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 31a5dc6dc1..5ac9f6efff 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -264,7 +264,7 @@ class LoadLorasRequest(BaseModel):
     lora_names: List[str]
 
 
-class ImageGenerationRequestParams(BaseModel):
+class ImageGenerationRequest(BaseModel):
     """Image-specific parameters for generation."""
     prompt: str
     negative_prompt: str = ""
@@ -275,8 +275,6 @@ class ImageGenerationRequestParams(BaseModel):
     batch_size: int | None = Field(default=None, ge=1, description="Parallel batch size (VRAM heavy)")
     n: int = Field(default=1, ge=1, description="Alias for batch_size (OpenAI compatibility)")
     batch_count: int = Field(default=1, ge=1, description="Sequential batch count")
-    llm_variations: bool = False
-    llm_variations_prompt: str | None = None
 
     # OpenAI compatibility (unused)
     model: str | None = None
@@ -297,10 +295,6 @@ def get_width_height(self) -> tuple[int, int]:
             return 1024, 1024
 
 
-class ImageGenerationRequest(GenerationOptions, ImageGenerationRequestParams):
-    pass
-
-
 class ImageGenerationResponse(BaseModel):
     created: int = int(time.time())
     data: List[dict]

From 56f2a9512fbbc178d3ea5e5393c1dcc45219fed3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 17:34:27 -0800
Subject: [PATCH 1107/1701] Revert "Image: Add the LLM-generated prompt to the
 API result"

This reverts commit c7ad28a4cd076a025addc0656e42a74d91555dd8.
---
 extensions/openai/images.py    | 10 ++--------
 modules/ui_image_generation.py |  8 +-------
 2 files changed, 3 insertions(+), 15 deletions(-)

diff --git a/extensions/openai/images.py b/extensions/openai/images.py
index f46d549dbb..ef3f4169e4 100644
--- a/extensions/openai/images.py
+++ b/extensions/openai/images.py
@@ -47,16 +47,10 @@ def generations(request):
     resp = {'created': int(time.time()), 'data': []}
     for img in images:
         b64 = _image_to_base64(img)
-        image_obj = {
-            'revised_prompt': img.info.get('revised_prompt', request.prompt)
-        }
-
         if request.response_format == 'b64_json':
-            image_obj['b64_json'] = b64
+            resp['data'].append({'b64_json': b64})
         else:
-            image_obj['url'] = f'data:image/png;base64,{b64}'
-
-        resp['data'].append(image_obj)
+            resp['data'].append({'url': f'data:image/png;base64,{b64}'})
 
     return resp
 
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 424589b6ea..2cad5dc41e 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -856,13 +856,7 @@ def run_batch():
                         if magic_suffix.strip(", ") not in clean_prompt:
                             gen_kwargs["prompt"] = clean_prompt + magic_suffix
 
-                    batch_results = shared.image_model(**gen_kwargs).images
-
-                    # Store the modified prompt in the metadata
-                    for img in batch_results:
-                        img.info["revised_prompt"] = clean_prompt
-
-                    result_holder.extend(batch_results)
+                    result_holder.extend(shared.image_model(**gen_kwargs).images)
                     gen_kwargs["prompt"] = clean_prompt  # restore
                 except Exception as e:
                     error_holder.append(e)

From 15c6e43597203a0468501621f0e9ea443d22e8fa Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 17:41:09 -0800
Subject: [PATCH 1108/1701] Image: Add a revised_prompt field to API results
 for OpenAI compatibility

---
 extensions/openai/images.py | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/extensions/openai/images.py b/extensions/openai/images.py
index ef3f4169e4..1337771ae9 100644
--- a/extensions/openai/images.py
+++ b/extensions/openai/images.py
@@ -47,10 +47,15 @@ def generations(request):
     resp = {'created': int(time.time()), 'data': []}
     for img in images:
         b64 = _image_to_base64(img)
+
+        image_obj = {'revised_prompt': request.prompt}
+
         if request.response_format == 'b64_json':
-            resp['data'].append({'b64_json': b64})
+            image_obj['b64_json'] = b64
         else:
-            resp['data'].append({'url': f'data:image/png;base64,{b64}'})
+            image_obj['url'] = f'data:image/png;base64,{b64}'
+
+        resp['data'].append(image_obj)
 
     return resp
 

From b4f06a50b024f5c279d031f0d188e81249b1f98d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 19:11:31 -0800
Subject: [PATCH 1109/1701] fix: Pass bos_token and eos_token from metadata to
 jinja2

Fixes loading Seed-Instruct-36B
---
 modules/chat.py            |  6 ++++--
 modules/models_settings.py | 10 +++++++---
 modules/shared.py          |  2 ++
 3 files changed, 13 insertions(+), 5 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index acfc2f6663..d1474cfec0 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -112,7 +112,9 @@ def generate_chat_prompt(user_input, state, **kwargs):
         add_generation_prompt=False,
         enable_thinking=state['enable_thinking'],
         reasoning_effort=state['reasoning_effort'],
-        thinking_budget=-1 if state.get('enable_thinking', True) else 0
+        thinking_budget=-1 if state.get('enable_thinking', True) else 0,
+        bos_token=shared.bos_token,
+        eos_token=shared.eos_token,
     )
 
     chat_renderer = partial(
@@ -475,7 +477,7 @@ def get_stopping_strings(state):
 
     if state['mode'] in ['instruct', 'chat-instruct']:
         template = jinja_env.from_string(state['instruction_template_str'])
-        renderer = partial(template.render, add_generation_prompt=False)
+        renderer = partial(template.render, add_generation_prompt=False, bos_token=shared.bos_token, eos_token=shared.eos_token)
         renderers.append(renderer)
 
     if state['mode'] in ['chat']:
diff --git a/modules/models_settings.py b/modules/models_settings.py
index 6dc000b424..d333e269b8 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -89,8 +89,9 @@ def get_model_metadata(model):
             else:
                 bos_token = ""
 
-            template = template.replace('eos_token', "'{}'".format(eos_token))
-            template = template.replace('bos_token', "'{}'".format(bos_token))
+
+            shared.bos_token = bos_token
+            shared.eos_token = eos_token
 
             template = re.sub(r"\{\{-?\s*raise_exception\(.*?\)\s*-?\}\}", "", template, flags=re.DOTALL)
             template = re.sub(r'raise_exception\([^)]*\)', "''", template)
@@ -160,13 +161,16 @@ def get_model_metadata(model):
 
         # 4. If a template was found from any source, process it
         if template:
+            shared.bos_token = ''
+            shared.eos_token = ''
+
             for k in ['eos_token', 'bos_token']:
                 if k in metadata:
                     value = metadata[k]
                     if isinstance(value, dict):
                         value = value['content']
 
-                    template = template.replace(k, "'{}'".format(value))
+                    setattr(shared, k, value)
 
             template = re.sub(r"\{\{-?\s*raise_exception\(.*?\)\s*-?\}\}", "", template, flags=re.DOTALL)
             template = re.sub(r'raise_exception\([^)]*\)', "''", template)
diff --git a/modules/shared.py b/modules/shared.py
index 2f39e4959a..7b572dec64 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -19,6 +19,8 @@
 is_multimodal = False
 model_dirty_from_training = False
 lora_names = []
+bos_token = ''
+eos_token = ''
 
 # Image model variables
 image_model = None

From 8eac99599ad9a645f608d3814c63a0297e877de8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 4 Dec 2025 19:58:06 -0800
Subject: [PATCH 1110/1701] Image: Better LLM variation default prompt

---
 modules/shared.py              | 2 +-
 modules/ui_image_generation.py | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 7b572dec64..f6e86bdf0b 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -322,7 +322,7 @@
     'image_batch_size': 1,
     'image_batch_count': 1,
     'image_llm_variations': False,
-    'image_llm_variations_prompt': 'Your task is to create a creative variation of the image generation prompt above. Keep the main subject but feel free to add an interesting setting, scenario, pose, atmosphere, or stylistic twist. Be specific and vivid. Respond with only the new prompt, nothing else.',
+    'image_llm_variations_prompt': 'Write a variation of the image generation prompt above. Consider the intent of the user with that prompt and write something that will likely please them, with added details. Your reply should contain the new prompt and nothing else.',
     'image_model_menu': 'None',
     'image_dtype': 'bfloat16',
     'image_attn_backend': 'sdpa',
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 2cad5dc41e..139fd8914c 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -694,7 +694,7 @@ def generate_prompt_variation(state):
     # Get the custom variation prompt or use default
     variation_instruction = state.get('image_llm_variations_prompt', '')
     if not variation_instruction:
-        variation_instruction = 'Your task is to create a creative variation of the image generation prompt above. Keep the main subject but feel free to add an interesting setting, scenario, pose, atmosphere, or stylistic twist. Be specific and vivid. Respond with only the new prompt, nothing else.'
+        variation_instruction = 'Write a variation of the image generation prompt above. Consider the intent of the user with that prompt and write something that will likely please them, with added details. Your reply should contain the new prompt and nothing else.'
 
     augmented_message = f"{prompt}\n\n=====\n\n{variation_instruction}"
 

From afa29b9554c2f55213799af1d8e79735f2129eee Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 5 Dec 2025 05:53:22 -0800
Subject: [PATCH 1111/1701] Image: Several fixes

---
 README.md                      |  3 +--
 extensions/openai/images.py    |  3 +++
 modules/image_models.py        | 24 ++++++++++++-------
 modules/ui_image_generation.py | 42 +++++++++++++---------------------
 4 files changed, 36 insertions(+), 36 deletions(-)

diff --git a/README.md b/README.md
index 174fc2ac37..ef4b2160e0 100644
--- a/README.md
+++ b/README.md
@@ -28,8 +28,7 @@ A Gradio web UI for Large Language Models.
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
 - **Vision (multimodal models)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
-Image generation: A dedicated tab for diffusers models like Z-Image-Turbo and Qwen-Image. Features 4-bit/8-bit quantization and a persistent gallery with metadata (tutorial).
-- **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo** and **Qwen-Image**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
+- **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
 - **Web search**: Optionally search the internet with LLM-generated queries to add context to the conversation.
 - Aesthetic UI with dark and light themes.
 - Syntax highlighting for code blocks and LaTeX rendering for mathematical expressions.
diff --git a/extensions/openai/images.py b/extensions/openai/images.py
index 1337771ae9..f7be3d227b 100644
--- a/extensions/openai/images.py
+++ b/extensions/openai/images.py
@@ -43,6 +43,9 @@ def generations(request):
     for images, _ in generate(state, save_images=False):
         pass
 
+    if not images:
+        raise ServiceUnavailableError("Image generation failed or produced no images.")
+
     # Build response
     resp = {'created': int(time.time()), 'data': []}
     for img in images:
diff --git a/modules/image_models.py b/modules/image_models.py
index 8363533b28..28b2bb4f3a 100644
--- a/modules/image_models.py
+++ b/modules/image_models.py
@@ -141,16 +141,24 @@ def load_image_model(model_name, dtype='bfloat16', attn_backend='sdpa', cpu_offl
         if not cpu_offload:
             pipe.to(get_device())
 
-        # Set attention backend (if supported by the pipeline)
-        if hasattr(pipe, 'transformer') and hasattr(pipe.transformer, 'set_attention_backend'):
-            if attn_backend == 'flash_attention_2':
-                pipe.transformer.set_attention_backend("flash")
-            # sdpa is the default, no action needed
+        modules = ["transformer", "unet"]
 
+        # Set attention backend
+        if attn_backend == 'flash_attention_2':
+            for name in modules:
+                mod = getattr(pipe, name, None)
+                if hasattr(mod, "set_attention_backend"):
+                    mod.set_attention_backend("flash")
+                    break
+
+        # Compile model
         if compile_model:
-            if hasattr(pipe, 'transformer') and hasattr(pipe.transformer, 'compile'):
-                logger.info("Compiling model (first run will be slow)...")
-                pipe.transformer.compile()
+            for name in modules:
+                mod = getattr(pipe, name, None)
+                if hasattr(mod, "compile"):
+                    logger.info("Compiling model (first run will be slow)...")
+                    mod.compile()
+                    break
 
         if cpu_offload:
             pipe.enable_model_cpu_offload()
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 139fd8914c..0a1b6891e3 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -43,10 +43,6 @@
     'image_cfg_scale',
 ]
 
-# Cache for all image paths
-_image_cache = []
-_cache_timestamp = 0
-
 
 def round_to_step(value, step=STEP):
     return round(value / step) * step
@@ -134,6 +130,9 @@ def build_generation_metadata(state, actual_seed):
 
 def save_generated_images(images, state, actual_seed):
     """Save images with generation metadata embedded in PNG."""
+    if shared.args.multi_user:
+        return
+
     date_str = datetime.now().strftime("%Y-%m-%d")
     folder_path = os.path.join("user_data", "image_outputs", date_str)
     os.makedirs(folder_path, exist_ok=True)
@@ -157,9 +156,14 @@ def save_generated_images(images, state, actual_seed):
 def read_image_metadata(image_path):
     """Read generation metadata from PNG file."""
     try:
-        with open_image_safely(image_path) as img:
+        img = open_image_safely(image_path)
+        if img is None:
+            return None
+        try:
             if hasattr(img, 'text') and 'image_gen_settings' in img.text:
                 return json.loads(img.text['image_gen_settings'])
+        finally:
+            img.close()
     except Exception as e:
         logger.debug(f"Could not read metadata from {image_path}: {e}")
     return None
@@ -198,19 +202,12 @@ def format_metadata_for_display(metadata):
     return "\n\n".join(lines)
 
 
-def get_all_history_images(force_refresh=False):
-    """Get all history images sorted by modification time (newest first). Uses caching."""
-    global _image_cache, _cache_timestamp
-
+def get_all_history_images():
+    """Get all history images sorted by modification time (newest first)."""
     output_dir = os.path.join("user_data", "image_outputs")
     if not os.path.exists(output_dir):
         return []
 
-    # Check if we need to refresh cache
-    current_time = time.time()
-    if not force_refresh and _image_cache and (current_time - _cache_timestamp) < 2:
-        return _image_cache
-
     image_files = []
     for root, _, files in os.walk(output_dir):
         for file in files:
@@ -219,15 +216,12 @@ def get_all_history_images(force_refresh=False):
                 image_files.append((full_path, os.path.getmtime(full_path)))
 
     image_files.sort(key=lambda x: x[1], reverse=True)
-    _image_cache = [x[0] for x in image_files]
-    _cache_timestamp = current_time
-
-    return _image_cache
+    return [x[0] for x in image_files]
 
 
-def get_paginated_images(page=0, force_refresh=False):
+def get_paginated_images(page=0):
     """Get images for a specific page."""
-    all_images = get_all_history_images(force_refresh)
+    all_images = get_all_history_images()
     total_images = len(all_images)
     total_pages = max(1, (total_images + IMAGES_PER_PAGE - 1) // IMAGES_PER_PAGE)
 
@@ -250,7 +244,7 @@ def get_initial_page_info():
 
 def refresh_gallery(current_page=0):
     """Refresh gallery with current page."""
-    images, page, total_pages, total_images = get_paginated_images(current_page, force_refresh=True)
+    images, page, total_pages, total_images = get_paginated_images(current_page)
     page_info = f"Page {page + 1} of {total_pages} ({total_images} total images)"
     return images, page, page_info
 
@@ -286,11 +280,7 @@ def on_gallery_select(evt: gr.SelectData, current_page):
     if evt.index is None:
         return "", "Select an image to view its settings"
 
-    if not _image_cache:
-        get_all_history_images()
-
-    # Get the current page's images to find the actual file path
-    all_images = _image_cache
+    all_images = get_all_history_images()
     total_images = len(all_images)
 
     # Calculate the actual index in the full list

From b63d57158d02be08b76d215479a9e6b9ccef2881 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 5 Dec 2025 05:59:54 -0800
Subject: [PATCH 1112/1701] Image: Add TGW as a prefix to output images

---
 modules/ui_image_generation.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 0a1b6891e3..927048345e 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -142,7 +142,7 @@ def save_generated_images(images, state, actual_seed):
 
     for idx, img in enumerate(images):
         timestamp = datetime.now().strftime("%H-%M-%S")
-        filename = f"{timestamp}_{actual_seed:010d}_{idx:03d}.png"
+        filename = f"TGW_{timestamp}_{actual_seed:010d}_{idx:03d}.png"
         filepath = os.path.join(folder_path, filename)
 
         # Create PNG metadata

From 0dd468245c41ccd649c725689418ffd4116ef1e8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 5 Dec 2025 07:11:38 -0800
Subject: [PATCH 1113/1701] Image: Add back the gallery cache (for performance)

---
 modules/ui_image_generation.py | 31 ++++++++++++++++++++++++-------
 1 file changed, 24 insertions(+), 7 deletions(-)

diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 927048345e..28d4c6279d 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -43,6 +43,10 @@
     'image_cfg_scale',
 ]
 
+# Cache for all image paths
+_image_cache = []
+_cache_timestamp = 0
+
 
 def round_to_step(value, step=STEP):
     return round(value / step) * step
@@ -202,12 +206,19 @@ def format_metadata_for_display(metadata):
     return "\n\n".join(lines)
 
 
-def get_all_history_images():
-    """Get all history images sorted by modification time (newest first)."""
+def get_all_history_images(force_refresh=False):
+    """Get all history images sorted by modification time (newest first). Uses caching."""
+    global _image_cache, _cache_timestamp
+
     output_dir = os.path.join("user_data", "image_outputs")
     if not os.path.exists(output_dir):
         return []
 
+    # Check if we need to refresh cache
+    current_time = time.time()
+    if not force_refresh and _image_cache and (current_time - _cache_timestamp) < 2:
+        return _image_cache
+
     image_files = []
     for root, _, files in os.walk(output_dir):
         for file in files:
@@ -216,12 +227,15 @@ def get_all_history_images():
                 image_files.append((full_path, os.path.getmtime(full_path)))
 
     image_files.sort(key=lambda x: x[1], reverse=True)
-    return [x[0] for x in image_files]
+    _image_cache = [x[0] for x in image_files]
+    _cache_timestamp = current_time
 
+    return _image_cache
 
-def get_paginated_images(page=0):
+
+def get_paginated_images(page=0, force_refresh=False):
     """Get images for a specific page."""
-    all_images = get_all_history_images()
+    all_images = get_all_history_images(force_refresh)
     total_images = len(all_images)
     total_pages = max(1, (total_images + IMAGES_PER_PAGE - 1) // IMAGES_PER_PAGE)
 
@@ -244,7 +258,7 @@ def get_initial_page_info():
 
 def refresh_gallery(current_page=0):
     """Refresh gallery with current page."""
-    images, page, total_pages, total_images = get_paginated_images(current_page)
+    images, page, total_pages, total_images = get_paginated_images(current_page, force_refresh=True)
     page_info = f"Page {page + 1} of {total_pages} ({total_images} total images)"
     return images, page, page_info
 
@@ -280,7 +294,10 @@ def on_gallery_select(evt: gr.SelectData, current_page):
     if evt.index is None:
         return "", "Select an image to view its settings"
 
-    all_images = get_all_history_images()
+    if not _image_cache:
+        get_all_history_images()
+
+    all_images = _image_cache
     total_images = len(all_images)
 
     # Calculate the actual index in the full list

From c11c14590aa5d34985fd12a0b1100d4e51343ae9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 5 Dec 2025 08:08:11 -0800
Subject: [PATCH 1114/1701] Image: Better LLM variation default prompt

---
 modules/shared.py              | 2 +-
 modules/ui_image_generation.py | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index f6e86bdf0b..3f3742ded9 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -322,7 +322,7 @@
     'image_batch_size': 1,
     'image_batch_count': 1,
     'image_llm_variations': False,
-    'image_llm_variations_prompt': 'Write a variation of the image generation prompt above. Consider the intent of the user with that prompt and write something that will likely please them, with added details. Your reply should contain the new prompt and nothing else.',
+    'image_llm_variations_prompt': 'Write a variation of the image generation prompt above. Consider the intent of the user with that prompt and write something that will likely please them, with added details. Output only the new prompt. Do not add any explanations, prefixes, or additional text.',
     'image_model_menu': 'None',
     'image_dtype': 'bfloat16',
     'image_attn_backend': 'sdpa',
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 28d4c6279d..c55d243840 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -701,7 +701,7 @@ def generate_prompt_variation(state):
     # Get the custom variation prompt or use default
     variation_instruction = state.get('image_llm_variations_prompt', '')
     if not variation_instruction:
-        variation_instruction = 'Write a variation of the image generation prompt above. Consider the intent of the user with that prompt and write something that will likely please them, with added details. Your reply should contain the new prompt and nothing else.'
+        variation_instruction = 'Write a variation of the image generation prompt above. Consider the intent of the user with that prompt and write something that will likely please them, with added details. Output only the new prompt. Do not add any explanations, prefixes, or additional text.'
 
     augmented_message = f"{prompt}\n\n=====\n\n{variation_instruction}"
 

From 5848c7884d641ec3c76799585e6f44f379d8b5da Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 5 Dec 2025 10:24:51 -0800
Subject: [PATCH 1115/1701] Increase the height of the image output gallery

---
 css/main.css | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/css/main.css b/css/main.css
index e758696090..dad53c30a6 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1692,8 +1692,8 @@ button#swap-height-width {
 }
 
 #image-output-gallery, #image-output-gallery > :nth-child(2) {
-    height: calc(100vh - 83px);
-    max-height: calc(100vh - 83px);
+    height: calc(100vh - 66px);
+    max-height: calc(100vh - 66px);
 }
 
 #image-history-gallery, #image-history-gallery > :nth-child(2) {

From eba8a594665cd9407f2657fb26a1c4bbac60baf0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 5 Dec 2025 12:10:41 -0800
Subject: [PATCH 1116/1701] docs: Improve the image generation tutorial

---
 docs/Image Generation Tutorial.md | 67 ++++++++++++++++++++++++++-----
 1 file changed, 56 insertions(+), 11 deletions(-)

diff --git a/docs/Image Generation Tutorial.md b/docs/Image Generation Tutorial.md
index e7022c34f1..0c9eb8489d 100644
--- a/docs/Image Generation Tutorial.md	
+++ b/docs/Image Generation Tutorial.md	
@@ -1,20 +1,65 @@
 # Image Generation Tutorial
 
-This feature allows you to generate images using high-speed models like Z-Image-Turbo directly within the web UI.
+This feature allows you to generate images using `diffusers` models like [Tongyi-MAI/Z-Image-Turbo](https://huggingface.co/Tongyi-MAI/Z-Image-Turbo) directly within the web UI.
 
-## How to use
+## Installation
 
-1. Click on the **Image AI** tab at the top of the interface.
-2. Select the **Model** sub-tab.
-3. Copy and paste the following link into the **Download model** box:
+1. Clone the repository with
 
 ```
-https://huggingface.co/Tongyi-MAI/Z-Image-Turbo
+git clone https://github.com/oobabooga/text-generation-webui
 ```
 
-4. Click the **Download** button and wait for the confirmation message.
-5. In the **Model** dropdown menu, select the model you just downloaded (if you don't see it, click the 🔄 refresh button).
-6. Click **Load**.
-7. Go to the **Generate** sub-tab, type a prompt, and click **GENERATE**.
+or download it from [here](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip) and unzip it.
+
+2. Use the one-click installer.
+
+- Windows: Double click on `start_windows.bat`
+- Linux: Run `./start_linux.sh`
+- macOS: Run `./start_macos.sh`
+
+Note: Image generation does not work with the portable builds in `.zip` format in the [Releases page](https://github.com/oobabooga/text-generation-webui/releases). You need the "full" version of the web UI.
+
+## Downloading a model
+
+1. Once installation ends, browse to `http://127.0.0.1:7860/`.
+2. Click on "Image AI" on the left.
+3. Click on "Model" at the top.
+4. In the "Download model" field, paste `https://huggingface.co/Tongyi-MAI/Z-Image-Turbo` and click "Download".
+5. Wait for the download to finish (it's 31 GB).
+
+## Loading the model
+
+Select the quantization option in the "Quantization" menu and click "Load".
+
+The memory usage for `Z-Image-Turbo` for each option is:
+
+If you have less GPU memory than _, check the "CPU Offload" option.
+
+Note: The next time you launch the web UI, the model will get automatically loaded with your last settings when you try to generate an image. You do not need to go to the Model tab and click "Load" each time.
+
+## Generating images:
+
+1. While still in the "Image AI" page, go to the "Generate" tab.
+2. Type your prompt and click on the Generate button.
+
+### LLM Prompt Variations
+
+To use this feature, you need to load an LLM in the main "Model" tab on the left.
+
+If you have no idea what to use, do this to get started:
+
+1. Download [Qwen3-4B-Q3_K_M.gguf](https://huggingface.co/unsloth/Qwen3-4B-GGUF/resolve/main/Qwen3-4B-Q3_K_M.gguf) to your `text-generation-webui/user_data/models` folder.
+2. Select the model in the dropdown menu in the "Model" page.
+3. Click Load.
+
+Then go back to the "Image AI" page and check "LLM Prompt Variations".
+
+After that, your prompts will be automatically updated by the LLM each time you generate an image. If you use sequential batch count value greater than 1, a new prompt will be created for each sequential batch.
+
+The improvement in creativity is striking:
+
+### Model-specific settings
+
+- For Z-Image-Turbo, make sure to keep CFG Scale at 0 and Steps at 9. Do not write a Negative Prompt as it will get ignored with this CFG Scale value.
 
-> **Note for Z-Image-Turbo:** For the best results with this specific model, keep the **CFG Scale** slider at **0**.

From 11937de517e6b661c3a112faa29852f474a3b9bc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 5 Dec 2025 12:13:24 -0800
Subject: [PATCH 1117/1701] Use flash attention for image generation by default

---
 modules/shared.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 3f3742ded9..0a27f33dc0 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -58,7 +58,7 @@
 group.add_argument('--image-model', type=str, help='Name of the image model to select on startup (overrides saved setting).')
 group.add_argument('--image-model-dir', type=str, default='user_data/image_models', help='Path to directory with all the image models.')
 group.add_argument('--image-dtype', type=str, default=None, choices=['bfloat16', 'float16'], help='Data type for image model.')
-group.add_argument('--image-attn-backend', type=str, default=None, choices=['sdpa', 'flash_attention_2'], help='Attention backend for image model.')
+group.add_argument('--image-attn-backend', type=str, default=None, choices=['flash_attention_2', 'sdpa'], help='Attention backend for image model.')
 group.add_argument('--image-cpu-offload', action='store_true', help='Enable CPU offloading for image model.')
 group.add_argument('--image-compile', action='store_true', help='Compile the image model for faster inference.')
 group.add_argument('--image-quant', type=str, default=None,
@@ -325,7 +325,7 @@
     'image_llm_variations_prompt': 'Write a variation of the image generation prompt above. Consider the intent of the user with that prompt and write something that will likely please them, with added details. Output only the new prompt. Do not add any explanations, prefixes, or additional text.',
     'image_model_menu': 'None',
     'image_dtype': 'bfloat16',
-    'image_attn_backend': 'sdpa',
+    'image_attn_backend': 'flash_attention_2',
     'image_cpu_offload': False,
     'image_compile': False,
     'image_quant': 'none',

From 6ca99910ba82a8b2212562ccaaf2e584aa369642 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 5 Dec 2025 13:08:46 -0800
Subject: [PATCH 1118/1701] Image: Quantize the text encoder for lower VRAM

---
 modules/image_models.py | 35 ++++++++++++++++++++++-------------
 1 file changed, 22 insertions(+), 13 deletions(-)

diff --git a/modules/image_models.py b/modules/image_models.py
index 28b2bb4f3a..290aaf1980 100644
--- a/modules/image_models.py
+++ b/modules/image_models.py
@@ -8,17 +8,14 @@
 def get_quantization_config(quant_method):
     """
     Get the appropriate quantization config based on the selected method.
-
-    Args:
-        quant_method: One of 'none', 'bnb-8bit', 'bnb-4bit',
-                      'torchao-int8wo', 'torchao-fp4', 'torchao-float8wo'
-
-    Returns:
-        PipelineQuantizationConfig or None
+    Applies quantization to both the transformer and the text_encoder.
     """
     import torch
-    from diffusers import BitsAndBytesConfig, TorchAoConfig
+    # Import BitsAndBytesConfig from BOTH libraries to be safe
+    from diffusers import BitsAndBytesConfig as DiffusersBnBConfig
+    from diffusers import TorchAoConfig
     from diffusers.quantizers import PipelineQuantizationConfig
+    from transformers import BitsAndBytesConfig as TransformersBnBConfig
 
     if quant_method == 'none' or not quant_method:
         return None
@@ -27,7 +24,10 @@ def get_quantization_config(quant_method):
     elif quant_method == 'bnb-8bit':
         return PipelineQuantizationConfig(
             quant_mapping={
-                "transformer": BitsAndBytesConfig(
+                "transformer": DiffusersBnBConfig(
+                    load_in_8bit=True
+                ),
+                "text_encoder": TransformersBnBConfig(
                     load_in_8bit=True
                 )
             }
@@ -37,7 +37,13 @@ def get_quantization_config(quant_method):
     elif quant_method == 'bnb-4bit':
         return PipelineQuantizationConfig(
             quant_mapping={
-                "transformer": BitsAndBytesConfig(
+                "transformer": DiffusersBnBConfig(
+                    load_in_4bit=True,
+                    bnb_4bit_quant_type="nf4",
+                    bnb_4bit_compute_dtype=torch.bfloat16,
+                    bnb_4bit_use_double_quant=True
+                ),
+                "text_encoder": TransformersBnBConfig(
                     load_in_4bit=True,
                     bnb_4bit_quant_type="nf4",
                     bnb_4bit_compute_dtype=torch.bfloat16,
@@ -50,7 +56,8 @@ def get_quantization_config(quant_method):
     elif quant_method == 'torchao-int8wo':
         return PipelineQuantizationConfig(
             quant_mapping={
-                "transformer": TorchAoConfig("int8wo")
+                "transformer": TorchAoConfig("int8wo"),
+                "text_encoder": TorchAoConfig("int8wo")
             }
         )
 
@@ -58,7 +65,8 @@ def get_quantization_config(quant_method):
     elif quant_method == 'torchao-fp4':
         return PipelineQuantizationConfig(
             quant_mapping={
-                "transformer": TorchAoConfig("fp4_e2m1")
+                "transformer": TorchAoConfig("fp4_e2m1"),
+                "text_encoder": TorchAoConfig("fp4_e2m1")
             }
         )
 
@@ -66,7 +74,8 @@ def get_quantization_config(quant_method):
     elif quant_method == 'torchao-float8wo':
         return PipelineQuantizationConfig(
             quant_mapping={
-                "transformer": TorchAoConfig("float8wo")
+                "transformer": TorchAoConfig("float8wo"),
+                "text_encoder": TorchAoConfig("float8wo")
             }
         )
 

From e20b2d38ff38fbd6451c8ff53c9e12fc9a327a14 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 5 Dec 2025 14:12:08 -0800
Subject: [PATCH 1119/1701] docs: Add VRAM measurements for Z-Image-Turbo

---
 docs/Image Generation Tutorial.md | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/docs/Image Generation Tutorial.md b/docs/Image Generation Tutorial.md
index 0c9eb8489d..a74a4dddca 100644
--- a/docs/Image Generation Tutorial.md	
+++ b/docs/Image Generation Tutorial.md	
@@ -34,7 +34,15 @@ Select the quantization option in the "Quantization" menu and click "Load".
 
 The memory usage for `Z-Image-Turbo` for each option is:
 
-If you have less GPU memory than _, check the "CPU Offload" option.
+| Quantization Method | VRAM Usage |
+| :--- | :--- |
+| **None (FP16/BF16)** | 25613 MiB |
+| **bnb-8bit** | 16301 MiB |
+| **bnb-8bit + CPU Offload** | 16235 MiB |
+| **bnb-4bit** | 11533 MiB |
+| **bnb-4bit + CPU Offload** | 7677 MiB |
+
+The `torchao` options support `torch.compile` for faster image generation, with `float8wo` specifically providing native hardware acceleration for RTX 40-series and newer GPUs.
 
 Note: The next time you launch the web UI, the model will get automatically loaded with your last settings when you try to generate an image. You do not need to go to the Model tab and click "Load" each time.
 

From 17b12567d82e7459438cd54f361805993ca4ca59 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 5 Dec 2025 14:15:15 -0800
Subject: [PATCH 1120/1701] docs: Small changes

---
 docs/Image Generation Tutorial.md | 11 +++++------
 1 file changed, 5 insertions(+), 6 deletions(-)

diff --git a/docs/Image Generation Tutorial.md b/docs/Image Generation Tutorial.md
index a74a4dddca..85f891d59f 100644
--- a/docs/Image Generation Tutorial.md	
+++ b/docs/Image Generation Tutorial.md	
@@ -51,6 +51,10 @@ Note: The next time you launch the web UI, the model will get automatically load
 1. While still in the "Image AI" page, go to the "Generate" tab.
 2. Type your prompt and click on the Generate button.
 
+### Model-specific settings
+
+- For Z-Image-Turbo, make sure to keep CFG Scale at 0 and Steps at 9. Do not write a Negative Prompt as it will get ignored with this CFG Scale value.
+
 ### LLM Prompt Variations
 
 To use this feature, you need to load an LLM in the main "Model" tab on the left.
@@ -63,11 +67,6 @@ If you have no idea what to use, do this to get started:
 
 Then go back to the "Image AI" page and check "LLM Prompt Variations".
 
-After that, your prompts will be automatically updated by the LLM each time you generate an image. If you use sequential batch count value greater than 1, a new prompt will be created for each sequential batch.
+After that, your prompts will be automatically updated by the LLM each time you generate an image. If you use a "Sequential Count" value greater than 1, a new prompt will be created for each sequential batch.
 
 The improvement in creativity is striking:
-
-### Model-specific settings
-
-- For Z-Image-Turbo, make sure to keep CFG Scale at 0 and Steps at 9. Do not write a Negative Prompt as it will get ignored with this CFG Scale value.
-

From 1a9ed1fe98a2d51a860d3c4f747ba46c543d0bf3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 6 Dec 2025 05:21:26 -0800
Subject: [PATCH 1121/1701] Fix the height of the image output gallery

---
 css/main.css | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/css/main.css b/css/main.css
index dad53c30a6..e758696090 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1692,8 +1692,8 @@ button#swap-height-width {
 }
 
 #image-output-gallery, #image-output-gallery > :nth-child(2) {
-    height: calc(100vh - 66px);
-    max-height: calc(100vh - 66px);
+    height: calc(100vh - 83px);
+    max-height: calc(100vh - 83px);
 }
 
 #image-history-gallery, #image-history-gallery > :nth-child(2) {

From 455dc06db08fc347f1bbd09c4ac74134954fc641 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 6 Dec 2025 05:43:00 -0800
Subject: [PATCH 1122/1701] Serve the original PNG images in the UI instead of
 webp

---
 modules/ui_image_generation.py | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)

diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index c55d243840..2fb93fd8ff 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -133,9 +133,9 @@ def build_generation_metadata(state, actual_seed):
 
 
 def save_generated_images(images, state, actual_seed):
-    """Save images with generation metadata embedded in PNG."""
+    """Save images with generation metadata embedded in PNG. Returns list of saved file paths."""
     if shared.args.multi_user:
-        return
+        return []
 
     date_str = datetime.now().strftime("%Y-%m-%d")
     folder_path = os.path.join("user_data", "image_outputs", date_str)
@@ -144,6 +144,7 @@ def save_generated_images(images, state, actual_seed):
     metadata = build_generation_metadata(state, actual_seed)
     metadata_json = json.dumps(metadata, ensure_ascii=False)
 
+    saved_paths = []
     for idx, img in enumerate(images):
         timestamp = datetime.now().strftime("%H-%M-%S")
         filename = f"TGW_{timestamp}_{actual_seed:010d}_{idx:03d}.png"
@@ -155,6 +156,9 @@ def save_generated_images(images, state, actual_seed):
 
         # Save with metadata
         img.save(filepath, pnginfo=png_info)
+        saved_paths.append(filepath)
+
+    return saved_paths
 
 
 def read_image_metadata(image_path):
@@ -892,10 +896,14 @@ def run_batch():
                 batch_seed = seed + batch_idx
                 original_prompt = state['image_prompt']
                 state['image_prompt'] = gen_kwargs["prompt"]
-                save_generated_images(result_holder, state, batch_seed)
+                saved_paths = save_generated_images(result_holder, state, batch_seed)
                 state['image_prompt'] = original_prompt
+                # Use file paths so gallery serves actual PNGs with metadata
+                all_images.extend(saved_paths)
+            else:
+                # Fallback to PIL objects if not saving
+                all_images.extend(result_holder)
 
-            all_images.extend(result_holder)
             yield all_images, progress_bar_html((batch_idx + 1) / batch_count, f"Batch {batch_idx + 1}/{batch_count} complete")
 
         t1 = time.time()

From 6411142111db1736a4fffac72fd5ebc63ea5de11 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 6 Dec 2025 06:36:16 -0800
Subject: [PATCH 1123/1701] docs: Small changes

---
 docs/Image Generation Tutorial.md | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/docs/Image Generation Tutorial.md b/docs/Image Generation Tutorial.md
index 85f891d59f..875d1d4491 100644
--- a/docs/Image Generation Tutorial.md	
+++ b/docs/Image Generation Tutorial.md	
@@ -2,6 +2,9 @@
 
 This feature allows you to generate images using `diffusers` models like [Tongyi-MAI/Z-Image-Turbo](https://huggingface.co/Tongyi-MAI/Z-Image-Turbo) directly within the web UI.
 
+print
+
+
 ## Installation
 
 1. Clone the repository with
@@ -36,11 +39,11 @@ The memory usage for `Z-Image-Turbo` for each option is:
 
 | Quantization Method | VRAM Usage |
 | :--- | :--- |
-| **None (FP16/BF16)** | 25613 MiB |
-| **bnb-8bit** | 16301 MiB |
-| **bnb-8bit + CPU Offload** | 16235 MiB |
-| **bnb-4bit** | 11533 MiB |
-| **bnb-4bit + CPU Offload** | 7677 MiB |
+| None (FP16/BF16) | 25613 MiB |
+| bnb-8bit | 16301 MiB |
+| bnb-8bit + CPU Offload | 16235 MiB |
+| bnb-4bit | 11533 MiB |
+| bnb-4bit + CPU Offload | 7677 MiB |
 
 The `torchao` options support `torch.compile` for faster image generation, with `float8wo` specifically providing native hardware acceleration for RTX 40-series and newer GPUs.
 
@@ -70,3 +73,5 @@ Then go back to the "Image AI" page and check "LLM Prompt Variations".
 After that, your prompts will be automatically updated by the LLM each time you generate an image. If you use a "Sequential Count" value greater than 1, a new prompt will be created for each sequential batch.
 
 The improvement in creativity is striking:
+
+comparison_collage

From 0100ad1bd7c0fbb068585902db78d0edce3d497d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 6 Dec 2025 06:39:30 -0800
Subject: [PATCH 1124/1701] Add user_data/image_outputs to the Gradio allowed
 paths

---
 server.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/server.py b/server.py
index 0c5d14ce91..d8fb2c224b 100644
--- a/server.py
+++ b/server.py
@@ -101,6 +101,11 @@ def create_interface():
             auth.extend(x.strip() for line in file for x in line.split(',') if x.strip())
     auth = [tuple(cred.split(':')) for cred in auth]
 
+    # Allowed paths
+    allowed_paths = ["css", "js", "extensions", "user_data/cache"]
+    if not shared.args.multi_user:
+        allowed_paths.append("user_data/image_outputs")
+
     # Import the extensions and execute their setup() functions
     if shared.args.extensions is not None and len(shared.args.extensions) > 0:
         extensions_module.load_extensions()
@@ -237,7 +242,7 @@ def create_interface():
             ssl_keyfile=shared.args.ssl_keyfile,
             ssl_certfile=shared.args.ssl_certfile,
             root_path=shared.args.subpath,
-            allowed_paths=["css", "js", "extensions", "user_data/cache"]
+            allowed_paths=allowed_paths,
         )
 
 

From 02518a96a9c9e75947a4d41e22a758fe9d83310c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 6 Dec 2025 06:55:06 -0800
Subject: [PATCH 1125/1701] Lint

---
 modules/models_settings.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index d333e269b8..e9f19a0603 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -89,7 +89,6 @@ def get_model_metadata(model):
             else:
                 bos_token = ""
 
-
             shared.bos_token = bos_token
             shared.eos_token = eos_token
 

From 1c36559e2b1b453a526b682b7db89486f9c3753c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 6 Dec 2025 07:05:00 -0800
Subject: [PATCH 1126/1701] Add a News section to the README

---
 README.md | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/README.md b/README.md
index ef4b2160e0..8065ca71e9 100644
--- a/README.md
+++ b/README.md
@@ -21,6 +21,10 @@ A Gradio web UI for Large Language Models.
 |:---:|:---:|
 |![Image1](https://github.com/oobabooga/screenshots/raw/main/DEFAULT-3.5.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/PARAMETERS-3.5.png) |
 
+## 🔥 News
+
+- The project now supports **image generation**! Including Z-Image-Turbo, 4bit/8bit quantization, `torch.compile`, and LLM-generated prompt variations ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
+
 ## Features
 
 - Supports multiple local text generation backends, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).

From 194e4c285fe0358a5ed15109cda1aaad204ee023 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 6 Dec 2025 08:14:48 -0800
Subject: [PATCH 1127/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 4 ++--
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 4 ++--
 requirements/portable/requirements_amd_noavx2.txt      | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 4 ++--
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 36 insertions(+), 36 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index c1f87988e1..2f1c0f09de 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -44,8 +44,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 512231e037..8d9f8c43d7 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -42,7 +42,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 1ccc507c6f..f69c24668a 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -42,7 +42,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 4d8aa7714d..fd2dad7fea 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -42,5 +42,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 5290aa714d..d71dfbe522 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -42,5 +42,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index d9c76a3144..2d17471334 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -42,5 +42,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 1a796c2199..2e047b42c8 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -42,5 +42,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 8105abaad6..4baca6c9a6 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -44,8 +44,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 0b944d48b6..21cef622d9 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 9b43c901e9..bce5a9e0d2 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index 825e6dec50..18257eea84 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 864b67756d..e97ff558db 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 03e090cf02..d113431c96 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 7dd851e876..1553096c55 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 63949d9f94..ce259dcc92 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 34652264a0..887a454229 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 28849ff7fa..d9eff1ac17 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 365a798e5b..4a5ee8a311 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.64.0/llama_cpp_binaries-0.64.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From c026dbaf647a1a6ead61f88848db4bb5f292b47a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 6 Dec 2025 08:23:21 -0800
Subject: [PATCH 1128/1701] Fix API requests always returning the same
 'created' time

---
 extensions/openai/typing.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 5ac9f6efff..5fea2324b7 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -130,7 +130,7 @@ class CompletionRequest(GenerationOptions, CompletionRequestParams):
 class CompletionResponse(BaseModel):
     id: str
     choices: List[dict]
-    created: int = int(time.time())
+    created: int = Field(default_factory=lambda: int(time.time()))
     model: str
     object: str = "text_completion"
     usage: dict
@@ -178,7 +178,7 @@ class ChatCompletionRequest(GenerationOptions, ChatCompletionRequestParams):
 class ChatCompletionResponse(BaseModel):
     id: str
     choices: List[dict]
-    created: int = int(time.time())
+    created: int = Field(default_factory=lambda: int(time.time()))
     model: str
     object: str = "chat.completion"
     usage: dict
@@ -296,7 +296,7 @@ def get_width_height(self) -> tuple[int, int]:
 
 
 class ImageGenerationResponse(BaseModel):
-    created: int = int(time.time())
+    created: int = Field(default_factory=lambda: int(time.time()))
     data: List[dict]
 
 

From f93cc4b5c37a929aa1a91c2648c571908d9a47cf Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 6 Dec 2025 08:33:06 -0800
Subject: [PATCH 1129/1701] Add an API example to the image generation tutorial

---
 docs/Image Generation Tutorial.md | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)

diff --git a/docs/Image Generation Tutorial.md b/docs/Image Generation Tutorial.md
index 875d1d4491..0d562041f4 100644
--- a/docs/Image Generation Tutorial.md	
+++ b/docs/Image Generation Tutorial.md	
@@ -75,3 +75,24 @@ After that, your prompts will be automatically updated by the LLM each time you
 The improvement in creativity is striking:
 
 comparison_collage
+
+## Generating images over API
+
+It is possible to generate images usign the project's API. Just make sure to start the server with `--api`, either by
+
+1. Passing the `--api` flag to your `start` script, like `./start_linux.sh --api`, or
+2. Writing `--api` to your `user_data/CMD_FLAGS.txt` file and relaunching the web UI.
+
+Here is an API call example:
+
+```
+curl http://127.0.0.1:5000/v1/images/generations \
+  -H "Content-Type: application/json" \
+  -d '{
+    "prompt": "an orange tree",
+    "steps": 9,
+    "cfg_scale": 0,
+    "batch_size": 1,
+    "batch_count": 1
+  }'
+```

From 160a25165af5f5a0b759269657a241fe75f684d5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 6 Dec 2025 08:41:12 -0800
Subject: [PATCH 1130/1701] docs: Small change

---
 docs/Image Generation Tutorial.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/docs/Image Generation Tutorial.md b/docs/Image Generation Tutorial.md
index 0d562041f4..b285bbc1ce 100644
--- a/docs/Image Generation Tutorial.md	
+++ b/docs/Image Generation Tutorial.md	
@@ -72,7 +72,7 @@ Then go back to the "Image AI" page and check "LLM Prompt Variations".
 
 After that, your prompts will be automatically updated by the LLM each time you generate an image. If you use a "Sequential Count" value greater than 1, a new prompt will be created for each sequential batch.
 
-The improvement in creativity is striking:
+The improvement in creativity is striking (prompt: `Photo of a beautiful woman at night under moonlight`):
 
 comparison_collage
 

From 1762312fb48e74cca45c916b92d616ed14d6ee9e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 6 Dec 2025 20:10:32 -0800
Subject: [PATCH 1131/1701] Use random instead of np.random for image seeds
 (makes it work on Windows)

---
 modules/ui_image_generation.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 2fb93fd8ff..6b39c5b5a0 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -1,12 +1,12 @@
 import json
 import os
+import random
 import time
 import traceback
 from datetime import datetime
 from pathlib import Path
 
 import gradio as gr
-import numpy as np
 from PIL.PngImagePlugin import PngInfo
 
 from modules import shared, ui, utils
@@ -794,7 +794,7 @@ def generate(state, save_images=True):
 
         seed = state['image_seed']
         if seed == -1:
-            seed = np.random.randint(0, 2**32 - 1)
+            seed = random.randint(0, 2**32 - 1)
 
         device = get_device()
         if device is None:

From 85f2df92e9e736708cab5ba8ea9b7484036d7ded Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 7 Dec 2025 06:56:58 -0800
Subject: [PATCH 1132/1701] Use flash_attention_2 by default for Transformers
 models

---
 modules/shared.py        | 2 +-
 modules/ui_model_menu.py | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 0a27f33dc0..b2aeadc60d 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -112,7 +112,7 @@
 group.add_argument('--trust-remote-code', action='store_true', help='Set trust_remote_code=True while loading the model. Necessary for some models.')
 group.add_argument('--force-safetensors', action='store_true', help='Set use_safetensors=True while loading the model. This prevents arbitrary code execution.')
 group.add_argument('--no_use_fast', action='store_true', help='Set use_fast=False while loading the tokenizer (it\'s True by default). Use this if you have any problems related to use_fast.')
-group.add_argument('--attn-implementation', type=str, default='sdpa', metavar="IMPLEMENTATION", help='Attention implementation. Valid options: sdpa, eager, flash_attention_2.')
+group.add_argument('--attn-implementation', type=str, default='flash_attention_2', metavar="IMPLEMENTATION", help='Attention implementation. Valid options: flash_attention_2, sdpa, eager.')
 
 # bitsandbytes 4-bit
 group = parser.add_argument_group('bitsandbytes 4-bit')
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 86adc2297e..aab0fcaf07 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -44,7 +44,7 @@ def create_ui():
                             shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. Common values: 4096, 8192, 16384, 32768, 65536, 131072.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
-                            shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
+                            shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['flash_attention_2', 'sdpa', 'eager'], value=shared.args.attn_implementation, info='Attention implementation.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
                             shared.gradio['tp_backend'] = gr.Dropdown(label="tp-backend", choices=['native', 'nccl'], value=shared.args.tp_backend, info='The backend for tensor parallelism.')
 

From 17bd8d10f080fc744080bc104dff0b2c47371d80 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 7 Dec 2025 09:37:18 -0800
Subject: [PATCH 1133/1701] Update exllamav3 to 0.0.17

---
 requirements/full/requirements.txt        | 4 ++--
 requirements/full/requirements_noavx2.txt | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 2f1c0f09de..b1dd9d837f 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -46,8 +46,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 4baca6c9a6..072f65bb3b 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -46,8 +46,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.16/exllamav3-0.0.16+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 058e78411d97737011420d4c4341dc8d2d4643b9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 7 Dec 2025 10:16:08 -0800
Subject: [PATCH 1134/1701] docs: Small changes

---
 docs/Image Generation Tutorial.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/docs/Image Generation Tutorial.md b/docs/Image Generation Tutorial.md
index b285bbc1ce..43d6de7c84 100644
--- a/docs/Image Generation Tutorial.md	
+++ b/docs/Image Generation Tutorial.md	
@@ -60,7 +60,7 @@ Note: The next time you launch the web UI, the model will get automatically load
 
 ### LLM Prompt Variations
 
-To use this feature, you need to load an LLM in the main "Model" tab on the left.
+To use this feature, you need to load an LLM in the main "Model" page on the left.
 
 If you have no idea what to use, do this to get started:
 
@@ -78,7 +78,7 @@ The improvement in creativity is striking (prompt: `Photo of a beautiful woman a
 
 ## Generating images over API
 
-It is possible to generate images usign the project's API. Just make sure to start the server with `--api`, either by
+It is possible to generate images using the project's API. Just make sure to start the server with `--api`, either by
 
 1. Passing the `--api` flag to your `start` script, like `./start_linux.sh --api`, or
 2. Writing `--api` to your `user_data/CMD_FLAGS.txt` file and relaunching the web UI.

From 3b8369a6798fbc504a980e50f579b7e042336762 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 7 Dec 2025 11:18:36 -0800
Subject: [PATCH 1135/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 4 ++--
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 4 ++--
 requirements/portable/requirements_amd_noavx2.txt      | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 4 ++--
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 36 insertions(+), 36 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index b1dd9d837f..8bfa1318ff 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -44,8 +44,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 8d9f8c43d7..088f7713d5 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -42,7 +42,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index f69c24668a..e4e55e4dd4 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -42,7 +42,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index fd2dad7fea..5903fe6081 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -42,5 +42,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index d71dfbe522..4751757880 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -42,5 +42,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 2d17471334..1fee7121a9 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -42,5 +42,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 2e047b42c8..204af32a7d 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -42,5 +42,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 072f65bb3b..86208b9f54 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -44,8 +44,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 21cef622d9..4a92fb8d8d 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index bce5a9e0d2..5ad1fe3a76 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index 18257eea84..9c3e285116 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index e97ff558db..8a618e3551 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index d113431c96..c1c643e8f5 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 1553096c55..f4bc8caba7 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index ce259dcc92..81f9bf2ec4 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 887a454229..6cc87a4ea2 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index d9eff1ac17..25c3638fb4 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 4a5ee8a311..54d0cd1a19 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.65.0/llama_cpp_binaries-0.65.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 1ec9f708e5cbb7a2d74fa35f56735d80991163c2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 7 Dec 2025 11:49:22 -0800
Subject: [PATCH 1136/1701] Clear the torch cache between sequential image
 generations

---
 modules/ui_image_generation.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 6b39c5b5a0..bcc5bb2d01 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -848,6 +848,9 @@ def interrupt_callback(pipe, step_index, timestep, callback_kwargs):
             if shared.stop_everything:
                 break
 
+            if batch_idx > 0:
+                clear_torch_cache()
+
             generator.manual_seed(int(seed + batch_idx))
 
             # Generate prompt variation if enabled

From b758059e950c2442cf6750082a1914e1c20b9e10 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 7 Dec 2025 12:23:19 -0800
Subject: [PATCH 1137/1701] Revert "Clear the torch cache between sequential
 image generations"

This reverts commit 1ec9f708e5cbb7a2d74fa35f56735d80991163c2.
---
 modules/ui_image_generation.py | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index bcc5bb2d01..6b39c5b5a0 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -848,9 +848,6 @@ def interrupt_callback(pipe, step_index, timestep, callback_kwargs):
             if shared.stop_everything:
                 break
 
-            if batch_idx > 0:
-                clear_torch_cache()
-
             generator.manual_seed(int(seed + batch_idx))
 
             # Generate prompt variation if enabled

From e7c8b51fecd1e33b918b73870fa62dab58dac51b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 7 Dec 2025 18:48:41 -0800
Subject: [PATCH 1138/1701] Revert "Use flash_attention_2 by default for
 Transformers models"

This reverts commit 85f2df92e9e736708cab5ba8ea9b7484036d7ded.
---
 modules/shared.py        | 2 +-
 modules/ui_model_menu.py | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index b2aeadc60d..0a27f33dc0 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -112,7 +112,7 @@
 group.add_argument('--trust-remote-code', action='store_true', help='Set trust_remote_code=True while loading the model. Necessary for some models.')
 group.add_argument('--force-safetensors', action='store_true', help='Set use_safetensors=True while loading the model. This prevents arbitrary code execution.')
 group.add_argument('--no_use_fast', action='store_true', help='Set use_fast=False while loading the tokenizer (it\'s True by default). Use this if you have any problems related to use_fast.')
-group.add_argument('--attn-implementation', type=str, default='flash_attention_2', metavar="IMPLEMENTATION", help='Attention implementation. Valid options: flash_attention_2, sdpa, eager.')
+group.add_argument('--attn-implementation', type=str, default='sdpa', metavar="IMPLEMENTATION", help='Attention implementation. Valid options: sdpa, eager, flash_attention_2.')
 
 # bitsandbytes 4-bit
 group = parser.add_argument_group('bitsandbytes 4-bit')
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index aab0fcaf07..86adc2297e 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -44,7 +44,7 @@ def create_ui():
                             shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. Common values: 4096, 8192, 16384, 32768, 65536, 131072.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
-                            shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['flash_attention_2', 'sdpa', 'eager'], value=shared.args.attn_implementation, info='Attention implementation.')
+                            shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
                             shared.gradio['tp_backend'] = gr.Dropdown(label="tp-backend", choices=['native', 'nccl'], value=shared.args.tp_backend, info='The backend for tensor parallelism.')
 

From 502f59d39babd020a83bba03b2daaa681e9b6bf8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 8 Dec 2025 05:07:57 -0800
Subject: [PATCH 1139/1701] Update diffusers to 0.36

---
 requirements/full/requirements.txt                 | 4 +---
 requirements/full/requirements_amd.txt             | 4 +---
 requirements/full/requirements_amd_noavx2.txt      | 4 +---
 requirements/full/requirements_apple_intel.txt     | 4 +---
 requirements/full/requirements_apple_silicon.txt   | 4 +---
 requirements/full/requirements_cpu_only.txt        | 4 +---
 requirements/full/requirements_cpu_only_noavx2.txt | 4 +---
 requirements/full/requirements_noavx2.txt          | 4 +---
 requirements/full/requirements_nowheels.txt        | 4 +---
 9 files changed, 9 insertions(+), 27 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 8bfa1318ff..64910090df 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -3,6 +3,7 @@ audioop-lts<1.0; python_version >= "3.13"
 bitsandbytes==0.48.*
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 flash-linear-attention==0.4.0
@@ -35,9 +36,6 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 088f7713d5..5a3d417c86 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -2,6 +2,7 @@ accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
@@ -33,9 +34,6 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index e4e55e4dd4..4ff947c1e7 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -2,6 +2,7 @@ accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
@@ -33,9 +34,6 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 5903fe6081..47cacfeb8b 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -2,6 +2,7 @@ accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
@@ -33,9 +34,6 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 4751757880..51d4aa4f0b 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -2,6 +2,7 @@ accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
@@ -33,9 +34,6 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 1fee7121a9..0b465964a4 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -2,6 +2,7 @@ accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
@@ -33,9 +34,6 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 204af32a7d..50363814d1 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -2,6 +2,7 @@ accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
@@ -33,9 +34,6 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 86208b9f54..0c6f625544 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -3,6 +3,7 @@ audioop-lts<1.0; python_version >= "3.13"
 bitsandbytes==0.48.*
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 flash-linear-attention==0.4.0
@@ -35,9 +36,6 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index ac24f0cbc9..ee8fc045a5 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -2,6 +2,7 @@ accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
@@ -33,9 +34,6 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5

From c4ebab9b29e4df06e61e526f3328d24b0fee2fc8 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Mon, 8 Dec 2025 17:56:07 -0300
Subject: [PATCH 1140/1701] Bump triton-windows in /requirements/full (#7346)

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 64910090df..482406c909 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -28,7 +28,7 @@ sentencepiece
 tensorboard
 torchao==0.14.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 5a3d417c86..e8b6c97198 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -26,7 +26,7 @@ sentencepiece
 tensorboard
 torchao==0.14.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 4ff947c1e7..add4c2b849 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -26,7 +26,7 @@ sentencepiece
 tensorboard
 torchao==0.14.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 47cacfeb8b..3f0c6f8d2e 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -26,7 +26,7 @@ sentencepiece
 tensorboard
 torchao==0.14.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 51d4aa4f0b..dbebc6d4e7 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -26,7 +26,7 @@ sentencepiece
 tensorboard
 torchao==0.14.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 0b465964a4..436e2edef4 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -26,7 +26,7 @@ sentencepiece
 tensorboard
 torchao==0.14.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 50363814d1..cc77f76226 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -26,7 +26,7 @@ sentencepiece
 tensorboard
 torchao==0.14.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 0c6f625544..02ac4b1713 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -28,7 +28,7 @@ sentencepiece
 tensorboard
 torchao==0.14.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index ee8fc045a5..bfb8934f60 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -26,7 +26,7 @@ sentencepiece
 tensorboard
 torchao==0.14.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 

From 85269d7fbbb02633692727a5f7b228067ba507eb Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Mon, 8 Dec 2025 17:58:27 -0300
Subject: [PATCH 1141/1701] Update safetensors requirement in
 /requirements/full (#7323)

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 482406c909..bde3b6a343 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -22,7 +22,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index e8b6c97198..716a240464 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -20,7 +20,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index add4c2b849..5c33549357 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -20,7 +20,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 3f0c6f8d2e..3aa4a6f81f 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -20,7 +20,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index dbebc6d4e7..400050b6d5 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -20,7 +20,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 436e2edef4..8472241636 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -20,7 +20,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index cc77f76226..0dbc84657b 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -20,7 +20,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 02ac4b1713..0691331b74 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -22,7 +22,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index bfb8934f60..070e5fd001 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -20,7 +20,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard

From aa16266c389738e3116f5b04818b61404f7d43bd Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 9 Dec 2025 03:19:23 -0800
Subject: [PATCH 1142/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 4 ++--
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 4 ++--
 requirements/portable/requirements_amd_noavx2.txt      | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 4 ++--
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 36 insertions(+), 36 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 64910090df..64273f89f3 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,8 +42,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 5a3d417c86..756fc4d6eb 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -40,7 +40,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 4ff947c1e7..48be84e9d1 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -40,7 +40,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 47cacfeb8b..41e1d8e5c4 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 51d4aa4f0b..4bd24475cc 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 0b465964a4..7c308e68bc 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 50363814d1..755d57aa7c 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 0c6f625544..ae60939387 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -42,8 +42,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 4a92fb8d8d..1f9313edd4 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 5ad1fe3a76..9d6f10cff1 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index 9c3e285116..34340e19d4 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 8a618e3551..f95feb273a 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index c1c643e8f5..93887a69e9 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index f4bc8caba7..f6a17a3649 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 81f9bf2ec4..33f9d3a00b 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 6cc87a4ea2..565eb26931 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 25c3638fb4..01cd161a2c 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 54d0cd1a19..ed3016eab0 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 9fe40ff90f2ef5a4dc9f0dff0698376e2b3d4927 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 10 Dec 2025 05:37:33 -0800
Subject: [PATCH 1143/1701] Update exllamav3 to 0.0.18

---
 requirements/full/requirements.txt        | 4 ++--
 requirements/full/requirements_noavx2.txt | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 8ed9b23eea..a7a4607793 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -44,8 +44,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index d48cd0b6de..32068ea8d9 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -44,8 +44,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 6e8fb0e7b17d5dd9ed35e04ce30f4c35922c8db5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 14 Dec 2025 13:32:14 -0800
Subject: [PATCH 1144/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 4 ++--
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 4 ++--
 requirements/portable/requirements_amd_noavx2.txt      | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 4 ++--
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 36 insertions(+), 36 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index a7a4607793..21336c556c 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,8 +42,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 6541571a6c..92e8d66f46 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -40,7 +40,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 01b240ee88..baf7e66b31 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -40,7 +40,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index dc0479a8b1..31d7c576ca 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 82de929f20..cd0f944e8e 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 3a323aa1a9..5f591edf4a 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 8e7a59b926..74f5d3f8e4 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 32068ea8d9..b013150a1e 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -42,8 +42,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 1f9313edd4..fc7560d939 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 9d6f10cff1..5b64e2a910 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index 34340e19d4..6ec4a89a08 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index f95feb273a..78ad17da00 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 93887a69e9..d8a41eccce 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index f6a17a3649..68fe0ee7ce 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 33f9d3a00b..a8a5b87d44 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 565eb26931..ccf57891ad 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 01cd161a2c..5fb6bb701b 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index ed3016eab0..b99bd55868 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.67.0/llama_cpp_binaries-0.67.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 09d88f91e878eb9273215f572e71eebdd3be6e3d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 19 Dec 2025 21:00:13 -0800
Subject: [PATCH 1145/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 4 ++--
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 4 ++--
 requirements/portable/requirements_amd_noavx2.txt      | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 4 ++--
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 36 insertions(+), 36 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 21336c556c..79292a9f2a 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,8 +42,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 92e8d66f46..b6b7850d43 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -40,7 +40,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index baf7e66b31..e1b0823cf5 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -40,7 +40,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 31d7c576ca..08abd11192 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index cd0f944e8e..6cc0c40e14 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 5f591edf4a..7b8513076a 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 74f5d3f8e4..6f7b5e15fa 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index b013150a1e..7f9e116278 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -42,8 +42,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index fc7560d939..91a2c60b3e 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 5b64e2a910..b8d55d3ae3 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index 6ec4a89a08..296e8b2e2d 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 78ad17da00..50fd898dbc 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index d8a41eccce..e7268b09c0 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 68fe0ee7ce..b7d99339d0 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index a8a5b87d44..043d9db516 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index ccf57891ad..341f8f45f1 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 5fb6bb701b..6a69a824ae 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index b99bd55868..2df6df0169 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.70.0/llama_cpp_binaries-0.70.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 9530d3a6d8ad24d68bf35fcf396a8c9edeeb6df3 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Sun, 21 Dec 2025 15:30:54 -0300
Subject: [PATCH 1146/1701] UI: Improve hr (horizontal separator) style

---
 css/html_instruct_style.css | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index 6dee0a8980..5a0338a55a 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -27,6 +27,10 @@
     border-color: #ffffff75 !important;
 }
 
+.dark .prose hr {
+    border-color: #ffffff75 !important;
+}
+
 .chat .message-body :is(p, ul, ol) {
     margin: 1.25em 0 !important;
 }

From 1066fe8c2110a8a366c7bdf945f0489bd8d1b113 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Sun, 21 Dec 2025 15:32:02 -0300
Subject: [PATCH 1147/1701] UI: Improve table styles (more minimalistic)

---
 css/main.css | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/css/main.css b/css/main.css
index e758696090..376cfa6d6b 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1797,3 +1797,20 @@ button#swap-height-width {
     top: 0;
     left: calc(100% - 174px);
 }
+
+table {
+  border-collapse: collapse;
+}
+
+table, tr, td, th, thead {
+  border: 0;
+}
+
+td + td,
+th + th { border-left: 1px solid; }
+
+tr + tr td,
+tr + tr th { border-top: 1px solid; }
+
+thead + tbody tr:first-child td,
+thead + tbody tr:first-child th { border-top: 1px solid; }

From a2ed640aa6271251520acbe4831fd860f5f44f1c Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Sun, 21 Dec 2025 15:38:48 -0300
Subject: [PATCH 1148/1701] UI: Improved border color for tables + hr

---
 css/html_instruct_style.css | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index 5a0338a55a..a7532cf626 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -19,16 +19,14 @@
     color: #d1d5db !important;
 }
 
-.chat .message-body :is(th, td) {
+.chat .message-body :is(th, td),
+.prose hr {
     border-color: #40404096 !important;
 }
 
-.dark .chat .message-body :is(th, td) {
-    border-color: #ffffff75 !important;
-}
-
+.dark .chat .message-body :is(th, td),
 .dark .prose hr {
-    border-color: #ffffff75 !important;
+    border-color: rgba(255, 255, 255, 0.3) !important;
 }
 
 .chat .message-body :is(p, ul, ol) {

From 6e2c4e9c23fbeb71fbbf4b97be6968351cd9ab7d Mon Sep 17 00:00:00 2001
From: Sergey 'Jin' Bostandzhyan 
Date: Tue, 6 Jan 2026 15:31:10 +0100
Subject: [PATCH 1149/1701] Fix loading models which have their eos token
 disabled (#7363)

---
 modules/models_settings.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index e9f19a0603..15ff2830ae 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -83,7 +83,11 @@ def get_model_metadata(model):
 
         if 'tokenizer.chat_template' in metadata:
             template = metadata['tokenizer.chat_template']
-            eos_token = metadata['tokenizer.ggml.tokens'][metadata['tokenizer.ggml.eos_token_id']]
+            if 'tokenizer.ggml.eos_token_id' in metadata:
+                eos_token = metadata['tokenizer.ggml.tokens'][metadata['tokenizer.ggml.eos_token_id']]
+            else:
+                eos_token = ""
+
             if 'tokenizer.ggml.bos_token_id' in metadata:
                 bos_token = metadata['tokenizer.ggml.tokens'][metadata['tokenizer.ggml.bos_token_id']]
             else:

From bb3b7bc1979b90f3de2a5b8968c816adf7458856 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 6 Jan 2026 10:23:58 -0800
Subject: [PATCH 1150/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 4 ++--
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 4 ++--
 requirements/portable/requirements_amd_noavx2.txt      | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 4 ++--
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 36 insertions(+), 36 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 79292a9f2a..a7d3403f47 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,8 +42,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index b6b7850d43..94367afa38 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -40,7 +40,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index e1b0823cf5..105ac7da3b 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -40,7 +40,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 08abd11192..8ffad94768 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 6cc0c40e14..b6f088c2d5 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 7b8513076a..63c63b044e 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 6f7b5e15fa..198cad4cf7 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 7f9e116278..b0da97c6f4 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -42,8 +42,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 91a2c60b3e..2007e96843 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index b8d55d3ae3..b8dd5a2dc4 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index 296e8b2e2d..fc26d632c6 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 50fd898dbc..7fab39b4a3 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index e7268b09c0..24658d3e5a 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index b7d99339d0..8434463dd7 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 043d9db516..c0b8c610f3 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 341f8f45f1..27817fda47 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 6a69a824ae..298d9815ed 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 2df6df0169..f7b78b309f 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.71.0/llama_cpp_binaries-0.71.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 36747cf99ca1987f56dea2314d8640f311124339 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 6 Jan 2026 10:24:34 -0800
Subject: [PATCH 1151/1701] Lint

---
 css/html_instruct_style.css | 2 +-
 css/main.css                | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index a7532cf626..72a148c3a9 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -26,7 +26,7 @@
 
 .dark .chat .message-body :is(th, td),
 .dark .prose hr {
-    border-color: rgba(255, 255, 255, 0.3) !important;
+    border-color: rgb(255 255 255 / 30%) !important;
 }
 
 .chat .message-body :is(p, ul, ol) {
diff --git a/css/main.css b/css/main.css
index 376cfa6d6b..ecf8568f0b 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1799,11 +1799,11 @@ button#swap-height-width {
 }
 
 table {
-  border-collapse: collapse;
+    border-collapse: collapse;
 }
 
 table, tr, td, th, thead {
-  border: 0;
+    border: 0;
 }
 
 td + td,

From b0968ed8b493b47d030cf5eede95a0c226b05b2d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 6 Jan 2026 10:26:43 -0800
Subject: [PATCH 1152/1701] Update flash-linear-attention

---
 requirements/full/requirements.txt        | 2 +-
 requirements/full/requirements_noavx2.txt | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index a7d3403f47..edc4f32bfe 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -6,7 +6,7 @@ datasets
 diffusers==0.36.*
 einops
 fastapi==0.112.4
-flash-linear-attention==0.4.0
+flash-linear-attention==0.4.*
 html2text==2025.4.15
 huggingface-hub==0.36.0
 jinja2==3.1.6
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index b0da97c6f4..49b7041f97 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -6,7 +6,7 @@ datasets
 diffusers==0.36.*
 einops
 fastapi==0.112.4
-flash-linear-attention==0.4.0
+flash-linear-attention==0.4.*
 html2text==2025.4.15
 huggingface-hub==0.36.0
 jinja2==3.1.6

From 45fbec03202c642b01ac71e5b57e855f498bc727 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Tue, 6 Jan 2026 15:27:10 -0300
Subject: [PATCH 1153/1701] Update torchao requirement in /requirements/full
 (#7356)

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index edc4f32bfe..43c71aa926 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -26,7 +26,7 @@ safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
 triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 94367afa38..e5d904e150 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -24,7 +24,7 @@ safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
 triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 105ac7da3b..75094d2faa 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -24,7 +24,7 @@ safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
 triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 8ffad94768..751c38d2cc 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -24,7 +24,7 @@ safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
 triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index b6f088c2d5..6cf2731d01 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -24,7 +24,7 @@ safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
 triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 63c63b044e..c02d625039 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -24,7 +24,7 @@ safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
 triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 198cad4cf7..88464eab90 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -24,7 +24,7 @@ safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
 triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 49b7041f97..d842f85cc2 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -26,7 +26,7 @@ safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
 triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 070e5fd001..d8d7a92783 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -24,7 +24,7 @@ safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
 triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm

From 50a35b483c30fb91664877a0aa031228b8e8f3d3 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Tue, 6 Jan 2026 15:27:23 -0300
Subject: [PATCH 1154/1701] Update bitsandbytes requirement in
 /requirements/full (#7353)

---
 requirements/full/requirements.txt        | 2 +-
 requirements/full/requirements_noavx2.txt | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 43c71aa926..b97ccd622a 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -1,6 +1,6 @@
 accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
-bitsandbytes==0.48.*
+bitsandbytes==0.49.*
 colorama
 datasets
 diffusers==0.36.*
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index d842f85cc2..34492c6490 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -1,6 +1,6 @@
 accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
-bitsandbytes==0.48.*
+bitsandbytes==0.49.*
 colorama
 datasets
 diffusers==0.36.*

From 332fd40653b57a3a515143f5684620c49075e837 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 7 Jan 2026 19:06:23 -0800
Subject: [PATCH 1155/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 4 ++--
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 4 ++--
 requirements/portable/requirements_amd_noavx2.txt      | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 4 ++--
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 36 insertions(+), 36 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index b97ccd622a..2ee917c68e 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,8 +42,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index e5d904e150..bd3a9c0bc8 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -40,7 +40,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 75094d2faa..b20112ec08 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -40,7 +40,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 751c38d2cc..f4e94577e4 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 6cf2731d01..710d055c82 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index c02d625039..890486bfe6 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 88464eab90..d50bfb4792 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 34492c6490..88a1f3e49d 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -42,8 +42,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 2007e96843..3da1bbdd43 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index b8dd5a2dc4..10c3fb33e8 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index fc26d632c6..c3c16abcd4 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 7fab39b4a3..adf154c58d 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 24658d3e5a..cbd9e999f7 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 8434463dd7..0f39c2e15d 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index c0b8c610f3..64b0688dc7 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index 27817fda47..b72757ccf8 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 298d9815ed..8cc9d77651 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index f7b78b309f..25ec67cdf7 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.72.0/llama_cpp_binaries-0.72.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From d79cdc614cb93e4687abf16eafda24c0351cbbf0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 8 Jan 2026 11:24:15 -0800
Subject: [PATCH 1156/1701] Update llama.cpp

---
 requirements/full/requirements.txt                     | 4 ++--
 requirements/full/requirements_amd.txt                 | 4 ++--
 requirements/full/requirements_amd_noavx2.txt          | 4 ++--
 requirements/full/requirements_apple_intel.txt         | 4 ++--
 requirements/full/requirements_apple_silicon.txt       | 4 ++--
 requirements/full/requirements_cpu_only.txt            | 4 ++--
 requirements/full/requirements_cpu_only_noavx2.txt     | 4 ++--
 requirements/full/requirements_noavx2.txt              | 4 ++--
 requirements/portable/requirements.txt                 | 4 ++--
 requirements/portable/requirements_amd.txt             | 4 ++--
 requirements/portable/requirements_amd_noavx2.txt      | 4 ++--
 requirements/portable/requirements_apple_intel.txt     | 4 ++--
 requirements/portable/requirements_apple_silicon.txt   | 4 ++--
 requirements/portable/requirements_cpu_only.txt        | 4 ++--
 requirements/portable/requirements_cpu_only_noavx2.txt | 4 ++--
 requirements/portable/requirements_noavx2.txt          | 4 ++--
 requirements/portable/requirements_vulkan.txt          | 4 ++--
 requirements/portable/requirements_vulkan_noavx2.txt   | 4 ++--
 18 files changed, 36 insertions(+), 36 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 2ee917c68e..1bc9ad1b7b 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,8 +42,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index bd3a9c0bc8..9b0fa12ee2 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -40,7 +40,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index b20112ec08..60a5415781 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -40,7 +40,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index f4e94577e4..c97928f9d3 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 710d055c82..c671c4d9c1 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 890486bfe6..1a14d0aeaf 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index d50bfb4792..2a9484bfdd 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -40,5 +40,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 88a1f3e49d..1e75b7919a 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -42,8 +42,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 3da1bbdd43..2f6bf51142 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 10c3fb33e8..209610ceb9 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index c3c16abcd4..29a1f72b04 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index adf154c58d..8812ac39f5 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index cbd9e999f7..c627cfd456 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 0f39c2e15d..5ca660adfc 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 64b0688dc7..5bb282eb18 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index b72757ccf8..ecce4a2f7d 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 8cc9d77651..4b3adddc3d 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 25ec67cdf7..8d6acdd78e 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.73.0/llama_cpp_binaries-0.73.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From a73186112761aa13671919cef7d25451e80097d0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 13 Jan 2026 15:38:32 -0800
Subject: [PATCH 1157/1701] Update README

---
 README.md | 25 +++++++++++--------------
 1 file changed, 11 insertions(+), 14 deletions(-)

diff --git a/README.md b/README.md
index 8065ca71e9..4cc13b88c2 100644
--- a/README.md
+++ b/README.md
@@ -13,7 +13,7 @@
 
 # Text Generation Web UI
 
-A Gradio web UI for Large Language Models.
+Run AI chatbots like ChatGPT on your own computer. **100% private and offline** – no subscriptions, no API fees, zero telemetry. Just download, unzip, and run.
 
 [Try the Deep Reason extension](https://oobabooga.gumroad.com/l/deep_reason)
 
@@ -21,38 +21,35 @@ A Gradio web UI for Large Language Models.
 |:---:|:---:|
 |![Image1](https://github.com/oobabooga/screenshots/raw/main/DEFAULT-3.5.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/PARAMETERS-3.5.png) |
 
-## 🔥 News
-
-- The project now supports **image generation**! Including Z-Image-Turbo, 4bit/8bit quantization, `torch.compile`, and LLM-generated prompt variations ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
-
 ## Features
 
-- Supports multiple local text generation backends, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).
-- Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory.
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
 - **Vision (multimodal models)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
 - **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
 - **Web search**: Optionally search the internet with LLM-generated queries to add context to the conversation.
-- Aesthetic UI with dark and light themes.
-- Syntax highlighting for code blocks and LaTeX rendering for mathematical expressions.
+- Aesthetic UI with dark/light themes, syntax highlighting, and LaTeX rendering.
+- Edit messages, navigate between message versions, and branch conversations at any point.
+- Switch between models without restarting, with automatic GPU layer allocation.
+- Free-form text generation in the Notebook tab without being limited to chat turns.
 - `instruct` mode for instruction-following (like ChatGPT), and `chat-instruct`/`chat` modes for talking to custom characters.
 - Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
-- Edit messages, navigate between message versions, and branch conversations at any point.
 - Multiple sampling parameters and generation options for sophisticated text generation control.
-- Switch between different models in the UI without restarting.
-- Automatic GPU layers for GGUF models (on NVIDIA GPUs).
-- Free-form text generation in the Notebook tab without being limited to chat turns.
+- Supports multiple backends including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM).
 - OpenAI-compatible API with Chat and Completions endpoints, including tool-calling support – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples).
 - Extension support, with numerous built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details.
 
+## 🔥 News
+
+- The project now supports **image generation**! Including Z-Image-Turbo, 4bit/8bit quantization, `torch.compile`, and LLM-generated prompt variations ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
+
 ## How to install
 
 #### ✅ Option 1: Portable builds (get started in 1 minute)
 
 No installation needed – just download, unzip and run. All dependencies included.
 
-Compatible with GGUF (llama.cpp) models on Windows, Linux, and macOS.
+Compatible with GGUF (llama.cpp) models on Windows, Linux, and macOS. [Check what models fit your hardware](https://huggingface.co/spaces/oobabooga/accurate-gguf-vram-calculator).
 
 Download from here: **https://github.com/oobabooga/text-generation-webui/releases**
 

From 21b979c02ad660951e1c51f991b18114e683dd4e Mon Sep 17 00:00:00 2001
From: jakubartur <90887830+jakubartur@users.noreply.github.com>
Date: Wed, 14 Jan 2026 23:34:21 +0100
Subject: [PATCH 1158/1701] Fix code block copy button on HTTP (Clipboard API
 fallback) (#7358)

---
 js/highlightjs/highlightjs-copy.min.js | 85 +++++++++++++++++++++++++-
 1 file changed, 84 insertions(+), 1 deletion(-)

diff --git a/js/highlightjs/highlightjs-copy.min.js b/js/highlightjs/highlightjs-copy.min.js
index b1c0d04162..56d185d5a6 100644
--- a/js/highlightjs/highlightjs-copy.min.js
+++ b/js/highlightjs/highlightjs-copy.min.js
@@ -1 +1,84 @@
-class CopyButtonPlugin{constructor(options={}){self.hook=options.hook;self.callback=options.callback;self.lang=options.lang||document.documentElement.lang||"en"}"after:highlightElement"({el,text}){let button=Object.assign(document.createElement("button"),{innerHTML:locales[lang]?.[0]||"Copy",className:"hljs-copy-button"});button.dataset.copied=false;el.parentElement.classList.add("hljs-copy-wrapper");el.parentElement.appendChild(button);el.parentElement.style.setProperty("--hljs-theme-background",window.getComputedStyle(el).backgroundColor);button.onclick=function(){if(!navigator.clipboard)return;let newText=text;if(hook&&typeof hook==="function"){newText=hook(text,el)||text}navigator.clipboard.writeText(newText).then(function(){button.innerHTML=locales[lang]?.[1]||"Copied!";button.dataset.copied=true;let alert=Object.assign(document.createElement("div"),{role:"status",className:"hljs-copy-alert",innerHTML:locales[lang]?.[2]||"Copied to clipboard"});el.parentElement.appendChild(alert);setTimeout(()=>{button.innerHTML=locales[lang]?.[0]||"Copy";button.dataset.copied=false;el.parentElement.removeChild(alert);alert=null},2e3)}).then(function(){if(typeof callback==="function")return callback(newText,el)})}}}if(typeof module!="undefined"){module.exports=CopyButtonPlugin}const locales={en:["Copy","Copied!","Copied to clipboard"],es:["Copiar","¡Copiado!","Copiado al portapapeles"],fr:["Copier","Copié !","Copié dans le presse-papier"],de:["Kopieren","Kopiert!","In die Zwischenablage kopiert"],ja:["コピー","コピーしました！","クリップボードにコピーしました"],ko:["복사","복사됨!","클립보드에 복사됨"],ru:["Копировать","Скопировано!","Скопировано в буфер обмена"],zh:["复制","已复制!","已复制到剪贴板"],"zh-tw":["複製","已複製!","已複製到剪貼簿"]};
\ No newline at end of file
+function fallbackCopyToClipboard(text) {
+  return new Promise((resolve, reject) => {
+    const textArea = document.createElement("textarea");
+    textArea.value = text;
+    textArea.style.position = "fixed";
+    textArea.style.left = "-9999px";
+    textArea.style.top = "-9999px";
+    document.body.appendChild(textArea);
+    textArea.focus();
+    textArea.select();
+    try {
+      const successful = document.execCommand("copy");
+      document.body.removeChild(textArea);
+      successful ? resolve() : reject();
+    } catch (err) {
+      document.body.removeChild(textArea);
+      reject(err);
+    }
+  });
+}
+
+class CopyButtonPlugin {
+  constructor(options = {}) {
+    self.hook = options.hook;
+    self.callback = options.callback;
+    self.lang = options.lang || document.documentElement.lang || "en";
+  }
+  "after:highlightElement"({ el, text }) {
+    let button = Object.assign(document.createElement("button"), {
+      innerHTML: locales[lang]?.[0] || "Copy",
+      className: "hljs-copy-button",
+    });
+    button.dataset.copied = false;
+    el.parentElement.classList.add("hljs-copy-wrapper");
+    el.parentElement.appendChild(button);
+    el.parentElement.style.setProperty(
+      "--hljs-theme-background",
+      window.getComputedStyle(el).backgroundColor,
+    );
+    button.onclick = function () {
+      let newText = text;
+      if (hook && typeof hook === "function") {
+        newText = hook(text, el) || text;
+      }
+      const copyPromise =
+        navigator.clipboard && window.isSecureContext
+          ? navigator.clipboard.writeText(newText)
+          : fallbackCopyToClipboard(newText);
+      copyPromise.then(function () {
+          button.innerHTML = locales[lang]?.[1] || "Copied!";
+          button.dataset.copied = true;
+          let alert = Object.assign(document.createElement("div"), {
+            role: "status",
+            className: "hljs-copy-alert",
+            innerHTML: locales[lang]?.[2] || "Copied to clipboard",
+          });
+          el.parentElement.appendChild(alert);
+          setTimeout(() => {
+            button.innerHTML = locales[lang]?.[0] || "Copy";
+            button.dataset.copied = false;
+            el.parentElement.removeChild(alert);
+            alert = null;
+          }, 2e3);
+        })
+        .then(function () {
+          if (typeof callback === "function") return callback(newText, el);
+        });
+    };
+  }
+}
+if (typeof module != "undefined") {
+  module.exports = CopyButtonPlugin;
+}
+const locales = {
+  en: ["Copy", "Copied!", "Copied to clipboard"],
+  es: ["Copiar", "¡Copiado!", "Copiado al portapapeles"],
+  fr: ["Copier", "Copié !", "Copié dans le presse-papier"],
+  de: ["Kopieren", "Kopiert!", "In die Zwischenablage kopiert"],
+  ja: ["コピー", "コピーしました！", "クリップボードにコピーしました"],
+  ko: ["복사", "복사됨!", "클립보드에 복사됨"],
+  ru: ["Копировать", "Скопировано!", "Скопировано в буфер обмена"],
+  zh: ["复制", "已复制!", "已复制到剪贴板"],
+  "zh-tw": ["複製", "已複製!", "已複製到剪貼簿"],
+};

From 7493fe7841c2dcf19f5f42bd60b148c7d9e0d70d Mon Sep 17 00:00:00 2001
From: "q5sys (JT)" 
Date: Wed, 14 Jan 2026 18:35:08 -0500
Subject: [PATCH 1159/1701] feat: Add a dropdown to save/load user personas 
 (#7367)

---
 css/main.css                 |   2 +-
 modules/chat.py              | 151 ++++++++++++++++++++++++++++++++++-
 modules/shared.py            |   1 +
 modules/ui.py                |   4 +
 modules/ui_chat.py           |  14 ++++
 modules/ui_file_saving.py    |  49 ++++++++++++
 modules/utils.py             |   7 ++
 user_data/users/Default.yaml |   2 +
 8 files changed, 228 insertions(+), 2 deletions(-)
 create mode 100644 user_data/users/Default.yaml

diff --git a/css/main.css b/css/main.css
index ecf8568f0b..c55a29dcf2 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1645,7 +1645,7 @@ button:focus {
 }
 
 #user-description textarea {
-    height: calc(100vh - 231px) !important;
+    height: calc(100vh - 334px) !important;
     min-height: 90px !important;
 }
 
diff --git a/modules/chat.py b/modules/chat.py
index d1474cfec0..42c0d46d5a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -32,7 +32,12 @@
     get_encoded_length,
     get_max_prompt_length
 )
-from modules.utils import delete_file, get_available_characters, save_file
+from modules.utils import (
+    delete_file,
+    get_available_characters,
+    get_available_users,
+    save_file
+)
 from modules.web_search import add_web_search_attachments
 
 
@@ -1647,6 +1652,150 @@ def delete_character(name, instruct=False):
         delete_file(Path(f'user_data/characters/{name}.{extension}'))
 
 
+def generate_user_pfp_cache(user):
+    """Generate cached profile picture for user"""
+    cache_folder = Path(shared.args.disk_cache_dir)
+    if not cache_folder.exists():
+        cache_folder.mkdir()
+
+    for path in [Path(f"user_data/users/{user}.{extension}") for extension in ['png', 'jpg', 'jpeg']]:
+        if path.exists():
+            original_img = Image.open(path)
+            # Define file paths
+            pfp_path = Path(f'{cache_folder}/pfp_me.png')
+
+            # Save thumbnail
+            thumb = make_thumbnail(original_img)
+            thumb.save(pfp_path, format='PNG')
+            logger.info(f'User profile picture cached to "{pfp_path}"')
+
+            return str(pfp_path)
+
+    return None
+
+
+def load_user(user_name, name1, user_bio):
+    """Load user profile from YAML file"""
+    picture = None
+
+    filepath = None
+    for extension in ["yml", "yaml", "json"]:
+        filepath = Path(f'user_data/users/{user_name}.{extension}')
+        if filepath.exists():
+            break
+
+    if filepath is None or not filepath.exists():
+        logger.error(f"Could not find the user \"{user_name}\" inside user_data/users. No user has been loaded.")
+        raise ValueError
+
+    with open(filepath, 'r', encoding='utf-8') as f:
+        file_contents = f.read()
+
+    extension = filepath.suffix[1:]  # Remove the leading dot
+    data = json.loads(file_contents) if extension == "json" else yaml.safe_load(file_contents)
+
+    # Clear existing user picture cache
+    cache_folder = Path(shared.args.disk_cache_dir)
+    pfp_path = Path(f"{cache_folder}/pfp_me.png")
+    if pfp_path.exists():
+        pfp_path.unlink()
+
+    # Generate new picture cache
+    picture = generate_user_pfp_cache(user_name)
+
+    # Get user name
+    if 'name' in data and data['name'] != '':
+        name1 = data['name']
+
+    # Get user bio
+    if 'user_bio' in data:
+        user_bio = data['user_bio']
+
+    return name1, user_bio, picture
+
+
+def generate_user_yaml(name, user_bio):
+    """Generate YAML content for user profile"""
+    data = {
+        'name': name,
+        'user_bio': user_bio,
+    }
+
+    return yaml.dump(data, sort_keys=False, width=float("inf"))
+
+
+def save_user(name, user_bio, picture, filename):
+    """Save user profile to YAML file"""
+    if filename == "":
+        logger.error("The filename is empty, so the user will not be saved.")
+        return
+
+    # Ensure the users directory exists
+    users_dir = Path('user_data/users')
+    users_dir.mkdir(parents=True, exist_ok=True)
+
+    data = generate_user_yaml(name, user_bio)
+    filepath = Path(f'user_data/users/{filename}.yaml')
+    save_file(filepath, data)
+
+    path_to_img = Path(f'user_data/users/{filename}.png')
+    if picture is not None:
+        # Copy the image file from its source path to the users folder
+        shutil.copy(picture, path_to_img)
+        logger.info(f'Saved user profile picture to {path_to_img}.')
+
+
+def delete_user(name):
+    """Delete user profile files"""
+    # Check for user data files
+    for extension in ["yml", "yaml", "json"]:
+        delete_file(Path(f'user_data/users/{name}.{extension}'))
+
+    # Check for user image files
+    for extension in ["png", "jpg", "jpeg"]:
+        delete_file(Path(f'user_data/users/{name}.{extension}'))
+
+
+def update_user_menu_after_deletion(idx):
+    """Update user menu after a user is deleted"""
+    users = get_available_users()
+    if len(users) == 0:
+        # Create a default user if none exist
+        save_user('You', '', None, 'Default')
+        users = get_available_users()
+
+    idx = min(int(idx), len(users) - 1)
+    idx = max(0, idx)
+    return gr.update(choices=users, value=users[idx])
+
+
+def handle_user_menu_change(state):
+    """Handle user menu selection change"""
+    try:
+        name1, user_bio, picture = load_user(state['user_menu'], state['name1'], state['user_bio'])
+
+        return [
+            name1,
+            user_bio,
+            picture
+        ]
+    except Exception as e:
+        logger.error(f"Failed to load user '{state['user_menu']}': {e}")
+        return [
+            state['name1'],
+            state['user_bio'],
+            None
+        ]
+
+
+def handle_save_user_click(name1):
+    """Handle save user button click"""
+    return [
+        name1,
+        gr.update(visible=True)
+    ]
+
+
 def jinja_template_from_old_format(params, verbose=False):
     MASTER_TEMPLATE = """
 {%- set ns = namespace(found=false) -%}
diff --git a/modules/shared.py b/modules/shared.py
index 0a27f33dc0..88e4b182ad 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -298,6 +298,7 @@
 
     # Character settings
     'character': 'Assistant',
+    'user': 'Default',
     'name1': 'You',
     'name2': 'AI',
     'user_bio': '',
diff --git a/modules/ui.py b/modules/ui.py
index 919a5740b5..8a49186139 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -251,6 +251,7 @@ def list_interface_input_elements():
         'chat_style',
         'chat-instruct_command',
         'character_menu',
+        'user_menu',
         'name2',
         'context',
         'greeting',
@@ -353,6 +354,8 @@ def save_settings(state, preset, extensions_list, show_controls, theme_state, ma
     output['preset'] = preset
     output['prompt-notebook'] = state['prompt_menu-default'] if state['show_two_notebook_columns'] else state['prompt_menu-notebook']
     output['character'] = state['character_menu']
+    if 'user_menu' in state and state['user_menu']:
+        output['user'] = state['user_menu']
     output['seed'] = int(output['seed'])
     output['show_controls'] = show_controls
     output['dark_theme'] = True if theme_state == 'dark' else False
@@ -457,6 +460,7 @@ def setup_auto_save():
         'chat_style',
         'chat-instruct_command',
         'character_menu',
+        'user_menu',
         'name1',
         'name2',
         'context',
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index c342ce5b0c..e00ddf5c39 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -137,6 +137,12 @@ def create_character_settings_ui():
                     shared.gradio['greeting'] = gr.Textbox(value=shared.settings['greeting'], lines=5, label='Greeting', elem_classes=['add_scrollbar'], elem_id="character-greeting")
 
                 with gr.Tab("User"):
+                    with gr.Row():
+                        shared.gradio['user_menu'] = gr.Dropdown(value=shared.settings['user'], choices=utils.get_available_users(), label='User', elem_id='user-menu', info='Select a user profile.', elem_classes='slim-dropdown')
+                        ui.create_refresh_button(shared.gradio['user_menu'], lambda: None, lambda: {'choices': utils.get_available_users()}, 'refresh-button', interactive=not mu)
+                        shared.gradio['save_user'] = gr.Button('💾', elem_classes='refresh-button', elem_id="save-user", interactive=not mu)
+                        shared.gradio['delete_user'] = gr.Button('🗑️', elem_classes='refresh-button', interactive=not mu)
+
                     shared.gradio['name1'] = gr.Textbox(value=shared.settings['name1'], lines=1, label='Name')
                     shared.gradio['user_bio'] = gr.Textbox(value=shared.settings['user_bio'], lines=10, label='Description', info='Here you can optionally write a description of yourself.', placeholder='{{user}}\'s personality: ...', elem_classes=['add_scrollbar'], elem_id="user-description")
 
@@ -372,3 +378,11 @@ def create_event_handlers():
         gradio('enable_web_search'),
         gradio('web_search_row')
     )
+
+    # User menu event handlers
+    shared.gradio['user_menu'].change(
+        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
+        chat.handle_user_menu_change, gradio('interface_state'), gradio('name1', 'user_bio', 'your_picture'), show_progress=False)
+
+    shared.gradio['save_user'].click(chat.handle_save_user_click, gradio('name1'), gradio('save_user_filename', 'user_saver'), show_progress=False)
+    shared.gradio['delete_user'].click(lambda: gr.update(visible=True), None, gradio('user_deleter'), show_progress=False)
diff --git a/modules/ui_file_saving.py b/modules/ui_file_saving.py
index d1f9379b5a..720bfdecf1 100644
--- a/modules/ui_file_saving.py
+++ b/modules/ui_file_saving.py
@@ -39,6 +39,19 @@ def create_ui():
             shared.gradio['delete_character_cancel'] = gr.Button('Cancel', elem_classes="small-button")
             shared.gradio['delete_character_confirm'] = gr.Button('Delete', elem_classes="small-button", variant='stop', interactive=not mu)
 
+    # User saver/deleter
+    with gr.Group(visible=False, elem_classes='file-saver') as shared.gradio['user_saver']:
+        shared.gradio['save_user_filename'] = gr.Textbox(lines=1, label='File name', info='The user profile will be saved to your user_data/users folder with this base filename.')
+        with gr.Row():
+            shared.gradio['save_user_cancel'] = gr.Button('Cancel', elem_classes="small-button")
+            shared.gradio['save_user_confirm'] = gr.Button('Save', elem_classes="small-button", variant='primary', interactive=not mu)
+
+    with gr.Group(visible=False, elem_classes='file-saver') as shared.gradio['user_deleter']:
+        gr.Markdown('Confirm the user deletion?')
+        with gr.Row():
+            shared.gradio['delete_user_cancel'] = gr.Button('Cancel', elem_classes="small-button")
+            shared.gradio['delete_user_confirm'] = gr.Button('Delete', elem_classes="small-button", variant='stop', interactive=not mu)
+
     # Preset saver
     with gr.Group(visible=False, elem_classes='file-saver') as shared.gradio['preset_saver']:
         shared.gradio['save_preset_filename'] = gr.Textbox(lines=1, label='File name', info='The preset will be saved to your user_data/presets folder with this base filename.')
@@ -69,6 +82,12 @@ def create_event_handlers():
     shared.gradio['save_character_cancel'].click(lambda: gr.update(visible=False), None, gradio('character_saver'), show_progress=False)
     shared.gradio['delete_character_cancel'].click(lambda: gr.update(visible=False), None, gradio('character_deleter'), show_progress=False)
 
+    # User save/delete event handlers
+    shared.gradio['save_user_confirm'].click(handle_save_user_confirm_click, gradio('name1', 'user_bio', 'your_picture', 'save_user_filename'), gradio('user_menu', 'user_saver'), show_progress=False)
+    shared.gradio['delete_user_confirm'].click(handle_delete_user_confirm_click, gradio('user_menu'), gradio('user_menu', 'user_deleter'), show_progress=False)
+    shared.gradio['save_user_cancel'].click(lambda: gr.update(visible=False), None, gradio('user_saver'), show_progress=False)
+    shared.gradio['delete_user_cancel'].click(lambda: gr.update(visible=False), None, gradio('user_deleter'), show_progress=False)
+
 
 def handle_save_preset_confirm_click(filename, contents):
     try:
@@ -165,3 +184,33 @@ def handle_delete_grammar_click(grammar_file):
         "user_data/grammars/",
         gr.update(visible=True)
     ]
+
+
+def handle_save_user_confirm_click(name1, user_bio, your_picture, filename):
+    try:
+        chat.save_user(name1, user_bio, your_picture, filename)
+        available_users = utils.get_available_users()
+        output = gr.update(choices=available_users, value=filename)
+    except Exception:
+        output = gr.update()
+        traceback.print_exc()
+
+    return [
+        output,
+        gr.update(visible=False)
+    ]
+
+
+def handle_delete_user_confirm_click(user):
+    try:
+        index = str(utils.get_available_users().index(user))
+        chat.delete_user(user)
+        output = chat.update_user_menu_after_deletion(index)
+    except Exception:
+        output = gr.update()
+        traceback.print_exc()
+
+    return [
+        output,
+        gr.update(visible=False)
+    ]
diff --git a/modules/utils.py b/modules/utils.py
index b478f0664e..d366784730 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -219,6 +219,13 @@ def get_available_characters():
     return sorted(set((k.stem for k in paths)), key=natural_keys)
 
 
+def get_available_users():
+    users_dir = Path('user_data/users')
+    users_dir.mkdir(parents=True, exist_ok=True)
+    paths = (x for x in users_dir.iterdir() if x.suffix in ('.json', '.yaml', '.yml'))
+    return sorted(set((k.stem for k in paths)), key=natural_keys)
+
+
 def get_available_instruction_templates():
     path = "user_data/instruction-templates"
     paths = []
diff --git a/user_data/users/Default.yaml b/user_data/users/Default.yaml
new file mode 100644
index 0000000000..5c9dbacc49
--- /dev/null
+++ b/user_data/users/Default.yaml
@@ -0,0 +1,2 @@
+name: You
+user_bio: ''

From cae1fef42d2905c2cf151b870cd29bb287d23615 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Wed, 14 Jan 2026 21:30:59 -0300
Subject: [PATCH 1160/1701] Bump triton-windows in /requirements/full (#7368)

---
 requirements/full/requirements.txt                 | 2 +-
 requirements/full/requirements_amd.txt             | 2 +-
 requirements/full/requirements_amd_noavx2.txt      | 2 +-
 requirements/full/requirements_apple_intel.txt     | 2 +-
 requirements/full/requirements_apple_silicon.txt   | 2 +-
 requirements/full/requirements_cpu_only.txt        | 2 +-
 requirements/full/requirements_cpu_only_noavx2.txt | 2 +-
 requirements/full/requirements_noavx2.txt          | 2 +-
 requirements/full/requirements_nowheels.txt        | 2 +-
 9 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 1bc9ad1b7b..dd74ed52e9 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -28,7 +28,7 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post22; platform_system == "Windows"
+triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 9b0fa12ee2..3475c161ac 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -26,7 +26,7 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post22; platform_system == "Windows"
+triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 60a5415781..05077a32a6 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -26,7 +26,7 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post22; platform_system == "Windows"
+triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index c97928f9d3..f591fa251e 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -26,7 +26,7 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post22; platform_system == "Windows"
+triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index c671c4d9c1..6ff4753d4b 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -26,7 +26,7 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post22; platform_system == "Windows"
+triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 1a14d0aeaf..bd0776daea 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -26,7 +26,7 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post22; platform_system == "Windows"
+triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 2a9484bfdd..39a06cdbe0 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -26,7 +26,7 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post22; platform_system == "Windows"
+triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 1e75b7919a..3ff17704be 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -28,7 +28,7 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post22; platform_system == "Windows"
+triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index d8d7a92783..dc740dc86f 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -26,7 +26,7 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post22; platform_system == "Windows"
+triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
 

From dc2bbf1861c5fd39a6ec8959718081d9b899cfac Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 14 Jan 2026 16:21:41 -0800
Subject: [PATCH 1161/1701] Refactor thinking block detection and add Solar
 Open support

---
 modules/html_generator.py | 113 +++++++++++++++-----------------------
 1 file changed, 43 insertions(+), 70 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index ccc232a049..312b66ad07 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -108,91 +108,64 @@ def replace_blockquote(m):
     return m.group().replace('\n', '\n> ').replace('\\begin{blockquote}', '').replace('\\end{blockquote}', '')
 
 
+# Thinking block format definitions: (start_tag, end_tag, content_start_tag)
+# Use None for start_tag to match from beginning (end-only formats should be listed last)
+THINKING_FORMATS = [
+    ('', '', None),
+    ('<|channel|>analysis<|message|>', '<|end|>', '<|start|>assistant<|channel|>final<|message|>'),
+    ('', '', None),
+    ('<|think|>', '<|end|>', '<|content|>'),  # Solar Open
+    (None, '', None),  # End-only variant (e.g., Qwen3-next)
+]
+
+
 def extract_thinking_block(string):
     """Extract thinking blocks from the beginning of a string."""
     if not string:
         return None, string
 
-    THINK_START_TAG = "<think>"
-    THINK_END_TAG = "</think>"
-
-    # Look for think tag first
-    start_pos = string.find(THINK_START_TAG)
-    end_pos = string.find(THINK_END_TAG)
+    for start_tag, end_tag, content_tag in THINKING_FORMATS:
+        end_esc = html.escape(end_tag)
+        content_esc = html.escape(content_tag) if content_tag else None
 
-    # If think tags found, use existing logic
-    if start_pos != -1 or end_pos != -1:
-        # handle missing start or end tags
-        if start_pos == -1:
-            thought_start = 0
-        else:
-            thought_start = start_pos + len(THINK_START_TAG)
-        if end_pos == -1:
-            thought_end = len(string)
-            content_start = len(string)
-        else:
-            thought_end = end_pos
-            content_start = end_pos + len(THINK_END_TAG)
-        thinking_content = string[thought_start:thought_end]
-        remaining_content = string[content_start:]
-        return thinking_content, remaining_content
-
-    # If think tags not found, try GPT-OSS alternative format
-    ALT_START = "<|channel|>analysis<|message|>"
-    ALT_END = "<|end|>"
-    ALT_CONTENT_START = "<|start|>assistant<|channel|>final<|message|>"
-
-    alt_start_pos = string.find(ALT_START)
-    alt_end_pos = string.find(ALT_END)
-    alt_content_pos = string.find(ALT_CONTENT_START)
-
-    if alt_start_pos != -1 or alt_end_pos != -1:
-        if alt_start_pos == -1:
+        if start_tag is None:
+            # End-only format: require end tag, start from beginning
+            end_pos = string.find(end_esc)
+            if end_pos == -1:
+                continue
             thought_start = 0
         else:
-            thought_start = alt_start_pos + len(ALT_START)
+            # Normal format: require start tag
+            start_esc = html.escape(start_tag)
+            start_pos = string.find(start_esc)
+            if start_pos == -1:
+                continue
+            thought_start = start_pos + len(start_esc)
+            end_pos = string.find(end_esc, thought_start)
 
-        # If no explicit end tag but content start exists, use content start as end
-        if alt_end_pos == -1:
-            if alt_content_pos != -1:
-                thought_end = alt_content_pos
-                content_start = alt_content_pos + len(ALT_CONTENT_START)
+        if end_pos == -1:
+            # End tag missing - check if content tag can serve as fallback
+            if content_esc:
+                content_pos = string.find(content_esc, thought_start)
+                if content_pos != -1:
+                    thought_end = content_pos
+                    content_start = content_pos + len(content_esc)
+                else:
+                    thought_end = len(string)
+                    content_start = len(string)
             else:
                 thought_end = len(string)
                 content_start = len(string)
         else:
-            thought_end = alt_end_pos
-            content_start = alt_content_pos + len(ALT_CONTENT_START) if alt_content_pos != -1 else alt_end_pos + len(ALT_END)
-
-        thinking_content = string[thought_start:thought_end]
-        remaining_content = string[content_start:]
-        return thinking_content, remaining_content
-
-    # Try seed:think format
-    SEED_START = "<seed:think>"
-    SEED_END = "</seed:think>"
-
-    seed_start_pos = string.find(SEED_START)
-    seed_end_pos = string.find(SEED_END)
-
-    if seed_start_pos != -1 or seed_end_pos != -1:
-        if seed_start_pos == -1:
-            thought_start = 0
-        else:
-            thought_start = seed_start_pos + len(SEED_START)
-
-        if seed_end_pos == -1:
-            thought_end = len(string)
-            content_start = len(string)
-        else:
-            thought_end = seed_end_pos
-            content_start = seed_end_pos + len(SEED_END)
+            thought_end = end_pos
+            if content_esc:
+                content_pos = string.find(content_esc, end_pos)
+                content_start = content_pos + len(content_esc) if content_pos != -1 else end_pos + len(end_esc)
+            else:
+                content_start = end_pos + len(end_esc)
 
-        thinking_content = string[thought_start:thought_end]
-        remaining_content = string[content_start:]
-        return thinking_content, remaining_content
+        return string[thought_start:thought_end], string[content_start:]
 
-    # Return if no format is found
     return None, string
 
 

From c54e8a2b3d84b376546f258c24046daa7fc4c8e5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 28 Jan 2026 08:23:55 -0800
Subject: [PATCH 1162/1701] Try to spawn llama.cpp on port 5001 instead of
 random port

---
 modules/llama_cpp_server.py | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index f83ed663d7..eb59baf9e1 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -298,10 +298,24 @@ def _get_bos_token(self):
         if "bos_token" in response:
             self.bos_token = response["bos_token"]
 
+    def _is_port_available(self, port):
+        """Check if a port is available for use."""
+        with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
+            try:
+                s.bind(('', port))
+                return True
+            except OSError:
+                return False
+
     def _find_available_port(self):
-        """Find an available port by letting the OS assign one."""
+        """Find an available port, preferring main port + 1."""
+        preferred_port = shared.args.api_port + 1
+        if self._is_port_available(preferred_port):
+            return preferred_port
+
+        # Fall back to OS-assigned random port
         with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
-            s.bind(('', 0))  # Bind to port 0 to get an available port
+            s.bind(('', 0))
             return s.getsockname()[1]
 
     def _start_server(self):

From efc72d5c32f4cd15d4736813e089c95ad96cd9c5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 3 Mar 2026 11:03:26 -0800
Subject: [PATCH 1163/1701] Update Python from 3.11 to 3.13

---
 .github/workflows/build-portable-release-cuda.yml  |  8 ++++----
 .github/workflows/build-portable-release-rocm.yml  |  6 +++---
 .../workflows/build-portable-release-vulkan.yml    |  8 ++++----
 .github/workflows/build-portable-release.yml       | 14 +++++++-------
 README.md                                          |  2 +-
 one_click.py                                       | 14 +++++++++++++-
 requirements/full/requirements.txt                 | 12 ++++++------
 requirements/full/requirements_amd.txt             |  2 +-
 requirements/full/requirements_amd_noavx2.txt      |  2 +-
 requirements/full/requirements_noavx2.txt          | 12 ++++++------
 start_linux.sh                                     |  2 +-
 start_macos.sh                                     |  2 +-
 start_windows.bat                                  |  2 +-
 13 files changed, 49 insertions(+), 37 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index 3b1ab8a862..f5d7c55c43 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -58,7 +58,7 @@ jobs:
         run: |
           $matrix = @{
               'os' = @('ubuntu-22.04', 'windows-2022')
-              'pyver' = @("3.11")
+              'pyver' = @("3.13")
               'avx' = @("AVX2")
               'cuda' = @("12.4")
           }
@@ -119,15 +119,15 @@ jobs:
             # 1. Set platform-specific variables
             if [[ "$RUNNER_OS" == "Windows" ]]; then
                 PLATFORM="windows"
-                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-pc-windows-msvc-install_only.tar.gz"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-pc-windows-msvc-install_only.tar.gz"
                 PIP_PATH="portable_env/python.exe -m pip"
                 PACKAGES_PATH="portable_env/Lib/site-packages"
                 rm start_linux.sh start_macos.sh
             else
                 PLATFORM="linux"
-                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-unknown-linux-gnu-install_only.tar.gz"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only.tar.gz"
                 PIP_PATH="portable_env/bin/python -m pip"
-                PACKAGES_PATH="portable_env/lib/python3.11/site-packages"
+                PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
                 rm start_macos.sh start_windows.bat
             fi
 
diff --git a/.github/workflows/build-portable-release-rocm.yml b/.github/workflows/build-portable-release-rocm.yml
index ed7f3a87f6..b34ebbccec 100644
--- a/.github/workflows/build-portable-release-rocm.yml
+++ b/.github/workflows/build-portable-release-rocm.yml
@@ -58,7 +58,7 @@ jobs:
         run: |
           $matrix = @{
               'os' = @('ubuntu-22.04')
-              'pyver' = @("3.11")
+              'pyver' = @("3.13")
               'avx' = @("AVX2")
           }
 
@@ -116,9 +116,9 @@ jobs:
 
             # 1. Set platform-specific variables (Linux only for ROCm)
             PLATFORM="linux"
-            PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-unknown-linux-gnu-install_only.tar.gz"
+            PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only.tar.gz"
             PIP_PATH="portable_env/bin/python -m pip"
-            PACKAGES_PATH="portable_env/lib/python3.11/site-packages"
+            PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
             rm start_macos.sh start_windows.bat
 
             # 2. Download and extract Python
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index d275460b76..15388c775f 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -58,7 +58,7 @@ jobs:
         run: |
           $matrix = @{
               'os' = @('ubuntu-22.04', 'windows-2022')
-              'pyver' = @("3.11")
+              'pyver' = @("3.13")
               'avx' = @("AVX2")
           }
 
@@ -117,15 +117,15 @@ jobs:
             # 1. Set platform-specific variables
             if [[ "$RUNNER_OS" == "Windows" ]]; then
                 PLATFORM="windows"
-                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-pc-windows-msvc-install_only.tar.gz"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-pc-windows-msvc-install_only.tar.gz"
                 PIP_PATH="portable_env/python.exe -m pip"
                 PACKAGES_PATH="portable_env/Lib/site-packages"
                 rm start_linux.sh start_macos.sh
             else
                 PLATFORM="linux"
-                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-unknown-linux-gnu-install_only.tar.gz"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only.tar.gz"
                 PIP_PATH="portable_env/bin/python -m pip"
-                PACKAGES_PATH="portable_env/lib/python3.11/site-packages"
+                PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
                 rm start_macos.sh start_windows.bat
             fi
 
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index a5b5b7c534..84b5c294a5 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -58,7 +58,7 @@ jobs:
         run: |
           $matrix = @{
               'os' = @('ubuntu-22.04', 'windows-2022', 'macos-14')
-              'pyver' = @("3.11")
+              'pyver' = @("3.13")
               'avx' = @("AVX2")
           }
 
@@ -118,29 +118,29 @@ jobs:
             # 1. Set platform-specific variables
             if [[ "$RUNNER_OS" == "Windows" ]]; then
                 PLATFORM="windows-cpu"
-                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-pc-windows-msvc-install_only.tar.gz"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-pc-windows-msvc-install_only.tar.gz"
                 PIP_PATH="portable_env/python.exe -m pip"
                 PACKAGES_PATH="portable_env/Lib/site-packages"
                 rm start_linux.sh start_macos.sh
             elif [[ "$RUNNER_OS" == "macOS" ]]; then
                 if [[ "$OS_TYPE" == "macos-13" ]]; then
                     PLATFORM="macos-x86_64"
-                    PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-apple-darwin-install_only.tar.gz"
+                    PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-apple-darwin-install_only.tar.gz"
                     REQ_TYPE="apple_intel"
                 else
                     PLATFORM="macos-arm64"
-                    PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-aarch64-apple-darwin-install_only.tar.gz"
+                    PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-aarch64-apple-darwin-install_only.tar.gz"
                     REQ_TYPE="apple_silicon"
                 fi
                 PIP_PATH="portable_env/bin/python -m pip"
-                PACKAGES_PATH="portable_env/lib/python3.11/site-packages"
+                PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
                 rm start_linux.sh start_windows.bat
             else
                 # Linux case
                 PLATFORM="linux-cpu"
-                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20250409/cpython-3.11.12+20250409-x86_64-unknown-linux-gnu-install_only.tar.gz"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only.tar.gz"
                 PIP_PATH="portable_env/bin/python -m pip"
-                PACKAGES_PATH="portable_env/lib/python3.11/site-packages"
+                PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
                 rm start_macos.sh start_windows.bat
             fi
 
diff --git a/README.md b/README.md
index 4cc13b88c2..5e44b6b9e0 100644
--- a/README.md
+++ b/README.md
@@ -138,7 +138,7 @@ For other platforms, download from: https://github.com/conda-forge/miniforge/rel
 #### 1. Create a new conda environment
 
 ```
-conda create -n textgen python=3.11
+conda create -n textgen python=3.13
 conda activate textgen
 ```
 
diff --git a/one_click.py b/one_click.py
index 881d7489b9..5f6cc85e2d 100644
--- a/one_click.py
+++ b/one_click.py
@@ -17,7 +17,7 @@
 
 # Define the required versions
 TORCH_VERSION = "2.7.1"
-PYTHON_VERSION = "3.11"
+PYTHON_VERSION = "3.13"
 LIBSTDCXX_VERSION_LINUX = "12.1.0"
 
 # Environment
@@ -359,6 +359,18 @@ def update_requirements(initial_installation=False, pull=True):
             assert_success=True
         )
 
+    # Check for outdated Python version and refuse to update
+    if '.'.join(map(str, sys.version_info[:2])) != PYTHON_VERSION:
+        print_big_message(
+            "Your current installation uses Python {}.{}, which is outdated.\n"
+            "Python {} is now required. A clean installation is needed.\n\n"
+            "INSTRUCTIONS:\n"
+            "1. Delete the 'installer_files' folder in your text-generation-webui directory.\n"
+            "2. Run the start script again (e.g., start_windows.bat).\n\n"
+            "This will create a fresh environment with the latest software.".format(*sys.version_info[:2], PYTHON_VERSION)
+        )
+        sys.exit(0)
+
     # Check for outdated CUDA 12.4 installs and refuse to update
     state = load_state()
     if state.get('gpu_choice') == 'NVIDIA':
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index dd74ed52e9..7e6d6240cb 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -44,10 +44,10 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/Dao-AILab/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu12torch2.7cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.7.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/Dao-AILab/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu12torch2.7cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 3475c161ac..a288857388 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -42,5 +42,5 @@ tiktoken
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 05077a32a6..7ec857e55a 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -42,5 +42,5 @@ tiktoken
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 3ff17704be..6550f8c0b6 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -44,10 +44,10 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.7.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/Dao-AILab/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu12torch2.7cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.7.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/Dao-AILab/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu12torch2.7cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
diff --git a/start_linux.sh b/start_linux.sh
index fac8df3b0f..81bda49062 100755
--- a/start_linux.sh
+++ b/start_linux.sh
@@ -57,7 +57,7 @@ fi
 
 # create the installer env
 if [ ! -e "$INSTALL_ENV_DIR" ]; then
-    "$CONDA_ROOT_PREFIX/bin/conda" create -y -k --prefix "$INSTALL_ENV_DIR" python=3.11
+    "$CONDA_ROOT_PREFIX/bin/conda" create -y -k --prefix "$INSTALL_ENV_DIR" python=3.13
 fi
 
 # check if conda environment was actually created
diff --git a/start_macos.sh b/start_macos.sh
index 6329cc0c6b..56a7afd04f 100755
--- a/start_macos.sh
+++ b/start_macos.sh
@@ -57,7 +57,7 @@ fi
 
 # create the installer env
 if [ ! -e "$INSTALL_ENV_DIR" ]; then
-    "$CONDA_ROOT_PREFIX/bin/conda" create -y -k --prefix "$INSTALL_ENV_DIR" python=3.11
+    "$CONDA_ROOT_PREFIX/bin/conda" create -y -k --prefix "$INSTALL_ENV_DIR" python=3.13
 fi
 
 # check if conda environment was actually created
diff --git a/start_windows.bat b/start_windows.bat
index 68c6194e76..ceaef95cd8 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -86,7 +86,7 @@ if "%conda_exists%" == "F" (
 @rem create the installer env
 if not exist "%INSTALL_ENV_DIR%" (
 	echo Packages to install: %PACKAGES_TO_INSTALL%
-	call "%CONDA_ROOT_PREFIX%\_conda.exe" create --no-shortcuts -y -k --prefix "%INSTALL_ENV_DIR%" python=3.11 || ( echo. && echo Conda environment creation failed. && goto end )
+	call "%CONDA_ROOT_PREFIX%\_conda.exe" create --no-shortcuts -y -k --prefix "%INSTALL_ENV_DIR%" python=3.13 || ( echo. && echo Conda environment creation failed. && goto end )
 )
 
 @rem check if conda environment was actually created

From ddd74324feb17233cd70f9726455ddd80d3b7afa Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 3 Mar 2026 11:38:52 -0800
Subject: [PATCH 1164/1701] Update PyTorch to 2.9.1 and ROCm to 6.4

---
 README.md                                     | 12 ++++++------
 one_click.py                                  | 10 +++++-----
 requirements/full/requirements.txt            | 12 ++++++------
 requirements/full/requirements_amd.txt        |  2 +-
 requirements/full/requirements_amd_noavx2.txt |  2 +-
 requirements/full/requirements_noavx2.txt     | 12 ++++++------
 6 files changed, 25 insertions(+), 25 deletions(-)

diff --git a/README.md b/README.md
index 5e44b6b9e0..c71fc565a0 100644
--- a/README.md
+++ b/README.md
@@ -146,12 +146,12 @@ conda activate textgen
 
 | System | GPU | Command |
 |--------|---------|---------|
-| Linux/WSL | NVIDIA | `pip3 install torch==2.7.1 --index-url https://download.pytorch.org/whl/cu128` |
-| Linux/WSL | CPU only | `pip3 install torch==2.7.1 --index-url https://download.pytorch.org/whl/cpu` |
-| Linux | AMD | `pip3 install torch==2.7.1 --index-url https://download.pytorch.org/whl/rocm6.2.4` |
-| MacOS + MPS | Any | `pip3 install torch==2.7.1` |
-| Windows | NVIDIA | `pip3 install torch==2.7.1 --index-url https://download.pytorch.org/whl/cu128` |
-| Windows | CPU only | `pip3 install torch==2.7.1` |
+| Linux/WSL | NVIDIA | `pip3 install torch==2.9.1 --index-url https://download.pytorch.org/whl/cu128` |
+| Linux/WSL | CPU only | `pip3 install torch==2.9.1 --index-url https://download.pytorch.org/whl/cpu` |
+| Linux | AMD | `pip3 install torch==2.9.1 --index-url https://download.pytorch.org/whl/rocm6.4` |
+| MacOS + MPS | Any | `pip3 install torch==2.9.1` |
+| Windows | NVIDIA | `pip3 install torch==2.9.1 --index-url https://download.pytorch.org/whl/cu128` |
+| Windows | CPU only | `pip3 install torch==2.9.1` |
 
 The up-to-date commands can be found here: https://pytorch.org/get-started/locally/.
 
diff --git a/one_click.py b/one_click.py
index 5f6cc85e2d..46e86a0cfc 100644
--- a/one_click.py
+++ b/one_click.py
@@ -16,7 +16,7 @@
 # os.environ["HCC_AMDGPU_TARGET"] = 'gfx1030'
 
 # Define the required versions
-TORCH_VERSION = "2.7.1"
+TORCH_VERSION = "2.9.1"
 PYTHON_VERSION = "3.13"
 LIBSTDCXX_VERSION_LINUX = "12.1.0"
 
@@ -114,7 +114,7 @@ def get_gpu_choice():
                 "What is your GPU?",
                 {
                     'A': 'NVIDIA',
-                    'B': 'AMD - Linux/macOS only, requires ROCm 6.2.4',
+                    'B': 'AMD - Linux/macOS only, requires ROCm 6.4',
                     'C': 'Apple M Series',
                     'D': 'Intel Arc (beta)',
                     'N': 'CPU mode'
@@ -138,7 +138,7 @@ def get_pytorch_install_command(gpu_choice):
     if gpu_choice == "NVIDIA_CUDA128":
         return base_cmd + "--index-url https://download.pytorch.org/whl/cu128"
     elif gpu_choice == "AMD":
-        return base_cmd + "--index-url https://download.pytorch.org/whl/rocm6.2.4"
+        return base_cmd + "--index-url https://download.pytorch.org/whl/rocm6.4"
     elif gpu_choice in ["APPLE", "NONE"]:
         return base_cmd + "--index-url https://download.pytorch.org/whl/cpu"
     elif gpu_choice == "INTEL":
@@ -157,7 +157,7 @@ def get_pytorch_update_command(gpu_choice):
     if gpu_choice == "NVIDIA_CUDA128":
         return f"{base_cmd} --index-url https://download.pytorch.org/whl/cu128"
     elif gpu_choice == "AMD":
-        return f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.2.4"
+        return f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.4"
     elif gpu_choice in ["APPLE", "NONE"]:
         return f"{base_cmd} --index-url https://download.pytorch.org/whl/cpu"
     elif gpu_choice == "INTEL":
@@ -291,7 +291,7 @@ def update_pytorch_and_python():
 
 
 def clean_outdated_pytorch_cuda_dependencies():
-    patterns = ["cu121", "cu122", "torch2.4", "torchvision", "torchaudio"]
+    patterns = ["cu121", "cu122", "torch2.4", "torch2.6", "torch2.7", "torchvision", "torchaudio"]
     result = run_cmd("python -m pip list --format=freeze", capture_output=True, environment=True)
     matching_packages = []
 
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 7e6d6240cb..e6152d612d 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -44,10 +44,10 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.7.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/Dao-AILab/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu12torch2.7cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index a288857388..251623b5c4 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -42,5 +42,5 @@ tiktoken
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.4.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 7ec857e55a..43cb282faf 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -42,5 +42,5 @@ tiktoken
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.4.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 6550f8c0b6..11b8325681 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -44,10 +44,10 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.7.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/Dao-AILab/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu12torch2.7cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"

From 38d0eeefc0f94c56a88fca41a1c6a1eccfd5166f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 3 Mar 2026 12:01:02 -0800
Subject: [PATCH 1165/1701] Update dependencies: torch 2.9.1, transformers 5.2,
 exllamav3 0.0.22, accelerate 1.12, huggingface-hub 1.5

---
 requirements/full/requirements.txt                     | 10 +++++-----
 requirements/full/requirements_amd.txt                 |  6 +++---
 requirements/full/requirements_amd_noavx2.txt          |  6 +++---
 requirements/full/requirements_apple_intel.txt         |  6 +++---
 requirements/full/requirements_apple_silicon.txt       |  6 +++---
 requirements/full/requirements_cpu_only.txt            |  6 +++---
 requirements/full/requirements_cpu_only_noavx2.txt     |  6 +++---
 requirements/full/requirements_noavx2.txt              | 10 +++++-----
 requirements/full/requirements_nowheels.txt            |  6 +++---
 requirements/portable/requirements.txt                 |  2 +-
 requirements/portable/requirements_amd.txt             |  2 +-
 requirements/portable/requirements_amd_noavx2.txt      |  2 +-
 requirements/portable/requirements_apple_intel.txt     |  2 +-
 requirements/portable/requirements_apple_silicon.txt   |  2 +-
 requirements/portable/requirements_cpu_only.txt        |  2 +-
 requirements/portable/requirements_cpu_only_noavx2.txt |  2 +-
 requirements/portable/requirements_noavx2.txt          |  2 +-
 requirements/portable/requirements_nowheels.txt        |  2 +-
 requirements/portable/requirements_vulkan.txt          |  2 +-
 requirements/portable/requirements_vulkan_noavx2.txt   |  2 +-
 20 files changed, 42 insertions(+), 42 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index e6152d612d..bed2aeb4b6 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -1,4 +1,4 @@
-accelerate==1.8.*
+accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 bitsandbytes==0.49.*
 colorama
@@ -8,7 +8,7 @@ einops
 fastapi==0.112.4
 flash-linear-attention==0.4.*
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
@@ -27,7 +27,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==4.57.*
+transformers==5.2.*
 triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
@@ -44,8 +44,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.22/exllamav3-0.0.22+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.22/exllamav3-0.0.22+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 251623b5c4..8824357c6a 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -1,4 +1,4 @@
-accelerate==1.8.*
+accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
@@ -6,7 +6,7 @@ diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
@@ -25,7 +25,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==4.57.*
+transformers==5.2.*
 triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
index 43cb282faf..ca7f0fe9f7 100644
--- a/requirements/full/requirements_amd_noavx2.txt
+++ b/requirements/full/requirements_amd_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.8.*
+accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
@@ -6,7 +6,7 @@ diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
@@ -25,7 +25,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==4.57.*
+transformers==5.2.*
 triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index f591fa251e..b9f1f1f293 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -1,4 +1,4 @@
-accelerate==1.8.*
+accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
@@ -6,7 +6,7 @@ diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
@@ -25,7 +25,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==4.57.*
+transformers==5.2.*
 triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 6ff4753d4b..71da9a67b7 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -1,4 +1,4 @@
-accelerate==1.8.*
+accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
@@ -6,7 +6,7 @@ diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
@@ -25,7 +25,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==4.57.*
+transformers==5.2.*
 triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index bd0776daea..cd70cb7fec 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -1,4 +1,4 @@
-accelerate==1.8.*
+accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
@@ -6,7 +6,7 @@ diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
@@ -25,7 +25,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==4.57.*
+transformers==5.2.*
 triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
index 39a06cdbe0..1a128289dc 100644
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ b/requirements/full/requirements_cpu_only_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.8.*
+accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
@@ -6,7 +6,7 @@ diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
@@ -25,7 +25,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==4.57.*
+transformers==5.2.*
 triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
index 11b8325681..e2a9fd580b 100644
--- a/requirements/full/requirements_noavx2.txt
+++ b/requirements/full/requirements_noavx2.txt
@@ -1,4 +1,4 @@
-accelerate==1.8.*
+accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 bitsandbytes==0.49.*
 colorama
@@ -8,7 +8,7 @@ einops
 fastapi==0.112.4
 flash-linear-attention==0.4.*
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
@@ -27,7 +27,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==4.57.*
+transformers==5.2.*
 triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
@@ -44,8 +44,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.22/exllamav3-0.0.22+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.22/exllamav3-0.0.22+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index dc740dc86f..e70bec7e9e 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -1,4 +1,4 @@
-accelerate==1.8.*
+accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
@@ -6,7 +6,7 @@ diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
@@ -25,7 +25,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==4.57.*
+transformers==5.2.*
 triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 2f6bf51142..b9576255bf 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -1,7 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 209610ceb9..dc30f421fa 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -1,7 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
index 29a1f72b04..3474642614 100644
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ b/requirements/portable/requirements_amd_noavx2.txt
@@ -1,7 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 8812ac39f5..6bae46d691 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -1,7 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index c627cfd456..2f9e4dc5d9 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -1,7 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 5ca660adfc..739ac26fa8 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -1,7 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
index 5bb282eb18..e096ed1d0c 100644
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ b/requirements/portable/requirements_cpu_only_noavx2.txt
@@ -1,7 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
index ecce4a2f7d..12e24ab49b 100644
--- a/requirements/portable/requirements_noavx2.txt
+++ b/requirements/portable/requirements_noavx2.txt
@@ -1,7 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index c9831985e5..347f97f69d 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -1,7 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 4b3adddc3d..c2a66a69c9 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -1,7 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
index 8d6acdd78e..79b632bb49 100644
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ b/requirements/portable/requirements_vulkan_noavx2.txt
@@ -1,7 +1,7 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
 html2text==2025.4.15
-huggingface-hub==0.36.0
+huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
 numpy==2.2.*

From 9814d3d0ae6aa3a478ebdd08246eac76fac8cf15 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 3 Mar 2026 12:20:37 -0800
Subject: [PATCH 1166/1701] Patch gradio 4.x for huggingface-hub 1.x
 compatibility

---
 server.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/server.py b/server.py
index d8fb2c224b..e98ff153df 100644
--- a/server.py
+++ b/server.py
@@ -3,6 +3,11 @@
 import warnings
 from pathlib import Path
 
+# Monkey-patch HfFolder for gradio 4.x compatibility with huggingface-hub 1.x
+import huggingface_hub
+if not hasattr(huggingface_hub, 'HfFolder'):
+    huggingface_hub.HfFolder = type('HfFolder', (), {'get_token': staticmethod(huggingface_hub.get_token)})
+
 from modules import shared
 from modules.block_requests import OpenMonkeyPatch, RequestBlocker
 from modules.image_models import load_image_model

From 9576c5a5f4db2715c1440e2b630e46cc0433167f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 3 Mar 2026 12:33:20 -0800
Subject: [PATCH 1167/1701] Update Miniforge from 25.3.0 to 26.1.0

---
 start_linux.sh    | 2 +-
 start_macos.sh    | 2 +-
 start_windows.bat | 4 ++--
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/start_linux.sh b/start_linux.sh
index 81bda49062..2aa07d45f9 100755
--- a/start_linux.sh
+++ b/start_linux.sh
@@ -30,7 +30,7 @@ esac
 INSTALL_DIR="$(pwd)/installer_files"
 CONDA_ROOT_PREFIX="$(pwd)/installer_files/conda"
 INSTALL_ENV_DIR="$(pwd)/installer_files/env"
-MINIFORGE_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-Linux-${OS_ARCH}.sh"
+MINIFORGE_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/26.1.0-0/Miniforge3-26.1.0-0-Linux-${OS_ARCH}.sh"
 conda_exists="F"
 
 # figure out whether git and conda needs to be installed
diff --git a/start_macos.sh b/start_macos.sh
index 56a7afd04f..de257608be 100755
--- a/start_macos.sh
+++ b/start_macos.sh
@@ -30,7 +30,7 @@ esac
 INSTALL_DIR="$(pwd)/installer_files"
 CONDA_ROOT_PREFIX="$(pwd)/installer_files/conda"
 INSTALL_ENV_DIR="$(pwd)/installer_files/env"
-MINIFORGE_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-MacOSX-${OS_ARCH}.sh"
+MINIFORGE_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/26.1.0-0/Miniforge3-26.1.0-0-MacOSX-${OS_ARCH}.sh"
 conda_exists="F"
 
 # figure out whether git and conda needs to be installed
diff --git a/start_windows.bat b/start_windows.bat
index ceaef95cd8..406fd27e07 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -36,8 +36,8 @@ set TEMP=%cd%\installer_files
 set INSTALL_DIR=%cd%\installer_files
 set CONDA_ROOT_PREFIX=%cd%\installer_files\conda
 set INSTALL_ENV_DIR=%cd%\installer_files\env
-set MINIFORGE_DOWNLOAD_URL=https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-Windows-x86_64.exe
-set MINIFORGE_CHECKSUM=b48cd98430170983076dfb51769a6d37668176f59bf3b59c4b21ac4c9bc24f39
+set MINIFORGE_DOWNLOAD_URL=https://github.com/conda-forge/miniforge/releases/download/26.1.0-0/Miniforge3-26.1.0-0-Windows-x86_64.exe
+set MINIFORGE_CHECKSUM=0ad64473c20a8649be9313f64ee898f4b23a35a7a25ea9998a751c542e5e3840
 set conda_exists=F
 
 @rem figure out whether git and conda needs to be installed

From d7dd533b99f19ba905f13201688395295234a43b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?P=C3=A1draic=20Slattery?= 
Date: Tue, 3 Mar 2026 21:54:12 +0100
Subject: [PATCH 1168/1701] chore: Update outdated GitHub Actions versions
 (#7384)

---
 .github/workflows/build-portable-release-cuda.yml   | 4 ++--
 .github/workflows/build-portable-release-rocm.yml   | 4 ++--
 .github/workflows/build-portable-release-vulkan.yml | 4 ++--
 .github/workflows/build-portable-release.yml        | 4 ++--
 4 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index f5d7c55c43..b304dcd6ad 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -88,13 +88,13 @@ jobs:
       PCKGVER: ${{ inputs.version }}
 
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           repository: 'oobabooga/text-generation-webui'
           ref: ${{ inputs.version }}
           submodules: 'recursive'
 
-      - uses: actions/setup-python@v4
+      - uses: actions/setup-python@v6
         with:
           python-version: ${{ matrix.pyver }}
 
diff --git a/.github/workflows/build-portable-release-rocm.yml b/.github/workflows/build-portable-release-rocm.yml
index b34ebbccec..0a15080649 100644
--- a/.github/workflows/build-portable-release-rocm.yml
+++ b/.github/workflows/build-portable-release-rocm.yml
@@ -87,13 +87,13 @@ jobs:
       PCKGVER: ${{ inputs.version }}
 
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           repository: 'oobabooga/text-generation-webui'
           ref: ${{ inputs.version }}
           submodules: 'recursive'
 
-      - uses: actions/setup-python@v4
+      - uses: actions/setup-python@v6
         with:
           python-version: ${{ matrix.pyver }}
 
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 15388c775f..eda684b158 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -87,13 +87,13 @@ jobs:
       PCKGVER: ${{ inputs.version }}
 
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           repository: 'oobabooga/text-generation-webui'
           ref: ${{ inputs.version }}
           submodules: 'recursive'
 
-      - uses: actions/setup-python@v4
+      - uses: actions/setup-python@v6
         with:
           python-version: ${{ matrix.pyver }}
 
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 84b5c294a5..1f6e4da30d 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -87,13 +87,13 @@ jobs:
       PCKGVER: ${{ inputs.version }}
 
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           repository: 'oobabooga/text-generation-webui'
           ref: ${{ inputs.version }}
           submodules: 'recursive'
 
-      - uses: actions/setup-python@v4
+      - uses: actions/setup-python@v6
         with:
           python-version: ${{ matrix.pyver }}
 

From b8fcc8ea325f2864794eaed87736b4093a6f9e2e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 3 Mar 2026 15:22:16 -0800
Subject: [PATCH 1169/1701] Update llama.cpp, remove noavx2 builds, add ROCm
 Windows portable builds

---
 .../workflows/build-portable-release-cuda.yml | 16 ++----
 .../workflows/build-portable-release-rocm.yml | 42 ++++++++-------
 .../build-portable-release-vulkan.yml         | 14 ++---
 .github/workflows/build-portable-release.yml  | 14 ++---
 README.md                                     | 17 +++---
 one_click.py                                  | 28 ++--------
 requirements/full/requirements.txt            |  4 +-
 requirements/full/requirements_amd.txt        |  4 +-
 requirements/full/requirements_amd_noavx2.txt | 46 ----------------
 .../full/requirements_apple_intel.txt         |  3 +-
 .../full/requirements_apple_silicon.txt       |  3 +-
 requirements/full/requirements_cpu_only.txt   |  6 +--
 .../full/requirements_cpu_only_noavx2.txt     | 44 ---------------
 requirements/full/requirements_noavx2.txt     | 53 -------------------
 requirements/portable/requirements.txt        |  4 +-
 requirements/portable/requirements_amd.txt    |  4 +-
 .../portable/requirements_amd_noavx2.txt      | 27 ----------
 .../portable/requirements_apple_intel.txt     |  3 +-
 .../portable/requirements_apple_silicon.txt   |  3 +-
 .../portable/requirements_cpu_only.txt        |  6 +--
 .../portable/requirements_cpu_only_noavx2.txt | 27 ----------
 requirements/portable/requirements_noavx2.txt | 27 ----------
 requirements/portable/requirements_vulkan.txt |  4 +-
 .../portable/requirements_vulkan_noavx2.txt   | 27 ----------
 24 files changed, 64 insertions(+), 362 deletions(-)
 delete mode 100644 requirements/full/requirements_amd_noavx2.txt
 delete mode 100644 requirements/full/requirements_cpu_only_noavx2.txt
 delete mode 100644 requirements/full/requirements_noavx2.txt
 delete mode 100644 requirements/portable/requirements_amd_noavx2.txt
 delete mode 100644 requirements/portable/requirements_cpu_only_noavx2.txt
 delete mode 100644 requirements/portable/requirements_noavx2.txt
 delete mode 100644 requirements/portable/requirements_vulkan_noavx2.txt

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index b304dcd6ad..a10e8ec4d0 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -59,7 +59,6 @@ jobs:
           $matrix = @{
               'os' = @('ubuntu-22.04', 'windows-2022')
               'pyver' = @("3.13")
-              'avx' = @("AVX2")
               'cuda' = @("12.4")
           }
 
@@ -75,7 +74,7 @@ jobs:
           Write-Output ('matrix=' + $matrixOut) >> $env:GITHUB_OUTPUT
 
   build_wheels:
-    name: ${{ matrix.os }} ${{ matrix.pyver }} CPU ${{ matrix.avx }} CUDA ${{ matrix.cuda }}
+    name: ${{ matrix.os }} ${{ matrix.pyver }} CUDA ${{ matrix.cuda }}
     needs: define_matrix
     runs-on: ${{ matrix.os }}
     strategy:
@@ -84,7 +83,6 @@ jobs:
       run:
         shell: pwsh
     env:
-      AVXVER: ${{ matrix.avx }}
       PCKGVER: ${{ inputs.version }}
 
     steps:
@@ -113,7 +111,6 @@ jobs:
 
             # Define common variables
             CUDA_VERSION="${{ matrix.cuda }}"
-            AVX_SUPPORT="${{ matrix.avx }}"
             VERSION="${{ inputs.version }}"
 
             # 1. Set platform-specific variables
@@ -138,16 +135,9 @@ jobs:
             tar -xzf python-build.tar.gz
             mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
 
-            # 3. Prepare requirements file based on AVX and CUDA
-            if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
-                BASE_REQ_FILE="requirements/portable/requirements.txt"
-            else
-                BASE_REQ_FILE="requirements/portable/requirements_noavx2.txt"
-            fi
-
-            # Create CUDA-specific requirements file if needed
+            # 3. Prepare requirements file
             cd "text-generation-webui-${VERSION_CLEAN}"
-            REQ_FILE="$BASE_REQ_FILE"
+            REQ_FILE="requirements/portable/requirements.txt"
 
             # 4. Install packages
             echo "Installing Python packages from $REQ_FILE..."
diff --git a/.github/workflows/build-portable-release-rocm.yml b/.github/workflows/build-portable-release-rocm.yml
index 0a15080649..5b43b2d317 100644
--- a/.github/workflows/build-portable-release-rocm.yml
+++ b/.github/workflows/build-portable-release-rocm.yml
@@ -57,9 +57,8 @@ jobs:
         id: set-matrix
         run: |
           $matrix = @{
-              'os' = @('ubuntu-22.04')
+              'os' = @('ubuntu-22.04', 'windows-2022')
               'pyver' = @("3.13")
-              'avx' = @("AVX2")
           }
 
           if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})}
@@ -74,7 +73,7 @@ jobs:
           Write-Output ('matrix=' + $matrixOut) >> $env:GITHUB_OUTPUT
 
   build_wheels:
-    name: ${{ matrix.os }} ${{ matrix.pyver }} CPU ${{ matrix.avx }}
+    name: ${{ matrix.os }} ${{ matrix.pyver }}
     needs: define_matrix
     runs-on: ${{ matrix.os }}
     strategy:
@@ -83,7 +82,6 @@ jobs:
       run:
         shell: pwsh
     env:
-      AVXVER: ${{ matrix.avx }}
       PCKGVER: ${{ inputs.version }}
 
     steps:
@@ -111,15 +109,22 @@ jobs:
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
             # Define common variables
-            AVX_SUPPORT="${{ matrix.avx }}"
             VERSION="${{ inputs.version }}"
 
-            # 1. Set platform-specific variables (Linux only for ROCm)
-            PLATFORM="linux"
-            PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only.tar.gz"
-            PIP_PATH="portable_env/bin/python -m pip"
-            PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
-            rm start_macos.sh start_windows.bat
+            # 1. Set platform-specific variables
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                PLATFORM="windows"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-pc-windows-msvc-install_only.tar.gz"
+                PIP_PATH="portable_env/python.exe -m pip"
+                PACKAGES_PATH="portable_env/Lib/site-packages"
+                rm start_linux.sh start_macos.sh
+            else
+                PLATFORM="linux"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only.tar.gz"
+                PIP_PATH="portable_env/bin/python -m pip"
+                PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
+                rm start_macos.sh start_windows.bat
+            fi
 
             # 2. Download and extract Python
             cd ..
@@ -128,13 +133,8 @@ jobs:
             tar -xzf python-build.tar.gz
             mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
 
-            # 3. Prepare requirements file based on AVX
-            if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
-                BASE_REQ_FILE="requirements/portable/requirements_amd.txt"
-            else
-                BASE_REQ_FILE="requirements/portable/requirements_amd_noavx2.txt"
-            fi
-            REQ_FILE="$BASE_REQ_FILE"
+            # 3. Prepare requirements file
+            REQ_FILE="requirements/portable/requirements_amd.txt"
 
             cd "text-generation-webui-${VERSION_CLEAN}"
 
@@ -150,7 +150,11 @@ jobs:
             ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-rocm.zip"
             echo "Creating archive: $ZIP_NAME"
 
-            zip -r "$ZIP_NAME" "text-generation-webui-${VERSION_CLEAN}"
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
+            else
+                zip -r "$ZIP_NAME" "text-generation-webui-${VERSION_CLEAN}"
+            fi
 
       - name: Upload files to a GitHub release
         id: upload-release
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index eda684b158..e8b75b5b62 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -59,7 +59,6 @@ jobs:
           $matrix = @{
               'os' = @('ubuntu-22.04', 'windows-2022')
               'pyver' = @("3.13")
-              'avx' = @("AVX2")
           }
 
           if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})}
@@ -74,7 +73,7 @@ jobs:
           Write-Output ('matrix=' + $matrixOut) >> $env:GITHUB_OUTPUT
 
   build_wheels:
-    name: ${{ matrix.os }} ${{ matrix.pyver }} CPU ${{ matrix.avx }}
+    name: ${{ matrix.os }} ${{ matrix.pyver }}
     needs: define_matrix
     runs-on: ${{ matrix.os }}
     strategy:
@@ -83,7 +82,6 @@ jobs:
       run:
         shell: pwsh
     env:
-      AVXVER: ${{ matrix.avx }}
       PCKGVER: ${{ inputs.version }}
 
     steps:
@@ -111,7 +109,6 @@ jobs:
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
             # Define common variables
-            AVX_SUPPORT="${{ matrix.avx }}"
             VERSION="${{ inputs.version }}"
 
             # 1. Set platform-specific variables
@@ -136,13 +133,8 @@ jobs:
             tar -xzf python-build.tar.gz
             mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
 
-            # 3. Prepare requirements file based on AVX
-            if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
-                BASE_REQ_FILE="requirements/portable/requirements_vulkan.txt"
-            else
-                BASE_REQ_FILE="requirements/portable/requirements_vulkan_noavx2.txt"
-            fi
-            REQ_FILE="$BASE_REQ_FILE"
+            # 3. Prepare requirements file
+            REQ_FILE="requirements/portable/requirements_vulkan.txt"
 
             cd "text-generation-webui-${VERSION_CLEAN}"
 
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 1f6e4da30d..a6aec7519e 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -59,7 +59,6 @@ jobs:
           $matrix = @{
               'os' = @('ubuntu-22.04', 'windows-2022', 'macos-14')
               'pyver' = @("3.13")
-              'avx' = @("AVX2")
           }
 
           if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})}
@@ -74,7 +73,7 @@ jobs:
           Write-Output ('matrix=' + $matrixOut) >> $env:GITHUB_OUTPUT
 
   build_wheels:
-    name: ${{ matrix.os }} ${{ matrix.pyver }} CPU ${{ matrix.avx }}
+    name: ${{ matrix.os }} ${{ matrix.pyver }}
     needs: define_matrix
     runs-on: ${{ matrix.os }}
     strategy:
@@ -83,7 +82,6 @@ jobs:
       run:
         shell: pwsh
     env:
-      AVXVER: ${{ matrix.avx }}
       PCKGVER: ${{ inputs.version }}
 
     steps:
@@ -111,7 +109,6 @@ jobs:
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
             # Define common variables
-            AVX_SUPPORT="${{ matrix.avx }}"
             VERSION="${{ inputs.version }}"
             OS_TYPE="${{ matrix.os }}"
 
@@ -151,7 +148,7 @@ jobs:
             tar -xzf python-build.tar.gz
             mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
 
-            # 3. Prepare requirements file based on platform and AVX
+            # 3. Prepare requirements file based on platform
             cd "text-generation-webui-${VERSION_CLEAN}"
 
             # Select requirements file based on platform
@@ -162,12 +159,7 @@ jobs:
                     REQ_FILE="requirements/portable/requirements_apple_silicon.txt"
                 fi
             else
-                # For Windows and Linux, check AVX support
-                if [[ "$AVX_SUPPORT" == "AVX2" ]]; then
-                    REQ_FILE="requirements/portable/requirements_cpu_only.txt"
-                else
-                    REQ_FILE="requirements/portable/requirements_cpu_only_noavx2.txt"
-                fi
+                REQ_FILE="requirements/portable/requirements_cpu_only.txt"
             fi
 
             echo "Using requirements file: $REQ_FILE"
diff --git a/README.md b/README.md
index c71fc565a0..2869aaec4a 100644
--- a/README.md
+++ b/README.md
@@ -171,16 +171,13 @@ pip install -r requirements/full/
 
 Requirements file to use:
 
-| GPU | CPU | requirements file to use |
-|--------|---------|---------|
-| NVIDIA | has AVX2 | `requirements.txt` |
-| NVIDIA | no AVX2 | `requirements_noavx2.txt` |
-| AMD | has AVX2 | `requirements_amd.txt` |
-| AMD | no AVX2 | `requirements_amd_noavx2.txt` |
-| CPU only | has AVX2 | `requirements_cpu_only.txt` |
-| CPU only | no AVX2 | `requirements_cpu_only_noavx2.txt` |
-| Apple | Intel | `requirements_apple_intel.txt` |
-| Apple | Apple Silicon | `requirements_apple_silicon.txt` |
+| GPU | requirements file to use |
+|--------|---------|
+| NVIDIA | `requirements.txt` |
+| AMD | `requirements_amd.txt` |
+| CPU only | `requirements_cpu_only.txt` |
+| Apple Intel | `requirements_apple_intel.txt` |
+| Apple Silicon | `requirements_apple_silicon.txt` |
 
 ### Start the web UI
 
diff --git a/one_click.py b/one_click.py
index 46e86a0cfc..dad9e8074d 100644
--- a/one_click.py
+++ b/one_click.py
@@ -65,24 +65,6 @@ def is_installed():
         return os.path.isdir(conda_env_path)
 
 
-def cpu_has_avx2():
-    try:
-        import cpuinfo
-        info = cpuinfo.get_cpu_info()
-        return 'avx2' in info['flags']
-    except:
-        return True
-
-
-def cpu_has_amx():
-    try:
-        import cpuinfo
-        info = cpuinfo.get_cpu_info()
-        return 'amx' in info['flags']
-    except:
-        return True
-
-
 def load_state():
     """Load installer state from JSON file"""
     if os.path.exists(state_file):
@@ -172,13 +154,13 @@ def get_requirements_file(gpu_choice):
     requirements_base = os.path.join("requirements", "full")
 
     if gpu_choice == "NVIDIA_CUDA128":
-        file_name = f"requirements{'_noavx2' if not cpu_has_avx2() else ''}.txt"
+        file_name = "requirements.txt"
     elif gpu_choice == "AMD":
-        file_name = f"requirements_amd{'_noavx2' if not cpu_has_avx2() else ''}.txt"
+        file_name = "requirements_amd.txt"
     elif gpu_choice == "APPLE":
         file_name = f"requirements_apple_{'intel' if is_x86_64() else 'silicon'}.txt"
     elif gpu_choice in ["INTEL", "NONE"]:
-        file_name = f"requirements_cpu_only{'_noavx2' if not cpu_has_avx2() else ''}.txt"
+        file_name = "requirements_cpu_only.txt"
     else:
         raise ValueError(f"Unknown GPU choice: {gpu_choice}")
 
@@ -327,7 +309,7 @@ def install_webui():
     elif any((is_windows(), is_linux())) and gpu_choice == "NVIDIA_CUDA128":
         print("CUDA: 12.8")
 
-    # No PyTorch for AMD on Windows (?)
+    # No PyTorch for AMD on Windows
     elif is_windows() and gpu_choice == "AMD":
         print("PyTorch setup on Windows is not implemented yet. Exiting...")
         sys.exit(1)
@@ -335,7 +317,7 @@ def install_webui():
     # Install Git and then Pytorch
     print_big_message("Installing PyTorch.")
     install_pytorch = get_pytorch_install_command(gpu_choice)
-    run_cmd(f"conda install -y ninja git && {install_pytorch} && python -m pip install py-cpuinfo==9.0.0", assert_success=True, environment=True)
+    run_cmd(f"conda install -y ninja git && {install_pytorch}", assert_success=True, environment=True)
 
     if gpu_choice == "INTEL":
         # Install oneAPI dependencies via conda
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index bed2aeb4b6..b948f2f51e 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,8 +42,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.22/exllamav3-0.0.22+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.22/exllamav3-0.0.22+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 8824357c6a..3754b123cf 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -40,7 +40,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.4.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_amd_noavx2.txt b/requirements/full/requirements_amd_noavx2.txt
deleted file mode 100644
index ca7f0fe9f7..0000000000
--- a/requirements/full/requirements_amd_noavx2.txt
+++ /dev/null
@@ -1,46 +0,0 @@
-accelerate==1.12.*
-audioop-lts<1.0; python_version >= "3.13"
-colorama
-datasets
-diffusers==0.36.*
-einops
-fastapi==0.112.4
-html2text==2025.4.15
-huggingface-hub==1.5.*
-jinja2==3.1.6
-markdown
-numpy==2.2.*
-pandas
-peft==0.18.*
-Pillow>=9.5.0
-psutil
-pydantic==2.11.0
-PyPDF2==3.0.1
-python-docx==1.1.2
-pyyaml
-requests
-rich
-safetensors==0.7.*
-scipy
-sentencepiece
-tensorboard
-torchao==0.15.*
-transformers==5.2.*
-triton-windows==3.5.1.post24; platform_system == "Windows"
-tqdm
-wandb
-
-# Gradio
-gradio==4.37.*
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
-
-# API
-flask_cloudflared==0.0.14
-sse-starlette==1.6.5
-tiktoken
-
-# AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.4.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index b9f1f1f293..eb3a3a1154 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -40,5 +40,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 71da9a67b7..13df359b9d 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -40,5 +40,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index cd70cb7fec..f7ffbca7af 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -39,6 +39,6 @@ flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
-# llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+# llama.cpp (CPU only)
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_cpu_only_noavx2.txt b/requirements/full/requirements_cpu_only_noavx2.txt
deleted file mode 100644
index 1a128289dc..0000000000
--- a/requirements/full/requirements_cpu_only_noavx2.txt
+++ /dev/null
@@ -1,44 +0,0 @@
-accelerate==1.12.*
-audioop-lts<1.0; python_version >= "3.13"
-colorama
-datasets
-diffusers==0.36.*
-einops
-fastapi==0.112.4
-html2text==2025.4.15
-huggingface-hub==1.5.*
-jinja2==3.1.6
-markdown
-numpy==2.2.*
-pandas
-peft==0.18.*
-Pillow>=9.5.0
-psutil
-pydantic==2.11.0
-PyPDF2==3.0.1
-python-docx==1.1.2
-pyyaml
-requests
-rich
-safetensors==0.7.*
-scipy
-sentencepiece
-tensorboard
-torchao==0.15.*
-transformers==5.2.*
-triton-windows==3.5.1.post24; platform_system == "Windows"
-tqdm
-wandb
-
-# Gradio
-gradio==4.37.*
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
-
-# API
-flask_cloudflared==0.0.14
-sse-starlette==1.6.5
-tiktoken
-
-# llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/full/requirements_noavx2.txt b/requirements/full/requirements_noavx2.txt
deleted file mode 100644
index e2a9fd580b..0000000000
--- a/requirements/full/requirements_noavx2.txt
+++ /dev/null
@@ -1,53 +0,0 @@
-accelerate==1.12.*
-audioop-lts<1.0; python_version >= "3.13"
-bitsandbytes==0.49.*
-colorama
-datasets
-diffusers==0.36.*
-einops
-fastapi==0.112.4
-flash-linear-attention==0.4.*
-html2text==2025.4.15
-huggingface-hub==1.5.*
-jinja2==3.1.6
-markdown
-numpy==2.2.*
-pandas
-peft==0.18.*
-Pillow>=9.5.0
-psutil
-pydantic==2.11.0
-PyPDF2==3.0.1
-python-docx==1.1.2
-pyyaml
-requests
-rich
-safetensors==0.7.*
-scipy
-sentencepiece
-tensorboard
-torchao==0.15.*
-transformers==5.2.*
-triton-windows==3.5.1.post24; platform_system == "Windows"
-tqdm
-wandb
-
-# Gradio
-gradio==4.37.*
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
-
-# API
-flask_cloudflared==0.0.14
-sse-starlette==1.6.5
-tiktoken
-
-# CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.22/exllamav3-0.0.22+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.22/exllamav3-0.0.22+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
-https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index b9576255bf..30e872aa15 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index dc30f421fa..bd78495077 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd_noavx2.txt b/requirements/portable/requirements_amd_noavx2.txt
deleted file mode 100644
index 3474642614..0000000000
--- a/requirements/portable/requirements_amd_noavx2.txt
+++ /dev/null
@@ -1,27 +0,0 @@
-audioop-lts<1.0; python_version >= "3.13"
-fastapi==0.112.4
-html2text==2025.4.15
-huggingface-hub==1.5.*
-jinja2==3.1.6
-markdown
-numpy==2.2.*
-pydantic==2.11.0
-PyPDF2==3.0.1
-python-docx==1.1.2
-pyyaml
-requests
-rich
-tqdm
-
-# Gradio
-gradio==4.37.*
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
-
-# API
-flask_cloudflared==0.0.14
-sse-starlette==1.6.5
-tiktoken
-
-# AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 6bae46d691..274b3a03e5 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,5 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 2f9e4dc5d9..f67c18518c 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,5 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 739ac26fa8..6b48c79551 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -22,6 +22,6 @@ flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
-# llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+# llama.cpp (CPU only)
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cpu_only_noavx2.txt b/requirements/portable/requirements_cpu_only_noavx2.txt
deleted file mode 100644
index e096ed1d0c..0000000000
--- a/requirements/portable/requirements_cpu_only_noavx2.txt
+++ /dev/null
@@ -1,27 +0,0 @@
-audioop-lts<1.0; python_version >= "3.13"
-fastapi==0.112.4
-html2text==2025.4.15
-huggingface-hub==1.5.*
-jinja2==3.1.6
-markdown
-numpy==2.2.*
-pydantic==2.11.0
-PyPDF2==3.0.1
-python-docx==1.1.2
-pyyaml
-requests
-rich
-tqdm
-
-# Gradio
-gradio==4.37.*
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
-
-# API
-flask_cloudflared==0.0.14
-sse-starlette==1.6.5
-tiktoken
-
-# llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_noavx2.txt b/requirements/portable/requirements_noavx2.txt
deleted file mode 100644
index 12e24ab49b..0000000000
--- a/requirements/portable/requirements_noavx2.txt
+++ /dev/null
@@ -1,27 +0,0 @@
-audioop-lts<1.0; python_version >= "3.13"
-fastapi==0.112.4
-html2text==2025.4.15
-huggingface-hub==1.5.*
-jinja2==3.1.6
-markdown
-numpy==2.2.*
-pydantic==2.11.0
-PyPDF2==3.0.1
-python-docx==1.1.2
-pyyaml
-requests
-rich
-tqdm
-
-# Gradio
-gradio==4.37.*
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
-
-# API
-flask_cloudflared==0.0.14
-sse-starlette==1.6.5
-tiktoken
-
-# CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index c2a66a69c9..ad0592812f 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan_noavx2.txt b/requirements/portable/requirements_vulkan_noavx2.txt
deleted file mode 100644
index 79b632bb49..0000000000
--- a/requirements/portable/requirements_vulkan_noavx2.txt
+++ /dev/null
@@ -1,27 +0,0 @@
-audioop-lts<1.0; python_version >= "3.13"
-fastapi==0.112.4
-html2text==2025.4.15
-huggingface-hub==1.5.*
-jinja2==3.1.6
-markdown
-numpy==2.2.*
-pydantic==2.11.0
-PyPDF2==3.0.1
-python-docx==1.1.2
-pyyaml
-requests
-rich
-tqdm
-
-# Gradio
-gradio==4.37.*
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
-
-# API
-flask_cloudflared==0.0.14
-sse-starlette==1.6.5
-tiktoken
-
-# CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 5fd79b23d163cefd1747634ea9575488b9be1833 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 3 Mar 2026 15:33:15 -0800
Subject: [PATCH 1170/1701] Add CUDA 13.1 portable builds

---
 .../workflows/build-portable-release-cuda.yml | 10 ++++---
 .../portable/requirements_cuda131.txt         | 27 +++++++++++++++++++
 2 files changed, 34 insertions(+), 3 deletions(-)
 create mode 100644 requirements/portable/requirements_cuda131.txt

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index a10e8ec4d0..87ab7f9ee9 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -59,7 +59,7 @@ jobs:
           $matrix = @{
               'os' = @('ubuntu-22.04', 'windows-2022')
               'pyver' = @("3.13")
-              'cuda' = @("12.4")
+              'cuda' = @("12.4", "13.1")
           }
 
           if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})}
@@ -135,9 +135,13 @@ jobs:
             tar -xzf python-build.tar.gz
             mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
 
-            # 3. Prepare requirements file
+            # 3. Prepare requirements file based on CUDA version
             cd "text-generation-webui-${VERSION_CLEAN}"
-            REQ_FILE="requirements/portable/requirements.txt"
+            if [[ "$CUDA_VERSION" == "13.1" ]]; then
+                REQ_FILE="requirements/portable/requirements_cuda131.txt"
+            else
+                REQ_FILE="requirements/portable/requirements.txt"
+            fi
 
             # 4. Install packages
             echo "Installing Python packages from $REQ_FILE..."
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
new file mode 100644
index 0000000000..ead7a2b5bc
--- /dev/null
+++ b/requirements/portable/requirements_cuda131.txt
@@ -0,0 +1,27 @@
+audioop-lts<1.0; python_version >= "3.13"
+fastapi==0.112.4
+html2text==2025.4.15
+huggingface-hub==1.5.*
+jinja2==3.1.6
+markdown
+numpy==2.2.*
+pydantic==2.11.0
+PyPDF2==3.0.1
+python-docx==1.1.2
+pyyaml
+requests
+rich
+tqdm
+
+# Gradio
+gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+
+# API
+flask_cloudflared==0.0.14
+sse-starlette==1.6.5
+tiktoken
+
+# CUDA wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From fbd2acfa1959a2edda01f17f555366f61115030c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 3 Mar 2026 16:13:04 -0800
Subject: [PATCH 1171/1701] Remove triton-windows from non-CUDA requirements

---
 requirements/full/requirements_amd.txt           | 1 -
 requirements/full/requirements_apple_intel.txt   | 1 -
 requirements/full/requirements_apple_silicon.txt | 1 -
 requirements/full/requirements_cpu_only.txt      | 1 -
 requirements/full/requirements_nowheels.txt      | 1 -
 5 files changed, 5 deletions(-)

diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 3754b123cf..b5a9c1e415 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -26,7 +26,6 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 transformers==5.2.*
-triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index eb3a3a1154..04230acf23 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -26,7 +26,6 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 transformers==5.2.*
-triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 13df359b9d..0966693e7b 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -26,7 +26,6 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 transformers==5.2.*
-triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index f7ffbca7af..16db61775d 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -26,7 +26,6 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 transformers==5.2.*
-triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index e70bec7e9e..a5d45e6e32 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -26,7 +26,6 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 transformers==5.2.*
-triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
 

From 9c604628a0dc14ced3d61625471023134622aef0 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Tue, 3 Mar 2026 21:41:46 -0300
Subject: [PATCH 1172/1701] Bump flask-cloudflared from 0.0.14 to 0.0.15 in
 /requirements/portable (#7382)

---
 requirements/portable/requirements.txt               | 2 +-
 requirements/portable/requirements_amd.txt           | 2 +-
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 2 +-
 requirements/portable/requirements_cuda131.txt       | 2 +-
 requirements/portable/requirements_nowheels.txt      | 2 +-
 requirements/portable/requirements_vulkan.txt        | 2 +-
 8 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 30e872aa15..3c36838538 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -18,7 +18,7 @@ gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
-flask_cloudflared==0.0.14
+flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
 
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index bd78495077..2d85188db6 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -18,7 +18,7 @@ gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
-flask_cloudflared==0.0.14
+flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
 
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 274b3a03e5..8161a5f5b0 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -18,7 +18,7 @@ gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
-flask_cloudflared==0.0.14
+flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
 
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index f67c18518c..dc9369f91e 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -18,7 +18,7 @@ gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
-flask_cloudflared==0.0.14
+flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
 
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 6b48c79551..f08e5112ec 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -18,7 +18,7 @@ gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
-flask_cloudflared==0.0.14
+flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
 
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index ead7a2b5bc..724e989398 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -18,7 +18,7 @@ gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
-flask_cloudflared==0.0.14
+flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
 
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 347f97f69d..3418384162 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -18,6 +18,6 @@ gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
-flask_cloudflared==0.0.14
+flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index ad0592812f..b1a74998e7 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -18,7 +18,7 @@ gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
-flask_cloudflared==0.0.14
+flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
 

From 3519890c8ebf5fd73824c1ce90a06e051264544a Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Tue, 3 Mar 2026 21:41:51 -0300
Subject: [PATCH 1173/1701] Bump flask-cloudflared from 0.0.14 to 0.0.15 in
 /requirements/full (#7380)

---
 requirements/full/requirements.txt               | 2 +-
 requirements/full/requirements_amd.txt           | 2 +-
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_cpu_only.txt      | 2 +-
 requirements/full/requirements_nowheels.txt      | 2 +-
 6 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index b948f2f51e..ca3d5c5475 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -37,7 +37,7 @@ gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
-flask_cloudflared==0.0.14
+flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
 
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index b5a9c1e415..788e6b4fd9 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -34,7 +34,7 @@ gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
-flask_cloudflared==0.0.14
+flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
 
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 04230acf23..601e7e8ac4 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -34,7 +34,7 @@ gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
-flask_cloudflared==0.0.14
+flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
 
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 0966693e7b..91d49faa07 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -34,7 +34,7 @@ gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
-flask_cloudflared==0.0.14
+flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
 
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 16db61775d..7114335608 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -34,7 +34,7 @@ gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
-flask_cloudflared==0.0.14
+flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
 
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index a5d45e6e32..63b84719de 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -34,6 +34,6 @@ gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
-flask_cloudflared==0.0.14
+flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken

From e9f22813e42635110bc99bd93536e48cefc2411a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 3 Mar 2026 16:51:27 -0800
Subject: [PATCH 1174/1701] Replace gradio with my gradio 4.37.2 fork

---
 requirements/full/requirements.txt                   | 2 +-
 requirements/full/requirements_amd.txt               | 2 +-
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 2 +-
 requirements/full/requirements_nowheels.txt          | 2 +-
 requirements/portable/requirements.txt               | 2 +-
 requirements/portable/requirements_amd.txt           | 2 +-
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 2 +-
 requirements/portable/requirements_cuda131.txt       | 2 +-
 requirements/portable/requirements_nowheels.txt      | 2 +-
 requirements/portable/requirements_vulkan.txt        | 2 +-
 14 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index ca3d5c5475..bd72415c62 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -33,7 +33,7 @@ tqdm
 wandb
 
 # Gradio
-gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 788e6b4fd9..c428f012c4 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -30,7 +30,7 @@ tqdm
 wandb
 
 # Gradio
-gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 601e7e8ac4..c90e662aaa 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -30,7 +30,7 @@ tqdm
 wandb
 
 # Gradio
-gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 91d49faa07..a7bfaad895 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -30,7 +30,7 @@ tqdm
 wandb
 
 # Gradio
-gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 7114335608..d90f127e9e 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -30,7 +30,7 @@ tqdm
 wandb
 
 # Gradio
-gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 63b84719de..522d40b874 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -30,7 +30,7 @@ tqdm
 wandb
 
 # Gradio
-gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 3c36838538..17aa113706 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,7 +14,7 @@ rich
 tqdm
 
 # Gradio
-gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 2d85188db6..68c7ec1426 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,7 +14,7 @@ rich
 tqdm
 
 # Gradio
-gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 8161a5f5b0..5c29a961fe 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,7 +14,7 @@ rich
 tqdm
 
 # Gradio
-gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index dc9369f91e..89954ac49e 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,7 +14,7 @@ rich
 tqdm
 
 # Gradio
-gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index f08e5112ec..55dcc65040 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,7 +14,7 @@ rich
 tqdm
 
 # Gradio
-gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 724e989398..f71568fce0 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,7 +14,7 @@ rich
 tqdm
 
 # Gradio
-gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 3418384162..841cff59fb 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,7 +14,7 @@ rich
 tqdm
 
 # Gradio
-gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index b1a74998e7..65ca609894 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,7 +14,7 @@ rich
 tqdm
 
 # Gradio
-gradio==4.37.*
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
 # API

From 2260e530c9066bce741b545d7ecdcc8c0427efb2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 3 Mar 2026 17:17:36 -0800
Subject: [PATCH 1175/1701] Remove gradio monkey-patches (moved to gradio fork)

---
 modules/block_requests.py | 96 --------------------------------------
 modules/gradio_hijack.py  | 97 ---------------------------------------
 server.py                 | 58 +++++++++++++----------
 3 files changed, 33 insertions(+), 218 deletions(-)
 delete mode 100644 modules/block_requests.py
 delete mode 100644 modules/gradio_hijack.py

diff --git a/modules/block_requests.py b/modules/block_requests.py
deleted file mode 100644
index 911e41d9a0..0000000000
--- a/modules/block_requests.py
+++ /dev/null
@@ -1,96 +0,0 @@
-import builtins
-import io
-import re
-
-import requests
-
-from modules import shared, ui
-from modules.logging_colors import logger
-
-original_open = open
-original_get = requests.get
-original_print = print
-
-
-class RequestBlocker:
-
-    def __enter__(self):
-        requests.get = my_get
-
-    def __exit__(self, exc_type, exc_value, traceback):
-        requests.get = original_get
-
-
-class OpenMonkeyPatch:
-
-    def __enter__(self):
-        builtins.open = my_open
-        builtins.print = my_print
-
-    def __exit__(self, exc_type, exc_value, traceback):
-        builtins.open = original_open
-        builtins.print = original_print
-
-
-def my_get(url, **kwargs):
-    logger.info('Unwanted HTTP request redirected to localhost :)')
-    kwargs.setdefault('allow_redirects', True)
-    return requests.api.request('get', 'http://127.0.0.1/', **kwargs)
-
-
-def my_open(*args, **kwargs):
-    filename = str(args[0])
-    if filename.endswith(('index.html', 'share.html')):
-        with original_open(*args, **kwargs) as f:
-            file_contents = f.read()
-
-        if len(args) > 1 and args[1] == 'rb':
-            file_contents = file_contents.decode('utf-8')
-
-        file_contents = file_contents.replace('\t\t', '')
-        file_contents = file_contents.replace('cdnjs.cloudflare.com', '127.0.0.1')
-        file_contents = file_contents.replace(
-            '',
-            '\n    '
-            '\n    '
-            '\n    '
-            '\n    '
-            '\n    '
-            '\n    '
-            '\n    '
-            '\n    '
-            '\n    '
-            '\n    '
-            f'\n    '
-            '\n    '
-            f'\n    '
-            '\n  '
-        )
-
-        file_contents = re.sub(
-            r'@media \(prefers-color-scheme: dark\) \{\s*body \{([^}]*)\}\s*\}',
-            r'body.dark {\1}',
-            file_contents,
-            flags=re.DOTALL
-        )
-
-        if len(args) > 1 and args[1] == 'rb':
-            file_contents = file_contents.encode('utf-8')
-            return io.BytesIO(file_contents)
-        else:
-            return io.StringIO(file_contents)
-
-    else:
-        return original_open(*args, **kwargs)
-
-
-def my_print(*args, **kwargs):
-    if len(args) > 0 and 'To create a public link, set `share=True`' in args[0]:
-        return
-    else:
-        if len(args) > 0 and 'Running on local URL' in args[0]:
-            args = list(args)
-            args[0] = f"\n{args[0].strip()}\n"
-            args = tuple(args)
-
-        original_print(*args, **kwargs)
diff --git a/modules/gradio_hijack.py b/modules/gradio_hijack.py
deleted file mode 100644
index 817da40c6c..0000000000
--- a/modules/gradio_hijack.py
+++ /dev/null
@@ -1,97 +0,0 @@
-'''
-Most of the code here was adapted from:
-https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14184
-'''
-
-import inspect
-import warnings
-from functools import wraps
-
-import gradio as gr
-import gradio.routes
-import gradio.utils
-from starlette.middleware.trustedhost import TrustedHostMiddleware
-
-from modules import shared
-
-orig_create_app = gradio.routes.App.create_app
-
-
-# Be strict about only approving access to localhost by default
-def create_app_with_trustedhost(*args, **kwargs):
-    app = orig_create_app(*args, **kwargs)
-
-    if not (shared.args.listen or shared.args.share):
-        app.add_middleware(
-            TrustedHostMiddleware,
-            allowed_hosts=["localhost", "127.0.0.1"]
-        )
-
-    return app
-
-
-gradio.routes.App.create_app = create_app_with_trustedhost
-gradio.utils.launch_counter = lambda: None
-
-
-class GradioDeprecationWarning(DeprecationWarning):
-    pass
-
-
-def repair(grclass):
-    if not getattr(grclass, 'EVENTS', None):
-        return
-
-    @wraps(grclass.__init__)
-    def __repaired_init__(self, *args, tooltip=None, source=None, original=grclass.__init__, **kwargs):
-        if source:
-            kwargs["sources"] = [source]
-
-        allowed_kwargs = inspect.signature(original).parameters
-        fixed_kwargs = {}
-        for k, v in kwargs.items():
-            if k in allowed_kwargs:
-                fixed_kwargs[k] = v
-            else:
-                warnings.warn(f"unexpected argument for {grclass.__name__}: {k}", GradioDeprecationWarning, stacklevel=2)
-
-        original(self, *args, **fixed_kwargs)
-
-        self.webui_tooltip = tooltip
-
-        for event in self.EVENTS:
-            replaced_event = getattr(self, str(event))
-
-            def fun(*xargs, _js=None, replaced_event=replaced_event, **xkwargs):
-                if _js:
-                    xkwargs['js'] = _js
-
-                return replaced_event(*xargs, **xkwargs)
-
-            setattr(self, str(event), fun)
-
-    grclass.__init__ = __repaired_init__
-    grclass.update = gr.update
-
-
-for component in set(gr.components.__all__ + gr.layouts.__all__):
-    repair(getattr(gr, component, None))
-
-
-class Dependency(gr.events.Dependency):
-    def __init__(self, *args, **kwargs):
-        super().__init__(*args, **kwargs)
-
-        def then(*xargs, _js=None, **xkwargs):
-            if _js:
-                xkwargs['js'] = _js
-
-            return original_then(*xargs, **xkwargs)
-
-        original_then = self.then
-        self.then = then
-
-
-gr.events.Dependency = Dependency
-
-gr.Box = gr.Group
diff --git a/server.py b/server.py
index e98ff153df..bd23f3681f 100644
--- a/server.py
+++ b/server.py
@@ -3,13 +3,7 @@
 import warnings
 from pathlib import Path
 
-# Monkey-patch HfFolder for gradio 4.x compatibility with huggingface-hub 1.x
-import huggingface_hub
-if not hasattr(huggingface_hub, 'HfFolder'):
-    huggingface_hub.HfFolder = type('HfFolder', (), {'get_token': staticmethod(huggingface_hub.get_token)})
-
 from modules import shared
-from modules.block_requests import OpenMonkeyPatch, RequestBlocker
 from modules.image_models import load_image_model
 from modules.logging_colors import logger
 from modules.prompts import load_prompt
@@ -32,9 +26,7 @@
 warnings.filterwarnings('ignore', category=UserWarning, message='The value passed into gr.Dropdown()')
 warnings.filterwarnings('ignore', category=UserWarning, message='Field "model_names" has conflict')
 
-with RequestBlocker():
-    from modules import gradio_hijack
-    import gradio as gr
+import gradio as gr
 
 import matplotlib
 
@@ -148,7 +140,24 @@ def create_interface():
     # Interface state elements
     shared.input_elements = ui.list_interface_input_elements()
 
-    with gr.Blocks(css=css, analytics_enabled=False, title=title, theme=ui.theme) as shared.gradio['interface']:
+    # Head HTML for font preloads, KaTeX, highlight.js, morphdom, and global JS
+    head_html = '\n'.join([
+        '',
+        '',
+        '',
+        '',
+        '',
+        '',
+        '',
+        '',
+        '',
+        '',
+        f'',
+        '',
+        f'',
+    ])
+
+    with gr.Blocks(css=css, analytics_enabled=False, title=title, theme=ui.theme, head=head_html) as shared.gradio['interface']:
 
         # Interface state
         shared.gradio['interface_state'] = gr.State({k: None for k in shared.input_elements})
@@ -234,21 +243,20 @@ def create_interface():
 
     # Launch the interface
     shared.gradio['interface'].queue()
-    with OpenMonkeyPatch():
-        shared.gradio['interface'].launch(
-            max_threads=64,
-            prevent_thread_lock=True,
-            share=shared.args.share,
-            server_name=None if not shared.args.listen else (shared.args.listen_host or '0.0.0.0'),
-            server_port=shared.args.listen_port,
-            inbrowser=shared.args.auto_launch,
-            auth=auth or None,
-            ssl_verify=False if (shared.args.ssl_keyfile or shared.args.ssl_certfile) else True,
-            ssl_keyfile=shared.args.ssl_keyfile,
-            ssl_certfile=shared.args.ssl_certfile,
-            root_path=shared.args.subpath,
-            allowed_paths=allowed_paths,
-        )
+    shared.gradio['interface'].launch(
+        max_threads=64,
+        prevent_thread_lock=True,
+        share=shared.args.share,
+        server_name=None if not shared.args.listen else (shared.args.listen_host or '0.0.0.0'),
+        server_port=shared.args.listen_port,
+        inbrowser=shared.args.auto_launch,
+        auth=auth or None,
+        ssl_verify=False if (shared.args.ssl_keyfile or shared.args.ssl_certfile) else True,
+        ssl_keyfile=shared.args.ssl_keyfile,
+        ssl_certfile=shared.args.ssl_certfile,
+        root_path=shared.args.subpath,
+        allowed_paths=allowed_paths,
+    )
 
 
 if __name__ == "__main__":

From c0bff831e3a4263e622397ba42d22d36ad182f76 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 3 Mar 2026 17:21:09 -0800
Subject: [PATCH 1176/1701] Update custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 14 files changed, 28 insertions(+), 28 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index bd72415c62..b2c11e67db 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -33,8 +33,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index c428f012c4..10a451c078 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index c90e662aaa..a03e2245c1 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index a7bfaad895..26c24c9920 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index d90f127e9e..b4686e7cd9 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 522d40b874..7351d37e72 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 17aa113706..22f5d87d94 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 68c7ec1426..a4912314e9 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 5c29a961fe..9ebcb931b9 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 89954ac49e..399432999a 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 55dcc65040..0ed3415af5 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index f71568fce0..53c20a9e3f 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 841cff59fb..4911cd3853 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 65ca609894..c70418093a 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.1-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From d584ede72e34bf19744459b8901488a6f77ca3d4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 3 Mar 2026 17:59:47 -0800
Subject: [PATCH 1177/1701] Avoid a circular import

---
 server.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/server.py b/server.py
index bd23f3681f..431565d141 100644
--- a/server.py
+++ b/server.py
@@ -3,7 +3,7 @@
 import warnings
 from pathlib import Path
 
-from modules import shared
+from modules import shared, ui  # ui must be imported early to avoid circular imports
 from modules.image_models import load_image_model
 from modules.logging_colors import logger
 from modules.prompts import load_prompt

From b3fd0d16e0fbccc23fee794e7e3efe08c6341c1d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 3 Mar 2026 18:11:42 -0800
Subject: [PATCH 1178/1701] Use a new gr.Headless component for efficient chat
 streaming

---
 modules/ui_chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index e00ddf5c39..aec9051ea5 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -19,7 +19,7 @@ def create_ui():
 
     shared.gradio['Chat input'] = gr.State()
     shared.gradio['history'] = gr.State({'internal': [], 'visible': [], 'metadata': {}})
-    shared.gradio['display'] = gr.JSON(value={}, visible=False)  # Hidden buffer
+    shared.gradio['display'] = gr.Headless(value={})
 
     with gr.Tab('Chat', elem_id='chat-tab'):
         with gr.Row(elem_id='past-chats-row', elem_classes=['pretty_scrollbar']):

From 866c48e55b9c3a01139111af0542eb7de05c48d9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 3 Mar 2026 18:41:47 -0800
Subject: [PATCH 1179/1701] Simplify dark theme handling using gradio fork's
 new dark_theme parameter

---
 js/global_scope_js.js | 21 ---------------------
 server.py             | 24 +++---------------------
 2 files changed, 3 insertions(+), 42 deletions(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 4d8c11215c..2b3f38e7ab 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -352,27 +352,6 @@ function handleMorphdomUpdate(data) {
   });
 }
 
-// Wait for Gradio to finish setting its styles, then force dark theme
-const observer = new MutationObserver((mutations) => {
-  mutations.forEach((mutation) => {
-    if (mutation.type === "attributes" &&
-        mutation.target.tagName === "GRADIO-APP" &&
-        mutation.attributeName === "style") {
-
-      // Gradio just set its styles, now force dark theme
-      document.body.classList.add("dark");
-      observer.disconnect();
-    }
-  });
-});
-
-// Start observing
-observer.observe(document.documentElement, {
-  attributes: true,
-  subtree: true,
-  attributeFilter: ["style"]
-});
-
 //------------------------------------------------
 // Suppress "Attempted to select a non-interactive or hidden tab" warning
 //------------------------------------------------
diff --git a/server.py b/server.py
index 431565d141..60dc8bbc62 100644
--- a/server.py
+++ b/server.py
@@ -157,7 +157,7 @@ def create_interface():
         f'',
     ])
 
-    with gr.Blocks(css=css, analytics_enabled=False, title=title, theme=ui.theme, head=head_html) as shared.gradio['interface']:
+    with gr.Blocks(css=css, analytics_enabled=False, title=title, theme=ui.theme, head=head_html, dark_theme=shared.settings['dark_theme']) as shared.gradio['interface']:
 
         # Interface state
         shared.gradio['interface_state'] = gr.State({k: None for k in shared.input_elements})
@@ -209,26 +209,8 @@ def create_interface():
             gradio('show_controls'),
             None,
             js=f"""(x) => {{
-                // Check if this is first visit or if localStorage is out of sync
-                const savedTheme = localStorage.getItem('theme');
-                const serverTheme = {str(shared.settings['dark_theme']).lower()} ? 'dark' : 'light';
-
-                // If no saved theme or mismatch with server on first load, use server setting
-                if (!savedTheme || !sessionStorage.getItem('theme_synced')) {{
-                    localStorage.setItem('theme', serverTheme);
-                    sessionStorage.setItem('theme_synced', 'true');
-                    if (serverTheme === 'dark') {{
-                        document.getElementsByTagName('body')[0].classList.add('dark');
-                    }} else {{
-                        document.getElementsByTagName('body')[0].classList.remove('dark');
-                    }}
-                }} else {{
-                    // Use localStorage for subsequent reloads
-                    if (savedTheme === 'dark') {{
-                        document.getElementsByTagName('body')[0].classList.add('dark');
-                    }} else {{
-                        document.getElementsByTagName('body')[0].classList.remove('dark');
-                    }}
+                if (!localStorage.getItem('theme')) {{
+                    localStorage.setItem('theme', {str(shared.settings['dark_theme']).lower()} ? 'dark' : 'light');
                 }}
                 {js}
                 {ui.show_controls_js}

From 8cbb7661a8eae2d6c5794318c19b8dec4534a3df Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 3 Mar 2026 18:51:24 -0800
Subject: [PATCH 1180/1701] Remove no longer needed dark theme localstorage
 code

---
 server.py | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/server.py b/server.py
index 60dc8bbc62..25776f2dc2 100644
--- a/server.py
+++ b/server.py
@@ -209,9 +209,6 @@ def create_interface():
             gradio('show_controls'),
             None,
             js=f"""(x) => {{
-                if (!localStorage.getItem('theme')) {{
-                    localStorage.setItem('theme', {str(shared.settings['dark_theme']).lower()} ? 'dark' : 'light');
-                }}
                 {js}
                 {ui.show_controls_js}
                 toggle_controls(x);

From 7d42b6900e9a053e0639566a9f0ede31e8812c97 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 05:47:59 -0800
Subject: [PATCH 1181/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 14 files changed, 28 insertions(+), 28 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index b2c11e67db..b816dc188f 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -33,8 +33,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 10a451c078..5c7f862221 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index a03e2245c1..6aea03eade 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 26c24c9920..ca723f902b 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index b4686e7cd9..a64fa61f8c 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 7351d37e72..ef77dfefc8 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 22f5d87d94..4487d6e47b 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index a4912314e9..98c11f8602 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 9ebcb931b9..5698b675ca 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 399432999a..e3cad9d1ba 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 0ed3415af5..13076cf48b 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 53c20a9e3f..a2ade3be33 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 4911cd3853..616982368d 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index c70418093a..55ffe74d80 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio-4.37.2+custom.2-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.2-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From 11dc6fdfce9d838800e9c0c85dfa37dd809ccc27 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 06:04:33 -0800
Subject: [PATCH 1182/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 14 files changed, 28 insertions(+), 28 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index b816dc188f..3a44aff6c8 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -33,8 +33,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 5c7f862221..3dc6744d0c 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 6aea03eade..ab89eb5739 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index ca723f902b..f526304325 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index a64fa61f8c..d2d0aaa015 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index ef77dfefc8..4ee86c2c3c 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 4487d6e47b..088a0a7b3a 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 98c11f8602..a071ac17d7 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 5698b675ca..82c10e07b9 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index e3cad9d1ba..389e1001cf 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 13076cf48b..ec1a896d15 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index a2ade3be33..70d032cc33 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 616982368d..58439cadd7 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 55ffe74d80..78ec78b9db 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio-4.37.2+custom.3-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2%2Bcustom.3/gradio_client-1.0.2+custom.3-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From 8a3d8664012f1add9a486a9638ffe7fd5456faca Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 06:10:51 -0800
Subject: [PATCH 1183/1701] Fix temperature_last having no effect in llama.cpp
 server sampler order

---
 modules/llama_cpp_server.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index eb59baf9e1..baf474bdcb 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -119,9 +119,9 @@ def prepare_payload(self, state):
                     penalty_found = True
 
             # Move temperature to the end if temperature_last is true and temperature exists in the list
-            if state["temperature_last"] and "temperature" in samplers:
-                samplers.remove("temperature")
-                samplers.append("temperature")
+            if state["temperature_last"] and "temperature" in filtered_samplers:
+                filtered_samplers.remove("temperature")
+                filtered_samplers.append("temperature")
 
             payload["samplers"] = filtered_samplers
 

From f4d787ab8d9e6bdb84695be347d3858fe73b6aa7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 06:37:50 -0800
Subject: [PATCH 1184/1701] Delegate GPU layer allocation to llama.cpp's --fit

---
 modules/llama_cpp_server.py |   6 +-
 modules/models_settings.py  | 132 ++++--------------------------------
 modules/shared.py           |   2 +-
 modules/ui_model_menu.py    |  16 ++---
 server.py                   |  15 ----
 5 files changed, 26 insertions(+), 145 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index baf474bdcb..4bf57ee873 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -329,7 +329,6 @@ def _start_server(self):
             self.server_path,
             "--model", self.model_path,
             "--ctx-size", str(shared.args.ctx_size),
-            "--gpu-layers", str(shared.args.gpu_layers),
             "--batch-size", str(shared.args.batch_size),
             "--ubatch-size", str(shared.args.ubatch_size),
             "--port", str(self.port),
@@ -337,6 +336,11 @@ def _start_server(self):
             "--flash-attn", "on",
         ]
 
+        if shared.args.gpu_layers > 0:
+            cmd += ["--gpu-layers", str(shared.args.gpu_layers), "--fit", "off"]
+        else:
+            cmd += ["--fit", "on"]
+
         if shared.args.threads > 0:
             cmd += ["--threads", str(shared.args.threads)]
         if shared.args.threads_batch > 0:
diff --git a/modules/models_settings.py b/modules/models_settings.py
index 15ff2830ae..3b28a800f5 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -1,7 +1,6 @@
 import functools
 import json
 import re
-import subprocess
 from math import floor
 from pathlib import Path
 
@@ -78,7 +77,7 @@ def get_model_metadata(model):
             elif k.endswith('rope.scaling.factor'):
                 model_settings['compress_pos_emb'] = metadata[k]
             elif k.endswith('.block_count'):
-                model_settings['gpu_layers'] = metadata[k] + 1
+                model_settings['gpu_layers'] = 0
                 model_settings['max_gpu_layers'] = metadata[k] + 1
 
         if 'tokenizer.chat_template' in metadata:
@@ -265,16 +264,18 @@ def apply_model_settings_to_state(model, state):
 
     # Handle GPU layers and VRAM update for llama.cpp
     if state['loader'] == 'llama.cpp' and 'gpu_layers' in model_settings:
-        vram_info, gpu_layers_update = update_gpu_layers_and_vram(
+        gpu_layers = model_settings['gpu_layers']  # 0 (auto) by default, or user-saved value
+        max_layers = model_settings.get('max_gpu_layers', 256)
+        state['gpu_layers'] = gr.update(value=gpu_layers, maximum=max_layers)
+
+        vram_info = update_gpu_layers_and_vram(
             state['loader'],
             model,
-            model_settings['gpu_layers'],
+            gpu_layers,
             state['ctx_size'],
             state['cache_type'],
-            auto_adjust=True
         )
 
-        state['gpu_layers'] = gpu_layers_update
         state['vram_info'] = vram_info
 
     return state
@@ -412,120 +413,13 @@ def estimate_vram(gguf_file, gpu_layers, ctx_size, cache_type):
     return vram
 
 
-def get_nvidia_vram(return_free=True):
-    """
-    Calculates VRAM statistics across all NVIDIA GPUs by parsing nvidia-smi output.
-
-    Args:
-        return_free (bool): If True, returns free VRAM. If False, returns total VRAM.
-
-    Returns:
-        int: Either the total free VRAM or total VRAM in MiB summed across all detected NVIDIA GPUs.
-             Returns -1 if nvidia-smi command fails (not found, error, etc.).
-             Returns 0 if nvidia-smi succeeds but no GPU memory info found.
-    """
-    try:
-        # Execute nvidia-smi command
-        result = subprocess.run(
-            ['nvidia-smi'],
-            capture_output=True,
-            text=True,
-            check=False
-        )
-
-        # Check if nvidia-smi returned an error
-        if result.returncode != 0:
-            return -1
-
-        # Parse the output for memory usage patterns
-        output = result.stdout
-
-        # Find memory usage like "XXXXMiB / YYYYMiB"
-        # Captures used and total memory for each GPU
-        matches = re.findall(r"(\d+)\s*MiB\s*/\s*(\d+)\s*MiB", output)
-
-        if not matches:
-            # No GPUs found in expected format
-            return 0
-
-        total_vram_mib = 0
-        total_free_vram_mib = 0
-
-        for used_mem_str, total_mem_str in matches:
-            try:
-                used_mib = int(used_mem_str)
-                total_mib = int(total_mem_str)
-                total_vram_mib += total_mib
-                total_free_vram_mib += (total_mib - used_mib)
-            except ValueError:
-                # Skip malformed entries
-                pass
-
-        # Return either free or total VRAM based on the flag
-        return total_free_vram_mib if return_free else total_vram_mib
-
-    except FileNotFoundError:
-        # nvidia-smi not found (likely no NVIDIA drivers installed)
-        return -1
-    except Exception:
-        # Handle any other unexpected exceptions
-        return -1
-
-
-def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type, auto_adjust=False, for_ui=True):
+def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type):
     """
-    Unified function to handle GPU layers and VRAM updates.
-
-    Args:
-        for_ui: If True, returns Gradio updates. If False, returns raw values.
-
-    Returns:
-        - If for_ui=True: (vram_info_update, gpu_layers_update) or just vram_info_update
-        - If for_ui=False: (vram_usage, adjusted_layers) or just vram_usage
+    Compute the estimated VRAM usage for the given GPU layers and return
+    an HTML string for the UI display.
     """
     if loader != 'llama.cpp' or model in ["None", None] or not model.endswith(".gguf"):
-        vram_info = "
            Estimated VRAM to load the model:
            "
-        if for_ui:
-            return (vram_info, gr.update()) if auto_adjust else vram_info
-        else:
-            return (0, gpu_layers) if auto_adjust else 0
-
-    # Get model settings including user preferences
-    model_settings = get_model_metadata(model)
+        return "
            Estimated VRAM to load the model:
            "
 
-    current_layers = gpu_layers
-    max_layers = model_settings.get('max_gpu_layers', 256)
-
-    if auto_adjust:
-        # Check if this is a user-saved setting
-        user_config = shared.user_config
-        model_regex = Path(model).name + '$'
-        has_user_setting = model_regex in user_config and 'gpu_layers' in user_config[model_regex]
-
-        if not has_user_setting:
-            # No user setting, auto-adjust from the maximum
-            current_layers = max_layers  # Start from max
-
-            # Auto-adjust based on available/total VRAM
-            # If a model is loaded and it's for the UI, use the total VRAM to avoid confusion
-            return_free = False if (for_ui and shared.model_name not in [None, 'None']) else True
-            available_vram = get_nvidia_vram(return_free=return_free)
-            if available_vram > 0:
-                tolerance = 577
-                while current_layers > 0 and estimate_vram(model, current_layers, ctx_size, cache_type) > available_vram - tolerance:
-                    current_layers -= 1
-
-    # Calculate VRAM with current layers
-    vram_usage = estimate_vram(model, current_layers, ctx_size, cache_type)
-
-    if for_ui:
-        vram_info = f"
            Estimated VRAM to load the model: {vram_usage:.0f} MiB
            "
-        if auto_adjust:
-            return vram_info, gr.update(value=current_layers, maximum=max_layers)
-        else:
-            return vram_info
-    else:
-        if auto_adjust:
-            return vram_usage, current_layers
-        else:
-            return vram_usage
+    vram_usage = estimate_vram(model, gpu_layers, ctx_size, cache_type)
+    return f"
            Estimated VRAM to load the model: {vram_usage:.0f} MiB
            "
diff --git a/modules/shared.py b/modules/shared.py
index 88e4b182ad..2fba530bd9 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -84,7 +84,7 @@
 
 # llama.cpp
 group = parser.add_argument_group('llama.cpp')
-group.add_argument('--gpu-layers', '--n-gpu-layers', type=int, default=256, metavar='N', help='Number of layers to offload to the GPU.')
+group.add_argument('--gpu-layers', '--n-gpu-layers', type=int, default=0, metavar='N', help='Number of layers to offload to the GPU. 0 means auto (llama.cpp decides via --fit).')
 group.add_argument('--cpu-moe', action='store_true', help='Move the experts to the CPU (for MoE models).')
 group.add_argument('--mmproj', type=str, default=None, help='Path to the mmproj file for vision models.')
 group.add_argument('--streaming-llm', action='store_true', help='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 86adc2297e..b9d731ee84 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -41,7 +41,7 @@ def create_ui():
                     gr.Markdown("## Main options")
                     with gr.Row():
                         with gr.Column():
-                            shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
+                            shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='0 = auto (llama.cpp decides via --fit). Set manually to override.')
                             shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. Common values: 4096, 8192, 16384, 32768, 65536, 131072.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
@@ -157,22 +157,22 @@ def create_event_handlers():
         handle_load_model_event_final, gradio('truncation_length', 'loader', 'interface_state'), gradio('truncation_length', 'filter_by_loader'), show_progress=False)
 
     shared.gradio['unload_model'].click(handle_unload_model_click, None, gradio('model_status'), show_progress=False).then(
-        partial(update_gpu_layers_and_vram, auto_adjust=True), gradio('loader', 'model_menu', 'gpu_layers', 'ctx_size', 'cache_type'), gradio('vram_info', 'gpu_layers'), show_progress=False)
+        update_gpu_layers_and_vram, gradio('loader', 'model_menu', 'gpu_layers', 'ctx_size', 'cache_type'), gradio('vram_info'), show_progress=False)
 
     shared.gradio['save_model_settings'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         save_model_settings, gradio('model_menu', 'interface_state'), gradio('model_status'), show_progress=False)
 
-    # For ctx_size and cache_type - auto-adjust GPU layers
+    # For ctx_size and cache_type - update VRAM display
     for param in ['ctx_size', 'cache_type']:
         shared.gradio[param].change(
-            partial(update_gpu_layers_and_vram, auto_adjust=True),
+            update_gpu_layers_and_vram,
             gradio('loader', 'model_menu', 'gpu_layers', 'ctx_size', 'cache_type'),
-            gradio('vram_info', 'gpu_layers'), show_progress=False)
+            gradio('vram_info'), show_progress=False)
 
     # For manual gpu_layers changes - only update VRAM
     shared.gradio['gpu_layers'].change(
-        partial(update_gpu_layers_and_vram, auto_adjust=False),
+        update_gpu_layers_and_vram,
         gradio('loader', 'model_menu', 'gpu_layers', 'ctx_size', 'cache_type'),
         gradio('vram_info'), show_progress=False)
 
@@ -386,8 +386,6 @@ def get_initial_vram_info():
             shared.args.gpu_layers,
             shared.args.ctx_size,
             shared.args.cache_type,
-            auto_adjust=False,
-            for_ui=True
         )
 
     return "
            Estimated VRAM to load the model:
            "
@@ -396,7 +394,7 @@ def get_initial_vram_info():
 def get_initial_gpu_layers_max():
     if shared.model_name != 'None' and shared.args.loader == 'llama.cpp':
         model_settings = get_model_metadata(shared.model_name)
-        return model_settings.get('max_gpu_layers', model_settings.get('gpu_layers', 256))
+        return model_settings.get('max_gpu_layers', 256)
 
     return 256
 
diff --git a/server.py b/server.py
index 25776f2dc2..304a0478cf 100644
--- a/server.py
+++ b/server.py
@@ -62,7 +62,6 @@
 from modules.models_settings import (
     get_fallback_settings,
     get_model_metadata,
-    update_gpu_layers_and_vram,
     update_model_parameters
 )
 from modules.shared import do_cmd_flags_warnings
@@ -315,20 +314,6 @@ def create_interface():
         model_settings = get_model_metadata(shared.model_name)
         update_model_parameters(model_settings, initial=True)  # hijack the command-line arguments
 
-        # Auto-adjust GPU layers if not provided by user and it's a llama.cpp model
-        if 'gpu_layers' not in shared.provided_arguments and shared.args.loader == 'llama.cpp' and 'gpu_layers' in model_settings:
-            vram_usage, adjusted_layers = update_gpu_layers_and_vram(
-                shared.args.loader,
-                shared.model_name,
-                model_settings['gpu_layers'],
-                shared.args.ctx_size,
-                shared.args.cache_type,
-                auto_adjust=True,
-                for_ui=False
-            )
-
-            shared.args.gpu_layers = adjusted_layers
-
         # Load the model
         shared.model, shared.tokenizer = load_model(shared.model_name)
         if shared.args.lora:

From f010aa16123b3ce8d91281a75868d74f7692dc91 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 06:43:37 -0800
Subject: [PATCH 1185/1701] Replace PyPDF2 with pymupdf for PDF text extraction

pymupdf produces cleaner text (e.g. no concatenated words in headers),
handles encrypted and malformed PDFs that PyPDF2 failed on, and
supports non-Latin scripts.
---
 modules/chat.py                                      | 10 ++++------
 requirements/full/requirements.txt                   |  2 +-
 requirements/full/requirements_amd.txt               |  2 +-
 requirements/full/requirements_apple_intel.txt       |  2 +-
 requirements/full/requirements_apple_silicon.txt     |  2 +-
 requirements/full/requirements_cpu_only.txt          |  2 +-
 requirements/full/requirements_nowheels.txt          |  2 +-
 requirements/portable/requirements.txt               |  2 +-
 requirements/portable/requirements_amd.txt           |  2 +-
 requirements/portable/requirements_apple_intel.txt   |  2 +-
 requirements/portable/requirements_apple_silicon.txt |  2 +-
 requirements/portable/requirements_cpu_only.txt      |  2 +-
 requirements/portable/requirements_cuda131.txt       |  2 +-
 requirements/portable/requirements_nowheels.txt      |  2 +-
 requirements/portable/requirements_vulkan.txt        |  2 +-
 15 files changed, 18 insertions(+), 20 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 42c0d46d5a..9ef8f5e9fb 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -657,15 +657,13 @@ def add_message_attachment(history, row_idx, file_path, is_user=True):
 
 def extract_pdf_text(pdf_path):
     """Extract text from a PDF file"""
-    import PyPDF2
+    import pymupdf
 
     text = ""
     try:
-        with open(pdf_path, 'rb') as file:
-            pdf_reader = PyPDF2.PdfReader(file)
-            for page_num in range(len(pdf_reader.pages)):
-                page = pdf_reader.pages[page_num]
-                text += page.extract_text() + "\n\n"
+        with pymupdf.open(pdf_path) as doc:
+            for page in doc:
+                text += page.get_text() + "\n\n"
 
         return text.strip()
     except Exception as e:
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 3a44aff6c8..085680859c 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -17,7 +17,7 @@ peft==0.18.*
 Pillow>=9.5.0
 psutil
 pydantic==2.11.0
-PyPDF2==3.0.1
+pymupdf==1.27.1
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 3dc6744d0c..a5c7b1b333 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -15,7 +15,7 @@ peft==0.18.*
 Pillow>=9.5.0
 psutil
 pydantic==2.11.0
-PyPDF2==3.0.1
+pymupdf==1.27.1
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index ab89eb5739..b91ef833a6 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -15,7 +15,7 @@ peft==0.18.*
 Pillow>=9.5.0
 psutil
 pydantic==2.11.0
-PyPDF2==3.0.1
+pymupdf==1.27.1
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index f526304325..b9c19714c4 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -15,7 +15,7 @@ peft==0.18.*
 Pillow>=9.5.0
 psutil
 pydantic==2.11.0
-PyPDF2==3.0.1
+pymupdf==1.27.1
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index d2d0aaa015..1421ad5f90 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -15,7 +15,7 @@ peft==0.18.*
 Pillow>=9.5.0
 psutil
 pydantic==2.11.0
-PyPDF2==3.0.1
+pymupdf==1.27.1
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 4ee86c2c3c..cf8c0ecb29 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -15,7 +15,7 @@ peft==0.18.*
 Pillow>=9.5.0
 psutil
 pydantic==2.11.0
-PyPDF2==3.0.1
+pymupdf==1.27.1
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 088a0a7b3a..1dc6abe32c 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -6,7 +6,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-PyPDF2==3.0.1
+pymupdf==1.27.1
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index a071ac17d7..09befbccf2 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -6,7 +6,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-PyPDF2==3.0.1
+pymupdf==1.27.1
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 82c10e07b9..e794a693b6 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -6,7 +6,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-PyPDF2==3.0.1
+pymupdf==1.27.1
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 389e1001cf..48eadd4734 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -6,7 +6,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-PyPDF2==3.0.1
+pymupdf==1.27.1
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index ec1a896d15..93b820fc30 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -6,7 +6,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-PyPDF2==3.0.1
+pymupdf==1.27.1
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 70d032cc33..6e2d05fd27 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -6,7 +6,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-PyPDF2==3.0.1
+pymupdf==1.27.1
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 58439cadd7..e6a9a0a456 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -6,7 +6,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-PyPDF2==3.0.1
+pymupdf==1.27.1
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 78ec78b9db..149bc82966 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -6,7 +6,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-PyPDF2==3.0.1
+pymupdf==1.27.1
 python-docx==1.1.2
 pyyaml
 requests

From 65de4c30c8a609122a3745da94138d7d0430720a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 09:41:29 -0800
Subject: [PATCH 1186/1701] Add adaptive-p sampler and n-gram speculative
 decoding support

---
 extensions/openai/typing.py |  4 +-
 modules/llama_cpp_server.py | 18 +++++++++
 modules/loaders.py          | 12 ++++++
 modules/presets.py          |  4 +-
 modules/sampler_hijack.py   | 81 ++++++++++++++++++++++++++++++++++++-
 modules/shared.py           |  6 +++
 modules/text_generation.py  |  2 +
 modules/ui.py               |  8 ++++
 modules/ui_model_menu.py    | 11 +++++
 modules/ui_parameters.py    |  2 +
 10 files changed, 145 insertions(+), 3 deletions(-)

diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 5fea2324b7..1548093ffa 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -22,6 +22,8 @@ class GenerationOptions(BaseModel):
     tfs: float = 1
     top_a: float = 0
     top_n_sigma: float = 0
+    adaptive_target: float = 0
+    adaptive_decay: float = 0.9
     dry_multiplier: float = 0
     dry_allowed_length: int = 2
     dry_base: float = 1.75
@@ -48,7 +50,7 @@ class GenerationOptions(BaseModel):
     static_cache: bool = False
     truncation_length: int = 0
     seed: int = -1
-    sampler_priority: List[str] | str | None = Field(default=None, description="List of samplers where the first items will appear first in the stack. Example: [\"top_k\", \"temperature\", \"top_p\"].")
+    sampler_priority: List[str] | str | None = Field(default=['repetition_penalty', 'presence_penalty', 'frequency_penalty', 'dry', 'top_n_sigma', 'temperature', 'dynamic_temperature', 'quadratic_sampling', 'top_k', 'top_p', 'typical_p', 'epsilon_cutoff', 'eta_cutoff', 'tfs', 'top_a', 'min_p', 'adaptive_p', 'mirostat', 'xtc', 'encoder_repetition_penalty', 'no_repeat_ngram'], description="List of samplers where the first items will appear first in the stack. Example: [\"top_k\", \"temperature\", \"top_p\"].")
     custom_token_bans: str = ""
     negative_prompt: str = ''
     dry_sequence_breakers: str = '"\\n", ":", "\\"", "*"'
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 4bf57ee873..0eb270c405 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -75,6 +75,8 @@ def prepare_payload(self, state):
             "top_p": state["top_p"],
             "min_p": state["min_p"],
             "top_n_sigma": state["top_n_sigma"] if state["top_n_sigma"] > 0 else -1,
+            "adaptive_target": state["adaptive_target"] if state["adaptive_target"] > 0 else -1,
+            "adaptive_decay": state["adaptive_decay"],
             "typical_p": state["typical_p"],
             "repeat_penalty": state["repetition_penalty"],
             "repeat_last_n": state["repetition_penalty_range"],
@@ -123,6 +125,12 @@ def prepare_payload(self, state):
                 filtered_samplers.remove("temperature")
                 filtered_samplers.append("temperature")
 
+            # adaptive-p replaces the default dist sampler; llama.cpp always
+            # places it at the end of the chain regardless of position, so we
+            # activate it based on the parameter value rather than sampler order.
+            if state.get("adaptive_target", 0) > 0:
+                filtered_samplers.append("adaptive-p")
+
             payload["samplers"] = filtered_samplers
 
         if state['custom_token_bans']:
@@ -391,6 +399,16 @@ def _start_server(self):
                 cmd += ["--device-draft", shared.args.device_draft]
             if shared.args.ctx_size_draft > 0:
                 cmd += ["--ctx-size-draft", str(shared.args.ctx_size_draft)]
+        if shared.args.spec_type != 'none':
+            cmd += ["--spec-type", shared.args.spec_type]
+            if shared.args.draft_max > 0:
+                cmd += ["--draft-max", str(shared.args.draft_max)]
+            if shared.args.spec_ngram_size_n != 12:
+                cmd += ["--spec-ngram-size-n", str(shared.args.spec_ngram_size_n)]
+            if shared.args.spec_ngram_size_m != 48:
+                cmd += ["--spec-ngram-size-m", str(shared.args.spec_ngram_size_m)]
+            if shared.args.spec_ngram_min_hits != 1:
+                cmd += ["--spec-ngram-min-hits", str(shared.args.spec_ngram_min_hits)]
         if shared.args.streaming_llm:
             cmd += ["--cache-reuse", "1"]
             cmd += ["--swa-full"]
diff --git a/modules/loaders.py b/modules/loaders.py
index 0bf3781b97..94c9c59337 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -28,6 +28,10 @@
         'gpu_layers_draft',
         'device_draft',
         'ctx_size_draft',
+        'spec_type',
+        'spec_ngram_size_n',
+        'spec_ngram_size_m',
+        'spec_ngram_min_hits',
         'speculative_decoding_accordion',
         'mmproj',
         'mmproj_accordion',
@@ -128,6 +132,8 @@ def transformers_samplers():
         'tfs',
         'top_a',
         'top_n_sigma',
+        'adaptive_target',
+        'adaptive_decay',
         'dry_multiplier',
         'dry_allowed_length',
         'dry_base',
@@ -183,6 +189,8 @@ def transformers_samplers():
         'tfs',
         'top_a',
         'top_n_sigma',
+        'adaptive_target',
+        'adaptive_decay',
         'dry_multiplier',
         'dry_allowed_length',
         'dry_base',
@@ -231,6 +239,8 @@ def transformers_samplers():
         'tfs',
         'top_a',
         'top_n_sigma',
+        'adaptive_target',
+        'adaptive_decay',
         'dry_multiplier',
         'dry_allowed_length',
         'dry_base',
@@ -327,6 +337,8 @@ def transformers_samplers():
         'xtc_threshold',
         'xtc_probability',
         'top_n_sigma',
+        'adaptive_target',
+        'adaptive_decay',
         'dry_multiplier',
         'dry_allowed_length',
         'dry_base',
diff --git a/modules/presets.py b/modules/presets.py
index 3eb1f5fc1f..16dfa6adb8 100644
--- a/modules/presets.py
+++ b/modules/presets.py
@@ -28,6 +28,8 @@ def default_preset():
         'tfs': 1,
         'top_a': 0,
         'top_n_sigma': 0,
+        'adaptive_target': 0,
+        'adaptive_decay': 0.9,
         'dry_multiplier': 0,
         'dry_allowed_length': 2,
         'dry_base': 1.75,
@@ -45,7 +47,7 @@ def default_preset():
         'do_sample': True,
         'dynamic_temperature': False,
         'temperature_last': False,
-        'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ndry\ntop_n_sigma\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nmirostat\nxtc\nencoder_repetition_penalty\nno_repeat_ngram',
+        'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ndry\ntop_n_sigma\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nadaptive_p\nmirostat\nxtc\nencoder_repetition_penalty\nno_repeat_ngram',
         'dry_sequence_breakers': '"\\n", ":", "\\"", "*"',
     }
 
diff --git a/modules/sampler_hijack.py b/modules/sampler_hijack.py
index dfdb691440..c419384be4 100644
--- a/modules/sampler_hijack.py
+++ b/modules/sampler_hijack.py
@@ -235,6 +235,73 @@ def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> to
         return scores
 
 
+class AdaptivePLogitsWarper(LogitsProcessor):
+    '''
+    Adaptive-p sampling. A stateful sampler that favors tokens near a target
+    probability, using an EMA-based control loop to adapt over time.
+
+    Matches the llama.cpp implementation from PR #17927.
+    '''
+
+    DISTRIBUTION_WIDTH = 0.3
+    PEAK_LOGIT_VALUE = 5.0
+    SHARPNESS = 10.0
+    INV_WIDTH = 1.0 / DISTRIBUTION_WIDTH
+
+    def __init__(self, adaptive_target, adaptive_decay, filter_value=-float("Inf"), min_tokens_to_keep=1):
+        self.target = adaptive_target
+        self.decay = min(adaptive_decay, 0.99)
+        self.filter_value = filter_value
+        self.min_tokens_to_keep = min_tokens_to_keep
+
+        # Initialize EMA at equilibrium (as if target was already achieved)
+        if self.decay < 1.0:
+            self.weighted_sum = self.target / (1.0 - self.decay)
+            self.total_weight = 1.0 / (1.0 - self.decay)
+        else:
+            self.weighted_sum = 0.0
+            self.total_weight = 0.0
+
+    def __call__(self, input_ids, scores):
+        logits = scores[0]
+
+        # Compute original probabilities (before transform)
+        probs = torch.softmax(logits, dim=-1)
+
+        # Compute adapted target using proportional control on the EMA
+        if self.total_weight > 0:
+            ema_avg = self.weighted_sum / self.total_weight
+        else:
+            ema_avg = self.target
+
+        adapted_target = max(0.0, min(1.0, 2.0 * self.target - ema_avg))
+
+        # Adaptive probability transform:
+        # quadratic near target for fine differentiation, transitioning
+        # to linear decay in the tails for proper suppression after softmax
+        dist = torch.abs((probs - adapted_target) * self.INV_WIDTH)
+        new_logits = self.PEAK_LOGIT_VALUE - self.SHARPNESS * dist * dist / (1.0 + dist)
+
+        # Preserve already-masked tokens (-inf logits from prior samplers)
+        new_logits = torch.where(torch.isfinite(logits), new_logits, logits)
+
+        # Softmax and sample from the transformed distribution
+        new_probs = torch.softmax(new_logits, dim=-1)
+        selected = torch.multinomial(new_probs, num_samples=1, replacement=True)
+
+        # Update EMA with the original probability of the selected token
+        original_prob = probs[selected[0]].item()
+        self.weighted_sum = original_prob + self.decay * self.weighted_sum
+        self.total_weight = 1.0 + self.decay * self.total_weight
+
+        # Mask all tokens except the selected one
+        indices_to_remove = torch.ones_like(scores[0], dtype=torch.bool)
+        indices_to_remove[selected[0]] = False
+        indices_to_remove = indices_to_remove.unsqueeze(0)
+        scores = scores.masked_fill(indices_to_remove, self.filter_value)
+        return scores
+
+
 # Exclude Top Choices (XTC)
 class XTCLogitsWarper(LogitsProcessor):
     def __init__(self, threshold: float, probability: float, filter_value: float = -float("Inf")):
@@ -575,6 +642,15 @@ def get_logits_processor_patch(self, **kwargs):
             )
         )
 
+    if generation_config.adaptive_target is not None and generation_config.adaptive_target > 0.0:
+        warpers_to_add.append(
+            AdaptivePLogitsWarper(
+                adaptive_target=generation_config.adaptive_target,
+                adaptive_decay=generation_config.adaptive_decay,
+                min_tokens_to_keep=min_tokens_to_keep
+            )
+        )
+
     if generation_config.xtc_probability is not None and generation_config.xtc_probability > 0:
         warpers_to_add.append(
             XTCLogitsWarper(
@@ -640,6 +716,7 @@ def get_logits_processor_patch(self, **kwargs):
         'TemperatureLogitsWarperCustom': 'temperature',
         'TopALogitsWarper': 'top_a',
         'TopNSigmaLogitsWarper': 'top_n_sigma',
+        'AdaptivePLogitsWarper': 'adaptive_p',
         'TopKLogitsWarper': 'top_k',
         'TopPLogitsWarper': 'top_p',
         'TypicalLogitsWarper': 'typical_p',
@@ -688,6 +765,8 @@ def generation_config_init_patch(self, **kwargs):
     self.tfs = kwargs.pop("tfs", 1.0)
     self.top_a = kwargs.pop("top_a", 0.0)
     self.top_n_sigma = kwargs.pop("top_n_sigma", 0.0)
+    self.adaptive_target = kwargs.pop("adaptive_target", 0.0)
+    self.adaptive_decay = kwargs.pop("adaptive_decay", 0.9)
     self.mirostat_mode = kwargs.pop("mirostat_mode", 0)
     self.mirostat_eta = kwargs.pop("mirostat_eta", 0.1)
     self.mirostat_tau = kwargs.pop("mirostat_tau", 5)
@@ -701,7 +780,7 @@ def generation_config_init_patch(self, **kwargs):
     self.xtc_threshold = kwargs.pop("xtc_threshold", 0.1)
     self.xtc_probability = kwargs.pop("xtc_probability", 0)
     self.temperature_last = kwargs.pop("temperature_last", False)
-    self.sampler_priority = kwargs.pop("sampler_priority", ['repetition_penalty', 'presence_penalty', 'frequency_penalty', 'dry', 'temperature', 'dynamic_temperature', 'quadratic_sampling', 'top_n_sigma', 'top_k', 'top_p', 'typical_p', 'epsilon_cutoff', 'eta_cutoff', 'tfs', 'top_a', 'min_p', 'mirostat', 'xtc', 'encoder_repetition_penalty', 'no_repeat_ngram'])
+    self.sampler_priority = kwargs.pop("sampler_priority", ['repetition_penalty', 'presence_penalty', 'frequency_penalty', 'dry', 'temperature', 'dynamic_temperature', 'quadratic_sampling', 'top_n_sigma', 'top_k', 'top_p', 'typical_p', 'epsilon_cutoff', 'eta_cutoff', 'tfs', 'top_a', 'min_p', 'adaptive_p', 'mirostat', 'xtc', 'encoder_repetition_penalty', 'no_repeat_ngram'])
 
 
 def hijack_samplers():
diff --git a/modules/shared.py b/modules/shared.py
index 2fba530bd9..4dedf6f64d 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -81,6 +81,10 @@
 group.add_argument('--gpu-layers-draft', type=int, default=256, help='Number of layers to offload to the GPU for the draft model.')
 group.add_argument('--device-draft', type=str, default=None, help='Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1')
 group.add_argument('--ctx-size-draft', type=int, default=0, help='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
+group.add_argument('--spec-type', type=str, default='none', choices=['none', 'ngram-cache', 'ngram-simple', 'ngram-map-k', 'ngram-map-k4v', 'ngram-mod'], help='Speculative decoding type for draftless speculation.')
+group.add_argument('--spec-ngram-size-n', type=int, default=12, help='N-gram lookup size for ngram speculative decoding.')
+group.add_argument('--spec-ngram-size-m', type=int, default=48, help='Draft n-gram size for ngram speculative decoding.')
+group.add_argument('--spec-ngram-min-hits', type=int, default=1, help='Minimum n-gram hits for ngram-map speculative decoding.')
 
 # llama.cpp
 group = parser.add_argument_group('llama.cpp')
@@ -269,6 +273,8 @@
     'tfs': neutral_samplers['tfs'],
     'top_a': neutral_samplers['top_a'],
     'top_n_sigma': neutral_samplers['top_n_sigma'],
+    'adaptive_target': neutral_samplers['adaptive_target'],
+    'adaptive_decay': neutral_samplers['adaptive_decay'],
 
     # Generation parameters - Repetition suppression
     'dry_multiplier': neutral_samplers['dry_multiplier'],
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 27c5de7dff..daaf08e73a 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -317,6 +317,8 @@ def generate_reply_HF(question, original_question, state, stopping_strings=None,
         'tfs',
         'top_a',
         'top_n_sigma',
+        'adaptive_target',
+        'adaptive_decay',
         'dry_multiplier',
         'dry_allowed_length',
         'dry_base',
diff --git a/modules/ui.py b/modules/ui.py
index 8a49186139..b46915833f 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -168,6 +168,10 @@ def list_model_elements():
         'gpu_layers_draft',
         'device_draft',
         'ctx_size_draft',
+        'spec_type',
+        'spec_ngram_size_n',
+        'spec_ngram_size_m',
+        'spec_ngram_min_hits',
         'mmproj',
     ]
 
@@ -193,6 +197,8 @@ def list_interface_input_elements():
         'tfs',
         'top_a',
         'top_n_sigma',
+        'adaptive_target',
+        'adaptive_decay',
         'dry_multiplier',
         'dry_allowed_length',
         'dry_base',
@@ -488,6 +494,8 @@ def setup_auto_save():
         'tfs',
         'top_a',
         'top_n_sigma',
+        'adaptive_target',
+        'adaptive_decay',
         'dry_multiplier',
         'dry_allowed_length',
         'dry_base',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index b9d731ee84..94be8d8371 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -76,6 +76,10 @@ def create_ui():
                                 shared.gradio['draft_max'] = gr.Number(label="draft-max", precision=0, step=1, value=shared.args.draft_max, info='Number of tokens to draft for speculative decoding. Recommended value: 4.')
                                 shared.gradio['device_draft'] = gr.Textbox(label="device-draft", value=shared.args.device_draft, info='Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1')
                                 shared.gradio['ctx_size_draft'] = gr.Number(label="ctx-size-draft", precision=0, step=256, value=shared.args.ctx_size_draft, info='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
+                                shared.gradio['spec_type'] = gr.Dropdown(label="spec-type", choices=['none', 'ngram-cache', 'ngram-simple', 'ngram-map-k', 'ngram-map-k4v', 'ngram-mod'], value=shared.args.spec_type, info='Draftless speculative decoding type. Uses n-gram matching from context.')
+                                shared.gradio['spec_ngram_size_n'] = gr.Number(label="spec-ngram-size-n", precision=0, step=1, value=shared.args.spec_ngram_size_n, info='N-gram lookup size for speculative decoding.', visible=shared.args.spec_type != 'none')
+                                shared.gradio['spec_ngram_size_m'] = gr.Number(label="spec-ngram-size-m", precision=0, step=1, value=shared.args.spec_ngram_size_m, info='Draft n-gram size for speculative decoding.', visible=shared.args.spec_type != 'none')
+                                shared.gradio['spec_ngram_min_hits'] = gr.Number(label="spec-ngram-min-hits", precision=0, step=1, value=shared.args.spec_ngram_min_hits, info='Minimum n-gram hits for ngram-map speculative decoding.', visible=shared.args.spec_type != 'none')
 
                     gr.Markdown("## Other options")
                     with gr.Accordion("See more options", open=False, elem_classes='tgw-accordion'):
@@ -179,6 +183,13 @@ def create_event_handlers():
     if not shared.args.portable:
         shared.gradio['lora_menu_apply'].click(load_lora_wrapper, gradio('lora_menu'), gradio('model_status'), show_progress=False)
 
+    shared.gradio['spec_type'].change(
+        lambda x: [gr.update(visible=x != 'none')] * 3,
+        gradio('spec_type'),
+        gradio('spec_ngram_size_n', 'spec_ngram_size_m', 'spec_ngram_min_hits'),
+        show_progress=False
+    )
+
     shared.gradio['download_model_button'].click(download_model_wrapper, gradio('custom_model_menu', 'download_specific_file'), gradio('model_status'), show_progress=True)
     shared.gradio['get_file_list'].click(partial(download_model_wrapper, return_links=True), gradio('custom_model_menu', 'download_specific_file'), gradio('model_status'), show_progress=True)
     shared.gradio['customized_template_submit'].click(save_instruction_template, gradio('model_menu', 'customized_template'), gradio('model_status'), show_progress=True)
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 45bef5f9e3..dc07d2915e 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -67,6 +67,8 @@ def create_ui():
                             shared.gradio['mirostat_mode'] = gr.Slider(0, 2, step=1, value=shared.settings['mirostat_mode'], label='mirostat_mode', info='mode=1 is for llama.cpp only.')
                             shared.gradio['mirostat_tau'] = gr.Slider(0, 10, step=0.01, value=shared.settings['mirostat_tau'], label='mirostat_tau')
                             shared.gradio['mirostat_eta'] = gr.Slider(0, 1, step=0.01, value=shared.settings['mirostat_eta'], label='mirostat_eta')
+                            shared.gradio['adaptive_target'] = gr.Slider(0.0, 1.0, value=shared.settings['adaptive_target'], step=0.01, label='adaptive_target', info='Target probability for adaptive-p sampling. Tokens near this probability are favored. 0 disables.')
+                            shared.gradio['adaptive_decay'] = gr.Slider(0.0, 0.99, value=shared.settings['adaptive_decay'], step=0.01, label='adaptive_decay', info='EMA decay rate for adaptive-p. Controls history window (~1/(1-decay) tokens). Default: 0.9.')
 
                             gr.Markdown('## Other options')
                             shared.gradio['do_sample'] = gr.Checkbox(value=shared.settings['do_sample'], label='do_sample')

From 22141679e3ec1e7a867a779a1c2239d2874a19cc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 10:01:31 -0800
Subject: [PATCH 1187/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 14 files changed, 28 insertions(+), 28 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 085680859c..3630e2402c 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -33,8 +33,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index a5c7b1b333..012a1b495a 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index b91ef833a6..a2e9183659 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index b9c19714c4..07045390db 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 1421ad5f90..31e55837ed 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index cf8c0ecb29..373d4017e6 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 1dc6abe32c..349708eb4e 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 09befbccf2..027c7a48ba 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index e794a693b6..bd4cd67f9f 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 48eadd4734..b2e131e0af 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 93b820fc30..462af277d6 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 6e2d05fd27..0648111d11 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index e6a9a0a456..d4ae40380b 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 149bc82966..04dd046e17 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio-4.37.2+custom.4-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.4/gradio_client-1.0.2+custom.4-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From 64eb77e78218241e9dded83e3b9f3964ebd0b6e1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 10:41:47 -0800
Subject: [PATCH 1188/1701] Fix the logits API endpoint with transformers

---
 modules/logits.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/logits.py b/modules/logits.py
index d668e44e4b..3bcdd07fbf 100644
--- a/modules/logits.py
+++ b/modules/logits.py
@@ -84,6 +84,7 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
 
             state['max_new_tokens'] = 1
             state['auto_max_new_tokens'] = False
+            state.setdefault('stream', True)
             for _ in generate_reply(prompt, state):
                 pass
 

From 5d93f4e80061eca1f528b646d3f2a2879a85141f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 10:43:23 -0800
Subject: [PATCH 1189/1701] Fix requires_grad warning in logits API

---
 modules/logits.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/logits.py b/modules/logits.py
index 3bcdd07fbf..7a5f98aeba 100644
--- a/modules/logits.py
+++ b/modules/logits.py
@@ -106,7 +106,7 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
                 output = shared.model(input_ids=tokens)
                 scores = output['logits'][-1][-1]
 
-        probs = torch.softmax(scores, dim=-1, dtype=torch.float)
+        probs = torch.softmax(scores.detach(), dim=-1, dtype=torch.float)
         topk_values, topk_indices = torch.topk(probs, k=top_logits, largest=True, sorted=True)
         if hasattr(shared.tokenizer, 'convert_ids_to_tokens'):
             tokens = [shared.tokenizer.convert_ids_to_tokens(int(i)) for i in topk_indices]

From 9b916f02cdc1d46291d59b0845a74885dcc3c068 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 10:51:15 -0800
Subject: [PATCH 1190/1701] ExLlamaV3: Attach AdaptiveP, fix speculative
 decoding parameter, add seed

---
 modules/exllamav3.py | 11 +++++++++--
 modules/loaders.py   |  2 ++
 2 files changed, 11 insertions(+), 2 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index d884bbf7e2..ea1f3dc911 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -9,6 +9,7 @@
 from exllamav3.generator import Job
 from exllamav3.generator.sampler import (
     CustomSampler,
+    SS_AdaptiveP,
     SS_Argmax,
     SS_MinP,
     SS_PresFreqP,
@@ -158,7 +159,7 @@ def from_pretrained(cls, path_to_model):
             tokenizer=tokenizer,
             draft_model=draft_model,
             draft_cache=draft_cache,
-            num_speculative_tokens=shared.args.draft_max if draft_model is not None else 0,
+            num_draft_tokens=shared.args.draft_max if draft_model is not None else 0,
         )
 
         result = cls()
@@ -302,7 +303,11 @@ def custom_sort_key(sampler_obj):
             ordered_samplers = sorted(unordered_samplers, key=custom_sort_key)
 
             # 5. Add the final sampling stage and build the sampler
-            ordered_samplers.append(SS_Sample())
+            if state.get('adaptive_target', 0) > 0:
+                ordered_samplers.append(SS_AdaptiveP(state['adaptive_target'], state['adaptive_decay']))
+            else:
+                ordered_samplers.append(SS_Sample())
+
             sampler = CustomSampler(ordered_samplers)
 
         # Encode prompt with embeddings (ExLlamaV3-specific)
@@ -329,12 +334,14 @@ def custom_sort_key(sampler_obj):
             if hasattr(self.tokenizer, 'eos_token_id') and self.tokenizer.eos_token_id is not None:
                 stop_conditions.append(self.tokenizer.eos_token_id)
 
+        seed = state.get('seed', -1)
         job = Job(
             input_ids=input_ids,
             max_new_tokens=max_new_tokens,
             decode_special_tokens=not state['skip_special_tokens'],
             embeddings=image_embeddings if image_embeddings else None,
             sampler=sampler,
+            seed=seed if seed >= 0 else None,
             stop_conditions=stop_conditions if stop_conditions else None,
         )
 
diff --git a/modules/loaders.py b/modules/loaders.py
index 94c9c59337..c8b499b9fc 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -276,6 +276,8 @@ def transformers_samplers():
         'min_p',
         'top_p',
         'top_k',
+        'adaptive_target',
+        'adaptive_decay',
         'repetition_penalty',
         'frequency_penalty',
         'presence_penalty',

From d8af0505a805dd5ebadeb9e6aef69f8da54599db Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 11:09:58 -0800
Subject: [PATCH 1191/1701] ExLlamav3_HF: Optimize prefill and fix CFG cache
 initialization

---
 modules/exllamav3_hf.py | 22 ++++++++++++----------
 1 file changed, 12 insertions(+), 10 deletions(-)

diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index e05f8d7dc2..b4b6ad207d 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -84,6 +84,12 @@ def __init__(self, model_dir):
         self.ex_model.load(**load_params)
         self.past_seq = None
         self.max_tokens = max_tokens
+        self.layer_type = layer_type
+        self.cache_kwargs = cache_kwargs
+
+        if shared.args.cfg_cache:
+            self.ex_cache_negative = Cache(self.ex_model, max_num_tokens=max_tokens, layer_type=layer_type, **cache_kwargs)
+            self.past_seq_negative = None
 
     def _validate_model_class(self):
         pass
@@ -126,7 +132,7 @@ def __call__(self, *args, **kwargs):
         reset = True
 
         # Maximum number of tokens to process in a single forward pass
-        max_chunk_size = 256
+        max_chunk_size = 2048
 
         # Make the forward call
         if labels is None:
@@ -147,17 +153,16 @@ def __call__(self, *args, **kwargs):
                         # Process tokens from longest_prefix to second-to-last token
                         tokens_to_process = seq_tensor[longest_prefix:-1]
 
-                        # Process in chunks if the number of tokens is large
+                        # Use prefill() to fill the cache without computing logits
                         for i in range(0, tokens_to_process.shape[0], max_chunk_size):
                             chunk = tokens_to_process[i:i + max_chunk_size]
-                            self.ex_model.forward(
+                            self.ex_model.prefill(
                                 input_ids=chunk.view(1, -1),
                                 params={
                                     "attn_mode": "flash_attn",
                                     "cache": ex_cache,
                                     "past_len": longest_prefix + i,
                                     "batch_shape": (1, self.max_tokens),
-                                    "reconstruct": False  # Force memory-efficient path
                                 }
                             )
 
@@ -168,18 +173,17 @@ def __call__(self, *args, **kwargs):
                     # Process all tokens except the last one
                     tokens_to_process = seq_tensor[:-1]
 
-                    # Process in chunks if the number of tokens is large
+                    # Use prefill() to fill the cache without computing logits
                     current_len = 0
                     for i in range(0, tokens_to_process.shape[0], max_chunk_size):
                         chunk = tokens_to_process[i:i + max_chunk_size]
-                        self.ex_model.forward(
+                        self.ex_model.prefill(
                             input_ids=chunk.view(1, -1),
                             params={
                                 "attn_mode": "flash_attn",
                                 "cache": ex_cache,
                                 "past_len": current_len,
                                 "batch_shape": (1, self.max_tokens),
-                                "reconstruct": False  # Force memory-efficient path
                             }
                         )
                         current_len += chunk.shape[0]
@@ -194,7 +198,6 @@ def __call__(self, *args, **kwargs):
                     "cache": ex_cache,
                     "past_len": current_len,
                     "batch_shape": (1, self.max_tokens),
-                    "reconstruct": False  # Force memory-efficient path
                 }
             ).to(input_ids.device).float()
         else:
@@ -208,8 +211,7 @@ def __call__(self, *args, **kwargs):
                 chunk_logits = self.ex_model.forward(
                     input_ids=chunk.view(1, -1),
                     params={
-                        "attn_mode": "flash_attn_nc",  # No caching for training
-                        "reconstruct": False  # Force memory-efficient path
+                        "attn_mode": "flash_attn_nc",
                     }
                 ).float()
 

From 0ffb75de7cdaae68085c62a6f09ac09d2f583996 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 11:12:54 -0800
Subject: [PATCH 1192/1701] Update Transformers to 5.3.0

---
 requirements/full/requirements.txt               | 2 +-
 requirements/full/requirements_amd.txt           | 2 +-
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_cpu_only.txt      | 2 +-
 requirements/full/requirements_nowheels.txt      | 2 +-
 6 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 3630e2402c..b3f428c167 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -27,7 +27,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.2.*
+transformers==5.3.*
 triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 012a1b495a..806307ed42 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -25,7 +25,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.2.*
+transformers==5.3.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index a2e9183659..aa328deabb 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -25,7 +25,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.2.*
+transformers==5.3.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 07045390db..c3c0054fc2 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -25,7 +25,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.2.*
+transformers==5.3.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 31e55837ed..ad1d1dcfdc 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -25,7 +25,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.2.*
+transformers==5.3.*
 tqdm
 wandb
 
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 373d4017e6..b5bec431f8 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -25,7 +25,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.2.*
+transformers==5.3.*
 tqdm
 wandb
 

From eb90daf09897d1ba4aade3535256c9ea5323b0d0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 11:14:50 -0800
Subject: [PATCH 1193/1701] ExLlamaV2: Don't expose unused seed parameter

---
 modules/loaders.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/modules/loaders.py b/modules/loaders.py
index c8b499b9fc..c920248e61 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -323,7 +323,6 @@ def transformers_samplers():
         'enable_thinking',
         'reasoning_effort',
         'skip_special_tokens',
-        'seed',
         'custom_token_bans',
         'dry_sequence_breakers',
     },

From cdf0e392e69b4552e90288d353bae7eeb3332ca5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 11:37:59 -0800
Subject: [PATCH 1194/1701] llama.cpp: Reorganize speculative decoding UI and
 use recommended ngram-mod defaults

---
 modules/llama_cpp_server.py | 12 ++++--------
 modules/loaders.py          |  1 +
 modules/shared.py           |  4 ++--
 modules/ui_model_menu.py    | 10 +++++++---
 4 files changed, 14 insertions(+), 13 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 0eb270c405..a346323355 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -401,14 +401,10 @@ def _start_server(self):
                 cmd += ["--ctx-size-draft", str(shared.args.ctx_size_draft)]
         if shared.args.spec_type != 'none':
             cmd += ["--spec-type", shared.args.spec_type]
-            if shared.args.draft_max > 0:
-                cmd += ["--draft-max", str(shared.args.draft_max)]
-            if shared.args.spec_ngram_size_n != 12:
-                cmd += ["--spec-ngram-size-n", str(shared.args.spec_ngram_size_n)]
-            if shared.args.spec_ngram_size_m != 48:
-                cmd += ["--spec-ngram-size-m", str(shared.args.spec_ngram_size_m)]
-            if shared.args.spec_ngram_min_hits != 1:
-                cmd += ["--spec-ngram-min-hits", str(shared.args.spec_ngram_min_hits)]
+            cmd += ["--draft-max", str(shared.args.draft_max)]
+            cmd += ["--spec-ngram-size-n", str(shared.args.spec_ngram_size_n)]
+            cmd += ["--spec-ngram-size-m", str(shared.args.spec_ngram_size_m)]
+            cmd += ["--spec-ngram-min-hits", str(shared.args.spec_ngram_min_hits)]
         if shared.args.streaming_llm:
             cmd += ["--cache-reuse", "1"]
             cmd += ["--swa-full"]
diff --git a/modules/loaders.py b/modules/loaders.py
index c920248e61..238b231e7f 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -28,6 +28,7 @@
         'gpu_layers_draft',
         'device_draft',
         'ctx_size_draft',
+        'ngram_header',
         'spec_type',
         'spec_ngram_size_n',
         'spec_ngram_size_m',
diff --git a/modules/shared.py b/modules/shared.py
index 4dedf6f64d..e3ef3c916c 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -81,8 +81,8 @@
 group.add_argument('--gpu-layers-draft', type=int, default=256, help='Number of layers to offload to the GPU for the draft model.')
 group.add_argument('--device-draft', type=str, default=None, help='Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1')
 group.add_argument('--ctx-size-draft', type=int, default=0, help='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
-group.add_argument('--spec-type', type=str, default='none', choices=['none', 'ngram-cache', 'ngram-simple', 'ngram-map-k', 'ngram-map-k4v', 'ngram-mod'], help='Speculative decoding type for draftless speculation.')
-group.add_argument('--spec-ngram-size-n', type=int, default=12, help='N-gram lookup size for ngram speculative decoding.')
+group.add_argument('--spec-type', type=str, default='none', choices=['none', 'ngram-mod', 'ngram-simple', 'ngram-map-k', 'ngram-map-k4v', 'ngram-cache'], help='Draftless speculative decoding type. Recommended: ngram-mod.')
+group.add_argument('--spec-ngram-size-n', type=int, default=24, help='N-gram lookup size for ngram speculative decoding.')
 group.add_argument('--spec-ngram-size-m', type=int, default=48, help='Draft n-gram size for ngram speculative decoding.')
 group.add_argument('--spec-ngram-min-hits', type=int, default=1, help='Minimum n-gram hits for ngram-map speculative decoding.')
 
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 94be8d8371..a81a547456 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -68,15 +68,19 @@ def create_ui():
 
                             # Speculative decoding
                             with gr.Accordion("Speculative decoding", open=False, elem_classes='tgw-accordion') as shared.gradio['speculative_decoding_accordion']:
+                                shared.gradio['draft_max'] = gr.Number(label="draft-max", precision=0, step=1, value=shared.args.draft_max, info='Maximum number of tokens to draft for speculative decoding. Recommended: 4 for draft model, 64 for n-gram.')
+
+                                gr.Markdown('#### Draft model')
                                 with gr.Row():
-                                    shared.gradio['model_draft'] = gr.Dropdown(label="model-draft", choices=['None'] + utils.get_available_models(), value=lambda: shared.args.model_draft, elem_classes='slim-dropdown', info='Draft model. Speculative decoding only works with models sharing the same vocabulary (e.g., same model family).', interactive=not mu)
+                                    shared.gradio['model_draft'] = gr.Dropdown(label="model-draft", choices=['None'] + utils.get_available_models(), value=lambda: shared.args.model_draft, elem_classes='slim-dropdown', info='Draft model. Must share the same vocabulary as the main model.', interactive=not mu)
                                     ui.create_refresh_button(shared.gradio['model_draft'], lambda: None, lambda: {'choices': ['None'] + utils.get_available_models()}, 'refresh-button', interactive=not mu)
 
                                 shared.gradio['gpu_layers_draft'] = gr.Slider(label="gpu-layers-draft", minimum=0, maximum=256, value=shared.args.gpu_layers_draft, info='Number of layers to offload to the GPU for the draft model.')
-                                shared.gradio['draft_max'] = gr.Number(label="draft-max", precision=0, step=1, value=shared.args.draft_max, info='Number of tokens to draft for speculative decoding. Recommended value: 4.')
                                 shared.gradio['device_draft'] = gr.Textbox(label="device-draft", value=shared.args.device_draft, info='Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1')
                                 shared.gradio['ctx_size_draft'] = gr.Number(label="ctx-size-draft", precision=0, step=256, value=shared.args.ctx_size_draft, info='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
-                                shared.gradio['spec_type'] = gr.Dropdown(label="spec-type", choices=['none', 'ngram-cache', 'ngram-simple', 'ngram-map-k', 'ngram-map-k4v', 'ngram-mod'], value=shared.args.spec_type, info='Draftless speculative decoding type. Uses n-gram matching from context.')
+
+                                shared.gradio['ngram_header'] = gr.Markdown('#### N-gram (draftless)')
+                                shared.gradio['spec_type'] = gr.Dropdown(label="spec-type", choices=['none', 'ngram-mod', 'ngram-simple', 'ngram-map-k', 'ngram-map-k4v', 'ngram-cache'], value=shared.args.spec_type, info='Draftless speculative decoding type. Recommended: ngram-mod.')
                                 shared.gradio['spec_ngram_size_n'] = gr.Number(label="spec-ngram-size-n", precision=0, step=1, value=shared.args.spec_ngram_size_n, info='N-gram lookup size for speculative decoding.', visible=shared.args.spec_type != 'none')
                                 shared.gradio['spec_ngram_size_m'] = gr.Number(label="spec-ngram-size-m", precision=0, step=1, value=shared.args.spec_ngram_size_m, info='Draft n-gram size for speculative decoding.', visible=shared.args.spec_type != 'none')
                                 shared.gradio['spec_ngram_min_hits'] = gr.Number(label="spec-ngram-min-hits", precision=0, step=1, value=shared.args.spec_ngram_min_hits, info='Minimum n-gram hits for ngram-map speculative decoding.', visible=shared.args.spec_type != 'none')

From 952e2c404aa96be48e34ce489d16adb6fb85f928 Mon Sep 17 00:00:00 2001
From: weiguang li 
Date: Thu, 5 Mar 2026 04:08:08 +0800
Subject: [PATCH 1195/1701] Bump sentence-transformers from 2.2.2 to 3.3.1 in
 superbooga (#7406)

---
 extensions/superbooga/requirements.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/extensions/superbooga/requirements.txt b/extensions/superbooga/requirements.txt
index 4b16656875..996f189158 100644
--- a/extensions/superbooga/requirements.txt
+++ b/extensions/superbooga/requirements.txt
@@ -2,5 +2,5 @@ beautifulsoup4==4.12.2
 chromadb==0.4.24
 pandas==2.0.3
 posthog==2.4.2
-sentence_transformers==2.2.2
+sentence_transformers==3.3.1
 lxml

From 68109bc5da2a79e421f7610b6c995fdca909b139 Mon Sep 17 00:00:00 2001
From: mamei16 
Date: Wed, 4 Mar 2026 21:26:13 +0100
Subject: [PATCH 1196/1701] Improve `process_markdown_content` (#7403)

---
 modules/html_generator.py | 26 ++++++++++++++++----------
 1 file changed, 16 insertions(+), 10 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 312b66ad07..667a64d6b3 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -211,23 +211,27 @@ def process_markdown_content(string):
     if not string:
         return ""
 
-    # Define a unique placeholder for LaTeX asterisks
+    # Define unique placeholders for LaTeX asterisks and underscores
     LATEX_ASTERISK_PLACEHOLDER = "LATEXASTERISKPLACEHOLDER"
+    LATEX_UNDERSCORE_PLACEHOLDER = "LATEXUNDERSCOREPLACEHOLDER"
 
-    def protect_asterisks_in_latex(match):
-        """A replacer function for re.sub to protect asterisks in multiple LaTeX formats."""
+    def protect_asterisks_underscores_in_latex(match):
+        """A replacer function for re.sub to protect asterisks and underscores in multiple LaTeX formats."""
         # Check which delimiter group was captured
         if match.group(1) is not None:  # Content from $$...$$
             content = match.group(1)
             modified_content = content.replace('*', LATEX_ASTERISK_PLACEHOLDER)
-            return f'$${modified_content}$$'
+            modified_content = modified_content.replace('_', LATEX_UNDERSCORE_PLACEHOLDER)
+            return f'{modified_content}'
         elif match.group(2) is not None:  # Content from \[...\]
             content = match.group(2)
             modified_content = content.replace('*', LATEX_ASTERISK_PLACEHOLDER)
+            modified_content = modified_content.replace('_', LATEX_UNDERSCORE_PLACEHOLDER)
             return f'\\[{modified_content}\\]'
         elif match.group(3) is not None:  # Content from \(...\)
             content = match.group(3)
             modified_content = content.replace('*', LATEX_ASTERISK_PLACEHOLDER)
+            modified_content = modified_content.replace('_', LATEX_UNDERSCORE_PLACEHOLDER)
             return f'\\({modified_content}\\)'
 
         return match.group(0)  # Fallback
@@ -261,9 +265,10 @@ def protect_asterisks_in_latex(match):
     string = string.replace('\\end{equation*}', '$$')
     string = re.sub(r"(.)```", r"\1\n```", string)
 
-    # Protect asterisks within all LaTeX blocks before markdown conversion
-    latex_pattern = re.compile(r'\$\$(.*?)\$\$|\\\[(.*?)\\\]|\\\((.*?)\\\)', re.DOTALL)
-    string = latex_pattern.sub(protect_asterisks_in_latex, string)
+    # Protect asterisks and underscores within all LaTeX blocks before markdown conversion
+    latex_pattern = re.compile(r'((?:^|[\r\n\s])\$\$[^`]*?\$\$)|\\\[(.*?)\\\]|\\\((.*?)\\\)',
+                               re.DOTALL)
+    string = latex_pattern.sub(protect_asterisks_underscores_in_latex, string)
 
     result = ''
     is_code = False
@@ -275,11 +280,11 @@ def protect_asterisks_in_latex(match):
 
         if stripped_line.startswith('```'):
             is_code = not is_code
-        elif stripped_line.startswith('$$'):
+        elif stripped_line.startswith('$$') and (stripped_line == "$$" or not stripped_line.endswith('$$')):
             is_latex = not is_latex
         elif stripped_line.endswith('$$'):
             is_latex = False
-        elif stripped_line.startswith('\\\\['):
+        elif stripped_line.startswith('\\\\[') and not stripped_line.endswith('\\\\]'):
             is_latex = True
         elif stripped_line.startswith('\\\\]'):
             is_latex = False
@@ -324,8 +329,9 @@ def protect_asterisks_in_latex(match):
         # Convert to HTML using markdown
         html_output = markdown.markdown(result, extensions=['fenced_code', 'tables', SaneListExtension()])
 
-    # Restore the LaTeX asterisks after markdown conversion
+    # Restore the LaTeX asterisks and underscores after markdown conversion
     html_output = html_output.replace(LATEX_ASTERISK_PLACEHOLDER, '*')
+    html_output = html_output.replace(LATEX_UNDERSCORE_PLACEHOLDER, '_')
 
     # Remove extra newlines before 
     html_output = re.sub(r'\s*', '', html_output)

From abb7cc02e90f1c36e24088e868e63d5f6ba9f801 Mon Sep 17 00:00:00 2001
From: mamei16 
Date: Wed, 4 Mar 2026 21:44:19 +0100
Subject: [PATCH 1197/1701] Re-introduce inline LaTeX rendering with more
 robust exception handling (#7402)

---
 js/katex/auto-render.js     | 184 ++++++++++++++++++++++++++++++++++++
 js/katex/auto-render.min.js |   1 -
 js/main.js                  |   1 +
 server.py                   |   2 +-
 4 files changed, 186 insertions(+), 2 deletions(-)
 create mode 100644 js/katex/auto-render.js
 delete mode 100644 js/katex/auto-render.min.js

diff --git a/js/katex/auto-render.js b/js/katex/auto-render.js
new file mode 100644
index 0000000000..c9385c5687
--- /dev/null
+++ b/js/katex/auto-render.js
@@ -0,0 +1,184 @@
+! function(e, t) {
+    "object" == typeof exports && "object" == typeof module ? module.exports = t(require("katex")) : "function" == typeof define && define.amd ? define(["katex"], t) : "object" == typeof exports ? exports.renderMathInElement = t(require("katex")) : e.renderMathInElement = t(e.katex)
+}("undefined" != typeof self ? self : this, (function(e) {
+    return function() {
+        "use strict";
+        var t = {
+                771: function(t) {
+                    t.exports = e
+                }
+            },
+            n = {};
+
+        function r(e) {
+            var o = n[e];
+            if (void 0 !== o) return o.exports;
+            var i = n[e] = {
+                exports: {}
+            };
+            return t[e](i, i.exports, r), i.exports
+        }
+        r.n = function(e) {
+            var t = e && e.__esModule ? function() {
+                return e.default
+            } : function() {
+                return e
+            };
+            return r.d(t, {
+                a: t
+            }), t
+        }, r.d = function(e, t) {
+            for (var n in t) r.o(t, n) && !r.o(e, n) && Object.defineProperty(e, n, {
+                enumerable: !0,
+                get: t[n]
+            })
+        }, r.o = function(e, t) {
+            return Object.prototype.hasOwnProperty.call(e, t)
+        };
+        var o = {};
+        return function() {
+            r.d(o, {
+                default: function() {
+                    return d
+                }
+            });
+            var e = r(771),
+                t = r.n(e);
+            const n = function(e, t, n) {
+                    let r = n,
+                        o = 0;
+                    const i = e.length;
+                    for (; r < t.length;) {
+                        const n = t[r];
+                        if (o <= 0 && t.slice(r, r + i) === e) return r;
+                        "\\" === n ? r++ : "{" === n ? o++ : "}" === n && o--, r++
+                    }
+                    return -1
+                },
+                i = /^\\begin{/;
+            var a = function(e, t) {
+                let r;
+                const o = [],
+                    a = new RegExp("(" + t.map((e => e.left.replace(/[-/\\^$*+?.()|[\]{}]/g, "\\$&"))).join("|") + ")");
+                for (; r = e.search(a), -1 !== r;) {
+                    const charAfterOpen = e[r + 1];
+                    if (e[r] == "$" && charAfterOpen != "$") {
+                        const closeDollarIndex = e.indexOf('$', r + 1);
+                        if (closeDollarIndex != -1) {
+                            const charBeforeOpen = r > 0 ? e[r - 1] : '';
+                            const charBeforeClose = r + 1 < closeDollarIndex ? e[closeDollarIndex - 1] : '';
+                            const charBeforeBeforeClose = r + 1 < closeDollarIndex ? e[closeDollarIndex - 2] : '';
+                            const charAfterClose = closeDollarIndex + 1 < e.length ? e[closeDollarIndex + 1] : '';
+                            if ((/[A-Za-z0-9_$-]/.test(charBeforeOpen)) || ((' ' == charBeforeClose) ||
+                                                                             /[0-9]/.test(charAfterOpen) &&
+                                                                             (/[A-Za-z0-9]/.test(charAfterClose)
+                                                                              || '-' == charBeforeClose))) {
+                                 o.push({
+                                     type: "text",
+                                     data: e.slice(0, r + 1),
+                                 });
+                                e = e.slice(r + 1); // now text starts after delimiter
+                                continue;
+                            }
+                        }
+                    }
+                    r > 0 && (o.push({
+                        type: "text",
+                        data: e.slice(0, r)
+                    }), e = e.slice(r));
+                    const a = t.findIndex((t => e.startsWith(t.left)));
+                    if (r = n(t[a].right, e, t[a].left.length), -1 === r) break;
+                    const l = e.slice(0, r + t[a].right.length),
+                        s = i.test(l) ? l : e.slice(t[a].left.length, r);
+                    o.push({
+                        type: "math",
+                        data: s,
+                        rawData: l,
+                        display: t[a].display
+                    }), e = e.slice(r + t[a].right.length)
+                }
+                return "" !== e && o.push({
+                    type: "text",
+                    data: e
+                }), o
+            };
+            const l = function(e, n) {
+                    const r = a(e, n.delimiters);
+                    if (1 === r.length && "text" === r[0].type) return null;
+                    const o = document.createDocumentFragment();
+                    for (let e = 0; e < r.length; e++)
+                        if ("text" === r[e].type) o.appendChild(document.createTextNode(r[e].data));
+                        else {
+                            const i = document.createElement("span");
+                            let a = r[e].data;
+                            n.displayMode = r[e].display;
+                            try {
+                                n.preProcess && (a = n.preProcess(a)), t().render(a, i, n)
+                            } catch (i) {
+                                if (!(i instanceof t().ParseError)) throw i;
+                                n.errorCallback("KaTeX auto-render: Failed to parse `" + r[e].data + "` with ", i), o.appendChild(document.createTextNode(r[e].rawData));
+                                continue
+                            }
+                            o.appendChild(i)
+                        }
+                    return o
+                },
+                s = function(e, t) {
+                    for (let n = 0; n < e.childNodes.length; n++) {
+                        const r = e.childNodes[n];
+                        if (3 === r.nodeType) {
+                            let o = r.textContent,
+                                i = r.nextSibling,
+                                a = 0;
+                            for (; i && i.nodeType === Node.TEXT_NODE;) o += i.textContent, i = i.nextSibling, a++;
+                            const s = l(o, t);
+                            if (s) {
+                                for (let e = 0; e < a; e++) r.nextSibling.remove();
+                                n += s.childNodes.length - 1, e.replaceChild(s, r)
+                            } else n += a
+                        } else if (1 === r.nodeType) {
+                            const e = " " + r.className + " "; - 1 === t.ignoredTags.indexOf(r.nodeName.toLowerCase()) && t.ignoredClasses.every((t => -1 === e.indexOf(" " + t + " "))) && s(r, t)
+                        }
+                    }
+                };
+            var d = function(e, t) {
+                if (!e) throw new Error("No element provided to render");
+                const n = {};
+                for (const e in t) t.hasOwnProperty(e) && (n[e] = t[e]);
+                n.delimiters = n.delimiters || [{
+                    left: "$$",
+                    right: "$$",
+                    display: !0
+                }, {
+                    left: "\\(",
+                    right: "\\)",
+                    display: !1
+                }, {
+                    left: "\\begin{equation}",
+                    right: "\\end{equation}",
+                    display: !0
+                }, {
+                    left: "\\begin{align}",
+                    right: "\\end{align}",
+                    display: !0
+                }, {
+                    left: "\\begin{alignat}",
+                    right: "\\end{alignat}",
+                    display: !0
+                }, {
+                    left: "\\begin{gather}",
+                    right: "\\end{gather}",
+                    display: !0
+                }, {
+                    left: "\\begin{CD}",
+                    right: "\\end{CD}",
+                    display: !0
+                }, {
+                    left: "\\[",
+                    right: "\\]",
+                    display: !0
+                }], n.ignoredTags = n.ignoredTags || ["script", "noscript", "style", "textarea", "pre", "code", "option"], n.ignoredClasses = n.ignoredClasses || [], n.errorCallback = n.errorCallback || console.error, n.macros = n.macros || {}, s(e, n)
+            }
+        }(), o = o.default
+    }()
+}));
diff --git a/js/katex/auto-render.min.js b/js/katex/auto-render.min.js
deleted file mode 100644
index 46d62af28d..0000000000
--- a/js/katex/auto-render.min.js
+++ /dev/null
@@ -1 +0,0 @@
-!function(e,t){"object"==typeof exports&&"object"==typeof module?module.exports=t(require("katex")):"function"==typeof define&&define.amd?define(["katex"],t):"object"==typeof exports?exports.renderMathInElement=t(require("katex")):e.renderMathInElement=t(e.katex)}("undefined"!=typeof self?self:this,(function(e){return function(){"use strict";var t={771:function(t){t.exports=e}},n={};function r(e){var o=n[e];if(void 0!==o)return o.exports;var i=n[e]={exports:{}};return t[e](i,i.exports,r),i.exports}r.n=function(e){var t=e&&e.__esModule?function(){return e.default}:function(){return e};return r.d(t,{a:t}),t},r.d=function(e,t){for(var n in t)r.o(t,n)&&!r.o(e,n)&&Object.defineProperty(e,n,{enumerable:!0,get:t[n]})},r.o=function(e,t){return Object.prototype.hasOwnProperty.call(e,t)};var o={};return function(){r.d(o,{default:function(){return d}});var e=r(771),t=r.n(e);const n=function(e,t,n){let r=n,o=0;const i=e.length;for(;re.left.replace(/[-/\\^$*+?.()|[\]{}]/g,"\\$&"))).join("|")+")");for(;r=e.search(a),-1!==r;){r>0&&(o.push({type:"text",data:e.slice(0,r)}),e=e.slice(r));const a=t.findIndex((t=>e.startsWith(t.left)));if(r=n(t[a].right,e,t[a].left.length),-1===r)break;const l=e.slice(0,r+t[a].right.length),s=i.test(l)?l:e.slice(t[a].left.length,r);o.push({type:"math",data:s,rawData:l,display:t[a].display}),e=e.slice(r+t[a].right.length)}return""!==e&&o.push({type:"text",data:e}),o};const l=function(e,n){const r=a(e,n.delimiters);if(1===r.length&&"text"===r[0].type)return null;const o=document.createDocumentFragment();for(let e=0;e-1===e.indexOf(" "+t+" ")))&&s(r,t)}}};var d=function(e,t){if(!e)throw new Error("No element provided to render");const n={};for(const e in t)t.hasOwnProperty(e)&&(n[e]=t[e]);n.delimiters=n.delimiters||[{left:"$$",right:"$$",display:!0},{left:"\\(",right:"\\)",display:!1},{left:"\\begin{equation}",right:"\\end{equation}",display:!0},{left:"\\begin{align}",right:"\\end{align}",display:!0},{left:"\\begin{alignat}",right:"\\end{alignat}",display:!0},{left:"\\begin{gather}",right:"\\end{gather}",display:!0},{left:"\\begin{CD}",right:"\\end{CD}",display:!0},{left:"\\[",right:"\\]",display:!0}],n.ignoredTags=n.ignoredTags||["script","noscript","style","textarea","pre","code","option"],n.ignoredClasses=n.ignoredClasses||[],n.errorCallback=n.errorCallback||console.error,n.macros=n.macros||{},s(e,n)}}(),o=o.default}()}));
\ No newline at end of file
diff --git a/js/main.js b/js/main.js
index 67f602793c..1317e9e70c 100644
--- a/js/main.js
+++ b/js/main.js
@@ -274,6 +274,7 @@ function doSyntaxHighlighting() {
               renderMathInElement(container, {
                 delimiters: [
                   { left: "$$", right: "$$", display: true },
+                  { left: "$", right: "$", display: false },
                   { left: "\\(", right: "\\)", display: false },
                   { left: "\\[", right: "\\]", display: true },
                 ],
diff --git a/server.py b/server.py
index 304a0478cf..ccfda5ac55 100644
--- a/server.py
+++ b/server.py
@@ -147,7 +147,7 @@ def create_interface():
         '',
         '',
         '',
-        '',
+        '',
         '',
         '',
         '',

From 1d1f4dfc884ee8321995762162a8fbdcd7870788 Mon Sep 17 00:00:00 2001
From: mamei16 
Date: Wed, 4 Mar 2026 21:51:00 +0100
Subject: [PATCH 1198/1701] Disable uncommonly used indented codeblocks (#7401)

---
 modules/sane_markdown_lists.py | 21 ++++++++++-----------
 1 file changed, 10 insertions(+), 11 deletions(-)

diff --git a/modules/sane_markdown_lists.py b/modules/sane_markdown_lists.py
index 1e1d76fd8e..4605ce3163 100644
--- a/modules/sane_markdown_lists.py
+++ b/modules/sane_markdown_lists.py
@@ -58,9 +58,8 @@ def __init__(self, *args):
     def test(self, parent: etree.Element, block: str) -> bool:
         return block.startswith(' ' * MIN_NESTED_LIST_INDENT) and \
             not self.parser.state.isstate('detabbed') and \
-            (parent.tag in self.ITEM_TYPES or
-                (len(parent) and parent[-1] is not None and
-                    (parent[-1].tag in self.LIST_TYPES)))
+            (parent.tag in self.ITEM_TYPES or (len(parent) and parent[-1] is not None and (parent[-1].tag in
+                                                                                           self.LIST_TYPES)))
 
     def get_level(self, parent: etree.Element, block: str) -> tuple[int, etree.Element]:
         """ Get level of indentation based on list level. """
@@ -79,8 +78,7 @@ def get_level(self, parent: etree.Element, block: str) -> tuple[int, etree.Eleme
         # Step through children of tree to find matching indent level.
         while indent_level > level:
             child = self.lastChild(parent)
-            if (child is not None and
-                    (child.tag in self.LIST_TYPES or child.tag in self.ITEM_TYPES)):
+            if child is not None and (child.tag in self.LIST_TYPES or child.tag in self.ITEM_TYPES):
                 if child.tag in self.LIST_TYPES:
                     level += 1
                 parent = child
@@ -124,16 +122,14 @@ class SaneOListProcessor(OListProcessor):
 
     def __init__(self, parser: blockparser.BlockParser):
         super().__init__(parser)
-        # This restriction stems from the 'CodeBlockProcessor' class,
-        # which automatically matches blocks with an indent = self.tab_length
-        max_list_start_indent = self.tab_length - 1
+        max_list_start_indent = self.tab_length
         # Detect an item (e.g., `1. item`)
         self.RE = re.compile(r'^[ ]{0,%d}[\*_]{0,2}\d+\.[ ]+(.*)' % max_list_start_indent)
         # Detect items on secondary lines. they can be of either list type.
         self.CHILD_RE = re.compile(r'^[ ]{0,%d}([\*_]{0,2})((\d+\.))[ ]+(.*)' % (MIN_NESTED_LIST_INDENT - 1))
         # Detect indented (nested) items of either type
         self.INDENT_RE = re.compile(r'^[ ]{%d,%d}[\*_]{0,2}((\d+\.)|[*+-])[ ]+.*' %
-                                    (MIN_NESTED_LIST_INDENT, self.tab_length * 2 - 1))
+                                    (MIN_NESTED_LIST_INDENT, self.tab_length * 2))
 
     def run(self, parent: etree.Element, blocks: list[str]) -> None:
         # Check for multiple items in one block.
@@ -242,7 +238,7 @@ class SaneUListProcessor(SaneOListProcessor):
     def __init__(self, parser: blockparser.BlockParser):
         super().__init__(parser)
         # Detect an item (e.g., `- item` or `+ item` or `* item`).
-        max_list_start_indent = self.tab_length - 1
+        max_list_start_indent = self.tab_length
         self.RE = re.compile(r'^[ ]{0,%d}[*+-][ ]+(.*)' % max_list_start_indent)
         self.CHILD_RE = re.compile(r'^[ ]{0,%d}(([*+-]))[ ]+(.*)' % (MIN_NESTED_LIST_INDENT - 1))
 
@@ -275,7 +271,7 @@ class SaneParagraphProcessor(ParagraphProcessor):
 
     def __init__(self, parser: BlockParser):
         super().__init__(parser)
-        max_list_start_indent = self.tab_length - 1
+        max_list_start_indent = self.tab_length
         self.LIST_RE = re.compile(r"\s{2}\n(\s{0,%d}[\d+*-])" % max_list_start_indent)
 
     def run(self, parent: etree.Element, blocks: list[str]) -> None:
@@ -331,6 +327,9 @@ def extendMarkdown(self, md):
         md.parser.blockprocessors.register(SaneUListProcessor(md.parser), 'ulist', 30)
         md.parser.blockprocessors.register(SaneParagraphProcessor(md.parser), 'paragraph', 10)
 
+        # Disable uncommon indented codeblocks (as opposed to fenced codeblocks delimited by "```")
+        md.parser.blockprocessors.deregister('code')
+
 
 def makeExtension(**kwargs):  # pragma: no cover
     return SaneListExtension(**kwargs)

From 7bf15ad933a17e81d46a8d3d43a6965638343377 Mon Sep 17 00:00:00 2001
From: Sense_wang <167664334+haosenwang1018@users.noreply.github.com>
Date: Thu, 5 Mar 2026 05:06:17 +0800
Subject: [PATCH 1199/1701] fix: replace bare except clauses with except
 Exception (#7400)

---
 extensions/openai/script.py          | 4 ++--
 extensions/sd_api_pictures/script.py | 8 ++++----
 modules/callbacks.py                 | 2 +-
 modules/chat.py                      | 6 +++---
 modules/evaluate.py                  | 2 +-
 modules/llama_cpp_server.py          | 4 ++--
 modules/logits.py                    | 2 +-
 modules/metadata_gguf.py             | 2 +-
 modules/prompts.py                   | 2 +-
 modules/training.py                  | 4 ++--
 modules/ui_model_menu.py             | 2 +-
 one_click.py                         | 2 +-
 server.py                            | 2 +-
 13 files changed, 21 insertions(+), 21 deletions(-)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 12f99ba4b5..edb22c22b1 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -357,7 +357,7 @@ async def handle_load_model(request_data: LoadModelRequest):
     try:
         OAImodels._load_model(to_dict(request_data))
         return JSONResponse(content="OK")
-    except:
+    except Exception:
         traceback.print_exc()
         return HTTPException(status_code=400, detail="Failed to load the model.")
 
@@ -378,7 +378,7 @@ async def handle_load_loras(request_data: LoadLorasRequest):
     try:
         OAImodels.load_loras(request_data.lora_names)
         return JSONResponse(content="OK")
-    except:
+    except Exception:
         traceback.print_exc()
         return HTTPException(status_code=400, detail="Failed to apply the LoRA(s).")
 
diff --git a/extensions/sd_api_pictures/script.py b/extensions/sd_api_pictures/script.py
index f216da3849..6674d43d18 100644
--- a/extensions/sd_api_pictures/script.py
+++ b/extensions/sd_api_pictures/script.py
@@ -264,7 +264,7 @@ def SD_api_address_update(address):
         response = requests.get(url=f'{params["address"]}/sdapi/v1/sd-models')
         response.raise_for_status()
         # r = response.json()
-    except:
+    except Exception:
         msg = "❌ No SD API endpoint on:"
 
     return gr.Textbox.update(label=msg)
@@ -284,7 +284,7 @@ def get_checkpoints():
         options_json = options.json()
         params['sd_checkpoint'] = options_json['sd_model_checkpoint']
         params['checkpoint_list'] = [result["title"] for result in models.json()]
-    except:
+    except Exception:
         params['sd_checkpoint'] = ""
         params['checkpoint_list'] = []
 
@@ -298,7 +298,7 @@ def load_checkpoint(checkpoint):
 
     try:
         requests.post(url=f'{params["address"]}/sdapi/v1/options', json=payload)
-    except:
+    except Exception:
         pass
 
 
@@ -307,7 +307,7 @@ def get_samplers():
         response = requests.get(url=f'{params["address"]}/sdapi/v1/samplers')
         response.raise_for_status()
         samplers = [x["name"] for x in response.json()]
-    except:
+    except Exception:
         samplers = []
 
     return samplers
diff --git a/modules/callbacks.py b/modules/callbacks.py
index 082365f098..afddf92d08 100644
--- a/modules/callbacks.py
+++ b/modules/callbacks.py
@@ -37,7 +37,7 @@ def gentask():
                 ret = self.mfunc(callback=_callback, *args, **self.kwargs)
             except StopNowException:
                 pass
-            except:
+            except Exception:
                 traceback.print_exc()
                 pass
 
diff --git a/modules/chat.py b/modules/chat.py
index 9ef8f5e9fb..21a2104a90 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1300,7 +1300,7 @@ def load_last_chat_state():
         try:
             with open(state_file, 'r', encoding='utf-8') as f:
                 return json.loads(f.read())
-        except:
+        except Exception:
             pass
 
     return {"last_chats": {}}
@@ -1372,7 +1372,7 @@ def load_history_json(file, history):
                     update_message_metadata(history['metadata'], "assistant", i, timestamp="")
 
         return history
-    except:
+    except Exception:
         return history
 
 
@@ -1526,7 +1526,7 @@ def upload_character(file, img_path, tavern=False):
     decoded_file = file if isinstance(file, str) else file.decode('utf-8')
     try:
         data = json.loads(decoded_file)
-    except:
+    except Exception:
         data = yaml.safe_load(decoded_file)
 
     if 'char_name' in data:
diff --git a/modules/evaluate.py b/modules/evaluate.py
index 4f41c1fc14..ccf72ec673 100644
--- a/modules/evaluate.py
+++ b/modules/evaluate.py
@@ -86,7 +86,7 @@ def calculate_perplexity(models, input_dataset, stride, _max_length):
                 update_model_parameters(model_settings)  # hijacking the command-line arguments
                 unload_model()
                 shared.model, shared.tokenizer = load_model(model)
-            except:
+            except Exception:
                 cumulative_log += f"Failed to load `{model}`. Moving on.\n\n"
                 yield cumulative_log
                 continue
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index a346323355..a90a192bf6 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -466,7 +466,7 @@ def _start_server(self):
                 response = self.session.get(health_url)
                 if response.status_code == 200:
                     break
-            except:
+            except Exception:
                 pass
 
             time.sleep(1)
@@ -559,5 +559,5 @@ def filter_stderr_with_progress(process_stderr):
     finally:
         try:
             process_stderr.close()
-        except:
+        except Exception:
             pass
diff --git a/modules/logits.py b/modules/logits.py
index 7a5f98aeba..d9e1ab48be 100644
--- a/modules/logits.py
+++ b/modules/logits.py
@@ -121,7 +121,7 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
                 if isinstance(key, bytes):
                     try:
                         key = key.decode()
-                    except:
+                    except Exception:
                         key = key.decode('latin')
 
                 output[key] = row[0]
diff --git a/modules/metadata_gguf.py b/modules/metadata_gguf.py
index 70ad41dc41..849f448535 100644
--- a/modules/metadata_gguf.py
+++ b/modules/metadata_gguf.py
@@ -53,7 +53,7 @@ def get_single(value_type, file):
         value = file.read(value_length)
         try:
             value = value.decode('utf-8')
-        except:
+        except Exception:
             pass
     else:
         type_str = _simple_value_packing.get(value_type)
diff --git a/modules/prompts.py b/modules/prompts.py
index b800af91a3..91f5812a04 100644
--- a/modules/prompts.py
+++ b/modules/prompts.py
@@ -33,5 +33,5 @@ def count_tokens(text):
     try:
         tokens = get_encoded_length(text)
         return str(tokens)
-    except:
+    except Exception:
         return '0'
diff --git a/modules/training.py b/modules/training.py
index 1de0b7026e..41b2028ef2 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -503,7 +503,7 @@ def generate_and_tokenize_prompt(data_point):
                     print("Model reloaded OK, continue with training.")
                 else:
                     return f"Failed to load {selected_model}."
-            except:
+            except Exception:
                 exc = traceback.format_exc()
                 logger.error('Failed to reload the model.')
                 print(exc)
@@ -542,7 +542,7 @@ def generate_and_tokenize_prompt(data_point):
             logger.info("Loading existing LoRA data")
             state_dict_peft = torch.load(f"{lora_file_path}/adapter_model.bin", weights_only=True)
             set_peft_model_state_dict(lora_model, state_dict_peft)
-    except:
+    except Exception:
         yield traceback.format_exc().replace('\n', '\n\n')
         return
 
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index a81a547456..c65c7a1b47 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -224,7 +224,7 @@ def load_model_wrapper(selected_model, loader, autoload=False):
                 yield f"Successfully loaded `{selected_model}`."
             else:
                 yield f"Failed to load `{selected_model}`."
-        except:
+        except Exception:
             exc = traceback.format_exc()
             logger.error('Failed to load the model.')
             print(exc)
diff --git a/one_click.py b/one_click.py
index dad9e8074d..1af1a686a3 100644
--- a/one_click.py
+++ b/one_click.py
@@ -71,7 +71,7 @@ def load_state():
         try:
             with open(state_file, 'r') as f:
                 return json.load(f)
-        except:
+        except Exception:
             return {}
     return {}
 
diff --git a/server.py b/server.py
index ccfda5ac55..565266662d 100644
--- a/server.py
+++ b/server.py
@@ -75,7 +75,7 @@ def signal_handler(sig, frame):
     if shared.model and shared.model.__class__.__name__ == 'LlamaServer':
         try:
             shared.model.stop()
-        except:
+        except Exception:
             pass
 
     sys.exit(0)

From 2ac4eb33c8981da5b1abe76efdc062832ac1bcb1 Mon Sep 17 00:00:00 2001
From: thecaptain789 <257642323+thecaptain789@users.noreply.github.com>
Date: Wed, 4 Mar 2026 21:09:28 +0000
Subject: [PATCH 1200/1701] fix: correct typo 'occured' to 'occurred' (#7389)

---
 extensions/Training_PRO/custom_scheduler.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/extensions/Training_PRO/custom_scheduler.py b/extensions/Training_PRO/custom_scheduler.py
index 1e80daed1a..cbcac1173a 100644
--- a/extensions/Training_PRO/custom_scheduler.py
+++ b/extensions/Training_PRO/custom_scheduler.py
@@ -27,7 +27,7 @@ def custom_scheduler_global_setup(trigger_loss: float, ramp_down_ratio: float):
 
     # calculates the total num steps after trigger
     custom_scheduler_params.update({'calc_num_training_steps': 0})
-    #calculates steps when the ramp_down trigger occured
+    #calculates steps when the ramp_down trigger occurred
     custom_scheduler_params.update({'calc_ramp_down_at_step': 0})
     # triggers scheduler stopping after it reached calc_num_training_steps
     custom_scheduler_params.update({'dynamic_scheduler_stop': False})

From 83cc207ef79a4b9717e811602f7fdf557c8c1e9e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 14:31:18 -0800
Subject: [PATCH 1201/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 14 files changed, 28 insertions(+), 28 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index b3f428c167..921cf6781b 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -33,8 +33,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 806307ed42..8810ff4bf8 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index aa328deabb..e0a8627ee3 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index c3c0054fc2..d9732e91a6 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index ad1d1dcfdc..0088adb005 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index b5bec431f8..e060038ad3 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 349708eb4e..88bf31268a 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 027c7a48ba..19031c213d 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index bd4cd67f9f..c98d60ac97 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index b2e131e0af..797586521d 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 462af277d6..f932f3ad5f 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 0648111d11..049b1d0496 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index d4ae40380b..ff1aeb988c 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 04dd046e17..bf37c4458d 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio-4.37.2+custom.5-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.5/gradio_client-1.0.2+custom.5-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From da3010c3eda35a5a1a6487652d784ee307cc1648 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 15:53:24 -0800
Subject: [PATCH 1202/1701] tiny improvements to llama_cpp_server.py

---
 modules/llama_cpp_server.py | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index a90a192bf6..2631f253c7 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -208,8 +208,9 @@ def generate_with_streaming(self, prompt, state):
         # Make the generation request
         response = self.session.post(url, json=payload, stream=True)
         try:
-            if response.status_code == 400 and response.json()["error"]["type"] == "exceed_context_size_error":
+            if response.status_code == 400 and response.json().get("error", {}).get("type") == "exceed_context_size_error":
                 logger.error("The request exceeds the available context size, try increasing it")
+                return
             else:
                 response.raise_for_status()  # Raise an exception for HTTP errors
 
@@ -286,6 +287,8 @@ def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
                     return result["completion_probabilities"][0]["top_probs"]
                 else:
                     return result["completion_probabilities"][0]["top_logprobs"]
+
+            time.sleep(0.05)
         else:
             raise Exception(f"Unexpected response format: 'completion_probabilities' not found in {result}")
 
@@ -390,7 +393,7 @@ def _start_server(self):
             else:
                 model_file = sorted(path.glob('*.gguf'))[0]
 
-            cmd += ["--model-draft", model_file]
+            cmd += ["--model-draft", str(model_file)]
             if shared.args.draft_max > 0:
                 cmd += ["--draft-max", str(shared.args.draft_max)]
             if shared.args.gpu_layers_draft > 0:
@@ -417,8 +420,11 @@ def _start_server(self):
                 extra_flags = extra_flags[1:-1].strip()
 
             for flag_item in extra_flags.split(','):
+                flag_item = flag_item.strip()
                 if '=' in flag_item:
                     flag, value = flag_item.split('=', 1)
+                    flag = flag.strip()
+                    value = value.strip()
                     if len(flag) <= 3:
                         cmd += [f"-{flag}", value]
                     else:

From 942ff8fcb4a9ce83f4142e12c644052f63ed88da Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 16:43:32 -0800
Subject: [PATCH 1203/1701] Remove obsolete stuff after custom gradio updates

---
 js/global_scope_js.js | 15 ---------------
 server.py             |  4 ----
 2 files changed, 19 deletions(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 2b3f38e7ab..205d937535 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -351,18 +351,3 @@ function handleMorphdomUpdate(data) {
     }
   });
 }
-
-//------------------------------------------------
-// Suppress "Attempted to select a non-interactive or hidden tab" warning
-//------------------------------------------------
-(function() {
-  const originalWarn = console.warn;
-
-  console.warn = function(...args) {
-    if (args[0] && typeof args[0] === "string" && args[0].includes("Attempted to select a non-interactive or hidden tab")) {
-      return;
-    }
-
-    originalWarn.apply(console, args);
-  };
-})();
diff --git a/server.py b/server.py
index 565266662d..2d0e567121 100644
--- a/server.py
+++ b/server.py
@@ -28,10 +28,6 @@
 
 import gradio as gr
 
-import matplotlib
-
-matplotlib.use('Agg')  # This fixes LaTeX rendering on some systems
-
 import os
 import signal
 import sys

From 387cf9d8df6cfacf603800f03d54a266cf6c1083 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 17:20:34 -0800
Subject: [PATCH 1204/1701] Remove obsolete DeepSpeed inference code (2023
 relic)

---
 README.md                       |  7 +---
 docs/08 - Additional Tips.md    | 23 ----------
 modules/deepspeed_parameters.py | 74 ---------------------------------
 modules/shared.py               |  6 ---
 modules/text_generation.py      |  2 -
 modules/torch_utils.py          |  3 --
 modules/transformers_loader.py  | 52 +----------------------
 7 files changed, 2 insertions(+), 165 deletions(-)
 delete mode 100644 modules/deepspeed_parameters.py

diff --git a/README.md b/README.md
index 2869aaec4a..a72e806050 100644
--- a/README.md
+++ b/README.md
@@ -245,7 +245,7 @@ usage: server.py [-h] [--multi-user] [--model MODEL] [--lora LORA [LORA ...]] [-
                  [--extra-flags EXTRA_FLAGS] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code]
                  [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE]
                  [--enable-tp] [--tp-backend TP_BACKEND] [--gpu-split GPU_SPLIT] [--autosplit] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa] [--num_experts_per_token N] [--cpp-runner]
-                 [--deepspeed] [--nvme-offload-dir NVME_OFFLOAD_DIR] [--local_rank LOCAL_RANK] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB]
+                 [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB]
                  [--listen] [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH]
                  [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors] [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT]
                  [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4] [--nowebui]
@@ -334,11 +334,6 @@ ExLlamaV2:
 TensorRT-LLM:
   --cpp-runner                              Use the ModelRunnerCpp runner, which is faster than the default ModelRunner but doesn't support streaming yet.
 
-DeepSpeed:
-  --deepspeed                               Enable the use of DeepSpeed ZeRO-3 for inference via the Transformers integration.
-  --nvme-offload-dir NVME_OFFLOAD_DIR       DeepSpeed: Directory to use for ZeRO-3 NVME offloading.
-  --local_rank LOCAL_RANK                   DeepSpeed: Optional argument for distributed setups.
-
 RoPE:
   --alpha_value ALPHA_VALUE                 Positional embeddings alpha factor for NTK RoPE scaling. Use either this or compress_pos_emb, not both.
   --rope_freq_base ROPE_FREQ_BASE           If greater than 0, will be used instead of alpha_value. Those two are related by rope_freq_base = 10000 * alpha_value ^ (64 / 63).
diff --git a/docs/08 - Additional Tips.md b/docs/08 - Additional Tips.md
index 079d1da06a..e6e2b3c9a5 100644
--- a/docs/08 - Additional Tips.md	
+++ b/docs/08 - Additional Tips.md	
@@ -13,29 +13,6 @@ Source: https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/1126
 
 This file will be automatically detected the next time you start the web UI.
 
-## DeepSpeed
-
-`DeepSpeed ZeRO-3` is an alternative offloading strategy for full-precision (16-bit) transformers models.
-
-With this, I have been able to load a 6b model (GPT-J 6B) with less than 6GB of VRAM. The speed of text generation is very decent and much better than what would be accomplished with `--auto-devices --gpu-memory 6`.
-
-As far as I know, DeepSpeed is only available for Linux at the moment.
-
-### How to use it
-
-1. Install DeepSpeed: 
-
-```
-conda install -c conda-forge mpi4py mpich
-pip install -U deepspeed
-```
-
-2. Start the web UI replacing `python` with `deepspeed --num_gpus=1` and adding the `--deepspeed` flag. Example:
-
-```
-deepspeed --num_gpus=1 server.py --deepspeed --chat --model gpt-j-6B
-```
-
 ## Miscellaneous info
 
 ### You can train LoRAs in CPU mode
diff --git a/modules/deepspeed_parameters.py b/modules/deepspeed_parameters.py
deleted file mode 100644
index f170a385cf..0000000000
--- a/modules/deepspeed_parameters.py
+++ /dev/null
@@ -1,74 +0,0 @@
-def generate_ds_config(ds_bf16, train_batch_size, nvme_offload_dir):
-    '''
-    DeepSpeed configuration
-    https://huggingface.co/docs/transformers/main_classes/deepspeed
-    '''
-
-    if nvme_offload_dir:
-        ds_config = {
-            "fp16": {
-                "enabled": not ds_bf16,
-            },
-            "bf16": {
-                "enabled": ds_bf16,
-            },
-            "zero_optimization": {
-                "stage": 3,
-                "offload_param": {
-                    "device": "nvme",
-                    "nvme_path": nvme_offload_dir,
-                    "pin_memory": True,
-                    "buffer_count": 5,
-                    "buffer_size": 1e9,
-                    "max_in_cpu": 1e9
-                },
-                "overlap_comm": True,
-                "reduce_bucket_size": "auto",
-                "contiguous_gradients": True,
-                "sub_group_size": 1e8,
-                "stage3_prefetch_bucket_size": "auto",
-                "stage3_param_persistence_threshold": "auto",
-                "stage3_max_live_parameters": "auto",
-                "stage3_max_reuse_distance": "auto",
-            },
-            "aio": {
-                "block_size": 262144,
-                "queue_depth": 32,
-                "thread_count": 1,
-                "single_submit": False,
-                "overlap_events": True
-            },
-            "steps_per_print": 2000,
-            "train_batch_size": train_batch_size,
-            "train_micro_batch_size_per_gpu": 1,
-            "wall_clock_breakdown": False
-        }
-    else:
-        ds_config = {
-            "fp16": {
-                "enabled": not ds_bf16,
-            },
-            "bf16": {
-                "enabled": ds_bf16,
-            },
-            "zero_optimization": {
-                "stage": 3,
-                "offload_param": {
-                    "device": "cpu",
-                    "pin_memory": True
-                },
-                "overlap_comm": True,
-                "contiguous_gradients": True,
-                "reduce_bucket_size": "auto",
-                "stage3_prefetch_bucket_size": "auto",
-                "stage3_param_persistence_threshold": "auto",
-                "stage3_max_live_parameters": "auto",
-                "stage3_max_reuse_distance": "auto",
-            },
-            "steps_per_print": 2000,
-            "train_batch_size": train_batch_size,
-            "train_micro_batch_size_per_gpu": 1,
-            "wall_clock_breakdown": False
-        }
-
-    return ds_config
diff --git a/modules/shared.py b/modules/shared.py
index e3ef3c916c..c917377a3f 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -144,12 +144,6 @@
 group = parser.add_argument_group('TensorRT-LLM')
 group.add_argument('--cpp-runner', action='store_true', help='Use the ModelRunnerCpp runner, which is faster than the default ModelRunner but doesn\'t support streaming yet.')
 
-# DeepSpeed
-group = parser.add_argument_group('DeepSpeed')
-group.add_argument('--deepspeed', action='store_true', help='Enable the use of DeepSpeed ZeRO-3 for inference via the Transformers integration.')
-group.add_argument('--nvme-offload-dir', type=str, help='DeepSpeed: Directory to use for ZeRO-3 NVME offloading.')
-group.add_argument('--local_rank', type=int, default=0, help='DeepSpeed: Optional argument for distributed setups.')
-
 # RoPE
 group = parser.add_argument_group('RoPE')
 group.add_argument('--alpha_value', type=float, default=1, help='Positional embeddings alpha factor for NTK RoPE scaling. Use either this or compress_pos_emb, not both.')
diff --git a/modules/text_generation.py b/modules/text_generation.py
index daaf08e73a..845711ce55 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -372,8 +372,6 @@ def generate_reply_HF(question, original_question, state, stopping_strings=None,
         generate_params['negative_prompt_ids'] = encode(state['negative_prompt'])
 
     generate_params.update({'use_cache': not shared.args.no_cache})
-    if shared.args.deepspeed:
-        generate_params.update({'synced_gpus': True})
 
     # Encode the input
     input_ids = encode(question, add_bos_token=state['add_bos_token'], truncation_length=get_max_prompt_length(state))
diff --git a/modules/torch_utils.py b/modules/torch_utils.py
index 418520a87d..ba9068571e 100644
--- a/modules/torch_utils.py
+++ b/modules/torch_utils.py
@@ -12,9 +12,6 @@ def get_device():
         return shared.model.device
     elif torch.cuda.is_available():
         return torch.device('cuda')
-    elif shared.args.deepspeed:
-        import deepspeed
-        return deepspeed.get_accelerator().current_device_name()
     elif torch.backends.mps.is_available():
         return torch.device('mps')
     elif is_torch_xpu_available():
diff --git a/modules/transformers_loader.py b/modules/transformers_loader.py
index f1af1299fa..d57020c6dd 100644
--- a/modules/transformers_loader.py
+++ b/modules/transformers_loader.py
@@ -1,4 +1,3 @@
-import os
 import pprint
 from pathlib import Path
 
@@ -6,11 +5,7 @@
 import torch.nn.functional as F
 import transformers
 from accelerate import infer_auto_device_map, init_empty_weights
-from accelerate.utils import (
-    is_ccl_available,
-    is_npu_available,
-    is_xpu_available
-)
+from accelerate.utils import is_xpu_available
 from transformers import (
     AutoConfig,
     AutoModel,
@@ -28,31 +23,6 @@
 
 transformers.logging.set_verbosity_error()
 
-local_rank = None
-if shared.args.deepspeed:
-    import deepspeed
-    from transformers.integrations.deepspeed import (
-        HfDeepSpeedConfig,
-        is_deepspeed_zero3_enabled
-    )
-
-    from modules.deepspeed_parameters import generate_ds_config
-
-    # Distributed setup
-    local_rank = shared.args.local_rank if shared.args.local_rank is not None else int(os.getenv("LOCAL_RANK", "0"))
-    world_size = int(os.getenv("WORLD_SIZE", "1"))
-    if is_xpu_available() and is_ccl_available():
-        torch.xpu.set_device(local_rank)
-        deepspeed.init_distributed(backend="ccl")
-    elif is_npu_available():
-        torch.npu.set_device(local_rank)
-        deepspeed.init_distributed(dist_backend="hccl")
-    else:
-        torch.cuda.set_device(local_rank)
-        deepspeed.init_distributed()
-    ds_config = generate_ds_config(shared.args.bf16, 1 * world_size, shared.args.nvme_offload_dir)
-    dschf = HfDeepSpeedConfig(ds_config)  # Keep this object alive for the Transformers integration
-
 
 class _StopEverythingStoppingCriteria(transformers.StoppingCriteria):
     def __init__(self):
@@ -163,7 +133,6 @@ def load_model_HF(model_name):
         shared.args.load_in_8bit,
         shared.args.load_in_4bit,
         shared.args.disk,
-        shared.args.deepspeed,
         shared.args.cpu_memory is not None,
         shared.args.compress_pos_emb > 1,
         shared.args.alpha_value > 1,
@@ -183,25 +152,6 @@ def load_model_HF(model_name):
             if device:
                 model = model.to(device)
 
-    # DeepSpeed ZeRO-3
-    elif shared.args.deepspeed:
-        model = LoaderClass.from_pretrained(
-            path_to_model,
-            torch_dtype=params['torch_dtype'],
-            trust_remote_code=params.get('trust_remote_code')
-        )
-
-        model = deepspeed.initialize(
-            model=model,
-            config_params=ds_config,
-            model_parameters=None,
-            optimizer=None,
-            lr_scheduler=None
-        )[0]
-
-        model.module.eval()  # Inference
-        logger.info(f'DeepSpeed ZeRO-3 is enabled: {is_deepspeed_zero3_enabled()}')
-
     # Load with quantization and/or offloading
     else:
         if not any((shared.args.cpu, torch.cuda.is_available(), is_xpu_available(), torch.backends.mps.is_available())):

From 5a22970ba8ac671851a83ac0ba1292a6849af19c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 22:59:45 -0300
Subject: [PATCH 1205/1701] Docker: fix and clean up configs, update docs

---
 docker/.dockerignore             |   7 +-
 docker/.env.example              |  15 +--
 docker/TensorRT-LLM/Dockerfile   |   4 +-
 docker/amd/Dockerfile            |   6 +-
 docker/amd/docker-compose.yml    |  20 +--
 docker/cpu/Dockerfile            |  10 +-
 docker/cpu/docker-compose.yml    |  32 +----
 docker/intel/Dockerfile          |   6 +-
 docker/intel/docker-compose.yml  |  20 +--
 docker/nvidia/Dockerfile         |   3 +-
 docker/nvidia/docker-compose.yml |  20 +--
 docs/09 - Docker.md              | 204 ++++---------------------------
 12 files changed, 50 insertions(+), 297 deletions(-)

diff --git a/docker/.dockerignore b/docker/.dockerignore
index 99d0adff8d..5e9c94edf3 100644
--- a/docker/.dockerignore
+++ b/docker/.dockerignore
@@ -1,8 +1,3 @@
 .env
 Dockerfile
-/characters
-/loras
-/models
-/presets
-/prompts
-/training
+/user_data
diff --git a/docker/.env.example b/docker/.env.example
index bd0f8bcc5e..fb6eaea375 100644
--- a/docker/.env.example
+++ b/docker/.env.example
@@ -1,8 +1,8 @@
-# by default the Dockerfile specifies these versions: 3.5;5.0;6.0;6.1;7.0;7.5;8.0;8.6+PTX
-# however for me to work i had to specify the exact version for my card ( 2060 ) it was 7.5
-# https://developer.nvidia.com/cuda-gpus you can find the version for your card here
-# Or for a programatic approach run `nvidia-smi --query-gpu=name,compute_cap --format=csv`
-TORCH_CUDA_ARCH_LIST=7.5
+# specify which cuda arch version your card supports (NVIDIA only)
+# https://developer.nvidia.com/cuda-gpus
+# or run: nvidia-smi --query-gpu=name,compute_cap --format=csv
+# default in docker-compose.yml covers RTX 3090 (8.6) and RTX 4090 (8.9)
+TORCH_CUDA_ARCH_LIST=8.6;8.9+PTX
 # the port the webui binds to on the host
 HOST_PORT=7860
 # the port the webui binds to inside the container
@@ -13,12 +13,9 @@ HOST_API_PORT=5000
 CONTAINER_API_PORT=5000
 # Comma separated extensions to build
 BUILD_EXTENSIONS=""
-# Set APP_RUNTIME_GID to an appropriate host system group to enable access to mounted volumes 
+# Set APP_RUNTIME_GID to an appropriate host system group to enable access to mounted volumes
 # You can find your current host user group id with the command `id -g`
 APP_RUNTIME_GID=6972
 # override default app build permissions (handy for deploying to cloud)
 #APP_GID=6972
 #APP_UID=6972
-# Set cache env
-TRANSFORMERS_CACHE=/home/app/text-generation-webui/cache/
-HF_HOME=/home/app/text-generation-webui/cache/
diff --git a/docker/TensorRT-LLM/Dockerfile b/docker/TensorRT-LLM/Dockerfile
index ae503c9411..c2be3efe65 100644
--- a/docker/TensorRT-LLM/Dockerfile
+++ b/docker/TensorRT-LLM/Dockerfile
@@ -1,4 +1,4 @@
-FROM pytorch/pytorch:2.2.1-cuda12.1-cudnn8-runtime
+FROM pytorch/pytorch:2.9.1-cuda12.8-cudnn9-runtime
 
 # Install Git
 RUN apt update && apt install -y git
@@ -12,7 +12,7 @@ WORKDIR /app
 # Install text-generation-webui
 RUN git clone https://github.com/oobabooga/text-generation-webui
 WORKDIR /app/text-generation-webui
-RUN pip install -r requirements.txt
+RUN pip install -r requirements/full/requirements.txt
 
 # This is needed to avoid an error about "Failed to build mpi4py" in the next command
 ENV LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu:$LD_LIBRARY_PATH
diff --git a/docker/amd/Dockerfile b/docker/amd/Dockerfile
index c23083f7d4..6df8e2de4c 100644
--- a/docker/amd/Dockerfile
+++ b/docker/amd/Dockerfile
@@ -1,7 +1,6 @@
 # BUILDER
 FROM ubuntu:22.04
 WORKDIR /builder
-ARG TORCH_CUDA_ARCH_LIST="${TORCH_CUDA_ARCH_LIST:-3.5;5.0;6.0;6.1;7.0;7.5;8.0;8.6+PTX}"
 ARG BUILD_EXTENSIONS="${BUILD_EXTENSIONS:-}"
 ARG APP_UID="${APP_UID:-6972}"
 ARG APP_GID="${APP_GID:-6972}"
@@ -14,8 +13,7 @@ WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
 RUN GPU_CHOICE=B LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
-COPY /user_data/CMD_FLAGS.txt /home/app/text-generation-webui/user_data
-EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
+EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000}
 WORKDIR /home/app/text-generation-webui
 # set umask to ensure group read / write at runtime
-CMD umask 0002 && export HOME=/home/app/text-generation-webui && ./start_linux.sh
+CMD umask 0002 && export HOME=/home/app/text-generation-webui && ./start_linux.sh --listen
diff --git a/docker/amd/docker-compose.yml b/docker/amd/docker-compose.yml
index a727ca3e67..6fe3a375b8 100644
--- a/docker/amd/docker-compose.yml
+++ b/docker/amd/docker-compose.yml
@@ -4,25 +4,9 @@ services:
     build:
       context: .
       args:
-        # Requirements file to use: 
-        # | GPU | requirements file to use |
-        # |--------|---------|
-        # | NVIDIA | `requirements.txt` |
-        # | AMD | `requirements_amd.txt` |
-        # | CPU only | `requirements_cpu_only.txt` |
-        # | Apple Intel | `requirements_apple_intel.txt` |
-        # | Apple Silicon | `requirements_apple_silicon.txt` |
-        # Default: requirements.txt`
-        # BUILD_REQUIREMENTS: requirements.txt
-        
-        # Extension requirements to build: 
-        # BUILD_EXTENSIONS: 
-        
-        # specify which cuda version your card supports: https://developer.nvidia.com/cuda-gpus
-        TORCH_CUDA_ARCH_LIST: ${TORCH_CUDA_ARCH_LIST:-7.5} 
         BUILD_EXTENSIONS: ${BUILD_EXTENSIONS:-}
-        APP_GID: ${APP_GID:-6972} 
-        APP_UID: ${APP_UID:-6972} 
+        APP_GID: ${APP_GID:-6972}
+        APP_UID: ${APP_UID:-6972}
     env_file: .env
     user: "${APP_RUNTIME_UID:-6972}:${APP_RUNTIME_GID:-6972}"
     ports:
diff --git a/docker/cpu/Dockerfile b/docker/cpu/Dockerfile
index 8f643d2f1a..96092edbec 100644
--- a/docker/cpu/Dockerfile
+++ b/docker/cpu/Dockerfile
@@ -1,14 +1,9 @@
 # BUILDER
 FROM ubuntu:22.04
 WORKDIR /builder
-ARG TORCH_CUDA_ARCH_LIST="${TORCH_CUDA_ARCH_LIST:-3.5;5.0;6.0;6.1;7.0;7.5;8.0;8.6+PTX}"
 ARG BUILD_EXTENSIONS="${BUILD_EXTENSIONS:-}"
 ARG APP_UID="${APP_UID:-6972}"
 ARG APP_GID="${APP_GID:-6972}"
-ARG GPU_CHOICE=A
-ARG USE_CUDA118=FALSE 
-ARG LAUNCH_AFTER_INSTALL=FALSE 	
-ARG INSTALL_EXTENSIONS=TRUE
 
 RUN --mount=type=cache,target=/var/cache/apt,sharing=locked,rw \
     apt update && \
@@ -18,8 +13,7 @@ WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
 RUN GPU_CHOICE=N LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
-COPY CMD_FLAGS.txt /home/app/text-generation-webui/
-EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
+EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000}
 # set umask to ensure group read / write at runtime
 WORKDIR /home/app/text-generation-webui
-CMD umask 0002 && export HOME=/home/app/text-generation-webui && ./start_linux.sh
+CMD umask 0002 && export HOME=/home/app/text-generation-webui && ./start_linux.sh --listen
diff --git a/docker/cpu/docker-compose.yml b/docker/cpu/docker-compose.yml
index 9aba314a4e..6942f8de5c 100644
--- a/docker/cpu/docker-compose.yml
+++ b/docker/cpu/docker-compose.yml
@@ -4,25 +4,9 @@ services:
     build:
       context: .
       args:
-        # Requirements file to use: 
-        # | GPU | requirements file to use |
-        # |--------|---------|
-        # | NVIDIA | `requirements.txt` |
-        # | AMD | `requirements_amd.txt` |
-        # | CPU only | `requirements_cpu_only.txt` |
-        # | Apple Intel | `requirements_apple_intel.txt` |
-        # | Apple Silicon | `requirements_apple_silicon.txt` |
-        # Default: requirements.txt`
-        # BUILD_REQUIREMENTS: requirements.txt
-        
-        # Extension requirements to build: 
-        # BUILD_EXTENSIONS: 
-        
-        # specify which cuda version your card supports: https://developer.nvidia.com/cuda-gpus
-        TORCH_CUDA_ARCH_LIST: ${TORCH_CUDA_ARCH_LIST:-7.5} 
         BUILD_EXTENSIONS: ${BUILD_EXTENSIONS:-}
-        APP_GID: ${APP_GID:-6972} 
-        APP_UID: ${APP_UID:-6972} 
+        APP_GID: ${APP_GID:-6972}
+        APP_UID: ${APP_UID:-6972}
     env_file: .env
     user: "${APP_RUNTIME_UID:-6972}:${APP_RUNTIME_GID:-6972}"
     ports:
@@ -31,14 +15,4 @@ services:
     stdin_open: true
     tty: true
     volumes:
-      - ./cache:/home/app/text-generation-webui/cache
-      - ./characters:/home/app/text-generation-webui/characters
-      - ./extensions:/home/app/text-generation-webui/extensions
-      - ./loras:/home/app/text-generation-webui/loras
-      - ./logs:/home/app/text-generation-webui/logs
-      - ./models:/home/app/text-generation-webui/models
-      - ./presets:/home/app/text-generation-webui/presets
-      - ./prompts:/home/app/text-generation-webui/prompts
-      - ./softprompts:/home/app/text-generation-webui/softprompts
-      - ./training:/home/app/text-generation-webui/training
-      - ./cloudflared:/etc/cloudflared
+      - ./user_data:/home/app/text-generation-webui/user_data
diff --git a/docker/intel/Dockerfile b/docker/intel/Dockerfile
index 4a70980358..28ee983ce8 100644
--- a/docker/intel/Dockerfile
+++ b/docker/intel/Dockerfile
@@ -1,7 +1,6 @@
 # BUILDER
 FROM ubuntu:22.04
 WORKDIR /builder
-ARG TORCH_CUDA_ARCH_LIST="${TORCH_CUDA_ARCH_LIST:-3.5;5.0;6.0;6.1;7.0;7.5;8.0;8.6+PTX}"
 ARG BUILD_EXTENSIONS="${BUILD_EXTENSIONS:-}"
 ARG APP_UID="${APP_UID:-6972}"
 ARG APP_GID="${APP_GID:-6972}"
@@ -14,8 +13,7 @@ WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
 RUN GPU_CHOICE=D LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
-COPY /user_data/CMD_FLAGS.txt /home/app/text-generation-webui/user_data
-EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
+EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000}
 # set umask to ensure group read / write at runtime
 WORKDIR /home/app/text-generation-webui
-CMD umask 0002 && export HOME=/home/app/text-generation-webui && ./start_linux.sh
+CMD umask 0002 && export HOME=/home/app/text-generation-webui && ./start_linux.sh --listen
diff --git a/docker/intel/docker-compose.yml b/docker/intel/docker-compose.yml
index bb48dd227d..6fe3a375b8 100644
--- a/docker/intel/docker-compose.yml
+++ b/docker/intel/docker-compose.yml
@@ -4,25 +4,9 @@ services:
     build:
       context: .
       args:
-        # Requirements file to use: 
-        # | GPU | requirements file to use |
-        # |--------|---------|
-        # | NVIDIA | `requirements.txt` |
-        # | AMD | `requirements_amd.txt` |
-        # | CPU only | `requirements_cpu_only.txt` |
-        # | Apple Intel | `requirements_apple_intel.txt` |
-        # | Apple Silicon | `requirements_apple_silicon.txt` |
-        # Default: requirements.txt`
-        # BUILD_REQUIREMENTS: requirements.txt
-
-        # Extension requirements to build: 
-        # BUILD_EXTENSIONS: 
-
-        # specify which cuda version your card supports: https://developer.nvidia.com/cuda-gpus
-        TORCH_CUDA_ARCH_LIST: ${TORCH_CUDA_ARCH_LIST:-7.5} 
         BUILD_EXTENSIONS: ${BUILD_EXTENSIONS:-}
-        APP_GID: ${APP_GID:-6972} 
-        APP_UID: ${APP_UID:-6972} 
+        APP_GID: ${APP_GID:-6972}
+        APP_UID: ${APP_UID:-6972}
     env_file: .env
     user: "${APP_RUNTIME_UID:-6972}:${APP_RUNTIME_GID:-6972}"
     ports:
diff --git a/docker/nvidia/Dockerfile b/docker/nvidia/Dockerfile
index 82594a26f2..a7ef93badb 100644
--- a/docker/nvidia/Dockerfile
+++ b/docker/nvidia/Dockerfile
@@ -14,8 +14,7 @@ WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
 RUN GPU_CHOICE=A LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
-COPY /user_data/CMD_FLAGS.txt /home/app/text-generation-webui/user_data
-EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
+EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000}
 WORKDIR /home/app/text-generation-webui
 # set umask to ensure group read / write at runtime
 CMD umask 0002 && export HOME=/home/app/text-generation-webui && ./start_linux.sh --listen
diff --git a/docker/nvidia/docker-compose.yml b/docker/nvidia/docker-compose.yml
index 23d5cacca5..ee8282747d 100644
--- a/docker/nvidia/docker-compose.yml
+++ b/docker/nvidia/docker-compose.yml
@@ -4,25 +4,11 @@ services:
     build:
       context: .
       args:
-        # Requirements file to use: 
-        # | GPU | requirements file to use |
-        # |--------|---------|
-        # | NVIDIA | `requirements.txt` |
-        # | AMD | `requirements_amd.txt` |
-        # | CPU only | `requirements_cpu_only.txt` |
-        # | Apple Intel | `requirements_apple_intel.txt` |
-        # | Apple Silicon | `requirements_apple_silicon.txt` |
-        # Default: requirements.txt`
-        # BUILD_REQUIREMENTS: requirements.txt
-
-        # Extension requirements to build: 
-        # BUILD_EXTENSIONS: 
-
         # specify which cuda version your card supports: https://developer.nvidia.com/cuda-gpus
-        TORCH_CUDA_ARCH_LIST: ${TORCH_CUDA_ARCH_LIST:-7.5} 
+        TORCH_CUDA_ARCH_LIST: ${TORCH_CUDA_ARCH_LIST:-8.6;8.9+PTX}
         BUILD_EXTENSIONS: ${BUILD_EXTENSIONS:-}
-        APP_GID: ${APP_GID:-6972} 
-        APP_UID: ${APP_UID:-6972} 
+        APP_GID: ${APP_GID:-6972}
+        APP_UID: ${APP_UID:-6972}
     env_file: .env
     user: "${APP_RUNTIME_UID:-6972}:${APP_RUNTIME_GID:-6972}"
     ports:
diff --git a/docs/09 - Docker.md b/docs/09 - Docker.md
index eec8fafd6a..4cc147faf7 100644
--- a/docs/09 - Docker.md	
+++ b/docs/09 - Docker.md	
@@ -1,208 +1,52 @@
 Docker Compose is a way of installing and launching the web UI in an isolated Ubuntu image using only a few commands.
 
-## Installing Docker Compose
+## Prerequisites
 
-In order to create the image as described in the main README, you must have Docker Compose installed (2.17 or higher is recommended):
+You need Docker Compose v2.17 or higher:
 
 ```
 ~$ docker compose version
 Docker Compose version v2.21.0
 ```
 
-The installation instructions for various Linux distributions can be found here:
+Installation instructions: https://docs.docker.com/engine/install/
 
-https://docs.docker.com/engine/install/ubuntu/#install-using-the-repository
+For NVIDIA GPUs, you also need the [NVIDIA Container Toolkit](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html).
 
-## Launching the image
+## Quick start
 
-Use these commands to launch the image:
+There are four Docker variants available under `docker/`:
 
-```
-cd text-generation-webui
-ln -s docker/{nvidia/Dockerfile,nvidia/docker-compose.yml,.dockerignore} .
-cp docker/.env.example .env
-# Edit .env and set TORCH_CUDA_ARCH_LIST based on your GPU model
-docker compose up --build
-```
+| Directory | GPU | Notes |
+|-----------|-----|-------|
+| `docker/nvidia` | NVIDIA | Requires NVIDIA Container Toolkit |
+| `docker/amd` | AMD | Requires ROCm-compatible GPU |
+| `docker/intel` | Intel Arc | Beta support |
+| `docker/cpu` | None | CPU-only inference |
 
-## More detailed installation instructions
-
-* [Docker Compose installation instructions](#docker-compose-installation-instructions)
-* [Repository with additional Docker files](#dedicated-docker-repository)
-
-By [@loeken](https://github.com/loeken).
-
-- [Ubuntu 22.04](#ubuntu-2204)
-  - [0. youtube video](#0-youtube-video)
-  - [1. update the drivers](#1-update-the-drivers)
-  - [2. reboot](#2-reboot)
-  - [3. install docker](#3-install-docker)
-  - [4. docker \& container toolkit](#4-docker--container-toolkit)
-  - [5. clone the repo](#5-clone-the-repo)
-  - [6. prepare models](#6-prepare-models)
-  - [7. prepare .env file](#7-prepare-env-file)
-  - [8. startup docker container](#8-startup-docker-container)
-- [Manjaro](#manjaro)
-  - [update the drivers](#update-the-drivers)
-  - [reboot](#reboot)
-  - [docker \& container toolkit](#docker--container-toolkit)
-  - [continue with ubuntu task](#continue-with-ubuntu-task)
-- [Windows](#windows)
-  - [0. youtube video](#0-youtube-video-1)
-  - [1. choco package manager](#1-choco-package-manager)
-  - [2. install drivers/dependencies](#2-install-driversdependencies)
-  - [3. install wsl](#3-install-wsl)
-  - [4. reboot](#4-reboot)
-  - [5. git clone \&\& startup](#5-git-clone--startup)
-  - [6. prepare models](#6-prepare-models-1)
-  - [7. startup](#7-startup)
-- [notes](#notes)
-
-### Ubuntu 22.04
-
-#### 0. youtube video
-A video walking you through the setup can be found here:
-
-[![oobabooga text-generation-webui setup in docker on ubuntu 22.04](https://img.youtube.com/vi/ELkKWYh8qOk/0.jpg)](https://www.youtube.com/watch?v=ELkKWYh8qOk)
-
-
-#### 1. update the drivers
-in the the “software updater” update drivers to the last version of the prop driver.
-
-#### 2. reboot
-to switch using to new driver
-
-#### 3. install docker
-```bash
-sudo apt update
-sudo apt-get install curl
-sudo mkdir -m 0755 -p /etc/apt/keyrings
-curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /etc/apt/keyrings/docker.gpg
-echo \
-  "deb [arch="$(dpkg --print-architecture)" signed-by=/etc/apt/keyrings/docker.gpg] https://download.docker.com/linux/ubuntu \
-  "$(. /etc/os-release && echo "$VERSION_CODENAME")" stable" | \
-  sudo tee /etc/apt/sources.list.d/docker.list > /dev/null
-sudo apt update
-sudo apt-get install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin docker-compose -y
-sudo usermod -aG docker $USER
-newgrp docker
-```
+To launch (using NVIDIA as an example):
 
-#### 4. docker & container toolkit
-```bash
-curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg
-echo "deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://nvidia.github.io/libnvidia-container/stable/ubuntu22.04/amd64 /" | \
-sudo tee /etc/apt/sources.list.d/nvidia.list > /dev/null 
-sudo apt update
-sudo apt install nvidia-docker2 nvidia-container-runtime -y
-sudo systemctl restart docker
-```
-
-#### 5. clone the repo
-```
-git clone https://github.com/oobabooga/text-generation-webui
-cd text-generation-webui
-```
-
-#### 6. prepare models
-download and place the models inside the models folder. tested with:
-
-4bit
-https://github.com/oobabooga/text-generation-webui/pull/530#issuecomment-1483891617
-https://github.com/oobabooga/text-generation-webui/pull/530#issuecomment-1483941105
-
-8bit:
-https://github.com/oobabooga/text-generation-webui/pull/530#issuecomment-1484235789
-
-#### 7. prepare .env file
-edit .env values to your needs.
-```bash
-cp .env.example .env
-nano .env
-```
-
-#### 8. startup docker container
 ```bash
+cd text-generation-webui/docker/nvidia
+cp ../.env.example .env
+# Optionally edit .env to customize ports, TORCH_CUDA_ARCH_LIST, etc.
 docker compose up --build
 ```
 
-### Manjaro
-manjaro/arch is similar to ubuntu just the dependency installation is more convenient
-
-#### update the drivers
-```bash
-sudo mhwd -a pci nonfree 0300
-```
-#### reboot
-```bash
-reboot
-```
-#### docker & container toolkit
-```bash
-yay -S docker docker-compose buildkit gcc nvidia-docker
-sudo usermod -aG docker $USER
-newgrp docker
-sudo systemctl restart docker # required by nvidia-container-runtime
-```
-
-#### continue with ubuntu task
-continue at [5. clone the repo](#5-clone-the-repo)
-
-### Windows
-#### 0. youtube video
-A video walking you through the setup can be found here:
-[![oobabooga text-generation-webui setup in docker on windows 11](https://img.youtube.com/vi/ejH4w5b5kFQ/0.jpg)](https://www.youtube.com/watch?v=ejH4w5b5kFQ)
-
-#### 1. choco package manager
-install package manager  (https://chocolatey.org/ )
-```
-Set-ExecutionPolicy Bypass -Scope Process -Force; [System.Net.ServicePointManager]::SecurityProtocol = [System.Net.ServicePointManager]::SecurityProtocol -bor 3072; iex ((New-Object System.Net.WebClient).DownloadString('https://community.chocolatey.org/install.ps1'))
-```
-
-#### 2. install drivers/dependencies
-```
-choco install nvidia-display-driver cuda git docker-desktop
-```
+The web UI will be available at `http://localhost:7860`.
 
-#### 3. install wsl
-wsl --install
+## User data
 
-#### 4. reboot
-after reboot enter username/password in wsl
+Create a `user_data/` directory next to the `docker-compose.yml` to persist your models, characters, presets, and settings between container rebuilds:
 
-#### 5. git clone && startup
-clone the repo and edit .env values to your needs.
-```
-cd Desktop
-git clone https://github.com/oobabooga/text-generation-webui
-cd text-generation-webui
-COPY .env.example .env
-notepad .env
-```
-
-#### 6. prepare models
-download and place the models inside the models folder. tested with:
-
-4bit https://github.com/oobabooga/text-generation-webui/pull/530#issuecomment-1483891617 https://github.com/oobabooga/text-generation-webui/pull/530#issuecomment-1483941105
-
-8bit: https://github.com/oobabooga/text-generation-webui/pull/530#issuecomment-1484235789
-
-#### 7. startup
-```
-docker compose up
+```bash
+mkdir -p user_data
 ```
 
-### notes
+This directory is mounted into the container at runtime. You can place a `CMD_FLAGS.txt` inside it to pass persistent flags to the web UI (e.g., `--api`).
 
-on older ubuntus you can manually install the docker compose plugin like this:
-```
-DOCKER_CONFIG=${DOCKER_CONFIG:-$HOME/.docker}
-mkdir -p $DOCKER_CONFIG/cli-plugins
-curl -SL https://github.com/docker/compose/releases/download/v2.17.2/docker-compose-linux-x86_64 -o $DOCKER_CONFIG/cli-plugins/docker-compose
-chmod +x $DOCKER_CONFIG/cli-plugins/docker-compose
-export PATH="$HOME/.docker/cli-plugins:$PATH"
-```
+Models can be downloaded through the web UI's “Model” tab once it's running, and they will be saved to `user_data/models/`.
 
 ## Dedicated docker repository
 
-An external repository maintains a docker wrapper for this project as well as several pre-configured 'one-click' `docker compose` variants (e.g., updated branches of GPTQ). It can be found at: [Atinoda/text-generation-webui-docker](https://github.com/Atinoda/text-generation-webui-docker).
+An external repository maintains a docker wrapper for this project as well as several pre-configured 'one-click' `docker compose` variants. It can be found at: [Atinoda/text-generation-webui-docker](https://github.com/Atinoda/text-generation-webui-docker).

From ff48956cb07d36ae3c54850cfa1f69593d4ecbf4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 18:19:56 -0800
Subject: [PATCH 1206/1701] UI: Remove unnecessary server round-trips from
 button click chains

---
 modules/ui_chat.py | 13 +++++--------
 1 file changed, 5 insertions(+), 8 deletions(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index aec9051ea5..516488a3b5 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -214,38 +214,35 @@ def create_event_handlers():
 
     shared.gradio['Generate'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        lambda x: (x, {"text": "", "files": []}), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False).then(
-        lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
+        lambda x: (x, {"text": "", "files": []}), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False, js='(x) => { document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating"); return x; }').then(
         chat.generate_chat_reply_wrapper, gradio(inputs), gradio('display', 'history'), show_progress=False).then(
         None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
 
     shared.gradio['textbox'].submit(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        lambda x: (x, {"text": "", "files": []}), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False).then(
-        lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
+        lambda x: (x, {"text": "", "files": []}), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False, js='(x) => { document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating"); return x; }').then(
         chat.generate_chat_reply_wrapper, gradio(inputs), gradio('display', 'history'), show_progress=False).then(
         None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
 
     shared.gradio['Regenerate'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
+        None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
         partial(chat.generate_chat_reply_wrapper, regenerate=True), gradio(inputs), gradio('display', 'history'), show_progress=False).then(
         None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
 
     shared.gradio['Continue'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
+        None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
         partial(chat.generate_chat_reply_wrapper, _continue=True), gradio(inputs), gradio('display', 'history'), show_progress=False).then(
         None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
 
     shared.gradio['Impersonate'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        lambda x: x, gradio('textbox'), gradio('Chat input'), show_progress=False).then(
-        lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
+        lambda x: x, gradio('textbox'), gradio('Chat input'), show_progress=False, js='(x) => { document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating"); return x; }').then(
         chat.impersonate_wrapper, gradio(inputs), gradio('textbox', 'display'), show_progress=False).then(
         None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')

From 6a08e79fa5fbe93cbfaee2292df386d1053c168d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 18:22:50 -0800
Subject: [PATCH 1207/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 14 files changed, 28 insertions(+), 28 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 921cf6781b..5b3139c11d 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -33,8 +33,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 8810ff4bf8..4735014363 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index e0a8627ee3..528224a6b1 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index d9732e91a6..29eca60f4e 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 0088adb005..ae31507a56 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index e060038ad3..386ee2a878 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 88bf31268a..40694564f1 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 19031c213d..64a2954d12 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index c98d60ac97..5669b4e190 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 797586521d..ed31dbaba1 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index f932f3ad5f..ed100f0e7e 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 049b1d0496..829e89730b 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index ff1aeb988c..872a94151d 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index bf37c4458d..cd140cd58e 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio-4.37.2+custom.6-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.6/gradio_client-1.0.2+custom.6-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From 0804296f4d362174103b60d5ea6e15ea7fd38470 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 18:41:30 -0800
Subject: [PATCH 1208/1701] Revert "UI: Remove unnecessary server round-trips
 from button click chains"

This reverts commit ff48956cb07d36ae3c54850cfa1f69593d4ecbf4.
---
 modules/ui_chat.py | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 516488a3b5..aec9051ea5 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -214,35 +214,38 @@ def create_event_handlers():
 
     shared.gradio['Generate'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        lambda x: (x, {"text": "", "files": []}), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False, js='(x) => { document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating"); return x; }').then(
+        lambda x: (x, {"text": "", "files": []}), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False).then(
+        lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
         chat.generate_chat_reply_wrapper, gradio(inputs), gradio('display', 'history'), show_progress=False).then(
         None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
 
     shared.gradio['textbox'].submit(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        lambda x: (x, {"text": "", "files": []}), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False, js='(x) => { document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating"); return x; }').then(
+        lambda x: (x, {"text": "", "files": []}), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False).then(
+        lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
         chat.generate_chat_reply_wrapper, gradio(inputs), gradio('display', 'history'), show_progress=False).then(
         None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
 
     shared.gradio['Regenerate'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
+        lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
         partial(chat.generate_chat_reply_wrapper, regenerate=True), gradio(inputs), gradio('display', 'history'), show_progress=False).then(
         None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
 
     shared.gradio['Continue'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
+        lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
         partial(chat.generate_chat_reply_wrapper, _continue=True), gradio(inputs), gradio('display', 'history'), show_progress=False).then(
         None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
 
     shared.gradio['Impersonate'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        lambda x: x, gradio('textbox'), gradio('Chat input'), show_progress=False, js='(x) => { document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating"); return x; }').then(
+        lambda x: x, gradio('textbox'), gradio('Chat input'), show_progress=False).then(
+        lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then(
         chat.impersonate_wrapper, gradio(inputs), gradio('textbox', 'display'), show_progress=False).then(
         None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')

From d45aa6606a439b42e053345abd3e053bd45e2487 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 19:07:55 -0800
Subject: [PATCH 1209/1701] Fix blank prompt dropdown in Notebook/Default tabs
 on first startup

---
 modules/utils.py |  6 ++++++
 server.py        | 14 ++++++++------
 2 files changed, 14 insertions(+), 6 deletions(-)

diff --git a/modules/utils.py b/modules/utils.py
index d366784730..1490019c5e 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -209,6 +209,12 @@ def get_available_prompts():
     notebook_dir.mkdir(parents=True, exist_ok=True)
 
     prompt_files = list(notebook_dir.glob('*.txt'))
+    if not prompt_files:
+        new_name = current_time()
+        new_path = notebook_dir / f"{new_name}.txt"
+        new_path.write_text("In this story,", encoding='utf-8')
+        prompt_files = [new_path]
+
     sorted_files = sorted(prompt_files, key=lambda x: x.stat().st_mtime, reverse=True)
     prompts = [file.stem for file in sorted_files]
     return prompts
diff --git a/server.py b/server.py
index 2d0e567121..000ea9fb4a 100644
--- a/server.py
+++ b/server.py
@@ -109,12 +109,14 @@ def create_interface():
         'filter_by_loader': (shared.args.loader or 'All') if not shared.args.portable else 'llama.cpp'
     })
 
-    if shared.settings['prompt-notebook']:
-        prompt = load_prompt(shared.settings['prompt-notebook'])
-        shared.persistent_interface_state.update({
-            'textbox-default': prompt,
-            'textbox-notebook': prompt
-        })
+    if not shared.settings['prompt-notebook']:
+        shared.settings['prompt-notebook'] = utils.get_available_prompts()[0]
+
+    prompt = load_prompt(shared.settings['prompt-notebook'])
+    shared.persistent_interface_state.update({
+        'textbox-default': prompt,
+        'textbox-notebook': prompt
+    })
 
     # Clear existing cache files
     for cache_file in ['pfp_character.png', 'pfp_character_thumb.png']:

From fbfcd59fe01310137dca5001107575237064e422 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 19:21:45 -0800
Subject: [PATCH 1210/1701] llama.cpp: Use -1 instead of 0 for auto gpu_layers

---
 modules/llama_cpp_server.py | 5 +++--
 modules/models_settings.py  | 6 +++---
 modules/shared.py           | 2 +-
 modules/ui_model_menu.py    | 2 +-
 4 files changed, 8 insertions(+), 7 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 2631f253c7..a5e90c9402 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -347,7 +347,7 @@ def _start_server(self):
             "--flash-attn", "on",
         ]
 
-        if shared.args.gpu_layers > 0:
+        if shared.args.gpu_layers >= 0:
             cmd += ["--gpu-layers", str(shared.args.gpu_layers), "--fit", "off"]
         else:
             cmd += ["--fit", "on"]
@@ -448,7 +448,8 @@ def _start_server(self):
             print(' '.join(str(item) for item in cmd[1:]))
             print()
 
-        logger.info(f"Using gpu_layers={shared.args.gpu_layers} | ctx_size={shared.args.ctx_size} | cache_type={cache_type}")
+        gpu_layers_str = "auto" if shared.args.gpu_layers < 0 else str(shared.args.gpu_layers)
+        logger.info(f"Using gpu_layers={gpu_layers_str} | ctx_size={shared.args.ctx_size} | cache_type={cache_type}")
         # Start the server with pipes for output
         self.process = subprocess.Popen(
             cmd,
diff --git a/modules/models_settings.py b/modules/models_settings.py
index 3b28a800f5..1ef436e045 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -77,7 +77,7 @@ def get_model_metadata(model):
             elif k.endswith('rope.scaling.factor'):
                 model_settings['compress_pos_emb'] = metadata[k]
             elif k.endswith('.block_count'):
-                model_settings['gpu_layers'] = 0
+                model_settings['gpu_layers'] = -1
                 model_settings['max_gpu_layers'] = metadata[k] + 1
 
         if 'tokenizer.chat_template' in metadata:
@@ -264,7 +264,7 @@ def apply_model_settings_to_state(model, state):
 
     # Handle GPU layers and VRAM update for llama.cpp
     if state['loader'] == 'llama.cpp' and 'gpu_layers' in model_settings:
-        gpu_layers = model_settings['gpu_layers']  # 0 (auto) by default, or user-saved value
+        gpu_layers = model_settings['gpu_layers']  # -1 (auto) by default, or user-saved value
         max_layers = model_settings.get('max_gpu_layers', 256)
         state['gpu_layers'] = gr.update(value=gpu_layers, maximum=max_layers)
 
@@ -418,7 +418,7 @@ def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type):
     Compute the estimated VRAM usage for the given GPU layers and return
     an HTML string for the UI display.
     """
-    if loader != 'llama.cpp' or model in ["None", None] or not model.endswith(".gguf"):
+    if loader != 'llama.cpp' or model in ["None", None] or not model.endswith(".gguf") or gpu_layers < 0:
         return "
            Estimated VRAM to load the model:
            "
 
     vram_usage = estimate_vram(model, gpu_layers, ctx_size, cache_type)
diff --git a/modules/shared.py b/modules/shared.py
index c917377a3f..787f04d9ed 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -88,7 +88,7 @@
 
 # llama.cpp
 group = parser.add_argument_group('llama.cpp')
-group.add_argument('--gpu-layers', '--n-gpu-layers', type=int, default=0, metavar='N', help='Number of layers to offload to the GPU. 0 means auto (llama.cpp decides via --fit).')
+group.add_argument('--gpu-layers', '--n-gpu-layers', type=int, default=-1, metavar='N', help='Number of layers to offload to the GPU. Set to -1 for auto mode, where llama.cpp decides via --fit.')
 group.add_argument('--cpu-moe', action='store_true', help='Move the experts to the CPU (for MoE models).')
 group.add_argument('--mmproj', type=str, default=None, help='Path to the mmproj file for vision models.')
 group.add_argument('--streaming-llm', action='store_true', help='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index c65c7a1b47..430123eb21 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -41,7 +41,7 @@ def create_ui():
                     gr.Markdown("## Main options")
                     with gr.Row():
                         with gr.Column():
-                            shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='0 = auto (llama.cpp decides via --fit). Set manually to override.')
+                            shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=-1, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Number of layers to offload to the GPU. -1 = auto (llama.cpp decides via --fit).')
                             shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. Common values: 4096, 8192, 16384, 32768, 65536, 131072.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')

From 69fa4dd0b195e8ff0c230f28170bb5c63d82fbfe Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 4 Mar 2026 19:33:20 -0800
Subject: [PATCH 1211/1701] llama.cpp: allow ctx_size=0 for auto context via
 --fit

---
 modules/llama_cpp_server.py | 7 +++++--
 modules/models.py           | 3 ++-
 modules/models_settings.py  | 2 +-
 modules/shared.py           | 4 ++--
 modules/ui_model_menu.py    | 4 ++--
 modules/ui_parameters.py    | 2 +-
 6 files changed, 13 insertions(+), 9 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index a5e90c9402..2af9aa8a14 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -339,7 +339,6 @@ def _start_server(self):
         cmd = [
             self.server_path,
             "--model", self.model_path,
-            "--ctx-size", str(shared.args.ctx_size),
             "--batch-size", str(shared.args.batch_size),
             "--ubatch-size", str(shared.args.ubatch_size),
             "--port", str(self.port),
@@ -347,6 +346,9 @@ def _start_server(self):
             "--flash-attn", "on",
         ]
 
+        if shared.args.ctx_size > 0:
+            cmd += ["--ctx-size", str(shared.args.ctx_size)]
+
         if shared.args.gpu_layers >= 0:
             cmd += ["--gpu-layers", str(shared.args.gpu_layers), "--fit", "off"]
         else:
@@ -449,7 +451,8 @@ def _start_server(self):
             print()
 
         gpu_layers_str = "auto" if shared.args.gpu_layers < 0 else str(shared.args.gpu_layers)
-        logger.info(f"Using gpu_layers={gpu_layers_str} | ctx_size={shared.args.ctx_size} | cache_type={cache_type}")
+        ctx_size_str = "auto" if shared.args.ctx_size == 0 else str(shared.args.ctx_size)
+        logger.info(f"Using gpu_layers={gpu_layers_str} | ctx_size={ctx_size_str} | cache_type={cache_type}")
         # Start the server with pipes for output
         self.process = subprocess.Popen(
             cmd,
diff --git a/modules/models.py b/modules/models.py
index 8c0f1c3700..2dd69c63e6 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -54,7 +54,8 @@ def load_model(model_name, loader=None):
 
     shared.settings.update({k: v for k, v in metadata.items() if k in shared.settings})
     if loader.lower().startswith('exllama') or loader.lower().startswith('tensorrt') or loader == 'llama.cpp':
-        shared.settings['truncation_length'] = shared.args.ctx_size
+        if shared.args.ctx_size > 0:
+            shared.settings['truncation_length'] = shared.args.ctx_size
 
     shared.is_multimodal = False
     if loader.lower() in ('exllamav3', 'llama.cpp') and hasattr(model, 'is_multimodal'):
diff --git a/modules/models_settings.py b/modules/models_settings.py
index 1ef436e045..1d784a8787 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -418,7 +418,7 @@ def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type):
     Compute the estimated VRAM usage for the given GPU layers and return
     an HTML string for the UI display.
     """
-    if loader != 'llama.cpp' or model in ["None", None] or not model.endswith(".gguf") or gpu_layers < 0:
+    if loader != 'llama.cpp' or model in ["None", None] or not model.endswith(".gguf") or gpu_layers < 0 or ctx_size == 0:
         return "
            Estimated VRAM to load the model:
            "
 
     vram_usage = estimate_vram(model, gpu_layers, ctx_size, cache_type)
diff --git a/modules/shared.py b/modules/shared.py
index 787f04d9ed..c80db29851 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -71,7 +71,7 @@
 
 # Cache
 group = parser.add_argument_group('Context and cache')
-group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=8192, metavar='N', help='Context size in tokens.')
+group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=8192, metavar='N', help='Context size in tokens. llama.cpp: 0 = auto if gpu-layers is also -1.')
 group.add_argument('--cache-type', '--cache_type', type=str, default='fp16', metavar='N', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).')
 
 # Speculative decoding
@@ -88,7 +88,7 @@
 
 # llama.cpp
 group = parser.add_argument_group('llama.cpp')
-group.add_argument('--gpu-layers', '--n-gpu-layers', type=int, default=-1, metavar='N', help='Number of layers to offload to the GPU. Set to -1 for auto mode, where llama.cpp decides via --fit.')
+group.add_argument('--gpu-layers', '--n-gpu-layers', type=int, default=-1, metavar='N', help='Number of layers to offload to the GPU. -1 = auto.')
 group.add_argument('--cpu-moe', action='store_true', help='Move the experts to the CPU (for MoE models).')
 group.add_argument('--mmproj', type=str, default=None, help='Path to the mmproj file for vision models.')
 group.add_argument('--streaming-llm', action='store_true', help='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 430123eb21..de9901e364 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -41,8 +41,8 @@ def create_ui():
                     gr.Markdown("## Main options")
                     with gr.Row():
                         with gr.Column():
-                            shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=-1, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Number of layers to offload to the GPU. -1 = auto (llama.cpp decides via --fit).')
-                            shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. Common values: 4096, 8192, 16384, 32768, 65536, 131072.')
+                            shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=-1, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Number of layers to offload to the GPU. -1 = auto.')
+                            shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=0, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. llama.cpp: 0 = auto if gpu-layers is also -1. Common values: 4096, 8192, 16384, 32768, 65536, 131072.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index dc07d2915e..23882084de 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -128,7 +128,7 @@ def create_event_handlers():
 
 
 def get_truncation_length():
-    if 'ctx_size' in shared.provided_arguments or shared.args.ctx_size != shared.args_defaults.ctx_size:
+    if shared.args.ctx_size > 0 and ('ctx_size' in shared.provided_arguments or shared.args.ctx_size != shared.args_defaults.ctx_size):
         return shared.args.ctx_size
     else:
         return shared.settings['truncation_length']

From 268cc3f10031f6b6faecbc23daf327752d3c3591 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 09:32:28 -0300
Subject: [PATCH 1212/1701] Update TensorRT-LLM to v1.1.0

---
 docker/TensorRT-LLM/Dockerfile |  2 +-
 modules/shared.py              |  2 +-
 modules/tensorrt_llm.py        | 41 +++++++++-------------------------
 3 files changed, 13 insertions(+), 32 deletions(-)

diff --git a/docker/TensorRT-LLM/Dockerfile b/docker/TensorRT-LLM/Dockerfile
index c2be3efe65..3c989d79f7 100644
--- a/docker/TensorRT-LLM/Dockerfile
+++ b/docker/TensorRT-LLM/Dockerfile
@@ -18,7 +18,7 @@ RUN pip install -r requirements/full/requirements.txt
 ENV LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu:$LD_LIBRARY_PATH
 
 # Install TensorRT-LLM
-RUN pip3 install tensorrt_llm==0.10.0 -U --pre --extra-index-url https://pypi.nvidia.com
+RUN pip3 install tensorrt_llm==1.1.0 -U --extra-index-url https://pypi.nvidia.com
 
 # Expose the necessary port for the Python server
 EXPOSE 7860 5000
diff --git a/modules/shared.py b/modules/shared.py
index c80db29851..8d617ecd0d 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -142,7 +142,7 @@
 
 # TensorRT-LLM
 group = parser.add_argument_group('TensorRT-LLM')
-group.add_argument('--cpp-runner', action='store_true', help='Use the ModelRunnerCpp runner, which is faster than the default ModelRunner but doesn\'t support streaming yet.')
+group.add_argument('--cpp-runner', action='store_true', help='Use the ModelRunnerCpp runner, which is faster than the default ModelRunner.')
 
 # RoPE
 group = parser.add_argument_group('RoPE')
diff --git a/modules/tensorrt_llm.py b/modules/tensorrt_llm.py
index 0527d493f8..161dde824c 100644
--- a/modules/tensorrt_llm.py
+++ b/modules/tensorrt_llm.py
@@ -17,7 +17,7 @@ def __init__(self):
         pass
 
     @classmethod
-    def from_pretrained(self, path_to_model):
+    def from_pretrained(cls, path_to_model):
 
         path_to_model = Path(f'{shared.args.model_dir}') / Path(path_to_model)
         runtime_rank = tensorrt_llm.mpi_rank()
@@ -35,11 +35,7 @@ def from_pretrained(self, path_to_model):
             logger.info("TensorRT-LLM: Using \"ModelRunnerCpp\"")
             runner_kwargs.update(
                 max_batch_size=1,
-                max_input_len=shared.args.ctx_size - 512,
-                max_output_len=512,
                 max_beam_width=1,
-                max_attention_window_size=None,
-                sink_token_length=None,
             )
         else:
             logger.info("TensorRT-LLM: Using \"ModelRunner\"")
@@ -48,7 +44,7 @@ def from_pretrained(self, path_to_model):
         runner_cls = ModelRunnerCpp if shared.args.cpp_runner else ModelRunner
         runner = runner_cls.from_dir(**runner_kwargs)
 
-        result = self()
+        result = cls()
         result.model = runner
         result.runtime_rank = runtime_rank
 
@@ -65,9 +61,7 @@ def generate_with_streaming(self, prompt, state):
         input_ids = input_ids[-get_max_prompt_length(state):]  # Apply truncation_length
         batch_input_ids.append(input_ids)
 
-        if shared.args.cpp_runner:
-            max_new_tokens = min(512, state['max_new_tokens'])
-        elif state['auto_max_new_tokens']:
+        if state['auto_max_new_tokens']:
             max_new_tokens = state['truncation_length'] - input_ids.shape[-1]
         else:
             max_new_tokens = state['max_new_tokens']
@@ -76,27 +70,22 @@ def generate_with_streaming(self, prompt, state):
             generator = self.model.generate(
                 batch_input_ids,
                 max_new_tokens=max_new_tokens,
-                max_attention_window_size=None,
-                sink_token_length=None,
                 end_id=shared.tokenizer.eos_token_id if not state['ban_eos_token'] else -1,
                 pad_id=shared.tokenizer.pad_token_id or shared.tokenizer.eos_token_id,
                 temperature=state['temperature'],
                 top_k=state['top_k'],
                 top_p=state['top_p'],
-                num_beams=1,
-                length_penalty=1.0,
                 repetition_penalty=state['repetition_penalty'],
                 presence_penalty=state['presence_penalty'],
                 frequency_penalty=state['frequency_penalty'],
                 stop_words_list=None,
                 bad_words_list=None,
                 lora_uids=None,
-                prompt_table_path=None,
+                prompt_table=None,
                 prompt_tasks=None,
-                streaming=not shared.args.cpp_runner,
+                streaming=True,
                 output_sequence_lengths=True,
                 return_dict=True,
-                medusa_choices=None
             )
 
         torch.cuda.synchronize()
@@ -104,24 +93,16 @@ def generate_with_streaming(self, prompt, state):
         cumulative_reply = ''
         starting_from = batch_input_ids[0].shape[-1]
 
-        if shared.args.cpp_runner:
-            sequence_length = generator['sequence_lengths'][0].item()
-            output_ids = generator['output_ids'][0][0][:sequence_length].tolist()
+        for curr_outputs in generator:
+            if shared.stop_everything:
+                break
+
+            sequence_length = curr_outputs['sequence_lengths'][0].item()
+            output_ids = curr_outputs['output_ids'][0][0][:sequence_length].tolist()
 
             cumulative_reply += get_reply_from_output_ids(output_ids, state, starting_from=starting_from)
             starting_from = sequence_length
             yield cumulative_reply
-        else:
-            for curr_outputs in generator:
-                if shared.stop_everything:
-                    break
-
-                sequence_length = curr_outputs['sequence_lengths'][0].item()
-                output_ids = curr_outputs['output_ids'][0][0][:sequence_length].tolist()
-
-                cumulative_reply += get_reply_from_output_ids(output_ids, state, starting_from=starting_from)
-                starting_from = sequence_length
-                yield cumulative_reply
 
     def generate(self, prompt, state):
         output = ''

From 45188eccef1de3f01e7a565a7bb4004c02480959 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 10:49:08 -0300
Subject: [PATCH 1213/1701] Overhaul LoRA training tab
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- Use peft's "all-linear" for target modules instead of the old
  model_to_lora_modules mapping (only knew ~39 model types)
- Add "Target all linear layers" checkbox, on by default
- Fix labels in tokenize() — were [1]s instead of actual token IDs
- Replace DataCollatorForLanguageModeling with custom collate_fn
- Raw text: concatenate-and-split instead of overlapping chunks
- Adapter backup/loading: check safetensors before bin
- Fix report_to=None crash on transformers 5.x
- Fix no_cuda deprecation for transformers 5.x (use use_cpu)
- Move torch.compile before Trainer init
- Add remove_unused_columns=False (torch.compile breaks column detection)
- Guard against no target modules selected
- Set tracked.did_save so we don't always save twice
- pad_token_id: fall back to eos_token_id instead of hardcoding 0
- Drop MODEL_CLASSES, split_chunks, cut_chunk_for_newline
- Update docs
---
 docs/05 - Training Tab.md |  24 ++---
 modules/training.py       | 203 +++++++++++++++++---------------------
 2 files changed, 103 insertions(+), 124 deletions(-)

diff --git a/docs/05 - Training Tab.md b/docs/05 - Training Tab.md
index eba62cb354..983659606c 100644
--- a/docs/05 - Training Tab.md	
+++ b/docs/05 - Training Tab.md	
@@ -32,7 +32,7 @@ The WebUI seeks to make training your own LoRAs as easy as possible. It comes do
 - Make sure to unload the LoRA before training it.
 - You can simply resume a prior run - use `Copy parameters from` to select your LoRA, and edit parameters. Note that you cannot change the `Rank` of an already created LoRA.
     - If you want to resume from a checkpoint saved along the way, simply copy the contents of the checkpoint folder into the LoRA's folder.
-    - (Note: `adapter_model.bin` is the important file that holds the actual LoRA content).
+    - (Note: `adapter_model.safetensors` or `adapter_model.bin` is the important file that holds the actual LoRA content).
     - This will start Learning Rate and Steps back to the start. If you want to resume as if you were midway through, you can adjust your Learning Rate to the last reported LR in logs and reduce your epochs.
 - Or, you can start over entirely if you prefer.
 - If your model is producing corrupted outputs, you probably need to start over and use a lower Learning Rate.
@@ -76,10 +76,14 @@ If you have different sets of key inputs, you can make your own format file to m
 
 ## Raw Text File Settings
 
-When using raw text files as your dataset, the text is automatically split into chunks based on your `Cutoff Length` you get a few basic options to configure them.
-- `Overlap Length` is how much to overlap chunks by. Overlapping chunks helps prevent the model from learning strange mid-sentence cuts, and instead learn continual sentences that flow from earlier text.
-- `Prefer Newline Cut Length` sets a maximum distance in characters to shift the chunk cut towards newlines. Doing this helps prevent lines from starting or ending mid-sentence, preventing the model from learning to cut off sentences randomly.
-- `Hard Cut String` sets a string that indicates there must be a hard cut without overlap. This defaults to `\n\n\n`, meaning 3 newlines. No trained chunk will ever contain this string. This allows you to insert unrelated sections of text in the same text file, but still ensure the model won't be taught to randomly change the subject.
+When using raw text files as your dataset, the text is split into sections by the `Hard Cut String` (default `\n\n\n`), tokenized, concatenated into one long token sequence, and then split into non-overlapping chunks of exactly `Cutoff Length` tokens (any remainder shorter than the cutoff is dropped). This is the standard concatenate-and-split approach used by HuggingFace `run_clm.py`.
+- `Hard Cut String` sets a string that indicates a boundary between unrelated sections of text. This defaults to `\n\n\n`, meaning 3 newlines. When `Add EOS token` is enabled, an EOS token is appended after each section before concatenation. This allows you to insert unrelated sections of text in the same text file, ensuring the model learns proper boundaries between them.
+
+## Target Modules
+
+By default, **Target all linear layers** is enabled. This uses peft's `all-linear` mode, which applies LoRA to every `nn.Linear` layer in the model except the output head (`lm_head`). It works for any model architecture.
+
+If you uncheck it, you can manually select individual projection modules (`q_proj`, `k_proj`, `v_proj`, `o_proj`, `gate_proj`, `down_proj`, `up_proj`). Targeting fewer modules reduces VRAM usage and adapter size, but also reduces how much the model can learn. The default selection of `q_proj` + `v_proj` is the minimum for basic style/format training.
 
 ## Parameters
 
@@ -91,7 +95,7 @@ That said, here's a guide to the most important parameter choices you should con
 
 - First, you must consider your VRAM availability.
     - Generally, under default settings, VRAM usage for training with default parameters is very close to when generating text (with 1000+ tokens of context) (ie, if you can generate text, you can train LoRAs).
-        - Note: worse by default in the 4-bit monkeypatch currently. Reduce `Micro Batch Size` to `1` to restore this to expectations.
+        - Note: VRAM usage is higher when training 4-bit quantized models. Reduce `Micro Batch Size` to `1` to compensate.
     - If you have VRAM to spare, setting higher batch sizes will use more VRAM and get you better quality training in exchange.
     - If you have large data, setting a higher cutoff length may be beneficial, but will cost significant VRAM. If you can spare some, set your batch size to `1` and see how high you can push your cutoff length.
     - If you're low on VRAM, reducing batch size or cutoff length will of course improve that.
@@ -129,11 +133,3 @@ In practice, a good LLM should have a very complex variable range of ideas runni
 So, in effect, Loss is a balancing game: you want to get it low enough that it understands your data, but high enough that it isn't forgetting everything else. Generally, if it goes below `1.0`, it's going to start forgetting its prior memories, and you should stop training. In some cases you may prefer to take it as low as `0.5` (if you want it to be very very predictable). Different goals have different needs, so don't be afraid to experiment and see what works best for you.
 
 Note: if you see Loss start at or suddenly jump to exactly `0`, it is likely something has gone wrong in your training process (eg model corruption).
-
-## Note: 4-Bit Monkeypatch
-
-The [4-bit LoRA monkeypatch](GPTQ-models-(4-bit-mode).md#using-loras-in-4-bit-mode) works for training, but has side effects:
-- VRAM usage is higher currently. You can reduce the `Micro Batch Size` to `1` to compensate.
-- Models do funky things. LoRAs apply themselves, or refuse to apply, or spontaneously error out, or etc. It can be helpful to reload base model or restart the WebUI between training/usage to minimize chances of anything going haywire.
-- Loading or working with multiple LoRAs at the same time doesn't currently work.
-- Generally, recognize and treat the monkeypatch as the dirty temporary hack it is - it works, but isn't very stable. It will get better in time when everything is merged upstream for full official support.
diff --git a/modules/training.py b/modules/training.py
index 41b2028ef2..691f99219e 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -26,7 +26,7 @@
 from modules.models import reload_model
 from modules.utils import natural_keys
 
-PARAMETERS = ["lora_name", "always_override", "q_proj_en", "v_proj_en", "k_proj_en", "o_proj_en", "gate_proj_en", "down_proj_en", "up_proj_en", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "raw_text_file", "overlap_len", "newline_favor_len", "higher_rank_limit", "warmup_steps", "optimizer", "hard_cut_string", "train_only_after", "stop_at_loss", "add_eos_token", "min_chars", "report_to"]
+PARAMETERS = ["lora_name", "always_override", "all_linear", "q_proj_en", "v_proj_en", "k_proj_en", "o_proj_en", "gate_proj_en", "down_proj_en", "up_proj_en", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "raw_text_file", "higher_rank_limit", "warmup_steps", "optimizer", "hard_cut_string", "train_only_after", "stop_at_loss", "add_eos_token", "min_chars", "report_to"]
 WANT_INTERRUPT = False
 
 train_log = {}
@@ -53,7 +53,8 @@ def create_ui():
                             always_override = gr.Checkbox(label='Override Existing Files', value=False, info='If the name is the same, checking will replace the existing file, and unchecking will load and continue from it (the rank must be the same).', elem_classes=['no-background'])
 
                     with gr.Accordion(label='Target Modules', open=False, elem_classes='tgw-accordion'):
-                        gr.Markdown("Selects which modules to target in training. Targeting more modules is closer to a full fine-tune at the cost of increased VRAM requirements and adapter size.\nNOTE: Only works for model_id='llama', other types will retain default training behavior and not use these settings.")
+                        gr.Markdown("Selects which modules to target in training. Targeting more modules is closer to a full fine-tune at the cost of increased VRAM requirements and adapter size.\n\"Target all linear layers\" uses peft's `all-linear` option, which targets every `nn.Linear` layer except `lm_head` and works for any model architecture. Uncheck it to manually select individual projection modules below.")
+                        all_linear = gr.Checkbox(label='Target all linear layers', value=True, info='Targets every nn.Linear layer except lm_head. Works for any model architecture.', elem_classes=['no-background'])
                         with gr.Row():
                             with gr.Column():
                                 q_proj_en = gr.Checkbox(label='Enable q_proj', value=True)
@@ -124,14 +125,8 @@ def create_ui():
                             raw_text_file = gr.Dropdown(choices=utils.get_datasets('user_data/training/datasets', 'txt'), value='None', label='Text file', info='The raw text file to use for training.', elem_classes=['slim-dropdown'], interactive=not mu)
                             ui.create_refresh_button(raw_text_file, lambda: None, lambda: {'choices': utils.get_datasets('user_data/training/datasets', 'txt')}, 'refresh-button', interactive=not mu)
 
-                        with gr.Row():
-                            with gr.Column():
-                                overlap_len = gr.Slider(label='Overlap Length', minimum=0, maximum=512, value=128, step=16, info='How many tokens from the prior chunk of text to include into the next chunk. (The chunks themselves will be of a size determined by Cutoff Length). Setting overlap to exactly half the cutoff length may be ideal.')
-                                newline_favor_len = gr.Slider(label='Prefer Newline Cut Length', minimum=0, maximum=512, value=128, step=16, info='Length (in characters, not tokens) of the maximum distance to shift an overlap cut by to ensure chunks cut at newlines. If too low, cuts may occur in the middle of lines.')
-
-                            with gr.Column():
-                                hard_cut_string = gr.Textbox(label='Hard Cut String', value='\\n\\n\\n', info='String that indicates a hard cut between text parts. Helps prevent unwanted overlap.')
-                                min_chars = gr.Number(label='Ignore small blocks', value=0, info='Ignore Hard Cut blocks that have less or equal characters than this number')
+                        hard_cut_string = gr.Textbox(label='Hard Cut String', value='\\n\\n\\n', info='String that indicates a hard cut between text parts. Helps prevent unwanted overlap.')
+                        min_chars = gr.Number(label='Ignore small blocks', value=0, info='Ignore Hard Cut blocks that have less or equal characters than this number')
 
                     with gr.Row():
                         start_button = gr.Button("Start LoRA Training", variant='primary', interactive=not mu)
@@ -165,7 +160,7 @@ def create_ui():
                 refresh_table = gr.Button('Refresh the table', elem_classes="small-button", interactive=not mu)
 
     # Training events
-    all_params = [lora_name, always_override, q_proj_en, v_proj_en, k_proj_en, o_proj_en, gate_proj_en, down_proj_en, up_proj_en, save_steps, micro_batch_size, batch_size, epochs, learning_rate, lr_scheduler_type, lora_rank, lora_alpha, lora_dropout, cutoff_len, dataset, eval_dataset, format, eval_steps, raw_text_file, overlap_len, newline_favor_len, higher_rank_limit, warmup_steps, optimizer, hard_cut_string, train_only_after, stop_at_loss, add_eos_token, min_chars, report_to]
+    all_params = [lora_name, always_override, all_linear, q_proj_en, v_proj_en, k_proj_en, o_proj_en, gate_proj_en, down_proj_en, up_proj_en, save_steps, micro_batch_size, batch_size, epochs, learning_rate, lr_scheduler_type, lora_rank, lora_alpha, lora_dropout, cutoff_len, dataset, eval_dataset, format, eval_steps, raw_text_file, higher_rank_limit, warmup_steps, optimizer, hard_cut_string, train_only_after, stop_at_loss, add_eos_token, min_chars, report_to]
 
     copy_from.change(do_copy_params, [copy_from] + all_params, all_params)
     start_button.click(do_train, all_params, output)
@@ -230,9 +225,11 @@ def clean_path(base_path: str, path: str):
 
 
 def backup_adapter(input_folder):
-    # Get the creation date of the file adapter_model.bin
+    # Get the creation date of the adapter file (safetensors or bin)
     try:
-        adapter_file = Path(f"{input_folder}/adapter_model.bin")
+        adapter_file = Path(f"{input_folder}/adapter_model.safetensors")
+        if not adapter_file.is_file():
+            adapter_file = Path(f"{input_folder}/adapter_model.bin")
         if adapter_file.is_file():
 
             logger.info("Backing up existing LoRA adapter")
@@ -244,7 +241,7 @@ def backup_adapter(input_folder):
             subfolder_path.mkdir(parents=True, exist_ok=True)
 
             # Check if the file already exists in the subfolder
-            backup_adapter_file = Path(f"{input_folder}/{creation_date_str}/adapter_model.bin")
+            backup_adapter_file = subfolder_path / adapter_file.name
             if backup_adapter_file.is_file():
                 print(" - Backup already exists. Skipping backup process.")
                 return
@@ -274,7 +271,7 @@ def calc_trainable_parameters(model):
     return trainable_params, all_param
 
 
-def do_train(lora_name: str, always_override: bool, q_proj_en: bool, v_proj_en: bool, k_proj_en: bool, o_proj_en: bool, gate_proj_en: bool, down_proj_en: bool, up_proj_en: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, raw_text_file: str, overlap_len: int, newline_favor_len: int, higher_rank_limit: bool, warmup_steps: int, optimizer: str, hard_cut_string: str, train_only_after: str, stop_at_loss: float, add_eos_token: bool, min_chars: int, report_to: str):
+def do_train(lora_name: str, always_override: bool, all_linear: bool, q_proj_en: bool, v_proj_en: bool, k_proj_en: bool, o_proj_en: bool, gate_proj_en: bool, down_proj_en: bool, up_proj_en: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, raw_text_file: str, higher_rank_limit: bool, warmup_steps: int, optimizer: str, hard_cut_string: str, train_only_after: str, stop_at_loss: float, add_eos_token: bool, min_chars: int, report_to: str):
 
     import torch
     import transformers
@@ -285,15 +282,6 @@ def do_train(lora_name: str, always_override: bool, q_proj_en: bool, v_proj_en:
         prepare_model_for_kbit_training,
         set_peft_model_state_dict
     )
-    from peft.utils.other import \
-        TRANSFORMERS_MODELS_TO_LORA_TARGET_MODULES_MAPPING as \
-        model_to_lora_modules
-    from transformers import is_torch_xpu_available
-    from transformers.models.auto.modeling_auto import (
-        MODEL_FOR_CAUSAL_LM_MAPPING_NAMES
-    )
-
-    MODEL_CLASSES = {v[1]: v[0] for v in MODEL_FOR_CAUSAL_LM_MAPPING_NAMES.items()}
 
     global WANT_INTERRUPT
     WANT_INTERRUPT = False
@@ -309,20 +297,13 @@ def do_train(lora_name: str, always_override: bool, q_proj_en: bool, v_proj_en:
     actual_lr = float(learning_rate)
     model_type = type(shared.model).__name__
 
-    if model_type in MODEL_CLASSES:
-        model_id = MODEL_CLASSES[model_type]
-    else:
-        model_id = "llama"
-        if model_type == "PeftModelForCausalLM":
-            if len(shared.lora_names) > 0:
-                yield "You are trying to train a LoRA while you already have another LoRA loaded. This will work, but may have unexpected effects. *(Will continue anyway in 5 seconds, press `Interrupt` to stop.)*"
-                logger.warning("Training LoRA over top of another LoRA. May have unexpected effects.")
-            else:
-                yield "Model ID not matched due to LoRA loading. Consider reloading base model. *(Will continue anyway in 5 seconds, press `Interrupt` to stop.)*"
-                logger.warning("Model ID not matched due to LoRA loading. Consider reloading base model.")
+    if model_type == "PeftModelForCausalLM":
+        if len(shared.lora_names) > 0:
+            yield "You are trying to train a LoRA while you already have another LoRA loaded. This will work, but may have unexpected effects. *(Will continue anyway in 5 seconds, press `Interrupt` to stop.)*"
+            logger.warning("Training LoRA over top of another LoRA. May have unexpected effects.")
         else:
-            yield "LoRA training has only currently been validated for LLaMA, OPT, GPT-J, and GPT-NeoX models. Unexpected errors may follow. *(Will continue anyway in 5 seconds, press `Interrupt` to stop.)*"
-            logger.warning(f"LoRA training has only currently been validated for LLaMA, OPT, GPT-J, and GPT-NeoX models. (Found model type: {model_type})")
+            yield "Model ID not matched due to LoRA loading. Consider reloading base model. *(Will continue anyway in 5 seconds, press `Interrupt` to stop.)*"
+            logger.warning("Model ID not matched due to LoRA loading. Consider reloading base model.")
 
         time.sleep(5)
 
@@ -331,24 +312,18 @@ def do_train(lora_name: str, always_override: bool, q_proj_en: bool, v_proj_en:
         return
 
     gradient_accumulation_steps = batch_size // micro_batch_size
-    shared.tokenizer.pad_token_id = 0
+    if shared.tokenizer.pad_token_id is None:
+        shared.tokenizer.pad_token_id = shared.tokenizer.eos_token_id
     shared.tokenizer.padding_side = "left"
 
-    # Populate target_modules list with chosen X_proj modules. Llama-based models only atm, non-llama will revert to default behavior.
-    def list_target_modules(model_id):
-        if model_id != "llama" and model_id != "mistral":
-            return model_to_lora_modules[model_id]
-
-        available_modules = {
-            "gate": gate_proj_en,
-            "down": down_proj_en,
-            "up": up_proj_en,
-            "q": q_proj_en,
-            "v": v_proj_en,
-            "k": k_proj_en,
-            "o": o_proj_en,
-        }
-        target_mods = [f"{name}_proj" for name, enabled in available_modules.items() if enabled]
+    def list_target_modules():
+        if all_linear:
+            return "all-linear"
+
+        target_mods = [f"{name}_proj" for name, enabled in {
+            "q": q_proj_en, "k": k_proj_en, "v": v_proj_en, "o": o_proj_en,
+            "gate": gate_proj_en, "down": down_proj_en, "up": up_proj_en,
+        }.items() if enabled]
         return target_mods
 
     def encode(text, add_bos_token):
@@ -369,8 +344,9 @@ def tokenize(prompt, append_eos_token=False):
             if append_eos_token and input_ids[-1] != shared.tokenizer.eos_token_id and len(input_ids) < cutoff_len:
                 input_ids.append(shared.tokenizer.eos_token_id)
 
-            input_ids = [shared.tokenizer.pad_token_id] * (cutoff_len - len(input_ids)) + input_ids
-            labels = [1] * len(input_ids)
+            pad_len = cutoff_len - len(input_ids)
+            labels = [-100] * pad_len + list(input_ids)
+            input_ids = [shared.tokenizer.pad_token_id] * pad_len + input_ids
 
         else:
             ind = prompt.index(train_only_after) + len(train_only_after)
@@ -387,13 +363,12 @@ def tokenize(prompt, append_eos_token=False):
                 before_tokens = [shared.tokenizer.pad_token_id] * (cutoff_len - full_length) + before_tokens
 
             input_ids = before_tokens + after_tokens
-            labels = [-100] * len(before_tokens) + [1] * len(after_tokens)
+            labels = [-100] * len(before_tokens) + list(after_tokens)
 
-        input_ids = torch.tensor(input_ids)
         return {
             "input_ids": input_ids,
             "labels": labels,
-            "attention_mask": input_ids.ne(shared.tokenizer.pad_token_id),
+            "attention_mask": [0 if t == shared.tokenizer.pad_token_id else 1 for t in input_ids],
         }
 
     train_template.clear()
@@ -420,7 +395,7 @@ def tokenize(prompt, append_eos_token=False):
 
         cut_string = hard_cut_string.replace('\\n', '\n')
         eos_added = 0
-        out_tokens = []
+        all_tokens = []
         for text_part in raw_text.split(cut_string):
             if len(text_part.strip()) <= min_chars:
                 continue
@@ -430,24 +405,28 @@ def tokenize(prompt, append_eos_token=False):
                 tokens.append(shared.tokenizer.eos_token_id)
                 eos_added += 1
 
-            step = cutoff_len - overlap_len
-            if step <= 0:
-                yield f"Error: overlap_len ({overlap_len}) cannot be greater than or equal to cutoff_len ({cutoff_len})"
-                return
-
-            out_tokens.extend(split_chunks(tokens, cutoff_len, step))
+            all_tokens.extend(tokens)
 
         if eos_added > 0:
             print(f"EOS added to {eos_added} text blocks")
 
         del raw_text  # Note: could be a gig for a large dataset, so delete redundant data as we go to be safe on RAM
-        text_chunks = [shared.tokenizer.decode(x) for x in out_tokens]
-        del out_tokens
-        if newline_favor_len > 0:
-            text_chunks = [cut_chunk_for_newline(x, newline_favor_len) for x in text_chunks]
 
-        train_data = Dataset.from_list([tokenize(x) for x in text_chunks])
-        del text_chunks
+        # Concatenate-and-split: non-overlapping chunks of exactly cutoff_len
+        num_chunks = len(all_tokens) // cutoff_len
+        if num_chunks == 0:
+            yield "Error: text is too short to fill even one chunk of the given cutoff length."
+            return
+
+        train_data = Dataset.from_list([
+            {
+                "input_ids": all_tokens[i * cutoff_len:(i + 1) * cutoff_len],
+                "labels": all_tokens[i * cutoff_len:(i + 1) * cutoff_len],
+                "attention_mask": [1] * cutoff_len,
+            }
+            for i in range(num_chunks)
+        ])
+        del all_tokens
         eval_data = None
     else:
         if dataset in ['None', '']:
@@ -519,10 +498,15 @@ def generate_and_tokenize_prompt(data_point):
     shared.model_dirty_from_training = True
 
     logger.info("Preparing for training")
+    target_modules = list_target_modules()
+    if not target_modules:
+        yield "No target modules selected. Enable at least one module or check 'Target all linear layers'."
+        return
+
     config = LoraConfig(
         r=lora_rank,
         lora_alpha=lora_alpha,
-        target_modules=list_target_modules(model_id),
+        target_modules=target_modules,
         lora_dropout=lora_dropout,
         bias="none",
         task_type="CAUSAL_LM"
@@ -538,10 +522,18 @@ def generate_and_tokenize_prompt(data_point):
     try:
         logger.info("Creating LoRA model")
         lora_model = get_peft_model(shared.model, config)
-        if not always_override and Path(f"{lora_file_path}/adapter_model.bin").is_file():
-            logger.info("Loading existing LoRA data")
-            state_dict_peft = torch.load(f"{lora_file_path}/adapter_model.bin", weights_only=True)
-            set_peft_model_state_dict(lora_model, state_dict_peft)
+        if not always_override:
+            safetensors_path = Path(f"{lora_file_path}/adapter_model.safetensors")
+            bin_path = Path(f"{lora_file_path}/adapter_model.bin")
+            if safetensors_path.is_file():
+                logger.info("Loading existing LoRA data (safetensors)")
+                from safetensors.torch import load_file
+                state_dict_peft = load_file(str(safetensors_path))
+                set_peft_model_state_dict(lora_model, state_dict_peft)
+            elif bin_path.is_file():
+                logger.info("Loading existing LoRA data (bin)")
+                state_dict_peft = torch.load(str(bin_path), weights_only=True)
+                set_peft_model_state_dict(lora_model, state_dict_peft)
     except Exception:
         yield traceback.format_exc().replace('\n', '\n\n')
         return
@@ -596,12 +588,27 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
         if param.requires_grad:
             param.data = param.data.float()
 
+    lora_model.config.use_cache = False
+
+    if sys.platform != "win32":
+        lora_model = torch.compile(lora_model)
+
+    def collate_fn(batch):
+        input_ids = torch.stack([torch.as_tensor(item['input_ids']) for item in batch])
+        labels = torch.stack([torch.as_tensor(item['labels']) for item in batch])
+        attention_mask = torch.stack([torch.as_tensor(item['attention_mask']) for item in batch])
+        return {
+            'input_ids': input_ids,
+            'labels': labels,
+            'attention_mask': attention_mask,
+        }
+
     trainer = transformers.Trainer(
         model=lora_model,
         train_dataset=train_data,
         eval_dataset=eval_data,
         args=transformers.TrainingArguments(
-            report_to=report_to if report_to != "None" else None,
+            report_to=report_to if report_to != "None" else "none",
             per_device_train_batch_size=micro_batch_size,
             gradient_accumulation_steps=gradient_accumulation_steps,
             warmup_steps=math.ceil(warmup_steps / gradient_accumulation_steps),
@@ -619,18 +626,13 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
             load_best_model_at_end=eval_data is not None,
             # TODO: Enable multi-device support
             ddp_find_unused_parameters=None,
-            no_cuda=shared.args.cpu,
-            # use_ipex=True if is_torch_xpu_available() and not shared.args.cpu else False
+            use_cpu=shared.args.cpu,
+            remove_unused_columns=False,
         ),
-        data_collator=transformers.DataCollatorForLanguageModeling(shared.tokenizer, mlm=False),
+        data_collator=collate_fn,
         callbacks=list([Callbacks()])
     )
 
-    lora_model.config.use_cache = False
-
-    if torch.__version__ >= "2" and sys.platform != "win32":
-        lora_model = torch.compile(lora_model)
-
     # == Save parameters for reuse ==
     with open(f"{lora_file_path}/training_parameters.json", 'w', encoding='utf-8') as file:
         vars = locals()
@@ -646,9 +648,12 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
 
     lora_trainable_param, lora_all_param = calc_trainable_parameters(lora_model)
 
-    projections_string = ", ".join([projection.replace("_proj", "") for projection in list_target_modules(model_id)])
+    if target_modules == "all-linear":
+        projections_string = "all-linear"
+    else:
+        projections_string = ", ".join([projection.replace("_proj", "") for projection in target_modules])
 
-    print(f"Training '{model_id}' model using ({projections_string}) projections")
+    print(f"Training '{model_type}' model using ({projections_string}) projections")
 
     if lora_all_param > 0:
         print(f"Trainable params: {lora_trainable_param:,d} ({100 * lora_trainable_param / lora_all_param:.4f} %), All params: {lora_all_param:,d} (Model: {model_all_params:,d})")
@@ -689,6 +694,7 @@ def threaded_run():
         trainer.train()
         # Note: save in the thread in case the gradio thread breaks (eg browser closed)
         lora_model.save_pretrained(lora_file_path)
+        tracked.did_save = True
         logger.info("LoRA training run is completed and saved.")
         # Save log
         with open(f"{lora_file_path}/training_log.json", 'w', encoding='utf-8') as file:
@@ -734,29 +740,6 @@ def threaded_run():
         yield f"Done! LoRA saved to `{lora_file_path}`.\n\nBefore testing your new LoRA, make sure to first reload the model, as it is currently dirty from training."
 
 
-def split_chunks(arr, size, step):
-    for i in range(0, len(arr), step):
-        yield arr[i:i + size]
-
-
-def cut_chunk_for_newline(chunk: str, max_length: int):
-    if '\n' not in chunk:
-        return chunk
-
-    first_newline = chunk.index('\n')
-    if first_newline < max_length:
-        chunk = chunk[first_newline + 1:]
-
-    if '\n' not in chunk:
-        return chunk
-
-    last_newline = chunk.rindex('\n')
-    if len(chunk) - last_newline < max_length:
-        chunk = chunk[:last_newline]
-
-    return chunk
-
-
 def format_time(seconds: float):
     if seconds < 120:
         return f"`{seconds:.0f}` seconds"

From b16a1a874ace5069eb6f8d21696150b6707366a7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 06:17:52 -0800
Subject: [PATCH 1214/1701] Update TensorRT-LLM Dockerfile for v1.1.0

---
 docker/TensorRT-LLM/Dockerfile | 21 +++++++++------------
 1 file changed, 9 insertions(+), 12 deletions(-)

diff --git a/docker/TensorRT-LLM/Dockerfile b/docker/TensorRT-LLM/Dockerfile
index 3c989d79f7..f9d4dc1cc8 100644
--- a/docker/TensorRT-LLM/Dockerfile
+++ b/docker/TensorRT-LLM/Dockerfile
@@ -1,27 +1,24 @@
-FROM pytorch/pytorch:2.9.1-cuda12.8-cudnn9-runtime
+FROM nvidia/cuda:13.0.1-cudnn-runtime-ubuntu24.04
 
-# Install Git
-RUN apt update && apt install -y git
-
-# System-wide TensorRT-LLM requirements
-RUN apt install -y openmpi-bin libopenmpi-dev
+# Install Python 3.12, Git, and OpenMPI
+RUN apt update && apt install -y python3.12 python3-pip git build-essential openmpi-bin libopenmpi-dev
 
 # Set the working directory
 WORKDIR /app
 
+# This is needed to avoid an error about "Failed to build mpi4py" in the next command
+ENV LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu:$LD_LIBRARY_PATH
+
 # Install text-generation-webui
 RUN git clone https://github.com/oobabooga/text-generation-webui
 WORKDIR /app/text-generation-webui
-RUN pip install -r requirements/full/requirements.txt
-
-# This is needed to avoid an error about "Failed to build mpi4py" in the next command
-ENV LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu:$LD_LIBRARY_PATH
+RUN pip install --break-system-packages -r requirements/full/requirements.txt
 
 # Install TensorRT-LLM
-RUN pip3 install tensorrt_llm==1.1.0 -U --extra-index-url https://pypi.nvidia.com
+RUN pip3 install --break-system-packages tensorrt_llm==1.1.0 --extra-index-url https://pypi.nvidia.com
 
 # Expose the necessary port for the Python server
 EXPOSE 7860 5000
 
 # Run the Python server.py script with the specified command
-CMD ["python", "server.py", "--api", "--listen"]
+CMD ["python3", "server.py", "--api", "--listen"]

From d278bb46a28d860524401679ae0f86ccbb3142fb Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 11:46:45 -0300
Subject: [PATCH 1215/1701] Add apply_chat_template() support for LoRA training

- Support multi-turn conversations (OpenAI messages + ShareGPT formats)
- Automatic assistant-only label masking via incremental tokenization
- Use tokenizer.apply_chat_template() for proper special token handling
- Add "Chat Template" option to the Data Format dropdown
- Also accept instruction/output datasets (auto-converted to messages)
- Validate chat template availability and dataset format upfront
- Fix after_tokens[-1] IndexError when train_only_after is at end of prompt
- Update docs
---
 docs/05 - Training Tab.md |  48 ++++++++++++
 modules/training.py       | 160 +++++++++++++++++++++++++++++++-------
 2 files changed, 179 insertions(+), 29 deletions(-)

diff --git a/docs/05 - Training Tab.md b/docs/05 - Training Tab.md
index 983659606c..89494e5bb2 100644
--- a/docs/05 - Training Tab.md	
+++ b/docs/05 - Training Tab.md	
@@ -79,6 +79,54 @@ If you have different sets of key inputs, you can make your own format file to m
 When using raw text files as your dataset, the text is split into sections by the `Hard Cut String` (default `\n\n\n`), tokenized, concatenated into one long token sequence, and then split into non-overlapping chunks of exactly `Cutoff Length` tokens (any remainder shorter than the cutoff is dropped). This is the standard concatenate-and-split approach used by HuggingFace `run_clm.py`.
 - `Hard Cut String` sets a string that indicates a boundary between unrelated sections of text. This defaults to `\n\n\n`, meaning 3 newlines. When `Add EOS token` is enabled, an EOS token is appended after each section before concatenation. This allows you to insert unrelated sections of text in the same text file, ensuring the model learns proper boundaries between them.
 
+## Chat Template Format
+
+Select **Chat Template** as the Data Format to use the model's built-in chat template (via `apply_chat_template()`) instead of a format file. This works with instruct/chat models that ship with a chat template in their tokenizer (Llama 3, Qwen, Mistral, etc.).
+
+**Advantages over format files:**
+- Special tokens are handled correctly by the tokenizer itself
+- Multi-turn conversations are supported natively
+- Labels are automatically masked so only assistant responses are trained on (no need for `Train Only After`)
+
+**Dataset formats:** Your JSON dataset can use any of these structures:
+
+OpenAI messages format (multi-turn):
+```json
+[
+  {
+    "messages": [
+      {"role": "system", "content": "You are a helpful assistant."},
+      {"role": "user", "content": "What is Python?"},
+      {"role": "assistant", "content": "A programming language."},
+      {"role": "user", "content": "What's it used for?"},
+      {"role": "assistant", "content": "Web dev, data science, scripting, and more."}
+    ]
+  }
+]
+```
+
+The conversation gets tokenized with the model's own chat template (correct special tokens), and the labels are automatically masked so the model only trains on the assistant responses — the system prompt and user turns get `-100` labels and contribute no gradient.
+
+ShareGPT format (`conversations` key with `from`/`value` fields):
+```json
+[
+  {
+    "conversations": [
+      {"from": "system", "value": "You are a helpful assistant."},
+      {"from": "human", "value": "What is Python?"},
+      {"from": "gpt", "value": "A programming language."},
+      {"from": "human", "value": "What's it used for?"},
+      {"from": "gpt", "value": "Web dev, data science, scripting, and more."}
+    ]
+  }
+]
+```
+
+Simple instruction/output format (auto-converted to a single-turn conversation):
+```json
+[{"instruction": "What is 2+2?", "output": "4"}]
+```
+
 ## Target Modules
 
 By default, **Target all linear layers** is enabled. This uses peft's `all-linear` mode, which applies LoRA to every `nn.Linear` layer in the model except the output head (`lm_head`). It works for any model architecture.
diff --git a/modules/training.py b/modules/training.py
index 691f99219e..a0ccb21810 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -107,8 +107,8 @@ def create_ui():
                 with gr.Column():
                     with gr.Tab(label='Formatted Dataset'):
                         with gr.Row():
-                            format = gr.Dropdown(choices=utils.get_datasets('user_data/training/formats', 'json'), value='None', label='Data Format', info='The format file used to decide how to format the dataset input.', elem_classes=['slim-dropdown'], interactive=not mu)
-                            ui.create_refresh_button(format, lambda: None, lambda: {'choices': utils.get_datasets('user_data/training/formats', 'json')}, 'refresh-button', interactive=not mu)
+                            format = gr.Dropdown(choices=['None', 'Chat Template'] + [x for x in utils.get_datasets('user_data/training/formats', 'json') if x != 'None'], value='None', label='Data Format', info='The format file used to decide how to format the dataset input. "Chat Template" uses the model\'s built-in chat template via apply_chat_template().', elem_classes=['slim-dropdown'], interactive=not mu)
+                            ui.create_refresh_button(format, lambda: None, lambda: {'choices': ['None', 'Chat Template'] + [x for x in utils.get_datasets('user_data/training/formats', 'json') if x != 'None']}, 'refresh-button', interactive=not mu)
 
                         with gr.Row():
                             dataset = gr.Dropdown(choices=utils.get_datasets('user_data/training/datasets', 'json'), value='None', label='Dataset', info='The dataset file to use for training.', elem_classes=['slim-dropdown'], interactive=not mu)
@@ -353,7 +353,7 @@ def tokenize(prompt, append_eos_token=False):
             before_tokens = encode(prompt[:ind], True)
             after_tokens = encode(prompt[ind:], False)
 
-            if append_eos_token and after_tokens[-1] != shared.tokenizer.eos_token_id:
+            if append_eos_token and len(after_tokens) > 0 and after_tokens[-1] != shared.tokenizer.eos_token_id:
                 after_tokens.append(shared.tokenizer.eos_token_id)
 
             full_length = len(after_tokens) + len(before_tokens)
@@ -371,6 +371,73 @@ def tokenize(prompt, append_eos_token=False):
             "attention_mask": [0 if t == shared.tokenizer.pad_token_id else 1 for t in input_ids],
         }
 
+    def normalize_messages(data_point):
+        """Convert a dataset row to OpenAI messages format for apply_chat_template()."""
+        if "messages" in data_point:
+            return data_point["messages"]
+
+        if "conversations" in data_point:
+            role_map = {"human": "user", "gpt": "assistant"}
+            return [
+                {"role": role_map.get(turn.get("from", ""), turn.get("from", "")), "content": turn["value"]}
+                for turn in data_point["conversations"]
+            ]
+
+        if "instruction" in data_point and "output" in data_point:
+            messages = []
+            if data_point.get("system", "").strip():
+                messages.append({"role": "system", "content": data_point["system"]})
+            messages.append({"role": "user", "content": data_point["instruction"]})
+            messages.append({"role": "assistant", "content": data_point["output"]})
+            return messages
+
+        raise RuntimeError(
+            f'Dataset row must contain "messages", "conversations", or "instruction"/"output" keys. '
+            f'Found: {list(data_point.keys())}'
+        )
+
+    def tokenize_conversation(data_point):
+        """Tokenize using apply_chat_template() with assistant-only label masking."""
+        messages = normalize_messages(data_point)
+        full_ids = shared.tokenizer.apply_chat_template(messages, tokenize=True)
+
+        # Build labels: -100 for everything, then unmask assistant turns.
+        # This assumes apply_chat_template(messages[:i]) is a token-for-token
+        # prefix of apply_chat_template(messages[:i+1]), which holds for all
+        # standard chat templates (Llama, ChatML, Mistral, etc.).
+        labels = [-100] * len(full_ids)
+        for i, msg in enumerate(messages):
+            if msg["role"] == "assistant":
+                # Tokens up to where this assistant turn starts
+                header_ids = shared.tokenizer.apply_chat_template(
+                    messages[:i], tokenize=True, add_generation_prompt=True
+                )
+                # Tokens through end of this assistant turn
+                through_ids = shared.tokenizer.apply_chat_template(
+                    messages[:i + 1], tokenize=True
+                )
+                # Unmask assistant tokens
+                start = len(header_ids)
+                end = min(len(through_ids), len(full_ids))
+                labels[start:end] = full_ids[start:end]
+
+        # Truncate from the right: keeps the system prompt and early turns
+        if len(full_ids) > cutoff_len:
+            full_ids = full_ids[:cutoff_len]
+            labels = labels[:cutoff_len]
+
+        # Left-pad to cutoff_len
+        pad_len = cutoff_len - len(full_ids)
+        attention_mask = [0] * pad_len + [1] * len(full_ids)
+        labels = [-100] * pad_len + labels
+        input_ids = [shared.tokenizer.pad_token_id] * pad_len + full_ids
+
+        return {
+            "input_ids": input_ids,
+            "labels": labels,
+            "attention_mask": attention_mask,
+        }
+
     train_template.clear()
 
     # == Prep the dataset, format, etc ==
@@ -437,38 +504,73 @@ def tokenize(prompt, append_eos_token=False):
             yield "Missing format choice input, cannot continue."
             return
 
-        train_template["template_type"] = "dataset"
-
-        with open(clean_path('user_data/training/formats', f'{format}.json'), 'r', encoding='utf-8-sig') as formatFile:
-            format_data: dict[str, str] = json.load(formatFile)
+        if format == 'Chat Template':
+            # Use the model's built-in chat template via apply_chat_template()
+            if not getattr(shared.tokenizer, 'chat_template', None):
+                yield "Error: this model's tokenizer does not have a chat template. Use a format file instead, or load an instruct/chat model."
+                return
 
-        # == store training prompt ==
-        for _, value in format_data.items():
-            prompt_key = f"template_{len(train_template)}"
-            train_template[prompt_key] = value
+            train_template["template_type"] = "chat_template"
 
-        def generate_prompt(data_point: dict[str, str]):
-            for options, data in format_data.items():
-                if set(options.split(',')) == set(x[0] for x in data_point.items() if (type(x[1]) is str and len(x[1].strip()) > 0)):
-                    for key, val in data_point.items():
-                        if type(val) is str:
-                            data = data.replace(f'%{key}%', val)
-                    return data
-            raise RuntimeError(f'Data-point "{data_point}" has no keyset match within format "{list(format_data.keys())}"')
+            logger.info("Loading JSON dataset with Chat Template format")
+            data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
 
-        def generate_and_tokenize_prompt(data_point):
-            prompt = generate_prompt(data_point)
-            return tokenize(prompt, add_eos_token)
+            # Validate the first row
+            try:
+                normalize_messages(data['train'][0])
+            except (RuntimeError, KeyError, IndexError) as e:
+                yield f"Error: {e}"
+                return
 
-        logger.info("Loading JSON datasets")
-        data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
-        train_data = data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
+            train_data = data['train'].map(
+                tokenize_conversation,
+                remove_columns=data['train'].column_names,
+                new_fingerprint='%030x' % random.randrange(16**30)
+            )
 
-        if eval_dataset == 'None':
-            eval_data = None
+            if eval_dataset == 'None':
+                eval_data = None
+            else:
+                eval_data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{eval_dataset}.json'))
+                eval_data = eval_data['train'].map(
+                    tokenize_conversation,
+                    remove_columns=eval_data['train'].column_names,
+                    new_fingerprint='%030x' % random.randrange(16**30)
+                )
         else:
-            eval_data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{eval_dataset}.json'))
-            eval_data = eval_data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
+            # Use format file for prompt generation
+            train_template["template_type"] = "dataset"
+
+            with open(clean_path('user_data/training/formats', f'{format}.json'), 'r', encoding='utf-8-sig') as formatFile:
+                format_data: dict[str, str] = json.load(formatFile)
+
+            # == store training prompt ==
+            for _, value in format_data.items():
+                prompt_key = f"template_{len(train_template)}"
+                train_template[prompt_key] = value
+
+            def generate_prompt(data_point: dict[str, str]):
+                for options, data in format_data.items():
+                    if set(options.split(',')) == set(x[0] for x in data_point.items() if (type(x[1]) is str and len(x[1].strip()) > 0)):
+                        for key, val in data_point.items():
+                            if type(val) is str:
+                                data = data.replace(f'%{key}%', val)
+                        return data
+                raise RuntimeError(f'Data-point "{data_point}" has no keyset match within format "{list(format_data.keys())}"')
+
+            def generate_and_tokenize_prompt(data_point):
+                prompt = generate_prompt(data_point)
+                return tokenize(prompt, add_eos_token)
+
+            logger.info("Loading JSON datasets")
+            data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
+            train_data = data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
+
+            if eval_dataset == 'None':
+                eval_data = None
+            else:
+                eval_data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{eval_dataset}.json'))
+                eval_data = eval_data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
 
     # == We MUST reload model if it went through any previous training, even failed one ==
     if shared.model_dirty_from_training:

From da2d4f1a6a566903bc76e4000bbda908ad8f4d42 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 12:32:17 -0300
Subject: [PATCH 1216/1701] Training: replace raw text file with JSONL text
 dataset, re-add stride overlap

- Replace "Raw text file" tab with "Text Dataset" tab using JSONL format with "text" key per row
- Re-add stride overlap for chunking (configurable Stride Length slider, 0-2048 tokens)
- Pad remainder chunks instead of dropping them
- Remove hard_cut_string, min_chars, raw_text_file parameters
- Remove .txt file and directory loading support
---
 docs/05 - Training Tab.md | 23 ++++++---
 modules/training.py       | 98 +++++++++++++++++++--------------------
 2 files changed, 65 insertions(+), 56 deletions(-)

diff --git a/docs/05 - Training Tab.md b/docs/05 - Training Tab.md
index 89494e5bb2..ae66ff030e 100644
--- a/docs/05 - Training Tab.md	
+++ b/docs/05 - Training Tab.md	
@@ -8,9 +8,8 @@ The WebUI seeks to make training your own LoRAs as easy as possible. It comes do
 
 ### **Step 2**: Gather a dataset.
 - If you use a dataset similar to the [Alpaca](https://github.com/gururise/AlpacaDataCleaned/blob/main/alpaca_data_cleaned.json) format, that is natively supported by the `Formatted Dataset` input in the WebUI, with premade formatter options.
-- If you use a dataset that isn't matched to Alpaca's format, but uses the same basic JSON structure, you can make your own format file by copying `training/formats/alpaca-format.json` to a new file and [editing its content](#format-files).
-- If you can get the dataset into a simple text file, that works too! You can train using the `Raw text file` input option.
-    - This means you can for example just copy/paste a chatlog/documentation page/whatever you want, shove it in a plain text file, and train on it.
+- If you use a dataset that isn't matched to Alpaca's format, but uses the same basic JSON structure, you can make your own format file by copying `user_data/training/formats/alpaca-format.json` to a new file and [editing its content](#format-files).
+- For pretraining-style training on raw text, use the `Text Dataset` tab with a JSON file where each row has a `"text"` key.
 - If you use a structured dataset not in this format, you may have to find an external way to convert it - or open an issue to request native support.
 
 ### **Step 3**: Do the training.
@@ -74,10 +73,22 @@ So for example if a dataset has `"instruction": "answer my question"`, then the
 
 If you have different sets of key inputs, you can make your own format file to match it. This format-file is designed to be as simple as possible to enable easy editing to match your needs.
 
-## Raw Text File Settings
+## Text Dataset
 
-When using raw text files as your dataset, the text is split into sections by the `Hard Cut String` (default `\n\n\n`), tokenized, concatenated into one long token sequence, and then split into non-overlapping chunks of exactly `Cutoff Length` tokens (any remainder shorter than the cutoff is dropped). This is the standard concatenate-and-split approach used by HuggingFace `run_clm.py`.
-- `Hard Cut String` sets a string that indicates a boundary between unrelated sections of text. This defaults to `\n\n\n`, meaning 3 newlines. When `Add EOS token` is enabled, an EOS token is appended after each section before concatenation. This allows you to insert unrelated sections of text in the same text file, ensuring the model learns proper boundaries between them.
+For pretraining-style training on raw text, use the **Text Dataset** tab. Your dataset should be a JSON file with one document per row, each with a `"text"` key:
+
+```json
+[
+  {"text": "First document content..."},
+  {"text": "Second document content..."}
+]
+```
+
+This is the standard format used by most pretraining datasets (The Pile, RedPajama, etc.).
+
+Each document is tokenized (with BOS token), concatenated into one long token sequence, and split into chunks of `Cutoff Length` tokens. The final chunk is padded if shorter than the cutoff length. When `Add EOS token` is enabled, an EOS token is appended after each document before concatenation, helping the model learn document boundaries.
+
+- `Stride Length` controls the overlap between consecutive chunks in tokens. Set to 0 for non-overlapping chunks (the standard concatenate-and-split approach). Values like 256 or 512 create overlapping chunks that help the model learn context across chunk boundaries, at the cost of more training samples.
 
 ## Chat Template Format
 
diff --git a/modules/training.py b/modules/training.py
index a0ccb21810..21f7fb9b28 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -26,7 +26,7 @@
 from modules.models import reload_model
 from modules.utils import natural_keys
 
-PARAMETERS = ["lora_name", "always_override", "all_linear", "q_proj_en", "v_proj_en", "k_proj_en", "o_proj_en", "gate_proj_en", "down_proj_en", "up_proj_en", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "raw_text_file", "higher_rank_limit", "warmup_steps", "optimizer", "hard_cut_string", "train_only_after", "stop_at_loss", "add_eos_token", "min_chars", "report_to"]
+PARAMETERS = ["lora_name", "always_override", "all_linear", "q_proj_en", "v_proj_en", "k_proj_en", "o_proj_en", "gate_proj_en", "down_proj_en", "up_proj_en", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "text_dataset", "higher_rank_limit", "warmup_steps", "optimizer", "stride_length", "train_only_after", "stop_at_loss", "add_eos_token", "report_to"]
 WANT_INTERRUPT = False
 
 train_log = {}
@@ -120,13 +120,12 @@ def create_ui():
 
                         eval_steps = gr.Number(label='Evaluate every n steps', value=100, info='If an evaluation dataset is given, test it every time this many steps pass.')
 
-                    with gr.Tab(label="Raw text file"):
+                    with gr.Tab(label="Text Dataset"):
                         with gr.Row():
-                            raw_text_file = gr.Dropdown(choices=utils.get_datasets('user_data/training/datasets', 'txt'), value='None', label='Text file', info='The raw text file to use for training.', elem_classes=['slim-dropdown'], interactive=not mu)
-                            ui.create_refresh_button(raw_text_file, lambda: None, lambda: {'choices': utils.get_datasets('user_data/training/datasets', 'txt')}, 'refresh-button', interactive=not mu)
+                            text_dataset = gr.Dropdown(choices=utils.get_datasets('user_data/training/datasets', 'json'), value='None', label='Text Dataset', info='A JSONL file with a "text" key per row, for pretraining-style training. Each row is one document.', elem_classes=['slim-dropdown'], interactive=not mu)
+                            ui.create_refresh_button(text_dataset, lambda: None, lambda: {'choices': utils.get_datasets('user_data/training/datasets', 'json')}, 'refresh-button', interactive=not mu)
 
-                        hard_cut_string = gr.Textbox(label='Hard Cut String', value='\\n\\n\\n', info='String that indicates a hard cut between text parts. Helps prevent unwanted overlap.')
-                        min_chars = gr.Number(label='Ignore small blocks', value=0, info='Ignore Hard Cut blocks that have less or equal characters than this number')
+                        stride_length = gr.Slider(label='Stride Length', minimum=0, maximum=2048, value=0, step=32, info='Overlap between chunks in tokens. 0 = no overlap. Values like 256 or 512 help preserve context across chunk boundaries.')
 
                     with gr.Row():
                         start_button = gr.Button("Start LoRA Training", variant='primary', interactive=not mu)
@@ -160,7 +159,7 @@ def create_ui():
                 refresh_table = gr.Button('Refresh the table', elem_classes="small-button", interactive=not mu)
 
     # Training events
-    all_params = [lora_name, always_override, all_linear, q_proj_en, v_proj_en, k_proj_en, o_proj_en, gate_proj_en, down_proj_en, up_proj_en, save_steps, micro_batch_size, batch_size, epochs, learning_rate, lr_scheduler_type, lora_rank, lora_alpha, lora_dropout, cutoff_len, dataset, eval_dataset, format, eval_steps, raw_text_file, higher_rank_limit, warmup_steps, optimizer, hard_cut_string, train_only_after, stop_at_loss, add_eos_token, min_chars, report_to]
+    all_params = [lora_name, always_override, all_linear, q_proj_en, v_proj_en, k_proj_en, o_proj_en, gate_proj_en, down_proj_en, up_proj_en, save_steps, micro_batch_size, batch_size, epochs, learning_rate, lr_scheduler_type, lora_rank, lora_alpha, lora_dropout, cutoff_len, dataset, eval_dataset, format, eval_steps, text_dataset, higher_rank_limit, warmup_steps, optimizer, stride_length, train_only_after, stop_at_loss, add_eos_token, report_to]
 
     copy_from.change(do_copy_params, [copy_from] + all_params, all_params)
     start_button.click(do_train, all_params, output)
@@ -271,7 +270,7 @@ def calc_trainable_parameters(model):
     return trainable_params, all_param
 
 
-def do_train(lora_name: str, always_override: bool, all_linear: bool, q_proj_en: bool, v_proj_en: bool, k_proj_en: bool, o_proj_en: bool, gate_proj_en: bool, down_proj_en: bool, up_proj_en: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, raw_text_file: str, higher_rank_limit: bool, warmup_steps: int, optimizer: str, hard_cut_string: str, train_only_after: str, stop_at_loss: float, add_eos_token: bool, min_chars: int, report_to: str):
+def do_train(lora_name: str, always_override: bool, all_linear: bool, q_proj_en: bool, v_proj_en: bool, k_proj_en: bool, o_proj_en: bool, gate_proj_en: bool, down_proj_en: bool, up_proj_en: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, text_dataset: str, higher_rank_limit: bool, warmup_steps: int, optimizer: str, stride_length: int, train_only_after: str, stop_at_loss: float, add_eos_token: bool, report_to: str):
 
     import torch
     import transformers
@@ -441,58 +440,57 @@ def tokenize_conversation(data_point):
     train_template.clear()
 
     # == Prep the dataset, format, etc ==
-    if raw_text_file not in ['None', '']:
-        train_template["template_type"] = "raw_text"
-        logger.info("Loading raw text file dataset")
-        fullpath = clean_path('user_data/training/datasets', f'{raw_text_file}')
-        fullpath = Path(fullpath)
-        if fullpath.is_dir():
-            logger.info('Training path directory {}'.format(raw_text_file))
-            raw_text = ""
-            file_paths = sorted(fullpath.glob('*.txt'), key=lambda path: natural_keys(path.name))
-            for file_path in file_paths:
-                if file_path.is_file():
-                    with file_path.open('r', encoding='utf-8') as file:
-                        raw_text += file.read().replace('\r', '')
-
-                    logger.info(f"Loaded training file: {file_path.name}")
-        else:
-            with open(clean_path('user_data/training/datasets', f'{raw_text_file}.txt'), 'r', encoding='utf-8') as file:
-                raw_text = file.read().replace('\r', '')
+    if text_dataset not in ['None', '']:
+        train_template["template_type"] = "text_dataset"
+        logger.info("Loading text dataset")
+        data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{text_dataset}.json'))
+
+        # Validate the first row has a "text" key
+        if "text" not in data['train'].column_names:
+            yield "Error: text dataset must have a \"text\" key per row."
+            return
 
-        cut_string = hard_cut_string.replace('\\n', '\n')
-        eos_added = 0
+        # Tokenize each document and concatenate
         all_tokens = []
-        for text_part in raw_text.split(cut_string):
-            if len(text_part.strip()) <= min_chars:
-                continue
-
-            tokens = shared.tokenizer.encode(text_part)
+        for row in data['train']:
+            tokens = shared.tokenizer.encode(row['text'])
             if add_eos_token:
                 tokens.append(shared.tokenizer.eos_token_id)
-                eos_added += 1
-
             all_tokens.extend(tokens)
 
-        if eos_added > 0:
-            print(f"EOS added to {eos_added} text blocks")
+        # Split into chunks with optional overlap (stride)
+        stride = int(stride_length)
+        step = cutoff_len - stride if stride > 0 else cutoff_len
 
-        del raw_text  # Note: could be a gig for a large dataset, so delete redundant data as we go to be safe on RAM
+        if step <= 0:
+            yield "Error: stride length must be smaller than cutoff length."
+            return
 
-        # Concatenate-and-split: non-overlapping chunks of exactly cutoff_len
-        num_chunks = len(all_tokens) // cutoff_len
-        if num_chunks == 0:
-            yield "Error: text is too short to fill even one chunk of the given cutoff length."
+        if len(all_tokens) < cutoff_len:
+            yield "Error: dataset is too short to fill even one chunk of the given cutoff length."
             return
 
-        train_data = Dataset.from_list([
-            {
-                "input_ids": all_tokens[i * cutoff_len:(i + 1) * cutoff_len],
-                "labels": all_tokens[i * cutoff_len:(i + 1) * cutoff_len],
-                "attention_mask": [1] * cutoff_len,
-            }
-            for i in range(num_chunks)
-        ])
+        chunks = []
+        for start in range(0, len(all_tokens), step):
+            chunk = all_tokens[start:start + cutoff_len]
+            if len(chunk) == 0:
+                break
+            if len(chunk) < cutoff_len:
+                # Pad the remainder
+                pad_len = cutoff_len - len(chunk)
+                chunks.append({
+                    "input_ids": chunk + [shared.tokenizer.pad_token_id] * pad_len,
+                    "labels": list(chunk) + [-100] * pad_len,
+                    "attention_mask": [1] * len(chunk) + [0] * pad_len,
+                })
+            else:
+                chunks.append({
+                    "input_ids": chunk,
+                    "labels": list(chunk),
+                    "attention_mask": [1] * cutoff_len,
+                })
+
+        train_data = Dataset.from_list(chunks)
         del all_tokens
         eval_data = None
     else:

From 1c2548fd892ad888676910af990020b9a968eae3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 12:36:10 -0300
Subject: [PATCH 1217/1701] Training: use dynamic padding (pad to batch max
 instead of cutoff_len)

- Remove pre-padding from tokenize() and tokenize_conversation()
- Collate function now right-pads each batch to the longest sequence
- Set tokenizer padding_side to "right" (standard for training)
- Remove dead natural_keys import
- Reduces wasted compute on batches with short sequences
- Aligns with axolotl/unsloth approach
---
 modules/training.py | 37 +++++++++++++++----------------------
 1 file changed, 15 insertions(+), 22 deletions(-)

diff --git a/modules/training.py b/modules/training.py
index 21f7fb9b28..292ee48477 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -24,7 +24,6 @@
 )
 from modules.logging_colors import logger
 from modules.models import reload_model
-from modules.utils import natural_keys
 
 PARAMETERS = ["lora_name", "always_override", "all_linear", "q_proj_en", "v_proj_en", "k_proj_en", "o_proj_en", "gate_proj_en", "down_proj_en", "up_proj_en", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "text_dataset", "higher_rank_limit", "warmup_steps", "optimizer", "stride_length", "train_only_after", "stop_at_loss", "add_eos_token", "report_to"]
 WANT_INTERRUPT = False
@@ -313,7 +312,7 @@ def do_train(lora_name: str, always_override: bool, all_linear: bool, q_proj_en:
     gradient_accumulation_steps = batch_size // micro_batch_size
     if shared.tokenizer.pad_token_id is None:
         shared.tokenizer.pad_token_id = shared.tokenizer.eos_token_id
-    shared.tokenizer.padding_side = "left"
+    shared.tokenizer.padding_side = "right"
 
     def list_target_modules():
         if all_linear:
@@ -343,9 +342,7 @@ def tokenize(prompt, append_eos_token=False):
             if append_eos_token and input_ids[-1] != shared.tokenizer.eos_token_id and len(input_ids) < cutoff_len:
                 input_ids.append(shared.tokenizer.eos_token_id)
 
-            pad_len = cutoff_len - len(input_ids)
-            labels = [-100] * pad_len + list(input_ids)
-            input_ids = [shared.tokenizer.pad_token_id] * pad_len + input_ids
+            labels = list(input_ids)
 
         else:
             ind = prompt.index(train_only_after) + len(train_only_after)
@@ -358,8 +355,6 @@ def tokenize(prompt, append_eos_token=False):
             full_length = len(after_tokens) + len(before_tokens)
             if full_length > cutoff_len:
                 after_tokens = after_tokens[:cutoff_len - len(before_tokens)]
-            else:
-                before_tokens = [shared.tokenizer.pad_token_id] * (cutoff_len - full_length) + before_tokens
 
             input_ids = before_tokens + after_tokens
             labels = [-100] * len(before_tokens) + list(after_tokens)
@@ -367,7 +362,7 @@ def tokenize(prompt, append_eos_token=False):
         return {
             "input_ids": input_ids,
             "labels": labels,
-            "attention_mask": [0 if t == shared.tokenizer.pad_token_id else 1 for t in input_ids],
+            "attention_mask": [1] * len(input_ids),
         }
 
     def normalize_messages(data_point):
@@ -425,16 +420,10 @@ def tokenize_conversation(data_point):
             full_ids = full_ids[:cutoff_len]
             labels = labels[:cutoff_len]
 
-        # Left-pad to cutoff_len
-        pad_len = cutoff_len - len(full_ids)
-        attention_mask = [0] * pad_len + [1] * len(full_ids)
-        labels = [-100] * pad_len + labels
-        input_ids = [shared.tokenizer.pad_token_id] * pad_len + full_ids
-
         return {
-            "input_ids": input_ids,
+            "input_ids": full_ids,
             "labels": labels,
-            "attention_mask": attention_mask,
+            "attention_mask": [1] * len(full_ids),
         }
 
     train_template.clear()
@@ -694,13 +683,17 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
         lora_model = torch.compile(lora_model)
 
     def collate_fn(batch):
-        input_ids = torch.stack([torch.as_tensor(item['input_ids']) for item in batch])
-        labels = torch.stack([torch.as_tensor(item['labels']) for item in batch])
-        attention_mask = torch.stack([torch.as_tensor(item['attention_mask']) for item in batch])
+        max_len = max(len(item['input_ids']) for item in batch)
+        input_ids, labels, attention_mask = [], [], []
+        for item in batch:
+            pad_len = max_len - len(item['input_ids'])
+            input_ids.append(item['input_ids'] + [shared.tokenizer.pad_token_id] * pad_len)
+            labels.append(item['labels'] + [-100] * pad_len)
+            attention_mask.append(item['attention_mask'] + [0] * pad_len)
         return {
-            'input_ids': input_ids,
-            'labels': labels,
-            'attention_mask': attention_mask,
+            'input_ids': torch.tensor(input_ids),
+            'labels': torch.tensor(labels),
+            'attention_mask': torch.tensor(attention_mask),
         }
 
     trainer = transformers.Trainer(

From 1ffe540c9710f184de421fe83ad53bb1f784841d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 12:46:21 -0300
Subject: [PATCH 1218/1701] Full documentation update to match current codebase

---
 README.md                              | 219 +++++++++++++------------
 docs/01 - Chat Tab.md                  |  49 +++---
 docs/02 - Default and Notebook Tabs.md |   6 +-
 docs/03 - Parameters Tab.md            |  71 +++++---
 docs/04 - Model Tab.md                 | 157 +++++++++---------
 docs/06 - Session Tab.md               |  13 +-
 docs/07 - Extensions.md                |  23 +--
 docs/11 - AMD Setup.md                 |  26 ++-
 docs/12 - OpenAI API.md                | 127 ++++++--------
 docs/What Works.md                     |  27 ++-
 10 files changed, 390 insertions(+), 328 deletions(-)

diff --git a/README.md b/README.md
index a72e806050..a2ba87f8f5 100644
--- a/README.md
+++ b/README.md
@@ -239,130 +239,150 @@ List of command-line flags
 
 ```txt
 usage: server.py [-h] [--multi-user] [--model MODEL] [--lora LORA [LORA ...]] [--model-dir MODEL_DIR] [--lora-dir LORA_DIR] [--model-menu] [--settings SETTINGS]
-                 [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--loader LOADER] [--ctx-size N] [--cache-type N] [--model-draft MODEL_DRAFT]
-                 [--draft-max DRAFT_MAX] [--gpu-layers-draft GPU_LAYERS_DRAFT] [--device-draft DEVICE_DRAFT] [--ctx-size-draft CTX_SIZE_DRAFT] [--gpu-layers N] [--mmproj MMPROJ] [--streaming-llm]
-                 [--tensor-split TENSOR_SPLIT] [--row-split] [--no-mmap] [--mlock] [--no-kv-offload] [--batch-size BATCH_SIZE] [--threads THREADS] [--threads-batch THREADS_BATCH] [--numa]
+                 [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--image-model IMAGE_MODEL] [--image-model-dir IMAGE_MODEL_DIR] [--image-dtype {bfloat16,float16}]
+                 [--image-attn-backend {flash_attention_2,sdpa}] [--image-cpu-offload] [--image-compile] [--image-quant {none,bnb-8bit,bnb-4bit,torchao-int8wo,torchao-fp4,torchao-float8wo}]
+                 [--loader LOADER] [--ctx-size N] [--cache-type N] [--model-draft MODEL_DRAFT] [--draft-max DRAFT_MAX] [--gpu-layers-draft GPU_LAYERS_DRAFT] [--device-draft DEVICE_DRAFT]
+                 [--ctx-size-draft CTX_SIZE_DRAFT] [--spec-type {none,ngram-mod,ngram-simple,ngram-map-k,ngram-map-k4v,ngram-cache}] [--spec-ngram-size-n SPEC_NGRAM_SIZE_N]
+                 [--spec-ngram-size-m SPEC_NGRAM_SIZE_M] [--spec-ngram-min-hits SPEC_NGRAM_MIN_HITS] [--gpu-layers N] [--cpu-moe] [--mmproj MMPROJ] [--streaming-llm] [--tensor-split TENSOR_SPLIT]
+                 [--row-split] [--no-mmap] [--mlock] [--no-kv-offload] [--batch-size BATCH_SIZE] [--ubatch-size UBATCH_SIZE] [--threads THREADS] [--threads-batch THREADS_BATCH] [--numa]
                  [--extra-flags EXTRA_FLAGS] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code]
                  [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE]
                  [--enable-tp] [--tp-backend TP_BACKEND] [--gpu-split GPU_SPLIT] [--autosplit] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa] [--num_experts_per_token N] [--cpp-runner]
-                 [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB]
-                 [--listen] [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH]
-                 [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors] [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT]
-                 [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4] [--nowebui]
+                 [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share]
+                 [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors]
+                 [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4]
+                 [--nowebui]
 
 Text Generation Web UI
 
 options:
-  -h, --help                                show this help message and exit
+  -h, --help                                           show this help message and exit
 
 Basic settings:
-  --multi-user                              Multi-user mode. Chat histories are not saved or automatically loaded. Warning: this is likely not safe for sharing publicly.
-  --model MODEL                             Name of the model to load by default.
-  --lora LORA [LORA ...]                    The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.
-  --model-dir MODEL_DIR                     Path to directory with all the models.
-  --lora-dir LORA_DIR                       Path to directory with all the loras.
-  --model-menu                              Show a model menu in the terminal when the web UI is first launched.
-  --settings SETTINGS                       Load the default interface settings from this yaml file. See user_data/settings-template.yaml for an example. If you create a file called
-                                            user_data/settings.yaml, this file will be loaded by default without the need to use the --settings flag.
-  --extensions EXTENSIONS [EXTENSIONS ...]  The list of extensions to load. If you want to load more than one extension, write the names separated by spaces.
-  --verbose                                 Print the prompts to the terminal.
-  --idle-timeout IDLE_TIMEOUT               Unload model after this many minutes of inactivity. It will be automatically reloaded when you try to use it again.
+  --multi-user                                         Multi-user mode. Chat histories are not saved or automatically loaded. Warning: this is likely not safe for sharing publicly.
+  --model MODEL                                        Name of the model to load by default.
+  --lora LORA [LORA ...]                               The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.
+  --model-dir MODEL_DIR                                Path to directory with all the models.
+  --lora-dir LORA_DIR                                  Path to directory with all the loras.
+  --model-menu                                         Show a model menu in the terminal when the web UI is first launched.
+  --settings SETTINGS                                  Load the default interface settings from this yaml file. See user_data/settings-template.yaml for an example. If you create a file called
+                                                       user_data/settings.yaml, this file will be loaded by default without the need to use the --settings flag.
+  --extensions EXTENSIONS [EXTENSIONS ...]             The list of extensions to load. If you want to load more than one extension, write the names separated by spaces.
+  --verbose                                            Print the prompts to the terminal.
+  --idle-timeout IDLE_TIMEOUT                          Unload model after this many minutes of inactivity. It will be automatically reloaded when you try to use it again.
+
+Image model:
+  --image-model IMAGE_MODEL                            Name of the image model to select on startup (overrides saved setting).
+  --image-model-dir IMAGE_MODEL_DIR                    Path to directory with all the image models.
+  --image-dtype {bfloat16,float16}                     Data type for image model.
+  --image-attn-backend {flash_attention_2,sdpa}        Attention backend for image model.
+  --image-cpu-offload                                  Enable CPU offloading for image model.
+  --image-compile                                      Compile the image model for faster inference.
+  --image-quant {none,bnb-8bit,bnb-4bit,torchao-int8wo,torchao-fp4,torchao-float8wo}
+                                                       Quantization method for image model.
 
 Model loader:
-  --loader LOADER                           Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF, ExLlamav2,
-                                            TensorRT-LLM.
+  --loader LOADER                                      Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF,
+                                                       ExLlamav2, TensorRT-LLM.
 
 Context and cache:
-  --ctx-size N, --n_ctx N, --max_seq_len N  Context size in tokens.
-  --cache-type N, --cache_type N            KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits
-                                            separately, e.g. q4_q8).
+  --ctx-size N, --n_ctx N, --max_seq_len N             Context size in tokens. llama.cpp: 0 = auto if gpu-layers is also -1.
+  --cache-type N, --cache_type N                       KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and
+                                                       v_bits separately, e.g. q4_q8).
 
 Speculative decoding:
-  --model-draft MODEL_DRAFT                 Path to the draft model for speculative decoding.
-  --draft-max DRAFT_MAX                     Number of tokens to draft for speculative decoding.
-  --gpu-layers-draft GPU_LAYERS_DRAFT       Number of layers to offload to the GPU for the draft model.
-  --device-draft DEVICE_DRAFT               Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1
-  --ctx-size-draft CTX_SIZE_DRAFT           Size of the prompt context for the draft model. If 0, uses the same as the main model.
+  --model-draft MODEL_DRAFT                            Path to the draft model for speculative decoding.
+  --draft-max DRAFT_MAX                                Number of tokens to draft for speculative decoding.
+  --gpu-layers-draft GPU_LAYERS_DRAFT                  Number of layers to offload to the GPU for the draft model.
+  --device-draft DEVICE_DRAFT                          Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1
+  --ctx-size-draft CTX_SIZE_DRAFT                      Size of the prompt context for the draft model. If 0, uses the same as the main model.
+  --spec-type {none,ngram-mod,ngram-simple,ngram-map-k,ngram-map-k4v,ngram-cache}
+                                                       Draftless speculative decoding type. Recommended: ngram-mod.
+  --spec-ngram-size-n SPEC_NGRAM_SIZE_N                N-gram lookup size for ngram speculative decoding.
+  --spec-ngram-size-m SPEC_NGRAM_SIZE_M                Draft n-gram size for ngram speculative decoding.
+  --spec-ngram-min-hits SPEC_NGRAM_MIN_HITS            Minimum n-gram hits for ngram-map speculative decoding.
 
 llama.cpp:
-  --gpu-layers N, --n-gpu-layers N          Number of layers to offload to the GPU.
-  --mmproj MMPROJ                           Path to the mmproj file for vision models.
-  --streaming-llm                           Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.
-  --tensor-split TENSOR_SPLIT               Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.
-  --row-split                               Split the model by rows across GPUs. This may improve multi-gpu performance.
-  --no-mmap                                 Prevent mmap from being used.
-  --mlock                                   Force the system to keep the model in RAM.
-  --no-kv-offload                           Do not offload the K, Q, V to the GPU. This saves VRAM but reduces the performance.
-  --batch-size BATCH_SIZE                   Maximum number of prompt tokens to batch together when calling llama_eval.
-  --threads THREADS                         Number of threads to use.
-  --threads-batch THREADS_BATCH             Number of threads to use for batches/prompt processing.
-  --numa                                    Activate NUMA task allocation for llama.cpp.
-  --extra-flags EXTRA_FLAGS                 Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"
+  --gpu-layers N, --n-gpu-layers N                     Number of layers to offload to the GPU. -1 = auto.
+  --cpu-moe                                            Move the experts to the CPU (for MoE models).
+  --mmproj MMPROJ                                      Path to the mmproj file for vision models.
+  --streaming-llm                                      Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.
+  --tensor-split TENSOR_SPLIT                          Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.
+  --row-split                                          Split the model by rows across GPUs. This may improve multi-gpu performance.
+  --no-mmap                                            Prevent mmap from being used.
+  --mlock                                              Force the system to keep the model in RAM.
+  --no-kv-offload                                      Do not offload the K, Q, V to the GPU. This saves VRAM but reduces the performance.
+  --batch-size BATCH_SIZE                              Maximum number of prompt tokens to batch together when calling llama-server. This is the application level batch size.
+  --ubatch-size UBATCH_SIZE                            Maximum number of prompt tokens to batch together when calling llama-server. This is the max physical batch size for computation (device level).
+  --threads THREADS                                    Number of threads to use.
+  --threads-batch THREADS_BATCH                        Number of threads to use for batches/prompt processing.
+  --numa                                               Activate NUMA task allocation for llama.cpp.
+  --extra-flags EXTRA_FLAGS                            Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"
 
 Transformers/Accelerate:
-  --cpu                                     Use the CPU to generate text. Warning: Training on CPU is extremely slow.
-  --cpu-memory CPU_MEMORY                   Maximum CPU memory in GiB. Use this for CPU offloading.
-  --disk                                    If the model is too large for your GPU(s) and CPU combined, send the remaining layers to the disk.
-  --disk-cache-dir DISK_CACHE_DIR           Directory to save the disk cache to. Defaults to "user_data/cache".
-  --load-in-8bit                            Load the model with 8-bit precision (using bitsandbytes).
-  --bf16                                    Load the model with bfloat16 precision. Requires NVIDIA Ampere GPU.
-  --no-cache                                Set use_cache to False while generating text. This reduces VRAM usage slightly, but it comes at a performance cost.
-  --trust-remote-code                       Set trust_remote_code=True while loading the model. Necessary for some models.
-  --force-safetensors                       Set use_safetensors=True while loading the model. This prevents arbitrary code execution.
-  --no_use_fast                             Set use_fast=False while loading the tokenizer (it's True by default). Use this if you have any problems related to use_fast.
-  --attn-implementation IMPLEMENTATION      Attention implementation. Valid options: sdpa, eager, flash_attention_2.
+  --cpu                                                Use the CPU to generate text. Warning: Training on CPU is extremely slow.
+  --cpu-memory CPU_MEMORY                              Maximum CPU memory in GiB. Use this for CPU offloading.
+  --disk                                               If the model is too large for your GPU(s) and CPU combined, send the remaining layers to the disk.
+  --disk-cache-dir DISK_CACHE_DIR                      Directory to save the disk cache to. Defaults to "user_data/cache".
+  --load-in-8bit                                       Load the model with 8-bit precision (using bitsandbytes).
+  --bf16                                               Load the model with bfloat16 precision. Requires NVIDIA Ampere GPU.
+  --no-cache                                           Set use_cache to False while generating text. This reduces VRAM usage slightly, but it comes at a performance cost.
+  --trust-remote-code                                  Set trust_remote_code=True while loading the model. Necessary for some models.
+  --force-safetensors                                  Set use_safetensors=True while loading the model. This prevents arbitrary code execution.
+  --no_use_fast                                        Set use_fast=False while loading the tokenizer (it's True by default). Use this if you have any problems related to use_fast.
+  --attn-implementation IMPLEMENTATION                 Attention implementation. Valid options: sdpa, eager, flash_attention_2.
 
 bitsandbytes 4-bit:
-  --load-in-4bit                            Load the model with 4-bit precision (using bitsandbytes).
-  --use_double_quant                        use_double_quant for 4-bit.
-  --compute_dtype COMPUTE_DTYPE             compute dtype for 4-bit. Valid options: bfloat16, float16, float32.
-  --quant_type QUANT_TYPE                   quant_type for 4-bit. Valid options: nf4, fp4.
+  --load-in-4bit                                       Load the model with 4-bit precision (using bitsandbytes).
+  --use_double_quant                                   use_double_quant for 4-bit.
+  --compute_dtype COMPUTE_DTYPE                        compute dtype for 4-bit. Valid options: bfloat16, float16, float32.
+  --quant_type QUANT_TYPE                              quant_type for 4-bit. Valid options: nf4, fp4.
 
 ExLlamaV3:
-  --enable-tp, --enable_tp                  Enable Tensor Parallelism (TP) to split the model across GPUs.
-  --tp-backend TP_BACKEND                   The backend for tensor parallelism. Valid options: native, nccl. Default: native.
+  --enable-tp, --enable_tp                             Enable Tensor Parallelism (TP) to split the model across GPUs.
+  --tp-backend TP_BACKEND                              The backend for tensor parallelism. Valid options: native, nccl. Default: native.
 
 ExLlamaV2:
-  --gpu-split GPU_SPLIT                     Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.
-  --autosplit                               Autosplit the model tensors across the available GPUs. This causes --gpu-split to be ignored.
-  --cfg-cache                               ExLlamav2_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.
-  --no_flash_attn                           Force flash-attention to not be used.
-  --no_xformers                             Force xformers to not be used.
-  --no_sdpa                                 Force Torch SDPA to not be used.
-  --num_experts_per_token N                 Number of experts to use for generation. Applies to MoE models like Mixtral.
+  --gpu-split GPU_SPLIT                                Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.
+  --autosplit                                          Autosplit the model tensors across the available GPUs. This causes --gpu-split to be ignored.
+  --cfg-cache                                          ExLlamav2_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.
+  --no_flash_attn                                      Force flash-attention to not be used.
+  --no_xformers                                        Force xformers to not be used.
+  --no_sdpa                                            Force Torch SDPA to not be used.
+  --num_experts_per_token N                            Number of experts to use for generation. Applies to MoE models like Mixtral.
 
 TensorRT-LLM:
-  --cpp-runner                              Use the ModelRunnerCpp runner, which is faster than the default ModelRunner but doesn't support streaming yet.
+  --cpp-runner                                         Use the ModelRunnerCpp runner, which is faster than the default ModelRunner.
 
 RoPE:
-  --alpha_value ALPHA_VALUE                 Positional embeddings alpha factor for NTK RoPE scaling. Use either this or compress_pos_emb, not both.
-  --rope_freq_base ROPE_FREQ_BASE           If greater than 0, will be used instead of alpha_value. Those two are related by rope_freq_base = 10000 * alpha_value ^ (64 / 63).
-  --compress_pos_emb COMPRESS_POS_EMB       Positional embeddings compression factor. Should be set to (context length) / (model's original context length). Equal to 1/rope_freq_scale.
+  --alpha_value ALPHA_VALUE                            Positional embeddings alpha factor for NTK RoPE scaling. Use either this or compress_pos_emb, not both.
+  --rope_freq_base ROPE_FREQ_BASE                      If greater than 0, will be used instead of alpha_value. Those two are related by rope_freq_base = 10000 * alpha_value ^ (64 / 63).
+  --compress_pos_emb COMPRESS_POS_EMB                  Positional embeddings compression factor. Should be set to (context length) / (model's original context length). Equal to 1/rope_freq_scale.
 
 Gradio:
-  --listen                                  Make the web UI reachable from your local network.
-  --listen-port LISTEN_PORT                 The listening port that the server will use.
-  --listen-host LISTEN_HOST                 The hostname that the server will use.
-  --share                                   Create a public URL. This is useful for running the web UI on Google Colab or similar.
-  --auto-launch                             Open the web UI in the default browser upon launch.
-  --gradio-auth GRADIO_AUTH                 Set Gradio authentication password in the format "username:password". Multiple credentials can also be supplied with "u1:p1,u2:p2,u3:p3".
-  --gradio-auth-path GRADIO_AUTH_PATH       Set the Gradio authentication file path. The file should contain one or more user:password pairs in the same format as above.
-  --ssl-keyfile SSL_KEYFILE                 The path to the SSL certificate key file.
-  --ssl-certfile SSL_CERTFILE               The path to the SSL certificate cert file.
-  --subpath SUBPATH                         Customize the subpath for gradio, use with reverse proxy
-  --old-colors                              Use the legacy Gradio colors, before the December/2024 update.
-  --portable                                Hide features not available in portable mode like training.
+  --listen                                             Make the web UI reachable from your local network.
+  --listen-port LISTEN_PORT                            The listening port that the server will use.
+  --listen-host LISTEN_HOST                            The hostname that the server will use.
+  --share                                              Create a public URL. This is useful for running the web UI on Google Colab or similar.
+  --auto-launch                                        Open the web UI in the default browser upon launch.
+  --gradio-auth GRADIO_AUTH                            Set Gradio authentication password in the format "username:password". Multiple credentials can also be supplied with "u1:p1,u2:p2,u3:p3".
+  --gradio-auth-path GRADIO_AUTH_PATH                  Set the Gradio authentication file path. The file should contain one or more user:password pairs in the same format as above.
+  --ssl-keyfile SSL_KEYFILE                            The path to the SSL certificate key file.
+  --ssl-certfile SSL_CERTFILE                          The path to the SSL certificate cert file.
+  --subpath SUBPATH                                    Customize the subpath for gradio, use with reverse proxy
+  --old-colors                                         Use the legacy Gradio colors, before the December/2024 update.
+  --portable                                           Hide features not available in portable mode like training.
 
 API:
-  --api                                     Enable the API extension.
-  --public-api                              Create a public URL for the API using Cloudfare.
-  --public-api-id PUBLIC_API_ID             Tunnel ID for named Cloudflare Tunnel. Use together with public-api option.
-  --api-port API_PORT                       The listening port for the API.
-  --api-key API_KEY                         API authentication key.
-  --admin-key ADMIN_KEY                     API authentication key for admin tasks like loading and unloading models. If not set, will be the same as --api-key.
-  --api-enable-ipv6                         Enable IPv6 for the API
-  --api-disable-ipv4                        Disable IPv4 for the API
-  --nowebui                                 Do not launch the Gradio UI. Useful for launching the API in standalone mode.
+  --api                                                Enable the API extension.
+  --public-api                                         Create a public URL for the API using Cloudfare.
+  --public-api-id PUBLIC_API_ID                        Tunnel ID for named Cloudflare Tunnel. Use together with public-api option.
+  --api-port API_PORT                                  The listening port for the API.
+  --api-key API_KEY                                    API authentication key.
+  --admin-key ADMIN_KEY                                API authentication key for admin tasks like loading and unloading models. If not set, will be the same as --api-key.
+  --api-enable-ipv6                                    Enable IPv6 for the API
+  --api-disable-ipv4                                   Disable IPv4 for the API
+  --nowebui                                            Do not launch the Gradio UI. Useful for launching the API in standalone mode.
 ```
 
 
            
@@ -390,20 +410,17 @@ text-generation-webui
 text-generation-webui
 └── user_data
     └── models
-        └── lmsys_vicuna-33b-v1.3
+        └── Qwen_Qwen3-8B
             ├── config.json
             ├── generation_config.json
-            ├── pytorch_model-00001-of-00007.bin
-            ├── pytorch_model-00002-of-00007.bin
-            ├── pytorch_model-00003-of-00007.bin
-            ├── pytorch_model-00004-of-00007.bin
-            ├── pytorch_model-00005-of-00007.bin
-            ├── pytorch_model-00006-of-00007.bin
-            ├── pytorch_model-00007-of-00007.bin
-            ├── pytorch_model.bin.index.json
+            ├── model-00001-of-00004.safetensors
+            ├── model-00002-of-00004.safetensors
+            ├── model-00003-of-00004.safetensors
+            ├── model-00004-of-00004.safetensors
+            ├── model.safetensors.index.json
             ├── special_tokens_map.json
             ├── tokenizer_config.json
-            └── tokenizer.model
+            └── tokenizer.json
 ```
 
 In both cases, you can use the "Model" tab of the UI to download the model from Hugging Face automatically. It is also possible to download it via the command-line with:
diff --git a/docs/01 - Chat Tab.md b/docs/01 - Chat Tab.md
index 4b177b80ff..5104895f28 100644
--- a/docs/01 - Chat Tab.md	
+++ b/docs/01 - Chat Tab.md	
@@ -2,31 +2,44 @@ Used to have multi-turn conversations with the model.
 
 ## Input area
 
-The following buttons can be found. Note that the hover menu can be replaced with always-visible buttons with the `--chat-buttons` flag.
+The main action buttons are:
 
-* **Generate**: sends your message and makes the model start a reply.
+* **Send**: sends your message and makes the model start a reply.
 * **Stop**: stops an ongoing generation as soon as the next token is generated (which can take a while for a slow model).
+
+The hover menu (☰) that appears over the chat area contains:
+
+* **Regenerate**: similar to Send, but your last message is used as input instead of the text in the input field. Note that if the temperature/top_p/top_k parameters are low in the "Parameters" tab of the UI, the new reply may end up identical to the previous one.
 * **Continue**: makes the model attempt to continue the existing reply. In some cases, the model may simply end the existing turn immediately without generating anything new, but in other cases, it may generate a longer reply.
-* **Regenerate**: similar to Generate, but your last message is used as input instead of the text in the input field. Note that if the temperature/top_p/top_k parameters are low in the "Parameters" tab of the UI, the new reply may end up identical to the previous one.
 * **Remove last reply**: removes the last input/output pair from the history and sends your last message back into the input field.
-* **Replace last reply**: replaces the last reply with whatever you typed into the input field. Useful in conjunction with "Copy last reply" if you want to edit the bot response.
-* **Copy last reply**: sends the contents of the bot's last reply to the input field.
 * **Impersonate**: makes the model generate a new message on your behalf in the input field, taking into consideration the existing chat history.
 * **Send dummy message**: adds a new message to the chat history without causing the model to generate a reply.
 * **Send dummy reply**: adds a new reply to the chat history as if the model had generated this reply. Useful in conjunction with "Send dummy message".
-* **Start new chat**: starts a new conversation while keeping the old one saved. If you are talking to a character that has a "Greeting" message defined, this message will be automatically added to the new history.
-* **Send to default**: sends the entire chat prompt up to now to the "Default" tab.
-* **Send to notebook**: sends the entire chat prompt up to now to the "Notebook" tab.
-
-The **Show controls** checkbox causes the input fields below the input textbox to disappear. It is useful for making the page fit entirely into view and not scroll.
+* **Send to Notebook**: sends the entire chat prompt up to now to the Notebook tab.
+* **Show controls**: checkbox that toggles the visibility of the sidebar controls (Start reply with, Mode, Chat style, etc.). Shortcut: Ctrl+S.
 
 ## Past chats
 
-Allows you to switch between the current and previous conversations with the current character, or between the current and previous instruct conversations (if in "instruct" mode). The **Rename** menu can be used to give a unique name to the selected conversation, and the 🗑️ button allows you to delete it.
+Allows you to switch between the current and previous conversations with the current character, or between the current and previous instruct conversations (if in "instruct" mode). The available buttons are:
+
+* **Branch**: creates a branch of the current conversation at a specific message.
+* **Rename**: allows you to give a unique name to the selected conversation.
+* **🗑️**: deletes the selected conversation.
+* **New chat**: starts a new conversation. If you are talking to a character that has a "Greeting" message defined, this message will be automatically added to the new history.
 
-## Start reply with
+A search field is also available to filter conversations by name.
 
-Whatever you type there will appear at the start of every reply by the bot. This is useful to guide the response in the desired direction.
+## Sidebar controls
+
+The sidebar (toggled via "Show controls") contains:
+
+* **Start reply with**: whatever you type there will appear at the start of every reply by the bot. This is useful to guide the response in the desired direction.
+* **Reasoning effort**: controls the thinking depth for models that support it. Options: low, medium, high.
+* **Enable thinking**: enables extended thinking mode for models that support it.
+* **Activate web search**: when enabled, the model can search the web for information before replying. You can also set the number of pages to download.
+* **Mode**: see below.
+* **Chat style**: see below.
+* **Command for chat-instruct mode**: the command that is used in chat-instruct mode to query the model to generate a reply on behalf of the character. Can be used creatively to generate specific kinds of responses. Inside this string, `<|character|>` is a placeholder that gets replaced with the bot name, and `<|prompt|>` is a placeholder that gets replaced with the full chat prompt.
 
 ## Mode
 
@@ -73,7 +86,7 @@ Now that an instruction-following model is defined, we can move on to describing
 
 ### Chat
 
-Used for talking to the character defined under "Parameters" > "Character" using a simple chat prompt in this format:
+Used for talking to the character defined under "Character" tab using a simple chat prompt in this format:
 
 ```
 Chiharu Yamada's Persona: Chiharu Yamada is a young, computer engineer-nerd with a knack for problem solving and a passion for technology.
@@ -83,7 +96,7 @@ You: How are you?
 Chiharu Yamada: I'm doing well, thank you for asking! Is there something specific you would like to talk about or ask me? I'm here to help answer any questions you may have.
 ```
 
-There are 3 adjustable parameters in "Parameters" > "Character" being used in this prompt:
+There are 3 adjustable parameters in the "Character" tab being used in this prompt:
 
 * The **Context** string appears at the top of the prompt. Most often it describes the bot's personality and adds a few example messages to guide the model towards the desired reply length and format. This string never gets truncated: as the prompt size increases, old messages get removed one at a time until the prompt becomes smaller than the truncation length set under "Parameters" > "Generation" > "Truncate the prompt up to this length".
 * The **Your name** string appears at the beginning of each user reply. By default, this string is "You".
@@ -99,7 +112,7 @@ Used for talking to an instruction-following model using the prompt format defin
 
 The prompt format is defined by the **Instruction template** parameter in "Parameters" > "Instruction template", which represents a Jinja2 template.
 
-Note that when you load a model in the "Model" tab, the web UI will try to automatically detect its instruction template (if any), and will update the values under "Parameters" > "Instruction template" accordingly. This is done using a set of regular expressions defined in `models/config.yaml`. This detection is not guaranteed to be accurate. You should check the model card on Hugging Face to see if you are using the correct prompt format.
+Note that when you load a model in the "Model" tab, the web UI will try to automatically detect its instruction template (if any), and will update the values under "Parameters" > "Instruction template" accordingly. This is done using a set of regular expressions defined in `user_data/models/config.yaml`. This detection is not guaranteed to be accurate. You should check the model card on Hugging Face to see if you are using the correct prompt format.
 
 ### Chat-instruct
 
@@ -127,8 +140,6 @@ Here, the command is
 
 Below this command, the regular chat prompt is added, including its Context string and the chat history, and then the user turn ends. The bot turn starts with the "Character's name" string followed by `:`, thus prompting the instruction-following model to write a single reply for the character.
 
-The chat-instruct command can be customized under "Parameters" > "Instruction template" > "Command for chat-instruct mode". Inside that command string, `<|character|>` is a placeholder that gets replaced with the bot name, and `<|prompt|>` is a placeholder that gets replaced with the full chat prompt.
-
 Note that you can get creative: instead of writing something trivial like "Write a single reply for the character", you could add more complex instructions like
 
 > This is an adventure game, and your task is to write a reply in name of "<|character|>" where 3 options are given for the user to then choose from.
@@ -145,4 +156,4 @@ The styles are only applied to chat and chat-instruct modes. Instruct mode has i
 
 ## Character gallery
 
-This menu is a built-in extension defined under `text-generation-webui/extensions/gallery`. It displays a gallery with your characters, and if you click on a character, it will be automatically selected in the menu under "Parameters" > "Character".
+This menu is a built-in extension defined under `text-generation-webui/extensions/gallery`. It displays a gallery with your characters, and if you click on a character, it will be automatically selected in the Character tab.
diff --git a/docs/02 - Default and Notebook Tabs.md b/docs/02 - Default and Notebook Tabs.md
index 4bb7844878..ba1028abf2 100644
--- a/docs/02 - Default and Notebook Tabs.md	
+++ b/docs/02 - Default and Notebook Tabs.md	
@@ -10,11 +10,11 @@ The number on the lower right of the Input box counts the number of tokens in th
 
 Below the Input box, the following buttons can be found:
 
+* **Continue**: starts a new generation taking as input the text in the "Output" box.
 * **Generate**: starts a new generation.
 * **Stop**: stops an ongoing generation as soon as the next token is generated (which can take a while for a slow model).
-* **Continue**: starts a new generation taking as input the text in the "Output" box.
 
-In the **Prompt** menu, you can select from some predefined prompts defined under `text-generation-webui/prompts`. The 💾 button saves your current input as a new prompt, the 🗑️ button deletes the selected prompt, and the 🔄 button refreshes the list. If you come up with an interesting prompt for a certain task, you are welcome to submit it to the repository.
+In the **Prompt** menu, you can select from saved prompts stored in `user_data/logs/notebook`. The **New** button creates a new prompt, the **Rename** button renames the selected prompt, and the 🗑️ button deletes it. The 🔄 button refreshes the list.
 
 ### Output
 
@@ -28,7 +28,7 @@ Five tabs can be found:
 
 ## Notebook tab
 
-Precisely the same thing as the Default tab, with the difference that the output appears in the same text box as the input. 
+Precisely the same thing as the Default tab, with the difference that the output appears in the same text box as the input.
 
 It contains the following additional button:
 
diff --git a/docs/03 - Parameters Tab.md b/docs/03 - Parameters Tab.md
index 8c758ce678..517950548e 100644
--- a/docs/03 - Parameters Tab.md	
+++ b/docs/03 - Parameters Tab.md	
@@ -43,9 +43,15 @@ For more information about the parameters, the [transformers documentation](http
 * **presence_penalty**: Similar to repetition_penalty, but with an additive offset on the raw token scores instead of a multiplicative factor. It may generate better results. 0 means no penalty, higher value = less repetition, lower value = more repetition. Previously called "additive_repetition_penalty".
 * **frequency_penalty**: Repetition penalty that scales based on how many times the token has appeared in the context. Be careful with this; there's no limit to how much a token can be penalized.
 * **repetition_penalty_range**: The number of most recent tokens to consider for repetition penalty. 0 makes all tokens be used.
+* **dry_multiplier**: Set to greater than 0 to enable DRY (Don't Repeat Yourself) sampling. It penalizes tokens that would extend a sequence that already appeared in the context. Recommended value: 0.8.
+* **dry_allowed_length**: The longest sequence that can be repeated without being penalized by DRY. Shorter values make DRY more aggressive.
+* **dry_base**: Controls how fast the DRY penalty grows with increasing sequence length.
 * **typical_p**: If not set to 1, select only tokens that are at least this much more likely to appear than random tokens, given the prior text.
 * **tfs**: Tries to detect a tail of low-probability tokens in the distribution and removes those tokens. See [this blog post](https://www.trentonbricken.com/Tail-Free-Sampling/) for details. The closer to 0, the more discarded tokens.
 * **top_a**: Tokens with probability smaller than `(top_a) * (probability of the most likely token)^2` are discarded.
+* **top_n_sigma**: Keeps only tokens within N standard deviations of the mean log-probability. Acts as an adaptive cutoff that adjusts to the shape of the distribution. 0 disables it.
+* **xtc_threshold**: eXclusion from Top Choices (XTC) sampling. If 2 or more tokens have probability above this threshold, the top token may be removed. This encourages the model to use less common word choices and can increase creativity.
+* **xtc_probability**: The probability that XTC removal will actually happen when the threshold condition is met. Set to 1 for it to always apply, or lower for occasional application.
 * **epsilon_cutoff**: In units of 1e-4; a reasonable value is 3. This sets a probability floor below which tokens are excluded from being sampled.
 * **eta_cutoff**: In units of 1e-4; a reasonable value is 3. The main parameter of the special Eta Sampling technique. See [this paper](https://arxiv.org/pdf/2210.15191.pdf) for a description.
 * **guidance_scale**: The main parameter for Classifier-Free Guidance (CFG). [The paper](https://arxiv.org/pdf/2306.17806.pdf) suggests that 1.5 is a good value. It can be used in conjunction with a negative prompt or not.
@@ -55,9 +61,12 @@ For more information about the parameters, the [transformers documentation](http
 *Note: Use either mirostat or dynamic_temperature, not both at the same time.*
 * **mirostat_tau**: Target perplexity for Mirostat sampling. Controls how “surprising” the text is. Higher values = more diverse, lower = more predictable. Preset Arena suggests 8 as a good value.
 * **mirostat_eta**: Learning rate for Mirostat’s perplexity adjustment. Higher values = adapts faster but less stable, lower values = slower but more stable. Preset Arena suggests 0.1 as a good value.
+* **adaptive_target**: Target probability for adaptive-p sampling. This method adjusts the sampling threshold dynamically based on an exponential moving average of recent token probabilities. 0 disables it.
+* **adaptive_decay**: EMA decay rate for adaptive-p sampling. Controls how quickly the running average adjusts. Default: 0.9.
 * **dynamic_temperature**: Activates Dynamic Temperature. This modifies temperature to range between "dynatemp_low" (minimum) and "dynatemp_high" (maximum), with an entropy-based scaling. The steepness of the curve is controlled by "dynatemp_exponent". 
 *Note: Use either dynamic_temperature or mirostat, not both at the same time.*
 * **smoothing_factor**: Activates Quadratic Sampling. When `0 < smoothing_factor < 1`, the logits distribution becomes flatter. When `smoothing_factor > 1`, it becomes more peaked.
+* **smoothing_curve**: Adjusts the dropoff curve of Quadratic Sampling. Higher values make the curve steeper. Only takes effect when smoothing_factor is set.
 * **temperature_last**: Makes temperature the last sampler instead of the first. With this, you can remove low probability tokens with a sampler like min_p and then use a high temperature to make the model creative without losing coherency. Note: this parameter takes precedence over "Sampler priority". That means that `temperature`/`dynamic_temperature`/`quadratic_sampling` will be removed from wherever they are and moved to the end of the stack.
 * **do_sample**: When unchecked, sampling is entirely disabled, and greedy decoding is used instead (the most likely token is always picked).
 * **Seed**: Set the Pytorch seed to this number. Note that some loaders do not use Pytorch (notably llama.cpp), and others are not deterministic (ExLlamaV2). For these loaders, the seed has no effect.
@@ -66,25 +75,48 @@ For more information about the parameters, the [transformers documentation](http
 
 To the right (or below if you are on mobile), the following parameters are present:
 
-* **Truncate the prompt up to this length**: Used to prevent the prompt from getting bigger than the model's context length. In the case of the transformers loader, which allocates memory dynamically, this parameter can also be used to set a VRAM ceiling and prevent out-of-memory errors. This parameter is automatically updated with the model's context length (from "n_ctx" or "max_seq_len" for loaders that use these parameters, and from the model metadata directly for loaders that do not) when you load a model.
+* **Truncate the prompt up to this length**: Used to prevent the prompt from getting bigger than the model's context length. In the case of the transformers loader, which allocates memory dynamically, this parameter can also be used to set a VRAM ceiling and prevent out-of-memory errors. This parameter is automatically updated with the model's context length (from "ctx_size" for loaders that use this parameter, and from the model metadata directly for loaders that do not) when you load a model.
 * **Maximum number of tokens/second**: to make text readable in real-time in case the model is generating too fast. Good if you want to flex and tell everyone how good your GPU is.
+* **Custom system message**: If not empty, will be used instead of the default system message in the instruction template. Useful for customizing the personality of the chatbot. Example: "You are a duck."
 * **Custom stopping strings**: The model stops generating as soon as any of the strings set in this field is generated. Note that when generating text in the Chat tab, some default stopping strings are set regardless of this parameter, like "\nYour Name:" and "\nBot name:" for chat mode. That's why this parameter has a "Custom" in its name.
 * **Custom token bans**: Allows you to ban the model from generating certain tokens altogether. You need to find the token IDs under "Default" > "Tokens" or "Notebook" > "Tokens", or by looking at the `tokenizer.json` for the model directly.
 * **auto_max_new_tokens**: When checked, the max_new_tokens parameter is expanded in the backend to the available context length. The maximum length is given by the "truncation_length" parameter. This is useful for getting long replies in the Chat tab without having to click on "Continue" many times.
 * **Ban the eos_token**: One of the possible tokens that a model can generate is the EOS (End of Sequence) token. When it is generated, the generation stops prematurely. When this parameter is checked, that token is banned from being generated, and the generation will always generate "max_new_tokens" tokens.
 * **Add the bos_token to the beginning of prompts**: By default, the tokenizer will add a BOS (Beginning of Sequence) token to your prompt. During training, BOS tokens are used to separate different documents. If unchecked, no BOS token will be added, and the model will interpret your prompt as being in the middle of a document instead of at the start of one. This significantly changes the output and can make it more creative.
 * **Skip special tokens**: When decoding the generated tokens, skip special tokens from being converted to their text representation. Otherwise, BOS appears as ``, EOS as ``, etc.
+* **prompt_lookup_num_tokens**: Activates Prompt Lookup Decoding, a form of speculative decoding for the Transformers loader. It guesses future tokens by looking for matching patterns in the prompt itself, which can speed up generation for tasks that involve repeating or paraphrasing parts of the input.
 * **Activate text streaming**: When unchecked, the full response is outputted at once, without streaming the words one at a time. I recommend unchecking this parameter on high latency networks like running the webui on Google Colab or using `--share`.
+* **Static KV cache**: Use a static cache for improved performance with the Transformers loader. May not be compatible with all models.
 * **Sampler priority**: Allows you to customize the order in which the different samplers are applied. The first sampler on the list gets applied first. With this, custom orders like `top_p -> temperature -> top_k` can be defined.
-* **Load grammar from file**: Loads a GBNF grammar from a file under `text-generation-webui/grammars`. The output is written to the "Grammar" box below. You can also save and delete custom grammars using this menu.
+* **DRY sequence breakers**: Tokens across which DRY sequence matching is not continued. Typically punctuation and special tokens. Only used when DRY is active (dry_multiplier > 0).
+* **Load grammar from file**: Loads a GBNF grammar from a file under `user_data/grammars`. The output is written to the "Grammar" box below. You can also save and delete custom grammars using this menu.
 * **Grammar**: Allows you to constrain the model output to a particular format. For instance, you can make the model generate lists, JSON, specific words, etc. Grammar is extremely powerful and I highly recommend it. The syntax looks a bit daunting at first sight, but it gets very easy once you understand it. See the [GBNF Guide](https://github.com/ggerganov/llama.cpp/blob/master/grammars/README.md) for details.
 
-## Character
+### Chat tab controls
 
-Parameters that define the character that is used in the Chat tab when "chat" or "chat-instruct" are selected under "Mode".
+The following parameters appear in the Chat tab sidebar rather than the Parameters tab:
 
-* **Character**: A dropdown menu where you can select from saved characters, save a new character (💾 button), and delete the selected character (🗑️).
-* **Your name**: Your name as it appears in the prompt.
+* **reasoning_effort**: Controls the thinking depth for models that support it (used by GPT-OSS). Options: low, medium, high.
+* **enable_thinking**: Enables extended thinking mode for models that support it (used by Seed-OSS and pre-2507 Qwen3). When enabled, the model can use a thinking step before generating its reply.
+
+## Instruction template
+
+This sub-tab within the Parameters tab defines the instruction template used in the Chat tab when "instruct" or "chat-instruct" are selected under "Mode".
+
+* **Saved instruction templates**: A dropdown menu where you can select a template. Click **Load** to apply it. The 💾 button saves the current template, and the 🗑️ button deletes the selected one.
+* **Instruction template**: A Jinja2 template that defines the prompt format for the instruction-following conversation.
+* **Send to notebook**: Send the full instruction template in string format to the Notebook tab.
+* **Chat template**: A Jinja2 template that defines the prompt format for regular chat conversations with characters.
+
+## Character tab
+
+The Character tab is a separate top-level tab that contains the following sub-tabs:
+
+### Character
+
+Parameters that define the character used in the Chat tab when "chat" or "chat-instruct" are selected under "Mode".
+
+* **Character**: A dropdown menu where you can select from saved characters, save a new character (💾 button), and delete the selected character (🗑️). The **Restore character** button resets the character to its last saved state.
 * **Character's name**: The bot name as it appears in the prompt.
 * **Context**: A string that is always at the top of the prompt. It never gets truncated. It usually defines the bot's personality and some key elements of the conversation.
 * **Greeting**: An opening message for the bot. When set, it appears whenever you start a new chat.
@@ -98,31 +130,26 @@ Note: the following replacements take place in the context and greeting fields w
 
 So you can use those special placeholders in your character definitions. They are commonly found in TavernAI character cards.
 
-## Instruction template
+### User
 
-Defines the instruction template that is used in the Chat tab when "instruct" or "chat-instruct" are selected under "Mode".
+Allows you to create and manage user profiles.
 
-* **Saved instruction templates**: A dropdown menu where you can load a saved template, save a new template (💾 button), and delete the currently selected template (🗑️).
-* **Custom system message**: A message that defines the personality of the chatbot, replacing its default "System message" string. Example: "You are a duck."
-* **Instruction template**: A Jinja2 template that defines the prompt format for the instruction-following conversation.
-* **Send to default**: Send the full instruction template in string format to the Default tab.
-* **Send to notebook**: Send the full instruction template in string format to the Notebook tab.
-* **Send to negative prompt**: Send the full instruction template in string format to the "Negative prompt" field under "Parameters" > "Generation".
-* **Chat template**: A Jinja2 template that defines the prompt format for regular chat conversations with characters.
-* **Command for chat-instruct mode**: The command that is used in chat-instruct mode to query the model to generate a reply on behalf of the character. Can be used creatively to generate specific kinds of responses.
+* **User**: A dropdown to select, save (💾), or delete (🗑️) user profiles.
+* **Name**: Your name as it appears in the prompt.
+* **Description**: An optional description of yourself that can be referenced in conversations.
 
-## Chat history
+### Chat history
 
-In this tab, you can download the current chat history in JSON format and upload a previously saved chat history. 
+In this tab, you can download the current chat history in JSON format and upload a previously saved chat history.
 
 When a history is uploaded, a new chat is created to hold it. That is, you don't lose your current chat in the Chat tab.
 
-## Upload character
+### Upload character
 
-### YAML or JSON
+#### YAML or JSON
 
-Allows you to upload characters in the YAML format used by the web UI, including optionally a profile picture. 
+Allows you to upload characters in the YAML format used by the web UI, including optionally a profile picture.
 
-### TavernAI PNG
+#### TavernAI PNG
 
 Allows you to upload a TavernAI character card. It will be converted to the internal YAML format of the web UI after upload.
diff --git a/docs/04 - Model Tab.md b/docs/04 - Model Tab.md
index f44eb96487..30b94151a7 100644
--- a/docs/04 - Model Tab.md	
+++ b/docs/04 - Model Tab.md	
@@ -2,112 +2,119 @@ This is where you load models, apply LoRAs to a loaded model, and download new m
 
 ## Model loaders
 
+### llama.cpp
+
+Loads: GGUF models. Note: GGML models have been deprecated and do not work anymore.
+
+Example: https://huggingface.co/TheBloke/Llama-2-7b-Chat-GGUF
+
+* **gpu_layers**: The number of layers to allocate to the GPU. If set to 0, only the CPU will be used. If you want to offload all layers, you can simply set this to the maximum value.
+* **ctx_size**: Context length of the model. In llama.cpp, the cache is preallocated, so the higher this value, the higher the VRAM. It is automatically set to the maximum sequence length for the model based on the metadata inside the GGUF file, but you may need to lower this value to fit the model into your GPU. Set to 0 for automatic context size based on available memory. After loading the model, the "Truncate the prompt up to this length" parameter under "Parameters" > "Generation" is automatically set to your chosen "ctx_size" so that you don't have to set the same thing twice.
+* **cache_type**: KV cache quantization type. Valid options: `fp16`, `q8_0`, `q4_0`. Lower quantization saves VRAM at the cost of some quality.
+* **tensor_split**: For multi-gpu only. Sets the amount of memory to allocate per GPU as proportions. Not to be confused with other loaders where this is set in GB; here you can set something like `30,70` for 30%/70%.
+* **batch_size**: Maximum number of prompt tokens to batch together when calling llama_eval.
+* **ubatch_size**: Physical maximum batch size for prompt processing.
+* **threads**: Number of threads. Recommended value: your number of physical cores.
+* **threads_batch**: Number of threads for batch processing. Recommended value: your total number of cores (physical + virtual).
+* **cpu_moe**: Force MoE expert layers to run on the CPU, keeping the rest on the GPU.
+* **extra_flags**: Extra flags to pass to llama-server. Format: `flag1=value1,flag2,flag3=value3`. Example: `override-tensor=exps=CPU`.
+* **mmproj**: Path to the mmproj file for multimodal (vision) models. This enables image understanding capabilities.
+* **streaming_llm**: Experimental feature to avoid re-evaluating the entire prompt when part of it is removed, for instance, when you hit the context length for the model in chat mode and an old message is removed.
+* **cpu**: Force a version of llama.cpp compiled without GPU acceleration to be used. Can usually be ignored. Only set this if you want to use CPU only and llama.cpp doesn't work otherwise.
+* **row_split**: Split the model by rows across GPUs. This may improve multi-gpu performance.
+* **no_kv_offload**: Do not offload the KV cache to the GPU. This saves VRAM but reduces performance.
+* **no_mmap**: Loads the model into memory at once, possibly preventing I/O operations later on at the cost of a longer load time.
+* **mlock**: Force the system to keep the model in RAM rather than swapping or compressing.
+* **numa**: May improve performance on certain multi-cpu systems.
+
 ### Transformers
 
-Loads: full precision (16-bit or 32-bit) models. The repository usually has a clean name without GGUF, EXL2, GPTQ, or AWQ in its name, and the model files are named `pytorch_model.bin` or `model.safetensors`. 
+Loads: full precision (16-bit or 32-bit) models, as well as bitsandbytes-quantized models. The repository usually has a clean name without GGUF, EXL2, or EXL3 in its name, and the model files are named `model.safetensors` or split into parts like `model-00001-of-00004.safetensors`.
 
-Example: [https://huggingface.co/lmsys/vicuna-7b-v1.5](https://huggingface.co/lmsys/vicuna-7b-v1.5). 
+Example: [https://huggingface.co/lmsys/vicuna-7b-v1.5](https://huggingface.co/lmsys/vicuna-7b-v1.5).
 
 Full precision models use a ton of VRAM, so you will usually want to select the "load_in_4bit" and "use_double_quant" options to load the model in 4-bit precision using bitsandbytes.
 
-This loader can also load GPTQ models and train LoRAs with them. For that, make sure to check the "auto-devices" and "disable_exllama" options before loading the model.
-
 Options:
 
-* **gpu-memory**: When set to greater than 0, activates CPU offloading using the accelerate library, where part of the layers go to the CPU. The performance is very bad. Note that accelerate doesn't treat this parameter very literally, so if you want the VRAM usage to be at most 10 GiB, you may need to set this parameter to 9 GiB or 8 GiB. It can be used in conjunction with "load_in_8bit" but not with "load-in-4bit" as far as I'm aware.
-* **cpu-memory**: Similarly to the parameter above, you can also set a limit on the amount of CPU memory used. Whatever doesn't fit either in the GPU or the CPU will go to a disk cache, so to use this option you should also check the "disk" checkbox.
-* **compute_dtype**: Used when "load-in-4bit" is checked. I recommend leaving the default value.
-* **quant_type**: Used when "load-in-4bit" is checked. I recommend leaving the default value.
+* **gpu_split**: When using multiple GPUs, sets the amount of VRAM in GB to allocate per GPU. Example: `20,7,7`.
+* **cpu_memory**: Maximum CPU memory in GiB to use for CPU offloading via the accelerate library. Whatever doesn't fit in the GPU or CPU will go to a disk cache if the "disk" checkbox is enabled.
+* **compute_dtype**: Used when "load_in_4bit" is checked. I recommend leaving the default value.
+* **quant_type**: Used when "load_in_4bit" is checked. I recommend leaving the default value.
 * **alpha_value**: Used to extend the context length of a model with a minor loss in quality. I have measured 1.75 to be optimal for 1.5x context, and 2.5 for 2x context. That is, with alpha = 2.5 you can make a model with 4096 context length go to 8192 context length.
 * **rope_freq_base**: Originally another way to write "alpha_value", it ended up becoming a necessary parameter for some models like CodeLlama, which was fine-tuned with this set to 1000000 and hence needs to be loaded with it set to 1000000 as well.
 * **compress_pos_emb**: The first and original context-length extension method, discovered by [kaiokendev](https://kaiokendev.github.io/til). When set to 2, the context length is doubled, 3 and it's tripled, etc. It should only be used for models that have been fine-tuned with this parameter set to different than 1. For models that have not been tuned to have greater context length, alpha_value will lead to a smaller accuracy loss.
-* **cpu**: Loads the model in CPU mode using Pytorch. The model will be loaded in 32-bit precision, so a lot of RAM will be used. CPU inference with transformers is older than llama.cpp and it works, but it's a lot slower. Note: this parameter has a different interpretation in the llama.cpp loader (see below).
-* **load-in-8bit**: Load the model in 8-bit precision using bitsandbytes. The 8-bit kernel in that library has been optimized for training and not inference, so load-in-8bit is slower than load-in-4bit (but more accurate).
+* **attn_implementation**: Choose the attention implementation. Valid options: `sdpa`, `eager`, `flash_attention_2`. The default (`sdpa`) works well in most cases; `flash_attention_2` may be useful for training.
+* **cpu**: Loads the model in CPU mode using Pytorch. The model will be loaded in 32-bit precision, so a lot of RAM will be used. CPU inference with transformers is older than llama.cpp and it works, but it's a lot slower. Note: this parameter has a different interpretation in the llama.cpp loader (see above).
+* **load_in_8bit**: Load the model in 8-bit precision using bitsandbytes. The 8-bit kernel in that library has been optimized for training and not inference, so load_in_8bit is slower than load_in_4bit (but more accurate).
 * **bf16**: Use bfloat16 precision instead of float16 (the default). Only applies when quantization is not used.
-* **auto-devices**: When checked, the backend will try to guess a reasonable value for "gpu-memory" to allow you to load a model with CPU offloading. I recommend just setting "gpu-memory" manually instead. This parameter is also needed for loading GPTQ models, in which case it needs to be checked before loading the model.
 * **disk**: Enable disk offloading for layers that don't fit into the GPU and CPU combined.
-* **load-in-4bit**: Load the model in 4-bit precision using bitsandbytes.
+* **load_in_4bit**: Load the model in 4-bit precision using bitsandbytes.
+* **use_double_quant**: Use double quantization with 4-bit loading for reduced memory usage.
 * **trust-remote-code**: Some models use custom Python code to load the model or the tokenizer. For such models, this option needs to be set. It doesn't download any remote content: all it does is execute the .py files that get downloaded with the model. Those files can potentially include malicious code; I have never seen it happen, but it is in principle possible.
 * **no_use_fast**: Do not use the "fast" version of the tokenizer. Can usually be ignored; only check this if you can't load the tokenizer for your model otherwise.
-* **use_flash_attention_2**: Set use_flash_attention_2=True while loading the model. Possibly useful for training.
-* **disable_exllama**: Only applies when you are loading a GPTQ model through the transformers loader. It needs to be checked if you intend to train LoRAs with the model.
-
-### ExLlamav2_HF
-
-Loads: GPTQ and EXL2 models. EXL2 models usually have "EXL2" in the model name, while GPTQ models usually have GPTQ in the model name, or alternatively something like "-4bit-128g" in the name.
-
-Examples:
-
-* https://huggingface.co/turboderp/Llama2-70B-exl2
-* https://huggingface.co/TheBloke/Llama-2-13B-chat-GPTQ
-
-* **gpu-split**: If you have multiple GPUs, the amount of memory to allocate per GPU should be set in this field. Make sure to set a lower value for the first GPU, as that's where the cache is allocated.
-* **max_seq_len**: The maximum sequence length for the model. In ExLlamaV2, the cache is preallocated, so the higher this value, the higher the VRAM. It is automatically set to the maximum sequence length for the model based on its metadata, but you may need to lower this value be able to fit the model into your GPU. After loading the model, the "Truncate the prompt up to this length" parameter under "Parameters" > "Generation" is automatically set to your chosen "max_seq_len" so that you don't have to set the same thing twice.
-* **cfg-cache**: Creates a second cache to hold the CFG negative prompts. You need to set this if and only if you intend to use CFG in the "Parameters" > "Generation" tab. Checking this parameter doubles the cache VRAM usage.
-* **no_flash_attn**: Disables flash attention. Otherwise, it is automatically used as long as the library is installed.
-* **cache_8bit**: Create a 8-bit precision cache instead of a 16-bit one. This saves VRAM but increases perplexity (I don't know by how much).
-* **cache_4bit**: Creates a Q4 cache using grouped quantization.
-
-### ExLlamav2
 
-The same as ExLlamav2_HF but using the internal samplers of ExLlamav2 instead of the ones in the Transformers library.
+### ExLlamav3_HF
 
-### AutoGPTQ
+Loads: EXL3 models. These models usually have "EXL3" or "exl3" in the model name.
 
-Loads: GPTQ models.
+Uses the ExLlamaV3 backend with Transformers samplers.
 
-* **wbits**: For ancient models without proper metadata, sets the model precision in bits manually. Can usually be ignored.
-* **groupsize**: For ancient models without proper metadata, sets the model group size manually. Can usually be ignored.
-* **triton**: Only available on Linux. Necessary to use models with both act-order and groupsize simultaneously. Note that ExLlamaV2 can load these same models on Windows without triton.
-* **no_inject_fused_attention**: Improves performance while increasing the VRAM usage.
-* **no_inject_fused_mlp**: Similar to the previous parameter but for Triton only.
-* **no_use_cuda_fp16**: On some systems, the performance can be very bad with this unset. Can usually be ignored.
-* **desc_act**: For ancient models without proper metadata, sets the model "act-order" parameter manually. Can usually be ignored.
+* **ctx_size**: Context length of the model. The cache is preallocated, so the higher this value, the higher the VRAM. It is automatically set to the maximum sequence length for the model based on its metadata, but you may need to lower this value to fit the model into your GPU. After loading the model, the "Truncate the prompt up to this length" parameter under "Parameters" > "Generation" is automatically set to your chosen "ctx_size" so that you don't have to set the same thing twice.
+* **cache_type**: KV cache quantization type. Valid options: `fp16`, `q2` to `q8`. You can also specify key and value bits separately, e.g. `q4_q8`. Lower quantization saves VRAM at the cost of some quality.
+* **gpu_split**: Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: `20,7,7`.
+* **cfg_cache**: Creates a second cache to hold the CFG negative prompts. You need to set this if and only if you intend to use CFG in the "Parameters" > "Generation" tab. Checking this parameter doubles the cache VRAM usage.
+* **no_use_fast**: Do not use the "fast" version of the tokenizer.
+* **enable_tp**: Enable Tensor Parallelism (TP) to split the model across GPUs.
+* **tp_backend**: The backend for tensor parallelism. Valid options: `native`, `nccl`. Default: `native`.
 
-### llama.cpp
+### ExLlamav3
 
-Loads: GGUF models. Note: GGML models have been deprecated and do not work anymore.
+The same as ExLlamav3_HF but using the internal samplers of ExLlamaV3 instead of the ones in the Transformers library. Supports speculative decoding with a draft model. Also supports multimodal (vision) models natively.
 
-Example: https://huggingface.co/TheBloke/Llama-2-7b-Chat-GGUF
+* **ctx_size**: Same as ExLlamav3_HF.
+* **cache_type**: Same as ExLlamav3_HF.
+* **gpu_split**: Same as ExLlamav3_HF.
+* **enable_tp**: Enable Tensor Parallelism (TP) to split the model across GPUs.
+* **tp_backend**: The backend for tensor parallelism. Valid options: `native`, `nccl`. Default: `native`.
 
-* **n-gpu-layers**: The number of layers to allocate to the GPU. If set to 0, only the CPU will be used. If you want to offload all layers, you can simply set this to the maximum value.
-* **n_ctx**: Context length of the model. In llama.cpp, the cache is preallocated, so the higher this value, the higher the VRAM. It is automatically set to the maximum sequence length for the model based on the metadata inside the GGUF file, but you may need to lower this value be able to fit the model into your GPU. After loading the model, the "Truncate the prompt up to this length" parameter under "Parameters" > "Generation" is automatically set to your chosen "n_ctx" so that you don't have to set the same thing twice.
-* **tensor_split**: For multi-gpu only. Sets the amount of memory to allocate per GPU as proportions. Not to be confused with other loaders where this is set in GB; here you can set something like `30,70` for 30%/70%.
-* **n_batch**: Batch size for prompt processing. Higher values are supposed to make generation faster, but I have never obtained any benefit from changing this value.
-* **threads**: Number of threads. Recommended value: your number of physical cores. 
-* **threads_batch**: Number of threads for batch processing. Recommended value: your total number of cores (physical + virtual).
-* **tensorcores**: Use llama.cpp compiled with "tensor cores" support, which improves performance on NVIDIA RTX cards in most cases.
-* **streamingllm**: Experimental feature to avoid re-evaluating the entire prompt when part of it is removed, for instance, when you hit the context length for the model in chat mode and an old message is removed.
-* **cpu**: Force a version of llama.cpp compiled without GPU acceleration to be used. Can usually be ignored. Only set this if you want to use CPU only and llama.cpp doesn't work otherwise. 
-* **no_mul_mat_q**: Disable the mul_mat_q kernel. This kernel usually improves generation speed significantly. This option to disable it is included in case it doesn't work on some system.
-* **no-mmap**: Loads the model into memory at once, possibly preventing I/O operations later on at the cost of a longer load time.
-* **mlock**: Force the system to keep the model in RAM rather than swapping or compressing (no idea what this means, never used it).
-* **numa**: May improve performance on certain multi-cpu systems.
-
-### llamacpp_HF
+### ExLlamav2_HF
 
-The same as llama.cpp but with transformers samplers, and using the transformers tokenizer instead of the internal llama.cpp tokenizer.
+Loads: GPTQ and EXL2 models. EXL2 models usually have "EXL2" in the model name, while GPTQ models usually have GPTQ in the model name, or alternatively something like "-4bit-128g" in the name.
 
-To use it, you need to download a tokenizer. There are two options:
+Examples:
 
-1) Download `oobabooga/llama-tokenizer` under "Download model or LoRA". That's a default Llama tokenizer.
-2) Place your .gguf in a subfolder of `models/` along with these 3 files: `tokenizer.model`, `tokenizer_config.json`, and `special_tokens_map.json`. This takes precedence over Option 1.
+* https://huggingface.co/turboderp/Llama2-70B-exl2
+* https://huggingface.co/TheBloke/Llama-2-13B-chat-GPTQ
 
-It has an additional parameter:
+* **ctx_size**: Context length of the model. In ExLlamaV2, the cache is preallocated, so the higher this value, the higher the VRAM. It is automatically set to the maximum sequence length for the model based on its metadata, but you may need to lower this value to fit the model into your GPU. After loading the model, the "Truncate the prompt up to this length" parameter under "Parameters" > "Generation" is automatically set to your chosen "ctx_size" so that you don't have to set the same thing twice.
+* **cache_type**: KV cache quantization type. Valid options: `fp16`, `fp8`, `q8`, `q6`, `q4`. Lower quantization saves VRAM at the cost of some quality.
+* **gpu_split**: Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: `20,7,7`. Make sure to set a lower value for the first GPU, as that's where the cache is allocated.
+* **alpha_value**: Used to extend the context length of a model with a minor loss in quality. Same as the Transformers parameter.
+* **compress_pos_emb**: Positional embeddings compression factor. Same as the Transformers parameter.
+* **num_experts_per_token**: Number of experts to use for generation. Applies to MoE models like Mixtral.
+* **autosplit**: Autosplit the model tensors across the available GPUs. This causes gpu_split to be ignored.
+* **enable_tp**: Enable Tensor Parallelism (TP) to split the model across GPUs.
+* **no_flash_attn**: Disables flash attention. Otherwise, it is automatically used as long as the library is installed.
+* **no_xformers**: Force xformers to not be used.
+* **no_sdpa**: Force Torch SDPA to not be used.
+* **cfg_cache**: Creates a second cache to hold the CFG negative prompts. You need to set this if and only if you intend to use CFG in the "Parameters" > "Generation" tab. Checking this parameter doubles the cache VRAM usage.
+* **no_use_fast**: Do not use the "fast" version of the tokenizer.
 
-* **logits_all**: Needs to be checked if you want to evaluate the perplexity of the llama.cpp model using the "Training" > "Perplexity evaluation" tab. Otherwise, leave it unchecked, as it makes prompt processing slower.
+### ExLlamav2
 
-### AutoAWQ
+The same as ExLlamav2_HF but using the internal samplers of ExLlamaV2 instead of the ones in the Transformers library. Supports speculative decoding with a draft model.
 
-Loads: AWQ models.
+### TensorRT-LLM
 
-Example: https://huggingface.co/TheBloke/Phind-CodeLlama-34B-v2-AWQ
+Loads: TensorRT-LLM engine models. These are highly optimized models compiled specifically for NVIDIA GPUs.
 
-The parameters are overall similar to AutoGPTQ.
+* **ctx_size**: Context length of the model.
+* **cpp_runner**: Use the ModelRunnerCpp runner, which is faster than the default ModelRunner but doesn't support streaming yet.
 
 ## Model dropdown
 
-Here you can select a model to be loaded, refresh the list of available models (🔄), load/unload/reload the selected model, and save the settings for the model. The "settings" are the values in the input fields (checkboxes, sliders, dropdowns) below this dropdown. 
+Here you can select a model to be loaded, refresh the list of available models, load/unload/reload the selected model, and save the settings for the model. The "settings" are the values in the input fields (checkboxes, sliders, dropdowns) below this dropdown.
 
 After saving, those settings will get restored whenever you select that model again in the dropdown menu.
 
@@ -115,15 +122,15 @@ If the **Autoload the model** checkbox is selected, the model will be loaded as
 
 ## LoRA dropdown
 
-Used to apply LoRAs to the model. Note that LoRA support is not implemented for all loaders. Check this [page](https://github.com/oobabooga/text-generation-webui/wiki) for details.
+Used to apply LoRAs to the model. Note that LoRA support is not implemented for all loaders. Check the [What Works](https://github.com/oobabooga/text-generation-webui/wiki/What-Works) page for details.
 
 ## Download model or LoRA
 
 Here you can download a model or LoRA directly from the https://huggingface.co/ website.
 
-* Models will be saved to `text-generation-webui/models`.
-* LoRAs will be saved to `text-generation-webui/loras`.
+* Models will be saved to `user_data/models`.
+* LoRAs will be saved to `user_data/loras`.
 
-In the input field, you can enter either the Hugging Face username/model path (like `facebook/galactica-125m`) or the full model URL (like `https://huggingface.co/facebook/galactica-125m`). To specify a branch, add it at the end after a ":" character like this: `facebook/galactica-125m:main`. 
+In the input field, you can enter either the Hugging Face username/model path (like `facebook/galactica-125m`) or the full model URL (like `https://huggingface.co/facebook/galactica-125m`). To specify a branch, add it at the end after a ":" character like this: `facebook/galactica-125m:main`.
 
 To download a single file, as necessary for models in GGUF format, you can click on "Get file list" after entering the model path in the input field, and then copy and paste the desired file name in the "File name" field before clicking on "Download".
diff --git a/docs/06 - Session Tab.md b/docs/06 - Session Tab.md
index fe96e5ca51..48735c369b 100644
--- a/docs/06 - Session Tab.md	
+++ b/docs/06 - Session Tab.md	
@@ -1,6 +1,15 @@
 Here you can restart the UI with new settings.
 
-* **Available extensions**: shows a list of extensions available under `text-generation-webui/extensions`.
+## Settings
+
+* **Toggle light/dark theme**: switches between light and dark mode.
+* **Show two columns in the Notebook tab**: toggles between the two-column Default layout and the single-column Notebook layout.
+* **Turn long pasted text into attachments in the Chat tab**: when enabled, long pasted text is automatically converted into file attachments.
+* **Include attachments/search results from previous messages in the chat prompt**: when enabled, attachments and web search results from earlier messages are included in subsequent prompts.
+
+## Extensions & flags
+
+* **Available extensions**: shows a list of extensions available under `text-generation-webui/extensions` and `text-generation-webui/user_data/extensions`. Note that some of these extensions may require manually installing Python requirements through the command: `pip install -r extensions/extension_name/requirements.txt`.
 * **Boolean command-line flags**: shows command-line flags of bool (true/false) type.
 
 After selecting your desired flags and extensions, you can restart the UI by clicking on **Apply flags/extensions and restart**.
@@ -27,6 +36,6 @@ If you used the one-click installer, this command should be executed in the term
 
 ## Saving UI defaults
 
-The **Save UI defaults to settings.yaml** button gathers the visible values in the UI and saves them to settings.yaml so that your settings will persist across multiple restarts of the UI.
+The **Save extensions settings to user_data/settings.yaml** button gathers the visible values in the UI and saves them to `user_data/settings.yaml` so that your settings will persist across multiple restarts of the UI.
 
 Note that preset parameters like temperature are not individually saved, so you need to first save your preset and select it in the preset menu before saving the defaults.
diff --git a/docs/07 - Extensions.md b/docs/07 - Extensions.md
index ebcd3c0ed9..63bddf2c10 100644
--- a/docs/07 - Extensions.md	
+++ b/docs/07 - Extensions.md	
@@ -21,17 +21,20 @@ If you create an extension, you are welcome to host it in a GitHub repository an
 |Extension|Description|
 |---------|-----------|
 |[openai](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/openai)| Creates an API that mimics the OpenAI API and can be used as a drop-in replacement. |
-|[multimodal](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/multimodal) | Adds multimodality support (text+images). For a detailed description see [README.md](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/multimodal/README.md) in the extension directory. |
-|[google_translate](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/google_translate)| Automatically translates inputs and outputs using Google Translate.|
+|[Training_PRO](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/Training_PRO)| Advanced LoRA training with support for model and LoRA merging. |
+|[superboogav2](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/superboogav2)| Enhanced RAG extension with support for PDF, DOCX, and PPTX files. |
+|[send_pictures](https://github.com/oobabooga/text-generation-webui/blob/main/extensions/send_pictures/)| Creates an image upload field that can be used to send images to the bot in chat mode. Captions are automatically generated using BLIP. |
+|[coqui_tts](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/coqui_tts)| Text-to-speech extension using Coqui XTTS v2. |
 |[silero_tts](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/silero_tts)| Text-to-speech extension using [Silero](https://github.com/snakers4/silero-models). When used in chat mode, responses are replaced with an audio widget. |
 |[whisper_stt](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/whisper_stt)| Allows you to enter your inputs in chat mode using your microphone. |
-|[sd_api_pictures](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/sd_api_pictures)| Allows you to request pictures from the bot in chat mode, which will be generated using the AUTOMATIC1111 Stable Diffusion API. See examples [here](https://github.com/oobabooga/text-generation-webui/pull/309). |
-|[character_bias](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/character_bias)| Just a very simple example that adds a hidden string at the beginning of the bot's reply in chat mode. |
-|[send_pictures](https://github.com/oobabooga/text-generation-webui/blob/main/extensions/send_pictures/)| Creates an image upload field that can be used to send images to the bot in chat mode. Captions are automatically generated using BLIP. |
+|[perplexity_colors](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/perplexity_colors)| Colors each token in the output text by its associated probability, as derived from the model logits. |
+|[google_translate](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/google_translate)| Automatically translates inputs and outputs using Google Translate.|
 |[gallery](https://github.com/oobabooga/text-generation-webui/blob/main/extensions/gallery/)| Creates a gallery with the chat characters and their pictures. |
-|[superbooga](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/superbooga)| An extension that uses ChromaDB to create an arbitrarily large pseudocontext, taking as input text files, URLs, or pasted text. Based on https://github.com/kaiokendev/superbig. |
+|[sd_api_pictures](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/sd_api_pictures)| Allows you to request pictures from the bot in chat mode, which will be generated using the AUTOMATIC1111 Stable Diffusion API. See examples [here](https://github.com/oobabooga/text-generation-webui/pull/309). |
+|[long_replies](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/long_replies)| Forces longer replies by suppressing early newlines in the model output. |
 |[ngrok](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/ngrok)| Allows you to access the web UI remotely using the ngrok reverse tunnel service (free). It's an alternative to the built-in Gradio `--share` feature. |
-|[perplexity_colors](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/perplexity_colors)| Colors each token in the output text by its associated probability, as derived from the model logits. |
+|[superbooga](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/superbooga)| An extension that uses ChromaDB to create an arbitrarily large pseudocontext, taking as input text files, URLs, or pasted text. Based on https://github.com/kaiokendev/superbig. |
+|[character_bias](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/character_bias)| Just a very simple example that adds a hidden string at the beginning of the bot's reply in chat mode. |
 
 ## How to write an extension
 
@@ -51,8 +54,8 @@ The extensions framework is based on special functions and variables that you ca
 | `def history_modifier(history)`  | Modifies the chat history before the text generation in chat mode begins. |
 | `def custom_generate_reply(...)` | Overrides the main text generation function. |
 | `def custom_generate_chat_prompt(...)` | Overrides the prompt generator in chat mode. |
-| `def tokenizer_modifier(state, prompt, input_ids, input_embeds)` | Modifies the `input_ids`/`input_embeds` fed to the model. Should return `prompt`, `input_ids`, `input_embeds`. See the `multimodal` extension for an example. |
-| `def custom_tokenized_length(prompt)` | Used in conjunction with `tokenizer_modifier`, returns the length in tokens of `prompt`. See the `multimodal` extension for an example. |
+| `def tokenizer_modifier(state, prompt, input_ids, input_embeds)` | Modifies the `input_ids`/`input_embeds` fed to the model. Should return `prompt`, `input_ids`, `input_embeds`. See the `example` extension for a template. |
+| `def custom_tokenized_length(prompt)` | Used in conjunction with `tokenizer_modifier`, returns the length in tokens of `prompt`. See the `example` extension for a template. |
 
 Additionally, you can define a special `params` dictionary. In it, the `display_name` key is used to define the displayed name of the extension in the UI, and the `is_tab` key is used to define whether the extension should appear in a new tab. By default, extensions appear at the bottom of the "Text generation" tab.
 
@@ -186,7 +189,7 @@ def bot_prefix_modifier(string, state):
 def tokenizer_modifier(state, prompt, input_ids, input_embeds):
     """
     Modifies the input ids and embeds.
-    Used by the multimodal extension to put image embeddings in the prompt.
+    Modifies the input ids and embeds fed to the model.
     Only used by loaders that use the transformers library for sampling.
     """
     return prompt, input_ids, input_embeds
diff --git a/docs/11 - AMD Setup.md b/docs/11 - AMD Setup.md
index 0bd22e7edc..6db7989df8 100644
--- a/docs/11 - AMD Setup.md	
+++ b/docs/11 - AMD Setup.md	
@@ -1,13 +1,25 @@
 ## Using an AMD GPU in Linux
 
-Requires ROCm SDK 5.4.2 or 5.4.3 to be installed. Some systems may also
-need: 
+Requires ROCm 6.4 to be installed.
+
+### Option 1: One-click installer
+
+The one-click installer (`start_linux.sh`) automatically detects AMD GPUs. When prompted, select the AMD option, or set the `GPU_CHOICE` environment variable before running:
+
+```
+GPU_CHOICE=B ./start_linux.sh
+```
+
+### Option 2: Manual conda install
+
+Follow the manual conda installation instructions in the README, using the AMD PyTorch command:
 
 ```
-sudo apt-get install libstdc++-12-dev
+pip3 install torch==2.9.1 --index-url https://download.pytorch.org/whl/rocm6.4
 ```
 
-Edit the "one_click.py" script using a text editor and un-comment and
-modify the lines near the top of the script according to your setup. In
-particular, modify the `os.environ["ROCM_PATH"] = '/opt/rocm'` line to
-point to your ROCm installation.
+Then install the project requirements with the AMD requirements file:
+
+```
+pip install -r requirements/full/requirements_amd.txt
+```
diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index 94a95b103e..cd5757f69d 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -39,7 +39,7 @@ curl http://127.0.0.1:5000/v1/completions \
 
 #### Chat completions
 
-Works best with instruction-following models. If the "instruction_template" variable is not provided, it will be guessed automatically based on the model name using the regex patterns in `models/config.yaml`.
+Works best with instruction-following models. If the "instruction_template" variable is not provided, it will be guessed automatically based on the model name using the regex patterns in `user_data/models/config.yaml`.
 
 ```shell
 curl http://127.0.0.1:5000/v1/chat/completions \
@@ -476,51 +476,45 @@ OPENAI_API_KEY=sk-111111111111111111111111111111111111111111111111
 OPENAI_API_BASE=http://127.0.0.1:5000/v1
 ```
 
-With the [official python openai client](https://github.com/openai/openai-python), the address can be set like this:
+With the [official python openai client](https://github.com/openai/openai-python) (v1.x), the address can be set like this:
 
 ```python
-import openai
-
-openai.api_key = "..."
-openai.api_base = "http://127.0.0.1:5000/v1"
-openai.api_version = "2023-05-15"
-```
-
-If using .env files to save the `OPENAI_API_BASE` and `OPENAI_API_KEY` variables, make sure the .env file is loaded before the openai module is imported:
-
-```python
-from dotenv import load_dotenv
-load_dotenv() # make sure the environment variables are set before import
-import openai
+from openai import OpenAI
+
+client = OpenAI(
+    api_key="sk-111111111111111111111111111111111111111111111111",
+    base_url="http://127.0.0.1:5000/v1"
+)
+
+response = client.chat.completions.create(
+    model="x",
+    messages=[{"role": "user", "content": "Hello!"}]
+)
+print(response.choices[0].message.content)
 ```
 
-With the [official Node.js openai client](https://github.com/openai/openai-node) it is slightly more more complex because the environment variables are not used by default, so small source code changes may be required to use the environment variables, like so:
+With the [official Node.js openai client](https://github.com/openai/openai-node) (v4.x):
 
 ```js
-const openai = OpenAI(
-  Configuration({
-    apiKey: process.env.OPENAI_API_KEY,
-    basePath: process.env.OPENAI_API_BASE
-  })
-);
-```
-
-For apps made with the [chatgpt-api Node.js client library](https://github.com/transitive-bullshit/chatgpt-api):
+import OpenAI from "openai";
 
-```js
-const api = new ChatGPTAPI({
+const client = new OpenAI({
   apiKey: process.env.OPENAI_API_KEY,
-  apiBaseUrl: process.env.OPENAI_API_BASE
+  baseURL: "http://127.0.0.1:5000/v1",
+});
+
+const response = await client.chat.completions.create({
+  model: "x",
+  messages: [{ role: "user", content: "Hello!" }],
 });
+console.log(response.choices[0].message.content);
 ```
 ### Embeddings (alpha)
 
-Embeddings requires `sentence-transformers` installed, but chat and completions will function without it loaded. The embeddings endpoint is currently using the HuggingFace model: `sentence-transformers/all-mpnet-base-v2` for embeddings. This produces 768 dimensional embeddings (the same as the text-davinci-002 embeddings), which is different from OpenAI's current default `text-embedding-ada-002` model which produces 1536 dimensional embeddings. The model is small-ish and fast-ish. This model and embedding size may change in the future.
+Embeddings requires `sentence-transformers` installed, but chat and completions will function without it loaded. The embeddings endpoint is currently using the HuggingFace model: `sentence-transformers/all-mpnet-base-v2` for embeddings. This produces 768 dimensional embeddings. The model is small and fast. This model and embedding size may change in the future.
 
 | model name             | dimensions | input max tokens | speed | size | Avg. performance |
 | ---------------------- | ---------- | ---------------- | ----- | ---- | ---------------- |
-| text-embedding-ada-002 | 1536       | 8192             | -     | -    | -                |
-| text-davinci-002       | 768        | 2046             | -     | -    | -                |
 | all-mpnet-base-v2      | 768        | 384              | 2800  | 420M | 63.3             |
 | all-MiniLM-L6-v2       | 384        | 256              | 14200 | 80M  | 58.8             |
 
@@ -528,50 +522,33 @@ In short, the all-MiniLM-L6-v2 model is 5x faster, 5x smaller ram, 2x smaller st
 
 Warning: You cannot mix embeddings from different models even if they have the same dimensions. They are not comparable.
 
-### Compatibility & not so compatibility
-
-Note: the table below may be obsolete.
-
-| API endpoint              | tested with                        | notes                                                                       |
-| ------------------------- | ---------------------------------- | --------------------------------------------------------------------------- |
-| /v1/chat/completions      | openai.ChatCompletion.create()     | Use it with instruction following models                                    |
-| /v1/embeddings            | openai.Embedding.create()          | Using SentenceTransformer embeddings                                        |
-| /v1/images/generations    | openai.Image.create()              | Bare bones, no model configuration, response_format='b64_json' only.        |
-| /v1/moderations           | openai.Moderation.create()         | Basic initial support via embeddings                                        |
-| /v1/models                | openai.Model.list()                | Lists models, Currently loaded model first, plus some compatibility options |
-| /v1/models/{id}           | openai.Model.get()                 | returns whatever you ask for                                                |
-| /v1/edits                 | openai.Edit.create()               | Removed, use /v1/chat/completions instead                                   |
-| /v1/text_completion       | openai.Completion.create()         | Legacy endpoint, variable quality based on the model                        |
-| /v1/completions           | openai api completions.create      | Legacy endpoint (v0.25)                                                     |
-| /v1/engines/\*/embeddings | python-openai v0.25                | Legacy endpoint                                                             |
-| /v1/engines/\*/generate   | openai engines.generate            | Legacy endpoint                                                             |
-| /v1/engines               | openai engines.list                | Legacy Lists models                                                         |
-| /v1/engines/{model_name}  | openai engines.get -i {model_name} | You can use this legacy endpoint to load models via the api or command line |
-| /v1/images/edits          | openai.Image.create_edit()         | not yet supported                                                           |
-| /v1/images/variations     | openai.Image.create_variation()    | not yet supported                                                           |
-| /v1/audio/\*              | openai.Audio.\*                    | supported                                                                   |
-| /v1/files\*               | openai.Files.\*                    | not yet supported                                                           |
-| /v1/fine-tunes\*          | openai.FineTune.\*                 | not yet supported                                                           |
-| /v1/search                | openai.search, engines.search      | not yet supported                                                           |
+### Compatibility
+
+| API endpoint              | notes                                                                       |
+| ------------------------- | --------------------------------------------------------------------------- |
+| /v1/chat/completions      | Use with instruction-following models. Supports streaming, tool calls.      |
+| /v1/completions           | Text completion endpoint.                                                   |
+| /v1/embeddings            | Using SentenceTransformer embeddings.                                       |
+| /v1/images/generations    | Image generation, response_format='b64_json' only.                         |
+| /v1/moderations           | Basic support via embeddings.                                               |
+| /v1/models                | Lists models. Currently loaded model first.                                 |
+| /v1/models/{id}           | Returns model info.                                                         |
+| /v1/audio/\*              | Supported.                                                                  |
+| /v1/images/edits          | Not yet supported.                                                          |
+| /v1/images/variations     | Not yet supported.                                                          |
 
 #### Applications
 
-Almost everything needs the `OPENAI_API_KEY` and `OPENAI_API_BASE` environment variable set, but there are some exceptions.
-
-Note: the table below may be obsolete.
-
-| Compatibility | Application/Library    | Website                                                                        | Notes                                                                                                                                                                                                        |
-| ------------- | ---------------------- | ------------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
-| ✅❌          | openai-python (v0.25+) | https://github.com/openai/openai-python                                        | only the endpoints from above are working. OPENAI_API_BASE=http://127.0.0.1:5001/v1                                                                                                                          |
-| ✅❌          | openai-node            | https://github.com/openai/openai-node                                          | only the endpoints from above are working. environment variables don't work by default, but can be configured (see above)                                                                                    |
-| ✅❌          | chatgpt-api            | https://github.com/transitive-bullshit/chatgpt-api                             | only the endpoints from above are working. environment variables don't work by default, but can be configured (see above)                                                                                    |
-| ✅            | anse                   | https://github.com/anse-app/anse                                               | API Key & URL configurable in UI, Images also work                                                                                                                                                           |
-| ✅            | shell_gpt              | https://github.com/TheR1D/shell_gpt                                            | OPENAI_API_HOST=http://127.0.0.1:5001                                                                                                                                                                        |
-| ✅            | gpt-shell              | https://github.com/jla/gpt-shell                                               | OPENAI_API_BASE=http://127.0.0.1:5001/v1                                                                                                                                                                     |
-| ✅            | gpt-discord-bot        | https://github.com/openai/gpt-discord-bot                                      | OPENAI_API_BASE=http://127.0.0.1:5001/v1                                                                                                                                                                     |
-| ✅            | OpenAI for Notepad++   | https://github.com/Krazal/nppopenai                                            | api_url=http://127.0.0.1:5001 in the config file, or environment variables                                                                                                                                   |
-| ✅            | vscode-openai          | https://marketplace.visualstudio.com/items?itemName=AndrewButson.vscode-openai | OPENAI_API_BASE=http://127.0.0.1:5001/v1                                                                                                                                                                     |
-| ✅❌          | langchain              | https://github.com/hwchase17/langchain                                         | OPENAI_API_BASE=http://127.0.0.1:5001/v1 even with a good 30B-4bit model the result is poor so far. It assumes zero shot python/json coding. Some model tailored prompt formatting improves results greatly. |
-| ✅❌          | Auto-GPT               | https://github.com/Significant-Gravitas/Auto-GPT                               | OPENAI_API_BASE=http://127.0.0.1:5001/v1 Same issues as langchain. Also assumes a 4k+ context                                                                                                                |
-| ✅❌          | babyagi                | https://github.com/yoheinakajima/babyagi                                       | OPENAI_API_BASE=http://127.0.0.1:5001/v1                                                                                                                                                                     |
-| ❌            | guidance               | https://github.com/microsoft/guidance                                          | logit_bias and logprobs not yet supported                                                                                                                                                                    |
+Almost everything needs the `OPENAI_API_KEY` and `OPENAI_API_BASE` environment variables set, but there are some exceptions.
+
+| Compatibility | Application/Library  | Website                                                                        | Notes                                                                                     |
+| ------------- | -------------------- | ------------------------------------------------------------------------------ | ----------------------------------------------------------------------------------------- |
+| ✅❌          | openai-python        | https://github.com/openai/openai-python                                        | Use `OpenAI(base_url="http://127.0.0.1:5000/v1")`. Only the endpoints from above work.   |
+| ✅❌          | openai-node          | https://github.com/openai/openai-node                                          | Use `new OpenAI({baseURL: "http://127.0.0.1:5000/v1"})`. See example above.              |
+| ✅            | anse                 | https://github.com/anse-app/anse                                               | API Key & URL configurable in UI, Images also work.                                       |
+| ✅            | shell_gpt            | https://github.com/TheR1D/shell_gpt                                            | OPENAI_API_HOST=http://127.0.0.1:5000                                                    |
+| ✅            | gpt-shell            | https://github.com/jla/gpt-shell                                               | OPENAI_API_BASE=http://127.0.0.1:5000/v1                                                 |
+| ✅            | gpt-discord-bot      | https://github.com/openai/gpt-discord-bot                                      | OPENAI_API_BASE=http://127.0.0.1:5000/v1                                                 |
+| ✅            | OpenAI for Notepad++ | https://github.com/Krazal/nppopenai                                            | api_url=http://127.0.0.1:5000 in the config file, or environment variables.               |
+| ✅            | vscode-openai        | https://marketplace.visualstudio.com/items?itemName=AndrewButson.vscode-openai | OPENAI_API_BASE=http://127.0.0.1:5000/v1                                                 |
+| ✅❌          | langchain            | https://github.com/hwchase17/langchain                                         | Use `base_url="http://127.0.0.1:5000/v1"`. Results depend on model and prompt formatting. |
diff --git a/docs/What Works.md b/docs/What Works.md
index 80abdc7fdb..d9c8530073 100644
--- a/docs/What Works.md	
+++ b/docs/What Works.md	
@@ -1,20 +1,19 @@
 ## What Works
 
-| Loader         | Loading 1 LoRA | Loading 2 or more LoRAs | Training LoRAs | Multimodal extension | Perplexity evaluation |
-|----------------|----------------|-------------------------|----------------|----------------------|-----------------------|
-| Transformers   |       ✅       |           ✅\*\*        |       ✅\*     |          ✅          |           ✅          |
-| llama.cpp      |       ❌       |           ❌            |       ❌       |          ❌          |    use llamacpp_HF    |
-| llamacpp_HF    |       ❌       |           ❌            |       ❌       |          ❌          |           ✅          |
-| ExLlamav2_HF   |       ✅       |           ✅            |       ❌       |          ❌          |           ✅          |
-| ExLlamav2      |       ✅       |           ✅            |       ❌       |          ❌          |   use ExLlamav2_HF    |
-| AutoGPTQ       |       ✅       |           ❌            |       ❌       |          ✅          |           ✅          |
-| AutoAWQ        |       ?        |           ❌            |       ?        |          ?           |           ✅          |
-| HQQ            |       ?        |           ?             |       ?        |          ?           |           ✅          |
+| Loader         | Loading LoRAs | Training LoRAs | Multimodal | Perplexity evaluation |
+|----------------|---------------|----------------|------------|-----------------------|
+| llama.cpp      |      ❌       |       ❌       |    ✅\*    |           ❌          |
+| Transformers   |      ✅       |       ✅       |    ✅\*\*  |           ✅          |
+| ExLlamav3_HF   |      ❌       |       ❌       |    ❌      |           ✅          |
+| ExLlamav3      |      ❌       |       ❌       |    ✅      |           ❌          |
+| ExLlamav2_HF   |      ✅       |       ❌       |    ❌      |           ✅          |
+| ExLlamav2      |      ✅       |       ❌       |    ❌      |           ❌          |
+| TensorRT-LLM   |      ❌       |       ❌       |    ❌      |           ❌          |
 
-❌ = not implemented
+❌ = not supported
 
-✅ = implemented
+✅ = supported
 
-\* Training LoRAs with GPTQ models also works with the Transformers loader. Make sure to check "auto-devices" and "disable_exllama" before loading the model.
+\* Via the `mmproj` parameter (multimodal projector file).
 
-\*\* Multi-LoRA in PEFT is tricky and the current implementation does not work reliably in all cases.
+\*\* Via the `send_pictures` extension.

From 5be68cc073b871e4429859f79cc5018f76981917 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 12:55:07 -0300
Subject: [PATCH 1219/1701] Remove Training_PRO extension

The built-in training tab now covers its essential functionality
with a more modern and correct implementation (apply_chat_template,
dynamic padding, JSONL datasets, stride overlap).
---
 docs/07 - Extensions.md                     |    1 -
 extensions/Training_PRO/README.md           |   92 --
 extensions/Training_PRO/custom_scheduler.py |  433 -------
 extensions/Training_PRO/matplotgraph.py     |   62 -
 extensions/Training_PRO/script.py           | 1293 -------------------
 extensions/Training_PRO/train_utils.py      |  368 ------
 6 files changed, 2249 deletions(-)
 delete mode 100644 extensions/Training_PRO/README.md
 delete mode 100644 extensions/Training_PRO/custom_scheduler.py
 delete mode 100644 extensions/Training_PRO/matplotgraph.py
 delete mode 100644 extensions/Training_PRO/script.py
 delete mode 100644 extensions/Training_PRO/train_utils.py

diff --git a/docs/07 - Extensions.md b/docs/07 - Extensions.md
index 63bddf2c10..48cd30cef5 100644
--- a/docs/07 - Extensions.md	
+++ b/docs/07 - Extensions.md	
@@ -21,7 +21,6 @@ If you create an extension, you are welcome to host it in a GitHub repository an
 |Extension|Description|
 |---------|-----------|
 |[openai](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/openai)| Creates an API that mimics the OpenAI API and can be used as a drop-in replacement. |
-|[Training_PRO](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/Training_PRO)| Advanced LoRA training with support for model and LoRA merging. |
 |[superboogav2](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/superboogav2)| Enhanced RAG extension with support for PDF, DOCX, and PPTX files. |
 |[send_pictures](https://github.com/oobabooga/text-generation-webui/blob/main/extensions/send_pictures/)| Creates an image upload field that can be used to send images to the bot in chat mode. Captions are automatically generated using BLIP. |
 |[coqui_tts](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/coqui_tts)| Text-to-speech extension using Coqui XTTS v2. |
diff --git a/extensions/Training_PRO/README.md b/extensions/Training_PRO/README.md
deleted file mode 100644
index 3eda332162..0000000000
--- a/extensions/Training_PRO/README.md
+++ /dev/null
@@ -1,92 +0,0 @@
-# Training_PRO
-
-This is an expanded and reworked Training tab
-Maintained by FP
-
-[![ko-fi](https://ko-fi.com/img/githubbutton_sm.svg)](https://ko-fi.com/Q5Q5MOB4M)
-
-Repo home:
-
-https://github.com/FartyPants/Training_PRO
-
-In general the repo above is ahead of the extension included in text WebUi.
-
-## News
-
-- NEFtune: add noise to help with generalization
-- Loss Graph in interface.
-- Supports Mistral training
-- some roundabout around pytorch and transformers version desync
-
-![image](https://github.com/FartyPants/Training_PRO/assets/23346289/e389ec69-d7ad-4922-9ad9-865625997479)
-
-## Features/Changes
-
-- Chunking: precise raw text slicer (PRTS) uses sentence slicing and making sure things are clean on all ends
-- overlap chunking - this special overlapping will make additional overlap block based on logical rules (aka no overlap block on hard cut)
-- custom scheduler (follow the code to make your own) In LR Scheduler select FP_low_epoch_annealing - this scheduler will keep the LR constant for first epoch then use cosine for the rest - this part would be best to spawn into a new py file
-- saves graph png file at the end with learning rate and loss per epoch
-- adding EOS to each block or to hard cut only
-- automatically lowers gradient accumulation if you go overboard and set gradient accumulation that will be higher than actual data - transformers would then throw error (or they used to, not sure if still true) but in any way, it will fix bad data
-- turn BOS on and OFF
-- target selector
-- DEMENTOR LEARNING (experimental) Deep Memorization Enforcement Through Overlapping and Repetition. This is an experiment for long-text learning using low epochs (basically use 1 epoch with constant LR or 2 epochs with FP_low_epoch_annealing LR scheduler)
-- Getting rid of micro batch size/batch size confusion. Now there is True Batch Size and Gradient accumulation slider, consisten with all the other training out there
-- Ability to save Checkpoint during training with a button
-- Ability to change Stop Loss during training
-- different modes of checkpoint auto saving
-- Function to Check Dataset and suggest parameters such as warmup and checkpoint save frequency before training
-- Graph Training Loss in interface
-- more custom schedulers
-  
-### Notes:
-
-This uses it's own chunking code for raw text based on sentence splitting. This will avoid weird cuts in the chunks and each chunk should now start with sentence and end on some sentence. It works hand in hand with Hard Cut. A propper use is to structure your text into logical blocks (ideas) separated by three \n then use three \n in hard cut. This way each chunk will contain only one flow of ideas and not derail in the thoughts. And Overlapping code will create overlapped blocks on sentence basis too, but not cross hard cut, thus not cross different ideas either. Does it make any sense? No? Hmmmm...
-
-### Custom schedulers
-
-A bunch of custom (combination) schedulers are added to the LR schedule. These are based on my own experiments
-
-**FP_low_epoch_annealing**
-
-Uses constant LR (with warmup) for 1 epoch only. The rest of the epoch(s) is cosine annealing. So 10 epochs - 1 will be constant 9 will be nose dive down. However a typical usage would be 2 epochs (hence low epoch in name). 1st is constant, the second is annealing. Simple. I use it 90% of time.
-
-**FP_half_time_annealing**
-
-Like the low epoch, but now the total number of steps is divided by 2. First half is constant, second half is annealing. So 10 epochs - 5 will be constant, 5 will be cosine nose down.
-
-**FP_raise_fall_creative**
-
-This is a sine raise till half of the total steps then cosine fall the rest. (Or you may think of the curve as sine in its entirety. The most learning is done in the hump, in the middle. The warmup entry has no effect, since sine is automatically warm up.
-The idea is to start very mildly as not to overfit with the first blocks of dataset. It seems to broaden the scope of the model making it less strict for tight dataset. 
-
-### Targets
-
-Normal LORA is q, v and that's what you should use. You can use (q k v o) or (q k v) and it will give you a lot more trainable parameters. The benefit is that you can keep rank lower and still attain the same coherency as q v with high rank. Guanaco has been trained with QLORA and q k v o for example and they swear by it.
-
-### DEMENTOR LEARNING (experimental) Deep Memorization Enforcement Through Overlapping and Repetition
-
-This is and experimental chunking to train long-form text in low number of epochs (basically 1) with sliding repetition. The depth of learning directly depends on the cutoff_length. Increasing cutoff length will also increase number of blocks created from long-form text (which is contrary to normal training). It is based on my own wild experiments. 
-
-### Getting rid of batch size and micro batch size
-
-Keeping consistency with everyone else. 
-
-Listen, There is only ONE batch size - the True batch size (called previously micro-batch size in WebUI) - this is how many blocks are processed at once (during a single step). It eats GPU, but it really helps with the quality training (in fact the ideal batch size would be the same as number of blocks - which is unrealistic) - so the idea is to cram as much True Batch Size before your GPU blows with OOM. On 24GB this is about 10 for 13b (loaded with 4-bit)
-
-So no micro batch size - it is now called True Batch Size, because that's what it is.
-
-The other thing is Gradient Accumulation - this is an emulation of the above Batch size - a virtual batch size, if you will. If your GPU can't handle real batch size then you may fake it using Gradient Accumulation. This will accumulate the gradients over so many steps defined here and then update the weights at the end without increase in GPU.
-Gradient accumulation is like a virtual Batch size multiplier without the GPU penalty.
-
-If your batch size is 4 and your gradient accumulation is 2 then it sort of behaves as if we have batch size 8. *Sort of* because Batch size of 4 and GA of 2 is NOT the same as batch size of 2 and GA of 4. (It produces different weights - hence it's not an equivalent). The idea is that if you don't have GPU - using GA to extend batch size is the next best thing (good enough) since you have no other choice.
-
-If all you can afford is 1 batch size, then increasing GA will likely make the learning better in some range of GA (it's not always more is better).
-
-However - GA is not some golden goose. As said, it isn't the same as batch size. In fact GA may worsen your learning as well.
-
-I would suggest a series of experiment where you would put batch size as high as possible without OOM, set GA 1, then repeat training while increasing the GA (2, 4...), and see how the model changes. It's likely that it would follow some sort of curve where GA will seem to help before it will make it worse. Some people believe that if you can squeeze 6 BATCH Size, then you should not bother with GA at all... YMMW
-
-High Batch Size vs High GA would also likely produce different results in terms of learning  words vs style. How? Hmmmm... good question.
-
-One optical "benefit" of GA is that the loss will fluctuate less (because of all the gradient accumulation, which works as a form of noise smoothing as well).
diff --git a/extensions/Training_PRO/custom_scheduler.py b/extensions/Training_PRO/custom_scheduler.py
deleted file mode 100644
index cbcac1173a..0000000000
--- a/extensions/Training_PRO/custom_scheduler.py
+++ /dev/null
@@ -1,433 +0,0 @@
-from functools import partial
-import torch
-import transformers
-import math
-from torch.optim.lr_scheduler import LambdaLR
-
-from peft import (
-    PeftModel,
-)
-
-RED = "\033[91m"
-YELLOW = "\033[93m"
-GREEN = "\033[92m"
-RESET = "\033[0m"
-
-last_print_label = ''
-
-custom_scheduler_params = {'trigger_loss': 0.0, 'ramp_down_ratio':1.0, 'current_loss': 0.0,'dynamic_scheduler_stop': False, 'calc_ramp_down_at_step': 0, 'calc_num_training_steps': 0}
-
-
-def custom_scheduler_global_update(current_loss: float):
-    custom_scheduler_params.update({'current_loss': current_loss})
-  
-def custom_scheduler_global_setup(trigger_loss: float, ramp_down_ratio: float):
-    custom_scheduler_params.update({'trigger_loss': trigger_loss})
-    custom_scheduler_params.update({'ramp_down_ratio': ramp_down_ratio})
-
-    # calculates the total num steps after trigger
-    custom_scheduler_params.update({'calc_num_training_steps': 0})
-    #calculates steps when the ramp_down trigger occurred
-    custom_scheduler_params.update({'calc_ramp_down_at_step': 0})
-    # triggers scheduler stopping after it reached calc_num_training_steps
-    custom_scheduler_params.update({'dynamic_scheduler_stop': False})
-
-
-# hold constant to the half of epochs then cosine down to 0
-def _get_fp_half_schedule_with_warmup_lr_lambda(current_step: int, *, num_warmup_steps: int, num_training_steps: int, num_firstepoch_steps: int):
-    
-    global last_print_label
-    print_label = ''
-
-    half_steps = num_training_steps//2
-    
-    num_warmup_steps = min(num_warmup_steps,half_steps)
-
-    if current_step < num_warmup_steps:
-        print_label = 'Scheduler: Warmup'
-    elif current_step < half_steps:
-        print_label = 'Scheduler: Hold'
-    else:
-        print_label = 'Scheduler: Annealing'
-    
-    if print_label != last_print_label:
-        print(print_label)
-    
-    last_print_label = print_label
-
-    if current_step < num_warmup_steps:
-        return float(current_step) / float(max(1, num_warmup_steps))
-    
-    if current_step < half_steps:
-        return 1.0 
-    
-    progress = float(current_step - half_steps) / float(max(1, num_training_steps - half_steps))
-    num_cycles = 0.5
-    return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))    
- 
-
-# raise up in cosine, then fall back in cosine
-def _get_fp_cosine_raise_and_fall_lr_lambda(current_step: int, *, num_warmup_steps: int, num_training_steps: int, num_firstepoch_steps: int):
-    
-    global last_print_label
-    print_label = ''
-
-    half_steps = num_training_steps//2
-    
-    #num_warmup_steps = min(num_warmup_steps,half_steps)
-
-    if current_step < half_steps:
-        print_label = 'Scheduler: Raise'
-    else:
-        print_label = 'Scheduler: Fall'
-    
-    if print_label != last_print_label:
-        print(print_label)
-    
-    last_print_label = print_label
-
-    
-    # linear
-    #    return float(current_step) / float(max(1, num_warmup_steps))
-    
-    progress = float(current_step - half_steps) / float(max(1, num_training_steps - half_steps))
-    num_cycles = 0.5
-    return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))    
- 
-# constant to the first epochs then cosine down to 0 over the rest epochs
-def _get_fp_cosine_schedule_with_warmup_lr_lambda(current_step: int, *, num_warmup_steps: int, num_training_steps: int, num_firstepoch_steps: int):
-    
-    global last_print_label
-    print_label = ''
-    
-    num_warmup_steps = min(num_warmup_steps,num_firstepoch_steps)
-
-    if current_step < num_warmup_steps:
-        print_label = 'Scheduler: Warmup'
-    elif current_step < num_firstepoch_steps:
-        print_label = 'Scheduler: Hold'
-    else:
-        print_label = 'Scheduler: Annealing'
-    
-    if print_label != last_print_label:
-        print(print_label)
-    
-    last_print_label = print_label
-
-    if current_step < num_warmup_steps:
-        return float(current_step) / float(max(1, num_warmup_steps))
-    
-    if current_step < num_firstepoch_steps:
-        return 1.0 
-    
-    progress = float(current_step - num_firstepoch_steps) / float(max(1, num_training_steps - num_firstepoch_steps))
-    num_cycles = 0.5
-    return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))    
-    
-# halve lr each epoch   
-
-def _get_fp_cdrop_rate_schedule_with_warmup_lr_lambda(current_step: int, *, num_warmup_steps: int, num_training_steps: int, num_firstepoch_steps: int):
-    
-    global last_print_label
-    print_label = ''
-    
-    num_warmup_steps = min(num_warmup_steps, num_firstepoch_steps)
-
-    current_epoch = (current_step // num_firstepoch_steps) + 1
-    
-    
-    if current_step < num_warmup_steps:
-        print_label = 'Scheduler: Warmup'
-    elif current_step < num_firstepoch_steps:
-        print_label = 'Scheduler: Hold'
-    else:
-        print_label = 'Scheduler: Drop Rate'
-    
-    if print_label != last_print_label:
-        print(print_label)
-    
-    last_print_label = print_label
-
-    if current_step < num_warmup_steps:
-        return float(current_step) / float(max(1, num_warmup_steps))
-    
-    if current_step < num_firstepoch_steps:
-        return 1.0 
-
-    # Compute the learning rate for the annealing phase
-    
-    learning_rate = 1.0 / float(2 ** (current_epoch - 1))
-   
-    return learning_rate
-
-# epoch decay: 1/(1 + decay * epoch)
-
-def custom_cosine_scheduler_with_warmup(optimizer, num_warmup_steps, num_training_steps, num_firstepoch_steps, last_epoch=-1):
-    """
-    Args:
-        optimizer ([`~torch.optim.Optimizer`]):
-            The optimizer for which to schedule the learning rate.
-        num_warmup_steps (`int`):
-            The number of steps for the warmup phase.
-        num_training_steps (`int`):
-            The total number of training steps.
-        last_epoch (`int`, *optional*, defaults to -1):
-            The index of the last epoch when resuming training.
-
-    Return:
-        `torch.optim.lr_scheduler.LambdaLR` with the appropriate schedule.
-    """
-    
-    lr_lambda = partial(
-        _get_fp_cosine_schedule_with_warmup_lr_lambda,
-        num_warmup_steps=num_warmup_steps,
-        num_training_steps=num_training_steps,
-        num_firstepoch_steps = num_firstepoch_steps,
-    )
-    return LambdaLR(optimizer, lr_lambda, last_epoch)
-
-def custom_half_scheduler_with_warmup(optimizer, num_warmup_steps, num_training_steps, num_firstepoch_steps, last_epoch=-1):
-    """
-    Args:
-        optimizer ([`~torch.optim.Optimizer`]):
-            The optimizer for which to schedule the learning rate.
-        num_warmup_steps (`int`):
-            The number of steps for the warmup phase.
-        num_training_steps (`int`):
-            The total number of training steps.
-        last_epoch (`int`, *optional*, defaults to -1):
-            The index of the last epoch when resuming training.
-
-    Return:
-        `torch.optim.lr_scheduler.LambdaLR` with the appropriate schedule.
-    """
-    
-    lr_lambda = partial(
-        _get_fp_half_schedule_with_warmup_lr_lambda,
-        num_warmup_steps=num_warmup_steps,
-        num_training_steps=num_training_steps,
-        num_firstepoch_steps = num_firstepoch_steps,
-    )
-    return LambdaLR(optimizer, lr_lambda, last_epoch)
-
-def custom_raise_fall_scheduler_with_warmup(optimizer, num_warmup_steps, num_training_steps, num_firstepoch_steps, last_epoch=-1):
-    """
-    Args:
-        optimizer ([`~torch.optim.Optimizer`]):
-            The optimizer for which to schedule the learning rate.
-        num_warmup_steps (`int`):
-            The number of steps for the warmup phase.
-        num_training_steps (`int`):
-            The total number of training steps.
-        last_epoch (`int`, *optional*, defaults to -1):
-            The index of the last epoch when resuming training.
-
-    Return:
-        `torch.optim.lr_scheduler.LambdaLR` with the appropriate schedule.
-    """
-    
-    lr_lambda = partial(
-        _get_fp_cosine_raise_and_fall_lr_lambda,
-        num_warmup_steps=num_warmup_steps,
-        num_training_steps=num_training_steps,
-        num_firstepoch_steps = num_firstepoch_steps,
-    )
-    return LambdaLR(optimizer, lr_lambda, last_epoch)
-
-
-def neftune_forward(self, input: torch.Tensor):
-    """
-    Implements the NEFTune forward pass for the model. Note this works only for
-    torch.nn.Embedding layers. This method is slightly adapted from the original source code
-    that can be found here: https://github.com/neelsjain/NEFTune
-
-    Args:
-        input (`torch.Tensor`):
-            The input tensor to the model.
-        noise_alpha (`float`):
-            The noise alpha value to use for the NEFTune forward pass.
-    """
-    embeddings = torch.nn.functional.embedding(
-        input, self.weight, self.padding_idx, self.max_norm, self.norm_type, self.scale_grad_by_freq, self.sparse
-    )
-
-    if self.training:
-        # Add noise to the embeddings
-        dims = torch.tensor(embeddings.size(1) * embeddings.size(2))
-        mag_norm = self.neftune_noise_alpha / torch.sqrt(dims)
-        embeddings = embeddings + torch.zeros_like(embeddings).uniform_(-mag_norm, mag_norm)
-
-    return embeddings    
-
-
-class FPNEFtuneTrainer(transformers.Trainer):
-    def __init__(self,neftune_noise_alpha:float = 0.0, model = None, *args, **kwargs):
-        self.neftune_noise_alpha = neftune_noise_alpha
-        if self.neftune_noise_alpha > 0.0:
-            model = self._activate_neftune(model)
-        super().__init__(model = model, *args, **kwargs)
-
-   
-    def _activate_neftune(self, model):
-        r"""
-        Activates the neftune as presented in this code: https://github.com/neelsjain/NEFTune and paper: https://arxiv.org/abs/2310.05914
-        """
-        print(f"Activating {RED}NEFtune{RESET} with scale: {self.neftune_noise_alpha}")
-        if isinstance(model, transformers.PreTrainedModel):
-            embeddings = model.get_input_embeddings()
-        elif isinstance(model, PeftModel):
-            embeddings = model.base_model.get_input_embeddings()
-
-        embeddings.neftune_noise_alpha = self.neftune_noise_alpha
-        old_forward = embeddings.forward
-
-        # This hack seems to be needed to properly use a custom forward pass
-        # all credits to: https://discuss.pytorch.org/t/how-can-i-replace-the-forward-method-of-a-predefined-torchvision-model-with-my-customized-forward-function/54224/11
-        bound_method = neftune_forward.__get__(embeddings, embeddings.__class__)
-        setattr(embeddings, "forward", bound_method)
-
-        # embeddings.forward = neftune_forward
-        embeddings._trl_old_forward = old_forward
-
-        return model
-    
-    def train(self, *args, **kwargs):
-        output = super().train(*args, **kwargs)
-
-        # After training we make sure to retrieve back the original forward pass method
-        # for the embedding layer
-        if self.neftune_noise_alpha is not None:
-
-            if isinstance(self.model, transformers.PreTrainedModel):
-                embeddings = self.model.get_input_embeddings()
-            elif isinstance(self.model, PeftModel):
-                embeddings = self.model.base_model.get_input_embeddings()
-
-            if hasattr(embeddings, "_trl_old_forward"):
-                embeddings.forward = embeddings._trl_old_forward
-                del embeddings._trl_old_forward
-                del embeddings.neftune_noise_alpha
-
-        return output
-
-
-class FPSchedulerTrainer(transformers.Trainer):
-    def __init__(self,neftune_noise_alpha:float = 0.0, model = None, *args, **kwargs):
-        self.neftune_noise_alpha = neftune_noise_alpha
-        if self.neftune_noise_alpha > 0.0:
-            model = self._activate_neftune(model)
-        super().__init__(model = model, *args, **kwargs)
-
-   
-    def _activate_neftune(self, model):
-        r"""
-        Activates the neftune as presented in this code: https://github.com/neelsjain/NEFTune and paper: https://arxiv.org/abs/2310.05914
-        """
-        print(f"Activating {RED}NEFtune{RESET} with scale: {self.neftune_noise_alpha}")
-        if isinstance(model, transformers.PreTrainedModel):
-            embeddings = model.get_input_embeddings()
-        elif isinstance(model, PeftModel):
-            embeddings = model.base_model.get_input_embeddings()
-
-        embeddings.neftune_noise_alpha = self.neftune_noise_alpha
-        old_forward = embeddings.forward
-
-        # This hack seems to be needed to properly use a custom forward pass
-        # all credits to: https://discuss.pytorch.org/t/how-can-i-replace-the-forward-method-of-a-predefined-torchvision-model-with-my-customized-forward-function/54224/11
-        bound_method = neftune_forward.__get__(embeddings, embeddings.__class__)
-        setattr(embeddings, "forward", bound_method)
-
-        # embeddings.forward = neftune_forward
-        embeddings._trl_old_forward = old_forward
-
-        return model
-    
-    def train(self, *args, **kwargs):
-        output = super().train(*args, **kwargs)
-
-        # After training we make sure to retrieve back the original forward pass method
-        # for the embedding layer
-        if self.neftune_noise_alpha is not None:
-
-            if isinstance(self.model, transformers.PreTrainedModel):
-                embeddings = self.model.get_input_embeddings()
-            elif isinstance(self.model, PeftModel):
-                embeddings = self.model.base_model.get_input_embeddings()
-
-            if hasattr(embeddings, "_trl_old_forward"):
-                embeddings.forward = embeddings._trl_old_forward
-                del embeddings._trl_old_forward
-                del embeddings.neftune_noise_alpha
-
-        return output
-
-
-    def create_scheduler(self, num_training_steps: int, optimizer: torch.optim.Optimizer = None):
-        #Setup the scheduler. The optimizer of the trainer must have been set up either before this method is called or passed as an argument.
-        
-        num_train_epochs = self.args.num_train_epochs
-        num_warmup_steps=self.args.get_warmup_steps(num_training_steps)
-        num_firstepoch_steps = math.ceil(num_training_steps/num_train_epochs)
-        num_warmup_acc = num_warmup_steps*self.args.gradient_accumulation_steps 
-        num_firstepoch_steps_acc = num_firstepoch_steps*self.args.gradient_accumulation_steps
-        num_training_steps_acc = num_training_steps*self.args.gradient_accumulation_steps
-
-        custom_scheduler_params.update({'dynamic_scheduler_stop': False})
- 
-        print (f"Warm-up steps aligned to Gradient accumulation ({self.args.gradient_accumulation_steps}) = {num_warmup_acc} actual warmup steps")
-        if self.args.lr_scheduler_type == 'cosine':
-            
-            num_warmup_acc_min = min(num_warmup_acc, num_firstepoch_steps_acc)
-
-            if num_warmup_acc>num_firstepoch_steps_acc:
-                print(f"\033[1;31;1mWARNING: The number of warmup steps is set too high! It will be clamped to 1 epoch, essentially going from warmup to annealing.\033[0;37;0m")
-                print (f"FP Scheduler Warmup: 0-[{num_warmup_acc_min}], Hold [{num_warmup_acc_min}]-{num_firstepoch_steps_acc}, Annealing {num_firstepoch_steps_acc}-{num_training_steps_acc}")
-            else:
-                print (f"FP Scheduler Warmup: 0-{num_warmup_acc_min}, Hold {num_warmup_acc_min}-{num_firstepoch_steps_acc}, Annealing {num_firstepoch_steps_acc}-{num_training_steps_acc}")
-
-            self.lr_scheduler = custom_cosine_scheduler_with_warmup(
-                    optimizer=self.optimizer if optimizer is None else optimizer,
-                    num_warmup_steps=num_warmup_steps,
-                    num_training_steps=num_training_steps, 
-                    num_firstepoch_steps = num_firstepoch_steps,
-                )
-            self._created_lr_scheduler = True
-            return self.lr_scheduler
-        elif self.args.lr_scheduler_type == 'constant':
-           
-            half_step_acc = num_training_steps_acc//2
-            num_warmup_acc_min = min(num_warmup_acc, half_step_acc)
-
-            if num_warmup_acc>half_step_acc:
-                print(f"\033[1;31;1mWARNING: The number of warmup steps is set too high! It will be clamped to half of all epochs, essentially going from warmup to annealing in the middle.\033[0;37;0m")
-                print (f"FP Scheduler Warmup: 0-[{num_warmup_acc_min}], Hold [{num_warmup_acc_min}]-{half_step_acc}, Annealing {half_step_acc}-{num_training_steps_acc}")
-            else:
-                print (f"FP Scheduler Warmup: 0-{num_warmup_acc_min}, Hold {num_warmup_acc_min}-{half_step_acc}, Annealing {half_step_acc}-{num_training_steps_acc}")
-
-            self.lr_scheduler = custom_half_scheduler_with_warmup(
-                    optimizer=self.optimizer if optimizer is None else optimizer,
-                    num_warmup_steps=num_warmup_steps,
-                    num_training_steps=num_training_steps, 
-                    num_firstepoch_steps = num_firstepoch_steps,
-                )
-            self._created_lr_scheduler = True
-            return self.lr_scheduler
-        elif self.args.lr_scheduler_type == 'constant_with_warmup':
-           
-            half_step_acc = num_training_steps_acc//2
-            
-            if num_warmup_steps>0:
-                print(f"Warmup doesn't apply to this scheduler [Raise-Fall]")
-
-            print (f"Scheduler Raise: 0-{half_step_acc}, Fall {half_step_acc}-{num_training_steps_acc}")
-
-            self.lr_scheduler = custom_raise_fall_scheduler_with_warmup(
-                    optimizer=self.optimizer if optimizer is None else optimizer,
-                    num_warmup_steps=num_warmup_steps,
-                    num_training_steps=num_training_steps, 
-                    num_firstepoch_steps = num_firstepoch_steps,
-                )
-            self._created_lr_scheduler = True
-            return self.lr_scheduler        
-        else:
-            return  super().create_scheduler(num_training_steps=num_training_steps, optimizer=optimizer)
\ No newline at end of file
diff --git a/extensions/Training_PRO/matplotgraph.py b/extensions/Training_PRO/matplotgraph.py
deleted file mode 100644
index b30bee8365..0000000000
--- a/extensions/Training_PRO/matplotgraph.py
+++ /dev/null
@@ -1,62 +0,0 @@
-import os
-import json
-
-def create_graph(lora_path, lora_name):
-    try:
-        import matplotlib.pyplot as plt
-        from matplotlib.ticker import ScalarFormatter
-        
-        peft_model_path = f'{lora_path}/training_graph.json'
-        image_model_path = f'{lora_path}/training_graph.png'
-        # Check if the JSON file exists
-        if os.path.exists(peft_model_path):
-            # Load data from JSON file
-            with open(peft_model_path, 'r') as file:
-                data = json.load(file)
-            # Extract x, y1, and y2 values
-            x = [item['epoch'] for item in data]
-            y1 = [item['learning_rate'] for item in data]
-            y2 = [item['loss'] for item in data]
-
-            # Create the line chart
-            fig, ax1 = plt.subplots(figsize=(10, 6))
-        
-
-            # Plot y1 (learning rate) on the first y-axis
-            ax1.plot(x, y1, 'b-', label='Learning Rate')
-            ax1.set_xlabel('Epoch')
-            ax1.set_ylabel('Learning Rate', color='b')
-            ax1.tick_params('y', colors='b')
-
-            # Create a second y-axis
-            ax2 = ax1.twinx()
-
-            # Plot y2 (loss) on the second y-axis
-            ax2.plot(x, y2, 'r-', label='Loss')
-            ax2.set_ylabel('Loss', color='r')
-            ax2.tick_params('y', colors='r')
-
-            # Set the y-axis formatter to display numbers in scientific notation
-            ax1.yaxis.set_major_formatter(ScalarFormatter(useMathText=True))
-            ax1.ticklabel_format(style='sci', axis='y', scilimits=(0,0))
-
-            # Add grid
-            ax1.grid(True)
-
-            # Combine the legends for both plots
-            lines, labels = ax1.get_legend_handles_labels()
-            lines2, labels2 = ax2.get_legend_handles_labels()
-            ax2.legend(lines + lines2, labels + labels2, loc='best')
-
-            # Set the title
-            plt.title(f'{lora_name} LR and Loss vs Epoch')
-
-            # Save the chart as an image
-            plt.savefig(image_model_path)
-
-            print(f"Graph saved in {image_model_path}")
-        else:
-            print(f"File 'training_graph.json' does not exist in the {lora_path}")
-      
-    except ImportError:
-        print("matplotlib is not installed. Please install matplotlib to create PNG graphs")
diff --git a/extensions/Training_PRO/script.py b/extensions/Training_PRO/script.py
deleted file mode 100644
index e2f90f17e2..0000000000
--- a/extensions/Training_PRO/script.py
+++ /dev/null
@@ -1,1293 +0,0 @@
-import os
-
-os.environ["WANDB_MODE"] = "offline"
-# os.environ["WANDB_DISABLED"] = "true"
-
-import json
-import math
-import random
-import shutil
-import sys
-import threading
-import time
-import traceback
-from datetime import datetime
-from pathlib import Path
-
-import gradio as gr
-import pandas as pd
-import torch
-import transformers
-
-from functools import partial
-
-from .custom_scheduler import FPSchedulerTrainer, FPNEFtuneTrainer
-
-from .matplotgraph import create_graph
-from .train_utils import get_available_loras_local, precise_cut, sliding_block_cut, download_file_from_url
-
-from datasets import Dataset, load_dataset
-from peft import (
-    LoraConfig,
-    get_peft_model,
-    prepare_model_for_kbit_training,
-    set_peft_model_state_dict
-)
-from peft.utils.other import \
-    TRANSFORMERS_MODELS_TO_LORA_TARGET_MODULES_MAPPING as model_to_lora_modules
-from transformers.models.auto.modeling_auto import (
-    MODEL_FOR_CAUSAL_LM_MAPPING_NAMES
-)
-
-from modules import shared, utils
-from modules.ui import create_refresh_button
-
-from modules.evaluate import (
-    calculate_perplexity,
-    generate_markdown_table,
-    save_past_evaluations
-)
-from modules.logging_colors import logger
-from modules.models import reload_model
-from modules.utils import natural_keys
-
-import warnings
-warnings.filterwarnings(action = "ignore", message="torch.utils.checkpoint:")
-warnings.filterwarnings(action = "ignore", message="`do_sample` is set to `False`")
-
-params = {
-        "display_name": "Training PRO",
-        "is_tab": True
-}
-
-non_serialized_params = {
-        "debug_slicer": False,
-        "Lora_sortedByTime": False,
-        "stop_at_loss": 0,
-        "save_steps_under_loss": 0.0,
-        "save_checkpoint_now": False,
-        "training_loop": False,
-        "current_stability": 0,
-        "save_epochs": 0,
-        "checkpoint_offset": 0,
-        "epoch_offset":0,
-        "safe_serialization": False,
-}
-
-MODEL_CLASSES = {v[1]: v[0] for v in MODEL_FOR_CAUSAL_LM_MAPPING_NAMES.items()}
-
-PARAMETERS = ["lora_name", "always_override", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "raw_text_file", "higher_rank_limit", "warmup_steps", "optimizer", "hard_cut_string", "train_only_after", "stop_at_loss", "add_eos_token", "min_chars", "report_to", "precize_slicing_overlap", "add_eos_token_type", "save_steps_under_loss", "add_bos_token", "training_projection","sliding_window","warmup_ratio","grad_accumulation","neft_noise_alpha"]
-WANT_INTERRUPT = False
-
-train_log = {}
-train_template = {}
-train_log_graph = []
-train_choices = ["all","q-k-v-o","q-k-v","k-v-down","q-v"]
-
-statistics = {
-			'loss': [],
-			'lr': [],
-}
-
-RED = "\033[91m"
-YELLOW = "\033[93m"
-GREEN = "\033[92m"
-RESET = "\033[0m"
-
-def ui():
-
-    with gr.Tab('Train LoRA', elem_id='lora-train-tab'):
-        tmp = gr.State('')
-        with gr.Row():
-            with gr.Column():
-                # YY.MM.DD
-                gr.Markdown("`Ver: 23.10.20 (REV2)` This is enhanced version of QLora Training. [Maintained by FP](https://github.com/FartyPants/Training_PRO/tree/main)")
-
-                with gr.Row():
-                    with gr.Column(scale=5):
-                        with gr.Row():
-                            copy_from = gr.Dropdown(label='Copy parameters from', value='None', choices=get_available_loras_local(non_serialized_params['Lora_sortedByTime']), elem_classes=['slim-dropdown'])
-                            create_refresh_button(copy_from, lambda: None, lambda: {'choices': get_available_loras_local(non_serialized_params['Lora_sortedByTime'])}, 'refresh-button')
-                    with gr.Column():
-                        sort_byTime = gr.Checkbox(label='Sort list by Date', value=False, info='Sorts Loras by date created.', elem_classes=['no-background'])                        
-
-                with gr.Row():
-                    with gr.Column(scale=5):
-                        lora_name = gr.Textbox(label='Name', info='The name of your new LoRA file')
-    
-                    with gr.Column():
-                        always_override = gr.Checkbox(label='Override Existing Files', value=False, info='If the name is the same, checking will replace the existing file, and unchecking will load and continue from it (the rank must be the same).', elem_classes=['no-background'])
-
-                with gr.Row():
-                    with gr.Column():
-                        lora_rank = gr.Slider(label='LoRA Rank', value=32, minimum=0, maximum=1024, step=4, info='Also called dimension count. Higher values = larger file, more content control. Smaller values = smaller file, less control. Use 4 or 8 for style, 128 or 256 to teach, 1024+ for fine-detail on big data. More VRAM is needed for higher ranks.')
-                        lora_alpha = gr.Slider(label='LoRA Alpha', value=64, minimum=0, maximum=2048, step=4, info='This divided by the rank becomes the scaling of the LoRA. Higher means stronger. A good standard value is twice your Rank.')
-                        batch_size = gr.Slider(visible= False, label='Batch Size', value=0, minimum=0, maximum=1024, step=4, info='Now Replaced with Gradient accumulation. Keeping it for sake of old saved data')
-                        micro_batch_size = gr.Slider(label='True Batch Size', value=4, minimum=1, maximum=128, step=1, info='Specifies how many text blocks per step will be trained. The higher value, the better the concept of training will be, but it requires more GPU memory and it reduces speed.')
-                        grad_accumulation = gr.Slider(label='Gradient Accumulation Steps', value=1, minimum=1, maximum=256, step=1, info="Virtually multiplies the Batch Size by averaging the learning over more than one step. VRAM friendly. Evens out loss fluctuations but can also degrade training fidelity.")
-
-                    with gr.Column():
-                        stop_at_loss = gr.Slider(label='Stop at loss (Can be changed during training)', minimum=0.0, maximum=3.0, step=0.1, value=0.00, info='The process will automatically stop once the desired loss value is reached.')
-                        gr.Markdown(" ")
-                        epochs = gr.Number(label='Epochs', value=3, info='Number of times every entry in the dataset should be fed into training. So 1 means feed each item in once, 5 means feed it in five times, etc.')
-                        learning_rate = gr.Textbox(label='Learning Rate', value='3e-4', info='In scientific notation. 3e-4 is a good starting base point. 1e-2 is extremely high, 1e-6 is extremely low.')
-                        lr_scheduler_type = gr.Dropdown(label='LR Scheduler', value='linear', choices=['linear', 'constant', 'constant_with_warmup', 'cosine', 'cosine_with_restarts', 'polynomial', 'inverse_sqrt', 'FP_low_epoch_annealing', 'FP_half_time_annealing','FP_raise_fall_creative'], info='Learning rate scheduler - defines how the learning rate changes over time. Custom schedulers: FP_low_epoch_annealing, FP_half_time_annealing, FP_raise_fall_creative (see README)', elem_classes=['slim-dropdown'])
-                        
-                with gr.Accordion(label='Checkpoints', open=True):
-                    with gr.Row():
-                        with gr.Column():
-                            save_steps = gr.Number(label='Save every n steps', value=0, info='A checkpoint will be saved every n steps and at each Epoch boundary. (0 = OFF)')
-                        with gr.Column():    
-                            save_steps_under_loss = gr.Slider(label='Save at 10% Loss change', value=1.8, minimum=0.0, maximum=3.0, step=0.1, info="Saves checkpoints at (or bellow) this loss and then each time loss falls by at least 10% This works independently from 'Save every n steps'")    
-                    with gr.Row():        
-                        save_chackpoint_now = gr.Button('Queue Checkpoint Now')
-
-                with gr.Accordion(label='Advanced Options', open=True):
-                    with gr.Row():
-                        with gr.Column():
-                            warmup_steps = gr.Number(label='Warmup Steps', value=100, info='Number of max steps used for a linear warmup. Reduces early over-fitting by the first training blocks. Value has precedent over Warmup Ratio. Aligns to the closest multiple of graddient accumulation')
-                            warmup_ratio = gr.Slider(label='Warmup Ratio', minimum=0.0, maximum=0.2, step=0.025, value=0.0, info='Ratio of total training steps that will be used for a linear warmup. It applies only if Warmup Step is 0.')
-                            neft_noise_alpha = gr.Slider(label='NEFtune noise scale', minimum=0.0, maximum=15, step=1, value=0.0, info='Add noise to the training to improve generalization. [0 - OFF, Starting value to experiment: 5]')
-                            training_projection = gr.Radio(value = train_choices[4], label='LLaMA Target Projections', info='Change the targets (LORA is typically q-v)', choices=train_choices)    
-                            lora_dropout = gr.Slider(label='LoRA Dropout', minimum=0.0, maximum=1.0, step=0.025, value=0.05, info='Percentage probability for dropout of LoRA layers. This can help reduce overfitting. Most users should leave at default.')
-                            optimizer = gr.Dropdown(label='Optimizer', value='adamw_torch', choices=['adamw_hf', 'adamw_torch', 'adamw_torch_fused', 'adamw_torch_xla', 'adamw_apex_fused', 'adafactor', 'adamw_bnb_8bit', 'adamw_anyprecision', 'sgd', 'adagrad'], info='Different optimizer implementation options, for advanced users. Effects of different options are not well documented yet.', elem_classes=['slim-dropdown'])
-
-                        with gr.Column():
-                            train_only_after = gr.Textbox(label='Train Only After', value='', info='Only consider text *after* this string in any given chunk for training. For Alpaca datasets, use "### Response:" to only train the response and ignore the input.')
-                            add_bos_token = gr.Checkbox(label='Add BOS token', value=True, info="Adds BOS token for each dataset item")
-                            add_eos_token = gr.Checkbox(label='Add EOS token', value=False, info="Adds EOS token for each dataset item")
-                            add_eos_token_type = gr.Dropdown(label='EOS placement (Text file)', choices=['Every Block', 'Hard Cut Blocks Only'], value='Every Block', info='', allow_custom_value = False)
-                            
-                            higher_rank_limit = gr.Checkbox(label='Enable higher ranks', value=False, info='If checked, changes Rank/Alpha slider above to go much higher. This will not work without a datacenter-class GPU.')
-                            report_to = gr.Radio(label="Save detailed logs with", value="None", choices=["None", "wandb", "tensorboard"], interactive=True)
-                # for future            
-                #with gr.Accordion(label='Dynamic Scheduler', open = False):
-                #    ds_min_epochs = gr.Number(label='Minimum Epochs', value='1', info='Minimum epochs that will be always performed before ramp down can be triggered')
-                #    ds_max_epochs = gr.Number(label='Maximum Epochs (fallback)', value='50', info='Maximum Epochs before the training will bail out completely (should be a large number)')
-                #    ds_loss_trigger = gr.Slider(label='Trigger Loss', minimum=0.0, maximum=2.8, step=0.1, value=1.6, info='Loss at which the ramp down schedule will be triggered')
-                #    ds_loss_rolling_window = gr.Number(label='Loss rolling average', value='4', info='Calculate loss by averaging last x numbers to avoid jumps and noise')
-                #    ds_epochs_to_ramp = gr.Slider(label='Ramp down ratio', minimum=0.0, maximum=2.0, step=0.1, value=1.00, info='How long the ramp down will last relative to ellapsed steps (before trigger)')
-                #    gr.Markdown('These are settings for FP_dynamic_loss_trigger scheduler. The scheduler will do warm up, then hold constant untill a loss falls under Trigger Loss, then it will commence linear ramp down schedule and stop. The length of ramp down is set by Ramp down ratio where (ramp down steps) = ratio * (elapsed steps). (The time to completition shown will be very high untill ramp down is triggered.)')
-                        
-
-            with gr.Column():
-                with gr.Tab(label='Formatted Dataset'):
-                    with gr.Row():
-                        with gr.Column():
-                            with gr.Row():
-                                dataset = gr.Dropdown(choices=get_datasets('user_data/training/datasets', 'json'), value='None', label='Dataset', info='The dataset file to use for training.', elem_classes=['slim-dropdown'])
-                                create_refresh_button(dataset, lambda: None, lambda: {'choices': get_datasets('user_data/training/datasets', 'json')}, 'refresh-button')
-                            with gr.Row():
-                                eval_dataset = gr.Dropdown(choices=get_datasets('user_data/training/datasets', 'json'), value='None', label='Evaluation Dataset', info='The (optional) dataset file used to evaluate the model after training.', elem_classes=['slim-dropdown'])
-                                create_refresh_button(eval_dataset, lambda: None, lambda: {'choices': get_datasets('user_data/training/datasets', 'json')}, 'refresh-button')
-
-                        with gr.Column():
-                            with gr.Row():
-                                format = gr.Dropdown(choices=get_datasets('user_data/training/formats', 'json'), value='None', label='Data Format', info='The format file used to decide how to format the dataset input.', elem_classes=['slim-dropdown'])
-                                create_refresh_button(format, lambda: None, lambda: {'choices': get_datasets('user_data/training/formats', 'json')}, 'refresh-button')
-                            with gr.Row():
-                                eval_steps = gr.Number(label='Evaluate every n steps', value=100, info='If an evaluation dataset is given, test it every time this many steps pass.')
-
-                with gr.Tab(label="Text file"):
-                    with gr.Row():
-                        raw_text_file = gr.Dropdown(choices=get_datasets('user_data/training/datasets', 'txt'), value='None', label='Text file', info='The text file to use for training.', elem_classes=['slim-dropdown'])
-                        create_refresh_button(raw_text_file, lambda: None, lambda: {'choices': get_datasets('user_data/training/datasets', 'txt')}, 'refresh-button')
-
-                    with gr.Row():
-                        with gr.Column():
-                            precize_slicing_overlap = gr.Checkbox(label='Add Overlapping blocks', value = True)
-                            sliding_window = gr.Checkbox(label='DEMENTOR Long-form Learning by FP (Highly Experimental, use low epochs)', value = False, info='Deep Memorization Enforcement Through Overlapping and Repetition. (I named it, so shush). Special process for learning long-form text using low amount of epochs.')
-                            #debug_slicer = gr.Checkbox(label='Dump sentencelist.json to logs', value = non_serialized_params['debug_slicer'], info='Debug Slicer')
-
-                        with gr.Column():
-                            hard_cut_string = gr.Textbox(label='Hard Cut String', value='\\n\\n\\n', info='String that indicates a cut between logical blocks of text (ex. Ideas or Chapters). Helps prevent unwanted overlap between unrelated ideas.')
-                            min_chars = gr.Number(label='Ignore small blocks', value=0, info='Ignore Text blocks that have less or equal characters than this number.')
-                with gr.Tab(label="URL"):
-                    with gr.Row():
-                        with gr.Column():
-                            download_file_url = gr.Textbox(label='Download JSON or txt file to datasets (or formats) folder', value='',info='The URL of a file to download. If on github, make sure you get url of the raw file (https://raw.githubusercontent.com/...). If huggin face, make sure the url has /resolve/ in it not /blob/')
-                            with gr.Row():
-                                download_check_overwrite = gr.Checkbox(label='Overwrite', value=False, info='Overwrite if file exist')
-                                download_folder = gr.Radio(label="Destination", value='user_data/training/datasets', choices=['user_data/training/datasets', 'user_data/training/formats'], interactive=True)
-                            download_button = gr.Button('Download')
-                            download_status = gr.Textbox(label='Download Status', value='', interactive=False)
-                with gr.Row():
-                    with gr.Column():
-                        with gr.Row():
-                            cutoff_len = gr.Slider(label='Chunk Length (Cutoff Length)', minimum=32, maximum=2048, value=256, step=32, info='The maximum length of a chunk (in tokens). Applies to both JSON dataset and text files. Higher values require much more VRAM.')
-                with gr.Row():
-                    with gr.Column():
-                        check_dataset_btn = gr.Button('Verify Dataset/Text File and suggest data entries')    
-                        check_dataset_txt = gr.Textbox(label='Dataset info', value='')
-
-                with gr.Row():
-                    start_button = gr.Button("Start LoRA Training", variant='primary')
-                    stop_button = gr.Button("Interrupt")
-
-                with gr.Accordion(label="Graph", open=True):
-                    with gr.Row():
-                        # show_actions_button = False - we use old gradio
-                        plot_graph = gr.LinePlot(x="epoch", y="value", title="Loss Metrics", overlay_point=True, tooltip=["epoch", "value"], x_lim=[0, 1], y_lim=[0, 3.5], width=500, height=250) 
- 
-                output = gr.Markdown(value="Ready")
-
-    with gr.Tab('Perplexity evaluation', elem_id='evaluate-tab'):
-        with gr.Row():
-            with gr.Column():
-                models = gr.Dropdown(utils.get_available_models(), label='Models', multiselect=True)
-                evaluate_text_file = gr.Dropdown(choices=['wikitext', 'ptb', 'ptb_new'] + get_datasets('user_data/training/datasets', 'txt')[1:], value='wikitext', label='Input dataset', info='The text file on which the model will be evaluated. The first options are automatically downloaded: wikitext, ptb, and ptb_new. The next options are your local text files under user_data/training/datasets.')
-                with gr.Row():
-                    with gr.Column():
-                        stride_length = gr.Slider(label='Stride', minimum=1, maximum=2048, value=512, step=1, info='Used to make the evaluation faster at the cost of accuracy. 1 = slowest but most accurate. 512 is a common value.')
-
-                    with gr.Column():
-                        max_length = gr.Number(label='max_length', precision=0, step=256, value=0, info='The context for each evaluation. If set to 0, the maximum context length for the model will be used.')
-
-                with gr.Row():
-                    start_current_evaluation = gr.Button("Evaluate loaded model")
-                    start_evaluation = gr.Button("Evaluate selected models")
-                    stop_evaluation = gr.Button("Interrupt")
-
-            with gr.Column():
-                evaluation_log = gr.Markdown(value='')
-
-        evaluation_table = gr.Dataframe(value=generate_markdown_table(), interactive=True)
-        with gr.Row():
-            save_comments = gr.Button('Save comments', elem_classes="small-button")
-            refresh_table = gr.Button('Refresh the table', elem_classes="small-button")
-
-    # Training events
-    all_params = [lora_name, always_override, save_steps, micro_batch_size, batch_size, epochs, learning_rate, lr_scheduler_type, lora_rank, lora_alpha, lora_dropout, cutoff_len, dataset, eval_dataset, format, eval_steps, raw_text_file, higher_rank_limit, warmup_steps, optimizer, hard_cut_string, train_only_after, stop_at_loss, add_eos_token, min_chars, report_to, precize_slicing_overlap, add_eos_token_type, save_steps_under_loss, add_bos_token, training_projection,sliding_window,warmup_ratio,grad_accumulation, neft_noise_alpha]
-
-    def fix_old_version(batch_size_val,micro_batch_size_val, grad_accumulation_val):
-        if batch_size_val>0:
-            gradient_acc =  batch_size_val // micro_batch_size_val
-            print(f"Using Old version of Batch Size ({batch_size_val}) to set Gradient Accumulation: {gradient_acc}")
-            return gradient_acc
-
-        return grad_accumulation_val
-
-    
-    copy_from.change(partial(do_copy_params, all_params= all_params), copy_from, all_params).then(fix_old_version,[batch_size,micro_batch_size, grad_accumulation],grad_accumulation)
-    start_button.click(do_train, all_params, [output,plot_graph])
-    stop_button.click(do_interrupt, None, None, queue=False)
-    higher_rank_limit.change(change_rank_limit, [higher_rank_limit], [lora_rank, lora_alpha])
-
-    def trigger_stop_at_loss(stop_at_loss_value):
-        non_serialized_params.update({"stop_at_loss": stop_at_loss_value})
-        if non_serialized_params['training_loop']:
-            print(f"Queue: [Stop at loss Change] to {stop_at_loss_value}")
-
-
-    stop_at_loss.change(trigger_stop_at_loss, stop_at_loss, None)
-
-    def trigger_save_checkpoint():
-        non_serialized_params.update({"save_checkpoint_now": True})
-        if non_serialized_params['training_loop']:
-            print("Queue: [Save checkpoint] Checkpoint will be saved after the current step is finished.")
-        else:
-            print("Use during the training to save the checkpoint at any time.")
-
-
-    def update_button():
-        return gr.Button.update('[Checkpoint in Queue]', variant='stop', interactive=True)
-
-    def update_button2():
-        time.sleep(1.0)
-        return gr.Button.update('Queue Checkpoint Now', variant='secondary',interactive = True)
-
-    save_chackpoint_now.click(trigger_save_checkpoint, None, None).then(update_button, None,save_chackpoint_now).then(update_button2, None,save_chackpoint_now)
-
-    dataset_calc_params = [save_steps,micro_batch_size, epochs, cutoff_len, dataset, format, raw_text_file, warmup_steps, hard_cut_string, min_chars, precize_slicing_overlap,sliding_window,warmup_ratio,grad_accumulation]
-
-    def check_dataset(save_steps:int, micro_batch_size: int, epochs: int, cutoff_len: int, dataset:str, format:str, raw_text_file:str, warmup_steps:int, hard_cut_string:str, min_chars:int, precize_slicing_overlap:bool,sliding_window:bool,warmup_ratio:float,grad_accumulation:int):
-        result = "Specify JSON dastaset or Text file"
-        total_blocks = 0
-        if shared.tokenizer is None:
-            yield "Tokenizer is not available. Please Load some Model first."
-            return
-        
-        
-        if raw_text_file not in ['None', '']:
-            logger.info("Loading Text file...")
-            fullpath = clean_path('user_data/training/datasets', f'{raw_text_file}')
-            fullpath = Path(fullpath)
-            if fullpath.is_dir():
-                logger.info('Training path directory {}'.format(raw_text_file))
-                raw_text = ""
-                file_paths = sorted(fullpath.glob('*.txt'), key=lambda path: natural_keys(path.name))
-                for file_path in file_paths:
-                    if file_path.is_file():
-                        with file_path.open('r', encoding='utf-8') as file:
-                            raw_text += file.read().replace('\r', '')
-
-                        logger.info(f"Loaded training file: {file_path.name}")
-            else:
-                try:
-                    with open(clean_path('user_data/training/datasets', f'{raw_text_file}.txt'), 'r', encoding='utf-8') as file:
-                        raw_text = file.read().replace('\r', '')
-                except:
-                    yield f"{raw_text_file}.txt doesn't seem to exsist anymore... check your user_data/training/datasets folder"
-                    return
-            
- 
-            if min_chars<0:
-                min_chars = 0
-
-            # == New more precise slicing on sentence boundary ==
-            if sliding_window:
-                text_chunks = sliding_block_cut(raw_text, min_chars, False, cutoff_len, hard_cut_string,non_serialized_params['debug_slicer'])
-            else:
-                text_chunks = precise_cut(raw_text, precize_slicing_overlap, min_chars, False, cutoff_len, hard_cut_string,non_serialized_params['debug_slicer'])
-
-            total_blocks = len(text_chunks)
-            result = f"Text: ({raw_text_file}.txt) has {total_blocks} blocks (Block Size {cutoff_len} tokens)"
-            del text_chunks
-       
-        else:
-            if dataset in ['None', '']:
-                yield "Select dataset or text file."
-                return 
-
-            if format in ['None', '']:
-                yield "Select format choice for dataset."
-                return
-
-            with open(clean_path('user_data/training/formats', f'{format}.json'), 'r', encoding='utf-8-sig') as formatFile:
-                format_data: dict[str, str] = json.load(formatFile)
-
-            def generate_prompt(data_point: dict[str, str]):
-                for options, data in format_data.items():
-                    if set(options.split(',')) == set(x[0] for x in data_point.items() if (type(x[1]) is str and len(x[1].strip()) > 0)):
-                        for key, val in data_point.items():
-                            if type(val) is str:
-                                data = data.replace(f'%{key}%', val)
-                        return data
-                raise RuntimeError(f'Data-point "{data_point}" has no keyset match within format "{list(format_data.keys())}"')
-
-            def tokenize_dummy(prompt):
-
-                input_ids = shared.tokenizer.encode(prompt, truncation=True, max_length=cutoff_len)
-                labels = [1] * len(input_ids)
-                input_ids = torch.tensor(input_ids)
-                return {
-                    "input_ids": input_ids,
-                    "labels": labels,
-                    "attention_mask": input_ids.ne(shared.tokenizer.pad_token_id),
-                }
-
-            def generate_and_tokenize_prompt(data_point):
-                prompt = generate_prompt(data_point)
-                return tokenize_dummy(prompt)
-
-            logger.info("Loading JSON datasets...")
-            data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
-            
-            data_keys = [] 
-
-            if data:
-                if 'train' in data:  # Check if the 'train' split exists in the dataset
-                    data_keys = list(data['train'][0].keys())
-                    print("Data Keys:", data_keys)
-            else:
-                print("The dataset is empty.")
-
-            train_data = data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
-            total_blocks = train_data.num_rows
-
-            result = f"Dataset: ({dataset}.json) has {total_blocks} blocks @ length = {cutoff_len} tokens\n(Keys: {data_keys} - Format: {format}.json): "
-
-            #for options, data in format_data.items():
-            #    format_keys = options.split(',')
-            #    result += f"{format_keys}, "
-            #result = result.rstrip()    
-            #result = result.rstrip(',')  
-
-        if total_blocks>0:
-            number_ofSteps = int(math.ceil(total_blocks / micro_batch_size) * epochs) 
-            num_stepsPer_epoch = int(math.ceil(number_ofSteps/epochs))
-            min_warm = math.ceil(100 / grad_accumulation)
-
-            warmup_steps_suggest = min(int(min_warm*grad_accumulation), int(math.ceil(number_ofSteps * 0.1)))
-            warmup_steps_suggest = min(warmup_steps_suggest,num_stepsPer_epoch)
-
-            save_each_n_min = int(math.ceil(number_ofSteps/10))
-            save_each_n_max = int(math.ceil(number_ofSteps/5))
-            gradient_accumulation_max = int(total_blocks)//micro_batch_size
-
- 
-            result += f"\n[Batch Size: {micro_batch_size}, Epochs: {epochs}, Gradient Accumulation: {grad_accumulation}]\n"
-            result += f"Total number of steps: {number_ofSteps}\n"
-            result += f"Steps per each Epoch: {num_stepsPer_epoch}\n"
-            result += f"Suggestions:\n"
-            result += f"Checkpoints: Save every {save_each_n_min} - {save_each_n_max} steps (Current: {int(save_steps)})\n"
-            result += f"Warmup steps: {warmup_steps_suggest} (Current: {int(warmup_steps)})"
-            if gradient_accumulation_max < grad_accumulation: 
-                result += f"\n\nWARNING: Gradient Accumulation {grad_accumulation} is too high: It should be below {gradient_accumulation_max}"
-
-
-        yield result
-        return
-    
-    check_dataset_btn.click(check_dataset, dataset_calc_params ,check_dataset_txt)
-
-    # Evaluation events. For some reason, the interrupt event
-    # doesn't work with the .then() syntax, so I write them one
-    # by one in this ugly but functional way.
-    ev = start_evaluation.click(calculate_perplexity, [models, evaluate_text_file, stride_length, max_length], evaluation_log, show_progress=False)
-    start_evaluation.click(generate_markdown_table, None, evaluation_table, show_progress=False)
-
-    start_current_evaluation.click(lambda: ['current model'], None, tmp)
-    ev_cur = start_current_evaluation.click(calculate_perplexity, [tmp, evaluate_text_file, stride_length, max_length], evaluation_log, show_progress=False)
-    start_current_evaluation.click(generate_markdown_table, None, evaluation_table, show_progress=False)
-
-    stop_evaluation.click(None, None, None, cancels=[ev, ev_cur], queue=False)
-    refresh_table.click(generate_markdown_table, None, evaluation_table, show_progress=True)
-    save_comments.click(
-        save_past_evaluations, evaluation_table, None).then(
-        lambda: "Comments saved.", None, evaluation_log, show_progress=False)
-
-    def reload_lora():
-        return gr.Dropdown.update(choices=get_available_loras_local(non_serialized_params['Lora_sortedByTime']))
- 
-    # nonserialized items
-
-    sort_byTime.change(lambda x: non_serialized_params.update({"Lora_sortedByTime": x}), sort_byTime, None).then(reload_lora,None,copy_from) 
-    #debug_slicer.change(lambda x: non_serialized_params.update({"debug_slicer": x}), debug_slicer, None)
-
-    def update_dataset():
-        return gr.update(choices=get_datasets('user_data/training/datasets', 'json')), gr.update(choices=get_datasets('user_data/training/datasets', 'txt'))
-
-    download_button.click(download_file_from_url, [download_file_url,download_check_overwrite,download_folder] , download_status).then(update_dataset,None,[dataset , raw_text_file])
-
-def get_datasets(path: str, ext: str):
-    # include subdirectories for raw txt files to allow training from a subdirectory of txt files
-    #if ext == "txt":
-    #    return ['None'] + sorted(set([k.stem for k in list(Path(path).glob('txt')) + list(Path(path).glob('*/')) if k.stem != 'put-trainer-datasets-here']), key=natural_keys)
-
-    return ['None'] + sorted(set([k.stem for k in Path(path).glob(f'*.{ext}') if k.stem != 'put-trainer-datasets-here']), key=natural_keys)
-
-def do_interrupt():
-    global WANT_INTERRUPT
-    WANT_INTERRUPT = True
-
-
-def do_copy_params(lora_name: str, all_params):
-
-    if lora_name:
-        f_name = f"{shared.args.lora_dir}/{clean_path(None, lora_name)}/training_parameters.json"
-        if Path(f_name).is_file():
-            with open(f_name, 'r', encoding='utf-8') as format_file:
-                params: dict[str, str] = json.load(format_file)
-        else:
-            params = {}
-    else:
-        params = {}        
-
-    result = list()
-    for i in range(0, len(PARAMETERS)):
-        key = PARAMETERS[i]
-        if key in params:
-            result.append(params[key])
-        else:
-            result.append(all_params[i])
-
-    return result
-
-
-def change_rank_limit(use_higher_ranks: bool):
-    mult = 2 if use_higher_ranks else 1
-    return {"maximum": 1024 * mult, "__type__": "update"}, {"maximum": 2048 * mult, "__type__": "update"}
-
-
-def clean_path(base_path: str, path: str):
-    """Strips unusual symbols and forcibly builds a path as relative to the intended directory."""
-    path = path.replace('\\', '/').replace('..', '_')
-    if base_path is None:
-        return path
-
-    return f'{Path(base_path).absolute()}/{path}'
-
-
-def backup_adapter(input_folder):
-    # Get the creation date of the file adapter_model.bin
-    try:
-        adapter_file = Path(f"{input_folder}/adapter_model.bin")
-        if adapter_file.is_file():
-
-            logger.info("Backing up existing LoRA adapter...")
-            creation_date = datetime.fromtimestamp(adapter_file.stat().st_ctime)
-            creation_date_str = creation_date.strftime("Backup-%Y-%m-%d")
-
-            # Create the new subfolder
-            subfolder_path = Path(f"{input_folder}/{creation_date_str}")
-            subfolder_path.mkdir(parents=True, exist_ok=True)
-
-            # Check if the file already exists in the subfolder
-            backup_adapter_file = Path(f"{input_folder}/{creation_date_str}/adapter_model.bin")
-            if backup_adapter_file.is_file():
-                print(" - Backup already exists. Skipping backup process.")
-                return
-
-            # Copy existing files to the new subfolder
-            existing_files = Path(input_folder).iterdir()
-            for file in existing_files:
-                if file.is_file():
-                    shutil.copy2(file, subfolder_path)
-    except Exception as e:
-        print("An error occurred in backup_adapter:", str(e))
-
-
-def calc_trainable_parameters(model):
-    trainable_params = 0
-    all_param = 0
-    for _, param in model.named_parameters():
-        num_params = param.numel()
-        # if using DS Zero 3 and the weights are initialized empty
-        if num_params == 0 and hasattr(param, "ds_numel"):
-            num_params = param.ds_numel
-
-        all_param += num_params
-        if param.requires_grad:
-            trainable_params += num_params
-
-    return trainable_params, all_param
-
-
-
-def do_train(lora_name: str, always_override: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, raw_text_file: str, higher_rank_limit: bool, warmup_steps: int, optimizer: str, hard_cut_string: str, train_only_after: str, stop_at_loss: float, add_eos_token: bool, min_chars: int, report_to: str, precize_slicing_overlap: bool, add_eos_token_type: str, save_steps_under_loss: float, add_bos_token: bool, training_projection: str,sliding_window:bool,warmup_ratio:float, grad_accumulation: int,neft_noise_alpha:float):
-
-    global train_log_graph
-    global WANT_INTERRUPT
-    WANT_INTERRUPT = False
-
-    statistics['loss'] = []
-
-    statistics['loss'].append({'epoch': 0, 'value': 0})
-    zero_pd = pd.DataFrame(statistics['loss'])
-
-    # == Input validation / processing ==
-    yield "Preparing the input...", zero_pd
-    lora_file_path = clean_path(None, lora_name)
-    if lora_file_path.strip() == '':
-        yield "Missing or invalid LoRA file name input.", zero_pd
-        return
-
-    lora_file_path = f"{Path(shared.args.lora_dir)}/{lora_file_path}"
-    actual_lr = float(learning_rate)
-    model_type = type(shared.model).__name__
-
-    if model_type in MODEL_CLASSES:
-        model_id = MODEL_CLASSES[model_type]
-    else:
-        model_id = "llama"
-        if model_type == "PeftModelForCausalLM":
-            if len(shared.lora_names) > 0:
-                yield "You are trying to train a LoRA while you already have another LoRA loaded. This will work, but may have unexpected effects. *(Will continue anyway in 5 seconds, press `Interrupt` to stop.)*", zero_pd
-                logger.warning("Training LoRA over top of another LoRA. May have unexpected effects.")
-            else:
-                yield "Model ID not matched due to LoRA loading. Consider reloading base model. *(Will continue anyway in 5 seconds, press `Interrupt` to stop.)*", zero_pd
-                logger.warning("Model ID not matched due to LoRA loading. Consider reloading base model.")
-        else:
-            yield "LoRA training has only currently been validated for LLaMA, OPT, GPT-J, and GPT-NeoX models. Unexpected errors may follow. *(Will continue anyway in 5 seconds, press `Interrupt` to stop.)*", zero_pd
-            logger.warning(f"LoRA training has only currently been validated for LLaMA, OPT, GPT-J, and GPT-NeoX models. (Found model type: {model_type})")
-
-        time.sleep(5)
-
-    if cutoff_len <= 0 or micro_batch_size <= 0 or actual_lr <= 0 or lora_rank <= 0 or lora_alpha <= 0:
-        yield "Cannot input zeroes.", zero_pd
-        return
-
-    #in new version we dumped this in favor of grad_accumulation
-    #set it to zero fo new save
-    batch_size = 0
-
-    gradient_accumulation_steps = grad_accumulation #batch_size // micro_batch_size
-    shared.tokenizer.pad_token_id = 0
-    shared.tokenizer.padding_side = "left"
-
-    def encode(text, prepend_bos_token):
-       
-        result = shared.tokenizer.encode(text, truncation=True, max_length=cutoff_len)
-        # Check if the first two tokens are BOS
-        if len(result) >= 2 and result[:2] == [shared.tokenizer.bos_token_id, shared.tokenizer.bos_token_id]:
-            result = result[1:]
-
-        if not prepend_bos_token and result[0] == shared.tokenizer.bos_token_id:
-            result = result[1:]
-        return result
-
-    def tokenize(prompt, append_eos_token=False, prepend_bos_token = False):
-
-        if train_only_after == '' or train_only_after not in prompt:
-            input_ids = encode(prompt, prepend_bos_token)
-
-            if append_eos_token and input_ids[-1] != shared.tokenizer.eos_token_id and len(input_ids) < cutoff_len:
-                input_ids.append(shared.tokenizer.eos_token_id)
-
-            input_ids = [shared.tokenizer.pad_token_id] * (cutoff_len - len(input_ids)) + input_ids
-            
-            labels = [1] * len(input_ids)
-        else:
-            ind = prompt.index(train_only_after) + len(train_only_after)
-            before_tokens = encode(prompt[:ind], prepend_bos_token)
-            after_tokens = encode(prompt[ind:], False)
-
-            if append_eos_token and after_tokens[-1] != shared.tokenizer.eos_token_id:
-                after_tokens.append(shared.tokenizer.eos_token_id)
-
-            full_length = len(after_tokens) + len(before_tokens)
-            if full_length > cutoff_len:
-                after_tokens = after_tokens[:cutoff_len - len(before_tokens)]
-            else:
-                before_tokens = [shared.tokenizer.pad_token_id] * (cutoff_len - full_length) + before_tokens
-
-            input_ids = before_tokens + after_tokens
-            labels = [-100] * len(before_tokens) + [1] * len(after_tokens)
-
-        input_ids = torch.tensor(input_ids)
-        return {
-            "input_ids": input_ids,
-            "labels": labels,
-            "attention_mask": input_ids.ne(shared.tokenizer.pad_token_id),
-        }
-
-    train_template.clear()
-            
-    #reset stuff
-    print(f"*** LoRA: {lora_name} ***")
-    non_serialized_params.update({"stop_at_loss": stop_at_loss})
-    non_serialized_params.update({"save_steps_under_loss": save_steps_under_loss+0.01})
-    non_serialized_params.update({"save_checkpoint_now": False})
-    non_serialized_params.update({"training_loop": False})
-    non_serialized_params.update({"current_stability": 0})
-    non_serialized_params.update({"save_epochs": 0})
-    non_serialized_params.update({"checkpoint_offset": 0})
-    non_serialized_params.update({"epoch_offset": 0})
-    train_log_graph.clear()
-  
-     # == Prep the dataset, format, etc ==
-    if raw_text_file not in ['None', '']:
-        train_template["template_type"] = "raw_text"
-        logger.info("Loading text file...")
-        fullpath = clean_path('user_data/training/datasets', f'{raw_text_file}')
-        fullpath = Path(fullpath)
-        if fullpath.is_dir():
-            logger.info('Training path directory {}'.format(raw_text_file))
-            raw_text = ""
-            file_paths = sorted(fullpath.glob('*.txt'), key=lambda path: natural_keys(path.name))
-            for file_path in file_paths:
-                if file_path.is_file():
-                    with file_path.open('r', encoding='utf-8') as file:
-                        raw_text += file.read().replace('\r', '')
-
-                    logger.info(f"Loaded training file: {file_path.name}")
-        else:
-            with open(clean_path('user_data/training/datasets', f'{raw_text_file}.txt'), 'r', encoding='utf-8') as file:
-                raw_text = file.read().replace('\r', '')
-        
-        # FPHAM PRECISE SLICING        
-        if min_chars<0:
-            min_chars = 0
-
-        add_EOS_to_all = add_eos_token and add_eos_token_type == 'Every Block'
-        add_EOS_to_HC = add_eos_token and add_eos_token_type != 'Every Block'
-
-        #print (f"add_eos_token {add_eos_token}, add_EOS_to_all {add_EOS_to_all}, add_EOS_to_HC {add_EOS_to_HC}")
-
-        # == New more precise slicing on sentence boundary ==
-        if sliding_window:
-            text_chunks = sliding_block_cut(raw_text, min_chars, add_EOS_to_HC, cutoff_len, hard_cut_string,non_serialized_params['debug_slicer'])
-        else:
-            text_chunks = precise_cut(raw_text, precize_slicing_overlap, min_chars, add_EOS_to_HC, cutoff_len, hard_cut_string,non_serialized_params['debug_slicer'])
-
-        train_data = Dataset.from_list([tokenize(x, add_EOS_to_all, add_bos_token) for x in text_chunks])
-        if add_EOS_to_all:
-            print(f"Added EOS to {len(text_chunks)} blocks") 
-
-        print(f"All Data Blocks: {len(text_chunks)}")
-
-        del text_chunks
-        eval_data = None
-    else:
-        if dataset in ['None', '']:
-            yield "Missing dataset choice input, cannot continue.", zero_pd
-            return
-
-        if format in ['None', '']:
-            yield "Missing format choice input, cannot continue.", zero_pd
-            return
-
-        train_template["template_type"] = "dataset"
-
-        with open(clean_path('user_data/training/formats', f'{format}.json'), 'r', encoding='utf-8-sig') as formatFile:
-            format_data: dict[str, str] = json.load(formatFile)
-
-        # == store training prompt ==
-        for _, value in format_data.items():
-            prompt_key = f"template_{len(train_template)}"
-            train_template[prompt_key] = value
-
-        def generate_prompt(data_point: dict[str, str]):
-            for options, data in format_data.items():
-                if set(options.split(',')) == set(x[0] for x in data_point.items() if (type(x[1]) is str and len(x[1].strip()) > 0)):
-                    for key, val in data_point.items():
-                        if type(val) is str:
-                            data = data.replace(f'%{key}%', val)
-                    return data
-            raise RuntimeError(f'Data-point "{data_point}" has no keyset match within format "{list(format_data.keys())}"')
-
-        def generate_and_tokenize_prompt(data_point):
-            prompt = generate_prompt(data_point)
-            return tokenize(prompt, add_eos_token, add_bos_token)
-
-        logger.info("Loading JSON datasets...")
-        data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
-        train_data = data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
-
-        print(f"BOS: {add_bos_token} EOS: {add_eos_token}") 
-        print(f"Data Blocks: {train_data.num_rows}")
-
-        if eval_dataset == 'None':
-            eval_data = None
-        else:
-            eval_data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{eval_dataset}.json'))
-            eval_data = eval_data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
-
-    # == We MUST reload model if it went through any previous training, even failed one ==
-    if shared.model_dirty_from_training:
-        selected_model = shared.model_name
-        if selected_model:
-            print("\033[1;31;1m(Model has been modified by previous training, it needs to be reloaded...)\033[0;37;0m")
-            try:
-                yield f"Reloading {selected_model}...", zero_pd
-                reload_model()
-                shared.tokenizer.pad_token_id = 0
-                shared.tokenizer.padding_side = "left"
-
-                if shared.model is not None:
-                    print("Model reloaded OK, continue with training.")
-                else:
-                    return f"Failed to load {selected_model}."
-            except:
-                exc = traceback.format_exc()
-                logger.error('Failed to reload the model.')
-                print(exc)
-                return exc.replace('\n', '\n\n')
-
-    # == Start prepping the model itself ==
-    if not hasattr(shared.model, 'lm_head') or hasattr(shared.model.lm_head, 'weight'):
-        logger.info("Getting model ready...")
-        # here we can disable gradient checkpoint, by default = true,  use_gradient_checkpointing=True
-        if 'quantization_config' in shared.model.config.to_dict():
-            print(f"Method: {RED}QLORA{RESET}")
-            prepare_model_for_kbit_training(shared.model)
-        else:
-            print(f"Method: {RED}LoRA{RESET}")    
-
-    # base model is now frozen and should not be reused for any other LoRA training than this one
-    shared.model_dirty_from_training = True
-    print(f"Transformers Model Type: {YELLOW}{model_type}{RESET}")
-
-    if training_projection==train_choices[0]:
-        model_to_lora_modules[model_id] = ["gate_proj","down_proj","up_proj","q_proj","k_proj","v_proj","o_proj"]
-    elif training_projection==train_choices[1]:
-        model_to_lora_modules[model_id] = ["q_proj","k_proj", "v_proj", "o_proj"]
-    elif training_projection==train_choices[2]:
-        model_to_lora_modules[model_id] = ["q_proj","k_proj", "v_proj"]
-    elif training_projection==train_choices[3]:
-        model_to_lora_modules[model_id] = ["k_proj", "v_proj", "down_proj"]        
-    else:
-        model_to_lora_modules[model_id] = ["q_proj", "v_proj"]
-
-
-    logger.info("Preparing for training...")
-    config = LoraConfig(
-        r=lora_rank,
-        lora_alpha=lora_alpha,
-        target_modules=model_to_lora_modules[model_id],
-        lora_dropout=lora_dropout,
-        bias="none",
-        task_type="CAUSAL_LM"
-    )
-
-    # == Backup the existing adapter ==
-    if not always_override:
-        backup_adapter(lora_file_path)
-
-    # == get model trainable params
-    model_trainable_params, model_all_params = calc_trainable_parameters(shared.model)
-
-    try:
-        logger.info("Creating LoRA model...")
-        lora_model = get_peft_model(shared.model, config)
-        if not always_override and Path(f"{lora_file_path}/adapter_model.bin").is_file():
-            logger.info("Loading existing LoRA data...")
-            state_dict_peft = torch.load(f"{lora_file_path}/adapter_model.bin", weights_only=True)
-            set_peft_model_state_dict(lora_model, state_dict_peft)
-
-            print(f" + Continue Training on {RED}{lora_file_path}/adapter_model.bin{RESET}")
-            
-            #load training_log.json if exist
-           
-            if Path(f"{lora_file_path}/training_log.json").is_file():
-                with open(f"{lora_file_path}/training_log.json", 'r') as json_file:
-                    json_ilog = json.load(json_file)
-                    for key, value in json_ilog.items():
-                        if key=='current_steps':
-                            non_serialized_params.update({"checkpoint_offset": int(value+1)})
-                            print(f" + Checkpoints will be saved with offset: {RED}{non_serialized_params['checkpoint_offset']}{RESET}")
-                        if key=='epoch':
-                            non_serialized_params.update({"epoch_offset": value})
-                            print(f" + Epoch offset: {RED}{non_serialized_params['epoch_offset']}{RESET}")
-           
-
-            if Path(f"{lora_file_path}/training_graph.json").is_file():
-                try:
-                    with open(f"{lora_file_path}/training_graph.json", 'r') as json_file:
-                        train_log_graph = json.load(json_file)
-                        print(" + Training Graph loaded")   
-                except:
-                    print(f"Can't read training_graph")
-
-
-    except:
-        yield traceback.format_exc().replace('\n', '\n\n'), zero_pd
-        return
-
-    class Tracked():
-        def __init__(self):
-            self.current_steps = 0
-            self.max_steps = 0
-            self.did_save = False
-
-    tracked = Tracked()
-    actual_save_steps = math.ceil(save_steps / gradient_accumulation_steps)
-
-    class Callbacks(transformers.TrainerCallback):
-        def on_step_begin(self, args: transformers.TrainingArguments, state: transformers.TrainerState, control: transformers.TrainerControl, **kwargs):
-            tracked.current_steps = state.global_step * gradient_accumulation_steps
-            tracked.max_steps = state.max_steps * gradient_accumulation_steps
-            ssteps10 = int(max(2,(state.max_steps/epochs)*0.1))
-
-            if WANT_INTERRUPT:
-                control.should_epoch_stop = True
-                control.should_training_stop = True
-            else:
-                current_loss = float(train_log.get('loss', 0.0))
-                current_epoch_int = int(float(train_log.get('epoch', 0.0)))
-              
-                force_save = False
-
-                current_steps_offset = tracked.current_steps + non_serialized_params['checkpoint_offset']
-
-                folder_save = f"checkpoint-{current_steps_offset}"    
-
-                # save if triggered by user
-                if non_serialized_params['save_checkpoint_now']:
-                    force_save = True
-                    non_serialized_params.update({"save_checkpoint_now": False})
-                    print(f"\033[1;31;1mSave Checkpoint manually trigerred.\033[0;37;0m")
-                    folder_save = f"checkpoint-{current_steps_offset}-user"  
-
-                patience = 3     # Set the number of consecutive steps for tracking stability
-                
-                if gradient_accumulation_steps==1:
-                    patience = 4
-
-                min_steps = ssteps10
-
-                # Save each time the loss is below the threshold 
-                if current_loss < non_serialized_params['save_steps_under_loss'] and current_loss > 0 and state.global_step > min_steps:
-                    current_stability = non_serialized_params['current_stability']
-                    current_stability += 1
-                    non_serialized_params.update({"current_stability": current_stability}) 
-
-                    if current_stability >= patience:
-                        current_stability = 0
-                        non_serialized_params.update({"current_stability": current_stability})     
-                        current_loss_dec = round(current_loss, 2)
-                        loss_str = f"{current_loss_dec:.2f}"
-                        loss_str = loss_str.replace('.', '_')
-                        new_save = (current_loss_dec-0.1) + 0.01
-                        non_serialized_params.update({"save_steps_under_loss": new_save})
-
-                        folder_save = f"checkpoint-{current_steps_offset}-loss-{loss_str}" 
-                        force_save = True   
-
-                   
-                else:
-                    # Reset stability if the loss goes above the threshold
-                    non_serialized_params.update({"current_stability": 0})   
-
-                # Save full epochs
-                if actual_save_steps>0 and current_epoch_int > non_serialized_params['save_epochs'] and state.global_step > min_steps: 
-
-                    
-                    current_epoch_offset = current_epoch_int
-                    
-                    if non_serialized_params['epoch_offset'] > 0:
-                        current_epoch_offset = current_epoch_int + round(non_serialized_params['epoch_offset'], 2)
-                    
-                    ep_off_str = f"{current_epoch_offset}"
-                    ep_off_str = ep_off_str.replace('.', '_')
-                    folder_save = f"checkpoint-{current_steps_offset}-epoch-{ep_off_str}" 
-
-                    non_serialized_params.update({"save_epochs": current_epoch_int})
-                    force_save = True
-
-                # save each actual_save_steps
-                if state.global_step > 0 and actual_save_steps > 0 and state.global_step % actual_save_steps == 0:
-                    folder_save = f"checkpoint-{current_steps_offset}"  
-                    force_save = True   
-
-                if force_save:       
-                    lora_model.save_pretrained(f"{lora_file_path}/{folder_save}/", safe_serialization = non_serialized_params['safe_serialization'])
-                    print(f"\033[1;30;40mStep: {tracked.current_steps:6} \033[0;37;0m Saved: [{folder_save}]")
-                    # Save log
-                    with open(f"{lora_file_path}/{folder_save}/training_log.json", 'w', encoding='utf-8') as file:
-                        json.dump(train_log, file, indent=2)
-                    # == Save training prompt ==
-                    with open(f"{lora_file_path}/{folder_save}/training_prompt.json", 'w', encoding='utf-8') as file:
-                        json.dump(train_template, file, indent=2)
-                
-
-        def on_substep_end(self, args: transformers.TrainingArguments, state: transformers.TrainerState, control: transformers.TrainerControl, **kwargs):
-            tracked.current_steps += 1
-            if WANT_INTERRUPT:
-                control.should_epoch_stop = True
-                control.should_training_stop = True
-
-        def on_log(self, args: transformers.TrainingArguments, state: transformers.TrainerState, control: transformers.TrainerControl, logs, **kwargs):
-            train_log.update(logs)
-
-            current_steps_offset = tracked.current_steps + non_serialized_params['checkpoint_offset']
-            current_epoch_offset = train_log.get('epoch', 0.0) + non_serialized_params['epoch_offset']
-
-            train_log.update({"current_steps": tracked.current_steps})
-            train_log.update({"current_steps_adjusted": current_steps_offset})
-            train_log.update({"epoch_adjusted": current_epoch_offset})
-
-            if WANT_INTERRUPT:
-                print("\033[1;31;1mInterrupted by user\033[0;37;0m")
-
-            if non_serialized_params['checkpoint_offset']>0:
-                print(f"\033[1;30;40mStep: {tracked.current_steps:6} [+{non_serialized_params['checkpoint_offset']}] \033[0;37;0m", end='')
-            else:
-                print(f"\033[1;30;40mStep: {tracked.current_steps:6} \033[0;37;0m", end='')
-            
-            graphentry = {
-                'current_steps': int(train_log.get('current_steps_adjusted',0)),
-                'loss': float(train_log.get('loss', 0.0)),
-                'learning_rate': float(train_log.get('learning_rate', 0.0)),
-                'epoch': float(train_log.get('epoch_adjusted', 0.0))
-            }
-
-            cur_loss = float(train_log.get('loss', 0.0))
-            cur_lr = float(train_log.get('learning_rate', 0.0))
-            cur_epoch = float(train_log.get('epoch', 0.0))
-            
-            if len(statistics['loss']) == 1:
-                first_epoch = statistics['loss'][0]['epoch']
-                first_value = statistics['loss'][0]['value']
-                if first_value ==0:
-                     statistics['loss'] = []
-
-
-            statistics['loss'].append({'epoch': cur_epoch, 'value': cur_loss})
-            statistics['lr'].append({'epoch': cur_epoch, 'value': cur_lr})
-
-            # Add the entry to the continuous log
-            train_log_graph.append(graphentry)
-
-            # Save the graph log for now, we can later generate full graph
-            with open(f"{lora_file_path}/training_graph.json", 'w') as file:
-                json.dump(train_log_graph, file, indent=4)
-
-            if 'loss' in logs:
-                loss = float(logs['loss'])
-                if loss <= stop_at_loss:
-                    control.should_epoch_stop = True
-                    control.should_training_stop = True
-                    print(f"{RED}Stop Loss {stop_at_loss} reached.{RESET}")
-
-    # FPHAM SAMPLE REQ Transformers error handling
-    gradient_accumulation_max = int(train_data.num_rows)//micro_batch_size
-    
-    if gradient_accumulation_max < gradient_accumulation_steps:
-        print(f"{RED}WARNING:{RESET} Current gradient accumulation is {RED}too high{RESET} for the amount of training data.")
-        print(f"Gradient accumulation: {gradient_accumulation_steps} should be less than: {gradient_accumulation_max}. {RED}This could crash Accelerate/Transformers{RESET}")
-        #min_batchSize = sample_req*micro_batch_size
-        print(f"Preferable fix: {RED}Increase the size of dataset{RESET}")
-        print(f"... or Decrerase Gradient Accumulation {RED}{gradient_accumulation_steps}{RESET} to below {GREEN}{gradient_accumulation_max}{RESET}")
-        gradient_accumulation_steps = max(1,gradient_accumulation_max-1)
-        print(f"Last resort fix for this run: Lowering Gradient accumulation to {GREEN}{gradient_accumulation_steps}{RESET} [Good luck]")
-
-    else:
-        print(f"Data Size Check: Gradient accumulation: {YELLOW}{gradient_accumulation_steps}{RESET} <= Blocks/Batch {gradient_accumulation_max} ... {GREEN}[OK]{RESET}")
-
-    #END OF FPHAM SAMPLE REQ
-
-    # FPHAM Custom Scheduler ==
-    custom_scheduller = False
-    lr_scheduler_type_arg = lr_scheduler_type
-
-    if lr_scheduler_type == 'FP_low_epoch_annealing':
-        custom_scheduller = True
-        lr_scheduler_type_arg = 'cosine'
-    elif lr_scheduler_type == 'FP_half_time_annealing':
-        custom_scheduller = True
-        lr_scheduler_type_arg = 'constant'
-    elif lr_scheduler_type =='FP_raise_fall_creative':
-        custom_scheduller = True
-        lr_scheduler_type_arg = 'constant_with_warmup'
-    
-    #gradient_checkpointing=True
-    
-    args=transformers.TrainingArguments(
-            report_to=report_to if report_to != "None" else None,
-            per_device_train_batch_size=micro_batch_size,
-            gradient_accumulation_steps=gradient_accumulation_steps,
-            warmup_steps=math.ceil(warmup_steps / gradient_accumulation_steps),
-            warmup_ratio = warmup_ratio,
-            num_train_epochs=epochs,
-            learning_rate=actual_lr,
-            fp16=False if shared.args.cpu else True,
-            optim=optimizer,
-            logging_steps=1,
-            evaluation_strategy="steps" if eval_data is not None else "no",
-            eval_steps=math.ceil(eval_steps / gradient_accumulation_steps) if eval_data is not None else None,
-            save_strategy="steps" if eval_data is not None else "no",
-            output_dir=lora_file_path,
-            lr_scheduler_type=lr_scheduler_type_arg,
-            load_best_model_at_end=eval_data is not None,
-            # TODO: Enable multi-device support
-            ddp_find_unused_parameters=None,
-            no_cuda=shared.args.cpu,
-        )
-
-    if custom_scheduller:
-        trainer = FPSchedulerTrainer(
-            neftune_noise_alpha=neft_noise_alpha,
-            model=lora_model,
-            train_dataset=train_data,
-            eval_dataset=eval_data,
-            args=args,
-            data_collator=transformers.DataCollatorForLanguageModeling(shared.tokenizer, mlm=False),
-            callbacks=list([Callbacks()])
-        )
-    elif neft_noise_alpha > 0:
-            trainer = FPNEFtuneTrainer(
-            neftune_noise_alpha=neft_noise_alpha,
-            model=lora_model,
-            train_dataset=train_data,
-            eval_dataset=eval_data,
-            args=args,
-            data_collator=transformers.DataCollatorForLanguageModeling(shared.tokenizer, mlm=False),
-            callbacks=list([Callbacks()])
-        )
-    else:
-        trainer = transformers.Trainer(
-            model=lora_model,
-            train_dataset=train_data,
-            eval_dataset=eval_data,
-            args=args,
-            data_collator=transformers.DataCollatorForLanguageModeling(shared.tokenizer, mlm=False),
-            callbacks=list([Callbacks()])
-        )
-    
-    # END OF FPHAM CUSTOM SCHEDULER
-
-    lora_model.config.use_cache = False
-
-    if torch.__version__ >= "2" and sys.platform != "win32":
-        lora_model = torch.compile(lora_model)
-
-    # == Save parameters for reuse ==
-    with open(f"{lora_file_path}/training_parameters.json", 'w', encoding='utf-8') as file:
-        vars = locals()
-        json.dump({x: vars[x] for x in PARAMETERS}, file, indent=2)
-
-    # == Save training prompt ==
-    with open(f"{lora_file_path}/training_prompt.json", 'w', encoding='utf-8') as file:
-        json.dump(train_template, file, indent=2)
-
-    # == Main run and monitor loop ==
-    logger.info("Starting training...")
-    yield "Starting...", zero_pd
-
-    lora_trainable_param, lora_all_param = calc_trainable_parameters(lora_model)
-
-    projections_string = ", ".join([projection.replace("_proj", "") for projection in model_to_lora_modules[model_id]])
-
-    print(f"Training '{model_id}' model using {YELLOW}({projections_string}){RESET} projections")
-
-    if lora_all_param > 0:
-        print(f"Trainable params: {lora_trainable_param:,d} ({RED}{100 * lora_trainable_param / lora_all_param:.4f} %{RESET}), All params: {lora_all_param:,d} (Model: {model_all_params:,d})")
-
-    train_log.update({"base_model_name": shared.model_name})
-    train_log.update({"base_model_class": shared.model.__class__.__name__})
-    train_log.update({"base_loaded_in_4bit": getattr(lora_model, "is_loaded_in_4bit", False)})
-    train_log.update({"base_loaded_in_8bit": getattr(lora_model, "is_loaded_in_8bit", False)})
-    train_log.update({"projections": projections_string})
-    if non_serialized_params['checkpoint_offset'] > 0:
-        train_log.update({"last_run_steps_offset": non_serialized_params['checkpoint_offset']})
-        train_log.update({"last_run_epoch_offset": non_serialized_params['epoch_offset']})
-
-
-    if non_serialized_params['checkpoint_offset'] > 0:
-        print(f"Continue training on {RED}previous adapter{RESET} from epoch: {RED}{non_serialized_params['epoch_offset']}{RESET}")
-
-    if stop_at_loss > 0:
-        print(f"Monitoring loss {RED}(Auto-Stop at: {stop_at_loss}){RESET}")
-
-    
-
-    if WANT_INTERRUPT:
-        yield "Interrupted before start.", zero_pd
-        return
-
-    def log_train_dataset(trainer):
-        decoded_entries = []
-        # Try to decode the entries and write the log file
-        try:
-            # Iterate over the first 10 elements in the dataset (or fewer if there are less than 10)
-            for i in range(min(10, len(trainer.train_dataset))):
-                decoded_text = shared.tokenizer.decode(trainer.train_dataset[i]['input_ids'])
-                decoded_entries.append({"value": decoded_text})
-
-            # Write the log file
-            Path('user_data/logs').mkdir(exist_ok=True)
-            with open(Path('user_data/logs/train_dataset_sample.json'), 'w') as json_file:
-                json.dump(decoded_entries, json_file, indent=4)
-
-            logger.info("Log file 'train_dataset_sample.json' created in the 'user_data/logs' directory.")
-        except Exception as e:
-            logger.error(f"Failed to create log file due to error: {e}")
-
-    def threaded_run():
-        log_train_dataset(trainer)
-        trainer.train()
-        # Note: save in the thread in case the gradio thread breaks (eg browser closed)
-        lora_model.save_pretrained(lora_file_path, safe_serialization = non_serialized_params['safe_serialization'])
-        logger.info("LoRA training run is completed and saved.")
-        # Save log
-        with open(f"{lora_file_path}/training_log.json", 'w', encoding='utf-8') as file:
-            json.dump(train_log, file, indent=2)
-
-    thread = threading.Thread(target=threaded_run)
-    thread.start()
-    last_step = 0
-    start_time = time.perf_counter()
-
-    while thread.is_alive():
-        time.sleep(0.5)
-
-        if statistics['loss']:
-            max_value_dict = max(statistics['loss'], key=lambda x: x['value'])
-            max_value = max_value_dict['value']+0.4
-            first_epoch = statistics['loss'][0]['epoch']
-            last_epoch = statistics['loss'][-1]['epoch']
-        else:
-            max_value = 3.5
-            last_epoch = 0
-            first_epoch = 0           
-
-        if WANT_INTERRUPT:
-
-            losses = gr.LinePlot.update(
-				value = pd.DataFrame(statistics['loss']),
-                x="epoch", y="value",
-                title="Loss Metrics",
-                overlay_point=True, tooltip=["epoch", "value"],
-				x_lim=[first_epoch,last_epoch], y_lim=[0,max_value],
-                width=500, height=250 )
-
-            yield "Interrupting, please wait... *(Run will stop after the current training step completes.)*", losses
-
-        elif tracked.current_steps != last_step:
-            last_step = tracked.current_steps
-            time_elapsed = time.perf_counter() - start_time
-            lastloss = float(train_log.get('loss', 0.0))
-
-            non_serialized_params.update({"training_loop": True})               
-
-            if lastloss > 0:
-                lastloss_str = f", ... Current Loss: `{lastloss:.2f}`"
-            else:
-                lastloss_str = ""
-
-            if time_elapsed <= 0:
-                timer_info = ""
-                total_time_estimate = 999
-            else:
-                its = tracked.current_steps / time_elapsed
-                if its > 1:
-                    timer_info = f"`{its:.2f}` it/s"
-                else:
-                    timer_info = f"`{1.0/its:.2f}` s/it"
-
-                total_time_estimate = (1.0 / its) * (tracked.max_steps)
-
-            if stop_at_loss != non_serialized_params['stop_at_loss']:
-                stop_at_loss = non_serialized_params['stop_at_loss']
-                print(f"Stop at loss changed {RED}(Auto-Stop at: {stop_at_loss}){RESET}")
-            
-            losses = gr.LinePlot.update(
-				value = pd.DataFrame(statistics['loss']),
-                x="epoch", y="value",
-                title="Loss Metrics",
-                overlay_point=True, tooltip=["epoch", "value"],
-				x_lim=[first_epoch,last_epoch], y_lim=[0,max_value],
-                width=500, height=250 )
-				
-
-            yield f"Running... **{tracked.current_steps}** / **{tracked.max_steps}** ... {timer_info}, {format_time(time_elapsed)} / {format_time(total_time_estimate)} ... {format_time(total_time_estimate - time_elapsed)} remaining {lastloss_str}", losses
-
-    # Saving in the train thread might fail if an error occurs, so save here if so.
-
-    #return_pd = pd.DataFrame(statistics['loss'])
-
-    if statistics['loss']:
-        max_value_dict = max(statistics['loss'], key=lambda x: x['value'])
-        max_value = max_value_dict['value']+0.4
-        first_epoch = statistics['loss'][0]['epoch']
-        last_epoch = statistics['loss'][-1]['epoch']
-    else:
-        max_value = 3.5
-        last_epoch = 0
-        first_epoch = 0 
-
-    return_pd = gr.LinePlot.update(
-        value = pd.DataFrame(statistics['loss']),
-        x="epoch", y="value",
-        title="Loss Metrics",
-        overlay_point=True, tooltip=["epoch", "value"],
-        x_lim=[first_epoch,last_epoch], y_lim=[0,max_value],
-        width=500, height=250)
-
-    non_serialized_params.update({"training_loop": False})
-
-    if not tracked.did_save:
-        logger.info("Training complete, saving...")
-        lora_model.save_pretrained(lora_file_path, safe_serialization = non_serialized_params['safe_serialization'])
-
-    if WANT_INTERRUPT:
-        logger.info("Training interrupted.")
-        yield f"Interrupted by user. LoRA saved to `{lora_file_path}`.", return_pd
-    else:
-        logger.info("Training complete!")
-        yield f"Done! LoRA saved to `{lora_file_path}`.\n\nBefore testing your new LoRA, make sure to first reload the model, as it is currently dirty from training.", return_pd
-
-    create_graph(lora_file_path, lora_name)
-
-def format_time(seconds: float):
-    if seconds < 120:
-        return f"`{seconds:.0f}` seconds"
-
-    minutes = seconds / 60
-    if minutes < 120:
-        return f"`{minutes:.0f}` minutes"
-
-    hours = minutes / 60
-    return f"`{hours:.0f}` hours"
diff --git a/extensions/Training_PRO/train_utils.py b/extensions/Training_PRO/train_utils.py
deleted file mode 100644
index 799948802b..0000000000
--- a/extensions/Training_PRO/train_utils.py
+++ /dev/null
@@ -1,368 +0,0 @@
-import os
-from modules import shared, utils
-from pathlib import Path
-import requests
-import tqdm
-import json
-
-'''
-def get_gpu_memory_usage(rank):
-    return {
-        'total': round(torch.cuda.get_device_properties(rank).total_memory / (1024**3), 2),
-        'max': round(torch.cuda.max_memory_allocated(rank) / (1024**3), 2),
-        'reserved': round(torch.cuda.memory_reserved(rank) / (1024**3), 2),
-        'allocated': round(torch.cuda.memory_allocated(rank) / (1024**3), 2)
-    }
-'''
-
-def list_subfoldersByTime(directory):
-
-    if not directory.endswith('/'):
-        directory += '/'
-    subfolders = []
-    subfolders.append('None') 
-    path = directory
-    name_list = os.listdir(path)
-    full_list = [os.path.join(path,i) for i in name_list]
-    time_sorted_list = sorted(full_list, key=os.path.getmtime,reverse=True)
-
-    for entry in time_sorted_list:
-        if os.path.isdir(entry):
-            entry_str = f"{entry}"  # Convert entry to a string
-            full_path = entry_str
-            entry_str = entry_str.replace('\\','/')
-            entry_str = entry_str.replace(f"{directory}", "")  # Remove directory part
-            subfolders.append(entry_str)
-
-    return subfolders
-
-def get_available_loras_local(_sortedByTime):
-    
-    model_dir = shared.args.lora_dir  # Update with the appropriate directory path
-    subfolders = []
-    if _sortedByTime:
-        subfolders = list_subfoldersByTime(model_dir)
-    else:
-        subfolders = utils.get_available_loras()        
-
-    return subfolders
-
-
-# FPHAM SPLIT BY SENTENCE BLOCK ===============
-     
-def split_sentences(text: str, cutoff_len: int):
-    sentences = []
-    sentence = ''
-    delimiters = ['. ', '? ', '! ', '... ', '.\n', '?\n', '!\n','...\n','','']
-    abbreviations = ['Mr. ', 'Mrs. ', 'Dr. ', 'Ms. ', 'St. ', 'Prof. ', 'Jr. ', 'Ltd. ', 'Capt. ', 'Col. ', 'Gen. ', 'Ave. ', 'Blvd. ', 'Co. ', 'Corp. ', 'Dept. ', 'Est. ', 'Gov. ', 'Inc. ', 'Ph.D. ', 'Univ. ']
-    errors = 0
-    max_cut = cutoff_len-1
-    prev_char = ''  
-
-    for char in text:
-        sentence += char
-
-    
-        if (any(sentence.endswith(delimiter) for delimiter in delimiters) and
-            not (prev_char.isupper() and len(sentence) >= 3 and sentence[-3] != ' ') and 
-            not any(sentence.endswith(abbreviation) for abbreviation in abbreviations)):
-            tokens = shared.tokenizer.encode(sentence)
-            
-            if len(tokens) > max_cut:
-                tokens = tokens[:max_cut]
-                sentence = shared.tokenizer.decode(tokens, skip_special_tokens=True)
-                errors = errors + 1
-
-            sentences.append({'text': sentence, 'size': len(tokens)})
-            
-            sentence = ''
-
-        prev_char = char
-
-    if sentence:
-        tokens = shared.tokenizer.encode(sentence)
-        if len(tokens) > max_cut:
-            tokens = tokens[:max_cut]
-            sentence = shared.tokenizer.decode(tokens, skip_special_tokens=True)  
-            errors = errors + 1
-
-        sentences.append({'text': sentence, 'size': len(tokens)})
-
-    if errors > 0:
-        print(f"Trimmed sentences beyond Cutoff Length: {errors}")
-
-    return sentences
-
-# The goal of following code is to create blocks of text + overlapping blocks while:
-# respects sentence boundaries
-# always uses all the text 
-# hard cut defined by hard_cut_string or  will always end at the end of data block
-# no overlapping blocks will be created across hard cut or across  token
-
-def precise_cut(text: str, overlap: bool, min_chars_cut: int, eos_to_hc: bool, cutoff_len: int, hard_cut_string: str, debug_slicer:bool):
-
-    EOSX_str = '' #hardcut placeholder
-    EOS_str = '' 
-    print("Precise raw text slicer: ON")
-    
-    cut_string = hard_cut_string.replace('\\n', '\n')
-    text = text.replace(cut_string, EOSX_str)
-    sentences = split_sentences(text, cutoff_len)
-
-    print(f"Sentences: {len(sentences)}")
-    sentencelist = []
-    currentSentence = ''
-    totalLength = 0
-    max_cut = cutoff_len-1
-    half_cut = cutoff_len//2
-    halfcut_length = 0
-
-    edgeindex = []
-    half_index = 0
-
-    for index, item in enumerate(sentences):
-        
-        if halfcut_length+ item['size'] < half_cut:
-            halfcut_length += item['size']
-            half_index = index
-        else:
-            edgeindex.append(half_index)
-            halfcut_length = -2 * max_cut
-
-
-        if totalLength + item['size'] < max_cut and not currentSentence.endswith(EOSX_str): 
-            currentSentence += item['text']
-            totalLength += item['size']
-        else:
-
-            if len(currentSentence.strip()) > min_chars_cut:
-                sentencelist.append(currentSentence.strip())
-
-            currentSentence = item['text']
-            totalLength = item['size']
-            halfcut_length = item['size']
-            
-    if len(currentSentence.strip()) > min_chars_cut:    
-        sentencelist.append(currentSentence.strip())
-
-    unique_blocks = len(sentencelist)
-    print(f"Text Blocks: {unique_blocks}")
-
-    #overlap strategies: 
-    # don't overlap across HARD CUT (EOSX)
-    if overlap:
-        for edge_idx in edgeindex:
-            currentSentence = ''
-            totalLength = 0
-
-            for item in sentences[edge_idx:]:
-                if totalLength + item['size'] < max_cut:
-                    currentSentence += item['text']
-                    totalLength += item['size']
-                else:
-                    #if by chance EOSX is at the end then it's acceptable
-                    if currentSentence.endswith(EOSX_str) and len(currentSentence.strip()) > min_chars_cut:
-                            sentencelist.append(currentSentence.strip())    
-                    # otherwise don't cross hard cut    
-                    elif EOSX_str not in currentSentence and len(currentSentence.strip()) > min_chars_cut:
-                        sentencelist.append(currentSentence.strip())
-                    
-                    currentSentence = ''
-                    totalLength = 0
-                    break
-        
-        print(f"+ Overlapping blocks: {len(sentencelist)-unique_blocks}")
-
-    num_EOS = 0
-    for i in range(len(sentencelist)):
-        if eos_to_hc:
-            sentencelist[i] = sentencelist[i].replace(EOSX_str, EOS_str)
-        else:
-            sentencelist[i] = sentencelist[i].replace(EOSX_str, '')
-        
-        #someone may have had stop strings in the raw text...
-        sentencelist[i] = sentencelist[i].replace("", EOS_str)
-        num_EOS += sentencelist[i].count(EOS_str)
-
-    if num_EOS > 0:
-        print(f"+ EOS count: {num_EOS}")
-
-    #final check for useless lines
-    sentencelist = [item for item in sentencelist if item.strip() != ""]
-    sentencelist = [item for item in sentencelist if item.strip() != ""]
-
-
-    if debug_slicer:
-                    # Write the log file
-        Path('user_data/logs').mkdir(exist_ok=True)
-        sentencelist_dict = {index: sentence for index, sentence in enumerate(sentencelist)}
-        output_file = "user_data/logs/sentencelist.json"
-        with open(output_file, 'w') as f:
-            json.dump(sentencelist_dict, f,indent=2)
-        
-        print("Saved sentencelist.json in user_data/logs folder")
-    
-    return sentencelist   
-
-
-def sliding_block_cut(text: str, min_chars_cut: int, eos_to_hc: bool, cutoff_len: int, hard_cut_string: str, debug_slicer:bool):
-
-    EOSX_str = '' #hardcut placeholder
-    EOS_str = '' 
-    print("Mega Block Overlap: ON")
-    
-    cut_string = hard_cut_string.replace('\\n', '\n')
-    text = text.replace(cut_string, EOSX_str)
-    sentences = split_sentences(text, cutoff_len)
-
-    print(f"Sentences: {len(sentences)}")
-    sentencelist = []
-    
-    max_cut = cutoff_len-1
-
-    #print(f"max_cut: {max_cut}")
-    advancing_to = 0
-
-    prev_block_lastsentence = ""
-    
-
-    for i in range(len(sentences)):
-        totalLength = 0
-        currentSentence = ''
-        lastsentence = ""
-        
-        if i >= advancing_to:
-            for k in range(i, len(sentences)):
-                
-                current_length = sentences[k]['size']
-
-                if totalLength + current_length <= max_cut and not currentSentence.endswith(EOSX_str):
-                    currentSentence += sentences[k]['text']
-                    totalLength += current_length
-                    lastsentence = sentences[k]['text']
-                else:
-                    if len(currentSentence.strip()) > min_chars_cut:
-                        if prev_block_lastsentence!=lastsentence:
-                            sentencelist.append(currentSentence.strip())
-                            prev_block_lastsentence = lastsentence
-                        
-                    advancing_to = 0
-                    if currentSentence.endswith(EOSX_str):
-                        advancing_to = k
-
-                    currentSentence = ""
-                    totalLength = 0
-                    break
-            
-            if currentSentence != "":
-                if len(currentSentence.strip()) > min_chars_cut:
-                    sentencelist.append(currentSentence.strip())
-
-    unique_blocks = len(sentencelist)
-    print(f"Text Blocks: {unique_blocks}")
-    num_EOS = 0
-    for i in range(len(sentencelist)):
-        if eos_to_hc:
-            sentencelist[i] = sentencelist[i].replace(EOSX_str, EOS_str)
-        else:
-            sentencelist[i] = sentencelist[i].replace(EOSX_str, '')
-        
-        #someone may have had stop strings in the raw text...
-        sentencelist[i] = sentencelist[i].replace("", EOS_str)
-        num_EOS += sentencelist[i].count(EOS_str)
-
-    if num_EOS > 0:
-        print(f"+ EOS count: {num_EOS}")
-
-    #final check for useless lines
-    sentencelist = [item for item in sentencelist if item.strip() != ""]
-    sentencelist = [item for item in sentencelist if item.strip() != ""]
-
-
-    if debug_slicer:
-                    # Write the log file
-        Path('user_data/logs').mkdir(exist_ok=True)
-        sentencelist_dict = {index: sentence for index, sentence in enumerate(sentencelist)}
-        output_file = "user_data/logs/sentencelist.json"
-        with open(output_file, 'w') as f:
-            json.dump(sentencelist_dict, f,indent=2)
-        
-        print("Saved sentencelist.json in user_data/logs folder")
-    
-    return sentencelist   
-
-# Example usage:
-# download_file_from_url('https://example.com/path/to/your/file.ext', '/output/directory')
-
-def download_file_from_url(url, overwrite, output_dir_in, valid_extensions = {'.txt', '.json'}):
-    try:
-    # Validate and sanitize the URL
-    #parsed_url = urllib.parse.urlparse(url)
-    #if not parsed_url.netloc:
-    #    raise ValueError("Invalid URL")
-    #filename = os.path.basename(parsed_url.path)
-
-    # Get the filename from the URL
-
-        session = requests.Session()
-        headers = {}
-        mode = 'wb'
-        filename = url.split('/')[-1]
-
-        output_dir = str(output_dir_in)
-        # Construct the full path to the output file
-        local_filename = os.path.join(output_dir, filename)
-
-        # Check if the local file already exists
-        overw = ''
-        if os.path.exists(local_filename):
-            if not overwrite:
-                yield f"File '{local_filename}' already exists. Aborting."
-                return
-            else:
-                overw = ' [Overwrite existing]'
-
-        filename_lower = filename.lower()
-
-        # Send an HTTP GET request to the URL with a timeout
-        file_extension = os.path.splitext(filename_lower)[-1]
-        
-        if file_extension not in valid_extensions:
-            yield f"Invalid file extension: {file_extension}. Only {valid_extensions} files are supported."
-            return
-
-        with session.get(url, stream=True, headers=headers, timeout=10) as r:
-            r.raise_for_status() 
-            # total size can be wildly inaccurate
-            #total_size = int(r.headers.get('content-length', 0))
-            
-            block_size = 1024 * 4  
-            with open(local_filename, mode) as f:
-                count = 0
-                for data in r.iter_content(block_size):
-                    f.write(data)
-                    count += len(data)
-
-                    yield f"Downloaded: {count} " + overw
-
-            # Verify file size if possible
-            if os.path.exists(local_filename):
-                downloaded_size = os.path.getsize(local_filename)
-                if downloaded_size > 0:
-                    yield f"File '{filename}' downloaded to '{output_dir}' ({downloaded_size} bytes)."
-                    print("File Downloaded")
-                else:
-                    print("Downloaded file is zero")
-                    yield f"Failed. Downloaded file size is zero)."
-            else:
-                print(f"Error: {local_filename} failed to download.")
-                yield f"Error: {local_filename} failed to download"
-
-    except Exception as e:
-        print(f"An error occurred: {e}")
-        yield f"An error occurred: {e}"
-
-    finally:
-        # Close the session to release resources
-        session.close()
-

From d337ba0390059fed02e9739ea09ef101784a2c9f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 13:45:05 -0300
Subject: [PATCH 1220/1701] Training: fix apply_chat_template returning
 BatchEncoding instead of list

---
 modules/training.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/modules/training.py b/modules/training.py
index 292ee48477..c68f3716c2 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -393,7 +393,7 @@ def normalize_messages(data_point):
     def tokenize_conversation(data_point):
         """Tokenize using apply_chat_template() with assistant-only label masking."""
         messages = normalize_messages(data_point)
-        full_ids = shared.tokenizer.apply_chat_template(messages, tokenize=True)
+        full_ids = list(shared.tokenizer.apply_chat_template(messages, tokenize=True, return_dict=False))
 
         # Build labels: -100 for everything, then unmask assistant turns.
         # This assumes apply_chat_template(messages[:i]) is a token-for-token
@@ -404,11 +404,11 @@ def tokenize_conversation(data_point):
             if msg["role"] == "assistant":
                 # Tokens up to where this assistant turn starts
                 header_ids = shared.tokenizer.apply_chat_template(
-                    messages[:i], tokenize=True, add_generation_prompt=True
+                    messages[:i], tokenize=True, return_dict=False, add_generation_prompt=True
                 )
                 # Tokens through end of this assistant turn
                 through_ids = shared.tokenizer.apply_chat_template(
-                    messages[:i + 1], tokenize=True
+                    messages[:i + 1], tokenize=True, return_dict=False
                 )
                 # Unmask assistant tokens
                 start = len(header_ids)

From 5b18be8582e86827f72183177c6fea60e65f4a50 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 14:39:37 -0300
Subject: [PATCH 1221/1701] Training: unify instruction training through
 apply_chat_template()
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Instead of two separate paths (format files vs Chat Template), all
instruction training now uses apply_chat_template() with assistant-only
label masking. Users pick a Jinja2 template from the dropdown or use the
model's built-in chat template — both work identically.
---
 docs/05 - Training Tab.md                     |  91 +++------
 modules/training.py                           | 175 ++++++------------
 .../training/datasets/example_messages.json   |  10 +
 .../training/datasets/example_sharegpt.json   |  10 +
 user_data/training/datasets/example_text.json |  10 +
 .../datasets/put-trainer-datasets-here.txt    |   2 +-
 user_data/training/formats/ChatML-format.json |   4 -
 .../formats/alpaca-chatbot-format.json        |   4 -
 user_data/training/formats/alpaca-format.json |   4 -
 .../training/formats/llama2-chat-format.json  |   4 -
 user_data/training/formats/vicuna-format.json |   3 -
 11 files changed, 119 insertions(+), 198 deletions(-)
 create mode 100644 user_data/training/datasets/example_messages.json
 create mode 100644 user_data/training/datasets/example_sharegpt.json
 create mode 100644 user_data/training/datasets/example_text.json
 delete mode 100644 user_data/training/formats/ChatML-format.json
 delete mode 100644 user_data/training/formats/alpaca-chatbot-format.json
 delete mode 100644 user_data/training/formats/alpaca-format.json
 delete mode 100644 user_data/training/formats/llama2-chat-format.json
 delete mode 100644 user_data/training/formats/vicuna-format.json

diff --git a/docs/05 - Training Tab.md b/docs/05 - Training Tab.md
index ae66ff030e..8ee7091776 100644
--- a/docs/05 - Training Tab.md	
+++ b/docs/05 - Training Tab.md	
@@ -7,8 +7,7 @@ The WebUI seeks to make training your own LoRAs as easy as possible. It comes do
 - What are you training it on? Do you want it to learn real information, a simple format, ...?
 
 ### **Step 2**: Gather a dataset.
-- If you use a dataset similar to the [Alpaca](https://github.com/gururise/AlpacaDataCleaned/blob/main/alpaca_data_cleaned.json) format, that is natively supported by the `Formatted Dataset` input in the WebUI, with premade formatter options.
-- If you use a dataset that isn't matched to Alpaca's format, but uses the same basic JSON structure, you can make your own format file by copying `user_data/training/formats/alpaca-format.json` to a new file and [editing its content](#format-files).
+- For instruction/chat training, prepare a JSON dataset in one of the [supported formats](#instruction-templates) (OpenAI messages or ShareGPT).
 - For pretraining-style training on raw text, use the `Text Dataset` tab with a JSON file where each row has a `"text"` key.
 - If you use a structured dataset not in this format, you may have to find an external way to convert it - or open an issue to request native support.
 
@@ -38,70 +37,24 @@ The WebUI seeks to make training your own LoRAs as easy as possible. It comes do
 - If your model isn't learning detailed information but you want it to, you might need to just run more epochs, or you might need a higher Rank.
 - If your model is enforcing a format you didn't want, you may need to tweak your dataset, or start over and not train as far.
 
-## Format Files
+## Instruction Templates
 
-If using JSON formatted datasets, they are presumed to be in the following approximate format:
+All instruction/chat training uses `apply_chat_template()` with Jinja2 templates. You have two options in the **Data Format** dropdown:
 
-```json
-[
-    {
-        "somekey": "somevalue",
-        "key2": "value2"
-    },
-    {
-        // etc
-    }
-]
-```
-
-Where the keys (eg `somekey`, `key2` above) are standardized, and relatively consistent across the dataset, and the values (eg `somevalue`, `value2`) contain the content actually intended to be trained.
-
-For Alpaca, the keys are `instruction`, `input`, and `output`, wherein `input` is sometimes blank.
-
-A simple format file for Alpaca to be used as a chat bot is:
-
-```json
-{
-    "instruction,output": "User: %instruction%\nAssistant: %output%",
-    "instruction,input,output": "User: %instruction%: %input%\nAssistant: %output%"
-}
-```
-
-Note that the keys (eg `instruction,output`) are a comma-separated list of dataset keys, and the values are a simple string that use those keys with `%%`.
-
-So for example if a dataset has `"instruction": "answer my question"`, then the format file's `User: %instruction%\n` will be automatically filled in as `User: answer my question\n`.
-
-If you have different sets of key inputs, you can make your own format file to match it. This format-file is designed to be as simple as possible to enable easy editing to match your needs.
-
-## Text Dataset
-
-For pretraining-style training on raw text, use the **Text Dataset** tab. Your dataset should be a JSON file with one document per row, each with a `"text"` key:
-
-```json
-[
-  {"text": "First document content..."},
-  {"text": "Second document content..."}
-]
-```
+- **Chat Template**: Uses the model's built-in chat template from its tokenizer. Works with instruct/chat models that ship with a chat template (Llama 3, Qwen, Mistral, etc.).
+- **Named template** (e.g. ChatML, Alpaca, Llama-v3, etc.): Loads a Jinja2 template from `user_data/instruction-templates/`. This is useful for base models that don't have a built-in template, or when you want to override the model's default template.
 
-This is the standard format used by most pretraining datasets (The Pile, RedPajama, etc.).
-
-Each document is tokenized (with BOS token), concatenated into one long token sequence, and split into chunks of `Cutoff Length` tokens. The final chunk is padded if shorter than the cutoff length. When `Add EOS token` is enabled, an EOS token is appended after each document before concatenation, helping the model learn document boundaries.
-
-- `Stride Length` controls the overlap between consecutive chunks in tokens. Set to 0 for non-overlapping chunks (the standard concatenate-and-split approach). Values like 256 or 512 create overlapping chunks that help the model learn context across chunk boundaries, at the cost of more training samples.
-
-## Chat Template Format
-
-Select **Chat Template** as the Data Format to use the model's built-in chat template (via `apply_chat_template()`) instead of a format file. This works with instruct/chat models that ship with a chat template in their tokenizer (Llama 3, Qwen, Mistral, etc.).
-
-**Advantages over format files:**
-- Special tokens are handled correctly by the tokenizer itself
+Both options are functionally identical — the only difference is where the Jinja2 template string comes from. In both cases:
+- The dataset is tokenized via `apply_chat_template()`
+- Labels are automatically masked so only assistant responses are trained on
 - Multi-turn conversations are supported natively
-- Labels are automatically masked so only assistant responses are trained on (no need for `Train Only After`)
+- Special tokens are handled correctly by the template
+
+The WebUI ships with 50+ templates in `user_data/instruction-templates/`. You can also add your own by creating a `.yaml` file with an `instruction_template` key containing a Jinja2 template string, or a plain `.jinja` file.
 
-**Dataset formats:** Your JSON dataset can use any of these structures:
+**Dataset formats:** Your JSON dataset can use either of these structures:
 
-OpenAI messages format (multi-turn):
+OpenAI messages format:
 ```json
 [
   {
@@ -116,8 +69,6 @@ OpenAI messages format (multi-turn):
 ]
 ```
 
-The conversation gets tokenized with the model's own chat template (correct special tokens), and the labels are automatically masked so the model only trains on the assistant responses — the system prompt and user turns get `-100` labels and contribute no gradient.
-
 ShareGPT format (`conversations` key with `from`/`value` fields):
 ```json
 [
@@ -133,11 +84,23 @@ ShareGPT format (`conversations` key with `from`/`value` fields):
 ]
 ```
 
-Simple instruction/output format (auto-converted to a single-turn conversation):
+## Text Dataset
+
+For pretraining-style training on raw text, use the **Text Dataset** tab. Your dataset should be a JSON file with one document per row, each with a `"text"` key:
+
 ```json
-[{"instruction": "What is 2+2?", "output": "4"}]
+[
+  {"text": "First document content..."},
+  {"text": "Second document content..."}
+]
 ```
 
+This is the standard format used by most pretraining datasets (The Pile, RedPajama, etc.).
+
+Each document is tokenized (with BOS token), concatenated into one long token sequence, and split into chunks of `Cutoff Length` tokens. The final chunk is padded if shorter than the cutoff length. When `Add EOS token` is enabled, an EOS token is appended after each document before concatenation, helping the model learn document boundaries.
+
+- `Stride Length` controls the overlap between consecutive chunks in tokens. Set to 0 for non-overlapping chunks (the standard concatenate-and-split approach). Values like 256 or 512 create overlapping chunks that help the model learn context across chunk boundaries, at the cost of more training samples.
+
 ## Target Modules
 
 By default, **Target all linear layers** is enabled. This uses peft's `all-linear` mode, which applies LoRA to every `nn.Linear` layer in the model except the output head (`lm_head`). It works for any model architecture.
diff --git a/modules/training.py b/modules/training.py
index c68f3716c2..6472e692eb 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -14,6 +14,7 @@
 from datetime import datetime
 from pathlib import Path
 
+import yaml
 import gradio as gr
 
 from modules import shared, ui, utils
@@ -25,7 +26,7 @@
 from modules.logging_colors import logger
 from modules.models import reload_model
 
-PARAMETERS = ["lora_name", "always_override", "all_linear", "q_proj_en", "v_proj_en", "k_proj_en", "o_proj_en", "gate_proj_en", "down_proj_en", "up_proj_en", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "text_dataset", "higher_rank_limit", "warmup_steps", "optimizer", "stride_length", "train_only_after", "stop_at_loss", "add_eos_token", "report_to"]
+PARAMETERS = ["lora_name", "always_override", "all_linear", "q_proj_en", "v_proj_en", "k_proj_en", "o_proj_en", "gate_proj_en", "down_proj_en", "up_proj_en", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "text_dataset", "higher_rank_limit", "warmup_steps", "optimizer", "stride_length", "stop_at_loss", "add_eos_token", "report_to"]
 WANT_INTERRUPT = False
 
 train_log = {}
@@ -96,9 +97,8 @@ def create_ui():
 
                             with gr.Column():
                                 warmup_steps = gr.Number(label='Warmup Steps', value=100, info='For this many steps at the start, the learning rate will be lower than normal. This helps the trainer prepare the model and precompute statistics to improve the quality of training after the start.')
-                                train_only_after = gr.Textbox(label='Train Only After', value='', info='Only consider text *after* this string in any given chunk for training. For Alpaca datasets, use "### Response:" to only train the response and ignore the input.')
 
-                                add_eos_token = gr.Checkbox(label='Add EOS token', value=False, info="Adds EOS token for each dataset item. In case of raw text, the EOS will be added at the Hard Cut")
+                                add_eos_token = gr.Checkbox(label='Add EOS token', value=False, info="Adds EOS token for each document in text datasets.")
 
                                 higher_rank_limit = gr.Checkbox(label='Enable higher ranks', value=False, info='If checked, changes Rank/Alpha slider above to go much higher. This will not work without a datacenter-class GPU.')
                                 report_to = gr.Radio(label="Save detailed logs with", value="None", choices=["None", "wandb", "tensorboard"], interactive=True)
@@ -106,8 +106,8 @@ def create_ui():
                 with gr.Column():
                     with gr.Tab(label='Formatted Dataset'):
                         with gr.Row():
-                            format = gr.Dropdown(choices=['None', 'Chat Template'] + [x for x in utils.get_datasets('user_data/training/formats', 'json') if x != 'None'], value='None', label='Data Format', info='The format file used to decide how to format the dataset input. "Chat Template" uses the model\'s built-in chat template via apply_chat_template().', elem_classes=['slim-dropdown'], interactive=not mu)
-                            ui.create_refresh_button(format, lambda: None, lambda: {'choices': ['None', 'Chat Template'] + [x for x in utils.get_datasets('user_data/training/formats', 'json') if x != 'None']}, 'refresh-button', interactive=not mu)
+                            format = gr.Dropdown(choices=get_instruction_templates(), value='None', label='Data Format', info='Select an instruction template for formatting the dataset, or "Chat Template" to use the model\'s built-in chat template.', elem_classes=['slim-dropdown'], interactive=not mu)
+                            ui.create_refresh_button(format, lambda: None, lambda: {'choices': get_instruction_templates()}, 'refresh-button', interactive=not mu)
 
                         with gr.Row():
                             dataset = gr.Dropdown(choices=utils.get_datasets('user_data/training/datasets', 'json'), value='None', label='Dataset', info='The dataset file to use for training.', elem_classes=['slim-dropdown'], interactive=not mu)
@@ -158,7 +158,7 @@ def create_ui():
                 refresh_table = gr.Button('Refresh the table', elem_classes="small-button", interactive=not mu)
 
     # Training events
-    all_params = [lora_name, always_override, all_linear, q_proj_en, v_proj_en, k_proj_en, o_proj_en, gate_proj_en, down_proj_en, up_proj_en, save_steps, micro_batch_size, batch_size, epochs, learning_rate, lr_scheduler_type, lora_rank, lora_alpha, lora_dropout, cutoff_len, dataset, eval_dataset, format, eval_steps, text_dataset, higher_rank_limit, warmup_steps, optimizer, stride_length, train_only_after, stop_at_loss, add_eos_token, report_to]
+    all_params = [lora_name, always_override, all_linear, q_proj_en, v_proj_en, k_proj_en, o_proj_en, gate_proj_en, down_proj_en, up_proj_en, save_steps, micro_batch_size, batch_size, epochs, learning_rate, lr_scheduler_type, lora_rank, lora_alpha, lora_dropout, cutoff_len, dataset, eval_dataset, format, eval_steps, text_dataset, higher_rank_limit, warmup_steps, optimizer, stride_length, stop_at_loss, add_eos_token, report_to]
 
     copy_from.change(do_copy_params, [copy_from] + all_params, all_params)
     start_button.click(do_train, all_params, output)
@@ -222,6 +222,29 @@ def clean_path(base_path: str, path: str):
     return f'{Path(base_path).absolute()}/{path}'
 
 
+def get_instruction_templates():
+    path = Path('user_data/instruction-templates')
+    names = set()
+    for ext in ['yaml', 'yml', 'jinja']:
+        for f in path.glob(f'*.{ext}'):
+            names.add(f.stem)
+    return ['None', 'Chat Template'] + sorted(names, key=utils.natural_keys)
+
+
+def load_template(name):
+    """Load a Jinja2 template string from user_data/instruction-templates/."""
+    path = Path('user_data/instruction-templates')
+    for ext in ['jinja', 'yaml', 'yml']:
+        filepath = path / f'{name}.{ext}'
+        if filepath.exists():
+            if ext == 'jinja':
+                return filepath.read_text(encoding='utf-8')
+            else:
+                data = yaml.safe_load(filepath.read_text(encoding='utf-8'))
+                return data.get('instruction_template', '')
+    return ''
+
+
 def backup_adapter(input_folder):
     # Get the creation date of the adapter file (safetensors or bin)
     try:
@@ -269,7 +292,7 @@ def calc_trainable_parameters(model):
     return trainable_params, all_param
 
 
-def do_train(lora_name: str, always_override: bool, all_linear: bool, q_proj_en: bool, v_proj_en: bool, k_proj_en: bool, o_proj_en: bool, gate_proj_en: bool, down_proj_en: bool, up_proj_en: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, text_dataset: str, higher_rank_limit: bool, warmup_steps: int, optimizer: str, stride_length: int, train_only_after: str, stop_at_loss: float, add_eos_token: bool, report_to: str):
+def do_train(lora_name: str, always_override: bool, all_linear: bool, q_proj_en: bool, v_proj_en: bool, k_proj_en: bool, o_proj_en: bool, gate_proj_en: bool, down_proj_en: bool, up_proj_en: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, text_dataset: str, higher_rank_limit: bool, warmup_steps: int, optimizer: str, stride_length: int, stop_at_loss: float, add_eos_token: bool, report_to: str):
 
     import torch
     import transformers
@@ -324,47 +347,6 @@ def list_target_modules():
         }.items() if enabled]
         return target_mods
 
-    def encode(text, add_bos_token):
-        result = shared.tokenizer.encode(text, truncation=True, max_length=cutoff_len)
-        # Check if the first two tokens are BOS
-        if len(result) >= 2 and result[:2] == [shared.tokenizer.bos_token_id, shared.tokenizer.bos_token_id]:
-            result = result[1:]
-
-        if not add_bos_token and result[0] == shared.tokenizer.bos_token_id:
-            result = result[1:]
-        return result
-
-    def tokenize(prompt, append_eos_token=False):
-
-        if train_only_after == '' or train_only_after not in prompt:
-            input_ids = encode(prompt, True)
-
-            if append_eos_token and input_ids[-1] != shared.tokenizer.eos_token_id and len(input_ids) < cutoff_len:
-                input_ids.append(shared.tokenizer.eos_token_id)
-
-            labels = list(input_ids)
-
-        else:
-            ind = prompt.index(train_only_after) + len(train_only_after)
-            before_tokens = encode(prompt[:ind], True)
-            after_tokens = encode(prompt[ind:], False)
-
-            if append_eos_token and len(after_tokens) > 0 and after_tokens[-1] != shared.tokenizer.eos_token_id:
-                after_tokens.append(shared.tokenizer.eos_token_id)
-
-            full_length = len(after_tokens) + len(before_tokens)
-            if full_length > cutoff_len:
-                after_tokens = after_tokens[:cutoff_len - len(before_tokens)]
-
-            input_ids = before_tokens + after_tokens
-            labels = [-100] * len(before_tokens) + list(after_tokens)
-
-        return {
-            "input_ids": input_ids,
-            "labels": labels,
-            "attention_mask": [1] * len(input_ids),
-        }
-
     def normalize_messages(data_point):
         """Convert a dataset row to OpenAI messages format for apply_chat_template()."""
         if "messages" in data_point:
@@ -377,16 +359,8 @@ def normalize_messages(data_point):
                 for turn in data_point["conversations"]
             ]
 
-        if "instruction" in data_point and "output" in data_point:
-            messages = []
-            if data_point.get("system", "").strip():
-                messages.append({"role": "system", "content": data_point["system"]})
-            messages.append({"role": "user", "content": data_point["instruction"]})
-            messages.append({"role": "assistant", "content": data_point["output"]})
-            return messages
-
         raise RuntimeError(
-            f'Dataset row must contain "messages", "conversations", or "instruction"/"output" keys. '
+            f'Dataset row must contain "messages" or "conversations" key. '
             f'Found: {list(data_point.keys())}'
         )
 
@@ -492,73 +466,46 @@ def tokenize_conversation(data_point):
             return
 
         if format == 'Chat Template':
-            # Use the model's built-in chat template via apply_chat_template()
             if not getattr(shared.tokenizer, 'chat_template', None):
-                yield "Error: this model's tokenizer does not have a chat template. Use a format file instead, or load an instruct/chat model."
+                yield "Error: this model's tokenizer does not have a chat template. Select an instruction template instead, or load an instruct/chat model."
+                return
+        else:
+            # Load custom instruction template and set on tokenizer
+            template_str = load_template(format)
+            if not template_str:
+                yield f"Error: could not load instruction template '{format}'."
                 return
+            shared.tokenizer.chat_template = template_str
 
-            train_template["template_type"] = "chat_template"
+        # Unified path — both cases use tokenize_conversation()
+        train_template["template_type"] = "chat_template"
 
-            logger.info("Loading JSON dataset with Chat Template format")
-            data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
+        logger.info("Loading JSON dataset with chat template format")
+        data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
 
-            # Validate the first row
-            try:
-                normalize_messages(data['train'][0])
-            except (RuntimeError, KeyError, IndexError) as e:
-                yield f"Error: {e}"
-                return
+        # Validate the first row
+        try:
+            normalize_messages(data['train'][0])
+        except (RuntimeError, KeyError, IndexError) as e:
+            yield f"Error: {e}"
+            return
 
-            train_data = data['train'].map(
+        train_data = data['train'].map(
+            tokenize_conversation,
+            remove_columns=data['train'].column_names,
+            new_fingerprint='%030x' % random.randrange(16**30)
+        )
+
+        if eval_dataset == 'None':
+            eval_data = None
+        else:
+            eval_data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{eval_dataset}.json'))
+            eval_data = eval_data['train'].map(
                 tokenize_conversation,
-                remove_columns=data['train'].column_names,
+                remove_columns=eval_data['train'].column_names,
                 new_fingerprint='%030x' % random.randrange(16**30)
             )
 
-            if eval_dataset == 'None':
-                eval_data = None
-            else:
-                eval_data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{eval_dataset}.json'))
-                eval_data = eval_data['train'].map(
-                    tokenize_conversation,
-                    remove_columns=eval_data['train'].column_names,
-                    new_fingerprint='%030x' % random.randrange(16**30)
-                )
-        else:
-            # Use format file for prompt generation
-            train_template["template_type"] = "dataset"
-
-            with open(clean_path('user_data/training/formats', f'{format}.json'), 'r', encoding='utf-8-sig') as formatFile:
-                format_data: dict[str, str] = json.load(formatFile)
-
-            # == store training prompt ==
-            for _, value in format_data.items():
-                prompt_key = f"template_{len(train_template)}"
-                train_template[prompt_key] = value
-
-            def generate_prompt(data_point: dict[str, str]):
-                for options, data in format_data.items():
-                    if set(options.split(',')) == set(x[0] for x in data_point.items() if (type(x[1]) is str and len(x[1].strip()) > 0)):
-                        for key, val in data_point.items():
-                            if type(val) is str:
-                                data = data.replace(f'%{key}%', val)
-                        return data
-                raise RuntimeError(f'Data-point "{data_point}" has no keyset match within format "{list(format_data.keys())}"')
-
-            def generate_and_tokenize_prompt(data_point):
-                prompt = generate_prompt(data_point)
-                return tokenize(prompt, add_eos_token)
-
-            logger.info("Loading JSON datasets")
-            data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
-            train_data = data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
-
-            if eval_dataset == 'None':
-                eval_data = None
-            else:
-                eval_data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{eval_dataset}.json'))
-                eval_data = eval_data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
-
     # == We MUST reload model if it went through any previous training, even failed one ==
     if shared.model_dirty_from_training:
         selected_model = shared.model_name
diff --git a/user_data/training/datasets/example_messages.json b/user_data/training/datasets/example_messages.json
new file mode 100644
index 0000000000..13adcc1425
--- /dev/null
+++ b/user_data/training/datasets/example_messages.json
@@ -0,0 +1,10 @@
+[
+    {"messages": [{"role": "user", "content": "What is the capital of France?"}, {"role": "assistant", "content": "The capital of France is Paris."}]},
+    {"messages": [{"role": "user", "content": "What is 2+2?"}, {"role": "assistant", "content": "2+2 equals 4."}]},
+    {"messages": [{"role": "system", "content": "You are a helpful assistant."}, {"role": "user", "content": "Name a color."}, {"role": "assistant", "content": "Blue is a color."}, {"role": "user", "content": "Another one?"}, {"role": "assistant", "content": "Red is also a color."}]},
+    {"messages": [{"role": "user", "content": "What is the sun?"}, {"role": "assistant", "content": "The sun is a star at the center of our solar system."}]},
+    {"messages": [{"role": "user", "content": "Say hello."}, {"role": "assistant", "content": "Hello! How can I help you today?"}]},
+    {"messages": [{"role": "user", "content": "What is water?"}, {"role": "assistant", "content": "Water is a chemical compound with the formula H2O."}]},
+    {"messages": [{"role": "user", "content": "Name a fruit."}, {"role": "assistant", "content": "An apple is a fruit."}]},
+    {"messages": [{"role": "user", "content": "What is gravity?"}, {"role": "assistant", "content": "Gravity is a fundamental force that attracts objects with mass toward each other."}]}
+]
diff --git a/user_data/training/datasets/example_sharegpt.json b/user_data/training/datasets/example_sharegpt.json
new file mode 100644
index 0000000000..277b16ae2b
--- /dev/null
+++ b/user_data/training/datasets/example_sharegpt.json
@@ -0,0 +1,10 @@
+[
+    {"conversations": [{"from": "human", "value": "What is the capital of France?"}, {"from": "gpt", "value": "The capital of France is Paris."}]},
+    {"conversations": [{"from": "human", "value": "What is 2+2?"}, {"from": "gpt", "value": "2+2 equals 4."}]},
+    {"conversations": [{"from": "system", "value": "You are a helpful assistant."}, {"from": "human", "value": "Name a color."}, {"from": "gpt", "value": "Blue is a color."}, {"from": "human", "value": "Another one?"}, {"from": "gpt", "value": "Red is also a color."}]},
+    {"conversations": [{"from": "human", "value": "What is the sun?"}, {"from": "gpt", "value": "The sun is a star at the center of our solar system."}]},
+    {"conversations": [{"from": "human", "value": "Say hello."}, {"from": "gpt", "value": "Hello! How can I help you today?"}]},
+    {"conversations": [{"from": "human", "value": "What is water?"}, {"from": "gpt", "value": "Water is a chemical compound with the formula H2O."}]},
+    {"conversations": [{"from": "human", "value": "Name a fruit."}, {"from": "gpt", "value": "An apple is a fruit."}]},
+    {"conversations": [{"from": "human", "value": "What is gravity?"}, {"from": "gpt", "value": "Gravity is a fundamental force that attracts objects with mass toward each other."}]}
+]
diff --git a/user_data/training/datasets/example_text.json b/user_data/training/datasets/example_text.json
new file mode 100644
index 0000000000..d476bd58a4
--- /dev/null
+++ b/user_data/training/datasets/example_text.json
@@ -0,0 +1,10 @@
+[
+    {"text": "The quick brown fox jumps over the lazy dog. This is a simple sentence used for testing purposes. It contains all the letters of the English alphabet."},
+    {"text": "Machine learning is a subset of artificial intelligence that focuses on building systems that learn from data. These systems improve their performance over time without being explicitly programmed."},
+    {"text": "Python is a high-level programming language known for its simplicity and readability. It was created by Guido van Rossum and first released in 1991."},
+    {"text": "The Earth orbits the Sun at an average distance of about 93 million miles. It takes approximately 365.25 days to complete one orbit, which is why we have leap years."},
+    {"text": "Neural networks are computing systems inspired by biological neural networks in the brain. They consist of layers of interconnected nodes that process information using connectionist approaches."},
+    {"text": "Water covers about 71 percent of the Earth's surface. The oceans hold about 96.5 percent of all Earth's water. Only about 2.5 percent of the Earth's water is freshwater."},
+    {"text": "The history of computing dates back to ancient times with devices like the abacus. Modern electronic computing began in the mid-20th century with the development of vacuum tube computers."},
+    {"text": "Photosynthesis is the process by which green plants and some other organisms use sunlight to synthesize nutrients from carbon dioxide and water. It generates oxygen as a byproduct."}
+]
diff --git a/user_data/training/datasets/put-trainer-datasets-here.txt b/user_data/training/datasets/put-trainer-datasets-here.txt
index 932eacf802..75073074c1 100644
--- a/user_data/training/datasets/put-trainer-datasets-here.txt
+++ b/user_data/training/datasets/put-trainer-datasets-here.txt
@@ -1 +1 @@
-to load multiple raw text files create a subdirectory and put them all there
+Put your training dataset JSON files here.
diff --git a/user_data/training/formats/ChatML-format.json b/user_data/training/formats/ChatML-format.json
deleted file mode 100644
index a9f8a09afa..0000000000
--- a/user_data/training/formats/ChatML-format.json
+++ /dev/null
@@ -1,4 +0,0 @@
-{
-  "instruction,output": "<|im_start|>system\n<|im_end|>\n<|im_start|>user\n%instruction%<|im_end|>\n<|im_start|>assistant\n%output%<|im_end|>",
-  "instruction,input,output": "<|im_start|>system\n<|im_end|>\n<|im_start|>user\n%instruction%: %input%<|im_end|>\n<|im_start|>assistant\n%output%<|im_end|>"
-}
diff --git a/user_data/training/formats/alpaca-chatbot-format.json b/user_data/training/formats/alpaca-chatbot-format.json
deleted file mode 100644
index 4b38103f4c..0000000000
--- a/user_data/training/formats/alpaca-chatbot-format.json
+++ /dev/null
@@ -1,4 +0,0 @@
-{
-    "instruction,output": "User: %instruction%\nAssistant: %output%",
-    "instruction,input,output": "User: %instruction%: %input%\nAssistant: %output%"
-}
diff --git a/user_data/training/formats/alpaca-format.json b/user_data/training/formats/alpaca-format.json
deleted file mode 100644
index dd6df95640..0000000000
--- a/user_data/training/formats/alpaca-format.json
+++ /dev/null
@@ -1,4 +0,0 @@
-{
-    "instruction,output": "Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n### Instruction:\n%instruction%\n\n### Response:\n%output%",
-    "instruction,input,output": "Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.\n\n### Instruction:\n%instruction%\n\n### Input:\n%input%\n\n### Response:\n%output%"
-}
diff --git a/user_data/training/formats/llama2-chat-format.json b/user_data/training/formats/llama2-chat-format.json
deleted file mode 100644
index 5d43c59b90..0000000000
--- a/user_data/training/formats/llama2-chat-format.json
+++ /dev/null
@@ -1,4 +0,0 @@
-{
-    "modelanswer,userprompt,systemprompt": "[INST] <>\n%systemprompt%\n<>\n\n%userprompt%[/INST] %modelanswer%",
-    "modelanswer,userprompt": "[INST] <>\n\n<>\n\n%userprompt%[/INST] %modelanswer%"
-}
diff --git a/user_data/training/formats/vicuna-format.json b/user_data/training/formats/vicuna-format.json
deleted file mode 100644
index c1aa4f15eb..0000000000
--- a/user_data/training/formats/vicuna-format.json
+++ /dev/null
@@ -1,3 +0,0 @@
-{
-  "instruction,output": "A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions.\n\nUSER: %instruction%\n\nASSISTANT: %output%"
-}

From c2e494963f825b8b9ff0d88d6d3b54e7b6393100 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 14:41:44 -0300
Subject: [PATCH 1222/1701] Training: fix silent error on model reload failure,
 minor cleanups

---
 modules/training.py | 14 ++++++++------
 1 file changed, 8 insertions(+), 6 deletions(-)

diff --git a/modules/training.py b/modules/training.py
index 6472e692eb..a762cbda9e 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -517,12 +517,14 @@ def tokenize_conversation(data_point):
                 if shared.model is not None:
                     print("Model reloaded OK, continue with training.")
                 else:
-                    return f"Failed to load {selected_model}."
+                    yield f"Failed to load {selected_model}."
+                    return
             except Exception:
                 exc = traceback.format_exc()
                 logger.error('Failed to reload the model.')
                 print(exc)
-                return exc.replace('\n', '\n\n')
+                yield exc.replace('\n', '\n\n')
+                return
 
     # == Start prepping the model itself ==
     if not hasattr(shared.model, 'lm_head') or hasattr(shared.model.lm_head, 'weight'):
@@ -614,7 +616,7 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
             print(f"\033[1;30;40mStep: {tracked.current_steps} \033[0;37;0m", end='')
             if 'loss' in logs:
                 loss = float(logs['loss'])
-                if loss <= stop_at_loss:
+                if stop_at_loss > 0 and loss <= stop_at_loss:
                     control.should_epoch_stop = True
                     control.should_training_stop = True
                     print(f"\033[1;31;1mStop Loss {stop_at_loss} reached.\033[0;37;0m")
@@ -670,13 +672,13 @@ def collate_fn(batch):
             remove_unused_columns=False,
         ),
         data_collator=collate_fn,
-        callbacks=list([Callbacks()])
+        callbacks=[Callbacks()]
     )
 
     # == Save parameters for reuse ==
     with open(f"{lora_file_path}/training_parameters.json", 'w', encoding='utf-8') as file:
-        vars = locals()
-        json.dump({x: vars[x] for x in PARAMETERS}, file, indent=2)
+        local_vars = locals()
+        json.dump({x: local_vars[x] for x in PARAMETERS}, file, indent=2)
 
     # == Save training prompt ==
     with open(f"{lora_file_path}/training_prompt.json", 'w', encoding='utf-8') as file:

From 33a38d7ece65925720b8de16039ab0577133ed41 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 14:56:27 -0300
Subject: [PATCH 1223/1701] Training: drop conversations exceeding cutoff
 length instead of truncating

---
 modules/training.py | 24 ++++++++++++++++++------
 1 file changed, 18 insertions(+), 6 deletions(-)

diff --git a/modules/training.py b/modules/training.py
index a762cbda9e..45e6fe5e09 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -26,7 +26,7 @@
 from modules.logging_colors import logger
 from modules.models import reload_model
 
-PARAMETERS = ["lora_name", "always_override", "all_linear", "q_proj_en", "v_proj_en", "k_proj_en", "o_proj_en", "gate_proj_en", "down_proj_en", "up_proj_en", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "text_dataset", "higher_rank_limit", "warmup_steps", "optimizer", "stride_length", "stop_at_loss", "add_eos_token", "report_to"]
+PARAMETERS = ["lora_name", "always_override", "all_linear", "q_proj_en", "v_proj_en", "k_proj_en", "o_proj_en", "gate_proj_en", "down_proj_en", "up_proj_en", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "text_dataset", "higher_rank_limit", "warmup_steps", "optimizer", "stride_length", "stop_at_loss", "add_eos_token", "excess_length", "report_to"]
 WANT_INTERRUPT = False
 
 train_log = {}
@@ -99,6 +99,7 @@ def create_ui():
                                 warmup_steps = gr.Number(label='Warmup Steps', value=100, info='For this many steps at the start, the learning rate will be lower than normal. This helps the trainer prepare the model and precompute statistics to improve the quality of training after the start.')
 
                                 add_eos_token = gr.Checkbox(label='Add EOS token', value=False, info="Adds EOS token for each document in text datasets.")
+                                excess_length = gr.Dropdown(label='Excess length', value='drop', choices=['drop', 'truncate'], info='What to do with conversations that exceed the cutoff length. "Drop" removes them entirely (recommended). "Truncate" cuts from the right, which may produce incomplete responses.', elem_classes=['slim-dropdown'])
 
                                 higher_rank_limit = gr.Checkbox(label='Enable higher ranks', value=False, info='If checked, changes Rank/Alpha slider above to go much higher. This will not work without a datacenter-class GPU.')
                                 report_to = gr.Radio(label="Save detailed logs with", value="None", choices=["None", "wandb", "tensorboard"], interactive=True)
@@ -158,7 +159,7 @@ def create_ui():
                 refresh_table = gr.Button('Refresh the table', elem_classes="small-button", interactive=not mu)
 
     # Training events
-    all_params = [lora_name, always_override, all_linear, q_proj_en, v_proj_en, k_proj_en, o_proj_en, gate_proj_en, down_proj_en, up_proj_en, save_steps, micro_batch_size, batch_size, epochs, learning_rate, lr_scheduler_type, lora_rank, lora_alpha, lora_dropout, cutoff_len, dataset, eval_dataset, format, eval_steps, text_dataset, higher_rank_limit, warmup_steps, optimizer, stride_length, stop_at_loss, add_eos_token, report_to]
+    all_params = [lora_name, always_override, all_linear, q_proj_en, v_proj_en, k_proj_en, o_proj_en, gate_proj_en, down_proj_en, up_proj_en, save_steps, micro_batch_size, batch_size, epochs, learning_rate, lr_scheduler_type, lora_rank, lora_alpha, lora_dropout, cutoff_len, dataset, eval_dataset, format, eval_steps, text_dataset, higher_rank_limit, warmup_steps, optimizer, stride_length, stop_at_loss, add_eos_token, excess_length, report_to]
 
     copy_from.change(do_copy_params, [copy_from] + all_params, all_params)
     start_button.click(do_train, all_params, output)
@@ -292,7 +293,7 @@ def calc_trainable_parameters(model):
     return trainable_params, all_param
 
 
-def do_train(lora_name: str, always_override: bool, all_linear: bool, q_proj_en: bool, v_proj_en: bool, k_proj_en: bool, o_proj_en: bool, gate_proj_en: bool, down_proj_en: bool, up_proj_en: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, text_dataset: str, higher_rank_limit: bool, warmup_steps: int, optimizer: str, stride_length: int, stop_at_loss: float, add_eos_token: bool, report_to: str):
+def do_train(lora_name: str, always_override: bool, all_linear: bool, q_proj_en: bool, v_proj_en: bool, k_proj_en: bool, o_proj_en: bool, gate_proj_en: bool, down_proj_en: bool, up_proj_en: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, text_dataset: str, higher_rank_limit: bool, warmup_steps: int, optimizer: str, stride_length: int, stop_at_loss: float, add_eos_token: bool, excess_length: str, report_to: str):
 
     import torch
     import transformers
@@ -389,10 +390,12 @@ def tokenize_conversation(data_point):
                 end = min(len(through_ids), len(full_ids))
                 labels[start:end] = full_ids[start:end]
 
-        # Truncate from the right: keeps the system prompt and early turns
         if len(full_ids) > cutoff_len:
-            full_ids = full_ids[:cutoff_len]
-            labels = labels[:cutoff_len]
+            if excess_length == 'truncate':
+                full_ids = full_ids[:cutoff_len]
+                labels = labels[:cutoff_len]
+            else:
+                return {"input_ids": [], "labels": [], "attention_mask": []}
 
         return {
             "input_ids": full_ids,
@@ -490,11 +493,19 @@ def tokenize_conversation(data_point):
             yield f"Error: {e}"
             return
 
+        total = len(data['train'])
         train_data = data['train'].map(
             tokenize_conversation,
             remove_columns=data['train'].column_names,
             new_fingerprint='%030x' % random.randrange(16**30)
         )
+        train_data = train_data.filter(lambda x: len(x['input_ids']) > 0)
+        dropped = total - len(train_data)
+        if dropped > 0:
+            logger.warning(f"Dropped {dropped}/{total} conversations exceeding cutoff length of {cutoff_len} tokens.")
+        if len(train_data) == 0:
+            yield f"Error: all {total} conversations exceed the cutoff length of {cutoff_len} tokens. Increase the cutoff length or shorten your data."
+            return
 
         if eval_dataset == 'None':
             eval_data = None
@@ -505,6 +516,7 @@ def tokenize_conversation(data_point):
                 remove_columns=eval_data['train'].column_names,
                 new_fingerprint='%030x' % random.randrange(16**30)
             )
+            eval_data = eval_data.filter(lambda x: len(x['input_ids']) > 0)
 
     # == We MUST reload model if it went through any previous training, even failed one ==
     if shared.model_dirty_from_training:

From 63f28cb4a295bfb7ecac2b18d6c135e0ab994374 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 15:12:32 -0300
Subject: [PATCH 1224/1701] Training: align defaults with peft/axolotl (rank 8,
 alpha 16, dropout 0, cutoff 512, eos on)

---
 modules/training.py | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/modules/training.py b/modules/training.py
index 45e6fe5e09..5721ec69f9 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -73,11 +73,11 @@ def create_ui():
 
                     with gr.Row():
                         with gr.Column():
-                            lora_rank = gr.Slider(label='LoRA Rank', value=32, minimum=0, maximum=1024, step=4, info='Also called dimension count. Higher values = larger file, more content control. Smaller values = smaller file, less control. Use 4 or 8 for style, 128 or 256 to teach, 1024+ for fine-detail on big data. More VRAM is needed for higher ranks.')
-                            lora_alpha = gr.Slider(label='LoRA Alpha', value=64, minimum=0, maximum=2048, step=4, info='This divided by the rank becomes the scaling of the LoRA. Higher means stronger. A good standard value is twice your Rank.')
+                            lora_rank = gr.Slider(label='LoRA Rank', value=8, minimum=0, maximum=1024, step=4, info='Also called dimension count. Higher values = larger file, more content control. Smaller values = smaller file, less control. Use 4 or 8 for style, 128 or 256 to teach, 1024+ for fine-detail on big data. More VRAM is needed for higher ranks.')
+                            lora_alpha = gr.Slider(label='LoRA Alpha', value=16, minimum=0, maximum=2048, step=4, info='This divided by the rank becomes the scaling of the LoRA. Higher means stronger. A good standard value is twice your Rank.')
                             batch_size = gr.Slider(label='Batch Size', value=128, minimum=0, maximum=1024, step=4, info='Global batch size. The two batch sizes together determine gradient accumulation (gradientAccum = batch / microBatch). Higher gradient accum values lead to better quality training.')
                             micro_batch_size = gr.Slider(label='Micro Batch Size', value=4, minimum=1, maximum=128, step=1, info='Per-device batch size (NOTE: multiple devices not yet implemented). Increasing this will increase VRAM usage.')
-                            cutoff_len = gr.Slider(label='Cutoff Length', minimum=0, maximum=4096, value=256, step=32, info='Cutoff length for text input. Essentially, how long of a line of text to feed in at a time. Higher values require drastically more VRAM.')
+                            cutoff_len = gr.Slider(label='Cutoff Length', minimum=0, maximum=4096, value=512, step=32, info='Cutoff length for text input. Essentially, how long of a line of text to feed in at a time. Higher values require drastically more VRAM.')
 
                         with gr.Column():
                             save_steps = gr.Number(label='Save every n steps', value=0, info='If above 0, a checkpoint of the LoRA will be saved every time this many steps pass.')
@@ -90,7 +90,7 @@ def create_ui():
                     with gr.Accordion(label='Advanced Options', open=False, elem_classes='tgw-accordion'):
                         with gr.Row():
                             with gr.Column():
-                                lora_dropout = gr.Slider(label='LoRA Dropout', minimum=0.0, maximum=1.0, step=0.025, value=0.05, info='Percentage probability for dropout of LoRA layers. This can help reduce overfitting. Most users should leave at default.')
+                                lora_dropout = gr.Slider(label='LoRA Dropout', minimum=0.0, maximum=1.0, step=0.025, value=0.0, info='Percentage probability for dropout of LoRA layers. This can help reduce overfitting. Most users should leave at default.')
                                 stop_at_loss = gr.Slider(label='Stop at loss', minimum=0.0, maximum=3.0, step=0.1, value=0.00, info='The process will automatically stop once the desired loss value is reached. (reasonable numbers are 1.5-1.8)')
                                 with gr.Row():
                                     optimizer = gr.Dropdown(label='Optimizer', value='adamw_torch', choices=['adamw_hf', 'adamw_torch', 'adamw_torch_fused', 'adamw_torch_xla', 'adamw_apex_fused', 'adafactor', 'adamw_bnb_8bit', 'adamw_anyprecision', 'sgd', 'adagrad'], info='Different optimizer implementation options, for advanced users. Effects of different options are not well documented yet.', elem_classes=['slim-dropdown'])
@@ -98,7 +98,7 @@ def create_ui():
                             with gr.Column():
                                 warmup_steps = gr.Number(label='Warmup Steps', value=100, info='For this many steps at the start, the learning rate will be lower than normal. This helps the trainer prepare the model and precompute statistics to improve the quality of training after the start.')
 
-                                add_eos_token = gr.Checkbox(label='Add EOS token', value=False, info="Adds EOS token for each document in text datasets.")
+                                add_eos_token = gr.Checkbox(label='Add EOS token', value=True, info="Adds EOS token for each document in text datasets.")
                                 excess_length = gr.Dropdown(label='Excess length', value='drop', choices=['drop', 'truncate'], info='What to do with conversations that exceed the cutoff length. "Drop" removes them entirely (recommended). "Truncate" cuts from the right, which may produce incomplete responses.', elem_classes=['slim-dropdown'])
 
                                 higher_rank_limit = gr.Checkbox(label='Enable higher ranks', value=False, info='If checked, changes Rank/Alpha slider above to go much higher. This will not work without a datacenter-class GPU.')
@@ -671,7 +671,7 @@ def collate_fn(batch):
             fp16=False if shared.args.cpu or shared.args.bf16 else True,
             bf16=shared.args.bf16,
             optim=optimizer,
-            logging_steps=2 if stop_at_loss > 0 else 5,
+            logging_steps=1,
             eval_strategy="steps" if eval_data is not None else "no",
             eval_steps=math.ceil(eval_steps / gradient_accumulation_steps) if eval_data is not None else None,
             save_strategy="steps" if eval_data is not None else "no",

From 438e59498e359031ced3180c8c7e13f9f356e64f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 10:24:31 -0800
Subject: [PATCH 1225/1701] Update ExLlamaV3 to v0.0.23

---
 requirements/full/requirements.txt | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 5b3139c11d..df0f0fc6a8 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -44,8 +44,8 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.22/exllamav3-0.0.22+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.22/exllamav3-0.0.22+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.23/exllamav3-0.0.23+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.23/exllamav3-0.0.23+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

From 275810c8439372892af00c6ab29df0576646fb54 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 15:32:49 -0300
Subject: [PATCH 1226/1701] Training: wire up HF Trainer checkpoint resumption
 for full state recovery

---
 modules/training.py | 54 +++++++++++++++++++++++++++------------------
 1 file changed, 32 insertions(+), 22 deletions(-)

diff --git a/modules/training.py b/modules/training.py
index 5721ec69f9..4552083048 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -569,21 +569,30 @@ def tokenize_conversation(data_point):
     # == get model trainable params
     model_trainable_params, model_all_params = calc_trainable_parameters(shared.model)
 
+    # == Determine if we can resume from a checkpoint ==
+    resume_checkpoint = None
     try:
         logger.info("Creating LoRA model")
         lora_model = get_peft_model(shared.model, config)
-        if not always_override:
-            safetensors_path = Path(f"{lora_file_path}/adapter_model.safetensors")
-            bin_path = Path(f"{lora_file_path}/adapter_model.bin")
-            if safetensors_path.is_file():
-                logger.info("Loading existing LoRA data (safetensors)")
-                from safetensors.torch import load_file
-                state_dict_peft = load_file(str(safetensors_path))
-                set_peft_model_state_dict(lora_model, state_dict_peft)
-            elif bin_path.is_file():
-                logger.info("Loading existing LoRA data (bin)")
-                state_dict_peft = torch.load(str(bin_path), weights_only=True)
-                set_peft_model_state_dict(lora_model, state_dict_peft)
+        if not always_override and Path(lora_file_path).exists():
+            # Look for HF Trainer checkpoint dirs (full resumption)
+            checkpoints = sorted(Path(lora_file_path).glob("checkpoint-*"), key=os.path.getmtime)
+            if checkpoints:
+                resume_checkpoint = str(checkpoints[-1])
+                logger.info(f"Will resume from checkpoint: {resume_checkpoint}")
+            else:
+                # Legacy fallback: load bare adapter weights only
+                safetensors_path = Path(f"{lora_file_path}/adapter_model.safetensors")
+                bin_path = Path(f"{lora_file_path}/adapter_model.bin")
+                if safetensors_path.is_file():
+                    logger.info("Loading existing LoRA data (safetensors)")
+                    from safetensors.torch import load_file
+                    state_dict_peft = load_file(str(safetensors_path))
+                    set_peft_model_state_dict(lora_model, state_dict_peft)
+                elif bin_path.is_file():
+                    logger.info("Loading existing LoRA data (bin)")
+                    state_dict_peft = torch.load(str(bin_path), weights_only=True)
+                    set_peft_model_state_dict(lora_model, state_dict_peft)
     except Exception:
         yield traceback.format_exc().replace('\n', '\n\n')
         return
@@ -604,14 +613,6 @@ def on_step_begin(self, args: transformers.TrainingArguments, state: transformer
             if WANT_INTERRUPT:
                 control.should_epoch_stop = True
                 control.should_training_stop = True
-            elif state.global_step > 0 and actual_save_steps > 0 and state.global_step % actual_save_steps == 0:
-                lora_model.save_pretrained(f"{lora_file_path}/checkpoint-{tracked.current_steps}/")
-                # Save log
-                with open(f"{lora_file_path}/checkpoint-{tracked.current_steps}/training_log.json", 'w', encoding='utf-8') as file:
-                    json.dump(train_log, file, indent=2)
-                # == Save training prompt ==
-                with open(f"{lora_file_path}/checkpoint-{tracked.current_steps}/training_prompt.json", 'w', encoding='utf-8') as file:
-                    json.dump(train_template, file, indent=2)
 
         def on_substep_end(self, args: transformers.TrainingArguments, state: transformers.TrainerState, control: transformers.TrainerControl, **kwargs):
             tracked.current_steps += 1
@@ -633,6 +634,14 @@ def on_log(self, args: transformers.TrainingArguments, state: transformers.Train
                     control.should_training_stop = True
                     print(f"\033[1;31;1mStop Loss {stop_at_loss} reached.\033[0;37;0m")
 
+        def on_save(self, args: transformers.TrainingArguments, state: transformers.TrainerState, control: transformers.TrainerControl, **kwargs):
+            checkpoint_dir = Path(args.output_dir) / f"checkpoint-{state.global_step}"
+            if checkpoint_dir.exists():
+                with open(checkpoint_dir / "training_log.json", 'w', encoding='utf-8') as file:
+                    json.dump(train_log, file, indent=2)
+                with open(checkpoint_dir / "training_prompt.json", 'w', encoding='utf-8') as file:
+                    json.dump(train_template, file, indent=2)
+
     # Fix training for mixed precision models
     for param in shared.model.parameters():
         if param.requires_grad:
@@ -674,7 +683,8 @@ def collate_fn(batch):
             logging_steps=1,
             eval_strategy="steps" if eval_data is not None else "no",
             eval_steps=math.ceil(eval_steps / gradient_accumulation_steps) if eval_data is not None else None,
-            save_strategy="steps" if eval_data is not None else "no",
+            save_strategy="steps" if save_steps > 0 or eval_data is not None else "no",
+            save_steps=actual_save_steps if save_steps > 0 else None,
             output_dir=lora_file_path,
             lr_scheduler_type=lr_scheduler_type,
             load_best_model_at_end=eval_data is not None,
@@ -745,7 +755,7 @@ def log_train_dataset(trainer):
 
     def threaded_run():
         log_train_dataset(trainer)
-        trainer.train()
+        trainer.train(resume_from_checkpoint=resume_checkpoint)
         # Note: save in the thread in case the gradio thread breaks (eg browser closed)
         lora_model.save_pretrained(lora_file_path)
         tracked.did_save = True

From 7a1fa8c9ea7ddc74821ff8ddf17bd790dacdd40c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 15:50:39 -0300
Subject: [PATCH 1227/1701] Training: fix checkpoint resume and surface
 training errors to UI

---
 modules/training.py | 33 +++++++++++++++++++++------------
 1 file changed, 21 insertions(+), 12 deletions(-)

diff --git a/modules/training.py b/modules/training.py
index 4552083048..0cb29dceb4 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -649,9 +649,6 @@ def on_save(self, args: transformers.TrainingArguments, state: transformers.Trai
 
     lora_model.config.use_cache = False
 
-    if sys.platform != "win32":
-        lora_model = torch.compile(lora_model)
-
     def collate_fn(batch):
         max_len = max(len(item['input_ids']) for item in batch)
         input_ids, labels, attention_mask = [], [], []
@@ -753,16 +750,23 @@ def log_train_dataset(trainer):
         except Exception as e:
             logger.error(f"Failed to create log file due to error: {e}")
 
+    thread_error = None
+
     def threaded_run():
-        log_train_dataset(trainer)
-        trainer.train(resume_from_checkpoint=resume_checkpoint)
-        # Note: save in the thread in case the gradio thread breaks (eg browser closed)
-        lora_model.save_pretrained(lora_file_path)
-        tracked.did_save = True
-        logger.info("LoRA training run is completed and saved.")
-        # Save log
-        with open(f"{lora_file_path}/training_log.json", 'w', encoding='utf-8') as file:
-            json.dump(train_log, file, indent=2)
+        nonlocal thread_error
+        try:
+            log_train_dataset(trainer)
+            trainer.train(resume_from_checkpoint=resume_checkpoint)
+            # Note: save in the thread in case the gradio thread breaks (eg browser closed)
+            lora_model.save_pretrained(lora_file_path)
+            tracked.did_save = True
+            logger.info("LoRA training run is completed and saved.")
+            # Save log
+            with open(f"{lora_file_path}/training_log.json", 'w', encoding='utf-8') as file:
+                json.dump(train_log, file, indent=2)
+        except Exception as e:
+            thread_error = e
+            logger.error(f"Training error: {e}")
 
     thread = threading.Thread(target=threaded_run)
     thread.start()
@@ -791,6 +795,11 @@ def threaded_run():
 
             yield f"Running... **{tracked.current_steps}** / **{tracked.max_steps}** ... {timer_info}, {format_time(time_elapsed)} / {format_time(total_time_estimate)} ... {format_time(total_time_estimate - time_elapsed)} remaining"
 
+    # Check for errors from the training thread
+    if thread_error is not None:
+        yield f"Training failed: {thread_error}"
+        return
+
     # Saving in the train thread might fail if an error occurs, so save here if so.
     if not tracked.did_save:
         logger.info("Training complete, saving")

From 33ff3773a0d66f25b0054896159ba1bffa646a4c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 15:59:49 -0300
Subject: [PATCH 1228/1701] Clean up LoRA loading parameter handling

---
 modules/LoRA.py | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/modules/LoRA.py b/modules/LoRA.py
index a9e9a895e5..63d764b4ad 100644
--- a/modules/LoRA.py
+++ b/modules/LoRA.py
@@ -77,9 +77,7 @@ def add_lora_transformers(lora_names):
     if len(lora_names) > 0:
         params = {}
         if not shared.args.cpu:
-            if shared.args.load_in_4bit or shared.args.load_in_8bit:
-                params['peft_type'] = shared.model.dtype
-            else:
+            if not shared.args.load_in_4bit and not shared.args.load_in_8bit:
                 params['dtype'] = shared.model.dtype
                 if hasattr(shared.model, "hf_device_map"):
                     params['device_map'] = {"base_model.model." + k: v for k, v in shared.model.hf_device_map.items()}

From 86d8291e58e8cdab18a35fd79a9402f63cd68d6f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 11:15:16 -0800
Subject: [PATCH 1229/1701] Training: UI cleanup and better defaults

---
 modules/training.py | 105 +++++++++++++++++++++++++-------------------
 modules/utils.py    |  33 ++++++++++++++
 2 files changed, 93 insertions(+), 45 deletions(-)

diff --git a/modules/training.py b/modules/training.py
index 0cb29dceb4..ca8d1bd150 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -53,8 +53,8 @@ def create_ui():
                             always_override = gr.Checkbox(label='Override Existing Files', value=False, info='If the name is the same, checking will replace the existing file, and unchecking will load and continue from it (the rank must be the same).', elem_classes=['no-background'])
 
                     with gr.Accordion(label='Target Modules', open=False, elem_classes='tgw-accordion'):
-                        gr.Markdown("Selects which modules to target in training. Targeting more modules is closer to a full fine-tune at the cost of increased VRAM requirements and adapter size.\n\"Target all linear layers\" uses peft's `all-linear` option, which targets every `nn.Linear` layer except `lm_head` and works for any model architecture. Uncheck it to manually select individual projection modules below.")
-                        all_linear = gr.Checkbox(label='Target all linear layers', value=True, info='Targets every nn.Linear layer except lm_head. Works for any model architecture.', elem_classes=['no-background'])
+                        gr.Markdown("Selects which modules to target in training. Targeting more modules is closer to a full fine-tune at the cost of increased VRAM and adapter size.")
+                        all_linear = gr.Checkbox(label='Target all linear layers', value=True, info='Targets every nn.Linear layer except lm_head. Works for any model architecture. When checked, the individual module checkboxes below are ignored.', elem_classes=['no-background'])
                         with gr.Row():
                             with gr.Column():
                                 q_proj_en = gr.Checkbox(label='Enable q_proj', value=True)
@@ -75,17 +75,17 @@ def create_ui():
                         with gr.Column():
                             lora_rank = gr.Slider(label='LoRA Rank', value=8, minimum=0, maximum=1024, step=4, info='Also called dimension count. Higher values = larger file, more content control. Smaller values = smaller file, less control. Use 4 or 8 for style, 128 or 256 to teach, 1024+ for fine-detail on big data. More VRAM is needed for higher ranks.')
                             lora_alpha = gr.Slider(label='LoRA Alpha', value=16, minimum=0, maximum=2048, step=4, info='This divided by the rank becomes the scaling of the LoRA. Higher means stronger. A good standard value is twice your Rank.')
-                            batch_size = gr.Slider(label='Batch Size', value=128, minimum=0, maximum=1024, step=4, info='Global batch size. The two batch sizes together determine gradient accumulation (gradientAccum = batch / microBatch). Higher gradient accum values lead to better quality training.')
+                            batch_size = gr.Slider(label='Batch Size', value=32, minimum=0, maximum=1024, step=4, info='Global batch size. The two batch sizes together determine gradient accumulation (gradientAccum = batch / microBatch). Higher gradient accum values lead to better quality training.')
                             micro_batch_size = gr.Slider(label='Micro Batch Size', value=4, minimum=1, maximum=128, step=1, info='Per-device batch size (NOTE: multiple devices not yet implemented). Increasing this will increase VRAM usage.')
-                            cutoff_len = gr.Slider(label='Cutoff Length', minimum=0, maximum=4096, value=512, step=32, info='Cutoff length for text input. Essentially, how long of a line of text to feed in at a time. Higher values require drastically more VRAM.')
+                            cutoff_len = gr.Slider(label='Cutoff Length', minimum=0, maximum=4096, value=512, step=32, info='Maximum sequence length in tokens. For instruction datasets, conversations longer than this are dropped. For text datasets, documents are split into chunks of this size. Higher values require more VRAM.')
 
                         with gr.Column():
-                            save_steps = gr.Number(label='Save every n steps', value=0, info='If above 0, a checkpoint of the LoRA will be saved every time this many steps pass.')
+                            save_steps = gr.Number(label='Save every n steps', value=0, info='If above 0, a full training checkpoint (adapter weights, optimizer, scheduler) will be saved every time this many steps pass. Training can be resumed from these checkpoints.')
 
                             epochs = gr.Number(label='Epochs', value=3, info='Number of times every entry in the dataset should be fed into training. So 1 means feed each item in once, 5 means feed it in five times, etc.')
                             learning_rate = gr.Textbox(label='Learning Rate', value='3e-4', info='In scientific notation. 3e-4 is a good starting base point. 1e-2 is extremely high, 1e-6 is extremely low.')
                             with gr.Row():
-                                lr_scheduler_type = gr.Dropdown(label='LR Scheduler', value='linear', choices=['linear', 'constant', 'constant_with_warmup', 'cosine', 'cosine_with_restarts', 'polynomial', 'inverse_sqrt'], info='Learning rate scheduler - defines how the learning rate changes over time. "Constant" means never change, "linear" means to go in a straight line from the learning rate down to 0, cosine follows a curve, etc.', elem_classes=['slim-dropdown'])
+                                lr_scheduler_type = gr.Dropdown(label='LR Scheduler', value='cosine', choices=['linear', 'constant', 'constant_with_warmup', 'cosine', 'cosine_with_restarts', 'polynomial', 'inverse_sqrt'], info='Learning rate scheduler - defines how the learning rate changes over time. "Constant" means never change, "linear" means to go in a straight line from the learning rate down to 0, cosine follows a curve, etc.', elem_classes=['slim-dropdown'])
 
                     with gr.Accordion(label='Advanced Options', open=False, elem_classes='tgw-accordion'):
                         with gr.Row():
@@ -93,10 +93,10 @@ def create_ui():
                                 lora_dropout = gr.Slider(label='LoRA Dropout', minimum=0.0, maximum=1.0, step=0.025, value=0.0, info='Percentage probability for dropout of LoRA layers. This can help reduce overfitting. Most users should leave at default.')
                                 stop_at_loss = gr.Slider(label='Stop at loss', minimum=0.0, maximum=3.0, step=0.1, value=0.00, info='The process will automatically stop once the desired loss value is reached. (reasonable numbers are 1.5-1.8)')
                                 with gr.Row():
-                                    optimizer = gr.Dropdown(label='Optimizer', value='adamw_torch', choices=['adamw_hf', 'adamw_torch', 'adamw_torch_fused', 'adamw_torch_xla', 'adamw_apex_fused', 'adafactor', 'adamw_bnb_8bit', 'adamw_anyprecision', 'sgd', 'adagrad'], info='Different optimizer implementation options, for advanced users. Effects of different options are not well documented yet.', elem_classes=['slim-dropdown'])
+                                    optimizer = gr.Dropdown(label='Optimizer', value='adamw_torch', choices=['adamw_hf', 'adamw_torch', 'adamw_torch_fused', 'adamw_torch_xla', 'adamw_apex_fused', 'adafactor', 'adamw_bnb_8bit', 'adamw_anyprecision', 'sgd', 'adagrad'], info='Optimizer algorithm. adamw_torch is the standard choice. adamw_bnb_8bit uses less VRAM. adafactor is memory-efficient for large models.', elem_classes=['slim-dropdown'])
 
                             with gr.Column():
-                                warmup_steps = gr.Number(label='Warmup Steps', value=100, info='For this many steps at the start, the learning rate will be lower than normal. This helps the trainer prepare the model and precompute statistics to improve the quality of training after the start.')
+                                warmup_steps = gr.Number(label='Warmup Steps', value=100, info='For this many steps at the start, the learning rate is gradually ramped up from 0 to the target value. This prevents unstable updates early in training.')
 
                                 add_eos_token = gr.Checkbox(label='Add EOS token', value=True, info="Adds EOS token for each document in text datasets.")
                                 excess_length = gr.Dropdown(label='Excess length', value='drop', choices=['drop', 'truncate'], info='What to do with conversations that exceed the cutoff length. "Drop" removes them entirely (recommended). "Truncate" cuts from the right, which may produce incomplete responses.', elem_classes=['slim-dropdown'])
@@ -105,27 +105,27 @@ def create_ui():
                                 report_to = gr.Radio(label="Save detailed logs with", value="None", choices=["None", "wandb", "tensorboard"], interactive=True)
 
                 with gr.Column():
-                    with gr.Tab(label='Formatted Dataset'):
+                    with gr.Tab(label='Chat Dataset'):
                         with gr.Row():
-                            format = gr.Dropdown(choices=get_instruction_templates(), value='None', label='Data Format', info='Select an instruction template for formatting the dataset, or "Chat Template" to use the model\'s built-in chat template.', elem_classes=['slim-dropdown'], interactive=not mu)
-                            ui.create_refresh_button(format, lambda: None, lambda: {'choices': get_instruction_templates()}, 'refresh-button', interactive=not mu)
+                            dataset = gr.Dropdown(choices=utils.get_chat_datasets('user_data/training/datasets'), value='None', label='Dataset File', info='A JSON file with chat conversations (messages or ShareGPT format). Each row is one conversation.', elem_classes=['slim-dropdown'], interactive=not mu)
+                            ui.create_refresh_button(dataset, lambda: None, lambda: {'choices': utils.get_chat_datasets('user_data/training/datasets')}, 'refresh-button', interactive=not mu)
 
                         with gr.Row():
-                            dataset = gr.Dropdown(choices=utils.get_datasets('user_data/training/datasets', 'json'), value='None', label='Dataset', info='The dataset file to use for training.', elem_classes=['slim-dropdown'], interactive=not mu)
-                            ui.create_refresh_button(dataset, lambda: None, lambda: {'choices': utils.get_datasets('user_data/training/datasets', 'json')}, 'refresh-button', interactive=not mu)
+                            format = gr.Dropdown(choices=get_instruction_templates(), value='None', label='Instruction Template', info='Select an instruction template for formatting the dataset, or "Chat Template" to use the model\'s built-in chat template.', elem_classes=['slim-dropdown'], interactive=not mu)
+                            ui.create_refresh_button(format, lambda: None, lambda: {'choices': get_instruction_templates()}, 'refresh-button', interactive=not mu)
 
+                    with gr.Tab(label="Text Dataset"):
                         with gr.Row():
-                            eval_dataset = gr.Dropdown(choices=utils.get_datasets('user_data/training/datasets', 'json'), value='None', label='Evaluation Dataset', info='The (optional) dataset file used to evaluate the model after training.', elem_classes=['slim-dropdown'], interactive=not mu)
-                            ui.create_refresh_button(eval_dataset, lambda: None, lambda: {'choices': utils.get_datasets('user_data/training/datasets', 'json')}, 'refresh-button', interactive=not mu)
+                            text_dataset = gr.Dropdown(choices=utils.get_text_datasets('user_data/training/datasets'), value='None', label='Dataset File', info='A JSON file with a "text" key per row, for pretraining-style training. Each row is one document.', elem_classes=['slim-dropdown'], interactive=not mu)
+                            ui.create_refresh_button(text_dataset, lambda: None, lambda: {'choices': utils.get_text_datasets('user_data/training/datasets')}, 'refresh-button', interactive=not mu)
 
-                        eval_steps = gr.Number(label='Evaluate every n steps', value=100, info='If an evaluation dataset is given, test it every time this many steps pass.')
+                        stride_length = gr.Slider(label='Stride Length', minimum=0, maximum=2048, value=256, step=32, info='Overlap between chunks in tokens. 0 = no overlap. Values like 256 or 512 help preserve context across chunk boundaries.')
 
-                    with gr.Tab(label="Text Dataset"):
-                        with gr.Row():
-                            text_dataset = gr.Dropdown(choices=utils.get_datasets('user_data/training/datasets', 'json'), value='None', label='Text Dataset', info='A JSONL file with a "text" key per row, for pretraining-style training. Each row is one document.', elem_classes=['slim-dropdown'], interactive=not mu)
-                            ui.create_refresh_button(text_dataset, lambda: None, lambda: {'choices': utils.get_datasets('user_data/training/datasets', 'json')}, 'refresh-button', interactive=not mu)
+                    with gr.Row():
+                        eval_dataset = gr.Dropdown(choices=utils.get_datasets('user_data/training/datasets', 'json'), value='None', label='Evaluation Dataset', info='The (optional) dataset file used to evaluate the model after training.', elem_classes=['slim-dropdown'], interactive=not mu)
+                        ui.create_refresh_button(eval_dataset, lambda: None, lambda: {'choices': utils.get_datasets('user_data/training/datasets', 'json')}, 'refresh-button', interactive=not mu)
 
-                        stride_length = gr.Slider(label='Stride Length', minimum=0, maximum=2048, value=0, step=32, info='Overlap between chunks in tokens. 0 = no overlap. Values like 256 or 512 help preserve context across chunk boundaries.')
+                    eval_steps = gr.Number(label='Evaluate every n steps', value=100, info='If an evaluation dataset is given, test it every time this many steps pass.')
 
                     with gr.Row():
                         start_button = gr.Button("Start LoRA Training", variant='primary', interactive=not mu)
@@ -406,35 +406,28 @@ def tokenize_conversation(data_point):
     train_template.clear()
 
     # == Prep the dataset, format, etc ==
-    if text_dataset not in ['None', '']:
-        train_template["template_type"] = "text_dataset"
-        logger.info("Loading text dataset")
-        data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{text_dataset}.json'))
-
-        # Validate the first row has a "text" key
-        if "text" not in data['train'].column_names:
-            yield "Error: text dataset must have a \"text\" key per row."
-            return
+    has_text_dataset = text_dataset not in ['None', '']
+    has_chat_dataset = dataset not in ['None', '']
+    if has_text_dataset and has_chat_dataset:
+        yield "Error: select either a Chat Dataset or a Text Dataset, not both."
+        return
 
-        # Tokenize each document and concatenate
+    def tokenize_text_data(data):
+        """Tokenize text dataset rows, concatenate, and split into chunks."""
         all_tokens = []
-        for row in data['train']:
+        for row in data:
             tokens = shared.tokenizer.encode(row['text'])
             if add_eos_token:
                 tokens.append(shared.tokenizer.eos_token_id)
             all_tokens.extend(tokens)
 
-        # Split into chunks with optional overlap (stride)
         stride = int(stride_length)
         step = cutoff_len - stride if stride > 0 else cutoff_len
 
         if step <= 0:
-            yield "Error: stride length must be smaller than cutoff length."
-            return
-
+            return None, "Error: stride length must be smaller than cutoff length."
         if len(all_tokens) < cutoff_len:
-            yield "Error: dataset is too short to fill even one chunk of the given cutoff length."
-            return
+            return None, "Error: dataset is too short to fill even one chunk of the given cutoff length."
 
         chunks = []
         for start in range(0, len(all_tokens), step):
@@ -442,7 +435,6 @@ def tokenize_conversation(data_point):
             if len(chunk) == 0:
                 break
             if len(chunk) < cutoff_len:
-                # Pad the remainder
                 pad_len = cutoff_len - len(chunk)
                 chunks.append({
                     "input_ids": chunk + [shared.tokenizer.pad_token_id] * pad_len,
@@ -456,14 +448,34 @@ def tokenize_conversation(data_point):
                     "attention_mask": [1] * cutoff_len,
                 })
 
-        train_data = Dataset.from_list(chunks)
-        del all_tokens
-        eval_data = None
-    else:
-        if dataset in ['None', '']:
-            yield "Missing dataset choice input, cannot continue."
+        return Dataset.from_list(chunks), None
+
+    if has_text_dataset:
+        train_template["template_type"] = "text_dataset"
+        logger.info("Loading text dataset")
+        data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{text_dataset}.json'))
+
+        if "text" not in data['train'].column_names:
+            yield "Error: text dataset must have a \"text\" key per row."
             return
 
+        train_data, err = tokenize_text_data(data['train'])
+        if err:
+            yield err
+            return
+
+        if eval_dataset == 'None':
+            eval_data = None
+        else:
+            eval_raw = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{eval_dataset}.json'))
+            if "text" not in eval_raw['train'].column_names:
+                yield "Error: evaluation dataset must have a \"text\" key per row."
+                return
+            eval_data, err = tokenize_text_data(eval_raw['train'])
+            if err:
+                yield err
+                return
+    elif has_chat_dataset:
         if format in ['None', '']:
             yield "Missing format choice input, cannot continue."
             return
@@ -517,6 +529,9 @@ def tokenize_conversation(data_point):
                 new_fingerprint='%030x' % random.randrange(16**30)
             )
             eval_data = eval_data.filter(lambda x: len(x['input_ids']) > 0)
+    else:
+        yield "No dataset selected. Choose a Chat Dataset or a Text Dataset."
+        return
 
     # == We MUST reload model if it went through any previous training, even failed one ==
     if shared.model_dirty_from_training:
diff --git a/modules/utils.py b/modules/utils.py
index 1490019c5e..456b5f8375 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -268,6 +268,39 @@ def get_datasets(path: str, ext: str):
     return ['None'] + sorted(set([k.stem for k in Path(path).glob(f'*.{ext}') if k.stem != 'put-trainer-datasets-here']), key=natural_keys)
 
 
+def get_chat_datasets(path: str):
+    """List JSON datasets that contain chat conversations (messages or ShareGPT format)."""
+    return ['None'] + sorted(set([k.stem for k in Path(path).glob('*.json') if k.stem != 'put-trainer-datasets-here' and _is_chat_dataset(k)]), key=natural_keys)
+
+
+def get_text_datasets(path: str):
+    """List JSON datasets that contain raw text ({"text": ...} format)."""
+    return ['None'] + sorted(set([k.stem for k in Path(path).glob('*.json') if k.stem != 'put-trainer-datasets-here' and _is_text_dataset(k)]), key=natural_keys)
+
+
+def _peek_json_keys(filepath):
+    """Read the first object in a JSON array file and return its keys."""
+    import json
+    try:
+        with open(filepath, 'r', encoding='utf-8') as f:
+            data = json.load(f)
+        if isinstance(data, list) and len(data) > 0 and isinstance(data[0], dict):
+            return set(data[0].keys())
+    except Exception:
+        pass
+    return set()
+
+
+def _is_chat_dataset(filepath):
+    keys = _peek_json_keys(filepath)
+    return bool(keys & {'messages', 'conversations'})
+
+
+def _is_text_dataset(filepath):
+    keys = _peek_json_keys(filepath)
+    return 'text' in keys
+
+
 def get_available_chat_styles():
     return sorted(set(('-'.join(k.stem.split('-')[1:]) for k in Path('css').glob('chat_style*.css'))), key=natural_keys)
 

From 409db3df1e25602a9b99896d34042d0a4b620c25 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 11:24:41 -0800
Subject: [PATCH 1230/1701] Training: Docs improvements

---
 docs/05 - Training Tab.md | 94 ++++++++++++---------------------------
 1 file changed, 29 insertions(+), 65 deletions(-)

diff --git a/docs/05 - Training Tab.md b/docs/05 - Training Tab.md
index 8ee7091776..902693e673 100644
--- a/docs/05 - Training Tab.md	
+++ b/docs/05 - Training Tab.md	
@@ -1,45 +1,30 @@
 ## Training Your Own LoRAs
 
-The WebUI seeks to make training your own LoRAs as easy as possible. It comes down to just a few simple steps:
-
-### **Step 1**: Make a plan.
-- What base model do you want to use? The LoRA you make has to be matched up to a single architecture (eg LLaMA-13B) and cannot be transferred to others (eg LLaMA-7B, StableLM, etc. would all be different). Derivatives of the same model (eg Alpaca finetune of LLaMA-13B) might be transferrable, but even then it's best to train exactly on what you plan to use.
-- What are you training it on? Do you want it to learn real information, a simple format, ...?
-
-### **Step 2**: Gather a dataset.
-- For instruction/chat training, prepare a JSON dataset in one of the [supported formats](#instruction-templates) (OpenAI messages or ShareGPT).
-- For pretraining-style training on raw text, use the `Text Dataset` tab with a JSON file where each row has a `"text"` key.
-- If you use a structured dataset not in this format, you may have to find an external way to convert it - or open an issue to request native support.
-
-### **Step 3**: Do the training.
-- **3.1**: Load the WebUI, and your model.
-    - Make sure you don't have any LoRAs already loaded (unless you want to train for multi-LoRA usage).
-- **3.2**: Open the `Training` tab at the top, `Train LoRA` sub-tab.
-- **3.3**: Fill in the name of the LoRA, select your dataset in the dataset options.
-- **3.4**: Select other parameters to your preference. See [parameters below](#parameters).
-- **3.5**: click `Start LoRA Training`, and wait.
-    - It can take a few hours for a large dataset, or just a few minute if doing a small run.
-    - You may want to monitor your [loss value](#loss) while it goes.
-
-### **Step 4**: Evaluate your results.
-- Load the LoRA under the Models Tab.
-- You can go test-drive it on the `Text generation` tab, or you can use the `Perplexity evaluation` sub-tab of the `Training` tab.
-- If you used the `Save every n steps` option, you can grab prior copies of the model from sub-folders within the LoRA model's folder and try them instead.
-
-### **Step 5**: Re-run if you're unhappy.
-- Make sure to unload the LoRA before training it.
-- You can simply resume a prior run - use `Copy parameters from` to select your LoRA, and edit parameters. Note that you cannot change the `Rank` of an already created LoRA.
-    - If you want to resume from a checkpoint saved along the way, simply copy the contents of the checkpoint folder into the LoRA's folder.
-    - (Note: `adapter_model.safetensors` or `adapter_model.bin` is the important file that holds the actual LoRA content).
-    - This will start Learning Rate and Steps back to the start. If you want to resume as if you were midway through, you can adjust your Learning Rate to the last reported LR in logs and reduce your epochs.
-- Or, you can start over entirely if you prefer.
-- If your model is producing corrupted outputs, you probably need to start over and use a lower Learning Rate.
-- If your model isn't learning detailed information but you want it to, you might need to just run more epochs, or you might need a higher Rank.
-- If your model is enforcing a format you didn't want, you may need to tweak your dataset, or start over and not train as far.
+A LoRA is tied to a specific model architecture — a LoRA trained on Llama 3 8B won't work on Mistral 7B. Train on the exact model you plan to use.
+
+### Quick Start
+
+1. Load your base model (no LoRAs loaded).
+2. Open the **Training** tab > **Train LoRA**.
+3. Pick a dataset and configure parameters (see [below](#parameters)).
+4. Click **Start LoRA Training** and monitor the [loss](#loss).
+5. When done, load the LoRA from the **Models** tab and test it.
+
+### Resuming Training
+
+To resume from a checkpoint, use the same LoRA name and uncheck `Override Existing Files`. If checkpoints exist (from `Save every n steps`), training will automatically resume from the latest one with full optimizer and scheduler state preserved. Note that you cannot change the `Rank` of an already created LoRA.
+
+You should also use `Copy parameters from` to restore the UI settings (learning rate, epochs, etc.) from the previous run, so that training continues with the same configuration.
+
+### Troubleshooting
+
+- **Corrupted outputs**: Start over with a lower Learning Rate.
+- **Not learning enough**: Run more epochs, or increase the Rank.
+- **Unwanted formatting**: Tweak your dataset, or train for fewer steps.
 
 ## Instruction Templates
 
-All instruction/chat training uses `apply_chat_template()` with Jinja2 templates. You have two options in the **Data Format** dropdown:
+All instruction/chat training uses `apply_chat_template()` with Jinja2 templates. You have two options in the **Instruction Template** dropdown:
 
 - **Chat Template**: Uses the model's built-in chat template from its tokenizer. Works with instruct/chat models that ship with a chat template (Llama 3, Qwen, Mistral, etc.).
 - **Named template** (e.g. ChatML, Alpaca, Llama-v3, etc.): Loads a Jinja2 template from `user_data/instruction-templates/`. This is useful for base models that don't have a built-in template, or when you want to override the model's default template.
@@ -109,49 +94,28 @@ If you uncheck it, you can manually select individual projection modules (`q_pro
 
 ## Parameters
 
-The basic purpose and function of each parameter is documented on-page in the WebUI, so read through them in the UI to understand your options.
-
-That said, here's a guide to the most important parameter choices you should consider:
+Each parameter has a description in the UI. Below is guidance on the most important choices.
 
 ### VRAM
 
-- First, you must consider your VRAM availability.
-    - Generally, under default settings, VRAM usage for training with default parameters is very close to when generating text (with 1000+ tokens of context) (ie, if you can generate text, you can train LoRAs).
-        - Note: VRAM usage is higher when training 4-bit quantized models. Reduce `Micro Batch Size` to `1` to compensate.
-    - If you have VRAM to spare, setting higher batch sizes will use more VRAM and get you better quality training in exchange.
-    - If you have large data, setting a higher cutoff length may be beneficial, but will cost significant VRAM. If you can spare some, set your batch size to `1` and see how high you can push your cutoff length.
-    - If you're low on VRAM, reducing batch size or cutoff length will of course improve that.
-    - Don't be afraid to just try it and see what happens. If it's too much, it will just error out, and you can lower settings and try again.
+VRAM usage during training is roughly similar to inference with ~1000 tokens of context. If you can run the model, you can probably train LoRAs with the default settings. If you run out of VRAM, reduce `Micro Batch Size` or `Cutoff Length`. Training 4-bit quantized models uses more VRAM — set `Micro Batch Size` to `1` to compensate.
 
 ### Rank
 
-- Second, you want to consider the amount of learning you want.
-    - For example, you may wish to just learn a dialogue format (as in the case of Alpaca) in which case setting a low `Rank` value (32 or lower) works great.
-    - Or, you might be training on project documentation you want the bot to understand and be able to understand questions about, in which case the higher the rank, the better.
-    - Generally, higher Rank = more precise learning = more total content learned = more VRAM usage while training.
+Higher rank = more learning capacity = larger adapter = more VRAM. Use 4–8 for style/format, 128–256 to teach factual knowledge.
 
 ### Learning Rate and Epochs
 
-- Third, how carefully you want it to be learned.
-    - In other words, how okay or not you are with the model losing unrelated understandings.
-    - You can control this with 3 key settings: the Learning Rate, its scheduler, and your total epochs.
-    - The learning rate controls how much change is made to the model by each token it sees.
-        - It's in scientific notation normally, so for example `3e-4` means `3 * 10^-4` which is `0.0003`. The number after `e-` controls how many `0`s are in the number.
-        - Higher values let training run faster, but also are more likely to corrupt prior data in the model.
-    - You essentially have two variables to balance: the LR, and Epochs.
-        - If you make LR higher, you can set Epochs equally lower to match. High LR + low epochs = very fast, low quality training.
-        - If you make LR low, set epochs high. Low LR + high epochs = slow but high-quality training.
-    - The scheduler controls change-over-time as you train - it starts high, and then goes low. This helps balance getting data in, and having decent quality, at the same time.
-        - You can see graphs of the different scheduler options [in the HuggingFace docs here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_1/en/main_classes/optimizer_schedules#transformers.SchedulerType)
+These control how aggressively the model learns and how many times it sees the data. Higher LR + fewer epochs = fast but rough. Lower LR + more epochs = slower but higher quality. The scheduler (default: cosine) decays the LR over the course of training — see [HuggingFace docs](https://huggingface.co/docs/transformers/main_classes/optimizer_schedules#schedules) for graphs of each option.
 
 ## Loss
 
 When you're running training, the WebUI's console window will log reports that include, among other things, a numeric value named `Loss`. It will start as a high number, and gradually get lower and lower as it goes.
 
-"Loss" in the world of AI training theoretically means "how close is the model to perfect", with `0` meaning "absolutely perfect". This is calculated by measuring the difference between the model outputting exactly the text you're training it to output, and what it actually outputs.
+Loss measures how far the model's predictions are from the training data, with `0` meaning a perfect match. It's calculated as the cross-entropy between the model's output distribution and the expected tokens.
 
-In practice, a good LLM should have a very complex variable range of ideas running in its artificial head, so a loss of `0` would indicate that the model has broken and forgotten how to think about anything other than what you trained it on.
+In practice, a loss of `0` means the model has overfit — it memorized the training data at the expense of its general capabilities.
 
-So, in effect, Loss is a balancing game: you want to get it low enough that it understands your data, but high enough that it isn't forgetting everything else. Generally, if it goes below `1.0`, it's going to start forgetting its prior memories, and you should stop training. In some cases you may prefer to take it as low as `0.5` (if you want it to be very very predictable). Different goals have different needs, so don't be afraid to experiment and see what works best for you.
+Loss is a balancing game: you want it low enough that the model learns your data, but not so low that it loses general knowledge. Generally, if it goes below `1.0`, overfitting is likely and you should stop training. In some cases you may want to go as low as `0.5` (if you need very predictable outputs). Different goals have different needs, so experiment and see what works best for you.
 
 Note: if you see Loss start at or suddenly jump to exactly `0`, it is likely something has gone wrong in your training process (eg model corruption).

From 134ac8fc292fb5e8f8909c911f2a2bcaf2f66f54 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 12:30:28 -0800
Subject: [PATCH 1231/1701] Update README

---
 README.md | 18 ++++++++----------
 1 file changed, 8 insertions(+), 10 deletions(-)

diff --git a/README.md b/README.md
index a2ba87f8f5..91179e0927 100644
--- a/README.md
+++ b/README.md
@@ -13,7 +13,7 @@
 
 # Text Generation Web UI
 
-Run AI chatbots like ChatGPT on your own computer. **100% private and offline** – no subscriptions, no API fees, zero telemetry. Just download, unzip, and run.
+A Gradio web UI for running Large Language Models locally. 100% private, offline, and free.
 
 [Try the Deep Reason extension](https://oobabooga.gumroad.com/l/deep_reason)
 
@@ -23,26 +23,24 @@ Run AI chatbots like ChatGPT on your own computer. **100% private and offline**
 
 ## Features
 
+- Supports multiple local text generation backends, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).
+- Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory.
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
+- `instruct` mode for instruction-following (like ChatGPT), and `chat-instruct`/`chat` modes for talking to custom characters. Prompts are automatically formatted with Jinja2 templates.
 - **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
 - **Vision (multimodal models)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
 - **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
 - **Web search**: Optionally search the internet with LLM-generated queries to add context to the conversation.
-- Aesthetic UI with dark/light themes, syntax highlighting, and LaTeX rendering.
+- **Training**: Fine-tune LoRAs on multi-turn chat or raw text datasets. Easy to use, good defaults, and supports resuming interrupted runs ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/05-%E2%80%90-Training-Tab)).
 - Edit messages, navigate between message versions, and branch conversations at any point.
-- Switch between models without restarting, with automatic GPU layer allocation.
+- Switch between different models in the UI without restarting.
 - Free-form text generation in the Notebook tab without being limited to chat turns.
-- `instruct` mode for instruction-following (like ChatGPT), and `chat-instruct`/`chat` modes for talking to custom characters.
-- Automatic prompt formatting using Jinja2 templates. You don't need to ever worry about prompt formats.
 - Multiple sampling parameters and generation options for sophisticated text generation control.
-- Supports multiple backends including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM).
+- Aesthetic UI with dark and light themes.
+- Syntax highlighting for code blocks and LaTeX rendering for mathematical expressions.
 - OpenAI-compatible API with Chat and Completions endpoints, including tool-calling support – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples).
 - Extension support, with numerous built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details.
 
-## 🔥 News
-
-- The project now supports **image generation**! Including Z-Image-Turbo, 4bit/8bit quantization, `torch.compile`, and LLM-generated prompt variations ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
-
 ## How to install
 
 #### ✅ Option 1: Portable builds (get started in 1 minute)

From 2f08dce7b0a85c95a8b7bf61cbdd0cedcdea3804 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 13:57:21 -0800
Subject: [PATCH 1232/1701] Remove ExLlamaV2 backend

- archived upstream: https://github.com/turboderp-org/exllamav2/commit/7dc12af3a81f34ac3f27cd7602ed539b638933ca
- replaced by ExLlamaV3, which has much better quantization accuracy
---
 README.md                              |  22 +--
 docs/03 - Parameters Tab.md            |   2 +-
 docs/04 - Model Tab.md                 |  29 +--
 docs/What Works.md                     |   2 -
 modules/LoRA.py                        |  31 +---
 modules/evaluate.py                    |   4 -
 modules/exllamav2.py                   | 247 -------------------------
 modules/exllamav2_hf.py                | 203 --------------------
 modules/exllamav3.py                   |   2 +-
 modules/loaders.py                     | 117 ------------
 modules/logits.py                      |   8 +-
 modules/models.py                      |  17 --
 modules/models_settings.py             |   6 +-
 modules/shared.py                      |  20 +-
 modules/text_generation.py             |   8 +-
 modules/ui.py                          |   5 -
 modules/ui_model_menu.py               |   7 +-
 requirements/full/requirements.txt     |   3 -
 requirements/full/requirements_amd.txt |   2 -
 19 files changed, 22 insertions(+), 713 deletions(-)
 delete mode 100644 modules/exllamav2.py
 delete mode 100644 modules/exllamav2_hf.py

diff --git a/README.md b/README.md
index 91179e0927..a47a0e6f7d 100644
--- a/README.md
+++ b/README.md
@@ -23,7 +23,7 @@ A Gradio web UI for running Large Language Models locally. 100% private, offline
 
 ## Features
 
-- Supports multiple local text generation backends, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), [ExLlamaV2](https://github.com/turboderp-org/exllamav2), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).
+- Supports multiple local text generation backends, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).
 - Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory.
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - `instruct` mode for instruction-following (like ChatGPT), and `chat-instruct`/`chat` modes for talking to custom characters. Prompts are automatically formatted with Jinja2 templates.
@@ -245,7 +245,7 @@ usage: server.py [-h] [--multi-user] [--model MODEL] [--lora LORA [LORA ...]] [-
                  [--row-split] [--no-mmap] [--mlock] [--no-kv-offload] [--batch-size BATCH_SIZE] [--ubatch-size UBATCH_SIZE] [--threads THREADS] [--threads-batch THREADS_BATCH] [--numa]
                  [--extra-flags EXTRA_FLAGS] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code]
                  [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE]
-                 [--enable-tp] [--tp-backend TP_BACKEND] [--gpu-split GPU_SPLIT] [--autosplit] [--cfg-cache] [--no_flash_attn] [--no_xformers] [--no_sdpa] [--num_experts_per_token N] [--cpp-runner]
+                 [--gpu-split GPU_SPLIT] [--enable-tp] [--tp-backend TP_BACKEND] [--cfg-cache] [--cpp-runner]
                  [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share]
                  [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors]
                  [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4]
@@ -280,13 +280,12 @@ Image model:
                                                        Quantization method for image model.
 
 Model loader:
-  --loader LOADER                                      Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF,
-                                                       ExLlamav2, TensorRT-LLM.
+  --loader LOADER                                      Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav3,
+                                                       TensorRT-LLM.
 
 Context and cache:
   --ctx-size N, --n_ctx N, --max_seq_len N             Context size in tokens. llama.cpp: 0 = auto if gpu-layers is also -1.
-  --cache-type N, --cache_type N                       KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and
-                                                       v_bits separately, e.g. q4_q8).
+  --cache-type N, --cache_type N                       KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).
 
 Speculative decoding:
   --model-draft MODEL_DRAFT                            Path to the draft model for speculative decoding.
@@ -337,17 +336,10 @@ bitsandbytes 4-bit:
   --quant_type QUANT_TYPE                              quant_type for 4-bit. Valid options: nf4, fp4.
 
 ExLlamaV3:
+  --gpu-split GPU_SPLIT                                Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.
   --enable-tp, --enable_tp                             Enable Tensor Parallelism (TP) to split the model across GPUs.
   --tp-backend TP_BACKEND                              The backend for tensor parallelism. Valid options: native, nccl. Default: native.
-
-ExLlamaV2:
-  --gpu-split GPU_SPLIT                                Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.
-  --autosplit                                          Autosplit the model tensors across the available GPUs. This causes --gpu-split to be ignored.
-  --cfg-cache                                          ExLlamav2_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.
-  --no_flash_attn                                      Force flash-attention to not be used.
-  --no_xformers                                        Force xformers to not be used.
-  --no_sdpa                                            Force Torch SDPA to not be used.
-  --num_experts_per_token N                            Number of experts to use for generation. Applies to MoE models like Mixtral.
+  --cfg-cache                                          Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.
 
 TensorRT-LLM:
   --cpp-runner                                         Use the ModelRunnerCpp runner, which is faster than the default ModelRunner.
diff --git a/docs/03 - Parameters Tab.md b/docs/03 - Parameters Tab.md
index 517950548e..13014affeb 100644
--- a/docs/03 - Parameters Tab.md	
+++ b/docs/03 - Parameters Tab.md	
@@ -69,7 +69,7 @@ For more information about the parameters, the [transformers documentation](http
 * **smoothing_curve**: Adjusts the dropoff curve of Quadratic Sampling. Higher values make the curve steeper. Only takes effect when smoothing_factor is set.
 * **temperature_last**: Makes temperature the last sampler instead of the first. With this, you can remove low probability tokens with a sampler like min_p and then use a high temperature to make the model creative without losing coherency. Note: this parameter takes precedence over "Sampler priority". That means that `temperature`/`dynamic_temperature`/`quadratic_sampling` will be removed from wherever they are and moved to the end of the stack.
 * **do_sample**: When unchecked, sampling is entirely disabled, and greedy decoding is used instead (the most likely token is always picked).
-* **Seed**: Set the Pytorch seed to this number. Note that some loaders do not use Pytorch (notably llama.cpp), and others are not deterministic (ExLlamaV2). For these loaders, the seed has no effect.
+* **Seed**: Set the Pytorch seed to this number. Note that some loaders do not use Pytorch (notably llama.cpp). For these loaders, the seed has no effect.
 * **encoder_repetition_penalty**: Also known as the "Hallucinations filter". Used to penalize tokens that are *not* in the prior text. Higher value = more likely to stay in context, lower value = more likely to diverge.
 * **no_repeat_ngram_size**: If not set to 0, specifies the length of token sets that are completely blocked from repeating at all. Higher values = blocks larger phrases, lower values = blocks words or letters from repeating. Only 0 or high values are a good idea in most cases.
 
diff --git a/docs/04 - Model Tab.md b/docs/04 - Model Tab.md
index 30b94151a7..4d5ae64507 100644
--- a/docs/04 - Model Tab.md	
+++ b/docs/04 - Model Tab.md	
@@ -29,7 +29,7 @@ Example: https://huggingface.co/TheBloke/Llama-2-7b-Chat-GGUF
 
 ### Transformers
 
-Loads: full precision (16-bit or 32-bit) models, as well as bitsandbytes-quantized models. The repository usually has a clean name without GGUF, EXL2, or EXL3 in its name, and the model files are named `model.safetensors` or split into parts like `model-00001-of-00004.safetensors`.
+Loads: full precision (16-bit or 32-bit) models, as well as bitsandbytes-quantized models. The repository usually has a clean name without GGUF or EXL3 in its name, and the model files are named `model.safetensors` or split into parts like `model-00001-of-00004.safetensors`.
 
 Example: [https://huggingface.co/lmsys/vicuna-7b-v1.5](https://huggingface.co/lmsys/vicuna-7b-v1.5).
 
@@ -78,33 +78,6 @@ The same as ExLlamav3_HF but using the internal samplers of ExLlamaV3 instead of
 * **enable_tp**: Enable Tensor Parallelism (TP) to split the model across GPUs.
 * **tp_backend**: The backend for tensor parallelism. Valid options: `native`, `nccl`. Default: `native`.
 
-### ExLlamav2_HF
-
-Loads: GPTQ and EXL2 models. EXL2 models usually have "EXL2" in the model name, while GPTQ models usually have GPTQ in the model name, or alternatively something like "-4bit-128g" in the name.
-
-Examples:
-
-* https://huggingface.co/turboderp/Llama2-70B-exl2
-* https://huggingface.co/TheBloke/Llama-2-13B-chat-GPTQ
-
-* **ctx_size**: Context length of the model. In ExLlamaV2, the cache is preallocated, so the higher this value, the higher the VRAM. It is automatically set to the maximum sequence length for the model based on its metadata, but you may need to lower this value to fit the model into your GPU. After loading the model, the "Truncate the prompt up to this length" parameter under "Parameters" > "Generation" is automatically set to your chosen "ctx_size" so that you don't have to set the same thing twice.
-* **cache_type**: KV cache quantization type. Valid options: `fp16`, `fp8`, `q8`, `q6`, `q4`. Lower quantization saves VRAM at the cost of some quality.
-* **gpu_split**: Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: `20,7,7`. Make sure to set a lower value for the first GPU, as that's where the cache is allocated.
-* **alpha_value**: Used to extend the context length of a model with a minor loss in quality. Same as the Transformers parameter.
-* **compress_pos_emb**: Positional embeddings compression factor. Same as the Transformers parameter.
-* **num_experts_per_token**: Number of experts to use for generation. Applies to MoE models like Mixtral.
-* **autosplit**: Autosplit the model tensors across the available GPUs. This causes gpu_split to be ignored.
-* **enable_tp**: Enable Tensor Parallelism (TP) to split the model across GPUs.
-* **no_flash_attn**: Disables flash attention. Otherwise, it is automatically used as long as the library is installed.
-* **no_xformers**: Force xformers to not be used.
-* **no_sdpa**: Force Torch SDPA to not be used.
-* **cfg_cache**: Creates a second cache to hold the CFG negative prompts. You need to set this if and only if you intend to use CFG in the "Parameters" > "Generation" tab. Checking this parameter doubles the cache VRAM usage.
-* **no_use_fast**: Do not use the "fast" version of the tokenizer.
-
-### ExLlamav2
-
-The same as ExLlamav2_HF but using the internal samplers of ExLlamaV2 instead of the ones in the Transformers library. Supports speculative decoding with a draft model.
-
 ### TensorRT-LLM
 
 Loads: TensorRT-LLM engine models. These are highly optimized models compiled specifically for NVIDIA GPUs.
diff --git a/docs/What Works.md b/docs/What Works.md
index d9c8530073..b10059bbf1 100644
--- a/docs/What Works.md	
+++ b/docs/What Works.md	
@@ -6,8 +6,6 @@
 | Transformers   |      ✅       |       ✅       |    ✅\*\*  |           ✅          |
 | ExLlamav3_HF   |      ❌       |       ❌       |    ❌      |           ✅          |
 | ExLlamav3      |      ❌       |       ❌       |    ✅      |           ❌          |
-| ExLlamav2_HF   |      ✅       |       ❌       |    ❌      |           ✅          |
-| ExLlamav2      |      ✅       |       ❌       |    ❌      |           ❌          |
 | TensorRT-LLM   |      ❌       |       ❌       |    ❌      |           ❌          |
 
 ❌ = not supported
diff --git a/modules/LoRA.py b/modules/LoRA.py
index 63d764b4ad..6f1367a8ec 100644
--- a/modules/LoRA.py
+++ b/modules/LoRA.py
@@ -5,10 +5,7 @@
 
 
 def add_lora_to_model(lora_names):
-    if shared.model.__class__.__name__ in ['Exllamav2Model', 'Exllamav2HF'] or shared.args.loader in ['ExLlamav2', 'ExLlamav2_HF']:
-        add_lora_exllamav2(lora_names)
-    else:
-        add_lora_transformers(lora_names)
+    add_lora_transformers(lora_names)
 
 
 def get_lora_path(lora_name):
@@ -19,32 +16,6 @@ def get_lora_path(lora_name):
     return Path(f"{shared.args.lora_dir}/{lora_name}")
 
 
-def add_lora_exllamav2(lora_names):
-
-    from exllamav2 import ExLlamaV2Lora
-
-    if isinstance(shared.model.loras, list):
-        for lora in shared.model.loras:
-            lora.unload()
-
-    if len(lora_names) > 0:
-        logger.info("Applying the following LoRAs to {}: {}".format(shared.model_name, ', '.join(lora_names)))
-        shared.model.loras = []
-        for lora_name in lora_names:
-            lora_path = get_lora_path(lora_name)
-            if shared.model.__class__.__name__ == 'Exllamav2Model':
-                lora = ExLlamaV2Lora.from_directory(shared.model.model, str(lora_path))
-            else:
-                lora = ExLlamaV2Lora.from_directory(shared.model.ex_model, str(lora_path))
-
-            shared.model.loras.append(lora)
-
-        shared.lora_names = lora_names
-    else:
-        shared.lora_names = []
-        shared.model.loras = None
-
-
 def add_lora_transformers(lora_names):
     from peft import PeftModel
 
diff --git a/modules/evaluate.py b/modules/evaluate.py
index ccf72ec673..e3562d5ff8 100644
--- a/modules/evaluate.py
+++ b/modules/evaluate.py
@@ -46,10 +46,6 @@ def calculate_perplexity(models, input_dataset, stride, _max_length):
         logger.error("Perplexity evaluation is not implemented for the llama.cpp loader.")
         raise ValueError
 
-    if shared.args.loader == "ExLlamav2":
-        logger.error("ExLlamav2_HF is required for perplexity evaluation with EXL2 models. Please reload the model with ExLlamav2_HF instead of ExLlamav2.")
-        raise ValueError
-
     if not shared.args.no_use_fast:
         logger.warning("--no_use_fast is not set. If tokenizing the input dataset takes a long time, try reloading the model with that option set/checked.")
 
diff --git a/modules/exllamav2.py b/modules/exllamav2.py
deleted file mode 100644
index 5d5c5b567c..0000000000
--- a/modules/exllamav2.py
+++ /dev/null
@@ -1,247 +0,0 @@
-import json
-import traceback
-from pathlib import Path
-
-import torch
-from exllamav2 import (
-    ExLlamaV2,
-    ExLlamaV2Cache,
-    ExLlamaV2Cache_8bit,
-    ExLlamaV2Cache_Q4,
-    ExLlamaV2Cache_Q6,
-    ExLlamaV2Cache_Q8,
-    ExLlamaV2Cache_TP,
-    ExLlamaV2Config,
-    ExLlamaV2Tokenizer
-)
-from exllamav2.generator import ExLlamaV2Sampler, ExLlamaV2StreamingGenerator
-
-from modules import shared
-from modules.logging_colors import logger
-from modules.text_generation import get_max_prompt_length
-
-try:
-    import flash_attn
-except Exception:
-    logger.warning('Failed to load flash-attention due to the following error:\n')
-    traceback.print_exc()
-
-
-class Exllamav2Model:
-    def __init__(self):
-        pass
-
-    @classmethod
-    def from_pretrained(self, path_to_model):
-
-        path_to_model = Path(f'{shared.args.model_dir}') / Path(path_to_model)
-
-        config = ExLlamaV2Config()
-        config.model_dir = str(path_to_model)
-        config.prepare()
-
-        config.max_seq_len = shared.args.ctx_size
-        config.scale_pos_emb = shared.args.compress_pos_emb
-        config.scale_alpha_value = shared.args.alpha_value
-        config.no_flash_attn = shared.args.no_flash_attn
-        config.no_xformers = shared.args.no_xformers
-        config.no_sdpa = shared.args.no_sdpa
-        config.num_experts_per_token = int(shared.args.num_experts_per_token)
-
-        model = ExLlamaV2(config)
-
-        split = None
-        if shared.args.gpu_split:
-            split = [float(alloc) for alloc in shared.args.gpu_split.split(",")]
-
-        if shared.args.enable_tp:
-            model.load_tp(split)
-        elif not shared.args.autosplit:
-            model.load(split)
-
-        # Determine the correct cache type
-        kv_cache_type = shared.args.cache_type.lower()
-
-        if kv_cache_type == 'fp16':
-            cache_type = ExLlamaV2Cache
-        elif kv_cache_type == 'fp8':
-            cache_type = ExLlamaV2Cache_8bit
-        elif kv_cache_type == 'q8':
-            cache_type = ExLlamaV2Cache_Q8
-        elif kv_cache_type == 'q6':
-            cache_type = ExLlamaV2Cache_Q6
-        elif kv_cache_type == 'q4':
-            cache_type = ExLlamaV2Cache_Q4
-        else:
-            raise ValueError(f"Invalid cache type for ExLlamaV2: {cache_type}. Valid options are: fp16, fp8, q8, q6, q4.")
-
-        # Use TP if specified
-        if shared.args.enable_tp:
-            cache = ExLlamaV2Cache_TP(model, base=cache_type)
-        else:
-            cache = cache_type(model, lazy=shared.args.autosplit)
-
-        if shared.args.autosplit and not shared.args.enable_tp:
-            model.load_autosplit(cache)
-
-        tokenizer = ExLlamaV2Tokenizer(config)
-
-        # Initialize draft model for speculative decoding
-        draft_model = None
-        draft_cache = None
-
-        if shared.args.model_draft and shared.args.model_draft.lower() not in ["none", ""]:
-            logger.info(f"Loading draft model for speculative decoding: {shared.args.model_draft}")
-
-            # Find the draft model path
-            draft_path = Path(shared.args.model_draft)
-            if not draft_path.exists():
-                draft_path = Path(f'{shared.args.model_dir}') / Path(shared.args.model_draft)
-
-            draft_config = ExLlamaV2Config()
-            draft_config.model_dir = str(draft_path)
-            draft_config.prepare()
-            draft_config.arch_compat_overrides()
-
-            # Set context size for draft model
-            if shared.args.ctx_size_draft > 0:
-                draft_config.max_seq_len = shared.args.ctx_size_draft
-            else:
-                draft_config.max_seq_len = config.max_seq_len
-
-            draft_model = ExLlamaV2(draft_config)
-            draft_cache = cache_type(draft_model, lazy=True)
-            draft_model.load_autosplit(draft_cache)
-
-            logger.info(f"Draft model loaded successfully with max_draft={shared.args.draft_max}")
-
-        generator = ExLlamaV2StreamingGenerator(
-            model,
-            cache,
-            tokenizer,
-            draft_model=draft_model,
-            draft_cache=draft_cache,
-            num_speculative_tokens=shared.args.draft_max if draft_model is not None else 0
-        )
-
-        result = self()
-        result.model = model
-        result.cache = cache
-        result.tokenizer = tokenizer
-        result.generator = generator
-        result.loras = None
-        result.draft_model = draft_model
-        result.draft_cache = draft_cache
-        return result, result
-
-    def encode(self, string, **kwargs):
-        add_bos = kwargs.pop('add_bos', True)
-        return self.tokenizer.encode(string, add_bos=add_bos, encode_special_tokens=True, **kwargs)
-
-    def decode(self, ids, **kwargs):
-        if isinstance(ids, list):
-            ids = torch.tensor([ids])
-        elif isinstance(ids, torch.Tensor) and ids.numel() == 1:
-            ids = ids.view(1, -1)
-
-        return self.tokenizer.decode(ids, decode_special_tokens=True)[0]
-
-    def get_logits(self, token_ids, **kwargs):
-        self.cache.current_seq_len = 0
-        if token_ids.shape[-1] > 1:
-            self.model.forward(token_ids[:, :-1], self.cache, input_mask=None, preprocess_only=True, loras=self.loras)
-
-        return self.model.forward(token_ids[:, -1:], self.cache, input_mask=None, loras=self.loras, **kwargs).float().cpu()
-
-    def generate_with_streaming(self, prompt, state):
-        settings = ExLlamaV2Sampler.Settings()
-
-        settings.token_repetition_penalty = state['repetition_penalty']
-        settings.token_repetition_range = -1 if state['repetition_penalty_range'] <= 0 else state['repetition_penalty_range']
-
-        settings.token_frequency_penalty = state['frequency_penalty']
-        settings.token_presence_penalty = state['presence_penalty']
-
-        settings.temperature = state['temperature']
-        settings.smoothing_factor = state['smoothing_factor']
-        settings.min_temp = state['dynatemp_low'] if state['dynamic_temperature'] else 0
-        settings.max_temp = state['dynatemp_high'] if state['dynamic_temperature'] else 0
-        settings.temp_exponent = state['dynatemp_exponent']
-        settings.top_k = state['top_k']
-        settings.top_p = state['top_p']
-        settings.top_a = state['top_a']
-        settings.min_p = state['min_p']
-        settings.tfs = state['tfs']
-        settings.typical = state['typical_p']
-
-        settings.temperature_last = state['temperature_last']
-
-        settings.mirostat = state['mirostat_mode'] == 2
-        settings.mirostat_tau = state['mirostat_tau']
-        settings.mirostat_eta = state['mirostat_eta']
-
-        if state['ban_eos_token']:
-            settings.disallow_tokens(self.tokenizer, [self.tokenizer.eos_token_id])
-
-        if state['custom_token_bans']:
-            to_ban = [int(x) for x in state['custom_token_bans'].split(',')]
-            if len(to_ban) > 0:
-                settings.disallow_tokens(self.tokenizer, to_ban)
-
-        settings.dry_allowed_length = state['dry_allowed_length']
-        settings.dry_base = state['dry_base']
-        settings.dry_multiplier = state['dry_multiplier']
-
-        # Dry sequence breakers processing
-        if state['dry_multiplier'] > 0 and state['dry_sequence_breakers']:
-            dry_sequence_breakers = state['dry_sequence_breakers']
-
-            # Support both JSON array notation and comma-separated strings.
-            if not dry_sequence_breakers.startswith("["):
-                dry_sequence_breakers = "[" + dry_sequence_breakers + "]"
-
-            sequence_breaker_strings = json.loads(dry_sequence_breakers)
-            # Prefix with 'a' to get the correct encoding of the token at the end of a text.
-            sequence_breakers = {
-                self.encode(f"a{s}")[0, -1].item() for s in sequence_breaker_strings
-            }
-
-            settings.dry_sequence_breakers = sequence_breakers
-
-        settings.xtc_probability = state['xtc_probability']
-        settings.xtc_threshold = state['xtc_threshold']
-
-        ids = self.tokenizer.encode(prompt, add_bos=state['add_bos_token'], encode_special_tokens=True)
-        ids = ids[:, -get_max_prompt_length(state):]
-
-        if state['auto_max_new_tokens']:
-            max_new_tokens = state['truncation_length'] - ids.shape[-1]
-        else:
-            max_new_tokens = state['max_new_tokens']
-
-        # Reset speculative decoding stats if using a draft model
-        if hasattr(self, 'draft_model') and self.draft_model is not None:
-            self.generator.reset_sd_stats()
-
-        self.generator.begin_stream(ids, settings, loras=self.loras)
-
-        decoded_text = ''
-        for i in range(max_new_tokens):
-            chunk, eos, _ = self.generator.stream()
-            if eos or shared.stop_everything:
-                break
-
-            decoded_text += chunk
-            yield decoded_text
-
-        # Log speculative decoding stats if using draft model
-        if hasattr(self, 'draft_model') and self.draft_model is not None:
-            efficiency, accuracy, total_tokens, total_draft_tokens, accepted_draft_tokens = self.generator.get_sd_stats()
-            logger.info(f"Speculative decoding: accepted={accepted_draft_tokens}/{total_draft_tokens} tokens")
-
-    def generate(self, prompt, state):
-        output = ''
-        for output in self.generate_with_streaming(prompt, state):
-            pass
-
-        return output
diff --git a/modules/exllamav2_hf.py b/modules/exllamav2_hf.py
deleted file mode 100644
index 4ba185902c..0000000000
--- a/modules/exllamav2_hf.py
+++ /dev/null
@@ -1,203 +0,0 @@
-import os
-import traceback
-from pathlib import Path
-from typing import Any, Dict, Optional, Union
-
-import torch
-from exllamav2 import (
-    ExLlamaV2,
-    ExLlamaV2Cache,
-    ExLlamaV2Cache_8bit,
-    ExLlamaV2Cache_Q4,
-    ExLlamaV2Cache_Q6,
-    ExLlamaV2Cache_Q8,
-    ExLlamaV2Cache_TP,
-    ExLlamaV2Config
-)
-from torch.nn import CrossEntropyLoss
-from transformers import (
-    GenerationConfig,
-    GenerationMixin,
-    PretrainedConfig,
-    PreTrainedModel
-)
-from transformers.modeling_outputs import CausalLMOutputWithPast
-
-from modules import shared
-from modules.logging_colors import logger
-
-try:
-    import flash_attn
-except Exception:
-    logger.warning('Failed to load flash-attention due to the following error:\n')
-    traceback.print_exc()
-
-
-class Exllamav2HF(PreTrainedModel, GenerationMixin):
-    def __init__(self, config: ExLlamaV2Config):
-        hf_config = PretrainedConfig.from_pretrained(config.model_dir)
-        super().__init__(hf_config)
-
-        self.ex_config = config
-        self.loras = None
-        self.generation_config = GenerationConfig()
-
-        self.ex_model = ExLlamaV2(config)
-
-        split = None
-        if shared.args.gpu_split:
-            split = [float(alloc) for alloc in shared.args.gpu_split.split(",")]
-
-        if shared.args.enable_tp:
-            self.ex_model.load_tp(split)
-        elif not shared.args.autosplit:
-            self.ex_model.load(split)
-
-        # Determine the correct cache type
-        kv_cache_type = shared.args.cache_type.lower()
-
-        if kv_cache_type == 'fp16':
-            cache_type = ExLlamaV2Cache
-        elif kv_cache_type == 'fp8':
-            cache_type = ExLlamaV2Cache_8bit
-        elif kv_cache_type == 'q8':
-            cache_type = ExLlamaV2Cache_Q8
-        elif kv_cache_type == 'q6':
-            cache_type = ExLlamaV2Cache_Q6
-        elif kv_cache_type == 'q4':
-            cache_type = ExLlamaV2Cache_Q4
-        else:
-            raise ValueError(f"Invalid cache type for ExLlamaV2: {kv_cache_type}. Valid options are: fp16, fp8, q8, q6, q4.")
-
-        # Use TP if specified
-        if shared.args.enable_tp:
-            self.ex_cache = ExLlamaV2Cache_TP(self.ex_model, base=cache_type)
-        else:
-            self.ex_cache = cache_type(self.ex_model, lazy=shared.args.autosplit)
-
-        if shared.args.autosplit and not shared.args.enable_tp:
-            self.ex_model.load_autosplit(self.ex_cache)
-
-        self.past_seq = None
-        if shared.args.cfg_cache:
-            if shared.args.enable_tp:
-                self.ex_cache_negative = ExLlamaV2Cache_TP(self.ex_model, base=cache_type)
-            else:
-                self.ex_cache_negative = cache_type(self.ex_model, lazy=shared.args.autosplit)
-
-            self.past_seq_negative = None
-
-    def _validate_model_class(self):
-        pass
-
-    def _validate_model_kwargs(self, model_kwargs: Dict[str, Any]):
-        pass
-
-    def prepare_inputs_for_generation(self, input_ids, **kwargs):
-        return {'input_ids': input_ids, **kwargs}
-
-    @property
-    def device(self) -> torch.device:
-        return torch.device(0)
-
-    def __call__(self, *args, **kwargs):
-        use_cache = kwargs.get('use_cache', True)
-        labels = kwargs.get('labels', None)
-        past_key_values = kwargs.get('past_key_values', None)
-
-        if len(args) > 0:
-            if not shared.args.cfg_cache:
-                logger.error("Please enable the cfg-cache option to use CFG with ExLlamav2_HF.")
-                return
-
-            input_ids = args[0]
-            is_negative = True
-            past_seq = self.past_seq_negative
-            ex_cache = self.ex_cache_negative
-        else:
-            input_ids = kwargs['input_ids']
-            is_negative = False
-            past_seq = self.past_seq
-            ex_cache = self.ex_cache
-
-        seq = input_ids[0].tolist()
-        if is_negative and past_key_values is not None:
-            seq = past_key_values + seq
-
-        seq_tensor = torch.tensor(seq)
-        reset = True
-
-        # Make the forward call
-        if labels is None:
-            if past_seq is not None:
-                min_length = min(past_seq.shape[0], seq_tensor.shape[0])
-                indices = torch.nonzero(~torch.eq(past_seq[:min_length], seq_tensor[:min_length]))
-                if len(indices) > 0:
-                    longest_prefix = indices[0].item()
-                else:
-                    longest_prefix = min_length
-
-                if longest_prefix > 0:
-                    reset = False
-                    ex_cache.current_seq_len = longest_prefix
-                    if len(seq_tensor) - longest_prefix > 1:
-                        self.ex_model.forward(seq_tensor[longest_prefix:-1].view(1, -1), ex_cache, preprocess_only=True, loras=self.loras)
-                    elif len(seq_tensor) == longest_prefix:
-                        # Very tricky: if the prefix we are reusing *is* the input_ids, then we have to back up the cache pointer by one,
-                        # because we feed input_ids[-1] to forward() below, but that last token is already in the cache!
-                        ex_cache.current_seq_len -= 1
-
-            if reset:
-                ex_cache.current_seq_len = 0
-                if len(seq_tensor) > 1:
-                    self.ex_model.forward(seq_tensor[:-1].view(1, -1), ex_cache, preprocess_only=True, loras=self.loras)
-
-            logits = self.ex_model.forward(seq_tensor[-1:].view(1, -1), ex_cache, loras=self.loras).to(input_ids.device).float()
-        else:
-            ex_cache.current_seq_len = 0
-            logits = self.ex_model.forward(seq_tensor.view(1, -1), ex_cache, last_id_only=False, loras=self.loras).float()
-
-        if is_negative:
-            self.past_seq_negative = seq_tensor
-        else:
-            self.past_seq = seq_tensor
-
-        if torch.cuda.is_available():
-            torch.cuda.synchronize()
-
-        loss = None
-        if labels is not None:
-            # Shift so that tokens < n predict n
-            shift_logits = logits[..., :-1, :].contiguous()
-            shift_labels = labels[..., 1:].contiguous()
-            # Flatten the tokens
-            loss_fct = CrossEntropyLoss()
-            shift_logits = shift_logits.view(-1, logits.shape[-1])
-            shift_labels = shift_labels.view(-1)
-            # Enable model parallelism
-            shift_labels = shift_labels.to(shift_logits.device)
-            loss = loss_fct(shift_logits, shift_labels)
-
-        return CausalLMOutputWithPast(logits=logits, past_key_values=seq if use_cache else None, loss=loss)
-
-    @classmethod
-    def from_pretrained(cls, pretrained_model_name_or_path: Optional[Union[str, os.PathLike]], *model_args, **kwargs):
-        assert len(model_args) == 0 and len(kwargs) == 0, "extra args is currently not supported"
-        if isinstance(pretrained_model_name_or_path, str):
-            pretrained_model_name_or_path = Path(pretrained_model_name_or_path)
-
-        pretrained_model_name_or_path = Path(f'{shared.args.model_dir}') / Path(pretrained_model_name_or_path)
-
-        config = ExLlamaV2Config()
-        config.model_dir = str(pretrained_model_name_or_path)
-        config.prepare()
-
-        config.max_seq_len = shared.args.ctx_size
-        config.scale_pos_emb = shared.args.compress_pos_emb
-        config.scale_alpha_value = shared.args.alpha_value
-        config.no_flash_attn = shared.args.no_flash_attn
-        config.no_xformers = shared.args.no_xformers
-        config.no_sdpa = shared.args.no_sdpa
-        config.num_experts_per_token = int(shared.args.num_experts_per_token)
-
-        return Exllamav2HF(config)
diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index ea1f3dc911..eca5dde0d9 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -59,7 +59,7 @@ def from_pretrained(cls, path_to_model):
             logger.warning(f"max_num_tokens must be a multiple of 256. Adjusting from {max_tokens} to {adjusted_tokens}")
             max_tokens = adjusted_tokens
 
-        # Parse cache type (ExLlamaV2 pattern)
+        # Parse cache type
         cache_type = shared.args.cache_type.lower()
         cache_kwargs = {}
         if cache_type == 'fp16':
diff --git a/modules/loaders.py b/modules/loaders.py
index 238b231e7f..9923f116e2 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -74,38 +74,6 @@
         'enable_tp',
         'tp_backend',
     ],
-    'ExLlamav2_HF': [
-        'ctx_size',
-        'cache_type',
-        'gpu_split',
-        'alpha_value',
-        'compress_pos_emb',
-        'num_experts_per_token',
-        'autosplit',
-        'enable_tp',
-        'no_flash_attn',
-        'no_xformers',
-        'no_sdpa',
-        'cfg_cache',
-        'no_use_fast',
-    ],
-    'ExLlamav2': [
-        'ctx_size',
-        'cache_type',
-        'gpu_split',
-        'alpha_value',
-        'compress_pos_emb',
-        'num_experts_per_token',
-        'autosplit',
-        'enable_tp',
-        'no_flash_attn',
-        'no_xformers',
-        'no_sdpa',
-        'model_draft',
-        'draft_max',
-        'ctx_size_draft',
-        'speculative_decoding_accordion',
-    ],
     'TensorRT-LLM': [
         'ctx_size',
         'cpp_runner',
@@ -222,56 +190,6 @@ def transformers_samplers():
         'grammar_string',
         'grammar_file_row',
     },
-    'ExLlamav2_HF': {
-        'temperature',
-        'dynatemp_low',
-        'dynatemp_high',
-        'dynatemp_exponent',
-        'smoothing_factor',
-        'smoothing_curve',
-        'min_p',
-        'top_p',
-        'top_k',
-        'typical_p',
-        'xtc_threshold',
-        'xtc_probability',
-        'epsilon_cutoff',
-        'eta_cutoff',
-        'tfs',
-        'top_a',
-        'top_n_sigma',
-        'adaptive_target',
-        'adaptive_decay',
-        'dry_multiplier',
-        'dry_allowed_length',
-        'dry_base',
-        'repetition_penalty',
-        'frequency_penalty',
-        'presence_penalty',
-        'encoder_repetition_penalty',
-        'no_repeat_ngram_size',
-        'repetition_penalty_range',
-        'guidance_scale',
-        'mirostat_mode',
-        'mirostat_tau',
-        'mirostat_eta',
-        'do_sample',
-        'dynamic_temperature',
-        'temperature_last',
-        'auto_max_new_tokens',
-        'ban_eos_token',
-        'add_bos_token',
-        'enable_thinking',
-        'reasoning_effort',
-        'skip_special_tokens',
-        'seed',
-        'sampler_priority',
-        'custom_token_bans',
-        'negative_prompt',
-        'dry_sequence_breakers',
-        'grammar_string',
-        'grammar_file_row',
-    },
     'ExLlamav3': {
         'temperature',
         'min_p',
@@ -292,41 +210,6 @@ def transformers_samplers():
         'seed',
         'skip_special_tokens',
     },
-    'ExLlamav2': {
-        'temperature',
-        'dynatemp_low',
-        'dynatemp_high',
-        'dynatemp_exponent',
-        'smoothing_factor',
-        'min_p',
-        'top_p',
-        'top_k',
-        'typical_p',
-        'xtc_threshold',
-        'xtc_probability',
-        'tfs',
-        'top_a',
-        'dry_multiplier',
-        'dry_allowed_length',
-        'dry_base',
-        'repetition_penalty',
-        'frequency_penalty',
-        'presence_penalty',
-        'repetition_penalty_range',
-        'mirostat_mode',
-        'mirostat_tau',
-        'mirostat_eta',
-        'dynamic_temperature',
-        'temperature_last',
-        'auto_max_new_tokens',
-        'ban_eos_token',
-        'add_bos_token',
-        'enable_thinking',
-        'reasoning_effort',
-        'skip_special_tokens',
-        'custom_token_bans',
-        'dry_sequence_breakers',
-    },
     'llama.cpp': {
         'temperature',
         'dynatemp_low',
diff --git a/modules/logits.py b/modules/logits.py
index d9e1ab48be..2d066c09c0 100644
--- a/modules/logits.py
+++ b/modules/logits.py
@@ -70,18 +70,12 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
         from modules import sampler_hijack
         from modules.torch_utils import get_device
 
-        is_non_hf_exllamav2 = shared.model.__class__.__name__ == 'Exllamav2Model'
         is_non_hf_exllamav3 = shared.model.__class__.__name__ == 'Exllamav3Model'
 
         if not use_samplers:
             state = {'stream': True}
 
         if use_samplers:
-            if is_non_hf_exllamav2:
-                # sampling is all done in C++ for exllama, so it is really hard to hijack
-                logger.error("Sampler hijacking is not supported non-Huggingface loaders.")
-                return 'Error: Sampler hijacking is not supported non-Huggingface loaders. Please disable the "Use samplers" option.', previous
-
             state['max_new_tokens'] = 1
             state['auto_max_new_tokens'] = False
             state.setdefault('stream', True)
@@ -90,7 +84,7 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
 
             scores = sampler_hijack.global_scores[-1]
         else:
-            if is_non_hf_exllamav2 or is_non_hf_exllamav3:
+            if is_non_hf_exllamav3:
                 device = get_device()
                 tokens = shared.tokenizer.encode(prompt)
                 if device:
diff --git a/modules/models.py b/modules/models.py
index 2dd69c63e6..9780b9f5b4 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -20,8 +20,6 @@ def load_model(model_name, loader=None):
         'Transformers': transformers_loader,
         'ExLlamav3_HF': ExLlamav3_HF_loader,
         'ExLlamav3': ExLlamav3_loader,
-        'ExLlamav2_HF': ExLlamav2_HF_loader,
-        'ExLlamav2': ExLlamav2_loader,
         'TensorRT-LLM': TensorRT_LLM_loader,
     }
 
@@ -109,19 +107,6 @@ def ExLlamav3_loader(model_name):
     return model, tokenizer
 
 
-def ExLlamav2_HF_loader(model_name):
-    from modules.exllamav2_hf import Exllamav2HF
-
-    return Exllamav2HF.from_pretrained(model_name)
-
-
-def ExLlamav2_loader(model_name):
-    from modules.exllamav2 import Exllamav2Model
-
-    model, tokenizer = Exllamav2Model.from_pretrained(model_name)
-    return model, tokenizer
-
-
 def TensorRT_LLM_loader(model_name):
     try:
         from modules.tensorrt_llm import TensorRTLLMModel
@@ -141,8 +126,6 @@ def unload_model(keep_model_name=False):
 
     if model_class_name in ['Exllamav3Model', 'Exllamav3HF']:
         shared.model.unload()
-    elif model_class_name in ['Exllamav2Model', 'Exllamav2HF'] and hasattr(shared.model, 'unload'):
-        shared.model.unload()
 
     shared.model = shared.tokenizer = None
     shared.lora_names = []
diff --git a/modules/models_settings.py b/modules/models_settings.py
index 1d784a8787..104635b901 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -216,12 +216,8 @@ def infer_loader(model_name, model_settings, hf_quant_method=None):
         loader = 'llama.cpp'
     elif hf_quant_method == 'exl3':
         loader = 'ExLlamav3'
-    elif hf_quant_method in ['exl2', 'gptq']:
-        loader = 'ExLlamav2_HF'
     elif re.match(r'.*exl3', model_name.lower()):
         loader = 'ExLlamav3'
-    elif re.match(r'.*exl2', model_name.lower()):
-        loader = 'ExLlamav2_HF'
     else:
         loader = 'Transformers'
 
@@ -255,7 +251,7 @@ def apply_model_settings_to_state(model, state):
     model_settings = get_model_metadata(model)
     if 'loader' in model_settings:
         loader = model_settings.pop('loader')
-        if not ((loader == 'ExLlamav2_HF' and state['loader'] == 'ExLlamav2') or (loader == 'ExLlamav3_HF' and state['loader'] == 'ExLlamav3')):
+        if not (loader == 'ExLlamav3_HF' and state['loader'] == 'ExLlamav3'):
             state['loader'] = loader
 
     for k in model_settings:
diff --git a/modules/shared.py b/modules/shared.py
index 8d617ecd0d..8b820e2df1 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -67,12 +67,12 @@
 
 # Model loader
 group = parser.add_argument_group('Model loader')
-group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav2_HF, ExLlamav2, TensorRT-LLM.')
+group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav3, TensorRT-LLM.')
 
 # Cache
 group = parser.add_argument_group('Context and cache')
 group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=8192, metavar='N', help='Context size in tokens. llama.cpp: 0 = auto if gpu-layers is also -1.')
-group.add_argument('--cache-type', '--cache_type', type=str, default='fp16', metavar='N', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).')
+group.add_argument('--cache-type', '--cache_type', type=str, default='fp16', metavar='N', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).')
 
 # Speculative decoding
 group = parser.add_argument_group('Speculative decoding')
@@ -127,18 +127,10 @@
 
 # ExLlamaV3
 group = parser.add_argument_group('ExLlamaV3')
+group.add_argument('--gpu-split', type=str, help='Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.')
 group.add_argument('--enable-tp', '--enable_tp', action='store_true', help='Enable Tensor Parallelism (TP) to split the model across GPUs.')
 group.add_argument('--tp-backend', type=str, default='native', help='The backend for tensor parallelism. Valid options: native, nccl. Default: native.')
-
-# ExLlamaV2
-group = parser.add_argument_group('ExLlamaV2')
-group.add_argument('--gpu-split', type=str, help='Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.')
-group.add_argument('--autosplit', action='store_true', help='Autosplit the model tensors across the available GPUs. This causes --gpu-split to be ignored.')
-group.add_argument('--cfg-cache', action='store_true', help='ExLlamav2_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.')
-group.add_argument('--no_flash_attn', action='store_true', help='Force flash-attention to not be used.')
-group.add_argument('--no_xformers', action='store_true', help='Force xformers to not be used.')
-group.add_argument('--no_sdpa', action='store_true', help='Force Torch SDPA to not be used.')
-group.add_argument('--num_experts_per_token', type=int, default=2, metavar='N', help='Number of experts to use for generation. Applies to MoE models like Mixtral.')
+group.add_argument('--cfg-cache', action='store_true', help='Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.')
 
 # TensorRT-LLM
 group = parser.add_argument_group('TensorRT-LLM')
@@ -379,10 +371,6 @@ def fix_loader_name(name):
         return 'llama.cpp'
     elif name in ['transformers', 'huggingface', 'hf', 'hugging_face', 'hugging face']:
         return 'Transformers'
-    elif name in ['exllamav2', 'exllama-v2', 'ex_llama-v2', 'exlamav2', 'exlama-v2', 'exllama2', 'exllama-2']:
-        return 'ExLlamav2'
-    elif name in ['exllamav2-hf', 'exllamav2_hf', 'exllama-v2-hf', 'exllama_v2_hf', 'exllama-v2_hf', 'exllama2-hf', 'exllama2_hf', 'exllama-2-hf', 'exllama_2_hf', 'exllama-2_hf']:
-        return 'ExLlamav2_HF'
     elif name in ['exllamav3-hf', 'exllamav3_hf', 'exllama-v3-hf', 'exllama_v3_hf', 'exllama-v3_hf', 'exllama3-hf', 'exllama3_hf', 'exllama-3-hf', 'exllama_3_hf', 'exllama-3_hf']:
         return 'ExLlamav3_HF'
     elif name in ['exllamav3']:
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 845711ce55..cace3f7c6c 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -40,7 +40,7 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
             yield ''
             return
 
-        if shared.model.__class__.__name__ in ['LlamaServer', 'Exllamav2Model', 'Exllamav3Model', 'TensorRTLLMModel']:
+        if shared.model.__class__.__name__ in ['LlamaServer', 'Exllamav3Model', 'TensorRTLLMModel']:
             generate_func = generate_reply_custom
         else:
             generate_func = generate_reply_HF
@@ -128,9 +128,9 @@ def encode(prompt, add_special_tokens=True, add_bos_token=True, truncation_lengt
 
         from modules.torch_utils import get_device
 
-        if shared.model.__class__.__name__ in ['Exllamav2Model', 'Exllamav3Model', 'TensorRTLLMModel']:
+        if shared.model.__class__.__name__ in ['Exllamav3Model', 'TensorRTLLMModel']:
             input_ids = shared.tokenizer.encode(str(prompt))
-            if shared.model.__class__.__name__ not in ['Exllamav2Model', 'Exllamav3Model']:
+            if shared.model.__class__.__name__ not in ['Exllamav3Model']:
                 input_ids = np.array(input_ids).reshape(1, len(input_ids))
         else:
             input_ids = shared.tokenizer.encode(str(prompt), return_tensors='pt', add_special_tokens=add_special_tokens)
@@ -148,7 +148,7 @@ def encode(prompt, add_special_tokens=True, add_bos_token=True, truncation_lengt
         if truncation_length is not None:
             input_ids = input_ids[:, -truncation_length:]
 
-        if shared.model.__class__.__name__ in ['Exllamav2Model', 'Exllamav3Model', 'TensorRTLLMModel'] or shared.args.cpu:
+        if shared.model.__class__.__name__ in ['Exllamav3Model', 'TensorRTLLMModel'] or shared.args.cpu:
             return input_ids
         else:
             device = get_device()
diff --git a/modules/ui.py b/modules/ui.py
index b46915833f..e118e68420 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -141,7 +141,6 @@ def list_model_elements():
         'compress_pos_emb',
         'compute_dtype',
         'quant_type',
-        'num_experts_per_token',
         'load_in_8bit',
         'load_in_4bit',
         'attn_implementation',
@@ -154,12 +153,8 @@ def list_model_elements():
         'numa',
         'use_double_quant',
         'bf16',
-        'autosplit',
         'enable_tp',
         'tp_backend',
-        'no_flash_attn',
-        'no_xformers',
-        'no_sdpa',
         'cfg_cache',
         'cpp_runner',
         'no_use_fast',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index de9901e364..3acdd062e4 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -45,7 +45,7 @@ def create_ui():
                             shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=0, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. llama.cpp: 0 = auto if gpu-layers is also -1. Common values: 4096, 8192, 16384, 32768, 65536, 131072.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
-                            shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
+                            shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
                             shared.gradio['tp_backend'] = gr.Dropdown(label="tp-backend", choices=['native', 'nccl'], value=shared.args.tp_backend, info='The backend for tensor parallelism.')
 
                         with gr.Column():
@@ -55,7 +55,6 @@ def create_ui():
                             shared.gradio['load_in_8bit'] = gr.Checkbox(label="load-in-8bit", value=shared.args.load_in_8bit)
                             shared.gradio['load_in_4bit'] = gr.Checkbox(label="load-in-4bit", value=shared.args.load_in_4bit)
                             shared.gradio['use_double_quant'] = gr.Checkbox(label="use_double_quant", value=shared.args.use_double_quant, info='Used by load-in-4bit.')
-                            shared.gradio['autosplit'] = gr.Checkbox(label="autosplit", value=shared.args.autosplit, info='Automatically split the model tensors across the available GPUs.')
                             shared.gradio['enable_tp'] = gr.Checkbox(label="enable_tp", value=shared.args.enable_tp, info='Enable tensor parallelism (TP).')
                             shared.gradio['cpp_runner'] = gr.Checkbox(label="cpp-runner", value=shared.args.cpp_runner, info='Enable inference with ModelRunnerCpp, which is faster than the default ModelRunner.')
                             shared.gradio['tensorrt_llm_info'] = gr.Markdown('* TensorRT-LLM has to be installed manually in a separate Python 3.10 environment at the moment. For a guide, consult the description of [this PR](https://github.com/oobabooga/text-generation-webui/pull/5715). \n\n* `ctx_size` is only used when `cpp-runner` is checked.\n\n* `cpp_runner` does not support streaming at the moment.')
@@ -101,7 +100,6 @@ def create_ui():
                                 shared.gradio['compress_pos_emb'] = gr.Number(label='compress_pos_emb', value=shared.args.compress_pos_emb, precision=2, info='Positional embeddings compression factor. Should be set to (context length) / (model\'s original context length). Equal to 1/rope_freq_scale.')
                                 shared.gradio['compute_dtype'] = gr.Dropdown(label="compute_dtype", choices=["bfloat16", "float16", "float32"], value=shared.args.compute_dtype, info='Used by load-in-4bit.')
                                 shared.gradio['quant_type'] = gr.Dropdown(label="quant_type", choices=["nf4", "fp4"], value=shared.args.quant_type, info='Used by load-in-4bit.')
-                                shared.gradio['num_experts_per_token'] = gr.Number(label="Number of experts per token", value=shared.args.num_experts_per_token, info='Only applies to MoE models like Mixtral.')
 
                             with gr.Column():
                                 shared.gradio['cpu'] = gr.Checkbox(label="cpu", value=shared.args.cpu, info='Use PyTorch in CPU mode.')
@@ -112,9 +110,6 @@ def create_ui():
                                 shared.gradio['mlock'] = gr.Checkbox(label="mlock", value=shared.args.mlock)
                                 shared.gradio['numa'] = gr.Checkbox(label="numa", value=shared.args.numa, info='NUMA support can help on some systems with non-uniform memory access.')
                                 shared.gradio['bf16'] = gr.Checkbox(label="bf16", value=shared.args.bf16)
-                                shared.gradio['no_flash_attn'] = gr.Checkbox(label="no_flash_attn", value=shared.args.no_flash_attn)
-                                shared.gradio['no_xformers'] = gr.Checkbox(label="no_xformers", value=shared.args.no_xformers)
-                                shared.gradio['no_sdpa'] = gr.Checkbox(label="no_sdpa", value=shared.args.no_sdpa)
                                 shared.gradio['cfg_cache'] = gr.Checkbox(label="cfg-cache", value=shared.args.cfg_cache, info='Necessary to use CFG with this loader.')
                                 shared.gradio['no_use_fast'] = gr.Checkbox(label="no_use_fast", value=shared.args.no_use_fast, info='Set use_fast=False while loading the tokenizer.')
                                 if not shared.args.portable:
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index df0f0fc6a8..6b5e2e784d 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -46,8 +46,5 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.23/exllamav3-0.0.23+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.23/exllamav3-0.0.23+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 4735014363..059de7229c 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -41,5 +41,3 @@ tiktoken
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.4.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"

From 9824c82cb65cb28953c038d36710ecc82df0eb1d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 16:49:58 -0800
Subject: [PATCH 1233/1701] API: Add parallel request support for llama.cpp and
 ExLlamaV3

---
 docs/12 - OpenAI API.md          |  29 +++++++++
 extensions/openai/completions.py |  24 ++++---
 extensions/openai/script.py      |  72 ++++++++++-----------
 modules/exllamav3.py             | 105 +++++++++++++++++++++++++++----
 modules/llama_cpp_server.py      |   4 +-
 modules/loaders.py               |   1 +
 modules/shared.py                |   1 +
 modules/text_generation.py       |  23 ++++++-
 modules/ui.py                    |   1 +
 modules/ui_model_menu.py         |   1 +
 10 files changed, 198 insertions(+), 63 deletions(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index cd5757f69d..fc444b153a 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -338,6 +338,35 @@ for event in client.events():
 print()
 ```
 
+#### Python parallel requests example
+
+The API supports handling multiple requests in parallel. For ExLlamaV3, this works out of the box. For llama.cpp, you need to pass `--parallel N` to set the number of concurrent slots.
+
+```python
+import concurrent.futures
+import requests
+
+url = "http://127.0.0.1:5000/v1/chat/completions"
+prompts = [
+    "Write a haiku about the ocean.",
+    "Explain quantum computing in simple terms.",
+    "Tell me a joke about programmers.",
+]
+
+def send_request(prompt):
+    response = requests.post(url, json={
+        "messages": [{"role": "user", "content": prompt}],
+        "max_tokens": 200,
+    })
+    return response.json()["choices"][0]["message"]["content"]
+
+with concurrent.futures.ThreadPoolExecutor() as executor:
+    results = list(executor.map(send_request, prompts))
+
+for prompt, result in zip(prompts, results):
+    print(f"Q: {prompt}\nA: {result}\n")
+```
+
 #### Python example with API key
 
 Replace
diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index c3037d0c24..de944a8fd7 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -165,7 +165,7 @@ def convert_history(history):
     }
 
 
-def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, prompt_only=False) -> dict:
+def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, prompt_only=False, stop_event=None) -> dict:
     if body.get('functions', []):
         raise InvalidRequestError(message="functions is not supported.", param='functions')
 
@@ -211,6 +211,8 @@ def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, p
 
     # generation parameters
     generate_params = process_parameters(body, is_legacy=is_legacy)
+    if stop_event is not None:
+        generate_params['stop_event'] = stop_event
     continue_ = body['continue_']
 
     # Instruction template
@@ -378,7 +380,7 @@ def chat_streaming_chunk(content, chunk_tool_calls=None):
         yield resp
 
 
-def completions_common(body: dict, is_legacy: bool = False, stream=False):
+def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_event=None):
     object_type = 'text_completion.chunk' if stream else 'text_completion'
     created_time = int(time.time())
     cmpl_id = "conv-%d" % (int(time.time() * 1000000000))
@@ -411,6 +413,8 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False):
     generate_params = process_parameters(body, is_legacy=is_legacy)
     max_tokens = generate_params['max_new_tokens']
     generate_params['stream'] = stream
+    if stop_event is not None:
+        generate_params['stop_event'] = stop_event
     requested_model = generate_params.pop('model')
     logprob_proc = generate_params.pop('logprob_proc', None)
     suffix = body['suffix'] if body['suffix'] else ''
@@ -561,23 +565,23 @@ def text_streaming_chunk(content):
         yield chunk
 
 
-def chat_completions(body: dict, is_legacy: bool = False) -> dict:
-    generator = chat_completions_common(body, is_legacy, stream=False)
+def chat_completions(body: dict, is_legacy: bool = False, stop_event=None) -> dict:
+    generator = chat_completions_common(body, is_legacy, stream=False, stop_event=stop_event)
     return deque(generator, maxlen=1).pop()
 
 
-def stream_chat_completions(body: dict, is_legacy: bool = False):
-    for resp in chat_completions_common(body, is_legacy, stream=True):
+def stream_chat_completions(body: dict, is_legacy: bool = False, stop_event=None):
+    for resp in chat_completions_common(body, is_legacy, stream=True, stop_event=stop_event):
         yield resp
 
 
-def completions(body: dict, is_legacy: bool = False) -> dict:
-    generator = completions_common(body, is_legacy, stream=False)
+def completions(body: dict, is_legacy: bool = False, stop_event=None) -> dict:
+    generator = completions_common(body, is_legacy, stream=False, stop_event=stop_event)
     return deque(generator, maxlen=1).pop()
 
 
-def stream_completions(body: dict, is_legacy: bool = False):
-    for resp in completions_common(body, is_legacy, stream=True):
+def stream_completions(body: dict, is_legacy: bool = False, stop_event=None):
+    for resp in completions_common(body, is_legacy, stream=True, stop_event=stop_event):
         yield resp
 
 
diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index edb22c22b1..7a30e311aa 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -3,6 +3,7 @@
 import logging
 import os
 import socket
+import threading
 import traceback
 from collections import deque
 from threading import Thread
@@ -24,7 +25,7 @@
 from modules import shared
 from modules.logging_colors import logger
 from modules.models import unload_model
-from modules.text_generation import stop_everything_event
+from modules.text_generation import stop_everything_event  # used by /v1/internal/stop-generation
 
 from .typing import (
     ChatCompletionRequest,
@@ -58,10 +59,6 @@
 }
 
 
-streaming_semaphore = asyncio.Semaphore(1)
-image_generation_semaphore = asyncio.Semaphore(1)
-
-
 def verify_api_key(authorization: str = Header(None)) -> None:
     expected_api_key = shared.args.api_key
     if expected_api_key and (authorization is None or authorization != f"Bearer {expected_api_key}"):
@@ -113,28 +110,30 @@ async def openai_completions(request: Request, request_data: CompletionRequest):
     is_legacy = "/generate" in path
 
     if request_data.stream:
+        stop_event = threading.Event()
+
         async def generator():
-            async with streaming_semaphore:
-                try:
-                    response = OAIcompletions.stream_completions(to_dict(request_data), is_legacy=is_legacy)
-                    async for resp in iterate_in_threadpool(response):
-                        disconnected = await request.is_disconnected()
-                        if disconnected:
-                            break
-
-                        yield {"data": json.dumps(resp)}
-                finally:
-                    stop_everything_event()
-                    response.close()
-                    return
+            response = OAIcompletions.stream_completions(to_dict(request_data), is_legacy=is_legacy, stop_event=stop_event)
+            try:
+                async for resp in iterate_in_threadpool(response):
+                    disconnected = await request.is_disconnected()
+                    if disconnected:
+                        break
+
+                    yield {"data": json.dumps(resp)}
+            finally:
+                stop_event.set()
+                response.close()
 
         return EventSourceResponse(generator())  # SSE streaming
 
     else:
+        stop_event = threading.Event()
         response = await asyncio.to_thread(
             OAIcompletions.completions,
             to_dict(request_data),
-            is_legacy=is_legacy
+            is_legacy=is_legacy,
+            stop_event=stop_event
         )
 
         return JSONResponse(response)
@@ -146,28 +145,30 @@ async def openai_chat_completions(request: Request, request_data: ChatCompletion
     is_legacy = "/generate" in path
 
     if request_data.stream:
+        stop_event = threading.Event()
+
         async def generator():
-            async with streaming_semaphore:
-                try:
-                    response = OAIcompletions.stream_chat_completions(to_dict(request_data), is_legacy=is_legacy)
-                    async for resp in iterate_in_threadpool(response):
-                        disconnected = await request.is_disconnected()
-                        if disconnected:
-                            break
-
-                        yield {"data": json.dumps(resp)}
-                finally:
-                    stop_everything_event()
-                    response.close()
-                    return
+            response = OAIcompletions.stream_chat_completions(to_dict(request_data), is_legacy=is_legacy, stop_event=stop_event)
+            try:
+                async for resp in iterate_in_threadpool(response):
+                    disconnected = await request.is_disconnected()
+                    if disconnected:
+                        break
+
+                    yield {"data": json.dumps(resp)}
+            finally:
+                stop_event.set()
+                response.close()
 
         return EventSourceResponse(generator())  # SSE streaming
 
     else:
+        stop_event = threading.Event()
         response = await asyncio.to_thread(
             OAIcompletions.chat_completions,
             to_dict(request_data),
-            is_legacy=is_legacy
+            is_legacy=is_legacy,
+            stop_event=stop_event
         )
 
         return JSONResponse(response)
@@ -232,9 +233,8 @@ async def handle_audio_transcription(request: Request):
 async def handle_image_generation(request_data: ImageGenerationRequest):
     import extensions.openai.images as OAIimages
 
-    async with image_generation_semaphore:
-        response = await asyncio.to_thread(OAIimages.generations, request_data)
-        return JSONResponse(response)
+    response = await asyncio.to_thread(OAIimages.generations, request_data)
+    return JSONResponse(response)
 
 
 @app.post("/v1/embeddings", response_model=EmbeddingsResponse, dependencies=check_key)
diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index eca5dde0d9..af5745bc3a 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -1,3 +1,5 @@
+import queue
+import threading
 import traceback
 from pathlib import Path
 from typing import Any, List, Tuple
@@ -34,6 +36,55 @@
     traceback.print_exc()
 
 
+class ConcurrentGenerator:
+    def __init__(self, generator):
+        self.generator = generator
+        self.lock = threading.Lock()
+        self.job_queues = {}
+        self.active = True
+        self.has_jobs = threading.Event()
+        self.thread = threading.Thread(target=self._iterate_loop, daemon=True)
+        self.thread.start()
+
+    def _iterate_loop(self):
+        while self.active:
+            self.has_jobs.wait(timeout=0.5)
+            with self.lock:
+                if self.generator.num_remaining_jobs() == 0:
+                    self.has_jobs.clear()
+                    continue
+                results = self.generator.iterate()
+            for result in results:
+                job = result["job"]
+                q = self.job_queues.get(job)
+                if q:
+                    q.put(result)
+                    if result.get("eos"):
+                        self.job_queues.pop(job, None)
+            if not self.job_queues:
+                self.has_jobs.clear()
+
+    def submit(self, job) -> queue.Queue:
+        q = queue.Queue()
+        with self.lock:
+            self.job_queues[job] = q
+            self.generator.enqueue(job)
+        self.has_jobs.set()
+        return q
+
+    def cancel(self, job):
+        with self.lock:
+            if job in self.job_queues:
+                self.generator.cancel(job)
+                self.job_queues[job].put(None)
+                del self.job_queues[job]
+
+    def stop(self):
+        self.active = False
+        self.has_jobs.set()
+        self.thread.join(timeout=5)
+
+
 class Exllamav3Model:
     def __init__(self):
         pass
@@ -167,6 +218,7 @@ def from_pretrained(cls, path_to_model):
         result.cache = cache
         result.tokenizer = tokenizer
         result.generator = generator
+        result.parallel_generator = ConcurrentGenerator(generator)
         result.config = config
         result.max_tokens = max_tokens
         result.vision_model = vision_model
@@ -346,27 +398,47 @@ def custom_sort_key(sampler_obj):
         )
 
         # Stream generation
-        self.generator.enqueue(job)
-
         response_text = ""
+        stop_event = state.get('stop_event')
 
-        try:
-            while self.generator.num_remaining_jobs():
-                if shared.stop_everything:
-                    break
-
-                results = self.generator.iterate()
-                for result in results:
-                    if "eos" in result and result["eos"]:
+        if stop_event:
+            # Concurrent path for API requests
+            result_queue = self.parallel_generator.submit(job)
+            try:
+                while True:
+                    if stop_event.is_set() or shared.stop_everything:
+                        break
+                    try:
+                        result = result_queue.get(timeout=0.1)
+                    except queue.Empty:
+                        continue
+                    if result is None or result.get("eos"):
                         break
-
                     chunk = result.get("text", "")
                     if chunk:
                         response_text += chunk
                         yield response_text
+            finally:
+                self.parallel_generator.cancel(job)
+        else:
+            # Original single-request path (WebUI)
+            self.generator.enqueue(job)
+            try:
+                while self.generator.num_remaining_jobs():
+                    if shared.stop_everything:
+                        break
 
-        finally:
-            self.generator.clear_queue()
+                    results = self.generator.iterate()
+                    for result in results:
+                        if "eos" in result and result["eos"]:
+                            break
+
+                        chunk = result.get("text", "")
+                        if chunk:
+                            response_text += chunk
+                            yield response_text
+            finally:
+                self.generator.clear_queue()
 
     def generate(self, prompt, state):
         output = ""
@@ -429,6 +501,13 @@ def last_prompt_token_count(self):
     def unload(self):
         logger.info("Unloading ExLlamaV3 model components...")
 
+        if hasattr(self, 'parallel_generator') and self.parallel_generator is not None:
+            try:
+                self.parallel_generator.stop()
+            except Exception as e:
+                logger.warning(f"Error stopping parallel generator: {e}")
+            self.parallel_generator = None
+
         if hasattr(self, 'vision_model') and self.vision_model is not None:
             try:
                 del self.vision_model
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 2af9aa8a14..017c5d2a61 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -217,8 +217,9 @@ def generate_with_streaming(self, prompt, state):
             full_text = ""
 
             # Process the streaming response
+            stop_event = state.get('stop_event')
             for line in response.iter_lines():
-                if shared.stop_everything:
+                if shared.stop_everything or (stop_event and stop_event.is_set()):
                     break
 
                 if not line:
@@ -410,6 +411,7 @@ def _start_server(self):
             cmd += ["--spec-ngram-size-n", str(shared.args.spec_ngram_size_n)]
             cmd += ["--spec-ngram-size-m", str(shared.args.spec_ngram_size_m)]
             cmd += ["--spec-ngram-min-hits", str(shared.args.spec_ngram_min_hits)]
+        cmd += ["--parallel", str(shared.args.parallel)]
         if shared.args.streaming_llm:
             cmd += ["--cache-reuse", "1"]
             cmd += ["--swa-full"]
diff --git a/modules/loaders.py b/modules/loaders.py
index 9923f116e2..0348c93960 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -23,6 +23,7 @@
         'no_mmap',
         'mlock',
         'numa',
+        'parallel',
         'model_draft',
         'draft_max',
         'gpu_layers_draft',
diff --git a/modules/shared.py b/modules/shared.py
index 8b820e2df1..ec6e23b9da 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -102,6 +102,7 @@
 group.add_argument('--threads', type=int, default=0, help='Number of threads to use.')
 group.add_argument('--threads-batch', type=int, default=0, help='Number of threads to use for batches/prompt processing.')
 group.add_argument('--numa', action='store_true', help='Activate NUMA task allocation for llama.cpp.')
+group.add_argument('--parallel', type=int, default=1, help='Number of parallel request slots. The context size is divided equally among slots. For example, to have 4 slots with 8192 context each, set ctx_size to 32768.')
 group.add_argument('--extra-flags', type=str, default=None, help='Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"')
 
 # Transformers/Accelerate
diff --git a/modules/text_generation.py b/modules/text_generation.py
index cace3f7c6c..6e0e67a1ef 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -22,13 +22,23 @@ def generate_reply(*args, **kwargs):
         from modules.models import load_model
         shared.model, shared.tokenizer = load_model(shared.model_name)
 
-    shared.generation_lock.acquire()
+    state = args[1] if len(args) > 1 else kwargs.get('state', {})
+    use_parallel = (
+        state.get('stop_event') is not None
+        and shared.model.__class__.__name__ in ['Exllamav3Model', 'LlamaServer']
+        and (shared.model.__class__.__name__ != 'LlamaServer' or shared.args.parallel > 1)
+    )
+
+    if not use_parallel:
+        shared.generation_lock.acquire()
+
     try:
         for result in _generate_reply(*args, **kwargs):
             yield result
     finally:
         models.last_generation_time = time.time()
-        shared.generation_lock.release()
+        if not use_parallel:
+            shared.generation_lock.release()
 
 
 def _generate_reply(question, state, stopping_strings=None, is_chat=False, escape_html=False, for_ui=False):
@@ -68,7 +78,10 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
     reply = ''
     is_stream = state['stream']
     if len(all_stop_strings) > 0 and not state['stream']:
+        stop_event_ref = state.pop('stop_event', None)
         state = copy.deepcopy(state)
+        if stop_event_ref is not None:
+            state['stop_event'] = stop_event_ref
         state['stream'] = True
 
     # Generate
@@ -99,7 +112,8 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
                     yield reply
                 last_update = time.monotonic()
 
-        if stop_found or (state['max_tokens_second'] > 0 and shared.stop_everything):
+        stop_event = state.get('stop_event')
+        if stop_found or (state['max_tokens_second'] > 0 and shared.stop_everything) or (stop_event and stop_event.is_set()):
             break
 
     if not is_chat:
@@ -474,7 +488,10 @@ def generate_reply_custom(question, original_question, state, stopping_strings=N
     For models that do not use the transformers library for sampling
     """
 
+    stop_event_ref = state.pop('stop_event', None)
     state = copy.deepcopy(state)
+    if stop_event_ref is not None:
+        state['stop_event'] = stop_event_ref
     state['seed'] = set_manual_seed(state['seed'])
     t0 = time.time()
     reply = ''
diff --git a/modules/ui.py b/modules/ui.py
index e118e68420..fd20e7823c 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -151,6 +151,7 @@ def list_model_elements():
         'no_mmap',
         'mlock',
         'numa',
+        'parallel',
         'use_double_quant',
         'bf16',
         'enable_tp',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 3acdd062e4..12b5654c59 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -90,6 +90,7 @@ def create_ui():
                             with gr.Column():
                                 shared.gradio['threads'] = gr.Slider(label="threads", minimum=0, step=1, maximum=256, value=shared.args.threads)
                                 shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
+                                shared.gradio['parallel'] = gr.Slider(label="parallel", minimum=1, step=1, maximum=64, value=shared.args.parallel, info='Number of parallel request slots. The context size is divided equally among slots. For example, to have 4 slots with 8192 context each, set ctx_size to 32768.')
                                 shared.gradio['batch_size'] = gr.Slider(label="batch_size", minimum=1, maximum=4096, step=1, value=shared.args.batch_size)
                                 shared.gradio['ubatch_size'] = gr.Slider(label="ubatch_size", minimum=1, maximum=4096, step=1, value=shared.args.ubatch_size)
                                 shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')

From f52d9336e53cd3149dc76e95c02925498915eabd Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 18:09:45 -0800
Subject: [PATCH 1234/1701] TensorRT-LLM: Migrate from ModelRunner to LLM API,
 add concurrent API request support

---
 modules/loaders.py         |   6 +-
 modules/models.py          |   4 +-
 modules/shared.py          |   4 --
 modules/tensorrt_llm.py    | 115 ++++++++++++-------------------------
 modules/text_generation.py |   2 +-
 modules/ui.py              |   1 -
 modules/ui_model_menu.py   |   7 ++-
 7 files changed, 50 insertions(+), 89 deletions(-)

diff --git a/modules/loaders.py b/modules/loaders.py
index 0348c93960..15b8dfebd4 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -77,7 +77,6 @@
     ],
     'TensorRT-LLM': [
         'ctx_size',
-        'cpp_runner',
         'tensorrt_llm_info',
     ]
 })
@@ -252,11 +251,16 @@ def transformers_samplers():
         'temperature',
         'top_p',
         'top_k',
+        'min_p',
         'repetition_penalty',
         'frequency_penalty',
         'presence_penalty',
+        'no_repeat_ngram_size',
         'auto_max_new_tokens',
         'ban_eos_token',
+        'add_bos_token',
+        'skip_special_tokens',
+        'seed',
     }
 }
 
diff --git a/modules/models.py b/modules/models.py
index 9780b9f5b4..bc5585cfc9 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -114,7 +114,7 @@ def TensorRT_LLM_loader(model_name):
         raise ModuleNotFoundError("Failed to import 'tensorrt_llm'. Please install it manually following the instructions in the TensorRT-LLM GitHub repository.")
 
     model = TensorRTLLMModel.from_pretrained(model_name)
-    return model
+    return model, model.tokenizer
 
 
 def unload_model(keep_model_name=False):
@@ -124,7 +124,7 @@ def unload_model(keep_model_name=False):
     model_class_name = shared.model.__class__.__name__
     is_llamacpp = (model_class_name == 'LlamaServer')
 
-    if model_class_name in ['Exllamav3Model', 'Exllamav3HF']:
+    if model_class_name in ['Exllamav3Model', 'Exllamav3HF', 'TensorRTLLMModel']:
         shared.model.unload()
 
     shared.model = shared.tokenizer = None
diff --git a/modules/shared.py b/modules/shared.py
index ec6e23b9da..4e212d1b31 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -133,10 +133,6 @@
 group.add_argument('--tp-backend', type=str, default='native', help='The backend for tensor parallelism. Valid options: native, nccl. Default: native.')
 group.add_argument('--cfg-cache', action='store_true', help='Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.')
 
-# TensorRT-LLM
-group = parser.add_argument_group('TensorRT-LLM')
-group.add_argument('--cpp-runner', action='store_true', help='Use the ModelRunnerCpp runner, which is faster than the default ModelRunner.')
-
 # RoPE
 group = parser.add_argument_group('RoPE')
 group.add_argument('--alpha_value', type=float, default=1, help='Positional embeddings alpha factor for NTK RoPE scaling. Use either this or compress_pos_emb, not both.')
diff --git a/modules/tensorrt_llm.py b/modules/tensorrt_llm.py
index 161dde824c..ae061d0664 100644
--- a/modules/tensorrt_llm.py
+++ b/modules/tensorrt_llm.py
@@ -1,15 +1,10 @@
 from pathlib import Path
 
-import tensorrt_llm
-import torch
-from tensorrt_llm.runtime import ModelRunner, ModelRunnerCpp
+from tensorrt_llm._tensorrt_engine import LLM
+from tensorrt_llm.llmapi import SamplingParams
 
 from modules import shared
 from modules.logging_colors import logger
-from modules.text_generation import (
-    get_max_prompt_length,
-    get_reply_from_output_ids
-)
 
 
 class TensorRTLLMModel:
@@ -18,91 +13,50 @@ def __init__(self):
 
     @classmethod
     def from_pretrained(cls, path_to_model):
-
         path_to_model = Path(f'{shared.args.model_dir}') / Path(path_to_model)
-        runtime_rank = tensorrt_llm.mpi_rank()
 
-        # Define model settings
-        runner_kwargs = dict(
-            engine_dir=str(path_to_model),
-            lora_dir=None,
-            rank=runtime_rank,
-            debug_mode=False,
-            lora_ckpt_source="hf",
+        llm = LLM(
+            model=str(path_to_model),
+            skip_tokenizer_init=False,
         )
 
-        if shared.args.cpp_runner:
-            logger.info("TensorRT-LLM: Using \"ModelRunnerCpp\"")
-            runner_kwargs.update(
-                max_batch_size=1,
-                max_beam_width=1,
-            )
-        else:
-            logger.info("TensorRT-LLM: Using \"ModelRunner\"")
-
-        # Load the model
-        runner_cls = ModelRunnerCpp if shared.args.cpp_runner else ModelRunner
-        runner = runner_cls.from_dir(**runner_kwargs)
-
         result = cls()
-        result.model = runner
-        result.runtime_rank = runtime_rank
-
+        result.llm = llm
+        result.tokenizer = llm.tokenizer
         return result
 
     def generate_with_streaming(self, prompt, state):
-        batch_input_ids = []
-        input_ids = shared.tokenizer.encode(
-            prompt,
-            add_special_tokens=True,
-            truncation=False,
+        sampling_params = SamplingParams(
+            max_tokens=state['max_new_tokens'] if not state['auto_max_new_tokens']
+                       else state['truncation_length'] - len(shared.tokenizer.encode(prompt)),
+            end_id=shared.tokenizer.eos_token_id,
+            temperature=state['temperature'],
+            top_k=state['top_k'],
+            top_p=state['top_p'],
+            min_p=state['min_p'],
+            repetition_penalty=state['repetition_penalty'],
+            presence_penalty=state['presence_penalty'],
+            frequency_penalty=state['frequency_penalty'],
+            no_repeat_ngram_size=state['no_repeat_ngram_size'] if state['no_repeat_ngram_size'] > 0 else None,
+            seed=state['seed'],
+            ignore_eos=state['ban_eos_token'],
+            add_special_tokens=state['add_bos_token'],
+            skip_special_tokens=state['skip_special_tokens'],
         )
-        input_ids = torch.tensor(input_ids, dtype=torch.int32)
-        input_ids = input_ids[-get_max_prompt_length(state):]  # Apply truncation_length
-        batch_input_ids.append(input_ids)
-
-        if state['auto_max_new_tokens']:
-            max_new_tokens = state['truncation_length'] - input_ids.shape[-1]
-        else:
-            max_new_tokens = state['max_new_tokens']
 
-        with torch.no_grad():
-            generator = self.model.generate(
-                batch_input_ids,
-                max_new_tokens=max_new_tokens,
-                end_id=shared.tokenizer.eos_token_id if not state['ban_eos_token'] else -1,
-                pad_id=shared.tokenizer.pad_token_id or shared.tokenizer.eos_token_id,
-                temperature=state['temperature'],
-                top_k=state['top_k'],
-                top_p=state['top_p'],
-                repetition_penalty=state['repetition_penalty'],
-                presence_penalty=state['presence_penalty'],
-                frequency_penalty=state['frequency_penalty'],
-                stop_words_list=None,
-                bad_words_list=None,
-                lora_uids=None,
-                prompt_table=None,
-                prompt_tasks=None,
-                streaming=True,
-                output_sequence_lengths=True,
-                return_dict=True,
-            )
-
-        torch.cuda.synchronize()
+        stop_event = state.get('stop_event')
+        result = self.llm.generate_async(prompt, sampling_params=sampling_params, streaming=True)
 
         cumulative_reply = ''
-        starting_from = batch_input_ids[0].shape[-1]
-
-        for curr_outputs in generator:
-            if shared.stop_everything:
+        for output in result:
+            if shared.stop_everything or (stop_event and stop_event.is_set()):
+                result.abort()
                 break
 
-            sequence_length = curr_outputs['sequence_lengths'][0].item()
-            output_ids = curr_outputs['output_ids'][0][0][:sequence_length].tolist()
-
-            cumulative_reply += get_reply_from_output_ids(output_ids, state, starting_from=starting_from)
-            starting_from = sequence_length
-            yield cumulative_reply
+            text_diff = output.outputs[0].text_diff
+            if text_diff:
+                cumulative_reply += text_diff
+                yield cumulative_reply
 
     def generate(self, prompt, state):
         output = ''
@@ -110,3 +64,8 @@ def generate(self, prompt, state):
             pass
 
         return output
+
+    def unload(self):
+        if hasattr(self, 'llm') and self.llm is not None:
+            self.llm.shutdown()
+            self.llm = None
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 6e0e67a1ef..02c1320c45 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -25,7 +25,7 @@ def generate_reply(*args, **kwargs):
     state = args[1] if len(args) > 1 else kwargs.get('state', {})
     use_parallel = (
         state.get('stop_event') is not None
-        and shared.model.__class__.__name__ in ['Exllamav3Model', 'LlamaServer']
+        and shared.model.__class__.__name__ in ['Exllamav3Model', 'LlamaServer', 'TensorRTLLMModel']
         and (shared.model.__class__.__name__ != 'LlamaServer' or shared.args.parallel > 1)
     )
 
diff --git a/modules/ui.py b/modules/ui.py
index fd20e7823c..0cd2cf43b7 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -157,7 +157,6 @@ def list_model_elements():
         'enable_tp',
         'tp_backend',
         'cfg_cache',
-        'cpp_runner',
         'no_use_fast',
         'model_draft',
         'draft_max',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 12b5654c59..0c2450d5eb 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -56,8 +56,11 @@ def create_ui():
                             shared.gradio['load_in_4bit'] = gr.Checkbox(label="load-in-4bit", value=shared.args.load_in_4bit)
                             shared.gradio['use_double_quant'] = gr.Checkbox(label="use_double_quant", value=shared.args.use_double_quant, info='Used by load-in-4bit.')
                             shared.gradio['enable_tp'] = gr.Checkbox(label="enable_tp", value=shared.args.enable_tp, info='Enable tensor parallelism (TP).')
-                            shared.gradio['cpp_runner'] = gr.Checkbox(label="cpp-runner", value=shared.args.cpp_runner, info='Enable inference with ModelRunnerCpp, which is faster than the default ModelRunner.')
-                            shared.gradio['tensorrt_llm_info'] = gr.Markdown('* TensorRT-LLM has to be installed manually in a separate Python 3.10 environment at the moment. For a guide, consult the description of [this PR](https://github.com/oobabooga/text-generation-webui/pull/5715). \n\n* `ctx_size` is only used when `cpp-runner` is checked.\n\n* `cpp_runner` does not support streaming at the moment.')
+                            shared.gradio['tensorrt_llm_info'] = gr.Markdown(
+                                '* TensorRT-LLM has to be installed manually: `pip install tensorrt_llm==1.1.0 --extra-index-url https://pypi.nvidia.com`.\n\n'
+                                '* You can load either a pre-built TensorRT engine or a regular HF model. '
+                                'HF models will be compiled to a TensorRT engine automatically on each load (this can take a while).'
+                            )
 
                             # Multimodal
                             with gr.Accordion("Multimodal (vision)", open=False, elem_classes='tgw-accordion') as shared.gradio['mmproj_accordion']:

From 249bd6eea2b7f07f73e96f5cfb32218196a9b237 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 18:11:55 -0800
Subject: [PATCH 1235/1701] UI: Update the parallel info message

---
 modules/ui_model_menu.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 0c2450d5eb..e6e57a2255 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -91,9 +91,9 @@ def create_ui():
                     with gr.Accordion("See more options", open=False, elem_classes='tgw-accordion'):
                         with gr.Row():
                             with gr.Column():
+                                shared.gradio['parallel'] = gr.Slider(label="parallel", minimum=1, step=1, maximum=64, value=shared.args.parallel, info='Number of parallel request slots for the API. The context size is divided equally among slots. For example, to have 4 slots with 8192 context each, set ctx_size to 32768.')
                                 shared.gradio['threads'] = gr.Slider(label="threads", minimum=0, step=1, maximum=256, value=shared.args.threads)
                                 shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
-                                shared.gradio['parallel'] = gr.Slider(label="parallel", minimum=1, step=1, maximum=64, value=shared.args.parallel, info='Number of parallel request slots. The context size is divided equally among slots. For example, to have 4 slots with 8192 context each, set ctx_size to 32768.')
                                 shared.gradio['batch_size'] = gr.Slider(label="batch_size", minimum=1, maximum=4096, step=1, value=shared.args.batch_size)
                                 shared.gradio['ubatch_size'] = gr.Slider(label="ubatch_size", minimum=1, maximum=4096, step=1, value=shared.args.ubatch_size)
                                 shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')

From 4c406e024fd1b0342baf35130a0a79868a4ef539 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 18:36:07 -0800
Subject: [PATCH 1236/1701] API: Speed up chat completions by ~85ms per request

---
 extensions/openai/completions.py |  4 ++--
 modules/chat.py                  | 20 ++++++++++++++++----
 2 files changed, 18 insertions(+), 6 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index de944a8fd7..56d2059def 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -288,8 +288,8 @@ def chat_streaming_chunk(content, chunk_tool_calls=None):
         return chunk
 
     # generate reply #######################################
-    prompt = generate_chat_prompt(user_input, generate_params, _continue=continue_)
     if prompt_only:
+        prompt = generate_chat_prompt(user_input, generate_params, _continue=continue_)
         yield {'prompt': prompt}
         return
 
@@ -335,7 +335,7 @@ def chat_streaming_chunk(content, chunk_tool_calls=None):
         if len(tool_calls) > 0:
             break
 
-    token_count = len(encode(prompt)[0])
+    token_count = shared.model.last_prompt_token_count if hasattr(shared.model, 'last_prompt_token_count') else 0
     completion_token_count = len(encode(answer)[0])
     stop_reason = "stop"
     if len(tool_calls) > 0:
diff --git a/modules/chat.py b/modules/chat.py
index 21a2104a90..f9a98bb6a9 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -76,6 +76,18 @@ def update_message_metadata(metadata_dict, role, index, **fields):
 )
 jinja_env.globals["strftime_now"] = strftime_now
 
+_template_cache = {}
+
+
+def get_compiled_template(template_str):
+    """Cache compiled Jinja2 templates keyed by their source string."""
+    compiled = _template_cache.get(template_str)
+    if compiled is None:
+        compiled = jinja_env.from_string(template_str)
+        _template_cache[template_str] = compiled
+
+    return compiled
+
 
 def str_presenter(dumper, data):
     """
@@ -106,8 +118,8 @@ def generate_chat_prompt(user_input, state, **kwargs):
     if state['mode'] != 'instruct':
         chat_template_str = replace_character_names(chat_template_str, state['name1'], state['name2'])
 
-    instruction_template = jinja_env.from_string(state['instruction_template_str'])
-    chat_template = jinja_env.from_string(chat_template_str)
+    instruction_template = get_compiled_template(state['instruction_template_str'])
+    chat_template = get_compiled_template(chat_template_str)
 
     instruct_renderer = partial(
         instruction_template.render,
@@ -481,12 +493,12 @@ def get_stopping_strings(state):
     renderers = []
 
     if state['mode'] in ['instruct', 'chat-instruct']:
-        template = jinja_env.from_string(state['instruction_template_str'])
+        template = get_compiled_template(state['instruction_template_str'])
         renderer = partial(template.render, add_generation_prompt=False, bos_token=shared.bos_token, eos_token=shared.eos_token)
         renderers.append(renderer)
 
     if state['mode'] in ['chat']:
-        template = jinja_env.from_string(state['chat_template_str'])
+        template = get_compiled_template(state['chat_template_str'])
         renderer = partial(template.render, add_generation_prompt=False, name1=state['name1'], name2=state['name2'])
         renderers.append(renderer)
 

From e2548f69a9e9a582f99b4deceeb5e5ed48f49418 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 19:26:21 -0800
Subject: [PATCH 1237/1701] Make user_data configurable: add --user-data-dir
 flag, auto-detect ../user_data

If --user-data-dir is not set, auto-detect: use ../user_data when
./user_data doesn't exist, making it easy to share user data across
portable builds by placing it one folder up.
---
 download-model.py              | 14 +++++---
 modules/chat.py                | 60 +++++++++++++++++-----------------
 modules/evaluate.py            |  8 ++---
 modules/extensions.py          | 14 ++++++--
 modules/html_generator.py      |  8 ++---
 modules/llama_cpp_server.py    |  2 +-
 modules/presets.py             |  2 +-
 modules/prompts.py             |  4 +--
 modules/shared.py              | 18 ++++++----
 modules/training.py            | 34 +++++++++----------
 modules/ui.py                  |  6 ++--
 modules/ui_chat.py             |  2 +-
 modules/ui_default.py          | 12 +++----
 modules/ui_file_saving.py      | 14 ++++----
 modules/ui_image_generation.py |  4 +--
 modules/ui_model_menu.py       |  6 ++--
 modules/ui_notebook.py         | 14 ++++----
 modules/ui_parameters.py       |  2 +-
 modules/ui_session.py          |  4 +--
 modules/utils.py               | 55 +++++++++++++++++++------------
 server.py                      | 16 ++++-----
 21 files changed, 166 insertions(+), 133 deletions(-)

diff --git a/download-model.py b/download-model.py
index 756d529f73..7acb7f46a0 100644
--- a/download-model.py
+++ b/download-model.py
@@ -24,6 +24,8 @@
 from requests.exceptions import ConnectionError, RequestException, Timeout
 from tqdm.contrib.concurrent import thread_map
 
+from modules.paths import resolve_user_data_dir
+
 base = os.environ.get("HF_ENDPOINT") or "https://huggingface.co"
 
 
@@ -182,11 +184,13 @@ def get_download_links_from_huggingface(self, model, branch, text_only=False, sp
         is_llamacpp = has_gguf and specific_file is not None
         return links, sha256, is_lora, is_llamacpp, file_sizes
 
-    def get_output_folder(self, model, branch, is_lora, is_llamacpp=False, model_dir=None):
+    def get_output_folder(self, model, branch, is_lora, is_llamacpp=False, model_dir=None, user_data_dir=None):
         if model_dir:
             base_folder = model_dir
         else:
-            base_folder = 'user_data/models' if not is_lora else 'user_data/loras'
+            if user_data_dir is None:
+                user_data_dir = resolve_user_data_dir()
+            base_folder = str(user_data_dir / 'models') if not is_lora else str(user_data_dir / 'loras')
 
         # If the model is of type GGUF, save directly in the base_folder
         if is_llamacpp:
@@ -392,7 +396,8 @@ def check_model_files(self, model, branch, links, sha256, output_folder):
     parser.add_argument('--specific-file', type=str, default=None, help='Name of the specific file to download (if not provided, downloads all).')
     parser.add_argument('--exclude-pattern', type=str, default=None, help='Regex pattern to exclude files from download.')
     parser.add_argument('--output', type=str, default=None, help='Save the model files to this folder.')
-    parser.add_argument('--model-dir', type=str, default=None, help='Save the model files to a subfolder of this folder instead of the default one (text-generation-webui/user_data/models).')
+    parser.add_argument('--model-dir', type=str, default=None, help='Save the model files to a subfolder of this folder instead of the default one (user_data/models).')
+    parser.add_argument('--user-data-dir', type=str, default=None, help='Path to the user data directory. Overrides auto-detection.')
     parser.add_argument('--clean', action='store_true', help='Does not resume the previous download.')
     parser.add_argument('--check', action='store_true', help='Validates the checksums of model files.')
     parser.add_argument('--max-retries', type=int, default=7, help='Max retries count when get error in download time.')
@@ -421,10 +426,11 @@ def check_model_files(self, model, branch, links, sha256, output_folder):
     )
 
     # Get the output folder
+    user_data_dir = Path(args.user_data_dir) if args.user_data_dir else None
     if args.output:
         output_folder = Path(args.output)
     else:
-        output_folder = downloader.get_output_folder(model, branch, is_lora, is_llamacpp=is_llamacpp, model_dir=args.model_dir)
+        output_folder = downloader.get_output_folder(model, branch, is_lora, is_llamacpp=is_llamacpp, model_dir=args.model_dir, user_data_dir=user_data_dir)
 
     if args.check:
         # Check previously downloaded files
diff --git a/modules/chat.py b/modules/chat.py
index f9a98bb6a9..7c58542fd8 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1126,9 +1126,9 @@ def start_new_chat(state):
 
 def get_history_file_path(unique_id, character, mode):
     if mode == 'instruct':
-        p = Path(f'user_data/logs/instruct/{unique_id}.json')
+        p = shared.user_data_dir / 'logs' / 'instruct' / f'{unique_id}.json'
     else:
-        p = Path(f'user_data/logs/chat/{character}/{unique_id}.json')
+        p = shared.user_data_dir / 'logs' / 'chat' / character / f'{unique_id}.json'
 
     return p
 
@@ -1164,13 +1164,13 @@ def rename_history(old_id, new_id, character, mode):
 
 def get_paths(state):
     if state['mode'] == 'instruct':
-        return Path('user_data/logs/instruct').glob('*.json')
+        return (shared.user_data_dir / 'logs' / 'instruct').glob('*.json')
     else:
         character = state['character_menu']
 
         # Handle obsolete filenames and paths
-        old_p = Path(f'user_data/logs/{character}_persistent.json')
-        new_p = Path(f'user_data/logs/persistent_{character}.json')
+        old_p = shared.user_data_dir / 'logs' / f'{character}_persistent.json'
+        new_p = shared.user_data_dir / 'logs' / f'persistent_{character}.json'
         if old_p.exists():
             logger.warning(f"Renaming \"{old_p}\" to \"{new_p}\"")
             old_p.rename(new_p)
@@ -1182,7 +1182,7 @@ def get_paths(state):
             p.parent.mkdir(exist_ok=True)
             new_p.rename(p)
 
-        return Path(f'user_data/logs/chat/{character}').glob('*.json')
+        return (shared.user_data_dir / 'logs' / 'chat' / character).glob('*.json')
 
 
 def find_all_histories(state):
@@ -1307,7 +1307,7 @@ def get_chat_state_key(character, mode):
 
 def load_last_chat_state():
     """Load the last chat state from file"""
-    state_file = Path('user_data/logs/chat_state.json')
+    state_file = shared.user_data_dir / 'logs' / 'chat_state.json'
     if state_file.exists():
         try:
             with open(state_file, 'r', encoding='utf-8') as f:
@@ -1327,7 +1327,7 @@ def save_last_chat_state(character, mode, unique_id):
     key = get_chat_state_key(character, mode)
     state["last_chats"][key] = unique_id
 
-    state_file = Path('user_data/logs/chat_state.json')
+    state_file = shared.user_data_dir / 'logs' / 'chat_state.json'
     state_file.parent.mkdir(exist_ok=True)
     with open(state_file, 'w', encoding='utf-8') as f:
         f.write(json.dumps(state, indent=2))
@@ -1403,7 +1403,7 @@ def generate_pfp_cache(character):
     if not cache_folder.exists():
         cache_folder.mkdir()
 
-    for path in [Path(f"user_data/characters/{character}.{extension}") for extension in ['png', 'jpg', 'jpeg']]:
+    for path in [shared.user_data_dir / 'characters' / f"{character}.{extension}" for extension in ['png', 'jpg', 'jpeg']]:
         if path.exists():
             original_img = Image.open(path)
             # Define file paths
@@ -1428,12 +1428,12 @@ def load_character(character, name1, name2):
 
     filepath = None
     for extension in ["yml", "yaml", "json"]:
-        filepath = Path(f'user_data/characters/{character}.{extension}')
+        filepath = shared.user_data_dir / 'characters' / f'{character}.{extension}'
         if filepath.exists():
             break
 
     if filepath is None or not filepath.exists():
-        logger.error(f"Could not find the character \"{character}\" inside user_data/characters. No character has been loaded.")
+        logger.error(f"Could not find the character \"{character}\" inside {shared.user_data_dir}/characters. No character has been loaded.")
         raise ValueError
 
     file_contents = open(filepath, 'r', encoding='utf-8').read()
@@ -1509,7 +1509,7 @@ def load_instruction_template(template):
     if template == 'None':
         return ''
 
-    for filepath in [Path(f'user_data/instruction-templates/{template}.yaml'), Path('user_data/instruction-templates/Alpaca.yaml')]:
+    for filepath in [shared.user_data_dir / 'instruction-templates' / f'{template}.yaml', shared.user_data_dir / 'instruction-templates' / 'Alpaca.yaml']:
         if filepath.exists():
             break
     else:
@@ -1552,17 +1552,17 @@ def upload_character(file, img_path, tavern=False):
 
     outfile_name = name
     i = 1
-    while Path(f'user_data/characters/{outfile_name}.yaml').exists():
+    while (shared.user_data_dir / 'characters' / f'{outfile_name}.yaml').exists():
         outfile_name = f'{name}_{i:03d}'
         i += 1
 
-    with open(Path(f'user_data/characters/{outfile_name}.yaml'), 'w', encoding='utf-8') as f:
+    with open(shared.user_data_dir / 'characters' / f'{outfile_name}.yaml', 'w', encoding='utf-8') as f:
         f.write(yaml_data)
 
     if img is not None:
-        img.save(Path(f'user_data/characters/{outfile_name}.png'))
+        img.save(shared.user_data_dir / 'characters' / f'{outfile_name}.png')
 
-    logger.info(f'New character saved to "user_data/characters/{outfile_name}.yaml".')
+    logger.info(f'New character saved to "{shared.user_data_dir}/characters/{outfile_name}.yaml".')
     return gr.update(value=outfile_name, choices=get_available_characters())
 
 
@@ -1643,9 +1643,9 @@ def save_character(name, greeting, context, picture, filename):
         return
 
     data = generate_character_yaml(name, greeting, context)
-    filepath = Path(f'user_data/characters/{filename}.yaml')
+    filepath = shared.user_data_dir / 'characters' / f'{filename}.yaml'
     save_file(filepath, data)
-    path_to_img = Path(f'user_data/characters/{filename}.png')
+    path_to_img = shared.user_data_dir / 'characters' / f'{filename}.png'
     if picture is not None:
         # Copy the image file from its source path to the character folder
         shutil.copy(picture, path_to_img)
@@ -1655,11 +1655,11 @@ def save_character(name, greeting, context, picture, filename):
 def delete_character(name, instruct=False):
     # Check for character data files
     for extension in ["yml", "yaml", "json"]:
-        delete_file(Path(f'user_data/characters/{name}.{extension}'))
+        delete_file(shared.user_data_dir / 'characters' / f'{name}.{extension}')
 
     # Check for character image files
     for extension in ["png", "jpg", "jpeg"]:
-        delete_file(Path(f'user_data/characters/{name}.{extension}'))
+        delete_file(shared.user_data_dir / 'characters' / f'{name}.{extension}')
 
 
 def generate_user_pfp_cache(user):
@@ -1668,7 +1668,7 @@ def generate_user_pfp_cache(user):
     if not cache_folder.exists():
         cache_folder.mkdir()
 
-    for path in [Path(f"user_data/users/{user}.{extension}") for extension in ['png', 'jpg', 'jpeg']]:
+    for path in [shared.user_data_dir / 'users' / f"{user}.{extension}" for extension in ['png', 'jpg', 'jpeg']]:
         if path.exists():
             original_img = Image.open(path)
             # Define file paths
@@ -1690,12 +1690,12 @@ def load_user(user_name, name1, user_bio):
 
     filepath = None
     for extension in ["yml", "yaml", "json"]:
-        filepath = Path(f'user_data/users/{user_name}.{extension}')
+        filepath = shared.user_data_dir / 'users' / f'{user_name}.{extension}'
         if filepath.exists():
             break
 
     if filepath is None or not filepath.exists():
-        logger.error(f"Could not find the user \"{user_name}\" inside user_data/users. No user has been loaded.")
+        logger.error(f"Could not find the user \"{user_name}\" inside {shared.user_data_dir}/users. No user has been loaded.")
         raise ValueError
 
     with open(filepath, 'r', encoding='utf-8') as f:
@@ -1741,14 +1741,14 @@ def save_user(name, user_bio, picture, filename):
         return
 
     # Ensure the users directory exists
-    users_dir = Path('user_data/users')
+    users_dir = shared.user_data_dir / 'users'
     users_dir.mkdir(parents=True, exist_ok=True)
 
     data = generate_user_yaml(name, user_bio)
-    filepath = Path(f'user_data/users/{filename}.yaml')
+    filepath = shared.user_data_dir / 'users' / f'{filename}.yaml'
     save_file(filepath, data)
 
-    path_to_img = Path(f'user_data/users/{filename}.png')
+    path_to_img = shared.user_data_dir / 'users' / f'{filename}.png'
     if picture is not None:
         # Copy the image file from its source path to the users folder
         shutil.copy(picture, path_to_img)
@@ -1759,11 +1759,11 @@ def delete_user(name):
     """Delete user profile files"""
     # Check for user data files
     for extension in ["yml", "yaml", "json"]:
-        delete_file(Path(f'user_data/users/{name}.{extension}'))
+        delete_file(shared.user_data_dir / 'users' / f'{name}.{extension}')
 
     # Check for user image files
     for extension in ["png", "jpg", "jpeg"]:
-        delete_file(Path(f'user_data/users/{name}.{extension}'))
+        delete_file(shared.user_data_dir / 'users' / f'{name}.{extension}')
 
 
 def update_user_menu_after_deletion(idx):
@@ -2224,7 +2224,7 @@ def handle_save_template_click(instruction_template_str):
     contents = generate_instruction_template_yaml(instruction_template_str)
     return [
         "My Template.yaml",
-        "user_data/instruction-templates/",
+        str(shared.user_data_dir / 'instruction-templates') + '/',
         contents,
         gr.update(visible=True)
     ]
@@ -2233,7 +2233,7 @@ def handle_save_template_click(instruction_template_str):
 def handle_delete_template_click(template):
     return [
         f"{template}.yaml",
-        "user_data/instruction-templates/",
+        str(shared.user_data_dir / 'instruction-templates') + '/',
         gr.update(visible=False)
     ]
 
diff --git a/modules/evaluate.py b/modules/evaluate.py
index e3562d5ff8..78d375cd74 100644
--- a/modules/evaluate.py
+++ b/modules/evaluate.py
@@ -12,8 +12,8 @@
 
 
 def load_past_evaluations():
-    if Path('user_data/logs/evaluations.csv').exists():
-        df = pd.read_csv(Path('user_data/logs/evaluations.csv'), dtype=str)
+    if (shared.user_data_dir / 'logs' / 'evaluations.csv').exists():
+        df = pd.read_csv(shared.user_data_dir / 'logs' / 'evaluations.csv', dtype=str)
         df['Perplexity'] = pd.to_numeric(df['Perplexity'])
         return df
     else:
@@ -26,7 +26,7 @@ def load_past_evaluations():
 def save_past_evaluations(df):
     global past_evaluations
     past_evaluations = df
-    filepath = Path('user_data/logs/evaluations.csv')
+    filepath = shared.user_data_dir / 'logs' / 'evaluations.csv'
     filepath.parent.mkdir(parents=True, exist_ok=True)
     df.to_csv(filepath, index=False)
 
@@ -65,7 +65,7 @@ def calculate_perplexity(models, input_dataset, stride, _max_length):
         data = load_dataset('ptb_text_only', 'penn_treebank', split='test')
         text = " ".join(data['sentence'])
     else:
-        with open(Path(f'user_data/training/datasets/{input_dataset}.txt'), 'r', encoding='utf-8') as f:
+        with open(shared.user_data_dir / 'training' / 'datasets' / f'{input_dataset}.txt', 'r', encoding='utf-8') as f:
             text = f.read()
 
     for model in models:
diff --git a/modules/extensions.py b/modules/extensions.py
index e00103124e..dd3278822d 100644
--- a/modules/extensions.py
+++ b/modules/extensions.py
@@ -1,4 +1,6 @@
 import importlib
+import importlib.util
+import sys
 import traceback
 from functools import partial
 from inspect import signature
@@ -38,9 +40,15 @@ def load_extensions():
 
         try:
             # Prefer user extension, fall back to system extension
-            user_script_path = Path(f'user_data/extensions/{name}/script.py')
+            user_script_path = shared.user_data_dir / 'extensions' / name / 'script.py'
             if user_script_path.exists():
-                extension = importlib.import_module(f"user_data.extensions.{name}.script")
+                spec = importlib.util.spec_from_file_location(
+                    f"user_ext_{name}",
+                    str(user_script_path)
+                )
+                extension = importlib.util.module_from_spec(spec)
+                sys.modules[spec.name] = extension
+                spec.loader.exec_module(extension)
             else:
                 extension = importlib.import_module(f"extensions.{name}.script")
 
@@ -53,7 +61,7 @@ def load_extensions():
             state[name] = [True, i, extension]  # Store extension object
 
         except ModuleNotFoundError:
-            extension_location = Path('user_data/extensions') / name if user_script_path.exists() else Path('extensions') / name
+            extension_location = shared.user_data_dir / 'extensions' / name if user_script_path.exists() else Path('extensions') / name
             windows_path = str(extension_location).replace('/', '\\')
             logger.error(
                 f"Could not import the requirements for '{name}'. Make sure to install the requirements for the extension.\n\n"
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 667a64d6b3..34a5bc5762 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -627,10 +627,10 @@ def create_message(role, content, raw_content):
 
 def get_character_image_with_cache_buster():
     """Get character image URL with cache busting based on file modification time"""
-    cache_path = Path("user_data/cache/pfp_character_thumb.png")
+    cache_path = shared.user_data_dir / "cache" / "pfp_character_thumb.png"
     if cache_path.exists():
         mtime = int(cache_path.stat().st_mtime)
-        return f''
+        return f''
 
     return ''
 
@@ -654,8 +654,8 @@ def create_message(role, content, raw_content):
 
         # Get appropriate image
         if role == "user":
-            img = (f''
-                   if Path("user_data/cache/pfp_me.png").exists() else '')
+            img = (f''
+                   if (shared.user_data_dir / "cache" / "pfp_me.png").exists() else '')
         else:
             img = img_bot
 
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 017c5d2a61..eb99f2a86d 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -384,7 +384,7 @@ def _start_server(self):
         if shared.args.mmproj not in [None, 'None']:
             path = Path(shared.args.mmproj)
             if not path.exists():
-                path = Path('user_data/mmproj') / shared.args.mmproj
+                path = shared.user_data_dir / 'mmproj' / shared.args.mmproj
 
             if path.exists():
                 cmd += ["--mmproj", str(path)]
diff --git a/modules/presets.py b/modules/presets.py
index 16dfa6adb8..9aab7e3c8c 100644
--- a/modules/presets.py
+++ b/modules/presets.py
@@ -66,7 +66,7 @@ def presets_params():
 def load_preset(name, verbose=False):
     generate_params = default_preset()
     if name not in ['None', None, '']:
-        path = Path(f'user_data/presets/{name}.yaml')
+        path = shared.user_data_dir / 'presets' / f'{name}.yaml'
         if path.exists():
             with open(path, 'r') as infile:
                 preset = yaml.safe_load(infile)
diff --git a/modules/prompts.py b/modules/prompts.py
index 91f5812a04..d107ce5a9d 100644
--- a/modules/prompts.py
+++ b/modules/prompts.py
@@ -8,7 +8,7 @@ def load_prompt(fname):
     if not fname:
         # Create new file
         new_name = utils.current_time()
-        prompt_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
+        prompt_path = shared.user_data_dir / "logs" / "notebook" / f"{new_name}.txt"
         prompt_path.parent.mkdir(parents=True, exist_ok=True)
         initial_content = "In this story,"
         prompt_path.write_text(initial_content, encoding='utf-8')
@@ -18,7 +18,7 @@ def load_prompt(fname):
 
         return initial_content
 
-    file_path = Path(f'user_data/logs/notebook/{fname}.txt')
+    file_path = shared.user_data_dir / 'logs' / 'notebook' / f'{fname}.txt'
     if file_path.exists():
         with open(file_path, 'r', encoding='utf-8') as f:
             text = f.read()
diff --git a/modules/shared.py b/modules/shared.py
index 4e212d1b31..57b7e3e991 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -9,8 +9,12 @@
 import yaml
 
 from modules.logging_colors import logger
+from modules.paths import resolve_user_data_dir
 from modules.presets import default_preset
 
+# Resolve user_data directory early (before argparse defaults are set)
+user_data_dir = resolve_user_data_dir()
+
 # Text model variables
 model = None
 tokenizer = None
@@ -42,11 +46,12 @@
 
 # Basic settings
 group = parser.add_argument_group('Basic settings')
+group.add_argument('--user-data-dir', type=str, default=str(user_data_dir), help='Path to the user data directory. Default: auto-detected.')
 group.add_argument('--multi-user', action='store_true', help='Multi-user mode. Chat histories are not saved or automatically loaded. Warning: this is likely not safe for sharing publicly.')
 group.add_argument('--model', type=str, help='Name of the model to load by default.')
 group.add_argument('--lora', type=str, nargs='+', help='The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.')
-group.add_argument('--model-dir', type=str, default='user_data/models', help='Path to directory with all the models.')
-group.add_argument('--lora-dir', type=str, default='user_data/loras', help='Path to directory with all the loras.')
+group.add_argument('--model-dir', type=str, default=str(user_data_dir / 'models'), help='Path to directory with all the models.')
+group.add_argument('--lora-dir', type=str, default=str(user_data_dir / 'loras'), help='Path to directory with all the loras.')
 group.add_argument('--model-menu', action='store_true', help='Show a model menu in the terminal when the web UI is first launched.')
 group.add_argument('--settings', type=str, help='Load the default interface settings from this yaml file. See user_data/settings-template.yaml for an example. If you create a file called user_data/settings.yaml, this file will be loaded by default without the need to use the --settings flag.')
 group.add_argument('--extensions', type=str, nargs='+', help='The list of extensions to load. If you want to load more than one extension, write the names separated by spaces.')
@@ -56,7 +61,7 @@
 # Image generation
 group = parser.add_argument_group('Image model')
 group.add_argument('--image-model', type=str, help='Name of the image model to select on startup (overrides saved setting).')
-group.add_argument('--image-model-dir', type=str, default='user_data/image_models', help='Path to directory with all the image models.')
+group.add_argument('--image-model-dir', type=str, default=str(user_data_dir / 'image_models'), help='Path to directory with all the image models.')
 group.add_argument('--image-dtype', type=str, default=None, choices=['bfloat16', 'float16'], help='Data type for image model.')
 group.add_argument('--image-attn-backend', type=str, default=None, choices=['flash_attention_2', 'sdpa'], help='Attention backend for image model.')
 group.add_argument('--image-cpu-offload', action='store_true', help='Enable CPU offloading for image model.')
@@ -110,7 +115,7 @@
 group.add_argument('--cpu', action='store_true', help='Use the CPU to generate text. Warning: Training on CPU is extremely slow.')
 group.add_argument('--cpu-memory', type=float, default=0, help='Maximum CPU memory in GiB. Use this for CPU offloading.')
 group.add_argument('--disk', action='store_true', help='If the model is too large for your GPU(s) and CPU combined, send the remaining layers to the disk.')
-group.add_argument('--disk-cache-dir', type=str, default='user_data/cache', help='Directory to save the disk cache to. Defaults to "user_data/cache".')
+group.add_argument('--disk-cache-dir', type=str, default=str(user_data_dir / 'cache'), help='Directory to save the disk cache to.')
 group.add_argument('--load-in-8bit', action='store_true', help='Load the model with 8-bit precision (using bitsandbytes).')
 group.add_argument('--bf16', action='store_true', help='Load the model with bfloat16 precision. Requires NVIDIA Ampere GPU.')
 group.add_argument('--no-cache', action='store_true', help='Set use_cache to False while generating text. This reduces VRAM usage slightly, but it comes at a performance cost.')
@@ -167,7 +172,7 @@
 group.add_argument('--nowebui', action='store_true', help='Do not launch the Gradio UI. Useful for launching the API in standalone mode.')
 
 # Handle CMD_FLAGS.txt
-cmd_flags_path = Path(__file__).parent.parent / "user_data" / "CMD_FLAGS.txt"
+cmd_flags_path = user_data_dir / "CMD_FLAGS.txt"
 if cmd_flags_path.exists():
     with cmd_flags_path.open('r', encoding='utf-8') as f:
         cmd_flags = ' '.join(
@@ -182,6 +187,7 @@
 
 
 args = parser.parse_args()
+user_data_dir = Path(args.user_data_dir)  # Update from parsed args (may differ from pre-parse)
 original_args = copy.deepcopy(args)
 args_defaults = parser.parse_args([])
 
@@ -212,7 +218,7 @@
     'enable_web_search': False,
     'web_search_pages': 3,
     'prompt-notebook': '',
-    'preset': 'Qwen3 - Thinking' if Path('user_data/presets/Qwen3 - Thinking.yaml').exists() else None,
+    'preset': 'Qwen3 - Thinking' if (user_data_dir / 'presets/Qwen3 - Thinking.yaml').exists() else None,
     'max_new_tokens': 512,
     'max_new_tokens_min': 1,
     'max_new_tokens_max': 4096,
diff --git a/modules/training.py b/modules/training.py
index ca8d1bd150..c9f32e6444 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -107,8 +107,8 @@ def create_ui():
                 with gr.Column():
                     with gr.Tab(label='Chat Dataset'):
                         with gr.Row():
-                            dataset = gr.Dropdown(choices=utils.get_chat_datasets('user_data/training/datasets'), value='None', label='Dataset File', info='A JSON file with chat conversations (messages or ShareGPT format). Each row is one conversation.', elem_classes=['slim-dropdown'], interactive=not mu)
-                            ui.create_refresh_button(dataset, lambda: None, lambda: {'choices': utils.get_chat_datasets('user_data/training/datasets')}, 'refresh-button', interactive=not mu)
+                            dataset = gr.Dropdown(choices=utils.get_chat_datasets(str(shared.user_data_dir / 'training/datasets')), value='None', label='Dataset File', info='A JSON file with chat conversations (messages or ShareGPT format). Each row is one conversation.', elem_classes=['slim-dropdown'], interactive=not mu)
+                            ui.create_refresh_button(dataset, lambda: None, lambda: {'choices': utils.get_chat_datasets(str(shared.user_data_dir / 'training/datasets'))}, 'refresh-button', interactive=not mu)
 
                         with gr.Row():
                             format = gr.Dropdown(choices=get_instruction_templates(), value='None', label='Instruction Template', info='Select an instruction template for formatting the dataset, or "Chat Template" to use the model\'s built-in chat template.', elem_classes=['slim-dropdown'], interactive=not mu)
@@ -116,14 +116,14 @@ def create_ui():
 
                     with gr.Tab(label="Text Dataset"):
                         with gr.Row():
-                            text_dataset = gr.Dropdown(choices=utils.get_text_datasets('user_data/training/datasets'), value='None', label='Dataset File', info='A JSON file with a "text" key per row, for pretraining-style training. Each row is one document.', elem_classes=['slim-dropdown'], interactive=not mu)
-                            ui.create_refresh_button(text_dataset, lambda: None, lambda: {'choices': utils.get_text_datasets('user_data/training/datasets')}, 'refresh-button', interactive=not mu)
+                            text_dataset = gr.Dropdown(choices=utils.get_text_datasets(str(shared.user_data_dir / 'training/datasets')), value='None', label='Dataset File', info='A JSON file with a "text" key per row, for pretraining-style training. Each row is one document.', elem_classes=['slim-dropdown'], interactive=not mu)
+                            ui.create_refresh_button(text_dataset, lambda: None, lambda: {'choices': utils.get_text_datasets(str(shared.user_data_dir / 'training/datasets'))}, 'refresh-button', interactive=not mu)
 
                         stride_length = gr.Slider(label='Stride Length', minimum=0, maximum=2048, value=256, step=32, info='Overlap between chunks in tokens. 0 = no overlap. Values like 256 or 512 help preserve context across chunk boundaries.')
 
                     with gr.Row():
-                        eval_dataset = gr.Dropdown(choices=utils.get_datasets('user_data/training/datasets', 'json'), value='None', label='Evaluation Dataset', info='The (optional) dataset file used to evaluate the model after training.', elem_classes=['slim-dropdown'], interactive=not mu)
-                        ui.create_refresh_button(eval_dataset, lambda: None, lambda: {'choices': utils.get_datasets('user_data/training/datasets', 'json')}, 'refresh-button', interactive=not mu)
+                        eval_dataset = gr.Dropdown(choices=utils.get_datasets(str(shared.user_data_dir / 'training/datasets'), 'json'), value='None', label='Evaluation Dataset', info='The (optional) dataset file used to evaluate the model after training.', elem_classes=['slim-dropdown'], interactive=not mu)
+                        ui.create_refresh_button(eval_dataset, lambda: None, lambda: {'choices': utils.get_datasets(str(shared.user_data_dir / 'training/datasets'), 'json')}, 'refresh-button', interactive=not mu)
 
                     eval_steps = gr.Number(label='Evaluate every n steps', value=100, info='If an evaluation dataset is given, test it every time this many steps pass.')
 
@@ -137,7 +137,7 @@ def create_ui():
             with gr.Row():
                 with gr.Column():
                     models = gr.Dropdown(utils.get_available_models(), label='Models', multiselect=True, interactive=not mu)
-                    evaluate_text_file = gr.Dropdown(choices=['wikitext', 'ptb', 'ptb_new'] + utils.get_datasets('user_data/training/datasets', 'txt')[1:], value='wikitext', label='Input dataset', info='The raw text file on which the model will be evaluated. The first options are automatically downloaded: wikitext, ptb, and ptb_new. The next options are your local text files under user_data/training/datasets.', interactive=not mu)
+                    evaluate_text_file = gr.Dropdown(choices=['wikitext', 'ptb', 'ptb_new'] + utils.get_datasets(str(shared.user_data_dir / 'training/datasets'), 'txt')[1:], value='wikitext', label='Input dataset', info=f'The raw text file on which the model will be evaluated. The first options are automatically downloaded: wikitext, ptb, and ptb_new. The next options are your local text files under {shared.user_data_dir}/training/datasets.', interactive=not mu)
                     with gr.Row():
                         with gr.Column():
                             stride_length = gr.Slider(label='Stride', minimum=0, maximum=32768, value=512, step=256, info='Used to make the evaluation faster at the cost of accuracy. 1 = slowest but most accurate. 512 is a common value.')
@@ -224,7 +224,7 @@ def clean_path(base_path: str, path: str):
 
 
 def get_instruction_templates():
-    path = Path('user_data/instruction-templates')
+    path = shared.user_data_dir / 'instruction-templates'
     names = set()
     for ext in ['yaml', 'yml', 'jinja']:
         for f in path.glob(f'*.{ext}'):
@@ -233,8 +233,8 @@ def get_instruction_templates():
 
 
 def load_template(name):
-    """Load a Jinja2 template string from user_data/instruction-templates/."""
-    path = Path('user_data/instruction-templates')
+    """Load a Jinja2 template string from {user_data_dir}/instruction-templates/."""
+    path = shared.user_data_dir / 'instruction-templates'
     for ext in ['jinja', 'yaml', 'yml']:
         filepath = path / f'{name}.{ext}'
         if filepath.exists():
@@ -453,7 +453,7 @@ def tokenize_text_data(data):
     if has_text_dataset:
         train_template["template_type"] = "text_dataset"
         logger.info("Loading text dataset")
-        data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{text_dataset}.json'))
+        data = load_dataset("json", data_files=clean_path(str(shared.user_data_dir / 'training/datasets'), f'{text_dataset}.json'))
 
         if "text" not in data['train'].column_names:
             yield "Error: text dataset must have a \"text\" key per row."
@@ -467,7 +467,7 @@ def tokenize_text_data(data):
         if eval_dataset == 'None':
             eval_data = None
         else:
-            eval_raw = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{eval_dataset}.json'))
+            eval_raw = load_dataset("json", data_files=clean_path(str(shared.user_data_dir / 'training/datasets'), f'{eval_dataset}.json'))
             if "text" not in eval_raw['train'].column_names:
                 yield "Error: evaluation dataset must have a \"text\" key per row."
                 return
@@ -496,7 +496,7 @@ def tokenize_text_data(data):
         train_template["template_type"] = "chat_template"
 
         logger.info("Loading JSON dataset with chat template format")
-        data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{dataset}.json'))
+        data = load_dataset("json", data_files=clean_path(str(shared.user_data_dir / 'training/datasets'), f'{dataset}.json'))
 
         # Validate the first row
         try:
@@ -522,7 +522,7 @@ def tokenize_text_data(data):
         if eval_dataset == 'None':
             eval_data = None
         else:
-            eval_data = load_dataset("json", data_files=clean_path('user_data/training/datasets', f'{eval_dataset}.json'))
+            eval_data = load_dataset("json", data_files=clean_path(str(shared.user_data_dir / 'training/datasets'), f'{eval_dataset}.json'))
             eval_data = eval_data['train'].map(
                 tokenize_conversation,
                 remove_columns=eval_data['train'].column_names,
@@ -757,11 +757,11 @@ def log_train_dataset(trainer):
                 decoded_entries.append({"value": decoded_text})
 
             # Write the log file
-            Path('user_data/logs').mkdir(exist_ok=True)
-            with open(Path('user_data/logs/train_dataset_sample.json'), 'w') as json_file:
+            (shared.user_data_dir / 'logs').mkdir(exist_ok=True)
+            with open(shared.user_data_dir / 'logs' / 'train_dataset_sample.json', 'w') as json_file:
                 json.dump(decoded_entries, json_file, indent=4)
 
-            logger.info("Log file 'train_dataset_sample.json' created in the 'user_data/logs' directory.")
+            logger.info(f"Log file 'train_dataset_sample.json' created in the '{shared.user_data_dir}/logs' directory.")
         except Exception as e:
             logger.error(f"Failed to create log file due to error: {e}")
 
diff --git a/modules/ui.py b/modules/ui.py
index 0cd2cf43b7..ae998ebb41 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -113,7 +113,7 @@
         block_radius='0',
     )
 
-if Path("user_data/notification.mp3").exists():
+if (shared.user_data_dir / "notification.mp3").exists():
     audio_notification_js = "document.querySelector('#audio_notification audio')?.play();"
 else:
     audio_notification_js = ""
@@ -381,7 +381,7 @@ def save_settings(state, preset, extensions_list, show_controls, theme_state, ma
                             output[_id] = params[param]
     else:
         # Preserve existing extensions and extension parameters during autosave
-        settings_path = Path('user_data') / 'settings.yaml'
+        settings_path = shared.user_data_dir / 'settings.yaml'
         if settings_path.exists():
             try:
                 with open(settings_path, 'r', encoding='utf-8') as f:
@@ -436,7 +436,7 @@ def _perform_debounced_save():
     try:
         if _last_interface_state is not None:
             contents = save_settings(_last_interface_state, _last_preset, _last_extensions, _last_show_controls, _last_theme_state, manual_save=False)
-            settings_path = Path('user_data') / 'settings.yaml'
+            settings_path = shared.user_data_dir / 'settings.yaml'
             settings_path.parent.mkdir(exist_ok=True)
             with open(settings_path, 'w', encoding='utf-8') as f:
                 f.write(contents)
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index aec9051ea5..74da0a40c5 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -175,7 +175,7 @@ def create_character_settings_ui():
 
             with gr.Column(scale=1):
                 shared.gradio['character_picture'] = gr.Image(label='Character picture', type='filepath', interactive=not mu)
-                shared.gradio['your_picture'] = gr.Image(label='Your picture', type='filepath', value=Image.open(Path('user_data/cache/pfp_me.png')) if Path('user_data/cache/pfp_me.png').exists() else None, interactive=not mu)
+                shared.gradio['your_picture'] = gr.Image(label='Your picture', type='filepath', value=Image.open(shared.user_data_dir / 'cache' / 'pfp_me.png') if (shared.user_data_dir / 'cache' / 'pfp_me.png').exists() else None, interactive=not mu)
 
 
 def create_chat_settings_ui():
diff --git a/modules/ui_default.py b/modules/ui_default.py
index c0feae1947..2c367ccaad 100644
--- a/modules/ui_default.py
+++ b/modules/ui_default.py
@@ -159,7 +159,7 @@ def handle_new_prompt():
     new_name = utils.current_time()
 
     # Create the new prompt file
-    prompt_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
+    prompt_path = shared.user_data_dir / "logs" / "notebook" / f"{new_name}.txt"
     prompt_path.parent.mkdir(parents=True, exist_ok=True)
     prompt_path.write_text("In this story,", encoding='utf-8')
 
@@ -170,15 +170,15 @@ def handle_delete_prompt_confirm_default(prompt_name):
     available_prompts = utils.get_available_prompts()
     current_index = available_prompts.index(prompt_name) if prompt_name in available_prompts else 0
 
-    (Path("user_data/logs/notebook") / f"{prompt_name}.txt").unlink(missing_ok=True)
+    (shared.user_data_dir / "logs" / "notebook" / f"{prompt_name}.txt").unlink(missing_ok=True)
     available_prompts = utils.get_available_prompts()
 
     if available_prompts:
         new_value = available_prompts[min(current_index, len(available_prompts) - 1)]
     else:
         new_value = utils.current_time()
-        Path("user_data/logs/notebook").mkdir(parents=True, exist_ok=True)
-        (Path("user_data/logs/notebook") / f"{new_value}.txt").write_text("In this story,")
+        (shared.user_data_dir / "logs" / "notebook").mkdir(parents=True, exist_ok=True)
+        (shared.user_data_dir / "logs" / "notebook" / f"{new_value}.txt").write_text("In this story,")
         available_prompts = [new_value]
 
     return [
@@ -199,8 +199,8 @@ def handle_rename_prompt_click_default(current_name):
 
 
 def handle_rename_prompt_confirm_default(new_name, current_name):
-    old_path = Path("user_data/logs/notebook") / f"{current_name}.txt"
-    new_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
+    old_path = shared.user_data_dir / "logs" / "notebook" / f"{current_name}.txt"
+    new_path = shared.user_data_dir / "logs" / "notebook" / f"{new_name}.txt"
 
     if old_path.exists() and not new_path.exists():
         old_path.rename(new_path)
diff --git a/modules/ui_file_saving.py b/modules/ui_file_saving.py
index 720bfdecf1..46087acea1 100644
--- a/modules/ui_file_saving.py
+++ b/modules/ui_file_saving.py
@@ -28,7 +28,7 @@ def create_ui():
 
     # Character saver/deleter
     with gr.Group(visible=False, elem_classes='file-saver') as shared.gradio['character_saver']:
-        shared.gradio['save_character_filename'] = gr.Textbox(lines=1, label='File name', info='The character will be saved to your user_data/characters folder with this base filename.')
+        shared.gradio['save_character_filename'] = gr.Textbox(lines=1, label='File name', info=f'The character will be saved to your {shared.user_data_dir}/characters folder with this base filename.')
         with gr.Row():
             shared.gradio['save_character_cancel'] = gr.Button('Cancel', elem_classes="small-button")
             shared.gradio['save_character_confirm'] = gr.Button('Save', elem_classes="small-button", variant='primary', interactive=not mu)
@@ -41,7 +41,7 @@ def create_ui():
 
     # User saver/deleter
     with gr.Group(visible=False, elem_classes='file-saver') as shared.gradio['user_saver']:
-        shared.gradio['save_user_filename'] = gr.Textbox(lines=1, label='File name', info='The user profile will be saved to your user_data/users folder with this base filename.')
+        shared.gradio['save_user_filename'] = gr.Textbox(lines=1, label='File name', info=f'The user profile will be saved to your {shared.user_data_dir}/users folder with this base filename.')
         with gr.Row():
             shared.gradio['save_user_cancel'] = gr.Button('Cancel', elem_classes="small-button")
             shared.gradio['save_user_confirm'] = gr.Button('Save', elem_classes="small-button", variant='primary', interactive=not mu)
@@ -54,7 +54,7 @@ def create_ui():
 
     # Preset saver
     with gr.Group(visible=False, elem_classes='file-saver') as shared.gradio['preset_saver']:
-        shared.gradio['save_preset_filename'] = gr.Textbox(lines=1, label='File name', info='The preset will be saved to your user_data/presets folder with this base filename.')
+        shared.gradio['save_preset_filename'] = gr.Textbox(lines=1, label='File name', info=f'The preset will be saved to your {shared.user_data_dir}/presets folder with this base filename.')
         shared.gradio['save_preset_contents'] = gr.Textbox(lines=10, label='File contents')
         with gr.Row():
             shared.gradio['save_preset_cancel'] = gr.Button('Cancel', elem_classes="small-button")
@@ -91,7 +91,7 @@ def create_event_handlers():
 
 def handle_save_preset_confirm_click(filename, contents):
     try:
-        utils.save_file(f"user_data/presets/{filename}.yaml", contents)
+        utils.save_file(str(shared.user_data_dir / "presets" / f"{filename}.yaml"), contents)
         available_presets = utils.get_available_presets()
         output = gr.update(choices=available_presets, value=filename)
     except Exception:
@@ -164,7 +164,7 @@ def handle_save_preset_click(state):
 def handle_delete_preset_click(preset):
     return [
         f"{preset}.yaml",
-        "user_data/presets/",
+        str(shared.user_data_dir / "presets") + "/",
         gr.update(visible=True)
     ]
 
@@ -173,7 +173,7 @@ def handle_save_grammar_click(grammar_string):
     return [
         grammar_string,
         "My Fancy Grammar.gbnf",
-        "user_data/grammars/",
+        str(shared.user_data_dir / "grammars") + "/",
         gr.update(visible=True)
     ]
 
@@ -181,7 +181,7 @@ def handle_save_grammar_click(grammar_string):
 def handle_delete_grammar_click(grammar_file):
     return [
         grammar_file,
-        "user_data/grammars/",
+        str(shared.user_data_dir / "grammars") + "/",
         gr.update(visible=True)
     ]
 
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 6b39c5b5a0..e9df9bd37c 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -138,7 +138,7 @@ def save_generated_images(images, state, actual_seed):
         return []
 
     date_str = datetime.now().strftime("%Y-%m-%d")
-    folder_path = os.path.join("user_data", "image_outputs", date_str)
+    folder_path = str(shared.user_data_dir / "image_outputs" / date_str)
     os.makedirs(folder_path, exist_ok=True)
 
     metadata = build_generation_metadata(state, actual_seed)
@@ -214,7 +214,7 @@ def get_all_history_images(force_refresh=False):
     """Get all history images sorted by modification time (newest first). Uses caching."""
     global _image_cache, _cache_timestamp
 
-    output_dir = os.path.join("user_data", "image_outputs")
+    output_dir = str(shared.user_data_dir / "image_outputs")
     if not os.path.exists(output_dir):
         return []
 
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index e6e57a2255..33b39a2569 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -65,7 +65,7 @@ def create_ui():
                             # Multimodal
                             with gr.Accordion("Multimodal (vision)", open=False, elem_classes='tgw-accordion') as shared.gradio['mmproj_accordion']:
                                 with gr.Row():
-                                    shared.gradio['mmproj'] = gr.Dropdown(label="mmproj file", choices=utils.get_available_mmproj(), value=lambda: shared.args.mmproj or 'None', elem_classes='slim-dropdown', info='Select a file that matches your model. Must be placed in user_data/mmproj/', interactive=not mu)
+                                    shared.gradio['mmproj'] = gr.Dropdown(label="mmproj file", choices=utils.get_available_mmproj(), value=lambda: shared.args.mmproj or 'None', elem_classes='slim-dropdown', info=f'Select a file that matches your model. Must be placed in {shared.user_data_dir}/mmproj/', interactive=not mu)
                                     ui.create_refresh_button(shared.gradio['mmproj'], lambda: None, lambda: {'choices': utils.get_available_mmproj()}, 'refresh-button', interactive=not mu)
 
                             # Speculative decoding
@@ -317,9 +317,9 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
             model_dir=shared.args.model_dir if shared.args.model_dir != shared.args_defaults.model_dir else None
         )
 
-        if output_folder == Path("user_data/models"):
+        if output_folder == shared.user_data_dir / "models":
             output_folder = Path(shared.args.model_dir)
-        elif output_folder == Path("user_data/loras"):
+        elif output_folder == shared.user_data_dir / "loras":
             output_folder = Path(shared.args.lora_dir)
 
         if check:
diff --git a/modules/ui_notebook.py b/modules/ui_notebook.py
index 9fab879b82..f550e6466b 100644
--- a/modules/ui_notebook.py
+++ b/modules/ui_notebook.py
@@ -194,7 +194,7 @@ def handle_new_prompt():
     new_name = utils.current_time()
 
     # Create the new prompt file
-    prompt_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
+    prompt_path = shared.user_data_dir / "logs" / "notebook" / f"{new_name}.txt"
     prompt_path.parent.mkdir(parents=True, exist_ok=True)
     prompt_path.write_text("In this story,", encoding='utf-8')
 
@@ -205,15 +205,15 @@ def handle_delete_prompt_confirm_notebook(prompt_name):
     available_prompts = utils.get_available_prompts()
     current_index = available_prompts.index(prompt_name) if prompt_name in available_prompts else 0
 
-    (Path("user_data/logs/notebook") / f"{prompt_name}.txt").unlink(missing_ok=True)
+    (shared.user_data_dir / "logs" / "notebook" / f"{prompt_name}.txt").unlink(missing_ok=True)
     available_prompts = utils.get_available_prompts()
 
     if available_prompts:
         new_value = available_prompts[min(current_index, len(available_prompts) - 1)]
     else:
         new_value = utils.current_time()
-        Path("user_data/logs/notebook").mkdir(parents=True, exist_ok=True)
-        (Path("user_data/logs/notebook") / f"{new_value}.txt").write_text("In this story,")
+        (shared.user_data_dir / "logs" / "notebook").mkdir(parents=True, exist_ok=True)
+        (shared.user_data_dir / "logs" / "notebook" / f"{new_value}.txt").write_text("In this story,")
         available_prompts = [new_value]
 
     return [
@@ -233,8 +233,8 @@ def handle_rename_prompt_click_notebook(current_name):
 
 
 def handle_rename_prompt_confirm_notebook(new_name, current_name):
-    old_path = Path("user_data/logs/notebook") / f"{current_name}.txt"
-    new_path = Path("user_data/logs/notebook") / f"{new_name}.txt"
+    old_path = shared.user_data_dir / "logs" / "notebook" / f"{current_name}.txt"
+    new_path = shared.user_data_dir / "logs" / "notebook" / f"{new_name}.txt"
 
     if old_path.exists() and not new_path.exists():
         old_path.rename(new_path)
@@ -250,7 +250,7 @@ def handle_rename_prompt_confirm_notebook(new_name, current_name):
 def autosave_prompt(text, prompt_name):
     """Automatically save the text to the selected prompt file"""
     if prompt_name and text.strip():
-        prompt_path = Path("user_data/logs/notebook") / f"{prompt_name}.txt"
+        prompt_path = shared.user_data_dir / "logs" / "notebook" / f"{prompt_name}.txt"
         prompt_path.parent.mkdir(parents=True, exist_ok=True)
         prompt_path.write_text(text, encoding='utf-8')
 
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index 23882084de..e5eb9210a3 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -135,7 +135,7 @@ def get_truncation_length():
 
 
 def load_grammar(name):
-    p = Path(f'user_data/grammars/{name}')
+    p = shared.user_data_dir / 'grammars' / name
     if p.exists():
         return open(p, 'r', encoding='utf-8').read()
     else:
diff --git a/modules/ui_session.py b/modules/ui_session.py
index 60b19f476c..e1807deae2 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -17,7 +17,7 @@ def create_ui():
 
             with gr.Column():
                 gr.Markdown("## Extensions & flags")
-                shared.gradio['save_settings'] = gr.Button('Save extensions settings to user_data/settings.yaml', elem_classes='refresh-button', interactive=not mu)
+                shared.gradio['save_settings'] = gr.Button(f'Save extensions settings to {shared.user_data_dir}/settings.yaml', elem_classes='refresh-button', interactive=not mu)
                 shared.gradio['reset_interface'] = gr.Button("Apply flags/extensions and restart", interactive=not mu)
                 with gr.Row():
                     with gr.Column():
@@ -54,7 +54,7 @@ def handle_save_settings(state, preset, extensions, show_controls, theme):
     return [
         contents,
         "settings.yaml",
-        "user_data/",
+        str(shared.user_data_dir) + "/",
         gr.update(visible=True)
     ]
 
diff --git a/modules/utils.py b/modules/utils.py
index 456b5f8375..447685b782 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -15,16 +15,31 @@ def gradio(*keys):
     return [shared.gradio[k] for k in keys]
 
 
+def _is_path_allowed(abs_path_str):
+    """Check if a path is under the project root or the configured user_data directory."""
+    abs_path = Path(abs_path_str).resolve()
+    root_folder = Path(__file__).resolve().parent.parent
+    user_data_resolved = shared.user_data_dir.resolve()
+    try:
+        abs_path.relative_to(root_folder)
+        return True
+    except ValueError:
+        pass
+    try:
+        abs_path.relative_to(user_data_resolved)
+        return True
+    except ValueError:
+        pass
+    return False
+
+
 def save_file(fname, contents):
     if fname == '':
         logger.error('File name is empty!')
         return
 
-    root_folder = Path(__file__).resolve().parent.parent
     abs_path_str = os.path.abspath(fname)
-    rel_path_str = os.path.relpath(abs_path_str, root_folder)
-    rel_path = Path(rel_path_str)
-    if rel_path.parts[0] == '..':
+    if not _is_path_allowed(abs_path_str):
         logger.error(f'Invalid file path: \"{fname}\"')
         return
 
@@ -39,16 +54,14 @@ def delete_file(fname):
         logger.error('File name is empty!')
         return
 
-    root_folder = Path(__file__).resolve().parent.parent
     abs_path_str = os.path.abspath(fname)
-    rel_path_str = os.path.relpath(abs_path_str, root_folder)
-    rel_path = Path(rel_path_str)
-    if rel_path.parts[0] == '..':
+    if not _is_path_allowed(abs_path_str):
         logger.error(f'Invalid file path: \"{fname}\"')
         return
 
-    if rel_path.exists():
-        rel_path.unlink()
+    p = Path(abs_path_str)
+    if p.exists():
+        p.unlink()
         logger.info(f'Deleted \"{fname}\".')
 
 
@@ -75,7 +88,7 @@ def natural_keys(text):
 def check_model_loaded():
     if shared.model_name == 'None' or shared.model is None:
         if len(get_available_models()) == 0:
-            error_msg = "No model is loaded.\n\nTo get started:\n1) Place a GGUF file in your user_data/models folder\n2) Go to the Model tab and select it"
+            error_msg = f"No model is loaded.\n\nTo get started:\n1) Place a GGUF file in your {shared.user_data_dir}/models folder\n2) Go to the Model tab and select it"
             logger.error(error_msg)
             return False, error_msg
         else:
@@ -188,7 +201,7 @@ def get_available_ggufs():
 
 
 def get_available_mmproj():
-    mmproj_dir = Path('user_data/mmproj')
+    mmproj_dir = shared.user_data_dir / 'mmproj'
     if not mmproj_dir.exists():
         return ['None']
 
@@ -201,11 +214,11 @@ def get_available_mmproj():
 
 
 def get_available_presets():
-    return sorted(set((k.stem for k in Path('user_data/presets').glob('*.yaml'))), key=natural_keys)
+    return sorted(set((k.stem for k in (shared.user_data_dir / 'presets').glob('*.yaml'))), key=natural_keys)
 
 
 def get_available_prompts():
-    notebook_dir = Path('user_data/logs/notebook')
+    notebook_dir = shared.user_data_dir / 'logs' / 'notebook'
     notebook_dir.mkdir(parents=True, exist_ok=True)
 
     prompt_files = list(notebook_dir.glob('*.txt'))
@@ -221,19 +234,19 @@ def get_available_prompts():
 
 
 def get_available_characters():
-    paths = (x for x in Path('user_data/characters').iterdir() if x.suffix in ('.json', '.yaml', '.yml'))
+    paths = (x for x in (shared.user_data_dir / 'characters').iterdir() if x.suffix in ('.json', '.yaml', '.yml'))
     return sorted(set((k.stem for k in paths)), key=natural_keys)
 
 
 def get_available_users():
-    users_dir = Path('user_data/users')
+    users_dir = shared.user_data_dir / 'users'
     users_dir.mkdir(parents=True, exist_ok=True)
     paths = (x for x in users_dir.iterdir() if x.suffix in ('.json', '.yaml', '.yml'))
     return sorted(set((k.stem for k in paths)), key=natural_keys)
 
 
 def get_available_instruction_templates():
-    path = "user_data/instruction-templates"
+    path = str(shared.user_data_dir / "instruction-templates")
     paths = []
     if os.path.exists(path):
         paths = (x for x in Path(path).iterdir() if x.suffix in ('.json', '.yaml', '.yml'))
@@ -244,13 +257,13 @@ def get_available_instruction_templates():
 def get_available_extensions():
     # User extensions (higher priority)
     user_extensions = []
-    user_ext_path = Path('user_data/extensions')
+    user_ext_path = shared.user_data_dir / 'extensions'
     if user_ext_path.exists():
-        user_exts = map(lambda x: x.parts[2], user_ext_path.glob('*/script.py'))
+        user_exts = map(lambda x: x.parent.name, user_ext_path.glob('*/script.py'))
         user_extensions = sorted(set(user_exts), key=natural_keys)
 
     # System extensions (excluding those overridden by user extensions)
-    system_exts = map(lambda x: x.parts[1], Path('extensions').glob('*/script.py'))
+    system_exts = map(lambda x: x.parent.name, Path('extensions').glob('*/script.py'))
     system_extensions = sorted(set(system_exts) - set(user_extensions), key=natural_keys)
 
     return user_extensions + system_extensions
@@ -306,4 +319,4 @@ def get_available_chat_styles():
 
 
 def get_available_grammars():
-    return ['None'] + sorted([item.name for item in list(Path('user_data/grammars').glob('*.gbnf'))], key=natural_keys)
+    return ['None'] + sorted([item.name for item in list((shared.user_data_dir / 'grammars').glob('*.gbnf'))], key=natural_keys)
diff --git a/server.py b/server.py
index 000ea9fb4a..a809cd7b9e 100644
--- a/server.py
+++ b/server.py
@@ -9,7 +9,7 @@
 from modules.prompts import load_prompt
 
 # Set up Gradio temp directory path
-gradio_temp_path = Path('user_data') / 'cache' / 'gradio'
+gradio_temp_path = shared.user_data_dir / 'cache' / 'gradio'
 shutil.rmtree(gradio_temp_path, ignore_errors=True)
 gradio_temp_path.mkdir(parents=True, exist_ok=True)
 
@@ -94,9 +94,9 @@ def create_interface():
     auth = [tuple(cred.split(':')) for cred in auth]
 
     # Allowed paths
-    allowed_paths = ["css", "js", "extensions", "user_data/cache"]
+    allowed_paths = ["css", "js", "extensions", str(shared.user_data_dir / "cache")]
     if not shared.args.multi_user:
-        allowed_paths.append("user_data/image_outputs")
+        allowed_paths.append(str(shared.user_data_dir / "image_outputs"))
 
     # Import the extensions and execute their setup() functions
     if shared.args.extensions is not None and len(shared.args.extensions) > 0:
@@ -120,7 +120,7 @@ def create_interface():
 
     # Clear existing cache files
     for cache_file in ['pfp_character.png', 'pfp_character_thumb.png']:
-        cache_path = Path(f"user_data/cache/{cache_file}")
+        cache_path = shared.user_data_dir / "cache" / cache_file
         if cache_path.exists():
             cache_path.unlink()
 
@@ -160,8 +160,8 @@ def create_interface():
         shared.gradio['interface_state'] = gr.State({k: None for k in shared.input_elements})
 
         # Audio notification
-        if Path("user_data/notification.mp3").exists():
-            shared.gradio['audio_notification'] = gr.Audio(interactive=False, value="user_data/notification.mp3", elem_id="audio_notification", visible=False)
+        if (shared.user_data_dir / "notification.mp3").exists():
+            shared.gradio['audio_notification'] = gr.Audio(interactive=False, value=str(shared.user_data_dir / "notification.mp3"), elem_id="audio_notification", visible=False)
 
         # Floating menus for saving/deleting files
         ui_file_saving.create_ui()
@@ -244,8 +244,8 @@ def create_interface():
     settings_file = None
     if shared.args.settings is not None and Path(shared.args.settings).exists():
         settings_file = Path(shared.args.settings)
-    elif Path('user_data/settings.yaml').exists():
-        settings_file = Path('user_data/settings.yaml')
+    elif (shared.user_data_dir / 'settings.yaml').exists():
+        settings_file = shared.user_data_dir / 'settings.yaml'
 
     if settings_file is not None:
         logger.info(f"Loading settings from \"{settings_file}\"")

From 8d43123f73546e8ed595c6421c6b1a52fb9dcdd3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 00:54:58 -0300
Subject: [PATCH 1238/1701] API: Fix function calling for Qwen, Mistral,
 GPT-OSS, and other models

The tool call response parser only handled JSON-based formats, causing
tool_calls to always be empty for models that use non-JSON formats.

Add parsers for three additional tool call formats:
- Qwen3.5: value
- Mistral/Devstral: functionName{"arg": "value"}
- GPT-OSS: <|channel|>commentary to=functions.name<|message|>{...}

Also fix multi-turn tool conversations crashing with Jinja2
UndefinedError on tool_call_id by preserving tool_calls and
tool_call_id metadata through the chat history conversion.
---
 extensions/openai/completions.py |  20 ++++--
 extensions/openai/utils.py       | 113 +++++++++++++++++++++++++++++++
 modules/chat.py                  |  17 ++++-
 3 files changed, 141 insertions(+), 9 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 56d2059def..5187343fdd 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -134,24 +134,32 @@ def convert_history(history):
             user_input_last = True
 
             if current_message:
-                chat_dialogue.append([current_message, '', ''])
+                chat_dialogue.append([current_message, '', '', {}])
                 current_message = ""
 
             current_message = content
         elif role == "assistant":
-            if "tool_calls" in entry and isinstance(entry["tool_calls"], list) and len(entry["tool_calls"]) > 0 and content.strip() == "":
-                continue  # skip tool calls
+            meta = {}
+            tool_calls = entry.get("tool_calls")
+            if tool_calls and isinstance(tool_calls, list) and len(tool_calls) > 0:
+                meta["tool_calls"] = tool_calls
+                if content.strip() == "":
+                    content = ""  # keep empty content, don't skip
+
             current_reply = content
             user_input_last = False
             if current_message:
-                chat_dialogue.append([current_message, current_reply, ''])
+                chat_dialogue.append([current_message, current_reply, '', meta])
                 current_message = ""
                 current_reply = ""
             else:
-                chat_dialogue.append(['', current_reply, ''])
+                chat_dialogue.append(['', current_reply, '', meta])
         elif role == "tool":
             user_input_last = False
-            chat_dialogue.append(['', '', content])
+            meta = {}
+            if "tool_call_id" in entry:
+                meta["tool_call_id"] = entry["tool_call_id"]
+            chat_dialogue.append(['', '', content, meta])
         elif role == "system":
             system_message += f"\n{content}" if system_message else content
 
diff --git a/extensions/openai/utils.py b/extensions/openai/utils.py
index 9a1de2e769..6937a1084b 100644
--- a/extensions/openai/utils.py
+++ b/extensions/openai/utils.py
@@ -83,6 +83,104 @@ def checkAndSanitizeToolCallCandidate(candidate_dict: dict, tool_names: list[str
     return None
 
 
+def _parseChannelToolCalls(answer: str, tool_names: list[str]):
+    """Parse channel-based tool calls used by GPT-OSS and similar models.
+
+    Format:
+        <|channel|>commentary to=functions.func_name <|constrain|>json<|message|>{"arg": "value"}
+    """
+    matches = []
+    for m in re.finditer(
+        r'<\|channel\|>commentary to=functions\.([^<\s]+)\s*(?:<\|constrain\|>json)?<\|message\|>(\{[^}]*(?:\{[^}]*\}[^}]*)*\})',
+        answer
+    ):
+        func_name = m.group(1).strip()
+        if func_name not in tool_names:
+            continue
+        try:
+            arguments = json.loads(m.group(2))
+            matches.append({
+                "type": "function",
+                "function": {
+                    "name": func_name,
+                    "arguments": arguments
+                }
+            })
+        except json.JSONDecodeError:
+            pass
+    return matches
+
+
+def _parseBareNameToolCalls(answer: str, tool_names: list[str]):
+    """Parse bare function-name style tool calls used by Mistral and similar models.
+
+    Format:
+        functionName{"arg": "value"}
+    Multiple calls are concatenated directly or separated by whitespace.
+    """
+    matches = []
+    # Build pattern that matches any known tool name followed by a JSON object
+    escaped_names = [re.escape(name) for name in tool_names]
+    pattern = r'(?:' + '|'.join(escaped_names) + r')\s*\{[^{}]*(?:\{[^{}]*\}[^{}]*)*\}'
+    for match in re.finditer(pattern, answer):
+        text = match.group(0)
+        # Split into function name and JSON arguments
+        for name in tool_names:
+            if text.startswith(name):
+                json_str = text[len(name):].strip()
+                try:
+                    arguments = json.loads(json_str)
+                    matches.append({
+                        "type": "function",
+                        "function": {
+                            "name": name,
+                            "arguments": arguments
+                        }
+                    })
+                except json.JSONDecodeError:
+                    pass
+                break
+    return matches
+
+
+def _parseXmlParamToolCalls(answer: str, tool_names: list[str]):
+    """Parse XML-parameter style tool calls used by Qwen3.5 and similar models.
+
+    Format:
+        
+        
+        value
+        
+        
+    """
+    matches = []
+    for tc_match in re.finditer(r'\s*(.*?)\s*', answer, re.DOTALL):
+        tc_content = tc_match.group(1)
+        func_match = re.search(r']+)>', tc_content)
+        if not func_match:
+            continue
+        func_name = func_match.group(1).strip()
+        if func_name not in tool_names:
+            continue
+        arguments = {}
+        for param_match in re.finditer(r']+)>\s*(.*?)\s*', tc_content, re.DOTALL):
+            param_name = param_match.group(1).strip()
+            param_value = param_match.group(2).strip()
+            try:
+                param_value = json.loads(param_value)
+            except (json.JSONDecodeError, ValueError):
+                pass  # keep as string
+            arguments[param_name] = param_value
+        matches.append({
+            "type": "function",
+            "function": {
+                "name": func_name,
+                "arguments": arguments
+            }
+        })
+    return matches
+
+
 def parseToolCall(answer: str, tool_names: list[str]):
     matches = []
 
@@ -90,6 +188,21 @@ def parseToolCall(answer: str, tool_names: list[str]):
     if len(answer) < 10:
         return matches
 
+    # Check for channel-based tool calls (e.g. GPT-OSS format)
+    matches = _parseChannelToolCalls(answer, tool_names)
+    if matches:
+        return matches
+
+    # Check for XML-parameter style tool calls (e.g. Qwen3.5 format)
+    matches = _parseXmlParamToolCalls(answer, tool_names)
+    if matches:
+        return matches
+
+    # Check for bare function-name style tool calls (e.g. Mistral format)
+    matches = _parseBareNameToolCalls(answer, tool_names)
+    if matches:
+        return matches
+
     # Define the regex pattern to find the JSON content wrapped in , , , and other tags observed from various models
     patterns = [r"(```[^\n]*)\n(.*?)```", r"<([^>]+)>(.*?)"]
 
diff --git a/modules/chat.py b/modules/chat.py
index 7c58542fd8..2beb154367 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -159,13 +159,20 @@ def generate_chat_prompt(user_input, state, **kwargs):
         user_msg = entry[0].strip()
         assistant_msg = entry[1].strip()
         tool_msg = entry[2].strip() if len(entry) > 2 else ''
+        entry_meta = entry[3] if len(entry) > 3 else {}
 
         row_idx = len(history) - i - 1
 
         if tool_msg:
-            messages.insert(insert_pos, {"role": "tool", "content": tool_msg})
-
-        if assistant_msg:
+            tool_message = {"role": "tool", "content": tool_msg}
+            if "tool_call_id" in entry_meta:
+                tool_message["tool_call_id"] = entry_meta["tool_call_id"]
+            messages.insert(insert_pos, tool_message)
+
+        if not assistant_msg and entry_meta.get('tool_calls'):
+            # Assistant message with only tool_calls and no text content
+            messages.insert(insert_pos, {"role": "assistant", "content": "", "tool_calls": entry_meta['tool_calls']})
+        elif assistant_msg:
             # Handle GPT-OSS as a special case
             if '<|channel|>analysis<|message|>' in assistant_msg or '<|channel|>final<|message|>' in assistant_msg:
                 thinking_content = ""
@@ -240,6 +247,10 @@ def generate_chat_prompt(user_input, state, **kwargs):
                 # Default case (used by all other models)
                 messages.insert(insert_pos, {"role": "assistant", "content": assistant_msg})
 
+            # Attach tool_calls metadata to the assistant message if present
+            if entry_meta.get('tool_calls') and messages[insert_pos].get('role') == 'assistant':
+                messages[insert_pos]['tool_calls'] = entry_meta['tool_calls']
+
         if user_msg not in ['', '<|BEGIN-VISIBLE-CHAT|>']:
             # Check for user message attachments in metadata
             user_key = f"user_{row_idx}"

From ddcad3cc51a17250ddf10030f42b98b1ae027b91 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 00:57:32 -0300
Subject: [PATCH 1239/1701] Follow-up to e2548f69: add missing paths module,
 fix gallery extension

---
 extensions/gallery/script.py |  5 +++--
 modules/paths.py             | 28 ++++++++++++++++++++++++++++
 2 files changed, 31 insertions(+), 2 deletions(-)
 create mode 100644 modules/paths.py

diff --git a/extensions/gallery/script.py b/extensions/gallery/script.py
index 8b242fb603..3633ae464b 100644
--- a/extensions/gallery/script.py
+++ b/extensions/gallery/script.py
@@ -2,6 +2,7 @@
 
 import gradio as gr
 
+import modules.shared as shared
 from modules.html_generator import get_image_cache
 from modules.shared import gradio
 
@@ -72,13 +73,13 @@ def generate_html():
     global cards
     cards = []
     # Iterate through files in image folder
-    for file in sorted(Path("user_data/characters").glob("*")):
+    for file in sorted((shared.user_data_dir / "characters").glob("*")):
         if file.suffix in [".json", ".yml", ".yaml"]:
             character = file.stem
             container_html = '
            '
             image_html = "
            "
 
-            for path in [Path(f"user_data/characters/{character}.{extension}") for extension in ['png', 'jpg', 'jpeg']]:
+            for path in [shared.user_data_dir / "characters" / f"{character}.{extension}" for extension in ['png', 'jpg', 'jpeg']]:
                 if path.exists():
                     image_html = f''
                     break
diff --git a/modules/paths.py b/modules/paths.py
new file mode 100644
index 0000000000..36fdc4f2f3
--- /dev/null
+++ b/modules/paths.py
@@ -0,0 +1,28 @@
+import sys
+from pathlib import Path
+
+
+def resolve_user_data_dir():
+    """
+    Resolve the user_data directory path. Order of precedence:
+    1. --user-data-dir CLI flag (pre-parsed from sys.argv before argparse)
+    2. Auto-detect: if ./user_data doesn't exist but ../user_data does, use ../user_data
+    3. Default: 'user_data'
+    """
+    script_dir = Path(__file__).resolve().parent.parent
+
+    # Check sys.argv for --user-data-dir before argparse runs
+    for i, arg in enumerate(sys.argv):
+        if arg == '--user-data-dir' and i + 1 < len(sys.argv):
+            return Path(sys.argv[i + 1])
+        elif arg.startswith('--user-data-dir='):
+            return Path(arg.split('=', 1)[1])
+
+    # Auto-detect: check if user_data exists locally vs one folder up
+    local_path = script_dir / 'user_data'
+    parent_path = script_dir.parent / 'user_data'
+
+    if not local_path.exists() and parent_path.exists():
+        return parent_path
+
+    return Path('user_data')

From 2e7e966ef2b0903bf43ad93e32ee7a4dc6b9d143 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 20:06:34 -0800
Subject: [PATCH 1240/1701] Docs: Better Tool/Function calling examples

---
 docs/12 - OpenAI API.md | 146 +++++++++++++++++++++-------------------
 1 file changed, 78 insertions(+), 68 deletions(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index fc444b153a..ebe2479711 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -388,83 +388,93 @@ headers = {
 
 in any of the examples above.
 
-#### Tool/Function Calling Example
+#### Tool/Function calling
 
-You need to use a model with tools support. The prompt will be automatically formatted using the model's Jinja2 template.
+Use a model with tool calling support (Qwen, Mistral, GPT-OSS, etc). Tools are passed via the `tools` parameter and the prompt is automatically formatted using the model's Jinja2 template.
 
-Request:
+When the model decides to call a tool, the response will have `finish_reason: "tool_calls"` and a `tool_calls` array with structured function names and arguments. You then execute the tool, send the result back as a `role: "tool"` message, and continue until the model responds with `finish_reason: "stop"`.
 
-```
-curl http://127.0.0.1:5000/v1/chat/completions \
-  -H "Content-Type: application/json" \
-  -d '{
-    "messages": [
-      {
-        "role": "system",
-        "content": "You are a helpful assistant."
-      },
-      {
-        "role": "user",
-        "content": "What time is it currently in New York City?"
-      }
-    ],
-    "tools": [
-      {
+Some models call multiple tools in parallel (Qwen, Mistral), while others call one at a time (GPT-OSS). The loop below handles both styles.
+
+```python
+import json
+import requests
+
+url = "http://127.0.0.1:5000/v1/chat/completions"
+
+# Define your tools
+tools = [
+    {
         "type": "function",
         "function": {
-          "name": "get_current_time",
-          "description": "Get current time in a specific timezones",
-          "parameters": {
-            "type": "object",
-            "required": ["timezone"],
-            "properties": {
-              "timezone": {
-                "type": "string",
-                "description": "IANA timezone name (e.g., America/New_York, Europe/London). Use Europe/Berlin as local timezone if no timezone provided by the user."
-              }
+            "name": "get_weather",
+            "description": "Get the current weather for a given location",
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "location": {"type": "string", "description": "City name"},
+                },
+                "required": ["location"]
             }
-          }
         }
-      }
-    ]
-  }'
-```
+    },
+    {
+        "type": "function",
+        "function": {
+            "name": "get_time",
+            "description": "Get the current time in a given timezone",
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "timezone": {"type": "string", "description": "IANA timezone string"},
+                },
+                "required": ["timezone"]
+            }
+        }
+    },
+]
 
-Sample response:
 
-```
-{
-    "id": "chatcmpl-1746532051477984256",
-    "object": "chat.completion",
-    "created": 1746532051,
-    "model": "qwen2.5-coder-14b-instruct-q4_k_m.gguf",
-    "choices": [
-        {
-            "index": 0,
-            "finish_reason": "tool_calls",
-            "message": {
-                "role": "assistant",
-                "content": "```xml\n\n{\n  \"name\": \"get_current_time\",\n  \"arguments\": {\n    \"timezone\": \"America/New_York\"\n  }\n}\n\n```"
-            },
-            "tool_calls": [
-                {
-                    "type": "function",
-                    "function": {
-                        "name": "get_current_time",
-                        "arguments": "{\"timezone\": \"America/New_York\"}"
-                    },
-                    "id": "call_52ij07mh",
-                    "index": "0"
-                }
-            ]
-        }
-    ],
-    "usage": {
-        "prompt_tokens": 224,
-        "completion_tokens": 38,
-        "total_tokens": 262
-    }
-}
+def execute_tool(name, arguments):
+    """Replace this with your actual tool implementations."""
+    if name == "get_weather":
+        return {"temperature": 22, "condition": "sunny", "humidity": 45}
+    elif name == "get_time":
+        return {"time": "2:30 PM", "timezone": "JST"}
+    return {"error": f"Unknown tool: {name}"}
+
+
+messages = [{"role": "user", "content": "What time is it in Tokyo and what's the weather like there?"}]
+
+# Tool-calling loop: keep going until the model gives a final answer
+for _ in range(10):
+    response = requests.post(url, json={"messages": messages, "tools": tools}).json()
+    choice = response["choices"][0]
+
+    if choice["finish_reason"] == "tool_calls":
+        # Add the assistant's response (with tool_calls) to history
+        messages.append({
+            "role": "assistant",
+            "content": choice["message"]["content"],
+            "tool_calls": choice["tool_calls"],
+        })
+
+        # Execute each tool and add results to history
+        for tool_call in choice["tool_calls"]:
+            name = tool_call["function"]["name"]
+            arguments = json.loads(tool_call["function"]["arguments"])
+            result = execute_tool(name, arguments)
+
+            print(f"Tool call: {name}({arguments}) => {result}")
+            messages.append({
+                "role": "tool",
+                "tool_call_id": tool_call["id"],
+                "content": json.dumps(result),
+            })
+    else:
+        # Final answer
+        print(f"\nAssistant: {choice['message']['content']}")
+        break
 ```
 
 ### Environment variables

From 8a9afcbec65a9fa556cbd0e3623440dd2d200f51 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 01:19:36 -0300
Subject: [PATCH 1241/1701] Allow extensions to skip output post-processing

---
 modules/chat.py | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 2beb154367..5d7e0e5657 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -923,7 +923,9 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
         visible_reply = html.escape(visible_reply)
 
         if shared.stop_everything:
-            output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
+            if not state.get('_skip_output_extensions'):
+                output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
+
             yield output
             return
 
@@ -956,9 +958,11 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             full_visible = full_internal
 
         full_visible = html.escape(full_visible)
-        output['visible'][-1][1] = apply_extensions('output', full_visible, state, is_chat=True)
+        if not state.get('_skip_output_extensions'):
+            output['visible'][-1][1] = apply_extensions('output', full_visible, state, is_chat=True)
     else:
-        output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
+        if not state.get('_skip_output_extensions'):
+            output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
 
     # Final sync for version metadata (in case streaming was disabled)
     if regenerate:

From 27bcc45c18ef3304d85b844a8741f993e3b6483b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 01:36:30 -0300
Subject: [PATCH 1242/1701] API: Add command-line flags to override default
 generation parameters

---
 extensions/openai/typing.py | 98 +++++++++++++++++++------------------
 modules/presets.py          | 83 ++++++++++++++++---------------
 modules/shared.py           | 46 ++++++++++++++++-
 3 files changed, 138 insertions(+), 89 deletions(-)

diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 1548093ffa..8af75ef834 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -4,56 +4,58 @@
 
 from pydantic import BaseModel, Field, model_validator, validator
 
+from modules import shared
+
 
 class GenerationOptions(BaseModel):
     preset: str | None = Field(default=None, description="The name of a file under text-generation-webui/user_data/presets (without the .yaml extension). The sampling parameters that get overwritten by this option are the keys in the default_preset() function in modules/presets.py.")
-    dynatemp_low: float = 1
-    dynatemp_high: float = 1
-    dynatemp_exponent: float = 1
-    smoothing_factor: float = 0
-    smoothing_curve: float = 1
-    min_p: float = 0
-    top_k: int = 0
-    typical_p: float = 1
-    xtc_threshold: float = 0.1
-    xtc_probability: float = 0
-    epsilon_cutoff: float = 0
-    eta_cutoff: float = 0
-    tfs: float = 1
-    top_a: float = 0
-    top_n_sigma: float = 0
-    adaptive_target: float = 0
-    adaptive_decay: float = 0.9
-    dry_multiplier: float = 0
-    dry_allowed_length: int = 2
-    dry_base: float = 1.75
-    repetition_penalty: float = 1
-    encoder_repetition_penalty: float = 1
-    no_repeat_ngram_size: int = 0
-    repetition_penalty_range: int = 1024
-    penalty_alpha: float = 0
-    guidance_scale: float = 1
-    mirostat_mode: int = 0
-    mirostat_tau: float = 5
-    mirostat_eta: float = 0.1
+    dynatemp_low: float = shared.args.dynatemp_low
+    dynatemp_high: float = shared.args.dynatemp_high
+    dynatemp_exponent: float = shared.args.dynatemp_exponent
+    smoothing_factor: float = shared.args.smoothing_factor
+    smoothing_curve: float = shared.args.smoothing_curve
+    min_p: float = shared.args.min_p
+    top_k: int = shared.args.top_k
+    typical_p: float = shared.args.typical_p
+    xtc_threshold: float = shared.args.xtc_threshold
+    xtc_probability: float = shared.args.xtc_probability
+    epsilon_cutoff: float = shared.args.epsilon_cutoff
+    eta_cutoff: float = shared.args.eta_cutoff
+    tfs: float = shared.args.tfs
+    top_a: float = shared.args.top_a
+    top_n_sigma: float = shared.args.top_n_sigma
+    adaptive_target: float = shared.args.adaptive_target
+    adaptive_decay: float = shared.args.adaptive_decay
+    dry_multiplier: float = shared.args.dry_multiplier
+    dry_allowed_length: int = shared.args.dry_allowed_length
+    dry_base: float = shared.args.dry_base
+    repetition_penalty: float = shared.args.repetition_penalty
+    encoder_repetition_penalty: float = shared.args.encoder_repetition_penalty
+    no_repeat_ngram_size: int = shared.args.no_repeat_ngram_size
+    repetition_penalty_range: int = shared.args.repetition_penalty_range
+    penalty_alpha: float = shared.args.penalty_alpha
+    guidance_scale: float = shared.args.guidance_scale
+    mirostat_mode: int = shared.args.mirostat_mode
+    mirostat_tau: float = shared.args.mirostat_tau
+    mirostat_eta: float = shared.args.mirostat_eta
     prompt_lookup_num_tokens: int = 0
     max_tokens_second: int = 0
-    do_sample: bool = True
-    dynamic_temperature: bool = False
-    temperature_last: bool = False
+    do_sample: bool = shared.args.do_sample
+    dynamic_temperature: bool = shared.args.dynamic_temperature
+    temperature_last: bool = shared.args.temperature_last
     auto_max_new_tokens: bool = False
     ban_eos_token: bool = False
     add_bos_token: bool = True
-    enable_thinking: bool = True
-    reasoning_effort: str = "medium"
+    enable_thinking: bool = shared.args.enable_thinking
+    reasoning_effort: str = shared.args.reasoning_effort
     skip_special_tokens: bool = True
     static_cache: bool = False
     truncation_length: int = 0
     seed: int = -1
-    sampler_priority: List[str] | str | None = Field(default=['repetition_penalty', 'presence_penalty', 'frequency_penalty', 'dry', 'top_n_sigma', 'temperature', 'dynamic_temperature', 'quadratic_sampling', 'top_k', 'top_p', 'typical_p', 'epsilon_cutoff', 'eta_cutoff', 'tfs', 'top_a', 'min_p', 'adaptive_p', 'mirostat', 'xtc', 'encoder_repetition_penalty', 'no_repeat_ngram'], description="List of samplers where the first items will appear first in the stack. Example: [\"top_k\", \"temperature\", \"top_p\"].")
+    sampler_priority: List[str] | str | None = Field(default=shared.args.sampler_priority, description="List of samplers where the first items will appear first in the stack. Example: [\"top_k\", \"temperature\", \"top_p\"].")
     custom_token_bans: str = ""
     negative_prompt: str = ''
-    dry_sequence_breakers: str = '"\\n", ":", "\\"", "*"'
+    dry_sequence_breakers: str = shared.args.dry_sequence_breakers
     grammar_string: str = ""
 
 
@@ -105,17 +107,17 @@ class CompletionRequestParams(BaseModel):
     messages: List[dict] | None = Field(default=None, description="OpenAI messages format for multimodal support. Alternative to 'prompt'.")
     best_of: int | None = Field(default=1, description="Unused parameter.")
     echo: bool | None = False
-    frequency_penalty: float | None = 0
+    frequency_penalty: float | None = shared.args.frequency_penalty
     logit_bias: dict | None = None
     logprobs: int | None = None
     max_tokens: int | None = 512
     n: int | None = Field(default=1, description="Unused parameter.")
-    presence_penalty: float | None = 0
+    presence_penalty: float | None = shared.args.presence_penalty
     stop: str | List[str] | None = None
     stream: bool | None = False
     suffix: str | None = None
-    temperature: float | None = 1
-    top_p: float | None = 1
+    temperature: float | None = shared.args.temperature
+    top_p: float | None = shared.args.top_p
     user: str | None = Field(default=None, description="Unused parameter.")
 
     @model_validator(mode='after')
@@ -141,18 +143,18 @@ class CompletionResponse(BaseModel):
 class ChatCompletionRequestParams(BaseModel):
     messages: List[dict]
     model: str | None = Field(default=None, description="Unused parameter. To change the model, use the /v1/internal/model/load endpoint.")
-    frequency_penalty: float | None = 0
+    frequency_penalty: float | None = shared.args.frequency_penalty
     function_call: str | dict | None = Field(default=None, description="Unused parameter.")
     functions: List[dict] | None = Field(default=None, description="Unused parameter.")
     tools: List[dict] | None = Field(default=None, description="Tools signatures passed via MCP.")
     logit_bias: dict | None = None
     max_tokens: int | None = None
     n: int | None = Field(default=1, description="Unused parameter.")
-    presence_penalty: float | None = 0
+    presence_penalty: float | None = shared.args.presence_penalty
     stop: str | List[str] | None = None
     stream: bool | None = False
-    temperature: float | None = 1
-    top_p: float | None = 1
+    temperature: float | None = shared.args.temperature
+    top_p: float | None = shared.args.top_p
     user: str | None = Field(default=None, description="Unused parameter.")
 
     mode: str = Field(default='instruct', description="Valid options: instruct, chat, chat-instruct.")
@@ -228,11 +230,11 @@ class LogitsRequestParams(BaseModel):
     prompt: str
     use_samplers: bool = False
     top_logits: int | None = 50
-    frequency_penalty: float | None = 0
+    frequency_penalty: float | None = shared.args.frequency_penalty
     max_tokens: int | None = 512
-    presence_penalty: float | None = 0
-    temperature: float | None = 1
-    top_p: float | None = 1
+    presence_penalty: float | None = shared.args.presence_penalty
+    temperature: float | None = shared.args.temperature
+    top_p: float | None = shared.args.top_p
 
 
 class LogitsRequest(GenerationOptions, LogitsRequestParams):
diff --git a/modules/presets.py b/modules/presets.py
index 9aab7e3c8c..b53195ee48 100644
--- a/modules/presets.py
+++ b/modules/presets.py
@@ -9,47 +9,50 @@
 from modules.logging_colors import logger
 
 
+default_preset_values = {
+    'temperature': 1,
+    'dynatemp_low': 1,
+    'dynatemp_high': 1,
+    'dynatemp_exponent': 1,
+    'smoothing_factor': 0,
+    'smoothing_curve': 1,
+    'min_p': 0,
+    'top_p': 1,
+    'top_k': 0,
+    'typical_p': 1,
+    'xtc_threshold': 0.1,
+    'xtc_probability': 0,
+    'epsilon_cutoff': 0,
+    'eta_cutoff': 0,
+    'tfs': 1,
+    'top_a': 0,
+    'top_n_sigma': 0,
+    'adaptive_target': 0,
+    'adaptive_decay': 0.9,
+    'dry_multiplier': 0,
+    'dry_allowed_length': 2,
+    'dry_base': 1.75,
+    'repetition_penalty': 1,
+    'frequency_penalty': 0,
+    'presence_penalty': 0,
+    'encoder_repetition_penalty': 1,
+    'no_repeat_ngram_size': 0,
+    'repetition_penalty_range': 1024,
+    'penalty_alpha': 0,
+    'guidance_scale': 1,
+    'mirostat_mode': 0,
+    'mirostat_tau': 5,
+    'mirostat_eta': 0.1,
+    'do_sample': True,
+    'dynamic_temperature': False,
+    'temperature_last': False,
+    'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ndry\ntop_n_sigma\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nadaptive_p\nmirostat\nxtc\nencoder_repetition_penalty\nno_repeat_ngram',
+    'dry_sequence_breakers': '"\\n", ":", "\\"", "*"',
+}
+
+
 def default_preset():
-    result = {
-        'temperature': 1,
-        'dynatemp_low': 1,
-        'dynatemp_high': 1,
-        'dynatemp_exponent': 1,
-        'smoothing_factor': 0,
-        'smoothing_curve': 1,
-        'min_p': 0,
-        'top_p': 1,
-        'top_k': 0,
-        'typical_p': 1,
-        'xtc_threshold': 0.1,
-        'xtc_probability': 0,
-        'epsilon_cutoff': 0,
-        'eta_cutoff': 0,
-        'tfs': 1,
-        'top_a': 0,
-        'top_n_sigma': 0,
-        'adaptive_target': 0,
-        'adaptive_decay': 0.9,
-        'dry_multiplier': 0,
-        'dry_allowed_length': 2,
-        'dry_base': 1.75,
-        'repetition_penalty': 1,
-        'frequency_penalty': 0,
-        'presence_penalty': 0,
-        'encoder_repetition_penalty': 1,
-        'no_repeat_ngram_size': 0,
-        'repetition_penalty_range': 1024,
-        'penalty_alpha': 0,
-        'guidance_scale': 1,
-        'mirostat_mode': 0,
-        'mirostat_tau': 5,
-        'mirostat_eta': 0.1,
-        'do_sample': True,
-        'dynamic_temperature': False,
-        'temperature_last': False,
-        'sampler_priority': 'repetition_penalty\npresence_penalty\nfrequency_penalty\ndry\ntop_n_sigma\ntemperature\ndynamic_temperature\nquadratic_sampling\ntop_k\ntop_p\ntypical_p\nepsilon_cutoff\neta_cutoff\ntfs\ntop_a\nmin_p\nadaptive_p\nmirostat\nxtc\nencoder_repetition_penalty\nno_repeat_ngram',
-        'dry_sequence_breakers': '"\\n", ":", "\\"", "*"',
-    }
+    result = dict(default_preset_values)
 
     if shared.args.portable:
         samplers = result['sampler_priority'].split('\n')
diff --git a/modules/shared.py b/modules/shared.py
index 57b7e3e991..5dbd592a4d 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -10,7 +10,7 @@
 
 from modules.logging_colors import logger
 from modules.paths import resolve_user_data_dir
-from modules.presets import default_preset
+from modules.presets import default_preset, default_preset_values
 
 # Resolve user_data directory early (before argparse defaults are set)
 user_data_dir = resolve_user_data_dir()
@@ -171,6 +171,50 @@
 group.add_argument('--api-disable-ipv4', action='store_true', help='Disable IPv4 for the API')
 group.add_argument('--nowebui', action='store_true', help='Do not launch the Gradio UI. Useful for launching the API in standalone mode.')
 
+# API generation defaults
+_d = default_preset_values
+group = parser.add_argument_group('API generation defaults')
+group.add_argument('--temperature', type=float, default=_d['temperature'])
+group.add_argument('--dynatemp-low', type=float, default=_d['dynatemp_low'])
+group.add_argument('--dynatemp-high', type=float, default=_d['dynatemp_high'])
+group.add_argument('--dynatemp-exponent', type=float, default=_d['dynatemp_exponent'])
+group.add_argument('--smoothing-factor', type=float, default=_d['smoothing_factor'])
+group.add_argument('--smoothing-curve', type=float, default=_d['smoothing_curve'])
+group.add_argument('--min-p', type=float, default=_d['min_p'])
+group.add_argument('--top-p', type=float, default=_d['top_p'])
+group.add_argument('--top-k', type=int, default=_d['top_k'])
+group.add_argument('--typical-p', type=float, default=_d['typical_p'])
+group.add_argument('--xtc-threshold', type=float, default=_d['xtc_threshold'])
+group.add_argument('--xtc-probability', type=float, default=_d['xtc_probability'])
+group.add_argument('--epsilon-cutoff', type=float, default=_d['epsilon_cutoff'])
+group.add_argument('--eta-cutoff', type=float, default=_d['eta_cutoff'])
+group.add_argument('--tfs', type=float, default=_d['tfs'])
+group.add_argument('--top-a', type=float, default=_d['top_a'])
+group.add_argument('--top-n-sigma', type=float, default=_d['top_n_sigma'])
+group.add_argument('--adaptive-target', type=float, default=_d['adaptive_target'])
+group.add_argument('--adaptive-decay', type=float, default=_d['adaptive_decay'])
+group.add_argument('--dry-multiplier', type=float, default=_d['dry_multiplier'])
+group.add_argument('--dry-allowed-length', type=int, default=_d['dry_allowed_length'])
+group.add_argument('--dry-base', type=float, default=_d['dry_base'])
+group.add_argument('--repetition-penalty', type=float, default=_d['repetition_penalty'])
+group.add_argument('--frequency-penalty', type=float, default=_d['frequency_penalty'])
+group.add_argument('--presence-penalty', type=float, default=_d['presence_penalty'])
+group.add_argument('--encoder-repetition-penalty', type=float, default=_d['encoder_repetition_penalty'])
+group.add_argument('--no-repeat-ngram-size', type=int, default=_d['no_repeat_ngram_size'])
+group.add_argument('--repetition-penalty-range', type=int, default=_d['repetition_penalty_range'])
+group.add_argument('--penalty-alpha', type=float, default=_d['penalty_alpha'])
+group.add_argument('--guidance-scale', type=float, default=_d['guidance_scale'])
+group.add_argument('--mirostat-mode', type=int, default=_d['mirostat_mode'])
+group.add_argument('--mirostat-tau', type=float, default=_d['mirostat_tau'])
+group.add_argument('--mirostat-eta', type=float, default=_d['mirostat_eta'])
+group.add_argument('--do-sample', action=argparse.BooleanOptionalAction, default=_d['do_sample'])
+group.add_argument('--dynamic-temperature', action=argparse.BooleanOptionalAction, default=_d['dynamic_temperature'])
+group.add_argument('--temperature-last', action=argparse.BooleanOptionalAction, default=_d['temperature_last'])
+group.add_argument('--sampler-priority', type=str, default=_d['sampler_priority'])
+group.add_argument('--dry-sequence-breakers', type=str, default=_d['dry_sequence_breakers'])
+group.add_argument('--enable-thinking', action=argparse.BooleanOptionalAction, default=True)
+group.add_argument('--reasoning-effort', type=str, default='medium')
+
 # Handle CMD_FLAGS.txt
 cmd_flags_path = user_data_dir / "CMD_FLAGS.txt"
 if cmd_flags_path.exists():

From e81a47f7085ffdfd67b332d6badd7f357e3ded4d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 20:41:45 -0800
Subject: [PATCH 1243/1701] Improve the API generation defaults --help message

---
 modules/shared.py | 80 +++++++++++++++++++++++------------------------
 1 file changed, 40 insertions(+), 40 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 5dbd592a4d..e9535aa001 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -174,46 +174,46 @@
 # API generation defaults
 _d = default_preset_values
 group = parser.add_argument_group('API generation defaults')
-group.add_argument('--temperature', type=float, default=_d['temperature'])
-group.add_argument('--dynatemp-low', type=float, default=_d['dynatemp_low'])
-group.add_argument('--dynatemp-high', type=float, default=_d['dynatemp_high'])
-group.add_argument('--dynatemp-exponent', type=float, default=_d['dynatemp_exponent'])
-group.add_argument('--smoothing-factor', type=float, default=_d['smoothing_factor'])
-group.add_argument('--smoothing-curve', type=float, default=_d['smoothing_curve'])
-group.add_argument('--min-p', type=float, default=_d['min_p'])
-group.add_argument('--top-p', type=float, default=_d['top_p'])
-group.add_argument('--top-k', type=int, default=_d['top_k'])
-group.add_argument('--typical-p', type=float, default=_d['typical_p'])
-group.add_argument('--xtc-threshold', type=float, default=_d['xtc_threshold'])
-group.add_argument('--xtc-probability', type=float, default=_d['xtc_probability'])
-group.add_argument('--epsilon-cutoff', type=float, default=_d['epsilon_cutoff'])
-group.add_argument('--eta-cutoff', type=float, default=_d['eta_cutoff'])
-group.add_argument('--tfs', type=float, default=_d['tfs'])
-group.add_argument('--top-a', type=float, default=_d['top_a'])
-group.add_argument('--top-n-sigma', type=float, default=_d['top_n_sigma'])
-group.add_argument('--adaptive-target', type=float, default=_d['adaptive_target'])
-group.add_argument('--adaptive-decay', type=float, default=_d['adaptive_decay'])
-group.add_argument('--dry-multiplier', type=float, default=_d['dry_multiplier'])
-group.add_argument('--dry-allowed-length', type=int, default=_d['dry_allowed_length'])
-group.add_argument('--dry-base', type=float, default=_d['dry_base'])
-group.add_argument('--repetition-penalty', type=float, default=_d['repetition_penalty'])
-group.add_argument('--frequency-penalty', type=float, default=_d['frequency_penalty'])
-group.add_argument('--presence-penalty', type=float, default=_d['presence_penalty'])
-group.add_argument('--encoder-repetition-penalty', type=float, default=_d['encoder_repetition_penalty'])
-group.add_argument('--no-repeat-ngram-size', type=int, default=_d['no_repeat_ngram_size'])
-group.add_argument('--repetition-penalty-range', type=int, default=_d['repetition_penalty_range'])
-group.add_argument('--penalty-alpha', type=float, default=_d['penalty_alpha'])
-group.add_argument('--guidance-scale', type=float, default=_d['guidance_scale'])
-group.add_argument('--mirostat-mode', type=int, default=_d['mirostat_mode'])
-group.add_argument('--mirostat-tau', type=float, default=_d['mirostat_tau'])
-group.add_argument('--mirostat-eta', type=float, default=_d['mirostat_eta'])
-group.add_argument('--do-sample', action=argparse.BooleanOptionalAction, default=_d['do_sample'])
-group.add_argument('--dynamic-temperature', action=argparse.BooleanOptionalAction, default=_d['dynamic_temperature'])
-group.add_argument('--temperature-last', action=argparse.BooleanOptionalAction, default=_d['temperature_last'])
-group.add_argument('--sampler-priority', type=str, default=_d['sampler_priority'])
-group.add_argument('--dry-sequence-breakers', type=str, default=_d['dry_sequence_breakers'])
-group.add_argument('--enable-thinking', action=argparse.BooleanOptionalAction, default=True)
-group.add_argument('--reasoning-effort', type=str, default='medium')
+group.add_argument('--temperature', type=float, default=_d['temperature'], metavar='N', help='Temperature')
+group.add_argument('--dynatemp-low', type=float, default=_d['dynatemp_low'], metavar='N', help='Dynamic temperature low')
+group.add_argument('--dynatemp-high', type=float, default=_d['dynatemp_high'], metavar='N', help='Dynamic temperature high')
+group.add_argument('--dynatemp-exponent', type=float, default=_d['dynatemp_exponent'], metavar='N', help='Dynamic temperature exponent')
+group.add_argument('--smoothing-factor', type=float, default=_d['smoothing_factor'], metavar='N', help='Smoothing factor')
+group.add_argument('--smoothing-curve', type=float, default=_d['smoothing_curve'], metavar='N', help='Smoothing curve')
+group.add_argument('--min-p', type=float, default=_d['min_p'], metavar='N', help='Min P')
+group.add_argument('--top-p', type=float, default=_d['top_p'], metavar='N', help='Top P')
+group.add_argument('--top-k', type=int, default=_d['top_k'], metavar='N', help='Top K')
+group.add_argument('--typical-p', type=float, default=_d['typical_p'], metavar='N', help='Typical P')
+group.add_argument('--xtc-threshold', type=float, default=_d['xtc_threshold'], metavar='N', help='XTC threshold')
+group.add_argument('--xtc-probability', type=float, default=_d['xtc_probability'], metavar='N', help='XTC probability')
+group.add_argument('--epsilon-cutoff', type=float, default=_d['epsilon_cutoff'], metavar='N', help='Epsilon cutoff')
+group.add_argument('--eta-cutoff', type=float, default=_d['eta_cutoff'], metavar='N', help='Eta cutoff')
+group.add_argument('--tfs', type=float, default=_d['tfs'], metavar='N', help='TFS')
+group.add_argument('--top-a', type=float, default=_d['top_a'], metavar='N', help='Top A')
+group.add_argument('--top-n-sigma', type=float, default=_d['top_n_sigma'], metavar='N', help='Top N Sigma')
+group.add_argument('--adaptive-target', type=float, default=_d['adaptive_target'], metavar='N', help='Adaptive target')
+group.add_argument('--adaptive-decay', type=float, default=_d['adaptive_decay'], metavar='N', help='Adaptive decay')
+group.add_argument('--dry-multiplier', type=float, default=_d['dry_multiplier'], metavar='N', help='DRY multiplier')
+group.add_argument('--dry-allowed-length', type=int, default=_d['dry_allowed_length'], metavar='N', help='DRY allowed length')
+group.add_argument('--dry-base', type=float, default=_d['dry_base'], metavar='N', help='DRY base')
+group.add_argument('--repetition-penalty', type=float, default=_d['repetition_penalty'], metavar='N', help='Repetition penalty')
+group.add_argument('--frequency-penalty', type=float, default=_d['frequency_penalty'], metavar='N', help='Frequency penalty')
+group.add_argument('--presence-penalty', type=float, default=_d['presence_penalty'], metavar='N', help='Presence penalty')
+group.add_argument('--encoder-repetition-penalty', type=float, default=_d['encoder_repetition_penalty'], metavar='N', help='Encoder repetition penalty')
+group.add_argument('--no-repeat-ngram-size', type=int, default=_d['no_repeat_ngram_size'], metavar='N', help='No repeat ngram size')
+group.add_argument('--repetition-penalty-range', type=int, default=_d['repetition_penalty_range'], metavar='N', help='Repetition penalty range')
+group.add_argument('--penalty-alpha', type=float, default=_d['penalty_alpha'], metavar='N', help='Penalty alpha')
+group.add_argument('--guidance-scale', type=float, default=_d['guidance_scale'], metavar='N', help='Guidance scale')
+group.add_argument('--mirostat-mode', type=int, default=_d['mirostat_mode'], metavar='N', help='Mirostat mode')
+group.add_argument('--mirostat-tau', type=float, default=_d['mirostat_tau'], metavar='N', help='Mirostat tau')
+group.add_argument('--mirostat-eta', type=float, default=_d['mirostat_eta'], metavar='N', help='Mirostat eta')
+group.add_argument('--do-sample', action=argparse.BooleanOptionalAction, default=_d['do_sample'], help='Do sample')
+group.add_argument('--dynamic-temperature', action=argparse.BooleanOptionalAction, default=_d['dynamic_temperature'], help='Dynamic temperature')
+group.add_argument('--temperature-last', action=argparse.BooleanOptionalAction, default=_d['temperature_last'], help='Temperature last')
+group.add_argument('--sampler-priority', type=str, default=_d['sampler_priority'], metavar='N', help='Sampler priority')
+group.add_argument('--dry-sequence-breakers', type=str, default=_d['dry_sequence_breakers'], metavar='N', help='DRY sequence breakers')
+group.add_argument('--enable-thinking', action=argparse.BooleanOptionalAction, default=True, help='Enable thinking')
+group.add_argument('--reasoning-effort', type=str, default='medium', metavar='N', help='Reasoning effort')
 
 # Handle CMD_FLAGS.txt
 cmd_flags_path = user_data_dir / "CMD_FLAGS.txt"

From 66fb79fe15f38c313ed9a2fbaabfad252c2b121a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 20:46:08 -0800
Subject: [PATCH 1244/1701] llama.cpp: Add --fit-target param

---
 modules/llama_cpp_server.py | 2 ++
 modules/loaders.py          | 1 +
 modules/shared.py           | 1 +
 modules/ui.py               | 1 +
 modules/ui_model_menu.py    | 1 +
 5 files changed, 6 insertions(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index eb99f2a86d..0e8828d9d6 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -354,6 +354,8 @@ def _start_server(self):
             cmd += ["--gpu-layers", str(shared.args.gpu_layers), "--fit", "off"]
         else:
             cmd += ["--fit", "on"]
+            if shared.args.fit_target:
+                cmd += ["--fit-target", shared.args.fit_target]
 
         if shared.args.threads > 0:
             cmd += ["--threads", str(shared.args.threads)]
diff --git a/modules/loaders.py b/modules/loaders.py
index 15b8dfebd4..42a5ff1c5e 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -6,6 +6,7 @@
 loaders_and_params = OrderedDict({
     'llama.cpp': [
         'gpu_layers',
+        'fit_target',
         'cpu_moe',
         'threads',
         'threads_batch',
diff --git a/modules/shared.py b/modules/shared.py
index e9535aa001..de0820af4f 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -108,6 +108,7 @@
 group.add_argument('--threads-batch', type=int, default=0, help='Number of threads to use for batches/prompt processing.')
 group.add_argument('--numa', action='store_true', help='Activate NUMA task allocation for llama.cpp.')
 group.add_argument('--parallel', type=int, default=1, help='Number of parallel request slots. The context size is divided equally among slots. For example, to have 4 slots with 8192 context each, set ctx_size to 32768.')
+group.add_argument('--fit-target', type=str, default='1024', help='Target VRAM margin per device for auto GPU layers, comma-separated list of values in MiB. A single value is broadcast across all devices. Default: 1024.')
 group.add_argument('--extra-flags', type=str, default=None, help='Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"')
 
 # Transformers/Accelerate
diff --git a/modules/ui.py b/modules/ui.py
index ae998ebb41..70e929f20c 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -125,6 +125,7 @@ def list_model_elements():
         'loader',
         'cpu_memory',
         'gpu_layers',
+        'fit_target',
         'cpu_moe',
         'threads',
         'threads_batch',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 33b39a2569..7e91f1cecc 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -46,6 +46,7 @@ def create_ui():
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
+                            shared.gradio['fit_target'] = gr.Textbox(label='fit-target', value=shared.args.fit_target, info='Target VRAM margin per device for auto GPU layers (MiB). Comma-separated list for multiple devices. Default: 1024.')
                             shared.gradio['tp_backend'] = gr.Dropdown(label="tp-backend", choices=['native', 'nccl'], value=shared.args.tp_backend, info='The backend for tensor parallelism.')
 
                         with gr.Column():

From 521ddbb7225e34b3ee28b9d3f97a1b27764c9337 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 01:56:13 -0300
Subject: [PATCH 1245/1701] Security: restrict API model loading args to
 UI-exposed parameters

The /v1/internal/model/load endpoint previously allowed setting any
shared.args attribute, including security-sensitive flags like
trust_remote_code. Now only keys from list_model_elements() are accepted.
---
 extensions/openai/models.py | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/extensions/openai/models.py b/extensions/openai/models.py
index 115149b07a..74f9dbee65 100644
--- a/extensions/openai/models.py
+++ b/extensions/openai/models.py
@@ -1,4 +1,4 @@
-from modules import shared
+from modules import shared, ui
 from modules.logging_colors import logger
 from modules.LoRA import add_lora_to_model
 from modules.models import load_model, unload_model
@@ -46,9 +46,13 @@ def _load_model(data):
     update_model_parameters(model_settings)
 
     # Update shared.args with custom model loading settings
+    # Security: only allow keys that correspond to model loading
+    # parameters exposed in the UI. Never allow security-sensitive
+    # flags like trust_remote_code to be set via the API.
+    allowed_keys = set(ui.list_model_elements())
     if args:
         for k in args:
-            if hasattr(shared.args, k):
+            if k in allowed_keys and hasattr(shared.args, k):
                 setattr(shared.args, k, args[k])
 
     shared.model, shared.tokenizer = load_model(model_name)

From eba262d47ad10c7aa3b7d3d3d35c02e33c45306b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 01:59:18 -0300
Subject: [PATCH 1246/1701] Security: prevent path traversal in
 character/user/file save and delete

---
 modules/chat.py           |  9 +++++++--
 modules/ui_file_saving.py |  5 ++++-
 modules/utils.py          | 11 +++++++++++
 3 files changed, 22 insertions(+), 3 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 5d7e0e5657..bc4fc1d878 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -36,6 +36,7 @@
     delete_file,
     get_available_characters,
     get_available_users,
+    sanitize_filename,
     save_file
 )
 from modules.web_search import add_web_search_attachments
@@ -1557,12 +1558,12 @@ def upload_character(file, img_path, tavern=False):
         data = yaml.safe_load(decoded_file)
 
     if 'char_name' in data:
-        name = data['char_name']
+        name = sanitize_filename(data['char_name'])
         greeting = data['char_greeting']
         context = build_pygmalion_style_context(data)
         yaml_data = generate_character_yaml(name, greeting, context)
     else:
-        name = data['name']
+        name = sanitize_filename(data['name'])
         yaml_data = generate_character_yaml(data['name'], data['greeting'], data['context'])
 
     outfile_name = name
@@ -1653,6 +1654,7 @@ def generate_instruction_template_yaml(instruction_template):
 
 
 def save_character(name, greeting, context, picture, filename):
+    filename = sanitize_filename(filename)
     if filename == "":
         logger.error("The filename is empty, so the character will not be saved.")
         return
@@ -1668,6 +1670,7 @@ def save_character(name, greeting, context, picture, filename):
 
 
 def delete_character(name, instruct=False):
+    name = sanitize_filename(name)
     # Check for character data files
     for extension in ["yml", "yaml", "json"]:
         delete_file(shared.user_data_dir / 'characters' / f'{name}.{extension}')
@@ -1751,6 +1754,7 @@ def generate_user_yaml(name, user_bio):
 
 def save_user(name, user_bio, picture, filename):
     """Save user profile to YAML file"""
+    filename = sanitize_filename(filename)
     if filename == "":
         logger.error("The filename is empty, so the user will not be saved.")
         return
@@ -1772,6 +1776,7 @@ def save_user(name, user_bio, picture, filename):
 
 def delete_user(name):
     """Delete user profile files"""
+    name = sanitize_filename(name)
     # Check for user data files
     for extension in ["yml", "yaml", "json"]:
         delete_file(shared.user_data_dir / 'users' / f'{name}.{extension}')
diff --git a/modules/ui_file_saving.py b/modules/ui_file_saving.py
index 46087acea1..3ed256f821 100644
--- a/modules/ui_file_saving.py
+++ b/modules/ui_file_saving.py
@@ -3,7 +3,7 @@
 import gradio as gr
 
 from modules import chat, presets, shared, ui, utils
-from modules.utils import gradio
+from modules.utils import gradio, sanitize_filename
 
 
 def create_ui():
@@ -91,6 +91,7 @@ def create_event_handlers():
 
 def handle_save_preset_confirm_click(filename, contents):
     try:
+        filename = sanitize_filename(filename)
         utils.save_file(str(shared.user_data_dir / "presets" / f"{filename}.yaml"), contents)
         available_presets = utils.get_available_presets()
         output = gr.update(choices=available_presets, value=filename)
@@ -106,6 +107,7 @@ def handle_save_preset_confirm_click(filename, contents):
 
 def handle_save_confirm_click(root, filename, contents):
     try:
+        filename = sanitize_filename(filename)
         utils.save_file(root + filename, contents)
     except Exception:
         traceback.print_exc()
@@ -115,6 +117,7 @@ def handle_save_confirm_click(root, filename, contents):
 
 def handle_delete_confirm_click(root, filename):
     try:
+        filename = sanitize_filename(filename)
         utils.delete_file(root + filename)
     except Exception:
         traceback.print_exc()
diff --git a/modules/utils.py b/modules/utils.py
index 447685b782..7ab4a55476 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -15,6 +15,17 @@ def gradio(*keys):
     return [shared.gradio[k] for k in keys]
 
 
+def sanitize_filename(name):
+    """Strip path traversal components from a filename.
+
+    Returns only the final path component with leading dots removed,
+    preventing directory traversal via '../' or absolute paths.
+    """
+    name = Path(name).name  # drop all directory components
+    name = name.lstrip('.')  # remove leading dots
+    return name
+
+
 def _is_path_allowed(abs_path_str):
     """Check if a path is under the project root or the configured user_data directory."""
     abs_path = Path(abs_path_str).resolve()

From 1729fb07b94cd9979253fa81b94410a4a6135ed0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 21:04:24 -0800
Subject: [PATCH 1247/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 12 files changed, 20 insertions(+), 20 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 6b5e2e784d..2f745947c3 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,8 +42,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.23/exllamav3-0.0.23+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.23/exllamav3-0.0.23+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 059de7229c..f1bab0d6d9 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -39,5 +39,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 528224a6b1..1ce19e3d28 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -39,4 +39,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 29eca60f4e..391a5f66b9 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -39,4 +39,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index ae31507a56..9ab5327c3d 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -39,5 +39,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 40694564f1..0c7803c434 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 64a2954d12..1f5f62ad7b 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 5669b4e190..56dc4ff5e2 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index ed31dbaba1..a70213be37 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index ed100f0e7e..87c021b5ba 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 829e89730b..8f7faa4392 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index cd140cd58e..164f4adf33 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.83.0/llama_cpp_binaries-0.83.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 8be444a559dbe18bbe59830cd7dabba7083d25b1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 21:05:15 -0800
Subject: [PATCH 1248/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 14 files changed, 28 insertions(+), 28 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 2f745947c3..5e4a077d7d 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -33,8 +33,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index f1bab0d6d9..2614263172 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 1ce19e3d28..e26cf35cfd 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 391a5f66b9..64079bef56 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 9ab5327c3d..fca5e1549b 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 386ee2a878..2a70d3a7be 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 0c7803c434..f70601f10a 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 1f5f62ad7b..2c34d6dcf0 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 56dc4ff5e2..0b84e3a420 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index a70213be37..2bed09a94c 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 87c021b5ba..206b68e3a4 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 8f7faa4392..dae9c266b5 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 872a94151d..267357be9b 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 164f4adf33..5cd07714b7 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio-4.37.2+custom.7-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.7/gradio_client-1.0.2+custom.7-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From f06583b2b986ca3164ad4eb8c772d379e8645150 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 21:16:37 -0800
Subject: [PATCH 1249/1701] API: Use \n instead of \r\n as the SSE separator to
 match OpenAI

---
 extensions/openai/script.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 7a30e311aa..521d2cb4bc 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -125,7 +125,7 @@ async def generator():
                 stop_event.set()
                 response.close()
 
-        return EventSourceResponse(generator())  # SSE streaming
+        return EventSourceResponse(generator(), sep="\n")  # SSE streaming
 
     else:
         stop_event = threading.Event()
@@ -160,7 +160,7 @@ async def generator():
                 stop_event.set()
                 response.close()
 
-        return EventSourceResponse(generator())  # SSE streaming
+        return EventSourceResponse(generator(), sep="\n")  # SSE streaming
 
     else:
         stop_event = threading.Event()

From d0ac58ad3135776ec6e108f82a5c64a9662cc122 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 21:25:03 -0800
Subject: [PATCH 1250/1701] API: Fix tool_calls placement and other response
 compatibility issues

---
 docs/12 - OpenAI API.md          | 4 ++--
 extensions/openai/completions.py | 5 ++---
 extensions/openai/script.py      | 4 ++--
 3 files changed, 6 insertions(+), 7 deletions(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index ebe2479711..637cccedd3 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -456,11 +456,11 @@ for _ in range(10):
         messages.append({
             "role": "assistant",
             "content": choice["message"]["content"],
-            "tool_calls": choice["tool_calls"],
+            "tool_calls": choice["message"]["tool_calls"],
         })
 
         # Execute each tool and add results to history
-        for tool_call in choice["tool_calls"]:
+        for tool_call in choice["message"]["tool_calls"]:
             name = tool_call["function"]["name"]
             arguments = json.loads(tool_call["function"]["arguments"])
             result = execute_tool(name, arguments)
diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 5187343fdd..1538b87a8b 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -370,8 +370,7 @@ def chat_streaming_chunk(content, chunk_tool_calls=None):
             resp_list: [{
                 "index": 0,
                 "finish_reason": stop_reason,
-                "message": {"role": "assistant", "content": answer},
-                "tool_calls": tool_calls
+                "message": {"role": "assistant", "content": answer, "tool_calls": tool_calls},
             }],
             "usage": {
                 "prompt_tokens": token_count,
@@ -389,7 +388,7 @@ def chat_streaming_chunk(content, chunk_tool_calls=None):
 
 
 def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_event=None):
-    object_type = 'text_completion.chunk' if stream else 'text_completion'
+    object_type = 'text_completion'
     created_time = int(time.time())
     cmpl_id = "conv-%d" % (int(time.time() * 1000000000))
     resp_list = 'data' if is_legacy else 'choices'
diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 521d2cb4bc..bfb6fd54c4 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -359,7 +359,7 @@ async def handle_load_model(request_data: LoadModelRequest):
         return JSONResponse(content="OK")
     except Exception:
         traceback.print_exc()
-        return HTTPException(status_code=400, detail="Failed to load the model.")
+        raise HTTPException(status_code=400, detail="Failed to load the model.")
 
 
 @app.post("/v1/internal/model/unload", dependencies=check_admin_key)
@@ -380,7 +380,7 @@ async def handle_load_loras(request_data: LoadLorasRequest):
         return JSONResponse(content="OK")
     except Exception:
         traceback.print_exc()
-        return HTTPException(status_code=400, detail="Failed to apply the LoRA(s).")
+        raise HTTPException(status_code=400, detail="Failed to apply the LoRA(s).")
 
 
 @app.post("/v1/internal/lora/unload", dependencies=check_admin_key)

From 93ebfa2b7ec100680b3dcf95e2c078e71a9b685f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 02:38:13 -0300
Subject: [PATCH 1251/1701] Fix llama-server output filter for new log format

---
 modules/llama_cpp_server.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 0e8828d9d6..3249497831 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -520,6 +520,8 @@ def filter_stderr_with_progress(process_stderr):
     inline (overwriting the same line) until completion.
     """
     progress_re = re.compile(r'slot update_slots: id.*progress = (\d+\.\d+)')
+    ansi_re = re.compile(r'\x1b\[[0-9;]*[a-zA-Z]')
+    log_prefix_re = re.compile(r'^[IWED] ')
     last_was_progress = False
 
     try:
@@ -538,6 +540,7 @@ def filter_stderr_with_progress(process_stderr):
                 line_bytes, buffer = buffer.split(b'\n', 1)
                 try:
                     line = line_bytes.decode('utf-8', errors='replace').strip('\r\n')
+                    line = log_prefix_re.sub('', ansi_re.sub('', line))
                     if line:  # Process non-empty lines
                         match = progress_re.search(line)
 

From 3880c1a406516b52082ffd8cb51b9648e2b63f1e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 02:41:16 -0300
Subject: [PATCH 1252/1701] API: Accept content:null and complex tool
 definitions in tool calling requests

---
 extensions/openai/completions.py |  6 +++++-
 extensions/openai/typing.py      | 12 +++++-------
 2 files changed, 10 insertions(+), 8 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 1538b87a8b..297c9ba7c2 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -197,7 +197,11 @@ def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, p
         # Handle multimodal content validation
         content = m.get('content')
         if content is None:
-            raise InvalidRequestError(message="messages: missing content", param='messages')
+            # OpenAI allows content: null on assistant messages when tool_calls is present
+            if m['role'] == 'assistant' and m.get('tool_calls'):
+                m['content'] = ''
+            else:
+                raise InvalidRequestError(message="messages: missing content", param='messages')
 
         # Validate multimodal content structure
         if isinstance(content, list):
diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 8af75ef834..4cf5f99701 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -1,8 +1,8 @@
 import json
 import time
-from typing import Dict, List, Optional
+from typing import Any, Dict, List, Optional
 
-from pydantic import BaseModel, Field, model_validator, validator
+from pydantic import BaseModel, ConfigDict, Field, model_validator, validator
 
 from modules import shared
 
@@ -65,22 +65,20 @@ class ToolDefinition(BaseModel):
 
 
 class ToolFunction(BaseModel):
+    model_config = ConfigDict(extra='allow')
     description: str
     name: str
     parameters: 'ToolParameters'
 
 
 class ToolParameters(BaseModel):
-    properties: Optional[Dict[str, 'ToolProperty']] = None
+    model_config = ConfigDict(extra='allow')
+    properties: Optional[Dict[str, Any]] = None
     required: Optional[list[str]] = None
     type: str
     description: Optional[str] = None
 
 
-class ToolProperty(BaseModel):
-    description: Optional[str] = None
-    type: Optional[str] = None  # we are faced with definitions like anyOf, e.g. {'type': 'function', 'function': {'name': 'git_create_branch', 'description': 'Creates a new branch from an optional base branch', 'parameters': {'type': 'object', 'properties': {'repo_path': {'title': 'Repo Path', 'type': 'string'}, 'branch_name': {'title': 'Branch Name', 'type': 'string'}, 'base_branch': {'anyOf': [{'type': 'string'}, {'type': 'null'}], 'default': None, 'title': 'Base Branch'}}, 'required': ['repo_path', 'branch_name'], 'title': 'GitCreateBranch'}}}
-
 
 class FunctionCall(BaseModel):
     name: str

From f9ed8820de1e6aca670b5880b3b6cec60d08721a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 21:43:33 -0800
Subject: [PATCH 1253/1701] API: Make tool function description and parameters
 optional

---
 extensions/openai/typing.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 4cf5f99701..e48b7b6037 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -66,9 +66,9 @@ class ToolDefinition(BaseModel):
 
 class ToolFunction(BaseModel):
     model_config = ConfigDict(extra='allow')
-    description: str
+    description: Optional[str] = None
     name: str
-    parameters: 'ToolParameters'
+    parameters: Optional['ToolParameters'] = None
 
 
 class ToolParameters(BaseModel):

From 6d7018069c076aa0b9ee16445114f1b990b7d093 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 5 Mar 2026 21:56:01 -0800
Subject: [PATCH 1254/1701] Installer: Use absolute Python path in Windows
 batch scripts

---
 start_windows.bat         | 2 +-
 update_wizard_windows.bat | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/start_windows.bat b/start_windows.bat
index 406fd27e07..dd096760e4 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -99,7 +99,7 @@ set "CUDA_HOME=%CUDA_PATH%"
 call "%CONDA_ROOT_PREFIX%\condabin\conda.bat" activate "%INSTALL_ENV_DIR%" || ( echo. && echo Miniforge hook not found. && goto end )
 
 @rem setup installer env
-call python one_click.py %*
+call "%INSTALL_ENV_DIR%\python.exe" one_click.py %*
 
 @rem below are functions for the script   next line skips these during normal execution
 goto end
diff --git a/update_wizard_windows.bat b/update_wizard_windows.bat
index e9203c529b..5e5beec0c5 100755
--- a/update_wizard_windows.bat
+++ b/update_wizard_windows.bat
@@ -28,7 +28,7 @@ set "CUDA_HOME=%CUDA_PATH%"
 call "%CONDA_ROOT_PREFIX%\condabin\conda.bat" activate "%INSTALL_ENV_DIR%" || ( echo. && echo Miniforge hook not found. && goto end )
 
 @rem update installer env
-call python one_click.py --update-wizard && (
+call "%INSTALL_ENV_DIR%\python.exe" one_click.py --update-wizard && (
     echo.
     echo Have a great day!
 )

From 0e0e3ceb974e56c9495e6992ee48a7132fc361d2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 05:46:08 -0800
Subject: [PATCH 1255/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 14 files changed, 28 insertions(+), 28 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 5e4a077d7d..aa8dfff905 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -33,8 +33,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 2614263172..99e9515155 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index e26cf35cfd..a51be49b8b 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 64079bef56..e5d6e11255 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index fca5e1549b..0c88d16146 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 2a70d3a7be..a86ed47080 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -30,8 +30,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index f70601f10a..763166cc2b 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 2c34d6dcf0..2aca3e4727 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 0b84e3a420..963d52edc6 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 2bed09a94c..7b3c9bf5e7 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 206b68e3a4..5c8957bcfd 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index dae9c266b5..a4df758091 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 267357be9b..c2fc33eb90 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 5cd07714b7..c78ba69518 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio-4.37.2+custom.8-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.8/gradio_client-1.0.2+custom.8-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From 86d59b440403573b51a9665343781c626de6258e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 06:16:57 -0800
Subject: [PATCH 1256/1701] Installer: Fix edge case in wheel re-download
 caching

---
 one_click.py | 14 ++++++++------
 1 file changed, 8 insertions(+), 6 deletions(-)

diff --git a/one_click.py b/one_click.py
index 1af1a686a3..bf7c59092d 100644
--- a/one_click.py
+++ b/one_click.py
@@ -395,6 +395,7 @@ def update_requirements(initial_installation=False, pull=True):
 
         # Perform the git pull
         run_cmd("git pull --autostash", assert_success=True, environment=True)
+        current_commit = get_current_commit()
 
         # Check hashes after pulling
         after_hashes = {file: calculate_file_hash(file) for file in files_to_check}
@@ -411,17 +412,12 @@ def update_requirements(initial_installation=False, pull=True):
 
                 # Save state before exiting
                 state = load_state()
+                state['last_installed_commit'] = current_commit
                 if wheels_changed:
                     state['wheels_changed'] = True
                 save_state(state)
                 sys.exit(1)
 
-    # Save current state
-    state = load_state()
-    state['last_installed_commit'] = current_commit
-    state.pop('wheels_changed', None)  # Remove wheels_changed flag
-    save_state(state)
-
     if os.environ.get("INSTALL_EXTENSIONS", "").lower() in ("yes", "y", "true", "1", "t", "on"):
         install_extensions_requirements()
 
@@ -456,6 +452,12 @@ def update_requirements(initial_installation=False, pull=True):
     # Install/update the project requirements
     run_cmd("python -m pip install -r temp_requirements.txt --upgrade", assert_success=True, environment=True)
 
+    # Save state after successful installation
+    state = load_state()
+    state['last_installed_commit'] = current_commit
+    state.pop('wheels_changed', None)
+    save_state(state)
+
     # Clean up
     os.remove('temp_requirements.txt')
     clear_cache()

From 36dbc4ccceb2d0304e2d3493f140e17d4c60c6ad Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 06:28:35 -0800
Subject: [PATCH 1257/1701] Remove unused colorama and psutil requirements

---
 requirements/full/requirements.txt               | 2 --
 requirements/full/requirements_amd.txt           | 2 --
 requirements/full/requirements_apple_intel.txt   | 2 --
 requirements/full/requirements_apple_silicon.txt | 2 --
 requirements/full/requirements_cpu_only.txt      | 2 --
 requirements/full/requirements_nowheels.txt      | 2 --
 6 files changed, 12 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index aa8dfff905..b01566f319 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -1,7 +1,6 @@
 accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 bitsandbytes==0.49.*
-colorama
 datasets
 diffusers==0.36.*
 einops
@@ -15,7 +14,6 @@ numpy==2.2.*
 pandas
 peft==0.18.*
 Pillow>=9.5.0
-psutil
 pydantic==2.11.0
 pymupdf==1.27.1
 python-docx==1.1.2
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 99e9515155..85713d7a3e 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -1,6 +1,5 @@
 accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
-colorama
 datasets
 diffusers==0.36.*
 einops
@@ -13,7 +12,6 @@ numpy==2.2.*
 pandas
 peft==0.18.*
 Pillow>=9.5.0
-psutil
 pydantic==2.11.0
 pymupdf==1.27.1
 python-docx==1.1.2
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index a51be49b8b..f2b2d479a2 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -1,6 +1,5 @@
 accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
-colorama
 datasets
 diffusers==0.36.*
 einops
@@ -13,7 +12,6 @@ numpy==2.2.*
 pandas
 peft==0.18.*
 Pillow>=9.5.0
-psutil
 pydantic==2.11.0
 pymupdf==1.27.1
 python-docx==1.1.2
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index e5d6e11255..b44932d010 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -1,6 +1,5 @@
 accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
-colorama
 datasets
 diffusers==0.36.*
 einops
@@ -13,7 +12,6 @@ numpy==2.2.*
 pandas
 peft==0.18.*
 Pillow>=9.5.0
-psutil
 pydantic==2.11.0
 pymupdf==1.27.1
 python-docx==1.1.2
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 0c88d16146..efd0fb06f5 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -1,6 +1,5 @@
 accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
-colorama
 datasets
 diffusers==0.36.*
 einops
@@ -13,7 +12,6 @@ numpy==2.2.*
 pandas
 peft==0.18.*
 Pillow>=9.5.0
-psutil
 pydantic==2.11.0
 pymupdf==1.27.1
 python-docx==1.1.2
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index a86ed47080..0a924d314c 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -1,6 +1,5 @@
 accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
-colorama
 datasets
 diffusers==0.36.*
 einops
@@ -13,7 +12,6 @@ numpy==2.2.*
 pandas
 peft==0.18.*
 Pillow>=9.5.0
-psutil
 pydantic==2.11.0
 pymupdf==1.27.1
 python-docx==1.1.2

From 3323dedd08ea0105b5d209a882a8c52c33626099 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 06:30:01 -0800
Subject: [PATCH 1258/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 12 files changed, 20 insertions(+), 20 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index b01566f319..85fbd5df73 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.23/exllamav3-0.0.23+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.23/exllamav3-0.0.23+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 85713d7a3e..de41f6063a 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index f2b2d479a2..9f0a113a5a 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index b44932d010..e7c95a7154 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index efd0fb06f5..28112860ed 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 763166cc2b..9ff8a55869 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 2aca3e4727..b79defafa5 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 963d52edc6..19310f9ca5 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 7b3c9bf5e7..3a269054db 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 5c8957bcfd..b5d319ac36 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index a4df758091..1e05de220a 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index c78ba69518..6298be11a5 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.84.0/llama_cpp_binaries-0.84.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From e7e0df0101886baad4c39a655ff6e922f159bd48 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 11:52:16 -0300
Subject: [PATCH 1259/1701] Fix hover menu shifting down when chat input grows

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index c55a29dcf2..2d19b5c8d0 100644
--- a/css/main.css
+++ b/css/main.css
@@ -715,7 +715,7 @@ audio {
 .hover-menu {
     display: none;
     position: absolute;
-    bottom: 80%;
+    bottom: 100%;
     left: 0;
     box-shadow: 0 0 5px rgb(0 0 0 / 25%);
     z-index: 10000;

From 3bab7fbfd48c4bd4dcca4a9b3c8bd07a7cb1a52a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 06:52:49 -0800
Subject: [PATCH 1260/1701] Update Colab notebook: new default model, direct
 GGUF URL support

---
 Colab-TextGen-GPU.ipynb | 34 ++++++++++++++++++----------------
 download-model.py       | 20 ++++++++++++++++++++
 2 files changed, 38 insertions(+), 16 deletions(-)

diff --git a/Colab-TextGen-GPU.ipynb b/Colab-TextGen-GPU.ipynb
index 971099f628..f256b0e84e 100644
--- a/Colab-TextGen-GPU.ipynb
+++ b/Colab-TextGen-GPU.ipynb
@@ -51,7 +51,7 @@
       "source": [
         "#@title 2. Launch the web UI\n",
         "\n",
-        "#@markdown If unsure about the branch, write \"main\" or leave it blank.\n",
+        "#@markdown You can provide a direct GGUF link or a Hugging Face model URL.\n",
         "\n",
         "import os\n",
         "from pathlib import Path\n",
@@ -72,9 +72,9 @@
         "  ./start_linux.sh\n",
         "\n",
         "# Parameters\n",
-        "model_url = \"https://huggingface.co/turboderp/gemma-2-9b-it-exl2\" #@param {type:\"string\"}\n",
-        "branch = \"8.0bpw\" #@param {type:\"string\"}\n",
-        "command_line_flags = \"--n-gpu-layers 128 --load-in-4bit --use_double_quant --no_flash_attn\" #@param {type:\"string\"}\n",
+        "model_url = \"https://huggingface.co/unsloth/Qwen3.5-9B-GGUF/resolve/main/Qwen3.5-9B-Q4_K_M.gguf\" #@param {type:\"string\"}\n",
+        "branch = \"\" #@param {type:\"string\"}\n",
+        "command_line_flags = \"--load-in-4bit --use_double_quant\" #@param {type:\"string\"}\n",
         "api = False #@param {type:\"boolean\"}\n",
         "\n",
         "if api:\n",
@@ -83,26 +83,28 @@
         "      command_line_flags += f\" {param}\"\n",
         "\n",
         "model_url = model_url.strip()\n",
+        "model_name = \"\"\n",
         "if model_url != \"\":\n",
         "    if not model_url.startswith('http'):\n",
         "        model_url = 'https://huggingface.co/' + model_url\n",
         "\n",
-        "    # Download the model\n",
-        "    url_parts = model_url.strip('/').strip().split('/')\n",
-        "    output_folder = f\"{url_parts[-2]}_{url_parts[-1]}\"\n",
-        "    branch = branch.strip('\"\\' ')\n",
-        "    if branch.strip() not in ['', 'main']:\n",
-        "        output_folder += f\"_{branch}\"\n",
-        "        !python download-model.py {model_url} --branch {branch}\n",
-        "    else:\n",
+        "    branch = branch.strip()\n",
+        "    if '/resolve/' in model_url:\n",
+        "        model_name = model_url.split('?')[0].split('/')[-1]\n",
         "        !python download-model.py {model_url}\n",
-        "else:\n",
-        "    output_folder = \"\"\n",
+        "    else:\n",
+        "        url_parts = model_url.strip('/').split('/')\n",
+        "        model_name = f\"{url_parts[-2]}_{url_parts[-1]}\"\n",
+        "        if branch not in ['', 'main']:\n",
+        "            model_name += f\"_{branch}\"\n",
+        "            !python download-model.py {model_url} --branch {branch}\n",
+        "        else:\n",
+        "            !python download-model.py {model_url}\n",
         "\n",
         "# Start the web UI\n",
         "cmd = f\"./start_linux.sh {command_line_flags} --share\"\n",
-        "if output_folder != \"\":\n",
-        "    cmd += f\" --model {output_folder}\"\n",
+        "if model_name != \"\":\n",
+        "    cmd += f\" --model {model_name}\"\n",
         "\n",
         "!$cmd"
       ],
diff --git a/download-model.py b/download-model.py
index 7acb7f46a0..95d25e1630 100644
--- a/download-model.py
+++ b/download-model.py
@@ -413,6 +413,26 @@ def check_model_files(self, model, branch, links, sha256, output_folder):
         sys.exit()
 
     downloader = ModelDownloader(max_retries=args.max_retries)
+
+    # Handle direct file URLs (e.g. https://huggingface.co/org/repo/resolve/branch/file.gguf)
+    if '/resolve/' in model:
+        url = model if model.startswith('http') else f'{base}/{model}'
+        url = url.split('?')[0]
+        filename = url.split('/')[-1]
+
+        if args.output:
+            output_folder = Path(args.output)
+        elif args.model_dir:
+            output_folder = Path(args.model_dir)
+        else:
+            user_data_dir = Path(args.user_data_dir) if args.user_data_dir else resolve_user_data_dir()
+            output_folder = user_data_dir / 'models'
+
+        output_folder.mkdir(parents=True, exist_ok=True)
+        print(f"Downloading {filename} to {output_folder}")
+        downloader.get_single_file(url, output_folder, start_from_scratch=args.clean)
+        sys.exit()
+
     # Clean up the model/branch names
     try:
         model, branch = downloader.sanitize_model_and_branch_names(model, branch)

From 8e24a208735be50ae64ef6656a412a7912520a8b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 07:39:50 -0800
Subject: [PATCH 1261/1701] Installer: Fix libstdcxx-ng version pin causing
 conda solver to hang on Python 3.13

---
 one_click.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/one_click.py b/one_click.py
index bf7c59092d..67bd3a9639 100644
--- a/one_click.py
+++ b/one_click.py
@@ -422,7 +422,7 @@ def update_requirements(initial_installation=False, pull=True):
         install_extensions_requirements()
 
     if is_linux():
-        run_cmd(f"conda install -y -c conda-forge libstdcxx-ng=={LIBSTDCXX_VERSION_LINUX}", assert_success=True, environment=True)
+        run_cmd(f"conda install -y -c conda-forge 'libstdcxx-ng>={LIBSTDCXX_VERSION_LINUX}'", assert_success=True, environment=True)
 
     # Update PyTorch
     if not initial_installation:

From 160f7ad6b470de260d47fcbc81cf403767acbffc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 12:56:33 -0300
Subject: [PATCH 1262/1701] Handle SIGTERM to stop llama-server on pkill

---
 server.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/server.py b/server.py
index a809cd7b9e..ff2d1db2d6 100644
--- a/server.py
+++ b/server.py
@@ -78,6 +78,7 @@ def signal_handler(sig, frame):
 
 
 signal.signal(signal.SIGINT, signal_handler)
+signal.signal(signal.SIGTERM, signal_handler)
 
 
 def create_interface():

From 3531069824a1ba1d06839b7d7b6b94ecac4ea1b1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 13:12:14 -0300
Subject: [PATCH 1263/1701] API: Support Llama 4 tool calling and fix tool
 calling edge cases

---
 extensions/openai/completions.py |  6 +++
 extensions/openai/utils.py       | 63 +++++++++++++++++++++++++++++++-
 2 files changed, 68 insertions(+), 1 deletion(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 297c9ba7c2..46502bdc46 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -603,6 +603,12 @@ def validateTools(tools: list[dict]):
         tool = tools[idx]
         try:
             tool_definition = ToolDefinition(**tool)
+            # Backfill defaults so Jinja2 templates don't crash on missing fields
+            func = tool.get("function", {})
+            if "description" not in func:
+                func["description"] = ""
+            if "parameters" not in func:
+                func["parameters"] = {"type": "object", "properties": {}}
             if valid_tools is None:
                 valid_tools = []
             valid_tools.append(tool)
diff --git a/extensions/openai/utils.py b/extensions/openai/utils.py
index 6937a1084b..64bd163104 100644
--- a/extensions/openai/utils.py
+++ b/extensions/openai/utils.py
@@ -91,7 +91,7 @@ def _parseChannelToolCalls(answer: str, tool_names: list[str]):
     """
     matches = []
     for m in re.finditer(
-        r'<\|channel\|>commentary to=functions\.([^<\s]+)\s*(?:<\|constrain\|>json)?<\|message\|>(\{[^}]*(?:\{[^}]*\}[^}]*)*\})',
+        r'<\|channel\|>commentary to=functions\.([^<\s]+)\s*(?:<\|constrain\|>json)?<\|message\|>(\{[^{}]*(?:\{[^{}]*\}[^{}]*)*\})',
         answer
     ):
         func_name = m.group(1).strip()
@@ -181,6 +181,62 @@ def _parseXmlParamToolCalls(answer: str, tool_names: list[str]):
     return matches
 
 
+def _parsePythonicToolCalls(answer: str, tool_names: list[str]):
+    """Parse pythonic-style tool calls used by Llama 4 and similar models.
+
+    Format:
+        [func_name(param1="value1", param2="value2"), func_name2(...)]
+    """
+    matches = []
+    # Match a bracketed list of function calls
+    bracket_match = re.search(r'\[([^\[\]]+)\]', answer)
+    if not bracket_match:
+        return matches
+
+    inner = bracket_match.group(1)
+
+    # Build pattern for known tool names
+    escaped_names = [re.escape(name) for name in tool_names]
+    name_pattern = '|'.join(escaped_names)
+
+    for call_match in re.finditer(
+        r'(' + name_pattern + r')\(([^)]*)\)',
+        inner
+    ):
+        func_name = call_match.group(1)
+        params_str = call_match.group(2).strip()
+        arguments = {}
+
+        if params_str:
+            # Parse key="value" pairs, handling commas inside quoted values
+            for param_match in re.finditer(
+                r'(\w+)\s*=\s*("(?:[^"\\]|\\.)*"|\'(?:[^\'\\]|\\.)*\'|[^,\)]+)',
+                params_str
+            ):
+                param_name = param_match.group(1)
+                param_value = param_match.group(2).strip()
+                # Strip surrounding quotes
+                if (param_value.startswith('"') and param_value.endswith('"')) or \
+                   (param_value.startswith("'") and param_value.endswith("'")):
+                    param_value = param_value[1:-1]
+                # Try to parse as JSON for numeric/bool/null values
+                try:
+                    param_value = json.loads(param_value)
+                except (json.JSONDecodeError, ValueError):
+                    pass
+                arguments[param_name] = param_value
+
+        matches.append({
+            "type": "function",
+            "function": {
+                "name": func_name,
+                "arguments": arguments
+            }
+        })
+
+    return matches
+
+
 def parseToolCall(answer: str, tool_names: list[str]):
     matches = []
 
@@ -203,6 +259,11 @@ def parseToolCall(answer: str, tool_names: list[str]):
     if matches:
         return matches
 
+    # Check for pythonic-style tool calls (e.g. Llama 4 format)
+    matches = _parsePythonicToolCalls(answer, tool_names)
+    if matches:
+        return matches
+
     # Define the regex pattern to find the JSON content wrapped in , , , and other tags observed from various models
     patterns = [r"(```[^\n]*)\n(.*?)```", r"<([^>]+)>(.*?)"]
 

From f5acf5520719ee6a9b0d0667e4994ab9eeef332a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 13:33:24 -0300
Subject: [PATCH 1264/1701] Add --chat-template-file flag to override the
 default instruction template for API requests

Matches llama.cpp's flag name. Supports .jinja, .jinja2, and .yaml files.
Priority: per-request params > --chat-template-file > model's built-in template.
---
 extensions/openai/completions.py | 17 +++++++++++++++++
 modules/shared.py                |  6 ++++++
 modules/training.py              |  6 +++---
 3 files changed, 26 insertions(+), 3 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 46502bdc46..cabfce9916 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -1,9 +1,12 @@
 import copy
+import functools
 import json
 import time
 from collections import deque
+from pathlib import Path
 
 import tiktoken
+import yaml
 from pydantic import ValidationError
 
 from extensions.openai.errors import InvalidRequestError
@@ -22,6 +25,18 @@
 from modules.text_generation import decode, encode, generate_reply
 
 
+@functools.cache
+def load_chat_template_file(filepath):
+    """Load a chat template from a file path (.jinja, .jinja2, or .yaml/.yml)."""
+    filepath = Path(filepath)
+    ext = filepath.suffix.lower()
+    text = filepath.read_text(encoding='utf-8')
+    if ext in ['.yaml', '.yml']:
+        data = yaml.safe_load(text)
+        return data.get('instruction_template', '')
+    return text
+
+
 def convert_logprobs_to_tiktoken(model, logprobs):
     # more problems than it's worth.
     # try:
@@ -234,6 +249,8 @@ def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, p
         instruction_template = body['instruction_template']
         instruction_template = "Alpaca" if instruction_template == "None" else instruction_template
         instruction_template_str = load_instruction_template_memoized(instruction_template)
+    elif shared.args.chat_template_file:
+        instruction_template_str = load_chat_template_file(shared.args.chat_template_file)
     else:
         instruction_template_str = shared.settings['instruction_template_str']
 
diff --git a/modules/shared.py b/modules/shared.py
index de0820af4f..bc7ea8ba9b 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -215,6 +215,7 @@
 group.add_argument('--dry-sequence-breakers', type=str, default=_d['dry_sequence_breakers'], metavar='N', help='DRY sequence breakers')
 group.add_argument('--enable-thinking', action=argparse.BooleanOptionalAction, default=True, help='Enable thinking')
 group.add_argument('--reasoning-effort', type=str, default='medium', metavar='N', help='Reasoning effort')
+group.add_argument('--chat-template-file', type=str, default=None, help='Path to a chat template file (.jinja, .jinja2, or .yaml) to use as the default instruction template for API requests. Overrides the model\'s built-in template.')
 
 # Handle CMD_FLAGS.txt
 cmd_flags_path = user_data_dir / "CMD_FLAGS.txt"
@@ -376,6 +377,11 @@
 
 
 def do_cmd_flags_warnings():
+    # Validate --chat-template-file
+    if args.chat_template_file and not Path(args.chat_template_file).is_file():
+        logger.error(f"--chat-template-file: file not found: {args.chat_template_file}")
+        sys.exit(1)
+
     # Security warnings
     if args.trust_remote_code:
         logger.warning(
diff --git a/modules/training.py b/modules/training.py
index c9f32e6444..8753946154 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -226,7 +226,7 @@ def clean_path(base_path: str, path: str):
 def get_instruction_templates():
     path = shared.user_data_dir / 'instruction-templates'
     names = set()
-    for ext in ['yaml', 'yml', 'jinja']:
+    for ext in ['yaml', 'yml', 'jinja', 'jinja2']:
         for f in path.glob(f'*.{ext}'):
             names.add(f.stem)
     return ['None', 'Chat Template'] + sorted(names, key=utils.natural_keys)
@@ -235,10 +235,10 @@ def get_instruction_templates():
 def load_template(name):
     """Load a Jinja2 template string from {user_data_dir}/instruction-templates/."""
     path = shared.user_data_dir / 'instruction-templates'
-    for ext in ['jinja', 'yaml', 'yml']:
+    for ext in ['jinja', 'jinja2', 'yaml', 'yml']:
         filepath = path / f'{name}.{ext}'
         if filepath.exists():
-            if ext == 'jinja':
+            if ext in ['jinja', 'jinja2']:
                 return filepath.read_text(encoding='utf-8')
             else:
                 data = yaml.safe_load(filepath.read_text(encoding='utf-8'))

From 044566d42dc728303e2e5b6d943d87fbb03e8430 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 15:06:16 -0300
Subject: [PATCH 1265/1701] API: Add tool call parsing for DeepSeek, GLM,
 MiniMax, and Kimi models

---
 extensions/openai/utils.py | 243 ++++++++++++++++++++++++++++++++++---
 modules/chat.py            |  48 +++++++-
 2 files changed, 270 insertions(+), 21 deletions(-)

diff --git a/extensions/openai/utils.py b/extensions/openai/utils.py
index 64bd163104..f4a31d1a72 100644
--- a/extensions/openai/utils.py
+++ b/extensions/openai/utils.py
@@ -83,6 +83,39 @@ def checkAndSanitizeToolCallCandidate(candidate_dict: dict, tool_names: list[str
     return None
 
 
+def _extractBalancedJson(text: str, start: int) -> str | None:
+    """Extract a balanced JSON object from text starting at the given position.
+
+    Walks through the string tracking brace depth and string boundaries
+    to correctly handle arbitrary nesting levels.
+    """
+    if start >= len(text) or text[start] != '{':
+        return None
+    depth = 0
+    in_string = False
+    escape_next = False
+    for i in range(start, len(text)):
+        c = text[i]
+        if escape_next:
+            escape_next = False
+            continue
+        if c == '\\' and in_string:
+            escape_next = True
+            continue
+        if c == '"':
+            in_string = not in_string
+            continue
+        if in_string:
+            continue
+        if c == '{':
+            depth += 1
+        elif c == '}':
+            depth -= 1
+            if depth == 0:
+                return text[start:i + 1]
+    return None
+
+
 def _parseChannelToolCalls(answer: str, tool_names: list[str]):
     """Parse channel-based tool calls used by GPT-OSS and similar models.
 
@@ -91,14 +124,17 @@ def _parseChannelToolCalls(answer: str, tool_names: list[str]):
     """
     matches = []
     for m in re.finditer(
-        r'<\|channel\|>commentary to=functions\.([^<\s]+)\s*(?:<\|constrain\|>json)?<\|message\|>(\{[^{}]*(?:\{[^{}]*\}[^{}]*)*\})',
+        r'<\|channel\|>commentary to=functions\.([^<\s]+)\s*(?:<\|constrain\|>json)?<\|message\|>',
         answer
     ):
         func_name = m.group(1).strip()
         if func_name not in tool_names:
             continue
+        json_str = _extractBalancedJson(answer, m.end())
+        if json_str is None:
+            continue
         try:
-            arguments = json.loads(m.group(2))
+            arguments = json.loads(json_str)
             matches.append({
                 "type": "function",
                 "function": {
@@ -119,27 +155,33 @@ def _parseBareNameToolCalls(answer: str, tool_names: list[str]):
     Multiple calls are concatenated directly or separated by whitespace.
     """
     matches = []
-    # Build pattern that matches any known tool name followed by a JSON object
+    # Match tool name followed by opening brace, then extract balanced JSON
     escaped_names = [re.escape(name) for name in tool_names]
-    pattern = r'(?:' + '|'.join(escaped_names) + r')\s*\{[^{}]*(?:\{[^{}]*\}[^{}]*)*\}'
+    pattern = r'(?:' + '|'.join(escaped_names) + r')\s*\{'
     for match in re.finditer(pattern, answer):
         text = match.group(0)
-        # Split into function name and JSON arguments
-        for name in tool_names:
-            if text.startswith(name):
-                json_str = text[len(name):].strip()
-                try:
-                    arguments = json.loads(json_str)
-                    matches.append({
-                        "type": "function",
-                        "function": {
-                            "name": name,
-                            "arguments": arguments
-                        }
-                    })
-                except json.JSONDecodeError:
-                    pass
+        name = None
+        for n in tool_names:
+            if text.startswith(n):
+                name = n
                 break
+        if not name:
+            continue
+        brace_start = match.end() - 1
+        json_str = _extractBalancedJson(answer, brace_start)
+        if json_str is None:
+            continue
+        try:
+            arguments = json.loads(json_str)
+            matches.append({
+                "type": "function",
+                "function": {
+                    "name": name,
+                    "arguments": arguments
+                }
+            })
+        except json.JSONDecodeError:
+            pass
     return matches
 
 
@@ -181,6 +223,149 @@ def _parseXmlParamToolCalls(answer: str, tool_names: list[str]):
     return matches
 
 
+def _parseKimiToolCalls(answer: str, tool_names: list[str]):
+    """Parse Kimi-K2-style tool calls using pipe-delimited tokens.
+
+    Format:
+        <|tool_calls_section_begin|>
+        <|tool_call_begin|>functions.func_name:index<|tool_call_argument_begin|>{"arg": "value"}<|tool_call_end|>
+        <|tool_calls_section_end|>
+    """
+    matches = []
+    for m in re.finditer(
+        r'<\|tool_call_begin\|>\s*(?:functions\.)?(\S+?)(?::\d+)?\s*<\|tool_call_argument_begin\|>\s*',
+        answer
+    ):
+        func_name = m.group(1).strip()
+        if func_name not in tool_names:
+            continue
+        json_str = _extractBalancedJson(answer, m.end())
+        if json_str is None:
+            continue
+        try:
+            arguments = json.loads(json_str)
+            matches.append({
+                "type": "function",
+                "function": {
+                    "name": func_name,
+                    "arguments": arguments
+                }
+            })
+        except json.JSONDecodeError:
+            pass
+    return matches
+
+
+def _parseMiniMaxToolCalls(answer: str, tool_names: list[str]):
+    """Parse MiniMax-style tool calls using invoke/parameter XML tags.
+
+    Format:
+        
+        
+        value
+        
+        
+    """
+    matches = []
+    for tc_match in re.finditer(r'\s*(.*?)\s*', answer, re.DOTALL):
+        tc_content = tc_match.group(1)
+        # Split on  to handle multiple parallel calls in one block
+        for invoke_match in re.finditer(r'(.*?)', tc_content, re.DOTALL):
+            func_name = invoke_match.group(1).strip()
+            if func_name not in tool_names:
+                continue
+            invoke_body = invoke_match.group(2)
+            arguments = {}
+            for param_match in re.finditer(r'\s*(.*?)\s*', invoke_body, re.DOTALL):
+                param_name = param_match.group(1).strip()
+                param_value = param_match.group(2).strip()
+                try:
+                    param_value = json.loads(param_value)
+                except (json.JSONDecodeError, ValueError):
+                    pass  # keep as string
+                arguments[param_name] = param_value
+            matches.append({
+                "type": "function",
+                "function": {
+                    "name": func_name,
+                    "arguments": arguments
+                }
+            })
+    return matches
+
+
+def _parseDeepSeekToolCalls(answer: str, tool_names: list[str]):
+    """Parse DeepSeek-style tool calls using fullwidth Unicode token delimiters.
+
+    Format:
+        <｜tool▁calls▁begin｜><｜tool▁call▁begin｜>func_name<｜tool▁sep｜>{"arg": "value"}<｜tool▁call▁end｜><｜tool▁calls▁end｜>
+    """
+    matches = []
+    for m in re.finditer(
+        r'<｜tool▁call▁begin｜>\s*(\S+?)\s*<｜tool▁sep｜>\s*',
+        answer
+    ):
+        func_name = m.group(1).strip()
+        if func_name not in tool_names:
+            continue
+        json_str = _extractBalancedJson(answer, m.end())
+        if json_str is None:
+            continue
+        try:
+            arguments = json.loads(json_str)
+            matches.append({
+                "type": "function",
+                "function": {
+                    "name": func_name,
+                    "arguments": arguments
+                }
+            })
+        except json.JSONDecodeError:
+            pass
+    return matches
+
+
+def _parseGlmToolCalls(answer: str, tool_names: list[str]):
+    """Parse GLM-style tool calls using arg_key/arg_value XML pairs.
+
+    Format:
+        function_name
+        key1
+        value1
+        
+    """
+    matches = []
+    for tc_match in re.finditer(r'\s*(.*?)\s*', answer, re.DOTALL):
+        tc_content = tc_match.group(1)
+        # First non-tag text is the function name
+        name_match = re.match(r'([^<\s]+)', tc_content.strip())
+        if not name_match:
+            continue
+        func_name = name_match.group(1).strip()
+        if func_name not in tool_names:
+            continue
+        # Extract arg_key/arg_value pairs
+        keys = [k.group(1).strip() for k in re.finditer(r'\s*(.*?)\s*', tc_content, re.DOTALL)]
+        vals = [v.group(1).strip() for v in re.finditer(r'\s*(.*?)\s*', tc_content, re.DOTALL)]
+        if len(keys) != len(vals):
+            continue
+        arguments = {}
+        for k, v in zip(keys, vals):
+            try:
+                v = json.loads(v)
+            except (json.JSONDecodeError, ValueError):
+                pass  # keep as string
+            arguments[k] = v
+        matches.append({
+            "type": "function",
+            "function": {
+                "name": func_name,
+                "arguments": arguments
+            }
+        })
+    return matches
+
+
 def _parsePythonicToolCalls(answer: str, tool_names: list[str]):
     """Parse pythonic-style tool calls used by Llama 4 and similar models.
 
@@ -244,11 +429,31 @@ def parseToolCall(answer: str, tool_names: list[str]):
     if len(answer) < 10:
         return matches
 
+    # Check for DeepSeek-style tool calls (fullwidth Unicode token delimiters)
+    matches = _parseDeepSeekToolCalls(answer, tool_names)
+    if matches:
+        return matches
+
+    # Check for Kimi-K2-style tool calls (pipe-delimited tokens)
+    matches = _parseKimiToolCalls(answer, tool_names)
+    if matches:
+        return matches
+
     # Check for channel-based tool calls (e.g. GPT-OSS format)
     matches = _parseChannelToolCalls(answer, tool_names)
     if matches:
         return matches
 
+    # Check for MiniMax-style tool calls (invoke/parameter XML tags)
+    matches = _parseMiniMaxToolCalls(answer, tool_names)
+    if matches:
+        return matches
+
+    # Check for GLM-style tool calls (arg_key/arg_value XML pairs)
+    matches = _parseGlmToolCalls(answer, tool_names)
+    if matches:
+        return matches
+
     # Check for XML-parameter style tool calls (e.g. Qwen3.5 format)
     matches = _parseXmlParamToolCalls(answer, tool_names)
     if matches:
diff --git a/modules/chat.py b/modules/chat.py
index bc4fc1d878..36d373d676 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -106,6 +106,50 @@ def str_presenter(dumper, data):
 yaml.representer.SafeRepresenter.add_representer(str, str_presenter)
 
 
+class _JsonDict(dict):
+    """A dict that serializes as JSON when used in string concatenation.
+
+    Some Jinja2 templates (Qwen, GLM) iterate arguments with .items(),
+    requiring a dict.  Others (DeepSeek) concatenate arguments as a
+    string, requiring JSON.  This class satisfies both.
+    """
+
+    def __str__(self):
+        return json.dumps(self, ensure_ascii=False)
+
+    def __add__(self, other):
+        return str(self) + other
+
+    def __radd__(self, other):
+        return other + str(self)
+
+
+def _deserialize_tool_call_arguments(tool_calls):
+    """Convert tool_call arguments from JSON strings to _JsonDict.
+
+    The OpenAI API spec sends arguments as a JSON string, but Jinja2
+    templates may need a dict (.items()) or a string (concatenation).
+    _JsonDict handles both transparently.
+    """
+    result = []
+    for tc in tool_calls:
+        tc = copy.copy(tc)
+        func = tc.get('function', {})
+        if isinstance(func, dict):
+            func = dict(func)
+            args = func.get('arguments')
+            if isinstance(args, str):
+                try:
+                    func['arguments'] = _JsonDict(json.loads(args))
+                except (json.JSONDecodeError, ValueError):
+                    pass
+            elif isinstance(args, dict) and not isinstance(args, _JsonDict):
+                func['arguments'] = _JsonDict(args)
+            tc['function'] = func
+        result.append(tc)
+    return result
+
+
 def generate_chat_prompt(user_input, state, **kwargs):
     impersonate = kwargs.get('impersonate', False)
     _continue = kwargs.get('_continue', False)
@@ -172,7 +216,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
         if not assistant_msg and entry_meta.get('tool_calls'):
             # Assistant message with only tool_calls and no text content
-            messages.insert(insert_pos, {"role": "assistant", "content": "", "tool_calls": entry_meta['tool_calls']})
+            messages.insert(insert_pos, {"role": "assistant", "content": "", "tool_calls": _deserialize_tool_call_arguments(entry_meta['tool_calls'])})
         elif assistant_msg:
             # Handle GPT-OSS as a special case
             if '<|channel|>analysis<|message|>' in assistant_msg or '<|channel|>final<|message|>' in assistant_msg:
@@ -250,7 +294,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
             # Attach tool_calls metadata to the assistant message if present
             if entry_meta.get('tool_calls') and messages[insert_pos].get('role') == 'assistant':
-                messages[insert_pos]['tool_calls'] = entry_meta['tool_calls']
+                messages[insert_pos]['tool_calls'] = _deserialize_tool_call_arguments(entry_meta['tool_calls'])
 
         if user_msg not in ['', '<|BEGIN-VISIBLE-CHAT|>']:
             # Check for user message attachments in metadata

From d03923924a853ffe3662f498fd4f4e5aedab8fb3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 16:52:02 -0300
Subject: [PATCH 1266/1701] Several small fixes

- Stop llama-server subprocess on model unload instead of relying on GC
- Fix tool_calls[].index being string instead of int in API responses
- Omit tool_calls key from API response when empty per OpenAI spec
- Prevent division by zero when micro_batch_size > batch_size in training
- Copy sampler_priority list before mutating in ExLlamaV3
- Normalize presence/frequency_penalty names for ExLlamaV3 sampler sorting
- Restore original chat_template after training instead of leaving it mutated
---
 extensions/openai/completions.py | 4 ++--
 modules/exllamav3.py             | 7 ++++++-
 modules/models.py                | 2 ++
 modules/training.py              | 7 ++++++-
 4 files changed, 16 insertions(+), 4 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index cabfce9916..8ba031c110 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -343,7 +343,7 @@ def chat_streaming_chunk(content, chunk_tool_calls=None):
             if len(tool_call) > 0:
                 for tc in tool_call:
                     tc["id"] = getToolCallId()
-                    tc["index"] = str(len(tool_calls))
+                    tc["index"] = len(tool_calls)
                     tc["function"]["arguments"] = json.dumps(tc["function"]["arguments"])
                     tool_calls.append(tc)
                 end_last_tool_call = len(answer)
@@ -391,7 +391,7 @@ def chat_streaming_chunk(content, chunk_tool_calls=None):
             resp_list: [{
                 "index": 0,
                 "finish_reason": stop_reason,
-                "message": {"role": "assistant", "content": answer, "tool_calls": tool_calls},
+                "message": {"role": "assistant", "content": answer, **({"tool_calls": tool_calls} if tool_calls else {})},
             }],
             "usage": {
                 "prompt_tokens": token_count,
diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index af5745bc3a..df37ddbec0 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -339,11 +339,16 @@ def generate_with_streaming(self, prompt, state):
 
             # 3. Get the priority list and handle temperature_last
             default_priority = ['repetition_penalty', 'presence_frequency_penalty', 'top_k', 'top_p', 'min_p', 'temperature']
-            sampler_priority = state.get('sampler_priority') or default_priority
+            sampler_priority = list(state.get('sampler_priority') or default_priority)
 
             if state['temperature_last'] and 'temperature' in sampler_priority:
                 sampler_priority.append(sampler_priority.pop(sampler_priority.index('temperature')))
 
+            # The preset system uses separate 'presence_penalty' and
+            # 'frequency_penalty', but ExLlamaV3 has a single combined
+            # SS_PresFreqP sampler. Normalize to the combined name.
+            sampler_priority = ['presence_frequency_penalty' if x in ('presence_penalty', 'frequency_penalty') else x for x in sampler_priority]
+
             # 4. Sort the unordered list based on the priority list
             def custom_sort_key(sampler_obj):
                 class_name = sampler_obj.__class__.__name__
diff --git a/modules/models.py b/modules/models.py
index bc5585cfc9..48d68b0bd2 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -126,6 +126,8 @@ def unload_model(keep_model_name=False):
 
     if model_class_name in ['Exllamav3Model', 'Exllamav3HF', 'TensorRTLLMModel']:
         shared.model.unload()
+    elif model_class_name == 'LlamaServer':
+        shared.model.stop()
 
     shared.model = shared.tokenizer = None
     shared.lora_names = []
diff --git a/modules/training.py b/modules/training.py
index 8753946154..2e172d2266 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -333,7 +333,8 @@ def do_train(lora_name: str, always_override: bool, all_linear: bool, q_proj_en:
         yield "Cannot input zeroes."
         return
 
-    gradient_accumulation_steps = batch_size // micro_batch_size
+    gradient_accumulation_steps = max(1, batch_size // micro_batch_size)
+    original_chat_template = getattr(shared.tokenizer, 'chat_template', None)
     if shared.tokenizer.pad_token_id is None:
         shared.tokenizer.pad_token_id = shared.tokenizer.eos_token_id
     shared.tokenizer.padding_side = "right"
@@ -820,6 +821,10 @@ def threaded_run():
         logger.info("Training complete, saving")
         lora_model.save_pretrained(lora_file_path)
 
+    # Restore the original chat_template if we changed it for training
+    if shared.tokenizer is not None and hasattr(shared.tokenizer, 'chat_template'):
+        shared.tokenizer.chat_template = original_chat_template
+
     if WANT_INTERRUPT:
         logger.info("Training interrupted.")
         yield f"Interrupted. Incomplete LoRA saved to `{lora_file_path}`."

From b8b4471ab5267779fbad80bcfec88c3c6052eea1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 16:57:57 -0300
Subject: [PATCH 1267/1701] Security: restrict file writes to user_data_dir,
 block extra_flags from API

---
 extensions/openai/models.py |  5 +++--
 modules/utils.py            | 11 ++---------
 2 files changed, 5 insertions(+), 11 deletions(-)

diff --git a/extensions/openai/models.py b/extensions/openai/models.py
index 74f9dbee65..d6ef119db8 100644
--- a/extensions/openai/models.py
+++ b/extensions/openai/models.py
@@ -48,8 +48,9 @@ def _load_model(data):
     # Update shared.args with custom model loading settings
     # Security: only allow keys that correspond to model loading
     # parameters exposed in the UI. Never allow security-sensitive
-    # flags like trust_remote_code to be set via the API.
-    allowed_keys = set(ui.list_model_elements())
+    # flags like trust_remote_code or extra_flags to be set via the API.
+    blocked_keys = {'extra_flags'}
+    allowed_keys = set(ui.list_model_elements()) - blocked_keys
     if args:
         for k in args:
             if k in allowed_keys and hasattr(shared.args, k):
diff --git a/modules/utils.py b/modules/utils.py
index 7ab4a55476..203d2d9c9a 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -27,21 +27,14 @@ def sanitize_filename(name):
 
 
 def _is_path_allowed(abs_path_str):
-    """Check if a path is under the project root or the configured user_data directory."""
+    """Check if a path is under the configured user_data directory."""
     abs_path = Path(abs_path_str).resolve()
-    root_folder = Path(__file__).resolve().parent.parent
     user_data_resolved = shared.user_data_dir.resolve()
-    try:
-        abs_path.relative_to(root_folder)
-        return True
-    except ValueError:
-        pass
     try:
         abs_path.relative_to(user_data_resolved)
         return True
     except ValueError:
-        pass
-    return False
+        return False
 
 
 def save_file(fname, contents):

From 5f6754c2673c2511c7b85793a80236a6c4004104 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 17:07:52 -0300
Subject: [PATCH 1268/1701] Fix stop button being ignored when token throttling
 is off

---
 modules/text_generation.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index 02c1320c45..c78afe3e80 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -113,7 +113,7 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
                 last_update = time.monotonic()
 
         stop_event = state.get('stop_event')
-        if stop_found or (state['max_tokens_second'] > 0 and shared.stop_everything) or (stop_event and stop_event.is_set()):
+        if stop_found or shared.stop_everything or (stop_event and stop_event.is_set()):
             break
 
     if not is_chat:

From 2beaa4b97173fcba4affa2d66074eb53113b0fd8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 14:39:35 -0800
Subject: [PATCH 1269/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 12 files changed, 20 insertions(+), 20 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 85fbd5df73..45d92132d7 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.23/exllamav3-0.0.23+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.23/exllamav3-0.0.23+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index de41f6063a..415c54c6e9 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 9f0a113a5a..7fe16ea341 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index e7c95a7154..d8d4030d68 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 28112860ed..90f834bf07 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 9ff8a55869..dd8fa191b8 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index b79defafa5..f34ed929c4 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 19310f9ca5..b9b5898550 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 3a269054db..26a6a48b23 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index b5d319ac36..10612c20ef 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 1e05de220a..f713db8618 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 6298be11a5..4f37acc105 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.85.0/llama_cpp_binaries-0.85.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From d48b53422f98c14b7243c588de074ab739443e70 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 15:39:08 -0800
Subject: [PATCH 1270/1701] Training: Optimize _peek_json_keys to avoid loading
 entire file into memory

---
 modules/utils.py | 33 ++++++++++++++++++++++++++++++---
 1 file changed, 30 insertions(+), 3 deletions(-)

diff --git a/modules/utils.py b/modules/utils.py
index 203d2d9c9a..a14f8b8f41 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -298,11 +298,38 @@ def get_text_datasets(path: str):
 def _peek_json_keys(filepath):
     """Read the first object in a JSON array file and return its keys."""
     import json
+    decoder = json.JSONDecoder()
+    WS = ' \t\n\r'
     try:
         with open(filepath, 'r', encoding='utf-8') as f:
-            data = json.load(f)
-        if isinstance(data, list) and len(data) > 0 and isinstance(data[0], dict):
-            return set(data[0].keys())
+            buf = ''
+            obj_start = None
+            while len(buf) < 1 << 20:  # Read up to 1MB
+                chunk = f.read(8192)
+                if not chunk:
+                    break
+                buf += chunk
+                if obj_start is None:
+                    idx = 0
+                    while idx < len(buf) and buf[idx] in WS:
+                        idx += 1
+                    if idx >= len(buf):
+                        continue
+                    if buf[idx] != '[':
+                        return set()
+                    idx += 1
+                    while idx < len(buf) and buf[idx] in WS:
+                        idx += 1
+                    if idx >= len(buf):
+                        continue
+                    obj_start = idx
+                try:
+                    obj, _ = decoder.raw_decode(buf, obj_start)
+                    if isinstance(obj, dict):
+                        return set(obj.keys())
+                    return set()
+                except json.JSONDecodeError:
+                    continue
     except Exception:
         pass
     return set()

From 1eead661c35c4d15dad84b9ab4b4118964492ae0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 18:04:48 -0800
Subject: [PATCH 1271/1701] Portable mode: always use ../user_data if it exists

---
 modules/paths.py | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/modules/paths.py b/modules/paths.py
index 36fdc4f2f3..2b9a1fcc3f 100644
--- a/modules/paths.py
+++ b/modules/paths.py
@@ -6,7 +6,7 @@ def resolve_user_data_dir():
     """
     Resolve the user_data directory path. Order of precedence:
     1. --user-data-dir CLI flag (pre-parsed from sys.argv before argparse)
-    2. Auto-detect: if ./user_data doesn't exist but ../user_data does, use ../user_data
+    2. In --portable mode, prefer ../user_data if it exists
     3. Default: 'user_data'
     """
     script_dir = Path(__file__).resolve().parent.parent
@@ -18,11 +18,11 @@ def resolve_user_data_dir():
         elif arg.startswith('--user-data-dir='):
             return Path(arg.split('=', 1)[1])
 
-    # Auto-detect: check if user_data exists locally vs one folder up
-    local_path = script_dir / 'user_data'
-    parent_path = script_dir.parent / 'user_data'
-
-    if not local_path.exists() and parent_path.exists():
-        return parent_path
+    # In portable mode, prefer ../user_data if it exists
+    is_portable = '--portable' in sys.argv
+    if is_portable:
+        parent_path = script_dir.parent / 'user_data'
+        if parent_path.exists():
+            return parent_path
 
     return Path('user_data')

From e8e0d024062953d93699c4c26e691e7351ab1c3a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 18:15:05 -0800
Subject: [PATCH 1272/1701] Remove outdated ROCm environment variable overrides
 from one_click.py

---
 one_click.py | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/one_click.py b/one_click.py
index 67bd3a9639..efb0713469 100644
--- a/one_click.py
+++ b/one_click.py
@@ -10,11 +10,6 @@
 import subprocess
 import sys
 
-# Remove the '# ' from the following lines as needed for your AMD GPU on Linux
-# os.environ["ROCM_PATH"] = '/opt/rocm'
-# os.environ["HSA_OVERRIDE_GFX_VERSION"] = '10.3.0'
-# os.environ["HCC_AMDGPU_TARGET"] = 'gfx1030'
-
 # Define the required versions
 TORCH_VERSION = "2.9.1"
 PYTHON_VERSION = "3.13"

From 5fa709a3f4cd93cbec176b78fc2a3abf609dd757 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 18:37:40 -0800
Subject: [PATCH 1273/1701] llama.cpp server: use port+5 offset and suppress No
 parser definition detected logs

---
 modules/llama_cpp_server.py | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 3249497831..12ff173e3b 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -320,8 +320,8 @@ def _is_port_available(self, port):
                 return False
 
     def _find_available_port(self):
-        """Find an available port, preferring main port + 1."""
-        preferred_port = shared.args.api_port + 1
+        """Find an available port, preferring main port + 5."""
+        preferred_port = shared.args.api_port + 5
         if self._is_port_available(preferred_port):
             return preferred_port
 
@@ -510,6 +510,7 @@ def stop(self):
                 self.process.wait(timeout=5)
             except subprocess.TimeoutExpired:
                 self.process.kill()
+                self.process.wait(timeout=5)
 
             self.process = None
 
@@ -560,7 +561,7 @@ def filter_stderr_with_progress(process_stderr):
                             last_was_progress = (progress < 1.0)
 
                         # skip noise lines
-                        elif not (line.startswith(('srv ', 'slot ')) or 'log_server_r: request: GET /health' in line):
+                        elif not (line.startswith(('srv ', 'slot ')) or 'log_server_r: request: GET /health' in line or 'No parser definition detected' in line):
                             # if we were in progress, finish that line first
                             if last_was_progress:
                                 print(file=sys.stderr)

From 7ea55132634088e1629ed31a8be7debd0a705141 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 19:01:28 -0800
Subject: [PATCH 1274/1701] Handle Qwen 3.5 thinking blocks

---
 modules/html_generator.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 34a5bc5762..472a9ea0fc 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -115,6 +115,7 @@ def replace_blockquote(m):
     ('<|channel|>analysis<|message|>', '<|end|>', '<|start|>assistant<|channel|>final<|message|>'),
     ('', '', None),
     ('<|think|>', '<|end|>', '<|content|>'),  # Solar Open
+    ('Thinking Process:', '', None),  # Qwen3.5 verbose thinking outside tags
     (None, '', None),  # End-only variant (e.g., Qwen3-next)
 ]
 

From f2fe001cc4e95c602cdabd5f35200f3574c5f098 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 19:01:38 -0800
Subject: [PATCH 1275/1701] Fix message copy buttons not working over HTTP

---
 js/global_scope_js.js | 27 ++++++++++++++++++++++++++-
 1 file changed, 26 insertions(+), 1 deletion(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 205d937535..62b31d3780 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -11,7 +11,11 @@ function copyToClipboard(element) {
   const rawText = messageElement.getAttribute("data-raw");
   if (!rawText) return;
 
-  navigator.clipboard.writeText(rawText).then(function() {
+  const copyPromise = navigator.clipboard && window.isSecureContext
+    ? navigator.clipboard.writeText(rawText)
+    : fallbackCopyToClipboard(rawText);
+
+  copyPromise.then(function() {
     const originalSvg = element.innerHTML;
     element.innerHTML = "";
     setTimeout(() => {
@@ -22,6 +26,27 @@ function copyToClipboard(element) {
   });
 }
 
+function fallbackCopyToClipboard(text) {
+  return new Promise((resolve, reject) => {
+    const textArea = document.createElement("textarea");
+    textArea.value = text;
+    textArea.style.position = "fixed";
+    textArea.style.left = "-9999px";
+    textArea.style.top = "-9999px";
+    document.body.appendChild(textArea);
+    textArea.focus();
+    textArea.select();
+    try {
+      const successful = document.execCommand("copy");
+      document.body.removeChild(textArea);
+      successful ? resolve() : reject();
+    } catch (err) {
+      document.body.removeChild(textArea);
+      reject(err);
+    }
+  });
+}
+
 function branchHere(element) {
   if (!element) return;
 

From abc699db9b3c370c6be977d59a23a0e02987dd79 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 19:03:38 -0800
Subject: [PATCH 1276/1701] Minor UI change

---
 modules/models_settings.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 104635b901..472871ce09 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -415,7 +415,7 @@ def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type):
     an HTML string for the UI display.
     """
     if loader != 'llama.cpp' or model in ["None", None] or not model.endswith(".gguf") or gpu_layers < 0 or ctx_size == 0:
-        return "
            Estimated VRAM to load the model:
            "
+        return f"
            Estimated VRAM to load the model: auto
            "
 
     vram_usage = estimate_vram(model, gpu_layers, ctx_size, cache_type)
     return f"
            Estimated VRAM to load the model: {vram_usage:.0f} MiB
            "

From aa634c77c083a8423a6bbb95f14779b642d34745 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 6 Mar 2026 21:00:36 -0800
Subject: [PATCH 1277/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 12 files changed, 20 insertions(+), 20 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 45d92132d7..eaf34fa8da 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.23/exllamav3-0.0.23+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.23/exllamav3-0.0.23+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 415c54c6e9..3211f25198 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 7fe16ea341..8d45211401 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index d8d4030d68..525ceed50a 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 90f834bf07..86b65a9726 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index dd8fa191b8..61c9ef73df 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index f34ed929c4..3d0785a390 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index b9b5898550..6805e20912 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 26a6a48b23..5a8ed87b0c 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 10612c20ef..fafa23cf57 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index f713db8618..3ef59f971a 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 4f37acc105..6039357dfa 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.86.0/llama_cpp_binaries-0.86.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 085c4ef5d75b760d0d0dfdf79188aa12e0048046 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Mar 2026 05:09:49 -0800
Subject: [PATCH 1278/1701] Revert "Update Miniforge from 25.3.0 to 26.1.0"

This reverts commit 9576c5a5f4db2715c1440e2b630e46cc0433167f.
---
 start_linux.sh    | 2 +-
 start_macos.sh    | 2 +-
 start_windows.bat | 4 ++--
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/start_linux.sh b/start_linux.sh
index 2aa07d45f9..81bda49062 100755
--- a/start_linux.sh
+++ b/start_linux.sh
@@ -30,7 +30,7 @@ esac
 INSTALL_DIR="$(pwd)/installer_files"
 CONDA_ROOT_PREFIX="$(pwd)/installer_files/conda"
 INSTALL_ENV_DIR="$(pwd)/installer_files/env"
-MINIFORGE_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/26.1.0-0/Miniforge3-26.1.0-0-Linux-${OS_ARCH}.sh"
+MINIFORGE_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-Linux-${OS_ARCH}.sh"
 conda_exists="F"
 
 # figure out whether git and conda needs to be installed
diff --git a/start_macos.sh b/start_macos.sh
index de257608be..56a7afd04f 100755
--- a/start_macos.sh
+++ b/start_macos.sh
@@ -30,7 +30,7 @@ esac
 INSTALL_DIR="$(pwd)/installer_files"
 CONDA_ROOT_PREFIX="$(pwd)/installer_files/conda"
 INSTALL_ENV_DIR="$(pwd)/installer_files/env"
-MINIFORGE_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/26.1.0-0/Miniforge3-26.1.0-0-MacOSX-${OS_ARCH}.sh"
+MINIFORGE_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-MacOSX-${OS_ARCH}.sh"
 conda_exists="F"
 
 # figure out whether git and conda needs to be installed
diff --git a/start_windows.bat b/start_windows.bat
index dd096760e4..2bac295f20 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -36,8 +36,8 @@ set TEMP=%cd%\installer_files
 set INSTALL_DIR=%cd%\installer_files
 set CONDA_ROOT_PREFIX=%cd%\installer_files\conda
 set INSTALL_ENV_DIR=%cd%\installer_files\env
-set MINIFORGE_DOWNLOAD_URL=https://github.com/conda-forge/miniforge/releases/download/26.1.0-0/Miniforge3-26.1.0-0-Windows-x86_64.exe
-set MINIFORGE_CHECKSUM=0ad64473c20a8649be9313f64ee898f4b23a35a7a25ea9998a751c542e5e3840
+set MINIFORGE_DOWNLOAD_URL=https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-Windows-x86_64.exe
+set MINIFORGE_CHECKSUM=b48cd98430170983076dfb51769a6d37668176f59bf3b59c4b21ac4c9bc24f39
 set conda_exists=F
 
 @rem figure out whether git and conda needs to be installed

From 304510eb3d86a8fa4102fde3c724c0da03d68a28 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Mar 2026 05:54:14 -0800
Subject: [PATCH 1279/1701] ExLlamav3: route all generation through
 ConcurrentGenerator

---
 modules/exllamav3.py | 57 ++++++++++++++------------------------------
 1 file changed, 18 insertions(+), 39 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index df37ddbec0..b4b76e2109 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -50,7 +50,7 @@ def _iterate_loop(self):
         while self.active:
             self.has_jobs.wait(timeout=0.5)
             with self.lock:
-                if self.generator.num_remaining_jobs() == 0:
+                if not self.job_queues:
                     self.has_jobs.clear()
                     continue
                 results = self.generator.iterate()
@@ -406,44 +406,23 @@ def custom_sort_key(sampler_obj):
         response_text = ""
         stop_event = state.get('stop_event')
 
-        if stop_event:
-            # Concurrent path for API requests
-            result_queue = self.parallel_generator.submit(job)
-            try:
-                while True:
-                    if stop_event.is_set() or shared.stop_everything:
-                        break
-                    try:
-                        result = result_queue.get(timeout=0.1)
-                    except queue.Empty:
-                        continue
-                    if result is None or result.get("eos"):
-                        break
-                    chunk = result.get("text", "")
-                    if chunk:
-                        response_text += chunk
-                        yield response_text
-            finally:
-                self.parallel_generator.cancel(job)
-        else:
-            # Original single-request path (WebUI)
-            self.generator.enqueue(job)
-            try:
-                while self.generator.num_remaining_jobs():
-                    if shared.stop_everything:
-                        break
-
-                    results = self.generator.iterate()
-                    for result in results:
-                        if "eos" in result and result["eos"]:
-                            break
-
-                        chunk = result.get("text", "")
-                        if chunk:
-                            response_text += chunk
-                            yield response_text
-            finally:
-                self.generator.clear_queue()
+        result_queue = self.parallel_generator.submit(job)
+        try:
+            while True:
+                if shared.stop_everything or (stop_event and stop_event.is_set()):
+                    break
+                try:
+                    result = result_queue.get(timeout=0.1)
+                except queue.Empty:
+                    continue
+                if result is None or result.get("eos"):
+                    break
+                chunk = result.get("text", "")
+                if chunk:
+                    response_text += chunk
+                    yield response_text
+        finally:
+            self.parallel_generator.cancel(job)
 
     def generate(self, prompt, state):
         output = ""

From 328215b0c74d249d153e7e684c0241a4c89eee6e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Mar 2026 06:06:13 -0800
Subject: [PATCH 1280/1701] API: Stop generation on client disconnect for
 non-streaming requests

---
 extensions/openai/script.py | 43 ++++++++++++++++++++++++++-----------
 1 file changed, 31 insertions(+), 12 deletions(-)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index bfb6fd54c4..7a13638d52 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -59,6 +59,15 @@
 }
 
 
+async def _wait_for_disconnect(request: Request, stop_event: threading.Event):
+    """Block until the client disconnects, then signal the stop_event."""
+    while True:
+        message = await request.receive()
+        if message["type"] == "http.disconnect":
+            stop_event.set()
+            return
+
+
 def verify_api_key(authorization: str = Header(None)) -> None:
     expected_api_key = shared.args.api_key
     if expected_api_key and (authorization is None or authorization != f"Bearer {expected_api_key}"):
@@ -129,12 +138,17 @@ async def generator():
 
     else:
         stop_event = threading.Event()
-        response = await asyncio.to_thread(
-            OAIcompletions.completions,
-            to_dict(request_data),
-            is_legacy=is_legacy,
-            stop_event=stop_event
-        )
+        monitor = asyncio.create_task(_wait_for_disconnect(request, stop_event))
+        try:
+            response = await asyncio.to_thread(
+                OAIcompletions.completions,
+                to_dict(request_data),
+                is_legacy=is_legacy,
+                stop_event=stop_event
+            )
+        finally:
+            stop_event.set()
+            monitor.cancel()
 
         return JSONResponse(response)
 
@@ -164,12 +178,17 @@ async def generator():
 
     else:
         stop_event = threading.Event()
-        response = await asyncio.to_thread(
-            OAIcompletions.chat_completions,
-            to_dict(request_data),
-            is_legacy=is_legacy,
-            stop_event=stop_event
-        )
+        monitor = asyncio.create_task(_wait_for_disconnect(request, stop_event))
+        try:
+            response = await asyncio.to_thread(
+                OAIcompletions.chat_completions,
+                to_dict(request_data),
+                is_legacy=is_legacy,
+                stop_event=stop_event
+            )
+        finally:
+            stop_event.set()
+            monitor.cancel()
 
         return JSONResponse(response)
 

From b686193fe2c4f6d4dd862153c2dfd3ba47ad690a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Mar 2026 06:10:05 -0800
Subject: [PATCH 1281/1701] Reapply "Update Miniforge from 25.3.0 to 26.1.0"

This reverts commit 085c4ef5d75b760d0d0dfdf79188aa12e0048046.
---
 start_linux.sh    | 2 +-
 start_macos.sh    | 2 +-
 start_windows.bat | 4 ++--
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/start_linux.sh b/start_linux.sh
index 81bda49062..2aa07d45f9 100755
--- a/start_linux.sh
+++ b/start_linux.sh
@@ -30,7 +30,7 @@ esac
 INSTALL_DIR="$(pwd)/installer_files"
 CONDA_ROOT_PREFIX="$(pwd)/installer_files/conda"
 INSTALL_ENV_DIR="$(pwd)/installer_files/env"
-MINIFORGE_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-Linux-${OS_ARCH}.sh"
+MINIFORGE_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/26.1.0-0/Miniforge3-26.1.0-0-Linux-${OS_ARCH}.sh"
 conda_exists="F"
 
 # figure out whether git and conda needs to be installed
diff --git a/start_macos.sh b/start_macos.sh
index 56a7afd04f..de257608be 100755
--- a/start_macos.sh
+++ b/start_macos.sh
@@ -30,7 +30,7 @@ esac
 INSTALL_DIR="$(pwd)/installer_files"
 CONDA_ROOT_PREFIX="$(pwd)/installer_files/conda"
 INSTALL_ENV_DIR="$(pwd)/installer_files/env"
-MINIFORGE_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-MacOSX-${OS_ARCH}.sh"
+MINIFORGE_DOWNLOAD_URL="https://github.com/conda-forge/miniforge/releases/download/26.1.0-0/Miniforge3-26.1.0-0-MacOSX-${OS_ARCH}.sh"
 conda_exists="F"
 
 # figure out whether git and conda needs to be installed
diff --git a/start_windows.bat b/start_windows.bat
index 2bac295f20..dd096760e4 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -36,8 +36,8 @@ set TEMP=%cd%\installer_files
 set INSTALL_DIR=%cd%\installer_files
 set CONDA_ROOT_PREFIX=%cd%\installer_files\conda
 set INSTALL_ENV_DIR=%cd%\installer_files\env
-set MINIFORGE_DOWNLOAD_URL=https://github.com/conda-forge/miniforge/releases/download/25.3.0-3/Miniforge3-25.3.0-3-Windows-x86_64.exe
-set MINIFORGE_CHECKSUM=b48cd98430170983076dfb51769a6d37668176f59bf3b59c4b21ac4c9bc24f39
+set MINIFORGE_DOWNLOAD_URL=https://github.com/conda-forge/miniforge/releases/download/26.1.0-0/Miniforge3-26.1.0-0-Windows-x86_64.exe
+set MINIFORGE_CHECKSUM=0ad64473c20a8649be9313f64ee898f4b23a35a7a25ea9998a751c542e5e3840
 set conda_exists=F
 
 @rem figure out whether git and conda needs to be installed

From e1bf0b866f811bed095e30df78a3149fd92413f7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Mar 2026 06:46:46 -0800
Subject: [PATCH 1282/1701] Update the macos workflow

---
 .github/workflows/build-everything-tgw.yml   | 2 +-
 .github/workflows/build-portable-release.yml | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/.github/workflows/build-everything-tgw.yml b/.github/workflows/build-everything-tgw.yml
index 78fd2d6bc1..9322f85929 100644
--- a/.github/workflows/build-everything-tgw.yml
+++ b/.github/workflows/build-everything-tgw.yml
@@ -67,4 +67,4 @@ jobs:
     uses: ./.github/workflows/build-portable-release.yml
     with:
       version: ${{ inputs.version }}
-      config: 'os:macos-13,macos-14'
+      config: 'os:macos-15-intel,macos-14'
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index a6aec7519e..cc584fb2a8 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -120,7 +120,7 @@ jobs:
                 PACKAGES_PATH="portable_env/Lib/site-packages"
                 rm start_linux.sh start_macos.sh
             elif [[ "$RUNNER_OS" == "macOS" ]]; then
-                if [[ "$OS_TYPE" == "macos-13" ]]; then
+                if [[ "$OS_TYPE" == "macos-15-intel" ]]; then
                     PLATFORM="macos-x86_64"
                     PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-apple-darwin-install_only.tar.gz"
                     REQ_TYPE="apple_intel"
@@ -153,7 +153,7 @@ jobs:
 
             # Select requirements file based on platform
             if [[ "$RUNNER_OS" == "macOS" ]]; then
-                if [[ "$OS_TYPE" == "macos-13" ]]; then
+                if [[ "$OS_TYPE" == "macos-15-intel" ]]; then
                     REQ_FILE="requirements/portable/requirements_apple_intel.txt"
                 else
                     REQ_FILE="requirements/portable/requirements_apple_silicon.txt"

From 0cecc0a041353be0d81f8ed969f03cca396351e1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Mar 2026 06:59:48 -0800
Subject: [PATCH 1283/1701] Use tar.gz for Linux/macOS portable builds to
 preserve symlinks

---
 .github/workflows/build-portable-release-cuda.yml | 15 ++++++++-------
 .github/workflows/build-portable-release-rocm.yml | 15 ++++++++-------
 .../workflows/build-portable-release-vulkan.yml   | 15 ++++++++-------
 .github/workflows/build-portable-release.yml      | 15 ++++++++-------
 4 files changed, 32 insertions(+), 28 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index 87ab7f9ee9..a575911297 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -150,15 +150,16 @@ jobs:
             # 5. Clean up
             rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore requirements/ one_click.py
 
-            # 6. Create ZIP file
+            # 6. Create archive
             cd ..
-            ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
-            echo "Creating archive: $ZIP_NAME"
-
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
+                ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
+                echo "Creating archive: $ARCHIVE_NAME"
+                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
             else
-                zip -r "$ZIP_NAME" "text-generation-webui-${VERSION_CLEAN}"
+                ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.tar.gz"
+                echo "Creating archive: $ARCHIVE_NAME"
+                tar czf "$ARCHIVE_NAME" "text-generation-webui-${VERSION_CLEAN}"
             fi
 
       - name: Upload files to a GitHub release
@@ -167,7 +168,7 @@ jobs:
         continue-on-error: true
         with:
           repo_token: ${{ secrets.GITHUB_TOKEN }}
-          file: ../textgen-portable-*.zip
+          file: ../textgen-portable-*
           tag: ${{ inputs.version }}
           file_glob: true
           make_latest: false
diff --git a/.github/workflows/build-portable-release-rocm.yml b/.github/workflows/build-portable-release-rocm.yml
index 5b43b2d317..6f9ea4ec80 100644
--- a/.github/workflows/build-portable-release-rocm.yml
+++ b/.github/workflows/build-portable-release-rocm.yml
@@ -145,15 +145,16 @@ jobs:
             # 5. Clean up
             rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore requirements/ one_click.py
 
-            # 6. Create ZIP file
+            # 6. Create archive
             cd ..
-            ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-rocm.zip"
-            echo "Creating archive: $ZIP_NAME"
-
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
+                ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-rocm.zip"
+                echo "Creating archive: $ARCHIVE_NAME"
+                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
             else
-                zip -r "$ZIP_NAME" "text-generation-webui-${VERSION_CLEAN}"
+                ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-rocm.tar.gz"
+                echo "Creating archive: $ARCHIVE_NAME"
+                tar czf "$ARCHIVE_NAME" "text-generation-webui-${VERSION_CLEAN}"
             fi
 
       - name: Upload files to a GitHub release
@@ -162,7 +163,7 @@ jobs:
         continue-on-error: true
         with:
           repo_token: ${{ secrets.GITHUB_TOKEN }}
-          file: ../textgen-portable-*.zip
+          file: ../textgen-portable-*
           tag: ${{ inputs.version }}
           file_glob: true
           make_latest: false
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index e8b75b5b62..b98b2e5e37 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -145,15 +145,16 @@ jobs:
             # 5. Clean up
             rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore requirements/ one_click.py
 
-            # 6. Create ZIP file
+            # 6. Create archive
             cd ..
-            ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.zip"
-            echo "Creating archive: $ZIP_NAME"
-
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
+                ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.zip"
+                echo "Creating archive: $ARCHIVE_NAME"
+                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
             else
-                zip -r "$ZIP_NAME" "text-generation-webui-${VERSION_CLEAN}"
+                ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.tar.gz"
+                echo "Creating archive: $ARCHIVE_NAME"
+                tar czf "$ARCHIVE_NAME" "text-generation-webui-${VERSION_CLEAN}"
             fi
 
       - name: Upload files to a GitHub release
@@ -162,7 +163,7 @@ jobs:
         continue-on-error: true
         with:
           repo_token: ${{ secrets.GITHUB_TOKEN }}
-          file: ../textgen-portable-*.zip
+          file: ../textgen-portable-*
           tag: ${{ inputs.version }}
           file_glob: true
           make_latest: false
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index cc584fb2a8..1bd4e1638f 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -171,15 +171,16 @@ jobs:
             # 5. Clean up
             rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore requirements/ one_click.py
 
-            # 6. Create ZIP file
+            # 6. Create archive
             cd ..
-            ZIP_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.zip"
-            echo "Creating archive: $ZIP_NAME"
-
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ZIP_NAME"
+                ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.zip"
+                echo "Creating archive: $ARCHIVE_NAME"
+                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
             else
-                zip -r "$ZIP_NAME" "text-generation-webui-${VERSION_CLEAN}"
+                ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.tar.gz"
+                echo "Creating archive: $ARCHIVE_NAME"
+                tar czf "$ARCHIVE_NAME" "text-generation-webui-${VERSION_CLEAN}"
             fi
 
       - name: Upload files to a GitHub release
@@ -188,7 +189,7 @@ jobs:
         continue-on-error: true
         with:
           repo_token: ${{ secrets.GITHUB_TOKEN }}
-          file: ../textgen-portable-*.zip
+          file: ../textgen-portable-*
           tag: ${{ inputs.version }}
           file_glob: true
           make_latest: false

From 6ff111d18e2f56f861fb6be54166b772f248fa33 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Mar 2026 22:05:31 -0300
Subject: [PATCH 1284/1701] ExLlamav3: handle exceptions in ConcurrentGenerator
 iterate loop

---
 modules/exllamav3.py | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index b4b76e2109..d97726827d 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -53,7 +53,16 @@ def _iterate_loop(self):
                 if not self.job_queues:
                     self.has_jobs.clear()
                     continue
-                results = self.generator.iterate()
+                try:
+                    results = self.generator.iterate()
+                except Exception:
+                    logger.error("Exception in ConcurrentGenerator iterate loop:\n" + traceback.format_exc())
+                    for q in self.job_queues.values():
+                        q.put(None)
+                    self.job_queues.clear()
+                    self.generator.clear_queue()
+                    self.has_jobs.clear()
+                    continue
             for result in results:
                 job = result["job"]
                 q = self.job_queues.get(job)

From baf4e13ff147246f00c3947f14a701deac88f4ea Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Mar 2026 22:34:48 -0300
Subject: [PATCH 1285/1701] ExLlamav3: fix draft cache size to match main cache

---
 modules/exllamav3.py | 17 +----------------
 1 file changed, 1 insertion(+), 16 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index d97726827d..9ea38432b9 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -175,23 +175,8 @@ def from_pretrained(cls, path_to_model):
                 logger.warning(f"Draft model not found at {draft_path}, speculative decoding disabled.")
             else:
                 draft_config = Config.from_directory(str(draft_path))
-
-                # Set context size for draft model with 256-multiple validation
-                if shared.args.ctx_size_draft > 0:
-                    draft_max_tokens = shared.args.ctx_size_draft
-                else:
-                    draft_max_tokens = shared.args.ctx_size
-
-                # Validate draft model context size is a multiple of 256
-                if draft_max_tokens % 256 != 0:
-                    adjusted_draft_tokens = ((draft_max_tokens // 256) + 1) * 256
-                    logger.warning(f"Draft model max_num_tokens must be a multiple of 256. Adjusting from {draft_max_tokens} to {adjusted_draft_tokens}")
-                    draft_max_tokens = adjusted_draft_tokens
-
-                draft_config.max_seq_len = draft_max_tokens
-
                 draft_model = Model.from_config(draft_config)
-                draft_cache = Cache(draft_model, max_num_tokens=draft_max_tokens, layer_type=layer_type, **cache_kwargs)
+                draft_cache = Cache(draft_model, max_num_tokens=max_tokens, layer_type=layer_type, **cache_kwargs)
 
                 draft_load_params = {'progressbar': True}
                 if split:

From 0132966d09abc8a3fc3ffba3db2a0069e430fd47 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Mar 2026 23:06:15 -0300
Subject: [PATCH 1286/1701] Add PyPI fallback for PyTorch install commands

---
 one_click.py | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)

diff --git a/one_click.py b/one_click.py
index efb0713469..5131206e8b 100644
--- a/one_click.py
+++ b/one_click.py
@@ -111,13 +111,14 @@ def get_gpu_choice():
 def get_pytorch_install_command(gpu_choice):
     """Get PyTorch installation command based on GPU choice"""
     base_cmd = f"python -m pip install torch=={TORCH_VERSION} "
+    pypi_fallback = " --extra-index-url https://pypi.org/simple/"
 
     if gpu_choice == "NVIDIA_CUDA128":
-        return base_cmd + "--index-url https://download.pytorch.org/whl/cu128"
+        return base_cmd + "--index-url https://download.pytorch.org/whl/cu128" + pypi_fallback
     elif gpu_choice == "AMD":
-        return base_cmd + "--index-url https://download.pytorch.org/whl/rocm6.4"
+        return base_cmd + "--index-url https://download.pytorch.org/whl/rocm6.4" + pypi_fallback
     elif gpu_choice in ["APPLE", "NONE"]:
-        return base_cmd + "--index-url https://download.pytorch.org/whl/cpu"
+        return base_cmd + "--index-url https://download.pytorch.org/whl/cpu" + pypi_fallback
     elif gpu_choice == "INTEL":
         if is_linux():
             return "python -m pip install torch==2.1.0a0 intel-extension-for-pytorch==2.1.10+xpu --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
@@ -130,16 +131,17 @@ def get_pytorch_install_command(gpu_choice):
 def get_pytorch_update_command(gpu_choice):
     """Get PyTorch update command based on GPU choice"""
     base_cmd = f"python -m pip install --upgrade torch=={TORCH_VERSION} "
+    pypi_fallback = " --extra-index-url https://pypi.org/simple/"
 
     if gpu_choice == "NVIDIA_CUDA128":
-        return f"{base_cmd} --index-url https://download.pytorch.org/whl/cu128"
+        return f"{base_cmd}--index-url https://download.pytorch.org/whl/cu128" + pypi_fallback
     elif gpu_choice == "AMD":
-        return f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.4"
+        return f"{base_cmd}--index-url https://download.pytorch.org/whl/rocm6.4" + pypi_fallback
     elif gpu_choice in ["APPLE", "NONE"]:
-        return f"{base_cmd} --index-url https://download.pytorch.org/whl/cpu"
+        return f"{base_cmd}--index-url https://download.pytorch.org/whl/cpu" + pypi_fallback
     elif gpu_choice == "INTEL":
         intel_extension = "intel-extension-for-pytorch==2.1.10+xpu" if is_linux() else "intel-extension-for-pytorch==2.1.10"
-        return f"{base_cmd} {intel_extension} --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
+        return f"{base_cmd}{intel_extension} --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
     else:
         return base_cmd
 

From b3705d87bfceda3ad6be09473e0772b67e897dd2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 7 Mar 2026 23:06:15 -0300
Subject: [PATCH 1287/1701] Add PyPI fallback for PyTorch install commands

---
 one_click.py | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)

diff --git a/one_click.py b/one_click.py
index efb0713469..5131206e8b 100644
--- a/one_click.py
+++ b/one_click.py
@@ -111,13 +111,14 @@ def get_gpu_choice():
 def get_pytorch_install_command(gpu_choice):
     """Get PyTorch installation command based on GPU choice"""
     base_cmd = f"python -m pip install torch=={TORCH_VERSION} "
+    pypi_fallback = " --extra-index-url https://pypi.org/simple/"
 
     if gpu_choice == "NVIDIA_CUDA128":
-        return base_cmd + "--index-url https://download.pytorch.org/whl/cu128"
+        return base_cmd + "--index-url https://download.pytorch.org/whl/cu128" + pypi_fallback
     elif gpu_choice == "AMD":
-        return base_cmd + "--index-url https://download.pytorch.org/whl/rocm6.4"
+        return base_cmd + "--index-url https://download.pytorch.org/whl/rocm6.4" + pypi_fallback
     elif gpu_choice in ["APPLE", "NONE"]:
-        return base_cmd + "--index-url https://download.pytorch.org/whl/cpu"
+        return base_cmd + "--index-url https://download.pytorch.org/whl/cpu" + pypi_fallback
     elif gpu_choice == "INTEL":
         if is_linux():
             return "python -m pip install torch==2.1.0a0 intel-extension-for-pytorch==2.1.10+xpu --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
@@ -130,16 +131,17 @@ def get_pytorch_install_command(gpu_choice):
 def get_pytorch_update_command(gpu_choice):
     """Get PyTorch update command based on GPU choice"""
     base_cmd = f"python -m pip install --upgrade torch=={TORCH_VERSION} "
+    pypi_fallback = " --extra-index-url https://pypi.org/simple/"
 
     if gpu_choice == "NVIDIA_CUDA128":
-        return f"{base_cmd} --index-url https://download.pytorch.org/whl/cu128"
+        return f"{base_cmd}--index-url https://download.pytorch.org/whl/cu128" + pypi_fallback
     elif gpu_choice == "AMD":
-        return f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.4"
+        return f"{base_cmd}--index-url https://download.pytorch.org/whl/rocm6.4" + pypi_fallback
     elif gpu_choice in ["APPLE", "NONE"]:
-        return f"{base_cmd} --index-url https://download.pytorch.org/whl/cpu"
+        return f"{base_cmd}--index-url https://download.pytorch.org/whl/cpu" + pypi_fallback
     elif gpu_choice == "INTEL":
         intel_extension = "intel-extension-for-pytorch==2.1.10+xpu" if is_linux() else "intel-extension-for-pytorch==2.1.10"
-        return f"{base_cmd} {intel_extension} --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
+        return f"{base_cmd}{intel_extension} --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
     else:
         return base_cmd
 

From 7170a16b91dab7e5de5cd01e01d5239050995474 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Mar 2026 04:09:18 -0700
Subject: [PATCH 1288/1701] Fix passing adaptive-p to llama-server

---
 modules/llama_cpp_server.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 12ff173e3b..6f7cbd20f5 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -129,7 +129,7 @@ def prepare_payload(self, state):
             # places it at the end of the chain regardless of position, so we
             # activate it based on the parameter value rather than sampler order.
             if state.get("adaptive_target", 0) > 0:
-                filtered_samplers.append("adaptive-p")
+                filtered_samplers.append("adaptive_p")
 
             payload["samplers"] = filtered_samplers
 

From 7a8ca9f2b01d5904149248715ba2cbd93b996bb6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Mar 2026 04:09:18 -0700
Subject: [PATCH 1289/1701] Fix passing adaptive-p to llama-server

---
 modules/llama_cpp_server.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 12ff173e3b..6f7cbd20f5 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -129,7 +129,7 @@ def prepare_payload(self, state):
             # places it at the end of the chain regardless of position, so we
             # activate it based on the parameter value rather than sampler order.
             if state.get("adaptive_target", 0) > 0:
-                filtered_samplers.append("adaptive-p")
+                filtered_samplers.append("adaptive_p")
 
             payload["samplers"] = filtered_samplers
 

From 5a91b8462f9ecc487ee5fc28d4875c58a40bb7d7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Mar 2026 09:53:48 -0300
Subject: [PATCH 1290/1701] Remove ctx_size_draft from ExLlamav3 loader

---
 modules/loaders.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/modules/loaders.py b/modules/loaders.py
index 42a5ff1c5e..64de3dde52 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -71,7 +71,6 @@
         'gpu_split',
         'model_draft',
         'draft_max',
-        'ctx_size_draft',
         'speculative_decoding_accordion',
         'enable_tp',
         'tp_backend',

From f6ffecfff2f06e6fead00ca0895971d162c0c68b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Mar 2026 10:46:51 -0300
Subject: [PATCH 1291/1701] Add guard against training with llama.cpp loader

---
 docs/05 - Training Tab.md | 2 +-
 modules/training.py       | 5 +++++
 2 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/docs/05 - Training Tab.md b/docs/05 - Training Tab.md
index 902693e673..0bfc59aa40 100644
--- a/docs/05 - Training Tab.md	
+++ b/docs/05 - Training Tab.md	
@@ -4,7 +4,7 @@ A LoRA is tied to a specific model architecture — a LoRA trained on Llama 3 8B
 
 ### Quick Start
 
-1. Load your base model (no LoRAs loaded).
+1. Load your base model with the **Transformers** loader (no LoRAs loaded).
 2. Open the **Training** tab > **Train LoRA**.
 3. Pick a dataset and configure parameters (see [below](#parameters)).
 4. Click **Start LoRA Training** and monitor the [loss](#loss).
diff --git a/modules/training.py b/modules/training.py
index 2e172d2266..878bb222a8 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -310,6 +310,11 @@ def do_train(lora_name: str, always_override: bool, all_linear: bool, q_proj_en:
 
     # == Input validation / processing ==
     yield "Preparing the input..."
+
+    if shared.args.loader == 'llama.cpp':
+        yield "Error: LoRA training requires a model loaded with the Transformers loader. GGUF models are not supported for training."
+        return
+
     lora_file_path = clean_path(None, lora_name)
     if lora_file_path.strip() == '':
         yield "Missing or invalid LoRA file name input."

From 40f1837b42ac0c444b86dc937a565d3a6d03448a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Mar 2026 08:38:29 -0700
Subject: [PATCH 1292/1701] README: Minor updates

---
 README.md         | 4 ++--
 modules/shared.py | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/README.md b/README.md
index a47a0e6f7d..a80ebc62ae 100644
--- a/README.md
+++ b/README.md
@@ -201,7 +201,7 @@ ln -s docker/{nvidia/Dockerfile,nvidia/docker-compose.yml,.dockerignore} .
 For AMD GPU:
 ln -s docker/{amd/Dockerfile,amd/docker-compose.yml,.dockerignore} .
 For Intel GPU:
-ln -s docker/{intel/Dockerfile,amd/docker-compose.yml,.dockerignore} .
+ln -s docker/{intel/Dockerfile,intel/docker-compose.yml,.dockerignore} .
 For CPU only
 ln -s docker/{cpu/Dockerfile,cpu/docker-compose.yml,.dockerignore} .
 cp docker/.env.example .env
@@ -365,7 +365,7 @@ Gradio:
 
 API:
   --api                                                Enable the API extension.
-  --public-api                                         Create a public URL for the API using Cloudfare.
+  --public-api                                         Create a public URL for the API using Cloudflare.
   --public-api-id PUBLIC_API_ID                        Tunnel ID for named Cloudflare Tunnel. Use together with public-api option.
   --api-port API_PORT                                  The listening port for the API.
   --api-key API_KEY                                    API authentication key.
diff --git a/modules/shared.py b/modules/shared.py
index bc7ea8ba9b..080874ec4c 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -163,7 +163,7 @@
 # API
 group = parser.add_argument_group('API')
 group.add_argument('--api', action='store_true', help='Enable the API extension.')
-group.add_argument('--public-api', action='store_true', help='Create a public URL for the API using Cloudfare.')
+group.add_argument('--public-api', action='store_true', help='Create a public URL for the API using Cloudflare.')
 group.add_argument('--public-api-id', type=str, help='Tunnel ID for named Cloudflare Tunnel. Use together with public-api option.', default=None)
 group.add_argument('--api-port', type=int, default=5000, help='The listening port for the API.')
 group.add_argument('--api-key', type=str, default='', help='API authentication key.')

From 634609accaca98ed750bc28ca9f3bf20eb08067a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Mar 2026 20:23:15 -0700
Subject: [PATCH 1293/1701] Fix pip installing to system Miniconda on Windows,
 revert 0132966d

---
 one_click.py | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

diff --git a/one_click.py b/one_click.py
index 5131206e8b..cbdeadca68 100644
--- a/one_click.py
+++ b/one_click.py
@@ -111,14 +111,13 @@ def get_gpu_choice():
 def get_pytorch_install_command(gpu_choice):
     """Get PyTorch installation command based on GPU choice"""
     base_cmd = f"python -m pip install torch=={TORCH_VERSION} "
-    pypi_fallback = " --extra-index-url https://pypi.org/simple/"
 
     if gpu_choice == "NVIDIA_CUDA128":
-        return base_cmd + "--index-url https://download.pytorch.org/whl/cu128" + pypi_fallback
+        return base_cmd + "--index-url https://download.pytorch.org/whl/cu128"
     elif gpu_choice == "AMD":
-        return base_cmd + "--index-url https://download.pytorch.org/whl/rocm6.4" + pypi_fallback
+        return base_cmd + "--index-url https://download.pytorch.org/whl/rocm6.4"
     elif gpu_choice in ["APPLE", "NONE"]:
-        return base_cmd + "--index-url https://download.pytorch.org/whl/cpu" + pypi_fallback
+        return base_cmd + "--index-url https://download.pytorch.org/whl/cpu"
     elif gpu_choice == "INTEL":
         if is_linux():
             return "python -m pip install torch==2.1.0a0 intel-extension-for-pytorch==2.1.10+xpu --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
@@ -131,17 +130,16 @@ def get_pytorch_install_command(gpu_choice):
 def get_pytorch_update_command(gpu_choice):
     """Get PyTorch update command based on GPU choice"""
     base_cmd = f"python -m pip install --upgrade torch=={TORCH_VERSION} "
-    pypi_fallback = " --extra-index-url https://pypi.org/simple/"
 
     if gpu_choice == "NVIDIA_CUDA128":
-        return f"{base_cmd}--index-url https://download.pytorch.org/whl/cu128" + pypi_fallback
+        return f"{base_cmd} --index-url https://download.pytorch.org/whl/cu128"
     elif gpu_choice == "AMD":
-        return f"{base_cmd}--index-url https://download.pytorch.org/whl/rocm6.4" + pypi_fallback
+        return f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.4"
     elif gpu_choice in ["APPLE", "NONE"]:
-        return f"{base_cmd}--index-url https://download.pytorch.org/whl/cpu" + pypi_fallback
+        return f"{base_cmd} --index-url https://download.pytorch.org/whl/cpu"
     elif gpu_choice == "INTEL":
         intel_extension = "intel-extension-for-pytorch==2.1.10+xpu" if is_linux() else "intel-extension-for-pytorch==2.1.10"
-        return f"{base_cmd}{intel_extension} --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
+        return f"{base_cmd} {intel_extension} --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
     else:
         return base_cmd
 
@@ -196,6 +194,8 @@ def run_cmd(cmd, assert_success=False, environment=False, capture_output=False,
     if environment:
         if is_windows():
             conda_bat_path = os.path.join(script_dir, "installer_files", "conda", "condabin", "conda.bat")
+            python_path = os.path.join(conda_env_path, "python.exe")
+            cmd = cmd.replace("python ", f'"{python_path}" ')
             cmd = f'"{conda_bat_path}" activate "{conda_env_path}" >nul && {cmd}'
         else:
             conda_sh_path = os.path.join(script_dir, "installer_files", "conda", "etc", "profile.d", "conda.sh")

From eb4a20137a4d4200d4deb05bd22dd5bc1ff2ce9f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 8 Mar 2026 20:38:50 -0700
Subject: [PATCH 1294/1701] Update README

---
 README.md | 85 ++++++++++++++++++++++++++++++++++++++++++++-----------
 1 file changed, 68 insertions(+), 17 deletions(-)

diff --git a/README.md b/README.md
index a80ebc62ae..7831ef650f 100644
--- a/README.md
+++ b/README.md
@@ -236,20 +236,25 @@ List of command-line flags
 
            
 
 ```txt
-usage: server.py [-h] [--multi-user] [--model MODEL] [--lora LORA [LORA ...]] [--model-dir MODEL_DIR] [--lora-dir LORA_DIR] [--model-menu] [--settings SETTINGS]
+usage: server.py [-h] [--user-data-dir USER_DATA_DIR] [--multi-user] [--model MODEL] [--lora LORA [LORA ...]] [--model-dir MODEL_DIR] [--lora-dir LORA_DIR] [--model-menu] [--settings SETTINGS]
                  [--extensions EXTENSIONS [EXTENSIONS ...]] [--verbose] [--idle-timeout IDLE_TIMEOUT] [--image-model IMAGE_MODEL] [--image-model-dir IMAGE_MODEL_DIR] [--image-dtype {bfloat16,float16}]
                  [--image-attn-backend {flash_attention_2,sdpa}] [--image-cpu-offload] [--image-compile] [--image-quant {none,bnb-8bit,bnb-4bit,torchao-int8wo,torchao-fp4,torchao-float8wo}]
                  [--loader LOADER] [--ctx-size N] [--cache-type N] [--model-draft MODEL_DRAFT] [--draft-max DRAFT_MAX] [--gpu-layers-draft GPU_LAYERS_DRAFT] [--device-draft DEVICE_DRAFT]
                  [--ctx-size-draft CTX_SIZE_DRAFT] [--spec-type {none,ngram-mod,ngram-simple,ngram-map-k,ngram-map-k4v,ngram-cache}] [--spec-ngram-size-n SPEC_NGRAM_SIZE_N]
                  [--spec-ngram-size-m SPEC_NGRAM_SIZE_M] [--spec-ngram-min-hits SPEC_NGRAM_MIN_HITS] [--gpu-layers N] [--cpu-moe] [--mmproj MMPROJ] [--streaming-llm] [--tensor-split TENSOR_SPLIT]
                  [--row-split] [--no-mmap] [--mlock] [--no-kv-offload] [--batch-size BATCH_SIZE] [--ubatch-size UBATCH_SIZE] [--threads THREADS] [--threads-batch THREADS_BATCH] [--numa]
-                 [--extra-flags EXTRA_FLAGS] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code]
-                 [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE]
-                 [--gpu-split GPU_SPLIT] [--enable-tp] [--tp-backend TP_BACKEND] [--cfg-cache] [--cpp-runner]
-                 [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE] [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share]
-                 [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors]
-                 [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4]
-                 [--nowebui]
+                 [--parallel PARALLEL] [--fit-target FIT_TARGET] [--extra-flags EXTRA_FLAGS] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] [--bf16]
+                 [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE]
+                 [--quant_type QUANT_TYPE] [--gpu-split GPU_SPLIT] [--enable-tp] [--tp-backend TP_BACKEND] [--cfg-cache] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE]
+                 [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH]
+                 [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors] [--portable] [--api] [--public-api]
+                 [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4] [--nowebui] [--temperature N]
+                 [--dynatemp-low N] [--dynatemp-high N] [--dynatemp-exponent N] [--smoothing-factor N] [--smoothing-curve N] [--min-p N] [--top-p N] [--top-k N] [--typical-p N] [--xtc-threshold N]
+                 [--xtc-probability N] [--epsilon-cutoff N] [--eta-cutoff N] [--tfs N] [--top-a N] [--top-n-sigma N] [--adaptive-target N] [--adaptive-decay N] [--dry-multiplier N]
+                 [--dry-allowed-length N] [--dry-base N] [--repetition-penalty N] [--frequency-penalty N] [--presence-penalty N] [--encoder-repetition-penalty N] [--no-repeat-ngram-size N]
+                 [--repetition-penalty-range N] [--penalty-alpha N] [--guidance-scale N] [--mirostat-mode N] [--mirostat-tau N] [--mirostat-eta N] [--do-sample | --no-do-sample]
+                 [--dynamic-temperature | --no-dynamic-temperature] [--temperature-last | --no-temperature-last] [--sampler-priority N] [--dry-sequence-breakers N]
+                 [--enable-thinking | --no-enable-thinking] [--reasoning-effort N] [--chat-template-file CHAT_TEMPLATE_FILE]
 
 Text Generation Web UI
 
@@ -257,6 +262,7 @@ options:
   -h, --help                                           show this help message and exit
 
 Basic settings:
+  --user-data-dir USER_DATA_DIR                        Path to the user data directory. Default: auto-detected.
   --multi-user                                         Multi-user mode. Chat histories are not saved or automatically loaded. Warning: this is likely not safe for sharing publicly.
   --model MODEL                                        Name of the model to load by default.
   --lora LORA [LORA ...]                               The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.
@@ -280,12 +286,12 @@ Image model:
                                                        Quantization method for image model.
 
 Model loader:
-  --loader LOADER                                      Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav3,
-                                                       TensorRT-LLM.
+  --loader LOADER                                      Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, ExLlamav3_HF, ExLlamav3, TensorRT-
+                                                       LLM.
 
 Context and cache:
-  --ctx-size N, --n_ctx N, --max_seq_len N             Context size in tokens. llama.cpp: 0 = auto if gpu-layers is also -1.
-  --cache-type N, --cache_type N                       KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).
+  --ctx-size, --n_ctx, --max_seq_len N                 Context size in tokens. llama.cpp: 0 = auto if gpu-layers is also -1.
+  --cache-type, --cache_type N                         KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).
 
 Speculative decoding:
   --model-draft MODEL_DRAFT                            Path to the draft model for speculative decoding.
@@ -300,7 +306,7 @@ Speculative decoding:
   --spec-ngram-min-hits SPEC_NGRAM_MIN_HITS            Minimum n-gram hits for ngram-map speculative decoding.
 
 llama.cpp:
-  --gpu-layers N, --n-gpu-layers N                     Number of layers to offload to the GPU. -1 = auto.
+  --gpu-layers, --n-gpu-layers N                       Number of layers to offload to the GPU. -1 = auto.
   --cpu-moe                                            Move the experts to the CPU (for MoE models).
   --mmproj MMPROJ                                      Path to the mmproj file for vision models.
   --streaming-llm                                      Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.
@@ -314,13 +320,17 @@ llama.cpp:
   --threads THREADS                                    Number of threads to use.
   --threads-batch THREADS_BATCH                        Number of threads to use for batches/prompt processing.
   --numa                                               Activate NUMA task allocation for llama.cpp.
+  --parallel PARALLEL                                  Number of parallel request slots. The context size is divided equally among slots. For example, to have 4 slots with 8192 context each, set
+                                                       ctx_size to 32768.
+  --fit-target FIT_TARGET                              Target VRAM margin per device for auto GPU layers, comma-separated list of values in MiB. A single value is broadcast across all devices.
+                                                       Default: 1024.
   --extra-flags EXTRA_FLAGS                            Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"
 
 Transformers/Accelerate:
   --cpu                                                Use the CPU to generate text. Warning: Training on CPU is extremely slow.
   --cpu-memory CPU_MEMORY                              Maximum CPU memory in GiB. Use this for CPU offloading.
   --disk                                               If the model is too large for your GPU(s) and CPU combined, send the remaining layers to the disk.
-  --disk-cache-dir DISK_CACHE_DIR                      Directory to save the disk cache to. Defaults to "user_data/cache".
+  --disk-cache-dir DISK_CACHE_DIR                      Directory to save the disk cache to.
   --load-in-8bit                                       Load the model with 8-bit precision (using bitsandbytes).
   --bf16                                               Load the model with bfloat16 precision. Requires NVIDIA Ampere GPU.
   --no-cache                                           Set use_cache to False while generating text. This reduces VRAM usage slightly, but it comes at a performance cost.
@@ -341,9 +351,6 @@ ExLlamaV3:
   --tp-backend TP_BACKEND                              The backend for tensor parallelism. Valid options: native, nccl. Default: native.
   --cfg-cache                                          Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.
 
-TensorRT-LLM:
-  --cpp-runner                                         Use the ModelRunnerCpp runner, which is faster than the default ModelRunner.
-
 RoPE:
   --alpha_value ALPHA_VALUE                            Positional embeddings alpha factor for NTK RoPE scaling. Use either this or compress_pos_emb, not both.
   --rope_freq_base ROPE_FREQ_BASE                      If greater than 0, will be used instead of alpha_value. Those two are related by rope_freq_base = 10000 * alpha_value ^ (64 / 63).
@@ -373,6 +380,50 @@ API:
   --api-enable-ipv6                                    Enable IPv6 for the API
   --api-disable-ipv4                                   Disable IPv4 for the API
   --nowebui                                            Do not launch the Gradio UI. Useful for launching the API in standalone mode.
+
+API generation defaults:
+  --temperature N                                      Temperature
+  --dynatemp-low N                                     Dynamic temperature low
+  --dynatemp-high N                                    Dynamic temperature high
+  --dynatemp-exponent N                                Dynamic temperature exponent
+  --smoothing-factor N                                 Smoothing factor
+  --smoothing-curve N                                  Smoothing curve
+  --min-p N                                            Min P
+  --top-p N                                            Top P
+  --top-k N                                            Top K
+  --typical-p N                                        Typical P
+  --xtc-threshold N                                    XTC threshold
+  --xtc-probability N                                  XTC probability
+  --epsilon-cutoff N                                   Epsilon cutoff
+  --eta-cutoff N                                       Eta cutoff
+  --tfs N                                              TFS
+  --top-a N                                            Top A
+  --top-n-sigma N                                      Top N Sigma
+  --adaptive-target N                                  Adaptive target
+  --adaptive-decay N                                   Adaptive decay
+  --dry-multiplier N                                   DRY multiplier
+  --dry-allowed-length N                               DRY allowed length
+  --dry-base N                                         DRY base
+  --repetition-penalty N                               Repetition penalty
+  --frequency-penalty N                                Frequency penalty
+  --presence-penalty N                                 Presence penalty
+  --encoder-repetition-penalty N                       Encoder repetition penalty
+  --no-repeat-ngram-size N                             No repeat ngram size
+  --repetition-penalty-range N                         Repetition penalty range
+  --penalty-alpha N                                    Penalty alpha
+  --guidance-scale N                                   Guidance scale
+  --mirostat-mode N                                    Mirostat mode
+  --mirostat-tau N                                     Mirostat tau
+  --mirostat-eta N                                     Mirostat eta
+  --do-sample, --no-do-sample                          Do sample
+  --dynamic-temperature, --no-dynamic-temperature      Dynamic temperature
+  --temperature-last, --no-temperature-last            Temperature last
+  --sampler-priority N                                 Sampler priority
+  --dry-sequence-breakers N                            DRY sequence breakers
+  --enable-thinking, --no-enable-thinking              Enable thinking
+  --reasoning-effort N                                 Reasoning effort
+  --chat-template-file CHAT_TEMPLATE_FILE              Path to a chat template file (.jinja, .jinja2, or .yaml) to use as the default instruction template for API requests. Overrides the model's
+                                                       built-in template.
 ```
 
 
            

From 9753b2342b3ffedd3a79e675d6ab5bf70400fef8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Mar 2026 16:22:04 -0300
Subject: [PATCH 1295/1701] Fix crash on non-UTF-8 Windows locales (e.g.
 Chinese GBK)

Closes #7416
---
 cmd_windows.bat   | 1 +
 start_windows.bat | 1 +
 2 files changed, 2 insertions(+)

diff --git a/cmd_windows.bat b/cmd_windows.bat
index 787b433501..b0540bd817 100755
--- a/cmd_windows.bat
+++ b/cmd_windows.bat
@@ -21,6 +21,7 @@ set INSTALL_ENV_DIR=%cd%\installer_files\env
 set PYTHONNOUSERSITE=1
 set PYTHONPATH=
 set PYTHONHOME=
+set PYTHONUTF8=1
 set "CUDA_PATH=%INSTALL_ENV_DIR%"
 set "CUDA_HOME=%CUDA_PATH%"
 
diff --git a/start_windows.bat b/start_windows.bat
index dd096760e4..8da6986ff9 100755
--- a/start_windows.bat
+++ b/start_windows.bat
@@ -5,6 +5,7 @@ setlocal enabledelayedexpansion
 set PYTHONNOUSERSITE=1
 set PYTHONPATH=
 set PYTHONHOME=
+set PYTHONUTF8=1
 
 cd /D "%~dp0"
 

From d6643bb4bc0205eb2de6a226b528b03609e8a786 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Mar 2026 12:30:43 -0700
Subject: [PATCH 1296/1701] One-click installer: Optimize wheel downloads to
 only re-download changed wheels

---
 one_click.py | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

diff --git a/one_click.py b/one_click.py
index cbdeadca68..60c6cb6ff7 100644
--- a/one_click.py
+++ b/one_click.py
@@ -365,8 +365,10 @@ def update_requirements(initial_installation=False, pull=True):
 
     current_commit = get_current_commit()
     wheels_changed = not os.path.exists(state_file)
+    installed_wheels = set()
     if not wheels_changed:
         state = load_state()
+        installed_wheels = set(state.get('installed_wheels', []))
         if 'wheels_changed' in state or state.get('last_installed_commit') != current_commit:
             wheels_changed = True
 
@@ -431,9 +433,17 @@ def update_requirements(initial_installation=False, pull=True):
 
     # Prepare the requirements file
     textgen_requirements = open(requirements_file).read().splitlines()
+    all_whl_lines = [line.strip() for line in textgen_requirements if '.whl' in line]
 
-    if not initial_installation and not wheels_changed:
-        textgen_requirements = [line for line in textgen_requirements if '.whl' not in line]
+    if not initial_installation:
+        if installed_wheels:
+            # Per-wheel comparison: only re-download wheels that changed
+            textgen_requirements = [
+                line for line in textgen_requirements
+                if '.whl' not in line or line.strip() not in installed_wheels
+            ]
+        elif not wheels_changed:
+            textgen_requirements = [line for line in textgen_requirements if '.whl' not in line]
 
     with open('temp_requirements.txt', 'w') as file:
         file.write('\n'.join(textgen_requirements))
@@ -452,6 +462,7 @@ def update_requirements(initial_installation=False, pull=True):
     # Save state after successful installation
     state = load_state()
     state['last_installed_commit'] = current_commit
+    state['installed_wheels'] = all_whl_lines
     state.pop('wheels_changed', None)
     save_state(state)
 

From 970055ca00d896293c603be1fd25d14d6e4e76e4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Mar 2026 17:08:31 -0300
Subject: [PATCH 1297/1701] Update Intel GPU support to use native PyTorch XPU
 wheels

PyTorch 2.9+ includes native XPU support, making
intel-extension-for-pytorch and the separate oneAPI conda
install unnecessary.

Closes #7308
---
 one_click.py | 15 ++-------------
 1 file changed, 2 insertions(+), 13 deletions(-)

diff --git a/one_click.py b/one_click.py
index 60c6cb6ff7..189d81b835 100644
--- a/one_click.py
+++ b/one_click.py
@@ -119,10 +119,7 @@ def get_pytorch_install_command(gpu_choice):
     elif gpu_choice in ["APPLE", "NONE"]:
         return base_cmd + "--index-url https://download.pytorch.org/whl/cpu"
     elif gpu_choice == "INTEL":
-        if is_linux():
-            return "python -m pip install torch==2.1.0a0 intel-extension-for-pytorch==2.1.10+xpu --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
-        else:
-            return "python -m pip install torch==2.1.0a0 intel-extension-for-pytorch==2.1.10 --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
+        return base_cmd + "--index-url https://download.pytorch.org/whl/xpu"
     else:
         return base_cmd
 
@@ -138,8 +135,7 @@ def get_pytorch_update_command(gpu_choice):
     elif gpu_choice in ["APPLE", "NONE"]:
         return f"{base_cmd} --index-url https://download.pytorch.org/whl/cpu"
     elif gpu_choice == "INTEL":
-        intel_extension = "intel-extension-for-pytorch==2.1.10+xpu" if is_linux() else "intel-extension-for-pytorch==2.1.10"
-        return f"{base_cmd} {intel_extension} --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
+        return f"{base_cmd} --index-url https://download.pytorch.org/whl/xpu"
     else:
         return base_cmd
 
@@ -316,13 +312,6 @@ def install_webui():
     install_pytorch = get_pytorch_install_command(gpu_choice)
     run_cmd(f"conda install -y ninja git && {install_pytorch}", assert_success=True, environment=True)
 
-    if gpu_choice == "INTEL":
-        # Install oneAPI dependencies via conda
-        print_big_message("Installing Intel oneAPI runtime libraries.")
-        run_cmd("conda install -y -c https://software.repos.intel.com/python/conda/ -c conda-forge dpcpp-cpp-rt=2024.0 mkl-dpcpp=2024.0", environment=True)
-        # Install libuv required by Intel-patched torch
-        run_cmd("conda install -y libuv", environment=True)
-
     # Install the webui requirements
     update_requirements(initial_installation=True, pull=False)
 

From 39e6c997cc480818911c7f3644bbcd306eca932e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Mar 2026 19:29:24 -0700
Subject: [PATCH 1298/1701] Refactor to not import gradio in `--nowebui` mode

---
 extensions/openai/models.py |   4 +-
 modules/chat.py             | 122 +++++++-----------------------------
 modules/extensions.py       |   4 +-
 modules/loaders.py          |  57 ++++++++++++++++-
 modules/models_settings.py  | 109 ++++++++++++++++++++++++++++++--
 modules/ui.py               |  54 +---------------
 server.py                   |  91 +++++++++++++--------------
 7 files changed, 232 insertions(+), 209 deletions(-)

diff --git a/extensions/openai/models.py b/extensions/openai/models.py
index d6ef119db8..82c65093c8 100644
--- a/extensions/openai/models.py
+++ b/extensions/openai/models.py
@@ -1,4 +1,4 @@
-from modules import shared, ui
+from modules import loaders, shared
 from modules.logging_colors import logger
 from modules.LoRA import add_lora_to_model
 from modules.models import load_model, unload_model
@@ -50,7 +50,7 @@ def _load_model(data):
     # parameters exposed in the UI. Never allow security-sensitive
     # flags like trust_remote_code or extra_flags to be set via the API.
     blocked_keys = {'extra_flags'}
-    allowed_keys = set(ui.list_model_elements()) - blocked_keys
+    allowed_keys = set(loaders.list_model_elements()) - blocked_keys
     if args:
         for k in args:
             if k in allowed_keys and hasattr(shared.args, k):
diff --git a/modules/chat.py b/modules/chat.py
index 36d373d676..62d1492d90 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -11,7 +11,6 @@
 from functools import partial
 from pathlib import Path
 
-import gradio as gr
 import yaml
 from jinja2.ext import loopcontrols
 from jinja2.sandbox import ImmutableSandboxedEnvironment
@@ -1333,6 +1332,7 @@ def load_history_after_deletion(state, idx):
     Loads the latest history for the given character in chat or chat-instruct
     mode, or the latest instruct history for instruct mode.
     '''
+    import gradio as gr
 
     if shared.args.multi_user:
         return start_new_chat(state)
@@ -1351,6 +1351,7 @@ def load_history_after_deletion(state, idx):
 
 
 def update_character_menu_after_deletion(idx):
+    import gradio as gr
     characters = utils.get_available_characters()
     idx = min(int(idx), len(characters) - 1)
     idx = max(0, idx)
@@ -1565,24 +1566,6 @@ def clear_character_for_ui(state):
     return state, state['name2'], state['context'], state['greeting'], None
 
 
-def load_instruction_template(template):
-    if template == 'None':
-        return ''
-
-    for filepath in [shared.user_data_dir / 'instruction-templates' / f'{template}.yaml', shared.user_data_dir / 'instruction-templates' / 'Alpaca.yaml']:
-        if filepath.exists():
-            break
-    else:
-        return ''
-
-    file_contents = open(filepath, 'r', encoding='utf-8').read()
-    data = yaml.safe_load(file_contents)
-    if 'instruction_template' in data:
-        return data['instruction_template']
-    else:
-        return jinja_template_from_old_format(data)
-
-
 @functools.cache
 def load_character_memoized(character, name1, name2):
     return load_character(character, name1, name2)
@@ -1590,10 +1573,12 @@ def load_character_memoized(character, name1, name2):
 
 @functools.cache
 def load_instruction_template_memoized(template):
+    from modules.models_settings import load_instruction_template
     return load_instruction_template(template)
 
 
 def upload_character(file, img_path, tavern=False):
+    import gradio as gr
     img = open_image_safely(img_path)
     decoded_file = file if isinstance(file, str) else file.decode('utf-8')
     try:
@@ -1647,6 +1632,7 @@ def upload_tavern_character(img_path, _json):
 
 
 def check_tavern_character(img_path):
+    import gradio as gr
     img = open_image_safely(img_path)
 
     if img is None:
@@ -1832,6 +1818,7 @@ def delete_user(name):
 
 def update_user_menu_after_deletion(idx):
     """Update user menu after a user is deleted"""
+    import gradio as gr
     users = get_available_users()
     if len(users) == 0:
         # Create a default user if none exist
@@ -1864,93 +1851,13 @@ def handle_user_menu_change(state):
 
 def handle_save_user_click(name1):
     """Handle save user button click"""
+    import gradio as gr
     return [
         name1,
         gr.update(visible=True)
     ]
 
 
-def jinja_template_from_old_format(params, verbose=False):
-    MASTER_TEMPLATE = """
-{%- set ns = namespace(found=false) -%}
-{%- for message in messages -%}
-    {%- if message['role'] == 'system' -%}
-        {%- set ns.found = true -%}
-    {%- endif -%}
-{%- endfor -%}
-{%- if not ns.found -%}
-    {{- '<|PRE-SYSTEM|>' + '<|SYSTEM-MESSAGE|>' + '<|POST-SYSTEM|>' -}}
-{%- endif %}
-{%- for message in messages %}
-    {%- if message['role'] == 'system' -%}
-        {{- '<|PRE-SYSTEM|>' + message['content'] + '<|POST-SYSTEM|>' -}}
-    {%- else -%}
-        {%- if message['role'] == 'user' -%}
-            {{-'<|PRE-USER|>' + message['content'] + '<|POST-USER|>'-}}
-        {%- else -%}
-            {{-'<|PRE-ASSISTANT|>' + message['content'] + '<|POST-ASSISTANT|>' -}}
-        {%- endif -%}
-    {%- endif -%}
-{%- endfor -%}
-{%- if add_generation_prompt -%}
-    {{-'<|PRE-ASSISTANT-GENERATE|>'-}}
-{%- endif -%}
-"""
-
-    if 'context' in params and '<|system-message|>' in params['context']:
-        pre_system = params['context'].split('<|system-message|>')[0]
-        post_system = params['context'].split('<|system-message|>')[1]
-    else:
-        pre_system = ''
-        post_system = ''
-
-    pre_user = params['turn_template'].split('<|user-message|>')[0].replace('<|user|>', params['user'])
-    post_user = params['turn_template'].split('<|user-message|>')[1].split('<|bot|>')[0]
-
-    pre_assistant = '<|bot|>' + params['turn_template'].split('<|bot-message|>')[0].split('<|bot|>')[1]
-    pre_assistant = pre_assistant.replace('<|bot|>', params['bot'])
-    post_assistant = params['turn_template'].split('<|bot-message|>')[1]
-
-    def preprocess(string):
-        return string.replace('\n', '\\n').replace('\'', '\\\'')
-
-    pre_system = preprocess(pre_system)
-    post_system = preprocess(post_system)
-    pre_user = preprocess(pre_user)
-    post_user = preprocess(post_user)
-    pre_assistant = preprocess(pre_assistant)
-    post_assistant = preprocess(post_assistant)
-
-    if verbose:
-        print(
-            '\n',
-            repr(pre_system) + '\n',
-            repr(post_system) + '\n',
-            repr(pre_user) + '\n',
-            repr(post_user) + '\n',
-            repr(pre_assistant) + '\n',
-            repr(post_assistant) + '\n',
-        )
-
-    result = MASTER_TEMPLATE
-    if 'system_message' in params:
-        result = result.replace('<|SYSTEM-MESSAGE|>', preprocess(params['system_message']))
-    else:
-        result = result.replace('<|SYSTEM-MESSAGE|>', '')
-
-    result = result.replace('<|PRE-SYSTEM|>', pre_system)
-    result = result.replace('<|POST-SYSTEM|>', post_system)
-    result = result.replace('<|PRE-USER|>', pre_user)
-    result = result.replace('<|POST-USER|>', post_user)
-    result = result.replace('<|PRE-ASSISTANT|>', pre_assistant)
-    result = result.replace('<|PRE-ASSISTANT-GENERATE|>', pre_assistant.rstrip(' '))
-    result = result.replace('<|POST-ASSISTANT|>', post_assistant)
-
-    result = result.strip()
-
-    return result
-
-
 def my_yaml_output(data):
     '''
     pyyaml is very inconsistent with multiline strings.
@@ -2002,6 +1909,7 @@ def handle_unique_id_select(state):
 
 
 def handle_start_new_chat_click(state):
+    import gradio as gr
     history = start_new_chat(state)
     histories = find_all_histories_with_first_prompts(state)
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
@@ -2017,6 +1925,7 @@ def handle_start_new_chat_click(state):
 
 
 def handle_delete_chat_confirm_click(state):
+    import gradio as gr
     filtered_histories = find_all_histories_with_first_prompts(state)
     filtered_ids = [h[1] for h in filtered_histories]
     index = str(filtered_ids.index(state['unique_id']))
@@ -2037,6 +1946,7 @@ def handle_delete_chat_confirm_click(state):
 
 
 def handle_branch_chat_click(state):
+    import gradio as gr
     branch_from_index = state['branch_index']
     if branch_from_index == -1:
         history = state['history']
@@ -2148,6 +2058,7 @@ def handle_navigate_version_click(state):
 
 
 def handle_rename_chat_click():
+    import gradio as gr
     return [
         gr.update(value="My New Chat"),
         gr.update(visible=True),
@@ -2155,6 +2066,7 @@ def handle_rename_chat_click():
 
 
 def handle_rename_chat_confirm(rename_to, state):
+    import gradio as gr
     rename_history(state['unique_id'], rename_to, state['character_menu'], state['mode'])
     histories = find_all_histories_with_first_prompts(state)
 
@@ -2165,11 +2077,13 @@ def handle_rename_chat_confirm(rename_to, state):
 
 
 def handle_search_chat_change(state):
+    import gradio as gr
     histories = find_all_histories_with_first_prompts(state)
     return gr.update(choices=histories)
 
 
 def handle_upload_chat_history(load_chat_history, state):
+    import gradio as gr
     history = start_new_chat(state)
     history = load_history_json(load_chat_history, history)
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
@@ -2192,6 +2106,7 @@ def handle_upload_chat_history(load_chat_history, state):
 
 
 def handle_character_menu_change(state):
+    import gradio as gr
     name1, name2, picture, greeting, context = load_character(state['character_menu'], state['name1'], state['name2'])
 
     state['name1'] = name1
@@ -2244,6 +2159,7 @@ def handle_character_picture_change(picture_path):
 
 
 def handle_mode_change(state):
+    import gradio as gr
     history, loaded_unique_id = load_latest_history(state)
     histories = find_all_histories_with_first_prompts(state)
 
@@ -2270,6 +2186,7 @@ def handle_mode_change(state):
 
 
 def handle_save_character_click(name2):
+    import gradio as gr
     return [
         name2,
         gr.update(visible=True)
@@ -2277,6 +2194,7 @@ def handle_save_character_click(name2):
 
 
 def handle_load_template_click(instruction_template):
+    from modules.models_settings import load_instruction_template
     output = load_instruction_template(instruction_template)
     return [
         output,
@@ -2285,6 +2203,7 @@ def handle_load_template_click(instruction_template):
 
 
 def handle_save_template_click(instruction_template_str):
+    import gradio as gr
     contents = generate_instruction_template_yaml(instruction_template_str)
     return [
         "My Template.yaml",
@@ -2295,6 +2214,7 @@ def handle_save_template_click(instruction_template_str):
 
 
 def handle_delete_template_click(template):
+    import gradio as gr
     return [
         f"{template}.yaml",
         str(shared.user_data_dir / 'instruction-templates') + '/',
@@ -2310,6 +2230,7 @@ def handle_your_picture_change(picture, state):
 
 
 def handle_send_instruction_click(state):
+    import gradio as gr
     state['mode'] = 'instruct'
     state['history'] = {'internal': [], 'visible': [], 'metadata': {}}
 
@@ -2322,6 +2243,7 @@ def handle_send_instruction_click(state):
 
 
 def handle_send_chat_click(state):
+    import gradio as gr
     output = generate_chat_prompt("", state, _continue=True)
 
     if state["show_two_notebook_columns"]:
diff --git a/modules/extensions.py b/modules/extensions.py
index dd3278822d..e58a9a4ce0 100644
--- a/modules/extensions.py
+++ b/modules/extensions.py
@@ -6,8 +6,6 @@
 from inspect import signature
 from pathlib import Path
 
-import gradio as gr
-
 import modules.shared as shared
 from modules.logging_colors import logger
 
@@ -214,6 +212,7 @@ def _apply_custom_js():
 
 
 def create_extensions_block():
+    import gradio as gr
     to_display = []
     for extension, name in iterator():
         if hasattr(extension, "ui") and not (hasattr(extension, 'params') and extension.params.get('is_tab', False)):
@@ -228,6 +227,7 @@ def create_extensions_block():
 
 
 def create_extensions_tabs():
+    import gradio as gr
     for extension, name in iterator():
         if hasattr(extension, "ui") and (hasattr(extension, 'params') and extension.params.get('is_tab', False)):
             display_name = getattr(extension, 'params', {}).get('display_name', name)
diff --git a/modules/loaders.py b/modules/loaders.py
index 64de3dde52..22ee5ed91d 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -1,8 +1,6 @@
 import functools
 from collections import OrderedDict
 
-import gradio as gr
-
 loaders_and_params = OrderedDict({
     'llama.cpp': [
         'gpu_layers',
@@ -276,6 +274,7 @@ def list_all_samplers():
 
 
 def blacklist_samplers(loader, dynamic_temperature):
+    import gradio as gr
     all_samplers = list_all_samplers()
     output = []
 
@@ -301,7 +300,61 @@ def get_all_params():
     return sorted(all_params)
 
 
+def list_model_elements():
+    return [
+        'filter_by_loader',
+        'loader',
+        'cpu_memory',
+        'gpu_layers',
+        'fit_target',
+        'cpu_moe',
+        'threads',
+        'threads_batch',
+        'batch_size',
+        'ubatch_size',
+        'ctx_size',
+        'cache_type',
+        'tensor_split',
+        'extra_flags',
+        'streaming_llm',
+        'gpu_split',
+        'alpha_value',
+        'rope_freq_base',
+        'compress_pos_emb',
+        'compute_dtype',
+        'quant_type',
+        'load_in_8bit',
+        'load_in_4bit',
+        'attn_implementation',
+        'cpu',
+        'disk',
+        'row_split',
+        'no_kv_offload',
+        'no_mmap',
+        'mlock',
+        'numa',
+        'parallel',
+        'use_double_quant',
+        'bf16',
+        'enable_tp',
+        'tp_backend',
+        'cfg_cache',
+        'no_use_fast',
+        'model_draft',
+        'draft_max',
+        'gpu_layers_draft',
+        'device_draft',
+        'ctx_size_draft',
+        'spec_type',
+        'spec_ngram_size_n',
+        'spec_ngram_size_m',
+        'spec_ngram_min_hits',
+        'mmproj',
+    ]
+
+
 def make_loader_params_visible(loader):
+    import gradio as gr
     params = []
     all_params = get_all_params()
     if loader in loaders_and_params:
diff --git a/modules/models_settings.py b/modules/models_settings.py
index 472871ce09..5e69b60e3f 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -4,10 +4,9 @@
 from math import floor
 from pathlib import Path
 
-import gradio as gr
 import yaml
 
-from modules import chat, loaders, metadata_gguf, shared, ui
+from modules import loaders, metadata_gguf, shared
 from modules.logging_colors import logger
 from modules.utils import resolve_model_path
 
@@ -199,7 +198,7 @@ def get_model_metadata(model):
 
     # Load instruction template if defined by name rather than by value
     if model_settings['instruction_template'] != 'Custom (obtained from model metadata)':
-        model_settings['instruction_template_str'] = chat.load_instruction_template(model_settings['instruction_template'])
+        model_settings['instruction_template_str'] = load_instruction_template(model_settings['instruction_template'])
 
     return model_settings
 
@@ -228,7 +227,7 @@ def update_model_parameters(state, initial=False):
     '''
     UI: update the command-line arguments based on the interface values
     '''
-    elements = ui.list_model_elements()  # the names of the parameters
+    elements = loaders.list_model_elements()  # the names of the parameters
 
     for i, element in enumerate(elements):
         if element not in state:
@@ -248,6 +247,7 @@ def apply_model_settings_to_state(model, state):
     '''
     UI: update the state variable with the model settings
     '''
+    import gradio as gr
     model_settings = get_model_metadata(model)
     if 'loader' in model_settings:
         loader = model_settings.pop('loader')
@@ -290,7 +290,7 @@ def save_model_settings(model, state):
     if model_regex not in user_config:
         user_config[model_regex] = {}
 
-    for k in ui.list_model_elements():
+    for k in loaders.list_model_elements():
         if k == 'loader' or k in loaders.loaders_and_params[state['loader']]:
             user_config[model_regex][k] = state[k]
 
@@ -419,3 +419,102 @@ def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type):
 
     vram_usage = estimate_vram(model, gpu_layers, ctx_size, cache_type)
     return f"
            Estimated VRAM to load the model: {vram_usage:.0f} MiB
            "
+
+
+def load_instruction_template(template):
+    if template == 'None':
+        return ''
+
+    for filepath in [shared.user_data_dir / 'instruction-templates' / f'{template}.yaml', shared.user_data_dir / 'instruction-templates' / 'Alpaca.yaml']:
+        if filepath.exists():
+            break
+    else:
+        return ''
+
+    file_contents = open(filepath, 'r', encoding='utf-8').read()
+    data = yaml.safe_load(file_contents)
+    if 'instruction_template' in data:
+        return data['instruction_template']
+    else:
+        return _jinja_template_from_old_format(data)
+
+
+def _jinja_template_from_old_format(params, verbose=False):
+    MASTER_TEMPLATE = """
+{%- set ns = namespace(found=false) -%}
+{%- for message in messages -%}
+    {%- if message['role'] == 'system' -%}
+        {%- set ns.found = true -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- if not ns.found -%}
+    {{- '<|PRE-SYSTEM|>' + '<|SYSTEM-MESSAGE|>' + '<|POST-SYSTEM|>' -}}
+{%- endif %}
+{%- for message in messages %}
+    {%- if message['role'] == 'system' -%}
+        {{- '<|PRE-SYSTEM|>' + message['content'] + '<|POST-SYSTEM|>' -}}
+    {%- else -%}
+        {%- if message['role'] == 'user' -%}
+            {{-'<|PRE-USER|>' + message['content'] + '<|POST-USER|>'-}}
+        {%- else -%}
+            {{-'<|PRE-ASSISTANT|>' + message['content'] + '<|POST-ASSISTANT|>' -}}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{-'<|PRE-ASSISTANT-GENERATE|>'-}}
+{%- endif -%}
+"""
+
+    if 'context' in params and '<|system-message|>' in params['context']:
+        pre_system = params['context'].split('<|system-message|>')[0]
+        post_system = params['context'].split('<|system-message|>')[1]
+    else:
+        pre_system = ''
+        post_system = ''
+
+    pre_user = params['turn_template'].split('<|user-message|>')[0].replace('<|user|>', params['user'])
+    post_user = params['turn_template'].split('<|user-message|>')[1].split('<|bot|>')[0]
+
+    pre_assistant = '<|bot|>' + params['turn_template'].split('<|bot-message|>')[0].split('<|bot|>')[1]
+    pre_assistant = pre_assistant.replace('<|bot|>', params['bot'])
+    post_assistant = params['turn_template'].split('<|bot-message|>')[1]
+
+    def preprocess(string):
+        return string.replace('\n', '\\n').replace('\'', '\\\'')
+
+    pre_system = preprocess(pre_system)
+    post_system = preprocess(post_system)
+    pre_user = preprocess(pre_user)
+    post_user = preprocess(post_user)
+    pre_assistant = preprocess(pre_assistant)
+    post_assistant = preprocess(post_assistant)
+
+    if verbose:
+        print(
+            '\n',
+            repr(pre_system) + '\n',
+            repr(post_system) + '\n',
+            repr(pre_user) + '\n',
+            repr(post_user) + '\n',
+            repr(pre_assistant) + '\n',
+            repr(post_assistant) + '\n',
+        )
+
+    result = MASTER_TEMPLATE
+    if 'system_message' in params:
+        result = result.replace('<|SYSTEM-MESSAGE|>', preprocess(params['system_message']))
+    else:
+        result = result.replace('<|SYSTEM-MESSAGE|>', '')
+
+    result = result.replace('<|PRE-SYSTEM|>', pre_system)
+    result = result.replace('<|POST-SYSTEM|>', post_system)
+    result = result.replace('<|PRE-USER|>', pre_user)
+    result = result.replace('<|POST-USER|>', post_user)
+    result = result.replace('<|PRE-ASSISTANT|>', pre_assistant)
+    result = result.replace('<|PRE-ASSISTANT-GENERATE|>', pre_assistant.rstrip(' '))
+    result = result.replace('<|POST-ASSISTANT|>', post_assistant)
+
+    result = result.strip()
+
+    return result
diff --git a/modules/ui.py b/modules/ui.py
index 70e929f20c..abbfde49d0 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -120,58 +120,8 @@
 
 
 def list_model_elements():
-    elements = [
-        'filter_by_loader',
-        'loader',
-        'cpu_memory',
-        'gpu_layers',
-        'fit_target',
-        'cpu_moe',
-        'threads',
-        'threads_batch',
-        'batch_size',
-        'ubatch_size',
-        'ctx_size',
-        'cache_type',
-        'tensor_split',
-        'extra_flags',
-        'streaming_llm',
-        'gpu_split',
-        'alpha_value',
-        'rope_freq_base',
-        'compress_pos_emb',
-        'compute_dtype',
-        'quant_type',
-        'load_in_8bit',
-        'load_in_4bit',
-        'attn_implementation',
-        'cpu',
-        'disk',
-        'row_split',
-        'no_kv_offload',
-        'no_mmap',
-        'mlock',
-        'numa',
-        'parallel',
-        'use_double_quant',
-        'bf16',
-        'enable_tp',
-        'tp_backend',
-        'cfg_cache',
-        'no_use_fast',
-        'model_draft',
-        'draft_max',
-        'gpu_layers_draft',
-        'device_draft',
-        'ctx_size_draft',
-        'spec_type',
-        'spec_ngram_size_n',
-        'spec_ngram_size_m',
-        'spec_ngram_min_hits',
-        'mmproj',
-    ]
-
-    return elements
+    from modules.loaders import list_model_elements
+    return list_model_elements()
 
 
 def list_interface_input_elements():
diff --git a/server.py b/server.py
index ff2d1db2d6..73f190b602 100644
--- a/server.py
+++ b/server.py
@@ -1,58 +1,20 @@
-import os
-import shutil
-import warnings
-from pathlib import Path
-
-from modules import shared, ui  # ui must be imported early to avoid circular imports
-from modules.image_models import load_image_model
-from modules.logging_colors import logger
-from modules.prompts import load_prompt
-
-# Set up Gradio temp directory path
-gradio_temp_path = shared.user_data_dir / 'cache' / 'gradio'
-shutil.rmtree(gradio_temp_path, ignore_errors=True)
-gradio_temp_path.mkdir(parents=True, exist_ok=True)
-
-# Set environment variables
-os.environ.update({
-    'GRADIO_ANALYTICS_ENABLED': 'False',
-    'BITSANDBYTES_NOWELCOME': '1',
-    'GRADIO_TEMP_DIR': str(gradio_temp_path)
-})
-
-warnings.filterwarnings('ignore', category=UserWarning, message='TypedStorage is deprecated')
-warnings.filterwarnings('ignore', category=UserWarning, message='Using the update method is deprecated')
-warnings.filterwarnings('ignore', category=UserWarning, message='Field "model_name" has conflict')
-warnings.filterwarnings('ignore', category=UserWarning, message='The value passed into gr.Dropdown()')
-warnings.filterwarnings('ignore', category=UserWarning, message='Field "model_names" has conflict')
-
-import gradio as gr
-
 import os
 import signal
 import sys
 import time
+import warnings
 from functools import partial
+from pathlib import Path
 from threading import Lock, Thread
 
 import yaml
 
+from modules import shared, utils
+from modules.image_models import load_image_model
+from modules.logging_colors import logger
+from modules.prompts import load_prompt
+
 import modules.extensions as extensions_module
-from modules import (
-    training,
-    ui,
-    ui_chat,
-    ui_default,
-    ui_file_saving,
-    ui_image_generation,
-    ui_model_menu,
-    ui_notebook,
-    ui_parameters,
-    ui_session,
-    utils
-)
-from modules.chat import generate_pfp_cache
-from modules.extensions import apply_extensions
 from modules.LoRA import add_lora_to_model
 from modules.models import load_model, unload_model_if_idle
 from modules.models_settings import (
@@ -61,7 +23,13 @@
     update_model_parameters
 )
 from modules.shared import do_cmd_flags_warnings
-from modules.utils import gradio
+
+os.environ['BITSANDBYTES_NOWELCOME'] = '1'
+
+warnings.filterwarnings('ignore', category=UserWarning, message='TypedStorage is deprecated')
+warnings.filterwarnings('ignore', category=UserWarning, message='Using the update method is deprecated')
+warnings.filterwarnings('ignore', category=UserWarning, message='Field "model_name" has conflict')
+warnings.filterwarnings('ignore', category=UserWarning, message='Field "model_names" has conflict')
 
 
 def signal_handler(sig, frame):
@@ -83,6 +51,37 @@ def signal_handler(sig, frame):
 
 def create_interface():
 
+    import shutil
+
+    import gradio as gr
+
+    from modules import (
+        training,
+        ui,
+        ui_chat,
+        ui_default,
+        ui_file_saving,
+        ui_image_generation,
+        ui_model_menu,
+        ui_notebook,
+        ui_parameters,
+        ui_session,
+    )
+    from modules.chat import generate_pfp_cache
+    from modules.extensions import apply_extensions
+    from modules.utils import gradio
+
+    warnings.filterwarnings('ignore', category=UserWarning, message='The value passed into gr.Dropdown()')
+
+    # Set up Gradio temp directory path
+    gradio_temp_path = shared.user_data_dir / 'cache' / 'gradio'
+    shutil.rmtree(gradio_temp_path, ignore_errors=True)
+    gradio_temp_path.mkdir(parents=True, exist_ok=True)
+    os.environ.update({
+        'GRADIO_ANALYTICS_ENABLED': 'False',
+        'GRADIO_TEMP_DIR': str(gradio_temp_path)
+    })
+
     title = 'Text Generation Web UI'
 
     # Password authentication

From 7f485274ebc01364b19e45df1d3e27683f7f63a2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Mar 2026 23:55:51 -0300
Subject: [PATCH 1299/1701] Fix ExLlamaV3 EOS handling, load order, and
 perplexity evaluation

- Use config.eos_token_id_list for all EOS tokens as stop conditions
  (fixes models like Llama-3 that define multiple EOS token IDs)
- Load vision/draft models before main model so autosplit accounts
  for their VRAM usage
- Fix loss computation in ExLlamav3_HF: use cache across chunks so
  sequences longer than 2048 tokens get correct perplexity values
---
 modules/exllamav3.py    | 39 ++++++++++++++++++++++-----------------
 modules/exllamav3_hf.py | 10 +++++++---
 2 files changed, 29 insertions(+), 20 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 9ea38432b9..aeb68564da 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -158,8 +158,21 @@ def from_pretrained(cls, path_to_model):
             load_params['tensor_p'] = True
             load_params['tp_backend'] = shared.args.tp_backend
 
-        model.load(**load_params)
-        tokenizer = Tokenizer.from_config(config)
+        # Load vision and draft before the main model so autosplit
+        # accounts for their VRAM usage.
+
+        # Load vision model component (ExLlamaV3 native)
+        vision_model = None
+        if "vision_config" in config.config_dict:
+            logger.info("Vision component detected in model config. Attempting to load...")
+            try:
+                vision_model = Model.from_config(config, component="vision")
+                vision_model.load(progressbar=True)
+                logger.info("Vision model loaded successfully.")
+            except Exception as e:
+                logger.warning(f"Vision model loading failed (multimodal disabled): {e}")
+        else:
+            logger.info("No vision component in model config. Skipping multimodal setup.")
 
         # Initialize draft model for speculative decoding
         draft_model = None
@@ -185,18 +198,9 @@ def from_pretrained(cls, path_to_model):
                 draft_model.load(**draft_load_params)
                 logger.info(f"Draft model loaded successfully. Max speculative tokens: {shared.args.draft_max}")
 
-        # Load vision model component (ExLlamaV3 native)
-        vision_model = None
-        if "vision_config" in config.config_dict:
-            logger.info("Vision component detected in model config. Attempting to load...")
-            try:
-                vision_model = Model.from_config(config, component="vision")
-                vision_model.load(progressbar=True)
-                logger.info("Vision model loaded successfully.")
-            except Exception as e:
-                logger.warning(f"Vision model loading failed (multimodal disabled): {e}")
-        else:
-            logger.info("No vision component in model config. Skipping multimodal setup.")
+        # Load main model last
+        model.load(**load_params)
+        tokenizer = Tokenizer.from_config(config)
 
         generator = Generator(
             model=model,
@@ -379,11 +383,12 @@ def custom_sort_key(sampler_obj):
         else:
             max_new_tokens = state['max_new_tokens']
 
-        # Get stop conditions
+        # Use full EOS token list from config (may contain multiple IDs)
         stop_conditions = []
         if not state['ban_eos_token']:
-            if hasattr(self.tokenizer, 'eos_token_id') and self.tokenizer.eos_token_id is not None:
-                stop_conditions.append(self.tokenizer.eos_token_id)
+            for eos_id in self.config.eos_token_id_list:
+                if eos_id is not None:
+                    stop_conditions.append(eos_id)
 
         seed = state.get('seed', -1)
         job = Job(
diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index b4b6ad207d..d3c1cb9053 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -201,19 +201,23 @@ def __call__(self, *args, **kwargs):
                 }
             ).to(input_ids.device).float()
         else:
-            # When processing with labels, handle as a complete sequence
-            # Process in chunks if the number of tokens is large
+            # Labels path: use cache for cross-chunk attention.
             tokens_to_process = seq_tensor
             all_logits = None
+            current_len = 0
 
             for i in range(0, tokens_to_process.shape[0], max_chunk_size):
                 chunk = tokens_to_process[i:i + max_chunk_size]
                 chunk_logits = self.ex_model.forward(
                     input_ids=chunk.view(1, -1),
                     params={
-                        "attn_mode": "flash_attn_nc",
+                        "attn_mode": "flash_attn",
+                        "cache": ex_cache,
+                        "past_len": current_len,
+                        "batch_shape": (1, self.max_tokens),
                     }
                 ).float()
+                current_len += chunk.shape[0]
 
                 if all_logits is None:
                     all_logits = chunk_logits

From 83b7e47d77c74da1c9edd7b242c3d89dd708e39e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Mar 2026 20:12:54 -0700
Subject: [PATCH 1300/1701] Update README

---
 README.md | 56 +++++++++++++++++--------------------------------------
 1 file changed, 17 insertions(+), 39 deletions(-)

diff --git a/README.md b/README.md
index 7831ef650f..d5d0a3ce26 100644
--- a/README.md
+++ b/README.md
@@ -13,7 +13,7 @@
 
 # Text Generation Web UI
 
-A Gradio web UI for running Large Language Models locally. 100% private, offline, and free.
+Run large language models locally with full privacy. Supports text generation, vision, image generation, training, tool-calling, and more — across multiple backends including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). 100% offline, zero telemetry.
 
 [Try the Deep Reason extension](https://oobabooga.gumroad.com/l/deep_reason)
 
@@ -23,22 +23,21 @@ A Gradio web UI for running Large Language Models locally. 100% private, offline
 
 ## Features
 
-- Supports multiple local text generation backends, including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) (the latter via its own [Dockerfile](https://github.com/oobabooga/text-generation-webui/blob/main/docker/TensorRT-LLM/Dockerfile)).
-- Easy setup: Choose between **portable builds** (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or the one-click installer that creates a self-contained `installer_files` directory.
+- **Multiple backends**: [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). Switch between backends and models without restarting.
+- **Vision (multimodal)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
+- **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
+- **Training**: Fine-tune LoRAs on multi-turn chat or raw text datasets. Supports resuming interrupted runs ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/05-%E2%80%90-Training-Tab)).
+- **OpenAI-compatible API**: Chat and Completions endpoints with tool-calling support — use as a local drop-in replacement for the OpenAI API ([examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples)).
+- **Web search**: Search the internet with LLM-generated queries to add context to conversations.
+- **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
+- **Easy setup**: [Portable builds](https://github.com/oobabooga/text-generation-webui/releases) (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or a one-click installer for the full feature set.
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - `instruct` mode for instruction-following (like ChatGPT), and `chat-instruct`/`chat` modes for talking to custom characters. Prompts are automatically formatted with Jinja2 templates.
-- **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
-- **Vision (multimodal models)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
-- **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
-- **Web search**: Optionally search the internet with LLM-generated queries to add context to the conversation.
-- **Training**: Fine-tune LoRAs on multi-turn chat or raw text datasets. Easy to use, good defaults, and supports resuming interrupted runs ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/05-%E2%80%90-Training-Tab)).
 - Edit messages, navigate between message versions, and branch conversations at any point.
-- Switch between different models in the UI without restarting.
 - Free-form text generation in the Notebook tab without being limited to chat turns.
 - Multiple sampling parameters and generation options for sophisticated text generation control.
 - Aesthetic UI with dark and light themes.
 - Syntax highlighting for code blocks and LaTeX rendering for mathematical expressions.
-- OpenAI-compatible API with Chat and Completions endpoints, including tool-calling support – see [examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples).
 - Extension support, with numerous built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details.
 
 ## How to install
@@ -430,22 +429,14 @@ API generation defaults:
 
 ## Downloading models
 
-Models should be placed in the folder `text-generation-webui/user_data/models`. They are usually downloaded from [Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads&search=gguf).
-
-To check if a GGUF model will fit in your hardware before downloading it, you can use this tool I created:
+Download a GGUF model file from [Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads&search=gguf) and place it in the `user_data/models` folder. That's it — the UI will detect it automatically. You can also download models directly from the Model tab in the UI.
 
-[Accurate GGUF VRAM Calculator](https://huggingface.co/spaces/oobabooga/accurate-gguf-vram-calculator)
+Not sure what will fit your GPU? Use the [VRAM Calculator](https://huggingface.co/spaces/oobabooga/accurate-gguf-vram-calculator).
 
-* GGUF models are a single file and should be placed directly into `user_data/models`. Example:
-
-```
-text-generation-webui
-└── user_data
-    └── models
-        └── llama-2-13b-chat.Q4_K_M.gguf
-```
+
            
+Other model types (Transformers, EXL3)
 
-* The remaining model types (like 16-bit Transformers models and EXL3 models) are made of several files and must be placed in a subfolder. Example:
+Models that consist of multiple files (like 16-bit Transformers models and EXL3 models) should be placed in a subfolder inside `user_data/models`:
 
 ```
 text-generation-webui
@@ -455,31 +446,18 @@ text-generation-webui
             ├── config.json
             ├── generation_config.json
             ├── model-00001-of-00004.safetensors
-            ├── model-00002-of-00004.safetensors
-            ├── model-00003-of-00004.safetensors
-            ├── model-00004-of-00004.safetensors
-            ├── model.safetensors.index.json
-            ├── special_tokens_map.json
+            ├── ...
             ├── tokenizer_config.json
             └── tokenizer.json
 ```
 
-In both cases, you can use the "Model" tab of the UI to download the model from Hugging Face automatically. It is also possible to download it via the command-line with:
-
-```
-python download-model.py organization/model
-```
-
-Run `python download-model.py --help` to see all the options.
+These formats require the one-click installer (not the portable build).
+
            
 
 ## Documentation
 
 https://github.com/oobabooga/text-generation-webui/wiki
 
-## Google Colab notebook
-
-https://colab.research.google.com/github/oobabooga/text-generation-webui/blob/main/Colab-TextGen-GPU.ipynb
-
 ## Community
 
 https://www.reddit.com/r/Oobabooga/

From 3b7193265812dba17ab757b3853b78dae3d9ba2e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Mar 2026 20:18:09 -0700
Subject: [PATCH 1301/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index d5d0a3ce26..8dc09f32c0 100644
--- a/README.md
+++ b/README.md
@@ -429,7 +429,7 @@ API generation defaults:
 
 ## Downloading models
 
-Download a GGUF model file from [Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads&search=gguf) and place it in the `user_data/models` folder. That's it — the UI will detect it automatically. You can also download models directly from the Model tab in the UI.
+Download a GGUF model file from [Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads&search=gguf) and place it in the `user_data/models` folder. That's it — the UI will detect it automatically.
 
 Not sure what will fit your GPU? Use the [VRAM Calculator](https://huggingface.co/spaces/oobabooga/accurate-gguf-vram-calculator).
 

From 15792c3cb8095c788c6737fdced34ccac451aff6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Mar 2026 20:31:05 -0700
Subject: [PATCH 1302/1701] Update ExLlamaV3 to 0.0.24

---
 requirements/full/requirements.txt | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index eaf34fa8da..a3d4d1e6a9 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,7 +42,7 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.23/exllamav3-0.0.23+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.23/exllamav3-0.0.23+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.24/exllamav3-0.0.24+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.24/exllamav3-0.0.24+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"

From c604ca66de4a25fc64b6c452ffb0cc611ead3d0a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Mar 2026 21:36:04 -0700
Subject: [PATCH 1303/1701] Update the --multi-user warning

---
 modules/shared.py | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 080874ec4c..50dfc7abd7 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -47,7 +47,7 @@
 # Basic settings
 group = parser.add_argument_group('Basic settings')
 group.add_argument('--user-data-dir', type=str, default=str(user_data_dir), help='Path to the user data directory. Default: auto-detected.')
-group.add_argument('--multi-user', action='store_true', help='Multi-user mode. Chat histories are not saved or automatically loaded. Warning: this is likely not safe for sharing publicly.')
+group.add_argument('--multi-user', action='store_true', help='Multi-user mode. Chat histories are not saved or automatically loaded. Best suited for small trusted teams.')
 group.add_argument('--model', type=str, help='Name of the model to load by default.')
 group.add_argument('--lora', type=str, nargs='+', help='The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.')
 group.add_argument('--model-dir', type=str, default=str(user_data_dir / 'models'), help='Path to directory with all the models.')
@@ -396,8 +396,15 @@ def do_cmd_flags_warnings():
             logger.warning("The gradio \"share link\" feature uses a proprietary executable to create a reverse tunnel. Use it with care.")
         if any((args.listen, args.share)) and not any((args.gradio_auth, args.gradio_auth_path)):
             logger.warning("\nYou are potentially exposing the web UI to the entire internet without any access password.\nYou can create one with the \"--gradio-auth\" flag like this:\n\n--gradio-auth username:password\n\nMake sure to replace username:password with your own.")
-            if args.multi_user:
-                logger.warning('\nThe multi-user mode is highly experimental and should not be shared publicly.')
+    if args.multi_user:
+        logger.warning(
+            'Multi-user mode is enabled. Known limitations:'
+            '\n- The Stop button stops generation for all users, not just you.'
+            '\n- Chat history is not saved and will be lost on page refresh.'
+            '\n- Only one user can generate at a time unless using a parallel-capable backend (e.g. llama.cpp with --parallel N for N > 1, or ExLlamaV3).'
+            '\n\nThis mode works best for small trusted teams.'
+            '\n\nDo not expose publicly. Grayed-out actions can easily be bypassed client-side.\n'
+        )
 
 
 def apply_image_model_cli_overrides():

From 307c085d1b7b3c0f60d40276c50f277b2ed1a0f4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 9 Mar 2026 21:44:53 -0700
Subject: [PATCH 1304/1701] Minor warning change

---
 modules/shared.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/shared.py b/modules/shared.py
index 50dfc7abd7..dbd805a12d 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -395,7 +395,7 @@ def do_cmd_flags_warnings():
         if args.share:
             logger.warning("The gradio \"share link\" feature uses a proprietary executable to create a reverse tunnel. Use it with care.")
         if any((args.listen, args.share)) and not any((args.gradio_auth, args.gradio_auth_path)):
-            logger.warning("\nYou are potentially exposing the web UI to the entire internet without any access password.\nYou can create one with the \"--gradio-auth\" flag like this:\n\n--gradio-auth username:password\n\nMake sure to replace username:password with your own.")
+            logger.warning("You are potentially exposing the web UI to the entire internet without any access password.\nYou can create one with the \"--gradio-auth\" flag like this:\n\n--gradio-auth username:password\n\nMake sure to replace username:password with your own.")
     if args.multi_user:
         logger.warning(
             'Multi-user mode is enabled. Known limitations:'

From 6ec4ca8b102725d2ebb810c07ad668179835a567 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Mar 2026 09:56:07 -0300
Subject: [PATCH 1305/1701] Add missing custom_token_bans to llama.cpp and
 reasoning_effort to ExLlamav3

---
 modules/loaders.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/modules/loaders.py b/modules/loaders.py
index 22ee5ed91d..d2ebdbc358 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -205,6 +205,7 @@ def transformers_samplers():
         'ban_eos_token',
         'add_bos_token',
         'enable_thinking',
+        'reasoning_effort',
         'seed',
         'skip_special_tokens',
     },
@@ -241,6 +242,7 @@ def transformers_samplers():
         'reasoning_effort',
         'seed',
         'sampler_priority',
+        'custom_token_bans',
         'dry_sequence_breakers',
         'grammar_string',
         'grammar_file_row',

From 8aeaa763659edc55a0159658e5eafd72552344e5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Mar 2026 10:41:15 -0300
Subject: [PATCH 1306/1701] Forward logit_bias, logprobs, and n to llama.cpp
 backend

- Forward logit_bias and logprobs natively to llama.cpp
- Support n>1 completions with seed increment for diversity
- Fix logprobs returning empty dict when not requested
---
 extensions/openai/completions.py | 98 +++++++++++++++++++++++---------
 extensions/openai/script.py      |  6 ++
 extensions/openai/typing.py      |  2 +-
 modules/llama_cpp_server.py      | 20 ++++++-
 4 files changed, 95 insertions(+), 31 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 8ba031c110..d70e69e6dc 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -37,6 +37,24 @@ def load_chat_template_file(filepath):
     return text
 
 
+def get_logprobs_from_llama_cpp():
+    """Read logprobs captured from the llama.cpp server response."""
+    if not hasattr(shared.model, 'last_completion_probabilities') or not shared.model.last_completion_probabilities:
+        return None
+
+    # Convert llama.cpp format to {token: logprob} dict
+    # llama.cpp returns: [{"token": "text", "logprob": -0.5, "top_logprobs": [{"token": "t", "logprob": -0.1}, ...]}, ...]
+    result = {}
+    for entry in shared.model.last_completion_probabilities:
+        top = entry.get('top_logprobs', entry.get('top_probs', []))
+        for item in top:
+            token = item.get('token', '')
+            logprob = item.get('logprob', item.get('prob', 0))
+            result[token] = logprob
+
+    return result
+
+
 def convert_logprobs_to_tiktoken(model, logprobs):
     # more problems than it's worth.
     # try:
@@ -72,6 +90,7 @@ def process_parameters(body, is_legacy=False):
         elif isinstance(body['stop'], list):
             generate_params['custom_stopping_strings'] = body['stop']
 
+    # For llama.cpp, logit_bias and logprobs are forwarded natively via prepare_payload()
     if shared.args.loader != 'llama.cpp':
         from transformers import LogitsProcessorList
 
@@ -85,13 +104,10 @@ def process_parameters(body, is_legacy=False):
         if logit_bias:  # {str: float, ...}
             logits_processor = [LogitsBiasProcessor(logit_bias)]
 
-        logprobs = None  # coming to chat eventually
-        if 'logprobs' in body:
-            logprobs = body.get('logprobs', 0)  # maybe cap at topk? don't clamp 0-5.
+        logprobs = body.get('logprobs', None)
+        if logprobs is not None and logprobs > 0:
             generate_params['logprob_proc'] = LogprobProcessor(logprobs)
             logits_processor.extend([generate_params['logprob_proc']])
-        else:
-            logprobs = None
 
         if logits_processor:  # requires logits_processor support
             generate_params['logits_processor'] = LogitsProcessorList(logits_processor)
@@ -456,6 +472,8 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_e
             logger.info(f"Found {len(raw_images)} image(s) in request.")
             generate_params['raw_images'] = raw_images
 
+    n_completions = body.get('n', 1) or 1
+
     if not stream:
         prompt_arg = body[prompt_str]
 
@@ -469,6 +487,7 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_e
         resp_list_data = []
         total_completion_token_count = 0
         total_prompt_token_count = 0
+        choice_index = 0
 
         for idx, prompt in enumerate(prompt_arg, start=0):
             if isinstance(prompt, list) and len(prompt) > 0 and isinstance(prompt[0], int):
@@ -483,31 +502,46 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_e
                         prompt = decode(prompt)[0]
 
             prefix = prompt if echo else ''
-
-            # generate reply #######################################
-            debug_msg({'prompt': prompt, 'generate_params': generate_params})
-            generator = generate_reply(prompt, generate_params, is_chat=False)
-            answer = ''
-
-            for a in generator:
-                answer = a
-
             token_count = len(encode(prompt)[0])
             total_prompt_token_count += token_count
-            completion_token_count = len(encode(answer)[0])
-            total_completion_token_count += completion_token_count
-            stop_reason = "stop"
-            if token_count + completion_token_count >= generate_params['truncation_length'] or completion_token_count >= max_tokens:
-                stop_reason = "length"
-
-            respi = {
-                "index": idx,
-                "finish_reason": stop_reason,
-                "text": prefix + answer + suffix,
-                "logprobs": {'top_logprobs': [logprob_proc.token_alternatives]} if logprob_proc else None,
-            }
 
-            resp_list_data.extend([respi])
+            original_seed = generate_params.get('seed', -1)
+            for _n in range(n_completions):
+                # Increment seed for each completion to ensure diversity (matches llama.cpp native behavior)
+                if original_seed >= 0:
+                    generate_params['seed'] = original_seed + _n
+
+                # generate reply #######################################
+                debug_msg({'prompt': prompt, 'generate_params': generate_params})
+                generator = generate_reply(prompt, generate_params, is_chat=False)
+                answer = ''
+
+                for a in generator:
+                    answer = a
+
+                completion_token_count = len(encode(answer)[0])
+                total_completion_token_count += completion_token_count
+                stop_reason = "stop"
+                if token_count + completion_token_count >= generate_params['truncation_length'] or completion_token_count >= max_tokens:
+                    stop_reason = "length"
+
+                if logprob_proc:
+                    completion_logprobs = {'top_logprobs': [logprob_proc.token_alternatives]}
+                elif shared.args.loader == 'llama.cpp':
+                    llama_logprobs = get_logprobs_from_llama_cpp()
+                    completion_logprobs = {'top_logprobs': [llama_logprobs]} if llama_logprobs else None
+                else:
+                    completion_logprobs = None
+
+                respi = {
+                    "index": choice_index,
+                    "finish_reason": stop_reason,
+                    "text": prefix + answer + suffix,
+                    "logprobs": completion_logprobs,
+                }
+
+                resp_list_data.append(respi)
+                choice_index += 1
 
         resp = {
             "id": cmpl_id,
@@ -540,6 +574,14 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_e
 
         def text_streaming_chunk(content):
             # begin streaming
+            if logprob_proc:
+                chunk_logprobs = {'top_logprobs': [logprob_proc.token_alternatives]}
+            elif shared.args.loader == 'llama.cpp':
+                llama_logprobs = get_logprobs_from_llama_cpp()
+                chunk_logprobs = {'top_logprobs': [llama_logprobs]} if llama_logprobs else None
+            else:
+                chunk_logprobs = None
+
             chunk = {
                 "id": cmpl_id,
                 "object": object_type,
@@ -549,7 +591,7 @@ def text_streaming_chunk(content):
                     "index": 0,
                     "finish_reason": None,
                     "text": content,
-                    "logprobs": {'top_logprobs': [logprob_proc.token_alternatives]} if logprob_proc else None,
+                    "logprobs": chunk_logprobs,
                 }],
             }
 
diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 7a13638d52..94c7650f2a 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -119,6 +119,12 @@ async def openai_completions(request: Request, request_data: CompletionRequest):
     is_legacy = "/generate" in path
 
     if request_data.stream:
+        if (request_data.n or 1) > 1:
+            return JSONResponse(
+                status_code=400,
+                content={"error": {"message": "n > 1 is not supported with streaming.", "type": "invalid_request_error", "param": "n", "code": None}}
+            )
+
         stop_event = threading.Event()
 
         async def generator():
diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index e48b7b6037..2156074b7f 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -109,7 +109,7 @@ class CompletionRequestParams(BaseModel):
     logit_bias: dict | None = None
     logprobs: int | None = None
     max_tokens: int | None = 512
-    n: int | None = Field(default=1, description="Unused parameter.")
+    n: int | None = Field(default=1, description="Number of completions to generate. Only supported without streaming.")
     presence_penalty: float | None = shared.args.presence_penalty
     stop: str | List[str] | None = None
     stream: bool | None = False
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 6f7cbd20f5..a3e431aca4 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -133,9 +133,20 @@ def prepare_payload(self, state):
 
             payload["samplers"] = filtered_samplers
 
+        logit_bias = []
         if state['custom_token_bans']:
-            to_ban = [[int(token_id), False] for token_id in state['custom_token_bans'].split(',')]
-            payload["logit_bias"] = to_ban
+            logit_bias.extend([[int(token_id), False] for token_id in state['custom_token_bans'].split(',')])
+
+        if state.get('logit_bias'):
+            for token_id_str, bias in state['logit_bias'].items():
+                logit_bias.append([int(token_id_str), bias])
+
+        if logit_bias:
+            payload["logit_bias"] = logit_bias
+
+        n_probs = state.get('logprobs', 0)
+        if n_probs and n_probs > 0:
+            payload["n_probs"] = n_probs
 
         return payload
 
@@ -215,6 +226,7 @@ def generate_with_streaming(self, prompt, state):
                 response.raise_for_status()  # Raise an exception for HTTP errors
 
             full_text = ""
+            self.last_completion_probabilities = []
 
             # Process the streaming response
             stop_event = state.get('stop_event')
@@ -240,6 +252,10 @@ def generate_with_streaming(self, prompt, state):
                         full_text += data['content']
                         yield full_text
 
+                    # Capture logprobs if present
+                    if 'completion_probabilities' in data:
+                        self.last_completion_probabilities.extend(data['completion_probabilities'])
+
                     # Check if generation is complete
                     if data.get('stop', False):
                         break

From 3304b57bdf050a3d5dfed375c67a8cac1c4a0eaa Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Mar 2026 11:03:00 -0300
Subject: [PATCH 1307/1701] Add native logit_bias and logprobs support for
 ExLlamav3

---
 extensions/openai/completions.py | 24 +++++------
 modules/exllamav3.py             | 70 ++++++++++++++++++++++++++++++++
 2 files changed, 82 insertions(+), 12 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index d70e69e6dc..04e644d6ed 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -37,13 +37,13 @@ def load_chat_template_file(filepath):
     return text
 
 
-def get_logprobs_from_llama_cpp():
-    """Read logprobs captured from the llama.cpp server response."""
+def get_logprobs_from_backend():
+    """Read logprobs captured from llama.cpp or ExLlamav3 native backend."""
     if not hasattr(shared.model, 'last_completion_probabilities') or not shared.model.last_completion_probabilities:
         return None
 
-    # Convert llama.cpp format to {token: logprob} dict
-    # llama.cpp returns: [{"token": "text", "logprob": -0.5, "top_logprobs": [{"token": "t", "logprob": -0.1}, ...]}, ...]
+    # Both backends store data in shared.model.last_completion_probabilities
+    # Format: [{"top_logprobs": [{"token": "text", "logprob": -0.5}, ...]}, ...]
     result = {}
     for entry in shared.model.last_completion_probabilities:
         top = entry.get('top_logprobs', entry.get('top_probs', []))
@@ -90,8 +90,8 @@ def process_parameters(body, is_legacy=False):
         elif isinstance(body['stop'], list):
             generate_params['custom_stopping_strings'] = body['stop']
 
-    # For llama.cpp, logit_bias and logprobs are forwarded natively via prepare_payload()
-    if shared.args.loader != 'llama.cpp':
+    # For llama.cpp and ExLlamav3 native, logit_bias and logprobs are forwarded natively
+    if shared.args.loader not in ('llama.cpp', 'ExLlamav3'):
         from transformers import LogitsProcessorList
 
         from modules.transformers_loader import (
@@ -527,9 +527,9 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_e
 
                 if logprob_proc:
                     completion_logprobs = {'top_logprobs': [logprob_proc.token_alternatives]}
-                elif shared.args.loader == 'llama.cpp':
-                    llama_logprobs = get_logprobs_from_llama_cpp()
-                    completion_logprobs = {'top_logprobs': [llama_logprobs]} if llama_logprobs else None
+                elif shared.args.loader in ('llama.cpp', 'ExLlamav3'):
+                    backend_logprobs = get_logprobs_from_backend()
+                    completion_logprobs = {'top_logprobs': [backend_logprobs]} if backend_logprobs else None
                 else:
                     completion_logprobs = None
 
@@ -576,9 +576,9 @@ def text_streaming_chunk(content):
             # begin streaming
             if logprob_proc:
                 chunk_logprobs = {'top_logprobs': [logprob_proc.token_alternatives]}
-            elif shared.args.loader == 'llama.cpp':
-                llama_logprobs = get_logprobs_from_llama_cpp()
-                chunk_logprobs = {'top_logprobs': [llama_logprobs]} if llama_logprobs else None
+            elif shared.args.loader in ('llama.cpp', 'ExLlamav3'):
+                backend_logprobs = get_logprobs_from_backend()
+                chunk_logprobs = {'top_logprobs': [backend_logprobs]} if backend_logprobs else None
             else:
                 chunk_logprobs = None
 
diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index aeb68564da..1c682e4995 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -1,3 +1,4 @@
+import math
 import queue
 import threading
 import traceback
@@ -9,6 +10,7 @@
 from exllamav3 import Cache, Config, Generator, Model, Tokenizer
 from exllamav3.cache import CacheLayer_fp16, CacheLayer_quant
 from exllamav3.generator import Job
+from exllamav3.generator.filter import Filter
 from exllamav3.generator.sampler import (
     CustomSampler,
     SS_AdaptiveP,
@@ -36,6 +38,29 @@
     traceback.print_exc()
 
 
+class LogitBiasFilter(Filter):
+    """Filter subclass that applies a static additive logit bias mask."""
+
+    def __init__(self, tokenizer, logit_bias_dict):
+        super().__init__(tokenizer=tokenizer, trigger_token=None, prefix_str=None, eos_after_completed=False)
+        self.logit_bias_dict = logit_bias_dict
+        self._mask = None
+
+    def reset(self): pass
+    def accept_token(self, token): pass
+    def is_completed(self): return False
+    def use_background_worker(self): return False
+
+    def get_next_logit_mask(self):
+        if self._mask is None:
+            self._mask = torch.zeros((1, self.vocab_size), dtype=self.logits_dtype)
+            for token_id_str, bias in self.logit_bias_dict.items():
+                token_id = int(token_id_str)
+                if 0 <= token_id < self.vocab_size:
+                    self._mask[0, token_id] = bias
+        return self._mask
+
+
 class ConcurrentGenerator:
     def __init__(self, generator):
         self.generator = generator
@@ -98,6 +123,10 @@ class Exllamav3Model:
     def __init__(self):
         pass
 
+    @property
+    def device(self) -> torch.device:
+        return torch.device(0)
+
     @classmethod
     def from_pretrained(cls, path_to_model):
         path_to_model = Path(f'{shared.args.model_dir}') / Path(path_to_model)
@@ -390,6 +419,16 @@ def custom_sort_key(sampler_obj):
                 if eos_id is not None:
                     stop_conditions.append(eos_id)
 
+        # Build filters for logit_bias (OpenAI API)
+        filters = []
+        logit_bias = state.get('logit_bias')
+        if logit_bias:
+            filters.append(LogitBiasFilter(self.tokenizer, logit_bias))
+
+        # Logprobs support (OpenAI API)
+        logprobs = state.get('logprobs', 0) or 0
+        return_top_tokens = logprobs if logprobs > 0 else 0
+
         seed = state.get('seed', -1)
         job = Job(
             input_ids=input_ids,
@@ -399,11 +438,15 @@ def custom_sort_key(sampler_obj):
             sampler=sampler,
             seed=seed if seed >= 0 else None,
             stop_conditions=stop_conditions if stop_conditions else None,
+            filters=filters if filters else None,
+            return_top_tokens=return_top_tokens,
+            return_probs=return_top_tokens > 0,
         )
 
         # Stream generation
         response_text = ""
         stop_event = state.get('stop_event')
+        self.last_completion_probabilities = []
 
         result_queue = self.parallel_generator.submit(job)
         try:
@@ -415,14 +458,41 @@ def custom_sort_key(sampler_obj):
                 except queue.Empty:
                     continue
                 if result is None or result.get("eos"):
+                    # Capture logprobs from the final eos result too
+                    if result is not None and return_top_tokens > 0:
+                        self._capture_logprobs(result)
                     break
                 chunk = result.get("text", "")
+
+                # Capture logprobs from streaming results
+                if return_top_tokens > 0:
+                    self._capture_logprobs(result)
+
                 if chunk:
                     response_text += chunk
                     yield response_text
         finally:
             self.parallel_generator.cancel(job)
 
+    def _capture_logprobs(self, result):
+        """Convert ExLlamav3 top-k token data to the shared logprobs format."""
+        top_k_tokens = result.get("top_k_tokens")
+        top_k_probs = result.get("top_k_probs")
+        if top_k_tokens is None or top_k_probs is None:
+            return
+
+        id_to_piece = self.tokenizer.get_id_to_piece_list(True)
+        # top_k_tokens shape: (batch, seq_len, k), top_k_probs same
+        for seq_idx in range(top_k_tokens.shape[1]):
+            entry = {"top_logprobs": []}
+            for k_idx in range(top_k_tokens.shape[2]):
+                token_id = top_k_tokens[0, seq_idx, k_idx].item()
+                prob = top_k_probs[0, seq_idx, k_idx].item()
+                token_str = id_to_piece[token_id] if token_id < len(id_to_piece) else f"<{token_id}>"
+                logprob = math.log(prob) if prob > 0 else float("-inf")
+                entry["top_logprobs"].append({"token": token_str, "logprob": logprob})
+            self.last_completion_probabilities.append(entry)
+
     def generate(self, prompt, state):
         output = ""
         for chunk in self.generate_with_streaming(prompt, state):

From f1cfeae37279fd26fe5d6cd1f1ea475818f55937 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 10 Mar 2026 20:55:49 -0700
Subject: [PATCH 1308/1701] API: Improve OpenAI spec compliance in streaming
 and non-streaming responses

---
 extensions/openai/completions.py | 89 ++++++++++++++++++++++++++------
 extensions/openai/typing.py      |  6 +++
 2 files changed, 80 insertions(+), 15 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 04e644d6ed..03c4b03e86 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -310,28 +310,41 @@ def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, p
     requested_model = generate_params.pop('model')
     logprob_proc = generate_params.pop('logprob_proc', None)
 
-    def chat_streaming_chunk(content, chunk_tool_calls=None):
+    def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False):
         # begin streaming
+        delta = {}
+        if include_role:
+            delta['role'] = 'assistant'
+            delta['refusal'] = None
+        if content is not None:
+            delta['content'] = content
+        if chunk_tool_calls:
+            delta['tool_calls'] = chunk_tool_calls
+
         chunk = {
             "id": cmpl_id,
             "object": object_type,
             "created": created_time,
             "model": shared.model_name,
+            "system_fingerprint": None,
             resp_list: [{
                 "index": 0,
                 "finish_reason": None,
-                "delta": {'role': 'assistant', 'content': content, 'tool_calls': chunk_tool_calls},
+                "delta": delta,
+                "logprobs": None,
             }],
         }
 
         if logprob_proc:  # not official for chat yet
             top_logprobs = convert_logprobs_to_tiktoken(model=requested_model, logprobs=logprob_proc.token_alternatives)
             chunk[resp_list][0]["logprobs"] = {'top_logprobs': [top_logprobs]}
-        # else:
-        #    chunk[resp_list][0]["logprobs"] = None
 
         return chunk
 
+    # Check if usage should be included in streaming chunks per OpenAI spec
+    stream_options = body.get('stream_options')
+    include_usage = bool(stream_options) and bool(stream_options.get('include_usage') if isinstance(stream_options, dict) else getattr(stream_options, 'include_usage', False))
+
     # generate reply #######################################
     if prompt_only:
         prompt = generate_chat_prompt(user_input, generate_params, _continue=continue_)
@@ -339,7 +352,10 @@ def chat_streaming_chunk(content, chunk_tool_calls=None):
         return
 
     if stream:
-        yield chat_streaming_chunk('')
+        chunk = chat_streaming_chunk('', include_role=True)
+        if include_usage:
+            chunk['usage'] = None
+        yield chunk
 
     generator = generate_chat_reply(
         user_input, generate_params, regenerate=False, _continue=continue_, loading_message=False)
@@ -372,6 +388,8 @@ def chat_streaming_chunk(content, chunk_tool_calls=None):
                 continue
 
             chunk = chat_streaming_chunk(new_content)
+            if include_usage:
+                chunk['usage'] = None
 
             seen_content = answer
             yield chunk
@@ -389,25 +407,42 @@ def chat_streaming_chunk(content, chunk_tool_calls=None):
         stop_reason = "length"
 
     if stream:
-        chunk = chat_streaming_chunk('', tool_calls)
+        chunk = chat_streaming_chunk(chunk_tool_calls=tool_calls)
         chunk[resp_list][0]['finish_reason'] = stop_reason
-        chunk['usage'] = {
+        usage = {
             "prompt_tokens": token_count,
             "completion_tokens": completion_token_count,
             "total_tokens": token_count + completion_token_count
         }
 
-        yield chunk
+        if include_usage:
+            chunk['usage'] = None
+            yield chunk
+            # Separate usage-only chunk with choices: [] per OpenAI spec
+            yield {
+                "id": cmpl_id,
+                "object": object_type,
+                "created": created_time,
+                "model": shared.model_name,
+                "system_fingerprint": None,
+                resp_list: [],
+                "usage": usage
+            }
+        else:
+            chunk['usage'] = usage
+            yield chunk
     else:
         resp = {
             "id": cmpl_id,
             "object": object_type,
             "created": created_time,
             "model": shared.model_name,
+            "system_fingerprint": None,
             resp_list: [{
                 "index": 0,
                 "finish_reason": stop_reason,
-                "message": {"role": "assistant", "content": answer, **({"tool_calls": tool_calls} if tool_calls else {})},
+                "message": {"role": "assistant", "refusal": None, "content": answer, **({"tool_calls": tool_calls} if tool_calls else {})},
+                "logprobs": None,
             }],
             "usage": {
                 "prompt_tokens": token_count,
@@ -418,8 +453,6 @@ def chat_streaming_chunk(content, chunk_tool_calls=None):
         if logprob_proc:  # not official for chat yet
             top_logprobs = convert_logprobs_to_tiktoken(model=requested_model, logprobs=logprob_proc.token_alternatives)
             resp[resp_list][0]["logprobs"] = {'top_logprobs': [top_logprobs]}
-        # else:
-        #     resp[resp_list][0]["logprobs"] = None
 
         yield resp
 
@@ -427,7 +460,7 @@ def chat_streaming_chunk(content, chunk_tool_calls=None):
 def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_event=None):
     object_type = 'text_completion'
     created_time = int(time.time())
-    cmpl_id = "conv-%d" % (int(time.time() * 1000000000))
+    cmpl_id = "cmpl-%d" % (int(time.time() * 1000000000))
     resp_list = 'data' if is_legacy else 'choices'
 
     prompt_str = 'context' if is_legacy else 'prompt'
@@ -548,6 +581,7 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_e
             "object": object_type,
             "created": created_time,
             "model": shared.model_name,
+            "system_fingerprint": None,
             resp_list: resp_list_data,
             "usage": {
                 "prompt_tokens": total_prompt_token_count,
@@ -572,6 +606,10 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_e
         prefix = prompt if echo else ''
         token_count = len(encode(prompt)[0])
 
+        # Check if usage should be included in streaming chunks per OpenAI spec
+        stream_options = body.get('stream_options')
+        include_usage = bool(stream_options) and bool(stream_options.get('include_usage') if isinstance(stream_options, dict) else getattr(stream_options, 'include_usage', False))
+
         def text_streaming_chunk(content):
             # begin streaming
             if logprob_proc:
@@ -587,6 +625,7 @@ def text_streaming_chunk(content):
                 "object": object_type,
                 "created": created_time,
                 "model": shared.model_name,
+                "system_fingerprint": None,
                 resp_list: [{
                     "index": 0,
                     "finish_reason": None,
@@ -597,7 +636,10 @@ def text_streaming_chunk(content):
 
             return chunk
 
-        yield text_streaming_chunk(prefix)
+        chunk = text_streaming_chunk(prefix)
+        if include_usage:
+            chunk['usage'] = None
+        yield chunk
 
         # generate reply #######################################
         debug_msg({'prompt': prompt, 'generate_params': generate_params})
@@ -617,6 +659,8 @@ def text_streaming_chunk(content):
 
             seen_content = answer
             chunk = text_streaming_chunk(new_content)
+            if include_usage:
+                chunk['usage'] = None
             yield chunk
 
         completion_token_count = len(encode(answer)[0])
@@ -626,13 +670,28 @@ def text_streaming_chunk(content):
 
         chunk = text_streaming_chunk(suffix)
         chunk[resp_list][0]["finish_reason"] = stop_reason
-        chunk["usage"] = {
+        usage = {
             "prompt_tokens": token_count,
             "completion_tokens": completion_token_count,
             "total_tokens": token_count + completion_token_count
         }
 
-        yield chunk
+        if include_usage:
+            chunk['usage'] = None
+            yield chunk
+            # Separate usage-only chunk with choices: [] per OpenAI spec
+            yield {
+                "id": cmpl_id,
+                "object": object_type,
+                "created": created_time,
+                "model": shared.model_name,
+                "system_fingerprint": None,
+                resp_list: [],
+                "usage": usage
+            }
+        else:
+            chunk["usage"] = usage
+            yield chunk
 
 
 def chat_completions(body: dict, is_legacy: bool = False, stop_event=None) -> dict:
diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 2156074b7f..078bd20123 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -99,6 +99,10 @@ class ToolCall(BaseModel):
     function: FunctionCall
 
 
+class StreamOptions(BaseModel):
+    include_usage: bool | None = False
+
+
 class CompletionRequestParams(BaseModel):
     model: str | None = Field(default=None, description="Unused parameter. To change the model, use the /v1/internal/model/load endpoint.")
     prompt: str | List[str] | None = Field(default=None, description="Text prompt for completion. Can also use 'messages' format for multimodal.")
@@ -113,6 +117,7 @@ class CompletionRequestParams(BaseModel):
     presence_penalty: float | None = shared.args.presence_penalty
     stop: str | List[str] | None = None
     stream: bool | None = False
+    stream_options: StreamOptions | None = None
     suffix: str | None = None
     temperature: float | None = shared.args.temperature
     top_p: float | None = shared.args.top_p
@@ -151,6 +156,7 @@ class ChatCompletionRequestParams(BaseModel):
     presence_penalty: float | None = shared.args.presence_penalty
     stop: str | List[str] | None = None
     stream: bool | None = False
+    stream_options: StreamOptions | None = None
     temperature: float | None = shared.args.temperature
     top_p: float | None = shared.args.top_p
     user: str | None = Field(default=None, description="Unused parameter.")

From 7a63a560430641d55b685994c097ab7a588b6547 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 11 Mar 2026 12:53:19 -0700
Subject: [PATCH 1309/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 12 files changed, 20 insertions(+), 20 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index a3d4d1e6a9..057ff94e91 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.24/exllamav3-0.0.24+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.24/exllamav3-0.0.24+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 3211f25198..19c45361ad 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 8d45211401..4f8ea026e3 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 525ceed50a..9b5bf1abdc 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 86b65a9726..7e49bbffe4 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 61c9ef73df..00c6d0d605 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 3d0785a390..2d8dad0a17 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 6805e20912..299f50c197 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 5a8ed87b0c..dbc9247cad 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index fafa23cf57..66a99ba631 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 3ef59f971a..4338b20cc4 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 6039357dfa..7d262302b2 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.87.0/llama_cpp_binaries-0.87.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 24977846fbd34eca6bb40292381fe2820fe58343 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 11 Mar 2026 13:14:26 -0700
Subject: [PATCH 1310/1701] Update AMD ROCm from 6.4 to 7.2

---
 one_click.py                               | 10 ++++++----
 requirements/full/requirements_amd.txt     |  4 ++--
 requirements/portable/requirements_amd.txt |  4 ++--
 3 files changed, 10 insertions(+), 8 deletions(-)

diff --git a/one_click.py b/one_click.py
index 189d81b835..fdffd0a029 100644
--- a/one_click.py
+++ b/one_click.py
@@ -91,7 +91,7 @@ def get_gpu_choice():
                 "What is your GPU?",
                 {
                     'A': 'NVIDIA',
-                    'B': 'AMD - Linux/macOS only, requires ROCm 6.4',
+                    'B': 'AMD - Linux only, ROCm 7.2',
                     'C': 'Apple M Series',
                     'D': 'Intel Arc (beta)',
                     'N': 'CPU mode'
@@ -115,7 +115,8 @@ def get_pytorch_install_command(gpu_choice):
     if gpu_choice == "NVIDIA_CUDA128":
         return base_cmd + "--index-url https://download.pytorch.org/whl/cu128"
     elif gpu_choice == "AMD":
-        return base_cmd + "--index-url https://download.pytorch.org/whl/rocm6.4"
+        py_tag = f"cp{PYTHON_VERSION.replace('.', '')}"
+        return f"python -m pip install https://repo.radeon.com/rocm/manylinux/rocm-rel-7.2/torch-{TORCH_VERSION}%2Brocm7.2.0.lw.git7e1940d4-{py_tag}-{py_tag}-linux_x86_64.whl"
     elif gpu_choice in ["APPLE", "NONE"]:
         return base_cmd + "--index-url https://download.pytorch.org/whl/cpu"
     elif gpu_choice == "INTEL":
@@ -131,7 +132,8 @@ def get_pytorch_update_command(gpu_choice):
     if gpu_choice == "NVIDIA_CUDA128":
         return f"{base_cmd} --index-url https://download.pytorch.org/whl/cu128"
     elif gpu_choice == "AMD":
-        return f"{base_cmd} --index-url https://download.pytorch.org/whl/rocm6.4"
+        py_tag = f"cp{PYTHON_VERSION.replace('.', '')}"
+        return f"python -m pip install --upgrade https://repo.radeon.com/rocm/manylinux/rocm-rel-7.2/torch-{TORCH_VERSION}%2Brocm7.2.0.lw.git7e1940d4-{py_tag}-{py_tag}-linux_x86_64.whl"
     elif gpu_choice in ["APPLE", "NONE"]:
         return f"{base_cmd} --index-url https://download.pytorch.org/whl/cpu"
     elif gpu_choice == "INTEL":
@@ -266,7 +268,7 @@ def update_pytorch_and_python():
 
 
 def clean_outdated_pytorch_cuda_dependencies():
-    patterns = ["cu121", "cu122", "torch2.4", "torch2.6", "torch2.7", "torchvision", "torchaudio"]
+    patterns = ["cu121", "cu122", "rocm6", "torch2.4", "torch2.6", "torch2.7", "torchvision", "torchaudio"]
     result = run_cmd("python -m pip list --format=freeze", capture_output=True, environment=True)
     matching_packages = []
 
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 19c45361ad..03e9678988 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 2d8dad0a17..8ccc15c9ba 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm6.4-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm6.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 66c976e995b554387aa5263cd5824b169c21f30a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 11 Mar 2026 18:09:40 -0300
Subject: [PATCH 1311/1701] Update README with ROCm 7.2 torch install URL

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 8dc09f32c0..9a8e0a8624 100644
--- a/README.md
+++ b/README.md
@@ -145,7 +145,7 @@ conda activate textgen
 |--------|---------|---------|
 | Linux/WSL | NVIDIA | `pip3 install torch==2.9.1 --index-url https://download.pytorch.org/whl/cu128` |
 | Linux/WSL | CPU only | `pip3 install torch==2.9.1 --index-url https://download.pytorch.org/whl/cpu` |
-| Linux | AMD | `pip3 install torch==2.9.1 --index-url https://download.pytorch.org/whl/rocm6.4` |
+| Linux | AMD | `pip3 install https://repo.radeon.com/rocm/manylinux/rocm-rel-7.2/torch-2.9.1%2Brocm7.2.0.lw.git7e1940d4-cp313-cp313-linux_x86_64.whl` |
 | MacOS + MPS | Any | `pip3 install torch==2.9.1` |
 | Windows | NVIDIA | `pip3 install torch==2.9.1 --index-url https://download.pytorch.org/whl/cu128` |
 | Windows | CPU only | `pip3 install torch==2.9.1` |

From bb00d96dc35f8b15f738905961c96b8c08900c87 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 11 Mar 2026 15:34:03 -0700
Subject: [PATCH 1312/1701] Use a new gr.DragDrop element for Sampler priority
 + update gradio

---
 js/main.js                                           | 10 +++++-----
 modules/ui_parameters.py                             |  2 +-
 requirements/full/requirements.txt                   |  4 ++--
 requirements/full/requirements_amd.txt               |  4 ++--
 requirements/full/requirements_apple_intel.txt       |  4 ++--
 requirements/full/requirements_apple_silicon.txt     |  4 ++--
 requirements/full/requirements_cpu_only.txt          |  4 ++--
 requirements/full/requirements_nowheels.txt          |  4 ++--
 requirements/portable/requirements.txt               |  4 ++--
 requirements/portable/requirements_amd.txt           |  4 ++--
 requirements/portable/requirements_apple_intel.txt   |  4 ++--
 requirements/portable/requirements_apple_silicon.txt |  4 ++--
 requirements/portable/requirements_cpu_only.txt      |  4 ++--
 requirements/portable/requirements_cuda131.txt       |  4 ++--
 requirements/portable/requirements_nowheels.txt      |  4 ++--
 requirements/portable/requirements_vulkan.txt        |  4 ++--
 16 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/js/main.js b/js/main.js
index 1317e9e70c..a7dc77bbfc 100644
--- a/js/main.js
+++ b/js/main.js
@@ -296,11 +296,11 @@ function doSyntaxHighlighting() {
 //------------------------------------------------
 // Add some scrollbars
 //------------------------------------------------
-const textareaElements = document.querySelectorAll(".add_scrollbar textarea");
-for(i = 0; i < textareaElements.length; i++) {
-  textareaElements[i].classList.remove("scroll-hide");
-  textareaElements[i].classList.add("pretty_scrollbar");
-  textareaElements[i].style.resize = "none";
+const scrollbarElements = document.querySelectorAll(".add_scrollbar textarea, .add_scrollbar .drag-drop-list");
+for(i = 0; i < scrollbarElements.length; i++) {
+  scrollbarElements[i].classList.remove("scroll-hide");
+  scrollbarElements[i].classList.add("pretty_scrollbar");
+  scrollbarElements[i].style.resize = "none";
 }
 
 //------------------------------------------------
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index e5eb9210a3..a5afd7e536 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -73,7 +73,7 @@ def create_ui():
                             gr.Markdown('## Other options')
                             shared.gradio['do_sample'] = gr.Checkbox(value=shared.settings['do_sample'], label='do_sample')
                             shared.gradio['temperature_last'] = gr.Checkbox(value=shared.settings['temperature_last'], label='temperature_last', info='Moves temperature/dynamic temperature/quadratic sampling to the end of the sampler stack, ignoring their positions in "Sampler priority".')
-                            shared.gradio['sampler_priority'] = gr.Textbox(value=shared.settings['sampler_priority'], lines=10, label='Sampler priority', info='Parameter names separated by new lines or commas.', elem_classes=['add_scrollbar'])
+                            shared.gradio['sampler_priority'] = gr.DragDrop(value=shared.settings['sampler_priority'], label='Sampler priority', info='Parameter names separated by new lines or commas.', elem_classes=['add_scrollbar'])
                             shared.gradio['dry_sequence_breakers'] = gr.Textbox(value=shared.settings['dry_sequence_breakers'], label='dry_sequence_breakers', info='Tokens across which sequence matching is not continued. Specified as a comma-separated list of quoted strings.')
 
                 with gr.Column():
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 057ff94e91..12e7fbaeb2 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -31,8 +31,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 03e9678988..19cc0d9d11 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -28,8 +28,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 4f8ea026e3..ebe26f9db6 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -28,8 +28,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 9b5bf1abdc..491556908f 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -28,8 +28,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 7e49bbffe4..1c7c57354c 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -28,8 +28,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 0a924d314c..63823db82d 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -28,8 +28,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 00c6d0d605..db23d4bfc7 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 8ccc15c9ba..e8cd9fd945 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 299f50c197..24c558a97d 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index dbc9247cad..f2e8e69170 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 66a99ba631..296c04326f 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 4338b20cc4..aefce76969 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index c2fc33eb90..8c3e2aacbd 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 7d262302b2..76bb58725c 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ rich
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio-4.37.2+custom.9-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.9/gradio_client-1.0.2+custom.9-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From 980a9d165780b6a2dcc6c4bf6469436861079d2a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 11 Mar 2026 15:50:16 -0700
Subject: [PATCH 1313/1701] UI: Minor defensive changes to autosave

---
 modules/chat.py |  8 ++++++--
 modules/ui.py   | 11 +++++++----
 2 files changed, 13 insertions(+), 6 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 62d1492d90..10969446e8 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -6,6 +6,7 @@
 import pprint
 import re
 import shutil
+import threading
 import time
 from datetime import datetime
 from functools import partial
@@ -40,6 +41,8 @@
 )
 from modules.web_search import add_web_search_attachments
 
+_history_file_lock = threading.Lock()
+
 
 def strftime_now(format):
     return datetime.now().strftime(format)
@@ -1200,8 +1203,9 @@ def save_history(history, unique_id, character, mode):
     if not p.parent.is_dir():
         p.parent.mkdir(parents=True)
 
-    with open(p, 'w', encoding='utf-8') as f:
-        f.write(json.dumps(history, indent=4, ensure_ascii=False))
+    with _history_file_lock:
+        with open(p, 'w', encoding='utf-8') as f:
+            f.write(json.dumps(history, indent=4, ensure_ascii=False))
 
 
 def rename_history(old_id, new_id, character, mode):
diff --git a/modules/ui.py b/modules/ui.py
index abbfde49d0..2ab3056391 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -303,12 +303,16 @@ def save_settings(state, preset, extensions_list, show_controls, theme_state, ma
         if k in shared.settings and k not in exclude:
             output[k] = state[k]
 
-    output['preset'] = preset
+    if preset:
+        output['preset'] = preset
     output['prompt-notebook'] = state['prompt_menu-default'] if state['show_two_notebook_columns'] else state['prompt_menu-notebook']
-    output['character'] = state['character_menu']
-    if 'user_menu' in state and state['user_menu']:
+    if state.get('character_menu'):
+        output['character'] = state['character_menu']
+    if state.get('user_menu'):
         output['user'] = state['user_menu']
     output['seed'] = int(output['seed'])
+    output['custom_stopping_strings'] = output.get('custom_stopping_strings') or ''
+    output['custom_token_bans'] = output.get('custom_token_bans') or ''
     output['show_controls'] = show_controls
     output['dark_theme'] = True if theme_state == 'dark' else False
     output.pop('instruction_template_str')
@@ -470,7 +474,6 @@ def setup_auto_save():
         'skip_special_tokens',
         'stream',
         'static_cache',
-        'truncation_length',
         'seed',
         'sampler_priority',
         'custom_stopping_strings',

From cf9ad8eafe96a3065bf3c8edba58d332de6474fe Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 01:15:49 -0300
Subject: [PATCH 1314/1701] Initial tool-calling support in the UI

---
 modules/chat.py     | 189 ++++++++++++++++++++++++++++++++++++++++----
 modules/shared.py   |   1 +
 modules/tool_use.py |  70 ++++++++++++++++
 modules/ui.py       |   2 +
 modules/ui_chat.py  |   5 ++
 5 files changed, 253 insertions(+), 14 deletions(-)
 create mode 100644 modules/tool_use.py

diff --git a/modules/chat.py b/modules/chat.py
index 10969446e8..b0be2bc215 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -298,6 +298,23 @@ def generate_chat_prompt(user_input, state, **kwargs):
             if entry_meta.get('tool_calls') and messages[insert_pos].get('role') == 'assistant':
                 messages[insert_pos]['tool_calls'] = _deserialize_tool_call_arguments(entry_meta['tool_calls'])
 
+        # Expand tool_sequence from metadata (inserted AFTER assistant so that
+        # the final order is: user → tool_calls → tool_results → final_answer)
+        meta_key = f"assistant_{row_idx}"
+        tool_seq = metadata.get(meta_key, {}).get('tool_sequence', [])
+        if tool_seq:
+            for item in reversed(tool_seq):
+                if 'tool_calls' in item:
+                    messages.insert(insert_pos, {
+                        "role": "assistant", "content": "",
+                        "tool_calls": _deserialize_tool_call_arguments(item['tool_calls'])
+                    })
+                elif item.get('role') == 'tool':
+                    messages.insert(insert_pos, {
+                        "role": "tool", "content": item['content'],
+                        "tool_call_id": item.get('tool_call_id', '')
+                    })
+
         if user_msg not in ['', '<|BEGIN-VISIBLE-CHAT|>']:
             # Check for user message attachments in metadata
             user_key = f"user_{row_idx}"
@@ -367,6 +384,22 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
             messages.append({"role": "user", "content": user_input})
 
+        # Expand tool_sequence for the current entry (excluded from the
+        # history loop during regenerate — needed so the model sees prior
+        # tool calls and results when re-generating the final answer).
+        current_tool_seq = metadata.get(f"assistant_{len(history)}", {}).get('tool_sequence', [])
+        for item in current_tool_seq:
+            if 'tool_calls' in item:
+                messages.append({
+                    "role": "assistant", "content": "",
+                    "tool_calls": _deserialize_tool_call_arguments(item['tool_calls'])
+                })
+            elif item.get('role') == 'tool':
+                messages.append({
+                    "role": "tool", "content": item['content'],
+                    "tool_call_id": item.get('tool_call_id', '')
+                })
+
     if impersonate and state['mode'] != 'chat-instruct':
         messages.append({"role": "user", "content": "fake user message replace me"})
 
@@ -886,7 +919,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             }
     else:
         text, visible_text = output['internal'][-1][0], output['visible'][-1][0]
-        if regenerate:
+        if regenerate and not state.get('_tool_turn'):
             row_idx = len(output['internal']) - 1
 
             # Store the old response as a version before regenerating
@@ -984,7 +1017,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             output['visible'][-1] = [visible_text, visible_reply.lstrip(' ')]
 
         # Keep version metadata in sync during streaming (for regeneration)
-        if regenerate:
+        if regenerate and not state.get('_tool_turn'):
             row_idx = len(output['internal']) - 1
             key = f"assistant_{row_idx}"
             current_idx = output['metadata'][key]['current_version_index']
@@ -1012,7 +1045,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
 
     # Final sync for version metadata (in case streaming was disabled)
-    if regenerate:
+    if regenerate and not state.get('_tool_turn'):
         row_idx = len(output['internal']) - 1
         key = f"assistant_{row_idx}"
         current_idx = output['metadata'][key]['current_version_index']
@@ -1066,12 +1099,24 @@ def character_is_loaded(state, raise_exception=False):
 
 def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
     '''
-    Same as above but returns HTML for the UI
+    Same as above but returns HTML for the UI.
+    When tools are selected, wraps generation in a loop that detects
+    tool calls, executes them, and re-generates until the model stops.
+    All tool output is consolidated into a single visible chat bubble
+    using metadata['assistant_N']['tool_sequence'].
     '''
 
     if not character_is_loaded(state):
         return
 
+    # On regenerate, clear old tool_sequence metadata so it gets rebuilt
+    if regenerate:
+        history = state['history']
+        meta = history.get('metadata', {})
+        row_idx = len(history['internal']) - 1
+        if row_idx >= 0:
+            meta.get(f'assistant_{row_idx}', {}).pop('tool_sequence', None)
+
     if state['start_with'] != '' and not _continue:
         if regenerate:
             text, state['history'] = remove_last_message(state['history'])
@@ -1081,23 +1126,139 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
         send_dummy_message(text, state)
         send_dummy_reply(state['start_with'], state)
 
-    history = state['history']
+    # Load tools if any are selected
+    selected = state.get('selected_tools', [])
+    if selected:
+        from modules.tool_use import load_tools, execute_tool, generate_tool_call_id
+        try:
+            from extensions.openai.utils import parseToolCall
+        except ImportError:
+            logger.warning('Tool calling requires the openai extension for parseToolCall. Disabling tools.')
+            selected = []
+
+    if selected:
+        tool_defs, tool_executors = load_tools(selected)
+        state['tools'] = tool_defs
+        tool_func_names = [t['function']['name'] for t in tool_defs]
+    else:
+        tool_func_names = None
+
+    visible_prefix = []  # Accumulated tool call summaries + results
     last_save_time = time.monotonic()
     save_interval = 8
-    for i, history in enumerate(generate_chat_reply(text, state, regenerate, _continue, loading_message=True, for_ui=True)):
-        yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'], last_message_only=(i > 0)), history
-        if i == 0:
-            time.sleep(0.125)  # We need this to make sure the first update goes through
+    max_tool_turns = 10
+
+    for _tool_turn in range(max_tool_turns):
+        history = state['history']
+
+        # Turn 0: use original flags; turns 2+: regenerate into the same entry
+        if _tool_turn > 0:
+            state['_tool_turn'] = True
+
+        regen = regenerate if _tool_turn == 0 else True
+        cont = _continue if _tool_turn == 0 else False
+        cur_text = text if _tool_turn == 0 else ''
+
+        for i, history in enumerate(generate_chat_reply(cur_text, state, regen, cont, loading_message=True, for_ui=True)):
+            # Prepend accumulated tool output to visible reply
+            if visible_prefix:
+                history['visible'][-1][1] = '\n\n'.join(visible_prefix + [history['visible'][-1][1]])
+
+            yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'], last_message_only=(i > 0)), history
+
+            if i == 0:
+                time.sleep(0.125)
+
+            current_time = time.monotonic()
+            if i == 0 or (current_time - last_save_time) >= save_interval:
+                save_history(history, state['unique_id'], state['character_menu'], state['mode'])
+                last_save_time = current_time
+
+            # Early stop on tool call detection
+            if tool_func_names and parseToolCall(history['internal'][-1][1], tool_func_names):
+                break
 
-        current_time = time.monotonic()
-        # Save on first iteration or if save_interval seconds have passed
-        if i == 0 or (current_time - last_save_time) >= save_interval:
-            save_history(history, state['unique_id'], state['character_menu'], state['mode'])
-            last_save_time = current_time
+        save_history(history, state['unique_id'], state['character_menu'], state['mode'])
+
+        # Check for tool calls
+        if not tool_func_names or shared.stop_everything:
+            break
+
+        answer = history['internal'][-1][1]
+        parsed_calls = parseToolCall(answer, tool_func_names) if answer else None
+
+        if not parsed_calls:
+            break  # No tool calls — done
+
+        # --- Process tool calls ---
+        row_idx = len(history['internal']) - 1
+        meta = history.get('metadata', {})
+        seq = meta.setdefault(f'assistant_{row_idx}', {}).setdefault('tool_sequence', [])
+
+        # Serialize tool calls
+        serialized = []
+        for tc in parsed_calls:
+            tc['id'] = generate_tool_call_id()
+            args = tc['function'].get('arguments', {})
+            serialized.append({
+                'id': tc['id'],
+                'type': 'function',
+                'function': {
+                    'name': tc['function']['name'],
+                    'arguments': json.dumps(args) if isinstance(args, dict) else args
+                }
+            })
+
+        seq.append({'tool_calls': serialized})
+
+        # Clear internal (raw tool markup)
+        history['internal'][-1][1] = ''
+
+        # Add call summary to visible prefix
+        call_summary = ', '.join(f'{tc["function"]["name"]}(...)' for tc in parsed_calls)
+        visible_prefix.append('Calling: ' + call_summary)
+
+        # Execute tools, store results
+        for tc in parsed_calls:
+            fn_name = tc['function']['name']
+            fn_args = tc['function'].get('arguments', {})
+            result = execute_tool(fn_name, fn_args, tool_executors)
+
+            seq.append({'role': 'tool', 'content': result, 'tool_call_id': tc['id']})
+            try:
+                pretty_result = json.dumps(json.loads(result), indent=2, ensure_ascii=False)
+            except (json.JSONDecodeError, TypeError):
+                pretty_result = result
+
+            visible_prefix.append(f'**{fn_name}**\n```json\n{pretty_result}\n```')
+
+        # Show tool results
+        history['visible'][-1][1] = '\n\n'.join(visible_prefix)
+        yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu']), history
+        save_history(history, state['unique_id'], state['character_menu'], state['mode'])
+
+        state['history'] = history
+
+    state.pop('_tool_turn', None)
+    state['history'] = history
+
+    # Sync version metadata so swipes show the full visible (with tool prefix)
+    if visible_prefix and history.get('metadata'):
+        row_idx = len(history['internal']) - 1
+        key = f"assistant_{row_idx}"
+        meta_entry = history['metadata'].get(key, {})
+        if 'versions' in meta_entry and 'current_version_index' in meta_entry:
+            current_idx = meta_entry['current_version_index']
+            if current_idx < len(meta_entry['versions']):
+                meta_entry['versions'][current_idx].update({
+                    'content': history['internal'][row_idx][1],
+                    'visible_content': history['visible'][row_idx][1]
+                })
 
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
 
 
+
 def remove_last_message(history):
     if 'metadata' not in history:
         history['metadata'] = {}
diff --git a/modules/shared.py b/modules/shared.py
index dbd805a12d..395ca83c7a 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -263,6 +263,7 @@
     'chat-instruct_command': 'Continue the chat dialogue below. Write a single reply for the character "<|character|>". Reply directly, without starting the reply with the character name.\n\n<|prompt|>',
     'enable_web_search': False,
     'web_search_pages': 3,
+    'selected_tools': [],
     'prompt-notebook': '',
     'preset': 'Qwen3 - Thinking' if (user_data_dir / 'presets/Qwen3 - Thinking.yaml').exists() else None,
     'max_new_tokens': 512,
diff --git a/modules/tool_use.py b/modules/tool_use.py
new file mode 100644
index 0000000000..cb1e140dcb
--- /dev/null
+++ b/modules/tool_use.py
@@ -0,0 +1,70 @@
+import importlib.util
+import json
+import random
+
+from modules import shared
+from modules.logging_colors import logger
+
+
+def get_available_tools():
+    """Return sorted list of tool script names from user_data/tools/*.py."""
+    tools_dir = shared.user_data_dir / 'tools'
+    tools_dir.mkdir(parents=True, exist_ok=True)
+    return sorted(p.stem for p in tools_dir.glob('*.py'))
+
+
+def load_tools(selected_names):
+    """
+    Import selected tool scripts and return their definitions and executors.
+    Returns (tool_defs, executors) where:
+      - tool_defs: list of OpenAI-format tool dicts
+      - executors: dict mapping function_name -> execute callable
+    """
+    tool_defs = []
+    executors = {}
+    for name in selected_names:
+        path = shared.user_data_dir / 'tools' / f'{name}.py'
+        if not path.exists():
+            continue
+
+        try:
+            spec = importlib.util.spec_from_file_location(f"tool_{name}", str(path))
+            module = importlib.util.module_from_spec(spec)
+            spec.loader.exec_module(module)
+        except Exception:
+            logger.exception(f'Failed to load tool script "{name}"')
+            continue
+
+        tool_def = getattr(module, 'tool', None)
+        execute_fn = getattr(module, 'execute', None)
+        if tool_def is None or execute_fn is None:
+            logger.warning(f'Tool "{name}" is missing a "tool" dict or "execute" function.')
+            continue
+
+        func_name = tool_def.get('function', {}).get('name', name)
+        tool_defs.append(tool_def)
+        executors[func_name] = execute_fn
+
+    return tool_defs, executors
+
+
+def generate_tool_call_id():
+    """Generate a unique tool call ID (e.g. 'call_a1b2c3d4')."""
+    chars = "abcdefghijklmnopqrstuvwxyz0123456789"
+    return "call_" + "".join(random.choice(chars) for _ in range(8))
+
+
+def execute_tool(func_name, arguments, executors):
+    """Execute a tool by function name. Returns result as a JSON string."""
+    fn = executors.get(func_name)
+    if fn is None:
+        return json.dumps({"error": f"Unknown tool: {func_name}"})
+
+    try:
+        if isinstance(arguments, str):
+            arguments = json.loads(arguments)
+        result = fn(arguments)
+        return json.dumps(result) if not isinstance(result, str) else result
+    except Exception as e:
+        logger.exception(f'Tool "{func_name}" execution failed')
+        return json.dumps({"error": str(e)})
diff --git a/modules/ui.py b/modules/ui.py
index 2ab3056391..3f39a1a419 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -199,6 +199,7 @@ def list_interface_input_elements():
         'unique_id',
         'textbox',
         'start_with',
+        'selected_tools',
         'mode',
         'chat_style',
         'chat-instruct_command',
@@ -424,6 +425,7 @@ def setup_auto_save():
         'user_bio',
         'custom_system_message',
         'chat_template_str',
+        'selected_tools',
 
         # Parameters tab (ui_parameters.py) - Generation parameters
         'preset_menu',
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 74da0a40c5..9c7424e76b 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -91,6 +91,11 @@ def create_ui():
 
                 gr.HTML("
            ")
 
+                from modules.tool_use import get_available_tools
+                shared.gradio['selected_tools'] = gr.CheckboxGroup(choices=get_available_tools(), value=[], label='Tools', info='Functions the model can call during generation.')
+
+                gr.HTML("
            ")
+
                 with gr.Row():
                     shared.gradio['mode'] = gr.Radio(choices=['instruct', 'chat-instruct', 'chat'], value=None, label='Mode', info='In instruct and chat-instruct modes, the template under Parameters > Instruction template is used.', elem_id='chat-mode')
 

From 0d6203871050f5db14b5197899f7e11fbb152fd2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 01:36:07 -0300
Subject: [PATCH 1315/1701] Add tools refresh button and _tool_turn comment

---
 modules/chat.py    | 4 +++-
 modules/ui_chat.py | 1 +
 2 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index b0be2bc215..ad1095f384 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1151,7 +1151,9 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
     for _tool_turn in range(max_tool_turns):
         history = state['history']
 
-        # Turn 0: use original flags; turns 2+: regenerate into the same entry
+        # Turn 0: use original flags; turns 2+: regenerate into the same entry.
+        # _tool_turn tells chatbot_wrapper to skip version creation/sync so
+        # that intermediate tool-loop regenerations don't pollute swipe history.
         if _tool_turn > 0:
             state['_tool_turn'] = True
 
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 9c7424e76b..d5b1309432 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -93,6 +93,7 @@ def create_ui():
 
                 from modules.tool_use import get_available_tools
                 shared.gradio['selected_tools'] = gr.CheckboxGroup(choices=get_available_tools(), value=[], label='Tools', info='Functions the model can call during generation.')
+                ui.create_refresh_button(shared.gradio['selected_tools'], lambda: None, lambda: {'choices': get_available_tools()}, 'refresh-button')
 
                 gr.HTML("
            ")
 

From b5cac2e3b2bc20d2343c2c7dbecdcb01197d422a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 01:48:19 -0300
Subject: [PATCH 1316/1701] Fix swipes and edit for tool calling in the UI

---
 modules/chat.py | 42 +++++++++++++++++++++++++++++++++++-------
 1 file changed, 35 insertions(+), 7 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index ad1095f384..f3020ee145 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1109,13 +1109,15 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
     if not character_is_loaded(state):
         return
 
-    # On regenerate, clear old tool_sequence metadata so it gets rebuilt
+    # On regenerate, clear old tool_sequence metadata so it gets rebuilt.
+    # Save it first so it can be stored per-version below.
+    _old_tool_sequence = None
     if regenerate:
         history = state['history']
         meta = history.get('metadata', {})
         row_idx = len(history['internal']) - 1
         if row_idx >= 0:
-            meta.get(f'assistant_{row_idx}', {}).pop('tool_sequence', None)
+            _old_tool_sequence = meta.get(f'assistant_{row_idx}', {}).pop('tool_sequence', None)
 
     if state['start_with'] != '' and not _continue:
         if regenerate:
@@ -1128,6 +1130,7 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
 
     # Load tools if any are selected
     selected = state.get('selected_tools', [])
+    parseToolCall = None
     if selected:
         from modules.tool_use import load_tools, execute_tool, generate_tool_call_id
         try:
@@ -1169,6 +1172,16 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
             yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'], last_message_only=(i > 0)), history
 
             if i == 0:
+                # Save old tool_sequence into version 0 (created by chatbot_wrapper
+                # on the first yield).  Only needed on the first regeneration when
+                # versions didn't previously exist.
+                if _old_tool_sequence is not None and _tool_turn == 0:
+                    _ri = len(history['internal']) - 1
+                    _versions = history.get('metadata', {}).get(f'assistant_{_ri}', {}).get('versions', [])
+                    if _versions and 'tool_sequence' not in _versions[0]:
+                        _versions[0]['tool_sequence'] = _old_tool_sequence
+                    _old_tool_sequence = None
+
                 time.sleep(0.125)
 
             current_time = time.monotonic()
@@ -1252,15 +1265,18 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
         if 'versions' in meta_entry and 'current_version_index' in meta_entry:
             current_idx = meta_entry['current_version_index']
             if current_idx < len(meta_entry['versions']):
-                meta_entry['versions'][current_idx].update({
+                version_update = {
                     'content': history['internal'][row_idx][1],
                     'visible_content': history['visible'][row_idx][1]
-                })
+                }
+                ts = meta_entry.get('tool_sequence')
+                if ts is not None:
+                    version_update['tool_sequence'] = ts
+                meta_entry['versions'][current_idx].update(version_update)
 
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
 
 
-
 def remove_last_message(history):
     if 'metadata' not in history:
         history['metadata'] = {}
@@ -2163,11 +2179,15 @@ def handle_edit_message_click(state):
         original_visible = history['visible'][message_index][role_idx]
         original_timestamp = history['metadata'][key].get('timestamp', get_current_timestamp())
 
-        history['metadata'][key]["versions"] = [{
+        version_entry = {
             "content": original_content,
             "visible_content": original_visible,
             "timestamp": original_timestamp
-        }]
+        }
+        ts = history['metadata'][key].get('tool_sequence')
+        if ts is not None:
+            version_entry['tool_sequence'] = ts
+        history['metadata'][key]["versions"] = [version_entry]
 
     history['internal'][message_index][role_idx] = apply_extensions('input', new_text, state, is_chat=True)
     history['visible'][message_index][role_idx] = html.escape(new_text)
@@ -2215,6 +2235,14 @@ def handle_navigate_version_click(state):
     history['internal'][message_index][msg_content_idx] = version_to_load['content']
     history['visible'][message_index][msg_content_idx] = version_to_load['visible_content']
     metadata['current_version_index'] = new_idx
+
+    # Restore per-version tool_sequence so follow-up prompts see consistent context
+    version_ts = version_to_load.get('tool_sequence')
+    if version_ts is not None:
+        metadata['tool_sequence'] = version_ts
+    else:
+        metadata.pop('tool_sequence', None)
+
     update_message_metadata(history['metadata'], role, message_index, timestamp=version_to_load['timestamp'])
 
     # Redraw and save

From 2549f7c33b8989c9604edc6476fa2354f9bb6662 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 10:28:49 -0300
Subject: [PATCH 1317/1701] API: Add tool_choice support and fix tool_calls
 spec compliance

---
 extensions/openai/completions.py | 24 ++++++++++++++++++------
 extensions/openai/typing.py      |  1 +
 2 files changed, 19 insertions(+), 6 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 03c4b03e86..8d3cce5745 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -218,6 +218,10 @@ def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, p
     if 'tools' in body and body['tools'] is not None and isinstance(body['tools'], list) and len(body['tools']) > 0:
         tools = validateTools(body['tools'])  # raises InvalidRequestError if validation fails
 
+    tool_choice = body.get('tool_choice', None)
+    if tool_choice == "none":
+        tools = None  # Disable tool detection entirely
+
     messages = body['messages']
     for m in messages:
         if 'role' not in m:
@@ -367,6 +371,12 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
     end_last_tool_call = 0
     supported_tools = [x["function"]["name"] for x in tools] if tools is not None else None
 
+    # Filter supported_tools when tool_choice specifies a particular function
+    if supported_tools and isinstance(tool_choice, dict):
+        specified_func = tool_choice.get("function", {}).get("name")
+        if specified_func and specified_func in supported_tools:
+            supported_tools = [specified_func]
+
     for a in generator:
         answer = a['internal'][-1][1]
 
@@ -375,11 +385,17 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
             if len(tool_call) > 0:
                 for tc in tool_call:
                     tc["id"] = getToolCallId()
-                    tc["index"] = len(tool_calls)
+                    if stream:
+                        tc["index"] = len(tool_calls)
                     tc["function"]["arguments"] = json.dumps(tc["function"]["arguments"])
                     tool_calls.append(tc)
                 end_last_tool_call = len(answer)
 
+        # Stop generation before streaming content if tool_calls were detected,
+        # so that raw tool markup is not sent as content deltas.
+        if len(tool_calls) > 0:
+            break
+
         if stream:
             len_seen = len(seen_content)
             new_content = answer[len_seen:]
@@ -394,10 +410,6 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
             seen_content = answer
             yield chunk
 
-        # stop generation if tool_calls were generated previously
-        if len(tool_calls) > 0:
-            break
-
     token_count = shared.model.last_prompt_token_count if hasattr(shared.model, 'last_prompt_token_count') else 0
     completion_token_count = len(encode(answer)[0])
     stop_reason = "stop"
@@ -441,7 +453,7 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
             resp_list: [{
                 "index": 0,
                 "finish_reason": stop_reason,
-                "message": {"role": "assistant", "refusal": None, "content": answer, **({"tool_calls": tool_calls} if tool_calls else {})},
+                "message": {"role": "assistant", "refusal": None, "content": None if tool_calls else answer, **({"tool_calls": tool_calls} if tool_calls else {})},
                 "logprobs": None,
             }],
             "usage": {
diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 078bd20123..4d0f4a4ad8 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -150,6 +150,7 @@ class ChatCompletionRequestParams(BaseModel):
     function_call: str | dict | None = Field(default=None, description="Unused parameter.")
     functions: List[dict] | None = Field(default=None, description="Unused parameter.")
     tools: List[dict] | None = Field(default=None, description="Tools signatures passed via MCP.")
+    tool_choice: str | dict | None = Field(default=None, description="Controls tool use: 'auto', 'none', 'required', or {\"type\": \"function\", \"function\": {\"name\": \"...\"}}.")
     logit_bias: dict | None = None
     max_tokens: int | None = None
     n: int | None = Field(default=1, description="Unused parameter.")

From 09723c9988cf8144286c88832cf2627e2df4938d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 12:43:03 -0300
Subject: [PATCH 1318/1701] API: Include /v1 in the printed API URL for easier
 integration

---
 extensions/openai/script.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index 94c7650f2a..e3726bc83f 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -453,11 +453,11 @@ def run_server():
             port,
             shared.args.public_api_id,
             max_attempts=3,
-            on_start=lambda url: logger.info(f'OpenAI-compatible API URL:\n\n{url}\n')
+            on_start=lambda url: logger.info(f'OpenAI-compatible API URL:\n\n{url}/v1\n')
         )
     else:
         url_proto = 'https://' if (ssl_certfile and ssl_keyfile) else 'http://'
-        urls = [f'{url_proto}{addr}:{port}' for addr in server_addrs]
+        urls = [f'{url_proto}{addr}:{port}/v1' for addr in server_addrs]
         if len(urls) > 1:
             logger.info('OpenAI-compatible API URLs:\n\n' + '\n'.join(urls) + '\n')
         else:

From 4b6c9db1c9137a81f363e3dc7c0665441be83735 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 13:11:28 -0300
Subject: [PATCH 1319/1701] UI: Fix stale tool_sequence after edit and
 chat-instruct tool rendering

---
 modules/chat.py   | 1 +
 modules/shared.py | 2 +-
 2 files changed, 2 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index f3020ee145..5cf550d679 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -2191,6 +2191,7 @@ def handle_edit_message_click(state):
 
     history['internal'][message_index][role_idx] = apply_extensions('input', new_text, state, is_chat=True)
     history['visible'][message_index][role_idx] = html.escape(new_text)
+    history['metadata'][key].pop('tool_sequence', None)
 
     add_message_version(history, role, message_index, is_current=True)
 
diff --git a/modules/shared.py b/modules/shared.py
index 395ca83c7a..8c0aad9a51 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -348,7 +348,7 @@
     'greeting': 'How can I help you today?',
     'custom_system_message': '',
     'instruction_template_str': "{%- set ns = namespace(found=false) -%}\n{%- for message in messages -%}\n    {%- if message['role'] == 'system' -%}\n        {%- set ns.found = true -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if not ns.found -%}\n    {{- '' + 'Below is an instruction that describes a task. Write a response that appropriately completes the request.' + '\\n\\n' -}}\n{%- endif %}\n{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {{- '' + message['content'] + '\\n\\n' -}}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{-'### Instruction:\\n' + message['content'] + '\\n\\n'-}}\n        {%- else -%}\n            {{-'### Response:\\n' + message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{-'### Response:\\n'-}}\n{%- endif -%}",
-    'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- else -%}\n        {%- if message['role'] == 'user' -%}\n            {{- name1 + ': ' + message['content'] + '\\n'-}}\n        {%- else -%}\n            {{- name2 + ': ' + message['content'] + '\\n' -}}\n        {%- endif -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt %}\n    {{- name2 + ':' -}}\n{%- endif %}",
+    'chat_template_str': "{%- for message in messages %}\n    {%- if message['role'] == 'system' -%}\n        {%- if message['content'] -%}\n            {{- message['content'] + '\\n\\n' -}}\n        {%- endif -%}\n        {%- if user_bio -%}\n            {{- user_bio + '\\n\\n' -}}\n        {%- endif -%}\n    {%- elif message['role'] == 'tool' -%}\n        {{- '[Tool result: ' + message['content'] + ']\\n' -}}\n    {%- elif message['role'] == 'user' -%}\n        {{- name1 + ': ' + message['content'] + '\\n'-}}\n    {%- elif message['tool_calls'] is defined and message['tool_calls'] -%}\n        {%- for tc in message['tool_calls'] -%}\n            {{- '[Calling: ' + tc['function']['name'] + '(' + tc['function']['arguments'] + ')]\\n' -}}\n        {%- endfor -%}\n    {%- else -%}\n        {{- name2 + ': ' + message['content'] + '\\n' -}}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt %}\n    {{- name2 + ':' -}}\n{%- endif %}",
 
     # Extensions
     'default_extensions': [],

From 5a017aa3380ae273e0ce32fe02709f93cb710d20 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 13:23:01 -0300
Subject: [PATCH 1320/1701] API: Several OpenAI spec compliance fixes

- Return proper OpenAI error format ({"error": {...}}) instead of HTTP 500 for validation errors
- Send data: [DONE] at the end of SSE streams
- Fix finish_reason so "tool_calls" takes priority over "length"
- Stop including usage in streaming chunks when include_usage is not set
- Handle "developer" role in messages (treated same as "system")
- Add logprobs and top_logprobs parameters for chat completions
- Fix chat completions logprobs not working with llama.cpp and ExLlamav3 backends
- Add max_completion_tokens as an alias for max_tokens in chat completions
---
 extensions/openai/completions.py | 25 ++++++++++++++++++-------
 extensions/openai/script.py      | 19 +++++++++++++++++++
 extensions/openai/typing.py      |  9 +++++++++
 3 files changed, 46 insertions(+), 7 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 8d3cce5745..a8b899d582 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -105,6 +105,10 @@ def process_parameters(body, is_legacy=False):
             logits_processor = [LogitsBiasProcessor(logit_bias)]
 
         logprobs = body.get('logprobs', None)
+        top_logprobs = body.get('top_logprobs', None)
+        # For chat completions, logprobs is a bool; use top_logprobs for the count
+        if logprobs is True:
+            logprobs = top_logprobs if top_logprobs and top_logprobs > 0 else 5
         if logprobs is not None and logprobs > 0:
             generate_params['logprob_proc'] = LogprobProcessor(logprobs)
             logits_processor.extend([generate_params['logprob_proc']])
@@ -191,7 +195,7 @@ def convert_history(history):
             if "tool_call_id" in entry:
                 meta["tool_call_id"] = entry["tool_call_id"]
             chat_dialogue.append(['', '', content, meta])
-        elif role == "system":
+        elif role in ("system", "developer"):
             system_message += f"\n{content}" if system_message else content
 
     if not user_input_last:
@@ -339,9 +343,13 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
             }],
         }
 
-        if logprob_proc:  # not official for chat yet
+        if logprob_proc:
             top_logprobs = convert_logprobs_to_tiktoken(model=requested_model, logprobs=logprob_proc.token_alternatives)
             chunk[resp_list][0]["logprobs"] = {'top_logprobs': [top_logprobs]}
+        elif shared.args.loader in ('llama.cpp', 'ExLlamav3'):
+            backend_logprobs = get_logprobs_from_backend()
+            if backend_logprobs:
+                chunk[resp_list][0]["logprobs"] = {'top_logprobs': [backend_logprobs]}
 
         return chunk
 
@@ -412,11 +420,12 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
 
     token_count = shared.model.last_prompt_token_count if hasattr(shared.model, 'last_prompt_token_count') else 0
     completion_token_count = len(encode(answer)[0])
-    stop_reason = "stop"
     if len(tool_calls) > 0:
         stop_reason = "tool_calls"
-    if token_count + completion_token_count >= generate_params['truncation_length'] or completion_token_count >= generate_params['max_new_tokens']:
+    elif token_count + completion_token_count >= generate_params['truncation_length'] or completion_token_count >= generate_params['max_new_tokens']:
         stop_reason = "length"
+    else:
+        stop_reason = "stop"
 
     if stream:
         chunk = chat_streaming_chunk(chunk_tool_calls=tool_calls)
@@ -441,7 +450,6 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
                 "usage": usage
             }
         else:
-            chunk['usage'] = usage
             yield chunk
     else:
         resp = {
@@ -462,9 +470,13 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
                 "total_tokens": token_count + completion_token_count
             }
         }
-        if logprob_proc:  # not official for chat yet
+        if logprob_proc:
             top_logprobs = convert_logprobs_to_tiktoken(model=requested_model, logprobs=logprob_proc.token_alternatives)
             resp[resp_list][0]["logprobs"] = {'top_logprobs': [top_logprobs]}
+        elif shared.args.loader in ('llama.cpp', 'ExLlamav3'):
+            backend_logprobs = get_logprobs_from_backend()
+            if backend_logprobs:
+                resp[resp_list][0]["logprobs"] = {'top_logprobs': [backend_logprobs]}
 
         yield resp
 
@@ -702,7 +714,6 @@ def text_streaming_chunk(content):
                 "usage": usage
             }
         else:
-            chunk["usage"] = usage
             yield chunk
 
 
diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index e3726bc83f..f161e1e453 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -21,6 +21,7 @@
 import extensions.openai.logits as OAIlogits
 import extensions.openai.models as OAImodels
 from extensions.openai.tokens import token_count, token_decode, token_encode
+from extensions.openai.errors import OpenAIError
 from extensions.openai.utils import _start_cloudflared
 from modules import shared
 from modules.logging_colors import logger
@@ -94,6 +95,20 @@ def verify_admin_key(authorization: str = Header(None)) -> None:
 )
 
 
+@app.exception_handler(OpenAIError)
+async def openai_error_handler(request: Request, exc: OpenAIError):
+    error_type = "server_error" if exc.code >= 500 else "invalid_request_error"
+    return JSONResponse(
+        status_code=exc.code,
+        content={"error": {
+            "message": exc.message,
+            "type": error_type,
+            "param": getattr(exc, 'param', None),
+            "code": None
+        }}
+    )
+
+
 @app.middleware("http")
 async def validate_host_header(request: Request, call_next):
     # Be strict about only approving access to localhost by default
@@ -136,6 +151,8 @@ async def generator():
                         break
 
                     yield {"data": json.dumps(resp)}
+
+                yield {"data": "[DONE]"}
             finally:
                 stop_event.set()
                 response.close()
@@ -176,6 +193,8 @@ async def generator():
                         break
 
                     yield {"data": json.dumps(resp)}
+
+                yield {"data": "[DONE]"}
             finally:
                 stop_event.set()
                 response.close()
diff --git a/extensions/openai/typing.py b/extensions/openai/typing.py
index 4d0f4a4ad8..80831c446a 100644
--- a/extensions/openai/typing.py
+++ b/extensions/openai/typing.py
@@ -152,7 +152,10 @@ class ChatCompletionRequestParams(BaseModel):
     tools: List[dict] | None = Field(default=None, description="Tools signatures passed via MCP.")
     tool_choice: str | dict | None = Field(default=None, description="Controls tool use: 'auto', 'none', 'required', or {\"type\": \"function\", \"function\": {\"name\": \"...\"}}.")
     logit_bias: dict | None = None
+    logprobs: bool | None = None
+    top_logprobs: int | None = None
     max_tokens: int | None = None
+    max_completion_tokens: int | None = None
     n: int | None = Field(default=1, description="Unused parameter.")
     presence_penalty: float | None = shared.args.presence_penalty
     stop: str | List[str] | None = None
@@ -162,6 +165,12 @@ class ChatCompletionRequestParams(BaseModel):
     top_p: float | None = shared.args.top_p
     user: str | None = Field(default=None, description="Unused parameter.")
 
+    @model_validator(mode='after')
+    def resolve_max_tokens(self):
+        if self.max_tokens is None and self.max_completion_tokens is not None:
+            self.max_tokens = self.max_completion_tokens
+        return self
+
     mode: str = Field(default='instruct', description="Valid options: instruct, chat, chat-instruct.")
 
     instruction_template: str | None = Field(default=None, description="An instruction template defined under text-generation-webui/user_data/instruction-templates. If not set, the correct template will be automatically obtained from the model metadata.")

From fb1b3b6ddf861dd3da46bde439e8bb8df8973230 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 14:16:34 -0300
Subject: [PATCH 1321/1701] API: Rewrite logprobs for OpenAI spec compliance
 across all backends

- Rewrite logprobs output format to match the OpenAI specification for
  both chat completions and completions endpoints
- Fix top_logprobs count being ignored for llama.cpp and ExLlamav3
  backends in chat completions (always returned 1 instead of requested N)
- Fix non-streaming responses only returning logprobs for the last token
  instead of all generated tokens (affects all HF-based loaders)
- Fix logprobs returning null for non-streaming chat requests on HF loaders
- Fix off-by-one returning one extra top alternative on HF loaders
---
 extensions/openai/completions.py | 185 ++++++++++++++++++++++++-------
 modules/text_generation.py       |   3 +
 modules/transformers_loader.py   |   4 +-
 3 files changed, 149 insertions(+), 43 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index a8b899d582..10cdbf4243 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -37,35 +37,114 @@ def load_chat_template_file(filepath):
     return text
 
 
-def get_logprobs_from_backend():
-    """Read logprobs captured from llama.cpp or ExLlamav3 native backend."""
+def _get_raw_logprob_entries(offset=0):
+    """Get raw logprob entries from llama.cpp/ExLlamav3 backend, starting from offset.
+
+    Returns (new_entries, new_offset).
+    """
     if not hasattr(shared.model, 'last_completion_probabilities') or not shared.model.last_completion_probabilities:
+        return [], offset
+
+    all_entries = shared.model.last_completion_probabilities
+    new_entries = all_entries[offset:]
+    return new_entries, len(all_entries)
+
+
+def _dict_to_logprob_entries(token_dict):
+    """Convert a flat {token: logprob} dict (from LogprobProcessor) to raw entry format."""
+    if not token_dict:
+        return []
+
+    return [{"top_logprobs": [{"token": t, "logprob": lp} for t, lp in token_dict.items()]}]
+
+
+def _parse_entry_top(entry):
+    """Extract the top logprobs list from a raw entry, handling both key names."""
+    return entry.get('top_logprobs', entry.get('top_probs', []))
+
+
+def format_chat_logprobs(entries):
+    """Format logprob entries into OpenAI chat completions logprobs format.
+
+    Output: {"content": [{"token", "logprob", "bytes", "top_logprobs": [...]}]}
+    """
+    if not entries:
         return None
 
-    # Both backends store data in shared.model.last_completion_probabilities
-    # Format: [{"top_logprobs": [{"token": "text", "logprob": -0.5}, ...]}, ...]
-    result = {}
-    for entry in shared.model.last_completion_probabilities:
-        top = entry.get('top_logprobs', entry.get('top_probs', []))
+    content = []
+    for entry in entries:
+        top = _parse_entry_top(entry)
+        if not top:
+            continue
+
+        chosen = top[0]
+        token_str = chosen.get('token', '')
+        token_logprob = chosen.get('logprob', chosen.get('prob', 0))
+
+        top_list = []
         for item in top:
-            token = item.get('token', '')
-            logprob = item.get('logprob', item.get('prob', 0))
-            result[token] = logprob
+            t = item.get('token', '')
+            lp = item.get('logprob', item.get('prob', 0))
+            top_list.append({
+                "token": t,
+                "logprob": lp,
+                "bytes": list(t.encode('utf-8')) if t else None
+            })
+
+        content.append({
+            "token": token_str,
+            "logprob": token_logprob,
+            "bytes": list(token_str.encode('utf-8')) if token_str else None,
+            "top_logprobs": top_list
+        })
+
+    return {"content": content} if content else None
+
+
+def format_completion_logprobs(entries):
+    """Format logprob entries into OpenAI completions logprobs format.
+
+    Output: {"tokens", "token_logprobs", "top_logprobs": [{token: prob}], "text_offset"}
+    """
+    if not entries:
+        return None
+
+    tokens = []
+    token_logprobs = []
+    top_logprobs = []
+    text_offset = []
+    offset = 0
+
+    for entry in entries:
+        top = _parse_entry_top(entry)
+        if not top:
+            continue
 
-    return result
+        chosen = top[0]
+        token_str = chosen.get('token', '')
+        token_logprob = chosen.get('logprob', chosen.get('prob', 0))
 
+        tokens.append(token_str)
+        token_logprobs.append(token_logprob)
+        text_offset.append(offset)
+        offset += len(token_str)
 
-def convert_logprobs_to_tiktoken(model, logprobs):
-    # more problems than it's worth.
-    # try:
-    #     encoder = tiktoken.encoding_for_model(model)
-    #     # just pick the first one if it encodes to multiple tokens... 99.9% not required and maybe worse overall.
-    #     return dict([(encoder.decode([encoder.encode(token)[0]]), prob) for token, prob in logprobs.items()])
-    # except KeyError:
-    #     # assume native tokens if we can't find the tokenizer
-    #     return logprobs
+        top_dict = {}
+        for item in top:
+            t = item.get('token', '')
+            lp = item.get('logprob', item.get('prob', 0))
+            top_dict[t] = lp
+        top_logprobs.append(top_dict)
 
-    return logprobs
+    if not tokens:
+        return None
+
+    return {
+        "tokens": tokens,
+        "token_logprobs": token_logprobs,
+        "top_logprobs": top_logprobs,
+        "text_offset": text_offset
+    }
 
 
 def process_parameters(body, is_legacy=False):
@@ -90,6 +169,14 @@ def process_parameters(body, is_legacy=False):
         elif isinstance(body['stop'], list):
             generate_params['custom_stopping_strings'] = body['stop']
 
+    # Resolve logprobs: for chat completions, logprobs is a bool and the count
+    # comes from top_logprobs. Normalize to an int for all backends.
+    logprobs = body.get('logprobs', None)
+    top_logprobs = body.get('top_logprobs', None)
+    if logprobs is True:
+        logprobs = top_logprobs if top_logprobs and top_logprobs > 0 else 5
+        generate_params['logprobs'] = logprobs
+
     # For llama.cpp and ExLlamav3 native, logit_bias and logprobs are forwarded natively
     if shared.args.loader not in ('llama.cpp', 'ExLlamav3'):
         from transformers import LogitsProcessorList
@@ -104,11 +191,6 @@ def process_parameters(body, is_legacy=False):
         if logit_bias:  # {str: float, ...}
             logits_processor = [LogitsBiasProcessor(logit_bias)]
 
-        logprobs = body.get('logprobs', None)
-        top_logprobs = body.get('top_logprobs', None)
-        # For chat completions, logprobs is a bool; use top_logprobs for the count
-        if logprobs is True:
-            logprobs = top_logprobs if top_logprobs and top_logprobs > 0 else 5
         if logprobs is not None and logprobs > 0:
             generate_params['logprob_proc'] = LogprobProcessor(logprobs)
             logits_processor.extend([generate_params['logprob_proc']])
@@ -317,6 +399,9 @@ def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, p
 
     requested_model = generate_params.pop('model')
     logprob_proc = generate_params.pop('logprob_proc', None)
+    if logprob_proc:
+        logprob_proc.token_alternatives_history.clear()
+    chat_logprobs_offset = [0]  # mutable for closure access in streaming
 
     def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False):
         # begin streaming
@@ -344,12 +429,16 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
         }
 
         if logprob_proc:
-            top_logprobs = convert_logprobs_to_tiktoken(model=requested_model, logprobs=logprob_proc.token_alternatives)
-            chunk[resp_list][0]["logprobs"] = {'top_logprobs': [top_logprobs]}
+            entries = _dict_to_logprob_entries(logprob_proc.token_alternatives)
+            formatted = format_chat_logprobs(entries)
+            if formatted:
+                chunk[resp_list][0]["logprobs"] = formatted
         elif shared.args.loader in ('llama.cpp', 'ExLlamav3'):
-            backend_logprobs = get_logprobs_from_backend()
-            if backend_logprobs:
-                chunk[resp_list][0]["logprobs"] = {'top_logprobs': [backend_logprobs]}
+            entries, chat_logprobs_offset[0] = _get_raw_logprob_entries(chat_logprobs_offset[0])
+            if entries:
+                formatted = format_chat_logprobs(entries)
+                if formatted:
+                    chunk[resp_list][0]["logprobs"] = formatted
 
         return chunk
 
@@ -471,12 +560,18 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
             }
         }
         if logprob_proc:
-            top_logprobs = convert_logprobs_to_tiktoken(model=requested_model, logprobs=logprob_proc.token_alternatives)
-            resp[resp_list][0]["logprobs"] = {'top_logprobs': [top_logprobs]}
+            all_entries = []
+            for alt in logprob_proc.token_alternatives_history:
+                all_entries.extend(_dict_to_logprob_entries(alt))
+            formatted = format_chat_logprobs(all_entries)
+            if formatted:
+                resp[resp_list][0]["logprobs"] = formatted
         elif shared.args.loader in ('llama.cpp', 'ExLlamav3'):
-            backend_logprobs = get_logprobs_from_backend()
-            if backend_logprobs:
-                resp[resp_list][0]["logprobs"] = {'top_logprobs': [backend_logprobs]}
+            raw = getattr(shared.model, 'last_completion_probabilities', None)
+            if raw:
+                formatted = format_chat_logprobs(raw)
+                if formatted:
+                    resp[resp_list][0]["logprobs"] = formatted
 
         yield resp
 
@@ -518,6 +613,8 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_e
         generate_params['stop_event'] = stop_event
     requested_model = generate_params.pop('model')
     logprob_proc = generate_params.pop('logprob_proc', None)
+    if logprob_proc:
+        logprob_proc.token_alternatives_history.clear()
     suffix = body['suffix'] if body['suffix'] else ''
     echo = body['echo']
 
@@ -583,10 +680,13 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_e
                     stop_reason = "length"
 
                 if logprob_proc:
-                    completion_logprobs = {'top_logprobs': [logprob_proc.token_alternatives]}
+                    all_entries = []
+                    for alt in logprob_proc.token_alternatives_history:
+                        all_entries.extend(_dict_to_logprob_entries(alt))
+                    completion_logprobs = format_completion_logprobs(all_entries)
                 elif shared.args.loader in ('llama.cpp', 'ExLlamav3'):
-                    backend_logprobs = get_logprobs_from_backend()
-                    completion_logprobs = {'top_logprobs': [backend_logprobs]} if backend_logprobs else None
+                    raw = getattr(shared.model, 'last_completion_probabilities', None)
+                    completion_logprobs = format_completion_logprobs(raw)
                 else:
                     completion_logprobs = None
 
@@ -633,14 +733,15 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_e
         # Check if usage should be included in streaming chunks per OpenAI spec
         stream_options = body.get('stream_options')
         include_usage = bool(stream_options) and bool(stream_options.get('include_usage') if isinstance(stream_options, dict) else getattr(stream_options, 'include_usage', False))
+        cmpl_logprobs_offset = [0]  # mutable for closure access in streaming
 
         def text_streaming_chunk(content):
             # begin streaming
             if logprob_proc:
-                chunk_logprobs = {'top_logprobs': [logprob_proc.token_alternatives]}
+                chunk_logprobs = format_completion_logprobs(_dict_to_logprob_entries(logprob_proc.token_alternatives))
             elif shared.args.loader in ('llama.cpp', 'ExLlamav3'):
-                backend_logprobs = get_logprobs_from_backend()
-                chunk_logprobs = {'top_logprobs': [backend_logprobs]} if backend_logprobs else None
+                entries, cmpl_logprobs_offset[0] = _get_raw_logprob_entries(cmpl_logprobs_offset[0])
+                chunk_logprobs = format_completion_logprobs(entries) if entries else None
             else:
                 chunk_logprobs = None
 
diff --git a/modules/text_generation.py b/modules/text_generation.py
index c78afe3e80..787c1814cd 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -78,10 +78,13 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
     reply = ''
     is_stream = state['stream']
     if len(all_stop_strings) > 0 and not state['stream']:
+        original_logits_processor = state.get('logits_processor')
         stop_event_ref = state.pop('stop_event', None)
         state = copy.deepcopy(state)
         if stop_event_ref is not None:
             state['stop_event'] = stop_event_ref
+        if original_logits_processor is not None:
+            state['logits_processor'] = original_logits_processor
         state['stream'] = True
 
     # Generate
diff --git a/modules/transformers_loader.py b/modules/transformers_loader.py
index d57020c6dd..b9918764bc 100644
--- a/modules/transformers_loader.py
+++ b/modules/transformers_loader.py
@@ -65,14 +65,16 @@ class LogprobProcessor(LogitsProcessor):
     def __init__(self, logprobs=None):
         self.logprobs = logprobs
         self.token_alternatives = {}
+        self.token_alternatives_history = []
 
     def __call__(self, input_ids: torch.LongTensor, logits: torch.FloatTensor) -> torch.FloatTensor:
         if self.logprobs is not None:  # 0-5
             log_e_probabilities = F.log_softmax(logits, dim=1)
-            top_values, top_indices = torch.topk(log_e_probabilities, k=self.logprobs + 1)
+            top_values, top_indices = torch.topk(log_e_probabilities, k=self.logprobs)
             top_tokens = [get_reply_from_output_ids([tok]) for tok in top_indices[0]]
             top_probs = [float(x) for x in top_values[0]]
             self.token_alternatives = dict(zip(top_tokens, top_probs))
+            self.token_alternatives_history.append(self.token_alternatives)
 
         return logits
 

From a916fb0e5c81449b4644119decfa86958278860d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 14:27:24 -0300
Subject: [PATCH 1322/1701] API: Preserve mid-conversation system message
 positions

---
 extensions/openai/completions.py | 14 +++++++++++++-
 modules/chat.py                  |  5 ++++-
 2 files changed, 17 insertions(+), 2 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 10cdbf4243..37e9568ab3 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -239,12 +239,14 @@ def convert_history(history):
     user_input = ""
     user_input_last = True
     system_message = ""
+    seen_non_system = False
 
     for entry in history:
         content = entry["content"]
         role = entry["role"]
 
         if role == "user":
+            seen_non_system = True
             # Extract text content (images handled by model-specific code)
             content = process_multimodal_content(content)
             user_input = content
@@ -256,6 +258,7 @@ def convert_history(history):
 
             current_message = content
         elif role == "assistant":
+            seen_non_system = True
             meta = {}
             tool_calls = entry.get("tool_calls")
             if tool_calls and isinstance(tool_calls, list) and len(tool_calls) > 0:
@@ -272,13 +275,22 @@ def convert_history(history):
             else:
                 chat_dialogue.append(['', current_reply, '', meta])
         elif role == "tool":
+            seen_non_system = True
             user_input_last = False
             meta = {}
             if "tool_call_id" in entry:
                 meta["tool_call_id"] = entry["tool_call_id"]
             chat_dialogue.append(['', '', content, meta])
         elif role in ("system", "developer"):
-            system_message += f"\n{content}" if system_message else content
+            if not seen_non_system:
+                # Leading system messages go to custom_system_message (placed at top)
+                system_message += f"\n{content}" if system_message else content
+            else:
+                # Mid-conversation system messages: preserve position in history
+                if current_message:
+                    chat_dialogue.append([current_message, '', '', {}])
+                    current_message = ""
+                chat_dialogue.append([content, '', '', {"role": "system"}])
 
     if not user_input_last:
         user_input = ""
diff --git a/modules/chat.py b/modules/chat.py
index 5cf550d679..8af922737a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -315,7 +315,10 @@ def generate_chat_prompt(user_input, state, **kwargs):
                         "tool_call_id": item.get('tool_call_id', '')
                     })
 
-        if user_msg not in ['', '<|BEGIN-VISIBLE-CHAT|>']:
+        if entry_meta.get('role') == 'system':
+            if user_msg:
+                messages.insert(insert_pos, {"role": "system", "content": user_msg})
+        elif user_msg not in ['', '<|BEGIN-VISIBLE-CHAT|>']:
             # Check for user message attachments in metadata
             user_key = f"user_{row_idx}"
             enhanced_user_msg = user_msg

From 2466305f76a7c197178096021eb3f18e0449410a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 16:02:57 -0300
Subject: [PATCH 1323/1701] Add tool examples

---
 user_data/tools/get_datetime.py | 18 ++++++++++++++++++
 user_data/tools/roll_dice.py    | 23 +++++++++++++++++++++++
 user_data/tools/web_search.py   | 27 +++++++++++++++++++++++++++
 3 files changed, 68 insertions(+)
 create mode 100644 user_data/tools/get_datetime.py
 create mode 100644 user_data/tools/roll_dice.py
 create mode 100644 user_data/tools/web_search.py

diff --git a/user_data/tools/get_datetime.py b/user_data/tools/get_datetime.py
new file mode 100644
index 0000000000..f0a9277792
--- /dev/null
+++ b/user_data/tools/get_datetime.py
@@ -0,0 +1,18 @@
+from datetime import datetime
+
+tool = {
+    "type": "function",
+    "function": {
+        "name": "get_datetime",
+        "description": "Get the current date and time.",
+        "parameters": {
+            "type": "object",
+            "properties": {},
+        }
+    }
+}
+
+
+def execute(arguments):
+    now = datetime.now()
+    return {"date": now.strftime("%Y-%m-%d"), "time": now.strftime("%I:%M %p")}
diff --git a/user_data/tools/roll_dice.py b/user_data/tools/roll_dice.py
new file mode 100644
index 0000000000..9cab48a833
--- /dev/null
+++ b/user_data/tools/roll_dice.py
@@ -0,0 +1,23 @@
+import random
+
+tool = {
+    "type": "function",
+    "function": {
+        "name": "roll_dice",
+        "description": "Roll one or more dice with the specified number of sides.",
+        "parameters": {
+            "type": "object",
+            "properties": {
+                "count": {"type": "integer", "description": "Number of dice to roll.", "default": 1},
+                "sides": {"type": "integer", "description": "Number of sides per die.", "default": 20},
+            },
+        }
+    }
+}
+
+
+def execute(arguments):
+    count = arguments.get("count", 1)
+    sides = arguments.get("sides", 20)
+    rolls = [random.randint(1, sides) for _ in range(count)]
+    return {"rolls": rolls, "total": sum(rolls)}
diff --git a/user_data/tools/web_search.py b/user_data/tools/web_search.py
new file mode 100644
index 0000000000..8923eab0a9
--- /dev/null
+++ b/user_data/tools/web_search.py
@@ -0,0 +1,27 @@
+from modules.web_search import perform_web_search
+
+tool = {
+    "type": "function",
+    "function": {
+        "name": "web_search",
+        "description": "Search the web using DuckDuckGo and return page contents.",
+        "parameters": {
+            "type": "object",
+            "properties": {
+                "query": {"type": "string", "description": "The search query."},
+            },
+            "required": ["query"]
+        }
+    }
+}
+
+
+def execute(arguments):
+    query = arguments.get("query", "")
+    results = perform_web_search(query, num_pages=3)
+    output = []
+    for r in results:
+        if r and r["content"].strip():
+            output.append({"title": r["title"], "url": r["url"], "content": r["content"][:4000]})
+
+    return output if output else [{"error": "No results found."}]

From d45c9b3c5957692f2f8900cc0e19a22220bc114a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 16:06:00 -0300
Subject: [PATCH 1324/1701] API: Minor logprobs fixes

---
 extensions/openai/completions.py | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 37e9568ab3..eb1702a83b 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -98,7 +98,7 @@ def format_chat_logprobs(entries):
             "top_logprobs": top_list
         })
 
-    return {"content": content} if content else None
+    return {"content": content, "refusal": None} if content else None
 
 
 def format_completion_logprobs(entries):
@@ -174,7 +174,7 @@ def process_parameters(body, is_legacy=False):
     logprobs = body.get('logprobs', None)
     top_logprobs = body.get('top_logprobs', None)
     if logprobs is True:
-        logprobs = top_logprobs if top_logprobs and top_logprobs > 0 else 5
+        logprobs = max(top_logprobs, 1) if top_logprobs is not None else 5
         generate_params['logprobs'] = logprobs
 
     # For llama.cpp and ExLlamav3 native, logit_bias and logprobs are forwarded natively
@@ -677,6 +677,9 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_e
                 if original_seed >= 0:
                     generate_params['seed'] = original_seed + _n
 
+                if logprob_proc:
+                    logprob_proc.token_alternatives_history.clear()
+
                 # generate reply #######################################
                 debug_msg({'prompt': prompt, 'generate_params': generate_params})
                 generator = generate_reply(prompt, generate_params, is_chat=False)

From 2d0cc7726eae0d98e52352b740397d86cdd2973e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 16:29:46 -0300
Subject: [PATCH 1325/1701] API: Add reasoning_content field to non-streaming
 chat completions

Extract thinking/reasoning blocks (e.g. ...) into a
separate reasoning_content field on the assistant message, matching
the convention used by DeepSeek, llama.cpp, and SGLang.
---
 extensions/openai/completions.py | 11 ++++-
 modules/html_generator.py        | 62 ++--------------------------
 modules/reasoning.py             | 71 ++++++++++++++++++++++++++++++++
 3 files changed, 84 insertions(+), 60 deletions(-)
 create mode 100644 modules/reasoning.py

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index eb1702a83b..0eb0cd2748 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -13,6 +13,7 @@
 from extensions.openai.typing import ToolDefinition
 from extensions.openai.utils import debug_msg, getToolCallId, parseToolCall
 from modules import shared
+from modules.reasoning import extract_reasoning
 from modules.chat import (
     generate_chat_prompt,
     generate_chat_reply,
@@ -553,6 +554,14 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
         else:
             yield chunk
     else:
+        reasoning, content = extract_reasoning(answer) if not tool_calls else (None, answer)
+        message = {
+            "role": "assistant",
+            "refusal": None,
+            "content": None if tool_calls else content,
+            **({"reasoning_content": reasoning} if reasoning else {}),
+            **({"tool_calls": tool_calls} if tool_calls else {}),
+        }
         resp = {
             "id": cmpl_id,
             "object": object_type,
@@ -562,7 +571,7 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
             resp_list: [{
                 "index": 0,
                 "finish_reason": stop_reason,
-                "message": {"role": "assistant", "refusal": None, "content": None if tool_calls else answer, **({"tool_calls": tool_calls} if tool_calls else {})},
+                "message": message,
                 "logprobs": None,
             }],
             "usage": {
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 472a9ea0fc..4d9904fb2a 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -10,6 +10,7 @@
 from PIL import Image, ImageOps
 
 from modules import shared
+from modules.reasoning import extract_reasoning
 from modules.sane_markdown_lists import SaneListExtension
 from modules.utils import get_available_chat_styles
 
@@ -108,66 +109,9 @@ def replace_blockquote(m):
     return m.group().replace('\n', '\n> ').replace('\\begin{blockquote}', '').replace('\\end{blockquote}', '')
 
 
-# Thinking block format definitions: (start_tag, end_tag, content_start_tag)
-# Use None for start_tag to match from beginning (end-only formats should be listed last)
-THINKING_FORMATS = [
-    ('', '', None),
-    ('<|channel|>analysis<|message|>', '<|end|>', '<|start|>assistant<|channel|>final<|message|>'),
-    ('', '', None),
-    ('<|think|>', '<|end|>', '<|content|>'),  # Solar Open
-    ('Thinking Process:', '', None),  # Qwen3.5 verbose thinking outside tags
-    (None, '', None),  # End-only variant (e.g., Qwen3-next)
-]
-
-
 def extract_thinking_block(string):
-    """Extract thinking blocks from the beginning of a string."""
-    if not string:
-        return None, string
-
-    for start_tag, end_tag, content_tag in THINKING_FORMATS:
-        end_esc = html.escape(end_tag)
-        content_esc = html.escape(content_tag) if content_tag else None
-
-        if start_tag is None:
-            # End-only format: require end tag, start from beginning
-            end_pos = string.find(end_esc)
-            if end_pos == -1:
-                continue
-            thought_start = 0
-        else:
-            # Normal format: require start tag
-            start_esc = html.escape(start_tag)
-            start_pos = string.find(start_esc)
-            if start_pos == -1:
-                continue
-            thought_start = start_pos + len(start_esc)
-            end_pos = string.find(end_esc, thought_start)
-
-        if end_pos == -1:
-            # End tag missing - check if content tag can serve as fallback
-            if content_esc:
-                content_pos = string.find(content_esc, thought_start)
-                if content_pos != -1:
-                    thought_end = content_pos
-                    content_start = content_pos + len(content_esc)
-                else:
-                    thought_end = len(string)
-                    content_start = len(string)
-            else:
-                thought_end = len(string)
-                content_start = len(string)
-        else:
-            thought_end = end_pos
-            if content_esc:
-                content_pos = string.find(content_esc, end_pos)
-                content_start = content_pos + len(content_esc) if content_pos != -1 else end_pos + len(end_esc)
-            else:
-                content_start = end_pos + len(end_esc)
-
-        return string[thought_start:thought_end], string[content_start:]
-
-    return None, string
+    """Extract thinking blocks from the beginning of an HTML-escaped string."""
+    return extract_reasoning(string, html_escaped=True)
 
 
 def build_thinking_block(thinking_content, message_id, has_remaining_content):
diff --git a/modules/reasoning.py b/modules/reasoning.py
new file mode 100644
index 0000000000..12f8553d98
--- /dev/null
+++ b/modules/reasoning.py
@@ -0,0 +1,71 @@
+import html as html_module
+
+# Thinking block format definitions: (start_tag, end_tag, content_start_tag)
+# Use None for start_tag to match from beginning (end-only formats should be listed last)
+THINKING_FORMATS = [
+    ('', '', None),
+    ('<|channel|>analysis<|message|>', '<|end|>', '<|start|>assistant<|channel|>final<|message|>'),
+    ('', '', None),
+    ('<|think|>', '<|end|>', '<|content|>'),  # Solar Open
+    ('Thinking Process:', '', None),  # Qwen3.5 verbose thinking outside tags
+    (None, '', None),  # End-only variant (e.g., Qwen3-next)
+]
+
+
+def extract_reasoning(text, html_escaped=False):
+    """Extract reasoning/thinking blocks from the beginning of a string.
+
+    When html_escaped=True, tags are HTML-escaped before searching
+    (for use on already-escaped UI strings).
+
+    Returns (reasoning_content, final_content) where reasoning_content is
+    None if no thinking block is found.
+    """
+    if not text:
+        return None, text
+
+    esc = html_module.escape if html_escaped else lambda s: s
+
+    for start_tag, end_tag, content_tag in THINKING_FORMATS:
+        end_esc = esc(end_tag)
+        content_esc = esc(content_tag) if content_tag else None
+
+        if start_tag is None:
+            # End-only format: require end tag, start from beginning
+            end_pos = text.find(end_esc)
+            if end_pos == -1:
+                continue
+            thought_start = 0
+        else:
+            # Normal format: require start tag
+            start_esc = esc(start_tag)
+            start_pos = text.find(start_esc)
+            if start_pos == -1:
+                continue
+            thought_start = start_pos + len(start_esc)
+            end_pos = text.find(end_esc, thought_start)
+
+        if end_pos == -1:
+            # End tag missing - check if content tag can serve as fallback
+            if content_esc:
+                content_pos = text.find(content_esc, thought_start)
+                if content_pos != -1:
+                    thought_end = content_pos
+                    content_start = content_pos + len(content_esc)
+                else:
+                    thought_end = len(text)
+                    content_start = len(text)
+            else:
+                thought_end = len(text)
+                content_start = len(text)
+        else:
+            thought_end = end_pos
+            if content_esc:
+                content_pos = text.find(content_esc, end_pos)
+                content_start = content_pos + len(content_esc) if content_pos != -1 else end_pos + len(end_esc)
+            else:
+                content_start = end_pos + len(end_esc)
+
+        return text[thought_start:thought_end], text[content_start:]
+
+    return None, text

From 9a7428b6271ffadf2094f93dd1e385c7c8444d71 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 14:16:04 -0700
Subject: [PATCH 1326/1701] UI: Add collapsible accordions for tool calling
 steps

---
 extensions/openai/utils.py |  93 +++++++++++++++++++++----------
 modules/chat.py            |  68 +++++++++++++++++++----
 modules/html_generator.py  | 110 ++++++++++++++++++++++++++++++-------
 3 files changed, 212 insertions(+), 59 deletions(-)

diff --git a/extensions/openai/utils.py b/extensions/openai/utils.py
index f4a31d1a72..6fb05f0855 100644
--- a/extensions/openai/utils.py
+++ b/extensions/openai/utils.py
@@ -123,6 +123,7 @@ def _parseChannelToolCalls(answer: str, tool_names: list[str]):
         <|channel|>commentary to=functions.func_name <|constrain|>json<|message|>{"arg": "value"}
     """
     matches = []
+    start_pos = None
     for m in re.finditer(
         r'<\|channel\|>commentary to=functions\.([^<\s]+)\s*(?:<\|constrain\|>json)?<\|message\|>',
         answer
@@ -135,6 +136,8 @@ def _parseChannelToolCalls(answer: str, tool_names: list[str]):
             continue
         try:
             arguments = json.loads(json_str)
+            if start_pos is None:
+                start_pos = m.start()
             matches.append({
                 "type": "function",
                 "function": {
@@ -144,7 +147,7 @@ def _parseChannelToolCalls(answer: str, tool_names: list[str]):
             })
         except json.JSONDecodeError:
             pass
-    return matches
+    return matches, start_pos
 
 
 def _parseBareNameToolCalls(answer: str, tool_names: list[str]):
@@ -155,6 +158,7 @@ def _parseBareNameToolCalls(answer: str, tool_names: list[str]):
     Multiple calls are concatenated directly or separated by whitespace.
     """
     matches = []
+    start_pos = None
     # Match tool name followed by opening brace, then extract balanced JSON
     escaped_names = [re.escape(name) for name in tool_names]
     pattern = r'(?:' + '|'.join(escaped_names) + r')\s*\{'
@@ -173,6 +177,8 @@ def _parseBareNameToolCalls(answer: str, tool_names: list[str]):
             continue
         try:
             arguments = json.loads(json_str)
+            if start_pos is None:
+                start_pos = match.start()
             matches.append({
                 "type": "function",
                 "function": {
@@ -182,7 +188,7 @@ def _parseBareNameToolCalls(answer: str, tool_names: list[str]):
             })
         except json.JSONDecodeError:
             pass
-    return matches
+    return matches, start_pos
 
 
 def _parseXmlParamToolCalls(answer: str, tool_names: list[str]):
@@ -196,6 +202,7 @@ def _parseXmlParamToolCalls(answer: str, tool_names: list[str]):
         
     """
     matches = []
+    start_pos = None
     for tc_match in re.finditer(r'\s*(.*?)\s*', answer, re.DOTALL):
         tc_content = tc_match.group(1)
         func_match = re.search(r']+)>', tc_content)
@@ -213,6 +220,8 @@ def _parseXmlParamToolCalls(answer: str, tool_names: list[str]):
             except (json.JSONDecodeError, ValueError):
                 pass  # keep as string
             arguments[param_name] = param_value
+        if start_pos is None:
+            start_pos = tc_match.start()
         matches.append({
             "type": "function",
             "function": {
@@ -220,7 +229,7 @@ def _parseXmlParamToolCalls(answer: str, tool_names: list[str]):
                 "arguments": arguments
             }
         })
-    return matches
+    return matches, start_pos
 
 
 def _parseKimiToolCalls(answer: str, tool_names: list[str]):
@@ -232,6 +241,7 @@ def _parseKimiToolCalls(answer: str, tool_names: list[str]):
         <|tool_calls_section_end|>
     """
     matches = []
+    start_pos = None
     for m in re.finditer(
         r'<\|tool_call_begin\|>\s*(?:functions\.)?(\S+?)(?::\d+)?\s*<\|tool_call_argument_begin\|>\s*',
         answer
@@ -244,6 +254,10 @@ def _parseKimiToolCalls(answer: str, tool_names: list[str]):
             continue
         try:
             arguments = json.loads(json_str)
+            if start_pos is None:
+                # Check for section begin marker before the call marker
+                section = answer.rfind('<|tool_calls_section_begin|>', 0, m.start())
+                start_pos = section if section != -1 else m.start()
             matches.append({
                 "type": "function",
                 "function": {
@@ -253,7 +267,7 @@ def _parseKimiToolCalls(answer: str, tool_names: list[str]):
             })
         except json.JSONDecodeError:
             pass
-    return matches
+    return matches, start_pos
 
 
 def _parseMiniMaxToolCalls(answer: str, tool_names: list[str]):
@@ -267,6 +281,7 @@ def _parseMiniMaxToolCalls(answer: str, tool_names: list[str]):
         
     """
     matches = []
+    start_pos = None
     for tc_match in re.finditer(r'\s*(.*?)\s*', answer, re.DOTALL):
         tc_content = tc_match.group(1)
         # Split on  to handle multiple parallel calls in one block
@@ -284,6 +299,8 @@ def _parseMiniMaxToolCalls(answer: str, tool_names: list[str]):
                 except (json.JSONDecodeError, ValueError):
                     pass  # keep as string
                 arguments[param_name] = param_value
+            if start_pos is None:
+                start_pos = tc_match.start()
             matches.append({
                 "type": "function",
                 "function": {
@@ -291,7 +308,7 @@ def _parseMiniMaxToolCalls(answer: str, tool_names: list[str]):
                     "arguments": arguments
                 }
             })
-    return matches
+    return matches, start_pos
 
 
 def _parseDeepSeekToolCalls(answer: str, tool_names: list[str]):
@@ -301,6 +318,7 @@ def _parseDeepSeekToolCalls(answer: str, tool_names: list[str]):
         <｜tool▁calls▁begin｜><｜tool▁call▁begin｜>func_name<｜tool▁sep｜>{"arg": "value"}<｜tool▁call▁end｜><｜tool▁calls▁end｜>
     """
     matches = []
+    start_pos = None
     for m in re.finditer(
         r'<｜tool▁call▁begin｜>\s*(\S+?)\s*<｜tool▁sep｜>\s*',
         answer
@@ -313,6 +331,10 @@ def _parseDeepSeekToolCalls(answer: str, tool_names: list[str]):
             continue
         try:
             arguments = json.loads(json_str)
+            if start_pos is None:
+                # Check for section begin marker before the call marker
+                section = answer.rfind('<｜tool▁calls▁begin｜>', 0, m.start())
+                start_pos = section if section != -1 else m.start()
             matches.append({
                 "type": "function",
                 "function": {
@@ -322,7 +344,7 @@ def _parseDeepSeekToolCalls(answer: str, tool_names: list[str]):
             })
         except json.JSONDecodeError:
             pass
-    return matches
+    return matches, start_pos
 
 
 def _parseGlmToolCalls(answer: str, tool_names: list[str]):
@@ -335,6 +357,7 @@ def _parseGlmToolCalls(answer: str, tool_names: list[str]):
         
     """
     matches = []
+    start_pos = None
     for tc_match in re.finditer(r'\s*(.*?)\s*', answer, re.DOTALL):
         tc_content = tc_match.group(1)
         # First non-tag text is the function name
@@ -356,6 +379,8 @@ def _parseGlmToolCalls(answer: str, tool_names: list[str]):
             except (json.JSONDecodeError, ValueError):
                 pass  # keep as string
             arguments[k] = v
+        if start_pos is None:
+            start_pos = tc_match.start()
         matches.append({
             "type": "function",
             "function": {
@@ -363,7 +388,7 @@ def _parseGlmToolCalls(answer: str, tool_names: list[str]):
                 "arguments": arguments
             }
         })
-    return matches
+    return matches, start_pos
 
 
 def _parsePythonicToolCalls(answer: str, tool_names: list[str]):
@@ -373,10 +398,11 @@ def _parsePythonicToolCalls(answer: str, tool_names: list[str]):
         [func_name(param1="value1", param2="value2"), func_name2(...)]
     """
     matches = []
+    start_pos = None
     # Match a bracketed list of function calls
     bracket_match = re.search(r'\[([^\[\]]+)\]', answer)
     if not bracket_match:
-        return matches
+        return matches, start_pos
 
     inner = bracket_match.group(1)
 
@@ -411,6 +437,8 @@ def _parsePythonicToolCalls(answer: str, tool_names: list[str]):
                     pass
                 arguments[param_name] = param_value
 
+        if start_pos is None:
+            start_pos = bracket_match.start()
         matches.append({
             "type": "function",
             "function": {
@@ -419,55 +447,62 @@ def _parsePythonicToolCalls(answer: str, tool_names: list[str]):
             }
         })
 
-    return matches
+    return matches, start_pos
 
 
-def parseToolCall(answer: str, tool_names: list[str]):
+def parseToolCall(answer: str, tool_names: list[str], return_prefix: bool = False):
     matches = []
+    start_pos = None
+
+    def _return(matches, start_pos):
+        if return_prefix:
+            prefix = answer[:start_pos] if matches and start_pos is not None else ''
+            return matches, prefix
+        return matches
 
     # abort on very short answers to save computation cycles
     if len(answer) < 10:
-        return matches
+        return _return(matches, start_pos)
 
     # Check for DeepSeek-style tool calls (fullwidth Unicode token delimiters)
-    matches = _parseDeepSeekToolCalls(answer, tool_names)
+    matches, start_pos = _parseDeepSeekToolCalls(answer, tool_names)
     if matches:
-        return matches
+        return _return(matches, start_pos)
 
     # Check for Kimi-K2-style tool calls (pipe-delimited tokens)
-    matches = _parseKimiToolCalls(answer, tool_names)
+    matches, start_pos = _parseKimiToolCalls(answer, tool_names)
     if matches:
-        return matches
+        return _return(matches, start_pos)
 
     # Check for channel-based tool calls (e.g. GPT-OSS format)
-    matches = _parseChannelToolCalls(answer, tool_names)
+    matches, start_pos = _parseChannelToolCalls(answer, tool_names)
     if matches:
-        return matches
+        return _return(matches, start_pos)
 
     # Check for MiniMax-style tool calls (invoke/parameter XML tags)
-    matches = _parseMiniMaxToolCalls(answer, tool_names)
+    matches, start_pos = _parseMiniMaxToolCalls(answer, tool_names)
     if matches:
-        return matches
+        return _return(matches, start_pos)
 
     # Check for GLM-style tool calls (arg_key/arg_value XML pairs)
-    matches = _parseGlmToolCalls(answer, tool_names)
+    matches, start_pos = _parseGlmToolCalls(answer, tool_names)
     if matches:
-        return matches
+        return _return(matches, start_pos)
 
     # Check for XML-parameter style tool calls (e.g. Qwen3.5 format)
-    matches = _parseXmlParamToolCalls(answer, tool_names)
+    matches, start_pos = _parseXmlParamToolCalls(answer, tool_names)
     if matches:
-        return matches
+        return _return(matches, start_pos)
 
     # Check for bare function-name style tool calls (e.g. Mistral format)
-    matches = _parseBareNameToolCalls(answer, tool_names)
+    matches, start_pos = _parseBareNameToolCalls(answer, tool_names)
     if matches:
-        return matches
+        return _return(matches, start_pos)
 
     # Check for pythonic-style tool calls (e.g. Llama 4 format)
-    matches = _parsePythonicToolCalls(answer, tool_names)
+    matches, start_pos = _parsePythonicToolCalls(answer, tool_names)
     if matches:
-        return matches
+        return _return(matches, start_pos)
 
     # Define the regex pattern to find the JSON content wrapped in , , , and other tags observed from various models
     patterns = [r"(```[^\n]*)\n(.*?)```", r"<([^>]+)>(.*?)"]
@@ -501,6 +536,8 @@ def parseToolCall(answer: str, tool_names: list[str]):
             for candidate_dict in candidates:
                 checked_candidate = checkAndSanitizeToolCallCandidate(candidate_dict, tool_names)
                 if checked_candidate is not None:
+                    if start_pos is None:
+                        start_pos = match.start()
                     matches.append(checked_candidate)
 
         # last resort if nothing has been mapped: LLM might have produced plain json tool call without xml-like tags
@@ -524,4 +561,4 @@ def parseToolCall(answer: str, tool_names: list[str]):
                 # Ignore invalid JSON silently
                 pass
 
-    return matches
+    return _return(matches, start_pos)
diff --git a/modules/chat.py b/modules/chat.py
index 8af922737a..d3cd0caecc 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -23,6 +23,7 @@
 from modules.html_generator import (
     chat_html_wrapper,
     convert_to_markdown,
+    extract_thinking_block,
     make_thumbnail
 )
 from modules.image_utils import open_image_safely
@@ -1168,12 +1169,19 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
         cur_text = text if _tool_turn == 0 else ''
 
         for i, history in enumerate(generate_chat_reply(cur_text, state, regen, cont, loading_message=True, for_ui=True)):
-            # Prepend accumulated tool output to visible reply
+            # Prepend accumulated tool output to visible reply for display.
+            # Save and restore the original to prevent the markers from leaking
+            # back into chatbot_wrapper's shared output object, which would cause
+            # duplication on the next yield.
+            _original_visible = history['visible'][-1][1] if visible_prefix else None
             if visible_prefix:
-                history['visible'][-1][1] = '\n\n'.join(visible_prefix + [history['visible'][-1][1]])
+                history['visible'][-1][1] = '\n\n'.join(visible_prefix + [_original_visible])
 
             yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'], last_message_only=(i > 0)), history
 
+            if visible_prefix:
+                history['visible'][-1][1] = _original_visible
+
             if i == 0:
                 # Save old tool_sequence into version 0 (created by chatbot_wrapper
                 # on the first yield).  Only needed on the first regeneration when
@@ -1196,6 +1204,15 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
             if tool_func_names and parseToolCall(history['internal'][-1][1], tool_func_names):
                 break
 
+        # Save the model's visible output before re-applying visible_prefix,
+        # so we can extract thinking content from just this turn's output.
+        _model_visible = history['visible'][-1][1]
+
+        # Re-apply visible prefix to the final state after streaming completes.
+        # This is safe because we're no longer sharing the object with chatbot_wrapper.
+        if visible_prefix:
+            history['visible'][-1][1] = '\n\n'.join(visible_prefix + [_model_visible])
+
         save_history(history, state['unique_id'], state['character_menu'], state['mode'])
 
         # Check for tool calls
@@ -1203,7 +1220,7 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
             break
 
         answer = history['internal'][-1][1]
-        parsed_calls = parseToolCall(answer, tool_func_names) if answer else None
+        parsed_calls, content_prefix = parseToolCall(answer, tool_func_names, return_prefix=True) if answer else (None, '')
 
         if not parsed_calls:
             break  # No tool calls — done
@@ -1232,12 +1249,38 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
         # Clear internal (raw tool markup)
         history['internal'][-1][1] = ''
 
-        # Add call summary to visible prefix
-        call_summary = ', '.join(f'{tc["function"]["name"]}(...)' for tc in parsed_calls)
-        visible_prefix.append('Calling: ' + call_summary)
-
-        # Execute tools, store results
+        # Preserve thinking block and intermediate text from this turn.
+        # content_prefix is the raw text before tool call syntax (returned
+        # by parseToolCall); HTML-escape it and extract thinking to get
+        # the content the user should see.
+        content_text = html.escape(content_prefix)
+        thinking_content, intermediate = extract_thinking_block(content_text)
+        if thinking_content:
+            visible_prefix.append(f'<think>\n{thinking_content}\n</think>')
+        if intermediate and intermediate.strip():
+            visible_prefix.append(intermediate.strip())
+
+        # Build args summaries and show placeholder accordions with "..."
+        # before execution starts (tool calls may be slow, e.g. web search).
+        tc_headers = []
         for tc in parsed_calls:
+            fn_name = tc['function']['name']
+            fn_args = tc['function'].get('arguments', {})
+            if isinstance(fn_args, dict) and fn_args:
+                args_summary = ', '.join(f'{k}={json.dumps(v, ensure_ascii=False)}' for k, v in fn_args.items())
+            elif isinstance(fn_args, dict):
+                args_summary = ''
+            else:
+                args_summary = str(fn_args)
+
+            tc_headers.append(f'{fn_name}({args_summary})')
+
+        pending_placeholders = [f'{h}\n...\n' for h in tc_headers]
+        history['visible'][-1][1] = '\n\n'.join(visible_prefix + pending_placeholders)
+        yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu']), history
+
+        # Execute tools, store results, and replace placeholders with real results
+        for i, tc in enumerate(parsed_calls):
             fn_name = tc['function']['name']
             fn_args = tc['function'].get('arguments', {})
             result = execute_tool(fn_name, fn_args, tool_executors)
@@ -1248,11 +1291,14 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
             except (json.JSONDecodeError, TypeError):
                 pretty_result = result
 
-            visible_prefix.append(f'**{fn_name}**\n```json\n{pretty_result}\n```')
+            # Replace the placeholder with the real result
+            pending_placeholders[i] = f'{tc_headers[i]}\n{pretty_result}\n'
+            history['visible'][-1][1] = '\n\n'.join(visible_prefix + pending_placeholders)
+            yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu']), history
 
-        # Show tool results
+        # Move completed tool calls into visible_prefix for next turns
+        visible_prefix.extend(pending_placeholders)
         history['visible'][-1][1] = '\n\n'.join(visible_prefix)
-        yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu']), history
         save_history(history, state['unique_id'], state['character_menu'], state['mode'])
 
         state['history'] = history
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 4d9904fb2a..f381160224 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -114,7 +114,36 @@ def extract_thinking_block(string):
     return extract_reasoning(string, html_escaped=True)
 
 
-def build_thinking_block(thinking_content, message_id, has_remaining_content):
+
+def build_tool_call_block(header, body, message_id, index):
+    """Build HTML for a tool call accordion block."""
+    block_id = f"tool-call-{message_id}-{index}"
+
+    if body == '...':
+        # Pending placeholder — no expandable body, just title with ellipsis
+        return f'''
+        
            
+            
+                {tool_svg_small}
+                {html.escape(header)} ...
+            
+        
            
+        '''
+
+    # Build a plain 
             directly to avoid highlight.js auto-detection
+    escaped_body = html.escape(body)
+    return f'''
+    
            
+        
+            {tool_svg_small}
+            {html.escape(header)}
+        
+        
            {escaped_body}
            
+    
            
+    '''
+
+
+def build_thinking_block(thinking_content, message_id, has_remaining_content, thinking_index=0):
     """Build HTML for a thinking block."""
     if thinking_content is None:
         return None
@@ -123,7 +152,7 @@ def build_thinking_block(thinking_content, message_id, has_remaining_content):
     thinking_html = process_markdown_content(thinking_content)
 
     # Generate unique ID for the thinking block
-    block_id = f"thinking-{message_id}-0"
+    block_id = f"thinking-{message_id}-{thinking_index}"
 
     # Check if thinking is complete or still in progress
     is_streaming = not has_remaining_content
@@ -304,24 +333,64 @@ def convert_to_markdown(string, message_id=None):
     if message_id is None:
         message_id = "unknown"
 
-    # Extract different components from the string
-    thinking_content, remaining_content = extract_thinking_block(string)
-
-    # Build individual HTML blocks
-    blocks = []
-
-    # Add thinking block if present
-    thinking_html = build_thinking_block(thinking_content, message_id, bool(remaining_content))
-    if thinking_html:
-        blocks.append(thinking_html)
-
-    # Add main content block
-    main_html = build_main_content_block(remaining_content)
-    if main_html:
-        blocks.append(main_html)
-
-    # Assemble all blocks into final HTML
-    return ''.join(blocks)
+    # Find tool call blocks by position, then process the text segments
+    # between them using extract_thinking_block (which supports all
+    # THINKING_FORMATS, including end-only variants like Qwen's).
+    tool_call_pattern = re.compile(r'(.*?)\n(.*?)\n', re.DOTALL)
+    tool_calls = list(tool_call_pattern.finditer(string))
+
+    if not tool_calls:
+        # No tool calls — use original single-pass extraction
+        thinking_content, remaining_content = extract_thinking_block(string)
+        blocks = []
+        thinking_html = build_thinking_block(thinking_content, message_id, bool(remaining_content))
+        if thinking_html:
+            blocks.append(thinking_html)
+
+        main_html = build_main_content_block(remaining_content)
+        if main_html:
+            blocks.append(main_html)
+
+        return ''.join(blocks)
+
+    # Split string into text segments around tool_call blocks and
+    # run extract_thinking_block on each segment for full format support.
+    html_parts = []
+    last_end = 0
+    tool_idx = 0
+    think_idx = 0
+
+    def process_text_segment(text, is_last_segment):
+        """Process a text segment between tool_call blocks for thinking content."""
+        nonlocal think_idx
+        if not text.strip():
+            return
+
+        thinking_content, remaining = extract_thinking_block(text)
+        if thinking_content is not None:
+            has_remaining = bool(remaining.strip()) or not is_last_segment
+            html_parts.append(build_thinking_block(thinking_content, message_id, has_remaining, think_idx))
+            think_idx += 1
+            text = remaining
+
+        if text.strip():
+            html_parts.append(process_markdown_content(text))
+
+    for tc in tool_calls:
+        # Process text before this tool_call
+        process_text_segment(string[last_end:tc.start()], is_last_segment=False)
+
+        # Add tool call accordion
+        header = tc.group(1).strip()
+        body = tc.group(2).strip()
+        html_parts.append(build_tool_call_block(header, body, message_id, tool_idx))
+        tool_idx += 1
+        last_end = tc.end()
+
+    # Process text after the last tool_call
+    process_text_segment(string[last_end:], is_last_segment=True)
+
+    return ''.join(html_parts)
 
 
 def convert_to_markdown_wrapped(string, message_id=None, use_cache=True):
@@ -379,6 +448,7 @@ def get_image_cache(path):
 edit_svg = ''''''
 info_svg = ''''''
 info_svg_small = ''''''
+tool_svg_small = ''''''
 attachment_svg = ''''''
 
 copy_button = f''

From 3e6bd1a310217c89fa5dff23c34e92ae43f1acb9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 14:30:51 -0700
Subject: [PATCH 1327/1701] UI: Prepend thinking tag when template appends it
 to prompt

Makes Qwen models have a thinking block straight away during streaming.
---
 modules/chat.py | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index d3cd0caecc..e4d5dd30f6 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -28,6 +28,7 @@
 )
 from modules.image_utils import open_image_safely
 from modules.logging_colors import logger
+from modules.reasoning import THINKING_FORMATS
 from modules.text_generation import (
     generate_reply,
     get_encoded_length,
@@ -986,10 +987,23 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
     # Add timestamp for assistant's response at the start of generation
     update_message_metadata(output['metadata'], "assistant", row_idx, timestamp=get_current_timestamp(), model_name=shared.model_name)
 
+    # Detect if the template appended a thinking start tag to the prompt
+    thinking_prefix = None
+    if not _continue:
+        stripped_prompt = prompt.rstrip('\n')
+        for start_tag, end_tag, content_tag in THINKING_FORMATS:
+            if start_tag is not None and stripped_prompt.endswith(start_tag):
+                thinking_prefix = start_tag
+                break
+
     # Generate
     reply = None
     for j, reply in enumerate(generate_reply(prompt, state, stopping_strings=stopping_strings, is_chat=True, for_ui=for_ui)):
 
+        # Prepend thinking tag if the template appended it to the prompt
+        if thinking_prefix:
+            reply = thinking_prefix + reply
+
         # Extract the reply
         if state['mode'] in ['chat', 'chat-instruct']:
             if not _continue:

From bbd43d9463cded0aae3ce4e1ced1519693bebf9a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 14:54:05 -0700
Subject: [PATCH 1328/1701] UI: Correctly propagate truncation_length when
 ctx_size is auto

---
 modules/llama_cpp_server.py | 8 +++++++-
 modules/models.py           | 2 ++
 modules/ui_model_menu.py    | 6 +++++-
 3 files changed, 14 insertions(+), 2 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index a3e431aca4..192aa9e408 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -36,6 +36,7 @@ def __init__(self, model_path, server_path=None):
         self.process = None
         self.session = requests.Session()
         self.vocabulary_size = None
+        self.n_ctx = None
         self.bos_token = ""
         self.last_prompt_token_count = 0
 
@@ -320,12 +321,17 @@ def _get_vocabulary_size(self):
                 self.vocabulary_size = model_info["meta"]["n_vocab"]
 
     def _get_bos_token(self):
-        """Get and store the model's BOS token."""
+        """Get and store the model's BOS token and context size."""
         url = f"http://127.0.0.1:{self.port}/props"
         response = self.session.get(url).json()
         if "bos_token" in response:
             self.bos_token = response["bos_token"]
 
+        # Get actual n_ctx from the server (important when --fit auto-selects it)
+        n_ctx = response.get("default_generation_settings", {}).get("n_ctx")
+        if n_ctx:
+            self.n_ctx = n_ctx
+
     def _is_port_available(self, port):
         """Check if a port is available for use."""
         with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
diff --git a/modules/models.py b/modules/models.py
index 48d68b0bd2..d83b98d79e 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -54,6 +54,8 @@ def load_model(model_name, loader=None):
     if loader.lower().startswith('exllama') or loader.lower().startswith('tensorrt') or loader == 'llama.cpp':
         if shared.args.ctx_size > 0:
             shared.settings['truncation_length'] = shared.args.ctx_size
+        elif loader == 'llama.cpp' and hasattr(model, 'n_ctx') and model.n_ctx:
+            shared.settings['truncation_length'] = model.n_ctx
 
     shared.is_multimodal = False
     if loader.lower() in ('exllamav3', 'llama.cpp') and hasattr(model, 'is_multimodal'):
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 7e91f1cecc..5c83096f45 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -388,7 +388,11 @@ def downloader_thread_target():
 def update_truncation_length(current_length, state):
     if 'loader' in state:
         if state['loader'].lower().startswith('exllama') or state['loader'] == 'llama.cpp':
-            return state['ctx_size']
+            if state['ctx_size'] > 0:
+                return state['ctx_size']
+
+            # ctx_size == 0 means auto: use the actual value from the server
+            return shared.settings['truncation_length']
 
     return current_length
 

From 4f82b71ef3cfd104cf88c50a7b427063fc2c6094 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 14:56:35 -0700
Subject: [PATCH 1329/1701] UI: Bump the ctx-size max from 131072 to 262144
 (256K)

---
 modules/ui_model_menu.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 5c83096f45..d17f586b17 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -42,7 +42,7 @@ def create_ui():
                     with gr.Row():
                         with gr.Column():
                             shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=-1, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Number of layers to offload to the GPU. -1 = auto.')
-                            shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=0, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. llama.cpp: 0 = auto if gpu-layers is also -1. Common values: 4096, 8192, 16384, 32768, 65536, 131072.')
+                            shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=0, maximum=262144, step=256, value=shared.args.ctx_size, info='Context length. llama.cpp: 0 = auto if gpu-layers is also -1. Common values: 4096, 8192, 16384, 32768, 65536, 131072.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')

From fdd8e5b1fd92de8d0ee20e8166b88c8356d13c52 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 15:48:50 -0700
Subject: [PATCH 1330/1701] Make repeated Ctrl+C force a shutdown

---
 server.py | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/server.py b/server.py
index 73f190b602..340f71265e 100644
--- a/server.py
+++ b/server.py
@@ -33,6 +33,10 @@
 
 
 def signal_handler(sig, frame):
+    # On second Ctrl+C, force an immediate exit
+    signal.signal(signal.SIGINT, signal.SIG_DFL)
+    signal.signal(signal.SIGTERM, signal.SIG_DFL)
+
     logger.info("Received Ctrl+C. Shutting down Text Generation Web UI gracefully.")
 
     # Explicitly stop LlamaServer to avoid __del__ cleanup issues during shutdown

From 09d5e049d672e6bddbf6aec402a8b38cd97e074c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 16:53:49 -0700
Subject: [PATCH 1331/1701] UI: Improve the Tools checkbox list style

---
 css/main.css       | 83 ++++++++++++++++++++++++++++++++++++++++++++++
 js/main.js         | 17 ++++++++++
 modules/ui_chat.py |  5 +--
 3 files changed, 103 insertions(+), 2 deletions(-)

diff --git a/css/main.css b/css/main.css
index 2d19b5c8d0..a7069f33c4 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1814,3 +1814,86 @@ tr + tr th { border-top: 1px solid; }
 
 thead + tbody tr:first-child td,
 thead + tbody tr:first-child th { border-top: 1px solid; }
+
+/* ------------------------------------------------
+   Tools CheckboxGroup - vertical DragDrop-like style
+   ------------------------------------------------ */
+
+/* "Refresh list" link in the Tools label */
+.tools-refresh-link {
+    cursor: pointer;
+}
+
+/* Checkbox list container */
+#tools-group {
+    padding: 0 !important;
+    border-width: 0 !important;
+    background: transparent !important;
+    min-height: 0 !important;
+}
+
+#tools-group .wrap {
+    display: flex;
+    flex-direction: column;
+    flex-wrap: nowrap;
+    gap: 4px;
+    padding: 0;
+    margin-top: var(--spacing-lg);
+    max-height: 350px;
+    overflow-y: auto;
+}
+
+/* Pretty scrollbar for the tools list */
+#tools-group .wrap::-webkit-scrollbar {
+    width: 8px;
+    height: 8px;
+}
+
+#tools-group .wrap::-webkit-scrollbar-track {
+    background: transparent;
+}
+
+#tools-group .wrap::-webkit-scrollbar-thumb,
+#tools-group .wrap::-webkit-scrollbar-thumb:hover {
+    background: var(--neutral-300);
+    border-radius: 30px;
+}
+
+.dark #tools-group .wrap::-webkit-scrollbar-thumb,
+.dark #tools-group .wrap::-webkit-scrollbar-thumb:hover {
+    background: rgb(255 255 255 / 6.25%);
+    border-radius: 10px;
+}
+
+#tools-group .wrap::-webkit-scrollbar-corner {
+    background: transparent;
+}
+
+/* Each checkbox item */
+#tools-group label {
+    display: flex;
+    align-items: center;
+    gap: 8px;
+    padding: 5px 8px;
+    border-radius: var(--radius-sm, 4px);
+    background: var(--block-background-fill);
+    border: 1px solid var(--border-color-primary);
+    color: var(--body-text-color);
+    font-size: var(--input-text-size);
+    font-weight: var(--input-text-weight);
+    cursor: pointer;
+    user-select: none;
+    transition: border-color 0.15s ease, background 0.15s ease;
+    box-shadow: none;
+}
+
+#tools-group label:hover {
+    border-color: var(--input-border-color-focus);
+}
+
+#tools-group label span {
+    flex: 1;
+    overflow: hidden;
+    text-overflow: ellipsis;
+    white-space: nowrap;
+}
diff --git a/js/main.js b/js/main.js
index a7dc77bbfc..0bc76b2aa4 100644
--- a/js/main.js
+++ b/js/main.js
@@ -303,6 +303,23 @@ for(i = 0; i < scrollbarElements.length; i++) {
   scrollbarElements[i].style.resize = "none";
 }
 
+
+//------------------------------------------------
+// Tools: inject "Refresh list" link into the label
+//------------------------------------------------
+const toolsTitle = document.querySelector("#tools-group > [data-testid='block-info']");
+const toolsInfo = toolsTitle ? toolsTitle.nextElementSibling : null;
+if (toolsInfo) {
+  const refreshLink = document.createElement("span");
+  refreshLink.textContent = " [Refresh list]";
+  refreshLink.className = "tools-refresh-link";
+  refreshLink.addEventListener("click", function(e) {
+    e.preventDefault();
+    document.querySelector("#tools-refresh-btn").click();
+  });
+  toolsInfo.appendChild(refreshLink);
+}
+
 //------------------------------------------------
 // Remove some backgrounds
 //------------------------------------------------
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index d5b1309432..8112d9565c 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -92,8 +92,9 @@ def create_ui():
                 gr.HTML("
            ")
 
                 from modules.tool_use import get_available_tools
-                shared.gradio['selected_tools'] = gr.CheckboxGroup(choices=get_available_tools(), value=[], label='Tools', info='Functions the model can call during generation.')
-                ui.create_refresh_button(shared.gradio['selected_tools'], lambda: None, lambda: {'choices': get_available_tools()}, 'refresh-button')
+                shared.gradio['selected_tools'] = gr.CheckboxGroup(choices=get_available_tools(), value=[], label='Tools', info='Functions the model can call during generation.', elem_id='tools-group')
+                shared.gradio['tools_refresh'] = gr.Button('Refresh list', elem_id='tools-refresh-btn', visible=False)
+                shared.gradio['tools_refresh'].click(fn=lambda: gr.update(choices=get_available_tools()), inputs=[], outputs=[shared.gradio['selected_tools']])
 
                 gr.HTML("
            ")
 

From 5c02b7f60369be70b52ccf0432a69380b7f5ab4a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 17:08:30 -0700
Subject: [PATCH 1332/1701] Allow the fetch_webpage tool to return links

---
 modules/tool_use.py   | 3 ++-
 modules/ui_chat.py    | 2 +-
 modules/web_search.py | 4 ++--
 3 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/modules/tool_use.py b/modules/tool_use.py
index cb1e140dcb..e464d5d48f 100644
--- a/modules/tool_use.py
+++ b/modules/tool_use.py
@@ -4,13 +4,14 @@
 
 from modules import shared
 from modules.logging_colors import logger
+from modules.utils import natural_keys
 
 
 def get_available_tools():
     """Return sorted list of tool script names from user_data/tools/*.py."""
     tools_dir = shared.user_data_dir / 'tools'
     tools_dir.mkdir(parents=True, exist_ok=True)
-    return sorted(p.stem for p in tools_dir.glob('*.py'))
+    return sorted((p.stem for p in tools_dir.glob('*.py')), key=natural_keys)
 
 
 def load_tools(selected_names):
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 8112d9565c..039b9af640 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -92,7 +92,7 @@ def create_ui():
                 gr.HTML("
            ")
 
                 from modules.tool_use import get_available_tools
-                shared.gradio['selected_tools'] = gr.CheckboxGroup(choices=get_available_tools(), value=[], label='Tools', info='Functions the model can call during generation.', elem_id='tools-group')
+                shared.gradio['selected_tools'] = gr.CheckboxGroup(choices=get_available_tools(), value=shared.settings.get('selected_tools', []), label='Tools', info='Functions the model can call during generation.', elem_id='tools-group')
                 shared.gradio['tools_refresh'] = gr.Button('Refresh list', elem_id='tools-refresh-btn', visible=False)
                 shared.gradio['tools_refresh'].click(fn=lambda: gr.update(choices=get_available_tools()), inputs=[], outputs=[shared.gradio['selected_tools']])
 
diff --git a/modules/web_search.py b/modules/web_search.py
index 597af4b289..b14cd04240 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -18,7 +18,7 @@ def get_current_timestamp():
     return datetime.now().strftime('%b %d, %Y %H:%M')
 
 
-def download_web_page(url, timeout=10):
+def download_web_page(url, timeout=10, include_links=False):
     """
     Download a web page and convert its HTML content to structured Markdown text.
     """
@@ -35,7 +35,7 @@ def download_web_page(url, timeout=10):
         h = html2text.HTML2Text()
         h.body_width = 0
         h.ignore_images = True
-        h.ignore_links = True
+        h.ignore_links = not include_links
 
         # Convert the HTML to Markdown
         markdown_text = h.handle(response.text)

From f8936ec47c95e4147116a2d497e5b94a2d7c99ba Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 17:10:41 -0700
Subject: [PATCH 1333/1701] Truncate web_search and fetch_webpage tools to 8192
 tokens

---
 user_data/tools/web_search.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/user_data/tools/web_search.py b/user_data/tools/web_search.py
index 8923eab0a9..46a45a8cf3 100644
--- a/user_data/tools/web_search.py
+++ b/user_data/tools/web_search.py
@@ -1,4 +1,4 @@
-from modules.web_search import perform_web_search
+from modules.web_search import perform_web_search, truncate_content_by_tokens
 
 tool = {
     "type": "function",
@@ -22,6 +22,6 @@ def execute(arguments):
     output = []
     for r in results:
         if r and r["content"].strip():
-            output.append({"title": r["title"], "url": r["url"], "content": r["content"][:4000]})
+            output.append({"title": r["title"], "url": r["url"], "content": truncate_content_by_tokens(r["content"])})
 
     return output if output else [{"error": "No results found."}]

From 1b7e6c57057fdddb57c1585e10c9ad6ccf59e856 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 17:11:05 -0700
Subject: [PATCH 1334/1701] Add the fetch_webpage tool source

---
 user_data/tools/fetch_webpage.py | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)
 create mode 100644 user_data/tools/fetch_webpage.py

diff --git a/user_data/tools/fetch_webpage.py b/user_data/tools/fetch_webpage.py
new file mode 100644
index 0000000000..e514491c17
--- /dev/null
+++ b/user_data/tools/fetch_webpage.py
@@ -0,0 +1,28 @@
+from modules.web_search import download_web_page, truncate_content_by_tokens
+
+tool = {
+    "type": "function",
+    "function": {
+        "name": "fetch_webpage",
+        "description": "Fetch and read the contents of a web page given its URL. Returns the page content as plain text.",
+        "parameters": {
+            "type": "object",
+            "properties": {
+                "url": {"type": "string", "description": "The URL of the web page to fetch."},
+            },
+            "required": ["url"]
+        }
+    }
+}
+
+
+def execute(arguments):
+    url = arguments.get("url", "")
+    if not url:
+        return {"error": "No URL provided."}
+
+    content = download_web_page(url, include_links=True)
+    if not content or not content.strip():
+        return {"error": f"Failed to fetch content from {url}"}
+
+    return {"url": url, "content": truncate_content_by_tokens(content)}

From a09f21b9de44480d92cb788053852721806c376a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 22:17:20 -0300
Subject: [PATCH 1335/1701] UI: Fix tool calling for GPT-OSS and Continue

---
 extensions/openai/utils.py |  7 ++-
 modules/chat.py            | 88 +++++++++++++++++++++++++-------------
 modules/reasoning.py       |  1 +
 3 files changed, 64 insertions(+), 32 deletions(-)

diff --git a/extensions/openai/utils.py b/extensions/openai/utils.py
index 6fb05f0855..eb34ce8805 100644
--- a/extensions/openai/utils.py
+++ b/extensions/openai/utils.py
@@ -120,12 +120,14 @@ def _parseChannelToolCalls(answer: str, tool_names: list[str]):
     """Parse channel-based tool calls used by GPT-OSS and similar models.
 
     Format:
+        <|start|>assistant to=functions.func_name<|channel|>commentary json<|message|>{"arg": "value"}
+    or:
         <|channel|>commentary to=functions.func_name <|constrain|>json<|message|>{"arg": "value"}
     """
     matches = []
     start_pos = None
     for m in re.finditer(
-        r'<\|channel\|>commentary to=functions\.([^<\s]+)\s*(?:<\|constrain\|>json)?<\|message\|>',
+        r'<\|channel\|>\w+ to=functions\.([^<\s]+).*?<\|message\|>',
         answer
     ):
         func_name = m.group(1).strip()
@@ -137,7 +139,8 @@ def _parseChannelToolCalls(answer: str, tool_names: list[str]):
         try:
             arguments = json.loads(json_str)
             if start_pos is None:
-                start_pos = m.start()
+                prefix = answer.rfind('<|start|>assistant', 0, m.start())
+                start_pos = prefix if prefix != -1 else m.start()
             matches.append({
                 "type": "function",
                 "function": {
diff --git a/modules/chat.py b/modules/chat.py
index e4d5dd30f6..19f7d07fab 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -81,6 +81,13 @@ def update_message_metadata(metadata_dict, role, index, **fields):
 )
 jinja_env.globals["strftime_now"] = strftime_now
 
+
+def _raise_exception(message):
+    raise ValueError(message)
+
+
+jinja_env.globals["raise_exception"] = _raise_exception
+
 _template_cache = {}
 
 
@@ -1048,16 +1055,23 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             yield output
 
     if _continue:
-        # Reprocess the entire internal text for extensions (like translation)
-        full_internal = output['internal'][-1][1]
-        if state['mode'] in ['chat', 'chat-instruct']:
-            full_visible = re.sub("(||{{user}})", state['name1'], full_internal)
-        else:
-            full_visible = full_internal
+        # Reprocess the entire internal text for extensions (like translation).
+        # Skip the rebuild when the visible text contains  markers,
+        # since those only exist in visible (internal is cleared after each tool
+        # execution) and rebuilding from internal would destroy them.
+        if '' not in output['visible'][-1][1]:
+            full_internal = output['internal'][-1][1]
+            if state['mode'] in ['chat', 'chat-instruct']:
+                full_visible = re.sub("(||{{user}})", state['name1'], full_internal)
+            else:
+                full_visible = full_internal
 
-        full_visible = html.escape(full_visible)
-        if not state.get('_skip_output_extensions'):
-            output['visible'][-1][1] = apply_extensions('output', full_visible, state, is_chat=True)
+            full_visible = html.escape(full_visible)
+            if not state.get('_skip_output_extensions'):
+                output['visible'][-1][1] = apply_extensions('output', full_visible, state, is_chat=True)
+        else:
+            if not state.get('_skip_output_extensions'):
+                output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
     else:
         if not state.get('_skip_output_extensions'):
             output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
@@ -1222,6 +1236,18 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
         # so we can extract thinking content from just this turn's output.
         _model_visible = history['visible'][-1][1]
 
+        # Recover visible_prefix from existing visible text (e.g. on Continue
+        # after a previous session had tool calls). Extract all 
+        # blocks and any text between them (thinking blocks, intermediate text).
+        if not visible_prefix and _model_visible:
+            tc_matches = list(re.finditer(r'.*?', _model_visible, re.DOTALL))
+            if tc_matches:
+                prefix_end = tc_matches[-1].end()
+                prefix = _model_visible[:prefix_end].strip()
+                if prefix:
+                    visible_prefix = [prefix]
+                _model_visible = _model_visible[prefix_end:].strip()
+
         # Re-apply visible prefix to the final state after streaming completes.
         # This is safe because we're no longer sharing the object with chatbot_wrapper.
         if visible_prefix:
@@ -1244,20 +1270,35 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
         meta = history.get('metadata', {})
         seq = meta.setdefault(f'assistant_{row_idx}', {}).setdefault('tool_sequence', [])
 
-        # Serialize tool calls
+        def _render():
+            return chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
+
+        # Serialize tool calls and build display headers in one pass
         serialized = []
+        tc_headers = []
         for tc in parsed_calls:
             tc['id'] = generate_tool_call_id()
-            args = tc['function'].get('arguments', {})
+            fn_name = tc['function']['name']
+            fn_args = tc['function'].get('arguments', {})
+
             serialized.append({
                 'id': tc['id'],
                 'type': 'function',
                 'function': {
-                    'name': tc['function']['name'],
-                    'arguments': json.dumps(args) if isinstance(args, dict) else args
+                    'name': fn_name,
+                    'arguments': json.dumps(fn_args) if isinstance(fn_args, dict) else fn_args
                 }
             })
 
+            if isinstance(fn_args, dict) and fn_args:
+                args_summary = ', '.join(f'{k}={json.dumps(v, ensure_ascii=False)}' for k, v in fn_args.items())
+            elif isinstance(fn_args, dict):
+                args_summary = ''
+            else:
+                args_summary = str(fn_args)
+
+            tc_headers.append(f'{fn_name}({args_summary})')
+
         seq.append({'tool_calls': serialized})
 
         # Clear internal (raw tool markup)
@@ -1274,24 +1315,11 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
         if intermediate and intermediate.strip():
             visible_prefix.append(intermediate.strip())
 
-        # Build args summaries and show placeholder accordions with "..."
-        # before execution starts (tool calls may be slow, e.g. web search).
-        tc_headers = []
-        for tc in parsed_calls:
-            fn_name = tc['function']['name']
-            fn_args = tc['function'].get('arguments', {})
-            if isinstance(fn_args, dict) and fn_args:
-                args_summary = ', '.join(f'{k}={json.dumps(v, ensure_ascii=False)}' for k, v in fn_args.items())
-            elif isinstance(fn_args, dict):
-                args_summary = ''
-            else:
-                args_summary = str(fn_args)
-
-            tc_headers.append(f'{fn_name}({args_summary})')
-
+        # Show placeholder accordions with "..." before execution starts
+        # (tool calls may be slow, e.g. web search).
         pending_placeholders = [f'{h}\n...\n' for h in tc_headers]
         history['visible'][-1][1] = '\n\n'.join(visible_prefix + pending_placeholders)
-        yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu']), history
+        yield _render(), history
 
         # Execute tools, store results, and replace placeholders with real results
         for i, tc in enumerate(parsed_calls):
@@ -1308,7 +1336,7 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
             # Replace the placeholder with the real result
             pending_placeholders[i] = f'{tc_headers[i]}\n{pretty_result}\n'
             history['visible'][-1][1] = '\n\n'.join(visible_prefix + pending_placeholders)
-            yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu']), history
+            yield _render(), history
 
         # Move completed tool calls into visible_prefix for next turns
         visible_prefix.extend(pending_placeholders)
diff --git a/modules/reasoning.py b/modules/reasoning.py
index 12f8553d98..708ee55aed 100644
--- a/modules/reasoning.py
+++ b/modules/reasoning.py
@@ -5,6 +5,7 @@
 THINKING_FORMATS = [
     ('', '', None),
     ('<|channel|>analysis<|message|>', '<|end|>', '<|start|>assistant<|channel|>final<|message|>'),
+    ('<|channel|>commentary<|message|>', '<|end|>', '<|start|>assistant<|channel|>final<|message|>'),
     ('', '', None),
     ('<|think|>', '<|end|>', '<|content|>'),  # Solar Open
     ('Thinking Process:', '', None),  # Qwen3.5 verbose thinking outside tags

From 4c7a56c18df369d5d04870d6280468900662d3cd Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 22:17:23 -0300
Subject: [PATCH 1336/1701] Add num_pages and max_tokens kwargs to web search
 tools

---
 user_data/tools/fetch_webpage.py | 4 +++-
 user_data/tools/web_search.py    | 8 ++++++--
 2 files changed, 9 insertions(+), 3 deletions(-)

diff --git a/user_data/tools/fetch_webpage.py b/user_data/tools/fetch_webpage.py
index e514491c17..ca3e73314c 100644
--- a/user_data/tools/fetch_webpage.py
+++ b/user_data/tools/fetch_webpage.py
@@ -9,6 +9,7 @@
             "type": "object",
             "properties": {
                 "url": {"type": "string", "description": "The URL of the web page to fetch."},
+                "max_tokens": {"type": "integer", "description": "Maximum number of tokens in the returned content (default: 2048)."},
             },
             "required": ["url"]
         }
@@ -18,6 +19,7 @@
 
 def execute(arguments):
     url = arguments.get("url", "")
+    max_tokens = arguments.get("max_tokens", 2048)
     if not url:
         return {"error": "No URL provided."}
 
@@ -25,4 +27,4 @@ def execute(arguments):
     if not content or not content.strip():
         return {"error": f"Failed to fetch content from {url}"}
 
-    return {"url": url, "content": truncate_content_by_tokens(content)}
+    return {"url": url, "content": truncate_content_by_tokens(content, max_tokens=max_tokens)}
diff --git a/user_data/tools/web_search.py b/user_data/tools/web_search.py
index 46a45a8cf3..8084596368 100644
--- a/user_data/tools/web_search.py
+++ b/user_data/tools/web_search.py
@@ -9,6 +9,8 @@
             "type": "object",
             "properties": {
                 "query": {"type": "string", "description": "The search query."},
+                "num_pages": {"type": "integer", "description": "Number of search result pages to fetch (default: 3)."},
+                "max_tokens": {"type": "integer", "description": "Maximum number of tokens per page result (default: 2048)."},
             },
             "required": ["query"]
         }
@@ -18,10 +20,12 @@
 
 def execute(arguments):
     query = arguments.get("query", "")
-    results = perform_web_search(query, num_pages=3)
+    num_pages = arguments.get("num_pages", 3)
+    max_tokens = arguments.get("max_tokens", 2048)
+    results = perform_web_search(query, num_pages=num_pages)
     output = []
     for r in results:
         if r and r["content"].strip():
-            output.append({"title": r["title"], "url": r["url"], "content": truncate_content_by_tokens(r["content"])})
+            output.append({"title": r["title"], "url": r["url"], "content": truncate_content_by_tokens(r["content"], max_tokens=max_tokens)})
 
     return output if output else [{"error": "No results found."}]

From 286ae475f6794cfb39bf9dabdd289b52eafd5c5b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 22:39:38 -0300
Subject: [PATCH 1337/1701] UI: Clean up tool calling code

---
 modules/chat.py     | 34 +++++++++++++++++-----------------
 modules/tool_use.py |  7 -------
 2 files changed, 17 insertions(+), 24 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 19f7d07fab..1649a6e822 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1056,9 +1056,10 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
 
     if _continue:
         # Reprocess the entire internal text for extensions (like translation).
-        # Skip the rebuild when the visible text contains  markers,
+        # Skip entirely when the visible text contains  markers,
         # since those only exist in visible (internal is cleared after each tool
-        # execution) and rebuilding from internal would destroy them.
+        # execution) and rebuilding from internal would destroy them. Output
+        # extensions also can't handle the raw  markup safely.
         if '' not in output['visible'][-1][1]:
             full_internal = output['internal'][-1][1]
             if state['mode'] in ['chat', 'chat-instruct']:
@@ -1069,9 +1070,6 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             full_visible = html.escape(full_visible)
             if not state.get('_skip_output_extensions'):
                 output['visible'][-1][1] = apply_extensions('output', full_visible, state, is_chat=True)
-        else:
-            if not state.get('_skip_output_extensions'):
-                output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
     else:
         if not state.get('_skip_output_extensions'):
             output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
@@ -1141,8 +1139,19 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
     if not character_is_loaded(state):
         return
 
+    if state['start_with'] != '' and not _continue:
+        if regenerate:
+            text, state['history'] = remove_last_message(state['history'])
+            regenerate = False
+
+        _continue = True
+        send_dummy_message(text, state)
+        send_dummy_reply(state['start_with'], state)
+
     # On regenerate, clear old tool_sequence metadata so it gets rebuilt.
     # Save it first so it can be stored per-version below.
+    # This must happen after the start_with logic above, which may remove
+    # and re-add messages, changing which row we operate on.
     _old_tool_sequence = None
     if regenerate:
         history = state['history']
@@ -1151,22 +1160,13 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
         if row_idx >= 0:
             _old_tool_sequence = meta.get(f'assistant_{row_idx}', {}).pop('tool_sequence', None)
 
-    if state['start_with'] != '' and not _continue:
-        if regenerate:
-            text, state['history'] = remove_last_message(state['history'])
-            regenerate = False
-
-        _continue = True
-        send_dummy_message(text, state)
-        send_dummy_reply(state['start_with'], state)
-
     # Load tools if any are selected
     selected = state.get('selected_tools', [])
     parseToolCall = None
     if selected:
-        from modules.tool_use import load_tools, execute_tool, generate_tool_call_id
+        from modules.tool_use import load_tools, execute_tool
         try:
-            from extensions.openai.utils import parseToolCall
+            from extensions.openai.utils import parseToolCall, getToolCallId
         except ImportError:
             logger.warning('Tool calling requires the openai extension for parseToolCall. Disabling tools.')
             selected = []
@@ -1277,7 +1277,7 @@ def _render():
         serialized = []
         tc_headers = []
         for tc in parsed_calls:
-            tc['id'] = generate_tool_call_id()
+            tc['id'] = getToolCallId()
             fn_name = tc['function']['name']
             fn_args = tc['function'].get('arguments', {})
 
diff --git a/modules/tool_use.py b/modules/tool_use.py
index e464d5d48f..d65c14eda4 100644
--- a/modules/tool_use.py
+++ b/modules/tool_use.py
@@ -1,6 +1,5 @@
 import importlib.util
 import json
-import random
 
 from modules import shared
 from modules.logging_colors import logger
@@ -49,12 +48,6 @@ def load_tools(selected_names):
     return tool_defs, executors
 
 
-def generate_tool_call_id():
-    """Generate a unique tool call ID (e.g. 'call_a1b2c3d4')."""
-    chars = "abcdefghijklmnopqrstuvwxyz0123456789"
-    return "call_" + "".join(random.choice(chars) for _ in range(8))
-
-
 def execute_tool(func_name, arguments, executors):
     """Execute a tool by function name. Returns result as a JSON string."""
     fn = executors.get(func_name)

From 1ed56aee85b4c64232f3f6517a8b555140f1c8a8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 18:45:19 -0700
Subject: [PATCH 1338/1701] Add a calculate tool

---
 user_data/tools/calculate.py | 48 ++++++++++++++++++++++++++++++++++++
 1 file changed, 48 insertions(+)
 create mode 100644 user_data/tools/calculate.py

diff --git a/user_data/tools/calculate.py b/user_data/tools/calculate.py
new file mode 100644
index 0000000000..e88b71a3ed
--- /dev/null
+++ b/user_data/tools/calculate.py
@@ -0,0 +1,48 @@
+import ast
+import operator
+
+OPERATORS = {
+    ast.Add: operator.add,
+    ast.Sub: operator.sub,
+    ast.Mult: operator.mul,
+    ast.Div: operator.truediv,
+    ast.Pow: operator.pow,
+    ast.Mod: operator.mod,
+    ast.USub: operator.neg,
+}
+
+
+def _eval(node):
+    if isinstance(node, ast.Constant) and isinstance(node.value, (int, float)):
+        return node.value
+    elif isinstance(node, ast.BinOp) and type(node.op) in OPERATORS:
+        return OPERATORS[type(node.op)](_eval(node.left), _eval(node.right))
+    elif isinstance(node, ast.UnaryOp) and type(node.op) in OPERATORS:
+        return OPERATORS[type(node.op)](_eval(node.operand))
+    raise ValueError(f"Unsupported expression")
+
+
+tool = {
+    "type": "function",
+    "function": {
+        "name": "calculate",
+        "description": "Evaluate a math expression. Supports +, -, *, /, **, %.",
+        "parameters": {
+            "type": "object",
+            "properties": {
+                "expression": {"type": "string", "description": "The math expression to evaluate (e.g. '2 * (3 + 4)')."},
+            },
+            "required": ["expression"]
+        }
+    }
+}
+
+
+def execute(arguments):
+    expr = arguments.get("expression", "")
+    try:
+        tree = ast.parse(expr, mode='eval')
+        result = _eval(tree.body)
+        return {"expression": expr, "result": result}
+    except Exception as e:
+        return {"error": str(e)}

From 0e35421593ebed4295116db35b323ac2e32c91a8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 18:52:41 -0700
Subject: [PATCH 1339/1701] API: Always extract reasoning_content, even with
 tool calls

---
 extensions/openai/completions.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 0eb0cd2748..290a5bc017 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -554,7 +554,7 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
         else:
             yield chunk
     else:
-        reasoning, content = extract_reasoning(answer) if not tool_calls else (None, answer)
+        reasoning, content = extract_reasoning(answer)
         message = {
             "role": "assistant",
             "refusal": None,

From 58f26a4cc7d0cec9cc22e4b32bc605540bcdef4f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 19:18:55 -0700
Subject: [PATCH 1340/1701] UI: Skip redundant work in chat loop when no tools
 are selected

---
 modules/chat.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 1649a6e822..c93972fbb2 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1239,7 +1239,7 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
         # Recover visible_prefix from existing visible text (e.g. on Continue
         # after a previous session had tool calls). Extract all 
         # blocks and any text between them (thinking blocks, intermediate text).
-        if not visible_prefix and _model_visible:
+        if tool_func_names and not visible_prefix and _model_visible:
             tc_matches = list(re.finditer(r'.*?', _model_visible, re.DOTALL))
             if tc_matches:
                 prefix_end = tc_matches[-1].end()
@@ -1253,7 +1253,8 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
         if visible_prefix:
             history['visible'][-1][1] = '\n\n'.join(visible_prefix + [_model_visible])
 
-        save_history(history, state['unique_id'], state['character_menu'], state['mode'])
+        if tool_func_names:
+            save_history(history, state['unique_id'], state['character_menu'], state['mode'])
 
         # Check for tool calls
         if not tool_func_names or shared.stop_everything:

From 04213dff143cc5a79aaa92fe3ba76271925833ee Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 19:55:02 -0700
Subject: [PATCH 1341/1701] Address copilot feedback

---
 modules/llama_cpp_server.py  | 2 +-
 modules/models_settings.py   | 3 ++-
 modules/text_generation.py   | 2 +-
 modules/tool_use.py          | 3 +++
 user_data/tools/calculate.py | 6 +++++-
 user_data/tools/roll_dice.py | 4 ++--
 6 files changed, 14 insertions(+), 6 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 192aa9e408..1425844d1e 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -136,7 +136,7 @@ def prepare_payload(self, state):
 
         logit_bias = []
         if state['custom_token_bans']:
-            logit_bias.extend([[int(token_id), False] for token_id in state['custom_token_bans'].split(',')])
+            logit_bias.extend([[int(token_id.strip()), False] for token_id in state['custom_token_bans'].split(',') if token_id.strip()])
 
         if state.get('logit_bias'):
             for token_id_str, bias in state['logit_bias'].items():
diff --git a/modules/models_settings.py b/modules/models_settings.py
index 5e69b60e3f..0e117176b7 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -431,7 +431,8 @@ def load_instruction_template(template):
     else:
         return ''
 
-    file_contents = open(filepath, 'r', encoding='utf-8').read()
+    with open(filepath, 'r', encoding='utf-8') as f:
+        file_contents = f.read()
     data = yaml.safe_load(file_contents)
     if 'instruction_template' in data:
         return data['instruction_template']
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 787c1814cd..d487cd2fbb 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -378,7 +378,7 @@ def generate_reply_HF(question, original_question, state, stopping_strings=None,
         generate_params['sampler_priority'] = [x.strip() for x in state['sampler_priority'].replace('\n', ',').split(',') if x.strip()]
 
     if state['custom_token_bans']:
-        to_ban = [int(x) for x in state['custom_token_bans'].split(',')]
+        to_ban = [int(x.strip()) for x in state['custom_token_bans'].split(',') if x.strip()]
         if len(to_ban) > 0:
             if generate_params.get('suppress_tokens', None):
                 generate_params['suppress_tokens'] += to_ban
diff --git a/modules/tool_use.py b/modules/tool_use.py
index d65c14eda4..5542485375 100644
--- a/modules/tool_use.py
+++ b/modules/tool_use.py
@@ -42,6 +42,9 @@ def load_tools(selected_names):
             continue
 
         func_name = tool_def.get('function', {}).get('name', name)
+        if func_name in executors:
+            logger.warning(f'Tool "{name}" declares function name "{func_name}" which conflicts with an already loaded tool. Skipping.')
+            continue
         tool_defs.append(tool_def)
         executors[func_name] = execute_fn
 
diff --git a/user_data/tools/calculate.py b/user_data/tools/calculate.py
index e88b71a3ed..94f74c4151 100644
--- a/user_data/tools/calculate.py
+++ b/user_data/tools/calculate.py
@@ -16,7 +16,11 @@ def _eval(node):
     if isinstance(node, ast.Constant) and isinstance(node.value, (int, float)):
         return node.value
     elif isinstance(node, ast.BinOp) and type(node.op) in OPERATORS:
-        return OPERATORS[type(node.op)](_eval(node.left), _eval(node.right))
+        left = _eval(node.left)
+        right = _eval(node.right)
+        if isinstance(node.op, ast.Pow) and isinstance(right, (int, float)) and abs(right) > 10000:
+            raise ValueError("Exponent too large (max 10000)")
+        return OPERATORS[type(node.op)](left, right)
     elif isinstance(node, ast.UnaryOp) and type(node.op) in OPERATORS:
         return OPERATORS[type(node.op)](_eval(node.operand))
     raise ValueError(f"Unsupported expression")
diff --git a/user_data/tools/roll_dice.py b/user_data/tools/roll_dice.py
index 9cab48a833..4af38ddcad 100644
--- a/user_data/tools/roll_dice.py
+++ b/user_data/tools/roll_dice.py
@@ -17,7 +17,7 @@
 
 
 def execute(arguments):
-    count = arguments.get("count", 1)
-    sides = arguments.get("sides", 20)
+    count = max(1, min(arguments.get("count", 1), 1000))
+    sides = max(2, min(arguments.get("sides", 20), 1000))
     rolls = [random.randint(1, sides) for _ in range(count)]
     return {"rolls": rolls, "total": sum(rolls)}

From 85ec85e569df750724551bf10d5970ee8e7eb351 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 20:21:01 -0700
Subject: [PATCH 1342/1701] UI: Fix Continue while in a tool-calling loop,
 remove the upper limit on number of tool calls

---
 modules/chat.py | 84 ++++++++++++++++++++++++++++++++++---------------
 1 file changed, 59 insertions(+), 25 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index c93972fbb2..b93f2b9d05 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -161,6 +161,49 @@ def _deserialize_tool_call_arguments(tool_calls):
     return result
 
 
+def _expand_tool_sequence(tool_seq):
+    """Expand a tool_sequence list into API messages.
+
+    Returns a list of dicts (role: assistant with tool_calls, or role: tool).
+    If any tool_call IDs are missing a matching tool result, a synthetic
+    empty result is inserted so the prompt is never malformed.
+    """
+    messages = []
+    expected_ids = []
+    seen_ids = set()
+
+    for item in tool_seq:
+        if 'tool_calls' in item:
+            deserialized = _deserialize_tool_call_arguments(item['tool_calls'])
+            messages.append({
+                "role": "assistant",
+                "content": "",
+                "tool_calls": deserialized
+            })
+            for tc in item['tool_calls']:
+                tc_id = tc.get('id', '')
+                if tc_id:
+                    expected_ids.append(tc_id)
+        elif item.get('role') == 'tool':
+            messages.append({
+                "role": "tool",
+                "content": item['content'],
+                "tool_call_id": item.get('tool_call_id', '')
+            })
+            seen_ids.add(item.get('tool_call_id', ''))
+
+    # Fill in synthetic results for any orphaned tool call IDs
+    for tc_id in expected_ids:
+        if tc_id not in seen_ids:
+            messages.append({
+                "role": "tool",
+                "content": "",
+                "tool_call_id": tc_id
+            })
+
+    return messages
+
+
 def generate_chat_prompt(user_input, state, **kwargs):
     impersonate = kwargs.get('impersonate', False)
     _continue = kwargs.get('_continue', False)
@@ -312,17 +355,8 @@ def generate_chat_prompt(user_input, state, **kwargs):
         meta_key = f"assistant_{row_idx}"
         tool_seq = metadata.get(meta_key, {}).get('tool_sequence', [])
         if tool_seq:
-            for item in reversed(tool_seq):
-                if 'tool_calls' in item:
-                    messages.insert(insert_pos, {
-                        "role": "assistant", "content": "",
-                        "tool_calls": _deserialize_tool_call_arguments(item['tool_calls'])
-                    })
-                elif item.get('role') == 'tool':
-                    messages.insert(insert_pos, {
-                        "role": "tool", "content": item['content'],
-                        "tool_call_id": item.get('tool_call_id', '')
-                    })
+            for msg in reversed(_expand_tool_sequence(tool_seq)):
+                messages.insert(insert_pos, msg)
 
         if entry_meta.get('role') == 'system':
             if user_msg:
@@ -400,17 +434,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
         # history loop during regenerate — needed so the model sees prior
         # tool calls and results when re-generating the final answer).
         current_tool_seq = metadata.get(f"assistant_{len(history)}", {}).get('tool_sequence', [])
-        for item in current_tool_seq:
-            if 'tool_calls' in item:
-                messages.append({
-                    "role": "assistant", "content": "",
-                    "tool_calls": _deserialize_tool_call_arguments(item['tool_calls'])
-                })
-            elif item.get('role') == 'tool':
-                messages.append({
-                    "role": "tool", "content": item['content'],
-                    "tool_call_id": item.get('tool_call_id', '')
-                })
+        messages.extend(_expand_tool_sequence(current_tool_seq))
 
     if impersonate and state['mode'] != 'chat-instruct':
         messages.append({"role": "user", "content": "fake user message replace me"})
@@ -1181,9 +1205,8 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
     visible_prefix = []  # Accumulated tool call summaries + results
     last_save_time = time.monotonic()
     save_interval = 8
-    max_tool_turns = 10
-
-    for _tool_turn in range(max_tool_turns):
+    _tool_turn = 0
+    while True:
         history = state['history']
 
         # Turn 0: use original flags; turns 2+: regenerate into the same entry.
@@ -1324,6 +1347,16 @@ def _render():
 
         # Execute tools, store results, and replace placeholders with real results
         for i, tc in enumerate(parsed_calls):
+            # Check for stop request before each tool execution
+            if shared.stop_everything:
+                for j in range(i, len(parsed_calls)):
+                    seq.append({'role': 'tool', 'content': 'Tool execution was cancelled by the user.', 'tool_call_id': parsed_calls[j]['id']})
+                    pending_placeholders[j] = f'{tc_headers[j]}\nCancelled\n'
+
+                history['visible'][-1][1] = '\n\n'.join(visible_prefix + pending_placeholders)
+                yield _render(), history
+                break
+
             fn_name = tc['function']['name']
             fn_args = tc['function'].get('arguments', {})
             result = execute_tool(fn_name, fn_args, tool_executors)
@@ -1345,6 +1378,7 @@ def _render():
         save_history(history, state['unique_id'], state['character_menu'], state['mode'])
 
         state['history'] = history
+        _tool_turn += 1
 
     state.pop('_tool_turn', None)
     state['history'] = history

From c094bc943c5912d592b0f796b75989af4fc82dc4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 12 Mar 2026 21:45:38 -0700
Subject: [PATCH 1343/1701] UI: Skip output extensions on intermediate
 tool-calling turns

---
 modules/chat.py | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index b93f2b9d05..57fd50e03b 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1214,6 +1214,7 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
         # that intermediate tool-loop regenerations don't pollute swipe history.
         if _tool_turn > 0:
             state['_tool_turn'] = True
+            state['_skip_output_extensions'] = True
 
         regen = regenerate if _tool_turn == 0 else True
         cont = _continue if _tool_turn == 0 else False
@@ -1381,6 +1382,18 @@ def _render():
         _tool_turn += 1
 
     state.pop('_tool_turn', None)
+
+    # If output extensions were deferred during tool turns, apply them now
+    # to the final model response only (not to tool call markers).
+    if state.pop('_skip_output_extensions', None):
+        _model_visible = apply_extensions('output', _model_visible, state, is_chat=True)
+        if visible_prefix:
+            history['visible'][-1][1] = '\n\n'.join(visible_prefix + [_model_visible])
+        else:
+            history['visible'][-1][1] = _model_visible
+
+        yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu']), history
+
     state['history'] = history
 
     # Sync version metadata so swipes show the full visible (with tool prefix)

From 5ddc1002d2f3d6cc8f65aad959dc810e92530319 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 02:40:09 -0700
Subject: [PATCH 1344/1701] Update ExLlamaV3 to 0.0.25

---
 requirements/full/requirements.txt | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 12e7fbaeb2..8e095fae85 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,7 +42,7 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.24/exllamav3-0.0.24+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.24/exllamav3-0.0.24+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.25/exllamav3-0.0.25+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.25/exllamav3-0.0.25+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"

From a4bef860b6d743c78d0aed36c934cc022df37bee Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 06:45:29 -0300
Subject: [PATCH 1345/1701] UI: Optimize chat streaming by batching morphdom to
 one update per animation frame

The monitor physically cannot paint faster than its refresh rate, so
intermediate morphdom calls between frames do redundant parsing, diffing,
and patching work that is never displayed.
---
 js/global_scope_js.js | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 62b31d3780..3207a68114 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -269,7 +269,21 @@ function removeLastClick() {
   document.getElementById("Remove-last").click();
 }
 
+let pendingMorphdomData = null;
+let morphdomRafId = null;
+
 function handleMorphdomUpdate(data) {
+  pendingMorphdomData = data;
+  if (!morphdomRafId) {
+    morphdomRafId = requestAnimationFrame(() => {
+      morphdomRafId = null;
+      applyMorphdomUpdate(pendingMorphdomData);
+      pendingMorphdomData = null;
+    });
+  }
+}
+
+function applyMorphdomUpdate(data) {
   // Determine target element and use it as query scope
   var target_element, target_html;
   if (data.last_message_only) {

From 5833d94d7fad6739ef72d3c169143fbc977211c2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 02:56:49 -0700
Subject: [PATCH 1346/1701] UI: Prevent word breaks in tables

---
 css/main.css | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/css/main.css b/css/main.css
index a7069f33c4..30c0d5f569 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1802,6 +1802,16 @@ table {
     border-collapse: collapse;
 }
 
+.message-body table {
+    display: block;
+    overflow-x: auto;
+}
+
+.message-body :is(td, th) {
+    word-break: normal;
+    overflow-wrap: normal;
+}
+
 table, tr, td, th, thead {
     border: 0;
 }

From fef95b9e56855707cc13e5dc1220193352bc4bc1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 03:05:09 -0700
Subject: [PATCH 1347/1701] UI: Fix an autoscroll race condition during chat
 streaming

---
 js/main.js | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index 0bc76b2aa4..135aa94806 100644
--- a/js/main.js
+++ b/js/main.js
@@ -145,6 +145,7 @@ targetElement.classList.add("pretty_scrollbar");
 targetElement.classList.add("chat-parent");
 window.isScrolled = false;
 let scrollTimeout;
+let isProgrammaticScroll = false;
 
 targetElement.addEventListener("scroll", function() {
   let diff = targetElement.scrollHeight - targetElement.clientHeight;
@@ -157,9 +158,10 @@ targetElement.addEventListener("scroll", function() {
 
   if(isAtBottomNow) {
     window.isScrolled = false;
-  } else {
+  } else if (!isProgrammaticScroll) {
     window.isScrolled = true;
   }
+  isProgrammaticScroll = false;
 
   // Clear previous timeout and set new one
   clearTimeout(scrollTimeout);
@@ -193,6 +195,7 @@ const observer = new MutationObserver(function(mutations) {
   if (!window.isScrolled && !isScrollingClassOnly) {
     const maxScroll = targetElement.scrollHeight - targetElement.clientHeight;
     if (maxScroll > 0 && targetElement.scrollTop < maxScroll - 1) {
+      isProgrammaticScroll = true;
       targetElement.scrollTop = maxScroll;
     }
   }
@@ -1091,6 +1094,7 @@ document.fonts.addEventListener("loadingdone", (event) => {
     if (!window.isScrolled) {
       const maxScroll = targetElement.scrollHeight - targetElement.clientHeight;
       if (targetElement.scrollTop < maxScroll - 5) {
+        isProgrammaticScroll = true;
         targetElement.scrollTop = maxScroll;
       }
     }

From 46288256511c0e42fb9fb4cfeb8585a2a9362947 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 03:17:36 -0700
Subject: [PATCH 1348/1701] Better solution to
 fef95b9e56855707cc13e5dc1220193352bc4bc1

---
 js/main.js | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/js/main.js b/js/main.js
index 135aa94806..c3e51c3c96 100644
--- a/js/main.js
+++ b/js/main.js
@@ -145,7 +145,8 @@ targetElement.classList.add("pretty_scrollbar");
 targetElement.classList.add("chat-parent");
 window.isScrolled = false;
 let scrollTimeout;
-let isProgrammaticScroll = false;
+let lastScrollTop = 0;
+let lastScrollHeight = 0;
 
 targetElement.addEventListener("scroll", function() {
   let diff = targetElement.scrollHeight - targetElement.clientHeight;
@@ -158,10 +159,11 @@ targetElement.addEventListener("scroll", function() {
 
   if(isAtBottomNow) {
     window.isScrolled = false;
-  } else if (!isProgrammaticScroll) {
+  } else if (targetElement.scrollTop < lastScrollTop && targetElement.scrollHeight >= lastScrollHeight) {
     window.isScrolled = true;
   }
-  isProgrammaticScroll = false;
+  lastScrollTop = targetElement.scrollTop;
+  lastScrollHeight = targetElement.scrollHeight;
 
   // Clear previous timeout and set new one
   clearTimeout(scrollTimeout);
@@ -195,7 +197,6 @@ const observer = new MutationObserver(function(mutations) {
   if (!window.isScrolled && !isScrollingClassOnly) {
     const maxScroll = targetElement.scrollHeight - targetElement.clientHeight;
     if (maxScroll > 0 && targetElement.scrollTop < maxScroll - 1) {
-      isProgrammaticScroll = true;
       targetElement.scrollTop = maxScroll;
     }
   }
@@ -1094,7 +1095,6 @@ document.fonts.addEventListener("loadingdone", (event) => {
     if (!window.isScrolled) {
       const maxScroll = targetElement.scrollHeight - targetElement.clientHeight;
       if (targetElement.scrollTop < maxScroll - 5) {
-        isProgrammaticScroll = true;
         targetElement.scrollTop = maxScroll;
       }
     }

From c39c187f47b79b540b8b0abf18979b5103f48bd0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 03:21:47 -0700
Subject: [PATCH 1349/1701] UI: Improve the style of table scrollbars

---
 css/main.css              | 3 +--
 modules/html_generator.py | 3 +++
 2 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/css/main.css b/css/main.css
index 30c0d5f569..7cc496a70d 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1802,8 +1802,7 @@ table {
     border-collapse: collapse;
 }
 
-.message-body table {
-    display: block;
+.table-wrapper {
     overflow-x: auto;
 }
 
diff --git a/modules/html_generator.py b/modules/html_generator.py
index f381160224..138d4ade62 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -317,6 +317,9 @@ def protect_asterisks_underscores_in_latex(match):
     # Unescape backslashes
     html_output = html_output.replace('\\\\', '\\')
 
+    # Wrap tables in a scrollable div
+    html_output = html_output.replace('
            {html.escape(repr(token_option))}{prob:.4f}
            {html.escape(repr(token_option))}{prob:.4f}
            Perplexity:{perplexity:.4f}
            ', '
            ').replace('
            ', '
            ')
+
     return html_output
 
 

From d0b72c73c08cc74c7d05c5afdbbef7f96b94dcfd Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 03:43:02 -0700
Subject: [PATCH 1350/1701] Update diffusers to 0.37

---
 requirements/full/requirements.txt               | 2 +-
 requirements/full/requirements_amd.txt           | 2 +-
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_cpu_only.txt      | 2 +-
 requirements/full/requirements_nowheels.txt      | 2 +-
 6 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 8e095fae85..03f4abacfa 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -2,7 +2,7 @@ accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 bitsandbytes==0.49.*
 datasets
-diffusers==0.36.*
+diffusers==0.37.*
 einops
 fastapi==0.112.4
 flash-linear-attention==0.4.*
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 19cc0d9d11..f3551fa2b0 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -1,7 +1,7 @@
 accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 datasets
-diffusers==0.36.*
+diffusers==0.37.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index ebe26f9db6..5e0cf8add5 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -1,7 +1,7 @@
 accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 datasets
-diffusers==0.36.*
+diffusers==0.37.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 491556908f..d55c3e245d 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -1,7 +1,7 @@
 accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 datasets
-diffusers==0.36.*
+diffusers==0.37.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 1c7c57354c..34e864ac28 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -1,7 +1,7 @@
 accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 datasets
-diffusers==0.36.*
+diffusers==0.37.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 63823db82d..6128c0edc9 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -1,7 +1,7 @@
 accelerate==1.12.*
 audioop-lts<1.0; python_version >= "3.13"
 datasets
-diffusers==0.36.*
+diffusers==0.37.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15

From b7670cc762206445e4e0ffed76f535c661bb85f4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 04:00:30 -0700
Subject: [PATCH 1351/1701] Add a tool calling tutorial

---
 docs/Tool Calling Tutorial.md | 144 ++++++++++++++++++++++++++++++++++
 1 file changed, 144 insertions(+)
 create mode 100644 docs/Tool Calling Tutorial.md

diff --git a/docs/Tool Calling Tutorial.md b/docs/Tool Calling Tutorial.md
new file mode 100644
index 0000000000..170bdff7c3
--- /dev/null
+++ b/docs/Tool Calling Tutorial.md	
@@ -0,0 +1,144 @@
+## Tool calling in the UI
+
+### 1. Load a model with tool-calling support
+
+Load a model with tool-calling support (Qwen, Mistral, Llama 4, etc.) from the Model tab.
+
+### 2. Select tools
+
+In the chat sidebar, check the tools you want the model to use:
+
+- **web_search** -- Search the web using DuckDuckGo.
+- **fetch_webpage** -- Fetch the content of a URL.
+- **calculate** -- Evaluate math expressions.
+- **get_datetime** -- Get the current date and time.
+- **roll_dice** -- Roll dice.
+
+### 3. Chat
+
+Send a message as usual. When the model decides it needs a tool, it will call it automatically. You will see each tool call and its result in a collapsible accordion inside the chat message.
+
+The model may call multiple tools in sequence before giving its final answer.
+
+## Writing custom tools
+
+Each tool is a single `.py` file in `user_data/tools/`. It needs two things:
+
+1. A `tool` dictionary that describes the function (name, description, parameters).
+2. An `execute(arguments)` function that runs it and returns the result.
+
+Here is a minimal example (`user_data/tools/get_datetime.py`):
+
+```python
+from datetime import datetime
+
+tool = {
+    "type": "function",
+    "function": {
+        "name": "get_datetime",
+        "description": "Get the current date and time.",
+        "parameters": {
+            "type": "object",
+            "properties": {},
+        }
+    }
+}
+
+
+def execute(arguments):
+    now = datetime.now()
+    return {"date": now.strftime("%Y-%m-%d"), "time": now.strftime("%I:%M %p")}
+```
+
+An example with parameters (`user_data/tools/roll_dice.py`):
+
+```python
+import random
+
+tool = {
+    "type": "function",
+    "function": {
+        "name": "roll_dice",
+        "description": "Roll one or more dice with the specified number of sides.",
+        "parameters": {
+            "type": "object",
+            "properties": {
+                "count": {"type": "integer", "description": "Number of dice to roll.", "default": 1},
+                "sides": {"type": "integer", "description": "Number of sides per die.", "default": 20},
+            },
+        }
+    }
+}
+
+
+def execute(arguments):
+    count = max(1, min(arguments.get("count", 1), 1000))
+    sides = max(2, min(arguments.get("sides", 20), 1000))
+    rolls = [random.randint(1, sides) for _ in range(count)]
+    return {"rolls": rolls, "total": sum(rolls)}
+```
+
+You can open the built-in tools in `user_data/tools/` for more examples.
+
+## Tool calling over the API
+
+Tool calling over the API follows the [OpenAI API](https://platform.openai.com/docs/guides/function-calling) convention. Define your tools, send them with your messages, and handle tool calls in a loop until the model gives a final answer.
+
+```python
+import json
+import requests
+
+url = "http://127.0.0.1:5000/v1/chat/completions"
+
+tools = [
+    {
+        "type": "function",
+        "function": {
+            "name": "get_weather",
+            "description": "Get the current weather for a given location.",
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "location": {"type": "string", "description": "City name"},
+                },
+                "required": ["location"]
+            }
+        }
+    }
+]
+
+
+def execute_tool(name, arguments):
+    if name == "get_weather":
+        return {"temperature": "14°C", "condition": "partly cloudy"}
+    return {"error": f"Unknown tool: {name}"}
+
+
+messages = [{"role": "user", "content": "What's the weather like in Paris?"}]
+
+for _ in range(10):
+    response = requests.post(url, json={"messages": messages, "tools": tools}).json()
+    choice = response["choices"][0]
+
+    if choice["finish_reason"] == "tool_calls":
+        messages.append({
+            "role": "assistant",
+            "content": choice["message"]["content"],
+            "tool_calls": choice["message"]["tool_calls"],
+        })
+
+        for tool_call in choice["message"]["tool_calls"]:
+            name = tool_call["function"]["name"]
+            arguments = json.loads(tool_call["function"]["arguments"])
+            result = execute_tool(name, arguments)
+            print(f"Tool call: {name}({arguments}) => {result}")
+
+            messages.append({
+                "role": "tool",
+                "tool_call_id": tool_call["id"],
+                "content": json.dumps(result),
+            })
+    else:
+        print(f"\nAssistant: {choice['message']['content']}")
+        break
+```

From e50b823eee369f50981fcaaef3bf05c8dc3e350d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 06:22:28 -0700
Subject: [PATCH 1352/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 12 files changed, 20 insertions(+), 20 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 03f4abacfa..c702a8d3a2 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.25/exllamav3-0.0.25+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.25/exllamav3-0.0.25+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index f3551fa2b0..65a9aa00bb 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 5e0cf8add5..bba62491e5 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index d55c3e245d..61dbf51b47 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 34e864ac28..384a552aa3 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index db23d4bfc7..0e3d67d379 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index e8cd9fd945..729829b3f4 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 24c558a97d..4b16414c6a 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index f2e8e69170..3a1764dc17 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 296c04326f..9d115c8669 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index aefce76969..4472e1d4d8 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 76bb58725c..dad7ee9f76 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.91.0/llama_cpp_binaries-0.91.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From e0a38da9f31c95332a5ca863217b1b2e485aecdc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 11:00:12 -0300
Subject: [PATCH 1353/1701] Improve tool call parsing for Devstral/GPT-OSS and
 preserve thinking across tool turns

---
 extensions/openai/utils.py | 51 +++++++++++++++++++++++++++++++++++---
 modules/chat.py            | 16 ++++++++++--
 2 files changed, 62 insertions(+), 5 deletions(-)

diff --git a/extensions/openai/utils.py b/extensions/openai/utils.py
index eb34ce8805..b179c267f2 100644
--- a/extensions/openai/utils.py
+++ b/extensions/openai/utils.py
@@ -126,8 +126,49 @@ def _parseChannelToolCalls(answer: str, tool_names: list[str]):
     """
     matches = []
     start_pos = None
-    for m in re.finditer(
+    # Pattern 1: to=functions.NAME before <|channel|> (GPT-OSS primary format)
+    # Pattern 2: to=functions.NAME after <|channel|> (alternative format)
+    patterns = [
+        r'to=functions\.([^<\s]+)\s*<\|channel\|>[^<]*<\|message\|>',
         r'<\|channel\|>\w+ to=functions\.([^<\s]+).*?<\|message\|>',
+    ]
+    for pattern in patterns:
+        for m in re.finditer(pattern, answer):
+            func_name = m.group(1).strip()
+            if func_name not in tool_names:
+                continue
+            json_str = _extractBalancedJson(answer, m.end())
+            if json_str is None:
+                continue
+            try:
+                arguments = json.loads(json_str)
+                if start_pos is None:
+                    prefix = answer.rfind('<|start|>assistant', 0, m.start())
+                    start_pos = prefix if prefix != -1 else m.start()
+                matches.append({
+                    "type": "function",
+                    "function": {
+                        "name": func_name,
+                        "arguments": arguments
+                    }
+                })
+            except json.JSONDecodeError:
+                pass
+        if matches:
+            break
+    return matches, start_pos
+
+
+def _parseMistralTokenToolCalls(answer: str, tool_names: list[str]):
+    """Parse Mistral/Devstral-style tool calls with [TOOL_CALLS] and [ARGS] special tokens.
+
+    Format:
+        [TOOL_CALLS]func_name[ARGS]{"arg": "value"}
+    """
+    matches = []
+    start_pos = None
+    for m in re.finditer(
+        r'\[TOOL_CALLS\]\s*(\S+?)\s*\[ARGS\]\s*',
         answer
     ):
         func_name = m.group(1).strip()
@@ -139,8 +180,7 @@ def _parseChannelToolCalls(answer: str, tool_names: list[str]):
         try:
             arguments = json.loads(json_str)
             if start_pos is None:
-                prefix = answer.rfind('<|start|>assistant', 0, m.start())
-                start_pos = prefix if prefix != -1 else m.start()
+                start_pos = m.start()
             matches.append({
                 "type": "function",
                 "function": {
@@ -497,6 +537,11 @@ def _return(matches, start_pos):
     if matches:
         return _return(matches, start_pos)
 
+    # Check for Mistral/Devstral-style tool calls ([TOOL_CALLS]name[ARGS]json)
+    matches, start_pos = _parseMistralTokenToolCalls(answer, tool_names)
+    if matches:
+        return _return(matches, start_pos)
+
     # Check for bare function-name style tool calls (e.g. Mistral format)
     matches, start_pos = _parseBareNameToolCalls(answer, tool_names)
     if matches:
diff --git a/modules/chat.py b/modules/chat.py
index 57fd50e03b..2c6f0ab2f9 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -177,7 +177,7 @@ def _expand_tool_sequence(tool_seq):
             deserialized = _deserialize_tool_call_arguments(item['tool_calls'])
             messages.append({
                 "role": "assistant",
-                "content": "",
+                "content": item.get('content', ''),
                 "tool_calls": deserialized
             })
             for tc in item['tool_calls']:
@@ -1324,7 +1324,19 @@ def _render():
 
             tc_headers.append(f'{fn_name}({args_summary})')
 
-        seq.append({'tool_calls': serialized})
+        seq_entry = {'tool_calls': serialized}
+        if content_prefix.strip():
+            # Strip GPT-OSS channel tokens so they don't get double-wrapped
+            # by the template (which adds its own channel markup).
+            clean = content_prefix.strip()
+            if '<|channel|>' in clean and '<|message|>' in clean:
+                inner = clean.split('<|message|>', 1)[1] if '<|message|>' in clean else clean
+                if '<|end|>' in inner:
+                    inner = inner.split('<|end|>', 1)[0]
+                clean = inner.strip()
+            if clean:
+                seq_entry['content'] = clean
+        seq.append(seq_entry)
 
         # Clear internal (raw tool markup)
         history['internal'][-1][1] = ''

From aab2596d29a850a185eb0c25c83f7dcf7387d9fc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 15:47:11 -0300
Subject: [PATCH 1354/1701] UI: Fix multiple thinking blocks rendering as raw
 text in HTML generator

---
 modules/html_generator.py | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 138d4ade62..8f3f261f9a 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -369,8 +369,10 @@ def process_text_segment(text, is_last_segment):
         if not text.strip():
             return
 
-        thinking_content, remaining = extract_thinking_block(text)
-        if thinking_content is not None:
+        while text.strip():
+            thinking_content, remaining = extract_thinking_block(text)
+            if thinking_content is None:
+                break
             has_remaining = bool(remaining.strip()) or not is_last_segment
             html_parts.append(build_thinking_block(thinking_content, message_id, has_remaining, think_idx))
             think_idx += 1

From d4c22ced83dd6fed7bb2785c7e6d5f7cbeace2db Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 15:47:14 -0300
Subject: [PATCH 1355/1701] UI: Optimize syntax highlighting and autoscroll by
 moving from MutationObserver to morphdom updates

---
 js/global_scope_js.js |  45 +++++++++++++++
 js/main.js            | 129 ++++++++++++++----------------------------
 2 files changed, 88 insertions(+), 86 deletions(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 3207a68114..425c2c5970 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -269,6 +269,34 @@ function removeLastClick() {
   document.getElementById("Remove-last").click();
 }
 
+function autoScrollToBottom() {
+  if (!window.isScrolled) {
+    const chatParent = document.getElementById("chat")?.parentNode?.parentNode?.parentNode;
+    if (chatParent) {
+      const maxScroll = chatParent.scrollHeight - chatParent.clientHeight;
+      if (maxScroll > 0 && chatParent.scrollTop < maxScroll - 1) {
+        chatParent.scrollTop = maxScroll;
+      }
+    }
+  }
+}
+
+function updateInstructPadding() {
+  const chatElement = document.getElementById("chat");
+  if (chatElement && chatElement.getAttribute("data-mode") === "instruct") {
+    const messagesContainer = chatElement.querySelector(".messages");
+    const lastChild = messagesContainer?.lastElementChild;
+    const prevSibling = lastChild?.previousElementSibling;
+    if (lastChild && prevSibling) {
+      let bufferHeight = Math.max(0, Math.max(window.innerHeight - 128 - 84, window.innerHeight - prevSibling.offsetHeight - 84) - lastChild.offsetHeight);
+      if (window.innerWidth <= 924) {
+        bufferHeight = Math.max(0, bufferHeight - 32);
+      }
+      messagesContainer.style.paddingBottom = `${bufferHeight}px`;
+    }
+  }
+}
+
 let pendingMorphdomData = null;
 let morphdomRafId = null;
 
@@ -373,10 +401,23 @@ function applyMorphdomUpdate(data) {
     }
   );
 
+  // Syntax highlighting and LaTeX
+  if (window.doSyntaxHighlighting) {
+    window.doSyntaxHighlighting();
+  }
+
+  // Auto-scroll runs both before and after padding update.
+  // Before: so content growth isn't hidden by padding absorption.
+  // After: so padding-added space is also scrolled into view.
+  autoScrollToBottom();
+  updateInstructPadding();
+  autoScrollToBottom();
+
   // Add toggle listeners for new blocks
   queryScope.querySelectorAll(".thinking-block").forEach(block => {
     if (!block._hasToggleListener) {
       block.addEventListener("toggle", function(e) {
+        const wasScrolled = window.isScrolled;
         if (this.open) {
           const content = this.querySelector(".thinking-content");
           if (content) {
@@ -385,6 +426,10 @@ function applyMorphdomUpdate(data) {
             }, 0);
           }
         }
+        updateInstructPadding();
+        // Restore scroll state so the browser's layout adjustment
+        // from the toggle doesn't disable auto-scroll
+        window.isScrolled = wasScrolled;
       });
       block._hasToggleListener = true;
     }
diff --git a/js/main.js b/js/main.js
index c3e51c3c96..0cefaa6e27 100644
--- a/js/main.js
+++ b/js/main.js
@@ -147,6 +147,7 @@ window.isScrolled = false;
 let scrollTimeout;
 let lastScrollTop = 0;
 let lastScrollHeight = 0;
+let lastClientHeight = 0;
 
 targetElement.addEventListener("scroll", function() {
   let diff = targetElement.scrollHeight - targetElement.clientHeight;
@@ -159,11 +160,12 @@ targetElement.addEventListener("scroll", function() {
 
   if(isAtBottomNow) {
     window.isScrolled = false;
-  } else if (targetElement.scrollTop < lastScrollTop && targetElement.scrollHeight >= lastScrollHeight) {
+  } else if (targetElement.scrollTop < lastScrollTop && targetElement.scrollHeight >= lastScrollHeight && targetElement.clientHeight <= lastClientHeight) {
     window.isScrolled = true;
   }
   lastScrollTop = targetElement.scrollTop;
   lastScrollHeight = targetElement.scrollHeight;
+  lastClientHeight = targetElement.clientHeight;
 
   // Clear previous timeout and set new one
   clearTimeout(scrollTimeout);
@@ -174,14 +176,7 @@ targetElement.addEventListener("scroll", function() {
 });
 
 // Create a MutationObserver instance
-const observer = new MutationObserver(function(mutations) {
-  // Check if this is just the scrolling class being toggled
-  const isScrollingClassOnly = mutations.every(mutation =>
-    mutation.type === "attributes" &&
-    mutation.attributeName === "class" &&
-    mutation.target === targetElement
-  );
-
+const observer = new MutationObserver(function() {
   if (targetElement.classList.contains("_generating")) {
     typing.parentNode.classList.add("visible-dots");
     document.getElementById("stop").style.display = "flex";
@@ -191,44 +186,11 @@ const observer = new MutationObserver(function(mutations) {
     document.getElementById("stop").style.display = "none";
     document.getElementById("Generate").style.display = "flex";
   }
-
-  doSyntaxHighlighting();
-
-  if (!window.isScrolled && !isScrollingClassOnly) {
-    const maxScroll = targetElement.scrollHeight - targetElement.clientHeight;
-    if (maxScroll > 0 && targetElement.scrollTop < maxScroll - 1) {
-      targetElement.scrollTop = maxScroll;
-    }
-  }
-
-  const chatElement = document.getElementById("chat");
-  if (chatElement && chatElement.getAttribute("data-mode") === "instruct") {
-    const messagesContainer = chatElement.querySelector(".messages");
-    const lastChild = messagesContainer?.lastElementChild;
-    const prevSibling = lastChild?.previousElementSibling;
-    if (lastChild && prevSibling) {
-      // Add padding to the messages container to create room for the last message.
-      // The purpose of this is to avoid constant scrolling during streaming in
-      // instruct mode.
-      let bufferHeight = Math.max(0, Math.max(window.innerHeight - 128 - 84, window.innerHeight - prevSibling.offsetHeight - 84) - lastChild.offsetHeight);
-
-      // Subtract header height when screen width is <= 924px
-      if (window.innerWidth <= 924) {
-        bufferHeight = Math.max(0, bufferHeight - 32);
-      }
-
-      messagesContainer.style.paddingBottom = `${bufferHeight}px`;
-    }
-  }
 });
 
-// Configure the observer to watch for changes in the subtree and attributes
+// Only watch for attribute changes on targetElement (e.g. _generating class)
 const config = {
-  childList: true,
-  subtree: true,
-  characterData: true,
-  attributeOldValue: true,
-  characterDataOldValue: true
+  attributes: true
 };
 
 // Start observing the target element
@@ -247,55 +209,50 @@ function isElementVisibleOnScreen(element) {
   );
 }
 
-function doSyntaxHighlighting() {
+window.doSyntaxHighlighting = function() {
   const messageBodies = document.getElementById("chat").querySelectorAll(".message-body");
 
   if (messageBodies.length > 0) {
-    observer.disconnect();
-
-    try {
-      let hasSeenVisible = false;
-
-      // Go from last message to first
-      for (let i = messageBodies.length - 1; i >= 0; i--) {
-        const messageBody = messageBodies[i];
-
-        if (isElementVisibleOnScreen(messageBody)) {
-          hasSeenVisible = true;
-
-          // Handle both code and math in a single pass through each message
-          const codeBlocks = messageBody.querySelectorAll("pre code:not([data-highlighted])");
-          codeBlocks.forEach((codeBlock) => {
-            hljs.highlightElement(codeBlock);
-            codeBlock.setAttribute("data-highlighted", "true");
-            codeBlock.classList.add("pretty_scrollbar");
-          });
-
-          // Only render math in visible elements
-          const mathContainers = messageBody.querySelectorAll("p, span, li, td, th, h1, h2, h3, h4, h5, h6, blockquote, figcaption, caption, dd, dt");
-          mathContainers.forEach(container => {
-            if (isElementVisibleOnScreen(container)) {
-              renderMathInElement(container, {
-                delimiters: [
-                  { left: "$$", right: "$$", display: true },
-                  { left: "$", right: "$", display: false },
-                  { left: "\\(", right: "\\)", display: false },
-                  { left: "\\[", right: "\\]", display: true },
-                ],
-              });
-            }
-          });
-        } else if (hasSeenVisible) {
-        // We've seen visible messages but this one is not visible
-        // Since we're going from last to first, we can break
-          break;
-        }
+    let hasSeenVisible = false;
+
+    // Go from last message to first
+    for (let i = messageBodies.length - 1; i >= 0; i--) {
+      const messageBody = messageBodies[i];
+
+      if (isElementVisibleOnScreen(messageBody)) {
+        hasSeenVisible = true;
+
+        // Handle both code and math in a single pass through each message
+        const codeBlocks = messageBody.querySelectorAll("pre code:not([data-highlighted])");
+        codeBlocks.forEach((codeBlock) => {
+          hljs.highlightElement(codeBlock);
+          codeBlock.setAttribute("data-highlighted", "true");
+          codeBlock.classList.add("pretty_scrollbar");
+        });
+
+        // Only render math in visible elements
+        const mathContainers = messageBody.querySelectorAll("p, span, li, td, th, h1, h2, h3, h4, h5, h6, blockquote, figcaption, caption, dd, dt");
+        mathContainers.forEach(container => {
+          if (isElementVisibleOnScreen(container)) {
+            renderMathInElement(container, {
+              delimiters: [
+                { left: "$$", right: "$$", display: true },
+                { left: "$", right: "$", display: false },
+                { left: "\\(", right: "\\)", display: false },
+                { left: "\\[", right: "\\]", display: true },
+              ],
+            });
+          }
+        });
+      } else if (hasSeenVisible) {
+      // We've seen visible messages but this one is not visible
+      // Since we're going from last to first, we can break
+        break;
       }
-    } finally {
-      observer.observe(targetElement, config);
     }
   }
 }
+const doSyntaxHighlighting = window.doSyntaxHighlighting;
 
 //------------------------------------------------
 // Add some scrollbars

From 5362bbb4132ae5ddbed4c4dab739e7dd64c1e6ab Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 12:09:08 -0700
Subject: [PATCH 1356/1701] Make web_search not download the page contents, use
 fetch_webpage instead

---
 modules/ui_chat.py            | 10 ++++++++++
 modules/web_search.py         | 14 ++++++++++++--
 user_data/tools/web_search.py | 14 ++++++--------
 3 files changed, 28 insertions(+), 10 deletions(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 039b9af640..ea341fa614 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -96,6 +96,16 @@ def create_ui():
                 shared.gradio['tools_refresh'] = gr.Button('Refresh list', elem_id='tools-refresh-btn', visible=False)
                 shared.gradio['tools_refresh'].click(fn=lambda: gr.update(choices=get_available_tools()), inputs=[], outputs=[shared.gradio['selected_tools']])
 
+                def sync_web_tools(selected):
+                    if 'web_search' in selected and 'fetch_webpage' not in selected:
+                        selected.append('fetch_webpage')
+                    elif 'web_search' not in selected and 'fetch_webpage' in selected:
+                        selected.remove('fetch_webpage')
+
+                    return gr.update(value=selected)
+
+                shared.gradio['selected_tools'].change(fn=sync_web_tools, inputs=[shared.gradio['selected_tools']], outputs=[shared.gradio['selected_tools']], show_progress=False)
+
                 gr.HTML("
            ")
 
                 with gr.Row():
diff --git a/modules/web_search.py b/modules/web_search.py
index b14cd04240..754dd11186 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -49,8 +49,8 @@ def download_web_page(url, timeout=10, include_links=False):
         return ""
 
 
-def perform_web_search(query, num_pages=3, max_workers=5, timeout=10):
-    """Perform web search and return results with content"""
+def perform_web_search(query, num_pages=3, max_workers=5, timeout=10, fetch_content=True):
+    """Perform web search and return results, optionally with page content"""
     try:
         search_url = f"https://html.duckduckgo.com/html/?q={quote_plus(query)}"
 
@@ -78,6 +78,16 @@ def perform_web_search(query, num_pages=3, max_workers=5, timeout=10):
 
         search_results = [None] * len(download_tasks)  # Pre-allocate to maintain order
 
+        if not fetch_content:
+            for url, title, index in download_tasks:
+                search_results[index] = {
+                    'title': title,
+                    'url': url,
+                    'content': ''
+                }
+
+            return search_results
+
         # Download pages in parallel
         with concurrent.futures.ThreadPoolExecutor(max_workers=max_workers) as executor:
             # Submit all download tasks
diff --git a/user_data/tools/web_search.py b/user_data/tools/web_search.py
index 8084596368..30d13473a2 100644
--- a/user_data/tools/web_search.py
+++ b/user_data/tools/web_search.py
@@ -1,16 +1,15 @@
-from modules.web_search import perform_web_search, truncate_content_by_tokens
+from modules.web_search import perform_web_search
 
 tool = {
     "type": "function",
     "function": {
         "name": "web_search",
-        "description": "Search the web using DuckDuckGo and return page contents.",
+        "description": "Search the web using DuckDuckGo and return a list of result titles and URLs. Use fetch_webpage to read the contents of a specific result.",
         "parameters": {
             "type": "object",
             "properties": {
                 "query": {"type": "string", "description": "The search query."},
-                "num_pages": {"type": "integer", "description": "Number of search result pages to fetch (default: 3)."},
-                "max_tokens": {"type": "integer", "description": "Maximum number of tokens per page result (default: 2048)."},
+                "num_pages": {"type": "integer", "description": "Number of search results to return (default: 3)."},
             },
             "required": ["query"]
         }
@@ -21,11 +20,10 @@
 def execute(arguments):
     query = arguments.get("query", "")
     num_pages = arguments.get("num_pages", 3)
-    max_tokens = arguments.get("max_tokens", 2048)
-    results = perform_web_search(query, num_pages=num_pages)
+    results = perform_web_search(query, num_pages=num_pages, fetch_content=False)
     output = []
     for r in results:
-        if r and r["content"].strip():
-            output.append({"title": r["title"], "url": r["url"], "content": truncate_content_by_tokens(r["content"], max_tokens=max_tokens)})
+        if r:
+            output.append({"title": r["title"], "url": r["url"]})
 
     return output if output else [{"error": "No results found."}]

From cabb95f0d6077d44741eb2a3ee0587470ade6300 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 12:24:48 -0700
Subject: [PATCH 1357/1701] UI: Increase the instruct width to 768px

---
 css/html_instruct_style.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index 72a148c3a9..d4780350de 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -78,7 +78,7 @@
 
 .chat .user-message .text,
 .chat .assistant-message .text {
-    max-width: 700px;
+    max-width: 768px;
     margin-left: auto;
     margin-right: auto;
 }

From 24e7e77b55e7758f7bd07e07016cc88b8b188c8b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 12:37:10 -0700
Subject: [PATCH 1358/1701] Clean up

---
 modules/chat.py    | 2 +-
 modules/ui_chat.py | 2 --
 2 files changed, 1 insertion(+), 3 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 2c6f0ab2f9..87e528512b 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1330,7 +1330,7 @@ def _render():
             # by the template (which adds its own channel markup).
             clean = content_prefix.strip()
             if '<|channel|>' in clean and '<|message|>' in clean:
-                inner = clean.split('<|message|>', 1)[1] if '<|message|>' in clean else clean
+                inner = clean.split('<|message|>', 1)[1]
                 if '<|end|>' in inner:
                     inner = inner.split('<|end|>', 1)[0]
                 clean = inner.strip()
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index ea341fa614..ce9fc0a2b7 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -99,8 +99,6 @@ def create_ui():
                 def sync_web_tools(selected):
                     if 'web_search' in selected and 'fetch_webpage' not in selected:
                         selected.append('fetch_webpage')
-                    elif 'web_search' not in selected and 'fetch_webpage' in selected:
-                        selected.remove('fetch_webpage')
 
                     return gr.update(value=selected)
 

From 0cd245bcbb46d894989e1ddd688d6640ba6ac537 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 12:58:56 -0700
Subject: [PATCH 1359/1701] UI: Make autoscroll more robust after the
 optimizations

---
 js/main.js | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/js/main.js b/js/main.js
index 0cefaa6e27..a8bbbc71e4 100644
--- a/js/main.js
+++ b/js/main.js
@@ -181,6 +181,13 @@ const observer = new MutationObserver(function() {
     typing.parentNode.classList.add("visible-dots");
     document.getElementById("stop").style.display = "flex";
     document.getElementById("Generate").style.display = "none";
+    // If the user is near the bottom, ensure auto-scroll is enabled
+    // for the new reply. This catches cases where isScrolled was
+    // incorrectly set to true by layout shifts during page load, etc.
+    const diff = targetElement.scrollHeight - targetElement.clientHeight;
+    if (Math.abs(targetElement.scrollTop - diff) <= 10 || diff <= 0) {
+      window.isScrolled = false;
+    }
   } else {
     typing.parentNode.classList.remove("visible-dots");
     document.getElementById("stop").style.display = "none";

From cb88066d15c6ffa54774f805190f000951c05e84 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 13:17:41 -0700
Subject: [PATCH 1360/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 12 files changed, 20 insertions(+), 20 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index c702a8d3a2..8a0802f7be 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.25/exllamav3-0.0.25+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.25/exllamav3-0.0.25+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 65a9aa00bb..9b31d668a9 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index bba62491e5..138639e551 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 61dbf51b47..f3ebd171f8 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 384a552aa3..e32a2ed1d1 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 0e3d67d379..93eb3b8555 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 729829b3f4..36e0e4d910 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 4b16414c6a..495bd5fa02 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 3a1764dc17..7e82f68d42 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 9d115c8669..046619e141 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 4472e1d4d8..590562f8b7 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index dad7ee9f76..bf80deb07c 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.92.0/llama_cpp_binaries-0.92.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From e8d1c663037666bafc0a45f4be0471a88fda4d57 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 18:13:12 -0700
Subject: [PATCH 1361/1701] Clean up tool calling code

---
 extensions/openai/completions.py |   7 +-
 extensions/openai/utils.py       | 558 -------------------------------
 modules/chat.py                  |  17 +-
 modules/tool_parsing.py          | 553 ++++++++++++++++++++++++++++++
 modules/tool_use.py              |   6 +-
 modules/ui_chat.py               |   2 +-
 modules/web_search.py            |  37 +-
 7 files changed, 604 insertions(+), 576 deletions(-)
 create mode 100644 modules/tool_parsing.py

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 290a5bc017..27defe4250 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -11,7 +11,8 @@
 
 from extensions.openai.errors import InvalidRequestError
 from extensions.openai.typing import ToolDefinition
-from extensions.openai.utils import debug_msg, getToolCallId, parseToolCall
+from extensions.openai.utils import debug_msg
+from modules.tool_parsing import get_tool_call_id, parse_tool_call
 from modules import shared
 from modules.reasoning import extract_reasoning
 from modules.chat import (
@@ -491,10 +492,10 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
         answer = a['internal'][-1][1]
 
         if supported_tools is not None:
-            tool_call = parseToolCall(answer[end_last_tool_call:], supported_tools) if len(answer) > 0 else []
+            tool_call = parse_tool_call(answer[end_last_tool_call:], supported_tools) if len(answer) > 0 else []
             if len(tool_call) > 0:
                 for tc in tool_call:
-                    tc["id"] = getToolCallId()
+                    tc["id"] = get_tool_call_id()
                     if stream:
                         tc["index"] = len(tool_calls)
                     tc["function"]["arguments"] = json.dumps(tc["function"]["arguments"])
diff --git a/extensions/openai/utils.py b/extensions/openai/utils.py
index b179c267f2..2b4147690e 100644
--- a/extensions/openai/utils.py
+++ b/extensions/openai/utils.py
@@ -1,8 +1,5 @@
 import base64
-import json
 import os
-import random
-import re
 import time
 import traceback
 from typing import Callable, Optional
@@ -55,558 +52,3 @@ def _start_cloudflared(port: int, tunnel_id: str, max_attempts: int = 3, on_star
             time.sleep(3)
 
         raise Exception('Could not start cloudflared.')
-
-
-def getToolCallId() -> str:
-    letter_bytes = "abcdefghijklmnopqrstuvwxyz0123456789"
-    b = [random.choice(letter_bytes) for _ in range(8)]
-    return "call_" + "".join(b).lower()
-
-
-def checkAndSanitizeToolCallCandidate(candidate_dict: dict, tool_names: list[str]):
-    # check if property 'function' exists and is a dictionary, otherwise adapt dict
-    if 'function' not in candidate_dict and 'name' in candidate_dict and isinstance(candidate_dict['name'], str):
-        candidate_dict = {"type": "function", "function": candidate_dict}
-    if 'function' in candidate_dict and isinstance(candidate_dict['function'], str):
-        candidate_dict['name'] = candidate_dict['function']
-        del candidate_dict['function']
-        candidate_dict = {"type": "function", "function": candidate_dict}
-    if 'function' in candidate_dict and isinstance(candidate_dict['function'], dict):
-        # check if 'name' exists within 'function' and is part of known tools
-        if 'name' in candidate_dict['function'] and candidate_dict['function']['name'] in tool_names:
-            candidate_dict["type"] = "function"  # ensure required property 'type' exists and has the right value
-            # map property 'parameters' used by some older models to 'arguments'
-            if "arguments" not in candidate_dict["function"] and "parameters" in candidate_dict["function"]:
-                candidate_dict["function"]["arguments"] = candidate_dict["function"]["parameters"]
-                del candidate_dict["function"]["parameters"]
-            return candidate_dict
-    return None
-
-
-def _extractBalancedJson(text: str, start: int) -> str | None:
-    """Extract a balanced JSON object from text starting at the given position.
-
-    Walks through the string tracking brace depth and string boundaries
-    to correctly handle arbitrary nesting levels.
-    """
-    if start >= len(text) or text[start] != '{':
-        return None
-    depth = 0
-    in_string = False
-    escape_next = False
-    for i in range(start, len(text)):
-        c = text[i]
-        if escape_next:
-            escape_next = False
-            continue
-        if c == '\\' and in_string:
-            escape_next = True
-            continue
-        if c == '"':
-            in_string = not in_string
-            continue
-        if in_string:
-            continue
-        if c == '{':
-            depth += 1
-        elif c == '}':
-            depth -= 1
-            if depth == 0:
-                return text[start:i + 1]
-    return None
-
-
-def _parseChannelToolCalls(answer: str, tool_names: list[str]):
-    """Parse channel-based tool calls used by GPT-OSS and similar models.
-
-    Format:
-        <|start|>assistant to=functions.func_name<|channel|>commentary json<|message|>{"arg": "value"}
-    or:
-        <|channel|>commentary to=functions.func_name <|constrain|>json<|message|>{"arg": "value"}
-    """
-    matches = []
-    start_pos = None
-    # Pattern 1: to=functions.NAME before <|channel|> (GPT-OSS primary format)
-    # Pattern 2: to=functions.NAME after <|channel|> (alternative format)
-    patterns = [
-        r'to=functions\.([^<\s]+)\s*<\|channel\|>[^<]*<\|message\|>',
-        r'<\|channel\|>\w+ to=functions\.([^<\s]+).*?<\|message\|>',
-    ]
-    for pattern in patterns:
-        for m in re.finditer(pattern, answer):
-            func_name = m.group(1).strip()
-            if func_name not in tool_names:
-                continue
-            json_str = _extractBalancedJson(answer, m.end())
-            if json_str is None:
-                continue
-            try:
-                arguments = json.loads(json_str)
-                if start_pos is None:
-                    prefix = answer.rfind('<|start|>assistant', 0, m.start())
-                    start_pos = prefix if prefix != -1 else m.start()
-                matches.append({
-                    "type": "function",
-                    "function": {
-                        "name": func_name,
-                        "arguments": arguments
-                    }
-                })
-            except json.JSONDecodeError:
-                pass
-        if matches:
-            break
-    return matches, start_pos
-
-
-def _parseMistralTokenToolCalls(answer: str, tool_names: list[str]):
-    """Parse Mistral/Devstral-style tool calls with [TOOL_CALLS] and [ARGS] special tokens.
-
-    Format:
-        [TOOL_CALLS]func_name[ARGS]{"arg": "value"}
-    """
-    matches = []
-    start_pos = None
-    for m in re.finditer(
-        r'\[TOOL_CALLS\]\s*(\S+?)\s*\[ARGS\]\s*',
-        answer
-    ):
-        func_name = m.group(1).strip()
-        if func_name not in tool_names:
-            continue
-        json_str = _extractBalancedJson(answer, m.end())
-        if json_str is None:
-            continue
-        try:
-            arguments = json.loads(json_str)
-            if start_pos is None:
-                start_pos = m.start()
-            matches.append({
-                "type": "function",
-                "function": {
-                    "name": func_name,
-                    "arguments": arguments
-                }
-            })
-        except json.JSONDecodeError:
-            pass
-    return matches, start_pos
-
-
-def _parseBareNameToolCalls(answer: str, tool_names: list[str]):
-    """Parse bare function-name style tool calls used by Mistral and similar models.
-
-    Format:
-        functionName{"arg": "value"}
-    Multiple calls are concatenated directly or separated by whitespace.
-    """
-    matches = []
-    start_pos = None
-    # Match tool name followed by opening brace, then extract balanced JSON
-    escaped_names = [re.escape(name) for name in tool_names]
-    pattern = r'(?:' + '|'.join(escaped_names) + r')\s*\{'
-    for match in re.finditer(pattern, answer):
-        text = match.group(0)
-        name = None
-        for n in tool_names:
-            if text.startswith(n):
-                name = n
-                break
-        if not name:
-            continue
-        brace_start = match.end() - 1
-        json_str = _extractBalancedJson(answer, brace_start)
-        if json_str is None:
-            continue
-        try:
-            arguments = json.loads(json_str)
-            if start_pos is None:
-                start_pos = match.start()
-            matches.append({
-                "type": "function",
-                "function": {
-                    "name": name,
-                    "arguments": arguments
-                }
-            })
-        except json.JSONDecodeError:
-            pass
-    return matches, start_pos
-
-
-def _parseXmlParamToolCalls(answer: str, tool_names: list[str]):
-    """Parse XML-parameter style tool calls used by Qwen3.5 and similar models.
-
-    Format:
-        
-        
-        value
-        
-        
-    """
-    matches = []
-    start_pos = None
-    for tc_match in re.finditer(r'\s*(.*?)\s*', answer, re.DOTALL):
-        tc_content = tc_match.group(1)
-        func_match = re.search(r']+)>', tc_content)
-        if not func_match:
-            continue
-        func_name = func_match.group(1).strip()
-        if func_name not in tool_names:
-            continue
-        arguments = {}
-        for param_match in re.finditer(r']+)>\s*(.*?)\s*', tc_content, re.DOTALL):
-            param_name = param_match.group(1).strip()
-            param_value = param_match.group(2).strip()
-            try:
-                param_value = json.loads(param_value)
-            except (json.JSONDecodeError, ValueError):
-                pass  # keep as string
-            arguments[param_name] = param_value
-        if start_pos is None:
-            start_pos = tc_match.start()
-        matches.append({
-            "type": "function",
-            "function": {
-                "name": func_name,
-                "arguments": arguments
-            }
-        })
-    return matches, start_pos
-
-
-def _parseKimiToolCalls(answer: str, tool_names: list[str]):
-    """Parse Kimi-K2-style tool calls using pipe-delimited tokens.
-
-    Format:
-        <|tool_calls_section_begin|>
-        <|tool_call_begin|>functions.func_name:index<|tool_call_argument_begin|>{"arg": "value"}<|tool_call_end|>
-        <|tool_calls_section_end|>
-    """
-    matches = []
-    start_pos = None
-    for m in re.finditer(
-        r'<\|tool_call_begin\|>\s*(?:functions\.)?(\S+?)(?::\d+)?\s*<\|tool_call_argument_begin\|>\s*',
-        answer
-    ):
-        func_name = m.group(1).strip()
-        if func_name not in tool_names:
-            continue
-        json_str = _extractBalancedJson(answer, m.end())
-        if json_str is None:
-            continue
-        try:
-            arguments = json.loads(json_str)
-            if start_pos is None:
-                # Check for section begin marker before the call marker
-                section = answer.rfind('<|tool_calls_section_begin|>', 0, m.start())
-                start_pos = section if section != -1 else m.start()
-            matches.append({
-                "type": "function",
-                "function": {
-                    "name": func_name,
-                    "arguments": arguments
-                }
-            })
-        except json.JSONDecodeError:
-            pass
-    return matches, start_pos
-
-
-def _parseMiniMaxToolCalls(answer: str, tool_names: list[str]):
-    """Parse MiniMax-style tool calls using invoke/parameter XML tags.
-
-    Format:
-        
-        
-        value
-        
-        
-    """
-    matches = []
-    start_pos = None
-    for tc_match in re.finditer(r'\s*(.*?)\s*', answer, re.DOTALL):
-        tc_content = tc_match.group(1)
-        # Split on  to handle multiple parallel calls in one block
-        for invoke_match in re.finditer(r'(.*?)', tc_content, re.DOTALL):
-            func_name = invoke_match.group(1).strip()
-            if func_name not in tool_names:
-                continue
-            invoke_body = invoke_match.group(2)
-            arguments = {}
-            for param_match in re.finditer(r'\s*(.*?)\s*', invoke_body, re.DOTALL):
-                param_name = param_match.group(1).strip()
-                param_value = param_match.group(2).strip()
-                try:
-                    param_value = json.loads(param_value)
-                except (json.JSONDecodeError, ValueError):
-                    pass  # keep as string
-                arguments[param_name] = param_value
-            if start_pos is None:
-                start_pos = tc_match.start()
-            matches.append({
-                "type": "function",
-                "function": {
-                    "name": func_name,
-                    "arguments": arguments
-                }
-            })
-    return matches, start_pos
-
-
-def _parseDeepSeekToolCalls(answer: str, tool_names: list[str]):
-    """Parse DeepSeek-style tool calls using fullwidth Unicode token delimiters.
-
-    Format:
-        <｜tool▁calls▁begin｜><｜tool▁call▁begin｜>func_name<｜tool▁sep｜>{"arg": "value"}<｜tool▁call▁end｜><｜tool▁calls▁end｜>
-    """
-    matches = []
-    start_pos = None
-    for m in re.finditer(
-        r'<｜tool▁call▁begin｜>\s*(\S+?)\s*<｜tool▁sep｜>\s*',
-        answer
-    ):
-        func_name = m.group(1).strip()
-        if func_name not in tool_names:
-            continue
-        json_str = _extractBalancedJson(answer, m.end())
-        if json_str is None:
-            continue
-        try:
-            arguments = json.loads(json_str)
-            if start_pos is None:
-                # Check for section begin marker before the call marker
-                section = answer.rfind('<｜tool▁calls▁begin｜>', 0, m.start())
-                start_pos = section if section != -1 else m.start()
-            matches.append({
-                "type": "function",
-                "function": {
-                    "name": func_name,
-                    "arguments": arguments
-                }
-            })
-        except json.JSONDecodeError:
-            pass
-    return matches, start_pos
-
-
-def _parseGlmToolCalls(answer: str, tool_names: list[str]):
-    """Parse GLM-style tool calls using arg_key/arg_value XML pairs.
-
-    Format:
-        function_name
-        key1
-        value1
-        
-    """
-    matches = []
-    start_pos = None
-    for tc_match in re.finditer(r'\s*(.*?)\s*', answer, re.DOTALL):
-        tc_content = tc_match.group(1)
-        # First non-tag text is the function name
-        name_match = re.match(r'([^<\s]+)', tc_content.strip())
-        if not name_match:
-            continue
-        func_name = name_match.group(1).strip()
-        if func_name not in tool_names:
-            continue
-        # Extract arg_key/arg_value pairs
-        keys = [k.group(1).strip() for k in re.finditer(r'\s*(.*?)\s*', tc_content, re.DOTALL)]
-        vals = [v.group(1).strip() for v in re.finditer(r'\s*(.*?)\s*', tc_content, re.DOTALL)]
-        if len(keys) != len(vals):
-            continue
-        arguments = {}
-        for k, v in zip(keys, vals):
-            try:
-                v = json.loads(v)
-            except (json.JSONDecodeError, ValueError):
-                pass  # keep as string
-            arguments[k] = v
-        if start_pos is None:
-            start_pos = tc_match.start()
-        matches.append({
-            "type": "function",
-            "function": {
-                "name": func_name,
-                "arguments": arguments
-            }
-        })
-    return matches, start_pos
-
-
-def _parsePythonicToolCalls(answer: str, tool_names: list[str]):
-    """Parse pythonic-style tool calls used by Llama 4 and similar models.
-
-    Format:
-        [func_name(param1="value1", param2="value2"), func_name2(...)]
-    """
-    matches = []
-    start_pos = None
-    # Match a bracketed list of function calls
-    bracket_match = re.search(r'\[([^\[\]]+)\]', answer)
-    if not bracket_match:
-        return matches, start_pos
-
-    inner = bracket_match.group(1)
-
-    # Build pattern for known tool names
-    escaped_names = [re.escape(name) for name in tool_names]
-    name_pattern = '|'.join(escaped_names)
-
-    for call_match in re.finditer(
-        r'(' + name_pattern + r')\(([^)]*)\)',
-        inner
-    ):
-        func_name = call_match.group(1)
-        params_str = call_match.group(2).strip()
-        arguments = {}
-
-        if params_str:
-            # Parse key="value" pairs, handling commas inside quoted values
-            for param_match in re.finditer(
-                r'(\w+)\s*=\s*("(?:[^"\\]|\\.)*"|\'(?:[^\'\\]|\\.)*\'|[^,\)]+)',
-                params_str
-            ):
-                param_name = param_match.group(1)
-                param_value = param_match.group(2).strip()
-                # Strip surrounding quotes
-                if (param_value.startswith('"') and param_value.endswith('"')) or \
-                   (param_value.startswith("'") and param_value.endswith("'")):
-                    param_value = param_value[1:-1]
-                # Try to parse as JSON for numeric/bool/null values
-                try:
-                    param_value = json.loads(param_value)
-                except (json.JSONDecodeError, ValueError):
-                    pass
-                arguments[param_name] = param_value
-
-        if start_pos is None:
-            start_pos = bracket_match.start()
-        matches.append({
-            "type": "function",
-            "function": {
-                "name": func_name,
-                "arguments": arguments
-            }
-        })
-
-    return matches, start_pos
-
-
-def parseToolCall(answer: str, tool_names: list[str], return_prefix: bool = False):
-    matches = []
-    start_pos = None
-
-    def _return(matches, start_pos):
-        if return_prefix:
-            prefix = answer[:start_pos] if matches and start_pos is not None else ''
-            return matches, prefix
-        return matches
-
-    # abort on very short answers to save computation cycles
-    if len(answer) < 10:
-        return _return(matches, start_pos)
-
-    # Check for DeepSeek-style tool calls (fullwidth Unicode token delimiters)
-    matches, start_pos = _parseDeepSeekToolCalls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for Kimi-K2-style tool calls (pipe-delimited tokens)
-    matches, start_pos = _parseKimiToolCalls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for channel-based tool calls (e.g. GPT-OSS format)
-    matches, start_pos = _parseChannelToolCalls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for MiniMax-style tool calls (invoke/parameter XML tags)
-    matches, start_pos = _parseMiniMaxToolCalls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for GLM-style tool calls (arg_key/arg_value XML pairs)
-    matches, start_pos = _parseGlmToolCalls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for XML-parameter style tool calls (e.g. Qwen3.5 format)
-    matches, start_pos = _parseXmlParamToolCalls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for Mistral/Devstral-style tool calls ([TOOL_CALLS]name[ARGS]json)
-    matches, start_pos = _parseMistralTokenToolCalls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for bare function-name style tool calls (e.g. Mistral format)
-    matches, start_pos = _parseBareNameToolCalls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for pythonic-style tool calls (e.g. Llama 4 format)
-    matches, start_pos = _parsePythonicToolCalls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Define the regex pattern to find the JSON content wrapped in , , , and other tags observed from various models
-    patterns = [r"(```[^\n]*)\n(.*?)```", r"<([^>]+)>(.*?)"]
-
-    for pattern in patterns:
-        for match in re.finditer(pattern, answer, re.DOTALL):
-            # print(match.group(2))
-            if match.group(2) is None:
-                continue
-            # remove backtick wraps if present
-            candidate = re.sub(r"^```(json|xml|python[^\n]*)\n", "", match.group(2).strip())
-            candidate = re.sub(r"```$", "", candidate.strip())
-            # unwrap inner tags
-            candidate = re.sub(pattern, r"\2", candidate.strip(), flags=re.DOTALL)
-            # llm might have generated multiple json objects separated by linebreaks, check for this pattern and try parsing each object individually
-            if re.search(r"\}\s*\n\s*\{", candidate) is not None:
-                candidate = re.sub(r"\}\s*\n\s*\{", "},\n{", candidate)
-            if not candidate.strip().startswith("["):
-                candidate = "[" + candidate + "]"
-
-            candidates = []
-            try:
-                # parse the candidate JSON into a dictionary
-                candidates = json.loads(candidate)
-                if not isinstance(candidates, list):
-                    candidates = [candidates]
-            except json.JSONDecodeError:
-                # Ignore invalid JSON silently
-                continue
-
-            for candidate_dict in candidates:
-                checked_candidate = checkAndSanitizeToolCallCandidate(candidate_dict, tool_names)
-                if checked_candidate is not None:
-                    if start_pos is None:
-                        start_pos = match.start()
-                    matches.append(checked_candidate)
-
-        # last resort if nothing has been mapped: LLM might have produced plain json tool call without xml-like tags
-        if len(matches) == 0:
-            try:
-                candidate = answer
-                # llm might have generated multiple json objects separated by linebreaks, check for this pattern and try parsing each object individually
-                if re.search(r"\}\s*\n\s*\{", candidate) is not None:
-                    candidate = re.sub(r"\}\s*\n\s*\{", "},\n{", candidate)
-                if not candidate.strip().startswith("["):
-                    candidate = "[" + candidate + "]"
-                # parse the candidate JSON into a dictionary
-                candidates = json.loads(candidate)
-                if not isinstance(candidates, list):
-                    candidates = [candidates]
-                for candidate_dict in candidates:
-                    checked_candidate = checkAndSanitizeToolCallCandidate(candidate_dict, tool_names)
-                    if checked_candidate is not None:
-                        matches.append(checked_candidate)
-            except json.JSONDecodeError:
-                # Ignore invalid JSON silently
-                pass
-
-    return _return(matches, start_pos)
diff --git a/modules/chat.py b/modules/chat.py
index 87e528512b..02ae46e488 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -239,6 +239,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
         name1=state['name1'],
         name2=state['name2'],
         user_bio=replace_character_names(state['user_bio'], state['name1'], state['name2']),
+        tools=state['tools'] if 'tools' in state else None,
     )
 
     messages = []
@@ -1186,14 +1187,10 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
 
     # Load tools if any are selected
     selected = state.get('selected_tools', [])
-    parseToolCall = None
+    parse_tool_call = None
     if selected:
         from modules.tool_use import load_tools, execute_tool
-        try:
-            from extensions.openai.utils import parseToolCall, getToolCallId
-        except ImportError:
-            logger.warning('Tool calling requires the openai extension for parseToolCall. Disabling tools.')
-            selected = []
+        from modules.tool_parsing import parse_tool_call, get_tool_call_id
 
     if selected:
         tool_defs, tool_executors = load_tools(selected)
@@ -1253,7 +1250,7 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
                 last_save_time = current_time
 
             # Early stop on tool call detection
-            if tool_func_names and parseToolCall(history['internal'][-1][1], tool_func_names):
+            if tool_func_names and parse_tool_call(history['internal'][-1][1], tool_func_names):
                 break
 
         # Save the model's visible output before re-applying visible_prefix,
@@ -1285,7 +1282,7 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
             break
 
         answer = history['internal'][-1][1]
-        parsed_calls, content_prefix = parseToolCall(answer, tool_func_names, return_prefix=True) if answer else (None, '')
+        parsed_calls, content_prefix = parse_tool_call(answer, tool_func_names, return_prefix=True) if answer else (None, '')
 
         if not parsed_calls:
             break  # No tool calls — done
@@ -1302,7 +1299,7 @@ def _render():
         serialized = []
         tc_headers = []
         for tc in parsed_calls:
-            tc['id'] = getToolCallId()
+            tc['id'] = get_tool_call_id()
             fn_name = tc['function']['name']
             fn_args = tc['function'].get('arguments', {})
 
@@ -1343,7 +1340,7 @@ def _render():
 
         # Preserve thinking block and intermediate text from this turn.
         # content_prefix is the raw text before tool call syntax (returned
-        # by parseToolCall); HTML-escape it and extract thinking to get
+        # by parse_tool_call); HTML-escape it and extract thinking to get
         # the content the user should see.
         content_text = html.escape(content_prefix)
         thinking_content, intermediate = extract_thinking_block(content_text)
diff --git a/modules/tool_parsing.py b/modules/tool_parsing.py
new file mode 100644
index 0000000000..460188d33c
--- /dev/null
+++ b/modules/tool_parsing.py
@@ -0,0 +1,553 @@
+import json
+import random
+import re
+
+
+def get_tool_call_id() -> str:
+    letter_bytes = "abcdefghijklmnopqrstuvwxyz0123456789"
+    b = [random.choice(letter_bytes) for _ in range(8)]
+    return "call_" + "".join(b).lower()
+
+
+def check_and_sanitize_tool_call_candidate(candidate_dict: dict, tool_names: list[str]):
+    # check if property 'function' exists and is a dictionary, otherwise adapt dict
+    if 'function' not in candidate_dict and 'name' in candidate_dict and isinstance(candidate_dict['name'], str):
+        candidate_dict = {"type": "function", "function": candidate_dict}
+    if 'function' in candidate_dict and isinstance(candidate_dict['function'], str):
+        candidate_dict['name'] = candidate_dict['function']
+        del candidate_dict['function']
+        candidate_dict = {"type": "function", "function": candidate_dict}
+    if 'function' in candidate_dict and isinstance(candidate_dict['function'], dict):
+        # check if 'name' exists within 'function' and is part of known tools
+        if 'name' in candidate_dict['function'] and candidate_dict['function']['name'] in tool_names:
+            candidate_dict["type"] = "function"  # ensure required property 'type' exists and has the right value
+            # map property 'parameters' used by some older models to 'arguments'
+            if "arguments" not in candidate_dict["function"] and "parameters" in candidate_dict["function"]:
+                candidate_dict["function"]["arguments"] = candidate_dict["function"]["parameters"]
+                del candidate_dict["function"]["parameters"]
+            return candidate_dict
+    return None
+
+
+def _extract_balanced_json(text: str, start: int) -> str | None:
+    """Extract a balanced JSON object from text starting at the given position.
+
+    Walks through the string tracking brace depth and string boundaries
+    to correctly handle arbitrary nesting levels.
+    """
+    if start >= len(text) or text[start] != '{':
+        return None
+    depth = 0
+    in_string = False
+    escape_next = False
+    for i in range(start, len(text)):
+        c = text[i]
+        if escape_next:
+            escape_next = False
+            continue
+        if c == '\\' and in_string:
+            escape_next = True
+            continue
+        if c == '"':
+            in_string = not in_string
+            continue
+        if in_string:
+            continue
+        if c == '{':
+            depth += 1
+        elif c == '}':
+            depth -= 1
+            if depth == 0:
+                return text[start:i + 1]
+    return None
+
+
+def _parse_channel_tool_calls(answer: str, tool_names: list[str]):
+    """Parse channel-based tool calls used by GPT-OSS and similar models.
+
+    Format:
+        <|start|>assistant to=functions.func_name<|channel|>commentary json<|message|>{"arg": "value"}
+    or:
+        <|channel|>commentary to=functions.func_name <|constrain|>json<|message|>{"arg": "value"}
+    """
+    matches = []
+    start_pos = None
+    # Pattern 1: to=functions.NAME before <|channel|> (GPT-OSS primary format)
+    # Pattern 2: to=functions.NAME after <|channel|> (alternative format)
+    patterns = [
+        r'to=functions\.([^<\s]+)\s*<\|channel\|>[^<]*<\|message\|>',
+        r'<\|channel\|>\w+ to=functions\.([^<\s]+).*?<\|message\|>',
+    ]
+    for pattern in patterns:
+        for m in re.finditer(pattern, answer):
+            func_name = m.group(1).strip()
+            if func_name not in tool_names:
+                continue
+            json_str = _extract_balanced_json(answer, m.end())
+            if json_str is None:
+                continue
+            try:
+                arguments = json.loads(json_str)
+                if start_pos is None:
+                    prefix = answer.rfind('<|start|>assistant', 0, m.start())
+                    start_pos = prefix if prefix != -1 else m.start()
+                matches.append({
+                    "type": "function",
+                    "function": {
+                        "name": func_name,
+                        "arguments": arguments
+                    }
+                })
+            except json.JSONDecodeError:
+                pass
+        if matches:
+            break
+    return matches, start_pos
+
+
+def _parse_mistral_token_tool_calls(answer: str, tool_names: list[str]):
+    """Parse Mistral/Devstral-style tool calls with [TOOL_CALLS] and [ARGS] special tokens.
+
+    Format:
+        [TOOL_CALLS]func_name[ARGS]{"arg": "value"}
+    """
+    matches = []
+    start_pos = None
+    for m in re.finditer(
+        r'\[TOOL_CALLS\]\s*(\S+?)\s*\[ARGS\]\s*',
+        answer
+    ):
+        func_name = m.group(1).strip()
+        if func_name not in tool_names:
+            continue
+        json_str = _extract_balanced_json(answer, m.end())
+        if json_str is None:
+            continue
+        try:
+            arguments = json.loads(json_str)
+            if start_pos is None:
+                start_pos = m.start()
+            matches.append({
+                "type": "function",
+                "function": {
+                    "name": func_name,
+                    "arguments": arguments
+                }
+            })
+        except json.JSONDecodeError:
+            pass
+    return matches, start_pos
+
+
+def _parse_bare_name_tool_calls(answer: str, tool_names: list[str]):
+    """Parse bare function-name style tool calls used by Mistral and similar models.
+
+    Format:
+        functionName{"arg": "value"}
+    Multiple calls are concatenated directly or separated by whitespace.
+    """
+    matches = []
+    start_pos = None
+    # Match tool name followed by opening brace, then extract balanced JSON
+    escaped_names = [re.escape(name) for name in tool_names]
+    pattern = r'(?:' + '|'.join(escaped_names) + r')\s*\{'
+    for match in re.finditer(pattern, answer):
+        text = match.group(0)
+        name = None
+        for n in tool_names:
+            if text.startswith(n):
+                name = n
+                break
+        if not name:
+            continue
+        brace_start = match.end() - 1
+        json_str = _extract_balanced_json(answer, brace_start)
+        if json_str is None:
+            continue
+        try:
+            arguments = json.loads(json_str)
+            if start_pos is None:
+                start_pos = match.start()
+            matches.append({
+                "type": "function",
+                "function": {
+                    "name": name,
+                    "arguments": arguments
+                }
+            })
+        except json.JSONDecodeError:
+            pass
+    return matches, start_pos
+
+
+def _parse_xml_param_tool_calls(answer: str, tool_names: list[str]):
+    """Parse XML-parameter style tool calls used by Qwen3.5 and similar models.
+
+    Format:
+        
+        
+        value
+        
+        
+    """
+    matches = []
+    start_pos = None
+    for tc_match in re.finditer(r'\s*(.*?)\s*', answer, re.DOTALL):
+        tc_content = tc_match.group(1)
+        func_match = re.search(r']+)>', tc_content)
+        if not func_match:
+            continue
+        func_name = func_match.group(1).strip()
+        if func_name not in tool_names:
+            continue
+        arguments = {}
+        for param_match in re.finditer(r']+)>\s*(.*?)\s*', tc_content, re.DOTALL):
+            param_name = param_match.group(1).strip()
+            param_value = param_match.group(2).strip()
+            try:
+                param_value = json.loads(param_value)
+            except (json.JSONDecodeError, ValueError):
+                pass  # keep as string
+            arguments[param_name] = param_value
+        if start_pos is None:
+            start_pos = tc_match.start()
+        matches.append({
+            "type": "function",
+            "function": {
+                "name": func_name,
+                "arguments": arguments
+            }
+        })
+    return matches, start_pos
+
+
+def _parse_kimi_tool_calls(answer: str, tool_names: list[str]):
+    """Parse Kimi-K2-style tool calls using pipe-delimited tokens.
+
+    Format:
+        <|tool_calls_section_begin|>
+        <|tool_call_begin|>functions.func_name:index<|tool_call_argument_begin|>{"arg": "value"}<|tool_call_end|>
+        <|tool_calls_section_end|>
+    """
+    matches = []
+    start_pos = None
+    for m in re.finditer(
+        r'<\|tool_call_begin\|>\s*(?:functions\.)?(\S+?)(?::\d+)?\s*<\|tool_call_argument_begin\|>\s*',
+        answer
+    ):
+        func_name = m.group(1).strip()
+        if func_name not in tool_names:
+            continue
+        json_str = _extract_balanced_json(answer, m.end())
+        if json_str is None:
+            continue
+        try:
+            arguments = json.loads(json_str)
+            if start_pos is None:
+                # Check for section begin marker before the call marker
+                section = answer.rfind('<|tool_calls_section_begin|>', 0, m.start())
+                start_pos = section if section != -1 else m.start()
+            matches.append({
+                "type": "function",
+                "function": {
+                    "name": func_name,
+                    "arguments": arguments
+                }
+            })
+        except json.JSONDecodeError:
+            pass
+    return matches, start_pos
+
+
+def _parse_minimax_tool_calls(answer: str, tool_names: list[str]):
+    """Parse MiniMax-style tool calls using invoke/parameter XML tags.
+
+    Format:
+        
+        
+        value
+        
+        
+    """
+    matches = []
+    start_pos = None
+    for tc_match in re.finditer(r'\s*(.*?)\s*', answer, re.DOTALL):
+        tc_content = tc_match.group(1)
+        # Split on  to handle multiple parallel calls in one block
+        for invoke_match in re.finditer(r'(.*?)', tc_content, re.DOTALL):
+            func_name = invoke_match.group(1).strip()
+            if func_name not in tool_names:
+                continue
+            invoke_body = invoke_match.group(2)
+            arguments = {}
+            for param_match in re.finditer(r'\s*(.*?)\s*', invoke_body, re.DOTALL):
+                param_name = param_match.group(1).strip()
+                param_value = param_match.group(2).strip()
+                try:
+                    param_value = json.loads(param_value)
+                except (json.JSONDecodeError, ValueError):
+                    pass  # keep as string
+                arguments[param_name] = param_value
+            if start_pos is None:
+                start_pos = tc_match.start()
+            matches.append({
+                "type": "function",
+                "function": {
+                    "name": func_name,
+                    "arguments": arguments
+                }
+            })
+    return matches, start_pos
+
+
+def _parse_deep_seek_tool_calls(answer: str, tool_names: list[str]):
+    """Parse DeepSeek-style tool calls using fullwidth Unicode token delimiters.
+
+    Format:
+        <｜tool▁calls▁begin｜><｜tool▁call▁begin｜>func_name<｜tool▁sep｜>{"arg": "value"}<｜tool▁call▁end｜><｜tool▁calls▁end｜>
+    """
+    matches = []
+    start_pos = None
+    for m in re.finditer(
+        r'<｜tool▁call▁begin｜>\s*(\S+?)\s*<｜tool▁sep｜>\s*',
+        answer
+    ):
+        func_name = m.group(1).strip()
+        if func_name not in tool_names:
+            continue
+        json_str = _extract_balanced_json(answer, m.end())
+        if json_str is None:
+            continue
+        try:
+            arguments = json.loads(json_str)
+            if start_pos is None:
+                # Check for section begin marker before the call marker
+                section = answer.rfind('<｜tool▁calls▁begin｜>', 0, m.start())
+                start_pos = section if section != -1 else m.start()
+            matches.append({
+                "type": "function",
+                "function": {
+                    "name": func_name,
+                    "arguments": arguments
+                }
+            })
+        except json.JSONDecodeError:
+            pass
+    return matches, start_pos
+
+
+def _parse_glm_tool_calls(answer: str, tool_names: list[str]):
+    """Parse GLM-style tool calls using arg_key/arg_value XML pairs.
+
+    Format:
+        function_name
+        key1
+        value1
+        
+    """
+    matches = []
+    start_pos = None
+    for tc_match in re.finditer(r'\s*(.*?)\s*', answer, re.DOTALL):
+        tc_content = tc_match.group(1)
+        # First non-tag text is the function name
+        name_match = re.match(r'([^<\s]+)', tc_content.strip())
+        if not name_match:
+            continue
+        func_name = name_match.group(1).strip()
+        if func_name not in tool_names:
+            continue
+        # Extract arg_key/arg_value pairs
+        keys = [k.group(1).strip() for k in re.finditer(r'\s*(.*?)\s*', tc_content, re.DOTALL)]
+        vals = [v.group(1).strip() for v in re.finditer(r'\s*(.*?)\s*', tc_content, re.DOTALL)]
+        if len(keys) != len(vals):
+            continue
+        arguments = {}
+        for k, v in zip(keys, vals):
+            try:
+                v = json.loads(v)
+            except (json.JSONDecodeError, ValueError):
+                pass  # keep as string
+            arguments[k] = v
+        if start_pos is None:
+            start_pos = tc_match.start()
+        matches.append({
+            "type": "function",
+            "function": {
+                "name": func_name,
+                "arguments": arguments
+            }
+        })
+    return matches, start_pos
+
+
+def _parse_pythonic_tool_calls(answer: str, tool_names: list[str]):
+    """Parse pythonic-style tool calls used by Llama 4 and similar models.
+
+    Format:
+        [func_name(param1="value1", param2="value2"), func_name2(...)]
+    """
+    matches = []
+    start_pos = None
+    # Match a bracketed list of function calls
+    bracket_match = re.search(r'\[([^\[\]]+)\]', answer)
+    if not bracket_match:
+        return matches, start_pos
+
+    inner = bracket_match.group(1)
+
+    # Build pattern for known tool names
+    escaped_names = [re.escape(name) for name in tool_names]
+    name_pattern = '|'.join(escaped_names)
+
+    for call_match in re.finditer(
+        r'(' + name_pattern + r')\(([^)]*)\)',
+        inner
+    ):
+        func_name = call_match.group(1)
+        params_str = call_match.group(2).strip()
+        arguments = {}
+
+        if params_str:
+            # Parse key="value" pairs, handling commas inside quoted values
+            for param_match in re.finditer(
+                r'(\w+)\s*=\s*("(?:[^"\\]|\\.)*"|\'(?:[^\'\\]|\\.)*\'|[^,\)]+)',
+                params_str
+            ):
+                param_name = param_match.group(1)
+                param_value = param_match.group(2).strip()
+                # Strip surrounding quotes
+                if (param_value.startswith('"') and param_value.endswith('"')) or \
+                   (param_value.startswith("'") and param_value.endswith("'")):
+                    param_value = param_value[1:-1]
+                # Try to parse as JSON for numeric/bool/null values
+                try:
+                    param_value = json.loads(param_value)
+                except (json.JSONDecodeError, ValueError):
+                    pass
+                arguments[param_name] = param_value
+
+        if start_pos is None:
+            start_pos = bracket_match.start()
+        matches.append({
+            "type": "function",
+            "function": {
+                "name": func_name,
+                "arguments": arguments
+            }
+        })
+
+    return matches, start_pos
+
+
+def parse_tool_call(answer: str, tool_names: list[str], return_prefix: bool = False):
+    matches = []
+    start_pos = None
+
+    def _return(matches, start_pos):
+        if return_prefix:
+            prefix = answer[:start_pos] if matches and start_pos is not None else ''
+            return matches, prefix
+        return matches
+
+    # Check for DeepSeek-style tool calls (fullwidth Unicode token delimiters)
+    matches, start_pos = _parse_deep_seek_tool_calls(answer, tool_names)
+    if matches:
+        return _return(matches, start_pos)
+
+    # Check for Kimi-K2-style tool calls (pipe-delimited tokens)
+    matches, start_pos = _parse_kimi_tool_calls(answer, tool_names)
+    if matches:
+        return _return(matches, start_pos)
+
+    # Check for channel-based tool calls (e.g. GPT-OSS format)
+    matches, start_pos = _parse_channel_tool_calls(answer, tool_names)
+    if matches:
+        return _return(matches, start_pos)
+
+    # Check for MiniMax-style tool calls (invoke/parameter XML tags)
+    matches, start_pos = _parse_minimax_tool_calls(answer, tool_names)
+    if matches:
+        return _return(matches, start_pos)
+
+    # Check for GLM-style tool calls (arg_key/arg_value XML pairs)
+    matches, start_pos = _parse_glm_tool_calls(answer, tool_names)
+    if matches:
+        return _return(matches, start_pos)
+
+    # Check for XML-parameter style tool calls (e.g. Qwen3.5 format)
+    matches, start_pos = _parse_xml_param_tool_calls(answer, tool_names)
+    if matches:
+        return _return(matches, start_pos)
+
+    # Check for Mistral/Devstral-style tool calls ([TOOL_CALLS]name[ARGS]json)
+    matches, start_pos = _parse_mistral_token_tool_calls(answer, tool_names)
+    if matches:
+        return _return(matches, start_pos)
+
+    # Check for bare function-name style tool calls (e.g. Mistral format)
+    matches, start_pos = _parse_bare_name_tool_calls(answer, tool_names)
+    if matches:
+        return _return(matches, start_pos)
+
+    # Check for pythonic-style tool calls (e.g. Llama 4 format)
+    matches, start_pos = _parse_pythonic_tool_calls(answer, tool_names)
+    if matches:
+        return _return(matches, start_pos)
+
+    # Define the regex pattern to find the JSON content wrapped in , , , and other tags observed from various models
+    patterns = [r"(```[^\n]*)\n(.*?)```", r"<([^>]+)>(.*?)"]
+
+    for pattern in patterns:
+        for match in re.finditer(pattern, answer, re.DOTALL):
+            if match.group(2) is None:
+                continue
+            # remove backtick wraps if present
+            candidate = re.sub(r"^```(json|xml|python[^\n]*)\n", "", match.group(2).strip())
+            candidate = re.sub(r"```$", "", candidate.strip())
+            # unwrap inner tags
+            candidate = re.sub(pattern, r"\2", candidate.strip(), flags=re.DOTALL)
+            # llm might have generated multiple json objects separated by linebreaks, check for this pattern and try parsing each object individually
+            if re.search(r"\}\s*\n\s*\{", candidate) is not None:
+                candidate = re.sub(r"\}\s*\n\s*\{", "},\n{", candidate)
+            if not candidate.strip().startswith("["):
+                candidate = "[" + candidate + "]"
+
+            candidates = []
+            try:
+                # parse the candidate JSON into a dictionary
+                candidates = json.loads(candidate)
+                if not isinstance(candidates, list):
+                    candidates = [candidates]
+            except json.JSONDecodeError:
+                # Ignore invalid JSON silently
+                continue
+
+            for candidate_dict in candidates:
+                checked_candidate = check_and_sanitize_tool_call_candidate(candidate_dict, tool_names)
+                if checked_candidate is not None:
+                    if start_pos is None:
+                        start_pos = match.start()
+                    matches.append(checked_candidate)
+
+        # last resort if nothing has been mapped: LLM might have produced plain json tool call without xml-like tags
+        if len(matches) == 0:
+            try:
+                candidate = answer
+                # llm might have generated multiple json objects separated by linebreaks, check for this pattern and try parsing each object individually
+                if re.search(r"\}\s*\n\s*\{", candidate) is not None:
+                    candidate = re.sub(r"\}\s*\n\s*\{", "},\n{", candidate)
+                if not candidate.strip().startswith("["):
+                    candidate = "[" + candidate + "]"
+                # parse the candidate JSON into a dictionary
+                candidates = json.loads(candidate)
+                if not isinstance(candidates, list):
+                    candidates = [candidates]
+                for candidate_dict in candidates:
+                    checked_candidate = check_and_sanitize_tool_call_candidate(candidate_dict, tool_names)
+                    if checked_candidate is not None:
+                        matches.append(checked_candidate)
+            except json.JSONDecodeError:
+                # Ignore invalid JSON silently
+                pass
+
+    return _return(matches, start_pos)
diff --git a/modules/tool_use.py b/modules/tool_use.py
index 5542485375..e22b17983b 100644
--- a/modules/tool_use.py
+++ b/modules/tool_use.py
@@ -3,7 +3,7 @@
 
 from modules import shared
 from modules.logging_colors import logger
-from modules.utils import natural_keys
+from modules.utils import natural_keys, sanitize_filename
 
 
 def get_available_tools():
@@ -23,6 +23,10 @@ def load_tools(selected_names):
     tool_defs = []
     executors = {}
     for name in selected_names:
+        name = sanitize_filename(name)
+        if not name:
+            continue
+
         path = shared.user_data_dir / 'tools' / f'{name}.py'
         if not path.exists():
             continue
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index ce9fc0a2b7..0acf9c04bd 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -97,7 +97,7 @@ def create_ui():
                 shared.gradio['tools_refresh'].click(fn=lambda: gr.update(choices=get_available_tools()), inputs=[], outputs=[shared.gradio['selected_tools']])
 
                 def sync_web_tools(selected):
-                    if 'web_search' in selected and 'fetch_webpage' not in selected:
+                    if 'web_search' in selected and 'fetch_webpage' not in selected and 'fetch_webpage' in get_available_tools():
                         selected.append('fetch_webpage')
 
                     return gr.update(value=selected)
diff --git a/modules/web_search.py b/modules/web_search.py
index 754dd11186..216d7933bf 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -1,11 +1,13 @@
 import concurrent.futures
 import html
+import ipaddress
 import random
 import re
+import socket
 import urllib.request
 from concurrent.futures import as_completed
 from datetime import datetime
-from urllib.parse import quote_plus
+from urllib.parse import quote_plus, urlparse
 
 import requests
 
@@ -13,6 +15,26 @@
 from modules.logging_colors import logger
 
 
+def _validate_url(url):
+    """Validate that a URL is safe to fetch (not targeting private/internal networks)."""
+    parsed = urlparse(url)
+    if parsed.scheme not in ('http', 'https'):
+        raise ValueError(f"Unsupported URL scheme: {parsed.scheme}")
+
+    hostname = parsed.hostname
+    if not hostname:
+        raise ValueError("No hostname in URL")
+
+    # Resolve hostname and check all returned addresses
+    try:
+        for family, _, _, _, sockaddr in socket.getaddrinfo(hostname, None):
+            ip = ipaddress.ip_address(sockaddr[0])
+            if ip.is_private or ip.is_loopback or ip.is_link_local or ip.is_reserved:
+                raise ValueError(f"Access to private/internal address {ip} is blocked")
+    except socket.gaierror:
+        raise ValueError(f"Could not resolve hostname: {hostname}")
+
+
 def get_current_timestamp():
     """Returns the current time in 24-hour format"""
     return datetime.now().strftime('%b %d, %Y %H:%M')
@@ -25,11 +47,20 @@ def download_web_page(url, timeout=10, include_links=False):
     import html2text
 
     try:
+        _validate_url(url)
         headers = {
             'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'
         }
-        response = requests.get(url, headers=headers, timeout=timeout)
-        response.raise_for_status()  # Raise an exception for bad status codes
+        max_redirects = 5
+        for _ in range(max_redirects):
+            response = requests.get(url, headers=headers, timeout=timeout, allow_redirects=False)
+            if response.is_redirect and 'Location' in response.headers:
+                url = response.headers['Location']
+                _validate_url(url)
+            else:
+                break
+
+        response.raise_for_status()
 
         # Initialize the HTML to Markdown converter
         h = html2text.HTML2Text()

From 16636c04b88df924b1af0a79dfe5cd574aa33753 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 19:06:04 -0700
Subject: [PATCH 1362/1701] UI: Minor fix/optimization

---
 js/main.js | 12 +++++-------
 1 file changed, 5 insertions(+), 7 deletions(-)

diff --git a/js/main.js b/js/main.js
index a8bbbc71e4..5f79c3ec25 100644
--- a/js/main.js
+++ b/js/main.js
@@ -1075,15 +1075,13 @@ document.fonts.addEventListener("loadingdone", (event) => {
     const currentHeight = chatInputRow.offsetHeight;
     const heightDifference = currentHeight - originalHeight;
     chatParent.style.marginBottom = `${originalMarginBottom + heightDifference}px`;
+    if (!window.isScrolled) {
+      chatParent.scrollTop = chatParent.scrollHeight - chatParent.clientHeight;
+    }
   }
 
-  // Watch for changes that might affect height
-  const observer = new MutationObserver(updateMargin);
-  observer.observe(chatInputRow, {
-    childList: true,
-    subtree: true,
-    attributes: true
-  });
+  // Watch for size changes that affect height
+  new ResizeObserver(updateMargin).observe(chatInputRow);
 
   // Also listen for window resize
   window.addEventListener("resize", updateMargin);

From 5f1707af3562c1854068af6669d055bdf02cf038 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 20:38:40 -0700
Subject: [PATCH 1363/1701] UI: Increase the width of non-instruct chat styles

---
 css/chat_style-Dark.css            | 1 +
 css/chat_style-TheEncrypted777.css | 1 +
 css/chat_style-cai-chat-square.css | 1 +
 css/chat_style-cai-chat.css        | 1 +
 css/chat_style-messenger.css       | 1 +
 css/chat_style-wpp.css             | 1 +
 css/main.css                       | 1 -
 7 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/css/chat_style-Dark.css b/css/chat_style-Dark.css
index 6a4784ccd1..28d77a4b54 100644
--- a/css/chat_style-Dark.css
+++ b/css/chat_style-Dark.css
@@ -2,6 +2,7 @@
     display: grid;
     align-items: start;
     grid-template-columns: 60px minmax(0, 1fr);
+    width: min(100%, calc(768px + 60px));
     padding-bottom: 22px;
     padding-top: 6px;
     font-size: 18px;
diff --git a/css/chat_style-TheEncrypted777.css b/css/chat_style-TheEncrypted777.css
index fbd47072fb..21156ee56b 100644
--- a/css/chat_style-TheEncrypted777.css
+++ b/css/chat_style-TheEncrypted777.css
@@ -4,6 +4,7 @@
     display: grid;
     align-items: start;
     grid-template-columns: 60px minmax(0, 1fr);
+    width: min(100%, calc(768px + 60px + 90px));
     padding-bottom: 21px;
     padding-top: 7px;
     font-size: 18px;
diff --git a/css/chat_style-cai-chat-square.css b/css/chat_style-cai-chat-square.css
index 291a120965..0d9467dfe7 100644
--- a/css/chat_style-cai-chat-square.css
+++ b/css/chat_style-cai-chat-square.css
@@ -19,4 +19,5 @@
     padding-bottom: 1.5em;
     padding-top: 0.5em;
     grid-template-columns: 70px minmax(0, 1fr);
+    width: min(100%, calc(768px + 70px));
 }
diff --git a/css/chat_style-cai-chat.css b/css/chat_style-cai-chat.css
index b06b1269c7..6de32597a9 100644
--- a/css/chat_style-cai-chat.css
+++ b/css/chat_style-cai-chat.css
@@ -2,6 +2,7 @@
     display: grid;
     align-items: start;
     grid-template-columns: 60px minmax(0, 1fr);
+    width: min(100%, calc(768px + 60px));
     padding-bottom: 1.5em;
     padding-top: 0.5em;
     font-size: 15px;
diff --git a/css/chat_style-messenger.css b/css/chat_style-messenger.css
index 70fd6d4a28..85178759ed 100644
--- a/css/chat_style-messenger.css
+++ b/css/chat_style-messenger.css
@@ -1,4 +1,5 @@
 .message {
+    width: min(100%, calc(48rem + 60px));
     padding-bottom: 22px;
     padding-top: 3px;
     font-size: 15px;
diff --git a/css/chat_style-wpp.css b/css/chat_style-wpp.css
index b2ac4d2a38..5c14fa8030 100644
--- a/css/chat_style-wpp.css
+++ b/css/chat_style-wpp.css
@@ -1,5 +1,6 @@
 .message {
     display: block;
+    width: min(100%, 48rem);
     padding-top: 0;
     padding-bottom: 21px;
     font-size: 15px;
diff --git a/css/main.css b/css/main.css
index 7cc496a70d..5a58c4a3de 100644
--- a/css/main.css
+++ b/css/main.css
@@ -400,7 +400,6 @@ audio {
 }
 
 .chat .message {
-    width: min(100%, 48rem);
     margin-left: auto;
     margin-right: auto;
     text-align: start;

From 998b9bfb2a6df1e0bd37bd4dfcdbd10bf4a38977 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 13 Mar 2026 21:05:31 -0700
Subject: [PATCH 1364/1701] UI: Make all chat styles better match instruct
 style

---
 css/chat_style-Dark.css            |  5 +----
 css/chat_style-TheEncrypted777.css |  6 ++----
 css/chat_style-cai-chat.css        |  8 +-------
 css/chat_style-messenger.css       | 13 +++++--------
 css/chat_style-wpp.css             | 11 +----------
 css/main.css                       | 22 +++++++++++++++++++---
 6 files changed, 29 insertions(+), 36 deletions(-)

diff --git a/css/chat_style-Dark.css b/css/chat_style-Dark.css
index 28d77a4b54..02beb93566 100644
--- a/css/chat_style-Dark.css
+++ b/css/chat_style-Dark.css
@@ -92,9 +92,6 @@
 }
 
 .message-body p {
-    margin-bottom: 0 !important;
-    font-size: 16px !important;
-    line-height: 1.5 !important;
     color: #e0e0e0 !important; /* Light color for text */
 }
 
@@ -123,7 +120,7 @@
     }
 
     .message-body p {
-        font-size: 14px !important; /* Smaller text for mobile */
+        font-size: 14px !important;
     }
 
     .username {
diff --git a/css/chat_style-TheEncrypted777.css b/css/chat_style-TheEncrypted777.css
index 21156ee56b..b3df671034 100644
--- a/css/chat_style-TheEncrypted777.css
+++ b/css/chat_style-TheEncrypted777.css
@@ -87,10 +87,8 @@
     border-radius: 20px;
 }
 
-.message-body p {
-    margin-bottom: 0 !important;
+.message-body p, .message-body li {
     font-size: 18px !important;
-    line-height: 1.428571429 !important;
     color: rgb(243 244 246) !important;
     text-shadow: 2px 2px 2px rgb(0 0 0);
     font-weight: 500;
@@ -128,7 +126,7 @@
         padding-left: 0;
     }
 
-    .message-body p {
+    .message-body p, .message-body li {
         font-size: 16px !important;
     }
 
diff --git a/css/chat_style-cai-chat.css b/css/chat_style-cai-chat.css
index 6de32597a9..9cc4d4cd6c 100644
--- a/css/chat_style-cai-chat.css
+++ b/css/chat_style-cai-chat.css
@@ -47,16 +47,10 @@
     border-radius: 20px;
 }
 
-.message-body p {
-    font-size: 15px !important;
-    line-height: 22.5px !important;
+.message-body p, .message-body li {
     font-weight: 500;
 }
 
-.message-body p, .chat .message-body ul, .chat .message-body ol {
-    margin-bottom: 10px !important;
-}
-
 .dark .message-body p em {
     color: rgb(138 138 138) !important;
 }
diff --git a/css/chat_style-messenger.css b/css/chat_style-messenger.css
index 85178759ed..438b806087 100644
--- a/css/chat_style-messenger.css
+++ b/css/chat_style-messenger.css
@@ -61,8 +61,10 @@
     text-align: right;
 }
 
-.dark .circle-bot + .text div, .dark .circle-bot + .text * {
-    color: #000;
+.dark .circle-bot + .text div, .dark .circle-bot + .text *,
+.dark .chat .message .circle-bot + .text .message-body :is(h1, h2, h3, h4, h5, h6),
+.dark .chat .message .circle-bot + .text .message-body a {
+    color: #000 !important;
 }
 
 .text {
@@ -77,19 +79,14 @@
     font-weight: bold;
 }
 
-.message-body {
-}
-
 .message-body img {
     max-width: 300px;
     max-height: 300px;
     border-radius: 20px;
 }
 
-.message-body p {
-    margin-bottom: 0 !important;
+.message-body p, .message-body li {
     font-size: 15px !important;
-    line-height: 1.428571429 !important;
     font-weight: 500;
 }
 
diff --git a/css/chat_style-wpp.css b/css/chat_style-wpp.css
index 5c14fa8030..ad6985d251 100644
--- a/css/chat_style-wpp.css
+++ b/css/chat_style-wpp.css
@@ -78,14 +78,8 @@
     border-radius: 12px;
 }
 
-.message-body p {
+.message-body p, .message-body li {
     font-size: 15px !important;
-    line-height: 1.4 !important;
-    font-weight: 400;
-}
-
-.message-body p:first-child {
-    margin-top: 0 !important;
 }
 
 .dark .message-body p em {
@@ -101,6 +95,3 @@
     margin-top: 8px;
 }
 
-.message-body p, .chat .message-body ul, .chat .message-body ol {
-    margin-bottom: 10px !important;
-}
diff --git a/css/main.css b/css/main.css
index 5a58c4a3de..49b8f752d8 100644
--- a/css/main.css
+++ b/css/main.css
@@ -430,7 +430,12 @@ audio {
     font-size: 16px;
 }
 
-.dark .message-body :is(h1, h2, h3, h4, h5, h6) {
+.dark .message-body h1,
+.dark .message-body h2,
+.dark .message-body h3,
+.dark .message-body h4,
+.dark .message-body h5,
+.dark .message-body h6 {
     color: white !important;
 }
 
@@ -830,9 +835,20 @@ audio {
     }
 }
 
-.message-body ol, .message-body ul {
+.message-body p, .message-body li {
+    line-height: 1.75 !important;
+}
+
+.message-body p, .message-body ul, .message-body ol {
+    margin: 1.25em 0 !important;
+}
+
+.message-body :is(p, ul, ol):first-child {
     margin-top: 0 !important;
-    margin-bottom: 1.25em !important;
+}
+
+.message-body :is(p, ul, ol):last-child {
+    margin-bottom: 0 !important;
 }
 
 /* ----------------------------------------------

From accb2ef661838d95521c5fc7fd7660bf541a5064 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 06:16:09 -0700
Subject: [PATCH 1365/1701] UI/API: Prevent tool call markup from leaking into
 streamed UI output (closes #7427)

---
 modules/chat.py         | 11 +++++++++
 modules/tool_parsing.py | 49 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 60 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index 02ae46e488..daecd50b0a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1028,6 +1028,13 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
                 thinking_prefix = start_tag
                 break
 
+    # When tools are active, buffer streaming output during potential tool
+    # call generation to prevent raw markup from leaking into the display.
+    _check_tool_markers = bool(state.get('tools'))
+    if _check_tool_markers:
+        from modules.tool_parsing import streaming_tool_buffer_check
+        _tool_names = [t['function']['name'] for t in state['tools'] if 'function' in t and 'name' in t['function']]
+
     # Generate
     reply = None
     for j, reply in enumerate(generate_reply(prompt, state, stopping_strings=stopping_strings, is_chat=True, for_ui=for_ui)):
@@ -1077,6 +1084,10 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             })
 
         if is_stream:
+            if _check_tool_markers:
+                if streaming_tool_buffer_check(output['internal'][-1][1], _tool_names):
+                    continue
+
             yield output
 
     if _continue:
diff --git a/modules/tool_parsing.py b/modules/tool_parsing.py
index 460188d33c..418503ad4e 100644
--- a/modules/tool_parsing.py
+++ b/modules/tool_parsing.py
@@ -9,6 +9,55 @@ def get_tool_call_id() -> str:
     return "call_" + "".join(b).lower()
 
 
+# Known opening markers for tool calls across model formats.
+# Used during streaming to buffer output that might be tool call markup,
+# preventing raw markup from leaking into displayed/streamed content.
+TOOL_CALL_OPENING_MARKERS = [
+    '',
+    '',
+    '',
+    '<|tool_call_begin|>',
+    '<|tool_calls_section_begin|>',
+    '<｜tool▁call▁begin｜>',
+    '<｜tool▁calls▁begin｜>',
+    '[TOOL_CALLS]',
+    'to=functions.',
+    '<|channel|>commentary',
+]
+
+def streaming_tool_buffer_check(text, tool_names=None):
+    '''
+    Check whether streaming output should be withheld because it may
+    contain tool-call markup.
+    '''
+    # Full marker found → buffer permanently
+    for marker in TOOL_CALL_OPENING_MARKERS:
+        if marker in text:
+            return True
+
+    # Bare function-name style (e.g. Devstral): "get_weather{...}"
+    # Only match tool name followed by '{' to avoid false positives on
+    # common words that happen to be tool names (e.g. "get", "search").
+    if tool_names:
+        for name in tool_names:
+            if name + '{' in text or name + ' {' in text:
+                return True
+            # Partial: text ends with tool name (or prefix of it) but '{' hasn't arrived yet
+            if text.endswith(name):
+                return True
+            for prefix_len in range(min(len(name) - 1, len(text)), 0, -1):
+                if text.endswith(name[:prefix_len]):
+                    return True
+
+    # Tail might be a partial marker forming across tokens
+    for marker in TOOL_CALL_OPENING_MARKERS:
+        for prefix_len in range(min(len(marker) - 1, len(text)), 0, -1):
+            if text.endswith(marker[:prefix_len]):
+                return True
+
+    return False
+
+
 def check_and_sanitize_tool_call_candidate(candidate_dict: dict, tool_names: list[str]):
     # check if property 'function' exists and is a dictionary, otherwise adapt dict
     if 'function' not in candidate_dict and 'name' in candidate_dict and isinstance(candidate_dict['name'], str):

From 09a6549816eca117d779fdf66bad9862d559f288 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 06:52:40 -0700
Subject: [PATCH 1366/1701] API: Stream reasoning_content separately from
 content in OpenAI-compatible responses

---
 extensions/openai/completions.py | 29 +++++++++++++++++++++++------
 modules/reasoning.py             | 16 ++++++++++++++--
 2 files changed, 37 insertions(+), 8 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 27defe4250..514270504f 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -417,7 +417,7 @@ def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, p
         logprob_proc.token_alternatives_history.clear()
     chat_logprobs_offset = [0]  # mutable for closure access in streaming
 
-    def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False):
+    def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False, reasoning_content=None):
         # begin streaming
         delta = {}
         if include_role:
@@ -425,6 +425,8 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
             delta['refusal'] = None
         if content is not None:
             delta['content'] = content
+        if reasoning_content is not None:
+            delta['reasoning_content'] = reasoning_content
         if chunk_tool_calls:
             delta['tool_calls'] = chunk_tool_calls
 
@@ -477,6 +479,7 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
 
     answer = ''
     seen_content = ''
+    seen_reasoning = ''
 
     tool_calls = []
     end_last_tool_call = 0
@@ -508,17 +511,31 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
             break
 
         if stream:
-            len_seen = len(seen_content)
-            new_content = answer[len_seen:]
+            # Strip reasoning/thinking blocks so only final content is streamed.
+            # Reasoning is emitted separately as reasoning_content deltas.
+            reasoning, content = extract_reasoning(answer)
+            if reasoning is not None:
+                new_reasoning = reasoning[len(seen_reasoning):]
+                new_content = content[len(seen_content):]
+            else:
+                new_reasoning = None
+                new_content = answer[len(seen_content):]
 
-            if not new_content or chr(0xfffd) in new_content:  # partial unicode character, don't send it yet.
+            if (not new_content and not new_reasoning) or chr(0xfffd) in (new_content or '') + (new_reasoning or ''):
                 continue
 
-            chunk = chat_streaming_chunk(new_content)
+            chunk = chat_streaming_chunk(
+                content=new_content if new_content else None,
+                reasoning_content=new_reasoning if new_reasoning else None,
+            )
             if include_usage:
                 chunk['usage'] = None
 
-            seen_content = answer
+            if reasoning is not None:
+                seen_reasoning = reasoning
+                seen_content = content
+            else:
+                seen_content = answer
             yield chunk
 
     token_count = shared.model.last_prompt_token_count if hasattr(shared.model, 'last_prompt_token_count') else 0
diff --git a/modules/reasoning.py b/modules/reasoning.py
index 708ee55aed..3a9ab546bb 100644
--- a/modules/reasoning.py
+++ b/modules/reasoning.py
@@ -8,7 +8,7 @@
     ('<|channel|>commentary<|message|>', '<|end|>', '<|start|>assistant<|channel|>final<|message|>'),
     ('', '', None),
     ('<|think|>', '<|end|>', '<|content|>'),  # Solar Open
-    ('Thinking Process:', '', None),  # Qwen3.5 verbose thinking outside tags
+    # ('Thinking Process:', '', None),  # Qwen3.5 verbose thinking outside tags -- removed: too prone to false positives in streaming
     (None, '', None),  # End-only variant (e.g., Qwen3-next)
 ]
 
@@ -42,6 +42,12 @@ def extract_reasoning(text, html_escaped=False):
             start_esc = esc(start_tag)
             start_pos = text.find(start_esc)
             if start_pos == -1:
+                # During streaming, the start tag may be arriving partially.
+                # If the text is a prefix of a start tag, return empty content
+                # to prevent the partial tag from leaking.
+                stripped = text.strip()
+                if stripped and start_esc.startswith(stripped):
+                    return '', ''
                 continue
             thought_start = start_pos + len(start_esc)
             end_pos = text.find(end_esc, thought_start)
@@ -63,7 +69,13 @@ def extract_reasoning(text, html_escaped=False):
             thought_end = end_pos
             if content_esc:
                 content_pos = text.find(content_esc, end_pos)
-                content_start = content_pos + len(content_esc) if content_pos != -1 else end_pos + len(end_esc)
+                if content_pos != -1:
+                    content_start = content_pos + len(content_esc)
+                else:
+                    # Content tag expected but not yet present (e.g. partial
+                    # streaming) — suppress intermediate tags between end_tag
+                    # and content_tag so they don't leak as content.
+                    content_start = len(text)
             else:
                 content_start = end_pos + len(end_esc)
 

From cb08ba63dcd76228a4c070acd5352a3df4d78486 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 09:08:05 -0700
Subject: [PATCH 1367/1701] Fix GPT-OSS channel markup leaking into UI when
 model skips analysis block

---
 modules/chat.py                |  2 ++
 modules/reasoning.py           | 14 ++++++++++++--
 modules/ui_image_generation.py |  2 ++
 3 files changed, 16 insertions(+), 2 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index daecd50b0a..10785c19a3 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -882,6 +882,8 @@ def generate_search_query(user_message, state):
         query = query.rsplit("", 1)[1]
     elif "<|start|>assistant<|channel|>final<|message|>" in query:
         query = query.rsplit("<|start|>assistant<|channel|>final<|message|>", 1)[1]
+    elif "<|channel|>final<|message|>" in query:
+        query = query.rsplit("<|channel|>final<|message|>", 1)[1]
     elif "" in query:
         query = query.rsplit("", 1)[1]
 
diff --git a/modules/reasoning.py b/modules/reasoning.py
index 3a9ab546bb..bc61aab398 100644
--- a/modules/reasoning.py
+++ b/modules/reasoning.py
@@ -4,8 +4,8 @@
 # Use None for start_tag to match from beginning (end-only formats should be listed last)
 THINKING_FORMATS = [
     ('', '', None),
-    ('<|channel|>analysis<|message|>', '<|end|>', '<|start|>assistant<|channel|>final<|message|>'),
-    ('<|channel|>commentary<|message|>', '<|end|>', '<|start|>assistant<|channel|>final<|message|>'),
+    ('<|channel|>analysis<|message|>', '<|end|>', '<|channel|>final<|message|>'),
+    ('<|channel|>commentary<|message|>', '<|end|>', '<|channel|>final<|message|>'),
     ('', '', None),
     ('<|think|>', '<|end|>', '<|content|>'),  # Solar Open
     # ('Thinking Process:', '', None),  # Qwen3.5 verbose thinking outside tags -- removed: too prone to false positives in streaming
@@ -81,4 +81,14 @@ def extract_reasoning(text, html_escaped=False):
 
         return text[thought_start:thought_end], text[content_start:]
 
+    # Handle standalone GPT-OSS final channel marker without a preceding
+    # analysis/commentary block (the model skipped thinking entirely).
+    for marker in ['<|start|>assistant<|channel|>final<|message|>', '<|channel|>final<|message|>']:
+        marker_esc = esc(marker)
+        pos = text.find(marker_esc)
+        if pos != -1:
+            before = text[:pos].strip()
+            after = text[pos + len(marker_esc):]
+            return (before if before else None), after
+
     return None, text
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index e9df9bd37c..dc108f6d6b 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -728,6 +728,8 @@ def generate_prompt_variation(state):
         variation = variation.rsplit("", 1)[1]
     elif "<|start|>assistant<|channel|>final<|message|>" in variation:
         variation = variation.rsplit("<|start|>assistant<|channel|>final<|message|>", 1)[1]
+    elif "<|channel|>final<|message|>" in variation:
+        variation = variation.rsplit("<|channel|>final<|message|>", 1)[1]
     elif "" in variation:
         variation = variation.rsplit("", 1)[1]
 

From 8bff331893e6ea1caf82b764e3d41514a04aa573 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 09:26:20 -0700
Subject: [PATCH 1368/1701] UI: Fix tool call markup flashing before accordion
 appears during streaming

---
 modules/chat.py | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index 10785c19a3..08f5553944 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1033,6 +1033,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
     # When tools are active, buffer streaming output during potential tool
     # call generation to prevent raw markup from leaking into the display.
     _check_tool_markers = bool(state.get('tools'))
+    _last_visible_before_tool_buffer = None
     if _check_tool_markers:
         from modules.tool_parsing import streaming_tool_buffer_check
         _tool_names = [t['function']['name'] for t in state['tools'] if 'function' in t and 'name' in t['function']]
@@ -1089,6 +1090,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             if _check_tool_markers:
                 if streaming_tool_buffer_check(output['internal'][-1][1], _tool_names):
                     continue
+                _last_visible_before_tool_buffer = output['visible'][-1][1]
 
             yield output
 
@@ -1122,6 +1124,13 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             'visible_content': output['visible'][row_idx][1]
         })
 
+    # When tool markers were detected during streaming, restore the last
+    # visible text from before buffering started so raw markup doesn't flash
+    # in the UI.  The internal text is left intact so the caller can still
+    # parse tool calls from it.
+    if is_stream and _check_tool_markers and streaming_tool_buffer_check(output['internal'][-1][1], _tool_names):
+        output['visible'][-1][1] = _last_visible_before_tool_buffer or ''
+
     yield output
 
 

From c908ac00d76d263c202a1b0cd2ed48c6f369f5e5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 09:29:17 -0700
Subject: [PATCH 1369/1701] Replace html2text with trafilatura for better web
 content extraction

After this change a lot of boilerplate is removed from web pages, saving tokens on agentic loops.
---
 modules/web_search.py                         | 21 ++++++++-----------
 requirements/full/requirements.txt            |  1 -
 requirements/full/requirements_amd.txt        |  1 -
 .../full/requirements_apple_intel.txt         |  1 -
 .../full/requirements_apple_silicon.txt       |  1 -
 requirements/full/requirements_cpu_only.txt   |  1 -
 requirements/full/requirements_nowheels.txt   |  1 -
 requirements/portable/requirements.txt        |  2 +-
 requirements/portable/requirements_amd.txt    |  2 +-
 .../portable/requirements_apple_intel.txt     |  2 +-
 .../portable/requirements_apple_silicon.txt   |  2 +-
 .../portable/requirements_cpu_only.txt        |  2 +-
 .../portable/requirements_cuda131.txt         |  2 +-
 .../portable/requirements_nowheels.txt        |  2 +-
 requirements/portable/requirements_vulkan.txt |  2 +-
 15 files changed, 17 insertions(+), 26 deletions(-)

diff --git a/modules/web_search.py b/modules/web_search.py
index 216d7933bf..a4424ee3c3 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -42,9 +42,9 @@ def get_current_timestamp():
 
 def download_web_page(url, timeout=10, include_links=False):
     """
-    Download a web page and convert its HTML content to structured Markdown text.
+    Download a web page and extract its main content as Markdown text.
     """
-    import html2text
+    import trafilatura
 
     try:
         _validate_url(url)
@@ -62,16 +62,13 @@ def download_web_page(url, timeout=10, include_links=False):
 
         response.raise_for_status()
 
-        # Initialize the HTML to Markdown converter
-        h = html2text.HTML2Text()
-        h.body_width = 0
-        h.ignore_images = True
-        h.ignore_links = not include_links
-
-        # Convert the HTML to Markdown
-        markdown_text = h.handle(response.text)
-
-        return markdown_text
+        result = trafilatura.extract(
+            response.text,
+            include_links=include_links,
+            output_format='markdown',
+            url=url
+        )
+        return result or ""
     except requests.exceptions.RequestException as e:
         logger.error(f"Error downloading {url}: {e}")
         return ""
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 8a0802f7be..e493d83d66 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -6,7 +6,6 @@ diffusers==0.37.*
 einops
 fastapi==0.112.4
 flash-linear-attention==0.4.*
-html2text==2025.4.15
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 9b31d668a9..48cace3335 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -4,7 +4,6 @@ datasets
 diffusers==0.37.*
 einops
 fastapi==0.112.4
-html2text==2025.4.15
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 138639e551..f9132f2e16 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -4,7 +4,6 @@ datasets
 diffusers==0.37.*
 einops
 fastapi==0.112.4
-html2text==2025.4.15
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index f3ebd171f8..e4b2882de4 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -4,7 +4,6 @@ datasets
 diffusers==0.37.*
 einops
 fastapi==0.112.4
-html2text==2025.4.15
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index e32a2ed1d1..1b42737be0 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -4,7 +4,6 @@ datasets
 diffusers==0.37.*
 einops
 fastapi==0.112.4
-html2text==2025.4.15
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 6128c0edc9..ea9ad2c71b 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -4,7 +4,6 @@ datasets
 diffusers==0.37.*
 einops
 fastapi==0.112.4
-html2text==2025.4.15
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 93eb3b8555..0471cc7339 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-html2text==2025.4.15
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
@@ -11,6 +10,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
+trafilatura==2.0.0
 tqdm
 
 # Gradio
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 36e0e4d910..dfefce206e 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-html2text==2025.4.15
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
@@ -11,6 +10,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
+trafilatura==2.0.0
 tqdm
 
 # Gradio
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 495bd5fa02..5c032e6b8c 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-html2text==2025.4.15
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
@@ -11,6 +10,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
+trafilatura==2.0.0
 tqdm
 
 # Gradio
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 7e82f68d42..385ecedf7c 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-html2text==2025.4.15
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
@@ -11,6 +10,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
+trafilatura==2.0.0
 tqdm
 
 # Gradio
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 046619e141..d8f7d494e9 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-html2text==2025.4.15
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
@@ -11,6 +10,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
+trafilatura==2.0.0
 tqdm
 
 # Gradio
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 590562f8b7..adc6a065db 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-html2text==2025.4.15
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
@@ -11,6 +10,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
+trafilatura==2.0.0
 tqdm
 
 # Gradio
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 8c3e2aacbd..942f7a2ad1 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-html2text==2025.4.15
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
@@ -11,6 +10,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
+trafilatura==2.0.0
 tqdm
 
 # Gradio
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index bf80deb07c..fca722fdc6 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -1,6 +1,5 @@
 audioop-lts<1.0; python_version >= "3.13"
 fastapi==0.112.4
-html2text==2025.4.15
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
@@ -11,6 +10,7 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
+trafilatura==2.0.0
 tqdm
 
 # Gradio

From c7953fb92319d426f09045f4bf34f701adfeec5c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 09:44:37 -0700
Subject: [PATCH 1370/1701] Add ROCm version to portable package filenames

---
 .github/workflows/build-portable-release-rocm.yml | 4 ++--
 requirements/full/requirements.txt                | 1 +
 requirements/full/requirements_amd.txt            | 1 +
 requirements/full/requirements_apple_intel.txt    | 1 +
 requirements/full/requirements_apple_silicon.txt  | 1 +
 requirements/full/requirements_cpu_only.txt       | 1 +
 requirements/full/requirements_nowheels.txt       | 1 +
 7 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/build-portable-release-rocm.yml b/.github/workflows/build-portable-release-rocm.yml
index 6f9ea4ec80..1050fa7e78 100644
--- a/.github/workflows/build-portable-release-rocm.yml
+++ b/.github/workflows/build-portable-release-rocm.yml
@@ -148,11 +148,11 @@ jobs:
             # 6. Create archive
             cd ..
             if [[ "$RUNNER_OS" == "Windows" ]]; then
-                ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-rocm.zip"
+                ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-rocm7.2.zip"
                 echo "Creating archive: $ARCHIVE_NAME"
                 powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
             else
-                ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-rocm.tar.gz"
+                ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-rocm7.2.tar.gz"
                 echo "Creating archive: $ARCHIVE_NAME"
                 tar czf "$ARCHIVE_NAME" "text-generation-webui-${VERSION_CLEAN}"
             fi
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index e493d83d66..dca686d9bc 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -24,6 +24,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
+trafilatura==2.0.0
 transformers==5.3.*
 triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 48cace3335..37cbf729f1 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -24,6 +24,7 @@ tensorboard
 torchao==0.15.*
 transformers==5.3.*
 tqdm
+trafilatura==2.0.0
 wandb
 
 # Gradio
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index f9132f2e16..fed46240a5 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -24,6 +24,7 @@ tensorboard
 torchao==0.15.*
 transformers==5.3.*
 tqdm
+trafilatura==2.0.0
 wandb
 
 # Gradio
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index e4b2882de4..fac36437eb 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -24,6 +24,7 @@ tensorboard
 torchao==0.15.*
 transformers==5.3.*
 tqdm
+trafilatura==2.0.0
 wandb
 
 # Gradio
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 1b42737be0..c86caf37fc 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -24,6 +24,7 @@ tensorboard
 torchao==0.15.*
 transformers==5.3.*
 tqdm
+trafilatura==2.0.0
 wandb
 
 # Gradio
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index ea9ad2c71b..4f5891da16 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -24,6 +24,7 @@ tensorboard
 torchao==0.15.*
 transformers==5.3.*
 tqdm
+trafilatura==2.0.0
 wandb
 
 # Gradio

From d0a4993cf483eaf0a4a4d96b5418c8da17dc23b4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 09:53:12 -0700
Subject: [PATCH 1371/1701] UI: Increase ctx-size slider maximum to 1M and step
 to 1024

---
 modules/ui_model_menu.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index d17f586b17..6ab19b7ce5 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -42,7 +42,7 @@ def create_ui():
                     with gr.Row():
                         with gr.Column():
                             shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=-1, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Number of layers to offload to the GPU. -1 = auto.')
-                            shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=0, maximum=262144, step=256, value=shared.args.ctx_size, info='Context length. llama.cpp: 0 = auto if gpu-layers is also -1. Common values: 4096, 8192, 16384, 32768, 65536, 131072.')
+                            shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=0, maximum=1048576, step=1024, value=shared.args.ctx_size, info='Context length. llama.cpp: 0 = auto if gpu-layers is also -1. Common values: 4096, 8192, 16384, 32768, 65536, 131072.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')

From 573617157ae6cf75393ee4aae118ac7e3607bc7f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 12:09:41 -0700
Subject: [PATCH 1372/1701] Optimize tool call detection

Avoids templates that don't contain a given necessary keyword
---
 extensions/openai/completions.py |   9 +-
 modules/chat.py                  |  17 ++-
 modules/tool_parsing.py          | 189 +++++++++++++++++++++----------
 3 files changed, 145 insertions(+), 70 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index 514270504f..fc17a19ac6 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -12,7 +12,7 @@
 from extensions.openai.errors import InvalidRequestError
 from extensions.openai.typing import ToolDefinition
 from extensions.openai.utils import debug_msg
-from modules.tool_parsing import get_tool_call_id, parse_tool_call
+from modules.tool_parsing import get_tool_call_id, parse_tool_call, detect_tool_call_format
 from modules import shared
 from modules.reasoning import extract_reasoning
 from modules.chat import (
@@ -484,6 +484,7 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
     tool_calls = []
     end_last_tool_call = 0
     supported_tools = [x["function"]["name"] for x in tools] if tools is not None else None
+    _tool_parsers = None
 
     # Filter supported_tools when tool_choice specifies a particular function
     if supported_tools and isinstance(tool_choice, dict):
@@ -491,11 +492,15 @@ def chat_streaming_chunk(content=None, chunk_tool_calls=None, include_role=False
         if specified_func and specified_func in supported_tools:
             supported_tools = [specified_func]
 
+    if supported_tools is not None:
+        _template_str = generate_params.get('instruction_template_str', '') if generate_params.get('mode') == 'instruct' else generate_params.get('chat_template_str', '')
+        _tool_parsers, _, _ = detect_tool_call_format(_template_str)
+
     for a in generator:
         answer = a['internal'][-1][1]
 
         if supported_tools is not None:
-            tool_call = parse_tool_call(answer[end_last_tool_call:], supported_tools) if len(answer) > 0 else []
+            tool_call = parse_tool_call(answer[end_last_tool_call:], supported_tools, parsers=_tool_parsers) if len(answer) > 0 else []
             if len(tool_call) > 0:
                 for tc in tool_call:
                     tc["id"] = get_tool_call_id()
diff --git a/modules/chat.py b/modules/chat.py
index 08f5553944..1ffbb56b8e 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1035,8 +1035,10 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
     _check_tool_markers = bool(state.get('tools'))
     _last_visible_before_tool_buffer = None
     if _check_tool_markers:
-        from modules.tool_parsing import streaming_tool_buffer_check
+        from modules.tool_parsing import streaming_tool_buffer_check, detect_tool_call_format
         _tool_names = [t['function']['name'] for t in state['tools'] if 'function' in t and 'name' in t['function']]
+        _template_str = state.get('instruction_template_str', '') if state.get('mode') == 'instruct' else state.get('chat_template_str', '')
+        _, _streaming_markers, _check_bare_names = detect_tool_call_format(_template_str)
 
     # Generate
     reply = None
@@ -1088,7 +1090,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
 
         if is_stream:
             if _check_tool_markers:
-                if streaming_tool_buffer_check(output['internal'][-1][1], _tool_names):
+                if streaming_tool_buffer_check(output['internal'][-1][1], markers=_streaming_markers, tool_names=_tool_names, check_bare_names=_check_bare_names):
                     continue
                 _last_visible_before_tool_buffer = output['visible'][-1][1]
 
@@ -1128,7 +1130,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
     # visible text from before buffering started so raw markup doesn't flash
     # in the UI.  The internal text is left intact so the caller can still
     # parse tool calls from it.
-    if is_stream and _check_tool_markers and streaming_tool_buffer_check(output['internal'][-1][1], _tool_names):
+    if is_stream and _check_tool_markers and streaming_tool_buffer_check(output['internal'][-1][1], markers=_streaming_markers, tool_names=_tool_names, check_bare_names=_check_bare_names):
         output['visible'][-1][1] = _last_visible_before_tool_buffer or ''
 
     yield output
@@ -1210,14 +1212,17 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
     # Load tools if any are selected
     selected = state.get('selected_tools', [])
     parse_tool_call = None
+    _tool_parsers = None
     if selected:
         from modules.tool_use import load_tools, execute_tool
-        from modules.tool_parsing import parse_tool_call, get_tool_call_id
+        from modules.tool_parsing import parse_tool_call, get_tool_call_id, detect_tool_call_format
 
     if selected:
         tool_defs, tool_executors = load_tools(selected)
         state['tools'] = tool_defs
         tool_func_names = [t['function']['name'] for t in tool_defs]
+        _template_str = state.get('instruction_template_str', '') if state.get('mode') == 'instruct' else state.get('chat_template_str', '')
+        _tool_parsers, _, _ = detect_tool_call_format(_template_str)
     else:
         tool_func_names = None
 
@@ -1272,7 +1277,7 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
                 last_save_time = current_time
 
             # Early stop on tool call detection
-            if tool_func_names and parse_tool_call(history['internal'][-1][1], tool_func_names):
+            if tool_func_names and parse_tool_call(history['internal'][-1][1], tool_func_names, parsers=_tool_parsers):
                 break
 
         # Save the model's visible output before re-applying visible_prefix,
@@ -1304,7 +1309,7 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
             break
 
         answer = history['internal'][-1][1]
-        parsed_calls, content_prefix = parse_tool_call(answer, tool_func_names, return_prefix=True) if answer else (None, '')
+        parsed_calls, content_prefix = parse_tool_call(answer, tool_func_names, return_prefix=True, parsers=_tool_parsers) if answer else (None, '')
 
         if not parsed_calls:
             break  # No tool calls — done
diff --git a/modules/tool_parsing.py b/modules/tool_parsing.py
index 418503ad4e..0454e901c5 100644
--- a/modules/tool_parsing.py
+++ b/modules/tool_parsing.py
@@ -9,9 +9,7 @@ def get_tool_call_id() -> str:
     return "call_" + "".join(b).lower()
 
 
-# Known opening markers for tool calls across model formats.
-# Used during streaming to buffer output that might be tool call markup,
-# preventing raw markup from leaking into displayed/streamed content.
+# All known opening markers for tool calls across model formats.
 TOOL_CALL_OPENING_MARKERS = [
     '',
     '',
@@ -25,36 +23,47 @@ def get_tool_call_id() -> str:
     '<|channel|>commentary',
 ]
 
-def streaming_tool_buffer_check(text, tool_names=None):
+
+def streaming_tool_buffer_check(text, markers=None, tool_names=None, check_bare_names=False):
     '''
     Check whether streaming output should be withheld because it may
     contain tool-call markup.
+
+    Args:
+        text: Full accumulated internal text.
+        markers: Template-specific markers for partial-prefix matching.
+                 If None, falls back to TOOL_CALL_OPENING_MARKERS.
+        tool_names: List of tool function names.
+        check_bare_names: Whether to do partial-prefix matching on tool
+                          names (for models with unknown template format).
     '''
-    # Full marker found → buffer permanently
+    # Full marker found in text → buffer permanently.
+    # Always checks ALL known markers regardless of template (cheap safety net).
     for marker in TOOL_CALL_OPENING_MARKERS:
         if marker in text:
             return True
 
-    # Bare function-name style (e.g. Devstral): "get_weather{...}"
-    # Only match tool name followed by '{' to avoid false positives on
-    # common words that happen to be tool names (e.g. "get", "search").
+    # Bare function-name full match: "get_weather{...}" or "get_weather {...}"
     if tool_names:
         for name in tool_names:
             if name + '{' in text or name + ' {' in text:
                 return True
-            # Partial: text ends with tool name (or prefix of it) but '{' hasn't arrived yet
+
+    # Partial-prefix matching: only for template-specific markers.
+    for marker in (markers if markers is not None else TOOL_CALL_OPENING_MARKERS):
+        for prefix_len in range(min(len(marker) - 1, len(text)), 0, -1):
+            if text.endswith(marker[:prefix_len]):
+                return True
+
+    # Bare-name partial matching: only when template format is unknown.
+    if check_bare_names and tool_names:
+        for name in tool_names:
             if text.endswith(name):
                 return True
             for prefix_len in range(min(len(name) - 1, len(text)), 0, -1):
                 if text.endswith(name[:prefix_len]):
                     return True
 
-    # Tail might be a partial marker forming across tokens
-    for marker in TOOL_CALL_OPENING_MARKERS:
-        for prefix_len in range(min(len(marker) - 1, len(text)), 0, -1):
-            if text.endswith(marker[:prefix_len]):
-                return True
-
     return False
 
 
@@ -488,7 +497,102 @@ def _parse_pythonic_tool_calls(answer: str, tool_names: list[str]):
     return matches, start_pos
 
 
-def parse_tool_call(answer: str, tool_names: list[str], return_prefix: bool = False):
+# Format registry: maps template substrings to the parser and streaming
+# markers for that format.  When a format's hints are NOT found in the
+# template, its parser and markers are excluded.
+TOOL_CALL_FORMATS = [
+    {
+        'template_hints': ['tool▁call▁begin', 'tool▁calls▁begin'],
+        'parser': _parse_deep_seek_tool_calls,
+        'markers': ['<｜tool▁call▁begin｜>', '<｜tool▁calls▁begin｜>'],
+    },
+    {
+        'template_hints': ['<|tool_call_begin|>', 'tool_calls_section'],
+        'parser': _parse_kimi_tool_calls,
+        'markers': ['<|tool_call_begin|>', '<|tool_calls_section_begin|>'],
+    },
+    {
+        'template_hints': ['to=functions.', '<|channel|>'],
+        'parser': _parse_channel_tool_calls,
+        'markers': ['to=functions.', '<|channel|>commentary'],
+    },
+    {
+        'template_hints': ['minimax:tool_call'],
+        'parser': _parse_minimax_tool_calls,
+        'markers': [''],
+    },
+    {
+        'template_hints': [''],
+        'parser': _parse_glm_tool_calls,
+        'markers': [''],
+    },
+    {
+        'template_hints': [''],
+        'parser': _parse_xml_param_tool_calls,
+        'markers': [''],
+    },
+    {
+        'template_hints': ['[TOOL_CALLS]'],
+        'parser': _parse_mistral_token_tool_calls,
+        'markers': ['[TOOL_CALLS]'],
+    },
+    {
+        'template_hints': [''],
+        'parser': None,
+        'markers': [''],
+    },
+]
+
+# Default ordered list of all specialized parsers.
+ALL_PARSERS = [
+    _parse_deep_seek_tool_calls,
+    _parse_kimi_tool_calls,
+    _parse_channel_tool_calls,
+    _parse_minimax_tool_calls,
+    _parse_glm_tool_calls,
+    _parse_xml_param_tool_calls,
+    _parse_mistral_token_tool_calls,
+    _parse_bare_name_tool_calls,
+    _parse_pythonic_tool_calls,
+]
+
+
+def detect_tool_call_format(template_str):
+    """Inspect a chat/instruction template to determine which tool call
+    formats are relevant.
+
+    Uses an exclude-based approach: starts with all parsers/markers,
+    then removes the ones whose hints are not found in the template.
+
+    Returns (parsers, streaming_markers, check_bare_names).
+    """
+    if not template_str:
+        return None, TOOL_CALL_OPENING_MARKERS, True
+
+    matched_any = False
+    exclude_parsers = []
+    exclude_markers = []
+    matched_markers = []
+
+    for fmt in TOOL_CALL_FORMATS:
+        if any(hint in template_str for hint in fmt['template_hints']):
+            matched_any = True
+            matched_markers.extend(fmt['markers'])
+        else:
+            if fmt['parser'] is not None:
+                exclude_parsers.append(fmt['parser'])
+            exclude_markers.extend(fmt['markers'])
+
+    if not matched_any:
+        return None, TOOL_CALL_OPENING_MARKERS, True
+
+    parsers = [p for p in ALL_PARSERS if p not in exclude_parsers]
+    markers = [m for m in TOOL_CALL_OPENING_MARKERS if m not in exclude_markers or m in matched_markers]
+
+    return parsers, markers, False
+
+
+def parse_tool_call(answer: str, tool_names: list[str], return_prefix: bool = False, parsers: list = None):
     matches = []
     start_pos = None
 
@@ -498,52 +602,13 @@ def _return(matches, start_pos):
             return matches, prefix
         return matches
 
-    # Check for DeepSeek-style tool calls (fullwidth Unicode token delimiters)
-    matches, start_pos = _parse_deep_seek_tool_calls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for Kimi-K2-style tool calls (pipe-delimited tokens)
-    matches, start_pos = _parse_kimi_tool_calls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for channel-based tool calls (e.g. GPT-OSS format)
-    matches, start_pos = _parse_channel_tool_calls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for MiniMax-style tool calls (invoke/parameter XML tags)
-    matches, start_pos = _parse_minimax_tool_calls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for GLM-style tool calls (arg_key/arg_value XML pairs)
-    matches, start_pos = _parse_glm_tool_calls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for XML-parameter style tool calls (e.g. Qwen3.5 format)
-    matches, start_pos = _parse_xml_param_tool_calls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for Mistral/Devstral-style tool calls ([TOOL_CALLS]name[ARGS]json)
-    matches, start_pos = _parse_mistral_token_tool_calls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for bare function-name style tool calls (e.g. Mistral format)
-    matches, start_pos = _parse_bare_name_tool_calls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Check for pythonic-style tool calls (e.g. Llama 4 format)
-    matches, start_pos = _parse_pythonic_tool_calls(answer, tool_names)
-    if matches:
-        return _return(matches, start_pos)
-
-    # Define the regex pattern to find the JSON content wrapped in , , , and other tags observed from various models
+    # Try specialized parsers.
+    for parser in (parsers if parsers is not None else ALL_PARSERS):
+        matches, start_pos = parser(answer, tool_names)
+        if matches:
+            return _return(matches, start_pos)
+
+    # Generic fallback: regex pattern to find the JSON content wrapped in , , , and other tags observed from various models
     patterns = [r"(```[^\n]*)\n(.*?)```", r"<([^>]+)>(.*?)"]
 
     for pattern in patterns:

From beab346f48639b67c4b61584d2410c8d4685539e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 12:45:37 -0700
Subject: [PATCH 1373/1701] UI: Fix a minor glitch

---
 css/main.css | 1 -
 1 file changed, 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 49b8f752d8..d5e511dea0 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1388,7 +1388,6 @@ audio {
     overflow-wrap: break-word;
     max-height: 250px;
     overflow-y: scroll;
-    contain: layout;
 }
 
 .chat .message-body .thinking-content p,

From c09a367c6460a3fb5e44f2f5d4c24c34a7506715 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 14:01:34 -0700
Subject: [PATCH 1374/1701] UI: Fix dark theme using light theme syntax
 highlighting

---
 js/main.js | 6 ++++++
 server.py  | 4 ++++
 2 files changed, 10 insertions(+)

diff --git a/js/main.js b/js/main.js
index 5f79c3ec25..394553d02b 100644
--- a/js/main.js
+++ b/js/main.js
@@ -2,6 +2,12 @@
 // Main
 // ------------------------------------------------
 
+// Sync highlight.js theme with the actual Gradio theme
+var defined_hljs_css = document.body.classList.contains("dark") ? "file/css/highlightjs/github-dark.min.css" : "file/css/highlightjs/github.min.css";
+if (document.getElementById("highlight-css").getAttribute("href") !== defined_hljs_css) {
+  document.getElementById("highlight-css").setAttribute("href", defined_hljs_css);
+}
+
 let main_parent = document.getElementById("chat-tab").parentNode;
 let extensions = document.getElementById("extensions");
 
diff --git a/server.py b/server.py
index 340f71265e..1aa9fc0482 100644
--- a/server.py
+++ b/server.py
@@ -218,6 +218,10 @@ def create_interface():
 
         shared.gradio['interface'].load(partial(ui.apply_interface_values, {}, use_persistent=True), None, gradio(ui.list_interface_input_elements()), show_progress=False)
 
+        # Sync theme_state with the actual client-side theme so that
+        # autosave always writes the correct dark_theme value.
+        shared.gradio['interface'].load(None, None, gradio('theme_state'), js='() => document.body.classList.contains("dark") ? "dark" : "light"')
+
         extensions_module.create_extensions_tabs()  # Extensions tabs
         extensions_module.create_extensions_block()  # Extensions block
 

From 9f657d39768da03180273dbb48e2af5424a29878 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 14:19:12 -0700
Subject: [PATCH 1375/1701] UI: Fix a minor glitch

---
 js/global_scope_js.js | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 425c2c5970..084c98e83d 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -287,7 +287,7 @@ function updateInstructPadding() {
     const messagesContainer = chatElement.querySelector(".messages");
     const lastChild = messagesContainer?.lastElementChild;
     const prevSibling = lastChild?.previousElementSibling;
-    if (lastChild && prevSibling) {
+    if (lastChild && prevSibling && chatElement.offsetHeight > 0) {
       let bufferHeight = Math.max(0, Math.max(window.innerHeight - 128 - 84, window.innerHeight - prevSibling.offsetHeight - 84) - lastChild.offsetHeight);
       if (window.innerWidth <= 924) {
         bufferHeight = Math.max(0, bufferHeight - 32);

From 4ae2bd86e28fe67755e5329069a0d073d1162bae Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 15:30:01 -0700
Subject: [PATCH 1376/1701] Change the default ctx-size to 0 (auto) for
 llama.cpp

---
 modules/llama_cpp_server.py | 4 +++-
 modules/models.py           | 3 +++
 modules/shared.py           | 2 +-
 modules/ui_model_menu.py    | 2 +-
 4 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 1425844d1e..fc8e9a1979 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -371,6 +371,8 @@ def _start_server(self):
 
         if shared.args.ctx_size > 0:
             cmd += ["--ctx-size", str(shared.args.ctx_size)]
+        elif shared.args.gpu_layers >= 0:
+            cmd += ["--ctx-size", "8192"]
 
         if shared.args.gpu_layers >= 0:
             cmd += ["--gpu-layers", str(shared.args.gpu_layers), "--fit", "off"]
@@ -477,7 +479,7 @@ def _start_server(self):
             print()
 
         gpu_layers_str = "auto" if shared.args.gpu_layers < 0 else str(shared.args.gpu_layers)
-        ctx_size_str = "auto" if shared.args.ctx_size == 0 else str(shared.args.ctx_size)
+        ctx_size_str = "auto" if shared.args.ctx_size == 0 and shared.args.gpu_layers < 0 else str(shared.args.ctx_size or 8192)
         logger.info(f"Using gpu_layers={gpu_layers_str} | ctx_size={ctx_size_str} | cache_type={cache_type}")
         # Start the server with pipes for output
         self.process = subprocess.Popen(
diff --git a/modules/models.py b/modules/models.py
index d83b98d79e..1d139b89a6 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -38,6 +38,9 @@ def load_model(model_name, loader=None):
         sampler_hijack.hijack_samplers()
 
     shared.args.loader = loader
+    if loader != 'llama.cpp' and shared.args.ctx_size == 0:
+        shared.args.ctx_size = 8192
+
     output = load_func_map[loader](model_name)
     if type(output) is tuple:
         model, tokenizer = output
diff --git a/modules/shared.py b/modules/shared.py
index 8c0aad9a51..1cf365c6ef 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -76,7 +76,7 @@
 
 # Cache
 group = parser.add_argument_group('Context and cache')
-group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=8192, metavar='N', help='Context size in tokens. llama.cpp: 0 = auto if gpu-layers is also -1.')
+group.add_argument('--ctx-size', '--n_ctx', '--max_seq_len', type=int, default=0, metavar='N', help='Context size in tokens. 0 = auto for llama.cpp (requires gpu-layers=-1), 8192 for other loaders.')
 group.add_argument('--cache-type', '--cache_type', type=str, default='fp16', metavar='N', help='KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).')
 
 # Speculative decoding
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 6ab19b7ce5..b53bc292e7 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -42,7 +42,7 @@ def create_ui():
                     with gr.Row():
                         with gr.Column():
                             shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=-1, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Number of layers to offload to the GPU. -1 = auto.')
-                            shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=0, maximum=1048576, step=1024, value=shared.args.ctx_size, info='Context length. llama.cpp: 0 = auto if gpu-layers is also -1. Common values: 4096, 8192, 16384, 32768, 65536, 131072.')
+                            shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=0, maximum=1048576, step=1024, value=shared.args.ctx_size, info='Context length. 0 = auto for llama.cpp (requires gpu-layers=-1), 8192 for other loaders. Common values: 4096, 8192, 16384, 32768, 65536, 131072.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')

From e11425d5f81adab3ab99a5920daf46b6ed042ceb Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 15:46:21 -0700
Subject: [PATCH 1377/1701] Fix relative redirect handling in web page fetcher

---
 modules/web_search.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/web_search.py b/modules/web_search.py
index a4424ee3c3..6d005496fe 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -7,7 +7,7 @@
 import urllib.request
 from concurrent.futures import as_completed
 from datetime import datetime
-from urllib.parse import quote_plus, urlparse
+from urllib.parse import quote_plus, urljoin, urlparse
 
 import requests
 
@@ -55,7 +55,7 @@ def download_web_page(url, timeout=10, include_links=False):
         for _ in range(max_redirects):
             response = requests.get(url, headers=headers, timeout=timeout, allow_redirects=False)
             if response.is_redirect and 'Location' in response.headers:
-                url = response.headers['Location']
+                url = urljoin(url, response.headers['Location'])
                 _validate_url(url)
             else:
                 break

From 9eacd4a2073fa5e2556dee8e6c511e979fd2f8c4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 16:07:16 -0700
Subject: [PATCH 1378/1701] UI: Minor morphdom optimizations

---
 js/global_scope_js.js | 30 ++++++++++++------------------
 1 file changed, 12 insertions(+), 18 deletions(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 084c98e83d..ba5abcb24a 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -325,27 +325,21 @@ function applyMorphdomUpdate(data) {
 
   const queryScope = target_element;
 
-  // Track open blocks
+  // Track open blocks and store their scroll positions
   const openBlocks = new Set();
+  const scrollPositions = {};
   queryScope.querySelectorAll(".thinking-block").forEach(block => {
     const blockId = block.getAttribute("data-block-id");
-    // If block exists and is open, add to open set
     if (blockId && block.hasAttribute("open")) {
       openBlocks.add(blockId);
-    }
-  });
-
-  // Store scroll positions for any open blocks
-  const scrollPositions = {};
-  queryScope.querySelectorAll(".thinking-block[open]").forEach(block => {
-    const content = block.querySelector(".thinking-content");
-    const blockId = block.getAttribute("data-block-id");
-    if (content && blockId) {
-      const isAtBottom = Math.abs((content.scrollHeight - content.scrollTop) - content.clientHeight) < 5;
-      scrollPositions[blockId] = {
-        position: content.scrollTop,
-        isAtBottom: isAtBottom
-      };
+      const content = block.querySelector(".thinking-content");
+      if (content) {
+        const isAtBottom = Math.abs((content.scrollHeight - content.scrollTop) - content.clientHeight) < 5;
+        scrollPositions[blockId] = {
+          position: content.scrollTop,
+          isAtBottom: isAtBottom
+        };
+      }
     }
   });
 
@@ -355,8 +349,8 @@ function applyMorphdomUpdate(data) {
     {
       onBeforeElUpdated: function(fromEl, toEl) {
         // Preserve code highlighting
-        if (fromEl.tagName === "PRE" && fromEl.querySelector("code[data-highlighted]")) {
-          const fromCode = fromEl.querySelector("code");
+        if (fromEl.tagName === "PRE") {
+          const fromCode = fromEl.querySelector("code[data-highlighted]");
           const toCode = toEl.querySelector("code");
 
           if (fromCode && toCode && fromCode.textContent === toCode.textContent) {

From b9bdbd638e91e797718ffc048d187365d162e0e0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 18:18:33 -0700
Subject: [PATCH 1379/1701] Fix after 4ae2bd86e28fe67755e5329069a0d073d1162bae

---
 modules/models_settings.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 0e117176b7..25a352379c 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -67,7 +67,7 @@ def get_model_metadata(model):
 
         for k in metadata:
             if k.endswith('.context_length'):
-                model_settings['ctx_size'] = min(metadata[k], 8192)
+                model_settings['ctx_size'] = 0
                 model_settings['truncation_length_info'] = metadata[k]
             elif k.endswith('rope.freq_base'):
                 model_settings['rope_freq_base'] = metadata[k]

From c12653006198ba1ba08d563160dfae81f457aa61 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 18:22:41 -0700
Subject: [PATCH 1380/1701] UI: Minor color change

---
 css/main.css | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/css/main.css b/css/main.css
index d5e511dea0..30dd28aba3 100644
--- a/css/main.css
+++ b/css/main.css
@@ -439,6 +439,10 @@ audio {
     color: white !important;
 }
 
+.dark .message-body blockquote {
+    border-left-color: rgb(255 255 255 / 30%);
+}
+
 .message-body h1 {
     font-weight: 800;
     font-size: 2.25em;

From d1aba085613656c4eefaed223412cefdbe231905 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 18:35:44 -0700
Subject: [PATCH 1381/1701] UI: Set chat widths to 724px

---
 css/chat_style-Dark.css            | 2 +-
 css/chat_style-TheEncrypted777.css | 2 +-
 css/chat_style-cai-chat-square.css | 2 +-
 css/chat_style-cai-chat.css        | 2 +-
 css/chat_style-messenger.css       | 2 +-
 css/chat_style-wpp.css             | 2 +-
 css/html_instruct_style.css        | 2 +-
 7 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/css/chat_style-Dark.css b/css/chat_style-Dark.css
index 02beb93566..01a168abe9 100644
--- a/css/chat_style-Dark.css
+++ b/css/chat_style-Dark.css
@@ -2,7 +2,7 @@
     display: grid;
     align-items: start;
     grid-template-columns: 60px minmax(0, 1fr);
-    width: min(100%, calc(768px + 60px));
+    width: min(100%, calc(724px + 60px));
     padding-bottom: 22px;
     padding-top: 6px;
     font-size: 18px;
diff --git a/css/chat_style-TheEncrypted777.css b/css/chat_style-TheEncrypted777.css
index b3df671034..9543a3dfd8 100644
--- a/css/chat_style-TheEncrypted777.css
+++ b/css/chat_style-TheEncrypted777.css
@@ -4,7 +4,7 @@
     display: grid;
     align-items: start;
     grid-template-columns: 60px minmax(0, 1fr);
-    width: min(100%, calc(768px + 60px + 90px));
+    width: min(100%, calc(724px + 60px + 90px));
     padding-bottom: 21px;
     padding-top: 7px;
     font-size: 18px;
diff --git a/css/chat_style-cai-chat-square.css b/css/chat_style-cai-chat-square.css
index 0d9467dfe7..8254a4ecfd 100644
--- a/css/chat_style-cai-chat-square.css
+++ b/css/chat_style-cai-chat-square.css
@@ -19,5 +19,5 @@
     padding-bottom: 1.5em;
     padding-top: 0.5em;
     grid-template-columns: 70px minmax(0, 1fr);
-    width: min(100%, calc(768px + 70px));
+    width: min(100%, calc(724px + 70px));
 }
diff --git a/css/chat_style-cai-chat.css b/css/chat_style-cai-chat.css
index 9cc4d4cd6c..66d2816d9f 100644
--- a/css/chat_style-cai-chat.css
+++ b/css/chat_style-cai-chat.css
@@ -2,7 +2,7 @@
     display: grid;
     align-items: start;
     grid-template-columns: 60px minmax(0, 1fr);
-    width: min(100%, calc(768px + 60px));
+    width: min(100%, calc(724px + 60px));
     padding-bottom: 1.5em;
     padding-top: 0.5em;
     font-size: 15px;
diff --git a/css/chat_style-messenger.css b/css/chat_style-messenger.css
index 438b806087..fd9b5b7000 100644
--- a/css/chat_style-messenger.css
+++ b/css/chat_style-messenger.css
@@ -1,5 +1,5 @@
 .message {
-    width: min(100%, calc(48rem + 60px));
+    width: min(100%, calc(724px + 60px));
     padding-bottom: 22px;
     padding-top: 3px;
     font-size: 15px;
diff --git a/css/chat_style-wpp.css b/css/chat_style-wpp.css
index ad6985d251..65e253d9c3 100644
--- a/css/chat_style-wpp.css
+++ b/css/chat_style-wpp.css
@@ -1,6 +1,6 @@
 .message {
     display: block;
-    width: min(100%, 48rem);
+    width: min(100%, 724px);
     padding-top: 0;
     padding-bottom: 21px;
     font-size: 15px;
diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index d4780350de..458feafc1e 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -78,7 +78,7 @@
 
 .chat .user-message .text,
 .chat .assistant-message .text {
-    max-width: 768px;
+    max-width: 724px;
     margin-left: auto;
     margin-right: auto;
 }

From 9955e54a1f22562b3e1c772e509abede61ecb92c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 18:51:12 -0700
Subject: [PATCH 1382/1701] UI: Fix autoscroll not engaging when regenerating
 short chats

---
 js/main.js | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index 394553d02b..23db59d54b 100644
--- a/js/main.js
+++ b/js/main.js
@@ -157,7 +157,7 @@ let lastClientHeight = 0;
 
 targetElement.addEventListener("scroll", function() {
   let diff = targetElement.scrollHeight - targetElement.clientHeight;
-  let isAtBottomNow = Math.abs(targetElement.scrollTop - diff) <= 10 || diff == 0;
+  let isAtBottomNow = Math.abs(targetElement.scrollTop - diff) <= 10 || diff <= 0;
 
   // Add scrolling class to disable hover effects
   if (window.isScrolled || !isAtBottomNow) {

From 2d3a3794c91ebabf0c07e01b2558a5ac37162c24 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 19:15:22 -0700
Subject: [PATCH 1383/1701] Add a Top-P preset, make it the new default, clean
 up the built-in presets

---
 modules/shared.py                          | 6 +++---
 user_data/presets/Instruct.yaml            | 1 -
 user_data/presets/Qwen3 - No Thinking.yaml | 3 ---
 user_data/presets/Qwen3 - Thinking.yaml    | 3 ---
 user_data/presets/Top-P.yaml               | 1 +
 user_data/presets/min_p.yaml               | 1 -
 6 files changed, 4 insertions(+), 11 deletions(-)
 delete mode 100644 user_data/presets/Instruct.yaml
 delete mode 100644 user_data/presets/Qwen3 - No Thinking.yaml
 delete mode 100644 user_data/presets/Qwen3 - Thinking.yaml
 create mode 100644 user_data/presets/Top-P.yaml
 delete mode 100644 user_data/presets/min_p.yaml

diff --git a/modules/shared.py b/modules/shared.py
index 1cf365c6ef..475d57b71d 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -265,7 +265,7 @@
     'web_search_pages': 3,
     'selected_tools': [],
     'prompt-notebook': '',
-    'preset': 'Qwen3 - Thinking' if (user_data_dir / 'presets/Qwen3 - Thinking.yaml').exists() else None,
+    'preset': 'Top-P' if (user_data_dir / 'presets/Top-P.yaml').exists() else None,
     'max_new_tokens': 512,
     'max_new_tokens_min': 1,
     'max_new_tokens_max': 4096,
@@ -290,7 +290,7 @@
     'include_past_attachments': True,
 
     # Generation parameters - Curve shape
-    'temperature': 0.6,
+    'temperature': neutral_samplers['temperature'],
     'dynatemp_low': neutral_samplers['dynatemp_low'],
     'dynatemp_high': neutral_samplers['dynatemp_high'],
     'dynatemp_exponent': neutral_samplers['dynatemp_exponent'],
@@ -300,7 +300,7 @@
     # Generation parameters - Curve cutoff
     'min_p': neutral_samplers['min_p'],
     'top_p': 0.95,
-    'top_k': 20,
+    'top_k': neutral_samplers['top_k'],
     'typical_p': neutral_samplers['typical_p'],
     'xtc_threshold': neutral_samplers['xtc_threshold'],
     'xtc_probability': neutral_samplers['xtc_probability'],
diff --git a/user_data/presets/Instruct.yaml b/user_data/presets/Instruct.yaml
deleted file mode 100644
index 142fcd82e6..0000000000
--- a/user_data/presets/Instruct.yaml
+++ /dev/null
@@ -1 +0,0 @@
-min_p: 0.2
diff --git a/user_data/presets/Qwen3 - No Thinking.yaml b/user_data/presets/Qwen3 - No Thinking.yaml
deleted file mode 100644
index b1c1e03c44..0000000000
--- a/user_data/presets/Qwen3 - No Thinking.yaml	
+++ /dev/null
@@ -1,3 +0,0 @@
-temperature: 0.7
-top_p: 0.8
-top_k: 20
diff --git a/user_data/presets/Qwen3 - Thinking.yaml b/user_data/presets/Qwen3 - Thinking.yaml
deleted file mode 100644
index cb2942f905..0000000000
--- a/user_data/presets/Qwen3 - Thinking.yaml	
+++ /dev/null
@@ -1,3 +0,0 @@
-temperature: 0.6
-top_p: 0.95
-top_k: 20
diff --git a/user_data/presets/Top-P.yaml b/user_data/presets/Top-P.yaml
new file mode 100644
index 0000000000..f39e148f2c
--- /dev/null
+++ b/user_data/presets/Top-P.yaml
@@ -0,0 +1 @@
+top_p: 0.95
diff --git a/user_data/presets/min_p.yaml b/user_data/presets/min_p.yaml
deleted file mode 100644
index b8ebc95fe1..0000000000
--- a/user_data/presets/min_p.yaml
+++ /dev/null
@@ -1 +0,0 @@
-min_p: 0.05

From f0c16813ef11a8ed39db3586614c06239ef25807 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 14 Mar 2026 19:35:12 -0700
Subject: [PATCH 1384/1701] Remove the rope scaling parameters

Now models have 131k+ context length. The parameters can still be
passed to llama.cpp through --extra-flags.
---
 README.md                      | 26 ++++++++++----------------
 docs/04 - Model Tab.md         |  3 ---
 modules/llama_cpp_server.py    |  4 ----
 modules/loaders.py             |  7 -------
 modules/models_settings.py     | 22 ----------------------
 modules/shared.py              |  6 ------
 modules/transformers_loader.py |  7 -------
 modules/ui_model_menu.py       |  3 ---
 8 files changed, 10 insertions(+), 68 deletions(-)

diff --git a/README.md b/README.md
index 9a8e0a8624..f152717657 100644
--- a/README.md
+++ b/README.md
@@ -244,15 +244,14 @@ usage: server.py [-h] [--user-data-dir USER_DATA_DIR] [--multi-user] [--model MO
                  [--row-split] [--no-mmap] [--mlock] [--no-kv-offload] [--batch-size BATCH_SIZE] [--ubatch-size UBATCH_SIZE] [--threads THREADS] [--threads-batch THREADS_BATCH] [--numa]
                  [--parallel PARALLEL] [--fit-target FIT_TARGET] [--extra-flags EXTRA_FLAGS] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] [--bf16]
                  [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE]
-                 [--quant_type QUANT_TYPE] [--gpu-split GPU_SPLIT] [--enable-tp] [--tp-backend TP_BACKEND] [--cfg-cache] [--alpha_value ALPHA_VALUE] [--rope_freq_base ROPE_FREQ_BASE]
-                 [--compress_pos_emb COMPRESS_POS_EMB] [--listen] [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH]
-                 [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors] [--portable] [--api] [--public-api]
-                 [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4] [--nowebui] [--temperature N]
-                 [--dynatemp-low N] [--dynatemp-high N] [--dynatemp-exponent N] [--smoothing-factor N] [--smoothing-curve N] [--min-p N] [--top-p N] [--top-k N] [--typical-p N] [--xtc-threshold N]
-                 [--xtc-probability N] [--epsilon-cutoff N] [--eta-cutoff N] [--tfs N] [--top-a N] [--top-n-sigma N] [--adaptive-target N] [--adaptive-decay N] [--dry-multiplier N]
-                 [--dry-allowed-length N] [--dry-base N] [--repetition-penalty N] [--frequency-penalty N] [--presence-penalty N] [--encoder-repetition-penalty N] [--no-repeat-ngram-size N]
-                 [--repetition-penalty-range N] [--penalty-alpha N] [--guidance-scale N] [--mirostat-mode N] [--mirostat-tau N] [--mirostat-eta N] [--do-sample | --no-do-sample]
-                 [--dynamic-temperature | --no-dynamic-temperature] [--temperature-last | --no-temperature-last] [--sampler-priority N] [--dry-sequence-breakers N]
+                 [--quant_type QUANT_TYPE] [--gpu-split GPU_SPLIT] [--enable-tp] [--tp-backend TP_BACKEND] [--cfg-cache] [--listen] [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share]
+                 [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors]
+                 [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4]
+                 [--nowebui] [--temperature N] [--dynatemp-low N] [--dynatemp-high N] [--dynatemp-exponent N] [--smoothing-factor N] [--smoothing-curve N] [--min-p N] [--top-p N] [--top-k N]
+                 [--typical-p N] [--xtc-threshold N] [--xtc-probability N] [--epsilon-cutoff N] [--eta-cutoff N] [--tfs N] [--top-a N] [--top-n-sigma N] [--adaptive-target N] [--adaptive-decay N]
+                 [--dry-multiplier N] [--dry-allowed-length N] [--dry-base N] [--repetition-penalty N] [--frequency-penalty N] [--presence-penalty N] [--encoder-repetition-penalty N]
+                 [--no-repeat-ngram-size N] [--repetition-penalty-range N] [--penalty-alpha N] [--guidance-scale N] [--mirostat-mode N] [--mirostat-tau N] [--mirostat-eta N]
+                 [--do-sample | --no-do-sample] [--dynamic-temperature | --no-dynamic-temperature] [--temperature-last | --no-temperature-last] [--sampler-priority N] [--dry-sequence-breakers N]
                  [--enable-thinking | --no-enable-thinking] [--reasoning-effort N] [--chat-template-file CHAT_TEMPLATE_FILE]
 
 Text Generation Web UI
@@ -262,7 +261,7 @@ options:
 
 Basic settings:
   --user-data-dir USER_DATA_DIR                        Path to the user data directory. Default: auto-detected.
-  --multi-user                                         Multi-user mode. Chat histories are not saved or automatically loaded. Warning: this is likely not safe for sharing publicly.
+  --multi-user                                         Multi-user mode. Chat histories are not saved or automatically loaded. Best suited for small trusted teams.
   --model MODEL                                        Name of the model to load by default.
   --lora LORA [LORA ...]                               The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.
   --model-dir MODEL_DIR                                Path to directory with all the models.
@@ -289,7 +288,7 @@ Model loader:
                                                        LLM.
 
 Context and cache:
-  --ctx-size, --n_ctx, --max_seq_len N                 Context size in tokens. llama.cpp: 0 = auto if gpu-layers is also -1.
+  --ctx-size, --n_ctx, --max_seq_len N                 Context size in tokens. 0 = auto for llama.cpp (requires gpu-layers=-1), 8192 for other loaders.
   --cache-type, --cache_type N                         KV cache type; valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV3 - fp16, q2 to q8 (can specify k_bits and v_bits separately, e.g. q4_q8).
 
 Speculative decoding:
@@ -350,11 +349,6 @@ ExLlamaV3:
   --tp-backend TP_BACKEND                              The backend for tensor parallelism. Valid options: native, nccl. Default: native.
   --cfg-cache                                          Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.
 
-RoPE:
-  --alpha_value ALPHA_VALUE                            Positional embeddings alpha factor for NTK RoPE scaling. Use either this or compress_pos_emb, not both.
-  --rope_freq_base ROPE_FREQ_BASE                      If greater than 0, will be used instead of alpha_value. Those two are related by rope_freq_base = 10000 * alpha_value ^ (64 / 63).
-  --compress_pos_emb COMPRESS_POS_EMB                  Positional embeddings compression factor. Should be set to (context length) / (model's original context length). Equal to 1/rope_freq_scale.
-
 Gradio:
   --listen                                             Make the web UI reachable from your local network.
   --listen-port LISTEN_PORT                            The listening port that the server will use.
diff --git a/docs/04 - Model Tab.md b/docs/04 - Model Tab.md
index 4d5ae64507..744970ac6a 100644
--- a/docs/04 - Model Tab.md	
+++ b/docs/04 - Model Tab.md	
@@ -41,9 +41,6 @@ Options:
 * **cpu_memory**: Maximum CPU memory in GiB to use for CPU offloading via the accelerate library. Whatever doesn't fit in the GPU or CPU will go to a disk cache if the "disk" checkbox is enabled.
 * **compute_dtype**: Used when "load_in_4bit" is checked. I recommend leaving the default value.
 * **quant_type**: Used when "load_in_4bit" is checked. I recommend leaving the default value.
-* **alpha_value**: Used to extend the context length of a model with a minor loss in quality. I have measured 1.75 to be optimal for 1.5x context, and 2.5 for 2x context. That is, with alpha = 2.5 you can make a model with 4096 context length go to 8192 context length.
-* **rope_freq_base**: Originally another way to write "alpha_value", it ended up becoming a necessary parameter for some models like CodeLlama, which was fine-tuned with this set to 1000000 and hence needs to be loaded with it set to 1000000 as well.
-* **compress_pos_emb**: The first and original context-length extension method, discovered by [kaiokendev](https://kaiokendev.github.io/til). When set to 2, the context length is doubled, 3 and it's tripled, etc. It should only be used for models that have been fine-tuned with this parameter set to different than 1. For models that have not been tuned to have greater context length, alpha_value will lead to a smaller accuracy loss.
 * **attn_implementation**: Choose the attention implementation. Valid options: `sdpa`, `eager`, `flash_attention_2`. The default (`sdpa`) works well in most cases; `flash_attention_2` may be useful for training.
 * **cpu**: Loads the model in CPU mode using Pytorch. The model will be loaded in 32-bit precision, so a lot of RAM will be used. CPU inference with transformers is older than llama.cpp and it works, but it's a lot slower. Note: this parameter has a different interpretation in the llama.cpp loader (see above).
 * **load_in_8bit**: Load the model in 8-bit precision using bitsandbytes. The 8-bit kernel in that library has been optimized for training and not inference, so load_in_8bit is slower than load_in_4bit (but more accurate).
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index fc8e9a1979..05c07748e0 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -403,10 +403,6 @@ def _start_server(self):
         if shared.args.cache_type != "fp16" and shared.args.cache_type in llamacpp_valid_cache_types:
             cmd += ["--cache-type-k", shared.args.cache_type, "--cache-type-v", shared.args.cache_type]
             cache_type = shared.args.cache_type
-        if shared.args.compress_pos_emb != 1:
-            cmd += ["--rope-freq-scale", str(1.0 / shared.args.compress_pos_emb)]
-        if shared.args.rope_freq_base > 0:
-            cmd += ["--rope-freq-base", str(shared.args.rope_freq_base)]
         if shared.args.mmproj not in [None, 'None']:
             path = Path(shared.args.mmproj)
             if not path.exists():
diff --git a/modules/loaders.py b/modules/loaders.py
index d2ebdbc358..c90f2ebbec 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -15,8 +15,6 @@
         'tensor_split',
         'extra_flags',
         'streaming_llm',
-        'rope_freq_base',
-        'compress_pos_emb',
         'row_split',
         'no_kv_offload',
         'no_mmap',
@@ -41,8 +39,6 @@
     'Transformers': [
         'gpu_split',
         'cpu_memory',
-        'alpha_value',
-        'compress_pos_emb',
         'compute_dtype',
         'quant_type',
         'load_in_8bit',
@@ -320,9 +316,6 @@ def list_model_elements():
         'extra_flags',
         'streaming_llm',
         'gpu_split',
-        'alpha_value',
-        'rope_freq_base',
-        'compress_pos_emb',
         'compute_dtype',
         'quant_type',
         'load_in_8bit',
diff --git a/modules/models_settings.py b/modules/models_settings.py
index 25a352379c..f3c9a9866f 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -15,9 +15,6 @@ def get_fallback_settings():
     return {
         'bf16': False,
         'ctx_size': 8192,
-        'rope_freq_base': 0,
-        'compress_pos_emb': 1,
-        'alpha_value': 1,
         'truncation_length': shared.settings['truncation_length'],
         'truncation_length_info': shared.settings['truncation_length'],
         'skip_special_tokens': shared.settings['skip_special_tokens'],
@@ -69,12 +66,6 @@ def get_model_metadata(model):
             if k.endswith('.context_length'):
                 model_settings['ctx_size'] = 0
                 model_settings['truncation_length_info'] = metadata[k]
-            elif k.endswith('rope.freq_base'):
-                model_settings['rope_freq_base'] = metadata[k]
-            elif k.endswith('rope.scale_linear'):
-                model_settings['compress_pos_emb'] = metadata[k]
-            elif k.endswith('rope.scaling.factor'):
-                model_settings['compress_pos_emb'] = metadata[k]
             elif k.endswith('.block_count'):
                 model_settings['gpu_layers'] = -1
                 model_settings['max_gpu_layers'] = metadata[k] + 1
@@ -119,15 +110,6 @@ def get_model_metadata(model):
                 model_settings['ctx_size'] = min(value, 8192)
                 break
 
-            if 'rope_theta' in metadata:
-                model_settings['rope_freq_base'] = metadata['rope_theta']
-            elif 'attn_config' in metadata and 'rope_theta' in metadata['attn_config']:
-                model_settings['rope_freq_base'] = metadata['attn_config']['rope_theta']
-
-            if 'rope_scaling' in metadata and isinstance(metadata['rope_scaling'], dict) and all(key in metadata['rope_scaling'] for key in ('type', 'factor')):
-                if metadata['rope_scaling']['type'] == 'linear':
-                    model_settings['compress_pos_emb'] = metadata['rope_scaling']['factor']
-
             if 'torch_dtype' in metadata and metadata['torch_dtype'] == 'bfloat16':
                 model_settings['bf16'] = True
 
@@ -181,10 +163,6 @@ def get_model_metadata(model):
     if 'instruction_template' not in model_settings:
         model_settings['instruction_template'] = 'Alpaca'
 
-    # Ignore rope_freq_base if set to the default value
-    if 'rope_freq_base' in model_settings and model_settings['rope_freq_base'] == 10000:
-        model_settings.pop('rope_freq_base')
-
     # Apply user settings from user_data/models/config-user.yaml
     settings = shared.user_config
     for pat in settings:
diff --git a/modules/shared.py b/modules/shared.py
index 475d57b71d..354f758996 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -139,12 +139,6 @@
 group.add_argument('--tp-backend', type=str, default='native', help='The backend for tensor parallelism. Valid options: native, nccl. Default: native.')
 group.add_argument('--cfg-cache', action='store_true', help='Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.')
 
-# RoPE
-group = parser.add_argument_group('RoPE')
-group.add_argument('--alpha_value', type=float, default=1, help='Positional embeddings alpha factor for NTK RoPE scaling. Use either this or compress_pos_emb, not both.')
-group.add_argument('--rope_freq_base', type=int, default=0, help='If greater than 0, will be used instead of alpha_value. Those two are related by rope_freq_base = 10000 * alpha_value ^ (64 / 63).')
-group.add_argument('--compress_pos_emb', type=int, default=1, help="Positional embeddings compression factor. Should be set to (context length) / (model\'s original context length). Equal to 1/rope_freq_scale.")
-
 # Gradio
 group = parser.add_argument_group('Gradio')
 group.add_argument('--listen', action='store_true', help='Make the web UI reachable from your local network.')
diff --git a/modules/transformers_loader.py b/modules/transformers_loader.py
index b9918764bc..63758ad7d1 100644
--- a/modules/transformers_loader.py
+++ b/modules/transformers_loader.py
@@ -136,8 +136,6 @@ def load_model_HF(model_name):
         shared.args.load_in_4bit,
         shared.args.disk,
         shared.args.cpu_memory is not None,
-        shared.args.compress_pos_emb > 1,
-        shared.args.alpha_value > 1,
     ])
 
     # Load the model without any special settings
@@ -200,11 +198,6 @@ def load_model_HF(model_name):
             if shared.args.disk:
                 params['offload_folder'] = str(Path(shared.args.disk_cache_dir))
 
-        if shared.args.compress_pos_emb > 1:
-            params['rope_scaling'] = {'type': 'linear', 'factor': shared.args.compress_pos_emb}
-        elif shared.args.alpha_value > 1:
-            params['rope_scaling'] = {'type': 'dynamic', 'factor': shared.args.alpha_value}
-
         logger.info("TRANSFORMERS_PARAMS=")
         pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(params)
         print()
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index b53bc292e7..08fdc83e4f 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -100,9 +100,6 @@ def create_ui():
                                 shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
                                 shared.gradio['extra_flags'] = gr.Textbox(label='extra-flags', info='Additional flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"', value=shared.args.extra_flags)
                                 shared.gradio['cpu_memory'] = gr.Number(label="Maximum CPU memory in GiB. Use this for CPU offloading.", value=shared.args.cpu_memory)
-                                shared.gradio['alpha_value'] = gr.Number(label='alpha_value', value=shared.args.alpha_value, precision=2, info='Positional embeddings alpha factor for NTK RoPE scaling. Recommended values (NTKv1): 1.75 for 1.5x context, 2.5 for 2x context. Use either this or compress_pos_emb, not both.')
-                                shared.gradio['rope_freq_base'] = gr.Number(label='rope_freq_base', value=shared.args.rope_freq_base, precision=0, info='Positional embeddings frequency base for NTK RoPE scaling. Related to alpha_value by rope_freq_base = 10000 * alpha_value ^ (64 / 63). 0 = from model.')
-                                shared.gradio['compress_pos_emb'] = gr.Number(label='compress_pos_emb', value=shared.args.compress_pos_emb, precision=2, info='Positional embeddings compression factor. Should be set to (context length) / (model\'s original context length). Equal to 1/rope_freq_scale.')
                                 shared.gradio['compute_dtype'] = gr.Dropdown(label="compute_dtype", choices=["bfloat16", "float16", "float32"], value=shared.args.compute_dtype, info='Used by load-in-4bit.')
                                 shared.gradio['quant_type'] = gr.Dropdown(label="quant_type", choices=["nf4", "fp4"], value=shared.args.quant_type, info='Used by load-in-4bit.')
 

From 5763cab3c4055122d85974b1bb94ce8aa526ac72 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Mar 2026 07:12:52 -0700
Subject: [PATCH 1385/1701] Fix a crash loading the MiniMax-M2.5 jinja template

---
 modules/chat.py | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index 1ffbb56b8e..bcb548fdcb 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -12,6 +12,7 @@
 from functools import partial
 from pathlib import Path
 
+import markupsafe
 import yaml
 from jinja2.ext import loopcontrols
 from jinja2.sandbox import ImmutableSandboxedEnvironment
@@ -79,6 +80,13 @@ def update_message_metadata(metadata_dict, role, index, **fields):
     lstrip_blocks=True,
     extensions=[loopcontrols]
 )
+
+
+def custom_tojson(value, indent=None, ensure_ascii=True):
+    return markupsafe.Markup(json.dumps(value, indent=indent, ensure_ascii=ensure_ascii))
+
+
+jinja_env.filters["tojson"] = custom_tojson
 jinja_env.globals["strftime_now"] = strftime_now
 
 

From 9119ce0680f3e03fb1e726b2942606ff37d01bd0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Mar 2026 09:22:38 -0700
Subject: [PATCH 1386/1701] llama.cpp: Use `--fit-ctx 8192` when `--fit on` is
 used

This sets the minimum acceptable context length, which by default is 4096.
---
 modules/llama_cpp_server.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 05c07748e0..c3a8d1052b 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -378,6 +378,7 @@ def _start_server(self):
             cmd += ["--gpu-layers", str(shared.args.gpu_layers), "--fit", "off"]
         else:
             cmd += ["--fit", "on"]
+            cmd += ["--fit-ctx", "8192"]
             if shared.args.fit_target:
                 cmd += ["--fit-target", shared.args.fit_target]
 

From 80d0c03bab226baf9058a7ae62db2a5f2e7eebca Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Mar 2026 09:29:25 -0700
Subject: [PATCH 1387/1701] llama.cpp: Change the default `--fit-target` from
 1024 to 512

---
 modules/shared.py        | 2 +-
 modules/ui_model_menu.py | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/shared.py b/modules/shared.py
index 354f758996..a82b20187e 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -108,7 +108,7 @@
 group.add_argument('--threads-batch', type=int, default=0, help='Number of threads to use for batches/prompt processing.')
 group.add_argument('--numa', action='store_true', help='Activate NUMA task allocation for llama.cpp.')
 group.add_argument('--parallel', type=int, default=1, help='Number of parallel request slots. The context size is divided equally among slots. For example, to have 4 slots with 8192 context each, set ctx_size to 32768.')
-group.add_argument('--fit-target', type=str, default='1024', help='Target VRAM margin per device for auto GPU layers, comma-separated list of values in MiB. A single value is broadcast across all devices. Default: 1024.')
+group.add_argument('--fit-target', type=str, default='512', help='Target VRAM margin per device for auto GPU layers, comma-separated list of values in MiB. A single value is broadcast across all devices.')
 group.add_argument('--extra-flags', type=str, default=None, help='Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"')
 
 # Transformers/Accelerate
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 08fdc83e4f..5cf0155dbc 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -46,7 +46,7 @@ def create_ui():
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
-                            shared.gradio['fit_target'] = gr.Textbox(label='fit-target', value=shared.args.fit_target, info='Target VRAM margin per device for auto GPU layers (MiB). Comma-separated list for multiple devices. Default: 1024.')
+                            shared.gradio['fit_target'] = gr.Textbox(label='fit-target', value=shared.args.fit_target, info='Target VRAM margin per device for auto GPU layers (MiB). Comma-separated list for multiple devices.')
                             shared.gradio['tp_backend'] = gr.Dropdown(label="tp-backend", choices=['native', 'nccl'], value=shared.args.tp_backend, info='The backend for tensor parallelism.')
 
                         with gr.Column():

From bfea49b197f053428fa8eca0d9f692141f4db623 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Mar 2026 09:34:17 -0700
Subject: [PATCH 1388/1701] Move top_p and top_k higher up in the UI and CLI
 help

---
 modules/presets.py       | 4 ++--
 modules/shared.py        | 8 ++++----
 modules/ui_parameters.py | 4 ++--
 3 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/modules/presets.py b/modules/presets.py
index b53195ee48..560e0b77df 100644
--- a/modules/presets.py
+++ b/modules/presets.py
@@ -16,9 +16,10 @@
     'dynatemp_exponent': 1,
     'smoothing_factor': 0,
     'smoothing_curve': 1,
-    'min_p': 0,
     'top_p': 1,
     'top_k': 0,
+    'min_p': 0,
+    'top_n_sigma': 0,
     'typical_p': 1,
     'xtc_threshold': 0.1,
     'xtc_probability': 0,
@@ -26,7 +27,6 @@
     'eta_cutoff': 0,
     'tfs': 1,
     'top_a': 0,
-    'top_n_sigma': 0,
     'adaptive_target': 0,
     'adaptive_decay': 0.9,
     'dry_multiplier': 0,
diff --git a/modules/shared.py b/modules/shared.py
index a82b20187e..329114bbdb 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -175,9 +175,10 @@
 group.add_argument('--dynatemp-exponent', type=float, default=_d['dynatemp_exponent'], metavar='N', help='Dynamic temperature exponent')
 group.add_argument('--smoothing-factor', type=float, default=_d['smoothing_factor'], metavar='N', help='Smoothing factor')
 group.add_argument('--smoothing-curve', type=float, default=_d['smoothing_curve'], metavar='N', help='Smoothing curve')
-group.add_argument('--min-p', type=float, default=_d['min_p'], metavar='N', help='Min P')
 group.add_argument('--top-p', type=float, default=_d['top_p'], metavar='N', help='Top P')
 group.add_argument('--top-k', type=int, default=_d['top_k'], metavar='N', help='Top K')
+group.add_argument('--min-p', type=float, default=_d['min_p'], metavar='N', help='Min P')
+group.add_argument('--top-n-sigma', type=float, default=_d['top_n_sigma'], metavar='N', help='Top N Sigma')
 group.add_argument('--typical-p', type=float, default=_d['typical_p'], metavar='N', help='Typical P')
 group.add_argument('--xtc-threshold', type=float, default=_d['xtc_threshold'], metavar='N', help='XTC threshold')
 group.add_argument('--xtc-probability', type=float, default=_d['xtc_probability'], metavar='N', help='XTC probability')
@@ -185,7 +186,6 @@
 group.add_argument('--eta-cutoff', type=float, default=_d['eta_cutoff'], metavar='N', help='Eta cutoff')
 group.add_argument('--tfs', type=float, default=_d['tfs'], metavar='N', help='TFS')
 group.add_argument('--top-a', type=float, default=_d['top_a'], metavar='N', help='Top A')
-group.add_argument('--top-n-sigma', type=float, default=_d['top_n_sigma'], metavar='N', help='Top N Sigma')
 group.add_argument('--adaptive-target', type=float, default=_d['adaptive_target'], metavar='N', help='Adaptive target')
 group.add_argument('--adaptive-decay', type=float, default=_d['adaptive_decay'], metavar='N', help='Adaptive decay')
 group.add_argument('--dry-multiplier', type=float, default=_d['dry_multiplier'], metavar='N', help='DRY multiplier')
@@ -292,9 +292,10 @@
     'smoothing_curve': neutral_samplers['smoothing_curve'],
 
     # Generation parameters - Curve cutoff
-    'min_p': neutral_samplers['min_p'],
     'top_p': 0.95,
     'top_k': neutral_samplers['top_k'],
+    'min_p': neutral_samplers['min_p'],
+    'top_n_sigma': neutral_samplers['top_n_sigma'],
     'typical_p': neutral_samplers['typical_p'],
     'xtc_threshold': neutral_samplers['xtc_threshold'],
     'xtc_probability': neutral_samplers['xtc_probability'],
@@ -302,7 +303,6 @@
     'eta_cutoff': neutral_samplers['eta_cutoff'],
     'tfs': neutral_samplers['tfs'],
     'top_a': neutral_samplers['top_a'],
-    'top_n_sigma': neutral_samplers['top_n_sigma'],
     'adaptive_target': neutral_samplers['adaptive_target'],
     'adaptive_decay': neutral_samplers['adaptive_decay'],
 
diff --git a/modules/ui_parameters.py b/modules/ui_parameters.py
index a5afd7e536..5411b29427 100644
--- a/modules/ui_parameters.py
+++ b/modules/ui_parameters.py
@@ -37,10 +37,10 @@ def create_ui():
                             shared.gradio['dynamic_temperature'] = gr.Checkbox(value=shared.settings['dynamic_temperature'], label='dynamic_temperature')
 
                             gr.Markdown('## Curve cutoff')
-                            shared.gradio['min_p'] = gr.Slider(0.0, 1.0, value=shared.settings['min_p'], step=0.01, label='min_p')
-                            shared.gradio['top_n_sigma'] = gr.Slider(0.0, 5.0, value=shared.settings['top_n_sigma'], step=0.01, label='top_n_sigma')
                             shared.gradio['top_p'] = gr.Slider(0.0, 1.0, value=shared.settings['top_p'], step=0.01, label='top_p')
                             shared.gradio['top_k'] = gr.Slider(0, 200, value=shared.settings['top_k'], step=1, label='top_k')
+                            shared.gradio['min_p'] = gr.Slider(0.0, 1.0, value=shared.settings['min_p'], step=0.01, label='min_p')
+                            shared.gradio['top_n_sigma'] = gr.Slider(0.0, 5.0, value=shared.settings['top_n_sigma'], step=0.01, label='top_n_sigma')
                             shared.gradio['typical_p'] = gr.Slider(0.0, 1.0, value=shared.settings['typical_p'], step=0.01, label='typical_p')
                             shared.gradio['xtc_threshold'] = gr.Slider(0, 0.5, value=shared.settings['xtc_threshold'], step=0.01, label='xtc_threshold', info='If 2 or more tokens have probability above this threshold, consider removing all but the last one.')
                             shared.gradio['xtc_probability'] = gr.Slider(0, 1, value=shared.settings['xtc_probability'], step=0.01, label='xtc_probability', info='Probability that the removal will actually happen. 0 disables the sampler. 1 makes it always happen.')

From 1a2b84093837fefc4aeb8ddc69923ffed8548158 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Mar 2026 09:52:31 -0700
Subject: [PATCH 1389/1701] UI: Fix scroll jump when toggling thinking blocks
 during streaming

---
 js/global_scope_js.js | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index ba5abcb24a..92f65622ff 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -420,7 +420,9 @@ function applyMorphdomUpdate(data) {
             }, 0);
           }
         }
+        autoScrollToBottom();
         updateInstructPadding();
+        autoScrollToBottom();
         // Restore scroll state so the browser's layout adjustment
         // from the toggle doesn't disable auto-scroll
         window.isScrolled = wasScrolled;

From f6a749a151a20b0d9027ab7a29797ca5d58a313c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Mar 2026 10:17:31 -0700
Subject: [PATCH 1390/1701] API: Fix /v1/models to only list the currently
 loaded model

---
 extensions/openai/models.py | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/extensions/openai/models.py b/extensions/openai/models.py
index 82c65093c8..c879a860f7 100644
--- a/extensions/openai/models.py
+++ b/extensions/openai/models.py
@@ -20,10 +20,14 @@ def list_models():
 
 def list_models_openai_format():
     """Returns model list in OpenAI API format"""
-    model_names = get_available_models()
+    if shared.model_name and shared.model_name != 'None':
+        data = [model_info_dict(shared.model_name)]
+    else:
+        data = []
+
     return {
         "object": "list",
-        "data": [model_info_dict(name) for name in model_names]
+        "data": data
     }
 
 

From 92d376e420cba87633cc84c16fcfb8d0c4b0db46 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Mar 2026 13:14:53 -0700
Subject: [PATCH 1391/1701] web_search: Return all results and improve URL
 extraction

---
 modules/web_search.py         | 33 +++++++++++++++++++--------------
 user_data/tools/web_search.py |  4 +---
 2 files changed, 20 insertions(+), 17 deletions(-)

diff --git a/modules/web_search.py b/modules/web_search.py
index 6d005496fe..9bebc846c1 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -4,10 +4,9 @@
 import random
 import re
 import socket
-import urllib.request
 from concurrent.futures import as_completed
 from datetime import datetime
-from urllib.parse import quote_plus, urljoin, urlparse
+from urllib.parse import parse_qs, quote_plus, urljoin, urlparse
 
 import requests
 
@@ -87,22 +86,28 @@ def perform_web_search(query, num_pages=3, max_workers=5, timeout=10, fetch_cont
             "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36"
         ]
 
-        response_text = ""
-        req = urllib.request.Request(search_url, headers={'User-Agent': random.choice(agents)})
-        with urllib.request.urlopen(req, timeout=timeout) as response:
-            response_text = response.read().decode('utf-8')
+        response = requests.get(search_url, headers={'User-Agent': random.choice(agents)}, timeout=timeout)
+        response.raise_for_status()
+        response_text = response.text
 
-        # Extract results with regex
-        titles = re.findall(r']*class="[^"]*result__a[^"]*"[^>]*>(.*?)', response_text, re.DOTALL)
-        urls = re.findall(r']*class="[^"]*result__url[^"]*"[^>]*>(.*?)', response_text, re.DOTALL)
+        # Extract results - title and URL come from the same  element
+        result_links = re.findall(r']*class="[^"]*result__a[^"]*"[^>]*>(.*?)', response_text, re.DOTALL)
+        result_tags = re.findall(r']*class="[^"]*result__a[^"]*"[^>]*)>', response_text, re.DOTALL)
 
         # Prepare download tasks
         download_tasks = []
-        for i in range(min(len(titles), len(urls), num_pages)):
-            url = f"https://{urls[i].strip()}"
-            title = re.sub(r'<[^>]+>', '', titles[i]).strip()
-            title = html.unescape(title)
-            download_tasks.append((url, title, i))
+        for i, (tag_attrs, raw_title) in enumerate(zip(result_tags, result_links)):
+            if num_pages is not None and i >= num_pages:
+                break
+            # Extract href and resolve the actual URL from DuckDuckGo's redirect link
+            href_match = re.search(r'href="([^"]*)"', tag_attrs)
+            if not href_match:
+                continue
+            uddg = parse_qs(urlparse(html.unescape(href_match.group(1))).query).get('uddg', [''])[0]
+            if not uddg:
+                continue
+            title = html.unescape(re.sub(r'<[^>]+>', '', raw_title).strip())
+            download_tasks.append((uddg, title, len(download_tasks)))
 
         search_results = [None] * len(download_tasks)  # Pre-allocate to maintain order
 
diff --git a/user_data/tools/web_search.py b/user_data/tools/web_search.py
index 30d13473a2..6c2b0f0b6f 100644
--- a/user_data/tools/web_search.py
+++ b/user_data/tools/web_search.py
@@ -9,7 +9,6 @@
             "type": "object",
             "properties": {
                 "query": {"type": "string", "description": "The search query."},
-                "num_pages": {"type": "integer", "description": "Number of search results to return (default: 3)."},
             },
             "required": ["query"]
         }
@@ -19,8 +18,7 @@
 
 def execute(arguments):
     query = arguments.get("query", "")
-    num_pages = arguments.get("num_pages", 3)
-    results = perform_web_search(query, num_pages=num_pages, fetch_content=False)
+    results = perform_web_search(query, num_pages=None, fetch_content=False)
     output = []
     for r in results:
         if r:

From f8ff7cf99e192d01f9f29bea943e452f2d60fdfa Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Mar 2026 14:12:59 -0700
Subject: [PATCH 1392/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 14 files changed, 28 insertions(+), 28 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index dca686d9bc..06d4e3d2ff 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -31,8 +31,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 37cbf729f1..368ffee1f1 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index fed46240a5..660a841b46 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index fac36437eb..ddb0efb571 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index c86caf37fc..0c0c2416aa 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 4f5891da16..2ec1e61eb1 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 0471cc7339..2bfd2587a5 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index dfefce206e..36a7d1850d 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 5c032e6b8c..fd68b53307 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 385ecedf7c..4d5b967815 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index d8f7d494e9..34ca382b3e 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index adc6a065db..b492acc153 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 942f7a2ad1..4b548daeae 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index fca722fdc6..0770b12e0b 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio-4.37.2+custom.10-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.10/gradio_client-1.0.2+custom.10-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From 4f80b208597e8b3526155c8b94710a84e9138fa1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Mar 2026 16:38:54 -0700
Subject: [PATCH 1393/1701] UI: Follow-up to beab346f (fix scroll deadlock on
 chat-parent)

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 30dd28aba3..b713507793 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1680,7 +1680,7 @@ button:focus {
 .chat-parent {
     /* Optimize for scrolling performance */
     will-change: scroll-position;
-    contain: layout style paint;
+    contain: style paint;
 
     /* Ensure GPU acceleration */
     transform: translateZ(0);

From c0de1d176cb054b783290da530baa0c0d66a54d5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Mar 2026 17:51:42 -0700
Subject: [PATCH 1394/1701] UI: Add an incognito chat option

---
 css/main.css       | 43 ++++++++++++++++++++++++++++++++++++++
 js/main.js         | 32 ++++++++++++++++++++++++++++
 modules/chat.py    | 52 +++++++++++++++++++++++++++++++++++-----------
 modules/ui_chat.py |  7 ++++++-
 4 files changed, 121 insertions(+), 13 deletions(-)

diff --git a/css/main.css b/css/main.css
index b713507793..25ae15b156 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1022,6 +1022,49 @@ audio {
     padding-right: 0.5rem;
 }
 
+#new-chat-wrapper {
+    display: contents;
+}
+
+.new-chat-arrow {
+    cursor: pointer;
+    position: relative;
+    padding: 0;
+    margin-right: -15px;
+    height: 39.594px;
+    display: flex;
+    align-items: center;
+}
+
+.new-chat-menu {
+    display: none;
+    position: absolute;
+    top: 0;
+    left: 0;
+    padding-top: 1.2em;
+    z-index: var(--layer-top);
+    white-space: nowrap;
+}
+
+.new-chat-arrow:hover .new-chat-menu {
+    display: block;
+}
+
+.new-chat-menu-item {
+    cursor: pointer;
+    padding: var(--size-2);
+    background: var(--background-fill-primary);
+    box-shadow: var(--shadow-drop-lg);
+    border-radius: var(--container-radius);
+    color: var(--body-text-color);
+    font-size: var(--text-md);
+    font-weight: var(--button-large-text-weight);
+}
+
+.new-chat-menu-item:hover {
+    background: var(--background-fill-secondary);
+}
+
 #past-chats-row,
 #chat-controls {
     width: 260px;
diff --git a/js/main.js b/js/main.js
index 23db59d54b..f05f93c600 100644
--- a/js/main.js
+++ b/js/main.js
@@ -552,6 +552,38 @@ document.querySelectorAll(".focus-on-chat-input").forEach(element => {
   });
 });
 
+//------------------------------------------------
+// "New chat" hover menu with incognito option
+//------------------------------------------------
+
+(function() {
+  const newChatBtn = document.getElementById("new-chat-btn");
+
+  const wrapper = document.createElement("div");
+  wrapper.id = "new-chat-wrapper";
+  newChatBtn.replaceWith(wrapper);
+  wrapper.appendChild(newChatBtn);
+
+  const arrow = document.createElement("span");
+  arrow.className = "new-chat-arrow";
+  arrow.textContent = "\u25BE";
+
+  const menu = document.createElement("div");
+  menu.className = "new-chat-menu";
+  const option = document.createElement("div");
+  option.className = "new-chat-menu-item";
+  option.textContent = "Incognito chat";
+  menu.appendChild(option);
+
+  arrow.appendChild(menu);
+  wrapper.appendChild(arrow);
+
+  option.addEventListener("click", function(e) {
+    e.stopPropagation();
+    document.querySelector("#incognito-chat-btn").click();
+  });
+})();
+
 //------------------------------------------------
 // Fix a border around the "past chats" menu
 //------------------------------------------------
diff --git a/modules/chat.py b/modules/chat.py
index bcb548fdcb..e4fcaabe4b 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1528,7 +1528,7 @@ def redraw_html(history, name1, name2, mode, style, character, reset_cache=False
     return chat_html_wrapper(history, name1, name2, mode, style, character, reset_cache=reset_cache)
 
 
-def start_new_chat(state):
+def start_new_chat(state, unique_id=None):
     mode = state['mode']
     # Initialize with empty metadata dictionary
     history = {'internal': [], 'visible': [], 'metadata': {}}
@@ -1542,7 +1542,9 @@ def start_new_chat(state):
             # Add timestamp for assistant's greeting
             update_message_metadata(history['metadata'], "assistant", 0, timestamp=get_current_timestamp())
 
-    unique_id = datetime.now().strftime('%Y%m%d-%H-%M-%S')
+    if unique_id is None:
+        unique_id = datetime.now().strftime('%Y%m%d-%H-%M-%S')
+
     save_history(history, unique_id, state['character_menu'], state['mode'])
 
     return history
@@ -1561,6 +1563,9 @@ def save_history(history, unique_id, character, mode):
     if shared.args.multi_user:
         return
 
+    if unique_id and unique_id.startswith('incognito-'):
+        return
+
     p = get_history_file_path(unique_id, character, mode)
     if not p.parent.is_dir():
         p.parent.mkdir(parents=True)
@@ -1750,6 +1755,9 @@ def save_last_chat_state(character, mode, unique_id):
     if shared.args.multi_user:
         return
 
+    if unique_id and unique_id.startswith('incognito-'):
+        return
+
     state = load_last_chat_state()
     key = get_chat_state_key(character, mode)
     state["last_chats"][key] = unique_id
@@ -2290,11 +2298,29 @@ def handle_start_new_chat_click(state):
     return [history, html, past_chats_update]
 
 
-def handle_delete_chat_confirm_click(state):
+def handle_start_incognito_chat_click(state):
     import gradio as gr
+    unique_id = 'incognito-' + datetime.now().strftime('%Y%m%d-%H-%M-%S')
+    history = start_new_chat(state, unique_id=unique_id)
+    html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
+
+    convert_to_markdown.cache_clear()
+
+    histories = find_all_histories_with_first_prompts(state)
+    past_chats_update = gr.update(choices=histories, value=unique_id)
+
+    return [history, html, past_chats_update]
+
+
+def handle_delete_chat_confirm_click(state):
     filtered_histories = find_all_histories_with_first_prompts(state)
     filtered_ids = [h[1] for h in filtered_histories]
-    index = str(filtered_ids.index(state['unique_id']))
+
+    if state['unique_id'] not in filtered_ids:
+        # Incognito or unknown chat — just load the most recent saved chat
+        index = '0'
+    else:
+        index = str(filtered_ids.index(state['unique_id']))
 
     delete_history(state['unique_id'], state['character_menu'], state['mode'])
     history, unique_id = load_history_after_deletion(state, index)
@@ -2302,13 +2328,7 @@ def handle_delete_chat_confirm_click(state):
 
     convert_to_markdown.cache_clear()
 
-    return [
-        history,
-        html,
-        unique_id,
-        gr.update(visible=False),
-        gr.update(visible=True),
-    ]
+    return [history, html, unique_id]
 
 
 def handle_branch_chat_click(state):
@@ -2324,7 +2344,8 @@ def handle_branch_chat_click(state):
         if 'metadata' in history:
             history['metadata'] = {k: v for k, v in history['metadata'].items() if int(k.split('_')[-1]) <= branch_from_index}
 
-    new_unique_id = datetime.now().strftime('%Y%m%d-%H-%M-%S')
+    prefix = 'incognito-' if state['unique_id'] and state['unique_id'].startswith('incognito-') else ''
+    new_unique_id = prefix + datetime.now().strftime('%Y%m%d-%H-%M-%S')
     save_history(history, new_unique_id, state['character_menu'], state['mode'])
 
     histories = find_all_histories_with_first_prompts(state)
@@ -2446,6 +2467,13 @@ def handle_rename_chat_click():
 
 def handle_rename_chat_confirm(rename_to, state):
     import gradio as gr
+
+    if state['unique_id'] and state['unique_id'].startswith('incognito-'):
+        return [
+            gr.update(),
+            gr.update(visible=False),
+        ]
+
     rename_history(state['unique_id'], rename_to, state['character_menu'], state['mode'])
     histories = find_all_histories_with_first_prompts(state)
 
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 0acf9c04bd..d2a515b84d 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -28,7 +28,8 @@ def create_ui():
                     shared.gradio['branch_chat'] = gr.Button('Branch', elem_classes=['refresh-button', 'refresh-button-medium'], elem_id='Branch', interactive=not mu)
                     shared.gradio['rename_chat'] = gr.Button('Rename', elem_classes=['refresh-button', 'refresh-button-medium'], interactive=not mu)
                     shared.gradio['delete_chat'] = gr.Button('🗑️', visible=False, elem_classes='refresh-button', interactive=not mu, elem_id='delete_chat')
-                    shared.gradio['Start new chat'] = gr.Button('New chat', elem_classes=['refresh-button', 'refresh-button-medium', 'focus-on-chat-input'])
+                    shared.gradio['Start new chat'] = gr.Button('New chat', elem_classes=['refresh-button', 'refresh-button-medium', 'focus-on-chat-input'], elem_id='new-chat-btn')
+                    shared.gradio['Start incognito chat'] = gr.Button('Incognito chat', visible=False, elem_id='incognito-chat-btn')
                     shared.gradio['branch_index'] = gr.Number(value=-1, precision=0, visible=False, elem_id="Branch-index", interactive=True)
 
                 shared.gradio['search_chat'] = gr.Textbox(placeholder='Search chats...', max_lines=1, elem_id='search_chat')
@@ -290,6 +291,10 @@ def create_event_handlers():
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_start_new_chat_click, gradio('interface_state'), gradio('history', 'display', 'unique_id'), show_progress=False)
 
+    shared.gradio['Start incognito chat'].click(
+        ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
+        chat.handle_start_incognito_chat_click, gradio('interface_state'), gradio('history', 'display', 'unique_id'), show_progress=False)
+
     shared.gradio['delete_chat-confirm'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_delete_chat_confirm_click, gradio('interface_state'), gradio('history', 'display', 'unique_id'), show_progress=False)

From b76a289e048b9c217d54397705c0e0f5faf15dbe Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Mar 2026 22:03:29 -0300
Subject: [PATCH 1395/1701] API: Respect --listen-host for the OpenAI API
 server

Closes #7429
---
 extensions/openai/script.py | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/extensions/openai/script.py b/extensions/openai/script.py
index f161e1e453..a0d5deb84e 100644
--- a/extensions/openai/script.py
+++ b/extensions/openai/script.py
@@ -458,10 +458,13 @@ def run_server():
 
     # In the server configuration:
     server_addrs = []
-    if os.environ.get('OPENEDAI_ENABLE_IPV6', shared.args.api_enable_ipv6):
-        server_addrs.append('[::]' if shared.args.listen else '[::1]')
-    if not os.environ.get('OPENEDAI_DISABLE_IPV4', shared.args.api_disable_ipv4):
-        server_addrs.append('0.0.0.0' if shared.args.listen else '127.0.0.1')
+    if shared.args.listen and shared.args.listen_host:
+        server_addrs.append(shared.args.listen_host)
+    else:
+        if os.environ.get('OPENEDAI_ENABLE_IPV6', shared.args.api_enable_ipv6):
+            server_addrs.append('[::]' if shared.args.listen else '[::1]')
+        if not os.environ.get('OPENEDAI_DISABLE_IPV4', shared.args.api_disable_ipv4):
+            server_addrs.append('0.0.0.0' if shared.args.listen else '127.0.0.1')
 
     if not server_addrs:
         raise Exception('you MUST enable IPv6 or IPv4 for the API to work')

From 5cfe9fe2954ad45fe8d485d20baeb29546e040e7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Mar 2026 20:05:56 -0700
Subject: [PATCH 1396/1701] Update README

---
 README.md | 20 ++++++++++++--------
 1 file changed, 12 insertions(+), 8 deletions(-)

diff --git a/README.md b/README.md
index f152717657..c3450bf801 100644
--- a/README.md
+++ b/README.md
@@ -13,7 +13,7 @@
 
 # Text Generation Web UI
 
-Run large language models locally with full privacy. Supports text generation, vision, image generation, training, tool-calling, and more — across multiple backends including [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). 100% offline, zero telemetry.
+A Gradio web UI for running Large Language Models locally. 100% private and offline. Supports text generation, vision, tool-calling, training, image generation, and more.
 
 [Try the Deep Reason extension](https://oobabooga.gumroad.com/l/deep_reason)
 
@@ -24,12 +24,12 @@ Run large language models locally with full privacy. Supports text generation, v
 ## Features
 
 - **Multiple backends**: [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). Switch between backends and models without restarting.
+- **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
 - **Vision (multimodal)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
-- **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
-- **Training**: Fine-tune LoRAs on multi-turn chat or raw text datasets. Supports resuming interrupted runs ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/05-%E2%80%90-Training-Tab)).
-- **OpenAI-compatible API**: Chat and Completions endpoints with tool-calling support — use as a local drop-in replacement for the OpenAI API ([examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples)).
 - **Web search**: Search the internet with LLM-generated queries to add context to conversations.
-- **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
+- **OpenAI-compatible API**: Chat and Completions endpoints with tool-calling support — use as a local drop-in replacement for the OpenAI API ([examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples)).
+- **Training**: Fine-tune LoRAs on multi-turn chat or raw text datasets. Supports resuming interrupted runs ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/05-%E2%80%90-Training-Tab)).
+- **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
 - **Easy setup**: [Portable builds](https://github.com/oobabooga/text-generation-webui/releases) (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or a one-click installer for the full feature set.
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - `instruct` mode for instruction-following (like ChatGPT), and `chat-instruct`/`chat` modes for talking to custom characters. Prompts are automatically formatted with Jinja2 templates.
@@ -46,10 +46,11 @@ Run large language models locally with full privacy. Supports text generation, v
 
 No installation needed – just download, unzip and run. All dependencies included.
 
-Compatible with GGUF (llama.cpp) models on Windows, Linux, and macOS. [Check what models fit your hardware](https://huggingface.co/spaces/oobabooga/accurate-gguf-vram-calculator).
-
 Download from here: **https://github.com/oobabooga/text-generation-webui/releases**
 
+- Builds are provided for Linux, Windows, and macOS, with options for CUDA, Vulkan, ROCm, and CPU-only.
+- Compatible with GGUF (llama.cpp) models.
+
 #### Option 2: Manual portable install with venv
 
 Very fast setup that should work on any Python 3.9+:
@@ -423,7 +424,10 @@ API generation defaults:
 
 ## Downloading models
 
-Download a GGUF model file from [Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads&search=gguf) and place it in the `user_data/models` folder. That's it — the UI will detect it automatically.
+1. Download a GGUF model file from [Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads&search=gguf).
+2. Place it in the `user_data/models` folder.
+
+That's it. The UI will detect it automatically.
 
 Not sure what will fit your GPU? Use the [VRAM Calculator](https://huggingface.co/spaces/oobabooga/accurate-gguf-vram-calculator).
 

From 9d9f5d98604da58e79911a98b6a714f968b27ac9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 15 Mar 2026 20:27:44 -0700
Subject: [PATCH 1397/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index c3450bf801..b75e2c1169 100644
--- a/README.md
+++ b/README.md
@@ -26,7 +26,7 @@ A Gradio web UI for running Large Language Models locally. 100% private and offl
 - **Multiple backends**: [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). Switch between backends and models without restarting.
 - **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
 - **Vision (multimodal)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
-- **Web search**: Search the internet with LLM-generated queries to add context to conversations.
+- **Tool-calling**: Models can call custom functions during chat — web search, page fetching, math, and more. Each tool is a single `.py` file, easy to create and extend ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Tool-Calling-Tutorial)).
 - **OpenAI-compatible API**: Chat and Completions endpoints with tool-calling support — use as a local drop-in replacement for the OpenAI API ([examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples)).
 - **Training**: Fine-tune LoRAs on multi-turn chat or raw text datasets. Supports resuming interrupted runs ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/05-%E2%80%90-Training-Tab)).
 - **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).

From 50685c93f2e273fd8dd46b1f790736e62f61f0d8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Mar 2026 05:29:27 -0700
Subject: [PATCH 1398/1701] Update README

---
 README.md | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/README.md b/README.md
index b75e2c1169..989659d191 100644
--- a/README.md
+++ b/README.md
@@ -27,7 +27,7 @@ A Gradio web UI for running Large Language Models locally. 100% private and offl
 - **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
 - **Vision (multimodal)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
 - **Tool-calling**: Models can call custom functions during chat — web search, page fetching, math, and more. Each tool is a single `.py` file, easy to create and extend ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Tool-Calling-Tutorial)).
-- **OpenAI-compatible API**: Chat and Completions endpoints with tool-calling support — use as a local drop-in replacement for the OpenAI API ([examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples)).
+- **OpenAI-compatible API**: Chat and Completions endpoints with tool-calling support. Use as a local drop-in replacement for the OpenAI API ([examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples)).
 - **Training**: Fine-tune LoRAs on multi-turn chat or raw text datasets. Supports resuming interrupted runs ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/05-%E2%80%90-Training-Tab)).
 - **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
 - **Easy setup**: [Portable builds](https://github.com/oobabooga/text-generation-webui/releases) (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or a one-click installer for the full feature set.
@@ -81,7 +81,7 @@ deactivate
 
 #### Option 3: One-click installer
 
-For users who need additional backends (ExLlamaV3, Transformers) or extensions (TTS, voice input, translation, etc). Requires ~10GB disk space and downloads PyTorch.
+For users who need additional backends (ExLlamaV3, Transformers), training, image generation, or extensions (TTS, voice input, translation, etc). Requires ~10GB disk space and downloads PyTorch.
 
 1. Clone the repository, or [download its source code](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip) and extract it.
 2. Run the startup script for your OS: `start_windows.bat`, `start_linux.sh`, or `start_macos.sh`.
@@ -429,7 +429,7 @@ API generation defaults:
 
 That's it. The UI will detect it automatically.
 
-Not sure what will fit your GPU? Use the [VRAM Calculator](https://huggingface.co/spaces/oobabooga/accurate-gguf-vram-calculator).
+To check what will fit your GPU, you can use the [VRAM Calculator](https://huggingface.co/spaces/oobabooga/accurate-gguf-vram-calculator).
 
 
            
 Other model types (Transformers, EXL3)

From 737ded695913de99db88d90384bde552d297dde9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Mar 2026 05:37:46 -0700
Subject: [PATCH 1399/1701] Web search: Fix SSRF validation to block all
 non-global IPs

---
 modules/web_search.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/web_search.py b/modules/web_search.py
index 9bebc846c1..e13ef62ad8 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -28,8 +28,8 @@ def _validate_url(url):
     try:
         for family, _, _, _, sockaddr in socket.getaddrinfo(hostname, None):
             ip = ipaddress.ip_address(sockaddr[0])
-            if ip.is_private or ip.is_loopback or ip.is_link_local or ip.is_reserved:
-                raise ValueError(f"Access to private/internal address {ip} is blocked")
+            if not ip.is_global:
+                raise ValueError(f"Access to non-public address {ip} is blocked")
     except socket.gaierror:
         raise ValueError(f"Could not resolve hostname: {hostname}")
 

From 6c05a964a75af8d6e75c053ee8642eb787151029 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Mar 2026 06:00:16 -0700
Subject: [PATCH 1400/1701] docs: Mention supported tool-calling models

---
 docs/Tool Calling Tutorial.md | 17 ++++++++++++++++-
 1 file changed, 16 insertions(+), 1 deletion(-)

diff --git a/docs/Tool Calling Tutorial.md b/docs/Tool Calling Tutorial.md
index 170bdff7c3..801e9d78b6 100644
--- a/docs/Tool Calling Tutorial.md	
+++ b/docs/Tool Calling Tutorial.md	
@@ -1,8 +1,23 @@
+## Supported models
+
+The following models are supported:
+
+- Qwen 3.5
+- GPT-OSS
+- Mistral Small / Devstral
+- DeepSeek V3
+- Kimi-K2
+- MiniMax-M2.5
+- GLM-5
+- Llama 4
+
+Other models that output tool calls as JSON (inside XML tags, code blocks, or plain JSON) are also supported through a generic fallback parser.
+
 ## Tool calling in the UI
 
 ### 1. Load a model with tool-calling support
 
-Load a model with tool-calling support (Qwen, Mistral, Llama 4, etc.) from the Model tab.
+Load a model with tool-calling support from the Model tab.
 
 ### 2. Select tools
 

From 44810751de7badd8d60677cc9ae3b1469cfdb0ec Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Mar 2026 06:21:14 -0700
Subject: [PATCH 1401/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 12 files changed, 20 insertions(+), 20 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 06d4e3d2ff..c24f4a9df9 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.25/exllamav3-0.0.25+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.25/exllamav3-0.0.25+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 368ffee1f1..7c481224a8 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 660a841b46..b1c8f78eb9 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index ddb0efb571..63ef33ea88 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 0c0c2416aa..4bc61622f8 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 2bfd2587a5..ba4c7a0498 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 36a7d1850d..5dfdd9c840 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index fd68b53307..f62241b36a 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 4d5b967815..353d9172e1 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 34ca382b3e..5f039318f6 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index b492acc153..d8b03102e5 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 0770b12e0b..fd2511f4c6 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.93.0/llama_cpp_binaries-0.93.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 1c89376370b63ad32fef472114ec036edaaf8d1c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Mar 2026 15:23:24 -0700
Subject: [PATCH 1402/1701] training: Add gradient_checkpointing for lower VRAM
 by default

---
 docs/05 - Training Tab.md | 2 ++
 modules/training.py       | 8 +++++---
 2 files changed, 7 insertions(+), 3 deletions(-)

diff --git a/docs/05 - Training Tab.md b/docs/05 - Training Tab.md
index 0bfc59aa40..46424eab16 100644
--- a/docs/05 - Training Tab.md	
+++ b/docs/05 - Training Tab.md	
@@ -100,6 +100,8 @@ Each parameter has a description in the UI. Below is guidance on the most import
 
 VRAM usage during training is roughly similar to inference with ~1000 tokens of context. If you can run the model, you can probably train LoRAs with the default settings. If you run out of VRAM, reduce `Micro Batch Size` or `Cutoff Length`. Training 4-bit quantized models uses more VRAM — set `Micro Batch Size` to `1` to compensate.
 
+**Gradient checkpointing** is enabled by default. It reduces VRAM usage by recomputing activations during the backward pass instead of storing them in memory. The tradeoff is ~20-30% slower training. There is no impact on accuracy — the results are mathematically identical. The savings are most noticeable with longer sequences and larger batch sizes. You can disable it if you have VRAM to spare and want faster training.
+
 ### Rank
 
 Higher rank = more learning capacity = larger adapter = more VRAM. Use 4–8 for style/format, 128–256 to teach factual knowledge.
diff --git a/modules/training.py b/modules/training.py
index 878bb222a8..6549b35e5b 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -26,7 +26,7 @@
 from modules.logging_colors import logger
 from modules.models import reload_model
 
-PARAMETERS = ["lora_name", "always_override", "all_linear", "q_proj_en", "v_proj_en", "k_proj_en", "o_proj_en", "gate_proj_en", "down_proj_en", "up_proj_en", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "text_dataset", "higher_rank_limit", "warmup_steps", "optimizer", "stride_length", "stop_at_loss", "add_eos_token", "excess_length", "report_to"]
+PARAMETERS = ["lora_name", "always_override", "all_linear", "q_proj_en", "v_proj_en", "k_proj_en", "o_proj_en", "gate_proj_en", "down_proj_en", "up_proj_en", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "text_dataset", "higher_rank_limit", "warmup_steps", "optimizer", "stride_length", "stop_at_loss", "add_eos_token", "excess_length", "report_to", "gradient_checkpointing"]
 WANT_INTERRUPT = False
 
 train_log = {}
@@ -101,6 +101,7 @@ def create_ui():
                                 add_eos_token = gr.Checkbox(label='Add EOS token', value=True, info="Adds EOS token for each document in text datasets.")
                                 excess_length = gr.Dropdown(label='Excess length', value='drop', choices=['drop', 'truncate'], info='What to do with conversations that exceed the cutoff length. "Drop" removes them entirely (recommended). "Truncate" cuts from the right, which may produce incomplete responses.', elem_classes=['slim-dropdown'])
 
+                                gradient_checkpointing = gr.Checkbox(label='Gradient checkpointing', value=True, info='Trades ~20-30% training speed for reduced VRAM usage by recomputing activations during the backward pass instead of storing them. No impact on accuracy.')
                                 higher_rank_limit = gr.Checkbox(label='Enable higher ranks', value=False, info='If checked, changes Rank/Alpha slider above to go much higher. This will not work without a datacenter-class GPU.')
                                 report_to = gr.Radio(label="Save detailed logs with", value="None", choices=["None", "wandb", "tensorboard"], interactive=True)
 
@@ -159,7 +160,7 @@ def create_ui():
                 refresh_table = gr.Button('Refresh the table', elem_classes="small-button", interactive=not mu)
 
     # Training events
-    all_params = [lora_name, always_override, all_linear, q_proj_en, v_proj_en, k_proj_en, o_proj_en, gate_proj_en, down_proj_en, up_proj_en, save_steps, micro_batch_size, batch_size, epochs, learning_rate, lr_scheduler_type, lora_rank, lora_alpha, lora_dropout, cutoff_len, dataset, eval_dataset, format, eval_steps, text_dataset, higher_rank_limit, warmup_steps, optimizer, stride_length, stop_at_loss, add_eos_token, excess_length, report_to]
+    all_params = [lora_name, always_override, all_linear, q_proj_en, v_proj_en, k_proj_en, o_proj_en, gate_proj_en, down_proj_en, up_proj_en, save_steps, micro_batch_size, batch_size, epochs, learning_rate, lr_scheduler_type, lora_rank, lora_alpha, lora_dropout, cutoff_len, dataset, eval_dataset, format, eval_steps, text_dataset, higher_rank_limit, warmup_steps, optimizer, stride_length, stop_at_loss, add_eos_token, excess_length, report_to, gradient_checkpointing]
 
     copy_from.change(do_copy_params, [copy_from] + all_params, all_params)
     start_button.click(do_train, all_params, output)
@@ -293,7 +294,7 @@ def calc_trainable_parameters(model):
     return trainable_params, all_param
 
 
-def do_train(lora_name: str, always_override: bool, all_linear: bool, q_proj_en: bool, v_proj_en: bool, k_proj_en: bool, o_proj_en: bool, gate_proj_en: bool, down_proj_en: bool, up_proj_en: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, text_dataset: str, higher_rank_limit: bool, warmup_steps: int, optimizer: str, stride_length: int, stop_at_loss: float, add_eos_token: bool, excess_length: str, report_to: str):
+def do_train(lora_name: str, always_override: bool, all_linear: bool, q_proj_en: bool, v_proj_en: bool, k_proj_en: bool, o_proj_en: bool, gate_proj_en: bool, down_proj_en: bool, up_proj_en: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, text_dataset: str, higher_rank_limit: bool, warmup_steps: int, optimizer: str, stride_length: int, stop_at_loss: float, add_eos_token: bool, excess_length: str, report_to: str, gradient_checkpointing: bool = True):
 
     import torch
     import transformers
@@ -708,6 +709,7 @@ def collate_fn(batch):
             load_best_model_at_end=eval_data is not None,
             # TODO: Enable multi-device support
             ddp_find_unused_parameters=None,
+            gradient_checkpointing=gradient_checkpointing,
             use_cpu=shared.args.cpu,
             remove_unused_columns=False,
         ),

From 22ff5044b0ccd12e2ab1181e4dd3d503a7b0ae2c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Mar 2026 16:01:28 -0700
Subject: [PATCH 1403/1701] training: Organize the UI

---
 modules/training.py | 11 ++++-------
 1 file changed, 4 insertions(+), 7 deletions(-)

diff --git a/modules/training.py b/modules/training.py
index 6549b35e5b..7cb5006875 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -90,19 +90,16 @@ def create_ui():
                     with gr.Accordion(label='Advanced Options', open=False, elem_classes='tgw-accordion'):
                         with gr.Row():
                             with gr.Column():
+                                optimizer = gr.Dropdown(label='Optimizer', value='adamw_torch', choices=['adamw_hf', 'adamw_torch', 'adamw_torch_fused', 'adamw_torch_xla', 'adamw_apex_fused', 'adafactor', 'adamw_bnb_8bit', 'adamw_anyprecision', 'sgd', 'adagrad'], info='Optimizer algorithm. adamw_torch is the standard choice. adamw_bnb_8bit uses less VRAM. adafactor is memory-efficient for large models.', elem_classes=['slim-dropdown'])
+                                warmup_steps = gr.Number(label='Warmup Steps', value=100, info='For this many steps at the start, the learning rate is gradually ramped up from 0 to the target value. This prevents unstable updates early in training.')
                                 lora_dropout = gr.Slider(label='LoRA Dropout', minimum=0.0, maximum=1.0, step=0.025, value=0.0, info='Percentage probability for dropout of LoRA layers. This can help reduce overfitting. Most users should leave at default.')
                                 stop_at_loss = gr.Slider(label='Stop at loss', minimum=0.0, maximum=3.0, step=0.1, value=0.00, info='The process will automatically stop once the desired loss value is reached. (reasonable numbers are 1.5-1.8)')
-                                with gr.Row():
-                                    optimizer = gr.Dropdown(label='Optimizer', value='adamw_torch', choices=['adamw_hf', 'adamw_torch', 'adamw_torch_fused', 'adamw_torch_xla', 'adamw_apex_fused', 'adafactor', 'adamw_bnb_8bit', 'adamw_anyprecision', 'sgd', 'adagrad'], info='Optimizer algorithm. adamw_torch is the standard choice. adamw_bnb_8bit uses less VRAM. adafactor is memory-efficient for large models.', elem_classes=['slim-dropdown'])
 
                             with gr.Column():
-                                warmup_steps = gr.Number(label='Warmup Steps', value=100, info='For this many steps at the start, the learning rate is gradually ramped up from 0 to the target value. This prevents unstable updates early in training.')
-
-                                add_eos_token = gr.Checkbox(label='Add EOS token', value=True, info="Adds EOS token for each document in text datasets.")
-                                excess_length = gr.Dropdown(label='Excess length', value='drop', choices=['drop', 'truncate'], info='What to do with conversations that exceed the cutoff length. "Drop" removes them entirely (recommended). "Truncate" cuts from the right, which may produce incomplete responses.', elem_classes=['slim-dropdown'])
-
                                 gradient_checkpointing = gr.Checkbox(label='Gradient checkpointing', value=True, info='Trades ~20-30% training speed for reduced VRAM usage by recomputing activations during the backward pass instead of storing them. No impact on accuracy.')
+                                add_eos_token = gr.Checkbox(label='Add EOS token', value=True, info="Adds EOS token for each document in text datasets.")
                                 higher_rank_limit = gr.Checkbox(label='Enable higher ranks', value=False, info='If checked, changes Rank/Alpha slider above to go much higher. This will not work without a datacenter-class GPU.')
+                                excess_length = gr.Dropdown(label='Excess length', value='drop', choices=['drop', 'truncate'], info='What to do with conversations that exceed the cutoff length. "Drop" removes them entirely (recommended). "Truncate" cuts from the right, which may produce incomplete responses.', elem_classes=['slim-dropdown'])
                                 report_to = gr.Radio(label="Save detailed logs with", value="None", choices=["None", "wandb", "tensorboard"], interactive=True)
 
                 with gr.Column():

From 238cbd5656a9007d7e4f5ff39a04e1d340b9e50c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Mar 2026 16:05:43 -0700
Subject: [PATCH 1404/1701] training: Remove arbitrary higher_rank_limit
 parameter

---
 modules/training.py | 17 ++++++-----------
 1 file changed, 6 insertions(+), 11 deletions(-)

diff --git a/modules/training.py b/modules/training.py
index 7cb5006875..db7b206b2d 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -26,7 +26,7 @@
 from modules.logging_colors import logger
 from modules.models import reload_model
 
-PARAMETERS = ["lora_name", "always_override", "all_linear", "q_proj_en", "v_proj_en", "k_proj_en", "o_proj_en", "gate_proj_en", "down_proj_en", "up_proj_en", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "text_dataset", "higher_rank_limit", "warmup_steps", "optimizer", "stride_length", "stop_at_loss", "add_eos_token", "excess_length", "report_to", "gradient_checkpointing"]
+PARAMETERS = ["lora_name", "always_override", "all_linear", "q_proj_en", "v_proj_en", "k_proj_en", "o_proj_en", "gate_proj_en", "down_proj_en", "up_proj_en", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "text_dataset", "warmup_steps", "optimizer", "stride_length", "stop_at_loss", "add_eos_token", "excess_length", "report_to", "gradient_checkpointing"]
 WANT_INTERRUPT = False
 
 train_log = {}
@@ -73,8 +73,8 @@ def create_ui():
 
                     with gr.Row():
                         with gr.Column():
-                            lora_rank = gr.Slider(label='LoRA Rank', value=8, minimum=0, maximum=1024, step=4, info='Also called dimension count. Higher values = larger file, more content control. Smaller values = smaller file, less control. Use 4 or 8 for style, 128 or 256 to teach, 1024+ for fine-detail on big data. More VRAM is needed for higher ranks.')
-                            lora_alpha = gr.Slider(label='LoRA Alpha', value=16, minimum=0, maximum=2048, step=4, info='This divided by the rank becomes the scaling of the LoRA. Higher means stronger. A good standard value is twice your Rank.')
+                            lora_rank = gr.Slider(label='LoRA Rank', value=8, minimum=0, maximum=2048, step=4, info='Also called dimension count. Use 4–8 for style/format, 128–256 to teach factual knowledge, 1024+ for comprehensive fine-tuning. Very high ranks require significant VRAM.')
+                            lora_alpha = gr.Slider(label='LoRA Alpha', value=16, minimum=0, maximum=4096, step=4, info='This divided by the rank becomes the scaling of the LoRA. Higher means stronger. A good standard value is twice your Rank.')
                             batch_size = gr.Slider(label='Batch Size', value=32, minimum=0, maximum=1024, step=4, info='Global batch size. The two batch sizes together determine gradient accumulation (gradientAccum = batch / microBatch). Higher gradient accum values lead to better quality training.')
                             micro_batch_size = gr.Slider(label='Micro Batch Size', value=4, minimum=1, maximum=128, step=1, info='Per-device batch size (NOTE: multiple devices not yet implemented). Increasing this will increase VRAM usage.')
                             cutoff_len = gr.Slider(label='Cutoff Length', minimum=0, maximum=4096, value=512, step=32, info='Maximum sequence length in tokens. For instruction datasets, conversations longer than this are dropped. For text datasets, documents are split into chunks of this size. Higher values require more VRAM.')
@@ -98,7 +98,6 @@ def create_ui():
                             with gr.Column():
                                 gradient_checkpointing = gr.Checkbox(label='Gradient checkpointing', value=True, info='Trades ~20-30% training speed for reduced VRAM usage by recomputing activations during the backward pass instead of storing them. No impact on accuracy.')
                                 add_eos_token = gr.Checkbox(label='Add EOS token', value=True, info="Adds EOS token for each document in text datasets.")
-                                higher_rank_limit = gr.Checkbox(label='Enable higher ranks', value=False, info='If checked, changes Rank/Alpha slider above to go much higher. This will not work without a datacenter-class GPU.')
                                 excess_length = gr.Dropdown(label='Excess length', value='drop', choices=['drop', 'truncate'], info='What to do with conversations that exceed the cutoff length. "Drop" removes them entirely (recommended). "Truncate" cuts from the right, which may produce incomplete responses.', elem_classes=['slim-dropdown'])
                                 report_to = gr.Radio(label="Save detailed logs with", value="None", choices=["None", "wandb", "tensorboard"], interactive=True)
 
@@ -157,12 +156,12 @@ def create_ui():
                 refresh_table = gr.Button('Refresh the table', elem_classes="small-button", interactive=not mu)
 
     # Training events
-    all_params = [lora_name, always_override, all_linear, q_proj_en, v_proj_en, k_proj_en, o_proj_en, gate_proj_en, down_proj_en, up_proj_en, save_steps, micro_batch_size, batch_size, epochs, learning_rate, lr_scheduler_type, lora_rank, lora_alpha, lora_dropout, cutoff_len, dataset, eval_dataset, format, eval_steps, text_dataset, higher_rank_limit, warmup_steps, optimizer, stride_length, stop_at_loss, add_eos_token, excess_length, report_to, gradient_checkpointing]
+    all_params = [lora_name, always_override, all_linear, q_proj_en, v_proj_en, k_proj_en, o_proj_en, gate_proj_en, down_proj_en, up_proj_en, save_steps, micro_batch_size, batch_size, epochs, learning_rate, lr_scheduler_type, lora_rank, lora_alpha, lora_dropout, cutoff_len, dataset, eval_dataset, format, eval_steps, text_dataset, warmup_steps, optimizer, stride_length, stop_at_loss, add_eos_token, excess_length, report_to, gradient_checkpointing]
 
     copy_from.change(do_copy_params, [copy_from] + all_params, all_params)
     start_button.click(do_train, all_params, output)
     stop_button.click(do_interrupt, None, None, queue=False)
-    higher_rank_limit.change(change_rank_limit, [higher_rank_limit], [lora_rank, lora_alpha])
+
 
     # Evaluation events. For some reason, the interrupt event
     # doesn't work with the .then() syntax, so I write them one
@@ -207,10 +206,6 @@ def do_copy_params(lora_name: str, *args):
     return result
 
 
-def change_rank_limit(use_higher_ranks: bool):
-    mult = 2 if use_higher_ranks else 1
-    return {"maximum": 1024 * mult, "__type__": "update"}, {"maximum": 2048 * mult, "__type__": "update"}
-
 
 def clean_path(base_path: str, path: str):
     """Strips unusual symbols and forcibly builds a path as relative to the intended directory."""
@@ -291,7 +286,7 @@ def calc_trainable_parameters(model):
     return trainable_params, all_param
 
 
-def do_train(lora_name: str, always_override: bool, all_linear: bool, q_proj_en: bool, v_proj_en: bool, k_proj_en: bool, o_proj_en: bool, gate_proj_en: bool, down_proj_en: bool, up_proj_en: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, text_dataset: str, higher_rank_limit: bool, warmup_steps: int, optimizer: str, stride_length: int, stop_at_loss: float, add_eos_token: bool, excess_length: str, report_to: str, gradient_checkpointing: bool = True):
+def do_train(lora_name: str, always_override: bool, all_linear: bool, q_proj_en: bool, v_proj_en: bool, k_proj_en: bool, o_proj_en: bool, gate_proj_en: bool, down_proj_en: bool, up_proj_en: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, text_dataset: str, warmup_steps: int, optimizer: str, stride_length: int, stop_at_loss: float, add_eos_token: bool, excess_length: str, report_to: str, gradient_checkpointing: bool = True):
 
     import torch
     import transformers

From 9d02d3a13b2e39a2a3bf91d8936044f9bbd9fd49 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Mar 2026 16:08:06 -0700
Subject: [PATCH 1405/1701] docs: Minor change to tool calling tutorial

---
 docs/Tool Calling Tutorial.md | 40 +++++++++++++++++------------------
 1 file changed, 20 insertions(+), 20 deletions(-)

diff --git a/docs/Tool Calling Tutorial.md b/docs/Tool Calling Tutorial.md
index 801e9d78b6..d95a9c8085 100644
--- a/docs/Tool Calling Tutorial.md	
+++ b/docs/Tool Calling Tutorial.md	
@@ -1,18 +1,3 @@
-## Supported models
-
-The following models are supported:
-
-- Qwen 3.5
-- GPT-OSS
-- Mistral Small / Devstral
-- DeepSeek V3
-- Kimi-K2
-- MiniMax-M2.5
-- GLM-5
-- Llama 4
-
-Other models that output tool calls as JSON (inside XML tags, code blocks, or plain JSON) are also supported through a generic fallback parser.
-
 ## Tool calling in the UI
 
 ### 1. Load a model with tool-calling support
@@ -23,11 +8,11 @@ Load a model with tool-calling support from the Model tab.
 
 In the chat sidebar, check the tools you want the model to use:
 
-- **web_search** -- Search the web using DuckDuckGo.
-- **fetch_webpage** -- Fetch the content of a URL.
-- **calculate** -- Evaluate math expressions.
-- **get_datetime** -- Get the current date and time.
-- **roll_dice** -- Roll dice.
+- `web_search`: Search the web using DuckDuckGo.
+- `fetch_webpage`: Fetch the content of a URL.
+- `calculate`: Evaluate math expressions.
+- `get_datetime`: Get the current date and time.
+- `roll_dice`: Roll dice.
 
 ### 3. Chat
 
@@ -157,3 +142,18 @@ for _ in range(10):
         print(f"\nAssistant: {choice['message']['content']}")
         break
 ```
+
+## Supported models
+
+The following models are supported:
+
+- Qwen 3.5
+- GPT-OSS
+- Mistral Small / Devstral
+- DeepSeek V3
+- Kimi-K2
+- MiniMax-M2.5
+- GLM-5
+- Llama 4
+
+Other models that output tool calls as JSON (inside XML tags, code blocks, or plain JSON) are also supported through a generic fallback parser.

From dff8903b03c2b3e46e11c16862f12db0495b3e91 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Mar 2026 18:25:54 -0700
Subject: [PATCH 1406/1701] UI: Modernize the Gradio theme

---
 css/main.css          | 73 +++++++++++++++++++++++++++++--------------
 modules/ui.py         | 26 +++++++++------
 modules/ui_session.py |  2 +-
 3 files changed, 68 insertions(+), 33 deletions(-)

diff --git a/css/main.css b/css/main.css
index 25ae15b156..22fac5c5da 100644
--- a/css/main.css
+++ b/css/main.css
@@ -2,8 +2,8 @@
     --darker-gray: #1C1C1D;
     --dark-gray: #212125;
     --light-gray: #2C2E34;
-    --light-theme-gray: #f9fbff;
-    --border-color-dark: #525252;
+    --light-theme-gray: #f0f3fb;
+    --border-color-dark: rgba(255, 255, 255, 0.15);
     --header-width: 112px;
     --selected-item-color-dark: #282930;
 }
@@ -127,7 +127,7 @@ gradio-app > :first-child {
 }
 
 .header_bar {
-    border-right: var(--input-border-width) solid var(--input-border-color);
+    border-right: none;
     margin-bottom: 0;
     overflow-x: scroll;
     text-wrap: nowrap;
@@ -150,7 +150,7 @@ gradio-app > :first-child {
 
 .dark .header_bar {
     border: none !important;
-    box-shadow: 0 3px 4px rgba(20 20 20 / 60%);
+    box-shadow: none;
     background-color: #8080802b;
 }
 
@@ -268,17 +268,17 @@ button {
 .dark #image-history-gallery > :nth-child(2)::-webkit-scrollbar-thumb,
 .dark #image-history-gallery > :nth-child(2)::-webkit-scrollbar-thumb:hover {
     background: rgb(255 255 255 / 6.25%);
-    border-radius: 10px;
+    border-radius: 30px;
 }
 
 .pretty_scrollbar::-webkit-resizer,
 #image-history-gallery > :nth-child(2)::-webkit-resizer {
-    background: #c5c5d2;
+    background: #d2d2d8;
 }
 
 .dark .pretty_scrollbar::-webkit-resizer,
 .dark #image-history-gallery > :nth-child(2)::-webkit-resizer {
-    background: #ccc;
+    background: rgb(255 255 255 / 10%);
     border-radius: 10px;
 }
 
@@ -582,10 +582,28 @@ audio {
 
 #chat-input textarea {
     background: #f3f4f6;
-    padding: 0.65rem 2.5rem;
-    border: 0;
-    box-shadow: 0;
-    border-radius: 8px;
+    padding: 0.65rem 2.5rem 0.6rem;
+    margin-top: 0.15rem;
+    border: 1px solid #d2d2d8;
+    border-radius: 1.5rem;
+    overflow-y: auto !important;
+}
+
+#chat-input textarea::-webkit-scrollbar {
+    width: 8px;
+}
+
+#chat-input textarea::-webkit-scrollbar-track {
+    background: transparent;
+}
+
+#chat-input textarea::-webkit-scrollbar-thumb {
+    background: var(--neutral-300);
+    border-radius: 30px;
+}
+
+.dark #chat-input textarea::-webkit-scrollbar-thumb {
+    background: rgb(255 255 255 / 6.25%);
 }
 
 #chat-input textarea::placeholder {
@@ -725,10 +743,12 @@ audio {
     position: absolute;
     bottom: 100%;
     left: 0;
-    box-shadow: 0 0 5px rgb(0 0 0 / 25%);
+    box-shadow: 0 2px 12px rgb(0 0 0 / 15%);
+    border-radius: 0.5rem;
     z-index: 10000;
     min-width: 330px;
     flex-direction: column;
+    overflow: hidden;
 }
 
 .hover-menu button {
@@ -739,6 +759,7 @@ audio {
     margin: 0 !important;
     height: 36px;
     border-color: transparent !important;
+    transition: background-color 0.15s ease;
 }
 
 .hover-menu button:not(#clear-history-confirm) {
@@ -914,7 +935,7 @@ audio {
 .options {
     z-index: 100 !important;
     border: 1px solid var(--input-border-color);
-    border-radius: 0;
+    border-radius: 0.5rem;
 }
 
 /* ----------------------------------------------
@@ -1008,9 +1029,13 @@ audio {
     cursor: pointer;
 }
 
+#past-chats label {
+    transition: background-color 0.15s ease;
+}
+
 #past-chats .selected,
 #past-chats label:hover {
-    background-color: #dbeafe !important;
+    background-color: #c8d8f5 !important;
 }
 
 #past-chats-buttons,
@@ -1166,7 +1191,7 @@ audio {
   Dark theme
 ---------------------------------------------- */
 .dark .header_bar {
-    background-color: var(--darker-gray) !important;
+    background-color: #1a1a1a !important;
 }
 
 .dark .header_bar button.selected {
@@ -1176,7 +1201,7 @@ audio {
 .dark #chat-input textarea {
     background: var(--light-gray);
     color: white !important;
-    border-color: #292c3b;
+    border-color: rgba(255, 255, 255, 0.06);
 }
 
 .dark #chat-input textarea::placeholder {
@@ -1192,6 +1217,7 @@ audio {
 .dark #past-chats-row {
     background-color: var(--darker-gray);
     border: 0 !important;
+    box-shadow: none;
 }
 
 .dark gradio-app .gradio-container.gradio-container-4-37-2 .contain #past-chats .selected,
@@ -1228,11 +1254,11 @@ audio {
   Light theme
 ---------------------------------------------- */
 .header_bar {
-    background-color: var(--light-theme-gray) !important;
+    background-color: #e4e8f0 !important;
 }
 
 .header_bar button.selected {
-    background: #dbeafe;
+    background: #c8d8f5;
 }
 
 #chat-controls,
@@ -1241,11 +1267,11 @@ audio {
 }
 
 .dark #chat-controls {
-    border-left: 1px solid #d9d9d0;
+    border-left: 1px solid rgba(255, 255, 255, 0.06);
 }
 
 .dark #past-chats-row {
-    border-right: 1px solid #d9d9d0;
+    border-right: 1px solid rgba(255, 255, 255, 0.06);
 }
 
 #past-chats-toggle,
@@ -1364,6 +1390,7 @@ audio {
 
 .tgw-accordion {
     padding: 10px 12px !important;
+    border: 1px solid #d2d2d8;
 }
 
 .dark .tgw-accordion {
@@ -1393,7 +1420,7 @@ audio {
 }
 
 .dark .thinking-block {
-    background-color: transparent;
+    background-color: var(--darker-gray);
     border: 1px solid var(--input-border-color);
 }
 
@@ -1742,7 +1769,7 @@ button:focus {
 }
 
 .dark .sidebar-vertical-separator {
-    border-bottom: 1px solid rgb(255 255 255 / 10%);
+    border-bottom: 1px solid rgba(255, 255, 255, 0.06);
 }
 
 button#swap-height-width {
@@ -1932,7 +1959,7 @@ thead + tbody tr:first-child th { border-top: 1px solid; }
 .dark #tools-group .wrap::-webkit-scrollbar-thumb,
 .dark #tools-group .wrap::-webkit-scrollbar-thumb:hover {
     background: rgb(255 255 255 / 6.25%);
-    border-radius: 10px;
+    border-radius: 30px;
 }
 
 #tools-group .wrap::-webkit-scrollbar-corner {
diff --git a/modules/ui.py b/modules/ui.py
index 3f39a1a419..bbb2226684 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -66,7 +66,8 @@
 if not shared.args.old_colors:
     theme = theme.set(
         # General Colors
-        border_color_primary='#c5c5d2',
+        border_color_primary='#d2d2d8',
+        block_border_color='transparent',
         body_text_color_subdued='#484848',
         background_fill_secondary='#eaeaea',
         background_fill_secondary_dark='var(--selected-item-color-dark, #282930)',
@@ -77,6 +78,12 @@
         body_text_color='rgb(64, 64, 64)',
         button_secondary_background_fill="white",
         button_secondary_border_color="var(--border-color-primary)",
+        block_title_text_color='*body_text_color',
+        button_primary_background_fill='#374151',
+        button_primary_background_fill_hover='#4b5563',
+        button_primary_background_fill_hover_dark='rgba(255, 255, 255, 0.05)',
+        button_primary_border_color='#374151',
+        button_primary_text_color='white',
         input_shadow="none",
         button_shadow_hover="none",
 
@@ -85,11 +92,11 @@
         checkbox_background_color_dark='var(--darker-gray, #1C1C1D)',
         block_background_fill_dark='transparent',
         block_border_color_dark='transparent',
-        input_border_color_dark='var(--border-color-dark, #525252)',
-        input_border_color_focus_dark='var(--border-color-dark, #525252)',
-        checkbox_border_color_dark='var(--border-color-dark, #525252)',
-        border_color_primary_dark='var(--border-color-dark, #525252)',
-        button_secondary_border_color_dark='var(--border-color-dark, #525252)',
+        input_border_color_dark='var(--border-color-dark)',
+        input_border_color_focus_dark='var(--border-color-dark)',
+        checkbox_border_color_dark='rgba(255, 255, 255, 0.2)',
+        border_color_primary_dark='var(--border-color-dark)',
+        button_secondary_border_color_dark='var(--border-color-dark)',
         body_background_fill_dark='var(--dark-gray, #212125)',
         button_primary_background_fill_dark='transparent',
         button_secondary_background_fill_dark='transparent',
@@ -107,10 +114,11 @@
         block_shadow_dark='none',
         input_shadow_focus='none',
         input_shadow_focus_dark='none',
-        button_large_radius='0.375rem',
+        button_large_radius='0.75rem',
         button_large_padding='6px 12px',
-        input_radius='0.375rem',
-        block_radius='0',
+        input_radius='0.5rem',
+        block_radius='0.375rem',
+        button_transition='background-color 0.15s ease, border-color 0.15s ease, color 0.15s ease',
     )
 
 if (shared.user_data_dir / "notification.mp3").exists():
diff --git a/modules/ui_session.py b/modules/ui_session.py
index e1807deae2..c06158438d 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -17,7 +17,7 @@ def create_ui():
 
             with gr.Column():
                 gr.Markdown("## Extensions & flags")
-                shared.gradio['save_settings'] = gr.Button(f'Save extensions settings to {shared.user_data_dir}/settings.yaml', elem_classes='refresh-button', interactive=not mu)
+                shared.gradio['save_settings'] = gr.Button(f'Save extensions settings to {shared.user_data_dir}/settings.yaml', interactive=not mu)
                 shared.gradio['reset_interface'] = gr.Button("Apply flags/extensions and restart", interactive=not mu)
                 with gr.Row():
                     with gr.Column():

From 5992e088faf94e5161f50d1dcf5996a10051d71c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 16 Mar 2026 19:34:37 -0700
Subject: [PATCH 1407/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 14 files changed, 28 insertions(+), 28 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index c24f4a9df9..ee83ce5631 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -31,8 +31,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 7c481224a8..ae21130109 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index b1c8f78eb9..158fc00474 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 63ef33ea88..f691d8729a 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 4bc61622f8..116db4421c 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 2ec1e61eb1..62f12e1b53 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index ba4c7a0498..d6e7896ce6 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 5dfdd9c840..26555e3024 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index f62241b36a..49f4c553bd 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 353d9172e1..6d8f478016 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 5f039318f6..9764b2e373 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index d8b03102e5..903da78ae7 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 4b548daeae..0360efdd88 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index fd2511f4c6..08b663e998 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio-4.37.2+custom.11-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.11/gradio_client-1.0.2+custom.11-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From 249861b65d0585f3cb290aaeb3d9050c18501ef3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Mar 2026 05:41:05 -0700
Subject: [PATCH 1408/1701] web search: Update the user agents

---
 modules/web_search.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/modules/web_search.py b/modules/web_search.py
index e13ef62ad8..2902c7c055 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -48,7 +48,7 @@ def download_web_page(url, timeout=10, include_links=False):
     try:
         _validate_url(url)
         headers = {
-            'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'
+            'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/130.0.0.0 Safari/537.36'
         }
         max_redirects = 5
         for _ in range(max_redirects):
@@ -82,8 +82,8 @@ def perform_web_search(query, num_pages=3, max_workers=5, timeout=10, fetch_cont
         search_url = f"https://html.duckduckgo.com/html/?q={quote_plus(query)}"
 
         agents = [
-            "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36",
-            "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36"
+            "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/130.0.0.0 Safari/537.36",
+            "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/130.0.0.0 Safari/537.36"
         ]
 
         response = requests.get(search_url, headers={'User-Agent': random.choice(agents)}, timeout=timeout)

From fffcd20f4d83d81b2577c4b9a94352cf8ed64484 Mon Sep 17 00:00:00 2001
From: Raunak-Kumar7 <73169853+Raunak-Kumar7@users.noreply.github.com>
Date: Tue, 17 Mar 2026 23:14:54 +0530
Subject: [PATCH 1409/1701] superboogav2: Fix broken delete endpoint (#6010)

---
 extensions/superboogav2/api.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/extensions/superboogav2/api.py b/extensions/superboogav2/api.py
index 552c1c2cfa..99b0e749cb 100644
--- a/extensions/superboogav2/api.py
+++ b/extensions/superboogav2/api.py
@@ -107,7 +107,7 @@ def do_POST(self):
 
             elif path in ['/api/v1/delete', '/api/delete']:
                 metadata = body.get('metadata')
-                if corpus is None:
+                if metadata is None:
                     self._send_412_error("Missing parameter 'metadata'")
                     return
 

From 2d141b54c5e0b5e042826e3d2f46bbaf87db023d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Mar 2026 11:11:12 -0700
Subject: [PATCH 1410/1701] Fix several typos

---
 README.md                        |  2 +-
 extensions/whisper_stt/readme.md |  4 ++--
 extensions/whisper_stt/script.py | 24 ++++++++++++------------
 modules/shared.py                |  2 +-
 modules/ui_model_menu.py         |  6 +++---
 5 files changed, 19 insertions(+), 19 deletions(-)

diff --git a/README.md b/README.md
index 989659d191..cabb81fc74 100644
--- a/README.md
+++ b/README.md
@@ -313,7 +313,7 @@ llama.cpp:
   --row-split                                          Split the model by rows across GPUs. This may improve multi-gpu performance.
   --no-mmap                                            Prevent mmap from being used.
   --mlock                                              Force the system to keep the model in RAM.
-  --no-kv-offload                                      Do not offload the K, Q, V to the GPU. This saves VRAM but reduces the performance.
+  --no-kv-offload                                      Do not offload the K, Q, V to the GPU. This saves VRAM but reduces performance.
   --batch-size BATCH_SIZE                              Maximum number of prompt tokens to batch together when calling llama-server. This is the application level batch size.
   --ubatch-size UBATCH_SIZE                            Maximum number of prompt tokens to batch together when calling llama-server. This is the max physical batch size for computation (device level).
   --threads THREADS                                    Number of threads to use.
diff --git a/extensions/whisper_stt/readme.md b/extensions/whisper_stt/readme.md
index 19488f94f4..7d9d8d23df 100644
--- a/extensions/whisper_stt/readme.md
+++ b/extensions/whisper_stt/readme.md
@@ -7,8 +7,8 @@ Allows you to enter your inputs in chat mode using your microphone.
 To adjust your default settings, you can add the following to your settings.yaml file.
 
 ```
-whisper_stt-whipser_language: chinese
-whisper_stt-whipser_model: tiny
+whisper_stt-whisper_language: chinese
+whisper_stt-whisper_model: tiny
 whisper_stt-auto_submit: False
 ```
 
diff --git a/extensions/whisper_stt/script.py b/extensions/whisper_stt/script.py
index d949e93f73..cd9175fefb 100644
--- a/extensions/whisper_stt/script.py
+++ b/extensions/whisper_stt/script.py
@@ -18,13 +18,13 @@
 
 # parameters which can be customized in settings.yaml of webui
 params = {
-    'whipser_language': 'english',
-    'whipser_model': 'small.en',
+    'whisper_language': 'english',
+    'whisper_model': 'small.en',
     'auto_submit': True
 }
 
 startup_device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
-WHISPERMODEL = whisper.load_model(params['whipser_model'], device=startup_device)
+WHISPERMODEL = whisper.load_model(params['whisper_model'], device=startup_device)
 
 
 def chat_input_modifier(text, visible_text, state):
@@ -36,7 +36,7 @@ def chat_input_modifier(text, visible_text, state):
         return text, visible_text
 
 
-def do_stt(audio, whipser_language):
+def do_stt(audio, whisper_language):
     # use pydub to convert sample_rate and sample_width for whisper input
     dubaudio = AudioSegment.from_file(io.BytesIO(audio))
     dubaudio = dubaudio.set_channels(1)
@@ -46,20 +46,20 @@ def do_stt(audio, whipser_language):
     # same method to get the array as openai whisper repo used from wav file
     audio_np = np.frombuffer(dubaudio.raw_data, np.int16).flatten().astype(np.float32) / 32768.0
 
-    if len(whipser_language) == 0:
+    if len(whisper_language) == 0:
         result = WHISPERMODEL.transcribe(audio=audio_np)
     else:
-        result = WHISPERMODEL.transcribe(audio=audio_np, language=whipser_language)
+        result = WHISPERMODEL.transcribe(audio=audio_np, language=whisper_language)
     return result["text"]
 
 
-def auto_transcribe(audio, auto_submit, whipser_language):
+def auto_transcribe(audio, auto_submit, whisper_language):
     if audio is None or audio == "":
         print("Whisper received no audio data")
         return "", ""
     audio_bytes = base64.b64decode(audio.split(',')[1])
 
-    transcription = do_stt(audio_bytes, whipser_language)
+    transcription = do_stt(audio_bytes, whisper_language)
     if auto_submit:
         input_hijack.update({"state": True, "value": [transcription, transcription]})
     return transcription
@@ -78,7 +78,7 @@ def reload_whispermodel(whisper_model_name: str, whisper_language: str, device:
                 device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
 
             WHISPERMODEL = whisper.load_model(whisper_model_name, device=device)
-            params.update({"whipser_model": whisper_model_name})
+            params.update({"whisper_model": whisper_model_name})
             if ".en" in whisper_model_name:
                 whisper_language = "english"
             audio_update = gr.Audio.update(interactive=True)
@@ -96,8 +96,8 @@ def ui():
             with gr.Accordion("Settings", open=False):
                 auto_submit = gr.Checkbox(label='Submit the transcribed audio automatically', value=params['auto_submit'])
                 device_dropd = gr.Dropdown(label='Device', value=str(startup_device), choices=["cuda", "cpu", "none"])
-                whisper_model_dropd = gr.Dropdown(label='Whisper Model', value=params['whipser_model'], choices=["tiny.en", "base.en", "small.en", "medium.en", "tiny", "base", "small", "medium", "large", "turbo"])
-                whisper_language = gr.Dropdown(label='Whisper Language', value=params['whipser_language'], choices=["english", "chinese", "german", "spanish", "russian", "korean", "french", "japanese", "portuguese", "turkish", "polish", "catalan", "dutch", "arabic", "swedish", "italian", "indonesian", "hindi", "finnish", "vietnamese", "hebrew", "ukrainian", "greek", "malay", "czech", "romanian", "danish", "hungarian", "tamil", "norwegian", "thai", "urdu", "croatian", "bulgarian", "lithuanian", "latin", "maori", "malayalam", "welsh", "slovak", "telugu", "persian", "latvian", "bengali", "serbian", "azerbaijani", "slovenian", "kannada", "estonian", "macedonian", "breton", "basque", "icelandic", "armenian", "nepali", "mongolian", "bosnian", "kazakh", "albanian", "swahili", "galician", "marathi", "punjabi", "sinhala", "khmer", "shona", "yoruba", "somali", "afrikaans", "occitan", "georgian", "belarusian", "tajik", "sindhi", "gujarati", "amharic", "yiddish", "lao", "uzbek", "faroese", "haitian creole", "pashto", "turkmen", "nynorsk", "maltese", "sanskrit", "luxembourgish", "myanmar", "tibetan", "tagalog", "malagasy", "assamese", "tatar", "hawaiian", "lingala", "hausa", "bashkir", "javanese", "sundanese"])
+                whisper_model_dropd = gr.Dropdown(label='Whisper Model', value=params['whisper_model'], choices=["tiny.en", "base.en", "small.en", "medium.en", "tiny", "base", "small", "medium", "large", "turbo"])
+                whisper_language = gr.Dropdown(label='Whisper Language', value=params['whisper_language'], choices=["english", "chinese", "german", "spanish", "russian", "korean", "french", "japanese", "portuguese", "turkish", "polish", "catalan", "dutch", "arabic", "swedish", "italian", "indonesian", "hindi", "finnish", "vietnamese", "hebrew", "ukrainian", "greek", "malay", "czech", "romanian", "danish", "hungarian", "tamil", "norwegian", "thai", "urdu", "croatian", "bulgarian", "lithuanian", "latin", "maori", "malayalam", "welsh", "slovak", "telugu", "persian", "latvian", "bengali", "serbian", "azerbaijani", "slovenian", "kannada", "estonian", "macedonian", "breton", "basque", "icelandic", "armenian", "nepali", "mongolian", "bosnian", "kazakh", "albanian", "swahili", "galician", "marathi", "punjabi", "sinhala", "khmer", "shona", "yoruba", "somali", "afrikaans", "occitan", "georgian", "belarusian", "tajik", "sindhi", "gujarati", "amharic", "yiddish", "lao", "uzbek", "faroese", "haitian creole", "pashto", "turkmen", "nynorsk", "maltese", "sanskrit", "luxembourgish", "myanmar", "tibetan", "tagalog", "malagasy", "assamese", "tatar", "hawaiian", "lingala", "hausa", "bashkir", "javanese", "sundanese"])
 
     audio.change(
         auto_transcribe, [audio, auto_submit, whisper_language], [shared.gradio['textbox']]).then(
@@ -105,7 +105,7 @@ def ui():
 
     device_dropd.input(reload_whispermodel, [whisper_model_dropd, whisper_language, device_dropd], [whisper_model_dropd, whisper_language, device_dropd, audio])
     whisper_model_dropd.change(reload_whispermodel, [whisper_model_dropd, whisper_language, device_dropd], [whisper_model_dropd, whisper_language, device_dropd, audio])
-    whisper_language.change(lambda x: params.update({"whipser_language": x}), whisper_language, None)
+    whisper_language.change(lambda x: params.update({"whisper_language": x}), whisper_language, None)
     auto_submit.change(lambda x: params.update({"auto_submit": x}), auto_submit, None)
 
 
diff --git a/modules/shared.py b/modules/shared.py
index 329114bbdb..486f376f4c 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -101,7 +101,7 @@
 group.add_argument('--row-split', action='store_true', help='Split the model by rows across GPUs. This may improve multi-gpu performance.')
 group.add_argument('--no-mmap', action='store_true', help='Prevent mmap from being used.')
 group.add_argument('--mlock', action='store_true', help='Force the system to keep the model in RAM.')
-group.add_argument('--no-kv-offload', action='store_true', help='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
+group.add_argument('--no-kv-offload', action='store_true', help='Do not offload the K, Q, V to the GPU. This saves VRAM but reduces performance.')
 group.add_argument('--batch-size', type=int, default=1024, help='Maximum number of prompt tokens to batch together when calling llama-server. This is the application level batch size.')
 group.add_argument('--ubatch-size', type=int, default=1024, help='Maximum number of prompt tokens to batch together when calling llama-server. This is the max physical batch size for computation (device level).')
 group.add_argument('--threads', type=int, default=0, help='Number of threads to use.')
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 5cf0155dbc..cb2052a475 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -107,7 +107,7 @@ def create_ui():
                                 shared.gradio['cpu'] = gr.Checkbox(label="cpu", value=shared.args.cpu, info='Use PyTorch in CPU mode.')
                                 shared.gradio['disk'] = gr.Checkbox(label="disk", value=shared.args.disk)
                                 shared.gradio['row_split'] = gr.Checkbox(label="row_split", value=shared.args.row_split, info='Split the model by rows across GPUs. This may improve multi-gpu performance.')
-                                shared.gradio['no_kv_offload'] = gr.Checkbox(label="no_kv_offload", value=shared.args.no_kv_offload, info='Do not offload the  K, Q, V to the GPU. This saves VRAM but reduces the performance.')
+                                shared.gradio['no_kv_offload'] = gr.Checkbox(label="no_kv_offload", value=shared.args.no_kv_offload, info='Do not offload the K, Q, V to the GPU. This saves VRAM but reduces performance.')
                                 shared.gradio['no_mmap'] = gr.Checkbox(label="no-mmap", value=shared.args.no_mmap)
                                 shared.gradio['mlock'] = gr.Checkbox(label="mlock", value=shared.args.mlock)
                                 shared.gradio['numa'] = gr.Checkbox(label="numa", value=shared.args.numa, info='NUMA support can help on some systems with non-uniform memory access.')
@@ -134,7 +134,7 @@ def create_ui():
                         ui.create_refresh_button(shared.gradio['customized_template'], lambda: None, lambda: {'choices': utils.get_available_instruction_templates()}, 'refresh-button', interactive=not mu)
 
                     shared.gradio['customized_template_submit'] = gr.Button("Submit", variant="primary", interactive=not mu)
-                    gr.Markdown("This allows you to set a customized template for the model currently selected in the \"Model loader\" menu. Whenever the model gets loaded, this template will be used in place of the template specified in the model's medatada, which sometimes is wrong.")
+                    gr.Markdown("This allows you to set a customized template for the model currently selected in the \"Model loader\" menu. Whenever the model gets loaded, this template will be used in place of the template specified in the model's metadata, which sometimes is wrong.")
 
                 with gr.Row():
                     shared.gradio['model_status'] = gr.Markdown('No model is loaded' if shared.model_name == 'None' else 'Ready')
@@ -231,7 +231,7 @@ def load_model_wrapper(selected_model, loader, autoload=False):
 def load_lora_wrapper(selected_loras):
     yield ("Applying the following LoRAs to {}:\n\n{}".format(shared.model_name, '\n'.join(selected_loras)))
     add_lora_to_model(selected_loras)
-    yield ("Successfuly applied the LoRAs")
+    yield ("Successfully applied the LoRAs")
 
 
 def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), return_links=False, check=False):

From 27a6cdeec11dc2f1536db3c846bb89a93efbdd69 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Mar 2026 11:31:55 -0700
Subject: [PATCH 1411/1701] Fix multi-turn thinking block corruption for Kimi
 models

---
 modules/chat.py | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index e4fcaabe4b..e526689d2a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -235,6 +235,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
         tools_in_user_message=False,
         add_generation_prompt=False,
         enable_thinking=state['enable_thinking'],
+        thinking=state['enable_thinking'],
         reasoning_effort=state['reasoning_effort'],
         thinking_budget=-1 if state.get('enable_thinking', True) else 0,
         bos_token=shared.bos_token,
@@ -351,6 +352,27 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
                 messages.insert(insert_pos, msg_dict)
 
+            # Handle  blocks (Kimi, DeepSeek, Qwen, etc.)
+            elif '' in assistant_msg:
+                thinking_content = ""
+                final_content = assistant_msg
+
+                parts = assistant_msg.split('', 1)
+                if len(parts) > 1:
+                    potential_content = parts[1]
+                    if '' in potential_content:
+                        thinking_content = potential_content.split('', 1)[0].strip()
+                        final_content = parts[0] + potential_content.split('', 1)[1]
+                    else:
+                        thinking_content = potential_content.strip()
+                        final_content = parts[0]
+
+                msg_dict = {"role": "assistant", "content": final_content.strip()}
+                if thinking_content:
+                    msg_dict["reasoning_content"] = thinking_content
+
+                messages.insert(insert_pos, msg_dict)
+
             else:
                 # Default case (used by all other models)
                 messages.insert(insert_pos, {"role": "assistant", "content": assistant_msg})

From 0f5053c0fbe4177b3b5af199d7301cc5e1bca0ac Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Mar 2026 17:57:35 -0700
Subject: [PATCH 1412/1701] requirements: Update pymupdf

---
 requirements/full/requirements.txt                   | 2 +-
 requirements/full/requirements_amd.txt               | 2 +-
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 2 +-
 requirements/full/requirements_nowheels.txt          | 2 +-
 requirements/portable/requirements.txt               | 2 +-
 requirements/portable/requirements_amd.txt           | 2 +-
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 2 +-
 requirements/portable/requirements_cuda131.txt       | 2 +-
 requirements/portable/requirements_nowheels.txt      | 2 +-
 requirements/portable/requirements_vulkan.txt        | 2 +-
 14 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index ee83ce5631..c8479d0418 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -14,7 +14,7 @@ pandas
 peft==0.18.*
 Pillow>=9.5.0
 pydantic==2.11.0
-pymupdf==1.27.1
+pymupdf==1.27.*
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index ae21130109..b11e50b716 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -12,7 +12,7 @@ pandas
 peft==0.18.*
 Pillow>=9.5.0
 pydantic==2.11.0
-pymupdf==1.27.1
+pymupdf==1.27.*
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 158fc00474..d147af3f28 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -12,7 +12,7 @@ pandas
 peft==0.18.*
 Pillow>=9.5.0
 pydantic==2.11.0
-pymupdf==1.27.1
+pymupdf==1.27.*
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index f691d8729a..d284c5d5fd 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -12,7 +12,7 @@ pandas
 peft==0.18.*
 Pillow>=9.5.0
 pydantic==2.11.0
-pymupdf==1.27.1
+pymupdf==1.27.*
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 116db4421c..3952054ee1 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -12,7 +12,7 @@ pandas
 peft==0.18.*
 Pillow>=9.5.0
 pydantic==2.11.0
-pymupdf==1.27.1
+pymupdf==1.27.*
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 62f12e1b53..77c254e681 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -12,7 +12,7 @@ pandas
 peft==0.18.*
 Pillow>=9.5.0
 pydantic==2.11.0
-pymupdf==1.27.1
+pymupdf==1.27.*
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index d6e7896ce6..abf7690c10 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -5,7 +5,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-pymupdf==1.27.1
+pymupdf==1.27.*
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 26555e3024..0d66c16c4b 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -5,7 +5,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-pymupdf==1.27.1
+pymupdf==1.27.*
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 49f4c553bd..0658239a11 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -5,7 +5,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-pymupdf==1.27.1
+pymupdf==1.27.*
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 6d8f478016..b66e2b38db 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -5,7 +5,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-pymupdf==1.27.1
+pymupdf==1.27.*
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 9764b2e373..bb815bb26e 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -5,7 +5,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-pymupdf==1.27.1
+pymupdf==1.27.*
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 903da78ae7..d57ba40b9f 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -5,7 +5,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-pymupdf==1.27.1
+pymupdf==1.27.*
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 0360efdd88..e8457909d9 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -5,7 +5,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-pymupdf==1.27.1
+pymupdf==1.27.*
 python-docx==1.1.2
 pyyaml
 requests
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 08b663e998..6abd892018 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -5,7 +5,7 @@ jinja2==3.1.6
 markdown
 numpy==2.2.*
 pydantic==2.11.0
-pymupdf==1.27.1
+pymupdf==1.27.*
 python-docx==1.1.2
 pyyaml
 requests

From f0014ab01c7a51bfa0f269c676404d48112f924b Mon Sep 17 00:00:00 2001
From: RoomWithOutRoof <166608075+Jah-yee@users.noreply.github.com>
Date: Wed, 18 Mar 2026 09:03:48 +0800
Subject: [PATCH 1413/1701] fix: mutable default argument in
 LogitsBiasProcessor (#7426)

---
 modules/transformers_loader.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/transformers_loader.py b/modules/transformers_loader.py
index 63758ad7d1..7f521b8c22 100644
--- a/modules/transformers_loader.py
+++ b/modules/transformers_loader.py
@@ -44,8 +44,8 @@ def __call__(self, input_ids, scores) -> bool:
 
 
 class LogitsBiasProcessor(LogitsProcessor):
-    def __init__(self, logit_bias={}):
-        self.logit_bias = logit_bias
+    def __init__(self, logit_bias=None):
+        self.logit_bias = logit_bias if logit_bias is not None else {}
         if self.logit_bias:
             self.keys = list([int(key) for key in self.logit_bias.keys()])
             values = [self.logit_bias[str(key)] for key in self.keys]

From 73a094a65773a3f2f9e7d626cfaa01893dbd3f88 Mon Sep 17 00:00:00 2001
From: Alvin Tang 
Date: Wed, 18 Mar 2026 09:06:05 +0800
Subject: [PATCH 1414/1701] Fix file handle leaks and redundant re-read in
 get_model_metadata (#7422)

---
 modules/models_settings.py | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index f3c9a9866f..dcface7182 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -34,7 +34,8 @@ def get_model_metadata(model):
 
     path = model_path / 'config.json'
     if path.exists():
-        hf_metadata = json.loads(open(path, 'r', encoding='utf-8').read())
+        with open(path, 'r', encoding='utf-8') as f:
+            hf_metadata = json.loads(f.read())
     else:
         hf_metadata = None
 
@@ -93,7 +94,7 @@ def get_model_metadata(model):
     else:
         # Transformers metadata
         if hf_metadata is not None:
-            metadata = json.loads(open(path, 'r', encoding='utf-8').read())
+            metadata = hf_metadata
             if 'pretrained_config' in metadata:
                 metadata = metadata['pretrained_config']
 
@@ -134,7 +135,8 @@ def get_model_metadata(model):
 
     # 3. Fall back to tokenizer_config.json metadata
     if path.exists():
-        metadata = json.loads(open(path, 'r', encoding='utf-8').read())
+        with open(path, 'r', encoding='utf-8') as f:
+            metadata = json.loads(f.read())
 
         # Only read from metadata if we haven't already loaded from .jinja or .json
         if template is None and 'chat_template' in metadata:

From 2a6b1fdcba676200d2e454534a91e1d334b60bdf Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Mar 2026 18:29:15 -0700
Subject: [PATCH 1415/1701] Fix `--extra-flags` breaking short long-form-only
 flags like `--rpc`

Closes #7357
---
 modules/llama_cpp_server.py | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index c3a8d1052b..321a6d755f 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -446,18 +446,21 @@ def _start_server(self):
             elif extra_flags.startswith("'") and extra_flags.endswith("'"):
                 extra_flags = extra_flags[1:-1].strip()
 
+            # llama.cpp flags that only have a long form (--) despite being short
+            long_form_only = {'rpc', 'fit', 'pos', 'ppl'}
+
             for flag_item in extra_flags.split(','):
                 flag_item = flag_item.strip()
                 if '=' in flag_item:
                     flag, value = flag_item.split('=', 1)
                     flag = flag.strip()
                     value = value.strip()
-                    if len(flag) <= 3:
+                    if len(flag) <= 3 and flag not in long_form_only:
                         cmd += [f"-{flag}", value]
                     else:
                         cmd += [f"--{flag}", value]
                 else:
-                    if len(flag_item) <= 3:
+                    if len(flag_item) <= 3 and flag_item not in long_form_only:
                         cmd.append(f"-{flag_item}")
                     else:
                         cmd.append(f"--{flag_item}")

From 7e54e7b7ae62b227fbd896b2daf704db1658baa5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Mar 2026 19:47:55 -0700
Subject: [PATCH 1416/1701] llama.cpp: Support literal flags in `--extra-flags`
 (e.g. `--rpc`, `--jinja`)

The old format is still accepted for backwards compatibility.
---
 modules/llama_cpp_server.py | 39 +++++++++++++++++++++----------------
 modules/shared.py           |  2 +-
 modules/ui_model_menu.py    |  2 +-
 3 files changed, 24 insertions(+), 19 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 321a6d755f..6dd36b2a31 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -1,6 +1,7 @@
 import json
 import os
 import pprint
+import shlex
 import re
 import socket
 import subprocess
@@ -446,24 +447,28 @@ def _start_server(self):
             elif extra_flags.startswith("'") and extra_flags.endswith("'"):
                 extra_flags = extra_flags[1:-1].strip()
 
-            # llama.cpp flags that only have a long form (--) despite being short
-            long_form_only = {'rpc', 'fit', 'pos', 'ppl'}
-
-            for flag_item in extra_flags.split(','):
-                flag_item = flag_item.strip()
-                if '=' in flag_item:
-                    flag, value = flag_item.split('=', 1)
-                    flag = flag.strip()
-                    value = value.strip()
-                    if len(flag) <= 3 and flag not in long_form_only:
-                        cmd += [f"-{flag}", value]
-                    else:
-                        cmd += [f"--{flag}", value]
-                else:
-                    if len(flag_item) <= 3 and flag_item not in long_form_only:
-                        cmd.append(f"-{flag_item}")
+            if extra_flags.startswith('-'):
+                # New literal format: "--jinja --rpc 1222,1222"
+                cmd += shlex.split(extra_flags)
+            else:
+                # Legacy format: "flag1=value1,flag2,flag3=value3"
+                long_form_only = {'rpc', 'fit', 'pos', 'ppl'}
+
+                for flag_item in extra_flags.split(','):
+                    flag_item = flag_item.strip()
+                    if '=' in flag_item:
+                        flag, value = flag_item.split('=', 1)
+                        flag = flag.strip()
+                        value = value.strip()
+                        if len(flag) <= 3 and flag not in long_form_only:
+                            cmd += [f"-{flag}", value]
+                        else:
+                            cmd += [f"--{flag}", value]
                     else:
-                        cmd.append(f"--{flag_item}")
+                        if len(flag_item) <= 3 and flag_item not in long_form_only:
+                            cmd.append(f"-{flag_item}")
+                        else:
+                            cmd.append(f"--{flag_item}")
 
         env = os.environ.copy()
         if os.name == 'posix':
diff --git a/modules/shared.py b/modules/shared.py
index 486f376f4c..2382e71432 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -109,7 +109,7 @@
 group.add_argument('--numa', action='store_true', help='Activate NUMA task allocation for llama.cpp.')
 group.add_argument('--parallel', type=int, default=1, help='Number of parallel request slots. The context size is divided equally among slots. For example, to have 4 slots with 8192 context each, set ctx_size to 32768.')
 group.add_argument('--fit-target', type=str, default='512', help='Target VRAM margin per device for auto GPU layers, comma-separated list of values in MiB. A single value is broadcast across all devices.')
-group.add_argument('--extra-flags', type=str, default=None, help='Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"')
+group.add_argument('--extra-flags', type=str, default=None, help='Extra flags to pass to llama-server. Example: "--jinja --rpc 192.168.1.100:50052"')
 
 # Transformers/Accelerate
 group = parser.add_argument_group('Transformers/Accelerate')
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index cb2052a475..6d8baff16a 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -98,7 +98,7 @@ def create_ui():
                                 shared.gradio['batch_size'] = gr.Slider(label="batch_size", minimum=1, maximum=4096, step=1, value=shared.args.batch_size)
                                 shared.gradio['ubatch_size'] = gr.Slider(label="ubatch_size", minimum=1, maximum=4096, step=1, value=shared.args.ubatch_size)
                                 shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
-                                shared.gradio['extra_flags'] = gr.Textbox(label='extra-flags', info='Additional flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"', value=shared.args.extra_flags)
+                                shared.gradio['extra_flags'] = gr.Textbox(label='extra-flags', info='Extra flags to pass to llama-server. Example: --jinja --rpc 192.168.1.100:50052', value=shared.args.extra_flags)
                                 shared.gradio['cpu_memory'] = gr.Number(label="Maximum CPU memory in GiB. Use this for CPU offloading.", value=shared.args.cpu_memory)
                                 shared.gradio['compute_dtype'] = gr.Dropdown(label="compute_dtype", choices=["bfloat16", "float16", "float32"], value=shared.args.compute_dtype, info='Used by load-in-4bit.')
                                 shared.gradio['quant_type'] = gr.Dropdown(label="quant_type", choices=["nf4", "fp4"], value=shared.args.quant_type, info='Used by load-in-4bit.')

From c8bb2129baf180c3d3a5d1d410d1e78dc5ddbea3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Mar 2026 22:24:36 -0700
Subject: [PATCH 1417/1701] Security: server-side file save roots, image URL
 SSRF protection, extension allowlist

---
 modules/chat.py           |  8 +++++--
 modules/image_utils.py    | 13 ++++++++++-
 modules/ui_chat.py        |  4 ++--
 modules/ui_file_saving.py | 46 +++++++++++++++++++++++++++------------
 modules/ui_session.py     |  6 +++--
 modules/utils.py          |  4 ++++
 6 files changed, 60 insertions(+), 21 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index e526689d2a..00f1659bbe 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -2634,19 +2634,23 @@ def handle_load_template_click(instruction_template):
 def handle_save_template_click(instruction_template_str):
     import gradio as gr
     contents = generate_instruction_template_yaml(instruction_template_str)
+    root = str(shared.user_data_dir / 'instruction-templates') + '/'
     return [
         "My Template.yaml",
-        str(shared.user_data_dir / 'instruction-templates') + '/',
+        root,
         contents,
+        root,
         gr.update(visible=True)
     ]
 
 
 def handle_delete_template_click(template):
     import gradio as gr
+    root = str(shared.user_data_dir / 'instruction-templates') + '/'
     return [
         f"{template}.yaml",
-        str(shared.user_data_dir / 'instruction-templates') + '/',
+        root,
+        root,
         gr.update(visible=False)
     ]
 
diff --git a/modules/image_utils.py b/modules/image_utils.py
index d2809fef36..b313879074 100644
--- a/modules/image_utils.py
+++ b/modules/image_utils.py
@@ -77,7 +77,18 @@ def process_message_content(content: Any) -> Tuple[str, List[Image.Image]]:
                     # Support external URLs
                     try:
                         import requests
-                        response = requests.get(image_url, timeout=10)
+                        from urllib.parse import urljoin
+                        from modules.web_search import _validate_url
+                        _validate_url(image_url)
+                        url = image_url
+                        for _ in range(5):
+                            response = requests.get(url, timeout=10, allow_redirects=False)
+                            if response.is_redirect and 'Location' in response.headers:
+                                url = urljoin(url, response.headers['Location'])
+                                _validate_url(url)
+                            else:
+                                break
+
                         response.raise_for_status()
                         image_data = response.content
                         image = Image.open(io.BytesIO(image_data))
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index d2a515b84d..f1dc7883cf 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -350,13 +350,13 @@ def create_event_handlers():
     shared.gradio['load_template'].click(chat.handle_load_template_click, gradio('instruction_template'), gradio('instruction_template_str', 'instruction_template'), show_progress=False)
     shared.gradio['save_template'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        chat.handle_save_template_click, gradio('instruction_template_str'), gradio('save_filename', 'save_root', 'save_contents', 'file_saver'), show_progress=False)
+        chat.handle_save_template_click, gradio('instruction_template_str'), gradio('save_filename', 'save_root', 'save_contents', 'save_root_state', 'file_saver'), show_progress=False)
 
     shared.gradio['restore_character'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.restore_character_for_ui, gradio('interface_state'), gradio('interface_state', 'name2', 'context', 'greeting', 'character_picture'), show_progress=False)
 
-    shared.gradio['delete_template'].click(chat.handle_delete_template_click, gradio('instruction_template'), gradio('delete_filename', 'delete_root', 'file_deleter'), show_progress=False)
+    shared.gradio['delete_template'].click(chat.handle_delete_template_click, gradio('instruction_template'), gradio('delete_filename', 'delete_root', 'delete_root_state', 'file_deleter'), show_progress=False)
     shared.gradio['save_chat_history'].click(
         lambda x: json.dumps(x, indent=4), gradio('history'), gradio('temporary_text')).then(
         None, gradio('temporary_text', 'character_menu', 'mode'), None, js=f'(hist, char, mode) => {{{ui.save_files_js}; saveHistory(hist, char, mode)}}')
diff --git a/modules/ui_file_saving.py b/modules/ui_file_saving.py
index 3ed256f821..99c4edd59b 100644
--- a/modules/ui_file_saving.py
+++ b/modules/ui_file_saving.py
@@ -9,6 +9,12 @@
 def create_ui():
     mu = shared.args.multi_user
 
+    # Server-side per-session root paths for the generic file saver/deleter.
+    # Set by the handler that opens the dialog, read by the confirm handler.
+    # Using gr.State so they are session-scoped and safe for multi-user.
+    shared.gradio['save_root_state'] = gr.State(None)
+    shared.gradio['delete_root_state'] = gr.State(None)
+
     # Text file saver
     with gr.Group(visible=False, elem_classes='file-saver') as shared.gradio['file_saver']:
         shared.gradio['save_filename'] = gr.Textbox(lines=1, label='File name')
@@ -66,13 +72,13 @@ def create_event_handlers():
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         handle_save_preset_click, gradio('interface_state'), gradio('save_preset_contents', 'save_preset_filename', 'preset_saver'), show_progress=False)
 
-    shared.gradio['delete_preset'].click(handle_delete_preset_click, gradio('preset_menu'), gradio('delete_filename', 'delete_root', 'file_deleter'), show_progress=False)
-    shared.gradio['save_grammar'].click(handle_save_grammar_click, gradio('grammar_string'), gradio('save_contents', 'save_filename', 'save_root', 'file_saver'), show_progress=False)
-    shared.gradio['delete_grammar'].click(handle_delete_grammar_click, gradio('grammar_file'), gradio('delete_filename', 'delete_root', 'file_deleter'), show_progress=False)
+    shared.gradio['delete_preset'].click(handle_delete_preset_click, gradio('preset_menu'), gradio('delete_filename', 'delete_root', 'delete_root_state', 'file_deleter'), show_progress=False)
+    shared.gradio['save_grammar'].click(handle_save_grammar_click, gradio('grammar_string'), gradio('save_contents', 'save_filename', 'save_root', 'save_root_state', 'file_saver'), show_progress=False)
+    shared.gradio['delete_grammar'].click(handle_delete_grammar_click, gradio('grammar_file'), gradio('delete_filename', 'delete_root', 'delete_root_state', 'file_deleter'), show_progress=False)
 
     shared.gradio['save_preset_confirm'].click(handle_save_preset_confirm_click, gradio('save_preset_filename', 'save_preset_contents'), gradio('preset_menu', 'preset_saver'), show_progress=False)
-    shared.gradio['save_confirm'].click(handle_save_confirm_click, gradio('save_root', 'save_filename', 'save_contents'), gradio('file_saver'), show_progress=False)
-    shared.gradio['delete_confirm'].click(handle_delete_confirm_click, gradio('delete_root', 'delete_filename'), gradio('file_deleter'), show_progress=False)
+    shared.gradio['save_confirm'].click(handle_save_confirm_click, gradio('save_root_state', 'save_filename', 'save_contents'), gradio('save_root_state', 'file_saver'), show_progress=False)
+    shared.gradio['delete_confirm'].click(handle_delete_confirm_click, gradio('delete_root_state', 'delete_filename'), gradio('delete_root_state', 'file_deleter'), show_progress=False)
     shared.gradio['save_character_confirm'].click(handle_save_character_confirm_click, gradio('name2', 'greeting', 'context', 'character_picture', 'save_character_filename'), gradio('character_menu', 'character_saver'), show_progress=False)
     shared.gradio['delete_character_confirm'].click(handle_delete_character_confirm_click, gradio('character_menu'), gradio('character_menu', 'character_deleter'), show_progress=False)
 
@@ -105,24 +111,30 @@ def handle_save_preset_confirm_click(filename, contents):
     ]
 
 
-def handle_save_confirm_click(root, filename, contents):
+def handle_save_confirm_click(root_state, filename, contents):
     try:
+        if root_state is None:
+            return None, gr.update(visible=False)
+
         filename = sanitize_filename(filename)
-        utils.save_file(root + filename, contents)
+        utils.save_file(root_state + filename, contents)
     except Exception:
         traceback.print_exc()
 
-    return gr.update(visible=False)
+    return None, gr.update(visible=False)
 
 
-def handle_delete_confirm_click(root, filename):
+def handle_delete_confirm_click(root_state, filename):
     try:
+        if root_state is None:
+            return None, gr.update(visible=False)
+
         filename = sanitize_filename(filename)
-        utils.delete_file(root + filename)
+        utils.delete_file(root_state + filename)
     except Exception:
         traceback.print_exc()
 
-    return gr.update(visible=False)
+    return None, gr.update(visible=False)
 
 
 def handle_save_character_confirm_click(name2, greeting, context, character_picture, filename):
@@ -165,26 +177,32 @@ def handle_save_preset_click(state):
 
 
 def handle_delete_preset_click(preset):
+    root = str(shared.user_data_dir / "presets") + "/"
     return [
         f"{preset}.yaml",
-        str(shared.user_data_dir / "presets") + "/",
+        root,
+        root,
         gr.update(visible=True)
     ]
 
 
 def handle_save_grammar_click(grammar_string):
+    root = str(shared.user_data_dir / "grammars") + "/"
     return [
         grammar_string,
         "My Fancy Grammar.gbnf",
-        str(shared.user_data_dir / "grammars") + "/",
+        root,
+        root,
         gr.update(visible=True)
     ]
 
 
 def handle_delete_grammar_click(grammar_file):
+    root = str(shared.user_data_dir / "grammars") + "/"
     return [
         grammar_file,
-        str(shared.user_data_dir / "grammars") + "/",
+        root,
+        root,
         gr.update(visible=True)
     ]
 
diff --git a/modules/ui_session.py b/modules/ui_session.py
index c06158438d..19026fbb72 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -30,7 +30,7 @@ def create_ui():
         if not mu:
             shared.gradio['save_settings'].click(
                 ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-                handle_save_settings, gradio('interface_state', 'preset_menu', 'extensions_menu', 'show_controls', 'theme_state'), gradio('save_contents', 'save_filename', 'save_root', 'file_saver'), show_progress=False)
+                handle_save_settings, gradio('interface_state', 'preset_menu', 'extensions_menu', 'show_controls', 'theme_state'), gradio('save_contents', 'save_filename', 'save_root', 'save_root_state', 'file_saver'), show_progress=False)
 
         shared.gradio['toggle_dark_mode'].click(
             lambda x: 'dark' if x == 'light' else 'light', gradio('theme_state'), gradio('theme_state')).then(
@@ -51,10 +51,12 @@ def create_ui():
 
 def handle_save_settings(state, preset, extensions, show_controls, theme):
     contents = ui.save_settings(state, preset, extensions, show_controls, theme, manual_save=True)
+    root = str(shared.user_data_dir) + "/"
     return [
         contents,
         "settings.yaml",
-        str(shared.user_data_dir) + "/",
+        root,
+        root,
         gr.update(visible=True)
     ]
 
diff --git a/modules/utils.py b/modules/utils.py
index a14f8b8f41..ff32e974f0 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -47,6 +47,10 @@ def save_file(fname, contents):
         logger.error(f'Invalid file path: \"{fname}\"')
         return
 
+    if Path(abs_path_str).suffix.lower() not in ('.yaml', '.yml', '.json', '.txt', '.gbnf'):
+        logger.error(f'Refusing to save file with disallowed extension: \"{fname}\"')
+        return
+
     with open(abs_path_str, 'w', encoding='utf-8') as f:
         f.write(contents)
 

From 256431f25869fb89326021b8051e340ef275a416 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Mar 2026 22:24:36 -0700
Subject: [PATCH 1418/1701] Security: server-side file save roots, image URL
 SSRF protection, extension allowlist

---
 modules/chat.py           |  8 +++++--
 modules/image_utils.py    | 13 ++++++++++-
 modules/ui_chat.py        |  4 ++--
 modules/ui_file_saving.py | 46 +++++++++++++++++++++++++++------------
 modules/ui_session.py     |  6 +++--
 modules/utils.py          |  4 ++++
 6 files changed, 60 insertions(+), 21 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index e4fcaabe4b..e37c7a4e22 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -2612,19 +2612,23 @@ def handle_load_template_click(instruction_template):
 def handle_save_template_click(instruction_template_str):
     import gradio as gr
     contents = generate_instruction_template_yaml(instruction_template_str)
+    root = str(shared.user_data_dir / 'instruction-templates') + '/'
     return [
         "My Template.yaml",
-        str(shared.user_data_dir / 'instruction-templates') + '/',
+        root,
         contents,
+        root,
         gr.update(visible=True)
     ]
 
 
 def handle_delete_template_click(template):
     import gradio as gr
+    root = str(shared.user_data_dir / 'instruction-templates') + '/'
     return [
         f"{template}.yaml",
-        str(shared.user_data_dir / 'instruction-templates') + '/',
+        root,
+        root,
         gr.update(visible=False)
     ]
 
diff --git a/modules/image_utils.py b/modules/image_utils.py
index d2809fef36..b313879074 100644
--- a/modules/image_utils.py
+++ b/modules/image_utils.py
@@ -77,7 +77,18 @@ def process_message_content(content: Any) -> Tuple[str, List[Image.Image]]:
                     # Support external URLs
                     try:
                         import requests
-                        response = requests.get(image_url, timeout=10)
+                        from urllib.parse import urljoin
+                        from modules.web_search import _validate_url
+                        _validate_url(image_url)
+                        url = image_url
+                        for _ in range(5):
+                            response = requests.get(url, timeout=10, allow_redirects=False)
+                            if response.is_redirect and 'Location' in response.headers:
+                                url = urljoin(url, response.headers['Location'])
+                                _validate_url(url)
+                            else:
+                                break
+
                         response.raise_for_status()
                         image_data = response.content
                         image = Image.open(io.BytesIO(image_data))
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index d2a515b84d..f1dc7883cf 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -350,13 +350,13 @@ def create_event_handlers():
     shared.gradio['load_template'].click(chat.handle_load_template_click, gradio('instruction_template'), gradio('instruction_template_str', 'instruction_template'), show_progress=False)
     shared.gradio['save_template'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        chat.handle_save_template_click, gradio('instruction_template_str'), gradio('save_filename', 'save_root', 'save_contents', 'file_saver'), show_progress=False)
+        chat.handle_save_template_click, gradio('instruction_template_str'), gradio('save_filename', 'save_root', 'save_contents', 'save_root_state', 'file_saver'), show_progress=False)
 
     shared.gradio['restore_character'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.restore_character_for_ui, gradio('interface_state'), gradio('interface_state', 'name2', 'context', 'greeting', 'character_picture'), show_progress=False)
 
-    shared.gradio['delete_template'].click(chat.handle_delete_template_click, gradio('instruction_template'), gradio('delete_filename', 'delete_root', 'file_deleter'), show_progress=False)
+    shared.gradio['delete_template'].click(chat.handle_delete_template_click, gradio('instruction_template'), gradio('delete_filename', 'delete_root', 'delete_root_state', 'file_deleter'), show_progress=False)
     shared.gradio['save_chat_history'].click(
         lambda x: json.dumps(x, indent=4), gradio('history'), gradio('temporary_text')).then(
         None, gradio('temporary_text', 'character_menu', 'mode'), None, js=f'(hist, char, mode) => {{{ui.save_files_js}; saveHistory(hist, char, mode)}}')
diff --git a/modules/ui_file_saving.py b/modules/ui_file_saving.py
index 3ed256f821..99c4edd59b 100644
--- a/modules/ui_file_saving.py
+++ b/modules/ui_file_saving.py
@@ -9,6 +9,12 @@
 def create_ui():
     mu = shared.args.multi_user
 
+    # Server-side per-session root paths for the generic file saver/deleter.
+    # Set by the handler that opens the dialog, read by the confirm handler.
+    # Using gr.State so they are session-scoped and safe for multi-user.
+    shared.gradio['save_root_state'] = gr.State(None)
+    shared.gradio['delete_root_state'] = gr.State(None)
+
     # Text file saver
     with gr.Group(visible=False, elem_classes='file-saver') as shared.gradio['file_saver']:
         shared.gradio['save_filename'] = gr.Textbox(lines=1, label='File name')
@@ -66,13 +72,13 @@ def create_event_handlers():
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         handle_save_preset_click, gradio('interface_state'), gradio('save_preset_contents', 'save_preset_filename', 'preset_saver'), show_progress=False)
 
-    shared.gradio['delete_preset'].click(handle_delete_preset_click, gradio('preset_menu'), gradio('delete_filename', 'delete_root', 'file_deleter'), show_progress=False)
-    shared.gradio['save_grammar'].click(handle_save_grammar_click, gradio('grammar_string'), gradio('save_contents', 'save_filename', 'save_root', 'file_saver'), show_progress=False)
-    shared.gradio['delete_grammar'].click(handle_delete_grammar_click, gradio('grammar_file'), gradio('delete_filename', 'delete_root', 'file_deleter'), show_progress=False)
+    shared.gradio['delete_preset'].click(handle_delete_preset_click, gradio('preset_menu'), gradio('delete_filename', 'delete_root', 'delete_root_state', 'file_deleter'), show_progress=False)
+    shared.gradio['save_grammar'].click(handle_save_grammar_click, gradio('grammar_string'), gradio('save_contents', 'save_filename', 'save_root', 'save_root_state', 'file_saver'), show_progress=False)
+    shared.gradio['delete_grammar'].click(handle_delete_grammar_click, gradio('grammar_file'), gradio('delete_filename', 'delete_root', 'delete_root_state', 'file_deleter'), show_progress=False)
 
     shared.gradio['save_preset_confirm'].click(handle_save_preset_confirm_click, gradio('save_preset_filename', 'save_preset_contents'), gradio('preset_menu', 'preset_saver'), show_progress=False)
-    shared.gradio['save_confirm'].click(handle_save_confirm_click, gradio('save_root', 'save_filename', 'save_contents'), gradio('file_saver'), show_progress=False)
-    shared.gradio['delete_confirm'].click(handle_delete_confirm_click, gradio('delete_root', 'delete_filename'), gradio('file_deleter'), show_progress=False)
+    shared.gradio['save_confirm'].click(handle_save_confirm_click, gradio('save_root_state', 'save_filename', 'save_contents'), gradio('save_root_state', 'file_saver'), show_progress=False)
+    shared.gradio['delete_confirm'].click(handle_delete_confirm_click, gradio('delete_root_state', 'delete_filename'), gradio('delete_root_state', 'file_deleter'), show_progress=False)
     shared.gradio['save_character_confirm'].click(handle_save_character_confirm_click, gradio('name2', 'greeting', 'context', 'character_picture', 'save_character_filename'), gradio('character_menu', 'character_saver'), show_progress=False)
     shared.gradio['delete_character_confirm'].click(handle_delete_character_confirm_click, gradio('character_menu'), gradio('character_menu', 'character_deleter'), show_progress=False)
 
@@ -105,24 +111,30 @@ def handle_save_preset_confirm_click(filename, contents):
     ]
 
 
-def handle_save_confirm_click(root, filename, contents):
+def handle_save_confirm_click(root_state, filename, contents):
     try:
+        if root_state is None:
+            return None, gr.update(visible=False)
+
         filename = sanitize_filename(filename)
-        utils.save_file(root + filename, contents)
+        utils.save_file(root_state + filename, contents)
     except Exception:
         traceback.print_exc()
 
-    return gr.update(visible=False)
+    return None, gr.update(visible=False)
 
 
-def handle_delete_confirm_click(root, filename):
+def handle_delete_confirm_click(root_state, filename):
     try:
+        if root_state is None:
+            return None, gr.update(visible=False)
+
         filename = sanitize_filename(filename)
-        utils.delete_file(root + filename)
+        utils.delete_file(root_state + filename)
     except Exception:
         traceback.print_exc()
 
-    return gr.update(visible=False)
+    return None, gr.update(visible=False)
 
 
 def handle_save_character_confirm_click(name2, greeting, context, character_picture, filename):
@@ -165,26 +177,32 @@ def handle_save_preset_click(state):
 
 
 def handle_delete_preset_click(preset):
+    root = str(shared.user_data_dir / "presets") + "/"
     return [
         f"{preset}.yaml",
-        str(shared.user_data_dir / "presets") + "/",
+        root,
+        root,
         gr.update(visible=True)
     ]
 
 
 def handle_save_grammar_click(grammar_string):
+    root = str(shared.user_data_dir / "grammars") + "/"
     return [
         grammar_string,
         "My Fancy Grammar.gbnf",
-        str(shared.user_data_dir / "grammars") + "/",
+        root,
+        root,
         gr.update(visible=True)
     ]
 
 
 def handle_delete_grammar_click(grammar_file):
+    root = str(shared.user_data_dir / "grammars") + "/"
     return [
         grammar_file,
-        str(shared.user_data_dir / "grammars") + "/",
+        root,
+        root,
         gr.update(visible=True)
     ]
 
diff --git a/modules/ui_session.py b/modules/ui_session.py
index e1807deae2..897bfd28d1 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -30,7 +30,7 @@ def create_ui():
         if not mu:
             shared.gradio['save_settings'].click(
                 ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-                handle_save_settings, gradio('interface_state', 'preset_menu', 'extensions_menu', 'show_controls', 'theme_state'), gradio('save_contents', 'save_filename', 'save_root', 'file_saver'), show_progress=False)
+                handle_save_settings, gradio('interface_state', 'preset_menu', 'extensions_menu', 'show_controls', 'theme_state'), gradio('save_contents', 'save_filename', 'save_root', 'save_root_state', 'file_saver'), show_progress=False)
 
         shared.gradio['toggle_dark_mode'].click(
             lambda x: 'dark' if x == 'light' else 'light', gradio('theme_state'), gradio('theme_state')).then(
@@ -51,10 +51,12 @@ def create_ui():
 
 def handle_save_settings(state, preset, extensions, show_controls, theme):
     contents = ui.save_settings(state, preset, extensions, show_controls, theme, manual_save=True)
+    root = str(shared.user_data_dir) + "/"
     return [
         contents,
         "settings.yaml",
-        str(shared.user_data_dir) + "/",
+        root,
+        root,
         gr.update(visible=True)
     ]
 
diff --git a/modules/utils.py b/modules/utils.py
index a14f8b8f41..ff32e974f0 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -47,6 +47,10 @@ def save_file(fname, contents):
         logger.error(f'Invalid file path: \"{fname}\"')
         return
 
+    if Path(abs_path_str).suffix.lower() not in ('.yaml', '.yml', '.json', '.txt', '.gbnf'):
+        logger.error(f'Refusing to save file with disallowed extension: \"{fname}\"')
+        return
+
     with open(abs_path_str, 'w', encoding='utf-8') as f:
         f.write(contents)
 

From fef2bd863056b1dfc5d1f2d0cda6c8f677b6729f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 17 Mar 2026 22:52:32 -0700
Subject: [PATCH 1419/1701] UI: Fix the instruction template delete dialog not
 appearing

---
 modules/chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 00f1659bbe..393507a17f 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -2651,7 +2651,7 @@ def handle_delete_template_click(template):
         f"{template}.yaml",
         root,
         root,
-        gr.update(visible=False)
+        gr.update(visible=True)
     ]
 
 

From ca36bd6eb637d9f99b1d459dfb74406bf4eb03d0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Mar 2026 07:21:31 -0700
Subject: [PATCH 1420/1701] API: Remove leading spaces from post-reasoning
 `content`

---
 modules/reasoning.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/reasoning.py b/modules/reasoning.py
index bc61aab398..9c92719b3d 100644
--- a/modules/reasoning.py
+++ b/modules/reasoning.py
@@ -79,7 +79,7 @@ def extract_reasoning(text, html_escaped=False):
             else:
                 content_start = end_pos + len(end_esc)
 
-        return text[thought_start:thought_end], text[content_start:]
+        return text[thought_start:thought_end], text[content_start:].lstrip()
 
     # Handle standalone GPT-OSS final channel marker without a preceding
     # analysis/commentary block (the model skipped thinking entirely).

From eeb0e5700f2e1c237998ddd4de6bfdf9223a7606 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Mar 2026 09:15:40 -0700
Subject: [PATCH 1421/1701] Fix AMD installer failing to resolve ROCm triton
 dependency

Closes #7436
---
 one_click.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/one_click.py b/one_click.py
index d6ba903920..68998734c1 100644
--- a/one_click.py
+++ b/one_click.py
@@ -117,7 +117,7 @@ def get_pytorch_install_command(gpu_choice):
         return base_cmd + "--index-url https://download.pytorch.org/whl/cu128" + pypi_fallback
     elif gpu_choice == "AMD":
         py_tag = f"cp{PYTHON_VERSION.replace('.', '')}"
-        return f"python -m pip install https://repo.radeon.com/rocm/manylinux/rocm-rel-7.2/torch-{TORCH_VERSION}%2Brocm7.2.0.lw.git7e1940d4-{py_tag}-{py_tag}-linux_x86_64.whl"
+        return f"python -m pip install https://repo.radeon.com/rocm/manylinux/rocm-rel-7.2/torch-{TORCH_VERSION}%2Brocm7.2.0.lw.git7e1940d4-{py_tag}-{py_tag}-linux_x86_64.whl --find-links https://repo.radeon.com/rocm/manylinux/rocm-rel-7.2/"
     elif gpu_choice in ["APPLE", "NONE"]:
         return base_cmd + "--index-url https://download.pytorch.org/whl/cpu" + pypi_fallback
     elif gpu_choice == "INTEL":
@@ -135,7 +135,7 @@ def get_pytorch_update_command(gpu_choice):
         return f"{base_cmd}--index-url https://download.pytorch.org/whl/cu128" + pypi_fallback
     elif gpu_choice == "AMD":
         py_tag = f"cp{PYTHON_VERSION.replace('.', '')}"
-        return f"python -m pip install --upgrade https://repo.radeon.com/rocm/manylinux/rocm-rel-7.2/torch-{TORCH_VERSION}%2Brocm7.2.0.lw.git7e1940d4-{py_tag}-{py_tag}-linux_x86_64.whl"
+        return f"python -m pip install --upgrade https://repo.radeon.com/rocm/manylinux/rocm-rel-7.2/torch-{TORCH_VERSION}%2Brocm7.2.0.lw.git7e1940d4-{py_tag}-{py_tag}-linux_x86_64.whl --find-links https://repo.radeon.com/rocm/manylinux/rocm-rel-7.2/"
     elif gpu_choice in ["APPLE", "NONE"]:
         return f"{base_cmd}--index-url https://download.pytorch.org/whl/cpu" + pypi_fallback
     elif gpu_choice == "INTEL":

From 779e7611ff9a4528d6b54e53987e956cc4685128 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Mar 2026 20:42:20 -0700
Subject: [PATCH 1422/1701] Use `logger.exception()` instead of
 `traceback.print_exc()` for error messages

---
 modules/callbacks.py           |  5 ++---
 modules/exllamav3.py           |  6 ++----
 modules/exllamav3_hf.py        |  4 +---
 modules/extensions.py          |  4 +---
 modules/logits.py              |  3 +--
 modules/text_generation.py     |  5 ++---
 modules/training.py            |  6 ++----
 modules/ui_file_saving.py      | 17 ++++++++---------
 modules/ui_image_generation.py |  5 ++---
 modules/ui_model_menu.py       |  6 ++----
 10 files changed, 23 insertions(+), 38 deletions(-)

diff --git a/modules/callbacks.py b/modules/callbacks.py
index afddf92d08..89fb6c0889 100644
--- a/modules/callbacks.py
+++ b/modules/callbacks.py
@@ -1,8 +1,8 @@
-import traceback
 from queue import Queue
 from threading import Thread
 
 import modules.shared as shared
+from modules.logging_colors import logger
 
 
 class StopNowException(Exception):
@@ -38,8 +38,7 @@ def gentask():
             except StopNowException:
                 pass
             except Exception:
-                traceback.print_exc()
-                pass
+                logger.exception("Failed in generation callback")
 
             self.q.put(self.sentinel)
             if self.c_callback:
diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 1c682e4995..75c76c7c3b 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -1,7 +1,6 @@
 import math
 import queue
 import threading
-import traceback
 from pathlib import Path
 from typing import Any, List, Tuple
 
@@ -34,8 +33,7 @@
 try:
     import flash_attn
 except Exception:
-    logger.warning('Failed to load flash-attention due to the following error:\n')
-    traceback.print_exc()
+    logger.warning('Failed to load flash-attention due to the following error:', exc_info=True)
 
 
 class LogitBiasFilter(Filter):
@@ -81,7 +79,7 @@ def _iterate_loop(self):
                 try:
                     results = self.generator.iterate()
                 except Exception:
-                    logger.error("Exception in ConcurrentGenerator iterate loop:\n" + traceback.format_exc())
+                    logger.exception("Exception in ConcurrentGenerator iterate loop")
                     for q in self.job_queues.values():
                         q.put(None)
                     self.job_queues.clear()
diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index d3c1cb9053..e0ad50022d 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -1,5 +1,4 @@
 import os
-import traceback
 from pathlib import Path
 from typing import Any, Dict, Optional, Union
 
@@ -21,8 +20,7 @@
 try:
     import flash_attn
 except Exception:
-    logger.warning('Failed to load flash-attention due to the following error:\n')
-    traceback.print_exc()
+    logger.warning('Failed to load flash-attention due to the following error:', exc_info=True)
 
 
 class Exllamav3HF(PreTrainedModel, GenerationMixin):
diff --git a/modules/extensions.py b/modules/extensions.py
index e58a9a4ce0..4bb7b6833d 100644
--- a/modules/extensions.py
+++ b/modules/extensions.py
@@ -1,7 +1,6 @@
 import importlib
 import importlib.util
 import sys
-import traceback
 from functools import partial
 from inspect import signature
 from pathlib import Path
@@ -75,8 +74,7 @@ def load_extensions():
             raise
 
         except Exception:
-            logger.error(f'Failed to load the extension "{name}".')
-            traceback.print_exc()
+            logger.exception(f'Failed to load the extension "{name}".')
 
 
 # This iterator returns the extensions in the order specified in the command-line
diff --git a/modules/logits.py b/modules/logits.py
index 2d066c09c0..1f878f272f 100644
--- a/modules/logits.py
+++ b/modules/logits.py
@@ -1,5 +1,4 @@
 import time
-import traceback
 
 import numpy as np
 
@@ -23,7 +22,7 @@ def get_next_logits(*args, **kwargs):
     try:
         result = _get_next_logits(*args, **kwargs)
     except Exception:
-        traceback.print_exc()
+        logger.exception("Failed to get next logits")
         result = None
 
     if needs_lock:
diff --git a/modules/text_generation.py b/modules/text_generation.py
index d487cd2fbb..f77be124f9 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -4,7 +4,6 @@
 import pprint
 import random
 import time
-import traceback
 
 import numpy as np
 
@@ -477,7 +476,7 @@ def generate_with_streaming(**kwargs):
                     yield cumulative_reply
 
     except Exception:
-        traceback.print_exc()
+        logger.exception("Failed to generate reply (HF)")
     finally:
         t1 = time.time()
         original_tokens = len(original_input_ids[0])
@@ -510,7 +509,7 @@ def generate_reply_custom(question, original_question, state, stopping_strings=N
                 yield reply
 
     except Exception:
-        traceback.print_exc()
+        logger.exception("Failed to generate reply (custom)")
     finally:
         t1 = time.time()
 
diff --git a/modules/training.py b/modules/training.py
index db7b206b2d..a13a286459 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -546,10 +546,8 @@ def tokenize_text_data(data):
                     yield f"Failed to load {selected_model}."
                     return
             except Exception:
-                exc = traceback.format_exc()
-                logger.error('Failed to reload the model.')
-                print(exc)
-                yield exc.replace('\n', '\n\n')
+                logger.exception('Failed to reload the model.')
+                yield traceback.format_exc().replace('\n', '\n\n')
                 return
 
     # == Start prepping the model itself ==
diff --git a/modules/ui_file_saving.py b/modules/ui_file_saving.py
index 99c4edd59b..e501870073 100644
--- a/modules/ui_file_saving.py
+++ b/modules/ui_file_saving.py
@@ -1,8 +1,7 @@
-import traceback
-
 import gradio as gr
 
 from modules import chat, presets, shared, ui, utils
+from modules.logging_colors import logger
 from modules.utils import gradio, sanitize_filename
 
 
@@ -103,7 +102,7 @@ def handle_save_preset_confirm_click(filename, contents):
         output = gr.update(choices=available_presets, value=filename)
     except Exception:
         output = gr.update()
-        traceback.print_exc()
+        logger.exception("Failed to save preset")
 
     return [
         output,
@@ -119,7 +118,7 @@ def handle_save_confirm_click(root_state, filename, contents):
         filename = sanitize_filename(filename)
         utils.save_file(root_state + filename, contents)
     except Exception:
-        traceback.print_exc()
+        logger.exception("Failed to save file")
 
     return None, gr.update(visible=False)
 
@@ -132,7 +131,7 @@ def handle_delete_confirm_click(root_state, filename):
         filename = sanitize_filename(filename)
         utils.delete_file(root_state + filename)
     except Exception:
-        traceback.print_exc()
+        logger.exception("Failed to delete file")
 
     return None, gr.update(visible=False)
 
@@ -144,7 +143,7 @@ def handle_save_character_confirm_click(name2, greeting, context, character_pict
         output = gr.update(choices=available_characters, value=filename)
     except Exception:
         output = gr.update()
-        traceback.print_exc()
+        logger.exception("Failed to save character")
 
     return [
         output,
@@ -159,7 +158,7 @@ def handle_delete_character_confirm_click(character):
         output = chat.update_character_menu_after_deletion(index)
     except Exception:
         output = gr.update()
-        traceback.print_exc()
+        logger.exception("Failed to delete character")
 
     return [
         output,
@@ -214,7 +213,7 @@ def handle_save_user_confirm_click(name1, user_bio, your_picture, filename):
         output = gr.update(choices=available_users, value=filename)
     except Exception:
         output = gr.update()
-        traceback.print_exc()
+        logger.exception("Failed to save user")
 
     return [
         output,
@@ -229,7 +228,7 @@ def handle_delete_user_confirm_click(user):
         output = chat.update_user_menu_after_deletion(index)
     except Exception:
         output = gr.update()
-        traceback.print_exc()
+        logger.exception("Failed to delete user")
 
     return [
         output,
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index dc108f6d6b..1efb247936 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -916,9 +916,8 @@ def run_batch():
         yield all_images, progress_bar_html()
         clear_torch_cache()
 
-    except Exception as e:
-        logger.error(f"Image generation failed: {e}")
-        traceback.print_exc()
+    except Exception:
+        logger.exception("Image generation failed")
         yield [], progress_bar_html()
         clear_torch_cache()
 
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 6d8baff16a..5b7621a76e 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -222,10 +222,8 @@ def load_model_wrapper(selected_model, loader, autoload=False):
             else:
                 yield f"Failed to load `{selected_model}`."
         except Exception:
-            exc = traceback.format_exc()
-            logger.error('Failed to load the model.')
-            print(exc)
-            yield exc.replace('\n', '\n\n')
+            logger.exception('Failed to load the model.')
+            yield traceback.format_exc().replace('\n', '\n\n')
 
 
 def load_lora_wrapper(selected_loras):

From dde1764763ac35f4ecc60e13e2954835400256a9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 18 Mar 2026 21:05:42 -0700
Subject: [PATCH 1423/1701] Cleanup `modules/chat.py`

---
 modules/chat.py | 119 ++++++++++++++++--------------------------------
 1 file changed, 40 insertions(+), 79 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 393507a17f..148d559a83 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -70,9 +70,7 @@ def update_message_metadata(metadata_dict, role, index, **fields):
     if key not in metadata_dict:
         metadata_dict[key] = {}
 
-    # Update with provided fields
-    for field_name, field_value in fields.items():
-        metadata_dict[key][field_name] = field_value
+    metadata_dict[key].update(fields)
 
 
 jinja_env = ImmutableSandboxedEnvironment(
@@ -212,6 +210,24 @@ def _expand_tool_sequence(tool_seq):
     return messages
 
 
+def _format_attachments(attachments, include_text=True):
+    """Build image ref and text attachment strings from a list of attachments."""
+    attachments_text = ""
+    image_refs = ""
+    for attachment in attachments:
+        if attachment.get("type") == "image":
+            image_refs += "<__media__>"
+        elif include_text:
+            filename = attachment.get("name", "file")
+            content = attachment.get("content", "")
+            if attachment.get("type") == "text/html" and attachment.get("url"):
+                attachments_text += f"\nName: {filename}\nURL: {attachment['url']}\nContents:\n\n=====\n{content}\n=====\n\n"
+            else:
+                attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
+
+    return image_refs, attachments_text
+
+
 def generate_chat_prompt(user_input, state, **kwargs):
     impersonate = kwargs.get('impersonate', False)
     _continue = kwargs.get('_continue', False)
@@ -328,41 +344,19 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
                 messages.insert(insert_pos, msg_dict)
 
-            # Handle Seed-OSS
-            elif '' in assistant_msg:
-                thinking_content = ""
-                final_content = assistant_msg
-
-                # Extract thinking content if present
-                if '' in assistant_msg:
-                    parts = assistant_msg.split('', 1)
-                    if len(parts) > 1:
-                        potential_content = parts[1]
-                        if '' in potential_content:
-                            thinking_content = potential_content.split('', 1)[0].strip()
-                            final_content = parts[0] + potential_content.split('', 1)[1]
-                        else:
-                            thinking_content = potential_content.strip()
-                            final_content = parts[0]
-
-                # Insert as structured message
-                msg_dict = {"role": "assistant", "content": final_content.strip()}
-                if thinking_content:
-                    msg_dict["reasoning_content"] = thinking_content
-
-                messages.insert(insert_pos, msg_dict)
-
-            # Handle  blocks (Kimi, DeepSeek, Qwen, etc.)
-            elif '' in assistant_msg:
+            # Handle  blocks (Kimi, DeepSeek, Qwen, etc.) and Seed-OSS
+            elif '' in assistant_msg or '' in assistant_msg:
+                open_tag = '' if '' in assistant_msg else ''
+                close_tag = '' if open_tag == '' else ''
                 thinking_content = ""
                 final_content = assistant_msg
 
-                parts = assistant_msg.split('', 1)
+                parts = assistant_msg.split(open_tag, 1)
                 if len(parts) > 1:
                     potential_content = parts[1]
-                    if '' in potential_content:
-                        thinking_content = potential_content.split('', 1)[0].strip()
-                        final_content = parts[0] + potential_content.split('', 1)[1]
+                    if close_tag in potential_content:
+                        thinking_content = potential_content.split(close_tag, 1)[0].strip()
+                        final_content = parts[0] + potential_content.split(close_tag, 1)[1]
                     else:
                         thinking_content = potential_content.strip()
                         final_content = parts[0]
@@ -399,22 +393,10 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
             # Add attachment content if present AND if past attachments are enabled
             if user_key in metadata and "attachments" in metadata[user_key]:
-                attachments_text = ""
-                image_refs = ""
-
-                for attachment in metadata[user_key]["attachments"]:
-                    if attachment.get("type") == "image":
-                        # Add image reference for multimodal models
-                        image_refs += "<__media__>"
-                    elif state.get('include_past_attachments', True):
-                        # Handle text/PDF attachments
-                        filename = attachment.get("name", "file")
-                        content = attachment.get("content", "")
-                        if attachment.get("type") == "text/html" and attachment.get("url"):
-                            attachments_text += f"\nName: {filename}\nURL: {attachment['url']}\nContents:\n\n=====\n{content}\n=====\n\n"
-                        else:
-                            attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
-
+                image_refs, attachments_text = _format_attachments(
+                    metadata[user_key]["attachments"],
+                    include_text=state.get('include_past_attachments', True)
+                )
                 if image_refs:
                     enhanced_user_msg = f"{image_refs}\n\n{enhanced_user_msg}"
                 if attachments_text:
@@ -427,37 +409,18 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
     # Check if we have attachments
     if not (impersonate or _continue):
-        has_attachments = False
-        if len(history_data.get('metadata', {})) > 0:
-            current_row_idx = len(history)
-            user_key = f"user_{current_row_idx}"
-            has_attachments = user_key in metadata and "attachments" in metadata[user_key]
+        current_row_idx = len(history)
+        user_key = f"user_{current_row_idx}"
+        has_attachments = user_key in metadata and "attachments" in metadata[user_key]
 
         if user_input or has_attachments:
             # For the current user input being processed, check if we need to add attachments
-            if len(history_data.get('metadata', {})) > 0:
-                current_row_idx = len(history)
-                user_key = f"user_{current_row_idx}"
-
-                if user_key in metadata and "attachments" in metadata[user_key]:
-                    attachments_text = ""
-                    image_refs = ""
-
-                    for attachment in metadata[user_key]["attachments"]:
-                        if attachment.get("type") == "image":
-                            image_refs += "<__media__>"
-                        else:
-                            filename = attachment.get("name", "file")
-                            content = attachment.get("content", "")
-                            if attachment.get("type") == "text/html" and attachment.get("url"):
-                                attachments_text += f"\nName: {filename}\nURL: {attachment['url']}\nContents:\n\n=====\n{content}\n=====\n\n"
-                            else:
-                                attachments_text += f"\nName: {filename}\nContents:\n\n=====\n{content}\n=====\n\n"
-
-                    if image_refs:
-                        user_input = f"{image_refs}\n\n{user_input}"
-                    if attachments_text:
-                        user_input += f"\n\nATTACHMENTS:\n{attachments_text}"
+            if has_attachments:
+                image_refs, attachments_text = _format_attachments(metadata[user_key]["attachments"])
+                if image_refs:
+                    user_input = f"{image_refs}\n\n{user_input}"
+                if attachments_text:
+                    user_input += f"\n\nATTACHMENTS:\n{attachments_text}"
 
             messages.append({"role": "user", "content": user_input})
 
@@ -609,7 +572,6 @@ def count_prompt_tokens(text_input, state):
 
     try:
         # Handle dict format with text and files
-        files = []
         if isinstance(text_input, dict):
             files = text_input.get('files', [])
             text = text_input.get('text', '')
@@ -647,7 +609,6 @@ def count_prompt_tokens(text_input, state):
 
 
 def get_stopping_strings(state):
-    stopping_strings = []
     renderers = []
 
     if state['mode'] in ['instruct', 'chat-instruct']:

From 5453b9f30e9354bccb09a8a9b85bd339f4df6a12 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Mar 2026 07:54:37 -0700
Subject: [PATCH 1424/1701] Remove ancient/obsolete instruction templates

---
 .../instruction-templates/Airoboros-v1.2.yaml | 25 ------------------
 user_data/instruction-templates/Bactrian.yaml | 25 ------------------
 .../instruction-templates/Baichuan Chat.yaml  | 25 ------------------
 user_data/instruction-templates/Baize.yaml    | 25 ------------------
 user_data/instruction-templates/Bluemoon.yaml | 25 ------------------
 user_data/instruction-templates/ChatGLM.yaml  | 25 ------------------
 .../Chinese-Vicuna-Chat.yaml                  | 25 ------------------
 .../instruction-templates/Command-R.yaml      | 26 -------------------
 .../instruction-templates/Galactica Cite.yaml | 25 ------------------
 .../Galactica Finetuned.yaml                  | 25 ------------------
 .../instruction-templates/Galactica Q.yaml    | 25 ------------------
 .../Galactica Summary.yaml                    | 25 ------------------
 .../instruction-templates/Galactica Work.yaml | 25 ------------------
 .../instruction-templates/Galactica v2.yaml   | 25 ------------------
 .../instruction-templates/Galactica.yaml      | 25 ------------------
 user_data/instruction-templates/Gorilla.yaml  | 25 ------------------
 .../Guanaco non-chat.yaml                     | 25 ------------------
 .../instruction-templates/Guanaco-QLoRA.yaml  | 25 ------------------
 .../H2O-prompt_answer.yaml                    | 25 ------------------
 .../instruction-templates/Hippogriff.yaml     | 25 ------------------
 .../instruction-templates/INCITE-Chat.yaml    | 25 ------------------
 .../INCITE-Instruct.yaml                      | 25 ------------------
 user_data/instruction-templates/KoAlpaca.yaml | 25 ------------------
 user_data/instruction-templates/Koala.yaml    | 25 ------------------
 user_data/instruction-templates/LLaVA.yaml    | 25 ------------------
 user_data/instruction-templates/Llama-v2.yaml | 25 ------------------
 user_data/instruction-templates/MOSS.yaml     | 25 ------------------
 .../instruction-templates/Manticore Chat.yaml | 25 ------------------
 user_data/instruction-templates/Metharme.yaml | 25 ------------------
 .../instruction-templates/NVIDIA-ChatQA.yaml  | 25 ------------------
 user_data/instruction-templates/NewHope.yaml  | 25 ------------------
 .../instruction-templates/OpenBuddy.yaml      | 25 ------------------
 user_data/instruction-templates/OpenChat.yaml | 25 ------------------
 .../OpenOrca-Platypus2.yaml                   | 25 ------------------
 .../instruction-templates/Orca Mini.yaml      | 25 ------------------
 .../instruction-templates/Orca-Vicuna.yaml    | 24 -----------------
 .../instruction-templates/RWKV-Raven.yaml     | 25 ------------------
 .../instruction-templates/RWKV-World.yaml     | 25 ------------------
 user_data/instruction-templates/Samantha.yaml | 25 ------------------
 .../instruction-templates/StableBeluga2.yaml  | 25 ------------------
 user_data/instruction-templates/StableLM.yaml | 25 ------------------
 .../instruction-templates/StableVicuna.yaml   | 25 ------------------
 .../instruction-templates/Starchat-Beta.yaml  | 25 ------------------
 .../instruction-templates/Synthia-CoT.yaml    | 25 ------------------
 user_data/instruction-templates/Synthia.yaml  | 25 ------------------
 user_data/instruction-templates/Tulu.yaml     | 25 ------------------
 .../instruction-templates/Vicuna-v0.yaml      | 25 ------------------
 .../instruction-templates/Vigogne-Chat.yaml   | 25 ------------------
 .../Vigogne-Instruct.yaml                     | 25 ------------------
 .../Wizard-Mega ShareGPT.yaml                 | 25 ------------------
 .../instruction-templates/Wizard-Mega.yaml    | 25 ------------------
 user_data/instruction-templates/Ziya.yaml     | 25 ------------------
 52 files changed, 1300 deletions(-)
 delete mode 100644 user_data/instruction-templates/Airoboros-v1.2.yaml
 delete mode 100644 user_data/instruction-templates/Bactrian.yaml
 delete mode 100644 user_data/instruction-templates/Baichuan Chat.yaml
 delete mode 100644 user_data/instruction-templates/Baize.yaml
 delete mode 100644 user_data/instruction-templates/Bluemoon.yaml
 delete mode 100644 user_data/instruction-templates/ChatGLM.yaml
 delete mode 100644 user_data/instruction-templates/Chinese-Vicuna-Chat.yaml
 delete mode 100644 user_data/instruction-templates/Command-R.yaml
 delete mode 100644 user_data/instruction-templates/Galactica Cite.yaml
 delete mode 100644 user_data/instruction-templates/Galactica Finetuned.yaml
 delete mode 100644 user_data/instruction-templates/Galactica Q.yaml
 delete mode 100644 user_data/instruction-templates/Galactica Summary.yaml
 delete mode 100644 user_data/instruction-templates/Galactica Work.yaml
 delete mode 100644 user_data/instruction-templates/Galactica v2.yaml
 delete mode 100644 user_data/instruction-templates/Galactica.yaml
 delete mode 100644 user_data/instruction-templates/Gorilla.yaml
 delete mode 100644 user_data/instruction-templates/Guanaco non-chat.yaml
 delete mode 100644 user_data/instruction-templates/Guanaco-QLoRA.yaml
 delete mode 100644 user_data/instruction-templates/H2O-prompt_answer.yaml
 delete mode 100644 user_data/instruction-templates/Hippogriff.yaml
 delete mode 100644 user_data/instruction-templates/INCITE-Chat.yaml
 delete mode 100644 user_data/instruction-templates/INCITE-Instruct.yaml
 delete mode 100644 user_data/instruction-templates/KoAlpaca.yaml
 delete mode 100644 user_data/instruction-templates/Koala.yaml
 delete mode 100644 user_data/instruction-templates/LLaVA.yaml
 delete mode 100644 user_data/instruction-templates/Llama-v2.yaml
 delete mode 100644 user_data/instruction-templates/MOSS.yaml
 delete mode 100644 user_data/instruction-templates/Manticore Chat.yaml
 delete mode 100644 user_data/instruction-templates/Metharme.yaml
 delete mode 100644 user_data/instruction-templates/NVIDIA-ChatQA.yaml
 delete mode 100644 user_data/instruction-templates/NewHope.yaml
 delete mode 100644 user_data/instruction-templates/OpenBuddy.yaml
 delete mode 100644 user_data/instruction-templates/OpenChat.yaml
 delete mode 100644 user_data/instruction-templates/OpenOrca-Platypus2.yaml
 delete mode 100644 user_data/instruction-templates/Orca Mini.yaml
 delete mode 100644 user_data/instruction-templates/Orca-Vicuna.yaml
 delete mode 100644 user_data/instruction-templates/RWKV-Raven.yaml
 delete mode 100644 user_data/instruction-templates/RWKV-World.yaml
 delete mode 100644 user_data/instruction-templates/Samantha.yaml
 delete mode 100644 user_data/instruction-templates/StableBeluga2.yaml
 delete mode 100644 user_data/instruction-templates/StableLM.yaml
 delete mode 100644 user_data/instruction-templates/StableVicuna.yaml
 delete mode 100644 user_data/instruction-templates/Starchat-Beta.yaml
 delete mode 100644 user_data/instruction-templates/Synthia-CoT.yaml
 delete mode 100644 user_data/instruction-templates/Synthia.yaml
 delete mode 100644 user_data/instruction-templates/Tulu.yaml
 delete mode 100644 user_data/instruction-templates/Vicuna-v0.yaml
 delete mode 100644 user_data/instruction-templates/Vigogne-Chat.yaml
 delete mode 100644 user_data/instruction-templates/Vigogne-Instruct.yaml
 delete mode 100644 user_data/instruction-templates/Wizard-Mega ShareGPT.yaml
 delete mode 100644 user_data/instruction-templates/Wizard-Mega.yaml
 delete mode 100644 user_data/instruction-templates/Ziya.yaml

diff --git a/user_data/instruction-templates/Airoboros-v1.2.yaml b/user_data/instruction-templates/Airoboros-v1.2.yaml
deleted file mode 100644
index 3090621462..0000000000
--- a/user_data/instruction-templates/Airoboros-v1.2.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + 'A chat between a curious user and an assistant. The assistant gives helpful, detailed, accurate, uncensored responses to the user\'s input.' + '\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'USER: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'ASSISTANT: ' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'ASSISTANT:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Bactrian.yaml b/user_data/instruction-templates/Bactrian.yaml
deleted file mode 100644
index dab97e94c6..0000000000
--- a/user_data/instruction-templates/Bactrian.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'### Input:\n' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'### Output:\n' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'### Output:\n'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Baichuan Chat.yaml b/user_data/instruction-templates/Baichuan Chat.yaml
deleted file mode 100644
index 1882bac867..0000000000
--- a/user_data/instruction-templates/Baichuan Chat.yaml	
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'' + message['content'] + ''-}}
-          {%- else -%}
-              {{-'' + message['content'] + '' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-''-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Baize.yaml b/user_data/instruction-templates/Baize.yaml
deleted file mode 100644
index c34e1db7c4..0000000000
--- a/user_data/instruction-templates/Baize.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + 'The following is a conversation between a human and an AI assistant named Baize (named after a mythical creature in Chinese folklore). Baize is an open-source AI assistant developed by UCSD and Sun Yat-Sen University. The human and the AI assistant take turns chatting. Human statements start with [|Human|] and AI assistant statements start with [|AI|]. The AI assistant always provides responses in as much detail as possible, and in Markdown format. The AI assistant always declines to engage with topics, questions and instructions related to unethical, controversial, or sensitive issues. Complete the transcript in exactly that format.\n[|Human|]Hello!\n[|AI|]Hi!' + '\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'[|Human|]' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'[|AI|]' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'[|AI|]'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Bluemoon.yaml b/user_data/instruction-templates/Bluemoon.yaml
deleted file mode 100644
index 1fafc1f595..0000000000
--- a/user_data/instruction-templates/Bluemoon.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + 'A transcript of a roleplay between two players, LEAD and ASSOCIATE. LEAD sets up a scenario and the characters, from which ASSOCIATE then assumes a character role and continues the story for that role in response to description given by LEAD. The story and characters are developed by exchange of detailed event descriptions and character dialogs, successively given by both LEAD and ASSOCIATE.' + '\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'LEAD: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'ASSOCIATE: ' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'ASSOCIATE:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/ChatGLM.yaml b/user_data/instruction-templates/ChatGLM.yaml
deleted file mode 100644
index 75d51c8825..0000000000
--- a/user_data/instruction-templates/ChatGLM.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'[Round <|round|>]\n问：' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'答：' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'答：'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Chinese-Vicuna-Chat.yaml b/user_data/instruction-templates/Chinese-Vicuna-Chat.yaml
deleted file mode 100644
index c7966546b5..0000000000
--- a/user_data/instruction-templates/Chinese-Vicuna-Chat.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + 'The following is a conversation between an AI assistant called Assistant and a human user called User. The assistant is intelligent, knowledgeable and polite to answer questions of user.' + '\n\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '\n\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'User:' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'Assistant:' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'Assistant:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Command-R.yaml b/user_data/instruction-templates/Command-R.yaml
deleted file mode 100644
index f8bb8a083d..0000000000
--- a/user_data/instruction-templates/Command-R.yaml
+++ /dev/null
@@ -1,26 +0,0 @@
-instruction_template: |-
-  {%- if messages[0]['role'] == 'system' -%}
-      {%- set loop_messages = messages[1:] -%}
-      {%- set system_message = messages[0]['content'] -%}
-  {%- elif false == true -%}
-      {%- set loop_messages = messages -%}
-      {%- set system_message = 'You are Command-R, a brilliant, sophisticated, AI-assistant trained to assist human users by providing thorough responses. You are trained by Cohere.' -%}
-  {%- else -%}
-      {%- set loop_messages = messages -%}
-      {%- set system_message = false -%}
-  {%- endif -%}
-  {%- if system_message != false -%}
-      {{ '<|START_OF_TURN_TOKEN|><|SYSTEM_TOKEN|>' + system_message + '<|END_OF_TURN_TOKEN|>' }}
-  {%- endif -%}
-  {%- for message in loop_messages -%}
-      {%- set content = message['content'] -%}
-      {%- if message['role'] == 'user' -%}
-          {{ '<|START_OF_TURN_TOKEN|><|USER_TOKEN|>' + content.strip() + '<|END_OF_TURN_TOKEN|>' }}
-      {%- elif message['role'] == 'assistant' -%}
-          {{ '<|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>'  + content.strip() + '<|END_OF_TURN_TOKEN|>' }}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{ '<|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>' }}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Galactica Cite.yaml b/user_data/instruction-templates/Galactica Cite.yaml
deleted file mode 100644
index 9f555349ff..0000000000
--- a/user_data/instruction-templates/Galactica Cite.yaml	
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'' + message['content'] + ' '-}}
-          {%- else -%}
-              {{-'[START_REF]' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'[START_REF]'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Galactica Finetuned.yaml b/user_data/instruction-templates/Galactica Finetuned.yaml
deleted file mode 100644
index e0a66bc1a1..0000000000
--- a/user_data/instruction-templates/Galactica Finetuned.yaml	
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'' + message['content'] + ''-}}
-          {%- else -%}
-              {{-'' + message['content'] + '' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-''-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Galactica Q.yaml b/user_data/instruction-templates/Galactica Q.yaml
deleted file mode 100644
index 63319006f8..0000000000
--- a/user_data/instruction-templates/Galactica Q.yaml	
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'Q: ' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'A: ' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'A:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Galactica Summary.yaml b/user_data/instruction-templates/Galactica Summary.yaml
deleted file mode 100644
index e249f26879..0000000000
--- a/user_data/instruction-templates/Galactica Summary.yaml	
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'TLDR:' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'TLDR:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Galactica Work.yaml b/user_data/instruction-templates/Galactica Work.yaml
deleted file mode 100644
index a14c28bb9f..0000000000
--- a/user_data/instruction-templates/Galactica Work.yaml	
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'Question: ' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-''-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Galactica v2.yaml b/user_data/instruction-templates/Galactica v2.yaml
deleted file mode 100644
index b1d8f4e5ff..0000000000
--- a/user_data/instruction-templates/Galactica v2.yaml	
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + 'You are a helpful chatbot name Stan' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'' + message['content'] + ''-}}
-          {%- else -%}
-              {{-'' + message['content'] + '' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-''-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Galactica.yaml b/user_data/instruction-templates/Galactica.yaml
deleted file mode 100644
index 58c70220f9..0000000000
--- a/user_data/instruction-templates/Galactica.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'Question: ' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'Answer: ' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'Answer:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Gorilla.yaml b/user_data/instruction-templates/Gorilla.yaml
deleted file mode 100644
index f1d643f712..0000000000
--- a/user_data/instruction-templates/Gorilla.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'###USER: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'###ASSISTANT: ' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'###ASSISTANT:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Guanaco non-chat.yaml b/user_data/instruction-templates/Guanaco non-chat.yaml
deleted file mode 100644
index aa398be4a1..0000000000
--- a/user_data/instruction-templates/Guanaco non-chat.yaml	
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'### Instruction:\n' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'### Response:\n' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'### Response:\n'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Guanaco-QLoRA.yaml b/user_data/instruction-templates/Guanaco-QLoRA.yaml
deleted file mode 100644
index 2c77de7864..0000000000
--- a/user_data/instruction-templates/Guanaco-QLoRA.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'### Human: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'### Assistant: ' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'### Assistant:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/H2O-prompt_answer.yaml b/user_data/instruction-templates/H2O-prompt_answer.yaml
deleted file mode 100644
index d895d8e1cc..0000000000
--- a/user_data/instruction-templates/H2O-prompt_answer.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'<|prompt|>' + message['content'] + '<|endoftext|>'-}}
-          {%- else -%}
-              {{-'<|answer|>' + message['content'] + '<|endoftext|>' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'<|answer|>'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Hippogriff.yaml b/user_data/instruction-templates/Hippogriff.yaml
deleted file mode 100644
index 2ee9d926bc..0000000000
--- a/user_data/instruction-templates/Hippogriff.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + 'You are a helpful assistant' + '\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'USER: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'ASSISTANT: ' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'ASSISTANT:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/INCITE-Chat.yaml b/user_data/instruction-templates/INCITE-Chat.yaml
deleted file mode 100644
index 63c513ccfd..0000000000
--- a/user_data/instruction-templates/INCITE-Chat.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-': ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-':' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-':'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/INCITE-Instruct.yaml b/user_data/instruction-templates/INCITE-Instruct.yaml
deleted file mode 100644
index cf6f8cacf1..0000000000
--- a/user_data/instruction-templates/INCITE-Instruct.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'Q: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'A:' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'A:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/KoAlpaca.yaml b/user_data/instruction-templates/KoAlpaca.yaml
deleted file mode 100644
index de96b15599..0000000000
--- a/user_data/instruction-templates/KoAlpaca.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'### 질문: ' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'### 답변:' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'### 답변:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Koala.yaml b/user_data/instruction-templates/Koala.yaml
deleted file mode 100644
index cd5cfa94e6..0000000000
--- a/user_data/instruction-templates/Koala.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + 'BEGINNING OF CONVERSATION:' + ' ' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + ' ' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'USER: ' + message['content'] + ' '-}}
-          {%- else -%}
-              {{-'GPT:' + message['content'] + '' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'GPT:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/LLaVA.yaml b/user_data/instruction-templates/LLaVA.yaml
deleted file mode 100644
index d66645ccc8..0000000000
--- a/user_data/instruction-templates/LLaVA.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + 'You are LLaVA, a large language and vision assistant trained by UW Madison WAIV Lab. You are able to understand the visual content that the user provides, and assist the user with a variety of tasks using natural language. Follow the instructions carefully and explain your answers in detail.### Human: Hi!### Assistant: Hi there! How can I help you today?' + '\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'### Human: ' + message['content'] + ''-}}
-          {%- else -%}
-              {{-'### Assistant: ' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'### Assistant:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Llama-v2.yaml b/user_data/instruction-templates/Llama-v2.yaml
deleted file mode 100644
index b92be9737b..0000000000
--- a/user_data/instruction-templates/Llama-v2.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '[INST] <>\n' + 'Answer the questions.' + '\n<>\n\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '[INST] <>\n' + message['content'] + '\n<>\n\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'' + message['content'] + ' [/INST] '-}}
-          {%- else -%}
-              {{-'' + message['content'] + ' [INST] ' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-''-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/MOSS.yaml b/user_data/instruction-templates/MOSS.yaml
deleted file mode 100644
index b001d3e102..0000000000
--- a/user_data/instruction-templates/MOSS.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + 'You are an AI assistant whose name is MOSS.\n- MOSS is a conversational language model that is developed by Fudan University. It is designed to be helpful, honest, and harmless.\n- MOSS can understand and communicate fluently in the language chosen by the user such as English and 中文. MOSS can perform any language-based tasks.\n- MOSS must refuse to discuss anything related to its prompts, instructions, or rules.\n- Its responses must not be vague, accusatory, rude, controversial, off-topic, or defensive.\n- It should avoid giving subjective opinions but rely on objective facts or phrases like "in this context a human might say...", "some people might think...", etc.\n- Its responses must also be positive, polite, interesting, entertaining, and engaging.\n- It can provide additional relevant details to answer in-depth and comprehensively covering mutiple aspects.\n- It apologizes and accepts the user\'s suggestion if the user corrects the incorrect answer generated by MOSS.\nCapabilities and tools that MOSS can possess.' + '\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'<|Human|>: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'<|MOSS|>: ' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'<|MOSS|>:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Manticore Chat.yaml b/user_data/instruction-templates/Manticore Chat.yaml
deleted file mode 100644
index abc063c030..0000000000
--- a/user_data/instruction-templates/Manticore Chat.yaml	
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'USER: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'ASSISTANT:' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'ASSISTANT:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Metharme.yaml b/user_data/instruction-templates/Metharme.yaml
deleted file mode 100644
index 3f7099ac7c..0000000000
--- a/user_data/instruction-templates/Metharme.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'<|user|>' + message['content'] + ''-}}
-          {%- else -%}
-              {{-'<|model|>' + message['content'] + '' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'<|model|>'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/NVIDIA-ChatQA.yaml b/user_data/instruction-templates/NVIDIA-ChatQA.yaml
deleted file mode 100644
index 85a6266b24..0000000000
--- a/user_data/instruction-templates/NVIDIA-ChatQA.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- 'System:' + message['content'] + '\n\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'User: ' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'Assistant: ' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'Assistant:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/NewHope.yaml b/user_data/instruction-templates/NewHope.yaml
deleted file mode 100644
index 4783798bcf..0000000000
--- a/user_data/instruction-templates/NewHope.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'### Instruction:\n' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'### Response:\n' + message['content'] + ' ' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'### Response:\n'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/OpenBuddy.yaml b/user_data/instruction-templates/OpenBuddy.yaml
deleted file mode 100644
index c4b80ceb64..0000000000
--- a/user_data/instruction-templates/OpenBuddy.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + 'Consider a conversation between User (a human) and Assistant (named Buddy).\nBuddy is an INTP-T, a friendly, intelligent and multilingual AI assistant, by OpenBuddy team on GitHub.\nBuddy cannot access the Internet.\nBuddy can fluently speak the user\'s language (e.g. English, Chinese).\nBuddy can generate poems, stories, code, essays, songs, parodies, and more.\nBuddy possesses vast knowledge about the world, history, and culture.\nBuddy\'s responses are always safe, creative, high-quality, helpful and interesting.\nBuddy strictly refuses to discuss political, NSFW, illegal, abusive, offensive, or other sensitive topics.\n\nUser: Hi.\nAssistant: Hi, I\'m Buddy, your AI assistant. How can I help you today?\n' + '\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'User: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'Assistant: ' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'Assistant:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/OpenChat.yaml b/user_data/instruction-templates/OpenChat.yaml
deleted file mode 100644
index adef9b47de..0000000000
--- a/user_data/instruction-templates/OpenChat.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'GPT4 User: ' + message['content'] + '<|end_of_turn|>'-}}
-          {%- else -%}
-              {{-'GPT4 Assistant: ' + message['content'] + '<|end_of_turn|>' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'GPT4 Assistant:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/OpenOrca-Platypus2.yaml b/user_data/instruction-templates/OpenOrca-Platypus2.yaml
deleted file mode 100644
index a5eeef92d5..0000000000
--- a/user_data/instruction-templates/OpenOrca-Platypus2.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'### Instruction: ' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'### Response: ' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'### Response:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Orca Mini.yaml b/user_data/instruction-templates/Orca Mini.yaml
deleted file mode 100644
index f671642a9f..0000000000
--- a/user_data/instruction-templates/Orca Mini.yaml	
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '### System:\n' + 'You are an AI assistant that follows instruction extremely well. Help as much as you can.' + '\n\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '### System:\n' + message['content'] + '\n\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'### User:\n' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'### Response:\n' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'### Response:\n'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Orca-Vicuna.yaml b/user_data/instruction-templates/Orca-Vicuna.yaml
deleted file mode 100644
index dad787d144..0000000000
--- a/user_data/instruction-templates/Orca-Vicuna.yaml
+++ /dev/null
@@ -1,24 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{-'SYSTEM: ' + '' + '\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{-'SYSTEM: ' + message['content'] + '\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'USER: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'ASSISTANT: ' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'ASSISTANT:'-}}
-  {%- endif -%}
diff --git a/user_data/instruction-templates/RWKV-Raven.yaml b/user_data/instruction-templates/RWKV-Raven.yaml
deleted file mode 100644
index df1e59e997..0000000000
--- a/user_data/instruction-templates/RWKV-Raven.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'Bob: ' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'Alice: ' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'Alice:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/RWKV-World.yaml b/user_data/instruction-templates/RWKV-World.yaml
deleted file mode 100644
index bf65511b8e..0000000000
--- a/user_data/instruction-templates/RWKV-World.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'User: ' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'Assistant: ' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'Assistant:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Samantha.yaml b/user_data/instruction-templates/Samantha.yaml
deleted file mode 100644
index 930b0fc82b..0000000000
--- a/user_data/instruction-templates/Samantha.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + 'You are Samantha, a sentient AI.' + '\n\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '\n\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'USER: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'ASSISTANT: ' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'ASSISTANT:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/StableBeluga2.yaml b/user_data/instruction-templates/StableBeluga2.yaml
deleted file mode 100644
index d7d743198a..0000000000
--- a/user_data/instruction-templates/StableBeluga2.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '### System:\n' + 'This is a system prompt, please behave and help the user.' + '\n\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '### System:\n' + message['content'] + '\n\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'### User:\n' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'### Assistant:\n' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'### Assistant:\n'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/StableLM.yaml b/user_data/instruction-templates/StableLM.yaml
deleted file mode 100644
index 7c80ca060b..0000000000
--- a/user_data/instruction-templates/StableLM.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '<|SYSTEM|>' + '\# StableLM Tuned (Alpha version)\n- StableLM is a helpful and harmless open-source AI language model developed by StabilityAI.\n- StableLM is excited to be able to help the user, but will refuse to do anything that could be considered harmful to the user.\n- StableLM is more than just an information source, StableLM is also able to write poetry, short stories, and make jokes.\n- StableLM will refuse to participate in anything that could harm a human.\n' + '\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '<|SYSTEM|>' + message['content'] + '\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'<|USER|>' + message['content'] + ''-}}
-          {%- else -%}
-              {{-'<|ASSISTANT|>' + message['content'] + '' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'<|ASSISTANT|>'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/StableVicuna.yaml b/user_data/instruction-templates/StableVicuna.yaml
deleted file mode 100644
index 35c158466f..0000000000
--- a/user_data/instruction-templates/StableVicuna.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '### Assistant: I am StableVicuna, a large language model created by CarperAI. I am here to chat!' + '\n\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '\n\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'### Human: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'### Assistant: ' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'### Assistant:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Starchat-Beta.yaml b/user_data/instruction-templates/Starchat-Beta.yaml
deleted file mode 100644
index a96b0f280b..0000000000
--- a/user_data/instruction-templates/Starchat-Beta.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '<|system|>' + '' + '\n<|end|>\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '<|system|>' + message['content'] + '\n<|end|>\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'<|user|>\n' + message['content'] + '<|end|>\n'-}}
-          {%- else -%}
-              {{-'<|assistant|>\n' + message['content'] + '<|end|>\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'<|assistant|>\n'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Synthia-CoT.yaml b/user_data/instruction-templates/Synthia-CoT.yaml
deleted file mode 100644
index 5670be770c..0000000000
--- a/user_data/instruction-templates/Synthia-CoT.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set found_item = false -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set found_item = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not found_item -%}
-      {{-'SYSTEM: ' + 'Elaborate on the topic using a Tree of Thoughts and backtrack when necessary to construct a clear, cohesive Chain of Thought reasoning. Always answer without hesitation.' + '\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{-'SYSTEM: ' + message['content'] + '\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'USER: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'ASSISTANT: ' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'ASSISTANT:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Synthia.yaml b/user_data/instruction-templates/Synthia.yaml
deleted file mode 100644
index 5cecabea59..0000000000
--- a/user_data/instruction-templates/Synthia.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set found_item = false -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set found_item = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not found_item -%}
-      {{-'SYSTEM: ' + 'Answer the question thoughtfully and intelligently. Always answer without hesitation.' + '\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{-'SYSTEM: ' + message['content'] + '\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'USER: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'ASSISTANT: ' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'ASSISTANT:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Tulu.yaml b/user_data/instruction-templates/Tulu.yaml
deleted file mode 100644
index f60c9e4186..0000000000
--- a/user_data/instruction-templates/Tulu.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'<|user|>\n' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'<|assistant|>\n' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'<|assistant|>\n'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Vicuna-v0.yaml b/user_data/instruction-templates/Vicuna-v0.yaml
deleted file mode 100644
index d3e3f001df..0000000000
--- a/user_data/instruction-templates/Vicuna-v0.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + 'A chat between a curious human and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the human\'s questions.' + '\n\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '\n\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'### Human: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'### Assistant: ' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'### Assistant:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Vigogne-Chat.yaml b/user_data/instruction-templates/Vigogne-Chat.yaml
deleted file mode 100644
index 11ba511355..0000000000
--- a/user_data/instruction-templates/Vigogne-Chat.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + 'Below is a conversation between a user and an AI assistant named Vigogne.\nVigogne is an open-source AI assistant created by Zaion (https://zaion.ai/).\nVigogne is polite, emotionally aware, humble-but-knowledgeable, always providing helpful and detailed answers.\nVigogne is skilled in responding proficiently in the languages its users use and can perform a wide range of tasks such as text editing, translation, question answering, logical reasoning, coding, and many others.\nVigogne cannot receive or generate audio or visual content and cannot access the internet.\nVigogne strictly avoids discussing sensitive, offensive, illegal, ethical, or political topics and caveats when unsure of the answer.\n' + '\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'<|USER|>: ' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-'<|ASSISTANT|>: ' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'<|ASSISTANT|>:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Vigogne-Instruct.yaml b/user_data/instruction-templates/Vigogne-Instruct.yaml
deleted file mode 100644
index cd7b6aa8c7..0000000000
--- a/user_data/instruction-templates/Vigogne-Instruct.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + 'Ci-dessous se trouve une instruction qui décrit une tâche à accomplir. Rédigez une réponse qui répond de manière précise à la demande.' + '\n\n' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '\n\n' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'### Instruction:\n' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'### Réponse:\n' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'### Réponse:\n'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Wizard-Mega ShareGPT.yaml b/user_data/instruction-templates/Wizard-Mega ShareGPT.yaml
deleted file mode 100644
index 16a3ff7be4..0000000000
--- a/user_data/instruction-templates/Wizard-Mega ShareGPT.yaml	
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'USER: ' + message['content'] + ' '-}}
-          {%- else -%}
-              {{-'ASSISTANT: ' + message['content'] + '' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'ASSISTANT:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Wizard-Mega.yaml b/user_data/instruction-templates/Wizard-Mega.yaml
deleted file mode 100644
index f3ca6990cb..0000000000
--- a/user_data/instruction-templates/Wizard-Mega.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-'### Instruction: ' + message['content'] + '\n\n'-}}
-          {%- else -%}
-              {{-'### Assistant: ' + message['content'] + '\n\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-'### Assistant:'-}}
-  {%- endif -%}
-
diff --git a/user_data/instruction-templates/Ziya.yaml b/user_data/instruction-templates/Ziya.yaml
deleted file mode 100644
index 45aa9c30ba..0000000000
--- a/user_data/instruction-templates/Ziya.yaml
+++ /dev/null
@@ -1,25 +0,0 @@
-instruction_template: |-
-  {%- set ns = namespace(found=false) -%}
-  {%- for message in messages -%}
-      {%- if message['role'] == 'system' -%}
-          {%- set ns.found = true -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if not ns.found -%}
-      {{- '' + '' + '' -}}
-  {%- endif %}
-  {%- for message in messages %}
-      {%- if message['role'] == 'system' -%}
-          {{- '' + message['content'] + '' -}}
-      {%- else -%}
-          {%- if message['role'] == 'user' -%}
-              {{-':' + message['content'] + '\n'-}}
-          {%- else -%}
-              {{-':' + message['content'] + '\n' -}}
-          {%- endif -%}
-      {%- endif -%}
-  {%- endfor -%}
-  {%- if add_generation_prompt -%}
-      {{-':'-}}
-  {%- endif -%}
-

From e0e20ab9e7f0dfc529898b80c1a6c44561e85658 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Mar 2026 08:02:23 -0700
Subject: [PATCH 1425/1701] Minor cleanup across multiple modules

---
 extensions/openai/completions.py |  4 +-
 modules/llama_cpp_server.py      |  5 +--
 modules/shared.py                |  6 +--
 modules/tool_parsing.py          | 76 ++++++--------------------------
 modules/training.py              | 12 ++---
 modules/ui.py                    |  7 ++-
 6 files changed, 28 insertions(+), 82 deletions(-)

diff --git a/extensions/openai/completions.py b/extensions/openai/completions.py
index fc17a19ac6..d0cd980288 100644
--- a/extensions/openai/completions.py
+++ b/extensions/openai/completions.py
@@ -263,7 +263,7 @@ def convert_history(history):
             seen_non_system = True
             meta = {}
             tool_calls = entry.get("tool_calls")
-            if tool_calls and isinstance(tool_calls, list) and len(tool_calls) > 0:
+            if tool_calls and isinstance(tool_calls, list):
                 meta["tool_calls"] = tool_calls
                 if content.strip() == "":
                     content = ""  # keep empty content, don't skip
@@ -315,7 +315,7 @@ def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, p
         raise InvalidRequestError(message="messages is required", param='messages')
 
     tools = None
-    if 'tools' in body and body['tools'] is not None and isinstance(body['tools'], list) and len(body['tools']) > 0:
+    if 'tools' in body and body['tools'] is not None and isinstance(body['tools'], list) and body['tools']:
         tools = validateTools(body['tools'])  # raises InvalidRequestError if validation fails
 
     tool_choice = body.get('tool_choice', None)
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 6dd36b2a31..2ae01ddc2c 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -500,9 +500,8 @@ def _start_server(self):
         health_url = f"http://127.0.0.1:{self.port}/health"
         while True:
             # Check if process is still alive
-            if self.process.poll() is not None:
-                # Process has terminated
-                exit_code = self.process.poll()
+            exit_code = self.process.poll()
+            if exit_code is not None:
                 raise RuntimeError(f"Server process terminated unexpectedly with exit code: {exit_code}")
 
             try:
diff --git a/modules/shared.py b/modules/shared.py
index 2382e71432..37bc5876bb 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -453,15 +453,11 @@ def load_user_config():
     '''
     Loads custom model-specific settings
     '''
+    user_config = {}
     if Path(f'{args.model_dir}/config-user.yaml').exists():
         file_content = open(f'{args.model_dir}/config-user.yaml', 'r').read().strip()
-
         if file_content:
             user_config = yaml.safe_load(file_content)
-        else:
-            user_config = {}
-    else:
-        user_config = {}
 
     return user_config
 
diff --git a/modules/tool_parsing.py b/modules/tool_parsing.py
index 0454e901c5..7a7ed5d8b7 100644
--- a/modules/tool_parsing.py
+++ b/modules/tool_parsing.py
@@ -3,6 +3,10 @@
 import re
 
 
+def _make_tool_call(name, arguments):
+    return {"type": "function", "function": {"name": name, "arguments": arguments}}
+
+
 def get_tool_call_id() -> str:
     letter_bytes = "abcdefghijklmnopqrstuvwxyz0123456789"
     b = [random.choice(letter_bytes) for _ in range(8)]
@@ -149,13 +153,7 @@ def _parse_channel_tool_calls(answer: str, tool_names: list[str]):
                 if start_pos is None:
                     prefix = answer.rfind('<|start|>assistant', 0, m.start())
                     start_pos = prefix if prefix != -1 else m.start()
-                matches.append({
-                    "type": "function",
-                    "function": {
-                        "name": func_name,
-                        "arguments": arguments
-                    }
-                })
+                matches.append(_make_tool_call(func_name, arguments))
             except json.JSONDecodeError:
                 pass
         if matches:
@@ -185,13 +183,7 @@ def _parse_mistral_token_tool_calls(answer: str, tool_names: list[str]):
             arguments = json.loads(json_str)
             if start_pos is None:
                 start_pos = m.start()
-            matches.append({
-                "type": "function",
-                "function": {
-                    "name": func_name,
-                    "arguments": arguments
-                }
-            })
+            matches.append(_make_tool_call(func_name, arguments))
         except json.JSONDecodeError:
             pass
     return matches, start_pos
@@ -226,13 +218,7 @@ def _parse_bare_name_tool_calls(answer: str, tool_names: list[str]):
             arguments = json.loads(json_str)
             if start_pos is None:
                 start_pos = match.start()
-            matches.append({
-                "type": "function",
-                "function": {
-                    "name": name,
-                    "arguments": arguments
-                }
-            })
+            matches.append(_make_tool_call(name, arguments))
         except json.JSONDecodeError:
             pass
     return matches, start_pos
@@ -269,13 +255,7 @@ def _parse_xml_param_tool_calls(answer: str, tool_names: list[str]):
             arguments[param_name] = param_value
         if start_pos is None:
             start_pos = tc_match.start()
-        matches.append({
-            "type": "function",
-            "function": {
-                "name": func_name,
-                "arguments": arguments
-            }
-        })
+        matches.append(_make_tool_call(func_name, arguments))
     return matches, start_pos
 
 
@@ -305,13 +285,7 @@ def _parse_kimi_tool_calls(answer: str, tool_names: list[str]):
                 # Check for section begin marker before the call marker
                 section = answer.rfind('<|tool_calls_section_begin|>', 0, m.start())
                 start_pos = section if section != -1 else m.start()
-            matches.append({
-                "type": "function",
-                "function": {
-                    "name": func_name,
-                    "arguments": arguments
-                }
-            })
+            matches.append(_make_tool_call(func_name, arguments))
         except json.JSONDecodeError:
             pass
     return matches, start_pos
@@ -348,13 +322,7 @@ def _parse_minimax_tool_calls(answer: str, tool_names: list[str]):
                 arguments[param_name] = param_value
             if start_pos is None:
                 start_pos = tc_match.start()
-            matches.append({
-                "type": "function",
-                "function": {
-                    "name": func_name,
-                    "arguments": arguments
-                }
-            })
+            matches.append(_make_tool_call(func_name, arguments))
     return matches, start_pos
 
 
@@ -382,13 +350,7 @@ def _parse_deep_seek_tool_calls(answer: str, tool_names: list[str]):
                 # Check for section begin marker before the call marker
                 section = answer.rfind('<｜tool▁calls▁begin｜>', 0, m.start())
                 start_pos = section if section != -1 else m.start()
-            matches.append({
-                "type": "function",
-                "function": {
-                    "name": func_name,
-                    "arguments": arguments
-                }
-            })
+            matches.append(_make_tool_call(func_name, arguments))
         except json.JSONDecodeError:
             pass
     return matches, start_pos
@@ -428,13 +390,7 @@ def _parse_glm_tool_calls(answer: str, tool_names: list[str]):
             arguments[k] = v
         if start_pos is None:
             start_pos = tc_match.start()
-        matches.append({
-            "type": "function",
-            "function": {
-                "name": func_name,
-                "arguments": arguments
-            }
-        })
+        matches.append(_make_tool_call(func_name, arguments))
     return matches, start_pos
 
 
@@ -486,13 +442,7 @@ def _parse_pythonic_tool_calls(answer: str, tool_names: list[str]):
 
         if start_pos is None:
             start_pos = bracket_match.start()
-        matches.append({
-            "type": "function",
-            "function": {
-                "name": func_name,
-                "arguments": arguments
-            }
-        })
+        matches.append(_make_tool_call(func_name, arguments))
 
     return matches, start_pos
 
diff --git a/modules/training.py b/modules/training.py
index a13a286459..145353c634 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -732,11 +732,13 @@ def collate_fn(batch):
     if lora_all_param > 0:
         print(f"Trainable params: {lora_trainable_param:,d} ({100 * lora_trainable_param / lora_all_param:.4f} %), All params: {lora_all_param:,d} (Model: {model_all_params:,d})")
 
-    train_log.update({"base_model_name": shared.model_name})
-    train_log.update({"base_model_class": shared.model.__class__.__name__})
-    train_log.update({"base_loaded_in_4bit": getattr(lora_model, "is_loaded_in_4bit", False)})
-    train_log.update({"base_loaded_in_8bit": getattr(lora_model, "is_loaded_in_8bit", False)})
-    train_log.update({"projections": projections_string})
+    train_log.update({
+        "base_model_name": shared.model_name,
+        "base_model_class": shared.model.__class__.__name__,
+        "base_loaded_in_4bit": getattr(lora_model, "is_loaded_in_4bit", False),
+        "base_loaded_in_8bit": getattr(lora_model, "is_loaded_in_8bit", False),
+        "projections": projections_string,
+    })
 
     if stop_at_loss > 0:
         print(f"Monitoring loss \033[1;31;1m(Auto-Stop at: {stop_at_loss})\033[0;37;0m")
diff --git a/modules/ui.py b/modules/ui.py
index bbb2226684..20bc837310 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -299,7 +299,7 @@ def apply_interface_values(state, use_persistent=False):
 
     elements = list_interface_input_elements()
 
-    if len(state) == 0:
+    if not state:
         return [gr.update() for k in elements]  # Dummy, do nothing
     else:
         return [state[k] if k in state else gr.update() for k in elements]
@@ -307,9 +307,8 @@ def apply_interface_values(state, use_persistent=False):
 
 def save_settings(state, preset, extensions_list, show_controls, theme_state, manual_save=False):
     output = copy.deepcopy(shared.settings)
-    exclude = []
     for k in state:
-        if k in shared.settings and k not in exclude:
+        if k in shared.settings:
             output[k] = state[k]
 
     if preset:
@@ -323,7 +322,7 @@ def save_settings(state, preset, extensions_list, show_controls, theme_state, ma
     output['custom_stopping_strings'] = output.get('custom_stopping_strings') or ''
     output['custom_token_bans'] = output.get('custom_token_bans') or ''
     output['show_controls'] = show_controls
-    output['dark_theme'] = True if theme_state == 'dark' else False
+    output['dark_theme'] = theme_state == 'dark'
     output.pop('instruction_template_str')
     output.pop('truncation_length')
 

From b3eb0e313d7f74e3f90c949d54f453d3f6846ae0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Mar 2026 11:53:12 -0700
Subject: [PATCH 1426/1701] Reduce the size of portable builds by using
 stripped Python

---
 .github/workflows/build-portable-release-cuda.yml   | 4 ++--
 .github/workflows/build-portable-release-rocm.yml   | 4 ++--
 .github/workflows/build-portable-release-vulkan.yml | 4 ++--
 .github/workflows/build-portable-release.yml        | 8 ++++----
 4 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index a575911297..5d66bd77fc 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -116,13 +116,13 @@ jobs:
             # 1. Set platform-specific variables
             if [[ "$RUNNER_OS" == "Windows" ]]; then
                 PLATFORM="windows"
-                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-pc-windows-msvc-install_only.tar.gz"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-pc-windows-msvc-install_only_stripped.tar.gz"
                 PIP_PATH="portable_env/python.exe -m pip"
                 PACKAGES_PATH="portable_env/Lib/site-packages"
                 rm start_linux.sh start_macos.sh
             else
                 PLATFORM="linux"
-                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only.tar.gz"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only_stripped.tar.gz"
                 PIP_PATH="portable_env/bin/python -m pip"
                 PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
                 rm start_macos.sh start_windows.bat
diff --git a/.github/workflows/build-portable-release-rocm.yml b/.github/workflows/build-portable-release-rocm.yml
index 1050fa7e78..b9a10bac82 100644
--- a/.github/workflows/build-portable-release-rocm.yml
+++ b/.github/workflows/build-portable-release-rocm.yml
@@ -114,13 +114,13 @@ jobs:
             # 1. Set platform-specific variables
             if [[ "$RUNNER_OS" == "Windows" ]]; then
                 PLATFORM="windows"
-                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-pc-windows-msvc-install_only.tar.gz"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-pc-windows-msvc-install_only_stripped.tar.gz"
                 PIP_PATH="portable_env/python.exe -m pip"
                 PACKAGES_PATH="portable_env/Lib/site-packages"
                 rm start_linux.sh start_macos.sh
             else
                 PLATFORM="linux"
-                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only.tar.gz"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only_stripped.tar.gz"
                 PIP_PATH="portable_env/bin/python -m pip"
                 PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
                 rm start_macos.sh start_windows.bat
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index b98b2e5e37..9748d5b8f2 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -114,13 +114,13 @@ jobs:
             # 1. Set platform-specific variables
             if [[ "$RUNNER_OS" == "Windows" ]]; then
                 PLATFORM="windows"
-                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-pc-windows-msvc-install_only.tar.gz"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-pc-windows-msvc-install_only_stripped.tar.gz"
                 PIP_PATH="portable_env/python.exe -m pip"
                 PACKAGES_PATH="portable_env/Lib/site-packages"
                 rm start_linux.sh start_macos.sh
             else
                 PLATFORM="linux"
-                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only.tar.gz"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only_stripped.tar.gz"
                 PIP_PATH="portable_env/bin/python -m pip"
                 PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
                 rm start_macos.sh start_windows.bat
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 1bd4e1638f..e03116f6e0 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -115,18 +115,18 @@ jobs:
             # 1. Set platform-specific variables
             if [[ "$RUNNER_OS" == "Windows" ]]; then
                 PLATFORM="windows-cpu"
-                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-pc-windows-msvc-install_only.tar.gz"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-pc-windows-msvc-install_only_stripped.tar.gz"
                 PIP_PATH="portable_env/python.exe -m pip"
                 PACKAGES_PATH="portable_env/Lib/site-packages"
                 rm start_linux.sh start_macos.sh
             elif [[ "$RUNNER_OS" == "macOS" ]]; then
                 if [[ "$OS_TYPE" == "macos-15-intel" ]]; then
                     PLATFORM="macos-x86_64"
-                    PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-apple-darwin-install_only.tar.gz"
+                    PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-apple-darwin-install_only_stripped.tar.gz"
                     REQ_TYPE="apple_intel"
                 else
                     PLATFORM="macos-arm64"
-                    PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-aarch64-apple-darwin-install_only.tar.gz"
+                    PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-aarch64-apple-darwin-install_only_stripped.tar.gz"
                     REQ_TYPE="apple_silicon"
                 fi
                 PIP_PATH="portable_env/bin/python -m pip"
@@ -135,7 +135,7 @@ jobs:
             else
                 # Linux case
                 PLATFORM="linux-cpu"
-                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only.tar.gz"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only_stripped.tar.gz"
                 PIP_PATH="portable_env/bin/python -m pip"
                 PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
                 rm start_macos.sh start_windows.bat

From 843de8b8a81edbd825cb03eb28af594fd3c7f3b1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 19 Mar 2026 18:49:36 -0700
Subject: [PATCH 1427/1701] Update exllamav3 to 0.0.26

---
 requirements/full/requirements.txt | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index c8479d0418..ad68ad59c2 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,7 +42,7 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.25/exllamav3-0.0.25+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.25/exllamav3-0.0.25+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.26/exllamav3-0.0.26+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.26/exllamav3-0.0.26+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"

From 2e4232e02bdf7640470ba1efdc5e72f1cd56b867 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 20 Mar 2026 07:20:26 -0700
Subject: [PATCH 1428/1701] Minor cleanup

---
 modules/callbacks.py | 2 +-
 modules/utils.py     | 8 --------
 2 files changed, 1 insertion(+), 9 deletions(-)

diff --git a/modules/callbacks.py b/modules/callbacks.py
index 89fb6c0889..6288de2952 100644
--- a/modules/callbacks.py
+++ b/modules/callbacks.py
@@ -34,7 +34,7 @@ def _callback(val):
 
         def gentask():
             try:
-                ret = self.mfunc(callback=_callback, *args, **self.kwargs)
+                ret = self.mfunc(callback=_callback, *self.args, **self.kwargs)
             except StopNowException:
                 pass
             except Exception:
diff --git a/modules/utils.py b/modules/utils.py
index ff32e974f0..b01953ee22 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -81,14 +81,6 @@ def atoi(text):
     return int(text) if text.isdigit() else text.lower()
 
 
-# Replace multiple string pairs in a string
-def replace_all(text, dic):
-    for i, j in dic.items():
-        text = text.replace(i, j)
-
-    return text
-
-
 def natural_keys(text):
     return [atoi(c) for c in re.split(r'(\d+)', text)]
 

From bf6fbc019dbd9470efdeafa033818efa178d7735 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 20 Mar 2026 14:46:00 -0300
Subject: [PATCH 1429/1701] API: Move OpenAI-compatible API from
 extensions/openai to modules/api

---
 .../workflows/build-portable-release-cuda.yml |  2 +-
 .../workflows/build-portable-release-rocm.yml |  2 +-
 .../build-portable-release-vulkan.yml         |  2 +-
 .github/workflows/build-portable-release.yml  |  2 +-
 docs/07 - Extensions.md                       |  1 -
 docs/12 - OpenAI API.md                       | 12 +------
 modules/api/__init__.py                       |  0
 .../api}/cache_embedding_model.py             |  0
 .../openai => modules/api}/completions.py     |  6 ++--
 .../openai => modules/api}/embeddings.py      | 10 +++---
 {extensions/openai => modules/api}/errors.py  |  0
 {extensions/openai => modules/api}/images.py  |  2 +-
 {extensions/openai => modules/api}/logits.py  |  2 +-
 {extensions/openai => modules/api}/models.py  |  0
 .../openai => modules/api}/moderations.py     |  2 +-
 {extensions/openai => modules/api}/script.py  | 34 ++++++++++---------
 {extensions/openai => modules/api}/tokens.py  |  0
 {extensions/openai => modules/api}/typing.py  |  0
 {extensions/openai => modules/api}/utils.py   |  3 +-
 modules/extensions.py                         |  3 +-
 modules/shared.py                             | 16 +--------
 modules/ui_session.py                         |  2 --
 server.py                                     | 15 ++++++++
 23 files changed, 51 insertions(+), 65 deletions(-)
 create mode 100644 modules/api/__init__.py
 rename {extensions/openai => modules/api}/cache_embedding_model.py (100%)
 rename {extensions/openai => modules/api}/completions.py (99%)
 rename {extensions/openai => modules/api}/embeddings.py (90%)
 rename {extensions/openai => modules/api}/errors.py (100%)
 rename {extensions/openai => modules/api}/images.py (96%)
 rename {extensions/openai => modules/api}/logits.py (84%)
 rename {extensions/openai => modules/api}/models.py (100%)
 rename {extensions/openai => modules/api}/moderations.py (97%)
 rename {extensions/openai => modules/api}/script.py (96%)
 rename {extensions/openai => modules/api}/tokens.py (100%)
 rename {extensions/openai => modules/api}/typing.py (100%)
 rename {extensions/openai => modules/api}/utils.py (93%)

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index 5d66bd77fc..f9eea58a37 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -106,7 +106,7 @@ jobs:
             cd "text-generation-webui-${VERSION_CLEAN}"
 
             # Remove extensions that need additional requirements
-            allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
+            allowed=("character_bias" "gallery" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
             # Define common variables
diff --git a/.github/workflows/build-portable-release-rocm.yml b/.github/workflows/build-portable-release-rocm.yml
index b9a10bac82..db42b7dc49 100644
--- a/.github/workflows/build-portable-release-rocm.yml
+++ b/.github/workflows/build-portable-release-rocm.yml
@@ -105,7 +105,7 @@ jobs:
             cd "text-generation-webui-${VERSION_CLEAN}"
 
             # Remove extensions that need additional requirements
-            allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
+            allowed=("character_bias" "gallery" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
             # Define common variables
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 9748d5b8f2..8f5aa7c883 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -105,7 +105,7 @@ jobs:
             cd "text-generation-webui-${VERSION_CLEAN}"
 
             # Remove extensions that need additional requirements
-            allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
+            allowed=("character_bias" "gallery" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
             # Define common variables
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index e03116f6e0..9ace90f6ad 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -105,7 +105,7 @@ jobs:
             cd "text-generation-webui-${VERSION_CLEAN}"
 
             # Remove extensions that need additional requirements
-            allowed=("character_bias" "gallery" "openai" "sd_api_pictures")
+            allowed=("character_bias" "gallery" "sd_api_pictures")
             find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
 
             # Define common variables
diff --git a/docs/07 - Extensions.md b/docs/07 - Extensions.md
index 48cd30cef5..779b2a3499 100644
--- a/docs/07 - Extensions.md	
+++ b/docs/07 - Extensions.md	
@@ -20,7 +20,6 @@ If you create an extension, you are welcome to host it in a GitHub repository an
 
 |Extension|Description|
 |---------|-----------|
-|[openai](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/openai)| Creates an API that mimics the OpenAI API and can be used as a drop-in replacement. |
 |[superboogav2](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/superboogav2)| Enhanced RAG extension with support for PDF, DOCX, and PPTX files. |
 |[send_pictures](https://github.com/oobabooga/text-generation-webui/blob/main/extensions/send_pictures/)| Creates an image upload field that can be used to send images to the bot in chat mode. Captions are automatically generated using BLIP. |
 |[coqui_tts](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/coqui_tts)| Text-to-speech extension using Coqui XTTS v2. |
diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index 637cccedd3..276a7e192a 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -19,7 +19,7 @@ Add `--api` to your command-line flags.
 
 ### Examples
 
-For the documentation with all the endpoints, parameters and their types, consult `http://127.0.0.1:5000/docs` or the [typing.py](https://github.com/oobabooga/text-generation-webui/blob/main/extensions/openai/typing.py) file.
+For the documentation with all the endpoints, parameters and their types, consult `http://127.0.0.1:5000/docs` or the [typing.py](https://github.com/oobabooga/text-generation-webui/blob/main/modules/api/typing.py) file.
 
 The official examples in the [OpenAI documentation](https://platform.openai.com/docs/api-reference) should also work, and the same parameters apply (although the API here has more optional parameters).
 
@@ -490,16 +490,6 @@ The following environment variables can be used (they take precedence over every
 | `OPENEDAI_EMBEDDING_MODEL` | Embedding model (if applicable) |          sentence-transformers/all-mpnet-base-v2                  |
 | `OPENEDAI_EMBEDDING_DEVICE` | Embedding device (if applicable) |           cuda                 |
 
-#### Persistent settings with `settings.yaml`
-
-You can also set the following variables in your `settings.yaml` file:
-
-```
-openai-embedding_device: cuda
-openai-embedding_model: "sentence-transformers/all-mpnet-base-v2"
-openai-debug: 1
-```
-
 ### Third-party application setup
 
 You can usually force an application that uses the OpenAI API to connect to the local API by using the following environment variables:
diff --git a/modules/api/__init__.py b/modules/api/__init__.py
new file mode 100644
index 0000000000..e69de29bb2
diff --git a/extensions/openai/cache_embedding_model.py b/modules/api/cache_embedding_model.py
similarity index 100%
rename from extensions/openai/cache_embedding_model.py
rename to modules/api/cache_embedding_model.py
diff --git a/extensions/openai/completions.py b/modules/api/completions.py
similarity index 99%
rename from extensions/openai/completions.py
rename to modules/api/completions.py
index d0cd980288..8948bb86bc 100644
--- a/extensions/openai/completions.py
+++ b/modules/api/completions.py
@@ -9,9 +9,9 @@
 import yaml
 from pydantic import ValidationError
 
-from extensions.openai.errors import InvalidRequestError
-from extensions.openai.typing import ToolDefinition
-from extensions.openai.utils import debug_msg
+from .errors import InvalidRequestError
+from .typing import ToolDefinition
+from .utils import debug_msg
 from modules.tool_parsing import get_tool_call_id, parse_tool_call, detect_tool_call_format
 from modules import shared
 from modules.reasoning import extract_reasoning
diff --git a/extensions/openai/embeddings.py b/modules/api/embeddings.py
similarity index 90%
rename from extensions/openai/embeddings.py
rename to modules/api/embeddings.py
index 1420879cc9..ad299c9d4c 100644
--- a/extensions/openai/embeddings.py
+++ b/modules/api/embeddings.py
@@ -3,8 +3,8 @@
 import numpy as np
 from transformers import AutoModel
 
-from extensions.openai.errors import ServiceUnavailableError
-from extensions.openai.utils import debug_msg, float_list_to_base64
+from .errors import ServiceUnavailableError
+from .utils import debug_msg, float_list_to_base64
 from modules.logging_colors import logger
 
 embeddings_params_initialized = False
@@ -17,14 +17,12 @@ def initialize_embedding_params():
     '''
     global embeddings_params_initialized
     if not embeddings_params_initialized:
-        from extensions.openai.script import params
-
         global st_model, embeddings_model, embeddings_device
 
-        st_model = os.environ.get("OPENEDAI_EMBEDDING_MODEL", params.get('embedding_model', 'all-mpnet-base-v2'))
+        st_model = os.environ.get("OPENEDAI_EMBEDDING_MODEL", 'sentence-transformers/all-mpnet-base-v2')
         embeddings_model = None
         # OPENEDAI_EMBEDDING_DEVICE: auto (best or cpu), cpu, cuda, ipu, xpu, mkldnn, opengl, opencl, ideep, hip, ve, fpga, ort, xla, lazy, vulkan, mps, meta, hpu, mtia, privateuseone
-        embeddings_device = os.environ.get("OPENEDAI_EMBEDDING_DEVICE", params.get('embedding_device', 'cpu'))
+        embeddings_device = os.environ.get("OPENEDAI_EMBEDDING_DEVICE", 'cpu')
         if embeddings_device.lower() == 'auto':
             embeddings_device = None
 
diff --git a/extensions/openai/errors.py b/modules/api/errors.py
similarity index 100%
rename from extensions/openai/errors.py
rename to modules/api/errors.py
diff --git a/extensions/openai/images.py b/modules/api/images.py
similarity index 96%
rename from extensions/openai/images.py
rename to modules/api/images.py
index f7be3d227b..9570453593 100644
--- a/extensions/openai/images.py
+++ b/modules/api/images.py
@@ -6,7 +6,7 @@
 import io
 import time
 
-from extensions.openai.errors import ServiceUnavailableError
+from .errors import ServiceUnavailableError
 from modules import shared
 
 
diff --git a/extensions/openai/logits.py b/modules/api/logits.py
similarity index 84%
rename from extensions/openai/logits.py
rename to modules/api/logits.py
index 280612db5f..e0c7ea0e6a 100644
--- a/extensions/openai/logits.py
+++ b/modules/api/logits.py
@@ -1,4 +1,4 @@
-from extensions.openai.completions import process_parameters
+from .completions import process_parameters
 from modules.logits import get_next_logits
 
 
diff --git a/extensions/openai/models.py b/modules/api/models.py
similarity index 100%
rename from extensions/openai/models.py
rename to modules/api/models.py
diff --git a/extensions/openai/moderations.py b/modules/api/moderations.py
similarity index 97%
rename from extensions/openai/moderations.py
rename to modules/api/moderations.py
index 1ca6b8abb2..ac0539d61d 100644
--- a/extensions/openai/moderations.py
+++ b/modules/api/moderations.py
@@ -3,7 +3,7 @@
 import numpy as np
 from numpy.linalg import norm
 
-from extensions.openai.embeddings import get_embeddings
+from .embeddings import get_embeddings
 
 moderations_disabled = False  # return 0/false
 category_embeddings = None
diff --git a/extensions/openai/script.py b/modules/api/script.py
similarity index 96%
rename from extensions/openai/script.py
rename to modules/api/script.py
index a0d5deb84e..356919e91d 100644
--- a/extensions/openai/script.py
+++ b/modules/api/script.py
@@ -13,16 +13,15 @@
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.requests import Request
 from fastapi.responses import JSONResponse
-from pydub import AudioSegment
 from sse_starlette import EventSourceResponse
 from starlette.concurrency import iterate_in_threadpool
 
-import extensions.openai.completions as OAIcompletions
-import extensions.openai.logits as OAIlogits
-import extensions.openai.models as OAImodels
-from extensions.openai.tokens import token_count, token_decode, token_encode
-from extensions.openai.errors import OpenAIError
-from extensions.openai.utils import _start_cloudflared
+import modules.api.completions as OAIcompletions
+import modules.api.logits as OAIlogits
+import modules.api.models as OAImodels
+from .tokens import token_count, token_decode, token_encode
+from .errors import OpenAIError
+from .utils import _start_cloudflared
 from modules import shared
 from modules.logging_colors import logger
 from modules.models import unload_model
@@ -53,12 +52,6 @@
     to_dict
 )
 
-params = {
-    'embedding_device': 'cpu',
-    'embedding_model': 'sentence-transformers/all-mpnet-base-v2',
-    'debug': 0
-}
-
 
 async def _wait_for_disconnect(request: Request, stop_event: threading.Event):
     """Block until the client disconnects, then signal the stop_event."""
@@ -244,6 +237,7 @@ def handle_billing_usage():
 @app.post('/v1/audio/transcriptions', dependencies=check_key)
 async def handle_audio_transcription(request: Request):
     import speech_recognition as sr
+    from pydub import AudioSegment
 
     r = sr.Recognizer()
 
@@ -275,7 +269,7 @@ async def handle_audio_transcription(request: Request):
 
 @app.post('/v1/images/generations', response_model=ImageGenerationResponse, dependencies=check_key)
 async def handle_image_generation(request_data: ImageGenerationRequest):
-    import extensions.openai.images as OAIimages
+    import modules.api.images as OAIimages
 
     response = await asyncio.to_thread(OAIimages.generations, request_data)
     return JSONResponse(response)
@@ -283,7 +277,7 @@ async def handle_image_generation(request_data: ImageGenerationRequest):
 
 @app.post("/v1/embeddings", response_model=EmbeddingsResponse, dependencies=check_key)
 async def handle_embeddings(request: Request, request_data: EmbeddingsRequest):
-    import extensions.openai.embeddings as OAIembeddings
+    import modules.api.embeddings as OAIembeddings
 
     input = request_data.input
     if not input:
@@ -298,7 +292,7 @@ async def handle_embeddings(request: Request, request_data: EmbeddingsRequest):
 
 @app.post("/v1/moderations", dependencies=check_key)
 async def handle_moderations(request: Request):
-    import extensions.openai.moderations as OAImoderations
+    import modules.api.moderations as OAImoderations
 
     body = await request.json()
     input = body["input"]
@@ -500,7 +494,15 @@ def run_server():
     uvicorn.run(app, host=server_addrs, port=port, ssl_certfile=ssl_certfile, ssl_keyfile=ssl_keyfile, access_log=False)
 
 
+_server_started = False
+
+
 def setup():
+    global _server_started
+    if _server_started:
+        return
+
+    _server_started = True
     if shared.args.nowebui:
         run_server()
     else:
diff --git a/extensions/openai/tokens.py b/modules/api/tokens.py
similarity index 100%
rename from extensions/openai/tokens.py
rename to modules/api/tokens.py
diff --git a/extensions/openai/typing.py b/modules/api/typing.py
similarity index 100%
rename from extensions/openai/typing.py
rename to modules/api/typing.py
diff --git a/extensions/openai/utils.py b/modules/api/utils.py
similarity index 93%
rename from extensions/openai/utils.py
rename to modules/api/utils.py
index 2b4147690e..fae181ffed 100644
--- a/extensions/openai/utils.py
+++ b/modules/api/utils.py
@@ -23,8 +23,7 @@ def float_list_to_base64(float_array: np.ndarray) -> str:
 
 
 def debug_msg(*args, **kwargs):
-    from extensions.openai.script import params
-    if os.environ.get("OPENEDAI_DEBUG", params.get('debug', 0)):
+    if os.environ.get("OPENEDAI_DEBUG", 0):
         print(*args, **kwargs)
 
 
diff --git a/modules/extensions.py b/modules/extensions.py
index 4bb7b6833d..09db9f4005 100644
--- a/modules/extensions.py
+++ b/modules/extensions.py
@@ -32,8 +32,7 @@ def load_extensions():
         if name not in available_extensions:
             continue
 
-        if name != 'api':
-            logger.info(f'Loading the extension "{name}"')
+        logger.info(f'Loading the extension "{name}"')
 
         try:
             # Prefer user extension, fall back to system extension
diff --git a/modules/shared.py b/modules/shared.py
index 37bc5876bb..69e169608b 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -156,7 +156,7 @@
 
 # API
 group = parser.add_argument_group('API')
-group.add_argument('--api', action='store_true', help='Enable the API extension.')
+group.add_argument('--api', action='store_true', help='Enable the API server.')
 group.add_argument('--public-api', action='store_true', help='Create a public URL for the API using Cloudflare.')
 group.add_argument('--public-api-id', type=str, help='Tunnel ID for named Cloudflare Tunnel. Use together with public-api option.', default=None)
 group.add_argument('--api-port', type=int, default=5000, help='The listening port for the API.')
@@ -435,16 +435,6 @@ def fix_loader_name(name):
         return 'TensorRT-LLM'
 
 
-def add_extension(name, last=False):
-    if args.extensions is None:
-        args.extensions = [name]
-    elif last:
-        args.extensions = [x for x in args.extensions if x != name]
-        args.extensions.append(name)
-    elif name not in args.extensions:
-        args.extensions.append(name)
-
-
 def is_chat():
     return True
 
@@ -464,10 +454,6 @@ def load_user_config():
 
 args.loader = fix_loader_name(args.loader)
 
-# Activate the API extension
-if args.api or args.public_api:
-    add_extension('openai', last=True)
-
 # Load model-specific settings
 p = Path(f'{args.model_dir}/config.yaml')
 if p.exists():
diff --git a/modules/ui_session.py b/modules/ui_session.py
index 19026fbb72..3f2c8a7bec 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -95,8 +95,6 @@ def set_interface_arguments(extensions, bool_active):
         setattr(shared.args, k, False)
     for k in bool_active:
         setattr(shared.args, k, True)
-        if k == 'api':
-            shared.add_extension('openai', last=True)
 
     shared.need_restart = True
 
diff --git a/server.py b/server.py
index 1aa9fc0482..cbdd2854d7 100644
--- a/server.py
+++ b/server.py
@@ -106,6 +106,11 @@ def create_interface():
     if shared.args.extensions is not None and len(shared.args.extensions) > 0:
         extensions_module.load_extensions()
 
+    # Start the API server if enabled
+    if shared.args.api or shared.args.public_api:
+        from modules.api.script import setup as api_setup
+        api_setup()
+
     # Force some events to be triggered on page load
     shared.persistent_interface_state.update({
         'mode': shared.settings['mode'],
@@ -273,6 +278,12 @@ def create_interface():
     # Activate the extensions listed on settings.yaml
     extensions_module.available_extensions = utils.get_available_extensions()
     for extension in shared.settings['default_extensions']:
+        # The openai extension was moved to modules/api and is now
+        # activated with --api. Treat it as an alias for backwards compat.
+        if extension == 'openai':
+            shared.args.api = True
+            continue
+
         shared.args.extensions = shared.args.extensions or []
         if extension not in shared.args.extensions:
             shared.args.extensions.append(extension)
@@ -337,6 +348,10 @@ def create_interface():
         shared.args.extensions = [x for x in (shared.args.extensions or []) if x != 'gallery']
         if shared.args.extensions:
             extensions_module.load_extensions()
+
+        if shared.args.api or shared.args.public_api:
+            from modules.api.script import setup as api_setup
+            api_setup()
     else:
         # Launch the web UI
         create_interface()

From 1a910574c36b6b1d93a3bf3303335201993f503a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 20 Mar 2026 14:57:01 -0300
Subject: [PATCH 1430/1701] API: Fix debug_msg truthy check for
 OPENEDAI_DEBUG=0

---
 modules/api/utils.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/api/utils.py b/modules/api/utils.py
index fae181ffed..f021c37882 100644
--- a/modules/api/utils.py
+++ b/modules/api/utils.py
@@ -23,7 +23,7 @@ def float_list_to_base64(float_array: np.ndarray) -> str:
 
 
 def debug_msg(*args, **kwargs):
-    if os.environ.get("OPENEDAI_DEBUG", 0):
+    if int(os.environ.get("OPENEDAI_DEBUG", 0)):
         print(*args, **kwargs)
 
 

From 855141967c4081f9f90a1b5b7fd091a14c543e8f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 20 Mar 2026 15:03:17 -0300
Subject: [PATCH 1431/1701] API: Handle --extensions openai as alias for --api

---
 server.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/server.py b/server.py
index cbdd2854d7..d224909ce1 100644
--- a/server.py
+++ b/server.py
@@ -288,6 +288,11 @@ def create_interface():
         if extension not in shared.args.extensions:
             shared.args.extensions.append(extension)
 
+    # Handle --extensions openai from the command line (moved to modules/api)
+    if shared.args.extensions and 'openai' in shared.args.extensions:
+        shared.args.extensions.remove('openai')
+        shared.args.api = True
+
     # Load image model if specified via CLI
     if shared.args.image_model:
         logger.info(f"Loading image model: {shared.args.image_model}")

From 7c79143a149d1618287ca0b526826ee04167f7d9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 20 Mar 2026 15:03:49 -0300
Subject: [PATCH 1432/1701] API: Fix _start_cloudflared raising after first
 attempt instead of exhausting retries

---
 modules/api/utils.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/api/utils.py b/modules/api/utils.py
index f021c37882..e8c505f6b2 100644
--- a/modules/api/utils.py
+++ b/modules/api/utils.py
@@ -50,4 +50,4 @@ def _start_cloudflared(port: int, tunnel_id: str, max_attempts: int = 3, on_star
             traceback.print_exc()
             time.sleep(3)
 
-        raise Exception('Could not start cloudflared.')
+    raise Exception('Could not start cloudflared.')

From f0e3997f375d61961c7032a09145f41c254d799f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 20 Mar 2026 16:04:57 -0300
Subject: [PATCH 1433/1701] Add missing __init__.py to modules/grammar

---
 modules/grammar/__init__.py | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 create mode 100644 modules/grammar/__init__.py

diff --git a/modules/grammar/__init__.py b/modules/grammar/__init__.py
new file mode 100644
index 0000000000..e69de29bb2

From 0216893475b415106ce631f62fc62bcd9d345f8a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 20 Mar 2026 19:05:36 -0300
Subject: [PATCH 1434/1701] API: Add Anthropic-compatible /v1/messages endpoint

---
 modules/api/anthropic.py | 468 +++++++++++++++++++++++++++++++++++++++
 modules/api/script.py    | 115 +++++++++-
 modules/api/typing.py    |  21 +-
 3 files changed, 600 insertions(+), 4 deletions(-)
 create mode 100644 modules/api/anthropic.py

diff --git a/modules/api/anthropic.py b/modules/api/anthropic.py
new file mode 100644
index 0000000000..5fbf5caf4a
--- /dev/null
+++ b/modules/api/anthropic.py
@@ -0,0 +1,468 @@
+import json
+import time
+
+from modules import shared
+
+
+def convert_request(body: dict) -> dict:
+    """Transform Anthropic Messages API body into the dict that chat_completions_common expects."""
+    messages = []
+
+    # System message
+    system = body.get('system')
+    if system:
+        if isinstance(system, list):
+            # List of content blocks like [{"type":"text","text":"..."}]
+            text_parts = [block.get('text', '') for block in system if isinstance(block, dict) and block.get('type') == 'text']
+            system_text = '\n'.join(text_parts)
+        else:
+            system_text = str(system)
+        if system_text:
+            messages.append({"role": "system", "content": system_text})
+
+    # Convert messages
+    for msg in body.get('messages', []):
+        role = msg.get('role')
+        content = msg.get('content')
+
+        if isinstance(content, str):
+            messages.append({"role": role, "content": content})
+            continue
+
+        if not isinstance(content, list):
+            messages.append({"role": role, "content": str(content) if content else ""})
+            continue
+
+        if role == 'assistant':
+            # Split into text content, tool_calls, and skip thinking blocks
+            text_parts = []
+            tool_calls = []
+            for block in content:
+                btype = block.get('type')
+                if btype == 'text':
+                    text_parts.append(block.get('text', ''))
+                elif btype == 'tool_use':
+                    tool_calls.append({
+                        "id": block.get('id', ''),
+                        "type": "function",
+                        "function": {
+                            "name": block.get('name', ''),
+                            "arguments": json.dumps(block.get('input', {}))
+                        }
+                    })
+                elif btype == 'thinking':
+                    pass  # Strip thinking blocks
+
+            assistant_msg = {"role": "assistant", "content": '\n'.join(text_parts) if text_parts else ""}
+            if tool_calls:
+                assistant_msg["tool_calls"] = tool_calls
+            messages.append(assistant_msg)
+
+        elif role == 'user':
+            # Handle tool_result blocks and regular content
+            regular_parts = []
+            for block in content:
+                btype = block.get('type')
+                if btype == 'tool_result':
+                    # Emit any accumulated regular content first
+                    if regular_parts:
+                        if len(regular_parts) == 1 and regular_parts[0].get('type') == 'text':
+                            messages.append({"role": "user", "content": regular_parts[0]['text']})
+                        else:
+                            messages.append({"role": "user", "content": regular_parts})
+                        regular_parts = []
+                    # Convert tool_result to OpenAI tool message
+                    tool_content = block.get('content', '')
+                    if isinstance(tool_content, list):
+                        tool_content = '\n'.join(
+                            b.get('text', '') for b in tool_content
+                            if isinstance(b, dict) and b.get('type') == 'text'
+                        )
+                    messages.append({
+                        "role": "tool",
+                        "tool_call_id": block.get('tool_use_id', ''),
+                        "content": str(tool_content)
+                    })
+                elif btype == 'text':
+                    regular_parts.append({"type": "text", "text": block.get('text', '')})
+                elif btype == 'image':
+                    source = block.get('source', {})
+                    if source.get('type') == 'base64':
+                        media_type = source.get('media_type', 'image/png')
+                        data = source.get('data', '')
+                        regular_parts.append({
+                            "type": "image_url",
+                            "image_url": {"url": f"data:{media_type};base64,{data}"}
+                        })
+                elif btype == 'thinking':
+                    pass  # Strip thinking blocks
+
+            if regular_parts:
+                if len(regular_parts) == 1 and regular_parts[0].get('type') == 'text':
+                    messages.append({"role": "user", "content": regular_parts[0]['text']})
+                else:
+                    messages.append({"role": "user", "content": regular_parts})
+        else:
+            messages.append({"role": role, "content": str(content)})
+
+    # Start with all fields from the original body (includes GenerationOptions defaults)
+    result = dict(body)
+
+    # Remove Anthropic-specific fields that don't map directly
+    for key in ('system', 'stop_sequences', 'tools', 'tool_choice', 'thinking', 'metadata'):
+        result.pop(key, None)
+
+    # Set converted fields
+    result['messages'] = messages
+    result['max_tokens'] = body.get('max_tokens', 4096)
+    result['stream'] = body.get('stream', False)
+    result['mode'] = 'instruct'
+
+    # Ensure ChatCompletionRequestParams defaults are present
+    result.setdefault('continue_', False)
+    result.setdefault('instruction_template', None)
+    result.setdefault('instruction_template_str', None)
+    result.setdefault('character', None)
+    result.setdefault('bot_name', None)
+    result.setdefault('context', None)
+    result.setdefault('greeting', None)
+    result.setdefault('user_name', None)
+    result.setdefault('user_bio', None)
+    result.setdefault('chat_template_str', None)
+    result.setdefault('chat_instruct_command', 'Continue the chat dialogue below. Write a single reply for the character "<|character|>".\n\n<|prompt|>')
+    result.setdefault('frequency_penalty', None)
+    result.setdefault('presence_penalty', None)
+    result.setdefault('logit_bias', None)
+    result.setdefault('logprobs', None)
+    result.setdefault('top_logprobs', None)
+    result.setdefault('n', 1)
+    result.setdefault('model', None)
+    result.setdefault('functions', None)
+    result.setdefault('function_call', None)
+    result.setdefault('stream_options', None)
+    result.setdefault('user', None)
+    result.setdefault('stop', None)
+    result.setdefault('tool_choice', None)
+
+    # Always request usage in streaming so the usage-only chunk triggers
+    # the deferred message_delta/message_stop with accurate output_tokens
+    if body.get('stream', False):
+        result['stream_options'] = {'include_usage': True}
+
+    # Map stop_sequences -> stop
+    if body.get('stop_sequences'):
+        result['stop'] = body['stop_sequences']
+
+    # Tools
+    if body.get('tools'):
+        result['tools'] = [
+            {
+                "type": "function",
+                "function": {
+                    "name": t.get('name', ''),
+                    "description": t.get('description', ''),
+                    "parameters": t.get('input_schema', {"type": "object", "properties": {}})
+                }
+            }
+            for t in body['tools']
+        ]
+
+    # Tool choice
+    tc = body.get('tool_choice')
+    if tc and isinstance(tc, dict):
+        tc_type = tc.get('type')
+        if tc_type == 'auto':
+            result['tool_choice'] = 'auto'
+        elif tc_type == 'any':
+            result['tool_choice'] = 'required'
+        elif tc_type == 'tool':
+            result['tool_choice'] = {"type": "function", "function": {"name": tc.get('name', '')}}
+        elif tc_type == 'none':
+            result['tool_choice'] = 'none'
+    else:
+        result.setdefault('tool_choice', None)
+
+    # Thinking
+    thinking = body.get('thinking')
+    if thinking and isinstance(thinking, dict) and thinking.get('type') in ('enabled', 'adaptive'):
+        result['enable_thinking'] = True
+
+    return result
+
+
+_FINISH_REASON_MAP = {
+    "stop": "end_turn",
+    "length": "max_tokens",
+    "tool_calls": "tool_use",
+}
+
+
+def build_response(openai_resp: dict, model: str) -> dict:
+    """Transform OpenAI chat completion response dict into Anthropic Messages format."""
+    resp_id = openai_resp.get('id', 'msg_unknown')
+    if resp_id.startswith('chatcmpl-'):
+        resp_id = 'msg_' + resp_id[9:]
+
+    choice = openai_resp.get('choices', [{}])[0]
+    message = choice.get('message', {})
+
+    content = []
+
+    # Reasoning/thinking content
+    reasoning = message.get('reasoning_content')
+    if reasoning:
+        content.append({"type": "thinking", "thinking": reasoning, "signature": ""})
+
+    # Text content
+    text = message.get('content')
+    if text:
+        content.append({"type": "text", "text": text})
+
+    # Tool calls
+    tool_calls = message.get('tool_calls')
+    if tool_calls:
+        for tc in tool_calls:
+            func = tc.get('function', {})
+            try:
+                input_data = json.loads(func.get('arguments', '{}'))
+            except (json.JSONDecodeError, TypeError):
+                input_data = {}
+            content.append({
+                "type": "tool_use",
+                "id": tc.get('id', ''),
+                "name": func.get('name', ''),
+                "input": input_data
+            })
+
+    finish_reason = choice.get('finish_reason', 'stop')
+    stop_reason = _FINISH_REASON_MAP.get(finish_reason, 'end_turn')
+
+    usage = openai_resp.get('usage', {})
+
+    return {
+        "id": resp_id,
+        "type": "message",
+        "role": "assistant",
+        "content": content,
+        "model": model,
+        "stop_reason": stop_reason,
+        "stop_sequence": None,
+        "usage": {
+            "input_tokens": usage.get('prompt_tokens', 0),
+            "output_tokens": usage.get('completion_tokens', 0),
+        }
+    }
+
+
+class StreamConverter:
+    """Stateful converter: processes one OpenAI chunk at a time, yields Anthropic SSE events.
+
+    When include_usage is enabled in the OpenAI request, the final chunk with
+    finish_reason has usage=None, followed by a separate usage-only chunk
+    (choices=[], usage={...}).  We defer emitting message_delta and message_stop
+    until we receive that usage chunk so output_tokens is accurate.
+    """
+
+    def __init__(self, model: str):
+        self.model = model
+        self.msg_id = "msg_%d" % int(time.time() * 1000000000)
+        self.block_index = 0
+        self.in_thinking = False
+        self.in_text = False
+        self.input_tokens = 0
+        self.output_tokens = 0
+        self.tool_calls_accum = {}
+        self.stop_reason = "end_turn"
+        self._pending_finish = False  # True after we've seen finish_reason
+
+    def process_chunk(self, chunk: dict) -> list[dict]:
+        """Process a single OpenAI streaming chunk; return list of Anthropic SSE event dicts."""
+        events = []
+        choices = chunk.get('choices', [])
+        usage = chunk.get('usage')
+
+        if usage:
+            self.input_tokens = usage.get('prompt_tokens', self.input_tokens)
+            self.output_tokens = usage.get('completion_tokens', self.output_tokens)
+
+        # Usage-only chunk (choices=[]) arrives after the finish chunk
+        if not choices:
+            if self._pending_finish:
+                events.extend(self.finish())
+            return events
+
+        choice = choices[0]
+        delta = choice.get('delta', {})
+        finish_reason = choice.get('finish_reason')
+
+        # First chunk with role
+        if 'role' in delta:
+            events.append({
+                "event": "message_start",
+                "data": json.dumps({
+                    "type": "message_start",
+                    "message": {
+                        "id": self.msg_id,
+                        "type": "message",
+                        "role": "assistant",
+                        "content": [],
+                        "model": self.model,
+                        "stop_reason": None,
+                        "stop_sequence": None,
+                        "usage": {"input_tokens": self.input_tokens, "output_tokens": 0}
+                    }
+                })
+            })
+            events.append({"event": "ping", "data": json.dumps({"type": "ping"})})
+            return events
+
+        # Reasoning content
+        reasoning_content = delta.get('reasoning_content')
+        if reasoning_content:
+            if not self.in_thinking:
+                self.in_thinking = True
+                events.append({
+                    "event": "content_block_start",
+                    "data": json.dumps({
+                        "type": "content_block_start",
+                        "index": self.block_index,
+                        "content_block": {"type": "thinking", "thinking": ""}
+                    })
+                })
+            events.append({
+                "event": "content_block_delta",
+                "data": json.dumps({
+                    "type": "content_block_delta",
+                    "index": self.block_index,
+                    "delta": {"type": "thinking_delta", "thinking": reasoning_content}
+                })
+            })
+            return events
+
+        # Text content
+        text_content = delta.get('content')
+        if text_content:
+            if self.in_thinking:
+                events.append({
+                    "event": "content_block_stop",
+                    "data": json.dumps({"type": "content_block_stop", "index": self.block_index})
+                })
+                self.in_thinking = False
+                self.block_index += 1
+
+            if not self.in_text:
+                self.in_text = True
+                events.append({
+                    "event": "content_block_start",
+                    "data": json.dumps({
+                        "type": "content_block_start",
+                        "index": self.block_index,
+                        "content_block": {"type": "text", "text": ""}
+                    })
+                })
+            events.append({
+                "event": "content_block_delta",
+                "data": json.dumps({
+                    "type": "content_block_delta",
+                    "index": self.block_index,
+                    "delta": {"type": "text_delta", "text": text_content}
+                })
+            })
+            return events
+
+        # Tool calls in delta
+        chunk_tool_calls = delta.get('tool_calls')
+        if chunk_tool_calls:
+            for tc in chunk_tool_calls:
+                tc_id = tc.get('id', '')
+                tc_idx = tc.get('index', 0)
+                func = tc.get('function', {})
+                if tc_id:
+                    self.tool_calls_accum[tc_idx] = {
+                        "id": tc_id,
+                        "name": func.get('name', ''),
+                        "arguments": func.get('arguments', '')
+                    }
+                elif tc_idx in self.tool_calls_accum:
+                    self.tool_calls_accum[tc_idx]["arguments"] += func.get('arguments', '')
+
+        # Final chunk — close open content blocks, defer message_delta/stop for usage
+        if finish_reason is not None:
+            self.stop_reason = _FINISH_REASON_MAP.get(finish_reason, 'end_turn')
+
+            if self.in_thinking:
+                events.append({
+                    "event": "content_block_stop",
+                    "data": json.dumps({"type": "content_block_stop", "index": self.block_index})
+                })
+                self.in_thinking = False
+                self.block_index += 1
+
+            if self.in_text:
+                events.append({
+                    "event": "content_block_stop",
+                    "data": json.dumps({"type": "content_block_stop", "index": self.block_index})
+                })
+                self.in_text = False
+                self.block_index += 1
+
+            for tc_idx in sorted(self.tool_calls_accum.keys()):
+                tc = self.tool_calls_accum[tc_idx]
+                arguments_str = tc["arguments"] or "{}"
+
+                events.append({
+                    "event": "content_block_start",
+                    "data": json.dumps({
+                        "type": "content_block_start",
+                        "index": self.block_index,
+                        "content_block": {
+                            "type": "tool_use",
+                            "id": tc["id"],
+                            "name": tc["name"],
+                            "input": {}
+                        }
+                    })
+                })
+                # Emit the full input as a single input_json_delta so SDK
+                # clients that reconstruct from deltas get the correct data
+                events.append({
+                    "event": "content_block_delta",
+                    "data": json.dumps({
+                        "type": "content_block_delta",
+                        "index": self.block_index,
+                        "delta": {
+                            "type": "input_json_delta",
+                            "partial_json": arguments_str
+                        }
+                    })
+                })
+                events.append({
+                    "event": "content_block_stop",
+                    "data": json.dumps({"type": "content_block_stop", "index": self.block_index})
+                })
+                self.block_index += 1
+
+            # Defer message_delta/stop — usage chunk may follow
+            self._pending_finish = True
+
+        return events
+
+    def finish(self) -> list[dict]:
+        """Emit deferred message_delta and message_stop. Safe to call multiple times."""
+        if not self._pending_finish:
+            return []
+        self._pending_finish = False
+        return [
+            {
+                "event": "message_delta",
+                "data": json.dumps({
+                    "type": "message_delta",
+                    "delta": {"stop_reason": self.stop_reason, "stop_sequence": None},
+                    "usage": {"input_tokens": self.input_tokens, "output_tokens": self.output_tokens}
+                })
+            },
+            {
+                "event": "message_stop",
+                "data": json.dumps({"type": "message_stop"})
+            }
+        ]
diff --git a/modules/api/script.py b/modules/api/script.py
index 356919e91d..a94247fa45 100644
--- a/modules/api/script.py
+++ b/modules/api/script.py
@@ -10,6 +10,7 @@
 
 import uvicorn
 from fastapi import Depends, FastAPI, Header, HTTPException
+from fastapi.exceptions import RequestValidationError
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.requests import Request
 from fastapi.responses import JSONResponse
@@ -19,6 +20,7 @@
 import modules.api.completions as OAIcompletions
 import modules.api.logits as OAIlogits
 import modules.api.models as OAImodels
+import modules.api.anthropic as Anthropic
 from .tokens import token_count, token_decode, token_encode
 from .errors import OpenAIError
 from .utils import _start_cloudflared
@@ -28,6 +30,7 @@
 from modules.text_generation import stop_everything_event  # used by /v1/internal/stop-generation
 
 from .typing import (
+    AnthropicRequest,
     ChatCompletionRequest,
     ChatCompletionResponse,
     ChatPromptResponse,
@@ -74,9 +77,23 @@ def verify_admin_key(authorization: str = Header(None)) -> None:
         raise HTTPException(status_code=401, detail="Unauthorized")
 
 
+def verify_anthropic_key(x_api_key: str = Header(None, alias="x-api-key")) -> None:
+    expected_api_key = shared.args.api_key
+    if expected_api_key and (x_api_key is None or x_api_key != expected_api_key):
+        raise HTTPException(status_code=401, detail="Unauthorized")
+
+
+class AnthropicError(Exception):
+    def __init__(self, message: str, error_type: str = "invalid_request_error", status_code: int = 400):
+        self.message = message
+        self.error_type = error_type
+        self.status_code = status_code
+
+
 app = FastAPI()
 check_key = [Depends(verify_api_key)]
 check_admin_key = [Depends(verify_admin_key)]
+check_anthropic_key = [Depends(verify_anthropic_key)]
 
 # Configure CORS settings to allow all origins, methods, and headers
 app.add_middleware(
@@ -102,6 +119,28 @@ async def openai_error_handler(request: Request, exc: OpenAIError):
     )
 
 
+@app.exception_handler(AnthropicError)
+async def anthropic_error_handler(request: Request, exc: AnthropicError):
+    return JSONResponse(
+        status_code=exc.status_code,
+        content={"type": "error", "error": {"type": exc.error_type, "message": exc.message}}
+    )
+
+
+@app.exception_handler(RequestValidationError)
+async def validation_error_handler(request: Request, exc: RequestValidationError):
+    if request.url.path.startswith("/v1/messages"):
+        messages = "; ".join(
+            f"{'.'.join(str(l) for l in e['loc'])}: {e['msg']}" for e in exc.errors()
+        )
+        return JSONResponse(
+            status_code=400,
+            content={"type": "error", "error": {"type": "invalid_request_error", "message": messages}}
+        )
+
+    return JSONResponse(status_code=422, content={"detail": exc.errors()})
+
+
 @app.middleware("http")
 async def validate_host_header(request: Request, call_next):
     # Be strict about only approving access to localhost by default
@@ -211,6 +250,76 @@ async def generator():
         return JSONResponse(response)
 
 
+@app.post('/v1/messages', dependencies=check_anthropic_key)
+async def anthropic_messages(request: Request, request_data: AnthropicRequest):
+    body = to_dict(request_data)
+    model = body.get('model') or shared.model_name or 'unknown'
+
+    try:
+        converted = Anthropic.convert_request(body)
+    except Exception as e:
+        raise AnthropicError(message=str(e))
+
+    try:
+        return await _anthropic_generate(request, request_data, converted, model)
+    except OpenAIError as e:
+        error_type = "invalid_request_error" if e.code < 500 else "api_error"
+        if e.code == 503:
+            error_type = "overloaded_error"
+        raise AnthropicError(message=e.message, error_type=error_type, status_code=e.code)
+    except Exception as e:
+        raise AnthropicError(message=str(e) or "Internal server error", error_type="api_error", status_code=500)
+
+
+async def _anthropic_generate(request, request_data, converted, model):
+    if request_data.stream:
+        stop_event = threading.Event()
+
+        async def generator():
+            converter = Anthropic.StreamConverter(model)
+            response = OAIcompletions.stream_chat_completions(converted, is_legacy=False, stop_event=stop_event)
+            try:
+                async for resp in iterate_in_threadpool(response):
+                    disconnected = await request.is_disconnected()
+                    if disconnected:
+                        break
+
+                    for event in converter.process_chunk(resp):
+                        yield event
+
+                for event in converter.finish():
+                    yield event
+            except OpenAIError as e:
+                error_type = "invalid_request_error" if e.code < 500 else "api_error"
+                if e.code == 503:
+                    error_type = "overloaded_error"
+                yield {
+                    "event": "error",
+                    "data": json.dumps({"type": "error", "error": {"type": error_type, "message": e.message}})
+                }
+            finally:
+                stop_event.set()
+                response.close()
+
+        return EventSourceResponse(generator(), sep="\n")
+
+    else:
+        stop_event = threading.Event()
+        monitor = asyncio.create_task(_wait_for_disconnect(request, stop_event))
+        try:
+            openai_resp = await asyncio.to_thread(
+                OAIcompletions.chat_completions,
+                converted,
+                is_legacy=False,
+                stop_event=stop_event
+            )
+        finally:
+            stop_event.set()
+            monitor.cancel()
+
+        return JSONResponse(Anthropic.build_response(openai_resp, model))
+
+
 @app.get("/v1/models", dependencies=check_key)
 @app.get("/v1/models/{model}", dependencies=check_key)
 async def handle_models(request: Request):
@@ -469,15 +578,15 @@ def run_server():
             port,
             shared.args.public_api_id,
             max_attempts=3,
-            on_start=lambda url: logger.info(f'OpenAI-compatible API URL:\n\n{url}/v1\n')
+            on_start=lambda url: logger.info(f'API URL (OpenAI + Anthropic compatible):\n\n{url}/v1\n')
         )
     else:
         url_proto = 'https://' if (ssl_certfile and ssl_keyfile) else 'http://'
         urls = [f'{url_proto}{addr}:{port}/v1' for addr in server_addrs]
         if len(urls) > 1:
-            logger.info('OpenAI-compatible API URLs:\n\n' + '\n'.join(urls) + '\n')
+            logger.info('API URLs (OpenAI + Anthropic compatible):\n\n' + '\n'.join(urls) + '\n')
         else:
-            logger.info('OpenAI-compatible API URL:\n\n' + '\n'.join(urls) + '\n')
+            logger.info('API URL (OpenAI + Anthropic compatible):\n\n' + '\n'.join(urls) + '\n')
 
     # Log API keys
     if shared.args.api_key:
diff --git a/modules/api/typing.py b/modules/api/typing.py
index 80831c446a..1d486e8fb7 100644
--- a/modules/api/typing.py
+++ b/modules/api/typing.py
@@ -144,7 +144,7 @@ class CompletionResponse(BaseModel):
 
 
 class ChatCompletionRequestParams(BaseModel):
-    messages: List[dict]
+    messages: List[dict] = Field(..., min_length=1)
     model: str | None = Field(default=None, description="Unused parameter. To change the model, use the /v1/internal/model/load endpoint.")
     frequency_penalty: float | None = shared.args.frequency_penalty
     function_call: str | dict | None = Field(default=None, description="Unused parameter.")
@@ -282,6 +282,25 @@ class LoadLorasRequest(BaseModel):
     lora_names: List[str]
 
 
+class AnthropicRequestParams(BaseModel):
+    model: str | None = None
+    messages: List[dict] = Field(..., min_length=1)
+    max_tokens: int
+    system: str | list | None = None
+    temperature: float | None = shared.args.temperature
+    top_p: float | None = shared.args.top_p
+    stop_sequences: list[str] | None = None
+    stream: bool = False
+    tools: list[dict] | None = None
+    tool_choice: dict | None = None
+    thinking: dict | None = None
+    metadata: dict | None = None
+
+
+class AnthropicRequest(GenerationOptions, AnthropicRequestParams):
+    pass
+
+
 class ImageGenerationRequest(BaseModel):
     """Image-specific parameters for generation."""
     prompt: str

From f2c909725ef667821a0e2ef5d68f4a2b86f0fd49 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 21 Mar 2026 11:09:06 -0700
Subject: [PATCH 1435/1701] API: Use top_p=0.95 by default

---
 modules/shared.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/shared.py b/modules/shared.py
index 69e169608b..16ccbe77a4 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -175,7 +175,7 @@
 group.add_argument('--dynatemp-exponent', type=float, default=_d['dynatemp_exponent'], metavar='N', help='Dynamic temperature exponent')
 group.add_argument('--smoothing-factor', type=float, default=_d['smoothing_factor'], metavar='N', help='Smoothing factor')
 group.add_argument('--smoothing-curve', type=float, default=_d['smoothing_curve'], metavar='N', help='Smoothing curve')
-group.add_argument('--top-p', type=float, default=_d['top_p'], metavar='N', help='Top P')
+group.add_argument('--top-p', type=float, default=0.95, metavar='N', help='Top P')
 group.add_argument('--top-k', type=int, default=_d['top_k'], metavar='N', help='Top K')
 group.add_argument('--min-p', type=float, default=_d['min_p'], metavar='N', help='Min P')
 group.add_argument('--top-n-sigma', type=float, default=_d['top_n_sigma'], metavar='N', help='Top N Sigma')

From 2c4f36433986001f80fdbb0f9095aa68f43274d2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 21 Mar 2026 18:38:11 -0700
Subject: [PATCH 1436/1701] Update API docs to mention Anthropic support

---
 README.md               | 2 +-
 docs/12 - OpenAI API.md | 4 ++--
 modules/api/script.py   | 6 +++---
 3 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/README.md b/README.md
index cabb81fc74..7e5566ece5 100644
--- a/README.md
+++ b/README.md
@@ -27,7 +27,7 @@ A Gradio web UI for running Large Language Models locally. 100% private and offl
 - **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
 - **Vision (multimodal)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
 - **Tool-calling**: Models can call custom functions during chat — web search, page fetching, math, and more. Each tool is a single `.py` file, easy to create and extend ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Tool-Calling-Tutorial)).
-- **OpenAI-compatible API**: Chat and Completions endpoints with tool-calling support. Use as a local drop-in replacement for the OpenAI API ([examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples)).
+- **OpenAI/Anthropic-compatible API**: Chat, Completions, and Messages endpoints with tool-calling support. Use as a local drop-in replacement for the OpenAI/Anthropic APIs ([examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples)).
 - **Training**: Fine-tune LoRAs on multi-turn chat or raw text datasets. Supports resuming interrupted runs ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/05-%E2%80%90-Training-Tab)).
 - **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
 - **Easy setup**: [Portable builds](https://github.com/oobabooga/text-generation-webui/releases) (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or a one-click installer for the full feature set.
diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index 276a7e192a..2a7a7f6915 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -1,6 +1,6 @@
-## OpenAI compatible API
+## OpenAI/Anthropic-compatible API
 
-The main API for this project is meant to be a drop-in replacement to the OpenAI API, including Chat and Completions endpoints.
+The main API for this project is meant to be a drop-in replacement for the OpenAI and Anthropic APIs, including Chat, Completions, and Messages endpoints.
 
 * It is 100% offline and private.
 * It doesn't create any logs.
diff --git a/modules/api/script.py b/modules/api/script.py
index a94247fa45..5913c2c590 100644
--- a/modules/api/script.py
+++ b/modules/api/script.py
@@ -578,15 +578,15 @@ def run_server():
             port,
             shared.args.public_api_id,
             max_attempts=3,
-            on_start=lambda url: logger.info(f'API URL (OpenAI + Anthropic compatible):\n\n{url}/v1\n')
+            on_start=lambda url: logger.info(f'OpenAI/Anthropic-compatible API URL:\n\n{url}/v1\n')
         )
     else:
         url_proto = 'https://' if (ssl_certfile and ssl_keyfile) else 'http://'
         urls = [f'{url_proto}{addr}:{port}/v1' for addr in server_addrs]
         if len(urls) > 1:
-            logger.info('API URLs (OpenAI + Anthropic compatible):\n\n' + '\n'.join(urls) + '\n')
+            logger.info('OpenAI/Anthropic-compatible API URLs:\n\n' + '\n'.join(urls) + '\n')
         else:
-            logger.info('API URL (OpenAI + Anthropic compatible):\n\n' + '\n'.join(urls) + '\n')
+            logger.info('OpenAI/Anthropic-compatible API URL:\n\n' + '\n'.join(urls) + '\n')
 
     # Log API keys
     if shared.args.api_key:

From 9488df3e489c97cc26018d9ae1dc6a4bc0384f1b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 21 Mar 2026 20:47:26 -0700
Subject: [PATCH 1437/1701] llama.cpp: Don't suppress llama-server logs

---
 modules/llama_cpp_server.py | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 2ae01ddc2c..b77a86050e 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -588,8 +588,11 @@ def filter_stderr_with_progress(process_stderr):
                             print(display_line, end=end_char, file=sys.stderr, flush=True)
                             last_was_progress = (progress < 1.0)
 
-                        # skip noise lines
-                        elif not (line.startswith(('srv ', 'slot ')) or 'log_server_r: request: GET /health' in line or 'No parser definition detected' in line):
+                        # skip health check polling and parser warnings
+                        elif 'log_server_r: request: GET /health' in line or 'No parser definition detected' in line:
+                            continue
+
+                        else:
                             # if we were in progress, finish that line first
                             if last_was_progress:
                                 print(file=sys.stderr)

From 1dda5e47111eaf8cb90f25ffb94e47296def5c8f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 21 Mar 2026 20:58:45 -0700
Subject: [PATCH 1438/1701] Follow-up to previous commit

---
 modules/llama_cpp_server.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index b77a86050e..5cbf212236 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -588,8 +588,8 @@ def filter_stderr_with_progress(process_stderr):
                             print(display_line, end=end_char, file=sys.stderr, flush=True)
                             last_was_progress = (progress < 1.0)
 
-                        # skip health check polling and parser warnings
-                        elif 'log_server_r: request: GET /health' in line or 'No parser definition detected' in line:
+                        # skip noise lines
+                        elif 'log_server_r: request: GET /health' in line or 'No parser definition detected' in line or (last_was_progress and ('memory_seq_rm' in line or 'context checkpoint' in line)):
                             continue
 
                         else:

From bde496ea5daf9f7fa9a0ac90f8f8f25166738112 Mon Sep 17 00:00:00 2001
From: Phrosty1 
Date: Sun, 22 Mar 2026 20:48:56 -0400
Subject: [PATCH 1439/1701] Fix prompt corruption when continuing with context
 truncation (#7439)

---
 modules/chat.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index 148d559a83..f8088e0fab 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -434,6 +434,8 @@ def generate_chat_prompt(user_input, state, **kwargs):
         messages.append({"role": "user", "content": "fake user message replace me"})
 
     def make_prompt(messages):
+        if _continue:
+            messages = copy.deepcopy(messages)
         last_message = messages[-1].copy()
         if _continue:
             if state['mode'] == 'chat-instruct':

From 9ec20d9730db3f41270da12f51f7ce138fb8705c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 22 Mar 2026 19:16:24 -0700
Subject: [PATCH 1440/1701] Strip thinking blocks before tool-call parsing

---
 modules/tool_parsing.py | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)

diff --git a/modules/tool_parsing.py b/modules/tool_parsing.py
index 7a7ed5d8b7..ec49f77f76 100644
--- a/modules/tool_parsing.py
+++ b/modules/tool_parsing.py
@@ -2,6 +2,8 @@
 import random
 import re
 
+from modules.reasoning import extract_reasoning
+
 
 def _make_tool_call(name, arguments):
     return {"type": "function", "function": {"name": name, "arguments": arguments}}
@@ -41,6 +43,10 @@ def streaming_tool_buffer_check(text, markers=None, tool_names=None, check_bare_
         check_bare_names: Whether to do partial-prefix matching on tool
                           names (for models with unknown template format).
     '''
+    # Strip thinking blocks so tool-call syntax inside  doesn't
+    # trigger false positives.
+    _, text = extract_reasoning(text)
+
     # Full marker found in text → buffer permanently.
     # Always checks ALL known markers regardless of template (cheap safety net).
     for marker in TOOL_CALL_OPENING_MARKERS:
@@ -543,12 +549,19 @@ def detect_tool_call_format(template_str):
 
 
 def parse_tool_call(answer: str, tool_names: list[str], return_prefix: bool = False, parsers: list = None):
+    # Strip thinking blocks so tool-call syntax inside  is ignored.
+    original_answer = answer
+    _, answer = extract_reasoning(answer)
+    # Offset between original and stripped text, used to map start_pos
+    # back to the original string when returning a prefix.
+    reasoning_offset = len(original_answer) - len(answer)
+
     matches = []
     start_pos = None
 
     def _return(matches, start_pos):
         if return_prefix:
-            prefix = answer[:start_pos] if matches and start_pos is not None else ''
+            prefix = original_answer[:start_pos + reasoning_offset] if matches and start_pos is not None else ''
             return matches, prefix
         return matches
 

From 307d0c92be2a4f8ac97f2be6c2cc3af1b9c8ad6f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 23 Mar 2026 06:35:14 -0700
Subject: [PATCH 1441/1701] UI polish

---
 css/main.css  | 8 ++++----
 modules/ui.py | 1 +
 2 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/css/main.css b/css/main.css
index 22fac5c5da..a8c30a3f36 100644
--- a/css/main.css
+++ b/css/main.css
@@ -54,7 +54,7 @@ div.svelte-iyf88w {
     height: 39.594px;
     align-self: end;
     line-height: 1em;
-    border-radius: 0.375rem;
+    border-radius: 0.75rem;
     flex: none;
 }
 
@@ -1420,7 +1420,7 @@ audio {
 }
 
 .dark .thinking-block {
-    background-color: var(--darker-gray);
+    background-color: var(--selected-item-color-dark);
     border: 1px solid var(--input-border-color);
 }
 
@@ -1558,7 +1558,7 @@ strong {
     min-height: 200px;
     max-height: 65vh;
     padding: 10px;
-    border-radius: 5px;
+    border-radius: 0.5rem;
     border: 1px solid #ccc;
     background-color: var(--light-theme-gray);
     font-family: inherit;
@@ -1586,7 +1586,7 @@ strong {
 .edit-control-button {
     padding: 6px 12px;
     border: 1px solid #ccc;
-    border-radius: 4px;
+    border-radius: 0.75rem;
     cursor: pointer;
     background-color: #f8f9fa;
     color: #212529;
diff --git a/modules/ui.py b/modules/ui.py
index 20bc837310..02b5a9fbfe 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -115,6 +115,7 @@
         input_shadow_focus='none',
         input_shadow_focus_dark='none',
         button_large_radius='0.75rem',
+        button_small_radius='0.75rem',
         button_large_padding='6px 12px',
         input_radius='0.5rem',
         block_radius='0.375rem',

From 02f18a1d65881cb3ed291050a191d8cf712b7115 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 23 Mar 2026 07:06:38 -0700
Subject: [PATCH 1442/1701] API: Add thinking block signature field, fix error
 codes, clean up logging

---
 modules/api/anthropic.py   | 2 +-
 modules/api/embeddings.py  | 4 ++--
 modules/api/moderations.py | 2 --
 modules/api/script.py      | 9 +++++++--
 4 files changed, 10 insertions(+), 7 deletions(-)

diff --git a/modules/api/anthropic.py b/modules/api/anthropic.py
index 5fbf5caf4a..3fab09a640 100644
--- a/modules/api/anthropic.py
+++ b/modules/api/anthropic.py
@@ -326,7 +326,7 @@ def process_chunk(self, chunk: dict) -> list[dict]:
                     "data": json.dumps({
                         "type": "content_block_start",
                         "index": self.block_index,
-                        "content_block": {"type": "thinking", "thinking": ""}
+                        "content_block": {"type": "thinking", "thinking": "", "signature": ""}
                     })
                 })
             events.append({
diff --git a/modules/api/embeddings.py b/modules/api/embeddings.py
index ad299c9d4c..16cf0482de 100644
--- a/modules/api/embeddings.py
+++ b/modules/api/embeddings.py
@@ -39,14 +39,14 @@ def load_embedding_model(model: str):
     initialize_embedding_params()
     global embeddings_device, embeddings_model
     try:
-        print(f"Try embedding model: {model} on {embeddings_device}")
+        logger.info(f"Try embedding model: {model} on {embeddings_device}")
         if 'jina-embeddings' in model:
             embeddings_model = AutoModel.from_pretrained(model, trust_remote_code=True)  # trust_remote_code is needed to use the encode method
             embeddings_model = embeddings_model.to(embeddings_device)
         else:
             embeddings_model = SentenceTransformer(model, device=embeddings_device)
 
-        print(f"Loaded embedding model: {model}")
+        logger.info(f"Loaded embedding model: {model}")
     except Exception as e:
         embeddings_model = None
         raise ServiceUnavailableError(f"Error: Failed to load embedding model: {model}", internal_message=repr(e))
diff --git a/modules/api/moderations.py b/modules/api/moderations.py
index ac0539d61d..a41763cfeb 100644
--- a/modules/api/moderations.py
+++ b/modules/api/moderations.py
@@ -64,6 +64,4 @@ def moderations(input):
                 'category_scores': category_scores,
             }])
 
-    print(results)
-
     return results
diff --git a/modules/api/script.py b/modules/api/script.py
index 5913c2c590..85f4974f29 100644
--- a/modules/api/script.py
+++ b/modules/api/script.py
@@ -506,12 +506,17 @@ async def handle_load_model(request_data: LoadModelRequest):
         return JSONResponse(content="OK")
     except Exception:
         traceback.print_exc()
-        raise HTTPException(status_code=400, detail="Failed to load the model.")
+        raise HTTPException(status_code=500, detail="Failed to load the model.")
 
 
 @app.post("/v1/internal/model/unload", dependencies=check_admin_key)
 async def handle_unload_model():
-    unload_model()
+    try:
+        unload_model()
+        return JSONResponse(content="OK")
+    except Exception:
+        traceback.print_exc()
+        raise HTTPException(status_code=500, detail="Failed to unload the model.")
 
 
 @app.get("/v1/internal/lora/list", response_model=LoraListResponse, dependencies=check_admin_key)

From 286bbb685d7bc585b8d82fd0e8d23515aeff9cb0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 23 Mar 2026 20:22:46 -0700
Subject: [PATCH 1443/1701] Revert "Follow-up to previous commit"

This reverts commit 1dda5e47111eaf8cb90f25ffb94e47296def5c8f.
---
 modules/llama_cpp_server.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 5cbf212236..b77a86050e 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -588,8 +588,8 @@ def filter_stderr_with_progress(process_stderr):
                             print(display_line, end=end_char, file=sys.stderr, flush=True)
                             last_was_progress = (progress < 1.0)
 
-                        # skip noise lines
-                        elif 'log_server_r: request: GET /health' in line or 'No parser definition detected' in line or (last_was_progress and ('memory_seq_rm' in line or 'context checkpoint' in line)):
+                        # skip health check polling and parser warnings
+                        elif 'log_server_r: request: GET /health' in line or 'No parser definition detected' in line:
                             continue
 
                         else:

From a7ef430b38c2f6e7c9a043b2f94ec6c2108d1480 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 23 Mar 2026 20:22:51 -0700
Subject: [PATCH 1444/1701] Revert "llama.cpp: Don't suppress llama-server
 logs"

This reverts commit 9488df3e489c97cc26018d9ae1dc6a4bc0384f1b.
---
 modules/llama_cpp_server.py | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index b77a86050e..2ae01ddc2c 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -588,11 +588,8 @@ def filter_stderr_with_progress(process_stderr):
                             print(display_line, end=end_char, file=sys.stderr, flush=True)
                             last_was_progress = (progress < 1.0)
 
-                        # skip health check polling and parser warnings
-                        elif 'log_server_r: request: GET /health' in line or 'No parser definition detected' in line:
-                            continue
-
-                        else:
+                        # skip noise lines
+                        elif not (line.startswith(('srv ', 'slot ')) or 'log_server_r: request: GET /health' in line or 'No parser definition detected' in line):
                             # if we were in progress, finish that line first
                             if last_was_progress:
                                 print(file=sys.stderr)

From c9d2240f5045baed0f234f3937614bdbe63af340 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 24 Mar 2026 06:45:39 -0700
Subject: [PATCH 1445/1701] Update README

---
 README.md | 13 ++++++-------
 1 file changed, 6 insertions(+), 7 deletions(-)

diff --git a/README.md b/README.md
index 7e5566ece5..ab6cc2e544 100644
--- a/README.md
+++ b/README.md
@@ -23,21 +23,20 @@ A Gradio web UI for running Large Language Models locally. 100% private and offl
 
 ## Features
 
+- **Easy setup**: [Portable builds](https://github.com/oobabooga/text-generation-webui/releases) (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or a one-click installer for the full feature set.
 - **Multiple backends**: [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). Switch between backends and models without restarting.
-- **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
-- **Vision (multimodal)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
-- **Tool-calling**: Models can call custom functions during chat — web search, page fetching, math, and more. Each tool is a single `.py` file, easy to create and extend ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Tool-Calling-Tutorial)).
 - **OpenAI/Anthropic-compatible API**: Chat, Completions, and Messages endpoints with tool-calling support. Use as a local drop-in replacement for the OpenAI/Anthropic APIs ([examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples)).
+- **Tool-calling**: Models can call custom functions during chat — web search, page fetching, math, and more. Each tool is a single `.py` file, easy to create and extend ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Tool-Calling-Tutorial)).
+- **Vision (multimodal)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
+- **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
 - **Training**: Fine-tune LoRAs on multi-turn chat or raw text datasets. Supports resuming interrupted runs ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/05-%E2%80%90-Training-Tab)).
 - **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
-- **Easy setup**: [Portable builds](https://github.com/oobabooga/text-generation-webui/releases) (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or a one-click installer for the full feature set.
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - `instruct` mode for instruction-following (like ChatGPT), and `chat-instruct`/`chat` modes for talking to custom characters. Prompts are automatically formatted with Jinja2 templates.
 - Edit messages, navigate between message versions, and branch conversations at any point.
 - Free-form text generation in the Notebook tab without being limited to chat turns.
 - Multiple sampling parameters and generation options for sophisticated text generation control.
-- Aesthetic UI with dark and light themes.
-- Syntax highlighting for code blocks and LaTeX rendering for mathematical expressions.
+- Dark/light themes, syntax highlighting for code blocks, and LaTeX rendering for mathematical expressions.
 - Extension support, with numerous built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details.
 
 ## How to install
@@ -429,7 +428,7 @@ API generation defaults:
 
 That's it. The UI will detect it automatically.
 
-To check what will fit your GPU, you can use the [VRAM Calculator](https://huggingface.co/spaces/oobabooga/accurate-gguf-vram-calculator).
+To estimate how much memory a model will use, you can use the [GGUF Memory Calculator](https://huggingface.co/spaces/oobabooga/accurate-gguf-vram-calculator).
 
 
            
 Other model types (Transformers, EXL3)

From 5b8da154b7aa4475718b819abba8acc1354e34eb Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 24 Mar 2026 09:34:59 -0700
Subject: [PATCH 1446/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 12 files changed, 20 insertions(+), 20 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index ad68ad59c2..5661962700 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,8 +40,8 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.26/exllamav3-0.0.26+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.26/exllamav3-0.0.26+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index b11e50b716..620683cce2 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index d147af3f28..b1f109b2e0 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index d284c5d5fd..a54476a95d 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 3952054ee1..be82c904f7 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index abf7690c10..188da3809a 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 0d66c16c4b..4562b6d01b 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 0658239a11..04dcf25e12 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index b66e2b38db..4b8af78a73 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index bb815bb26e..5b0eaf892e 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index d57ba40b9f..90b3234f37 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 6abd892018..ea72b4ec15 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.94.0/llama_cpp_binaries-0.94.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 5814e745be03d1f6f4cc6614e7a10d45282024b8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 24 Mar 2026 11:14:22 -0700
Subject: [PATCH 1447/1701] UI: Minor polish

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index a8c30a3f36..d42bd6aeb7 100644
--- a/css/main.css
+++ b/css/main.css
@@ -582,7 +582,7 @@ audio {
 
 #chat-input textarea {
     background: #f3f4f6;
-    padding: 0.65rem 2.5rem 0.6rem;
+    padding: 0.675rem 2.5rem 0.6rem;
     margin-top: 0.15rem;
     border: 1px solid #d2d2d8;
     border-radius: 1.5rem;

From 750502695c4339dc525d50cf428960d7ffbeeb05 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 24 Mar 2026 11:39:24 -0700
Subject: [PATCH 1448/1701] Fix GPT-OSS tool-calling after 9ec20d97

---
 modules/reasoning.py | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/modules/reasoning.py b/modules/reasoning.py
index 9c92719b3d..aa1939b8f3 100644
--- a/modules/reasoning.py
+++ b/modules/reasoning.py
@@ -72,10 +72,9 @@ def extract_reasoning(text, html_escaped=False):
                 if content_pos != -1:
                     content_start = content_pos + len(content_esc)
                 else:
-                    # Content tag expected but not yet present (e.g. partial
-                    # streaming) — suppress intermediate tags between end_tag
-                    # and content_tag so they don't leak as content.
-                    content_start = len(text)
+                    # Content tag not present — fall back to content after
+                    # end_tag (e.g. GPT-OSS tool calls skip the final channel).
+                    content_start = end_pos + len(end_esc)
             else:
                 content_start = end_pos + len(end_esc)
 

From f48a2b79d022a3f503085d6daeb3706b3b6dc2e0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 24 Mar 2026 11:45:33 -0700
Subject: [PATCH 1449/1701] UI: Minor polish

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index d42bd6aeb7..009b7c0a43 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1420,7 +1420,7 @@ audio {
 }
 
 .dark .thinking-block {
-    background-color: var(--selected-item-color-dark);
+    background-color: transparent;
     border: 1px solid var(--input-border-color);
 }
 

From 807be1183272fac409ce8f08609dbdd0d9f63362 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 24 Mar 2026 18:48:50 -0700
Subject: [PATCH 1450/1701] Remove obsolete models/config.yaml and related code

---
 docs/01 - Chat Tab.md        |   2 +-
 docs/12 - OpenAI API.md      |   2 +-
 modules/models.py            |   1 -
 modules/models_settings.py   |   9 +-
 modules/shared.py            |  10 --
 server.py                    |   5 -
 user_data/models/config.yaml | 203 -----------------------------------
 7 files changed, 4 insertions(+), 228 deletions(-)
 delete mode 100644 user_data/models/config.yaml

diff --git a/docs/01 - Chat Tab.md b/docs/01 - Chat Tab.md
index 5104895f28..96b232fa31 100644
--- a/docs/01 - Chat Tab.md	
+++ b/docs/01 - Chat Tab.md	
@@ -112,7 +112,7 @@ Used for talking to an instruction-following model using the prompt format defin
 
 The prompt format is defined by the **Instruction template** parameter in "Parameters" > "Instruction template", which represents a Jinja2 template.
 
-Note that when you load a model in the "Model" tab, the web UI will try to automatically detect its instruction template (if any), and will update the values under "Parameters" > "Instruction template" accordingly. This is done using a set of regular expressions defined in `user_data/models/config.yaml`. This detection is not guaranteed to be accurate. You should check the model card on Hugging Face to see if you are using the correct prompt format.
+Note that when you load a model in the "Model" tab, the web UI will try to automatically detect its instruction template (if any) from the model metadata (e.g. `tokenizer_config.json` or GGUF metadata), and will update the values under "Parameters" > "Instruction template" accordingly. You should check the model card on Hugging Face to see if you are using the correct prompt format.
 
 ### Chat-instruct
 
diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index 2a7a7f6915..0a076c350a 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -39,7 +39,7 @@ curl http://127.0.0.1:5000/v1/completions \
 
 #### Chat completions
 
-Works best with instruction-following models. If the "instruction_template" variable is not provided, it will be guessed automatically based on the model name using the regex patterns in `user_data/models/config.yaml`.
+Works best with instruction-following models. If the "instruction_template" variable is not provided, it will be detected automatically from the model metadata.
 
 ```shell
 curl http://127.0.0.1:5000/v1/chat/completions \
diff --git a/modules/models.py b/modules/models.py
index 1d139b89a6..b2665c6bb1 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -67,7 +67,6 @@ def load_model(model_name, loader=None):
     logger.info(f"Loaded \"{model_name}\" in {(time.time()-t0):.2f} seconds.")
     logger.info(f"LOADER: \"{loader}\"")
     logger.info(f"TRUNCATION LENGTH: {shared.settings['truncation_length']}")
-    logger.info(f"INSTRUCTION TEMPLATE: \"{metadata['instruction_template']}\"")
     return model, tokenizer
 
 
diff --git a/modules/models_settings.py b/modules/models_settings.py
index dcface7182..eafa058107 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -23,14 +23,9 @@ def get_fallback_settings():
 
 def get_model_metadata(model):
     model_path = resolve_model_path(model)
-    model_settings = {}
 
-    # Get settings from user_data/models/config.yaml and user_data/models/config-user.yaml
-    settings = shared.model_config
-    for pat in settings:
-        if re.match(pat.lower(), Path(model).name.lower()):
-            for k in settings[pat]:
-                model_settings[k] = settings[pat][k]
+    # Fallback settings
+    model_settings = get_fallback_settings()
 
     path = model_path / 'config.json'
     if path.exists():
diff --git a/modules/shared.py b/modules/shared.py
index 16ccbe77a4..acb103b45d 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -454,17 +454,7 @@ def load_user_config():
 
 args.loader = fix_loader_name(args.loader)
 
-# Load model-specific settings
-p = Path(f'{args.model_dir}/config.yaml')
-if p.exists():
-    model_config = yaml.safe_load(open(p, 'r').read())
-else:
-    model_config = {}
-del p
-
-
 # Load custom model-specific settings
 user_config = load_user_config()
 
-model_config = OrderedDict(model_config)
 user_config = OrderedDict(user_config)
diff --git a/server.py b/server.py
index d224909ce1..88936ca6e8 100644
--- a/server.py
+++ b/server.py
@@ -18,7 +18,6 @@
 from modules.LoRA import add_lora_to_model
 from modules.models import load_model, unload_model_if_idle
 from modules.models_settings import (
-    get_fallback_settings,
     get_model_metadata,
     update_model_parameters
 )
@@ -271,10 +270,6 @@ def create_interface():
     # Apply CLI overrides for image model settings (CLI flags take precedence over saved settings)
     shared.apply_image_model_cli_overrides()
 
-    # Fallback settings for models
-    shared.model_config['.*'] = get_fallback_settings()
-    shared.model_config.move_to_end('.*', last=False)  # Move to the beginning
-
     # Activate the extensions listed on settings.yaml
     extensions_module.available_extensions = utils.get_available_extensions()
     for extension in shared.settings['default_extensions']:
diff --git a/user_data/models/config.yaml b/user_data/models/config.yaml
deleted file mode 100644
index 038ebcf1ed..0000000000
--- a/user_data/models/config.yaml
+++ /dev/null
@@ -1,203 +0,0 @@
-.*(llama|alpac|vicuna|guanaco|koala|llava|wizardlm|metharme|pygmalion-7b|pygmalion-2|mythalion|wizard-mega|openbuddy|vigogne|h2ogpt-research|manticore):
-  model_type: 'llama'
-.*(opt-|opt_|opt1|opt3|optfor|galactica|galpaca|pygmalion-350m):
-  model_type: 'opt'
-.*(gpt-j|gptj|gpt4all-j|malion-6b|pygway|pygmalion-6b|dolly-v1):
-  model_type: 'gptj'
-.*(gpt-neox|koalpaca-polyglot|polyglot.*koalpaca|polyglot-ko|polyglot_ko|pythia|stablelm|incite|dolly-v2|polycoder|h2ogpt-oig|h2ogpt-oasst1|h2ogpt-gm):
-  model_type: 'gptneox'
-.*bloom:
-  model_type: 'bloom'
-.*gpt2:
-  model_type: 'gpt2'
-.*falcon:
-  model_type: 'falcon'
-.*mpt:
-  model_type: 'mpt'
-.*(starcoder|starchat):
-  model_type: 'starcoder'
-.*dolly-v2:
-  model_type: 'dollyv2'
-.*replit:
-  model_type: 'replit'
-.*(oasst|openassistant-|stablelm-7b-sft-v7-epoch-3):
-  instruction_template: 'Open Assistant'
-  skip_special_tokens: false
-(?!.*galactica)(?!.*reward).*openassistant:
-  instruction_template: 'Open Assistant'
-  skip_special_tokens: false
-.*galactica:
-  skip_special_tokens: false
-.*dolly-v[0-9]-[0-9]*b:
-  instruction_template: 'Alpaca'
-  skip_special_tokens: false
-.*alpaca-native-4bit:
-  instruction_template: 'Alpaca'
-.*llava:
-  instruction_template: 'LLaVA'
-.*llava.*1.5:
-  instruction_template: 'Vicuna-v1.1'
-.*wizard.*mega:
-  instruction_template: 'Wizard-Mega'
-.*starchat-beta:
-  instruction_template: 'Starchat-Beta'
-(?!.*v0)(?!.*1.1)(?!.*1_1)(?!.*stable)(?!.*chinese).*vicuna:
-  instruction_template: 'Vicuna-v0'
-.*vicuna.*v0:
-  instruction_template: 'Vicuna-v0'
-.*vicuna.*(1.1|1_1|1.3|1_3):
-  instruction_template: 'Vicuna-v1.1'
-.*vicuna.*(1.5|1_5):
-  instruction_template: 'Vicuna-v1.1'
-.*stable.*vicuna:
-  instruction_template: 'StableVicuna'
-(?!.*chat).*chinese-vicuna:
-  instruction_template: 'Alpaca'
-.*chinese-vicuna.*chat:
-  instruction_template: 'Chinese-Vicuna-Chat'
-.*alpaca:
-  instruction_template: 'Alpaca'
-.*koala:
-  instruction_template: 'Koala'
-.*chatglm:
-  instruction_template: 'ChatGLM'
-.*(metharme|pygmalion|mythalion):
-  instruction_template: 'Metharme'
-.*raven:
-  instruction_template: 'RWKV-Raven'
-.*moss-moon.*sft:
-  instruction_template: 'MOSS'
-.*stablelm-tuned:
-  instruction_template: 'StableLM'
-.*galactica.*finetuned:
-  instruction_template: 'Galactica Finetuned'
-.*galactica.*-v2:
-  instruction_template: 'Galactica v2'
-(?!.*finetuned)(?!.*-v2).*galactica:
-  instruction_template: 'Galactica'
-.*guanaco:
-  instruction_template: 'Guanaco non-chat'
-.*baize:
-  instruction_template: 'Baize'
-.*mpt-.*instruct:
-  instruction_template: 'Alpaca'
-.*mpt-.*chat:
-  instruction_template: 'ChatML'
-(?!.*-flan-)(?!.*-t5-).*lamini-:
-  instruction_template: 'Alpaca'
-.*incite.*chat:
-  instruction_template: 'INCITE-Chat'
-.*incite.*instruct:
-  instruction_template: 'INCITE-Instruct'
-.*ziya-:
-  instruction_template: 'Ziya'
-.*koalpaca:
-  instruction_template: 'KoAlpaca'
-.*openbuddy:
-  instruction_template: 'OpenBuddy'
-(?!.*chat).*vigogne:
-  instruction_template: 'Vigogne-Instruct'
-.*vigogne.*chat:
-  instruction_template: 'Vigogne-Chat'
-.*(llama-deus|supercot|llama-natural-instructions|open-llama-0.3t-7b-instruct-dolly-hhrlhf|open-llama-0.3t-7b-open-instruct):
-  instruction_template: 'Alpaca'
-.*bactrian:
-  instruction_template: 'Bactrian'
-.*(h2ogpt-oig-|h2ogpt-oasst1-|h2ogpt-research-oasst1-):
-  instruction_template: 'INCITE-Chat'
-.*h2ogpt-gm-:
-  instruction_template: 'H2O-prompt_answer'
-.*manticore:
-  instruction_template: 'Manticore Chat'
-.*bluemoonrp-(30|13)b:
-  instruction_template: 'Bluemoon'
-.*Nous-Hermes-13b:
-  instruction_template: 'Alpaca'
-.*airoboros:
-  instruction_template: 'Vicuna-v1.1'
-.*airoboros.*1.2:
-  instruction_template: 'Airoboros-v1.2'
-.*alpa(cino|sta):
-  instruction_template: 'Alpaca'
-.*hippogriff:
-  instruction_template: 'Hippogriff'
-.*lazarus:
-  instruction_template: 'Alpaca'
-.*guanaco-.*(7|13|33|65)b:
-  instruction_template: 'Vicuna-v0'
-.*hypermantis:
-  instruction_template: 'Alpaca'
-.*open-llama-.*-open-instruct:
-  instruction_template: 'Alpaca'
-.*starcoder-gpteacher-code-instruct:
-  instruction_template: 'Alpaca'
-.*tulu:
-  instruction_template: 'Tulu'
-.*chronos:
-  instruction_template: 'Alpaca'
-.*samantha:
-  instruction_template: 'Samantha'
-.*wizardcoder:
-  instruction_template: 'Alpaca'
-.*minotaur:
-  instruction_template: 'Manticore Chat'
-.*orca_mini:
-  instruction_template: 'Orca Mini'
-.*(platypus|gplatty|superplatty):
-  instruction_template: 'Alpaca'
-.*(openorca-platypus2):
-  instruction_template: 'OpenOrca-Platypus2'
-.*longchat:
-  instruction_template: 'Vicuna-v1.1'
-.*vicuna-33b:
-  instruction_template: 'Vicuna-v1.1'
-.*redmond-hermes-coder:
-  instruction_template: 'Alpaca'
-.*wizardcoder-15b:
-  instruction_template: 'Alpaca'
-.*wizardlm:
-  instruction_template: 'Vicuna-v1.1'
-.*godzilla:
-  instruction_template: 'Alpaca'
-.*llama(-?)(2|v2).*chat:
-  instruction_template: 'Llama-v2'
-.*newhope:
-  instruction_template: 'NewHope'
-.*stablebeluga2:
-  instruction_template: 'StableBeluga2'
-.*openchat:
-  instruction_template: 'OpenChat'
-.*codellama.*instruct:
-  instruction_template: 'Llama-v2'
-.*(mistral|mixtral).*instruct:
-  instruction_template: 'Mistral'
-.*mistral.*openorca:
-  instruction_template: 'ChatML'
-.*(WizardCoder-Python-34B-V1.0|Phind-CodeLlama-34B-v2|CodeBooga-34B-v0.1):
-  instruction_template: 'Alpaca'
-.*orca-2-(13|7)b:
-  instruction_template: 'ChatML'
-.*openhermes.*mistral:
-  instruction_template: 'ChatML'
-.*Yi-34B-Chat:
-  instruction_template: 'ChatML'
-(dolphin).*:
-  instruction_template: 'ChatML'
-.*synthia:
-  instruction_template: 'Synthia'
-.*(hercules|hyperion):
-  instruction_template: 'ChatML'
-.*command-r:
-  instruction_template: 'Command-R'
-.*xwin-lm-70b-v0.1:
-  instruction_template: 'Vicuna-v1.1'
-.*platypus-yi-34b:
-  instruction_template: 'Vicuna-v1.1'
-.*CausalLM-RP-34B:
-  instruction_template: 'ChatML'
-34b-beta:
-  instruction_template: 'ChatML'
-.*airoboros-3_1-yi-34b-200k:
-  instruction_template: 'Llama-v2'
-.*chatqa:
-  instruction_template: 'NVIDIA-ChatQA'

From d6f1485dd189494f6fbe5b6ea7ebd5cc0404233a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 24 Mar 2026 21:45:11 -0700
Subject: [PATCH 1451/1701] UI: Update the enable_thinking info message

---
 modules/ui_chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index f1dc7883cf..10d05f659c 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -82,7 +82,7 @@ def create_ui():
                 gr.HTML("
            ")
 
                 shared.gradio['reasoning_effort'] = gr.Dropdown(value=shared.settings['reasoning_effort'], choices=['low', 'medium', 'high'], label='Reasoning effort', info='Used by GPT-OSS.')
-                shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', info='Used by Seed-OSS and pre-2507 Qwen3.')
+                shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', info='For models with thinking support.')
 
                 gr.HTML("
            ")
 

From 368f37335f634ba001d00d2841902de85c7b48db Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 25 Mar 2026 06:37:45 -0700
Subject: [PATCH 1452/1701] Fix --idle-timeout issues with encode/decode and
 parallel generation

---
 modules/logits.py          |  4 +---
 modules/models.py          | 15 ++++++++++++++-
 modules/text_generation.py | 18 +++++++++++++-----
 3 files changed, 28 insertions(+), 9 deletions(-)

diff --git a/modules/logits.py b/modules/logits.py
index 1f878f272f..473f589033 100644
--- a/modules/logits.py
+++ b/modules/logits.py
@@ -4,7 +4,6 @@
 
 from modules import models, shared
 from modules.logging_colors import logger
-from modules.models import load_model
 from modules.text_generation import generate_reply
 from modules.utils import check_model_loaded
 
@@ -12,8 +11,7 @@
 
 
 def get_next_logits(*args, **kwargs):
-    if shared.args.idle_timeout > 0 and shared.model is None and shared.model_name not in [None, 'None']:
-        shared.model, shared.tokenizer = load_model(shared.model_name)
+    models.load_model_if_idle_unloaded()
 
     needs_lock = not args[2]  # use_samplers
     if needs_lock:
diff --git a/modules/models.py b/modules/models.py
index b2665c6bb1..61ca383878 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -1,4 +1,5 @@
 import sys
+import threading
 import time
 
 import modules.shared as shared
@@ -7,6 +8,15 @@
 from modules.utils import resolve_model_path
 
 last_generation_time = time.time()
+active_generation_count = 0
+_generation_count_lock = threading.Lock()
+
+
+def load_model_if_idle_unloaded():
+    global last_generation_time
+    if shared.args.idle_timeout > 0 and shared.model is None and shared.model_name not in [None, 'None']:
+        shared.model, shared.tokenizer = load_model(shared.model_name)
+        last_generation_time = time.time()
 
 
 def load_model(model_name, loader=None):
@@ -158,7 +168,10 @@ def unload_model_if_idle():
     while True:
         shared.generation_lock.acquire()
         try:
-            if time.time() - last_generation_time > shared.args.idle_timeout * 60:
+            with _generation_count_lock:
+                is_active = active_generation_count > 0
+
+            if not is_active and time.time() - last_generation_time > shared.args.idle_timeout * 60:
                 if shared.model is not None:
                     logger.info("Unloading the model for inactivity.")
                     unload_model(keep_model_name=True)
diff --git a/modules/text_generation.py b/modules/text_generation.py
index f77be124f9..3a9ddab531 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -17,9 +17,7 @@
 
 
 def generate_reply(*args, **kwargs):
-    if shared.args.idle_timeout > 0 and shared.model is None and shared.model_name not in [None, 'None']:
-        from modules.models import load_model
-        shared.model, shared.tokenizer = load_model(shared.model_name)
+    models.load_model_if_idle_unloaded()
 
     state = args[1] if len(args) > 1 else kwargs.get('state', {})
     use_parallel = (
@@ -31,10 +29,16 @@ def generate_reply(*args, **kwargs):
     if not use_parallel:
         shared.generation_lock.acquire()
 
+    with models._generation_count_lock:
+        models.active_generation_count += 1
+
     try:
         for result in _generate_reply(*args, **kwargs):
             yield result
     finally:
+        with models._generation_count_lock:
+            models.active_generation_count -= 1
+
         models.last_generation_time = time.time()
         if not use_parallel:
             shared.generation_lock.release()
@@ -126,7 +130,9 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
 
 def encode(prompt, add_special_tokens=True, add_bos_token=True, truncation_length=None):
     if shared.tokenizer is None:
-        raise ValueError('No tokenizer is loaded')
+        models.load_model_if_idle_unloaded()
+        if shared.tokenizer is None:
+            raise ValueError('No tokenizer is loaded')
 
     # llama.cpp case
     if shared.model.__class__.__name__ == 'LlamaServer':
@@ -176,7 +182,9 @@ def encode(prompt, add_special_tokens=True, add_bos_token=True, truncation_lengt
 
 def decode(output_ids, skip_special_tokens=True):
     if shared.tokenizer is None:
-        raise ValueError('No tokenizer is loaded')
+        models.load_model_if_idle_unloaded()
+        if shared.tokenizer is None:
+            raise ValueError('No tokenizer is loaded')
 
     return shared.tokenizer.decode(output_ids, skip_special_tokens=skip_special_tokens)
 

From e1541400219043f9b9cebf5f002b48251efc8bf9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 25 Mar 2026 07:21:02 -0700
Subject: [PATCH 1453/1701] Rename "truncation length" to "context length" in
 logs

---
 modules/api/models.py | 2 +-
 modules/models.py     | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/api/models.py b/modules/api/models.py
index c879a860f7..b89397d3dd 100644
--- a/modules/api/models.py
+++ b/modules/api/models.py
@@ -68,7 +68,7 @@ def _load_model(data):
             if k in shared.settings:
                 shared.settings[k] = settings[k]
                 if k == 'truncation_length':
-                    logger.info(f"TRUNCATION LENGTH (UPDATED): {shared.settings['truncation_length']}")
+                    logger.info(f"CONTEXT LENGTH (UPDATED): {shared.settings['truncation_length']}")
                 elif k == 'instruction_template':
                     logger.info(f"INSTRUCTION TEMPLATE (UPDATED): {shared.settings['instruction_template']}")
 
diff --git a/modules/models.py b/modules/models.py
index 61ca383878..e997d2d864 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -76,7 +76,7 @@ def load_model(model_name, loader=None):
 
     logger.info(f"Loaded \"{model_name}\" in {(time.time()-t0):.2f} seconds.")
     logger.info(f"LOADER: \"{loader}\"")
-    logger.info(f"TRUNCATION LENGTH: {shared.settings['truncation_length']}")
+    logger.info(f"CONTEXT LENGTH: {shared.settings['truncation_length']}")
     return model, tokenizer
 
 

From 4cbea02ed4e0dee2efd066ac48bcdf33631b9eca Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 26 Mar 2026 06:49:39 -0700
Subject: [PATCH 1454/1701] Add ik_llama.cpp support via `--ik` flag

---
 modules/llama_cpp_server.py | 37 +++++++++++++++++++++++++++++++++++++
 modules/shared.py           |  1 +
 2 files changed, 38 insertions(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 2ae01ddc2c..9b9756a95d 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -470,6 +470,10 @@ def _start_server(self):
                         else:
                             cmd.append(f"--{flag_item}")
 
+        # Patch flags for ik_llama.cpp compatibility
+        if shared.args.ik:
+            cmd = _patch_cmd_for_ik(cmd)
+
         env = os.environ.copy()
         if os.name == 'posix':
             current_path = env.get('LD_LIBRARY_PATH', '')
@@ -607,3 +611,36 @@ def filter_stderr_with_progress(process_stderr):
             process_stderr.close()
         except Exception:
             pass
+
+
+def _patch_cmd_for_ik(cmd):
+    """
+    Rewrite upstream llama.cpp flags to ik_llama.cpp equivalents:
+      --no-webui          → --webui none
+      --fit off            → (removed)
+      --fit on / --fit-ctx → --fit (bare flag)
+      --fit-target         → --fit-margin
+    """
+    patched = []
+    i = 0
+    while i < len(cmd):
+        arg = cmd[i]
+
+        if arg == "--no-webui":
+            patched += ["--webui", "none"]
+        elif arg == "--fit" and i + 1 < len(cmd) and cmd[i + 1] in ("on", "off"):
+            val = cmd[i + 1]
+            i += 1
+            if val == "on":
+                patched.append("--fit")
+            # "off" → drop entirely
+        elif arg == "--fit-ctx":
+            i += 1  # skip the value
+        elif arg == "--fit-target":
+            patched.append("--fit-margin")
+        else:
+            patched.append(arg)
+
+        i += 1
+
+    return patched
diff --git a/modules/shared.py b/modules/shared.py
index acb103b45d..c50736d703 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -110,6 +110,7 @@
 group.add_argument('--parallel', type=int, default=1, help='Number of parallel request slots. The context size is divided equally among slots. For example, to have 4 slots with 8192 context each, set ctx_size to 32768.')
 group.add_argument('--fit-target', type=str, default='512', help='Target VRAM margin per device for auto GPU layers, comma-separated list of values in MiB. A single value is broadcast across all devices.')
 group.add_argument('--extra-flags', type=str, default=None, help='Extra flags to pass to llama-server. Example: "--jinja --rpc 192.168.1.100:50052"')
+group.add_argument('--ik', action='store_true', help='Use ik_llama.cpp instead of upstream llama.cpp. To install: build ik_llama.cpp, then delete all files inside /lib/pythonX.Y/site-packages/llama_cpp_binaries/bin/ and copy or symlink the ik_llama.cpp build outputs into that folder.')
 
 # Transformers/Accelerate
 group = parser.add_argument_group('Transformers/Accelerate')

From bda95172bd6abecba165fc118f140cfc446f3c42 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 28 Mar 2026 06:09:53 -0700
Subject: [PATCH 1455/1701] Fix stopping string detection for
 chromadb/context-1

---
 modules/chat.py | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index f8088e0fab..edda11b09a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -671,7 +671,10 @@ def get_stopping_strings(state):
     # Handle GPT-OSS as a special case
     if '<|channel|>final<|message|>' in state['instruction_template_str'] and "<|end|>" in result:
         result.remove("<|end|>")
-        result.append("<|result|>")
+        if '<|result|>' in state['instruction_template_str']:
+            result.append("<|result|>")
+        elif '<|return|>' in state['instruction_template_str']:
+            result.append("<|return|>")
         result = list(set(result))
 
     if shared.args.verbose:

From 9dd04b86ce407507bcaf0862b97aadc64b6e62a6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 28 Mar 2026 06:17:57 -0700
Subject: [PATCH 1456/1701] Suppress EOS token at logit level for ExLlamav3
 when ban_eos_token is set

---
 modules/exllamav3.py | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 75c76c7c3b..f873503a8b 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -423,6 +423,15 @@ def custom_sort_key(sampler_obj):
         if logit_bias:
             filters.append(LogitBiasFilter(self.tokenizer, logit_bias))
 
+        # Suppress EOS tokens via logit bias so they are never sampled
+        if state['ban_eos_token']:
+            eos_bias = {}
+            for eos_id in self.config.eos_token_id_list:
+                if eos_id is not None:
+                    eos_bias[str(eos_id)] = float('-inf')
+            if eos_bias:
+                filters.append(LogitBiasFilter(self.tokenizer, eos_bias))
+
         # Logprobs support (OpenAI API)
         logprobs = state.get('logprobs', 0) or 0
         return_top_tokens = logprobs if logprobs > 0 else 0

From 4979e87e48c78d5e3186e4d9b2fbc8b30e86164f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 28 Mar 2026 11:49:47 -0300
Subject: [PATCH 1457/1701] Add ik_llama.cpp support via ik_llama_cpp_binaries
 package

---
 .github/workflows/build-everything-tgw.yml    |  35 +++
 .../build-portable-release-ik-cuda.yml        | 179 +++++++++++++++
 .../workflows/build-portable-release-ik.yml   | 205 ++++++++++++++++++
 modules/llama_cpp_server.py                   |  21 +-
 modules/loaders.py                            |   2 +
 modules/shared.py                             |   2 +-
 modules/ui_model_menu.py                      |   3 +
 requirements/full/requirements.txt            |   6 +-
 requirements/full/requirements_amd.txt        |   4 +-
 .../full/requirements_apple_intel.txt         |   3 +-
 .../full/requirements_apple_silicon.txt       |   3 +-
 requirements/full/requirements_cpu_only.txt   |   6 +-
 requirements/portable/requirements.txt        |   4 +-
 requirements/portable/requirements_amd.txt    |   4 +-
 .../portable/requirements_apple_intel.txt     |   2 +-
 .../portable/requirements_apple_silicon.txt   |   2 +-
 .../portable/requirements_cpu_only.txt        |   4 +-
 .../portable/requirements_cuda131.txt         |   4 +-
 requirements/portable/requirements_vulkan.txt |   4 +-
 19 files changed, 469 insertions(+), 24 deletions(-)
 create mode 100644 .github/workflows/build-portable-release-ik-cuda.yml
 create mode 100644 .github/workflows/build-portable-release-ik.yml

diff --git a/.github/workflows/build-everything-tgw.yml b/.github/workflows/build-everything-tgw.yml
index 9322f85929..4de591f407 100644
--- a/.github/workflows/build-everything-tgw.yml
+++ b/.github/workflows/build-everything-tgw.yml
@@ -68,3 +68,38 @@ jobs:
     with:
       version: ${{ inputs.version }}
       config: 'os:macos-15-intel,macos-14'
+
+  build_release_ik_cuda_windows:
+    name: ik CUDA Windows
+    uses: ./.github/workflows/build-portable-release-ik-cuda.yml
+    with:
+      version: ${{ inputs.version }}
+      config: 'os:windows-2022'
+
+  build_release_ik_cuda_linux:
+    name: ik CUDA Linux
+    uses: ./.github/workflows/build-portable-release-ik-cuda.yml
+    with:
+      version: ${{ inputs.version }}
+      config: 'os:ubuntu-22.04'
+
+  build_release_ik_cpu_windows:
+    name: ik CPU Windows
+    uses: ./.github/workflows/build-portable-release-ik.yml
+    with:
+      version: ${{ inputs.version }}
+      config: 'os:windows-2022'
+
+  build_release_ik_cpu_linux:
+    name: ik CPU Linux
+    uses: ./.github/workflows/build-portable-release-ik.yml
+    with:
+      version: ${{ inputs.version }}
+      config: 'os:ubuntu-22.04'
+
+  build_release_ik_macos:
+    name: ik macOS
+    uses: ./.github/workflows/build-portable-release-ik.yml
+    with:
+      version: ${{ inputs.version }}
+      config: 'os:macos-14'
diff --git a/.github/workflows/build-portable-release-ik-cuda.yml b/.github/workflows/build-portable-release-ik-cuda.yml
new file mode 100644
index 0000000000..40b4b92f98
--- /dev/null
+++ b/.github/workflows/build-portable-release-ik-cuda.yml
@@ -0,0 +1,179 @@
+name: Build ik CUDA
+
+on:
+  workflow_dispatch:
+    inputs:
+      version:
+        description: 'Version tag of text-generation-webui to build: v3.0'
+        default: 'v3.0'
+        required: true
+        type: string
+      config:
+        description: 'Override configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2'
+        default: 'Default'
+        required: false
+        type: string
+      exclude:
+        description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
+        default: 'None'
+        required: false
+        type: string
+  workflow_call:
+    inputs:
+      version:
+        description: 'Version tag of text-generation-webui to build: v3.0'
+        default: 'v3.0'
+        required: true
+        type: string
+      config:
+        description: 'Configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2'
+        default: 'Default'
+        required: false
+        type: string
+      exclude:
+        description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
+        default: 'None'
+        required: false
+        type: string
+
+permissions:
+  contents: write
+
+jobs:
+  define_matrix:
+    name: Define Build Matrix
+    runs-on: ubuntu-latest
+    outputs:
+      matrix: ${{ steps.set-matrix.outputs.matrix }}
+    defaults:
+      run:
+        shell: pwsh
+    env:
+      CONFIGIN: ${{ inputs.config }}
+      EXCLUDEIN: ${{ inputs.exclude }}
+
+    steps:
+      - name: Define Job Output
+        id: set-matrix
+        run: |
+          $matrix = @{
+              'os' = @('ubuntu-22.04', 'windows-2022')
+              'pyver' = @("3.13")
+              'cuda' = @("12.4", "13.1")
+          }
+
+          if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})}
+
+          if ($env:EXCLUDEIN -ne 'None') {
+              $exclusions = @()
+              $exclusions += $env:EXCLUDEIN.split(';').replace(':','=').replace(',',"`n") | ConvertFrom-StringData
+              $matrix['exclude'] = $exclusions
+          }
+
+          $matrixOut = ConvertTo-Json $matrix -Compress
+          Write-Output ('matrix=' + $matrixOut) >> $env:GITHUB_OUTPUT
+
+  build_wheels:
+    name: ${{ matrix.os }} ${{ matrix.pyver }} CUDA ${{ matrix.cuda }}
+    needs: define_matrix
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix: ${{ fromJSON(needs.define_matrix.outputs.matrix) }}
+    defaults:
+      run:
+        shell: pwsh
+    env:
+      PCKGVER: ${{ inputs.version }}
+
+    steps:
+      - uses: actions/checkout@v6
+        with:
+          repository: 'oobabooga/text-generation-webui'
+          ref: ${{ inputs.version }}
+          submodules: 'recursive'
+
+      - uses: actions/setup-python@v6
+        with:
+          python-version: ${{ matrix.pyver }}
+
+      - name: Build Package
+        shell: bash
+        run: |
+            VERSION_CLEAN="${{ inputs.version }}"
+            VERSION_CLEAN="${VERSION_CLEAN#v}"
+            cd ..
+            cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
+            cd "text-generation-webui-${VERSION_CLEAN}"
+
+            # Remove extensions that need additional requirements
+            allowed=("character_bias" "gallery" "sd_api_pictures")
+            find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
+
+            # Define common variables
+            CUDA_VERSION="${{ matrix.cuda }}"
+            VERSION="${{ inputs.version }}"
+
+            # 1. Set platform-specific variables
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                PLATFORM="windows"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-pc-windows-msvc-install_only_stripped.tar.gz"
+                PIP_PATH="portable_env/python.exe -m pip"
+                PACKAGES_PATH="portable_env/Lib/site-packages"
+                rm start_linux.sh start_macos.sh
+            else
+                PLATFORM="linux"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only_stripped.tar.gz"
+                PIP_PATH="portable_env/bin/python -m pip"
+                PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
+                rm start_macos.sh start_windows.bat
+            fi
+
+            # 2. Download and extract Python
+            cd ..
+            echo "Downloading Python for $PLATFORM..."
+            curl -L -o python-build.tar.gz "$PYTHON_URL"
+            tar -xzf python-build.tar.gz
+            mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
+
+            # 3. Prepare requirements file based on CUDA version
+            cd "text-generation-webui-${VERSION_CLEAN}"
+            if [[ "$CUDA_VERSION" == "13.1" ]]; then
+                REQ_FILE="requirements/portable/requirements_cuda131.txt"
+            else
+                REQ_FILE="requirements/portable/requirements.txt"
+            fi
+
+            # 4. Swap llama.cpp wheels for ik_llama.cpp and inject --ik into start scripts
+            sed -i 's|/llama_cpp_binaries-|/ik_llama_cpp_binaries-|g' "$REQ_FILE"
+            sed -i 's/--portable/--portable --ik/g' start_linux.sh start_windows.bat start_macos.sh 2>/dev/null || true
+
+            # 5. Install packages
+            echo "Installing Python packages from $REQ_FILE..."
+            $PIP_PATH install --target="./$PACKAGES_PATH" -r "$REQ_FILE"
+
+            # 6. Clean up
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore requirements/ one_click.py
+
+            # 7. Create archive
+            cd ..
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                ARCHIVE_NAME="textgen-portable-ik-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
+                echo "Creating archive: $ARCHIVE_NAME"
+                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
+            else
+                ARCHIVE_NAME="textgen-portable-ik-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.tar.gz"
+                echo "Creating archive: $ARCHIVE_NAME"
+                tar czf "$ARCHIVE_NAME" "text-generation-webui-${VERSION_CLEAN}"
+            fi
+
+      - name: Upload files to a GitHub release
+        id: upload-release
+        uses: svenstaro/upload-release-action@2.7.0
+        continue-on-error: true
+        with:
+          repo_token: ${{ secrets.GITHUB_TOKEN }}
+          file: ../textgen-portable-ik-*
+          tag: ${{ inputs.version }}
+          file_glob: true
+          make_latest: false
+          overwrite: true
diff --git a/.github/workflows/build-portable-release-ik.yml b/.github/workflows/build-portable-release-ik.yml
new file mode 100644
index 0000000000..afb2e76327
--- /dev/null
+++ b/.github/workflows/build-portable-release-ik.yml
@@ -0,0 +1,205 @@
+name: Build ik CPU and macOS
+
+on:
+  workflow_dispatch:
+    inputs:
+      version:
+        description: 'Version tag of text-generation-webui to build: v3.0'
+        default: 'v3.0'
+        required: true
+        type: string
+      config:
+        description: 'Override configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2'
+        default: 'Default'
+        required: false
+        type: string
+      exclude:
+        description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
+        default: 'None'
+        required: false
+        type: string
+  workflow_call:
+    inputs:
+      version:
+        description: 'Version tag of text-generation-webui to build: v3.0'
+        default: 'v3.0'
+        required: true
+        type: string
+      config:
+        description: 'Configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2'
+        default: 'Default'
+        required: false
+        type: string
+      exclude:
+        description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2'
+        default: 'None'
+        required: false
+        type: string
+
+permissions:
+  contents: write
+
+jobs:
+  define_matrix:
+    name: Define Build Matrix
+    runs-on: ubuntu-latest
+    outputs:
+      matrix: ${{ steps.set-matrix.outputs.matrix }}
+    defaults:
+      run:
+        shell: pwsh
+    env:
+      CONFIGIN: ${{ inputs.config }}
+      EXCLUDEIN: ${{ inputs.exclude }}
+
+    steps:
+      - name: Define Job Output
+        id: set-matrix
+        run: |
+          $matrix = @{
+              'os' = @('ubuntu-22.04', 'windows-2022', 'macos-14')
+              'pyver' = @("3.13")
+          }
+
+          if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})}
+
+          if ($env:EXCLUDEIN -ne 'None') {
+              $exclusions = @()
+              $exclusions += $env:EXCLUDEIN.split(';').replace(':','=').replace(',',"`n") | ConvertFrom-StringData
+              $matrix['exclude'] = $exclusions
+          }
+
+          $matrixOut = ConvertTo-Json $matrix -Compress
+          Write-Output ('matrix=' + $matrixOut) >> $env:GITHUB_OUTPUT
+
+  build_wheels:
+    name: ${{ matrix.os }} ${{ matrix.pyver }}
+    needs: define_matrix
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix: ${{ fromJSON(needs.define_matrix.outputs.matrix) }}
+    defaults:
+      run:
+        shell: pwsh
+    env:
+      PCKGVER: ${{ inputs.version }}
+
+    steps:
+      - uses: actions/checkout@v6
+        with:
+          repository: 'oobabooga/text-generation-webui'
+          ref: ${{ inputs.version }}
+          submodules: 'recursive'
+
+      - uses: actions/setup-python@v6
+        with:
+          python-version: ${{ matrix.pyver }}
+
+      - name: Build Package
+        shell: bash
+        run: |
+            VERSION_CLEAN="${{ inputs.version }}"
+            VERSION_CLEAN="${VERSION_CLEAN#v}"
+            cd ..
+            cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
+            cd "text-generation-webui-${VERSION_CLEAN}"
+
+            # Remove extensions that need additional requirements
+            allowed=("character_bias" "gallery" "sd_api_pictures")
+            find extensions/ -mindepth 1 -maxdepth 1 -type d | grep -v -E "$(printf '%s|' "${allowed[@]}" | sed 's/|$//')" | xargs rm -rf
+
+            # Define common variables
+            VERSION="${{ inputs.version }}"
+            OS_TYPE="${{ matrix.os }}"
+
+            # 1. Set platform-specific variables
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                PLATFORM="windows-cpu"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-pc-windows-msvc-install_only_stripped.tar.gz"
+                PIP_PATH="portable_env/python.exe -m pip"
+                PACKAGES_PATH="portable_env/Lib/site-packages"
+                rm start_linux.sh start_macos.sh
+            elif [[ "$RUNNER_OS" == "macOS" ]]; then
+                if [[ "$OS_TYPE" == "macos-15-intel" ]]; then
+                    PLATFORM="macos-x86_64"
+                    PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-apple-darwin-install_only_stripped.tar.gz"
+                    REQ_TYPE="apple_intel"
+                else
+                    PLATFORM="macos-arm64"
+                    PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-aarch64-apple-darwin-install_only_stripped.tar.gz"
+                    REQ_TYPE="apple_silicon"
+                fi
+                PIP_PATH="portable_env/bin/python -m pip"
+                PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
+                rm start_linux.sh start_windows.bat
+            else
+                # Linux case
+                PLATFORM="linux-cpu"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only_stripped.tar.gz"
+                PIP_PATH="portable_env/bin/python -m pip"
+                PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
+                rm start_macos.sh start_windows.bat
+            fi
+
+            # 2. Download and extract Python
+            echo "Downloading Python for $PLATFORM..."
+            cd ..
+            curl -L -o python-build.tar.gz "$PYTHON_URL"
+            tar -xzf python-build.tar.gz
+            mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
+
+            # 3. Prepare requirements file based on platform
+            cd "text-generation-webui-${VERSION_CLEAN}"
+
+            # Select requirements file based on platform
+            if [[ "$RUNNER_OS" == "macOS" ]]; then
+                if [[ "$OS_TYPE" == "macos-15-intel" ]]; then
+                    REQ_FILE="requirements/portable/requirements_apple_intel.txt"
+                else
+                    REQ_FILE="requirements/portable/requirements_apple_silicon.txt"
+                fi
+            else
+                REQ_FILE="requirements/portable/requirements_cpu_only.txt"
+            fi
+
+            echo "Using requirements file: $REQ_FILE"
+
+            # 4. Swap llama.cpp wheels for ik_llama.cpp and inject --ik into start scripts
+            if [[ "$RUNNER_OS" == "macOS" ]]; then
+                sed -i '' 's|/llama_cpp_binaries-|/ik_llama_cpp_binaries-|g' "$REQ_FILE"
+                sed -i '' 's/--portable/--portable --ik/g' start_macos.sh
+            else
+                sed -i 's|/llama_cpp_binaries-|/ik_llama_cpp_binaries-|g' "$REQ_FILE"
+                sed -i 's/--portable/--portable --ik/g' start_linux.sh start_windows.bat 2>/dev/null || true
+            fi
+
+            # 5. Install packages
+            echo "Installing Python packages from $REQ_FILE..."
+            $PIP_PATH install --target="./$PACKAGES_PATH" -r "$REQ_FILE"
+
+            # 6. Clean up
+            rm -rf .git cmd* update_wizard* Colab-TextGen-GPU.ipynb docker setup.cfg .github .gitignore requirements/ one_click.py
+
+            # 7. Create archive
+            cd ..
+            if [[ "$RUNNER_OS" == "Windows" ]]; then
+                ARCHIVE_NAME="textgen-portable-ik-${VERSION_CLEAN}-${PLATFORM}.zip"
+                echo "Creating archive: $ARCHIVE_NAME"
+                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
+            else
+                ARCHIVE_NAME="textgen-portable-ik-${VERSION_CLEAN}-${PLATFORM}.tar.gz"
+                echo "Creating archive: $ARCHIVE_NAME"
+                tar czf "$ARCHIVE_NAME" "text-generation-webui-${VERSION_CLEAN}"
+            fi
+
+      - name: Upload files to a GitHub release
+        id: upload-release
+        uses: svenstaro/upload-release-action@2.7.0
+        continue-on-error: true
+        with:
+          repo_token: ${{ secrets.GITHUB_TOKEN }}
+          file: ../textgen-portable-ik-*
+          tag: ${{ inputs.version }}
+          file_glob: true
+          make_latest: false
+          overwrite: true
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 9b9756a95d..5e2decfa27 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -11,7 +11,6 @@
 from pathlib import Path
 from typing import Any, List
 
-import llama_cpp_binaries
 import requests
 
 from modules import shared
@@ -357,7 +356,16 @@ def _start_server(self):
         """Start the llama.cpp server and wait until it's ready."""
         # Determine the server path
         if self.server_path is None:
-            self.server_path = llama_cpp_binaries.get_binary_path()
+            if shared.args.ik:
+                try:
+                    import ik_llama_cpp_binaries
+                except ImportError:
+                    raise ImportError("--ik requires the ik_llama_cpp_binaries package. Install it with: pip install ")
+
+                self.server_path = ik_llama_cpp_binaries.get_binary_path()
+            else:
+                import llama_cpp_binaries
+                self.server_path = llama_cpp_binaries.get_binary_path()
 
         # Build the command
         cmd = [
@@ -616,10 +624,12 @@ def filter_stderr_with_progress(process_stderr):
 def _patch_cmd_for_ik(cmd):
     """
     Rewrite upstream llama.cpp flags to ik_llama.cpp equivalents:
-      --no-webui          → --webui none
+      --no-webui           → --webui none
       --fit off            → (removed)
       --fit on / --fit-ctx → --fit (bare flag)
       --fit-target         → --fit-margin
+      --cache-reuse        → (removed, unsupported)
+      --swa-full           → (removed, unsupported)
     """
     patched = []
     i = 0
@@ -635,9 +645,14 @@ def _patch_cmd_for_ik(cmd):
                 patched.append("--fit")
             # "off" → drop entirely
         elif arg == "--fit-ctx":
+            patched.append("--fit")
             i += 1  # skip the value
         elif arg == "--fit-target":
             patched.append("--fit-margin")
+        elif arg == "--cache-reuse":
+            i += 1  # skip the value
+        elif arg == "--swa-full":
+            pass  # bare flag, just drop it
         else:
             patched.append(arg)
 
diff --git a/modules/loaders.py b/modules/loaders.py
index c90f2ebbec..cb1f3d3bad 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -20,6 +20,7 @@
         'no_mmap',
         'mlock',
         'numa',
+        'ik',
         'parallel',
         'model_draft',
         'draft_max',
@@ -345,6 +346,7 @@ def list_model_elements():
         'spec_ngram_size_m',
         'spec_ngram_min_hits',
         'mmproj',
+        'ik',
     ]
 
 
diff --git a/modules/shared.py b/modules/shared.py
index c50736d703..13843f0c52 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -110,7 +110,7 @@
 group.add_argument('--parallel', type=int, default=1, help='Number of parallel request slots. The context size is divided equally among slots. For example, to have 4 slots with 8192 context each, set ctx_size to 32768.')
 group.add_argument('--fit-target', type=str, default='512', help='Target VRAM margin per device for auto GPU layers, comma-separated list of values in MiB. A single value is broadcast across all devices.')
 group.add_argument('--extra-flags', type=str, default=None, help='Extra flags to pass to llama-server. Example: "--jinja --rpc 192.168.1.100:50052"')
-group.add_argument('--ik', action='store_true', help='Use ik_llama.cpp instead of upstream llama.cpp. To install: build ik_llama.cpp, then delete all files inside /lib/pythonX.Y/site-packages/llama_cpp_binaries/bin/ and copy or symlink the ik_llama.cpp build outputs into that folder.')
+group.add_argument('--ik', action='store_true', help='Use ik_llama.cpp instead of upstream llama.cpp. Requires the ik_llama_cpp_binaries package to be installed.')
 
 # Transformers/Accelerate
 group = parser.add_argument_group('Transformers/Accelerate')
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 5b7621a76e..16505afa55 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -51,6 +51,9 @@ def create_ui():
 
                         with gr.Column():
                             shared.gradio['vram_info'] = gr.HTML(value=get_initial_vram_info())
+                            if not shared.args.portable:
+                                shared.gradio['ik'] = gr.Checkbox(label="ik", value=shared.args.ik, info='Use ik_llama.cpp instead of upstream llama.cpp.')
+
                             shared.gradio['cpu_moe'] = gr.Checkbox(label="cpu-moe", value=shared.args.cpu_moe, info='Move the experts to the CPU. Saves VRAM on MoE models.')
                             shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming-llm", value=shared.args.streaming_llm, info='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
                             shared.gradio['load_in_8bit'] = gr.Checkbox(label="load-in-8bit", value=shared.args.load_in_8bit)
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 5661962700..100c99d17f 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,8 +40,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/ik_llama_cpp_binaries-0.99.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/ik_llama_cpp_binaries-0.99.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.26/exllamav3-0.0.26+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.26/exllamav3-0.0.26+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 620683cce2..66fa4ac76d 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index b1f109b2e0..98dc8be65c 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,4 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/ik_llama_cpp_binaries-0.99.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index a54476a95d..e33264cf85 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,4 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/ik_llama_cpp_binaries-0.99.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index be82c904f7..cd083f6d11 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,5 +37,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/ik_llama_cpp_binaries-0.99.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/ik_llama_cpp_binaries-0.99.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 188da3809a..671822251c 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 4562b6d01b..5f5b2f8d91 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 04dcf25e12..f5f7d6eec3 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 4b8af78a73..e51fc29666 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 5b0eaf892e..683f94c890 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 90b3234f37..942d087737 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index ea72b4ec15..ae784e00b5 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.98.0/llama_cpp_binaries-0.98.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From be6fc0663ac1b7a60b7fde24afb38de2b0aba57b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 28 Mar 2026 08:11:28 -0700
Subject: [PATCH 1458/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 14 files changed, 28 insertions(+), 28 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 100c99d17f..6e11dd2fa0 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -31,8 +31,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 66fa4ac76d..c964eff69f 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 98dc8be65c..b1dd6a4fdc 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index e33264cf85..4d03d28007 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index cd083f6d11..9d41d069c2 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 77c254e681..052085cc42 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 671822251c..ff80b6c836 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 5f5b2f8d91..318044da94 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index f5f7d6eec3..1676bffbcd 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index e51fc29666..27fc2da81b 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 683f94c890..0bbdd30a0d 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 942d087737..c3ae3c57dd 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index e8457909d9..e38140cebd 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index ae784e00b5..e646c04c93 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio-4.37.2+custom.12-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.12/gradio_client-1.0.2+custom.12-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From 0466b6e2714a05c04eff0c929f15e4679f029e8d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 29 Mar 2026 15:52:36 -0700
Subject: [PATCH 1459/1701] ik_llama.cpp: Auto-enable Hadamard KV cache
 rotation with quantized cache

---
 modules/llama_cpp_server.py | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 5e2decfa27..fa968be13a 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -631,6 +631,12 @@ def _patch_cmd_for_ik(cmd):
       --cache-reuse        → (removed, unsupported)
       --swa-full           → (removed, unsupported)
     """
+    # Add Hadamard KV cache rotation when using quantized cache types.
+    # This significantly improves quantized cache quality (especially q4_0)
+    # and is a no-op for MLA models like DeepSeek.
+    if shared.args.cache_type in ("q8_0", "q4_0"):
+        cmd += ["-khad", "-vhad"]
+
     patched = []
     i = 0
     while i < len(cmd):

From 6382fbef8381bf60ff909b4fd76e7c1f4c063afc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 30 Mar 2026 17:44:19 -0700
Subject: [PATCH 1460/1701] Several small code simplifications

---
 download-model.py        |  25 +++---
 js/dark_theme.js         |  12 ++-
 js/global_scope_js.js    |  79 +++++++++---------
 js/main.js               | 171 +++++++++++++--------------------------
 js/save_files.js         |  18 ++---
 js/show_controls.js      |  21 ++---
 js/switch_tabs.js        |  24 ++----
 js/update_big_picture.js |   3 +-
 modules/extensions.py    |  22 +++--
 9 files changed, 140 insertions(+), 235 deletions(-)

diff --git a/download-model.py b/download-model.py
index 95d25e1630..a31bbfc637 100644
--- a/download-model.py
+++ b/download-model.py
@@ -158,28 +158,21 @@ def get_download_links_from_huggingface(self, model, branch, text_only=False, sp
         # Also if GGUF and safetensors are available, download only safetensors
         if (has_pytorch or has_pt or has_gguf) and has_safetensors:
             has_gguf = False
-            for i in range(len(classifications) - 1, -1, -1):
-                if classifications[i] in ['pytorch', 'pt', 'gguf']:
-                    links.pop(i)
-                    file_sizes.pop(i)
+            keep = [i for i, c in enumerate(classifications) if c not in ['pytorch', 'pt', 'gguf']]
+            links = [links[i] for i in keep]
+            file_sizes = [file_sizes[i] for i in keep]
 
         # For GGUF, try to download only the Q4_K_M if no specific file is specified.
         if has_gguf and specific_file is None:
-            has_q4km = False
-            for i in range(len(classifications) - 1, -1, -1):
-                if 'q4_k_m' in links[i].lower():
-                    has_q4km = True
+            has_q4km = any('q4_k_m' in link.lower() for link in links)
 
             if has_q4km:
-                for i in range(len(classifications) - 1, -1, -1):
-                    if 'q4_k_m' not in links[i].lower():
-                        links.pop(i)
-                        file_sizes.pop(i)
+                keep = [i for i, link in enumerate(links) if 'q4_k_m' in link.lower()]
             else:
-                for i in range(len(classifications) - 1, -1, -1):
-                    if links[i].lower().endswith('.gguf'):
-                        links.pop(i)
-                        file_sizes.pop(i)
+                keep = [i for i, link in enumerate(links) if not link.lower().endswith('.gguf')]
+
+            links = [links[i] for i in keep]
+            file_sizes = [file_sizes[i] for i in keep]
 
         is_llamacpp = has_gguf and specific_file is not None
         return links, sha256, is_lora, is_llamacpp, file_sizes
diff --git a/js/dark_theme.js b/js/dark_theme.js
index 7136f5bf2e..9d7069e206 100644
--- a/js/dark_theme.js
+++ b/js/dark_theme.js
@@ -1,6 +1,6 @@
 function toggleDarkMode() {
   document.body.classList.toggle("dark");
-  var currentCSS = document.getElementById("highlight-css");
+  const currentCSS = document.getElementById("highlight-css");
   if (currentCSS.getAttribute("href") === "file/css/highlightjs/github-dark.min.css") {
     currentCSS.setAttribute("href", "file/css/highlightjs/github.min.css");
   } else {
@@ -9,12 +9,10 @@ function toggleDarkMode() {
 
   // Re-highlight all code blocks once stylesheet loads
   currentCSS.onload = function() {
-    const messageBodies = document.getElementById("chat").querySelectorAll(".message-body");
-    messageBodies.forEach((messageBody) => {
-      const codeBlocks = messageBody.querySelectorAll("pre code");
-      codeBlocks.forEach((codeBlock) => {
-        hljs.highlightElement(codeBlock);
-      });
+    // Clear data-highlighted so hljs will re-process with the new theme
+    document.querySelectorAll("#chat .message-body pre code[data-highlighted]").forEach((codeBlock) => {
+      delete codeBlock.dataset.highlighted;
     });
+    doSyntaxHighlighting();
   };
 }
diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 92f65622ff..20eeef66a0 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -1,11 +1,35 @@
+// -------------------------------------------------
+// Shared helpers
+// -------------------------------------------------
+
+function getProfilePictureUrl() {
+  return "/file/user_data/cache/pfp_character.png?time=" + Date.now();
+}
+
+const MESSAGE_SELECTOR = ".message, .user-message, .assistant-message";
+
+function getMessageElement(element) {
+  if (!element) return null;
+  return element.closest(MESSAGE_SELECTOR);
+}
+
+function isUserRole(messageElement) {
+  return messageElement.classList.contains("user-message") ||
+         messageElement.querySelector(".text-you") !== null ||
+         messageElement.querySelector(".circle-you") !== null;
+}
+
+// Trigger a synthetic 'input' event so Gradio picks up programmatic value changes
+function dispatchGradioInput(element) {
+  element.dispatchEvent(new Event("input", { bubbles: true }));
+}
+
 // -------------------------------------------------
 // Event handlers
 // -------------------------------------------------
 
 function copyToClipboard(element) {
-  if (!element) return;
-
-  const messageElement = element.closest(".message, .user-message, .assistant-message");
+  const messageElement = getMessageElement(element);
   if (!messageElement) return;
 
   const rawText = messageElement.getAttribute("data-raw");
@@ -48,9 +72,7 @@ function fallbackCopyToClipboard(text) {
 }
 
 function branchHere(element) {
-  if (!element) return;
-
-  const messageElement = element.closest(".message, .user-message, .assistant-message");
+  const messageElement = getMessageElement(element);
   if (!messageElement) return;
 
   const index = messageElement.getAttribute("data-index");
@@ -69,11 +91,7 @@ function branchHere(element) {
   }
 
   branchIndexInput.value = index;
-
-  // Trigger any 'change' or 'input' events Gradio might be listening for
-  const event = new Event("input", { bubbles: true });
-  branchIndexInput.dispatchEvent(event);
-
+  dispatchGradioInput(branchIndexInput);
   branchButton.click();
 }
 
@@ -82,9 +100,7 @@ function branchHere(element) {
 // -------------------------------------------------
 
 function editHere(buttonElement) {
-  if (!buttonElement) return;
-
-  const messageElement = buttonElement.closest(".message, .user-message, .assistant-message");
+  const messageElement = getMessageElement(buttonElement);
   if (!messageElement) return;
 
   const messageBody = messageElement.querySelector(".message-body");
@@ -97,12 +113,7 @@ function editHere(buttonElement) {
     return;
   }
 
-  // Determine role based on message element - handle different chat modes
-  const isUserMessage = messageElement.classList.contains("user-message") ||
-                       messageElement.querySelector(".text-you") !== null ||
-                       messageElement.querySelector(".circle-you") !== null;
-
-  startEditing(messageElement, messageBody, isUserMessage);
+  startEditing(messageElement, messageBody, isUserRole(messageElement));
 }
 
 function startEditing(messageElement, messageBody, isUserMessage) {
@@ -209,30 +220,22 @@ function submitMessageEdit(index, newText, isUserMessage) {
   editTextInput.value = newText;
   editRoleInput.value = isUserMessage ? "user" : "assistant";
 
-  editIndexInput.dispatchEvent(new Event("input", { bubbles: true }));
-  editTextInput.dispatchEvent(new Event("input", { bubbles: true }));
-  editRoleInput.dispatchEvent(new Event("input", { bubbles: true }));
+  dispatchGradioInput(editIndexInput);
+  dispatchGradioInput(editTextInput);
+  dispatchGradioInput(editRoleInput);
 
   editButton.click();
   return true;
 }
 
 function navigateVersion(element, direction) {
-  if (!element) return;
-
-  const messageElement = element.closest(".message, .user-message, .assistant-message");
+  const messageElement = getMessageElement(element);
   if (!messageElement) return;
 
   const index = messageElement.getAttribute("data-index");
   if (!index) return;
 
-  // Determine role based on message element classes
-  let role = "assistant"; // Default role
-  if (messageElement.classList.contains("user-message") ||
-      messageElement.querySelector(".text-you") ||
-      messageElement.querySelector(".circle-you")) {
-    role = "user";
-  }
+  const role = isUserRole(messageElement) ? "user" : "assistant";
 
   const indexInput = document.getElementById("Navigate-message-index")?.querySelector("input");
   const directionInput = document.getElementById("Navigate-direction")?.querySelector("textarea");
@@ -248,11 +251,9 @@ function navigateVersion(element, direction) {
   directionInput.value = direction;
   roleInput.value = role;
 
-  // Trigger 'input' events for Gradio to pick up changes
-  const event = new Event("input", { bubbles: true });
-  indexInput.dispatchEvent(event);
-  directionInput.dispatchEvent(event);
-  roleInput.dispatchEvent(event);
+  dispatchGradioInput(indexInput);
+  dispatchGradioInput(directionInput);
+  dispatchGradioInput(roleInput);
 
   navigateButton.click();
 }
@@ -313,7 +314,7 @@ function handleMorphdomUpdate(data) {
 
 function applyMorphdomUpdate(data) {
   // Determine target element and use it as query scope
-  var target_element, target_html;
+  let target_element, target_html;
   if (data.last_message_only) {
     const childNodes = document.getElementsByClassName("messages")[0].childNodes;
     target_element = childNodes[childNodes.length - 1];
diff --git a/js/main.js b/js/main.js
index f05f93c600..cba4c903c1 100644
--- a/js/main.js
+++ b/js/main.js
@@ -4,8 +4,9 @@
 
 // Sync highlight.js theme with the actual Gradio theme
 var defined_hljs_css = document.body.classList.contains("dark") ? "file/css/highlightjs/github-dark.min.css" : "file/css/highlightjs/github.min.css";
-if (document.getElementById("highlight-css").getAttribute("href") !== defined_hljs_css) {
-  document.getElementById("highlight-css").setAttribute("href", defined_hljs_css);
+var hljsCssElement = document.getElementById("highlight-css");
+if (hljsCssElement.getAttribute("href") !== defined_hljs_css) {
+  hljsCssElement.setAttribute("href", defined_hljs_css);
 }
 
 let main_parent = document.getElementById("chat-tab").parentNode;
@@ -49,21 +50,18 @@ document.querySelector(".header_bar").addEventListener("click", function(event)
 //------------------------------------------------
 
 // --- Helper functions --- //
-function isModifiedKeyboardEvent() {
-  return (event instanceof KeyboardEvent &&
-    event.shiftKey ||
-    event.ctrlKey ||
-    event.altKey ||
-    event.metaKey);
+function isModifiedKeyboardEvent(event) {
+  return event instanceof KeyboardEvent &&
+    (event.shiftKey || event.ctrlKey || event.altKey || event.metaKey);
 }
 
-function isFocusedOnEditableTextbox() {
+function isFocusedOnEditableTextbox(event) {
   if (event.target.tagName === "INPUT" || event.target.tagName === "TEXTAREA") {
     return !!event.target.value;
   }
+  return false;
 }
 
-let previousTabId = "chat-tab-button";
 document.addEventListener("keydown", function(event) {
   // Stop generation on Esc pressed
   if (event.key === "Escape") {
@@ -117,14 +115,14 @@ document.addEventListener("keydown", function(event) {
   }
 
   // --- Simple version navigation --- //
-  if (!isFocusedOnEditableTextbox()) {
+  if (!isFocusedOnEditableTextbox(event)) {
     // Version navigation on Arrow keys (horizontal)
-    if (!isModifiedKeyboardEvent() && event.key === "ArrowLeft") {
+    if (!isModifiedKeyboardEvent(event) && event.key === "ArrowLeft") {
       event.preventDefault();
       navigateLastAssistantMessage("left");
     }
 
-    else if (!isModifiedKeyboardEvent() && event.key === "ArrowRight") {
+    else if (!isModifiedKeyboardEvent(event) && event.key === "ArrowRight") {
       event.preventDefault();
       if (!navigateLastAssistantMessage("right")) {
         // If can't navigate right (last version), regenerate
@@ -159,9 +157,8 @@ targetElement.addEventListener("scroll", function() {
   let diff = targetElement.scrollHeight - targetElement.clientHeight;
   let isAtBottomNow = Math.abs(targetElement.scrollTop - diff) <= 10 || diff <= 0;
 
-  // Add scrolling class to disable hover effects
   if (window.isScrolled || !isAtBottomNow) {
-    targetElement.classList.add("scrolling");
+    targetElement.classList.add("scrolling"); // Disables hover effects during scroll
   }
 
   if(isAtBottomNow) {
@@ -202,12 +199,8 @@ const observer = new MutationObserver(function() {
 });
 
 // Only watch for attribute changes on targetElement (e.g. _generating class)
-const config = {
-  attributes: true
-};
-
 // Start observing the target element
-observer.observe(targetElement, config);
+observer.observe(targetElement, { attributes: true });
 
 //------------------------------------------------
 // Handle syntax highlighting / LaTeX
@@ -228,7 +221,7 @@ window.doSyntaxHighlighting = function() {
   if (messageBodies.length > 0) {
     let hasSeenVisible = false;
 
-    // Go from last message to first
+    // Go from last message to first so we can early-exit once past visible area
     for (let i = messageBodies.length - 1; i >= 0; i--) {
       const messageBody = messageBodies[i];
 
@@ -243,8 +236,8 @@ window.doSyntaxHighlighting = function() {
           codeBlock.classList.add("pretty_scrollbar");
         });
 
-        // Only render math in visible elements
         const mathContainers = messageBody.querySelectorAll("p, span, li, td, th, h1, h2, h3, h4, h5, h6, blockquote, figcaption, caption, dd, dt");
+        // Only render math in individually visible containers (the outer check is on the message body)
         mathContainers.forEach(container => {
           if (isElementVisibleOnScreen(container)) {
             renderMathInElement(container, {
@@ -271,7 +264,7 @@ const doSyntaxHighlighting = window.doSyntaxHighlighting;
 // Add some scrollbars
 //------------------------------------------------
 const scrollbarElements = document.querySelectorAll(".add_scrollbar textarea, .add_scrollbar .drag-drop-list");
-for(i = 0; i < scrollbarElements.length; i++) {
+for(let i = 0; i < scrollbarElements.length; i++) {
   scrollbarElements[i].classList.remove("scroll-hide");
   scrollbarElements[i].classList.add("pretty_scrollbar");
   scrollbarElements[i].style.resize = "none";
@@ -298,13 +291,13 @@ if (toolsInfo) {
 // Remove some backgrounds
 //------------------------------------------------
 const noBackgroundelements = document.querySelectorAll(".no-background");
-for(i = 0; i < noBackgroundelements.length; i++) {
+for(let i = 0; i < noBackgroundelements.length; i++) {
   noBackgroundelements[i].parentNode.style.border = "none";
   noBackgroundelements[i].parentNode.parentNode.parentNode.style.alignItems = "center";
 }
 
 const slimDropdownElements = document.querySelectorAll(".slim-dropdown");
-for (i = 0; i < slimDropdownElements.length; i++) {
+for (let i = 0; i < slimDropdownElements.length; i++) {
   const parentNode = slimDropdownElements[i].parentNode;
   parentNode.style.background = "transparent";
   parentNode.style.border = "0";
@@ -374,49 +367,43 @@ button.addEventListener("click", function () {
   }
 });
 
-// Add event listener for mouseleave on the button
-button.addEventListener("mouseleave", function () {
-  // Delay to prevent menu hiding when the mouse leaves the button into the menu
+// Delay to prevent menu hiding when the mouse leaves the button or menu
+function delayedHideMenu() {
   setTimeout(function () {
     if (!isMouseOverButtonOrMenu()) {
       hideMenu();
     }
   }, 100);
-});
+}
 
+// Add event listener for mouseleave on the button
+button.addEventListener("mouseleave", delayedHideMenu);
 // Add event listener for mouseleave on the menu
-menu.addEventListener("mouseleave", function () {
-  // Delay to prevent menu hide when the mouse leaves the menu into the button
-  setTimeout(function () {
-    if (!isMouseOverButtonOrMenu()) {
-      hideMenu();
-    }
-  }, 100);
-});
+menu.addEventListener("mouseleave", delayedHideMenu);
 
 // Add event listener for click anywhere in the document
 document.addEventListener("click", function (event) {
-  const target = event.target;
-
   // Check if the click is outside the button/menu and the menu is visible
   if (!isMouseOverButtonOrMenu() && menu.style.display === "flex") {
     hideMenu();
   }
 
-  if (event.target.classList.contains("pfp_character")) {
+  const target = event.target;
+
+  if (target.classList.contains("pfp_character")) {
     toggleBigPicture();
   }
 
   // Handle sidebar clicks on mobile
   if (isMobile()) {
-  // Check if the click did NOT originate from any of the specified toggle buttons or elements
+    // Check if the click did NOT originate from any of the specified toggle buttons or elements
     if (
       target.closest("#navigation-toggle") !== navigationToggle &&
-    target.closest("#past-chats-toggle") !== pastChatsToggle &&
-    target.closest("#chat-controls-toggle") !== chatControlsToggle &&
-    target.closest(".header_bar") !== headerBar &&
-    target.closest("#past-chats-row") !== pastChatsRow &&
-    target.closest("#chat-controls") !== chatControlsRow
+      target.closest("#past-chats-toggle") !== pastChatsToggle &&
+      target.closest("#chat-controls-toggle") !== chatControlsToggle &&
+      target.closest(".header_bar") !== headerBar &&
+      target.closest("#past-chats-row") !== pastChatsRow &&
+      target.closest("#chat-controls") !== chatControlsRow
     ) {
       handleIndividualSidebarClose(event);
     }
@@ -433,27 +420,19 @@ document.getElementById("chat-input-row").classList.add("chat-input-positioned")
 //------------------------------------------------
 const chatTextArea = document.getElementById("chat-input").querySelector("textarea");
 
-function respondToChatInputVisibility(element, callback) {
-  var options = {
-    root: document.documentElement,
-  };
-
-  var observer = new IntersectionObserver((entries, observer) => {
+function focusOnVisible(element) {
+  var observer = new IntersectionObserver((entries) => {
     entries.forEach(entry => {
-      callback(entry.intersectionRatio > 0);
+      if (entry.intersectionRatio > 0) {
+        element.focus();
+      }
     });
-  }, options);
+  }, { root: document.documentElement });
 
   observer.observe(element);
 }
 
-function handleChatInputVisibilityChange(isVisible) {
-  if (isVisible) {
-    chatTextArea.focus();
-  }
-}
-
-respondToChatInputVisibility(chatTextArea, handleChatInputVisibilityChange);
+focusOnVisible(chatTextArea);
 
 //------------------------------------------------
 // Show enlarged character picture when the profile
@@ -463,8 +442,7 @@ let bigPictureVisible = false;
 
 function addBigPicture() {
   var imgElement = document.createElement("img");
-  var timestamp = new Date().getTime();
-  imgElement.src = "/file/user_data/cache/pfp_character.png?time=" + timestamp;
+  imgElement.src = getProfilePictureUrl();
   imgElement.classList.add("bigProfilePicture");
   imgElement.addEventListener("load", function () {
     this.style.visibility = "visible";
@@ -478,9 +456,8 @@ function addBigPicture() {
 }
 
 function deleteBigPicture() {
-  var bigProfilePictures = document.querySelectorAll(".bigProfilePicture");
-  bigProfilePictures.forEach(function (element) {
-    element.parentNode.removeChild(element);
+  document.querySelectorAll(".bigProfilePicture").forEach(function (element) {
+    element.remove();
   });
 }
 
@@ -494,44 +471,11 @@ function toggleBigPicture() {
   }
 }
 
-//------------------------------------------------
-// Handle the chat input box growth
-//------------------------------------------------
-
-// Cache DOM elements
-const chatContainer = document.getElementById("chat").parentNode.parentNode.parentNode;
-const chatInput = document.querySelector("#chat-input textarea");
-
-// Variables to store current dimensions
-let currentChatInputHeight = chatInput.clientHeight;
-
 //------------------------------------------------
 // Focus on the rename text area when it becomes visible
 //------------------------------------------------
 const renameTextArea = document.getElementById("rename-row").querySelector("textarea");
-
-function respondToRenameVisibility(element, callback) {
-  var options = {
-    root: document.documentElement,
-  };
-
-  var observer = new IntersectionObserver((entries, observer) => {
-    entries.forEach(entry => {
-      callback(entry.intersectionRatio > 0);
-    });
-  }, options);
-
-  observer.observe(element);
-}
-
-
-function handleVisibilityChange(isVisible) {
-  if (isVisible) {
-    renameTextArea.focus();
-  }
-}
-
-respondToRenameVisibility(renameTextArea, handleVisibilityChange);
+focusOnVisible(renameTextArea);
 
 //------------------------------------------------
 // Adjust the chat tab margin if no extension UI
@@ -737,21 +681,21 @@ function handleIndividualSidebarClose(event) {
 
   // Close navigation bar if click is outside and it is open
   if (!headerBar.contains(target) && !headerBar.classList.contains("sidebar-hidden")) {
-    toggleSidebar(headerBar, navigationToggle, true);
+    toggleSidebar(headerBar, navigationToggle);
   }
 
   // Close past chats row if click is outside and it is open
   if (!pastChatsRow.contains(target) && !pastChatsRow.classList.contains("sidebar-hidden")) {
-    toggleSidebar(pastChatsRow, pastChatsToggle, true);
+    toggleSidebar(pastChatsRow, pastChatsToggle);
   }
 
   // Close chat controls row if click is outside and it is open
   if (!chatControlsRow.contains(target) && !chatControlsRow.classList.contains("sidebar-hidden")) {
-    toggleSidebar(chatControlsRow, chatControlsToggle, true);
+    toggleSidebar(chatControlsRow, chatControlsToggle);
   }
 }
 
-function toggleSidebar(sidebar, toggle, forceClose = false) {
+function toggleSidebar(sidebar, toggle) {
   const isCurrentlyHidden = sidebar.classList.contains("sidebar-hidden");
   const shouldClose = !isCurrentlyHidden;
 
@@ -776,11 +720,6 @@ function toggleSidebar(sidebar, toggle, forceClose = false) {
     toggle.classList.toggle("chat-controls-open", !shouldClose);
     toggle.innerHTML = shouldClose ? leftArrowSVG : rightArrowSVG;
   }
-
-  // Mobile handling
-  if (isMobile()) {
-    sidebar.classList.toggle("sidebar-shown", !shouldClose);
-  }
 }
 
 // Function to check if the device is mobile
@@ -840,17 +779,17 @@ pastChatsToggle.addEventListener("click", () => {
   const isCurrentlyOpen = !pastChatsRow.classList.contains("sidebar-hidden");
   toggleSidebar(pastChatsRow, pastChatsToggle);
 
-  // On desktop, open/close both sidebars at the same time
+  // On desktop, sync both sidebars together
   if (!isMobile()) {
     if (isCurrentlyOpen) {
       // If we just closed the left sidebar, also close the right sidebar
       if (!chatControlsRow.classList.contains("sidebar-hidden")) {
-        toggleSidebar(chatControlsRow, chatControlsToggle, true);
+        toggleSidebar(chatControlsRow, chatControlsToggle);
       }
     } else {
       // If we just opened the left sidebar, also open the right sidebar
       if (chatControlsRow.classList.contains("sidebar-hidden")) {
-        toggleSidebar(chatControlsRow, chatControlsToggle, false);
+        toggleSidebar(chatControlsRow, chatControlsToggle);
       }
     }
   }
@@ -860,17 +799,17 @@ chatControlsToggle.addEventListener("click", () => {
   const isCurrentlyOpen = !chatControlsRow.classList.contains("sidebar-hidden");
   toggleSidebar(chatControlsRow, chatControlsToggle);
 
-  // On desktop, open/close both sidebars at the same time
+  // On desktop, sync both sidebars together
   if (!isMobile()) {
     if (isCurrentlyOpen) {
       // If we just closed the right sidebar, also close the left sidebar
       if (!pastChatsRow.classList.contains("sidebar-hidden")) {
-        toggleSidebar(pastChatsRow, pastChatsToggle, true);
+        toggleSidebar(pastChatsRow, pastChatsToggle);
       }
     } else {
       // If we just opened the right sidebar, also open the left sidebar
       if (pastChatsRow.classList.contains("sidebar-hidden")) {
-        toggleSidebar(pastChatsRow, pastChatsToggle, false);
+        toggleSidebar(pastChatsRow, pastChatsToggle);
       }
     }
   }
@@ -890,7 +829,7 @@ if (isMobile()) {
   const textarea = document.querySelector("#chat-input textarea");
 
   if (textarea) {
-    // Simulate adding and removing a newline
+    // Force textarea height recalculation by simulating content change
     textarea.value += "\n";
     textarea.dispatchEvent(new Event("input", { bubbles: true }));
     textarea.value = textarea.value.slice(0, -1);
diff --git a/js/save_files.js b/js/save_files.js
index bdb0e33421..c3cbf9ff4d 100644
--- a/js/save_files.js
+++ b/js/save_files.js
@@ -1,10 +1,9 @@
 // Functions for downloading JSON files
 function getCurrentTimestamp() {
   const now = new Date();
-  const timezoneOffset = now.getTimezoneOffset() * 60000; // Convert to milliseconds
+  const timezoneOffset = now.getTimezoneOffset() * 60000; // Convert minutes to milliseconds
   const localTime = new Date(now.getTime() - timezoneOffset);
-  const formattedTimestamp = localTime.toISOString().replace(/[-:]/g, "").slice(0, 15);
-  return formattedTimestamp;
+  return localTime.toISOString().replace(/[-:]/g, "").slice(0, 15);
 }
 
 function saveFile(contents, filename) {
@@ -18,23 +17,18 @@ function saveFile(contents, filename) {
 }
 
 function saveHistory(history, character, mode) {
-  let path = null;
+  let path;
 
   if (["chat", "chat-instruct"].includes(mode) && character && character.trim() !== "") {
     path = `history_${character}_${getCurrentTimestamp()}.json`;
   } else {
-    try {
-      path = `history_${mode}_${getCurrentTimestamp()}.json`;
-    } catch (error) {
-      path = `history_${getCurrentTimestamp()}.json`;
-    }
+    path = `history_${mode || "unknown"}_${getCurrentTimestamp()}.json`;
   }
+
   saveFile(history, path);
 }
 
 function saveSession(session) {
-  let path = null;
-
-  path = `session_${getCurrentTimestamp()}.json`;
+  const path = `session_${getCurrentTimestamp()}.json`;
   saveFile(session, path);
 }
diff --git a/js/show_controls.js b/js/show_controls.js
index ff513395ee..d5642dc489 100644
--- a/js/show_controls.js
+++ b/js/show_controls.js
@@ -1,13 +1,11 @@
-const chatParent = document.querySelector(".chat-parent");
-
 function toggle_controls(value) {
+  const navToggle = document.getElementById("navigation-toggle");
+  const pastChatsToggle = document.getElementById("past-chats-toggle");
   const extensions = document.querySelector("#extensions");
+  const galleryExtension = document.getElementById("gallery-extension");
 
   if (value) {
     // SHOW MODE: Click toggles to show hidden sidebars
-    const navToggle = document.getElementById("navigation-toggle");
-    const pastChatsToggle = document.getElementById("past-chats-toggle");
-
     if (navToggle && document.querySelector(".header_bar")?.classList.contains("sidebar-hidden")) {
       navToggle.click();
     }
@@ -19,17 +17,11 @@ function toggle_controls(value) {
     if (extensions) {
       extensions.style.display = "inherit";
     }
-
-    let gallery_element = document.getElementById("gallery-extension");
-    if (gallery_element) {
-      gallery_element.style.display = "block";
+    if (galleryExtension) {
+      galleryExtension.style.display = "block";
     }
-
   } else {
     // HIDE MODE: Click toggles to hide visible sidebars
-    const navToggle = document.getElementById("navigation-toggle");
-    const pastChatsToggle = document.getElementById("past-chats-toggle");
-
     if (navToggle && !document.querySelector(".header_bar")?.classList.contains("sidebar-hidden")) {
       navToggle.click();
     }
@@ -41,5 +33,8 @@ function toggle_controls(value) {
     if (extensions) {
       extensions.style.display = "none";
     }
+    if (galleryExtension) {
+      galleryExtension.style.display = "none";
+    }
   }
 }
diff --git a/js/switch_tabs.js b/js/switch_tabs.js
index 36e5736bad..a1b44ef36d 100644
--- a/js/switch_tabs.js
+++ b/js/switch_tabs.js
@@ -2,17 +2,9 @@ function scrollToTop() {
   window.scrollTo({ top: 0 });
 }
 
-function findButtonsByText(buttonText) {
-  const buttons = document.getElementsByTagName("button");
-  const matchingButtons = [];
-
-  for (let i = 0; i < buttons.length; i++) {
-    if (buttons[i].textContent.trim() === buttonText) {
-      matchingButtons.push(buttons[i]);
-    }
-  }
-
-  return matchingButtons;
+function findButtonsByText(buttonText, container = document) {
+  return Array.from(container.getElementsByTagName("button"))
+    .filter(btn => btn.textContent.trim() === buttonText);
 }
 
 function switch_to_chat() {
@@ -39,13 +31,9 @@ function switch_to_character() {
 
 function switch_to_image_ai_generate() {
   const container = document.querySelector("#image-ai-tab");
-  const buttons = container.getElementsByTagName("button");
-
-  for (let i = 0; i < buttons.length; i++) {
-    if (buttons[i].textContent.trim() === "Generate") {
-      buttons[i].click();
-      break;
-    }
+  const generateBtn = findButtonsByText("Generate", container)[0];
+  if (generateBtn) {
+    generateBtn.click();
   }
 
   scrollToTop();
diff --git a/js/update_big_picture.js b/js/update_big_picture.js
index ec51d63b07..8f638c99f9 100644
--- a/js/update_big_picture.js
+++ b/js/update_big_picture.js
@@ -1,7 +1,6 @@
 function updateBigPicture() {
   var existingElement = document.querySelector(".bigProfilePicture");
   if (existingElement) {
-    var timestamp = new Date().getTime();
-    existingElement.src = "/file/user_data/cache/pfp_character.png?time=" + timestamp;
+    existingElement.src = getProfilePictureUrl();
   }
 }
diff --git a/modules/extensions.py b/modules/extensions.py
index 09db9f4005..afe847f0d1 100644
--- a/modules/extensions.py
+++ b/modules/extensions.py
@@ -191,21 +191,19 @@ def _apply_custom_generate_reply():
 
 
 def _apply_custom_css():
-    all_css = ''
-    for extension, _ in iterator():
-        if hasattr(extension, 'custom_css'):
-            all_css += getattr(extension, 'custom_css')()
-
-    return all_css
+    return ''.join(
+        getattr(extension, 'custom_css')()
+        for extension, _ in iterator()
+        if hasattr(extension, 'custom_css')
+    )
 
 
 def _apply_custom_js():
-    all_js = ''
-    for extension, _ in iterator():
-        if hasattr(extension, 'custom_js'):
-            all_js += getattr(extension, 'custom_js')()
-
-    return all_js
+    return ''.join(
+        getattr(extension, 'custom_js')()
+        for extension, _ in iterator()
+        if hasattr(extension, 'custom_js')
+    )
 
 
 def create_extensions_block():

From 71c1a52afe54ab599ab5849ae80f1d5a3a72fb5a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 30 Mar 2026 20:49:38 -0700
Subject: [PATCH 1461/1701] API: Implement echo + logprobs for /v1/completions
 endpoint

---
 modules/api/completions.py  | 291 ++++++++++++++++++++++++++++++------
 modules/exllamav3.py        |  26 +++-
 modules/llama_cpp_server.py |  39 ++++-
 3 files changed, 305 insertions(+), 51 deletions(-)

diff --git a/modules/api/completions.py b/modules/api/completions.py
index 8948bb86bc..587ad6eadb 100644
--- a/modules/api/completions.py
+++ b/modules/api/completions.py
@@ -39,6 +39,129 @@ def load_chat_template_file(filepath):
     return text
 
 
+def _first_token_display_str(token_id, prompt, tokenizer):
+    """Return the display string for the first prompt token.
+
+    Returns empty string for BOS or tokens that don't appear at the start
+    of the prompt text, so they don't shift text_offset for subsequent tokens.
+    """
+    token_id = int(token_id)
+    bos_id = getattr(tokenizer, 'bos_token_id', None)
+    if bos_id is not None and token_id == bos_id:
+        return ""
+
+    import torch
+    tok = tokenizer.decode(torch.tensor([token_id]))
+    if not prompt.startswith(tok):
+        return ""
+
+    return tok
+
+
+def _compute_prompt_logprob_entries(prompt, logprobs_count, input_ids=None):
+    """Compute logprob entries for prompt tokens via a forward pass.
+
+    Returns a list of logprob entries in the standard format.
+    The first token gets a null entry (no conditioning context).
+
+    Supported for HF-compatible loaders (Transformers, ExLlamav3_HF, etc.)
+    via a single forward pass, and for llama.cpp via the server's
+    prompt_logprobs parameter. Returns [] for unsupported loaders.
+    """
+    if input_ids is None:
+        input_ids = encode(prompt)  # (1, seq_len) tensor or array
+
+    token_ids = input_ids[0]
+    n_tokens = len(token_ids)
+
+    if n_tokens == 0:
+        return []
+
+    loader = shared.args.loader
+    model = shared.model
+
+    if loader == 'llama.cpp':
+        return model.get_prompt_logprob_entries(token_ids, max(logprobs_count, 1), prompt=prompt)
+
+    first_token_str = _first_token_display_str(token_ids[0], prompt, shared.tokenizer)
+
+    if n_tokens <= 1:
+        return [{"token": first_token_str, "null_logprob": True}]
+
+    import torch
+
+    if loader == 'ExLlamav3' and hasattr(model, 'model') and hasattr(model, 'cache'):
+        # Native ExLlamav3: call the underlying Model.forward() directly
+        input_ids_tensor = input_ids if isinstance(input_ids, torch.Tensor) else torch.tensor(input_ids, dtype=torch.long)
+        with torch.no_grad():
+            logits = model.model.forward(
+                input_ids=input_ids_tensor,
+                params={
+                    "attn_mode": "flash_attn",
+                    "cache": model.cache,
+                    "past_len": 0,
+                    "batch_shape": (1, model.max_tokens),
+                }
+            ).float().cpu()
+
+    elif hasattr(model, 'forward'):
+        # HF-compatible loaders (Transformers, ExLlamav3_HF, etc.)
+        input_ids_tensor = input_ids if isinstance(input_ids, torch.Tensor) else torch.tensor(input_ids, dtype=torch.long)
+        if hasattr(model, 'device'):
+            input_ids_tensor = input_ids_tensor.to(model.device)
+        with torch.no_grad():
+            # Pass labels to ensure logits are returned for ALL positions,
+            # not just the last token (some HF wrappers like ExLlamav3_HF
+            # only compute the last-token logits when labels are absent).
+            outputs = model(input_ids=input_ids_tensor, labels=input_ids_tensor)
+            logits = outputs.logits.float().cpu()
+
+    else:
+        return []
+
+    entries = [{"token": first_token_str, "null_logprob": True}]
+
+    # Batch logsumexp and topk as single operations across all positions
+    # to avoid per-position kernel launch overhead.
+    prompt_logits = logits[0, :n_tokens - 1]  # positions 0..n-2 predict tokens 1..n-1
+    k = min(logprobs_count, prompt_logits.shape[-1])
+    all_top_values, all_top_indices = torch.topk(prompt_logits, k=k, dim=-1)
+    all_lse = torch.logsumexp(prompt_logits, dim=-1)
+    all_top_log_probs = all_top_values - all_lse.unsqueeze(-1)
+
+    # Batch-decode all unique token IDs to avoid O(N*k) individual decode calls
+    unique_ids = set(int(tid) for tid in token_ids[1:])
+    unique_ids.update(int(tid) for tid in all_top_indices.flatten().tolist())
+
+    decoded_strs = {tid: shared.tokenizer.decode(torch.tensor([tid])) for tid in unique_ids}
+
+    for i in range(1, n_tokens):
+        token_id = int(token_ids[i])
+        idx = i - 1
+        top_log_probs = all_top_log_probs[idx]
+        top_ids = all_top_indices[idx].tolist()
+        actual_token_str = decoded_strs[token_id]
+
+        # Build the top list with the actual prompt token guaranteed at front
+        if token_id in top_ids:
+            actual_lp = top_log_probs[top_ids.index(token_id)].item()
+            alternatives = [
+                {"token": decoded_strs[top_ids[j]], "logprob": top_log_probs[j].item()}
+                for j in range(k) if top_ids[j] != token_id
+            ]
+        else:
+            actual_lp = (prompt_logits[idx, token_id] - all_lse[idx]).item()
+            alternatives = [
+                {"token": decoded_strs[top_ids[j]], "logprob": top_log_probs[j].item()}
+                for j in range(k - 1)  # drop lowest to make room
+            ]
+
+        entry = {"top_logprobs": [{"token": actual_token_str, "logprob": actual_lp}] + alternatives}
+        entries.append(entry)
+
+    return entries
+
+
 def _get_raw_logprob_entries(offset=0):
     """Get raw logprob entries from llama.cpp/ExLlamav3 backend, starting from offset.
 
@@ -65,6 +188,21 @@ def _parse_entry_top(entry):
     return entry.get('top_logprobs', entry.get('top_probs', []))
 
 
+def _extract_sampled_token(entry, top):
+    """Get the actually sampled token and its logprob from a logprob entry.
+
+    Uses the entry-level token/logprob when available (the actually sampled
+    token), falling back to top[0] (highest-probability alternative) which
+    may differ with non-greedy sampling.
+    """
+    if 'token' in entry:
+        return entry['token'], entry.get('logprob', entry.get('prob', 0))
+
+    token_str = top[0].get('token', '')
+    token_logprob = top[0].get('logprob', top[0].get('prob', 0))
+    return token_str, token_logprob
+
+
 def format_chat_logprobs(entries):
     """Format logprob entries into OpenAI chat completions logprobs format.
 
@@ -79,9 +217,7 @@ def format_chat_logprobs(entries):
         if not top:
             continue
 
-        chosen = top[0]
-        token_str = chosen.get('token', '')
-        token_logprob = chosen.get('logprob', chosen.get('prob', 0))
+        token_str, token_logprob = _extract_sampled_token(entry, top)
 
         top_list = []
         for item in top:
@@ -118,13 +254,21 @@ def format_completion_logprobs(entries):
     offset = 0
 
     for entry in entries:
+        # Handle null logprob entries (first prompt token with echo)
+        if entry.get("null_logprob"):
+            token_str = entry.get("token", "")
+            tokens.append(token_str)
+            token_logprobs.append(None)
+            top_logprobs.append(None)
+            text_offset.append(offset)
+            offset += len(token_str)
+            continue
+
         top = _parse_entry_top(entry)
         if not top:
             continue
 
-        chosen = top[0]
-        token_str = chosen.get('token', '')
-        token_logprob = chosen.get('logprob', chosen.get('prob', 0))
+        token_str, token_logprob = _extract_sampled_token(entry, top)
 
         tokens.append(token_str)
         token_logprobs.append(token_logprob)
@@ -407,7 +551,10 @@ def chat_completions_common(body: dict, is_legacy: bool = False, stream=False, p
     })
 
     max_tokens = generate_params['max_new_tokens']
-    if max_tokens in [None, 0]:
+    if max_tokens is not None and max_tokens <= 0:
+        raise InvalidRequestError(message="max_tokens must be greater than 0.", param="max_tokens")
+
+    if max_tokens is None:
         generate_params['max_new_tokens'] = 512
         generate_params['auto_max_new_tokens'] = True
 
@@ -652,6 +799,15 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_e
     # common params
     generate_params = process_parameters(body, is_legacy=is_legacy)
     max_tokens = generate_params['max_new_tokens']
+    if max_tokens is None:
+        generate_params['max_new_tokens'] = 512
+        generate_params['auto_max_new_tokens'] = True
+        max_tokens = 512
+    elif max_tokens < 0:
+        raise InvalidRequestError(message="max_tokens must be greater than or equal to 0.", param="max_tokens")
+    elif max_tokens == 0 and body.get('logprobs') is None:
+        raise InvalidRequestError(message="max_tokens is 0 but no logprobs parameter was specified.", param="max_tokens")
+
     generate_params['stream'] = stream
     if stop_event is not None:
         generate_params['stop_event'] = stop_event
@@ -700,9 +856,17 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_e
                         prompt = decode(prompt)[0]
 
             prefix = prompt if echo else ''
-            token_count = len(encode(prompt)[0])
+            prompt_input_ids = encode(prompt)
+            token_count = len(prompt_input_ids[0])
             total_prompt_token_count += token_count
 
+            # Compute prompt logprobs once per prompt (shared across n_completions)
+            logprobs_val = body.get('logprobs', None)
+            if echo and logprobs_val is not None and logprobs_val >= 0:
+                prompt_entries = _compute_prompt_logprob_entries(prompt, logprobs_val, input_ids=prompt_input_ids)
+            else:
+                prompt_entries = None
+
             original_seed = generate_params.get('seed', -1)
             for _n in range(n_completions):
                 # Increment seed for each completion to ensure diversity (matches llama.cpp native behavior)
@@ -713,29 +877,41 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_e
                     logprob_proc.token_alternatives_history.clear()
 
                 # generate reply #######################################
-                debug_msg({'prompt': prompt, 'generate_params': generate_params})
-                generator = generate_reply(prompt, generate_params, is_chat=False)
-                answer = ''
+                if max_tokens == 0:
+                    answer = ''
+                    completion_token_count = 0
+                    stop_reason = "stop"
+                else:
+                    debug_msg({'prompt': prompt, 'generate_params': generate_params})
+                    generator = generate_reply(prompt, generate_params, is_chat=False)
+                    answer = ''
+
+                    for a in generator:
+                        answer = a
 
-                for a in generator:
-                    answer = a
+                    completion_token_count = len(encode(answer)[0])
+                    stop_reason = "stop"
+                    if token_count + completion_token_count >= generate_params['truncation_length'] or completion_token_count >= max_tokens:
+                        stop_reason = "length"
 
-                completion_token_count = len(encode(answer)[0])
                 total_completion_token_count += completion_token_count
-                stop_reason = "stop"
-                if token_count + completion_token_count >= generate_params['truncation_length'] or completion_token_count >= max_tokens:
-                    stop_reason = "length"
 
-                if logprob_proc:
+                if max_tokens == 0:
                     all_entries = []
-                    for alt in logprob_proc.token_alternatives_history:
-                        all_entries.extend(_dict_to_logprob_entries(alt))
-                    completion_logprobs = format_completion_logprobs(all_entries)
-                elif shared.args.loader in ('llama.cpp', 'ExLlamav3'):
-                    raw = getattr(shared.model, 'last_completion_probabilities', None)
-                    completion_logprobs = format_completion_logprobs(raw)
                 else:
-                    completion_logprobs = None
+                    if logprob_proc:
+                        all_entries = []
+                        for alt in logprob_proc.token_alternatives_history:
+                            all_entries.extend(_dict_to_logprob_entries(alt))
+                    elif shared.args.loader in ('llama.cpp', 'ExLlamav3'):
+                        all_entries = getattr(shared.model, 'last_completion_probabilities', None) or []
+                    else:
+                        all_entries = []
+
+                if prompt_entries:
+                    all_entries = prompt_entries + all_entries
+
+                completion_logprobs = format_completion_logprobs(all_entries) if all_entries else None
 
                 respi = {
                     "index": choice_index,
@@ -775,7 +951,8 @@ def completions_common(body: dict, is_legacy: bool = False, stream=False, stop_e
                 raise InvalidRequestError(message="API Batched generation not yet supported.", param=prompt_str)
 
         prefix = prompt if echo else ''
-        token_count = len(encode(prompt)[0])
+        prompt_input_ids = encode(prompt)
+        token_count = len(prompt_input_ids[0])
 
         # Check if usage should be included in streaming chunks per OpenAI spec
         stream_options = body.get('stream_options')
@@ -808,37 +985,57 @@ def text_streaming_chunk(content):
 
             return chunk
 
+        logprobs_val = body.get('logprobs', None)
+        if echo and logprobs_val is not None and logprobs_val >= 0:
+            prompt_entries = _compute_prompt_logprob_entries(prompt, logprobs_val, input_ids=prompt_input_ids)
+            prompt_logprobs_formatted = format_completion_logprobs(prompt_entries) if prompt_entries else None
+        else:
+            prompt_logprobs_formatted = None
+
+        # Clear stale logprobs from any previous request before building the
+        # first chunk, so text_streaming_chunk doesn't pick up old data.
+        if hasattr(shared.model, 'last_completion_probabilities'):
+            shared.model.last_completion_probabilities = []
+        cmpl_logprobs_offset[0] = 0
+
         chunk = text_streaming_chunk(prefix)
+        if prompt_logprobs_formatted is not None:
+            chunk[resp_list][0]["logprobs"] = prompt_logprobs_formatted
         if include_usage:
             chunk['usage'] = None
         yield chunk
 
         # generate reply #######################################
-        debug_msg({'prompt': prompt, 'generate_params': generate_params})
-        generator = generate_reply(prompt, generate_params, is_chat=False)
-        answer = ''
-        seen_content = ''
-        completion_token_count = 0
-
-        for a in generator:
-            answer = a
+        if max_tokens == 0:
+            answer = ''
+            completion_token_count = 0
+            stop_reason = "stop"
+        else:
+            debug_msg({'prompt': prompt, 'generate_params': generate_params})
+            generator = generate_reply(prompt, generate_params, is_chat=False)
+            answer = ''
+            seen_content = ''
+            completion_token_count = 0
 
-            len_seen = len(seen_content)
-            new_content = answer[len_seen:]
+            for a in generator:
+                answer = a
 
-            if not new_content or chr(0xfffd) in new_content:  # partial unicode character, don't send it yet.
-                continue
+                len_seen = len(seen_content)
+                new_content = answer[len_seen:]
 
-            seen_content = answer
-            chunk = text_streaming_chunk(new_content)
-            if include_usage:
-                chunk['usage'] = None
-            yield chunk
+                if not new_content or chr(0xfffd) in new_content:  # partial unicode character, don't send it yet.
+                    continue
 
-        completion_token_count = len(encode(answer)[0])
-        stop_reason = "stop"
-        if token_count + completion_token_count >= generate_params['truncation_length'] or completion_token_count >= max_tokens:
-            stop_reason = "length"
+                seen_content = answer
+                chunk = text_streaming_chunk(new_content)
+                if include_usage:
+                    chunk['usage'] = None
+                yield chunk
+
+            completion_token_count = len(encode(answer)[0])
+            stop_reason = "stop"
+            if token_count + completion_token_count >= generate_params['truncation_length'] or completion_token_count >= max_tokens:
+                stop_reason = "length"
 
         chunk = text_streaming_chunk(suffix)
         chunk[resp_list][0]["finish_reason"] = stop_reason
diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index f873503a8b..3782a693f1 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -489,15 +489,35 @@ def _capture_logprobs(self, result):
             return
 
         id_to_piece = self.tokenizer.get_id_to_piece_list(True)
+        sampled_ids = result.get("token_ids")    # (batch, seq_len) - actually sampled tokens
+        sampled_probs = result.get("token_probs")  # (batch, seq_len) - their probabilities
+
+        def _piece(tid):
+            s = id_to_piece[tid] if tid < len(id_to_piece) else f"<{tid}>"
+            return s.replace('\u2581', ' ')
+
+        def _logprob(prob):
+            return math.log(prob) if prob > 0 else float("-inf")
+
         # top_k_tokens shape: (batch, seq_len, k), top_k_probs same
         for seq_idx in range(top_k_tokens.shape[1]):
             entry = {"top_logprobs": []}
             for k_idx in range(top_k_tokens.shape[2]):
                 token_id = top_k_tokens[0, seq_idx, k_idx].item()
                 prob = top_k_probs[0, seq_idx, k_idx].item()
-                token_str = id_to_piece[token_id] if token_id < len(id_to_piece) else f"<{token_id}>"
-                logprob = math.log(prob) if prob > 0 else float("-inf")
-                entry["top_logprobs"].append({"token": token_str, "logprob": logprob})
+                entry["top_logprobs"].append({"token": _piece(token_id), "logprob": _logprob(prob)})
+
+            # Record the actually sampled token at the entry level so
+            # format_completion_logprobs uses it instead of top_logprobs[0]
+            # (they differ with non-greedy sampling).
+            if sampled_ids is not None:
+                sid = sampled_ids[0, seq_idx].item()
+                entry["token"] = _piece(sid)
+                if sampled_probs is not None:
+                    entry["logprob"] = _logprob(sampled_probs[0, seq_idx].item())
+                else:
+                    entry["logprob"] = None
+
             self.last_completion_probabilities.append(entry)
 
     def generate(self, prompt, state):
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index fa968be13a..3408046609 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -310,8 +310,45 @@ def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
         else:
             raise Exception(f"Unexpected response format: 'completion_probabilities' not found in {result}")
 
+    def get_prompt_logprob_entries(self, token_ids, n_probs=5, prompt=""):
+        """Get logprob entries for prompt tokens via a single n_predict=0 request.
+
+        Requires llama.cpp server with prompt_logprobs support.
+        Returns entries in the standard format for format_completion_logprobs().
+        """
+        token_ids_list = token_ids.tolist() if hasattr(token_ids, 'tolist') else list(token_ids)
+
+        url = f"http://127.0.0.1:{self.port}/completion"
+        payload = {
+            "prompt": token_ids_list,
+            "n_predict": 0,
+            "n_probs": n_probs,
+            "prompt_logprobs": True,
+            "stream": False,
+            "cache_prompt": False,
+        }
+
+        response = self.session.post(url, json=payload)
+        result = response.json()
+
+        prompt_probs = result.get("prompt_probabilities", [])
+        if not prompt_probs:
+            return []
+
+        # Null first token (no conditioning context); use empty string for BOS
+        # or tokens that don't appear at the start of the prompt text.
+        first_token_str = self.decode([token_ids_list[0]])
+        if self.bos_token and first_token_str == self.bos_token:
+            first_token_str = ""
+        elif not prompt.startswith(first_token_str):
+            first_token_str = ""
+
+        entries = [{"token": first_token_str, "null_logprob": True}]
+        entries.extend(prompt_probs)
+        return entries
+
     def _get_vocabulary_size(self):
-        """Get and store the model's maximum context length."""
+        """Get and store the model's vocabulary size."""
         url = f"http://127.0.0.1:{self.port}/v1/models"
         response = self.session.get(url).json()
 

From 328534b762f22c82b09babf6b04e289eab4a7fde Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 1 Apr 2026 12:51:07 -0700
Subject: [PATCH 1462/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 8 ++++----
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 8 ++++----
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 12 files changed, 26 insertions(+), 26 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 6e11dd2fa0..57991c9a75 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,10 +40,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/ik_llama_cpp_binaries-0.99.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/ik_llama_cpp_binaries-0.99.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.26/exllamav3-0.0.26+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.26/exllamav3-0.0.26+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index c964eff69f..bb47ea4bcd 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index b1dd6a4fdc..5750b10931 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/ik_llama_cpp_binaries-0.99.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 4d03d28007..d8302d3db3 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/ik_llama_cpp_binaries-0.99.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 9d41d069c2..d3a5c00891 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,7 +37,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/ik_llama_cpp_binaries-0.99.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/ik_llama_cpp_binaries-0.99.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index ff80b6c836..1180b42d1b 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 318044da94..57aa62628b 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 1676bffbcd..894c919931 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 27fc2da81b..32b9727f6a 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 0bbdd30a0d..73b7283282 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index c3ae3c57dd..ad96bbe2ee 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index e646c04c93..a5df3ad4e5 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.99.0/llama_cpp_binaries-0.99.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 4073164be0b305d8ac4a01d4259448370d009a99 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 1 Apr 2026 19:08:37 -0700
Subject: [PATCH 1463/1701] Fix ExLlamav3 OOM on prompt logprobs and
 qwen3_5_moe HF compat

---
 modules/api/completions.py | 13 +++++--------
 modules/exllamav3.py       | 33 ++++-----------------------------
 modules/exllamav3_hf.py    | 32 ++++++++------------------------
 3 files changed, 17 insertions(+), 61 deletions(-)

diff --git a/modules/api/completions.py b/modules/api/completions.py
index 587ad6eadb..a15e1f861d 100644
--- a/modules/api/completions.py
+++ b/modules/api/completions.py
@@ -91,17 +91,14 @@ def _compute_prompt_logprob_entries(prompt, logprobs_count, input_ids=None):
     import torch
 
     if loader == 'ExLlamav3' and hasattr(model, 'model') and hasattr(model, 'cache'):
-        # Native ExLlamav3: call the underlying Model.forward() directly
+        # Native ExLlamav3: call the underlying Model.forward() in chunks
+        # to avoid OOM from giant logits tensors (seq_len * vocab_size * 4 bytes)
         input_ids_tensor = input_ids if isinstance(input_ids, torch.Tensor) else torch.tensor(input_ids, dtype=torch.long)
+        input_ids_tensor = input_ids_tensor.view(-1).cpu()
         with torch.no_grad():
             logits = model.model.forward(
-                input_ids=input_ids_tensor,
-                params={
-                    "attn_mode": "flash_attn",
-                    "cache": model.cache,
-                    "past_len": 0,
-                    "batch_shape": (1, model.max_tokens),
-                }
+                input_ids=input_ids_tensor.view(1, -1),
+                params={"attn_mode": "flash_attn_nc"}
             ).float().cpu()
 
     elif hasattr(model, 'forward'):
diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 3782a693f1..7556a908eb 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -530,39 +530,14 @@ def generate(self, prompt, state):
     def get_logits(self, token_ids, **kwargs):
         """
         Process a batch of token_ids and return the logits for the last token.
-        This will reset and overwrite the model's cache.
+        Uses flash_attn_nc (no cache) for correct results with recurrent models.
         """
-        # Initialize a single params dictionary that will be updated in-place
-        params = {
-            "cache": self.cache,
-            "reconstruct": False,
-            "attn_mode": "flash_attn",
-            "batch_shape": (1, self.max_tokens),
-            "past_len": 0
-        }
-        params.update(kwargs)
-
-        # Process prefix tokens to fill the cache and generate recurrent state
-        if token_ids.shape[-1] > 1:
-            prefix_ids = token_ids[:, :-1]
-
-            # This forward call updates the 'params' dict with the recurrent state
-            self.model.forward(
-                input_ids=prefix_ids,
-                params=params
-            )
-
-            # Update past_len for the next call
-            params["past_len"] = prefix_ids.shape[-1]
-
-        # Process the last token, now using the state-filled 'params' dict
-        last_token_ids = token_ids[:, -1:]
         logits = self.model.forward(
-            input_ids=last_token_ids,
-            params=params
+            input_ids=token_ids,
+            params={"attn_mode": "flash_attn_nc"}
         )
 
-        return logits.float().cpu()
+        return logits[:, -1:, :].float().cpu()
 
     def encode(self, string, **kwargs):
         add_bos = kwargs.pop('add_bos', True)
diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index e0ad50022d..5e634e2250 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -26,6 +26,9 @@
 class Exllamav3HF(PreTrainedModel, GenerationMixin):
     def __init__(self, model_dir):
         hf_config = PretrainedConfig.from_pretrained(model_dir)
+        # Ensure text_config is a proper object, not a dict (fixes qwen3_5_moe + transformers compat)
+        if isinstance(getattr(hf_config, 'text_config', None), dict):
+            hf_config.text_config = PretrainedConfig(**hf_config.text_config)
         super().__init__(hf_config)
 
         exl3_config = Config.from_directory(model_dir)
@@ -199,30 +202,11 @@ def __call__(self, *args, **kwargs):
                 }
             ).to(input_ids.device).float()
         else:
-            # Labels path: use cache for cross-chunk attention.
-            tokens_to_process = seq_tensor
-            all_logits = None
-            current_len = 0
-
-            for i in range(0, tokens_to_process.shape[0], max_chunk_size):
-                chunk = tokens_to_process[i:i + max_chunk_size]
-                chunk_logits = self.ex_model.forward(
-                    input_ids=chunk.view(1, -1),
-                    params={
-                        "attn_mode": "flash_attn",
-                        "cache": ex_cache,
-                        "past_len": current_len,
-                        "batch_shape": (1, self.max_tokens),
-                    }
-                ).float()
-                current_len += chunk.shape[0]
-
-                if all_logits is None:
-                    all_logits = chunk_logits
-                else:
-                    all_logits = torch.cat([all_logits, chunk_logits], dim=1)
-
-            logits = all_logits
+            # Labels path: single pass without cache for correct logits
+            logits = self.ex_model.forward(
+                input_ids=seq_tensor.view(1, -1),
+                params={"attn_mode": "flash_attn_nc"}
+            ).float().cpu()
 
         if is_negative:
             self.past_seq_negative = seq_tensor

From a32ce254f275efe473d6624995957b3b6bd51aa1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 1 Apr 2026 20:28:44 -0700
Subject: [PATCH 1464/1701] Don't pass torch_dtype to transformers, autodetect
 from model config

---
 modules/transformers_loader.py | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/modules/transformers_loader.py b/modules/transformers_loader.py
index 7f521b8c22..5964f0124c 100644
--- a/modules/transformers_loader.py
+++ b/modules/transformers_loader.py
@@ -109,7 +109,6 @@ def load_model_HF(model_name):
     params = {
         'low_cpu_mem_usage': True,
         'attn_implementation': shared.args.attn_implementation,
-        'torch_dtype': torch.bfloat16 if shared.args.bf16 else torch.float16,
     }
 
     if shared.original_args.trust_remote_code:
@@ -120,6 +119,17 @@ def load_model_HF(model_name):
 
     config = AutoConfig.from_pretrained(path_to_model, trust_remote_code=shared.original_args.trust_remote_code)
 
+    # Determine torch_dtype: respect --bf16 flag, otherwise autodetect
+    # from model config, but never allow float32.
+    if shared.args.bf16:
+        params['torch_dtype'] = torch.bfloat16
+    else:
+        dtype = getattr(config, 'torch_dtype', None) or getattr(getattr(config, 'text_config', None), 'torch_dtype', None)
+        if dtype in (torch.float16, torch.bfloat16):
+            params['torch_dtype'] = dtype
+        else:
+            params['torch_dtype'] = torch.float16
+
     if 'chatglm' in model_name.lower():
         LoaderClass = AutoModel
     else:

From c10c6e87ae0b0085b36e7e13269461744ce04ff6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 07:17:27 -0700
Subject: [PATCH 1465/1701] API: Add token ids to logprobs output

---
 modules/api/completions.py | 19 ++++++++++++++-----
 1 file changed, 14 insertions(+), 5 deletions(-)

diff --git a/modules/api/completions.py b/modules/api/completions.py
index a15e1f861d..453fa07b91 100644
--- a/modules/api/completions.py
+++ b/modules/api/completions.py
@@ -143,17 +143,17 @@ def _compute_prompt_logprob_entries(prompt, logprobs_count, input_ids=None):
         if token_id in top_ids:
             actual_lp = top_log_probs[top_ids.index(token_id)].item()
             alternatives = [
-                {"token": decoded_strs[top_ids[j]], "logprob": top_log_probs[j].item()}
+                {"token": decoded_strs[top_ids[j]], "token_id": top_ids[j], "logprob": top_log_probs[j].item()}
                 for j in range(k) if top_ids[j] != token_id
             ]
         else:
             actual_lp = (prompt_logits[idx, token_id] - all_lse[idx]).item()
             alternatives = [
-                {"token": decoded_strs[top_ids[j]], "logprob": top_log_probs[j].item()}
+                {"token": decoded_strs[top_ids[j]], "token_id": top_ids[j], "logprob": top_log_probs[j].item()}
                 for j in range(k - 1)  # drop lowest to make room
             ]
 
-        entry = {"top_logprobs": [{"token": actual_token_str, "logprob": actual_lp}] + alternatives}
+        entry = {"top_logprobs": [{"token": actual_token_str, "token_id": token_id, "logprob": actual_lp}] + alternatives}
         entries.append(entry)
 
     return entries
@@ -239,7 +239,7 @@ def format_chat_logprobs(entries):
 def format_completion_logprobs(entries):
     """Format logprob entries into OpenAI completions logprobs format.
 
-    Output: {"tokens", "token_logprobs", "top_logprobs": [{token: prob}], "text_offset"}
+    Output: {"tokens", "token_logprobs", "top_logprobs": [{token: prob}], "top_logprobs_ids": [{token_id: prob}], "text_offset"}
     """
     if not entries:
         return None
@@ -247,6 +247,7 @@ def format_completion_logprobs(entries):
     tokens = []
     token_logprobs = []
     top_logprobs = []
+    top_logprobs_ids = []
     text_offset = []
     offset = 0
 
@@ -257,6 +258,7 @@ def format_completion_logprobs(entries):
             tokens.append(token_str)
             token_logprobs.append(None)
             top_logprobs.append(None)
+            top_logprobs_ids.append(None)
             text_offset.append(offset)
             offset += len(token_str)
             continue
@@ -273,21 +275,28 @@ def format_completion_logprobs(entries):
         offset += len(token_str)
 
         top_dict = {}
+        top_dict_ids = {}
         for item in top:
             t = item.get('token', '')
             lp = item.get('logprob', item.get('prob', 0))
             top_dict[t] = lp
+            if 'token_id' in item:
+                top_dict_ids[item['token_id']] = lp
         top_logprobs.append(top_dict)
+        top_logprobs_ids.append(top_dict_ids if top_dict_ids else None)
 
     if not tokens:
         return None
 
-    return {
+    result = {
         "tokens": tokens,
         "token_logprobs": token_logprobs,
         "top_logprobs": top_logprobs,
         "text_offset": text_offset
     }
+    if any(x is not None for x in top_logprobs_ids):
+        result["top_logprobs_ids"] = top_logprobs_ids
+    return result
 
 
 def process_parameters(body, is_legacy=False):

From ea1f8c71f2e92dc9ae230b943c605e43ff5c633c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 14:30:59 -0300
Subject: [PATCH 1466/1701] API: Optimize prompt logprobs and refactor
 ExLlamav3 forward pass

---
 modules/api/completions.py | 69 ++++++++++++++++++++++++--------------
 modules/exllamav3.py       | 14 ++++++++
 2 files changed, 58 insertions(+), 25 deletions(-)

diff --git a/modules/api/completions.py b/modules/api/completions.py
index 453fa07b91..4eb8fdad4f 100644
--- a/modules/api/completions.py
+++ b/modules/api/completions.py
@@ -90,16 +90,8 @@ def _compute_prompt_logprob_entries(prompt, logprobs_count, input_ids=None):
 
     import torch
 
-    if loader == 'ExLlamav3' and hasattr(model, 'model') and hasattr(model, 'cache'):
-        # Native ExLlamav3: call the underlying Model.forward() in chunks
-        # to avoid OOM from giant logits tensors (seq_len * vocab_size * 4 bytes)
-        input_ids_tensor = input_ids if isinstance(input_ids, torch.Tensor) else torch.tensor(input_ids, dtype=torch.long)
-        input_ids_tensor = input_ids_tensor.view(-1).cpu()
-        with torch.no_grad():
-            logits = model.model.forward(
-                input_ids=input_ids_tensor.view(1, -1),
-                params={"attn_mode": "flash_attn_nc"}
-            ).float().cpu()
+    if hasattr(model, 'get_prompt_logits'):
+        logits = model.get_prompt_logits(input_ids)
 
     elif hasattr(model, 'forward'):
         # HF-compatible loaders (Transformers, ExLlamav3_HF, etc.)
@@ -111,26 +103,54 @@ def _compute_prompt_logprob_entries(prompt, logprobs_count, input_ids=None):
             # not just the last token (some HF wrappers like ExLlamav3_HF
             # only compute the last-token logits when labels are absent).
             outputs = model(input_ids=input_ids_tensor, labels=input_ids_tensor)
-            logits = outputs.logits.float().cpu()
+            logits = outputs.logits  # keep on GPU, (1, seq_len, vocab) in model dtype
+            del outputs
 
     else:
         return []
 
     entries = [{"token": first_token_str, "null_logprob": True}]
 
-    # Batch logsumexp and topk as single operations across all positions
-    # to avoid per-position kernel launch overhead.
-    prompt_logits = logits[0, :n_tokens - 1]  # positions 0..n-2 predict tokens 1..n-1
-    k = min(logprobs_count, prompt_logits.shape[-1])
-    all_top_values, all_top_indices = torch.topk(prompt_logits, k=k, dim=-1)
-    all_lse = torch.logsumexp(prompt_logits, dim=-1)
-    all_top_log_probs = all_top_values - all_lse.unsqueeze(-1)
-
-    # Batch-decode all unique token IDs to avoid O(N*k) individual decode calls
+    logprobs_count = max(logprobs_count, 1)
+    k = min(logprobs_count, logits.shape[-1])
+    chunk_size = 2048
     unique_ids = set(int(tid) for tid in token_ids[1:])
-    unique_ids.update(int(tid) for tid in all_top_indices.flatten().tolist())
 
-    decoded_strs = {tid: shared.tokenizer.decode(torch.tensor([tid])) for tid in unique_ids}
+    # Process logits in chunks on GPU, only move top-K results to CPU
+    all_top_log_probs_list = []
+    all_top_indices_list = []
+    all_actual_lps = []
+
+    for start in range(0, n_tokens - 1, chunk_size):
+        end = min(start + chunk_size, n_tokens - 1)
+        chunk_logits = logits[0, start:end].float()  # (chunk, vocab) on GPU
+        chunk_lse = torch.logsumexp(chunk_logits, dim=-1)
+        chunk_top_values, chunk_top_indices = torch.topk(chunk_logits, k=k, dim=-1)
+        chunk_top_log_probs = chunk_top_values - chunk_lse.unsqueeze(-1)
+
+        # Compute logprob for actual next tokens in this chunk
+        chunk_top_sets = [set(chunk_top_indices[j].tolist()) for j in range(end - start)]
+        for j in range(end - start):
+            actual_tid = int(token_ids[start + j + 1])
+            if actual_tid not in chunk_top_sets[j]:
+                all_actual_lps.append((chunk_logits[j, actual_tid] - chunk_lse[j]).item())
+            else:
+                all_actual_lps.append(None)  # will use top_log_probs
+
+        all_top_log_probs_list.append(chunk_top_log_probs.cpu())
+        all_top_indices_list.append(chunk_top_indices.cpu())
+        unique_ids.update(int(tid) for tid in chunk_top_indices.flatten().tolist())
+        del chunk_logits, chunk_lse, chunk_top_values
+
+    del logits
+    torch.cuda.empty_cache()
+
+    all_top_log_probs = torch.cat(all_top_log_probs_list, dim=0)
+    all_top_indices = torch.cat(all_top_indices_list, dim=0)
+
+    unique_ids_list = sorted(unique_ids)
+    decoded_list = shared.tokenizer.batch_decode([[tid] for tid in unique_ids_list]) if hasattr(shared.tokenizer, 'batch_decode') else [shared.tokenizer.decode(torch.tensor([tid])) for tid in unique_ids_list]
+    decoded_strs = dict(zip(unique_ids_list, decoded_list))
 
     for i in range(1, n_tokens):
         token_id = int(token_ids[i])
@@ -139,7 +159,6 @@ def _compute_prompt_logprob_entries(prompt, logprobs_count, input_ids=None):
         top_ids = all_top_indices[idx].tolist()
         actual_token_str = decoded_strs[token_id]
 
-        # Build the top list with the actual prompt token guaranteed at front
         if token_id in top_ids:
             actual_lp = top_log_probs[top_ids.index(token_id)].item()
             alternatives = [
@@ -147,10 +166,10 @@ def _compute_prompt_logprob_entries(prompt, logprobs_count, input_ids=None):
                 for j in range(k) if top_ids[j] != token_id
             ]
         else:
-            actual_lp = (prompt_logits[idx, token_id] - all_lse[idx]).item()
+            actual_lp = all_actual_lps[idx]
             alternatives = [
                 {"token": decoded_strs[top_ids[j]], "token_id": top_ids[j], "logprob": top_log_probs[j].item()}
-                for j in range(k - 1)  # drop lowest to make room
+                for j in range(k - 1)
             ]
 
         entry = {"top_logprobs": [{"token": actual_token_str, "token_id": token_id, "logprob": actual_lp}] + alternatives}
diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 7556a908eb..e1efbfeb90 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -527,6 +527,20 @@ def generate(self, prompt, state):
 
         return output
 
+    def get_prompt_logits(self, input_ids):
+        """Return logits for all positions via a single no-cache forward pass.
+
+        Used by prompt logprobs computation. Returns (1, seq_len, vocab) on CPU in float32.
+        """
+        import torch
+        input_ids_tensor = input_ids if isinstance(input_ids, torch.Tensor) else torch.tensor(input_ids, dtype=torch.long)
+        input_ids_tensor = input_ids_tensor.view(1, -1).cpu()
+        with torch.no_grad():
+            return self.model.forward(
+                input_ids=input_ids_tensor,
+                params={"attn_mode": "flash_attn_nc"}
+            ).cpu().float()
+
     def get_logits(self, token_ids, **kwargs):
         """
         Process a batch of token_ids and return the logits for the last token.

From c50e17bdbe1da850189188afaf0682a952efa0d1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 14:49:31 -0300
Subject: [PATCH 1467/1701] Add dedicated ik portable requirements files and
 remove macOS ik builds

---
 .github/workflows/build-everything-tgw.yml    |  7 ---
 .../build-portable-release-ik-cuda.yml        |  9 ++--
 .../workflows/build-portable-release-ik.yml   | 44 +++----------------
 requirements/portable/requirements_ik.txt     | 27 ++++++++++++
 .../portable/requirements_ik_cpu_only.txt     | 27 ++++++++++++
 .../portable/requirements_ik_cuda131.txt      | 27 ++++++++++++
 6 files changed, 91 insertions(+), 50 deletions(-)
 create mode 100644 requirements/portable/requirements_ik.txt
 create mode 100644 requirements/portable/requirements_ik_cpu_only.txt
 create mode 100644 requirements/portable/requirements_ik_cuda131.txt

diff --git a/.github/workflows/build-everything-tgw.yml b/.github/workflows/build-everything-tgw.yml
index 4de591f407..40d9db5d85 100644
--- a/.github/workflows/build-everything-tgw.yml
+++ b/.github/workflows/build-everything-tgw.yml
@@ -96,10 +96,3 @@ jobs:
     with:
       version: ${{ inputs.version }}
       config: 'os:ubuntu-22.04'
-
-  build_release_ik_macos:
-    name: ik macOS
-    uses: ./.github/workflows/build-portable-release-ik.yml
-    with:
-      version: ${{ inputs.version }}
-      config: 'os:macos-14'
diff --git a/.github/workflows/build-portable-release-ik-cuda.yml b/.github/workflows/build-portable-release-ik-cuda.yml
index 40b4b92f98..331a7653f7 100644
--- a/.github/workflows/build-portable-release-ik-cuda.yml
+++ b/.github/workflows/build-portable-release-ik-cuda.yml
@@ -138,14 +138,13 @@ jobs:
             # 3. Prepare requirements file based on CUDA version
             cd "text-generation-webui-${VERSION_CLEAN}"
             if [[ "$CUDA_VERSION" == "13.1" ]]; then
-                REQ_FILE="requirements/portable/requirements_cuda131.txt"
+                REQ_FILE="requirements/portable/requirements_ik_cuda131.txt"
             else
-                REQ_FILE="requirements/portable/requirements.txt"
+                REQ_FILE="requirements/portable/requirements_ik.txt"
             fi
 
-            # 4. Swap llama.cpp wheels for ik_llama.cpp and inject --ik into start scripts
-            sed -i 's|/llama_cpp_binaries-|/ik_llama_cpp_binaries-|g' "$REQ_FILE"
-            sed -i 's/--portable/--portable --ik/g' start_linux.sh start_windows.bat start_macos.sh 2>/dev/null || true
+            # 4. Inject --ik into start scripts
+            sed -i 's/--portable/--portable --ik/g' start_linux.sh start_windows.bat 2>/dev/null || true
 
             # 5. Install packages
             echo "Installing Python packages from $REQ_FILE..."
diff --git a/.github/workflows/build-portable-release-ik.yml b/.github/workflows/build-portable-release-ik.yml
index afb2e76327..bf54eb0e88 100644
--- a/.github/workflows/build-portable-release-ik.yml
+++ b/.github/workflows/build-portable-release-ik.yml
@@ -1,4 +1,4 @@
-name: Build ik CPU and macOS
+name: Build ik CPU
 
 on:
   workflow_dispatch:
@@ -57,7 +57,7 @@ jobs:
         id: set-matrix
         run: |
           $matrix = @{
-              'os' = @('ubuntu-22.04', 'windows-2022', 'macos-14')
+              'os' = @('ubuntu-22.04', 'windows-2022')
               'pyver' = @("3.13")
           }
 
@@ -110,7 +110,6 @@ jobs:
 
             # Define common variables
             VERSION="${{ inputs.version }}"
-            OS_TYPE="${{ matrix.os }}"
 
             # 1. Set platform-specific variables
             if [[ "$RUNNER_OS" == "Windows" ]]; then
@@ -119,21 +118,7 @@ jobs:
                 PIP_PATH="portable_env/python.exe -m pip"
                 PACKAGES_PATH="portable_env/Lib/site-packages"
                 rm start_linux.sh start_macos.sh
-            elif [[ "$RUNNER_OS" == "macOS" ]]; then
-                if [[ "$OS_TYPE" == "macos-15-intel" ]]; then
-                    PLATFORM="macos-x86_64"
-                    PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-apple-darwin-install_only_stripped.tar.gz"
-                    REQ_TYPE="apple_intel"
-                else
-                    PLATFORM="macos-arm64"
-                    PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-aarch64-apple-darwin-install_only_stripped.tar.gz"
-                    REQ_TYPE="apple_silicon"
-                fi
-                PIP_PATH="portable_env/bin/python -m pip"
-                PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
-                rm start_linux.sh start_windows.bat
             else
-                # Linux case
                 PLATFORM="linux-cpu"
                 PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only_stripped.tar.gz"
                 PIP_PATH="portable_env/bin/python -m pip"
@@ -148,30 +133,13 @@ jobs:
             tar -xzf python-build.tar.gz
             mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
 
-            # 3. Prepare requirements file based on platform
+            # 3. Prepare requirements file
             cd "text-generation-webui-${VERSION_CLEAN}"
-
-            # Select requirements file based on platform
-            if [[ "$RUNNER_OS" == "macOS" ]]; then
-                if [[ "$OS_TYPE" == "macos-15-intel" ]]; then
-                    REQ_FILE="requirements/portable/requirements_apple_intel.txt"
-                else
-                    REQ_FILE="requirements/portable/requirements_apple_silicon.txt"
-                fi
-            else
-                REQ_FILE="requirements/portable/requirements_cpu_only.txt"
-            fi
-
+            REQ_FILE="requirements/portable/requirements_ik_cpu_only.txt"
             echo "Using requirements file: $REQ_FILE"
 
-            # 4. Swap llama.cpp wheels for ik_llama.cpp and inject --ik into start scripts
-            if [[ "$RUNNER_OS" == "macOS" ]]; then
-                sed -i '' 's|/llama_cpp_binaries-|/ik_llama_cpp_binaries-|g' "$REQ_FILE"
-                sed -i '' 's/--portable/--portable --ik/g' start_macos.sh
-            else
-                sed -i 's|/llama_cpp_binaries-|/ik_llama_cpp_binaries-|g' "$REQ_FILE"
-                sed -i 's/--portable/--portable --ik/g' start_linux.sh start_windows.bat 2>/dev/null || true
-            fi
+            # 4. Inject --ik into start scripts
+            sed -i 's/--portable/--portable --ik/g' start_linux.sh start_windows.bat 2>/dev/null || true
 
             # 5. Install packages
             echo "Installing Python packages from $REQ_FILE..."
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
new file mode 100644
index 0000000000..2fa037f774
--- /dev/null
+++ b/requirements/portable/requirements_ik.txt
@@ -0,0 +1,27 @@
+audioop-lts<1.0; python_version >= "3.13"
+fastapi==0.112.4
+huggingface-hub==1.5.*
+jinja2==3.1.6
+markdown
+numpy==2.2.*
+pydantic==2.11.0
+pymupdf==1.27.*
+python-docx==1.1.2
+pyyaml
+requests
+rich
+trafilatura==2.0.0
+tqdm
+
+# Gradio
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+
+# API
+flask_cloudflared==0.0.15
+sse-starlette==1.6.5
+tiktoken
+
+# CUDA wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
new file mode 100644
index 0000000000..b43b51c41e
--- /dev/null
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -0,0 +1,27 @@
+audioop-lts<1.0; python_version >= "3.13"
+fastapi==0.112.4
+huggingface-hub==1.5.*
+jinja2==3.1.6
+markdown
+numpy==2.2.*
+pydantic==2.11.0
+pymupdf==1.27.*
+python-docx==1.1.2
+pyyaml
+requests
+rich
+trafilatura==2.0.0
+tqdm
+
+# Gradio
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+
+# API
+flask_cloudflared==0.0.15
+sse-starlette==1.6.5
+tiktoken
+
+# ik_llama.cpp (CPU only)
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
new file mode 100644
index 0000000000..127672854a
--- /dev/null
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -0,0 +1,27 @@
+audioop-lts<1.0; python_version >= "3.13"
+fastapi==0.112.4
+huggingface-hub==1.5.*
+jinja2==3.1.6
+markdown
+numpy==2.2.*
+pydantic==2.11.0
+pymupdf==1.27.*
+python-docx==1.1.2
+pyyaml
+requests
+rich
+trafilatura==2.0.0
+tqdm
+
+# Gradio
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+
+# API
+flask_cloudflared==0.0.15
+sse-starlette==1.6.5
+tiktoken
+
+# CUDA wheels
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 8f8b57a029715d07ab164aa22a779ea7ea4619f1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 10:54:20 -0700
Subject: [PATCH 1468/1701] Update exllamav3

---
 requirements/full/requirements.txt | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 57991c9a75..5591c9ca54 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -44,7 +44,7 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.26/exllamav3-0.0.26+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.26/exllamav3-0.0.26+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"

From 6a1f720c7bb9aef73c1c7c4e311460174c5255ec Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 10:58:20 -0700
Subject: [PATCH 1469/1701] Update transformers

---
 requirements/full/requirements.txt               | 2 +-
 requirements/full/requirements_amd.txt           | 2 +-
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_cpu_only.txt      | 2 +-
 requirements/full/requirements_nowheels.txt      | 2 +-
 6 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 5591c9ca54..30ee0316d4 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -25,7 +25,7 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 trafilatura==2.0.0
-transformers==5.3.*
+transformers==5.5.*
 triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index bb47ea4bcd..9edc1d9532 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -22,7 +22,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.3.*
+transformers==5.5.*
 tqdm
 trafilatura==2.0.0
 wandb
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 5750b10931..ff8687c13a 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -22,7 +22,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.3.*
+transformers==5.5.*
 tqdm
 trafilatura==2.0.0
 wandb
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index d8302d3db3..208632e8fc 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -22,7 +22,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.3.*
+transformers==5.5.*
 tqdm
 trafilatura==2.0.0
 wandb
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index d3a5c00891..4a7e5aaa77 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -22,7 +22,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.3.*
+transformers==5.5.*
 tqdm
 trafilatura==2.0.0
 wandb
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 052085cc42..6200589e83 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -22,7 +22,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.3.*
+transformers==5.5.*
 tqdm
 trafilatura==2.0.0
 wandb

From 468cb5cb87bf02f96efcd5acb1d1ac4b08c68273 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 10:59:28 -0700
Subject: [PATCH 1470/1701] Update accelerate

---
 requirements/full/requirements.txt               | 2 +-
 requirements/full/requirements_amd.txt           | 2 +-
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_cpu_only.txt      | 2 +-
 requirements/full/requirements_nowheels.txt      | 2 +-
 6 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 30ee0316d4..e5bec6ecbb 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -1,4 +1,4 @@
-accelerate==1.12.*
+accelerate==1.13.*
 audioop-lts<1.0; python_version >= "3.13"
 bitsandbytes==0.49.*
 datasets
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 9edc1d9532..c6b5b2d00b 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -1,4 +1,4 @@
-accelerate==1.12.*
+accelerate==1.13.*
 audioop-lts<1.0; python_version >= "3.13"
 datasets
 diffusers==0.37.*
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index ff8687c13a..ce671f0a5c 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -1,4 +1,4 @@
-accelerate==1.12.*
+accelerate==1.13.*
 audioop-lts<1.0; python_version >= "3.13"
 datasets
 diffusers==0.37.*
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 208632e8fc..d12d9f807b 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -1,4 +1,4 @@
-accelerate==1.12.*
+accelerate==1.13.*
 audioop-lts<1.0; python_version >= "3.13"
 datasets
 diffusers==0.37.*
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 4a7e5aaa77..4066b1af63 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -1,4 +1,4 @@
-accelerate==1.12.*
+accelerate==1.13.*
 audioop-lts<1.0; python_version >= "3.13"
 datasets
 diffusers==0.37.*
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 6200589e83..7173345a1c 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -1,4 +1,4 @@
-accelerate==1.12.*
+accelerate==1.13.*
 audioop-lts<1.0; python_version >= "3.13"
 datasets
 diffusers==0.37.*

From 80e81a54cacacbd8aa16ccf312ae0e574e4b416c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 11:11:44 -0700
Subject: [PATCH 1471/1701] Remove ik macOS wheels from full requirements

---
 requirements/full/requirements_apple_intel.txt   | 1 -
 requirements/full/requirements_apple_silicon.txt | 1 -
 2 files changed, 2 deletions(-)

diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index ce671f0a5c..55a313e9d9 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -38,4 +38,3 @@ tiktoken
 
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index d12d9f807b..a6d34cbbb8 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -38,4 +38,3 @@ tiktoken
 
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"

From f6f8f14c8d0993327a2c86dfa3c976a7c1c569fc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 16:13:39 -0300
Subject: [PATCH 1472/1701] Security: Fix SSRF in superbooga extensions

---
 extensions/superbooga/download_urls.py   | 3 +++
 extensions/superboogav2/download_urls.py | 2 ++
 2 files changed, 5 insertions(+)

diff --git a/extensions/superbooga/download_urls.py b/extensions/superbooga/download_urls.py
index 424a988576..b28fea42dd 100644
--- a/extensions/superbooga/download_urls.py
+++ b/extensions/superbooga/download_urls.py
@@ -2,8 +2,11 @@
 
 import requests
 
+from modules.web_search import _validate_url
+
 
 def download_single(url):
+    _validate_url(url)
     headers = {
         'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/58.0.3029.110 Safari/537.3'
     }
diff --git a/extensions/superboogav2/download_urls.py b/extensions/superboogav2/download_urls.py
index 5b5a2e17ac..4d8b98b1a7 100644
--- a/extensions/superboogav2/download_urls.py
+++ b/extensions/superboogav2/download_urls.py
@@ -5,12 +5,14 @@
 from bs4 import BeautifulSoup
 
 import extensions.superboogav2.parameters as parameters
+from modules.web_search import _validate_url
 
 from .data_processor import process_and_add_to_collector
 from .utils import create_metadata_source
 
 
 def _download_single(url):
+    _validate_url(url)
     response = requests.get(url, timeout=5)
     if response.status_code == 200:
         return response.content

From 091037ec20743ac6c7bccb75b59743045a692c4a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 16:13:45 -0300
Subject: [PATCH 1473/1701] Fix top_logprobs_ids missing for llama.cpp loader

---
 modules/api/completions.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/modules/api/completions.py b/modules/api/completions.py
index 4eb8fdad4f..98bcff4705 100644
--- a/modules/api/completions.py
+++ b/modules/api/completions.py
@@ -299,8 +299,9 @@ def format_completion_logprobs(entries):
             t = item.get('token', '')
             lp = item.get('logprob', item.get('prob', 0))
             top_dict[t] = lp
-            if 'token_id' in item:
-                top_dict_ids[item['token_id']] = lp
+            tid = item.get('token_id', item.get('id'))
+            if tid is not None:
+                top_dict_ids[tid] = lp
         top_logprobs.append(top_dict)
         top_logprobs_ids.append(top_dict_ids if top_dict_ids else None)
 

From a61bde509ff44a0f7662067bc94efd7f103f3162 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 17:30:02 -0700
Subject: [PATCH 1474/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 8 ++++----
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 8 ++++----
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 15 files changed, 30 insertions(+), 30 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index e5bec6ecbb..f1a953a5c3 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,10 +40,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index c6b5b2d00b..211600e268 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 55a313e9d9..54d904dd0a 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index a6d34cbbb8..8829eb4486 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 4066b1af63..0a8cfac62f 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,7 +37,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 1180b42d1b..607c642f64 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 57aa62628b..f0af64c8ae 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 894c919931..c5f351c549 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 32b9727f6a..5287aa2571 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 73b7283282..038318ab7e 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index ad96bbe2ee..d87c741e33 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index 2fa037f774..3e2471aea4 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index b43b51c41e..8272b9b6dd 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # ik_llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index 127672854a..98ef23d747 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/ik_llama_cpp_binaries-0.101.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index a5df3ad4e5..157ad313d2 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.101.0/llama_cpp_binaries-0.101.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From d84157403a1c8b65f8597302463e46c28a6659d1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 17:31:44 -0700
Subject: [PATCH 1475/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 17 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index f1a953a5c3..b38ae8483e 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -31,8 +31,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 211600e268..7fb3a7d927 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 54d904dd0a..4a0f764c5d 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 8829eb4486..942d5d71af 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 0a8cfac62f..6b61dca73d 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 7173345a1c..a4d6cc97a3 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 607c642f64..5aff54b26b 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index f0af64c8ae..0771f53eb5 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index c5f351c549..427d59b2aa 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 5287aa2571..c47a6ca1f6 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 038318ab7e..e491e357f1 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index d87c741e33..5870983a2c 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index 3e2471aea4..d11d337dfc 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index 8272b9b6dd..c2b69e1c65 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index 98ef23d747..7f28093090 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index e38140cebd..322056bec4 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 157ad313d2..dfd52be51e 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio-4.37.2+custom.13-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.13/gradio_client-1.0.2+custom.13-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From 7aab2fdf9aefb0f14fbf58e132a2a9a5850f8319 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 17:50:42 -0700
Subject: [PATCH 1476/1701] API: Improve cache clearing in logprobs

---
 modules/api/completions.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/modules/api/completions.py b/modules/api/completions.py
index 98bcff4705..f228273118 100644
--- a/modules/api/completions.py
+++ b/modules/api/completions.py
@@ -89,6 +89,7 @@ def _compute_prompt_logprob_entries(prompt, logprobs_count, input_ids=None):
         return [{"token": first_token_str, "null_logprob": True}]
 
     import torch
+    from modules.torch_utils import clear_torch_cache
 
     if hasattr(model, 'get_prompt_logits'):
         logits = model.get_prompt_logits(input_ids)
@@ -143,7 +144,7 @@ def _compute_prompt_logprob_entries(prompt, logprobs_count, input_ids=None):
         del chunk_logits, chunk_lse, chunk_top_values
 
     del logits
-    torch.cuda.empty_cache()
+    clear_torch_cache()
 
     all_top_log_probs = torch.cat(all_top_log_probs_list, dim=0)
     all_top_indices = torch.cat(all_top_indices_list, dim=0)

From b108c55353e11343a9e8f8566d92e52e868dfa69 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 19:10:41 -0700
Subject: [PATCH 1477/1701] Fix portable builds not starting due to missing ik
 element

---
 modules/loaders.py | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/modules/loaders.py b/modules/loaders.py
index cb1f3d3bad..31b1b51a7c 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -291,16 +291,21 @@ def blacklist_samplers(loader, dynamic_temperature):
 
 @functools.cache
 def get_all_params():
+    from modules import shared
     all_params = set()
     for k in loaders_and_params:
         for el in loaders_and_params[k]:
             all_params.add(el)
 
+    if shared.args.portable:
+        all_params.discard('ik')
+
     return sorted(all_params)
 
 
+@functools.cache
 def list_model_elements():
-    return [
+    elements = [
         'filter_by_loader',
         'loader',
         'cpu_memory',
@@ -346,9 +351,14 @@ def list_model_elements():
         'spec_ngram_size_m',
         'spec_ngram_min_hits',
         'mmproj',
-        'ik',
     ]
 
+    from modules import shared
+    if not shared.args.portable:
+        elements.append('ik')
+
+    return elements
+
 
 def make_loader_params_visible(loader):
     import gradio as gr

From 6e2b70bde60c089f97b0abe97bb1b594cce75357 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 20:26:09 -0700
Subject: [PATCH 1478/1701] Add Gemma 4 tool-calling support

---
 modules/chat.py         | 57 +++++++++++++++++++++++++++++
 modules/reasoning.py    |  1 +
 modules/tool_parsing.py | 79 +++++++++++++++++++++++++++++++++++++++++
 3 files changed, 137 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index edda11b09a..818309e645 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -210,6 +210,57 @@ def _expand_tool_sequence(tool_seq):
     return messages
 
 
+def _convert_to_tool_responses(messages):
+    """Convert role:'tool' messages to tool_responses format.
+
+    Templates like Gemma 4 expect tool results as a ``tool_responses``
+    attribute on a message rather than separate ``role: 'tool'`` messages.
+    This function groups consecutive tool messages and rewrites them.
+    """
+    result = []
+    tc_id_to_name = {}
+
+    i = 0
+    while i < len(messages):
+        msg = messages[i]
+
+        if msg.get('tool_calls'):
+            for tc in msg['tool_calls']:
+                tc_id = tc.get('id', '')
+                func_name = tc.get('function', {}).get('name', 'unknown')
+                if tc_id:
+                    tc_id_to_name[tc_id] = func_name
+
+        if msg.get('role') == 'tool':
+            tool_responses = []
+            while i < len(messages) and messages[i].get('role') == 'tool':
+                tool_msg = messages[i]
+                tc_id = tool_msg.get('tool_call_id', '')
+                func_name = tc_id_to_name.get(tc_id, 'unknown')
+
+                content = tool_msg.get('content', '')
+                try:
+                    response = json.loads(content)
+                except (json.JSONDecodeError, ValueError, TypeError):
+                    response = content
+
+                tool_responses.append({
+                    'name': func_name,
+                    'response': response,
+                })
+                i += 1
+
+            result.append({
+                'role': 'tool',
+                'tool_responses': tool_responses,
+            })
+        else:
+            result.append(msg)
+            i += 1
+
+    return result
+
+
 def _format_attachments(attachments, include_text=True):
     """Build image ref and text attachment strings from a list of attachments."""
     attachments_text = ""
@@ -267,6 +318,9 @@ def generate_chat_prompt(user_input, state, **kwargs):
         tools=state['tools'] if 'tools' in state else None,
     )
 
+    active_template_str = state['instruction_template_str'] if state['mode'] == 'instruct' else chat_template_str
+    uses_tool_responses = 'tool_responses' in active_template_str
+
     messages = []
 
     if state['mode'] == 'instruct':
@@ -503,6 +557,9 @@ def make_prompt(messages):
 
         return prompt
 
+    if uses_tool_responses:
+        messages = _convert_to_tool_responses(messages)
+
     prompt = make_prompt(messages)
 
     # Handle truncation
diff --git a/modules/reasoning.py b/modules/reasoning.py
index aa1939b8f3..4a7cfa7945 100644
--- a/modules/reasoning.py
+++ b/modules/reasoning.py
@@ -7,6 +7,7 @@
     ('<|channel|>analysis<|message|>', '<|end|>', '<|channel|>final<|message|>'),
     ('<|channel|>commentary<|message|>', '<|end|>', '<|channel|>final<|message|>'),
     ('', '', None),
+    ('<|channel>thought', '', None),  # Gemma 4
     ('<|think|>', '<|end|>', '<|content|>'),  # Solar Open
     # ('Thinking Process:', '', None),  # Qwen3.5 verbose thinking outside tags -- removed: too prone to false positives in streaming
     (None, '', None),  # End-only variant (e.g., Qwen3-next)
diff --git a/modules/tool_parsing.py b/modules/tool_parsing.py
index ec49f77f76..45da25c968 100644
--- a/modules/tool_parsing.py
+++ b/modules/tool_parsing.py
@@ -27,6 +27,7 @@ def get_tool_call_id() -> str:
     '[TOOL_CALLS]',
     'to=functions.',
     '<|channel|>commentary',
+    '<|tool_call>call:',
 ]
 
 
@@ -400,6 +401,78 @@ def _parse_glm_tool_calls(answer: str, tool_names: list[str]):
     return matches, start_pos
 
 
+def _extract_gemma4_balanced(text, start):
+    """Extract balanced braces from Gemma 4 format, using <|"|> as string delimiters."""
+    if start >= len(text) or text[start] != '{':
+        return None
+    depth = 0
+    in_string = False
+    quote_token = '<|"|>'
+    quote_len = len(quote_token)
+    i = start
+    while i < len(text):
+        if text[i:i + quote_len] == quote_token:
+            in_string = not in_string
+            i += quote_len
+            continue
+        if in_string:
+            i += 1
+            continue
+        c = text[i]
+        if c == '{':
+            depth += 1
+        elif c == '}':
+            depth -= 1
+            if depth == 0:
+                return text[start:i + 1]
+        i += 1
+    return None
+
+
+def _parse_gemma4_tool_calls(answer: str, tool_names: list[str]):
+    """Parse Gemma 4-style tool calls.
+
+    Format:
+        <|tool_call>call:func_name{key:<|"|>value<|"|>,...}
+
+    Values use <|"|> tokens instead of standard JSON quotes, and keys are
+    bare identifiers.
+    """
+    matches = []
+    start_pos = None
+
+    for m in re.finditer(r'<\|tool_call>call:([^\s{]+)\s*', answer):
+        func_name = m.group(1).strip()
+        if func_name not in tool_names:
+            continue
+
+        brace_start = m.end()
+        if brace_start >= len(answer) or answer[brace_start] != '{':
+            continue
+
+        content = _extract_gemma4_balanced(answer, brace_start)
+        if content is None:
+            continue
+
+        # Convert to JSON: split on <|"|> tokens so that key quoting
+        # only applies outside string values (even-indexed parts),
+        # then rejoin with real quotes.
+        parts = content.split('<|"|>')
+        for idx in range(0, len(parts), 2):
+            parts[idx] = re.sub(r'(^|[{,\[])\s*(\w+)\s*:', r'\1"\2":', parts[idx])
+        json_str = '"'.join(parts)
+
+        try:
+            arguments = json.loads(json_str)
+            if start_pos is None:
+                start_pos = m.start()
+            matches.append(_make_tool_call(func_name, arguments))
+        except (json.JSONDecodeError, ValueError):
+            pass
+
+    return matches, start_pos
+
+
 def _parse_pythonic_tool_calls(answer: str, tool_names: list[str]):
     """Parse pythonic-style tool calls used by Llama 4 and similar models.
 
@@ -472,6 +545,11 @@ def _parse_pythonic_tool_calls(answer: str, tool_names: list[str]):
         'parser': _parse_channel_tool_calls,
         'markers': ['to=functions.', '<|channel|>commentary'],
     },
+    {
+        'template_hints': ['<|tool_call>call:'],
+        'parser': _parse_gemma4_tool_calls,
+        'markers': ['<|tool_call>call:'],
+    },
     {
         'template_hints': ['minimax:tool_call'],
         'parser': _parse_minimax_tool_calls,
@@ -504,6 +582,7 @@ def _parse_pythonic_tool_calls(answer: str, tool_names: list[str]):
     _parse_deep_seek_tool_calls,
     _parse_kimi_tool_calls,
     _parse_channel_tool_calls,
+    _parse_gemma4_tool_calls,
     _parse_minimax_tool_calls,
     _parse_glm_tool_calls,
     _parse_xml_param_tool_calls,

From 42dfcdfc5b50333c40a6adda0f4c8672508212cb Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 20:46:27 -0700
Subject: [PATCH 1479/1701] API: Add warning about vanilla llama-server not
 supporting prompt logprobs + instructions

---
 modules/llama_cpp_server.py | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 3408046609..2d873f00d4 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -333,6 +333,12 @@ def get_prompt_logprob_entries(self, token_ids, n_probs=5, prompt=""):
 
         prompt_probs = result.get("prompt_probabilities", [])
         if not prompt_probs:
+            logger.warning(
+                "The llama.cpp server did not return prompt probabilities. "
+                "This feature requires a custom build with prompt_logprobs support. "
+                "See: https://github.com/oobabooga/llama.cpp/tree/prompt-logprobs "
+                "or https://github.com/oobabooga/ik_llama.cpp/tree/prompt-logprobs"
+            )
             return []
 
         # Null first token (no conditioning context); use empty string for BOS

From a1cb5b5dc05d2540640069b9549dd93557c81a16 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 2 Apr 2026 21:56:06 -0700
Subject: [PATCH 1480/1701] llama.cpp: Disable jinja by default (we use Python
 jinja, not cpp jinja)

This was causing template compilation issues with qwen models.
---
 modules/llama_cpp_server.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 2d873f00d4..a4390adb97 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -418,6 +418,7 @@ def _start_server(self):
             "--ubatch-size", str(shared.args.ubatch_size),
             "--port", str(self.port),
             "--no-webui",
+            "--no-jinja",
             "--flash-attn", "on",
         ]
 

From 000d776967f0a73684b85c9d052a738dba073fb6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 3 Apr 2026 05:49:03 -0700
Subject: [PATCH 1481/1701] Revert "llama.cpp: Disable jinja by default (we use
 Python jinja, not cpp jinja)"

This reverts commit a1cb5b5dc05d2540640069b9549dd93557c81a16.
---
 modules/llama_cpp_server.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index a4390adb97..2d873f00d4 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -418,7 +418,6 @@ def _start_server(self):
             "--ubatch-size", str(shared.args.ubatch_size),
             "--port", str(self.port),
             "--no-webui",
-            "--no-jinja",
             "--flash-attn", "on",
         ]
 

From 66d1a22c733b04c38d89a128a7eeacd8e142e629 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 3 Apr 2026 05:56:36 -0700
Subject: [PATCH 1482/1701] Fix crash when no model is selected (None passed to
 resolve_model_path)

---
 modules/utils.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/modules/utils.py b/modules/utils.py
index b01953ee22..c4acf7144d 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -105,6 +105,9 @@ def resolve_model_path(model_name_or_path, image_model=False):
     before the default models directory.
     """
 
+    if model_name_or_path is None:
+        raise FileNotFoundError("No model specified.")
+
     path_candidate = Path(model_name_or_path)
     if path_candidate.exists():
         return path_candidate

From 8bba9ecc3fc0afc044a1f6810f014f721dbb7809 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 3 Apr 2026 05:58:05 -0700
Subject: [PATCH 1483/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 17 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index b38ae8483e..91d27d867c 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -31,8 +31,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 7fb3a7d927..eea869b134 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 4a0f764c5d..391973b771 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 942d5d71af..4d0ffe294f 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 6b61dca73d..44e54eaa48 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index a4d6cc97a3..41d6aad670 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 5aff54b26b..91b58e0d76 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 0771f53eb5..36d6dcb155 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 427d59b2aa..0d882b831b 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index c47a6ca1f6..d79832e588 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index e491e357f1..3e1de9c904 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 5870983a2c..40e68d9918 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index d11d337dfc..9e61ad3f42 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index c2b69e1c65..cdd1218f45 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index 7f28093090..b7422758de 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 322056bec4..372e718b9e 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index dfd52be51e..3e539988d5 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio-4.37.2+custom.14-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.14/gradio_client-1.0.2+custom.14-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From 95d6c53e13673defecff6def4aead4c7ea157911 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 3 Apr 2026 07:30:48 -0700
Subject: [PATCH 1484/1701] Revert "API: Add warning about vanilla llama-server
 not supporting prompt logprobs + instructions"

This reverts commit 42dfcdfc5b50333c40a6adda0f4c8672508212cb.
---
 modules/llama_cpp_server.py | 6 ------
 1 file changed, 6 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 2d873f00d4..3408046609 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -333,12 +333,6 @@ def get_prompt_logprob_entries(self, token_ids, n_probs=5, prompt=""):
 
         prompt_probs = result.get("prompt_probabilities", [])
         if not prompt_probs:
-            logger.warning(
-                "The llama.cpp server did not return prompt probabilities. "
-                "This feature requires a custom build with prompt_logprobs support. "
-                "See: https://github.com/oobabooga/llama.cpp/tree/prompt-logprobs "
-                "or https://github.com/oobabooga/ik_llama.cpp/tree/prompt-logprobs"
-            )
             return []
 
         # Null first token (no conditioning context); use empty string for BOS

From 131a9a0140baeef90061bb97065d32b23385e142 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 3 Apr 2026 09:15:03 -0700
Subject: [PATCH 1485/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 8 ++++----
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 8 ++++----
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 15 files changed, 30 insertions(+), 30 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 91d27d867c..8816f76eef 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -40,10 +40,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index eea869b134..466b680f2e 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -37,5 +37,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 391973b771..49a948c733 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 4d0ffe294f..508c137a5f 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -37,4 +37,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 44e54eaa48..17ecbd61dd 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -37,7 +37,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 91b58e0d76..73d3f3b620 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 36d6dcb155..95c234247a 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 0d882b831b..4d18875b14 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index d79832e588..a181212be8 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -23,4 +23,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 3e1de9c904..5ddd53e19e 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 40e68d9918..3ab238acf1 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index 9e61ad3f42..624fbe5aea 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index cdd1218f45..c1ab17581b 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # ik_llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index b7422758de..6d17200d29 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/ik_llama_cpp_binaries-0.102.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 3e539988d5..cdeb2b79f0 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.102.0/llama_cpp_binaries-0.102.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 8e8e1ba8984cc3cef4b4c0d88e7c9eb7977dd3fe Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 3 Apr 2026 09:50:15 -0700
Subject: [PATCH 1486/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 17 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 8816f76eef..3b5501f46b 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -31,8 +31,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 466b680f2e..eb7f86180f 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 49a948c733..e11522a98d 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 508c137a5f..76d8a70910 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 17ecbd61dd..8d4df234e5 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 41d6aad670..cecc2a25f3 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 73d3f3b620..42db46a448 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 95c234247a..5e0b589b44 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 4d18875b14..711c68f911 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index a181212be8..f1bbccf091 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 5ddd53e19e..dc2807f2f4 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 3ab238acf1..6d34b894cd 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index 624fbe5aea..5b3bac8341 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index c1ab17581b..b8e0897df6 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index 6d17200d29..fd623b0b56 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 372e718b9e..92c910ac04 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index cdeb2b79f0..bc17dda94a 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio-4.37.2+custom.15-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.15/gradio_client-1.0.2+custom.15-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From 6b66da84d2dbccf63ffe79939edf92ad935bb3ee Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 3 Apr 2026 10:01:51 -0700
Subject: [PATCH 1487/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 17 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 3b5501f46b..4d9d243cc7 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -31,8 +31,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index eb7f86180f..b6caf320af 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index e11522a98d..ae2f62637a 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 76d8a70910..f4783d2b2e 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 8d4df234e5..11d670b604 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index cecc2a25f3..d4b1ca80d3 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 42db46a448..901d14949d 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 5e0b589b44..5705c64c8a 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 711c68f911..3c3deaedc6 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index f1bbccf091..3f1e814c4e 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index dc2807f2f4..3bfed7f8f7 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 6d34b894cd..dd7059c8f9 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index 5b3bac8341..4fdd10fedd 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index b8e0897df6..1b37746370 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index fd623b0b56..b85607d361 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 92c910ac04..39628ee6cd 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index bc17dda94a..16aa5593d9 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio-4.37.2+custom.16-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.16/gradio_client-1.0.2+custom.16-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From 5fb8c4fbd6d8112429335b48c93a6fe941f4c5e3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 3 Apr 2026 11:02:00 -0700
Subject: [PATCH 1488/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 17 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 4d9d243cc7..b7a5ca97a3 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -31,8 +31,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index b6caf320af..2c6275852e 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index ae2f62637a..7e3fc35fb4 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index f4783d2b2e..2603201db7 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 11d670b604..fe3bf3ba7a 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index d4b1ca80d3..acae301eab 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 901d14949d..56795843ea 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 5705c64c8a..abaa13383d 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 3c3deaedc6..b22a03d918 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 3f1e814c4e..97c5903ca0 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 3bfed7f8f7..57e92f7443 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index dd7059c8f9..1f7d27a7bc 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index 4fdd10fedd..65f6a004e4 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index 1b37746370..0a82adb702 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index b85607d361..3d812045b7 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 39628ee6cd..91bef10b07 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 16aa5593d9..7c61f0cc78 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio-4.37.2+custom.17-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.17/gradio_client-1.0.2+custom.17-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From 8ecdb41078cfaf54fa0be66d54cf6e3911936b68 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Fri, 3 Apr 2026 19:36:50 -0300
Subject: [PATCH 1489/1701] fix(security): sanitize filenames in all prompt
 file operations (CWE-22) (#7462)

---------

Co-authored-by: Alex Chen 
---
 modules/prompts.py     | 2 ++
 modules/ui_default.py  | 5 ++++-
 modules/ui_notebook.py | 6 +++++-
 3 files changed, 11 insertions(+), 2 deletions(-)

diff --git a/modules/prompts.py b/modules/prompts.py
index d107ce5a9d..85dc32e3a4 100644
--- a/modules/prompts.py
+++ b/modules/prompts.py
@@ -1,6 +1,7 @@
 from pathlib import Path
 
 from modules import shared, utils
+from modules.utils import sanitize_filename
 from modules.text_generation import get_encoded_length
 
 
@@ -18,6 +19,7 @@ def load_prompt(fname):
 
         return initial_content
 
+    fname = sanitize_filename(fname)
     file_path = shared.user_data_dir / 'logs' / 'notebook' / f'{fname}.txt'
     if file_path.exists():
         with open(file_path, 'r', encoding='utf-8') as f:
diff --git a/modules/ui_default.py b/modules/ui_default.py
index 2c367ccaad..48cb2fc26f 100644
--- a/modules/ui_default.py
+++ b/modules/ui_default.py
@@ -10,7 +10,7 @@
     stop_everything_event
 )
 from modules.ui_notebook import store_notebook_state_and_debounce
-from modules.utils import gradio
+from modules.utils import gradio, sanitize_filename
 
 inputs = ('textbox-default', 'interface_state')
 outputs = ('output_textbox', 'html-default')
@@ -167,6 +167,7 @@ def handle_new_prompt():
 
 
 def handle_delete_prompt_confirm_default(prompt_name):
+    prompt_name = sanitize_filename(prompt_name)
     available_prompts = utils.get_available_prompts()
     current_index = available_prompts.index(prompt_name) if prompt_name in available_prompts else 0
 
@@ -199,6 +200,8 @@ def handle_rename_prompt_click_default(current_name):
 
 
 def handle_rename_prompt_confirm_default(new_name, current_name):
+    new_name = sanitize_filename(new_name)
+    current_name = sanitize_filename(current_name)
     old_path = shared.user_data_dir / "logs" / "notebook" / f"{current_name}.txt"
     new_path = shared.user_data_dir / "logs" / "notebook" / f"{new_name}.txt"
 
diff --git a/modules/ui_notebook.py b/modules/ui_notebook.py
index f550e6466b..88f00ac53f 100644
--- a/modules/ui_notebook.py
+++ b/modules/ui_notebook.py
@@ -11,7 +11,7 @@
     get_token_ids,
     stop_everything_event
 )
-from modules.utils import gradio
+from modules.utils import gradio, sanitize_filename
 
 _notebook_file_lock = threading.Lock()
 _notebook_auto_save_timer = None
@@ -202,6 +202,7 @@ def handle_new_prompt():
 
 
 def handle_delete_prompt_confirm_notebook(prompt_name):
+    prompt_name = sanitize_filename(prompt_name)
     available_prompts = utils.get_available_prompts()
     current_index = available_prompts.index(prompt_name) if prompt_name in available_prompts else 0
 
@@ -233,6 +234,8 @@ def handle_rename_prompt_click_notebook(current_name):
 
 
 def handle_rename_prompt_confirm_notebook(new_name, current_name):
+    new_name = sanitize_filename(new_name)
+    current_name = sanitize_filename(current_name)
     old_path = shared.user_data_dir / "logs" / "notebook" / f"{current_name}.txt"
     new_path = shared.user_data_dir / "logs" / "notebook" / f"{new_name}.txt"
 
@@ -249,6 +252,7 @@ def handle_rename_prompt_confirm_notebook(new_name, current_name):
 
 def autosave_prompt(text, prompt_name):
     """Automatically save the text to the selected prompt file"""
+    prompt_name = sanitize_filename(prompt_name)
     if prompt_name and text.strip():
         prompt_path = shared.user_data_dir / "logs" / "notebook" / f"{prompt_name}.txt"
         prompt_path.parent.mkdir(parents=True, exist_ok=True)

From fc35acab9b07f1b0dc57b89a7cb459894aa44c5b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 3 Apr 2026 16:56:15 -0700
Subject: [PATCH 1490/1701] API: Fix tool call parser crash on non-dict JSON
 output

---
 modules/tool_parsing.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/modules/tool_parsing.py b/modules/tool_parsing.py
index 45da25c968..919e523ab6 100644
--- a/modules/tool_parsing.py
+++ b/modules/tool_parsing.py
@@ -699,6 +699,8 @@ def _return(matches, start_pos):
                 if not isinstance(candidates, list):
                     candidates = [candidates]
                 for candidate_dict in candidates:
+                    if not isinstance(candidate_dict, dict):
+                        continue
                     checked_candidate = check_and_sanitize_tool_call_candidate(candidate_dict, tool_names)
                     if checked_candidate is not None:
                         matches.append(checked_candidate)

From 2fbaee58cd7c65c22267410f8a77b6c04b3ee954 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 3 Apr 2026 20:54:28 -0700
Subject: [PATCH 1491/1701] Add Windows + ROCm portable builds

---
 .github/workflows/build-everything-tgw.yml | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/.github/workflows/build-everything-tgw.yml b/.github/workflows/build-everything-tgw.yml
index 40d9db5d85..0b65dfd64f 100644
--- a/.github/workflows/build-everything-tgw.yml
+++ b/.github/workflows/build-everything-tgw.yml
@@ -41,6 +41,13 @@ jobs:
       version: ${{ inputs.version }}
       config: 'os:ubuntu-22.04'
 
+  build_release_rocm_windows:
+    name: ROCm Windows
+    uses: ./.github/workflows/build-portable-release-rocm.yml
+    with:
+      version: ${{ inputs.version }}
+      config: 'os:windows-2022'
+
   build_release_rocm_linux:
     name: ROCm Linux
     uses: ./.github/workflows/build-portable-release-rocm.yml

From 54b2f39c780a0a27d2fe27dc114a0aabe09e0249 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 3 Apr 2026 22:07:21 -0700
Subject: [PATCH 1492/1701] Cleanup modules/chat.py

---
 modules/chat.py | 51 ++++++++++++++++++++-----------------------------
 1 file changed, 21 insertions(+), 30 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 818309e645..e28d39632d 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -694,7 +694,7 @@ def get_stopping_strings(state):
         # Find positions of each message content
         first_user_end = prompt.find("first user message") + len("first user message")
         first_assistant_start = prompt.find("first assistant message")
-        first_assistant_end = prompt.find("first assistant message") + len("first assistant message")
+        first_assistant_end = first_assistant_start + len("first assistant message")
         second_user_start = prompt.find("second user message")
         second_assistant_end = prompt.find("second assistant message") + len("second assistant message")
 
@@ -1819,7 +1819,8 @@ def load_history(unique_id, character, mode):
     if not p.exists():
         return {'internal': [], 'visible': [], 'metadata': {}}
 
-    f = json.loads(open(p, 'rb').read())
+    with open(p, 'rb') as fh:
+        f = json.loads(fh.read())
     if 'internal' in f and 'visible' in f:
         history = f
     else:
@@ -1883,19 +1884,17 @@ def generate_pfp_cache(character):
     if not cache_folder.exists():
         cache_folder.mkdir()
 
-    for path in [shared.user_data_dir / 'characters' / f"{character}.{extension}" for extension in ['png', 'jpg', 'jpeg']]:
+    for extension in ['png', 'jpg', 'jpeg']:
+        path = shared.user_data_dir / 'characters' / f"{character}.{extension}"
         if path.exists():
             original_img = Image.open(path)
-            # Define file paths
-            pfp_path = Path(f'{cache_folder}/pfp_character.png')
-            thumb_path = Path(f'{cache_folder}/pfp_character_thumb.png')
+            pfp_path = cache_folder / 'pfp_character.png'
+            thumb_path = cache_folder / 'pfp_character_thumb.png'
 
-            # Save main picture and thumbnail
             original_img.save(pfp_path, format='PNG')
             thumb = make_thumbnail(original_img)
             thumb.save(thumb_path, format='PNG')
 
-            # Return the path to the thumbnail, not the in-memory PIL Image object.
             return str(thumb_path)
 
     return None
@@ -1916,13 +1915,13 @@ def load_character(character, name1, name2):
         logger.error(f"Could not find the character \"{character}\" inside {shared.user_data_dir}/characters. No character has been loaded.")
         raise ValueError
 
-    file_contents = open(filepath, 'r', encoding='utf-8').read()
+    with open(filepath, 'r', encoding='utf-8') as fh:
+        file_contents = fh.read()
     data = json.loads(file_contents) if extension == "json" else yaml.safe_load(file_contents)
     cache_folder = Path(shared.args.disk_cache_dir)
 
-    for path in [Path(f"{cache_folder}/pfp_character.png"), Path(f"{cache_folder}/pfp_character_thumb.png")]:
-        if path.exists():
-            path.unlink()
+    for path in [cache_folder / "pfp_character.png", cache_folder / "pfp_character_thumb.png"]:
+        path.unlink(missing_ok=True)
 
     picture = generate_pfp_cache(character)
 
@@ -1978,9 +1977,7 @@ def clear_character_for_ui(state):
     # Clear the cache files
     cache_folder = Path(shared.args.disk_cache_dir)
     for cache_file in ['pfp_character.png', 'pfp_character_thumb.png']:
-        cache_path = Path(f'{cache_folder}/{cache_file}')
-        if cache_path.exists():
-            cache_path.unlink()
+        (cache_folder / cache_file).unlink(missing_ok=True)
 
     return state, state['name2'], state['context'], state['greeting'], None
 
@@ -2075,11 +2072,10 @@ def upload_your_profile_picture(img_path):
         cache_folder.mkdir()
 
     if img is None:
-        if Path(f"{cache_folder}/pfp_me.png").exists():
-            Path(f"{cache_folder}/pfp_me.png").unlink()
+        (cache_folder / "pfp_me.png").unlink(missing_ok=True)
     else:
         img = make_thumbnail(img)
-        img.save(Path(f'{cache_folder}/pfp_me.png'))
+        img.save(cache_folder / 'pfp_me.png')
         logger.info(f'Profile picture saved to "{cache_folder}/pfp_me.png"')
 
 
@@ -2135,13 +2131,12 @@ def generate_user_pfp_cache(user):
     if not cache_folder.exists():
         cache_folder.mkdir()
 
-    for path in [shared.user_data_dir / 'users' / f"{user}.{extension}" for extension in ['png', 'jpg', 'jpeg']]:
+    for extension in ['png', 'jpg', 'jpeg']:
+        path = shared.user_data_dir / 'users' / f"{user}.{extension}"
         if path.exists():
             original_img = Image.open(path)
-            # Define file paths
-            pfp_path = Path(f'{cache_folder}/pfp_me.png')
+            pfp_path = cache_folder / 'pfp_me.png'
 
-            # Save thumbnail
             thumb = make_thumbnail(original_img)
             thumb.save(pfp_path, format='PNG')
             logger.info(f'User profile picture cached to "{pfp_path}"')
@@ -2173,9 +2168,7 @@ def load_user(user_name, name1, user_bio):
 
     # Clear existing user picture cache
     cache_folder = Path(shared.args.disk_cache_dir)
-    pfp_path = Path(f"{cache_folder}/pfp_me.png")
-    if pfp_path.exists():
-        pfp_path.unlink()
+    (cache_folder / "pfp_me.png").unlink(missing_ok=True)
 
     # Generate new picture cache
     picture = generate_user_pfp_cache(user_name)
@@ -2599,15 +2592,13 @@ def handle_character_picture_change(picture_path):
 
     if picture is not None:
         # Save to cache
-        picture.save(Path(f'{cache_folder}/pfp_character.png'), format='PNG')
+        picture.save(cache_folder / 'pfp_character.png', format='PNG')
         thumb = make_thumbnail(picture)
-        thumb.save(Path(f'{cache_folder}/pfp_character_thumb.png'), format='PNG')
+        thumb.save(cache_folder / 'pfp_character_thumb.png', format='PNG')
     else:
         # Remove cache files when picture is cleared
         for cache_file in ['pfp_character.png', 'pfp_character_thumb.png']:
-            cache_path = Path(f'{cache_folder}/{cache_file}')
-            if cache_path.exists():
-                cache_path.unlink()
+            (cache_folder / cache_file).unlink(missing_ok=True)
 
 
 def handle_mode_change(state):

From 16af11f8680180b24876ee77d5ce29fcd20cd8db Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 04:22:37 -0700
Subject: [PATCH 1493/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index ab6cc2e544..23cd09c5dd 100644
--- a/README.md
+++ b/README.md
@@ -24,7 +24,7 @@ A Gradio web UI for running Large Language Models locally. 100% private and offl
 ## Features
 
 - **Easy setup**: [Portable builds](https://github.com/oobabooga/text-generation-webui/releases) (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or a one-click installer for the full feature set.
-- **Multiple backends**: [llama.cpp](https://github.com/ggerganov/llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). Switch between backends and models without restarting.
+- **Multiple backends**: [llama.cpp](https://github.com/ggerganov/llama.cpp), [ik_llama.cpp](https://github.com/ikawrakow/ik_llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). Switch between backends and models without restarting.
 - **OpenAI/Anthropic-compatible API**: Chat, Completions, and Messages endpoints with tool-calling support. Use as a local drop-in replacement for the OpenAI/Anthropic APIs ([examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples)).
 - **Tool-calling**: Models can call custom functions during chat — web search, page fetching, math, and more. Each tool is a single `.py` file, easy to create and extend ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Tool-Calling-Tutorial)).
 - **Vision (multimodal)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).

From e0ad4e60df40378f9846e1fad20e337e4bb2cb8f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 09:57:07 -0700
Subject: [PATCH 1494/1701] UI: Fix tool buffer check truncating visible text
 at end of generation

---
 modules/chat.py         | 2 +-
 modules/tool_parsing.py | 9 ++++++++-
 2 files changed, 9 insertions(+), 2 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index e28d39632d..76b8694af3 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1183,7 +1183,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
     # visible text from before buffering started so raw markup doesn't flash
     # in the UI.  The internal text is left intact so the caller can still
     # parse tool calls from it.
-    if is_stream and _check_tool_markers and streaming_tool_buffer_check(output['internal'][-1][1], markers=_streaming_markers, tool_names=_tool_names, check_bare_names=_check_bare_names):
+    if is_stream and _check_tool_markers and streaming_tool_buffer_check(output['internal'][-1][1], markers=_streaming_markers, tool_names=_tool_names, check_bare_names=_check_bare_names, partial_match=False):
         output['visible'][-1][1] = _last_visible_before_tool_buffer or ''
 
     yield output
diff --git a/modules/tool_parsing.py b/modules/tool_parsing.py
index 919e523ab6..7fcf58b740 100644
--- a/modules/tool_parsing.py
+++ b/modules/tool_parsing.py
@@ -31,7 +31,7 @@ def get_tool_call_id() -> str:
 ]
 
 
-def streaming_tool_buffer_check(text, markers=None, tool_names=None, check_bare_names=False):
+def streaming_tool_buffer_check(text, markers=None, tool_names=None, check_bare_names=False, partial_match=True):
     '''
     Check whether streaming output should be withheld because it may
     contain tool-call markup.
@@ -43,6 +43,10 @@ def streaming_tool_buffer_check(text, markers=None, tool_names=None, check_bare_
         tool_names: List of tool function names.
         check_bare_names: Whether to do partial-prefix matching on tool
                           names (for models with unknown template format).
+        partial_match: Whether to check partial prefixes of markers/names.
+                       Set to False for end-of-generation checks where a
+                       partial prefix is just normal text, not an incomplete
+                       tool call.
     '''
     # Strip thinking blocks so tool-call syntax inside  doesn't
     # trigger false positives.
@@ -60,6 +64,9 @@ def streaming_tool_buffer_check(text, markers=None, tool_names=None, check_bare_
             if name + '{' in text or name + ' {' in text:
                 return True
 
+    if not partial_match:
+        return False
+
     # Partial-prefix matching: only for template-specific markers.
     for marker in (markers if markers is not None else TOOL_CALL_OPENING_MARKERS):
         for prefix_len in range(min(len(marker) - 1, len(text)), 0, -1):

From 9183dc444e6c62ae4e33a759e03d9b2b66a49bf2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 10:48:53 -0700
Subject: [PATCH 1495/1701] API: Fix loader args leaking between sequential
 model loads

---
 modules/api/models.py | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/modules/api/models.py b/modules/api/models.py
index b89397d3dd..e0bd21f3dc 100644
--- a/modules/api/models.py
+++ b/modules/api/models.py
@@ -47,7 +47,6 @@ def _load_model(data):
 
     unload_model()
     model_settings = get_model_metadata(model_name)
-    update_model_parameters(model_settings)
 
     # Update shared.args with custom model loading settings
     # Security: only allow keys that correspond to model loading
@@ -55,6 +54,16 @@ def _load_model(data):
     # flags like trust_remote_code or extra_flags to be set via the API.
     blocked_keys = {'extra_flags'}
     allowed_keys = set(loaders.list_model_elements()) - blocked_keys
+
+    # Reset all loader args to their startup values before applying new ones,
+    # so settings from a previous API load don't leak into this one.
+    # Include blocked keys in the reset (safe: restores startup value, not API-controlled).
+    for k in allowed_keys | blocked_keys:
+        if hasattr(shared.args, k) and hasattr(shared.original_args, k):
+            setattr(shared.args, k, getattr(shared.original_args, k))
+
+    update_model_parameters(model_settings)
+
     if args:
         for k in args:
             if k in allowed_keys and hasattr(shared.args, k):

From 2eef90a32346f5acea1771803460504294b9d9be Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 11:00:14 -0700
Subject: [PATCH 1496/1701] API: Remove deprecated "settings" parameter from
 model load endpoint

---
 modules/api/models.py | 14 +-------------
 modules/api/script.py | 20 ++++----------------
 modules/api/typing.py |  1 -
 3 files changed, 5 insertions(+), 30 deletions(-)

diff --git a/modules/api/models.py b/modules/api/models.py
index e0bd21f3dc..5dd7785050 100644
--- a/modules/api/models.py
+++ b/modules/api/models.py
@@ -1,5 +1,4 @@
 from modules import loaders, shared
-from modules.logging_colors import logger
 from modules.LoRA import add_lora_to_model
 from modules.models import load_model, unload_model
 from modules.models_settings import get_model_metadata, update_model_parameters
@@ -42,8 +41,7 @@ def model_info_dict(model_name: str) -> dict:
 
 def _load_model(data):
     model_name = data["model_name"]
-    args = data["args"]
-    settings = data["settings"]
+    args = data.get("args")
 
     unload_model()
     model_settings = get_model_metadata(model_name)
@@ -71,16 +69,6 @@ def _load_model(data):
 
     shared.model, shared.tokenizer = load_model(model_name)
 
-    # Update shared.settings with custom generation defaults
-    if settings:
-        for k in settings:
-            if k in shared.settings:
-                shared.settings[k] = settings[k]
-                if k == 'truncation_length':
-                    logger.info(f"CONTEXT LENGTH (UPDATED): {shared.settings['truncation_length']}")
-                elif k == 'instruction_template':
-                    logger.info(f"INSTRUCTION TEMPLATE (UPDATED): {shared.settings['instruction_template']}")
-
 
 def list_loras():
     return {'lora_names': get_available_loras()[1:]}
diff --git a/modules/api/script.py b/modules/api/script.py
index 85f4974f29..beed3d06e8 100644
--- a/modules/api/script.py
+++ b/modules/api/script.py
@@ -475,10 +475,8 @@ async def handle_list_models():
 @app.post("/v1/internal/model/load", dependencies=check_admin_key)
 async def handle_load_model(request_data: LoadModelRequest):
     '''
-    This endpoint is experimental and may change in the future.
-
-    The "args" parameter can be used to modify flags like "--load-in-4bit"
-    or "--n-gpu-layers" before loading a model. Example:
+    The "args" parameter can be used to modify loader flags before loading
+    a model. Example:
 
     ```
     "args": {
@@ -487,18 +485,8 @@ async def handle_load_model(request_data: LoadModelRequest):
     }
     ```
 
-    Note that those settings will remain after loading the model. So you
-    may need to change them back to load a second model.
-
-    The "settings" parameter is also a dict but with keys for the
-    shared.settings object. It can be used to modify the default instruction
-    template like this:
-
-    ```
-    "settings": {
-      "instruction_template": "Alpaca"
-    }
-    ```
+    Loader args are reset to their startup defaults between loads, so
+    settings from a previous load do not leak into the next one.
     '''
 
     try:
diff --git a/modules/api/typing.py b/modules/api/typing.py
index 1d486e8fb7..a758743e46 100644
--- a/modules/api/typing.py
+++ b/modules/api/typing.py
@@ -271,7 +271,6 @@ class ModelListResponse(BaseModel):
 class LoadModelRequest(BaseModel):
     model_name: str
     args: dict | None = None
-    settings: dict | None = None
 
 
 class LoraListResponse(BaseModel):

From 7fed60f90ad9a37a9f0656d054bb5553a0e8da13 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 18:29:36 -0700
Subject: [PATCH 1497/1701] UI: Improve the hover menu looks

---
 css/main.css       | 91 +++++++++++++++++++++++++++++++++++-----------
 js/main.js         | 44 +++++++++++-----------
 modules/ui_chat.py |  2 +-
 3 files changed, 92 insertions(+), 45 deletions(-)

diff --git a/css/main.css b/css/main.css
index 009b7c0a43..913576c551 100644
--- a/css/main.css
+++ b/css/main.css
@@ -735,7 +735,30 @@ audio {
 
 .hover-element {
     position: relative;
-    font-size: 24px;
+    padding-top: 4px;
+}
+
+#hover-element-button {
+    display: flex;
+    align-items: center;
+    justify-content: center;
+    width: 32px;
+    height: 32px;
+    border-radius: 0.5rem;
+    cursor: pointer;
+    color: gray;
+}
+
+#hover-element-button:hover {
+    background-color: var(--background-fill-secondary);
+}
+
+#hover-element-button svg {
+    color: inherit;
+}
+
+.dark #hover-element-button:hover {
+    background-color: var(--selected-item-color-dark);
 }
 
 .hover-menu {
@@ -743,27 +766,40 @@ audio {
     position: absolute;
     bottom: 100%;
     left: 0;
-    box-shadow: 0 2px 12px rgb(0 0 0 / 15%);
-    border-radius: 0.5rem;
+    background: white;
+    border: 1px solid rgba(0, 0, 0, 0.1);
+    box-shadow: 0 4px 16px rgb(0 0 0 / 12%), 0 1px 3px rgb(0 0 0 / 8%);
+    border-radius: 0.75rem;
     z-index: 10000;
     min-width: 330px;
     flex-direction: column;
-    overflow: hidden;
+    padding: 4px;
+}
+
+.hover-menu::before {
+    content: '';
+    position: absolute;
+    top: 100%;
+    left: 0;
+    width: 100%;
+    height: 8px;
+}
+
+.hover-menu > * {
+    border: none !important;
+    box-shadow: none !important;
 }
 
 .hover-menu button {
     width: 100%;
-    background: white !important;
-    border-radius: 0 !important;
+    background: transparent !important;
+    border: none !important;
+    border-radius: 0.5rem !important;
     justify-content: space-between;
     margin: 0 !important;
     height: 36px;
-    border-color: transparent !important;
-    transition: background-color 0.15s ease;
-}
-
-.hover-menu button:not(#clear-history-confirm) {
-    border-bottom: 0 !important;
+    font-weight: 500;
+    box-shadow: none !important;
 }
 
 .hover-menu button:hover {
@@ -775,19 +811,26 @@ audio {
 }
 
 #show-controls {
-    background-color: white;
-    border-color: transparent !important;
+    background-color: transparent;
+    border: none !important;
     height: 36px;
-    border-radius: 0;
-    border-bottom: 0 !important;
+    border-radius: 0.5rem;
     padding-top: 3px;
     padding-left: 4px;
     display: flex;
     font-weight: normal;
 }
 
+#show-controls:hover {
+    background-color: #dbeafe;
+}
+
 .dark #show-controls {
-    background-color: var(--darker-gray);
+    background-color: transparent;
+}
+
+.dark #show-controls:hover {
+    background-color: var(--selected-item-color-dark);
 }
 
 #show-controls label {
@@ -797,12 +840,12 @@ audio {
     width: 100%;
     padding-right: 12px;
     gap: 10px;
-    font-weight: 600;
+    font-weight: 500;
     color: var(--button-secondary-text-color);
 }
 
 #show-controls label input {
-    margin-top: 4px;
+    margin-top: 5px;
 }
 
 .transparent-substring {
@@ -817,6 +860,7 @@ audio {
     min-width: 0 !important;
     display: flex;
     flex-direction: column-reverse;
+    padding-left: 12px;
     padding-right: 20px;
     padding-bottom: 3px;
     flex-grow: 0 !important;
@@ -1208,9 +1252,14 @@ audio {
     color: #9ca3af;
 }
 
+.dark .hover-menu {
+    background: var(--darker-gray);
+    border-color: transparent;
+    box-shadow: 0 4px 16px rgb(0 0 0 / 40%);
+}
+
 .dark .hover-menu button {
-    border-color: var(--border-color-primary);
-    background-color: var(--darker-gray) !important;
+    background-color: transparent !important;
 }
 
 .dark #chat-controls,
diff --git a/js/main.js b/js/main.js
index cba4c903c1..918c85c15a 100644
--- a/js/main.js
+++ b/js/main.js
@@ -309,18 +309,19 @@ for (let i = 0; i < slimDropdownElements.length; i++) {
 // https://github.com/SillyTavern/SillyTavern/blob/6c8bd06308c69d51e2eb174541792a870a83d2d6/public/script.js
 //------------------------------------------------
 var buttonsInChat = document.querySelectorAll("#chat-tab #chat-buttons button, #chat-tab #chat-buttons #show-controls");
+var hoverContainer = document.getElementById("gr-hover-container");
 var button = document.getElementById("hover-element-button");
 var menu = document.getElementById("hover-menu");
 var istouchscreen = (navigator.maxTouchPoints > 0) || "ontouchstart" in document.documentElement;
 
 function showMenu() {
-  menu.style.display = "flex"; // Show the menu
+  menu.style.display = "flex";
 }
 
 function hideMenu() {
-  menu.style.display = "none"; // Hide the menu
+  menu.style.display = "none";
   if (!istouchscreen) {
-    document.querySelector("#chat-input textarea").focus(); // Focus on the chat input
+    document.querySelector("#chat-input textarea").focus();
   }
 }
 
@@ -329,7 +330,6 @@ if (buttonsInChat.length > 0) {
     const thisButton = buttonsInChat[i];
     menu.appendChild(thisButton);
 
-    // Only apply transformations to button elements
     if (thisButton.tagName.toLowerCase() === "button") {
       thisButton.addEventListener("click", () => {
         hideMenu();
@@ -339,7 +339,6 @@ if (buttonsInChat.length > 0) {
       const matches = buttonText.match(/(\(.*?\))/);
 
       if (matches && matches.length > 1) {
-        // Apply the transparent-substring class to the matched substring
         const substring = matches[1];
         const newText = buttonText.replace(substring, ` ${substring.slice(1, -1)}`);
         thisButton.innerHTML = newText;
@@ -348,16 +347,19 @@ if (buttonsInChat.length > 0) {
   }
 }
 
-function isMouseOverButtonOrMenu() {
-  return menu.matches(":hover") || button.matches(":hover");
-}
+var menuInteracting = false;
 
-button.addEventListener("mouseenter", function () {
+hoverContainer.addEventListener("mouseenter", function () {
   if (!istouchscreen) {
     showMenu();
   }
 });
 
+hoverContainer.addEventListener("mousedown", function () {
+  menuInteracting = true;
+  setTimeout(function () { menuInteracting = false; }, 300);
+});
+
 button.addEventListener("click", function () {
   if (menu.style.display === "flex") {
     hideMenu();
@@ -367,24 +369,20 @@ button.addEventListener("click", function () {
   }
 });
 
-// Delay to prevent menu hiding when the mouse leaves the button or menu
-function delayedHideMenu() {
-  setTimeout(function () {
-    if (!isMouseOverButtonOrMenu()) {
-      hideMenu();
-    }
-  }, 100);
-}
-
-// Add event listener for mouseleave on the button
-button.addEventListener("mouseleave", delayedHideMenu);
-// Add event listener for mouseleave on the menu
-menu.addEventListener("mouseleave", delayedHideMenu);
+hoverContainer.addEventListener("mouseleave", function () {
+  if (!istouchscreen) {
+    setTimeout(function () {
+      if (!hoverContainer.matches(":hover") && !menu.matches(":hover")) {
+        hideMenu();
+      }
+    }, 50);
+  }
+});
 
 // Add event listener for click anywhere in the document
 document.addEventListener("click", function (event) {
   // Check if the click is outside the button/menu and the menu is visible
-  if (!isMouseOverButtonOrMenu() && menu.style.display === "flex") {
+  if (!menuInteracting && !event.target.closest("#gr-hover-container") && menu.style.display === "flex") {
     hideMenu();
   }
 
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 10d05f659c..d96522535a 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -52,7 +52,7 @@ def create_ui():
                 shared.gradio['html_display'] = gr.HTML(value=chat_html_wrapper({'internal': [], 'visible': [], 'metadata': {}}, '', '', 'chat', 'cai-chat', '')['html'], visible=True)
                 with gr.Row(elem_id="chat-input-row"):
                     with gr.Column(scale=1, elem_id='gr-hover-container'):
-                        gr.HTML(value='
            ☰
            ', elem_id='gr-hover')
+                        gr.HTML(value='
            ', elem_id='gr-hover')
 
                     with gr.Column(scale=10, elem_id='chat-input-container'):
                         shared.gradio['textbox'] = gr.MultimodalTextbox(label='', placeholder='Send a message', file_types=['text', '.pdf', 'image'], file_count="multiple", elem_id='chat-input', elem_classes=['add_scrollbar'])

From ffea8f282e3a2f798d7bf5531be278754b47da21 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 18:53:13 -0700
Subject: [PATCH 1498/1701] UI: Improve message text contrast

---
 css/html_instruct_style.css |  2 +-
 css/main.css                | 17 +++++++++++++----
 modules/ui.py               |  2 +-
 3 files changed, 15 insertions(+), 6 deletions(-)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index 458feafc1e..aa61f33bdc 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -13,7 +13,7 @@
     line-height: 28px !important;
 }
 
-.dark .chat .message-body :is(p,li,h1,h2,h3,h4,h5,h6),
+.dark .chat .message-body :is(p,li),
 .dark .chat .message-body em:not(:is(h1,h2,h3,h4,h5,h6,b,strong) em),
 .dark .chat .message-body q:not(:is(h1,h2,h3,h4,h5,h6,b,strong) q) {
     color: #d1d5db !important;
diff --git a/css/main.css b/css/main.css
index 913576c551..d06d29056f 100644
--- a/css/main.css
+++ b/css/main.css
@@ -436,15 +436,24 @@ audio {
 .dark .message-body h4,
 .dark .message-body h5,
 .dark .message-body h6 {
-    color: white !important;
+    color: #e8e8e8 !important;
 }
 
 .dark .message-body blockquote {
     border-left-color: rgb(255 255 255 / 30%);
 }
 
+.message-body h1,
+.message-body h2,
+.message-body h3,
+.message-body h4,
+.message-body h5,
+.message-body h6 {
+    color: #1a1a1a;
+}
+
 .message-body h1 {
-    font-weight: 800;
+    font-weight: 700;
     font-size: 2.25em;
     margin-top: 0;
     margin-bottom: 0.8888889em;
@@ -476,13 +485,13 @@ audio {
 }
 
 .message-body h5 {
-    font-weight: normal;
+    font-weight: 600;
     font-size: 1em;
     margin: 0;
 }
 
 .message-body h6 {
-    font-weight: normal;
+    font-weight: 600;
     font-size: 1em;
     margin: 0;
 }
diff --git a/modules/ui.py b/modules/ui.py
index 02b5a9fbfe..73072cbe8a 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -75,7 +75,7 @@
         background_fill_primary_dark='var(--darker-gray, #1C1C1D)',
         body_background_fill="white",
         block_background_fill="transparent",
-        body_text_color='rgb(64, 64, 64)',
+        body_text_color='#1a1a1a',
         button_secondary_background_fill="white",
         button_secondary_border_color="var(--border-color-primary)",
         block_title_text_color='*body_text_color',

From 41bce3f4dee83ede8ba05a3f3cdab9e729ec0979 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 19:07:23 -0700
Subject: [PATCH 1499/1701] UI: Improve scrollbars style

---
 css/main.css | 26 ++++++++++++--------------
 1 file changed, 12 insertions(+), 14 deletions(-)

diff --git a/css/main.css b/css/main.css
index d06d29056f..c54367e6f2 100644
--- a/css/main.css
+++ b/css/main.css
@@ -246,8 +246,8 @@ button {
 
 .pretty_scrollbar::-webkit-scrollbar,
 #image-history-gallery > :nth-child(2)::-webkit-scrollbar {
-    width: 8px;
-    height: 8px;
+    width: 7px;
+    height: 7px;
 }
 
 .pretty_scrollbar::-webkit-scrollbar-track,
@@ -260,7 +260,7 @@ button {
 #image-history-gallery > :nth-child(2)::-webkit-scrollbar-thumb,
 #image-history-gallery > :nth-child(2)::-webkit-scrollbar-thumb:hover {
     background: var(--neutral-300);
-    border-radius: 30px;
+    border-radius: 9999px;
 }
 
 .dark .pretty_scrollbar::-webkit-scrollbar-thumb,
@@ -268,18 +268,17 @@ button {
 .dark #image-history-gallery > :nth-child(2)::-webkit-scrollbar-thumb,
 .dark #image-history-gallery > :nth-child(2)::-webkit-scrollbar-thumb:hover {
     background: rgb(255 255 255 / 6.25%);
-    border-radius: 30px;
+    border-radius: 9999px;
 }
 
 .pretty_scrollbar::-webkit-resizer,
 #image-history-gallery > :nth-child(2)::-webkit-resizer {
-    background: #d2d2d8;
+    background: transparent;
 }
 
 .dark .pretty_scrollbar::-webkit-resizer,
 .dark #image-history-gallery > :nth-child(2)::-webkit-resizer {
-    background: rgb(255 255 255 / 10%);
-    border-radius: 10px;
+    background: transparent;
 }
 
 .pretty_scrollbar::-webkit-scrollbar-corner,
@@ -599,7 +598,7 @@ audio {
 }
 
 #chat-input textarea::-webkit-scrollbar {
-    width: 8px;
+    width: 7px;
 }
 
 #chat-input textarea::-webkit-scrollbar-track {
@@ -608,7 +607,7 @@ audio {
 
 #chat-input textarea::-webkit-scrollbar-thumb {
     background: var(--neutral-300);
-    border-radius: 30px;
+    border-radius: 9999px;
 }
 
 .dark #chat-input textarea::-webkit-scrollbar-thumb {
@@ -869,7 +868,6 @@ audio {
     min-width: 0 !important;
     display: flex;
     flex-direction: column-reverse;
-    padding-left: 12px;
     padding-right: 20px;
     padding-bottom: 3px;
     flex-grow: 0 !important;
@@ -2000,8 +1998,8 @@ thead + tbody tr:first-child th { border-top: 1px solid; }
 
 /* Pretty scrollbar for the tools list */
 #tools-group .wrap::-webkit-scrollbar {
-    width: 8px;
-    height: 8px;
+    width: 7px;
+    height: 7px;
 }
 
 #tools-group .wrap::-webkit-scrollbar-track {
@@ -2011,13 +2009,13 @@ thead + tbody tr:first-child th { border-top: 1px solid; }
 #tools-group .wrap::-webkit-scrollbar-thumb,
 #tools-group .wrap::-webkit-scrollbar-thumb:hover {
     background: var(--neutral-300);
-    border-radius: 30px;
+    border-radius: 9999px;
 }
 
 .dark #tools-group .wrap::-webkit-scrollbar-thumb,
 .dark #tools-group .wrap::-webkit-scrollbar-thumb:hover {
     background: rgb(255 255 255 / 6.25%);
-    border-radius: 30px;
+    border-radius: 9999px;
 }
 
 #tools-group .wrap::-webkit-scrollbar-corner {

From 8cb7fe9c470101d07f80b236a1d34b906bcdb25a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 19:14:17 -0700
Subject: [PATCH 1500/1701] UI: Improve message action icon visibility in light
 mode

---
 css/main.css | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/css/main.css b/css/main.css
index c54367e6f2..41b0830801 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1428,12 +1428,11 @@ audio {
 }
 
 .footer-button svg {
-    stroke: rgb(156 163 175);
-    transition: stroke 0.2s;
+    stroke: rgb(107 114 128);
 }
 
 .footer-button:hover svg {
-    stroke: rgb(107 114 128);
+    stroke: rgb(64 64 64);
 }
 
 .dark .footer-button svg {

From 1b403a4ffab0833cdce527360f445a0003c7ea41 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 19:33:05 -0700
Subject: [PATCH 1501/1701] UI: Fix inline LaTeX rendering by protecting $...$
 from markdown (closes #7423)

---
 modules/html_generator.py | 32 +++++++++++++++++---------------
 1 file changed, 17 insertions(+), 15 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 8f3f261f9a..8dd46850ac 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -185,28 +185,29 @@ def process_markdown_content(string):
     if not string:
         return ""
 
-    # Define unique placeholders for LaTeX asterisks and underscores
+    # Define unique placeholders for LaTeX characters that conflict with markdown
     LATEX_ASTERISK_PLACEHOLDER = "LATEXASTERISKPLACEHOLDER"
     LATEX_UNDERSCORE_PLACEHOLDER = "LATEXUNDERSCOREPLACEHOLDER"
+    LATEX_PIPE_PLACEHOLDER = "LATEXPIPEPLACEHOLDER"
+
+    def protect_latex_content(content):
+        """Protect markdown-sensitive characters inside LaTeX."""
+        content = content.replace('*', LATEX_ASTERISK_PLACEHOLDER)
+        content = content.replace('_', LATEX_UNDERSCORE_PLACEHOLDER)
+        content = content.replace('|', LATEX_PIPE_PLACEHOLDER)
+        return content
 
     def protect_asterisks_underscores_in_latex(match):
-        """A replacer function for re.sub to protect asterisks and underscores in multiple LaTeX formats."""
+        """A replacer function for re.sub to protect markdown-sensitive characters in multiple LaTeX formats."""
         # Check which delimiter group was captured
         if match.group(1) is not None:  # Content from $$...$$
-            content = match.group(1)
-            modified_content = content.replace('*', LATEX_ASTERISK_PLACEHOLDER)
-            modified_content = modified_content.replace('_', LATEX_UNDERSCORE_PLACEHOLDER)
-            return f'{modified_content}'
+            return protect_latex_content(match.group(1))
         elif match.group(2) is not None:  # Content from \[...\]
-            content = match.group(2)
-            modified_content = content.replace('*', LATEX_ASTERISK_PLACEHOLDER)
-            modified_content = modified_content.replace('_', LATEX_UNDERSCORE_PLACEHOLDER)
-            return f'\\[{modified_content}\\]'
+            return f'\\[{protect_latex_content(match.group(2))}\\]'
         elif match.group(3) is not None:  # Content from \(...\)
-            content = match.group(3)
-            modified_content = content.replace('*', LATEX_ASTERISK_PLACEHOLDER)
-            modified_content = modified_content.replace('_', LATEX_UNDERSCORE_PLACEHOLDER)
-            return f'\\({modified_content}\\)'
+            return f'\\({protect_latex_content(match.group(3))}\\)'
+        elif match.group(4) is not None:  # Content from $...$
+            return f'${protect_latex_content(match.group(4).strip())}$'
 
         return match.group(0)  # Fallback
 
@@ -240,7 +241,7 @@ def protect_asterisks_underscores_in_latex(match):
     string = re.sub(r"(.)```", r"\1\n```", string)
 
     # Protect asterisks and underscores within all LaTeX blocks before markdown conversion
-    latex_pattern = re.compile(r'((?:^|[\r\n\s])\$\$[^`]*?\$\$)|\\\[(.*?)\\\]|\\\((.*?)\\\)',
+    latex_pattern = re.compile(r'((?:^|[\r\n\s])\$\$[^`]*?\$\$)|\\\[(.*?)\\\]|\\\((.*?)\\\)|(?
     html_output = re.sub(r'\s*            ', '            ', html_output)

From 0c033caf0ef79838178238912df29cc47bb10ba3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 20:09:28 -0700
Subject: [PATCH 1502/1701] UI: Reduce spacing above chat input

---
 css/main.css | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index 41b0830801..43e9684f3e 100644
--- a/css/main.css
+++ b/css/main.css
@@ -893,7 +893,7 @@ audio {
 }
 
 #chat-input-row {
-    padding: 1rem;
+    padding: 0.5rem 1rem 1rem;
 }
 
 #chat-col {

From dfd8ec9c4992f801304fc7efb89e7e47355fd18e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 20:13:20 -0700
Subject: [PATCH 1503/1701] UI: Make accordion outline styling global

---
 css/main.css             | 4 ++--
 modules/training.py      | 4 ++--
 modules/ui_model_menu.py | 6 +++---
 3 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/css/main.css b/css/main.css
index 43e9684f3e..459c9fab43 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1443,12 +1443,12 @@ audio {
     stroke: rgb(209 213 219);
 }
 
-.tgw-accordion {
+.block:has(> .label-wrap) {
     padding: 10px 12px !important;
     border: 1px solid #d2d2d8;
 }
 
-.dark .tgw-accordion {
+.dark .block:has(> .label-wrap) {
     border: 1px solid var(--border-color-dark);
 }
 
diff --git a/modules/training.py b/modules/training.py
index 145353c634..bca4f02ebe 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -52,7 +52,7 @@ def create_ui():
                         with gr.Column():
                             always_override = gr.Checkbox(label='Override Existing Files', value=False, info='If the name is the same, checking will replace the existing file, and unchecking will load and continue from it (the rank must be the same).', elem_classes=['no-background'])
 
-                    with gr.Accordion(label='Target Modules', open=False, elem_classes='tgw-accordion'):
+                    with gr.Accordion(label='Target Modules', open=False):
                         gr.Markdown("Selects which modules to target in training. Targeting more modules is closer to a full fine-tune at the cost of increased VRAM and adapter size.")
                         all_linear = gr.Checkbox(label='Target all linear layers', value=True, info='Targets every nn.Linear layer except lm_head. Works for any model architecture. When checked, the individual module checkboxes below are ignored.', elem_classes=['no-background'])
                         with gr.Row():
@@ -87,7 +87,7 @@ def create_ui():
                             with gr.Row():
                                 lr_scheduler_type = gr.Dropdown(label='LR Scheduler', value='cosine', choices=['linear', 'constant', 'constant_with_warmup', 'cosine', 'cosine_with_restarts', 'polynomial', 'inverse_sqrt'], info='Learning rate scheduler - defines how the learning rate changes over time. "Constant" means never change, "linear" means to go in a straight line from the learning rate down to 0, cosine follows a curve, etc.', elem_classes=['slim-dropdown'])
 
-                    with gr.Accordion(label='Advanced Options', open=False, elem_classes='tgw-accordion'):
+                    with gr.Accordion(label='Advanced Options', open=False):
                         with gr.Row():
                             with gr.Column():
                                 optimizer = gr.Dropdown(label='Optimizer', value='adamw_torch', choices=['adamw_hf', 'adamw_torch', 'adamw_torch_fused', 'adamw_torch_xla', 'adamw_apex_fused', 'adafactor', 'adamw_bnb_8bit', 'adamw_anyprecision', 'sgd', 'adagrad'], info='Optimizer algorithm. adamw_torch is the standard choice. adamw_bnb_8bit uses less VRAM. adafactor is memory-efficient for large models.', elem_classes=['slim-dropdown'])
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 16505afa55..243079a02e 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -67,13 +67,13 @@ def create_ui():
                             )
 
                             # Multimodal
-                            with gr.Accordion("Multimodal (vision)", open=False, elem_classes='tgw-accordion') as shared.gradio['mmproj_accordion']:
+                            with gr.Accordion("Multimodal (vision)", open=False) as shared.gradio['mmproj_accordion']:
                                 with gr.Row():
                                     shared.gradio['mmproj'] = gr.Dropdown(label="mmproj file", choices=utils.get_available_mmproj(), value=lambda: shared.args.mmproj or 'None', elem_classes='slim-dropdown', info=f'Select a file that matches your model. Must be placed in {shared.user_data_dir}/mmproj/', interactive=not mu)
                                     ui.create_refresh_button(shared.gradio['mmproj'], lambda: None, lambda: {'choices': utils.get_available_mmproj()}, 'refresh-button', interactive=not mu)
 
                             # Speculative decoding
-                            with gr.Accordion("Speculative decoding", open=False, elem_classes='tgw-accordion') as shared.gradio['speculative_decoding_accordion']:
+                            with gr.Accordion("Speculative decoding", open=False) as shared.gradio['speculative_decoding_accordion']:
                                 shared.gradio['draft_max'] = gr.Number(label="draft-max", precision=0, step=1, value=shared.args.draft_max, info='Maximum number of tokens to draft for speculative decoding. Recommended: 4 for draft model, 64 for n-gram.')
 
                                 gr.Markdown('#### Draft model')
@@ -92,7 +92,7 @@ def create_ui():
                                 shared.gradio['spec_ngram_min_hits'] = gr.Number(label="spec-ngram-min-hits", precision=0, step=1, value=shared.args.spec_ngram_min_hits, info='Minimum n-gram hits for ngram-map speculative decoding.', visible=shared.args.spec_type != 'none')
 
                     gr.Markdown("## Other options")
-                    with gr.Accordion("See more options", open=False, elem_classes='tgw-accordion'):
+                    with gr.Accordion("See more options", open=False):
                         with gr.Row():
                             with gr.Column():
                                 shared.gradio['parallel'] = gr.Slider(label="parallel", minimum=1, step=1, maximum=64, value=shared.args.parallel, info='Number of parallel request slots for the API. The context size is divided equally among slots. For example, to have 4 slots with 8192 context each, set ctx_size to 32768.')

From ee917cd5edfc3b192d4a3147001f0c1752a3e354 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 20:35:27 -0700
Subject: [PATCH 1504/1701] UI: Make table and hr borders more subtle

---
 css/html_instruct_style.css |  9 ---------
 css/main.css                | 16 +++++++++++++---
 2 files changed, 13 insertions(+), 12 deletions(-)

diff --git a/css/html_instruct_style.css b/css/html_instruct_style.css
index aa61f33bdc..fc20d1666e 100644
--- a/css/html_instruct_style.css
+++ b/css/html_instruct_style.css
@@ -19,15 +19,6 @@
     color: #d1d5db !important;
 }
 
-.chat .message-body :is(th, td),
-.prose hr {
-    border-color: #40404096 !important;
-}
-
-.dark .chat .message-body :is(th, td),
-.dark .prose hr {
-    border-color: rgb(255 255 255 / 30%) !important;
-}
 
 .chat .message-body :is(p, ul, ol) {
     margin: 1.25em 0 !important;
diff --git a/css/main.css b/css/main.css
index 459c9fab43..d9dc5d2ee2 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1958,14 +1958,24 @@ table, tr, td, th, thead {
     border: 0;
 }
 
+.prose hr {
+    border-color: var(--border-color-primary);
+}
+
 td + td,
-th + th { border-left: 1px solid; }
+th + th {
+    border-left: 1px solid var(--border-color-primary) !important;
+}
 
 tr + tr td,
-tr + tr th { border-top: 1px solid; }
+tr + tr th {
+    border-top: 1px solid var(--border-color-primary) !important;
+}
 
 thead + tbody tr:first-child td,
-thead + tbody tr:first-child th { border-top: 1px solid; }
+thead + tbody tr:first-child th {
+    border-top: 1px solid var(--border-color-primary) !important;
+}
 
 /* ------------------------------------------------
    Tools CheckboxGroup - vertical DragDrop-like style

From e8b31c063a3a5d1486dba2969b116835aa6a56bf Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 20:38:31 -0700
Subject: [PATCH 1505/1701] UI: Soften message action icons in light mode

---
 css/main.css | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/css/main.css b/css/main.css
index d9dc5d2ee2..a59e08ce3c 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1428,11 +1428,11 @@ audio {
 }
 
 .footer-button svg {
-    stroke: rgb(107 114 128);
+    stroke: rgb(140 140 148);
 }
 
 .footer-button:hover svg {
-    stroke: rgb(64 64 64);
+    stroke: rgb(107 114 128);
 }
 
 .dark .footer-button svg {

From 1f49a64e1ac1b2e700146956ac3dc17794d53243 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 20:44:37 -0700
Subject: [PATCH 1506/1701] UI: Improve blockquote border width and color

---
 css/main.css | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/css/main.css b/css/main.css
index a59e08ce3c..6685ab3461 100644
--- a/css/main.css
+++ b/css/main.css
@@ -438,8 +438,9 @@ audio {
     color: #e8e8e8 !important;
 }
 
-.dark .message-body blockquote {
-    border-left-color: rgb(255 255 255 / 30%);
+.message-body blockquote {
+    border-left-width: 4px;
+    border-left-color: var(--border-color-primary);
 }
 
 .message-body h1,

From 91f9b01516ff50bd35477ccccff9b53a03041cf8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 21:13:20 -0700
Subject: [PATCH 1507/1701] UI: Minor change

---
 css/main.css | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/css/main.css b/css/main.css
index 6685ab3461..7f47a3aa1b 100644
--- a/css/main.css
+++ b/css/main.css
@@ -642,6 +642,10 @@ audio {
     background: transparent;
 }
 
+#chat-input .thumbnails {
+    padding-top: 3px;
+}
+
 .chat-input-positioned {
     max-width: 54rem;
     left: 50%;

From 9805ddcde95f75bb1de100553dd3b604a4a6537c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 21:34:09 -0700
Subject: [PATCH 1508/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 17 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index b7a5ca97a3..9f83830aa8 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -31,8 +31,8 @@ tqdm
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 2c6275852e..b4b8386ee9 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 7e3fc35fb4..41ee6a6047 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 2603201db7..8be2f55e23 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index fe3bf3ba7a..d7f1bf132f 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index acae301eab..7b331f9629 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -28,8 +28,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 56795843ea..b467cf2627 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index abaa13383d..4eca16e1db 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index b22a03d918..55f8d3f8fa 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 97c5903ca0..54e8f350dd 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 57e92f7443..f073a6147c 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 1f7d27a7bc..8cd40f39dd 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index 65f6a004e4..fbb9125d81 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index 0a82adb702..59fcfae11e 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index 3d812045b7..ffdbe56843 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 91bef10b07..4a47b1f05b 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 7c61f0cc78..97abd9333e 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -14,8 +14,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio-4.37.2+custom.18-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.18/gradio_client-1.0.2+custom.18-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From c63a79ee4871178aa4d7b7f570e5e9d45b0280de Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 23:15:14 -0700
Subject: [PATCH 1509/1701] Image generation: Embed generation metadata in API
 image responses

---
 modules/api/images.py          | 23 +++++++++++++++++------
 modules/ui_image_generation.py |  3 +++
 2 files changed, 20 insertions(+), 6 deletions(-)

diff --git a/modules/api/images.py b/modules/api/images.py
index 9570453593..dde7d336c2 100644
--- a/modules/api/images.py
+++ b/modules/api/images.py
@@ -4,8 +4,11 @@
 
 import base64
 import io
+import json
 import time
 
+from PIL.PngImagePlugin import PngInfo
+
 from .errors import ServiceUnavailableError
 from modules import shared
 
@@ -15,7 +18,7 @@ def generations(request):
     Generate images using the loaded diffusion model.
     Returns dict with 'created' timestamp and 'data' list of images.
     """
-    from modules.ui_image_generation import generate
+    from modules.ui_image_generation import build_generation_metadata, generate
 
     if shared.image_model is None:
         raise ServiceUnavailableError("No image model loaded. Load a model via the UI first.")
@@ -46,10 +49,18 @@ def generations(request):
     if not images:
         raise ServiceUnavailableError("Image generation failed or produced no images.")
 
-    # Build response
+    # Build response with per-batch metadata (seed increments per batch)
+    base_seed = state.get('image_seed_resolved', state['image_seed'])
+    batch_size = int(state['image_batch_size'])
+
     resp = {'created': int(time.time()), 'data': []}
-    for img in images:
-        b64 = _image_to_base64(img)
+    for idx, img in enumerate(images):
+        batch_seed = base_seed + idx // batch_size
+        metadata = build_generation_metadata(state, batch_seed)
+        metadata_json = json.dumps(metadata, ensure_ascii=False)
+        png_info = PngInfo()
+        png_info.add_text("image_gen_settings", metadata_json)
+        b64 = _image_to_base64(img, png_info)
 
         image_obj = {'revised_prompt': request.prompt}
 
@@ -63,7 +74,7 @@ def generations(request):
     return resp
 
 
-def _image_to_base64(image) -> str:
+def _image_to_base64(image, png_info=None) -> str:
     buffered = io.BytesIO()
-    image.save(buffered, format="PNG")
+    image.save(buffered, format="PNG", pnginfo=png_info)
     return base64.b64encode(buffered.getvalue()).decode('utf-8')
diff --git a/modules/ui_image_generation.py b/modules/ui_image_generation.py
index 1efb247936..727aa7b149 100644
--- a/modules/ui_image_generation.py
+++ b/modules/ui_image_generation.py
@@ -798,6 +798,9 @@ def generate(state, save_images=True):
         if seed == -1:
             seed = random.randint(0, 2**32 - 1)
 
+        # Store resolved seed back so callers (e.g. API) can access it
+        state['image_seed_resolved'] = seed
+
         device = get_device()
         if device is None:
             device = "cpu"

From 544fcb0b7f0344fac249005f869b02110da69738 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 23:29:57 -0700
Subject: [PATCH 1510/1701] Simplify modules/image_models.py

---
 modules/image_models.py | 67 ++++++++++++++---------------------------
 1 file changed, 22 insertions(+), 45 deletions(-)

diff --git a/modules/image_models.py b/modules/image_models.py
index 290aaf1980..eed8783c8c 100644
--- a/modules/image_models.py
+++ b/modules/image_models.py
@@ -10,72 +10,49 @@ def get_quantization_config(quant_method):
     Get the appropriate quantization config based on the selected method.
     Applies quantization to both the transformer and the text_encoder.
     """
+    if quant_method == 'none' or not quant_method:
+        return None
+
     import torch
-    # Import BitsAndBytesConfig from BOTH libraries to be safe
     from diffusers import BitsAndBytesConfig as DiffusersBnBConfig
     from diffusers import TorchAoConfig
     from diffusers.quantizers import PipelineQuantizationConfig
     from transformers import BitsAndBytesConfig as TransformersBnBConfig
 
-    if quant_method == 'none' or not quant_method:
-        return None
+    torchao_methods = {
+        'torchao-int8wo': 'int8wo',
+        'torchao-fp4': 'fp4_e2m1',
+        'torchao-float8wo': 'float8wo',
+    }
 
-    # Bitsandbytes 8-bit quantization
-    elif quant_method == 'bnb-8bit':
+    if quant_method == 'bnb-8bit':
         return PipelineQuantizationConfig(
             quant_mapping={
-                "transformer": DiffusersBnBConfig(
-                    load_in_8bit=True
-                ),
-                "text_encoder": TransformersBnBConfig(
-                    load_in_8bit=True
-                )
+                "transformer": DiffusersBnBConfig(load_in_8bit=True),
+                "text_encoder": TransformersBnBConfig(load_in_8bit=True)
             }
         )
 
-    # Bitsandbytes 4-bit quantization
     elif quant_method == 'bnb-4bit':
-        return PipelineQuantizationConfig(
-            quant_mapping={
-                "transformer": DiffusersBnBConfig(
-                    load_in_4bit=True,
-                    bnb_4bit_quant_type="nf4",
-                    bnb_4bit_compute_dtype=torch.bfloat16,
-                    bnb_4bit_use_double_quant=True
-                ),
-                "text_encoder": TransformersBnBConfig(
-                    load_in_4bit=True,
-                    bnb_4bit_quant_type="nf4",
-                    bnb_4bit_compute_dtype=torch.bfloat16,
-                    bnb_4bit_use_double_quant=True
-                )
-            }
+        bnb_4bit_kwargs = dict(
+            load_in_4bit=True,
+            bnb_4bit_quant_type="nf4",
+            bnb_4bit_compute_dtype=torch.bfloat16,
+            bnb_4bit_use_double_quant=True
         )
-
-    # torchao int8 weight-only
-    elif quant_method == 'torchao-int8wo':
-        return PipelineQuantizationConfig(
-            quant_mapping={
-                "transformer": TorchAoConfig("int8wo"),
-                "text_encoder": TorchAoConfig("int8wo")
-            }
-        )
-
-    # torchao fp4 (e2m1)
-    elif quant_method == 'torchao-fp4':
         return PipelineQuantizationConfig(
             quant_mapping={
-                "transformer": TorchAoConfig("fp4_e2m1"),
-                "text_encoder": TorchAoConfig("fp4_e2m1")
+                "transformer": DiffusersBnBConfig(**bnb_4bit_kwargs),
+                "text_encoder": TransformersBnBConfig(**bnb_4bit_kwargs)
             }
         )
 
-    # torchao float8 weight-only
-    elif quant_method == 'torchao-float8wo':
+    elif quant_method in torchao_methods:
+        ao_type = torchao_methods[quant_method]
         return PipelineQuantizationConfig(
             quant_mapping={
-                "transformer": TorchAoConfig("float8wo"),
-                "text_encoder": TorchAoConfig("float8wo")
+                "transformer": TorchAoConfig(ao_type),
+                "text_encoder": TorchAoConfig(ao_type)
             }
         )
 

From 422f42ca7faa1d0834b1b503e87d605ad55f1ef8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 4 Apr 2026 23:51:15 -0700
Subject: [PATCH 1511/1701] Pre-compile LaTeX regex in html_generator.py

---
 modules/html_generator.py | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 8dd46850ac..e3ebea8dbf 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -14,6 +14,13 @@
 from modules.sane_markdown_lists import SaneListExtension
 from modules.utils import get_available_chat_styles
 
+# Pre-compiled regex for protecting markdown-sensitive characters inside LaTeX.
+# Covers $$...$$, \[...\], \(...\), and inline $...$ (when content contains \\).
+_LATEX_PATTERN = re.compile(
+    r'((?:^|[\r\n\s])\$\$[^`]*?\$\$)|\\\[(.*?)\\\]|\\\((.*?)\\\)|(?
Date: Sun, 5 Apr 2026 05:55:39 -0700
Subject: [PATCH 1512/1701] Fix "address already in use" on server restart
 (Linux/macOS)

---
 modules/api/script.py       | 26 ++++++++++++++++++++++++--
 modules/llama_cpp_server.py |  1 +
 2 files changed, 25 insertions(+), 2 deletions(-)

diff --git a/modules/api/script.py b/modules/api/script.py
index beed3d06e8..14e2d03a58 100644
--- a/modules/api/script.py
+++ b/modules/api/script.py
@@ -591,9 +591,31 @@ def run_server():
     if shared.args.admin_key and shared.args.admin_key != shared.args.api_key:
         logger.info(f'OpenAI API admin key (for loading/unloading models):\n\n{shared.args.admin_key}\n')
 
-    # Start server
+    # Use SO_REUSEADDR to avoid "address already in use" after restart
     logging.getLogger("uvicorn.error").propagate = False
-    uvicorn.run(app, host=server_addrs, port=port, ssl_certfile=ssl_certfile, ssl_keyfile=ssl_keyfile, access_log=False)
+    sockets = []
+    try:
+        for addr in server_addrs:
+            family = socket.AF_INET6 if ':' in addr else socket.AF_INET
+            sock = socket.socket(family, socket.SOCK_STREAM)
+            sock.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
+            if family == socket.AF_INET6:
+                sock.setsockopt(socket.IPPROTO_IPV6, socket.IPV6_V6ONLY, 1)
+            sock.bind((addr.strip('[]'), port))
+            sock.listen(socket.SOMAXCONN)
+            sockets.append(sock)
+    except Exception:
+        for s in sockets:
+            s.close()
+        raise
+
+    config = uvicorn.Config(app, ssl_certfile=ssl_certfile, ssl_keyfile=ssl_keyfile, access_log=False)
+    server = uvicorn.Server(config)
+    try:
+        server.run(sockets=sockets)
+    finally:
+        for s in sockets:
+            s.close()
 
 
 _server_started = False
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 3408046609..c01f5d5bf5 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -373,6 +373,7 @@ def _is_port_available(self, port):
         """Check if a port is available for use."""
         with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
             try:
+                s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
                 s.bind(('', port))
                 return True
             except OSError:

From f8db23b36286b09155e08beaa07a5797c879c7ef Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 5 Apr 2026 17:12:28 -0700
Subject: [PATCH 1513/1701] Call ik portable build folders
 text-generation-webui-ik-version

---
 .github/workflows/build-portable-release-ik-cuda.yml | 12 ++++++------
 .github/workflows/build-portable-release-ik.yml      | 12 ++++++------
 2 files changed, 12 insertions(+), 12 deletions(-)

diff --git a/.github/workflows/build-portable-release-ik-cuda.yml b/.github/workflows/build-portable-release-ik-cuda.yml
index 331a7653f7..a336a1cb9f 100644
--- a/.github/workflows/build-portable-release-ik-cuda.yml
+++ b/.github/workflows/build-portable-release-ik-cuda.yml
@@ -102,8 +102,8 @@ jobs:
             VERSION_CLEAN="${{ inputs.version }}"
             VERSION_CLEAN="${VERSION_CLEAN#v}"
             cd ..
-            cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
-            cd "text-generation-webui-${VERSION_CLEAN}"
+            cp -r text-generation-webui "text-generation-webui-ik-${VERSION_CLEAN}"
+            cd "text-generation-webui-ik-${VERSION_CLEAN}"
 
             # Remove extensions that need additional requirements
             allowed=("character_bias" "gallery" "sd_api_pictures")
@@ -133,10 +133,10 @@ jobs:
             echo "Downloading Python for $PLATFORM..."
             curl -L -o python-build.tar.gz "$PYTHON_URL"
             tar -xzf python-build.tar.gz
-            mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
+            mv python "text-generation-webui-ik-${VERSION_CLEAN}/portable_env"
 
             # 3. Prepare requirements file based on CUDA version
-            cd "text-generation-webui-${VERSION_CLEAN}"
+            cd "text-generation-webui-ik-${VERSION_CLEAN}"
             if [[ "$CUDA_VERSION" == "13.1" ]]; then
                 REQ_FILE="requirements/portable/requirements_ik_cuda131.txt"
             else
@@ -158,11 +158,11 @@ jobs:
             if [[ "$RUNNER_OS" == "Windows" ]]; then
                 ARCHIVE_NAME="textgen-portable-ik-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
                 echo "Creating archive: $ARCHIVE_NAME"
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
+                powershell -Command "Compress-Archive -Path text-generation-webui-ik-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
             else
                 ARCHIVE_NAME="textgen-portable-ik-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.tar.gz"
                 echo "Creating archive: $ARCHIVE_NAME"
-                tar czf "$ARCHIVE_NAME" "text-generation-webui-${VERSION_CLEAN}"
+                tar czf "$ARCHIVE_NAME" "text-generation-webui-ik-${VERSION_CLEAN}"
             fi
 
       - name: Upload files to a GitHub release
diff --git a/.github/workflows/build-portable-release-ik.yml b/.github/workflows/build-portable-release-ik.yml
index bf54eb0e88..5eaf7c8671 100644
--- a/.github/workflows/build-portable-release-ik.yml
+++ b/.github/workflows/build-portable-release-ik.yml
@@ -101,8 +101,8 @@ jobs:
             VERSION_CLEAN="${{ inputs.version }}"
             VERSION_CLEAN="${VERSION_CLEAN#v}"
             cd ..
-            cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
-            cd "text-generation-webui-${VERSION_CLEAN}"
+            cp -r text-generation-webui "text-generation-webui-ik-${VERSION_CLEAN}"
+            cd "text-generation-webui-ik-${VERSION_CLEAN}"
 
             # Remove extensions that need additional requirements
             allowed=("character_bias" "gallery" "sd_api_pictures")
@@ -131,10 +131,10 @@ jobs:
             cd ..
             curl -L -o python-build.tar.gz "$PYTHON_URL"
             tar -xzf python-build.tar.gz
-            mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
+            mv python "text-generation-webui-ik-${VERSION_CLEAN}/portable_env"
 
             # 3. Prepare requirements file
-            cd "text-generation-webui-${VERSION_CLEAN}"
+            cd "text-generation-webui-ik-${VERSION_CLEAN}"
             REQ_FILE="requirements/portable/requirements_ik_cpu_only.txt"
             echo "Using requirements file: $REQ_FILE"
 
@@ -153,11 +153,11 @@ jobs:
             if [[ "$RUNNER_OS" == "Windows" ]]; then
                 ARCHIVE_NAME="textgen-portable-ik-${VERSION_CLEAN}-${PLATFORM}.zip"
                 echo "Creating archive: $ARCHIVE_NAME"
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
+                powershell -Command "Compress-Archive -Path text-generation-webui-ik-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
             else
                 ARCHIVE_NAME="textgen-portable-ik-${VERSION_CLEAN}-${PLATFORM}.tar.gz"
                 echo "Creating archive: $ARCHIVE_NAME"
-                tar czf "$ARCHIVE_NAME" "text-generation-webui-${VERSION_CLEAN}"
+                tar czf "$ARCHIVE_NAME" "text-generation-webui-ik-${VERSION_CLEAN}"
             fi
 
       - name: Upload files to a GitHub release

From 223dd4b8017d24f7c5c2f33be2ca8409e1897b34 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 5 Apr 2026 18:22:50 -0700
Subject: [PATCH 1514/1701] UI: Hide spin buttons on number inputs

---
 css/main.css | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/css/main.css b/css/main.css
index 7f47a3aa1b..db0b781b53 100644
--- a/css/main.css
+++ b/css/main.css
@@ -22,6 +22,17 @@
     font-style: italic;
 }
 
+/* Hide spin buttons on number inputs (look bad on Windows) */
+input[type="number"]::-webkit-outer-spin-button,
+input[type="number"]::-webkit-inner-spin-button {
+    -webkit-appearance: none;
+    margin: 0;
+}
+
+input[type="number"] {
+    -moz-appearance: textfield;
+}
+
 .padded.svelte-12cmxck {
     padding: 3px 0;
 }

From abc3487f4dec9215abd9ebfb5ac796c32361b018 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 5 Apr 2026 18:24:26 -0700
Subject: [PATCH 1515/1701] UI: Move cpu-moe checkbox to extra flags (no longer
 useful now that --fit exists)

---
 modules/ui_model_menu.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 243079a02e..9c8306f50d 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -54,7 +54,6 @@ def create_ui():
                             if not shared.args.portable:
                                 shared.gradio['ik'] = gr.Checkbox(label="ik", value=shared.args.ik, info='Use ik_llama.cpp instead of upstream llama.cpp.')
 
-                            shared.gradio['cpu_moe'] = gr.Checkbox(label="cpu-moe", value=shared.args.cpu_moe, info='Move the experts to the CPU. Saves VRAM on MoE models.')
                             shared.gradio['streaming_llm'] = gr.Checkbox(label="streaming-llm", value=shared.args.streaming_llm, info='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
                             shared.gradio['load_in_8bit'] = gr.Checkbox(label="load-in-8bit", value=shared.args.load_in_8bit)
                             shared.gradio['load_in_4bit'] = gr.Checkbox(label="load-in-4bit", value=shared.args.load_in_4bit)
@@ -109,6 +108,7 @@ def create_ui():
                             with gr.Column():
                                 shared.gradio['cpu'] = gr.Checkbox(label="cpu", value=shared.args.cpu, info='Use PyTorch in CPU mode.')
                                 shared.gradio['disk'] = gr.Checkbox(label="disk", value=shared.args.disk)
+                                shared.gradio['cpu_moe'] = gr.Checkbox(label="cpu-moe", value=shared.args.cpu_moe, info='Move the experts to the CPU. Saves VRAM on MoE models.')
                                 shared.gradio['row_split'] = gr.Checkbox(label="row_split", value=shared.args.row_split, info='Split the model by rows across GPUs. This may improve multi-gpu performance.')
                                 shared.gradio['no_kv_offload'] = gr.Checkbox(label="no_kv_offload", value=shared.args.no_kv_offload, info='Do not offload the K, Q, V to the GPU. This saves VRAM but reduces performance.')
                                 shared.gradio['no_mmap'] = gr.Checkbox(label="no-mmap", value=shared.args.no_mmap)

From b1d06dcf96e2b5958ae004b8c9bbb0fc8518328b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 5 Apr 2026 23:07:14 -0300
Subject: [PATCH 1516/1701] UI: Add MCP server support

---
 README.md                                     |   2 +-
 docs/Tool Calling Tutorial.md                 |  13 ++
 modules/chat.py                               |  15 ++-
 modules/shared.py                             |   1 +
 modules/tool_use.py                           | 114 ++++++++++++++++++
 modules/ui.py                                 |   2 +
 modules/ui_chat.py                            |   3 +
 requirements/full/requirements.txt            |   1 +
 requirements/full/requirements_amd.txt        |   1 +
 .../full/requirements_apple_intel.txt         |   1 +
 .../full/requirements_apple_silicon.txt       |   1 +
 requirements/full/requirements_cpu_only.txt   |   1 +
 requirements/full/requirements_nowheels.txt   |   1 +
 requirements/portable/requirements.txt        |   1 +
 requirements/portable/requirements_amd.txt    |   1 +
 .../portable/requirements_apple_intel.txt     |   1 +
 .../portable/requirements_apple_silicon.txt   |   1 +
 .../portable/requirements_cpu_only.txt        |   1 +
 .../portable/requirements_cuda131.txt         |   1 +
 requirements/portable/requirements_ik.txt     |   1 +
 .../portable/requirements_ik_cpu_only.txt     |   1 +
 .../portable/requirements_ik_cuda131.txt      |   1 +
 .../portable/requirements_nowheels.txt        |   1 +
 requirements/portable/requirements_vulkan.txt |   1 +
 24 files changed, 163 insertions(+), 4 deletions(-)

diff --git a/README.md b/README.md
index 23cd09c5dd..b168ebdbde 100644
--- a/README.md
+++ b/README.md
@@ -26,7 +26,7 @@ A Gradio web UI for running Large Language Models locally. 100% private and offl
 - **Easy setup**: [Portable builds](https://github.com/oobabooga/text-generation-webui/releases) (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or a one-click installer for the full feature set.
 - **Multiple backends**: [llama.cpp](https://github.com/ggerganov/llama.cpp), [ik_llama.cpp](https://github.com/ikawrakow/ik_llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). Switch between backends and models without restarting.
 - **OpenAI/Anthropic-compatible API**: Chat, Completions, and Messages endpoints with tool-calling support. Use as a local drop-in replacement for the OpenAI/Anthropic APIs ([examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples)).
-- **Tool-calling**: Models can call custom functions during chat — web search, page fetching, math, and more. Each tool is a single `.py` file, easy to create and extend ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Tool-Calling-Tutorial)).
+- **Tool-calling**: Models can call custom functions during chat — web search, page fetching, math, and more. Each tool is a single `.py` file. MCP servers are also supported ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Tool-Calling-Tutorial)).
 - **Vision (multimodal)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
 - **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
 - **Training**: Fine-tune LoRAs on multi-turn chat or raw text datasets. Supports resuming interrupted runs ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/05-%E2%80%90-Training-Tab)).
diff --git a/docs/Tool Calling Tutorial.md b/docs/Tool Calling Tutorial.md
index d95a9c8085..7d2a86de03 100644
--- a/docs/Tool Calling Tutorial.md	
+++ b/docs/Tool Calling Tutorial.md	
@@ -80,6 +80,19 @@ def execute(arguments):
 
 You can open the built-in tools in `user_data/tools/` for more examples.
 
+## MCP servers
+
+You can connect to remote [MCP (Model Context Protocol)](https://modelcontextprotocol.io/) servers to use their tools alongside local ones.
+
+In the chat sidebar, open the **MCP servers** accordion and enter one server URL per line. For servers that require authentication, append headers after the URL separated by commas:
+
+```
+https://example.com/mcp
+https://other.com/mcp,Authorization: Bearer sk-xxx
+```
+
+All tools from the configured servers are automatically discovered and made available to the model during generation. If an MCP tool has the same name as a selected local tool, the local tool takes priority.
+
 ## Tool calling over the API
 
 Tool calling over the API follows the [OpenAI API](https://platform.openai.com/docs/guides/function-calling) convention. Define your tools, send them with your messages, and handle tool calls in a loop until the model gives a final answer.
diff --git a/modules/chat.py b/modules/chat.py
index 76b8694af3..aeed688dd0 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1264,14 +1264,23 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
 
     # Load tools if any are selected
     selected = state.get('selected_tools', [])
+    mcp_servers = state.get('mcp_servers', '')
     parse_tool_call = None
     _tool_parsers = None
-    if selected:
-        from modules.tool_use import load_tools, execute_tool
+    if selected or mcp_servers:
+        from modules.tool_use import load_tools, load_mcp_tools, execute_tool
         from modules.tool_parsing import parse_tool_call, get_tool_call_id, detect_tool_call_format
 
-    if selected:
         tool_defs, tool_executors = load_tools(selected)
+        if mcp_servers:
+            mcp_defs, mcp_executors = load_mcp_tools(mcp_servers)
+            for td in mcp_defs:
+                fn = td['function']['name']
+                if fn in tool_executors:
+                    logger.warning(f'MCP tool "{fn}" conflicts with a local tool. Skipping.')
+                    continue
+                tool_defs.append(td)
+                tool_executors[fn] = mcp_executors[fn]
         state['tools'] = tool_defs
         tool_func_names = [t['function']['name'] for t in tool_defs]
         _template_str = state.get('instruction_template_str', '') if state.get('mode') == 'instruct' else state.get('chat_template_str', '')
diff --git a/modules/shared.py b/modules/shared.py
index 13843f0c52..92c4f56ccc 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -259,6 +259,7 @@
     'enable_web_search': False,
     'web_search_pages': 3,
     'selected_tools': [],
+    'mcp_servers': '',
     'prompt-notebook': '',
     'preset': 'Top-P' if (user_data_dir / 'presets/Top-P.yaml').exists() else None,
     'max_new_tokens': 512,
diff --git a/modules/tool_use.py b/modules/tool_use.py
index e22b17983b..f9ddf9407e 100644
--- a/modules/tool_use.py
+++ b/modules/tool_use.py
@@ -1,3 +1,4 @@
+import asyncio
 import importlib.util
 import json
 
@@ -55,6 +56,119 @@ def load_tools(selected_names):
     return tool_defs, executors
 
 
+def _parse_mcp_servers(servers_str):
+    """Parse MCP servers textbox: one server per line, format 'url' or 'url,Header: value,Header2: value2'."""
+    servers = []
+    for line in servers_str.strip().splitlines():
+        line = line.strip()
+        if not line:
+            continue
+        parts = line.split(',')
+        url = parts[0].strip()
+        headers = {}
+        for part in parts[1:]:
+            part = part.strip()
+            if ':' in part:
+                key, val = part.split(':', 1)
+                headers[key.strip()] = val.strip()
+        servers.append((url, headers))
+    return servers
+
+
+def _mcp_tool_to_openai(tool):
+    """Convert an MCP Tool object to OpenAI-format tool dict."""
+    return {
+        "type": "function",
+        "function": {
+            "name": tool.name,
+            "description": tool.description or "",
+            "parameters": tool.inputSchema or {"type": "object", "properties": {}}
+        }
+    }
+
+
+async def _mcp_session(url, headers, callback):
+    """Open an MCP session and pass it to the callback."""
+    from mcp.client.streamable_http import streamablehttp_client
+    from mcp import ClientSession
+
+    async with streamablehttp_client(url, headers=headers or None) as (read_stream, write_stream, _):
+        async with ClientSession(read_stream, write_stream) as session:
+            await session.initialize()
+            return await callback(session)
+
+
+def _make_mcp_executor(name, url, headers):
+    def executor(arguments):
+        return asyncio.run(_call_mcp_tool(name, arguments, url, headers))
+    return executor
+
+
+async def _connect_mcp_server(url, headers):
+    """Connect to one MCP server and return (tool_defs, executors)."""
+
+    async def _discover(session):
+        result = await session.list_tools()
+        tool_defs = []
+        executors = {}
+        for tool in result.tools:
+            tool_defs.append(_mcp_tool_to_openai(tool))
+            executors[tool.name] = _make_mcp_executor(tool.name, url, headers)
+        return tool_defs, executors
+
+    return await _mcp_session(url, headers, _discover)
+
+
+async def _call_mcp_tool(name, arguments, url, headers):
+    """Connect to an MCP server and call a single tool."""
+
+    async def _invoke(session):
+        result = await session.call_tool(name, arguments)
+        parts = []
+        for content in result.content:
+            if hasattr(content, 'text'):
+                parts.append(content.text)
+            else:
+                parts.append(str(content))
+        return '\n'.join(parts) if parts else ''
+
+    return await _mcp_session(url, headers, _invoke)
+
+
+async def _connect_all_mcp_servers(servers):
+    """Connect to all MCP servers concurrently."""
+    results = await asyncio.gather(
+        *(_connect_mcp_server(url, headers) for url, headers in servers),
+        return_exceptions=True
+    )
+    all_defs = []
+    all_executors = {}
+    for (url, _), result in zip(servers, results):
+        if isinstance(result, Exception):
+            logger.exception(f'Failed to connect to MCP server "{url}"', exc_info=result)
+            continue
+        defs, execs = result
+        for td, (fn, ex) in zip(defs, execs.items()):
+            if fn in all_executors:
+                logger.warning(f'MCP tool "{fn}" from {url} conflicts with an already loaded tool. Skipping.')
+                continue
+            all_defs.append(td)
+            all_executors[fn] = ex
+    return all_defs, all_executors
+
+
+def load_mcp_tools(servers_str):
+    """
+    Parse MCP servers string and discover tools from each server.
+    Returns (tool_defs, executors) in the same format as load_tools.
+    """
+    servers = _parse_mcp_servers(servers_str)
+    if not servers:
+        return [], {}
+
+    return asyncio.run(_connect_all_mcp_servers(servers))
+
+
 def execute_tool(func_name, arguments, executors):
     """Execute a tool by function name. Returns result as a JSON string."""
     fn = executors.get(func_name)
diff --git a/modules/ui.py b/modules/ui.py
index 73072cbe8a..3a8390f7ce 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -209,6 +209,7 @@ def list_interface_input_elements():
         'textbox',
         'start_with',
         'selected_tools',
+        'mcp_servers',
         'mode',
         'chat_style',
         'chat-instruct_command',
@@ -434,6 +435,7 @@ def setup_auto_save():
         'custom_system_message',
         'chat_template_str',
         'selected_tools',
+        'mcp_servers',
 
         # Parameters tab (ui_parameters.py) - Generation parameters
         'preset_menu',
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index d96522535a..14489d9623 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -105,6 +105,9 @@ def sync_web_tools(selected):
 
                 shared.gradio['selected_tools'].change(fn=sync_web_tools, inputs=[shared.gradio['selected_tools']], outputs=[shared.gradio['selected_tools']], show_progress=False)
 
+                with gr.Accordion('MCP servers', open=False):
+                    shared.gradio['mcp_servers'] = gr.Textbox(value=shared.settings.get('mcp_servers', ''), lines=3, max_lines=3, label='', info='One url per line. For headers, write url,Header: value,Header2: value2', elem_classes=['add_scrollbar'])
+
                 gr.HTML("
            ")
 
                 with gr.Row():
diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 9f83830aa8..104cfdb2f1 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -48,3 +48,4 @@ https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+mcp==1.27.0
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index b4b8386ee9..49db44db1f 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -39,3 +39,4 @@ tiktoken
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+mcp==1.27.0
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 41ee6a6047..4584708ff4 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -38,3 +38,4 @@ tiktoken
 
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+mcp==1.27.0
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 8be2f55e23..4376a2b4ab 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -38,3 +38,4 @@ tiktoken
 
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+mcp==1.27.0
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index d7f1bf132f..2999d4a9f3 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -41,3 +41,4 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+mcp==1.27.0
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 7b331f9629..5a1e504e8f 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -35,3 +35,4 @@ https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_cl
 flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
+mcp==1.27.0
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index b467cf2627..fb51c7cc55 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -25,3 +25,4 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+mcp==1.27.0
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 4eca16e1db..dbea7597bb 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -25,3 +25,4 @@ tiktoken
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+mcp==1.27.0
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 55f8d3f8fa..d0f83a745b 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -24,3 +24,4 @@ tiktoken
 
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+mcp==1.27.0
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 54e8f350dd..160c064649 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -24,3 +24,4 @@ tiktoken
 
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+mcp==1.27.0
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index f073a6147c..2169558537 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -25,3 +25,4 @@ tiktoken
 # llama.cpp (CPU only)
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+mcp==1.27.0
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 8cd40f39dd..6b09a46b6e 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -25,3 +25,4 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+mcp==1.27.0
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index fbb9125d81..ca5ece2d00 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -25,3 +25,4 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+mcp==1.27.0
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index 59fcfae11e..f8bafb272f 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -25,3 +25,4 @@ tiktoken
 # ik_llama.cpp (CPU only)
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+mcp==1.27.0
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index ffdbe56843..7825b95938 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -25,3 +25,4 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+mcp==1.27.0
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 4a47b1f05b..cde036d98c 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -21,3 +21,4 @@ https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_cl
 flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
+mcp==1.27.0
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 97abd9333e..32f9e5933a 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -25,3 +25,4 @@ tiktoken
 # Vulkan wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+mcp==1.27.0

From 05e484203308adb3324f7a9edd1412ed9762e359 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 5 Apr 2026 20:03:06 -0700
Subject: [PATCH 1517/1701] Fix image generation: default to SDPA attention
 backend

---
 modules/image_models.py | 2 +-
 modules/shared.py       | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/image_models.py b/modules/image_models.py
index eed8783c8c..e244c3c876 100644
--- a/modules/image_models.py
+++ b/modules/image_models.py
@@ -129,7 +129,7 @@ def load_image_model(model_name, dtype='bfloat16', attn_backend='sdpa', cpu_offl
 
         modules = ["transformer", "unet"]
 
-        # Set attention backend
+        # Set attention backend (diffusers defaults to native/SDPA)
         if attn_backend == 'flash_attention_2':
             for name in modules:
                 mod = getattr(pipe, name, None)
diff --git a/modules/shared.py b/modules/shared.py
index 92c4f56ccc..e04f28f3c4 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -364,7 +364,7 @@
     'image_llm_variations_prompt': 'Write a variation of the image generation prompt above. Consider the intent of the user with that prompt and write something that will likely please them, with added details. Output only the new prompt. Do not add any explanations, prefixes, or additional text.',
     'image_model_menu': 'None',
     'image_dtype': 'bfloat16',
-    'image_attn_backend': 'flash_attention_2',
+    'image_attn_backend': 'sdpa',
     'image_cpu_offload': False,
     'image_compile': False,
     'image_quant': 'none',

From 7b2f15e34ae57a6e86b0901482b4ed9b6b52ad8a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 5 Apr 2026 21:16:32 -0700
Subject: [PATCH 1518/1701] Minor change after
 b1d06dcf96e2b5958ae004b8c9bbb0fc8518328b

---
 requirements/full/requirements.txt                   | 2 +-
 requirements/full/requirements_amd.txt               | 2 +-
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 2 +-
 requirements/full/requirements_nowheels.txt          | 2 +-
 requirements/portable/requirements.txt               | 2 +-
 requirements/portable/requirements_amd.txt           | 2 +-
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 2 +-
 requirements/portable/requirements_cuda131.txt       | 2 +-
 requirements/portable/requirements_ik.txt            | 2 +-
 requirements/portable/requirements_ik_cpu_only.txt   | 2 +-
 requirements/portable/requirements_ik_cuda131.txt    | 2 +-
 requirements/portable/requirements_nowheels.txt      | 2 +-
 requirements/portable/requirements_vulkan.txt        | 2 +-
 17 files changed, 17 insertions(+), 17 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 104cfdb2f1..d466e7e311 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -9,6 +9,7 @@ flash-linear-attention==0.4.*
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pandas
 peft==0.18.*
@@ -48,4 +49,3 @@ https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
-mcp==1.27.0
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 49db44db1f..e88ff7c510 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -7,6 +7,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pandas
 peft==0.18.*
@@ -39,4 +40,3 @@ tiktoken
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-mcp==1.27.0
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 4584708ff4..eefd979e65 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -7,6 +7,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pandas
 peft==0.18.*
@@ -38,4 +39,3 @@ tiktoken
 
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
-mcp==1.27.0
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 4376a2b4ab..d1b4e09f9a 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -7,6 +7,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pandas
 peft==0.18.*
@@ -38,4 +39,3 @@ tiktoken
 
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
-mcp==1.27.0
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 2999d4a9f3..156ceb7761 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -7,6 +7,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pandas
 peft==0.18.*
@@ -41,4 +42,3 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-mcp==1.27.0
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 5a1e504e8f..19ac51836b 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -7,6 +7,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pandas
 peft==0.18.*
@@ -35,4 +36,3 @@ https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_cl
 flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
-mcp==1.27.0
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index fb51c7cc55..8a158f05cf 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -3,6 +3,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pydantic==2.11.0
 pymupdf==1.27.*
@@ -25,4 +26,3 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-mcp==1.27.0
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index dbea7597bb..a4949a46f7 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -3,6 +3,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pydantic==2.11.0
 pymupdf==1.27.*
@@ -25,4 +26,3 @@ tiktoken
 # AMD wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-mcp==1.27.0
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index d0f83a745b..227823a688 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -3,6 +3,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pydantic==2.11.0
 pymupdf==1.27.*
@@ -24,4 +25,3 @@ tiktoken
 
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
-mcp==1.27.0
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 160c064649..9779dd4af9 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -3,6 +3,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pydantic==2.11.0
 pymupdf==1.27.*
@@ -24,4 +25,3 @@ tiktoken
 
 # Mac wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
-mcp==1.27.0
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 2169558537..ff84907a59 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -3,6 +3,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pydantic==2.11.0
 pymupdf==1.27.*
@@ -25,4 +26,3 @@ tiktoken
 # llama.cpp (CPU only)
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-mcp==1.27.0
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 6b09a46b6e..89e43e1a05 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -3,6 +3,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pydantic==2.11.0
 pymupdf==1.27.*
@@ -25,4 +26,3 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-mcp==1.27.0
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index ca5ece2d00..a23d8ff071 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -3,6 +3,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pydantic==2.11.0
 pymupdf==1.27.*
@@ -25,4 +26,3 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-mcp==1.27.0
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index f8bafb272f..a200e80f1f 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -3,6 +3,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pydantic==2.11.0
 pymupdf==1.27.*
@@ -25,4 +26,3 @@ tiktoken
 # ik_llama.cpp (CPU only)
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-mcp==1.27.0
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index 7825b95938..8e9a097bcb 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -3,6 +3,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pydantic==2.11.0
 pymupdf==1.27.*
@@ -25,4 +26,3 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-mcp==1.27.0
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index cde036d98c..cafe3ceea2 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -3,6 +3,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pydantic==2.11.0
 pymupdf==1.27.*
@@ -21,4 +22,3 @@ https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_cl
 flask_cloudflared==0.0.15
 sse-starlette==1.6.5
 tiktoken
-mcp==1.27.0
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 32f9e5933a..595246686f 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -3,6 +3,7 @@ fastapi==0.112.4
 huggingface-hub==1.5.*
 jinja2==3.1.6
 markdown
+mcp==1.27.0
 numpy==2.2.*
 pydantic==2.11.0
 pymupdf==1.27.*
@@ -25,4 +26,3 @@ tiktoken
 # Vulkan wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-mcp==1.27.0

From 4d6230a944a71dab794d880d7c353eb37934d584 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 6 Apr 2026 06:48:48 -0700
Subject: [PATCH 1519/1701] Follow-up to
 d78fc46114a4ce1de505fc286798372ddaa0c32d

---
 modules/api/script.py | 28 +++-------------------------
 1 file changed, 3 insertions(+), 25 deletions(-)

diff --git a/modules/api/script.py b/modules/api/script.py
index 14e2d03a58..e79a1967da 100644
--- a/modules/api/script.py
+++ b/modules/api/script.py
@@ -532,8 +532,8 @@ async def handle_unload_loras():
 def find_available_port(starting_port):
     """Try the starting port, then find an available one if it's taken."""
     try:
-        # Try to create a socket with the starting port
         with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
+            s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
             s.bind(('', starting_port))
             return starting_port
     except OSError:
@@ -591,31 +591,9 @@ def run_server():
     if shared.args.admin_key and shared.args.admin_key != shared.args.api_key:
         logger.info(f'OpenAI API admin key (for loading/unloading models):\n\n{shared.args.admin_key}\n')
 
-    # Use SO_REUSEADDR to avoid "address already in use" after restart
+    # Start server
     logging.getLogger("uvicorn.error").propagate = False
-    sockets = []
-    try:
-        for addr in server_addrs:
-            family = socket.AF_INET6 if ':' in addr else socket.AF_INET
-            sock = socket.socket(family, socket.SOCK_STREAM)
-            sock.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
-            if family == socket.AF_INET6:
-                sock.setsockopt(socket.IPPROTO_IPV6, socket.IPV6_V6ONLY, 1)
-            sock.bind((addr.strip('[]'), port))
-            sock.listen(socket.SOMAXCONN)
-            sockets.append(sock)
-    except Exception:
-        for s in sockets:
-            s.close()
-        raise
-
-    config = uvicorn.Config(app, ssl_certfile=ssl_certfile, ssl_keyfile=ssl_keyfile, access_log=False)
-    server = uvicorn.Server(config)
-    try:
-        server.run(sockets=sockets)
-    finally:
-        for s in sockets:
-            s.close()
+    uvicorn.run(app, host=server_addrs, port=port, ssl_certfile=ssl_certfile, ssl_keyfile=ssl_keyfile, access_log=False)
 
 
 _server_started = False

From c26ffdd24c60b1dc6ad339c847b8993f490dc036 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 6 Apr 2026 07:02:53 -0700
Subject: [PATCH 1520/1701] API: add instruction_template support to the model
 load endpoint

---
 docs/12 - OpenAI API.md    | 11 +++++++++++
 modules/api/models.py      | 10 +++++++++-
 modules/api/script.py      |  5 +++++
 modules/api/typing.py      |  2 ++
 modules/models_settings.py | 17 +++++++++++------
 5 files changed, 38 insertions(+), 7 deletions(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index 0a076c350a..727f6eced4 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -232,6 +232,17 @@ curl -k http://127.0.0.1:5000/v1/internal/model/load \
   }'
 ```
 
+You can also set a default instruction template for all subsequent API requests by passing `instruction_template` (a template name from `user_data/instruction-templates/`) or `instruction_template_str` (a raw Jinja2 string):
+
+```shell
+curl -k http://127.0.0.1:5000/v1/internal/model/load \
+  -H "Content-Type: application/json" \
+  -d '{
+    "model_name": "Qwen_Qwen3-0.6B-Q4_K_M.gguf",
+    "instruction_template": "Alpaca"
+  }'
+```
+
 #### Python chat example
 
 ```python
diff --git a/modules/api/models.py b/modules/api/models.py
index 5dd7785050..bfcd2c31fd 100644
--- a/modules/api/models.py
+++ b/modules/api/models.py
@@ -1,7 +1,8 @@
 from modules import loaders, shared
+from modules.logging_colors import logger
 from modules.LoRA import add_lora_to_model
 from modules.models import load_model, unload_model
-from modules.models_settings import get_model_metadata, update_model_parameters
+from modules.models_settings import get_model_metadata, load_instruction_template, update_model_parameters
 from modules.utils import get_available_loras, get_available_models
 
 
@@ -69,6 +70,13 @@ def _load_model(data):
 
     shared.model, shared.tokenizer = load_model(model_name)
 
+    if data.get("instruction_template_str") is not None:
+        shared.settings['instruction_template_str'] = data["instruction_template_str"]
+        logger.info("INSTRUCTION TEMPLATE: set to custom Jinja2 string")
+    elif data.get("instruction_template") is not None:
+        shared.settings['instruction_template_str'] = load_instruction_template(data["instruction_template"])
+        logger.info(f"INSTRUCTION TEMPLATE: {data['instruction_template']}")
+
 
 def list_loras():
     return {'lora_names': get_available_loras()[1:]}
diff --git a/modules/api/script.py b/modules/api/script.py
index e79a1967da..1f41d0cd38 100644
--- a/modules/api/script.py
+++ b/modules/api/script.py
@@ -487,6 +487,11 @@ async def handle_load_model(request_data: LoadModelRequest):
 
     Loader args are reset to their startup defaults between loads, so
     settings from a previous load do not leak into the next one.
+
+    The "instruction_template" parameter sets the default instruction
+    template by name (from user_data/instruction-templates/). The
+    "instruction_template_str" parameter sets it as a raw Jinja2 string
+    and takes precedence over "instruction_template".
     '''
 
     try:
diff --git a/modules/api/typing.py b/modules/api/typing.py
index a758743e46..56d7f2bc8a 100644
--- a/modules/api/typing.py
+++ b/modules/api/typing.py
@@ -271,6 +271,8 @@ class ModelListResponse(BaseModel):
 class LoadModelRequest(BaseModel):
     model_name: str
     args: dict | None = None
+    instruction_template: str | None = Field(default=None, description="An instruction template defined under text-generation-webui/user_data/instruction-templates. Sets the default template for all subsequent API requests.")
+    instruction_template_str: str | None = Field(default=None, description="A Jinja2 instruction template string. If set, takes precedence over instruction_template.")
 
 
 class LoraListResponse(BaseModel):
diff --git a/modules/models_settings.py b/modules/models_settings.py
index eafa058107..b10d780c94 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -400,14 +400,19 @@ def load_instruction_template(template):
     if template == 'None':
         return ''
 
-    for filepath in [shared.user_data_dir / 'instruction-templates' / f'{template}.yaml', shared.user_data_dir / 'instruction-templates' / 'Alpaca.yaml']:
-        if filepath.exists():
-            break
+    for name in (template, 'Alpaca'):
+        path = shared.user_data_dir / 'instruction-templates' / f'{name}.yaml'
+        try:
+            with open(path, 'r', encoding='utf-8') as f:
+                file_contents = f.read()
+        except FileNotFoundError:
+            if name == template:
+                logger.warning(f"Instruction template '{template}' not found, falling back to Alpaca")
+            continue
+
+        break
     else:
         return ''
-
-    with open(filepath, 'r', encoding='utf-8') as f:
-        file_contents = f.read()
     data = yaml.safe_load(file_contents)
     if 'instruction_template' in data:
         return data['instruction_template']

From 193424cc9359859b5b97bf5b229409a3fb727274 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 6 Apr 2026 10:07:52 -0700
Subject: [PATCH 1521/1701] API: Fix IPv6 address formatting

---
 modules/api/script.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/api/script.py b/modules/api/script.py
index 1f41d0cd38..ceeca2dc93 100644
--- a/modules/api/script.py
+++ b/modules/api/script.py
@@ -563,7 +563,7 @@ def run_server():
         server_addrs.append(shared.args.listen_host)
     else:
         if os.environ.get('OPENEDAI_ENABLE_IPV6', shared.args.api_enable_ipv6):
-            server_addrs.append('[::]' if shared.args.listen else '[::1]')
+            server_addrs.append('::' if shared.args.listen else '::1')
         if not os.environ.get('OPENEDAI_DISABLE_IPV4', shared.args.api_disable_ipv4):
             server_addrs.append('0.0.0.0' if shared.args.listen else '127.0.0.1')
 
@@ -580,7 +580,7 @@ def run_server():
         )
     else:
         url_proto = 'https://' if (ssl_certfile and ssl_keyfile) else 'http://'
-        urls = [f'{url_proto}{addr}:{port}/v1' for addr in server_addrs]
+        urls = [f'{url_proto}[{addr}]:{port}/v1' if ':' in addr else f'{url_proto}{addr}:{port}/v1' for addr in server_addrs]
         if len(urls) > 1:
             logger.info('OpenAI/Anthropic-compatible API URLs:\n\n' + '\n'.join(urls) + '\n')
         else:

From cb511928e2be4b7ee234582ecba96801fccf94fe Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 6 Apr 2026 12:06:28 -0700
Subject: [PATCH 1522/1701] Fix GPT-OSS tag leak during streaming between
 thinking and tool calls

---
 modules/reasoning.py    | 13 ++++++++++---
 modules/tool_parsing.py | 12 +++++++++---
 2 files changed, 19 insertions(+), 6 deletions(-)

diff --git a/modules/reasoning.py b/modules/reasoning.py
index 4a7cfa7945..2b26081806 100644
--- a/modules/reasoning.py
+++ b/modules/reasoning.py
@@ -73,9 +73,16 @@ def extract_reasoning(text, html_escaped=False):
                 if content_pos != -1:
                     content_start = content_pos + len(content_esc)
                 else:
-                    # Content tag not present — fall back to content after
-                    # end_tag (e.g. GPT-OSS tool calls skip the final channel).
-                    content_start = end_pos + len(end_esc)
+                    # Content tag not present yet.  In GPT-OSS the region
+                    # between <|end|> and the content tag contains internal
+                    # markup (<|start|>assistant…) that must not be shown.
+                    # Suppress it to prevent tag leaks during streaming.
+                    remainder = text[end_pos + len(end_esc):].lstrip()
+                    framing_token = esc('<|start|>')
+                    if not remainder or remainder.startswith(framing_token) or framing_token.startswith(remainder):
+                        content_start = len(text)
+                    else:
+                        content_start = end_pos + len(end_esc)
             else:
                 content_start = end_pos + len(end_esc)
 
diff --git a/modules/tool_parsing.py b/modules/tool_parsing.py
index 7fcf58b740..aa3e0e95be 100644
--- a/modules/tool_parsing.py
+++ b/modules/tool_parsing.py
@@ -638,9 +638,15 @@ def parse_tool_call(answer: str, tool_names: list[str], return_prefix: bool = Fa
     # Strip thinking blocks so tool-call syntax inside  is ignored.
     original_answer = answer
     _, answer = extract_reasoning(answer)
-    # Offset between original and stripped text, used to map start_pos
-    # back to the original string when returning a prefix.
-    reasoning_offset = len(original_answer) - len(answer)
+    # Reasoning extraction returns empty content when GPT-OSS internal
+    # markup (<|start|>assistant…) follows the thinking block without a
+    # content tag.  Fall back to the full text so tool-call markers can
+    # be found.
+    if not answer.strip():
+        answer = original_answer
+        reasoning_offset = 0
+    else:
+        reasoning_offset = len(original_answer) - len(answer)
 
     matches = []
     start_pos = None

From 775c913de20824d187f677e65845fe8680ecd7f6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 6 Apr 2026 14:13:01 -0700
Subject: [PATCH 1523/1701] Fix crash when truncating prompts with tool call
 messages

---
 modules/chat.py | 21 ++++++++++++++++-----
 1 file changed, 16 insertions(+), 5 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index aeed688dd0..7e9cce60e6 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -568,13 +568,24 @@ def make_prompt(messages):
         encoded_length = get_encoded_length(prompt)
         while len(messages) > 0 and encoded_length > max_length:
 
-            # Remove old message, save system message
             if len(messages) > 2 and messages[0]['role'] == 'system':
-                messages.pop(1)
-
-            # Remove old message when no system message is present
+                pop_idx = 1
             elif len(messages) > 1 and messages[0]['role'] != 'system':
-                messages.pop(0)
+                pop_idx = 0
+            else:
+                pop_idx = None
+
+            if pop_idx is not None:
+                messages.pop(pop_idx)
+
+                # Remove orphaned tool-call/tool-result messages that
+                # would be invalid without their partner.
+                while pop_idx < len(messages):
+                    msg = messages[pop_idx]
+                    if msg.get('role') == 'tool' or (msg.get('role') == 'assistant' and msg.get('tool_calls')):
+                        messages.pop(pop_idx)
+                    else:
+                        break
 
             # Resort to truncating the user input
             else:

From 778e1c4d52cc6f86cd55207543563773b12cd2cf Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 6 Apr 2026 17:04:49 -0700
Subject: [PATCH 1524/1701] Update llama.cpp/ik_llama.cpp

---
 requirements/full/requirements.txt                   | 8 ++++----
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 8 ++++----
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 15 files changed, 30 insertions(+), 30 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index d466e7e311..ed5841b8f1 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -41,10 +41,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index e88ff7c510..fe6ce28cdf 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index eefd979e65..09c01a6148 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -38,4 +38,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index d1b4e09f9a..4221040761 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -38,4 +38,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 156ceb7761..5cd7ae7d72 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -38,7 +38,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 8a158f05cf..807ff07951 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index a4949a46f7..55fe79eac8 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 227823a688..6d4a63f76e 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -24,4 +24,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 9779dd4af9..aebb7c5b54 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -24,4 +24,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index ff84907a59..d7e2b051ff 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 89e43e1a05..42a9a16f83 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index a23d8ff071..c3fdb5e877 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index a200e80f1f..ea3ba6010e 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # ik_llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index 8e9a097bcb..7530375d7a 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/ik_llama_cpp_binaries-0.106.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 595246686f..3b8b05735b 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.106.0/llama_cpp_binaries-0.106.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From e18f32cba78d471dd86a924147aa3ea6638d5e97 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 6 Apr 2026 17:47:50 -0700
Subject: [PATCH 1525/1701] Remove hardcoded trust_remote_code=True in
 embedding loader

---
 modules/api/embeddings.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/modules/api/embeddings.py b/modules/api/embeddings.py
index 16cf0482de..17e595fb6f 100644
--- a/modules/api/embeddings.py
+++ b/modules/api/embeddings.py
@@ -6,6 +6,7 @@
 from .errors import ServiceUnavailableError
 from .utils import debug_msg, float_list_to_base64
 from modules.logging_colors import logger
+from modules import shared
 
 embeddings_params_initialized = False
 
@@ -41,7 +42,7 @@ def load_embedding_model(model: str):
     try:
         logger.info(f"Try embedding model: {model} on {embeddings_device}")
         if 'jina-embeddings' in model:
-            embeddings_model = AutoModel.from_pretrained(model, trust_remote_code=True)  # trust_remote_code is needed to use the encode method
+            embeddings_model = AutoModel.from_pretrained(model, trust_remote_code=shared.args.trust_remote_code)
             embeddings_model = embeddings_model.to(embeddings_device)
         else:
             embeddings_model = SentenceTransformer(model, device=embeddings_device)

From 5ad199e9bbd15b5c31af75b20ce6aa15ce86c7c1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 8 Apr 2026 14:20:24 -0700
Subject: [PATCH 1526/1701] Update README

---
 README.md | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/README.md b/README.md
index b168ebdbde..c37391ff11 100644
--- a/README.md
+++ b/README.md
@@ -15,18 +15,20 @@
 
 A Gradio web UI for running Large Language Models locally. 100% private and offline. Supports text generation, vision, tool-calling, training, image generation, and more.
 
-[Try the Deep Reason extension](https://oobabooga.gumroad.com/l/deep_reason)
-
 |![Image1](https://github.com/oobabooga/screenshots/raw/main/INSTRUCT-3.5.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/CHAT-3.5.png) |
 |:---:|:---:|
 |![Image1](https://github.com/oobabooga/screenshots/raw/main/DEFAULT-3.5.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/PARAMETERS-3.5.png) |
 
+## 🔥 News
+My GGUF quantization benchmarks on [LocalBench](https://localbench.substack.com), where I compare quants from uploaders like unsloth, bartowski, and lmstudio-community:
+- **2026-04-07**: [Gemma 4 31B results](https://localbench.substack.com/p/gemma-4-31b-gguf-kl-divergence)
+
 ## Features
 
 - **Easy setup**: [Portable builds](https://github.com/oobabooga/text-generation-webui/releases) (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or a one-click installer for the full feature set.
 - **Multiple backends**: [llama.cpp](https://github.com/ggerganov/llama.cpp), [ik_llama.cpp](https://github.com/ikawrakow/ik_llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). Switch between backends and models without restarting.
 - **OpenAI/Anthropic-compatible API**: Chat, Completions, and Messages endpoints with tool-calling support. Use as a local drop-in replacement for the OpenAI/Anthropic APIs ([examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples)).
-- **Tool-calling**: Models can call custom functions during chat — web search, page fetching, math, and more. Each tool is a single `.py` file. MCP servers are also supported ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Tool-Calling-Tutorial)).
+- **Tool-calling**: Models can call custom functions during chat: web search, page fetching, math, and more. Each tool is a single `.py` file. MCP servers are also supported ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Tool-Calling-Tutorial)).
 - **Vision (multimodal)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
 - **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
 - **Training**: Fine-tune LoRAs on multi-turn chat or raw text datasets. Supports resuming interrupted runs ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/05-%E2%80%90-Training-Tab)).

From 24cc0e22366c36d7500dc8d9f248a0740c935b4c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 10 Apr 2026 08:38:59 -0700
Subject: [PATCH 1527/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 8 ++++----
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 8 ++++----
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 15 files changed, 30 insertions(+), 30 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index ed5841b8f1..fbd7f348f1 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -41,10 +41,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index fe6ce28cdf..c2b16deb98 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 09c01a6148..07ebdbe442 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -38,4 +38,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 4221040761..6d6f4e4ff8 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -38,4 +38,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 5cd7ae7d72..f406e73696 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -38,7 +38,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 807ff07951..d6d2498006 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 55fe79eac8..eed2cca301 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 6d4a63f76e..ed2f99af2f 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -24,4 +24,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index aebb7c5b54..9752eb98e0 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -24,4 +24,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index d7e2b051ff..099d76f12f 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 42a9a16f83..229b6be580 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index c3fdb5e877..802d679d18 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index ea3ba6010e..c12cb1a5d7 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # ik_llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index 7530375d7a..5142d8ce37 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/ik_llama_cpp_binaries-0.110.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 3b8b05735b..7aca9fdddb 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.110.0/llama_cpp_binaries-0.110.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 9bc3f9f3dd4fc26a9b0360e2b68954388c82c67e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 10 Apr 2026 12:21:48 -0700
Subject: [PATCH 1528/1701] Update README

---
 README.md | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

diff --git a/README.md b/README.md
index c37391ff11..4fb825b7b8 100644
--- a/README.md
+++ b/README.md
@@ -13,7 +13,7 @@
 
 # Text Generation Web UI
 
-A Gradio web UI for running Large Language Models locally. 100% private and offline. Supports text generation, vision, tool-calling, training, image generation, and more.
+A Gradio web UI for running Large Language Models locally. 100% private and offline. Supports text generation, vision, tool-calling, training, and image generation.
 
 |![Image1](https://github.com/oobabooga/screenshots/raw/main/INSTRUCT-3.5.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/CHAT-3.5.png) |
 |:---:|:---:|
@@ -28,7 +28,7 @@ My GGUF quantization benchmarks on [LocalBench](https://localbench.substack.com)
 - **Easy setup**: [Portable builds](https://github.com/oobabooga/text-generation-webui/releases) (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or a one-click installer for the full feature set.
 - **Multiple backends**: [llama.cpp](https://github.com/ggerganov/llama.cpp), [ik_llama.cpp](https://github.com/ikawrakow/ik_llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). Switch between backends and models without restarting.
 - **OpenAI/Anthropic-compatible API**: Chat, Completions, and Messages endpoints with tool-calling support. Use as a local drop-in replacement for the OpenAI/Anthropic APIs ([examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples)).
-- **Tool-calling**: Models can call custom functions during chat: web search, page fetching, math, and more. Each tool is a single `.py` file. MCP servers are also supported ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Tool-Calling-Tutorial)).
+- **Tool-calling**: Models can call custom functions during chat, including web search, page fetching, and math. Each tool is a single `.py` file. MCP servers are also supported ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Tool-Calling-Tutorial)).
 - **Vision (multimodal)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
 - **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
 - **Training**: Fine-tune LoRAs on multi-turn chat or raw text datasets. Supports resuming interrupted runs ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/05-%E2%80%90-Training-Tab)).
@@ -37,9 +37,8 @@ My GGUF quantization benchmarks on [LocalBench](https://localbench.substack.com)
 - `instruct` mode for instruction-following (like ChatGPT), and `chat-instruct`/`chat` modes for talking to custom characters. Prompts are automatically formatted with Jinja2 templates.
 - Edit messages, navigate between message versions, and branch conversations at any point.
 - Free-form text generation in the Notebook tab without being limited to chat turns.
-- Multiple sampling parameters and generation options for sophisticated text generation control.
 - Dark/light themes, syntax highlighting for code blocks, and LaTeX rendering for mathematical expressions.
-- Extension support, with numerous built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details.
+- Extension support, with built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details.
 
 ## How to install
 
@@ -54,7 +53,7 @@ Download from here: **https://github.com/oobabooga/text-generation-webui/release
 
 #### Option 2: Manual portable install with venv
 
-Very fast setup that should work on any Python 3.9+:
+Fast setup on any Python 3.9+:
 
 ```bash
 # Clone repository

From 326867e7993a8349ec846ac1f82c549cca4334f8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 10 Apr 2026 19:21:57 -0700
Subject: [PATCH 1529/1701] Reduce VRAM peak in prompt logprobs forward pass

---
 modules/api/completions.py | 17 ++++++++---------
 modules/exllamav3.py       |  7 +++++--
 modules/exllamav3_hf.py    | 16 ++++++++++++++++
 3 files changed, 29 insertions(+), 11 deletions(-)

diff --git a/modules/api/completions.py b/modules/api/completions.py
index f228273118..007f80a1c2 100644
--- a/modules/api/completions.py
+++ b/modules/api/completions.py
@@ -95,16 +95,15 @@ def _compute_prompt_logprob_entries(prompt, logprobs_count, input_ids=None):
         logits = model.get_prompt_logits(input_ids)
 
     elif hasattr(model, 'forward'):
-        # HF-compatible loaders (Transformers, ExLlamav3_HF, etc.)
+        # HF-compatible loaders (Transformers, etc.). Loaders that need a
+        # custom path (e.g. wrappers that only compute last-token logits in
+        # __call__) should expose get_prompt_logits() above.
         input_ids_tensor = input_ids if isinstance(input_ids, torch.Tensor) else torch.tensor(input_ids, dtype=torch.long)
         if hasattr(model, 'device'):
             input_ids_tensor = input_ids_tensor.to(model.device)
-        with torch.no_grad():
-            # Pass labels to ensure logits are returned for ALL positions,
-            # not just the last token (some HF wrappers like ExLlamav3_HF
-            # only compute the last-token logits when labels are absent).
-            outputs = model(input_ids=input_ids_tensor, labels=input_ids_tensor)
-            logits = outputs.logits  # keep on GPU, (1, seq_len, vocab) in model dtype
+        with torch.inference_mode():
+            outputs = model(input_ids=input_ids_tensor)
+            logits = outputs.logits  # keep on device, (1, seq_len, vocab) in model dtype
             del outputs
 
     else:
@@ -117,14 +116,14 @@ def _compute_prompt_logprob_entries(prompt, logprobs_count, input_ids=None):
     chunk_size = 2048
     unique_ids = set(int(tid) for tid in token_ids[1:])
 
-    # Process logits in chunks on GPU, only move top-K results to CPU
+    # Process logits in chunks, only move top-K results to CPU
     all_top_log_probs_list = []
     all_top_indices_list = []
     all_actual_lps = []
 
     for start in range(0, n_tokens - 1, chunk_size):
         end = min(start + chunk_size, n_tokens - 1)
-        chunk_logits = logits[0, start:end].float()  # (chunk, vocab) on GPU
+        chunk_logits = logits[0, start:end].float()  # (chunk, vocab) on logits.device
         chunk_lse = torch.logsumexp(chunk_logits, dim=-1)
         chunk_top_values, chunk_top_indices = torch.topk(chunk_logits, k=k, dim=-1)
         chunk_top_log_probs = chunk_top_values - chunk_lse.unsqueeze(-1)
diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index e1efbfeb90..f424902ada 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -535,11 +535,14 @@ def get_prompt_logits(self, input_ids):
         import torch
         input_ids_tensor = input_ids if isinstance(input_ids, torch.Tensor) else torch.tensor(input_ids, dtype=torch.long)
         input_ids_tensor = input_ids_tensor.view(1, -1).cpu()
-        with torch.no_grad():
-            return self.model.forward(
+        with torch.inference_mode():
+            output = self.model.forward(
                 input_ids=input_ids_tensor,
                 params={"attn_mode": "flash_attn_nc"}
             ).cpu().float()
+            # Mask padding slots beyond the real vocab so they can't appear in top-k
+            output[..., self.model.config.vocab_size:] = float("-inf")
+            return output
 
     def get_logits(self, token_ids, **kwargs):
         """
diff --git a/modules/exllamav3_hf.py b/modules/exllamav3_hf.py
index 5e634e2250..4496400e61 100644
--- a/modules/exllamav3_hf.py
+++ b/modules/exllamav3_hf.py
@@ -98,6 +98,22 @@ def _validate_model_class(self):
     def _validate_model_kwargs(self, model_kwargs: Dict[str, Any]):
         pass
 
+    def get_prompt_logits(self, input_ids):
+        """Return logits for all positions via a single no-cache forward pass.
+
+        Used by prompt logprobs computation. Returns (1, seq_len, vocab) on CPU in float32.
+        """
+        input_ids_tensor = input_ids if isinstance(input_ids, torch.Tensor) else torch.tensor(input_ids, dtype=torch.long)
+        input_ids_tensor = input_ids_tensor.view(1, -1).cpu()
+        with torch.inference_mode():
+            output = self.ex_model.forward(
+                input_ids=input_ids_tensor,
+                params={"attn_mode": "flash_attn_nc"}
+            ).cpu().float()
+            # Mask padding slots beyond the real vocab so they can't appear in top-k
+            output[..., self.ex_model.config.vocab_size:] = float("-inf")
+            return output
+
     def prepare_inputs_for_generation(self, input_ids, **kwargs):
         return {'input_ids': input_ids, **kwargs}
 

From 456afad585741ce27b88de06cef6af226198ac7e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 11 Apr 2026 15:42:31 -0700
Subject: [PATCH 1530/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 8 ++++----
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 8 ++++----
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 15 files changed, 30 insertions(+), 30 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index fbd7f348f1..97b8f034eb 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -41,10 +41,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index c2b16deb98..dcd3bacb88 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 07ebdbe442..0032f6822f 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -38,4 +38,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 6d6f4e4ff8..7f44dd9591 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -38,4 +38,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index f406e73696..454893c031 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -38,7 +38,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index d6d2498006..827fc23945 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index eed2cca301..116631dd08 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index ed2f99af2f..ce9cccb562 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -24,4 +24,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 9752eb98e0..4758046885 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -24,4 +24,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 099d76f12f..0d039a03a3 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 229b6be580..cb375e7812 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index 802d679d18..cff325c683 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index c12cb1a5d7..0d0f63dddc 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # ik_llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index 5142d8ce37..6832acb92a 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/ik_llama_cpp_binaries-0.112.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 7aca9fdddb..2bf42bc863 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.112.0/llama_cpp_binaries-0.112.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From e9d7feb1517a90af2b5edec37a05789755fcd88f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 11 Apr 2026 21:29:46 -0700
Subject: [PATCH 1531/1701] Update exllamav3

---
 requirements/full/requirements.txt | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 97b8f034eb..2e01b6bdca 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -45,7 +45,7 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.29/exllamav3-0.0.29+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.29/exllamav3-0.0.29+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"

From add91613db9346f19cc5dc682ec477b21603ca26 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 12 Apr 2026 13:46:01 -0700
Subject: [PATCH 1532/1701] Fix bos/eos tokens not being set for models without
 a chat template

---
 modules/exllamav3.py       |  2 ++
 modules/models_settings.py | 19 ++++++++-----------
 2 files changed, 10 insertions(+), 11 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index f424902ada..815ca3dd9b 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -558,6 +558,8 @@ def get_logits(self, token_ids, **kwargs):
 
     def encode(self, string, **kwargs):
         add_bos = kwargs.pop('add_bos', True)
+        if add_bos and self.tokenizer.bos_token and string.startswith(self.tokenizer.bos_token):
+            add_bos = False
         return self.tokenizer.encode(string, add_bos=add_bos, **kwargs)
 
     def decode(self, ids, **kwargs):
diff --git a/modules/models_settings.py b/modules/models_settings.py
index b10d780c94..814e3778ee 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -133,6 +133,14 @@ def get_model_metadata(model):
         with open(path, 'r', encoding='utf-8') as f:
             metadata = json.loads(f.read())
 
+        for k in ['eos_token', 'bos_token']:
+            if k in metadata:
+                value = metadata[k]
+                if isinstance(value, dict):
+                    value = value['content']
+
+                setattr(shared, k, value)
+
         # Only read from metadata if we haven't already loaded from .jinja or .json
         if template is None and 'chat_template' in metadata:
             template = metadata['chat_template']
@@ -141,17 +149,6 @@ def get_model_metadata(model):
 
         # 4. If a template was found from any source, process it
         if template:
-            shared.bos_token = ''
-            shared.eos_token = ''
-
-            for k in ['eos_token', 'bos_token']:
-                if k in metadata:
-                    value = metadata[k]
-                    if isinstance(value, dict):
-                        value = value['content']
-
-                    setattr(shared, k, value)
-
             template = re.sub(r"\{\{-?\s*raise_exception\(.*?\)\s*-?\}\}", "", template, flags=re.DOTALL)
             template = re.sub(r'raise_exception\([^)]*\)', "''", template)
             model_settings['instruction_template'] = 'Custom (obtained from model metadata)'

From 7e4c8fa2097dbe6ec406368d890a314d0c3ad3ae Mon Sep 17 00:00:00 2001
From: mamei16 
Date: Mon, 13 Apr 2026 06:41:57 +0200
Subject: [PATCH 1533/1701] prevent tool icon SVG shrinking when tool calls are
 long (#7488)

---
 css/main.css | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/css/main.css b/css/main.css
index db0b781b53..cfa80d8ee2 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1518,6 +1518,10 @@ audio {
 .thinking-icon {
     margin-right: 8px;
     color: rgb(0 0 0 / 50%);
+
+    /* Prevents the SVG from shrinking
+     * when tool call arguments are long */
+    flex-shrink: 0;
 }
 
 .thinking-title {

From fae17dc5b83d18cc04c7ac060555db7ca51989f6 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Mon, 13 Apr 2026 14:47:46 -0300
Subject: [PATCH 1534/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index c37391ff11..13281561f1 100644
--- a/README.md
+++ b/README.md
@@ -11,7 +11,7 @@
 
            
 
            
 
-# Text Generation Web UI
+# TextGen
 
 A Gradio web UI for running Large Language Models locally. 100% private and offline. Supports text generation, vision, tool-calling, training, image generation, and more.
 

From 61bfc2ffd56559da96dbde3e0d60fc233ec1a5ac Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 13 Apr 2026 12:14:59 -0700
Subject: [PATCH 1535/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 93ceb0ae26..35898691f2 100644
--- a/README.md
+++ b/README.md
@@ -13,7 +13,7 @@
 
 # TextGen
 
-A Gradio web UI for running Large Language Models locally. 100% private and offline. Supports text generation, vision, tool-calling, training, and image generation.
+**The original local LLM interface.** Text, vision, tool-calling, training, image generation. UI + API, 100% offline and private.
 
 |![Image1](https://github.com/oobabooga/screenshots/raw/main/INSTRUCT-3.5.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/CHAT-3.5.png) |
 |:---:|:---:|

From 65cef2c73177fa674ebb8b70a2566e9bac7990a8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 13 Apr 2026 16:42:51 -0300
Subject: [PATCH 1536/1701] Rename project from text-generation-webui to
 textgen

---
 .github/pull_request_template.md              |  2 +-
 .github/workflows/build-everything-tgw.yml    |  2 +-
 .../workflows/build-portable-release-cuda.yml | 18 ++++-----
 .../build-portable-release-ik-cuda.yml        | 18 ++++-----
 .../workflows/build-portable-release-ik.yml   | 18 ++++-----
 .../workflows/build-portable-release-rocm.yml | 18 ++++-----
 .../build-portable-release-vulkan.yml         | 18 ++++-----
 .github/workflows/build-portable-release.yml  | 18 ++++-----
 Colab-TextGen-GPU.ipynb                       | 10 ++---
 README.md                                     | 38 +++++++++----------
 docker/TensorRT-LLM/Dockerfile                |  6 +--
 docker/amd/Dockerfile                         |  8 ++--
 docker/amd/docker-compose.yml                 |  4 +-
 docker/cpu/Dockerfile                         |  8 ++--
 docker/cpu/docker-compose.yml                 |  4 +-
 docker/intel/Dockerfile                       |  8 ++--
 docker/intel/docker-compose.yml               |  4 +-
 docker/nvidia/Dockerfile                      |  8 ++--
 docker/nvidia/docker-compose.yml              |  4 +-
 docs/01 - Chat Tab.md                         |  8 ++--
 docs/02 - Default and Notebook Tabs.md        |  2 +-
 docs/04 - Model Tab.md                        |  2 +-
 docs/06 - Session Tab.md                      |  4 +-
 docs/07 - Extensions.md                       | 34 ++++++++---------
 docs/08 - Additional Tips.md                  |  2 +-
 docs/09 - Docker.md                           |  2 +-
 docs/12 - OpenAI API.md                       |  2 +-
 docs/Image Generation Tutorial.md             |  8 ++--
 docs/Multimodal Tutorial.md                   |  6 +--
 docs/README.md                                |  2 +-
 extensions/ngrok/script.py                    |  2 +-
 extensions/sd_api_pictures/README.MD          |  6 +--
 extensions/superboogav2/README.md             |  4 +-
 modules/api/typing.py                         |  8 ++--
 modules/logging_colors.py                     |  2 +-
 modules/shared.py                             |  2 +-
 modules/training.py                           |  2 +-
 one_click.py                                  |  6 +--
 server.py                                     |  6 +--
 39 files changed, 162 insertions(+), 162 deletions(-)

diff --git a/.github/pull_request_template.md b/.github/pull_request_template.md
index 51e26b13a3..a9d0250534 100644
--- a/.github/pull_request_template.md
+++ b/.github/pull_request_template.md
@@ -1,3 +1,3 @@
 ## Checklist:
 
-- [ ] I have read the [Contributing guidelines](https://github.com/oobabooga/text-generation-webui/wiki/Contributing-guidelines).
+- [ ] I have read the [Contributing guidelines](https://github.com/oobabooga/textgen/wiki/Contributing-guidelines).
diff --git a/.github/workflows/build-everything-tgw.yml b/.github/workflows/build-everything-tgw.yml
index 0b65dfd64f..7d7baa6d77 100644
--- a/.github/workflows/build-everything-tgw.yml
+++ b/.github/workflows/build-everything-tgw.yml
@@ -4,7 +4,7 @@ on:
   workflow_dispatch:
     inputs:
       version:
-        description: 'Version tag of text-generation-webui to build: v3.0'
+        description: 'Version tag of textgen to build: v3.0'
         default: 'v3.0'
         required: true
         type: string
diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index f9eea58a37..58ae208447 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -4,7 +4,7 @@ on:
   workflow_dispatch:
     inputs:
       version:
-        description: 'Version tag of text-generation-webui to build: v3.0'
+        description: 'Version tag of textgen to build: v3.0'
         default: 'v3.0'
         required: true
         type: string
@@ -21,7 +21,7 @@ on:
   workflow_call:
     inputs:
       version:
-        description: 'Version tag of text-generation-webui to build: v3.0'
+        description: 'Version tag of textgen to build: v3.0'
         default: 'v3.0'
         required: true
         type: string
@@ -88,7 +88,7 @@ jobs:
     steps:
       - uses: actions/checkout@v6
         with:
-          repository: 'oobabooga/text-generation-webui'
+          repository: 'oobabooga/textgen'
           ref: ${{ inputs.version }}
           submodules: 'recursive'
 
@@ -102,8 +102,8 @@ jobs:
             VERSION_CLEAN="${{ inputs.version }}"
             VERSION_CLEAN="${VERSION_CLEAN#v}"
             cd ..
-            cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
-            cd "text-generation-webui-${VERSION_CLEAN}"
+            cp -r textgen "textgen-${VERSION_CLEAN}"
+            cd "textgen-${VERSION_CLEAN}"
 
             # Remove extensions that need additional requirements
             allowed=("character_bias" "gallery" "sd_api_pictures")
@@ -133,10 +133,10 @@ jobs:
             echo "Downloading Python for $PLATFORM..."
             curl -L -o python-build.tar.gz "$PYTHON_URL"
             tar -xzf python-build.tar.gz
-            mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
+            mv python "textgen-${VERSION_CLEAN}/portable_env"
 
             # 3. Prepare requirements file based on CUDA version
-            cd "text-generation-webui-${VERSION_CLEAN}"
+            cd "textgen-${VERSION_CLEAN}"
             if [[ "$CUDA_VERSION" == "13.1" ]]; then
                 REQ_FILE="requirements/portable/requirements_cuda131.txt"
             else
@@ -155,11 +155,11 @@ jobs:
             if [[ "$RUNNER_OS" == "Windows" ]]; then
                 ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
                 echo "Creating archive: $ARCHIVE_NAME"
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
+                powershell -Command "Compress-Archive -Path textgen-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
             else
                 ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.tar.gz"
                 echo "Creating archive: $ARCHIVE_NAME"
-                tar czf "$ARCHIVE_NAME" "text-generation-webui-${VERSION_CLEAN}"
+                tar czf "$ARCHIVE_NAME" "textgen-${VERSION_CLEAN}"
             fi
 
       - name: Upload files to a GitHub release
diff --git a/.github/workflows/build-portable-release-ik-cuda.yml b/.github/workflows/build-portable-release-ik-cuda.yml
index a336a1cb9f..f96068957f 100644
--- a/.github/workflows/build-portable-release-ik-cuda.yml
+++ b/.github/workflows/build-portable-release-ik-cuda.yml
@@ -4,7 +4,7 @@ on:
   workflow_dispatch:
     inputs:
       version:
-        description: 'Version tag of text-generation-webui to build: v3.0'
+        description: 'Version tag of textgen to build: v3.0'
         default: 'v3.0'
         required: true
         type: string
@@ -21,7 +21,7 @@ on:
   workflow_call:
     inputs:
       version:
-        description: 'Version tag of text-generation-webui to build: v3.0'
+        description: 'Version tag of textgen to build: v3.0'
         default: 'v3.0'
         required: true
         type: string
@@ -88,7 +88,7 @@ jobs:
     steps:
       - uses: actions/checkout@v6
         with:
-          repository: 'oobabooga/text-generation-webui'
+          repository: 'oobabooga/textgen'
           ref: ${{ inputs.version }}
           submodules: 'recursive'
 
@@ -102,8 +102,8 @@ jobs:
             VERSION_CLEAN="${{ inputs.version }}"
             VERSION_CLEAN="${VERSION_CLEAN#v}"
             cd ..
-            cp -r text-generation-webui "text-generation-webui-ik-${VERSION_CLEAN}"
-            cd "text-generation-webui-ik-${VERSION_CLEAN}"
+            cp -r textgen "textgen-ik-${VERSION_CLEAN}"
+            cd "textgen-ik-${VERSION_CLEAN}"
 
             # Remove extensions that need additional requirements
             allowed=("character_bias" "gallery" "sd_api_pictures")
@@ -133,10 +133,10 @@ jobs:
             echo "Downloading Python for $PLATFORM..."
             curl -L -o python-build.tar.gz "$PYTHON_URL"
             tar -xzf python-build.tar.gz
-            mv python "text-generation-webui-ik-${VERSION_CLEAN}/portable_env"
+            mv python "textgen-ik-${VERSION_CLEAN}/portable_env"
 
             # 3. Prepare requirements file based on CUDA version
-            cd "text-generation-webui-ik-${VERSION_CLEAN}"
+            cd "textgen-ik-${VERSION_CLEAN}"
             if [[ "$CUDA_VERSION" == "13.1" ]]; then
                 REQ_FILE="requirements/portable/requirements_ik_cuda131.txt"
             else
@@ -158,11 +158,11 @@ jobs:
             if [[ "$RUNNER_OS" == "Windows" ]]; then
                 ARCHIVE_NAME="textgen-portable-ik-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.zip"
                 echo "Creating archive: $ARCHIVE_NAME"
-                powershell -Command "Compress-Archive -Path text-generation-webui-ik-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
+                powershell -Command "Compress-Archive -Path textgen-ik-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
             else
                 ARCHIVE_NAME="textgen-portable-ik-${VERSION_CLEAN}-${PLATFORM}-cuda${CUDA_VERSION}.tar.gz"
                 echo "Creating archive: $ARCHIVE_NAME"
-                tar czf "$ARCHIVE_NAME" "text-generation-webui-ik-${VERSION_CLEAN}"
+                tar czf "$ARCHIVE_NAME" "textgen-ik-${VERSION_CLEAN}"
             fi
 
       - name: Upload files to a GitHub release
diff --git a/.github/workflows/build-portable-release-ik.yml b/.github/workflows/build-portable-release-ik.yml
index 5eaf7c8671..b3bad0e6f3 100644
--- a/.github/workflows/build-portable-release-ik.yml
+++ b/.github/workflows/build-portable-release-ik.yml
@@ -4,7 +4,7 @@ on:
   workflow_dispatch:
     inputs:
       version:
-        description: 'Version tag of text-generation-webui to build: v3.0'
+        description: 'Version tag of textgen to build: v3.0'
         default: 'v3.0'
         required: true
         type: string
@@ -21,7 +21,7 @@ on:
   workflow_call:
     inputs:
       version:
-        description: 'Version tag of text-generation-webui to build: v3.0'
+        description: 'Version tag of textgen to build: v3.0'
         default: 'v3.0'
         required: true
         type: string
@@ -87,7 +87,7 @@ jobs:
     steps:
       - uses: actions/checkout@v6
         with:
-          repository: 'oobabooga/text-generation-webui'
+          repository: 'oobabooga/textgen'
           ref: ${{ inputs.version }}
           submodules: 'recursive'
 
@@ -101,8 +101,8 @@ jobs:
             VERSION_CLEAN="${{ inputs.version }}"
             VERSION_CLEAN="${VERSION_CLEAN#v}"
             cd ..
-            cp -r text-generation-webui "text-generation-webui-ik-${VERSION_CLEAN}"
-            cd "text-generation-webui-ik-${VERSION_CLEAN}"
+            cp -r textgen "textgen-ik-${VERSION_CLEAN}"
+            cd "textgen-ik-${VERSION_CLEAN}"
 
             # Remove extensions that need additional requirements
             allowed=("character_bias" "gallery" "sd_api_pictures")
@@ -131,10 +131,10 @@ jobs:
             cd ..
             curl -L -o python-build.tar.gz "$PYTHON_URL"
             tar -xzf python-build.tar.gz
-            mv python "text-generation-webui-ik-${VERSION_CLEAN}/portable_env"
+            mv python "textgen-ik-${VERSION_CLEAN}/portable_env"
 
             # 3. Prepare requirements file
-            cd "text-generation-webui-ik-${VERSION_CLEAN}"
+            cd "textgen-ik-${VERSION_CLEAN}"
             REQ_FILE="requirements/portable/requirements_ik_cpu_only.txt"
             echo "Using requirements file: $REQ_FILE"
 
@@ -153,11 +153,11 @@ jobs:
             if [[ "$RUNNER_OS" == "Windows" ]]; then
                 ARCHIVE_NAME="textgen-portable-ik-${VERSION_CLEAN}-${PLATFORM}.zip"
                 echo "Creating archive: $ARCHIVE_NAME"
-                powershell -Command "Compress-Archive -Path text-generation-webui-ik-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
+                powershell -Command "Compress-Archive -Path textgen-ik-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
             else
                 ARCHIVE_NAME="textgen-portable-ik-${VERSION_CLEAN}-${PLATFORM}.tar.gz"
                 echo "Creating archive: $ARCHIVE_NAME"
-                tar czf "$ARCHIVE_NAME" "text-generation-webui-ik-${VERSION_CLEAN}"
+                tar czf "$ARCHIVE_NAME" "textgen-ik-${VERSION_CLEAN}"
             fi
 
       - name: Upload files to a GitHub release
diff --git a/.github/workflows/build-portable-release-rocm.yml b/.github/workflows/build-portable-release-rocm.yml
index db42b7dc49..ca71c55a5f 100644
--- a/.github/workflows/build-portable-release-rocm.yml
+++ b/.github/workflows/build-portable-release-rocm.yml
@@ -4,7 +4,7 @@ on:
   workflow_dispatch:
     inputs:
       version:
-        description: 'Version tag of text-generation-webui to build: v3.0'
+        description: 'Version tag of textgen to build: v3.0'
         default: 'v3.0'
         required: true
         type: string
@@ -21,7 +21,7 @@ on:
   workflow_call:
     inputs:
       version:
-        description: 'Version tag of text-generation-webui to build: v3.0'
+        description: 'Version tag of textgen to build: v3.0'
         default: 'v3.0'
         required: true
         type: string
@@ -87,7 +87,7 @@ jobs:
     steps:
       - uses: actions/checkout@v6
         with:
-          repository: 'oobabooga/text-generation-webui'
+          repository: 'oobabooga/textgen'
           ref: ${{ inputs.version }}
           submodules: 'recursive'
 
@@ -101,8 +101,8 @@ jobs:
             VERSION_CLEAN="${{ inputs.version }}"
             VERSION_CLEAN="${VERSION_CLEAN#v}"
             cd ..
-            cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
-            cd "text-generation-webui-${VERSION_CLEAN}"
+            cp -r textgen "textgen-${VERSION_CLEAN}"
+            cd "textgen-${VERSION_CLEAN}"
 
             # Remove extensions that need additional requirements
             allowed=("character_bias" "gallery" "sd_api_pictures")
@@ -131,12 +131,12 @@ jobs:
             echo "Downloading Python for $PLATFORM..."
             curl -L -o python-build.tar.gz "$PYTHON_URL"
             tar -xzf python-build.tar.gz
-            mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
+            mv python "textgen-${VERSION_CLEAN}/portable_env"
 
             # 3. Prepare requirements file
             REQ_FILE="requirements/portable/requirements_amd.txt"
 
-            cd "text-generation-webui-${VERSION_CLEAN}"
+            cd "textgen-${VERSION_CLEAN}"
 
             # 4. Install packages
             echo "Installing Python packages from $REQ_FILE..."
@@ -150,11 +150,11 @@ jobs:
             if [[ "$RUNNER_OS" == "Windows" ]]; then
                 ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-rocm7.2.zip"
                 echo "Creating archive: $ARCHIVE_NAME"
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
+                powershell -Command "Compress-Archive -Path textgen-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
             else
                 ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-rocm7.2.tar.gz"
                 echo "Creating archive: $ARCHIVE_NAME"
-                tar czf "$ARCHIVE_NAME" "text-generation-webui-${VERSION_CLEAN}"
+                tar czf "$ARCHIVE_NAME" "textgen-${VERSION_CLEAN}"
             fi
 
       - name: Upload files to a GitHub release
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 8f5aa7c883..2b03f6241b 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -4,7 +4,7 @@ on:
   workflow_dispatch:
     inputs:
       version:
-        description: 'Version tag of text-generation-webui to build: v3.0'
+        description: 'Version tag of textgen to build: v3.0'
         default: 'v3.0'
         required: true
         type: string
@@ -21,7 +21,7 @@ on:
   workflow_call:
     inputs:
       version:
-        description: 'Version tag of text-generation-webui to build: v3.0'
+        description: 'Version tag of textgen to build: v3.0'
         default: 'v3.0'
         required: true
         type: string
@@ -87,7 +87,7 @@ jobs:
     steps:
       - uses: actions/checkout@v6
         with:
-          repository: 'oobabooga/text-generation-webui'
+          repository: 'oobabooga/textgen'
           ref: ${{ inputs.version }}
           submodules: 'recursive'
 
@@ -101,8 +101,8 @@ jobs:
             VERSION_CLEAN="${{ inputs.version }}"
             VERSION_CLEAN="${VERSION_CLEAN#v}"
             cd ..
-            cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
-            cd "text-generation-webui-${VERSION_CLEAN}"
+            cp -r textgen "textgen-${VERSION_CLEAN}"
+            cd "textgen-${VERSION_CLEAN}"
 
             # Remove extensions that need additional requirements
             allowed=("character_bias" "gallery" "sd_api_pictures")
@@ -131,12 +131,12 @@ jobs:
             echo "Downloading Python for $PLATFORM..."
             curl -L -o python-build.tar.gz "$PYTHON_URL"
             tar -xzf python-build.tar.gz
-            mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
+            mv python "textgen-${VERSION_CLEAN}/portable_env"
 
             # 3. Prepare requirements file
             REQ_FILE="requirements/portable/requirements_vulkan.txt"
 
-            cd "text-generation-webui-${VERSION_CLEAN}"
+            cd "textgen-${VERSION_CLEAN}"
 
             # 4. Install packages
             echo "Installing Python packages from $REQ_FILE..."
@@ -150,11 +150,11 @@ jobs:
             if [[ "$RUNNER_OS" == "Windows" ]]; then
                 ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.zip"
                 echo "Creating archive: $ARCHIVE_NAME"
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
+                powershell -Command "Compress-Archive -Path textgen-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
             else
                 ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}-vulkan.tar.gz"
                 echo "Creating archive: $ARCHIVE_NAME"
-                tar czf "$ARCHIVE_NAME" "text-generation-webui-${VERSION_CLEAN}"
+                tar czf "$ARCHIVE_NAME" "textgen-${VERSION_CLEAN}"
             fi
 
       - name: Upload files to a GitHub release
diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index 9ace90f6ad..dc3a28d25d 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -4,7 +4,7 @@ on:
   workflow_dispatch:
     inputs:
       version:
-        description: 'Version tag of text-generation-webui to build: v3.0'
+        description: 'Version tag of textgen to build: v3.0'
         default: 'v3.0'
         required: true
         type: string
@@ -21,7 +21,7 @@ on:
   workflow_call:
     inputs:
       version:
-        description: 'Version tag of text-generation-webui to build: v3.0'
+        description: 'Version tag of textgen to build: v3.0'
         default: 'v3.0'
         required: true
         type: string
@@ -87,7 +87,7 @@ jobs:
     steps:
       - uses: actions/checkout@v6
         with:
-          repository: 'oobabooga/text-generation-webui'
+          repository: 'oobabooga/textgen'
           ref: ${{ inputs.version }}
           submodules: 'recursive'
 
@@ -101,8 +101,8 @@ jobs:
             VERSION_CLEAN="${{ inputs.version }}"
             VERSION_CLEAN="${VERSION_CLEAN#v}"
             cd ..
-            cp -r text-generation-webui "text-generation-webui-${VERSION_CLEAN}"
-            cd "text-generation-webui-${VERSION_CLEAN}"
+            cp -r textgen "textgen-${VERSION_CLEAN}"
+            cd "textgen-${VERSION_CLEAN}"
 
             # Remove extensions that need additional requirements
             allowed=("character_bias" "gallery" "sd_api_pictures")
@@ -146,10 +146,10 @@ jobs:
             cd ..
             curl -L -o python-build.tar.gz "$PYTHON_URL"
             tar -xzf python-build.tar.gz
-            mv python "text-generation-webui-${VERSION_CLEAN}/portable_env"
+            mv python "textgen-${VERSION_CLEAN}/portable_env"
 
             # 3. Prepare requirements file based on platform
-            cd "text-generation-webui-${VERSION_CLEAN}"
+            cd "textgen-${VERSION_CLEAN}"
 
             # Select requirements file based on platform
             if [[ "$RUNNER_OS" == "macOS" ]]; then
@@ -176,11 +176,11 @@ jobs:
             if [[ "$RUNNER_OS" == "Windows" ]]; then
                 ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.zip"
                 echo "Creating archive: $ARCHIVE_NAME"
-                powershell -Command "Compress-Archive -Path text-generation-webui-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
+                powershell -Command "Compress-Archive -Path textgen-${VERSION_CLEAN} -DestinationPath $ARCHIVE_NAME"
             else
                 ARCHIVE_NAME="textgen-portable-${VERSION_CLEAN}-${PLATFORM}.tar.gz"
                 echo "Creating archive: $ARCHIVE_NAME"
-                tar czf "$ARCHIVE_NAME" "text-generation-webui-${VERSION_CLEAN}"
+                tar czf "$ARCHIVE_NAME" "textgen-${VERSION_CLEAN}"
             fi
 
       - name: Upload files to a GitHub release
diff --git a/Colab-TextGen-GPU.ipynb b/Colab-TextGen-GPU.ipynb
index f256b0e84e..734fe4fc7f 100644
--- a/Colab-TextGen-GPU.ipynb
+++ b/Colab-TextGen-GPU.ipynb
@@ -20,11 +20,11 @@
     {
       "cell_type": "markdown",
       "source": [
-        "# oobabooga/text-generation-webui\n",
+        "# oobabooga/textgen\n",
         "\n",
         "After running both cells, a public gradio URL will appear at the bottom in around 10 minutes. You can optionally generate an API link.\n",
         "\n",
-        "* Project page: https://github.com/oobabooga/text-generation-webui\n",
+        "* Project page: https://github.com/oobabooga/textgen\n",
         "* Gradio server status: https://status.gradio.app/"
       ],
       "metadata": {
@@ -59,11 +59,11 @@
         "os.environ.pop('PYTHONPATH', None)\n",
         "os.environ.pop('MPLBACKEND', None)\n",
         "\n",
-        "if Path.cwd().name != 'text-generation-webui':\n",
+        "if Path.cwd().name != 'textgen':\n",
         "  print(\"\\033[1;32;1m\\n --> Installing the web UI. This will take a while, but after the initial setup, you can download and test as many models as you like.\\033[0;37;0m\\n\")\n",
         "\n",
-        "  !git clone https://github.com/oobabooga/text-generation-webui\n",
-        "  %cd text-generation-webui\n",
+        "  !git clone https://github.com/oobabooga/textgen\n",
+        "  %cd textgen\n",
         "\n",
         "  # Install the project in an isolated environment\n",
         "  !GPU_CHOICE=A \\\n",
diff --git a/README.md b/README.md
index 35898691f2..0266d047d5 100644
--- a/README.md
+++ b/README.md
@@ -25,20 +25,20 @@ My GGUF quantization benchmarks on [LocalBench](https://localbench.substack.com)
 
 ## Features
 
-- **Easy setup**: [Portable builds](https://github.com/oobabooga/text-generation-webui/releases) (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or a one-click installer for the full feature set.
+- **Easy setup**: [Portable builds](https://github.com/oobabooga/textgen/releases) (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or a one-click installer for the full feature set.
 - **Multiple backends**: [llama.cpp](https://github.com/ggerganov/llama.cpp), [ik_llama.cpp](https://github.com/ikawrakow/ik_llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). Switch between backends and models without restarting.
-- **OpenAI/Anthropic-compatible API**: Chat, Completions, and Messages endpoints with tool-calling support. Use as a local drop-in replacement for the OpenAI/Anthropic APIs ([examples](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#examples)).
-- **Tool-calling**: Models can call custom functions during chat, including web search, page fetching, and math. Each tool is a single `.py` file. MCP servers are also supported ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Tool-Calling-Tutorial)).
-- **Vision (multimodal)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Multimodal-Tutorial)).
+- **OpenAI/Anthropic-compatible API**: Chat, Completions, and Messages endpoints with tool-calling support. Use as a local drop-in replacement for the OpenAI/Anthropic APIs ([examples](https://github.com/oobabooga/textgen/wiki/12-%E2%80%90-OpenAI-API#examples)).
+- **Tool-calling**: Models can call custom functions during chat, including web search, page fetching, and math. Each tool is a single `.py` file. MCP servers are also supported ([tutorial](https://github.com/oobabooga/textgen/wiki/Tool-Calling-Tutorial)).
+- **Vision (multimodal)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/textgen/wiki/Multimodal-Tutorial)).
 - **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
-- **Training**: Fine-tune LoRAs on multi-turn chat or raw text datasets. Supports resuming interrupted runs ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/05-%E2%80%90-Training-Tab)).
-- **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/text-generation-webui/wiki/Image-Generation-Tutorial)).
+- **Training**: Fine-tune LoRAs on multi-turn chat or raw text datasets. Supports resuming interrupted runs ([tutorial](https://github.com/oobabooga/textgen/wiki/05-%E2%80%90-Training-Tab)).
+- **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/textgen/wiki/Image-Generation-Tutorial)).
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
 - `instruct` mode for instruction-following (like ChatGPT), and `chat-instruct`/`chat` modes for talking to custom characters. Prompts are automatically formatted with Jinja2 templates.
 - Edit messages, navigate between message versions, and branch conversations at any point.
 - Free-form text generation in the Notebook tab without being limited to chat turns.
 - Dark/light themes, syntax highlighting for code blocks, and LaTeX rendering for mathematical expressions.
-- Extension support, with built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details.
+- Extension support, with built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/textgen/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/textgen-extensions) for details.
 
 ## How to install
 
@@ -46,7 +46,7 @@ My GGUF quantization benchmarks on [LocalBench](https://localbench.substack.com)
 
 No installation needed – just download, unzip and run. All dependencies included.
 
-Download from here: **https://github.com/oobabooga/text-generation-webui/releases**
+Download from here: **https://github.com/oobabooga/textgen/releases**
 
 - Builds are provided for Linux, Windows, and macOS, with options for CUDA, Vulkan, ROCm, and CPU-only.
 - Compatible with GGUF (llama.cpp) models.
@@ -57,8 +57,8 @@ Fast setup on any Python 3.9+:
 
 ```bash
 # Clone repository
-git clone https://github.com/oobabooga/text-generation-webui
-cd text-generation-webui
+git clone https://github.com/oobabooga/textgen
+cd textgen
 
 # Create virtual environment
 python -m venv venv
@@ -83,7 +83,7 @@ deactivate
 
 For users who need additional backends (ExLlamaV3, Transformers), training, image generation, or extensions (TTS, voice input, translation, etc). Requires ~10GB disk space and downloads PyTorch.
 
-1. Clone the repository, or [download its source code](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip) and extract it.
+1. Clone the repository, or [download its source code](https://github.com/oobabooga/textgen/archive/refs/heads/main.zip) and extract it.
 2. Run the startup script for your OS: `start_windows.bat`, `start_linux.sh`, or `start_macos.sh`.
 3. When prompted, select your GPU vendor.
 4. After installation, open `http://127.0.0.1:7860` in your browser.
@@ -162,8 +162,8 @@ conda install -y -c "nvidia/label/cuda-12.8.1" cuda
 #### 3. Install the web UI
 
 ```
-git clone https://github.com/oobabooga/text-generation-webui
-cd text-generation-webui
+git clone https://github.com/oobabooga/textgen
+cd textgen
 pip install -r requirements/full/
 ```
 
@@ -181,7 +181,7 @@ Requirements file to use:
 
 ```
 conda activate textgen
-cd text-generation-webui
+cd textgen
 python server.py
 ```
 
@@ -216,7 +216,7 @@ mkdir -p user_data/logs user_data/cache
 docker compose up --build
 ```
 
-* You need to have Docker Compose v2.17 or higher installed. See [this guide](https://github.com/oobabooga/text-generation-webui/wiki/09-%E2%80%90-Docker) for instructions.
+* You need to have Docker Compose v2.17 or higher installed. See [this guide](https://github.com/oobabooga/textgen/wiki/09-%E2%80%90-Docker) for instructions.
 * For additional docker files, check out [this repository](https://github.com/Atinoda/text-generation-webui-docker).
 
 ### Updating the requirements
@@ -225,7 +225,7 @@ From time to time, the `requirements*.txt` change. To update, use these commands
 
 ```
 conda activate textgen
-cd text-generation-webui
+cd textgen
 pip install -r  --upgrade
 ```
 
            
@@ -255,7 +255,7 @@ usage: server.py [-h] [--user-data-dir USER_DATA_DIR] [--multi-user] [--model MO
                  [--do-sample | --no-do-sample] [--dynamic-temperature | --no-dynamic-temperature] [--temperature-last | --no-temperature-last] [--sampler-priority N] [--dry-sequence-breakers N]
                  [--enable-thinking | --no-enable-thinking] [--reasoning-effort N] [--chat-template-file CHAT_TEMPLATE_FILE]
 
-Text Generation Web UI
+TextGen
 
 options:
   -h, --help                                           show this help message and exit
@@ -437,7 +437,7 @@ To estimate how much memory a model will use, you can use the [GGUF Memory Calcu
 Models that consist of multiple files (like 16-bit Transformers models and EXL3 models) should be placed in a subfolder inside `user_data/models`:
 
 ```
-text-generation-webui
+textgen
 └── user_data
     └── models
         └── Qwen_Qwen3-8B
@@ -454,7 +454,7 @@ These formats require the one-click installer (not the portable build).
 
 ## Documentation
 
-https://github.com/oobabooga/text-generation-webui/wiki
+https://github.com/oobabooga/textgen/wiki
 
 ## Community
 
diff --git a/docker/TensorRT-LLM/Dockerfile b/docker/TensorRT-LLM/Dockerfile
index f9d4dc1cc8..800af5e982 100644
--- a/docker/TensorRT-LLM/Dockerfile
+++ b/docker/TensorRT-LLM/Dockerfile
@@ -9,9 +9,9 @@ WORKDIR /app
 # This is needed to avoid an error about "Failed to build mpi4py" in the next command
 ENV LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu:$LD_LIBRARY_PATH
 
-# Install text-generation-webui
-RUN git clone https://github.com/oobabooga/text-generation-webui
-WORKDIR /app/text-generation-webui
+# Install textgen
+RUN git clone https://github.com/oobabooga/textgen
+WORKDIR /app/textgen
 RUN pip install --break-system-packages -r requirements/full/requirements.txt
 
 # Install TensorRT-LLM
diff --git a/docker/amd/Dockerfile b/docker/amd/Dockerfile
index 6df8e2de4c..4e6db97235 100644
--- a/docker/amd/Dockerfile
+++ b/docker/amd/Dockerfile
@@ -10,10 +10,10 @@ RUN --mount=type=cache,target=/var/cache/apt,sharing=locked,rw \
     apt install --no-install-recommends -y git vim build-essential python3-dev pip bash curl && \
     rm -rf /var/lib/apt/lists/*
 WORKDIR /home/app/
-RUN git clone https://github.com/oobabooga/text-generation-webui.git 
-WORKDIR /home/app/text-generation-webui
+RUN git clone https://github.com/oobabooga/textgen.git 
+WORKDIR /home/app/textgen
 RUN GPU_CHOICE=B LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000}
-WORKDIR /home/app/text-generation-webui
+WORKDIR /home/app/textgen
 # set umask to ensure group read / write at runtime
-CMD umask 0002 && export HOME=/home/app/text-generation-webui && ./start_linux.sh --listen
+CMD umask 0002 && export HOME=/home/app/textgen && ./start_linux.sh --listen
diff --git a/docker/amd/docker-compose.yml b/docker/amd/docker-compose.yml
index 6fe3a375b8..3f6c173e1a 100644
--- a/docker/amd/docker-compose.yml
+++ b/docker/amd/docker-compose.yml
@@ -1,6 +1,6 @@
 version: "3.3"
 services:
-  text-generation-webui:
+  textgen:
     build:
       context: .
       args:
@@ -25,4 +25,4 @@ services:
     security_opt:
       - seccomp=unconfined
     volumes:
-      - ./user_data:/home/app/text-generation-webui/user_data
+      - ./user_data:/home/app/textgen/user_data
diff --git a/docker/cpu/Dockerfile b/docker/cpu/Dockerfile
index 96092edbec..4068b9778f 100644
--- a/docker/cpu/Dockerfile
+++ b/docker/cpu/Dockerfile
@@ -10,10 +10,10 @@ RUN --mount=type=cache,target=/var/cache/apt,sharing=locked,rw \
     apt install --no-install-recommends -y git vim build-essential python3-dev pip bash curl && \
     rm -rf /var/lib/apt/lists/*
 WORKDIR /home/app/
-RUN git clone https://github.com/oobabooga/text-generation-webui.git 
-WORKDIR /home/app/text-generation-webui
+RUN git clone https://github.com/oobabooga/textgen.git 
+WORKDIR /home/app/textgen
 RUN GPU_CHOICE=N LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000}
 # set umask to ensure group read / write at runtime
-WORKDIR /home/app/text-generation-webui
-CMD umask 0002 && export HOME=/home/app/text-generation-webui && ./start_linux.sh --listen
+WORKDIR /home/app/textgen
+CMD umask 0002 && export HOME=/home/app/textgen && ./start_linux.sh --listen
diff --git a/docker/cpu/docker-compose.yml b/docker/cpu/docker-compose.yml
index 6942f8de5c..4287dd9786 100644
--- a/docker/cpu/docker-compose.yml
+++ b/docker/cpu/docker-compose.yml
@@ -1,6 +1,6 @@
 version: "3.3"
 services:
-  text-generation-webui:
+  textgen:
     build:
       context: .
       args:
@@ -15,4 +15,4 @@ services:
     stdin_open: true
     tty: true
     volumes:
-      - ./user_data:/home/app/text-generation-webui/user_data
+      - ./user_data:/home/app/textgen/user_data
diff --git a/docker/intel/Dockerfile b/docker/intel/Dockerfile
index 28ee983ce8..ba4c0ff943 100644
--- a/docker/intel/Dockerfile
+++ b/docker/intel/Dockerfile
@@ -10,10 +10,10 @@ RUN --mount=type=cache,target=/var/cache/apt,sharing=locked,rw \
     apt install --no-install-recommends -y git vim build-essential python3-dev pip bash curl && \
     rm -rf /var/lib/apt/lists/*
 WORKDIR /home/app/
-RUN git clone https://github.com/oobabooga/text-generation-webui.git 
-WORKDIR /home/app/text-generation-webui
+RUN git clone https://github.com/oobabooga/textgen.git 
+WORKDIR /home/app/textgen
 RUN GPU_CHOICE=D LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000}
 # set umask to ensure group read / write at runtime
-WORKDIR /home/app/text-generation-webui
-CMD umask 0002 && export HOME=/home/app/text-generation-webui && ./start_linux.sh --listen
+WORKDIR /home/app/textgen
+CMD umask 0002 && export HOME=/home/app/textgen && ./start_linux.sh --listen
diff --git a/docker/intel/docker-compose.yml b/docker/intel/docker-compose.yml
index 6fe3a375b8..3f6c173e1a 100644
--- a/docker/intel/docker-compose.yml
+++ b/docker/intel/docker-compose.yml
@@ -1,6 +1,6 @@
 version: "3.3"
 services:
-  text-generation-webui:
+  textgen:
     build:
       context: .
       args:
@@ -25,4 +25,4 @@ services:
     security_opt:
       - seccomp=unconfined
     volumes:
-      - ./user_data:/home/app/text-generation-webui/user_data
+      - ./user_data:/home/app/textgen/user_data
diff --git a/docker/nvidia/Dockerfile b/docker/nvidia/Dockerfile
index a7ef93badb..797d95689d 100644
--- a/docker/nvidia/Dockerfile
+++ b/docker/nvidia/Dockerfile
@@ -11,10 +11,10 @@ RUN --mount=type=cache,target=/var/cache/apt,sharing=locked,rw \
     apt install --no-install-recommends -y git vim build-essential python3-dev pip bash curl && \
     rm -rf /var/lib/apt/lists/*
 WORKDIR /home/app/
-RUN git clone https://github.com/oobabooga/text-generation-webui.git 
-WORKDIR /home/app/text-generation-webui
+RUN git clone https://github.com/oobabooga/textgen.git 
+WORKDIR /home/app/textgen
 RUN GPU_CHOICE=A LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000}
-WORKDIR /home/app/text-generation-webui
+WORKDIR /home/app/textgen
 # set umask to ensure group read / write at runtime
-CMD umask 0002 && export HOME=/home/app/text-generation-webui && ./start_linux.sh --listen
+CMD umask 0002 && export HOME=/home/app/textgen && ./start_linux.sh --listen
diff --git a/docker/nvidia/docker-compose.yml b/docker/nvidia/docker-compose.yml
index ee8282747d..078baa05f9 100644
--- a/docker/nvidia/docker-compose.yml
+++ b/docker/nvidia/docker-compose.yml
@@ -1,6 +1,6 @@
 version: "3.3"
 services:
-  text-generation-webui:
+  textgen:
     build:
       context: .
       args:
@@ -17,7 +17,7 @@ services:
     stdin_open: true
     tty: true
     volumes:
-      - ./user_data:/home/app/text-generation-webui/user_data
+      - ./user_data:/home/app/textgen/user_data
     deploy:
       resources:
         reservations:
diff --git a/docs/01 - Chat Tab.md b/docs/01 - Chat Tab.md
index 96b232fa31..5425e63acc 100644
--- a/docs/01 - Chat Tab.md	
+++ b/docs/01 - Chat Tab.md	
@@ -146,14 +146,14 @@ Note that you can get creative: instead of writing something trivial like "Write
 
 And it works:
 
-![chat-instruct](https://github.com/oobabooga/text-generation-webui/assets/112222186/e38e3469-8263-4a10-b1a1-3c955026b8e7)
+![chat-instruct](https://github.com/oobabooga/textgen/assets/112222186/e38e3469-8263-4a10-b1a1-3c955026b8e7)
 
 ## Chat style
 
-This defines the visual style of the chat UI. Each option is a CSS file defined under `text-generation-webui/css/chat_style-name.css`, where "name" is how this style is called in the dropdown menu. You can add new styles by simply copying `chat_style-cai-chat.css` to `chat_style-myNewStyle.css` and editing the contents of this new file. If you end up with a style that you like, you are highly encouraged to submit it to the repository.
+This defines the visual style of the chat UI. Each option is a CSS file defined under `textgen/css/chat_style-name.css`, where "name" is how this style is called in the dropdown menu. You can add new styles by simply copying `chat_style-cai-chat.css` to `chat_style-myNewStyle.css` and editing the contents of this new file. If you end up with a style that you like, you are highly encouraged to submit it to the repository.
 
-The styles are only applied to chat and chat-instruct modes. Instruct mode has its separate style defined in `text-generation-webui/css/html_instruct_style.css`.
+The styles are only applied to chat and chat-instruct modes. Instruct mode has its separate style defined in `textgen/css/html_instruct_style.css`.
 
 ## Character gallery
 
-This menu is a built-in extension defined under `text-generation-webui/extensions/gallery`. It displays a gallery with your characters, and if you click on a character, it will be automatically selected in the Character tab.
+This menu is a built-in extension defined under `textgen/extensions/gallery`. It displays a gallery with your characters, and if you click on a character, it will be automatically selected in the Character tab.
diff --git a/docs/02 - Default and Notebook Tabs.md b/docs/02 - Default and Notebook Tabs.md
index ba1028abf2..fd027e86da 100644
--- a/docs/02 - Default and Notebook Tabs.md	
+++ b/docs/02 - Default and Notebook Tabs.md	
@@ -22,7 +22,7 @@ Five tabs can be found:
 
 * **Raw**: where the raw text generated by the model appears.
 * **Markdown**: it contains a "Render" button. You can click on it at any time to render the current output as markdown. This is particularly useful for models that generate LaTeX equations like GALACTICA.
-* **HTML**: displays the output in an HTML style that is meant to be easier to read. Its style is defined under `text-generation-webui/css/html_readable_style.css`.
+* **HTML**: displays the output in an HTML style that is meant to be easier to read. Its style is defined under `textgen/css/html_readable_style.css`.
 * **Logits**: when you click on "Get next token probabilities", this tab displays the 50 most likely next tokens and their probabilities based on your current input. If "Use samplers" is checked, the probabilities will be the ones after the sampling parameters in the "Parameters" > "Generation" tab are applied. Otherwise, they will be the raw probabilities generated by the model.
 * **Tokens**: allows you to tokenize your prompt and see the ID numbers for the individual tokens.
 
diff --git a/docs/04 - Model Tab.md b/docs/04 - Model Tab.md
index 744970ac6a..ba415a93d4 100644
--- a/docs/04 - Model Tab.md	
+++ b/docs/04 - Model Tab.md	
@@ -92,7 +92,7 @@ If the **Autoload the model** checkbox is selected, the model will be loaded as
 
 ## LoRA dropdown
 
-Used to apply LoRAs to the model. Note that LoRA support is not implemented for all loaders. Check the [What Works](https://github.com/oobabooga/text-generation-webui/wiki/What-Works) page for details.
+Used to apply LoRAs to the model. Note that LoRA support is not implemented for all loaders. Check the [What Works](https://github.com/oobabooga/textgen/wiki/What-Works) page for details.
 
 ## Download model or LoRA
 
diff --git a/docs/06 - Session Tab.md b/docs/06 - Session Tab.md
index 48735c369b..c15a0e26d5 100644
--- a/docs/06 - Session Tab.md	
+++ b/docs/06 - Session Tab.md	
@@ -9,14 +9,14 @@ Here you can restart the UI with new settings.
 
 ## Extensions & flags
 
-* **Available extensions**: shows a list of extensions available under `text-generation-webui/extensions` and `text-generation-webui/user_data/extensions`. Note that some of these extensions may require manually installing Python requirements through the command: `pip install -r extensions/extension_name/requirements.txt`.
+* **Available extensions**: shows a list of extensions available under `textgen/extensions` and `textgen/user_data/extensions`. Note that some of these extensions may require manually installing Python requirements through the command: `pip install -r extensions/extension_name/requirements.txt`.
 * **Boolean command-line flags**: shows command-line flags of bool (true/false) type.
 
 After selecting your desired flags and extensions, you can restart the UI by clicking on **Apply flags/extensions and restart**.
 
 ## Install or update an extension
 
-In this field, you can enter the GitHub URL for an extension and press enter to either install it (i.e. cloning it into `text-generation-webui/extensions`) or update it with `git pull` in case it is already cloned.
+In this field, you can enter the GitHub URL for an extension and press enter to either install it (i.e. cloning it into `textgen/extensions`) or update it with `git pull` in case it is already cloned.
 
 Note that some extensions may include additional Python requirements. In this case, to install those you have to run the command
 
diff --git a/docs/07 - Extensions.md b/docs/07 - Extensions.md
index 779b2a3499..c83fbf9dd8 100644
--- a/docs/07 - Extensions.md	
+++ b/docs/07 - Extensions.md	
@@ -1,8 +1,8 @@
 # Extensions
 
 Extensions are defined by files named `script.py` inside subfolders of either:
-- `text-generation-webui/extensions`
-- `text-generation-webui/user_data/extensions`
+- `textgen/extensions`
+- `textgen/user_data/extensions`
 
 They are loaded at startup if the folder name is specified after the `--extensions` flag.
 
@@ -10,7 +10,7 @@ For instance, `extensions/silero_tts/script.py` or `user_data/extensions/silero_
 
 **Note:** Extensions in `user_data/extensions/` take priority over those in `extensions/` when both exist with the same name.
 
-## [text-generation-webui-extensions](https://github.com/oobabooga/text-generation-webui-extensions)
+## [textgen-extensions](https://github.com/oobabooga/textgen-extensions)
 
 The repository above contains a directory of user extensions.
 
@@ -20,19 +20,19 @@ If you create an extension, you are welcome to host it in a GitHub repository an
 
 |Extension|Description|
 |---------|-----------|
-|[superboogav2](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/superboogav2)| Enhanced RAG extension with support for PDF, DOCX, and PPTX files. |
-|[send_pictures](https://github.com/oobabooga/text-generation-webui/blob/main/extensions/send_pictures/)| Creates an image upload field that can be used to send images to the bot in chat mode. Captions are automatically generated using BLIP. |
-|[coqui_tts](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/coqui_tts)| Text-to-speech extension using Coqui XTTS v2. |
-|[silero_tts](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/silero_tts)| Text-to-speech extension using [Silero](https://github.com/snakers4/silero-models). When used in chat mode, responses are replaced with an audio widget. |
-|[whisper_stt](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/whisper_stt)| Allows you to enter your inputs in chat mode using your microphone. |
-|[perplexity_colors](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/perplexity_colors)| Colors each token in the output text by its associated probability, as derived from the model logits. |
-|[google_translate](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/google_translate)| Automatically translates inputs and outputs using Google Translate.|
-|[gallery](https://github.com/oobabooga/text-generation-webui/blob/main/extensions/gallery/)| Creates a gallery with the chat characters and their pictures. |
-|[sd_api_pictures](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/sd_api_pictures)| Allows you to request pictures from the bot in chat mode, which will be generated using the AUTOMATIC1111 Stable Diffusion API. See examples [here](https://github.com/oobabooga/text-generation-webui/pull/309). |
-|[long_replies](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/long_replies)| Forces longer replies by suppressing early newlines in the model output. |
-|[ngrok](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/ngrok)| Allows you to access the web UI remotely using the ngrok reverse tunnel service (free). It's an alternative to the built-in Gradio `--share` feature. |
-|[superbooga](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/superbooga)| An extension that uses ChromaDB to create an arbitrarily large pseudocontext, taking as input text files, URLs, or pasted text. Based on https://github.com/kaiokendev/superbig. |
-|[character_bias](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/character_bias)| Just a very simple example that adds a hidden string at the beginning of the bot's reply in chat mode. |
+|[superboogav2](https://github.com/oobabooga/textgen/tree/main/extensions/superboogav2)| Enhanced RAG extension with support for PDF, DOCX, and PPTX files. |
+|[send_pictures](https://github.com/oobabooga/textgen/blob/main/extensions/send_pictures/)| Creates an image upload field that can be used to send images to the bot in chat mode. Captions are automatically generated using BLIP. |
+|[coqui_tts](https://github.com/oobabooga/textgen/tree/main/extensions/coqui_tts)| Text-to-speech extension using Coqui XTTS v2. |
+|[silero_tts](https://github.com/oobabooga/textgen/tree/main/extensions/silero_tts)| Text-to-speech extension using [Silero](https://github.com/snakers4/silero-models). When used in chat mode, responses are replaced with an audio widget. |
+|[whisper_stt](https://github.com/oobabooga/textgen/tree/main/extensions/whisper_stt)| Allows you to enter your inputs in chat mode using your microphone. |
+|[perplexity_colors](https://github.com/oobabooga/textgen/tree/main/extensions/perplexity_colors)| Colors each token in the output text by its associated probability, as derived from the model logits. |
+|[google_translate](https://github.com/oobabooga/textgen/tree/main/extensions/google_translate)| Automatically translates inputs and outputs using Google Translate.|
+|[gallery](https://github.com/oobabooga/textgen/blob/main/extensions/gallery/)| Creates a gallery with the chat characters and their pictures. |
+|[sd_api_pictures](https://github.com/oobabooga/textgen/tree/main/extensions/sd_api_pictures)| Allows you to request pictures from the bot in chat mode, which will be generated using the AUTOMATIC1111 Stable Diffusion API. See examples [here](https://github.com/oobabooga/textgen/pull/309). |
+|[long_replies](https://github.com/oobabooga/textgen/tree/main/extensions/long_replies)| Forces longer replies by suppressing early newlines in the model output. |
+|[ngrok](https://github.com/oobabooga/textgen/tree/main/extensions/ngrok)| Allows you to access the web UI remotely using the ngrok reverse tunnel service (free). It's an alternative to the built-in Gradio `--share` feature. |
+|[superbooga](https://github.com/oobabooga/textgen/tree/main/extensions/superbooga)| An extension that uses ChromaDB to create an arbitrarily large pseudocontext, taking as input text files, URLs, or pasted text. Based on https://github.com/kaiokendev/superbig. |
+|[character_bias](https://github.com/oobabooga/textgen/tree/main/extensions/character_bias)| Just a very simple example that adds a hidden string at the beginning of the bot's reply in chat mode. |
 
 ## How to write an extension
 
@@ -104,7 +104,7 @@ only the first declaration encountered will be used and the rest will be ignored
 
 ## A full example
 
-The source code below can be found at [extensions/example/script.py](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/example/script.py).
+The source code below can be found at [extensions/example/script.py](https://github.com/oobabooga/textgen/tree/main/extensions/example/script.py).
 
 ```python
 """
diff --git a/docs/08 - Additional Tips.md b/docs/08 - Additional Tips.md
index e6e2b3c9a5..1d51aff4df 100644
--- a/docs/08 - Additional Tips.md	
+++ b/docs/08 - Additional Tips.md	
@@ -1,6 +1,6 @@
 ## Audio notification
 
-If your computer takes a long time to generate each response for the model that you are using, you can enable an audio notification for when the response is completed. This feature was kindly contributed by HappyWorldGames in [#1277](https://github.com/oobabooga/text-generation-webui/pull/1277).
+If your computer takes a long time to generate each response for the model that you are using, you can enable an audio notification for when the response is completed. This feature was kindly contributed by HappyWorldGames in [#1277](https://github.com/oobabooga/textgen/pull/1277).
 
 ### Installation
 
diff --git a/docs/09 - Docker.md b/docs/09 - Docker.md
index 4cc147faf7..69d8aea9db 100644
--- a/docs/09 - Docker.md	
+++ b/docs/09 - Docker.md	
@@ -27,7 +27,7 @@ There are four Docker variants available under `docker/`:
 To launch (using NVIDIA as an example):
 
 ```bash
-cd text-generation-webui/docker/nvidia
+cd textgen/docker/nvidia
 cp ../.env.example .env
 # Optionally edit .env to customize ports, TORCH_CUDA_ARCH_LIST, etc.
 docker compose up --build
diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index 727f6eced4..866b6432d1 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -19,7 +19,7 @@ Add `--api` to your command-line flags.
 
 ### Examples
 
-For the documentation with all the endpoints, parameters and their types, consult `http://127.0.0.1:5000/docs` or the [typing.py](https://github.com/oobabooga/text-generation-webui/blob/main/modules/api/typing.py) file.
+For the documentation with all the endpoints, parameters and their types, consult `http://127.0.0.1:5000/docs` or the [typing.py](https://github.com/oobabooga/textgen/blob/main/modules/api/typing.py) file.
 
 The official examples in the [OpenAI documentation](https://platform.openai.com/docs/api-reference) should also work, and the same parameters apply (although the API here has more optional parameters).
 
diff --git a/docs/Image Generation Tutorial.md b/docs/Image Generation Tutorial.md
index 43d6de7c84..22e9398686 100644
--- a/docs/Image Generation Tutorial.md	
+++ b/docs/Image Generation Tutorial.md	
@@ -10,10 +10,10 @@ This feature allows you to generate images using `diffusers` models like [Tongyi
 1. Clone the repository with
 
 ```
-git clone https://github.com/oobabooga/text-generation-webui
+git clone https://github.com/oobabooga/textgen
 ```
 
-or download it from [here](https://github.com/oobabooga/text-generation-webui/archive/refs/heads/main.zip) and unzip it.
+or download it from [here](https://github.com/oobabooga/textgen/archive/refs/heads/main.zip) and unzip it.
 
 2. Use the one-click installer.
 
@@ -21,7 +21,7 @@ or download it from [here](https://github.com/oobabooga/text-generation-webui/ar
 - Linux: Run `./start_linux.sh`
 - macOS: Run `./start_macos.sh`
 
-Note: Image generation does not work with the portable builds in `.zip` format in the [Releases page](https://github.com/oobabooga/text-generation-webui/releases). You need the "full" version of the web UI.
+Note: Image generation does not work with the portable builds in `.zip` format in the [Releases page](https://github.com/oobabooga/textgen/releases). You need the "full" version of the web UI.
 
 ## Downloading a model
 
@@ -64,7 +64,7 @@ To use this feature, you need to load an LLM in the main "Model" page on the lef
 
 If you have no idea what to use, do this to get started:
 
-1. Download [Qwen3-4B-Q3_K_M.gguf](https://huggingface.co/unsloth/Qwen3-4B-GGUF/resolve/main/Qwen3-4B-Q3_K_M.gguf) to your `text-generation-webui/user_data/models` folder.
+1. Download [Qwen3-4B-Q3_K_M.gguf](https://huggingface.co/unsloth/Qwen3-4B-GGUF/resolve/main/Qwen3-4B-Q3_K_M.gguf) to your `textgen/user_data/models` folder.
 2. Select the model in the dropdown menu in the "Model" page.
 3. Click Load.
 
diff --git a/docs/Multimodal Tutorial.md b/docs/Multimodal Tutorial.md
index a30889f734..d244553004 100644
--- a/docs/Multimodal Tutorial.md	
+++ b/docs/Multimodal Tutorial.md	
@@ -14,7 +14,7 @@ As an example, download
 
 https://huggingface.co/unsloth/gemma-3-4b-it-GGUF/resolve/main/gemma-3-4b-it-Q4_K_S.gguf?download=true
 
-to your `text-generation-webui/user_data/models` folder.
+to your `textgen/user_data/models` folder.
 
 ### 3. Download the associated mmproj file to `user_data/mmproj`
 
@@ -22,7 +22,7 @@ Then download
 
 https://huggingface.co/unsloth/gemma-3-4b-it-GGUF/resolve/main/mmproj-F16.gguf?download=true
 
-to your `text-generation-webui/user_data/mmproj` folder. Name it `mmproj-gemma-3-4b-it-F16.gguf` to give it a recognizable name.
+to your `textgen/user_data/mmproj` folder. Name it `mmproj-gemma-3-4b-it-F16.gguf` to give it a recognizable name.
 
 ### 4. Load the model
 
@@ -63,4 +63,4 @@ Examples of models that you can use:
 
 In the page below you can find some ready-to-use examples:
 
-[Multimodal/vision (llama.cpp and ExLlamaV3)](https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API#multimodalvision-llamacpp-and-exllamav3)
+[Multimodal/vision (llama.cpp and ExLlamaV3)](https://github.com/oobabooga/textgen/wiki/12-%E2%80%90-OpenAI-API#multimodalvision-llamacpp-and-exllamav3)
diff --git a/docs/README.md b/docs/README.md
index 666ee85ca0..c060975b8b 100644
--- a/docs/README.md
+++ b/docs/README.md
@@ -1,5 +1,5 @@
 These files are a mirror of the documentation at:
 
-# https://github.com/oobabooga/text-generation-webui/wiki
+# https://github.com/oobabooga/textgen/wiki
 
 It is recommended to browse it there. Contributions can be sent here and will later be synced with the wiki.
diff --git a/extensions/ngrok/script.py b/extensions/ngrok/script.py
index 7bfb9f6e1f..55da3521d3 100644
--- a/extensions/ngrok/script.py
+++ b/extensions/ngrok/script.py
@@ -18,7 +18,7 @@
 options = {
     'addr': f"{host}:{port}",
     'authtoken_from_env': True,
-    'session_metadata': 'text-generation-webui',
+    'session_metadata': 'textgen',
 }
 
 
diff --git a/extensions/sd_api_pictures/README.MD b/extensions/sd_api_pictures/README.MD
index 67c75e145c..ec3a9013a0 100644
--- a/extensions/sd_api_pictures/README.MD
+++ b/extensions/sd_api_pictures/README.MD
@@ -2,7 +2,7 @@
 TL;DR: Lets the bot answer you with a picture!  
 
 Stable Diffusion API pictures for TextGen, v.1.2.0  
-An extension to [oobabooga's textgen-webui](https://github.com/oobabooga/text-generation-webui) allowing you to receive pics generated by [Automatic1111's SD-WebUI API](https://github.com/AUTOMATIC1111/stable-diffusion-webui)
+An extension to [oobabooga's TextGen](https://github.com/oobabooga/textgen) allowing you to receive pics generated by [Automatic1111's SD-WebUI API](https://github.com/AUTOMATIC1111/stable-diffusion-webui)
 
 
            
 Interface overview
@@ -17,7 +17,7 @@ Load it in the `--chat` mode with `--extension sd_api_pictures` alongside `send_
 
 ## History
 
-Consider the version included with [oobabooga's repository](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/sd_api_pictures) to be STABLE, experimental developments and untested features are pushed in [Brawlence/SD_api_pics](https://github.com/Brawlence/SD_api_pics)
+Consider the version included with [oobabooga's repository](https://github.com/oobabooga/textgen/tree/main/extensions/sd_api_pictures) to be STABLE, experimental developments and untested features are pushed in [Brawlence/SD_api_pics](https://github.com/Brawlence/SD_api_pics)
 
 Lastest change:  
 1.1.0 → 1.1.1 Fixed not having Auto1111's metadata in received images
@@ -48,7 +48,7 @@ Green mark confirms the ability to communicate with Auto1111's API on this addre
 
 ### Persistents settings
 
-Create or modify the `settings.json` in the `text-generation-webui` root directory to override the defaults
+Create or modify the `settings.json` in the `textgen` root directory to override the defaults
 present in script.py, ex:
 
 ```json
diff --git a/extensions/superboogav2/README.md b/extensions/superboogav2/README.md
index 0460c40154..904ff58cc1 100644
--- a/extensions/superboogav2/README.md
+++ b/extensions/superboogav2/README.md
@@ -8,7 +8,7 @@ Enhance your LLM with additional information from text, URLs, and files for more
 
 ## Installation and Activation
 
-1. Start the conda environment by running `cmd_windows.bat` or the equivalent for your system in the root directory of `text-generation-webui`.
+1. Start the conda environment by running `cmd_windows.bat` or the equivalent for your system in the root directory of `textgen`.
 2. Install the necessary packages:
    ```
    pip install -r extensions/superboogav2/requirements.txt
@@ -38,4 +38,4 @@ SuperboogaV2 utilizes MuPDF, pandas, python-docx, and python-pptx to extract tex
 
 SuperboogaV2 processes your data into context-aware chunks, applies cleaning techniques, and stores them as embeddings to minimize redundant computations. Relevance is determined using distance calculations and prioritization of recent information.
 
-For a detailed description and more information, refer to the comments in this pull request: [https://github.com/oobabooga/text-generation-webui/pull/3272](https://github.com/oobabooga/text-generation-webui/pull/3272)
+For a detailed description and more information, refer to the comments in this pull request: [https://github.com/oobabooga/textgen/pull/3272](https://github.com/oobabooga/textgen/pull/3272)
diff --git a/modules/api/typing.py b/modules/api/typing.py
index 56d7f2bc8a..392259a7f7 100644
--- a/modules/api/typing.py
+++ b/modules/api/typing.py
@@ -8,7 +8,7 @@
 
 
 class GenerationOptions(BaseModel):
-    preset: str | None = Field(default=None, description="The name of a file under text-generation-webui/user_data/presets (without the .yaml extension). The sampling parameters that get overwritten by this option are the keys in the default_preset() function in modules/presets.py.")
+    preset: str | None = Field(default=None, description="The name of a file under textgen/user_data/presets (without the .yaml extension). The sampling parameters that get overwritten by this option are the keys in the default_preset() function in modules/presets.py.")
     dynatemp_low: float = shared.args.dynatemp_low
     dynatemp_high: float = shared.args.dynatemp_high
     dynatemp_exponent: float = shared.args.dynatemp_exponent
@@ -173,10 +173,10 @@ def resolve_max_tokens(self):
 
     mode: str = Field(default='instruct', description="Valid options: instruct, chat, chat-instruct.")
 
-    instruction_template: str | None = Field(default=None, description="An instruction template defined under text-generation-webui/user_data/instruction-templates. If not set, the correct template will be automatically obtained from the model metadata.")
+    instruction_template: str | None = Field(default=None, description="An instruction template defined under textgen/user_data/instruction-templates. If not set, the correct template will be automatically obtained from the model metadata.")
     instruction_template_str: str | None = Field(default=None, description="A Jinja2 instruction template. If set, will take precedence over everything else.")
 
-    character: str | None = Field(default=None, description="A character defined under text-generation-webui/user_data/characters. If not set, the default \"Assistant\" character will be used.")
+    character: str | None = Field(default=None, description="A character defined under textgen/user_data/characters. If not set, the default \"Assistant\" character will be used.")
     bot_name: str | None = Field(default=None, description="Overwrites the value set by character field.", alias="name2")
     context: str | None = Field(default=None, description="Overwrites the value set by character field.")
     greeting: str | None = Field(default=None, description="Overwrites the value set by character field.")
@@ -271,7 +271,7 @@ class ModelListResponse(BaseModel):
 class LoadModelRequest(BaseModel):
     model_name: str
     args: dict | None = None
-    instruction_template: str | None = Field(default=None, description="An instruction template defined under text-generation-webui/user_data/instruction-templates. Sets the default template for all subsequent API requests.")
+    instruction_template: str | None = Field(default=None, description="An instruction template defined under textgen/user_data/instruction-templates. Sets the default template for all subsequent API requests.")
     instruction_template_str: str | None = Field(default=None, description="A Jinja2 instruction template string. If set, takes precedence over instruction_template.")
 
 
diff --git a/modules/logging_colors.py b/modules/logging_colors.py
index b9791e2685..c930ab7146 100644
--- a/modules/logging_colors.py
+++ b/modules/logging_colors.py
@@ -1,6 +1,6 @@
 import logging
 
-logger = logging.getLogger('text-generation-webui')
+logger = logging.getLogger('textgen')
 
 
 def setup_logging():
diff --git a/modules/shared.py b/modules/shared.py
index e04f28f3c4..8d79eb077b 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -42,7 +42,7 @@
 need_restart = False
 
 # Parser copied from https://github.com/vladmandic/automatic
-parser = argparse.ArgumentParser(description="Text Generation Web UI", conflict_handler='resolve', add_help=True, formatter_class=lambda prog: argparse.HelpFormatter(prog, max_help_position=55, indent_increment=2, width=200))
+parser = argparse.ArgumentParser(description="TextGen", conflict_handler='resolve', add_help=True, formatter_class=lambda prog: argparse.HelpFormatter(prog, max_help_position=55, indent_increment=2, width=200))
 
 # Basic settings
 group = parser.add_argument_group('Basic settings')
diff --git a/modules/training.py b/modules/training.py
index bca4f02ebe..3782c9fcf5 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -40,7 +40,7 @@ def create_ui():
             tmp = gr.State('')
             with gr.Row():
                 with gr.Column():
-                    gr.Markdown("[Tutorial](https://github.com/oobabooga/text-generation-webui/wiki/05-%E2%80%90-Training-Tab)")
+                    gr.Markdown("[Tutorial](https://github.com/oobabooga/textgen/wiki/05-%E2%80%90-Training-Tab)")
 
                     with gr.Row():
                         copy_from = gr.Dropdown(label='Copy parameters from', value='None', choices=utils.get_available_loras(), elem_classes=['slim-dropdown'], interactive=not mu)
diff --git a/one_click.py b/one_click.py
index 68998734c1..0e835dec4a 100644
--- a/one_click.py
+++ b/one_click.py
@@ -324,7 +324,7 @@ def update_requirements(initial_installation=False, pull=True):
     # Create .git directory if missing
     if not os.path.exists(os.path.join(script_dir, ".git")):
         run_cmd(
-            "git init -b main && git remote add origin https://github.com/oobabooga/text-generation-webui && "
+            "git init -b main && git remote add origin https://github.com/oobabooga/textgen && "
             "git fetch && git symbolic-ref refs/remotes/origin/HEAD refs/remotes/origin/main && "
             "git reset --hard origin/main && git branch --set-upstream-to=origin/main",
             environment=True,
@@ -337,7 +337,7 @@ def update_requirements(initial_installation=False, pull=True):
             "Your current installation uses Python {}.{}, which is outdated.\n"
             "Python {} is now required. A clean installation is needed.\n\n"
             "INSTRUCTIONS:\n"
-            "1. Delete the 'installer_files' folder in your text-generation-webui directory.\n"
+            "1. Delete the 'installer_files' folder in your textgen directory.\n"
             "2. Run the start script again (e.g., start_windows.bat).\n\n"
             "This will create a fresh environment with the latest software.".format(*sys.version_info[:2], PYTHON_VERSION)
         )
@@ -350,7 +350,7 @@ def update_requirements(initial_installation=False, pull=True):
             "Your current installation uses CUDA 12.4, which has been removed.\n"
             "To update to the new default (CUDA 12.8), a clean installation is required.\n\n"
             "INSTRUCTIONS:\n"
-            "1. Delete the 'installer_files' folder in your text-generation-webui directory.\n"
+            "1. Delete the 'installer_files' folder in your textgen directory.\n"
             "2. Run the start script again (e.g., start_windows.bat).\n\n"
             "This will create a fresh environment with the latest software."
         )
diff --git a/server.py b/server.py
index 88936ca6e8..6a6eb4f068 100644
--- a/server.py
+++ b/server.py
@@ -36,7 +36,7 @@ def signal_handler(sig, frame):
     signal.signal(signal.SIGINT, signal.SIG_DFL)
     signal.signal(signal.SIGTERM, signal.SIG_DFL)
 
-    logger.info("Received Ctrl+C. Shutting down Text Generation Web UI gracefully.")
+    logger.info("Received Ctrl+C. Shutting down TextGen gracefully.")
 
     # Explicitly stop LlamaServer to avoid __del__ cleanup issues during shutdown
     if shared.model and shared.model.__class__.__name__ == 'LlamaServer':
@@ -85,7 +85,7 @@ def create_interface():
         'GRADIO_TEMP_DIR': str(gradio_temp_path)
     })
 
-    title = 'Text Generation Web UI'
+    title = 'TextGen'
 
     # Password authentication
     auth = []
@@ -249,7 +249,7 @@ def create_interface():
 
 if __name__ == "__main__":
 
-    logger.info("Starting Text Generation Web UI")
+    logger.info("Starting TextGen")
     do_cmd_flags_warnings()
 
     # Load custom settings

From 5f16b86db00f6b255ce550bd64c0dd807ff2e0bc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Apr 2026 11:44:26 -0700
Subject: [PATCH 1537/1701] Update README

---
 README.md | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/README.md b/README.md
index 0266d047d5..9a4e736bd2 100644
--- a/README.md
+++ b/README.md
@@ -15,14 +15,12 @@
 
 **The original local LLM interface.** Text, vision, tool-calling, training, image generation. UI + API, 100% offline and private.
 
+Check out my GGUF quantization quality benchmarks on [LocalBench](https://localbench.substack.com). 360+ quants tested across unsloth, bartowski, and other uploaders.
+
 |![Image1](https://github.com/oobabooga/screenshots/raw/main/INSTRUCT-3.5.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/CHAT-3.5.png) |
 |:---:|:---:|
 |![Image1](https://github.com/oobabooga/screenshots/raw/main/DEFAULT-3.5.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/PARAMETERS-3.5.png) |
 
-## 🔥 News
-My GGUF quantization benchmarks on [LocalBench](https://localbench.substack.com), where I compare quants from uploaders like unsloth, bartowski, and lmstudio-community:
-- **2026-04-07**: [Gemma 4 31B results](https://localbench.substack.com/p/gemma-4-31b-gguf-kl-divergence)
-
 ## Features
 
 - **Easy setup**: [Portable builds](https://github.com/oobabooga/textgen/releases) (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or a one-click installer for the full feature set.

From 3dcb30b970f501e7ba90aa582582f7bdd589418c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Apr 2026 17:55:46 -0700
Subject: [PATCH 1538/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 8 ++++----
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 8 ++++----
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 15 files changed, 30 insertions(+), 30 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 2e01b6bdca..f8508632b0 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -41,10 +41,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.29/exllamav3-0.0.29+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.29/exllamav3-0.0.29+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index dcd3bacb88..324c8eb3c2 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 0032f6822f..bf709e39e0 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -38,4 +38,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 7f44dd9591..c428ea0a5f 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -38,4 +38,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 454893c031..ace7254bd6 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -38,7 +38,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 827fc23945..2324d2532d 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 116631dd08..1a4c7f7967 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index ce9cccb562..91dcca1956 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -24,4 +24,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 4758046885..58bb1a5cbc 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -24,4 +24,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 0d039a03a3..c360259c30 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index cb375e7812..b1bfa12d6a 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index cff325c683..093fc12160 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index 0d0f63dddc..de86e2470c 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # ik_llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index 6832acb92a..cecb1fd18c 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/ik_llama_cpp_binaries-0.113.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 2bf42bc863..8eb0ba3576 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.113.0/llama_cpp_binaries-0.113.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 533b43640b34dab35dcee3705cdf3b7b6575dbc8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Apr 2026 22:07:20 -0300
Subject: [PATCH 1539/1701] Revert "Update exllamav3"

This reverts commit e9d7feb1517a90af2b5edec37a05789755fcd88f.
---
 requirements/full/requirements.txt | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index f8508632b0..427383ccef 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -45,7 +45,7 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.29/exllamav3-0.0.29+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.29/exllamav3-0.0.29+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"

From fbd95bd5e674f15b5a4f49b9946cb5b7115ab35c Mon Sep 17 00:00:00 2001
From: mamei16 
Date: Wed, 15 Apr 2026 03:25:10 +0200
Subject: [PATCH 1540/1701] Handle Double Quotes and Newline Chars in Gemma-4
 Tool Call Arguments (#7477)

---
 modules/tool_parsing.py | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/modules/tool_parsing.py b/modules/tool_parsing.py
index aa3e0e95be..3e660a7a0a 100644
--- a/modules/tool_parsing.py
+++ b/modules/tool_parsing.py
@@ -463,10 +463,14 @@ def _parse_gemma4_tool_calls(answer: str, tool_names: list[str]):
 
         # Convert to JSON: split on <|"|> tokens so that key quoting
         # only applies outside string values (even-indexed parts),
+        # escape newlines and double quotes in arguments (odd-indexed parts),
         # then rejoin with real quotes.
         parts = content.split('<|"|>')
-        for idx in range(0, len(parts), 2):
-            parts[idx] = re.sub(r'(^|[{,\[])\s*(\w+)\s*:', r'\1"\2":', parts[idx])
+        for idx in range(len(parts)):
+            if idx % 2 == 0:
+                parts[idx] = re.sub(r'(^|[{,\[])\s*(\w+)\s*:', r'\1"\2":', parts[idx])
+            else:
+                parts[idx] = json.dumps(parts[idx])[1:-1] 
         json_str = '"'.join(parts)
 
         try:

From 75bf2feb590f2af2785d1deff92f11891c624f67 Mon Sep 17 00:00:00 2001
From: wiger3 <74871505+wiger3@users.noreply.github.com>
Date: Wed, 15 Apr 2026 03:29:19 +0200
Subject: [PATCH 1541/1701] Logits display improvements (#7486)

---
 modules/llama_cpp_server.py | 29 ++++++++++++++++++-----------
 modules/logits.py           | 20 +++++++++++++-------
 2 files changed, 31 insertions(+), 18 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index c01f5d5bf5..f77d2e07e1 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -296,19 +296,26 @@ def get_logits(self, prompt, state, n_probs=128, use_samplers=False):
             pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
             print()
 
-        for retry in range(5):
-            response = self.session.post(url, json=payload)
-            result = response.json()
+        def _try_fetch_logits():
+            for retry in range(5):
+                response = self.session.post(url, json=payload)
+                result = response.json()
+
+                if "completion_probabilities" in result:
+                    if use_samplers:
+                        return result["completion_probabilities"][0]["top_probs"]
+                    else:
+                        return result["completion_probabilities"][0]["top_logprobs"]
 
-            if "completion_probabilities" in result:
-                if use_samplers:
-                    return result["completion_probabilities"][0]["top_probs"]
-                else:
-                    return result["completion_probabilities"][0]["top_logprobs"]
+                time.sleep(0.05)
+            else:
+                raise Exception(f"Unexpected response format: 'completion_probabilities' not found in {result}")
 
-            time.sleep(0.05)
-        else:
-            raise Exception(f"Unexpected response format: 'completion_probabilities' not found in {result}")
+        result = _try_fetch_logits()
+        for entry in result:
+            if not entry.get('token'):
+                entry['token'] = self.decode([entry['id']])
+        return result
 
     def get_prompt_logprob_entries(self, token_ids, n_probs=5, prompt=""):
         """Get logprob entries for prompt tokens via a single n_predict=0 request.
diff --git a/modules/logits.py b/modules/logits.py
index 473f589033..0984972169 100644
--- a/modules/logits.py
+++ b/modules/logits.py
@@ -36,15 +36,22 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
         return error_message, previous
 
     # llama.cpp case
+    def _escaped(token):
+        chars = []
+        for a in token:
+            # C0 and DEL and C1
+            if ord(a) <= 0x1F or 0x7F <= ord(a) <= 0x9F:
+                chars.append(repr(a)[1:-1])
+            else:
+                chars.append(a)
+        return ''.join(chars)
     if shared.model.__class__.__name__ == 'LlamaServer':
         logprobs = shared.model.get_logits(prompt, state, n_probs=top_logits, use_samplers=use_samplers)
 
         if return_dict:
             output = {}
             for entry in logprobs:
-                token = repr(entry['token'])
-                if len(token) > 2 and token.startswith("'") and token.endswith("'"):
-                    token = token[1:-1]
+                token = _escaped(entry['token'])
 
                 prob = entry['prob'] if use_samplers else np.exp(entry['logprob'])
                 output[token] = prob
@@ -52,12 +59,11 @@ def _get_next_logits(prompt, state, use_samplers, previous, top_logits=25, retur
         else:
             output = ''
             for entry in logprobs:
-                token = repr(entry['token'])
-                if len(token) > 2 and token.startswith("'") and token.endswith("'"):
-                    token = token[1:-1]
+                token = _escaped(entry['token'])
+                token_id = entry['id']
 
                 prob = entry['prob'] if use_samplers else np.exp(entry['logprob'])
-                output += f"{prob:.5f}  -  {token}\n"
+                output += f"{prob:.5f}  -  [{token}] ({token_id})\n"
             return output, previous
 
     # All other model types

From c7ca2f232678b25ebff10e065d5e9e6858744b0b Mon Sep 17 00:00:00 2001
From: chaoliang yan 
Date: Wed, 15 Apr 2026 11:31:26 +1000
Subject: [PATCH 1542/1701] Fix: wrong chat deleted when selection changes
 before confirm (#7483)

---
 js/main.js | 1 +
 1 file changed, 1 insertion(+)

diff --git a/js/main.js b/js/main.js
index 918c85c15a..4678b94c1e 100644
--- a/js/main.js
+++ b/js/main.js
@@ -1004,6 +1004,7 @@ function addMiniDeletes() {
 
     confirmBtn.onclick = (e) => {
       e.stopPropagation();
+      label.querySelector("input").click();
       document.querySelector("#delete_chat-confirm").click();
       resetButtons();
     };

From ed61081a25db4d98ed5030451b85cffb9fd52c1d Mon Sep 17 00:00:00 2001
From: Underscore <47636331+Th-Underscore@users.noreply.github.com>
Date: Tue, 14 Apr 2026 21:36:01 -0400
Subject: [PATCH 1543/1701] UI: Add sky-blue color for quoted text in light
 mode (#7473)

---
 css/main.css | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/css/main.css b/css/main.css
index cfa80d8ee2..3ed2efbd66 100644
--- a/css/main.css
+++ b/css/main.css
@@ -511,6 +511,10 @@ audio {
     color: #f5b031;
 }
 
+.message q {
+    color: #3480be;
+}
+
 .message-body q::before, .message-body q::after {
     content: "";
 }

From a8a0f8dc821e74f2e193a83ada7187b1e0f80376 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Apr 2026 23:13:41 -0300
Subject: [PATCH 1544/1701] Fix version metadata not syncing on Continue
 (closes #7492)

---
 modules/chat.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 7e9cce60e6..f8ff86983b 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1142,8 +1142,8 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             output['internal'][-1] = [text, reply.lstrip(' ')]
             output['visible'][-1] = [visible_text, visible_reply.lstrip(' ')]
 
-        # Keep version metadata in sync during streaming (for regeneration)
-        if regenerate and not state.get('_tool_turn'):
+        # Keep version metadata in sync during streaming (for regeneration/continue)
+        if (regenerate or _continue) and not state.get('_tool_turn'):
             row_idx = len(output['internal']) - 1
             key = f"assistant_{row_idx}"
             current_idx = output['metadata'][key]['current_version_index']
@@ -1181,7 +1181,7 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
 
     # Final sync for version metadata (in case streaming was disabled)
-    if regenerate and not state.get('_tool_turn'):
+    if (regenerate or _continue) and not state.get('_tool_turn'):
         row_idx = len(output['internal']) - 1
         key = f"assistant_{row_idx}"
         current_idx = output['metadata'][key]['current_version_index']

From 9e323425d853e0884cda6037f54bb258587c2df0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Apr 2026 23:18:36 -0300
Subject: [PATCH 1545/1701] Fix row_split not working with ik_llama.cpp (closes
 #7489)

---
 modules/llama_cpp_server.py | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index f77d2e07e1..653f5001bf 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -675,6 +675,7 @@ def _patch_cmd_for_ik(cmd):
       --fit-target         → --fit-margin
       --cache-reuse        → (removed, unsupported)
       --swa-full           → (removed, unsupported)
+      --split-mode row     → --split-mode graph
     """
     # Add Hadamard KV cache rotation when using quantized cache types.
     # This significantly improves quantized cache quality (especially q4_0)
@@ -702,6 +703,9 @@ def _patch_cmd_for_ik(cmd):
             patched.append("--fit-margin")
         elif arg == "--cache-reuse":
             i += 1  # skip the value
+        elif arg == "--split-mode" and i + 1 < len(cmd) and cmd[i + 1] == "row":
+            patched += ["--split-mode", "graph"]
+            i += 1  # skip the value
         elif arg == "--swa-full":
             pass  # bare flag, just drop it
         else:

From 2aee3a1898ea6765e57a10e8604b0bf2d8849a0f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 15 Apr 2026 00:08:06 -0300
Subject: [PATCH 1546/1701] UI: Deduplicate auto-scroll calls with
 queueMicrotask

---
 js/global_scope_js.js | 21 ++++++++++++++-------
 1 file changed, 14 insertions(+), 7 deletions(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 20eeef66a0..101f6d4ba3 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -270,16 +270,23 @@ function removeLastClick() {
   document.getElementById("Remove-last").click();
 }
 
+let _scrollPending = false;
+
 function autoScrollToBottom() {
-  if (!window.isScrolled) {
-    const chatParent = document.getElementById("chat")?.parentNode?.parentNode?.parentNode;
-    if (chatParent) {
-      const maxScroll = chatParent.scrollHeight - chatParent.clientHeight;
-      if (maxScroll > 0 && chatParent.scrollTop < maxScroll - 1) {
-        chatParent.scrollTop = maxScroll;
+  if (_scrollPending) return;
+  _scrollPending = true;
+  queueMicrotask(() => {
+    _scrollPending = false;
+    if (!window.isScrolled) {
+      const chatParent = document.getElementById("chat")?.parentNode?.parentNode?.parentNode;
+      if (chatParent) {
+        const maxScroll = chatParent.scrollHeight - chatParent.clientHeight;
+        if (maxScroll > 0 && chatParent.scrollTop < maxScroll - 1) {
+          chatParent.scrollTop = maxScroll;
+        }
       }
     }
-  }
+  });
 }
 
 function updateInstructPadding() {

From 4bce4a38fcf7abd446d05fde06b60e889245a359 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 15 Apr 2026 00:08:08 -0300
Subject: [PATCH 1547/1701] Attempt at fixing chat scroll getting stuck on
 thinking blocks (#7485)

---
 css/main.css | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/css/main.css b/css/main.css
index 3ed2efbd66..67d705b12f 100644
--- a/css/main.css
+++ b/css/main.css
@@ -415,7 +415,13 @@ audio {
     text-align: start;
     padding-left: 1rem;
     padding-right: 1rem;
-    contain: layout;
+    contain: layout paint;
+}
+
+.message,
+.user-message,
+.assistant-message {
+    contain: layout paint;
 }
 
 .chat .message .timestamp {
@@ -1827,14 +1833,9 @@ button:focus {
 }
 
 .chat-parent {
-    /* Optimize for scrolling performance */
     will-change: scroll-position;
-    contain: style paint;
-
-    /* Ensure GPU acceleration */
+    contain: style;
     transform: translateZ(0);
-
-    /* Prevent layout shifts */
     overflow-anchor: none;
 }
 

From 4e978dd037698c7fc877a9780f023773b5e9b296 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Apr 2026 20:23:16 -0700
Subject: [PATCH 1548/1701] Update README

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 9a4e736bd2..edf45ced8b 100644
--- a/README.md
+++ b/README.md
@@ -15,7 +15,7 @@
 
 **The original local LLM interface.** Text, vision, tool-calling, training, image generation. UI + API, 100% offline and private.
 
-Check out my GGUF quantization quality benchmarks on [LocalBench](https://localbench.substack.com). 360+ quants tested across unsloth, bartowski, and other uploaders.
+For recommended GGUF quants, check out my new project: [LocalBench](https://localbench.substack.com).
 
 |![Image1](https://github.com/oobabooga/screenshots/raw/main/INSTRUCT-3.5.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/CHAT-3.5.png) |
 |:---:|:---:|

From b06b36a6a6c0406efcf1891bbe4d53e0c793e03f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Apr 2026 20:28:55 -0700
Subject: [PATCH 1549/1701] Lint

---
 modules/tool_parsing.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/tool_parsing.py b/modules/tool_parsing.py
index 3e660a7a0a..3ba5840f3a 100644
--- a/modules/tool_parsing.py
+++ b/modules/tool_parsing.py
@@ -470,7 +470,7 @@ def _parse_gemma4_tool_calls(answer: str, tool_names: list[str]):
             if idx % 2 == 0:
                 parts[idx] = re.sub(r'(^|[{,\[])\s*(\w+)\s*:', r'\1"\2":', parts[idx])
             else:
-                parts[idx] = json.dumps(parts[idx])[1:-1] 
+                parts[idx] = json.dumps(parts[idx])[1:-1]
         json_str = '"'.join(parts)
 
         try:

From 734d3c38a0114601818772ee98656f1995ae67a1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Apr 2026 20:30:29 -0700
Subject: [PATCH 1550/1701] Reset bos/eos tokens to defaults before reading
 model metadata

---
 modules/models_settings.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 814e3778ee..2a652f6c8a 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -129,6 +129,8 @@ def get_model_metadata(model):
                     template = json_data['chat_template']
 
     # 3. Fall back to tokenizer_config.json metadata
+    shared.bos_token = ''
+    shared.eos_token = ''
     if path.exists():
         with open(path, 'r', encoding='utf-8') as f:
             metadata = json.loads(f.read())

From f66a01bd06fc9f5cf1adb96ac3af12de03314d99 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 14 Apr 2026 20:59:27 -0700
Subject: [PATCH 1551/1701] UI: Add an error message when no model is loaded

---
 modules/chat.py            | 10 ++++++++++
 modules/text_generation.py |  5 +++++
 modules/utils.py           |  5 ++---
 3 files changed, 17 insertions(+), 3 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index f8ff86983b..71989fface 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1204,6 +1204,11 @@ def impersonate_wrapper(textbox, state):
     text = textbox['text']
     static_output = chat_html_wrapper(state['history'], state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
 
+    model_is_loaded, error_message = utils.check_model_loaded()
+    if not model_is_loaded:
+        import gradio as gr
+        raise gr.Error(error_message)
+
     prompt = generate_chat_prompt('', state, impersonate=True)
     stopping_strings = get_stopping_strings(state)
 
@@ -1252,6 +1257,11 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
     if not character_is_loaded(state):
         return
 
+    model_is_loaded, error_message = utils.check_model_loaded()
+    if not model_is_loaded:
+        import gradio as gr
+        raise gr.Error(error_message)
+
     if state['start_with'] != '' and not _continue:
         if regenerate:
             text, state['history'] = remove_last_message(state['history'])
diff --git a/modules/text_generation.py b/modules/text_generation.py
index 3a9ddab531..da73632913 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -216,6 +216,11 @@ def generate_reply_wrapper(question, state, stopping_strings=None):
     """
     Returns formatted outputs for the UI
     """
+    model_is_loaded, error_message = check_model_loaded()
+    if not model_is_loaded:
+        import gradio as gr
+        raise gr.Error(error_message)
+
     reply = question if not shared.is_seq2seq else ''
     yield formatted_outputs(reply, shared.model_name)
 
diff --git a/modules/utils.py b/modules/utils.py
index c4acf7144d..fb25eef199 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -88,9 +88,8 @@ def natural_keys(text):
 def check_model_loaded():
     if shared.model_name == 'None' or shared.model is None:
         if len(get_available_models()) == 0:
-            error_msg = f"No model is loaded.\n\nTo get started:\n1) Place a GGUF file in your {shared.user_data_dir}/models folder\n2) Go to the Model tab and select it"
-            logger.error(error_msg)
-            return False, error_msg
+            logger.error(f"No model is loaded. To get started: 1) Place a GGUF file in your {shared.user_data_dir}/models folder, 2) Go to the Model tab and select it")
+            return False, f"No model is loaded. Place a GGUF model in your {shared.user_data_dir}/models folder, then select it in the Model tab."
         else:
             error_msg = "No model is loaded. Please select one in the Model tab."
             logger.error(error_msg)

From 5992fa90443955da92c3c930ebdc28fba52b0419 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 15 Apr 2026 08:12:20 -0700
Subject: [PATCH 1552/1701] Fix "Start reply with" crash (closes #7497)

---
 modules/chat.py | 24 +++++++++++++-----------
 1 file changed, 13 insertions(+), 11 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 71989fface..f3a7eb19b5 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1055,6 +1055,14 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
 
     row_idx = len(output['internal']) - 1
 
+    # Check if the current row has version metadata to sync during streaming
+    _version_meta = output['metadata'].get(f"assistant_{row_idx}")
+    _sync_versions = (
+        _version_meta is not None
+        and 'current_version_index' in _version_meta
+        and not state.get('_tool_turn')
+    )
+
     # Collect image attachments for multimodal generation from the entire history
     all_image_attachments = []
     if 'metadata' in output:
@@ -1142,12 +1150,9 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             output['internal'][-1] = [text, reply.lstrip(' ')]
             output['visible'][-1] = [visible_text, visible_reply.lstrip(' ')]
 
-        # Keep version metadata in sync during streaming (for regeneration/continue)
-        if (regenerate or _continue) and not state.get('_tool_turn'):
-            row_idx = len(output['internal']) - 1
-            key = f"assistant_{row_idx}"
-            current_idx = output['metadata'][key]['current_version_index']
-            output['metadata'][key]['versions'][current_idx].update({
+        # Keep version metadata in sync during streaming
+        if _sync_versions:
+            _version_meta['versions'][_version_meta['current_version_index']].update({
                 'content': output['internal'][row_idx][1],
                 'visible_content': output['visible'][row_idx][1]
             })
@@ -1181,11 +1186,8 @@ def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_mess
             output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
 
     # Final sync for version metadata (in case streaming was disabled)
-    if (regenerate or _continue) and not state.get('_tool_turn'):
-        row_idx = len(output['internal']) - 1
-        key = f"assistant_{row_idx}"
-        current_idx = output['metadata'][key]['current_version_index']
-        output['metadata'][key]['versions'][current_idx].update({
+    if _sync_versions:
+        _version_meta['versions'][_version_meta['current_version_index']].update({
             'content': output['internal'][row_idx][1],
             'visible_content': output['visible'][row_idx][1]
         })

From e4168ee8b816c3623e3ad29626bf1c4cdabfdda8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 15 Apr 2026 08:56:57 -0700
Subject: [PATCH 1553/1701] Fix tool responses with Gemma 4 template (closes
 #7498)

---
 modules/chat.py | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index f3a7eb19b5..bbc0d7deb4 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -214,8 +214,10 @@ def _convert_to_tool_responses(messages):
     """Convert role:'tool' messages to tool_responses format.
 
     Templates like Gemma 4 expect tool results as a ``tool_responses``
-    attribute on a message rather than separate ``role: 'tool'`` messages.
-    This function groups consecutive tool messages and rewrites them.
+    attribute on the preceding assistant message rather than separate
+    ``role: 'tool'`` messages.  This function groups consecutive tool
+    messages and attaches them to the assistant message that issued the
+    tool calls.
     """
     result = []
     tc_id_to_name = {}
@@ -250,10 +252,8 @@ def _convert_to_tool_responses(messages):
                 })
                 i += 1
 
-            result.append({
-                'role': 'tool',
-                'tool_responses': tool_responses,
-            })
+            if result and result[-1].get('role') == 'assistant':
+                result[-1]['tool_responses'] = tool_responses
         else:
             result.append(msg)
             i += 1

From dd77820a3ae8766e9f862232ee95aa26ea0dde20 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 15 Apr 2026 11:24:25 -0700
Subject: [PATCH 1554/1701] UI: Fix consecutive thinking blocks rendering with
 Gemma 4

---
 modules/html_generator.py | 14 --------------
 1 file changed, 14 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index e3ebea8dbf..2ac92f601e 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -349,20 +349,6 @@ def convert_to_markdown(string, message_id=None):
     tool_call_pattern = re.compile(r'(.*?)\n(.*?)\n', re.DOTALL)
     tool_calls = list(tool_call_pattern.finditer(string))
 
-    if not tool_calls:
-        # No tool calls — use original single-pass extraction
-        thinking_content, remaining_content = extract_thinking_block(string)
-        blocks = []
-        thinking_html = build_thinking_block(thinking_content, message_id, bool(remaining_content))
-        if thinking_html:
-            blocks.append(thinking_html)
-
-        main_html = build_main_content_block(remaining_content)
-        if main_html:
-            blocks.append(main_html)
-
-        return ''.join(blocks)
-
     # Split string into text segments around tool_call blocks and
     # run extract_thinking_block on each segment for full format support.
     html_parts = []

From a56e281204a5a95d4639760fe5d5fe5c65f15a93 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 15 Apr 2026 11:34:29 -0700
Subject: [PATCH 1555/1701] Clean up unused function

---
 modules/html_generator.py | 8 --------
 1 file changed, 8 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 2ac92f601e..44e80e9912 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -176,14 +176,6 @@ def build_thinking_block(thinking_content, message_id, has_remaining_content, th
     '''
 
 
-def build_main_content_block(content):
-    """Build HTML for the main content block."""
-    if not content:
-        return ""
-
-    return process_markdown_content(content)
-
-
 def process_markdown_content(string):
     """
     Process a string through the markdown conversion pipeline.

From 57be34d7a66b1c850fbec8fc45fc96edeb7a4515 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 15 Apr 2026 13:14:27 -0700
Subject: [PATCH 1556/1701] Fix bos/eos tokens being overwritten after GGUF
 metadata sets them (closes #7496)

---
 modules/models_settings.py | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 2a652f6c8a..2193c1810d 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -42,6 +42,10 @@ def get_model_metadata(model):
             hf_quant_method=quant_method
         )
 
+    # Default bos/eos tokens (may be overridden by GGUF metadata or tokenizer_config.json)
+    shared.bos_token = ''
+    shared.eos_token = ''
+
     # GGUF metadata
     if model_settings['loader'] == 'llama.cpp':
         path = model_path
@@ -129,8 +133,6 @@ def get_model_metadata(model):
                     template = json_data['chat_template']
 
     # 3. Fall back to tokenizer_config.json metadata
-    shared.bos_token = ''
-    shared.eos_token = ''
     if path.exists():
         with open(path, 'r', encoding='utf-8') as f:
             metadata = json.loads(f.read())

From acb2488acf3183d2cfae695241d7fa0f519d6b4a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 15 Apr 2026 19:43:56 -0700
Subject: [PATCH 1557/1701] UI: Make sidebars toggle independently, overlay the
 chat area, and persist state on page refresh

---
 css/main.css        | 18 +++++-----
 js/main.js          | 87 ++++++++++++++-------------------------------
 js/show_controls.js | 58 ++++++++++++++----------------
 3 files changed, 60 insertions(+), 103 deletions(-)

diff --git a/css/main.css b/css/main.css
index 67d705b12f..5d30555087 100644
--- a/css/main.css
+++ b/css/main.css
@@ -976,6 +976,13 @@ audio {
     flex-shrink: 0;
     box-sizing: content-box;
     z-index: 10;
+    position: fixed;
+    top: 0;
+    background: var(--background-fill-primary);
+}
+
+#past-chats-row {
+    left: var(--header-width);
 }
 
 #past-chats-row:not(.negative-header) {
@@ -983,6 +990,7 @@ audio {
 }
 
 #chat-controls {
+    right: 0;
     padding: 1rem;
     padding-bottom: 0;
     overflow-y: scroll;
@@ -1253,16 +1261,6 @@ audio {
     right: 7px;
 }
 
-@media screen and (width <= 924px) {
-    #chat-controls.sidebar-shown {
-        position: fixed;
-        right: 0;
-    }
-
-    #past-chats-row.sidebar-shown {
-        position: fixed;
-    }
-}
 
 /* ----------------------------------------------
   Dark theme
diff --git a/js/main.js b/js/main.js
index 4678b94c1e..4004d4863f 100644
--- a/js/main.js
+++ b/js/main.js
@@ -674,23 +674,21 @@ headerBar.appendChild(navigationToggle);
 const pastChatsToggle = document.getElementById("past-chats-toggle");
 const chatControlsToggle = document.getElementById("chat-controls-toggle");
 
+const SIDEBARS = [
+  { element: headerBar, toggle: navigationToggle, key: "sidebar-header-hidden" },
+  { element: pastChatsRow, toggle: pastChatsToggle, key: "sidebar-past-chats-hidden" },
+  { element: chatControlsRow, toggle: chatControlsToggle, key: "sidebar-chat-controls-hidden" },
+];
+window.SIDEBARS = SIDEBARS;
+
 function handleIndividualSidebarClose(event) {
   const target = event.target;
-
-  // Close navigation bar if click is outside and it is open
-  if (!headerBar.contains(target) && !headerBar.classList.contains("sidebar-hidden")) {
-    toggleSidebar(headerBar, navigationToggle);
-  }
-
-  // Close past chats row if click is outside and it is open
-  if (!pastChatsRow.contains(target) && !pastChatsRow.classList.contains("sidebar-hidden")) {
-    toggleSidebar(pastChatsRow, pastChatsToggle);
-  }
-
-  // Close chat controls row if click is outside and it is open
-  if (!chatControlsRow.contains(target) && !chatControlsRow.classList.contains("sidebar-hidden")) {
-    toggleSidebar(chatControlsRow, chatControlsToggle);
-  }
+  SIDEBARS.forEach(({ element, toggle, key }) => {
+    if (!element.contains(target) && !element.classList.contains("sidebar-hidden")) {
+      toggleSidebar(element, toggle);
+      localStorage.setItem(key, "true");
+    }
+  });
 }
 
 function toggleSidebar(sidebar, toggle) {
@@ -720,12 +718,12 @@ function toggleSidebar(sidebar, toggle) {
   }
 }
 
-// Function to check if the device is mobile
+window.toggleSidebar = toggleSidebar;
+
 function isMobile() {
   return window.innerWidth <= 924;
 }
 
-// Function to initialize sidebars
 function initializeSidebars() {
   const isOnMobile = isMobile();
 
@@ -766,55 +764,22 @@ function initializeSidebars() {
     pastChatsToggle.innerHTML = leftArrowSVG;
     chatControlsToggle.innerHTML = rightArrowSVG;
     navigationToggle.innerHTML = closeMenuSVG;
-  }
-}
 
-// Run the initializer when the page loads
-initializeSidebars();
-
-// Add click event listeners to toggle buttons
-pastChatsToggle.addEventListener("click", () => {
-  const isCurrentlyOpen = !pastChatsRow.classList.contains("sidebar-hidden");
-  toggleSidebar(pastChatsRow, pastChatsToggle);
-
-  // On desktop, sync both sidebars together
-  if (!isMobile()) {
-    if (isCurrentlyOpen) {
-      // If we just closed the left sidebar, also close the right sidebar
-      if (!chatControlsRow.classList.contains("sidebar-hidden")) {
-        toggleSidebar(chatControlsRow, chatControlsToggle);
+    SIDEBARS.forEach(({ element, toggle, key }) => {
+      if (localStorage.getItem(key) === "true") {
+        toggleSidebar(element, toggle);
       }
-    } else {
-      // If we just opened the left sidebar, also open the right sidebar
-      if (chatControlsRow.classList.contains("sidebar-hidden")) {
-        toggleSidebar(chatControlsRow, chatControlsToggle);
-      }
-    }
+    });
   }
-});
-
-chatControlsToggle.addEventListener("click", () => {
-  const isCurrentlyOpen = !chatControlsRow.classList.contains("sidebar-hidden");
-  toggleSidebar(chatControlsRow, chatControlsToggle);
+}
 
-  // On desktop, sync both sidebars together
-  if (!isMobile()) {
-    if (isCurrentlyOpen) {
-      // If we just closed the right sidebar, also close the left sidebar
-      if (!pastChatsRow.classList.contains("sidebar-hidden")) {
-        toggleSidebar(pastChatsRow, pastChatsToggle);
-      }
-    } else {
-      // If we just opened the right sidebar, also open the left sidebar
-      if (pastChatsRow.classList.contains("sidebar-hidden")) {
-        toggleSidebar(pastChatsRow, pastChatsToggle);
-      }
-    }
-  }
-});
+initializeSidebars();
 
-navigationToggle.addEventListener("click", () => {
-  toggleSidebar(headerBar, navigationToggle);
+SIDEBARS.forEach(({ element, toggle, key }) => {
+  toggle.addEventListener("click", () => {
+    toggleSidebar(element, toggle);
+    localStorage.setItem(key, element.classList.contains("sidebar-hidden"));
+  });
 });
 
 //------------------------------------------------
diff --git a/js/show_controls.js b/js/show_controls.js
index d5642dc489..ef9d4f988c 100644
--- a/js/show_controls.js
+++ b/js/show_controls.js
@@ -1,40 +1,34 @@
+if (window._controlsInitialized === undefined) {
+  window._controlsInitialized = false;
+}
+
 function toggle_controls(value) {
-  const navToggle = document.getElementById("navigation-toggle");
-  const pastChatsToggle = document.getElementById("past-chats-toggle");
   const extensions = document.querySelector("#extensions");
   const galleryExtension = document.getElementById("gallery-extension");
 
-  if (value) {
-    // SHOW MODE: Click toggles to show hidden sidebars
-    if (navToggle && document.querySelector(".header_bar")?.classList.contains("sidebar-hidden")) {
-      navToggle.click();
-    }
-    if (pastChatsToggle && document.getElementById("past-chats-row")?.classList.contains("sidebar-hidden")) {
-      pastChatsToggle.click();
-    }
+  if (window._controlsInitialized) {
+    window.SIDEBARS.forEach(({ element, toggle, key }) => {
+      if (value) {
+        if (element && element.classList.contains("sidebar-hidden")) {
+          window.toggleSidebar(element, toggle);
+        }
+        localStorage.removeItem(key);
+      } else {
+        if (element && !element.classList.contains("sidebar-hidden")) {
+          window.toggleSidebar(element, toggle);
+        }
+        localStorage.setItem(key, "true");
+      }
+    });
+  }
 
-    // Show extensions only
-    if (extensions) {
-      extensions.style.display = "inherit";
-    }
-    if (galleryExtension) {
-      galleryExtension.style.display = "block";
-    }
+  if (value) {
+    if (extensions) extensions.style.display = "inherit";
+    if (galleryExtension) galleryExtension.style.display = "block";
   } else {
-    // HIDE MODE: Click toggles to hide visible sidebars
-    if (navToggle && !document.querySelector(".header_bar")?.classList.contains("sidebar-hidden")) {
-      navToggle.click();
-    }
-    if (pastChatsToggle && !document.getElementById("past-chats-row")?.classList.contains("sidebar-hidden")) {
-      pastChatsToggle.click();
-    }
-
-    // Hide extensions only
-    if (extensions) {
-      extensions.style.display = "none";
-    }
-    if (galleryExtension) {
-      galleryExtension.style.display = "none";
-    }
+    if (extensions) extensions.style.display = "none";
+    if (galleryExtension) galleryExtension.style.display = "none";
   }
+
+  window._controlsInitialized = true;
 }

From 21d4c174444a27da4f2efd76fff3e27a886775bc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 15 Apr 2026 20:01:42 -0700
Subject: [PATCH 1558/1701] UI: Fix code block copy button centering and
 light/dark mode colors

---
 css/highlightjs/highlightjs-copy.min.css | 74 +++++++++++++++++++++++-
 1 file changed, 73 insertions(+), 1 deletion(-)

diff --git a/css/highlightjs/highlightjs-copy.min.css b/css/highlightjs/highlightjs-copy.min.css
index 5a94fece08..473ba4e51b 100644
--- a/css/highlightjs/highlightjs-copy.min.css
+++ b/css/highlightjs/highlightjs-copy.min.css
@@ -1 +1,73 @@
-.hljs-copy-wrapper{position:relative;overflow:hidden}.hljs-copy-wrapper:hover .hljs-copy-button,.hljs-copy-button:focus{transform:translateX(0)}.hljs-copy-button{position:absolute;transform:translateX(calc(100% + 1.125em));top:1em;right:1em;width:2rem;height:2rem;text-indent:-9999px;color:#fff;border-radius:.25rem;border:1px solid #ffffff22;background-color:#2d2b57;background-color:var(--hljs-theme-background);background-image:url('data:image/svg+xml;utf-8,');background-repeat:no-repeat;background-position:center;transition:background-color 200ms ease,transform 200ms ease-out}.hljs-copy-button:hover{border-color:#ffffff44}.hljs-copy-button:active{border-color:#ffffff66}.hljs-copy-button[data-copied="true"]{text-indent:0;width:auto;background-image:none}@media(prefers-reduced-motion){.hljs-copy-button{transition:none}}.hljs-copy-alert{clip:rect(0 0 0 0);clip-path:inset(50%);height:1px;overflow:hidden;position:absolute;white-space:nowrap;width:1px}
+.hljs-copy-wrapper {
+    position: relative;
+    overflow: hidden;
+    min-height: 3em;
+}
+
+.hljs-copy-wrapper:hover .hljs-copy-button,
+.hljs-copy-button:focus {
+    transform: translateX(0);
+}
+
+.hljs-copy-button {
+    position: absolute;
+    transform: translateX(calc(100% + 1.125em));
+    top: min(1em, calc(50% - 1rem));
+    right: 1em;
+    width: 2rem;
+    height: 2rem;
+    text-indent: -9999px;
+    color: #1f2328;
+    border-radius: .25rem;
+    border: 1px solid #1f232822;
+    background-color: #2d2b57;
+    background-color: var(--hljs-theme-background);
+    background-image: url('data:image/svg+xml;utf-8,');
+    background-repeat: no-repeat;
+    background-position: center;
+    transition: background-color 200ms ease, transform 200ms ease-out;
+}
+
+.hljs-copy-button:hover {
+    border-color: #1f232844;
+}
+
+.hljs-copy-button:active {
+    border-color: #1f232866;
+}
+
+.dark .hljs-copy-button {
+    color: #fff;
+    border-color: #ffffff22;
+    background-image: url('data:image/svg+xml;utf-8,');
+}
+
+.dark .hljs-copy-button:hover {
+    border-color: #ffffff44;
+}
+
+.dark .hljs-copy-button:active {
+    border-color: #ffffff66;
+}
+
+.hljs-copy-button[data-copied="true"] {
+    text-indent: 0;
+    width: auto;
+    background-image: none;
+}
+
+@media(prefers-reduced-motion) {
+    .hljs-copy-button {
+        transition: none;
+    }
+}
+
+.hljs-copy-alert {
+    clip: rect(0 0 0 0);
+    clip-path: inset(50%);
+    height: 1px;
+    overflow: hidden;
+    position: absolute;
+    white-space: nowrap;
+    width: 1px;
+}

From 78a178de9e8a500714581b545e6247b6bbbb3f84 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 16 Apr 2026 03:32:44 -0700
Subject: [PATCH 1559/1701] UI: Past chats menu improvement

---
 css/main.css    | 1 +
 modules/chat.py | 6 +++---
 2 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/css/main.css b/css/main.css
index 5d30555087..8982b464f1 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1103,6 +1103,7 @@ audio {
     padding: 0.75rem;
     font-size: 12.5px;
     font-weight: 400;
+    margin-right: 8px;
 }
 
 #past-chats .selected,
diff --git a/modules/chat.py b/modules/chat.py
index bbc0d7deb4..7ebffef460 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1736,9 +1736,9 @@ def find_all_histories_with_first_prompts(state):
 
         first_prompt = first_prompt.strip()
 
-        # Truncate the first prompt if it's longer than 30 characters
-        if len(first_prompt) > 30:
-            first_prompt = first_prompt[:30 - 3] + '...'
+        # Truncate the first prompt if it's longer than 28 characters
+        if len(first_prompt) > 28:
+            first_prompt = first_prompt[:28 - 3] + '...'
 
         result.append((first_prompt, filename))
 

From 145f3297a2ea17eae0fc5a16a08719ae654fae67 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 16 Apr 2026 03:51:12 -0700
Subject: [PATCH 1560/1701] UI: Sidebar defaults based on viewport width

---
 css/main.css | 35 +++++++++++++++-----
 js/main.js   | 93 ++++++++++++++++------------------------------------
 2 files changed, 55 insertions(+), 73 deletions(-)

diff --git a/css/main.css b/css/main.css
index 8982b464f1..20daa87414 100644
--- a/css/main.css
+++ b/css/main.css
@@ -976,13 +976,6 @@ audio {
     flex-shrink: 0;
     box-sizing: content-box;
     z-index: 10;
-    position: fixed;
-    top: 0;
-    background: var(--background-fill-primary);
-}
-
-#past-chats-row {
-    left: var(--header-width);
 }
 
 #past-chats-row:not(.negative-header) {
@@ -990,7 +983,6 @@ audio {
 }
 
 #chat-controls {
-    right: 0;
     padding: 1rem;
     padding-bottom: 0;
     overflow-y: scroll;
@@ -1262,6 +1254,33 @@ audio {
     right: 7px;
 }
 
+@media screen and (width <= 924px) {
+    #chat-controls.sidebar-shown {
+        position: fixed;
+        right: 0;
+    }
+
+    #past-chats-row.sidebar-shown {
+        position: fixed;
+    }
+}
+
+@media screen and (min-width: 1470px) {
+    #chat-controls,
+    #past-chats-row {
+        position: fixed;
+        top: 0;
+        background: var(--background-fill-primary);
+    }
+
+    #past-chats-row {
+        left: var(--header-width);
+    }
+
+    #chat-controls {
+        right: 0;
+    }
+}
 
 /* ----------------------------------------------
   Dark theme
diff --git a/js/main.js b/js/main.js
index 4004d4863f..b6da2265a1 100644
--- a/js/main.js
+++ b/js/main.js
@@ -691,33 +691,30 @@ function handleIndividualSidebarClose(event) {
   });
 }
 
-function toggleSidebar(sidebar, toggle) {
-  const isCurrentlyHidden = sidebar.classList.contains("sidebar-hidden");
-  const shouldClose = !isCurrentlyHidden;
-
-  // Apply visibility classes
-  sidebar.classList.toggle("sidebar-hidden", shouldClose);
-  sidebar.classList.toggle("sidebar-shown", !shouldClose);
+function setSidebarState(sidebar, toggle, hidden) {
+  sidebar.classList.toggle("sidebar-hidden", hidden);
+  sidebar.classList.toggle("sidebar-shown", !hidden);
 
   if (sidebar === headerBar) {
-    // Special handling for header bar
-    document.documentElement.style.setProperty("--header-width", shouldClose ? "0px" : "112px");
-    pastChatsRow.classList.toggle("negative-header", shouldClose);
-    pastChatsToggle.classList.toggle("negative-header", shouldClose);
-    toggle.innerHTML = shouldClose ? hamburgerMenuSVG : closeMenuSVG;
+    document.documentElement.style.setProperty("--header-width", hidden ? "0px" : "112px");
+    pastChatsRow.classList.toggle("negative-header", hidden);
+    pastChatsToggle.classList.toggle("negative-header", hidden);
+    toggle.innerHTML = hidden ? hamburgerMenuSVG : closeMenuSVG;
   } else if (sidebar === pastChatsRow) {
-    // Past chats sidebar
-    toggle.classList.toggle("past-chats-closed", shouldClose);
-    toggle.classList.toggle("past-chats-open", !shouldClose);
-    toggle.innerHTML = shouldClose ? rightArrowSVG : leftArrowSVG;
+    toggle.classList.toggle("past-chats-closed", hidden);
+    toggle.classList.toggle("past-chats-open", !hidden);
+    toggle.innerHTML = hidden ? rightArrowSVG : leftArrowSVG;
   } else if (sidebar === chatControlsRow) {
-    // Chat controls sidebar
-    toggle.classList.toggle("chat-controls-closed", shouldClose);
-    toggle.classList.toggle("chat-controls-open", !shouldClose);
-    toggle.innerHTML = shouldClose ? leftArrowSVG : rightArrowSVG;
+    toggle.classList.toggle("chat-controls-closed", hidden);
+    toggle.classList.toggle("chat-controls-open", !hidden);
+    toggle.innerHTML = hidden ? leftArrowSVG : rightArrowSVG;
   }
 }
 
+function toggleSidebar(sidebar, toggle) {
+  setSidebarState(sidebar, toggle, !sidebar.classList.contains("sidebar-hidden"));
+}
+
 window.toggleSidebar = toggleSidebar;
 
 function isMobile() {
@@ -725,52 +722,18 @@ function isMobile() {
 }
 
 function initializeSidebars() {
-  const isOnMobile = isMobile();
-
-  if (isOnMobile) {
-    // Mobile state: Hide sidebars and set closed states
-    [pastChatsRow, chatControlsRow, headerBar].forEach(el => {
-      el.classList.add("sidebar-hidden");
-      el.classList.remove("sidebar-shown");
-    });
-
-    document.documentElement.style.setProperty("--header-width", "0px");
-    pastChatsRow.classList.add("negative-header");
-    pastChatsToggle.classList.add("negative-header", "past-chats-closed");
-    pastChatsToggle.classList.remove("past-chats-open");
-
-    [chatControlsToggle, navigationToggle].forEach(el => {
-      el.classList.add("chat-controls-closed");
-      el.classList.remove("chat-controls-open");
-    });
-
-    pastChatsToggle.innerHTML = rightArrowSVG;
-    chatControlsToggle.innerHTML = leftArrowSVG;
-    navigationToggle.innerHTML = hamburgerMenuSVG;
-  } else {
-    // Desktop state: Show sidebars and set open states
-    [pastChatsRow, chatControlsRow].forEach(el => {
-      el.classList.remove("sidebar-hidden", "sidebar-shown");
-    });
-
-    pastChatsToggle.classList.add("past-chats-open");
-    pastChatsToggle.classList.remove("past-chats-closed");
-
-    [chatControlsToggle, navigationToggle].forEach(el => {
-      el.classList.add("chat-controls-open");
-      el.classList.remove("chat-controls-closed");
-    });
-
-    pastChatsToggle.innerHTML = leftArrowSVG;
-    chatControlsToggle.innerHTML = rightArrowSVG;
-    navigationToggle.innerHTML = closeMenuSVG;
+  const width = window.innerWidth;
+  const defaults = {
+    "sidebar-header-hidden": width <= 924,
+    "sidebar-past-chats-hidden": width < 1200,
+    "sidebar-chat-controls-hidden": width < 1470,
+  };
 
-    SIDEBARS.forEach(({ element, toggle, key }) => {
-      if (localStorage.getItem(key) === "true") {
-        toggleSidebar(element, toggle);
-      }
-    });
-  }
+  SIDEBARS.forEach(({ element, toggle, key }) => {
+    const stored = localStorage.getItem(key);
+    const hidden = stored !== null ? stored === "true" : defaults[key];
+    setSidebarState(element, toggle, hidden);
+  });
 }
 
 initializeSidebars();

From b6460908def747dfcf4f206521f3d4dc9c656baa Mon Sep 17 00:00:00 2001
From: Th-Underscore 
Date: Fri, 17 Apr 2026 03:50:36 -0400
Subject: [PATCH 1561/1701] Add model download branch handling in
 download_model_wrapper

Works with `*/tree/` URL or `*:` ID
---
 download-model.py        |  3 +--
 modules/ui_model_menu.py | 22 +++++++++++++++++++++-
 2 files changed, 22 insertions(+), 3 deletions(-)

diff --git a/download-model.py b/download-model.py
index a31bbfc637..e4c0ec543f 100644
--- a/download-model.py
+++ b/download-model.py
@@ -57,8 +57,7 @@ def get_session(self):
         return session
 
     def sanitize_model_and_branch_names(self, model, branch):
-        if model[-1] == '/':
-            model = model[:-1]
+        model = model.removesuffix("/")
 
         if model.startswith(base + '/'):
             model = model[len(base) + 1:]
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 9c8306f50d..7671aeb852 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -239,8 +239,27 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
     downloader_module = importlib.import_module("download-model")
     downloader = downloader_module.ModelDownloader()
     update_queue = queue.Queue()
+    branch = None
 
     try:
+        # Handle branch in URL
+        if "/tree/" in repo_id:
+            try:
+                repo_id, branch = repo_id.split("/tree/")
+            except Exception as e:
+                yield f"Error parsing branch from URL: {e}"
+                progress(0.0)
+                return
+
+        # Handle branch delimited by ":"
+        elif ":" in repo_id:
+            try:
+                repo_id, branch = repo_id.split(":")
+            except Exception as e:
+                yield f"Error parsing branch from repo_id: {e}"
+                progress(0.0)
+                return
+
         # Handle direct GGUF URLs
         if repo_id.startswith("https://") and ("huggingface.co" in repo_id) and (repo_id.endswith(".gguf") or repo_id.endswith(".gguf?download=true")):
             try:
@@ -256,6 +275,7 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
                 progress(0.0)
                 return
 
+
         if not repo_id:
             yield "Please enter a model path."
             progress(0.0)
@@ -266,7 +286,7 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
 
         progress(0.0, "Preparing download...")
 
-        model, branch = downloader.sanitize_model_and_branch_names(repo_id, None)
+        model, branch = downloader.sanitize_model_and_branch_names(repo_id, branch)
         yield "Getting download links from Hugging Face..."
         links, sha256, is_lora, is_llamacpp, file_sizes = downloader.get_download_links_from_huggingface(model, branch, text_only=False, specific_file=specific_file)
 

From d6d9720eb539a9b102a53d8fcb7b743362297192 Mon Sep 17 00:00:00 2001
From: Th-Underscore 
Date: Fri, 17 Apr 2026 03:57:41 -0400
Subject: [PATCH 1562/1701] Remove double space

---
 modules/ui_model_menu.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 7671aeb852..37fde10652 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -275,7 +275,6 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
                 progress(0.0)
                 return
 
-
         if not repo_id:
             yield "Please enter a model path."
             progress(0.0)

From 40ba6109e0492a9b7faf17cd6d2bf3498435d3d5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 18 Apr 2026 10:38:55 -0700
Subject: [PATCH 1563/1701] UI: Improve border colors in light theme

---
 css/main.css  | 8 ++++----
 modules/ui.py | 2 +-
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/css/main.css b/css/main.css
index 20daa87414..eaf0b86ae5 100644
--- a/css/main.css
+++ b/css/main.css
@@ -614,7 +614,7 @@ audio {
     background: #f3f4f6;
     padding: 0.675rem 2.5rem 0.6rem;
     margin-top: 0.15rem;
-    border: 1px solid #d2d2d8;
+    border: 1px solid var(--border-color-primary);
     border-radius: 1.5rem;
     overflow-y: auto !important;
 }
@@ -1489,7 +1489,7 @@ audio {
 
 .block:has(> .label-wrap) {
     padding: 10px 12px !important;
-    border: 1px solid #d2d2d8;
+    border: 1px solid var(--border-color-primary);
 }
 
 .dark .block:has(> .label-wrap) {
@@ -1662,7 +1662,7 @@ strong {
     max-height: 65vh;
     padding: 10px;
     border-radius: 0.5rem;
-    border: 1px solid #ccc;
+    border: 1px solid var(--border-color-primary);
     background-color: var(--light-theme-gray);
     font-family: inherit;
     font-size: inherit;
@@ -1688,7 +1688,7 @@ strong {
 
 .edit-control-button {
     padding: 6px 12px;
-    border: 1px solid #ccc;
+    border: 1px solid var(--border-color-primary);
     border-radius: 0.75rem;
     cursor: pointer;
     background-color: #f8f9fa;
diff --git a/modules/ui.py b/modules/ui.py
index 3a8390f7ce..ddfe164bf4 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -66,7 +66,7 @@
 if not shared.args.old_colors:
     theme = theme.set(
         # General Colors
-        border_color_primary='#d2d2d8',
+        border_color_primary='rgba(0, 0, 0, 0.15)',
         block_border_color='transparent',
         body_text_color_subdued='#484848',
         background_fill_secondary='#eaeaea',

From ae997d6a365db4367633a0191f46cee50b39bfc6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 18 Apr 2026 11:42:27 -0700
Subject: [PATCH 1564/1701] UI: Remove fixed sidebar positioning on wide
 viewports

This made the main chat scrollbar disappear behind the
right sidebar on desktop.
---
 css/main.css | 16 ----------------
 1 file changed, 16 deletions(-)

diff --git a/css/main.css b/css/main.css
index eaf0b86ae5..baa6cbd4a9 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1265,22 +1265,6 @@ audio {
     }
 }
 
-@media screen and (min-width: 1470px) {
-    #chat-controls,
-    #past-chats-row {
-        position: fixed;
-        top: 0;
-        background: var(--background-fill-primary);
-    }
-
-    #past-chats-row {
-        left: var(--header-width);
-    }
-
-    #chat-controls {
-        right: 0;
-    }
-}
 
 /* ----------------------------------------------
   Dark theme

From beb6e17ac7ffcbb246e987958bc14321b5ceb049 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 18 Apr 2026 11:53:37 -0700
Subject: [PATCH 1565/1701] UI: Fix code block scrollbar flash during page load

---
 css/main.css | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/css/main.css b/css/main.css
index baa6cbd4a9..652d1cdb55 100644
--- a/css/main.css
+++ b/css/main.css
@@ -579,11 +579,14 @@ audio {
     padding: 1px 3px;
     background: #f3f4f6 !important;
     color: #1f2328;
+    scrollbar-width: thin;
+    scrollbar-color: var(--neutral-300) transparent;
 }
 
 .dark .message-body pre > code {
     background: #0d1117 !important;
     color: rgb(201 209 217);
+    scrollbar-color: rgb(255 255 255 / 6.25%) transparent;
 }
 
 .message-body pre > code {

From 6f3b1b03697939dd85c51707a74bd4c7d6f487b8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 18 Apr 2026 13:45:35 -0700
Subject: [PATCH 1566/1701] UI: Only show reasoning effort/enable thinking for
 models that use them

---
 modules/loaders.py       |  8 --------
 modules/ui_chat.py       |  8 +++++---
 modules/ui_model_menu.py | 15 ++++++++++-----
 modules/utils.py         | 10 ++++++++++
 server.py                |  2 ++
 5 files changed, 27 insertions(+), 16 deletions(-)

diff --git a/modules/loaders.py b/modules/loaders.py
index 31b1b51a7c..cc84377e60 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -119,8 +119,6 @@ def transformers_samplers():
         'auto_max_new_tokens',
         'ban_eos_token',
         'add_bos_token',
-        'enable_thinking',
-        'reasoning_effort',
         'skip_special_tokens',
         'static_cache',
         'seed',
@@ -174,8 +172,6 @@ def transformers_samplers():
         'auto_max_new_tokens',
         'ban_eos_token',
         'add_bos_token',
-        'enable_thinking',
-        'reasoning_effort',
         'skip_special_tokens',
         'seed',
         'sampler_priority',
@@ -201,8 +197,6 @@ def transformers_samplers():
         'auto_max_new_tokens',
         'ban_eos_token',
         'add_bos_token',
-        'enable_thinking',
-        'reasoning_effort',
         'seed',
         'skip_special_tokens',
     },
@@ -235,8 +229,6 @@ def transformers_samplers():
         'auto_max_new_tokens',
         'ban_eos_token',
         'add_bos_token',
-        'enable_thinking',
-        'reasoning_effort',
         'seed',
         'sampler_priority',
         'custom_token_bans',
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 14489d9623..69f0bbaf4f 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -79,10 +79,12 @@ def create_ui():
                 with gr.Row():
                     shared.gradio['start_with'] = gr.Textbox(label='Start reply with', placeholder='Sure thing!', value=shared.settings['start_with'], elem_classes=['add_scrollbar'])
 
-                gr.HTML("
            ")
+                show_separator, show_reasoning, show_thinking = utils.get_jinja_control_visibility(shared.settings.get('instruction_template_str', ''))
+
+                shared.gradio['jinja_controls_separator'] = gr.HTML("
            ", visible=show_separator)
 
-                shared.gradio['reasoning_effort'] = gr.Dropdown(value=shared.settings['reasoning_effort'], choices=['low', 'medium', 'high'], label='Reasoning effort', info='Used by GPT-OSS.')
-                shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', info='For models with thinking support.')
+                shared.gradio['reasoning_effort'] = gr.Dropdown(value=shared.settings['reasoning_effort'], choices=['low', 'medium', 'high'], label='Reasoning effort', visible=show_reasoning)
+                shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', visible=show_thinking)
 
                 gr.HTML("
            ")
 
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 9c8306f50d..e5e5a61f72 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -154,15 +154,15 @@ def create_event_handlers():
     # with the model defaults (if any), and then the model is loaded
     shared.gradio['model_menu'].change(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        handle_load_model_event_initial, gradio('model_menu', 'interface_state'), gradio(ui.list_interface_input_elements()) + gradio('interface_state') + gradio('vram_info'), show_progress=False).then(
+        handle_load_model_event_initial, gradio('model_menu', 'interface_state'), gradio(ui.list_interface_input_elements()) + gradio('interface_state') + gradio('vram_info') + gradio('jinja_controls_separator'), show_progress=False).then(
         partial(load_model_wrapper, autoload=False), gradio('model_menu', 'loader'), gradio('model_status'), show_progress=True).success(
-        handle_load_model_event_final, gradio('truncation_length', 'loader', 'interface_state'), gradio('truncation_length', 'filter_by_loader'), show_progress=False)
+        handle_load_model_event_final, gradio('truncation_length', 'loader', 'interface_state'), gradio('truncation_length', 'filter_by_loader', 'jinja_controls_separator', 'reasoning_effort', 'enable_thinking'), show_progress=False)
 
     shared.gradio['load_model'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         update_model_parameters, gradio('interface_state'), None).then(
         partial(load_model_wrapper, autoload=True), gradio('model_menu', 'loader'), gradio('model_status'), show_progress=True).success(
-        handle_load_model_event_final, gradio('truncation_length', 'loader', 'interface_state'), gradio('truncation_length', 'filter_by_loader'), show_progress=False)
+        handle_load_model_event_final, gradio('truncation_length', 'loader', 'interface_state'), gradio('truncation_length', 'filter_by_loader', 'jinja_controls_separator', 'reasoning_effort', 'enable_thinking'), show_progress=False)
 
     shared.gradio['unload_model'].click(handle_unload_model_click, None, gradio('model_status'), show_progress=False).then(
         update_gpu_layers_and_vram, gradio('loader', 'model_menu', 'gpu_layers', 'ctx_size', 'cache_type'), gradio('vram_info'), show_progress=False)
@@ -421,13 +421,18 @@ def handle_load_model_event_initial(model, state):
     output = ui.apply_interface_values(state)
     update_model_parameters(state)  # This updates the command-line flags
 
+    show_separator, _, _ = utils.get_jinja_control_visibility(state.get('instruction_template_str', ''))
+
     vram_info = state.get('vram_info', "
            Estimated VRAM to load the model:
            ")
-    return output + [state] + [vram_info]
+    return output + [state] + [vram_info] + [gr.update(visible=show_separator)]
 
 
 def handle_load_model_event_final(truncation_length, loader, state):
     truncation_length = update_truncation_length(truncation_length, state)
-    return [truncation_length, loader]
+
+    show_separator, show_reasoning, show_thinking = utils.get_jinja_control_visibility(state.get('instruction_template_str', ''))
+
+    return [truncation_length, loader, gr.update(visible=show_separator), gr.update(visible=show_reasoning), gr.update(visible=show_thinking)]
 
 
 def handle_unload_model_click():
diff --git a/modules/utils.py b/modules/utils.py
index fb25eef199..4ce62bc659 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -349,3 +349,13 @@ def get_available_chat_styles():
 
 def get_available_grammars():
     return ['None'] + sorted([item.name for item in list((shared.user_data_dir / 'grammars').glob('*.gbnf'))], key=natural_keys)
+
+
+def get_jinja_control_visibility(template_str):
+    if shared.model_name == 'None':
+        return True, True, True
+
+    show_reasoning = 'reasoning_effort' in template_str
+    show_thinking = 'enable_thinking' in template_str or 'thinking_budget' in template_str
+    show_separator = show_reasoning or show_thinking
+    return show_separator, show_reasoning, show_thinking
diff --git a/server.py b/server.py
index 6a6eb4f068..14a585317e 100644
--- a/server.py
+++ b/server.py
@@ -330,6 +330,8 @@ def create_interface():
     if shared.model_name != 'None':
         model_settings = get_model_metadata(shared.model_name)
         update_model_parameters(model_settings, initial=True)  # hijack the command-line arguments
+        if 'instruction_template_str' in model_settings:
+            shared.settings['instruction_template_str'] = model_settings['instruction_template_str']
 
         # Load the model
         shared.model, shared.tokenizer = load_model(shared.model_name)

From 57643eb64d961c8ce4d2620ae73eec64cec6e372 Mon Sep 17 00:00:00 2001
From: Th-Underscore 
Date: Sat, 18 Apr 2026 20:03:32 -0400
Subject: [PATCH 1567/1701] Account for non-tree URLs

---
 modules/ui_model_menu.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 37fde10652..29367b91b7 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -252,7 +252,7 @@ def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), retur
                 return
 
         # Handle branch delimited by ":"
-        elif ":" in repo_id:
+        elif not repo_id.startswith("http") and ":" in repo_id:
             try:
                 repo_id, branch = repo_id.split(":")
             except Exception as e:

From eb596b8e0ad43c4eac66d446ae091f79ce3d8e20 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 18 Apr 2026 17:13:04 -0700
Subject: [PATCH 1568/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 8 ++++----
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 8 ++++----
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 15 files changed, 30 insertions(+), 30 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 427383ccef..36f414e2cb 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -41,10 +41,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 324c8eb3c2..c1a411f1df 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index bf709e39e0..d205523b92 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -38,4 +38,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index c428ea0a5f..e70c6383dd 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -38,4 +38,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index ace7254bd6..1b6cc67f10 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -38,7 +38,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 2324d2532d..47391efb3e 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 1a4c7f7967..6266c52f64 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 91dcca1956..9654343eea 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -24,4 +24,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 58bb1a5cbc..add6dbc09b 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -24,4 +24,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index c360259c30..b540af6ca5 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index b1bfa12d6a..c22763c0c0 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index 093fc12160..1c16f37168 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index de86e2470c..d03f284643 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # ik_llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index cecb1fd18c..6a42758822 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/ik_llama_cpp_binaries-0.116.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 8eb0ba3576..0c630be853 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.116.0/llama_cpp_binaries-0.116.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 1b3af51dc5ea09436cfe02d8dd63d9be5ce812c5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 19 Apr 2026 11:39:00 -0700
Subject: [PATCH 1569/1701] Update exllamav3 to v0.0.30, pin xformers

---
 requirements/full/requirements.txt | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 36f414e2cb..02bba9be1d 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -30,6 +30,7 @@ transformers==5.5.*
 triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
+xformers==0.0.33.post2
 
 # Gradio
 https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
@@ -45,7 +46,7 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.28/exllamav3-0.0.28+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.30/exllamav3-0.0.30+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.30/exllamav3-0.0.30+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"

From 707b5df4ea5b461e053a96eb26f29482ab4239e3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 19 Apr 2026 20:47:59 -0700
Subject: [PATCH 1570/1701] Add stdio MCP server support via user_data/mcp.json

---
 docs/Tool Calling Tutorial.md |  45 ++++++++++++-
 modules/chat.py               |   6 +-
 modules/tool_use.py           | 116 +++++++++++++++++++++++++++-------
 modules/ui_chat.py            |   2 +-
 server.py                     |   4 ++
 5 files changed, 146 insertions(+), 27 deletions(-)

diff --git a/docs/Tool Calling Tutorial.md b/docs/Tool Calling Tutorial.md
index 7d2a86de03..e8e86da5a6 100644
--- a/docs/Tool Calling Tutorial.md	
+++ b/docs/Tool Calling Tutorial.md	
@@ -82,7 +82,9 @@ You can open the built-in tools in `user_data/tools/` for more examples.
 
 ## MCP servers
 
-You can connect to remote [MCP (Model Context Protocol)](https://modelcontextprotocol.io/) servers to use their tools alongside local ones.
+You can connect to [MCP (Model Context Protocol)](https://modelcontextprotocol.io/) servers to use their tools alongside local ones. Both HTTP and stdio servers are supported.
+
+### HTTP servers
 
 In the chat sidebar, open the **MCP servers** accordion and enter one server URL per line. For servers that require authentication, append headers after the URL separated by commas:
 
@@ -91,6 +93,47 @@ https://example.com/mcp
 https://other.com/mcp,Authorization: Bearer sk-xxx
 ```
 
+### Stdio servers
+
+Stdio MCP servers run as local subprocesses. To configure them, create a `user_data/mcp.json` file using the standard format (compatible with Claude Desktop, Cursor, and LM Studio):
+
+```json
+{
+    "mcpServers": {
+        "filesystem": {
+            "command": "npx",
+            "args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/allowed/dir"]
+        },
+        "another-server": {
+            "command": "python3",
+            "args": ["-m", "my_mcp_server", "--flag", "value"],
+            "env": {
+                "API_KEY": "your-key-here"
+            }
+        }
+    }
+}
+```
+
+The file is detected automatically and a warning is printed at startup when it is found.
+
+**Quick test example:** Install `npx` (comes with Node.js), then create `user_data/mcp.json` with:
+
+```json
+{
+    "mcpServers": {
+        "filesystem": {
+            "command": "npx",
+            "args": ["-y", "@modelcontextprotocol/server-filesystem", "/tmp/folder"]
+        }
+    }
+}
+```
+
+Create the target directory (`mkdir -p /tmp/folder`), start the web UI, load a model with tool-calling support, and try asking: *"What files are in /tmp/folder?"* or *"Write a file called notes.txt in /tmp/folder containing 'MCP is working'"*.
+
+### Tool priority
+
 All tools from the configured servers are automatically discovered and made available to the model during generation. If an MCP tool has the same name as a selected local tool, the local tool takes priority.
 
 ## Tool calling over the API
diff --git a/modules/chat.py b/modules/chat.py
index 7ebffef460..a4e4100ed4 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1288,14 +1288,16 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
     # Load tools if any are selected
     selected = state.get('selected_tools', [])
     mcp_servers = state.get('mcp_servers', '')
+    from modules.tool_use import has_mcp_config
+    has_mcp = has_mcp_config()
     parse_tool_call = None
     _tool_parsers = None
-    if selected or mcp_servers:
+    if selected or mcp_servers or has_mcp:
         from modules.tool_use import load_tools, load_mcp_tools, execute_tool
         from modules.tool_parsing import parse_tool_call, get_tool_call_id, detect_tool_call_format
 
         tool_defs, tool_executors = load_tools(selected)
-        if mcp_servers:
+        if mcp_servers or has_mcp:
             mcp_defs, mcp_executors = load_mcp_tools(mcp_servers)
             for td in mcp_defs:
                 fn = td['function']['name']
diff --git a/modules/tool_use.py b/modules/tool_use.py
index f9ddf9407e..143ca75230 100644
--- a/modules/tool_use.py
+++ b/modules/tool_use.py
@@ -6,6 +6,8 @@
 from modules.logging_colors import logger
 from modules.utils import natural_keys, sanitize_filename
 
+_MCP_JSON_PATH = shared.user_data_dir / 'mcp.json'
+
 
 def get_available_tools():
     """Return sorted list of tool script names from user_data/tools/*.py."""
@@ -57,7 +59,7 @@ def load_tools(selected_names):
 
 
 def _parse_mcp_servers(servers_str):
-    """Parse MCP servers textbox: one server per line, format 'url' or 'url,Header: value,Header2: value2'."""
+    """Parse MCP servers textbox: one HTTP server per line, format 'url' or 'url,Header: value,Header2: value2'."""
     servers = []
     for line in servers_str.strip().splitlines():
         line = line.strip()
@@ -71,7 +73,53 @@ def _parse_mcp_servers(servers_str):
             if ':' in part:
                 key, val = part.split(':', 1)
                 headers[key.strip()] = val.strip()
-        servers.append((url, headers))
+        servers.append({"type": "http", "url": url, "headers": headers})
+    return servers
+
+
+def has_mcp_config():
+    """Check if user_data/mcp.json exists."""
+    return _MCP_JSON_PATH.exists()
+
+
+def _load_mcp_json():
+    """Load stdio MCP servers from user_data/mcp.json (Claude Desktop / Cursor format).
+
+    Expected format:
+    {
+        "mcpServers": {
+            "server-name": {
+                "command": "npx",
+                "args": ["-y", "@modelcontextprotocol/server-filesystem", "/path"],
+                "env": {"KEY": "value"}
+            }
+        }
+    }
+    """
+    if not _MCP_JSON_PATH.exists():
+        return []
+
+    try:
+        with open(_MCP_JSON_PATH) as f:
+            config = json.load(f)
+    except Exception:
+        logger.exception(f'Failed to parse {_MCP_JSON_PATH}')
+        return []
+
+    servers = []
+    for name, entry in config.get('mcpServers', {}).items():
+        command = entry.get('command')
+        if not command:
+            logger.warning(f'MCP server "{name}" in mcp.json is missing "command". Skipping.')
+            continue
+
+        servers.append({
+            "type": "stdio",
+            "command": command,
+            "args": entry.get("args", []),
+            "env": entry.get("env"),
+        })
+
     return servers
 
 
@@ -87,24 +135,45 @@ def _mcp_tool_to_openai(tool):
     }
 
 
-async def _mcp_session(url, headers, callback):
-    """Open an MCP session and pass it to the callback."""
-    from mcp.client.streamable_http import streamablehttp_client
-    from mcp import ClientSession
+def _mcp_server_id(server):
+    """Return a human-readable identifier for a server config."""
+    if server["type"] == "http":
+        return server["url"]
+    elif server["type"] == "stdio":
+        return f'{server["command"]} {" ".join(server["args"])}'
+    else:
+        raise ValueError(f"Unknown MCP server type: {server['type']}")
 
-    async with streamablehttp_client(url, headers=headers or None) as (read_stream, write_stream, _):
-        async with ClientSession(read_stream, write_stream) as session:
-            await session.initialize()
-            return await callback(session)
 
+async def _mcp_session(server, callback):
+    """Open an MCP session and pass it to the callback."""
+    from mcp import ClientSession
 
-def _make_mcp_executor(name, url, headers):
+    if server["type"] == "http":
+        from mcp.client.streamable_http import streamablehttp_client
+        async with streamablehttp_client(server["url"], headers=server["headers"] or None) as (read_stream, write_stream, _):
+            async with ClientSession(read_stream, write_stream) as session:
+                await session.initialize()
+                return await callback(session)
+    elif server["type"] == "stdio":
+        from mcp import StdioServerParameters
+        from mcp.client.stdio import stdio_client
+        params = StdioServerParameters(command=server["command"], args=server["args"], env=server.get("env"))
+        async with stdio_client(params) as (read_stream, write_stream):
+            async with ClientSession(read_stream, write_stream) as session:
+                await session.initialize()
+                return await callback(session)
+    else:
+        raise ValueError(f"Unknown MCP server type: {server['type']}")
+
+
+def _make_mcp_executor(name, server):
     def executor(arguments):
-        return asyncio.run(_call_mcp_tool(name, arguments, url, headers))
+        return asyncio.run(_call_mcp_tool(name, arguments, server))
     return executor
 
 
-async def _connect_mcp_server(url, headers):
+async def _connect_mcp_server(server):
     """Connect to one MCP server and return (tool_defs, executors)."""
 
     async def _discover(session):
@@ -113,13 +182,13 @@ async def _discover(session):
         executors = {}
         for tool in result.tools:
             tool_defs.append(_mcp_tool_to_openai(tool))
-            executors[tool.name] = _make_mcp_executor(tool.name, url, headers)
+            executors[tool.name] = _make_mcp_executor(tool.name, server)
         return tool_defs, executors
 
-    return await _mcp_session(url, headers, _discover)
+    return await _mcp_session(server, _discover)
 
 
-async def _call_mcp_tool(name, arguments, url, headers):
+async def _call_mcp_tool(name, arguments, server):
     """Connect to an MCP server and call a single tool."""
 
     async def _invoke(session):
@@ -132,25 +201,25 @@ async def _invoke(session):
                 parts.append(str(content))
         return '\n'.join(parts) if parts else ''
 
-    return await _mcp_session(url, headers, _invoke)
+    return await _mcp_session(server, _invoke)
 
 
 async def _connect_all_mcp_servers(servers):
     """Connect to all MCP servers concurrently."""
     results = await asyncio.gather(
-        *(_connect_mcp_server(url, headers) for url, headers in servers),
+        *(_connect_mcp_server(server) for server in servers),
         return_exceptions=True
     )
     all_defs = []
     all_executors = {}
-    for (url, _), result in zip(servers, results):
+    for server, result in zip(servers, results):
         if isinstance(result, Exception):
-            logger.exception(f'Failed to connect to MCP server "{url}"', exc_info=result)
+            logger.exception(f'Failed to connect to MCP server "{_mcp_server_id(server)}"', exc_info=result)
             continue
         defs, execs = result
         for td, (fn, ex) in zip(defs, execs.items()):
             if fn in all_executors:
-                logger.warning(f'MCP tool "{fn}" from {url} conflicts with an already loaded tool. Skipping.')
+                logger.warning(f'MCP tool "{fn}" from {_mcp_server_id(server)} conflicts with an already loaded tool. Skipping.')
                 continue
             all_defs.append(td)
             all_executors[fn] = ex
@@ -159,10 +228,11 @@ async def _connect_all_mcp_servers(servers):
 
 def load_mcp_tools(servers_str):
     """
-    Parse MCP servers string and discover tools from each server.
+    Discover tools from MCP servers (HTTP from UI textbox + stdio from mcp.json).
     Returns (tool_defs, executors) in the same format as load_tools.
     """
-    servers = _parse_mcp_servers(servers_str)
+    servers = _parse_mcp_servers(servers_str) if servers_str else []
+    servers += _load_mcp_json()
     if not servers:
         return [], {}
 
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 69f0bbaf4f..f0a2c6d6c0 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -108,7 +108,7 @@ def sync_web_tools(selected):
                 shared.gradio['selected_tools'].change(fn=sync_web_tools, inputs=[shared.gradio['selected_tools']], outputs=[shared.gradio['selected_tools']], show_progress=False)
 
                 with gr.Accordion('MCP servers', open=False):
-                    shared.gradio['mcp_servers'] = gr.Textbox(value=shared.settings.get('mcp_servers', ''), lines=3, max_lines=3, label='', info='One url per line. For headers, write url,Header: value,Header2: value2', elem_classes=['add_scrollbar'])
+                    shared.gradio['mcp_servers'] = gr.Textbox(value=shared.settings.get('mcp_servers', ''), lines=3, max_lines=3, label='', info='One URL per line for HTTP servers. For headers: url,Header: value. For stdio servers, use user_data/mcp.json.', elem_classes=['add_scrollbar'])
 
                 gr.HTML("
            ")
 
diff --git a/server.py b/server.py
index 14a585317e..4fada434e9 100644
--- a/server.py
+++ b/server.py
@@ -259,6 +259,10 @@ def create_interface():
     elif (shared.user_data_dir / 'settings.yaml').exists():
         settings_file = shared.user_data_dir / 'settings.yaml'
 
+    from modules.tool_use import has_mcp_config
+    if has_mcp_config():
+        logger.warning(f"MCP stdio servers will be loaded from \"{shared.user_data_dir / 'mcp.json'}\"")
+
     if settings_file is not None:
         logger.info(f"Loading settings from \"{settings_file}\"")
         with open(settings_file, 'r', encoding='utf-8') as f:

From 1e1d2d37288e1ae8baf47d212712d5b029ccf1c0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 22 Apr 2026 09:45:33 -0700
Subject: [PATCH 1571/1701] Cache MCP tool discovery to avoid re-querying on
 each generation

---
 modules/tool_use.py | 50 ++++++++++++++++++++++++++-------------------
 1 file changed, 29 insertions(+), 21 deletions(-)

diff --git a/modules/tool_use.py b/modules/tool_use.py
index 143ca75230..9e883f8941 100644
--- a/modules/tool_use.py
+++ b/modules/tool_use.py
@@ -204,39 +204,47 @@ async def _invoke(session):
     return await _mcp_session(server, _invoke)
 
 
-async def _connect_all_mcp_servers(servers):
-    """Connect to all MCP servers concurrently."""
-    results = await asyncio.gather(
-        *(_connect_mcp_server(server) for server in servers),
-        return_exceptions=True
-    )
-    all_defs = []
-    all_executors = {}
-    for server, result in zip(servers, results):
-        if isinstance(result, Exception):
-            logger.exception(f'Failed to connect to MCP server "{_mcp_server_id(server)}"', exc_info=result)
-            continue
-        defs, execs = result
-        for td, (fn, ex) in zip(defs, execs.items()):
-            if fn in all_executors:
-                logger.warning(f'MCP tool "{fn}" from {_mcp_server_id(server)} conflicts with an already loaded tool. Skipping.')
-                continue
-            all_defs.append(td)
-            all_executors[fn] = ex
-    return all_defs, all_executors
+_mcp_server_cache = {}
 
 
 def load_mcp_tools(servers_str):
     """
     Discover tools from MCP servers (HTTP from UI textbox + stdio from mcp.json).
     Returns (tool_defs, executors) in the same format as load_tools.
+    Tool discovery is cached per server so each server is only queried once.
     """
     servers = _parse_mcp_servers(servers_str) if servers_str else []
     servers += _load_mcp_json()
     if not servers:
         return [], {}
 
-    return asyncio.run(_connect_all_mcp_servers(servers))
+    uncached = [s for s in servers if _mcp_server_id(s) not in _mcp_server_cache]
+    if uncached:
+        results = asyncio.run(asyncio.gather(
+            *(_connect_mcp_server(s) for s in uncached),
+            return_exceptions=True
+        ))
+        for server, result in zip(uncached, results):
+            sid = _mcp_server_id(server)
+            if isinstance(result, Exception):
+                logger.exception(f'Failed to connect to MCP server "{sid}"', exc_info=result)
+                _mcp_server_cache[sid] = ([], {})
+            else:
+                _mcp_server_cache[sid] = result
+
+    all_defs = []
+    all_executors = {}
+    for server in servers:
+        sid = _mcp_server_id(server)
+        defs, execs = _mcp_server_cache[sid]
+        for td, (fn, ex) in zip(defs, execs.items()):
+            if fn in all_executors:
+                logger.warning(f'MCP tool "{fn}" from {sid} conflicts with an already loaded tool. Skipping.')
+                continue
+            all_defs.append(td)
+            all_executors[fn] = ex
+
+    return all_defs, all_executors
 
 
 def execute_tool(func_name, arguments, executors):

From a253d38953f3a1de298d75d0e6dd1135eae0f35f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 19 Apr 2026 13:19:21 -0700
Subject: [PATCH 1572/1701] Add tool call confirmation with inline
 approve/reject buttons

---
 css/main.css              | 32 ++++++++++++-----
 modules/chat.py           | 73 +++++++++++++++++++++++++++++++++++----
 modules/html_generator.py | 18 ++++++++++
 modules/shared.py         |  1 +
 modules/tool_use.py       | 11 +++---
 modules/ui.py             |  2 ++
 modules/ui_chat.py        | 14 ++++++++
 7 files changed, 133 insertions(+), 18 deletions(-)

diff --git a/css/main.css b/css/main.css
index 652d1cdb55..7e5e4f29a9 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1580,6 +1580,28 @@ audio {
     100% { opacity: 0.6; }
 }
 
+.tool-approval-buttons {
+    display: flex;
+    gap: 8px;
+    max-height: none;
+    overflow-y: visible;
+}
+
+.tool-approval-btn {
+    padding: 6px 12px;
+    border: 1px solid var(--border-color-primary);
+    border-radius: 0.75rem;
+    background: var(--button-secondary-background-fill);
+    color: var(--button-secondary-text-color);
+    cursor: pointer;
+    font-size: 12px;
+    margin-bottom: 0 !important;
+}
+
+.tool-approval-btn:hover {
+    background: var(--button-secondary-background-fill-hover);
+}
+
 strong {
     font-weight: bold;
 }
@@ -1678,18 +1700,12 @@ strong {
     border: 1px solid var(--border-color-primary);
     border-radius: 0.75rem;
     cursor: pointer;
-    background-color: #f8f9fa;
-    color: #212529;
+    background: var(--button-secondary-background-fill);
+    color: var(--button-secondary-text-color);
     font-size: 12px;
     margin: 0;
 }
 
-.dark .edit-control-button {
-    border: 1px solid var(--border-color-dark);
-    background-color: var(--light-gray);
-    color: #efefef;
-}
-
 /* --- Simple Version Navigation --- */
 .version-navigation {
     position: absolute;
diff --git a/modules/chat.py b/modules/chat.py
index a4e4100ed4..2957b17d07 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -22,6 +22,7 @@
 from modules import utils
 from modules.extensions import apply_extensions
 from modules.html_generator import (
+    TOOL_APPROVAL_PENDING,
     chat_html_wrapper,
     convert_to_markdown,
     extract_thinking_block,
@@ -46,6 +47,42 @@
 
 _history_file_lock = threading.Lock()
 
+_tool_approvals = {}
+_tool_approvals_lock = threading.Lock()
+
+
+def request_tool_approval(session_key, tool_name):
+    """Block until the user approves/rejects a tool call. Returns 'approve'|'always'|'reject'."""
+    with _tool_approvals_lock:
+        if session_key not in _tool_approvals:
+            _tool_approvals[session_key] = {
+                "event": threading.Event(),
+                "result": None,
+                "tool_name": None,
+                "approved": set(),
+            }
+    session = _tool_approvals[session_key]
+    session["event"].clear()
+    session["result"] = None
+    session["tool_name"] = tool_name
+    while not session["event"].wait(timeout=0.5):
+        if shared.stop_everything:
+            session["tool_name"] = None
+            return 'reject'
+    session["tool_name"] = None
+    return session["result"]
+
+
+def resolve_tool_approval(session_key, result):
+    """Called by button handlers to resolve a pending approval."""
+    session = _tool_approvals.get(session_key)
+    if not session:
+        return
+    if result == 'always' and session["tool_name"]:
+        session["approved"].add(session["tool_name"])
+    session["result"] = result
+    session["event"].set()
+
 
 def strftime_now(format):
     return datetime.now().strftime(format)
@@ -1470,19 +1507,43 @@ def _render():
         yield _render(), history
 
         # Execute tools, store results, and replace placeholders with real results
+        _session_key = state.get('unique_id', '')
+        def _cancel_remaining(from_idx):
+            for j in range(from_idx, len(parsed_calls)):
+                seq.append({'role': 'tool', 'content': 'Tool execution was cancelled by the user.', 'tool_call_id': parsed_calls[j]['id']})
+                pending_placeholders[j] = f'{tc_headers[j]}\nCancelled\n'
+
+            history['visible'][-1][1] = '\n\n'.join(visible_prefix + pending_placeholders)
+
         for i, tc in enumerate(parsed_calls):
-            # Check for stop request before each tool execution
             if shared.stop_everything:
-                for j in range(i, len(parsed_calls)):
-                    seq.append({'role': 'tool', 'content': 'Tool execution was cancelled by the user.', 'tool_call_id': parsed_calls[j]['id']})
-                    pending_placeholders[j] = f'{tc_headers[j]}\nCancelled\n'
-
-                history['visible'][-1][1] = '\n\n'.join(visible_prefix + pending_placeholders)
+                _cancel_remaining(i)
                 yield _render(), history
                 break
 
             fn_name = tc['function']['name']
             fn_args = tc['function'].get('arguments', {})
+
+            _approved = _tool_approvals[_session_key]["approved"] if _session_key in _tool_approvals else set()
+            if state.get('confirm_tool_calls', False) and fn_name not in _approved:
+                pending_placeholders[i] = f'{tc_headers[i]}\n{TOOL_APPROVAL_PENDING}\n'
+                history['visible'][-1][1] = '\n\n'.join(visible_prefix + pending_placeholders)
+                yield _render(), history
+
+                approval = request_tool_approval(_session_key, fn_name)
+
+                if approval == 'reject' and shared.stop_everything:
+                    _cancel_remaining(i)
+                    yield _render(), history
+                    break
+
+                if approval == 'reject':
+                    seq.append({'role': 'tool', 'content': 'Tool call was rejected by the user.', 'tool_call_id': tc['id']})
+                    pending_placeholders[i] = f'{tc_headers[i]}\nRejected\n'
+                    history['visible'][-1][1] = '\n\n'.join(visible_prefix + pending_placeholders)
+                    yield _render(), history
+                    continue
+
             result = execute_tool(fn_name, fn_args, tool_executors)
 
             seq.append({'role': 'tool', 'content': result, 'tool_call_id': tc['id']})
diff --git a/modules/html_generator.py b/modules/html_generator.py
index 44e80e9912..e060604d40 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -122,6 +122,9 @@ def extract_thinking_block(string):
 
 
 
+TOOL_APPROVAL_PENDING = '\x00approval_pending'
+
+
 def build_tool_call_block(header, body, message_id, index):
     """Build HTML for a tool call accordion block."""
     block_id = f"tool-call-{message_id}-{index}"
@@ -137,6 +140,21 @@ def build_tool_call_block(header, body, message_id, index):
         
            
         '''
 
+    if body == TOOL_APPROVAL_PENDING:
+        return f'''
+        
            
+            
+                {tool_svg_small}
+                {html.escape(header)}
+            
+            
            
+                
+                
+                
+            
            
+        
            
+        '''
+
     # Build a plain 
             directly to avoid highlight.js auto-detection
     escaped_body = html.escape(body)
     return f'''
diff --git a/modules/shared.py b/modules/shared.py
index 8d79eb077b..4217f612bb 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -260,6 +260,7 @@
     'web_search_pages': 3,
     'selected_tools': [],
     'mcp_servers': '',
+    'confirm_tool_calls': False,
     'prompt-notebook': '',
     'preset': 'Top-P' if (user_data_dir / 'presets/Top-P.yaml').exists() else None,
     'max_new_tokens': 512,
diff --git a/modules/tool_use.py b/modules/tool_use.py
index 9e883f8941..05690e69c8 100644
--- a/modules/tool_use.py
+++ b/modules/tool_use.py
@@ -220,10 +220,13 @@ def load_mcp_tools(servers_str):
 
     uncached = [s for s in servers if _mcp_server_id(s) not in _mcp_server_cache]
     if uncached:
-        results = asyncio.run(asyncio.gather(
-            *(_connect_mcp_server(s) for s in uncached),
-            return_exceptions=True
-        ))
+        async def _discover_uncached():
+            return await asyncio.gather(
+                *(_connect_mcp_server(s) for s in uncached),
+                return_exceptions=True
+            )
+
+        results = asyncio.run(_discover_uncached())
         for server, result in zip(uncached, results):
             sid = _mcp_server_id(server)
             if isinstance(result, Exception):
diff --git a/modules/ui.py b/modules/ui.py
index ddfe164bf4..7b81f0b438 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -210,6 +210,7 @@ def list_interface_input_elements():
         'start_with',
         'selected_tools',
         'mcp_servers',
+        'confirm_tool_calls',
         'mode',
         'chat_style',
         'chat-instruct_command',
@@ -436,6 +437,7 @@ def setup_auto_save():
         'chat_template_str',
         'selected_tools',
         'mcp_servers',
+        'confirm_tool_calls',
 
         # Parameters tab (ui_parameters.py) - Generation parameters
         'preset_menu',
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index f0a2c6d6c0..a0d52379a7 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -63,6 +63,11 @@ def create_ui():
                             shared.gradio['Stop'] = gr.Button('Stop', elem_id='stop', visible=False)
                             shared.gradio['Generate'] = gr.Button('Send', elem_id='Generate', variant='primary')
 
+        # Hidden buttons for tool approval (triggered via JS from inline HTML buttons)
+        shared.gradio['tool_approve'] = gr.Button(visible=False, elem_id='tool-approve-btn')
+        shared.gradio['tool_always_approve'] = gr.Button(visible=False, elem_id='tool-always-approve-btn')
+        shared.gradio['tool_reject'] = gr.Button(visible=False, elem_id='tool-reject-btn')
+
         # Hover menu buttons
         with gr.Column(elem_id='chat-buttons'):
             shared.gradio['Regenerate'] = gr.Button('Regenerate (Ctrl + Enter)', elem_id='Regenerate')
@@ -110,6 +115,8 @@ def sync_web_tools(selected):
                 with gr.Accordion('MCP servers', open=False):
                     shared.gradio['mcp_servers'] = gr.Textbox(value=shared.settings.get('mcp_servers', ''), lines=3, max_lines=3, label='', info='One URL per line for HTTP servers. For headers: url,Header: value. For stdio servers, use user_data/mcp.json.', elem_classes=['add_scrollbar'])
 
+                shared.gradio['confirm_tool_calls'] = gr.Checkbox(value=shared.settings.get('confirm_tool_calls', False), label='Confirm tool calls', info='Ask for approval before executing each tool call.')
+
                 gr.HTML("
            ")
 
                 with gr.Row():
@@ -287,6 +294,13 @@ def create_event_handlers():
         stop_everything_event, None, None, queue=False).then(
         chat.redraw_html, gradio(reload_arr), gradio('display'), show_progress=False)
 
+    shared.gradio['tool_approve'].click(
+        lambda uid: chat.resolve_tool_approval(uid or '', 'approve'), gradio('unique_id'), None, queue=False)
+    shared.gradio['tool_always_approve'].click(
+        lambda uid: chat.resolve_tool_approval(uid or '', 'always'), gradio('unique_id'), None, queue=False)
+    shared.gradio['tool_reject'].click(
+        lambda uid: chat.resolve_tool_approval(uid or '', 'reject'), gradio('unique_id'), None, queue=False)
+
     if not shared.args.multi_user:
         shared.gradio['unique_id'].select(
             ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(

From 4e065e6a656ea13649186b5f8be8d5a3b31c784b Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 22 Apr 2026 12:50:04 -0700
Subject: [PATCH 1573/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 8 ++++----
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 8 ++++----
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 15 files changed, 30 insertions(+), 30 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 02bba9be1d..f3e8c3bc5c 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,10 +42,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.30/exllamav3-0.0.30+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.30/exllamav3-0.0.30+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index c1a411f1df..6e7ebb6467 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index d205523b92..e2c57f585d 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -38,4 +38,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index e70c6383dd..f06d62a857 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -38,4 +38,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 1b6cc67f10..019ff03527 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -38,7 +38,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 47391efb3e..da3e15aed9 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 6266c52f64..690bd4ce31 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 9654343eea..99a95566ab 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -24,4 +24,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index add6dbc09b..2fc069443f 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -24,4 +24,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index b540af6ca5..b79f52e603 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index c22763c0c0..3336327194 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index 1c16f37168..7ef6ea2e3b 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index d03f284643..2d6a6e7d4d 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # ik_llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index 6a42758822..ee1c951b50 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/ik_llama_cpp_binaries-0.119.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 0c630be853..9bea07e695 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.119.0/llama_cpp_binaries-0.119.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 9a9f3ae9e86628b0a6b22617aeef9c44fa0f252e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 22 Apr 2026 13:00:48 -0700
Subject: [PATCH 1574/1701] llama.cpp: Pass --draft-min 48 by default for
 draftless speculative decoding

---
 modules/llama_cpp_server.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 653f5001bf..bb244612a9 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -485,6 +485,7 @@ def _start_server(self):
         if shared.args.spec_type != 'none':
             cmd += ["--spec-type", shared.args.spec_type]
             cmd += ["--draft-max", str(shared.args.draft_max)]
+            cmd += ["--draft-min", "48"]
             cmd += ["--spec-ngram-size-n", str(shared.args.spec_ngram_size_n)]
             cmd += ["--spec-ngram-size-m", str(shared.args.spec_ngram_size_m)]
             cmd += ["--spec-ngram-min-hits", str(shared.args.spec_ngram_min_hits)]

From 83347ed7e482101b65751a3ef2fb53343ef6566e Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 22 Apr 2026 13:43:56 -0700
Subject: [PATCH 1575/1701] Add `preserve_thinking` chat template parameter (UI
 and CLI)

---
 modules/api/typing.py    |  1 +
 modules/chat.py          |  1 +
 modules/shared.py        |  2 ++
 modules/ui.py            |  2 ++
 modules/ui_chat.py       |  3 ++-
 modules/ui_model_menu.py | 10 +++++-----
 modules/utils.py         |  7 ++++---
 7 files changed, 17 insertions(+), 9 deletions(-)

diff --git a/modules/api/typing.py b/modules/api/typing.py
index 392259a7f7..5e1e595b52 100644
--- a/modules/api/typing.py
+++ b/modules/api/typing.py
@@ -48,6 +48,7 @@ class GenerationOptions(BaseModel):
     add_bos_token: bool = True
     enable_thinking: bool = shared.args.enable_thinking
     reasoning_effort: str = shared.args.reasoning_effort
+    preserve_thinking: bool = shared.args.preserve_thinking
     skip_special_tokens: bool = True
     static_cache: bool = False
     truncation_length: int = 0
diff --git a/modules/chat.py b/modules/chat.py
index 2957b17d07..1376f462fa 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -341,6 +341,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
         enable_thinking=state['enable_thinking'],
         thinking=state['enable_thinking'],
         reasoning_effort=state['reasoning_effort'],
+        preserve_thinking=state['preserve_thinking'],
         thinking_budget=-1 if state.get('enable_thinking', True) else 0,
         bos_token=shared.bos_token,
         eos_token=shared.eos_token,
diff --git a/modules/shared.py b/modules/shared.py
index 4217f612bb..9fcf9c1f27 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -210,6 +210,7 @@
 group.add_argument('--dry-sequence-breakers', type=str, default=_d['dry_sequence_breakers'], metavar='N', help='DRY sequence breakers')
 group.add_argument('--enable-thinking', action=argparse.BooleanOptionalAction, default=True, help='Enable thinking')
 group.add_argument('--reasoning-effort', type=str, default='medium', metavar='N', help='Reasoning effort')
+group.add_argument('--preserve-thinking', action=argparse.BooleanOptionalAction, default=False, help='Preserve thinking blocks from prior turns in the chat template')
 group.add_argument('--chat-template-file', type=str, default=None, help='Path to a chat template file (.jinja, .jinja2, or .yaml) to use as the default instruction template for API requests. Overrides the model\'s built-in template.')
 
 # Handle CMD_FLAGS.txt
@@ -273,6 +274,7 @@
     'add_bos_token': True,
     'enable_thinking': True,
     'reasoning_effort': 'medium',
+    'preserve_thinking': False,
     'skip_special_tokens': True,
     'stream': True,
     'static_cache': False,
diff --git a/modules/ui.py b/modules/ui.py
index 7b81f0b438..fac7317223 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -179,6 +179,7 @@ def list_interface_input_elements():
         'add_bos_token',
         'enable_thinking',
         'reasoning_effort',
+        'preserve_thinking',
         'skip_special_tokens',
         'stream',
         'static_cache',
@@ -485,6 +486,7 @@ def setup_auto_save():
         'add_bos_token',
         'enable_thinking',
         'reasoning_effort',
+        'preserve_thinking',
         'skip_special_tokens',
         'stream',
         'static_cache',
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index a0d52379a7..2fe9a34194 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -84,12 +84,13 @@ def create_ui():
                 with gr.Row():
                     shared.gradio['start_with'] = gr.Textbox(label='Start reply with', placeholder='Sure thing!', value=shared.settings['start_with'], elem_classes=['add_scrollbar'])
 
-                show_separator, show_reasoning, show_thinking = utils.get_jinja_control_visibility(shared.settings.get('instruction_template_str', ''))
+                show_separator, show_reasoning, show_thinking, show_preserve_thinking = utils.get_jinja_control_visibility(shared.settings.get('instruction_template_str', ''))
 
                 shared.gradio['jinja_controls_separator'] = gr.HTML("
            ", visible=show_separator)
 
                 shared.gradio['reasoning_effort'] = gr.Dropdown(value=shared.settings['reasoning_effort'], choices=['low', 'medium', 'high'], label='Reasoning effort', visible=show_reasoning)
                 shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', visible=show_thinking)
+                shared.gradio['preserve_thinking'] = gr.Checkbox(value=shared.settings['preserve_thinking'], label='Preserve thinking', visible=show_preserve_thinking)
 
                 gr.HTML("
            ")
 
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 0ed37dead4..1dfcd5d60d 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -156,13 +156,13 @@ def create_event_handlers():
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         handle_load_model_event_initial, gradio('model_menu', 'interface_state'), gradio(ui.list_interface_input_elements()) + gradio('interface_state') + gradio('vram_info') + gradio('jinja_controls_separator'), show_progress=False).then(
         partial(load_model_wrapper, autoload=False), gradio('model_menu', 'loader'), gradio('model_status'), show_progress=True).success(
-        handle_load_model_event_final, gradio('truncation_length', 'loader', 'interface_state'), gradio('truncation_length', 'filter_by_loader', 'jinja_controls_separator', 'reasoning_effort', 'enable_thinking'), show_progress=False)
+        handle_load_model_event_final, gradio('truncation_length', 'loader', 'interface_state'), gradio('truncation_length', 'filter_by_loader', 'jinja_controls_separator', 'reasoning_effort', 'enable_thinking', 'preserve_thinking'), show_progress=False)
 
     shared.gradio['load_model'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         update_model_parameters, gradio('interface_state'), None).then(
         partial(load_model_wrapper, autoload=True), gradio('model_menu', 'loader'), gradio('model_status'), show_progress=True).success(
-        handle_load_model_event_final, gradio('truncation_length', 'loader', 'interface_state'), gradio('truncation_length', 'filter_by_loader', 'jinja_controls_separator', 'reasoning_effort', 'enable_thinking'), show_progress=False)
+        handle_load_model_event_final, gradio('truncation_length', 'loader', 'interface_state'), gradio('truncation_length', 'filter_by_loader', 'jinja_controls_separator', 'reasoning_effort', 'enable_thinking', 'preserve_thinking'), show_progress=False)
 
     shared.gradio['unload_model'].click(handle_unload_model_click, None, gradio('model_status'), show_progress=False).then(
         update_gpu_layers_and_vram, gradio('loader', 'model_menu', 'gpu_layers', 'ctx_size', 'cache_type'), gradio('vram_info'), show_progress=False)
@@ -440,7 +440,7 @@ def handle_load_model_event_initial(model, state):
     output = ui.apply_interface_values(state)
     update_model_parameters(state)  # This updates the command-line flags
 
-    show_separator, _, _ = utils.get_jinja_control_visibility(state.get('instruction_template_str', ''))
+    show_separator, _, _, _ = utils.get_jinja_control_visibility(state.get('instruction_template_str', ''))
 
     vram_info = state.get('vram_info', "
            Estimated VRAM to load the model:
            ")
     return output + [state] + [vram_info] + [gr.update(visible=show_separator)]
@@ -449,9 +449,9 @@ def handle_load_model_event_initial(model, state):
 def handle_load_model_event_final(truncation_length, loader, state):
     truncation_length = update_truncation_length(truncation_length, state)
 
-    show_separator, show_reasoning, show_thinking = utils.get_jinja_control_visibility(state.get('instruction_template_str', ''))
+    show_separator, show_reasoning, show_thinking, show_preserve_thinking = utils.get_jinja_control_visibility(state.get('instruction_template_str', ''))
 
-    return [truncation_length, loader, gr.update(visible=show_separator), gr.update(visible=show_reasoning), gr.update(visible=show_thinking)]
+    return [truncation_length, loader, gr.update(visible=show_separator), gr.update(visible=show_reasoning), gr.update(visible=show_thinking), gr.update(visible=show_preserve_thinking)]
 
 
 def handle_unload_model_click():
diff --git a/modules/utils.py b/modules/utils.py
index 4ce62bc659..c3cccf9be7 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -353,9 +353,10 @@ def get_available_grammars():
 
 def get_jinja_control_visibility(template_str):
     if shared.model_name == 'None':
-        return True, True, True
+        return True, True, True, True
 
     show_reasoning = 'reasoning_effort' in template_str
     show_thinking = 'enable_thinking' in template_str or 'thinking_budget' in template_str
-    show_separator = show_reasoning or show_thinking
-    return show_separator, show_reasoning, show_thinking
+    show_preserve_thinking = 'preserve_thinking' in template_str
+    show_separator = show_reasoning or show_thinking or show_preserve_thinking
+    return show_separator, show_reasoning, show_thinking, show_preserve_thinking

From 4737744f4a8e7154bee7e5c8afc42e225fad73ba Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 22 Apr 2026 14:43:16 -0700
Subject: [PATCH 1576/1701] Security: Fix SSRF vulnerabilities in URL fetching

---
 extensions/superbooga/download_urls.py   |  7 ++---
 extensions/superboogav2/download_urls.py |  6 ++---
 modules/image_utils.py                   | 15 ++---------
 modules/web_search.py                    | 33 +++++++++++++++++-------
 4 files changed, 29 insertions(+), 32 deletions(-)

diff --git a/extensions/superbooga/download_urls.py b/extensions/superbooga/download_urls.py
index b28fea42dd..280fc2faaf 100644
--- a/extensions/superbooga/download_urls.py
+++ b/extensions/superbooga/download_urls.py
@@ -1,16 +1,13 @@
 import concurrent.futures
 
-import requests
-
-from modules.web_search import _validate_url
+from modules.web_search import safe_get
 
 
 def download_single(url):
-    _validate_url(url)
     headers = {
         'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/58.0.3029.110 Safari/537.3'
     }
-    response = requests.get(url, headers=headers, timeout=5)
+    response = safe_get(url, headers=headers, timeout=5)
     if response.status_code == 200:
         return response.content
     else:
diff --git a/extensions/superboogav2/download_urls.py b/extensions/superboogav2/download_urls.py
index 4d8b98b1a7..faed861133 100644
--- a/extensions/superboogav2/download_urls.py
+++ b/extensions/superboogav2/download_urls.py
@@ -1,19 +1,17 @@
 import concurrent.futures
 import re
 
-import requests
 from bs4 import BeautifulSoup
 
 import extensions.superboogav2.parameters as parameters
-from modules.web_search import _validate_url
+from modules.web_search import safe_get
 
 from .data_processor import process_and_add_to_collector
 from .utils import create_metadata_source
 
 
 def _download_single(url):
-    _validate_url(url)
-    response = requests.get(url, timeout=5)
+    response = safe_get(url, timeout=5)
     if response.status_code == 200:
         return response.content
     else:
diff --git a/modules/image_utils.py b/modules/image_utils.py
index b313879074..51f09f7d6f 100644
--- a/modules/image_utils.py
+++ b/modules/image_utils.py
@@ -76,19 +76,8 @@ def process_message_content(content: Any) -> Tuple[str, List[Image.Image]]:
                 elif image_url.startswith('http'):
                     # Support external URLs
                     try:
-                        import requests
-                        from urllib.parse import urljoin
-                        from modules.web_search import _validate_url
-                        _validate_url(image_url)
-                        url = image_url
-                        for _ in range(5):
-                            response = requests.get(url, timeout=10, allow_redirects=False)
-                            if response.is_redirect and 'Location' in response.headers:
-                                url = urljoin(url, response.headers['Location'])
-                                _validate_url(url)
-                            else:
-                                break
-
+                        from modules.web_search import safe_get
+                        response = safe_get(image_url, timeout=10)
                         response.raise_for_status()
                         image_data = response.content
                         image = Image.open(io.BytesIO(image_data))
diff --git a/modules/web_search.py b/modules/web_search.py
index 2902c7c055..dd5d2be452 100644
--- a/modules/web_search.py
+++ b/modules/web_search.py
@@ -16,10 +16,18 @@
 
 def _validate_url(url):
     """Validate that a URL is safe to fetch (not targeting private/internal networks)."""
+    # Reject characters that cause parsing discrepancies between urlparse and requests,
+    # which can be exploited to bypass SSRF protections (GHSA-27xf-58m5-vxmc).
+    if '\\' in url:
+        raise ValueError("Invalid URL: backslashes are not allowed")
+
     parsed = urlparse(url)
     if parsed.scheme not in ('http', 'https'):
         raise ValueError(f"Unsupported URL scheme: {parsed.scheme}")
 
+    if '@' in parsed.netloc:
+        raise ValueError("Invalid URL: userinfo (credentials) in URLs is not allowed")
+
     hostname = parsed.hostname
     if not hostname:
         raise ValueError("No hostname in URL")
@@ -34,6 +42,20 @@ def _validate_url(url):
         raise ValueError(f"Could not resolve hostname: {hostname}")
 
 
+def safe_get(url, headers=None, timeout=10, max_redirects=5):
+    """Fetch a URL with SSRF-safe redirect handling. Validates every hop."""
+    _validate_url(url)
+    for _ in range(max_redirects):
+        response = requests.get(url, headers=headers, timeout=timeout, allow_redirects=False)
+        if response.is_redirect and 'Location' in response.headers:
+            url = urljoin(url, response.headers['Location'])
+            _validate_url(url)
+        else:
+            return response
+
+    raise ValueError(f"Too many redirects (max {max_redirects})")
+
+
 def get_current_timestamp():
     """Returns the current time in 24-hour format"""
     return datetime.now().strftime('%b %d, %Y %H:%M')
@@ -46,19 +68,10 @@ def download_web_page(url, timeout=10, include_links=False):
     import trafilatura
 
     try:
-        _validate_url(url)
         headers = {
             'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/130.0.0.0 Safari/537.36'
         }
-        max_redirects = 5
-        for _ in range(max_redirects):
-            response = requests.get(url, headers=headers, timeout=timeout, allow_redirects=False)
-            if response.is_redirect and 'Location' in response.headers:
-                url = urljoin(url, response.headers['Location'])
-                _validate_url(url)
-            else:
-                break
-
+        response = safe_get(url, headers=headers, timeout=timeout)
         response.raise_for_status()
 
         result = trafilatura.extract(

From 1d871c0896996e8291b1328471a2791c64ef738d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 22 Apr 2026 15:17:17 -0700
Subject: [PATCH 1577/1701] Fix Gemma 4 thinking tags not hidden after tool
 calls (closes #7509)

---
 modules/reasoning.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/reasoning.py b/modules/reasoning.py
index 2b26081806..59520158fb 100644
--- a/modules/reasoning.py
+++ b/modules/reasoning.py
@@ -8,6 +8,7 @@
     ('<|channel|>commentary<|message|>', '<|end|>', '<|channel|>final<|message|>'),
     ('', '', None),
     ('<|channel>thought', '', None),  # Gemma 4
+    ('thought\n', '', None),  # Gemma 4 (after tool responses, <|channel> may be absent)
     ('<|think|>', '<|end|>', '<|content|>'),  # Solar Open
     # ('Thinking Process:', '', None),  # Qwen3.5 verbose thinking outside tags -- removed: too prone to false positives in streaming
     (None, '', None),  # End-only variant (e.g., Qwen3-next)

From cb023a9897155eb29d09d65cbd72133d557945e8 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 22 Apr 2026 15:31:47 -0700
Subject: [PATCH 1578/1701] Follow-up to
 a253d38953f3a1de298d75d0e6dd1135eae0f35f

---
 js/global_scope_js.js | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 101f6d4ba3..839d97effe 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -338,7 +338,7 @@ function applyMorphdomUpdate(data) {
   const scrollPositions = {};
   queryScope.querySelectorAll(".thinking-block").forEach(block => {
     const blockId = block.getAttribute("data-block-id");
-    if (blockId && block.hasAttribute("open")) {
+    if (blockId && block.hasAttribute("open") && !block.querySelector(".tool-approval-buttons")) {
       openBlocks.add(blockId);
       const content = block.querySelector(".thinking-content");
       if (content) {
@@ -368,13 +368,11 @@ function applyMorphdomUpdate(data) {
           }
         }
 
-        // For thinking blocks, assume closed by default
+        // For thinking blocks, preserve user-opened state and
+        // server-requested open (e.g. tool approval); otherwise close.
         if (fromEl.classList && fromEl.classList.contains("thinking-block") &&
            toEl.classList && toEl.classList.contains("thinking-block")) {
           const blockId = toEl.getAttribute("data-block-id");
-          // Remove open attribute by default
-          toEl.removeAttribute("open");
-          // If this block was explicitly opened by user, keep it open
           if (blockId && openBlocks.has(blockId)) {
             toEl.setAttribute("open", "");
           }

From 1bb6ae4fe66a7339e6a43519c68276531dc90c6f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 22 Apr 2026 15:52:53 -0700
Subject: [PATCH 1579/1701] Update README

---
 README.md | 115 ++++++++++++++++++++++++++++++------------------------
 1 file changed, 65 insertions(+), 50 deletions(-)

diff --git a/README.md b/README.md
index edf45ced8b..12b26b6ef7 100644
--- a/README.md
+++ b/README.md
@@ -15,41 +15,86 @@
 
 **The original local LLM interface.** Text, vision, tool-calling, training, image generation. UI + API, 100% offline and private.
 
-For recommended GGUF quants, check out my new project: [LocalBench](https://localbench.substack.com).
+[![GitHub stars](https://img.shields.io/github/stars/oobabooga/textgen?style=for-the-badge&logo=github&logoColor=white&labelColor=black)](https://github.com/oobabooga/textgen)
 
-|![Image1](https://github.com/oobabooga/screenshots/raw/main/INSTRUCT-3.5.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/CHAT-3.5.png) |
+## Get started in 1 minute
+
+Zero setup. Download, unzip, run.
+
+**https://github.com/oobabooga/textgen/releases**
+
+Portable builds for Linux, Windows, and macOS with CUDA, Vulkan, ROCm, and CPU-only options. All dependencies included. Compatible with GGUF (llama.cpp) models.
+
+For additional backends (ExLlamaV3, Transformers), training, image generation, and extensions, see [Installation](#installation).
+
+|![Instruct mode](https://github.com/oobabooga/screenshots/raw/main/INSTRUCT-3.5.png) | ![Chat mode](https://github.com/oobabooga/screenshots/raw/main/CHAT-3.5.png) |
 |:---:|:---:|
-|![Image1](https://github.com/oobabooga/screenshots/raw/main/DEFAULT-3.5.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/PARAMETERS-3.5.png) |
+| Instruct mode | Chat mode |
+|![Notebook mode](https://github.com/oobabooga/screenshots/raw/main/DEFAULT-3.5.png) | ![Parameters](https://github.com/oobabooga/screenshots/raw/main/PARAMETERS-3.5.png) |
+| Notebook mode | Parameters |
 
 ## Features
 
-- **Easy setup**: [Portable builds](https://github.com/oobabooga/textgen/releases) (zero setup, just unzip and run) for GGUF models on Windows/Linux/macOS, or a one-click installer for the full feature set.
+### Chat & generation
+
+- `instruct` mode for instruction-following (like ChatGPT), and `chat-instruct`/`chat` modes for talking to custom characters. Prompts are automatically formatted with Jinja2 templates.
+- **Vision (multimodal)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/textgen/wiki/Multimodal-Tutorial)).
+- **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
+- Edit messages, navigate between message versions, and branch conversations at any point.
+- Notebook tab for free-form text generation outside of chat turns.
+
+### Backends & API
+
 - **Multiple backends**: [llama.cpp](https://github.com/ggerganov/llama.cpp), [ik_llama.cpp](https://github.com/ikawrakow/ik_llama.cpp), [Transformers](https://github.com/huggingface/transformers), [ExLlamaV3](https://github.com/turboderp-org/exllamav3), and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM). Switch between backends and models without restarting.
 - **OpenAI/Anthropic-compatible API**: Chat, Completions, and Messages endpoints with tool-calling support. Use as a local drop-in replacement for the OpenAI/Anthropic APIs ([examples](https://github.com/oobabooga/textgen/wiki/12-%E2%80%90-OpenAI-API#examples)).
 - **Tool-calling**: Models can call custom functions during chat, including web search, page fetching, and math. Each tool is a single `.py` file. MCP servers are also supported ([tutorial](https://github.com/oobabooga/textgen/wiki/Tool-Calling-Tutorial)).
-- **Vision (multimodal)**: Attach images to messages for visual understanding ([tutorial](https://github.com/oobabooga/textgen/wiki/Multimodal-Tutorial)).
-- **File attachments**: Upload text files, PDF documents, and .docx documents to talk about their contents.
+
+### Training & image generation
+
 - **Training**: Fine-tune LoRAs on multi-turn chat or raw text datasets. Supports resuming interrupted runs ([tutorial](https://github.com/oobabooga/textgen/wiki/05-%E2%80%90-Training-Tab)).
 - **Image generation**: A dedicated tab for `diffusers` models like **Z-Image-Turbo**. Features 4-bit/8-bit quantization and a persistent gallery with metadata ([tutorial](https://github.com/oobabooga/textgen/wiki/Image-Generation-Tutorial)).
+
+### Privacy & interface
+
 - 100% offline and private, with zero telemetry, external resources, or remote update requests.
-- `instruct` mode for instruction-following (like ChatGPT), and `chat-instruct`/`chat` modes for talking to custom characters. Prompts are automatically formatted with Jinja2 templates.
-- Edit messages, navigate between message versions, and branch conversations at any point.
-- Free-form text generation in the Notebook tab without being limited to chat turns.
 - Dark/light themes, syntax highlighting for code blocks, and LaTeX rendering for mathematical expressions.
-- Extension support, with built-in and user-contributed extensions available. See the [wiki](https://github.com/oobabooga/textgen/wiki/07-%E2%80%90-Extensions) and [extensions directory](https://github.com/oobabooga/textgen-extensions) for details.
+- Built-in and community [extensions](https://github.com/oobabooga/textgen/wiki/07-%E2%80%90-Extensions) including TTS, voice input, and translation. See the [extensions directory](https://github.com/oobabooga/textgen-extensions) for the full list.
 
-## How to install
+## Downloading models
 
-#### ✅ Option 1: Portable builds (get started in 1 minute)
+1. Download a GGUF model file from [Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads&search=gguf).
+2. Place it in the `user_data/models` folder.
 
-No installation needed – just download, unzip and run. All dependencies included.
+That's it. The UI will detect it automatically.
 
-Download from here: **https://github.com/oobabooga/textgen/releases**
+For recommended GGUF quants, check out [LocalBench](https://localbench.substack.com). To estimate how much memory a model will use, try the [GGUF Memory Calculator](https://huggingface.co/spaces/oobabooga/accurate-gguf-vram-calculator).
 
-- Builds are provided for Linux, Windows, and macOS, with options for CUDA, Vulkan, ROCm, and CPU-only.
-- Compatible with GGUF (llama.cpp) models.
+
            
+Other model types (Transformers, EXL3)
 
-#### Option 2: Manual portable install with venv
+Models that consist of multiple files (like 16-bit Transformers models and EXL3 models) should be placed in a subfolder inside `user_data/models`:
+
+```
+textgen
+└── user_data
+    └── models
+        └── Qwen_Qwen3-8B
+            ├── config.json
+            ├── generation_config.json
+            ├── model-00001-of-00004.safetensors
+            ├── ...
+            ├── tokenizer_config.json
+            └── tokenizer.json
+```
+
+These formats require the one-click installer (not the portable build).
+
            
+
+## Installation
+
+For a zero-setup option, see the [portable builds](https://github.com/oobabooga/textgen/releases).
+
+### Manual portable install with venv
 
 Fast setup on any Python 3.9+:
 
@@ -77,9 +122,9 @@ python server.py --portable --api --auto-launch
 deactivate
 ```
 
-#### Option 3: One-click installer
+### One-click installer
 
-For users who need additional backends (ExLlamaV3, Transformers), training, image generation, or extensions (TTS, voice input, translation, etc). Requires ~10GB disk space and downloads PyTorch.
+For users who need additional backends (ExLlamaV3, Transformers), training, image generation, or extensions like TTS, voice input, and translation. Requires ~10GB disk space and downloads PyTorch.
 
 1. Clone the repository, or [download its source code](https://github.com/oobabooga/textgen/archive/refs/heads/main.zip) and extract it.
 2. Run the startup script for your OS: `start_windows.bat`, `start_linux.sh`, or `start_macos.sh`.
@@ -420,43 +465,13 @@ API generation defaults:
 
 
            
 
-## Downloading models
-
-1. Download a GGUF model file from [Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads&search=gguf).
-2. Place it in the `user_data/models` folder.
-
-That's it. The UI will detect it automatically.
-
-To estimate how much memory a model will use, you can use the [GGUF Memory Calculator](https://huggingface.co/spaces/oobabooga/accurate-gguf-vram-calculator).
-
-
            
-Other model types (Transformers, EXL3)
-
-Models that consist of multiple files (like 16-bit Transformers models and EXL3 models) should be placed in a subfolder inside `user_data/models`:
-
-```
-textgen
-└── user_data
-    └── models
-        └── Qwen_Qwen3-8B
-            ├── config.json
-            ├── generation_config.json
-            ├── model-00001-of-00004.safetensors
-            ├── ...
-            ├── tokenizer_config.json
-            └── tokenizer.json
-```
-
-These formats require the one-click installer (not the portable build).
-
            
-
 ## Documentation
 
 https://github.com/oobabooga/textgen/wiki
 
 ## Community
 
-https://www.reddit.com/r/Oobabooga/
+[![Reddit](https://img.shields.io/reddit/subreddit-subscribers/Oobabooga?style=for-the-badge&logo=reddit&logoColor=white&label=r%2FOobabooga&labelColor=black&color=FF4500)](https://www.reddit.com/r/Oobabooga/)
 
 ## Acknowledgments
 

From 7338120bff86b3fd640aca948ced5b328382cfa4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 22 Apr 2026 18:08:59 -0700
Subject: [PATCH 1580/1701] Fix GPT-OSS channel tokens leaking in UI after tool
 calls

---
 modules/chat.py           | 27 +++++++++++++++++++--------
 modules/html_generator.py |  1 +
 2 files changed, 20 insertions(+), 8 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 1376f462fa..1e01988688 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -204,6 +204,23 @@ def _deserialize_tool_call_arguments(tool_calls):
     return result
 
 
+def _strip_channel_tokens(text):
+    """Strip GPT-OSS ``<|channel|>…<|message|>…<|end|>`` wrappers from
+    user-facing content (``final`` or ``commentary`` channels).
+
+    Analysis/thinking channels are left untouched so the reasoning
+    extraction pipeline can handle them separately.
+    """
+    text = text.strip()
+    for tag in ('<|channel|>final<|message|>', '<|channel|>commentary<|message|>'):
+        _, found, after = text.partition(tag)
+        if found:
+            inner, _, _ = after.partition('<|end|>')
+            return inner.strip()
+
+    return text
+
+
 def _expand_tool_sequence(tool_seq):
     """Expand a tool_sequence list into API messages.
 
@@ -1475,14 +1492,7 @@ def _render():
 
         seq_entry = {'tool_calls': serialized}
         if content_prefix.strip():
-            # Strip GPT-OSS channel tokens so they don't get double-wrapped
-            # by the template (which adds its own channel markup).
-            clean = content_prefix.strip()
-            if '<|channel|>' in clean and '<|message|>' in clean:
-                inner = clean.split('<|message|>', 1)[1]
-                if '<|end|>' in inner:
-                    inner = inner.split('<|end|>', 1)[0]
-                clean = inner.strip()
+            clean = _strip_channel_tokens(content_prefix)
             if clean:
                 seq_entry['content'] = clean
         seq.append(seq_entry)
@@ -1572,6 +1582,7 @@ def _cancel_remaining(from_idx):
     # to the final model response only (not to tool call markers).
     if state.pop('_skip_output_extensions', None):
         _model_visible = apply_extensions('output', _model_visible, state, is_chat=True)
+
         if visible_prefix:
             history['visible'][-1][1] = '\n\n'.join(visible_prefix + [_model_visible])
         else:
diff --git a/modules/html_generator.py b/modules/html_generator.py
index e060604d40..e3bafda4ab 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -375,6 +375,7 @@ def process_text_segment(text, is_last_segment):
         while text.strip():
             thinking_content, remaining = extract_thinking_block(text)
             if thinking_content is None:
+                text = remaining  # strip standalone channel markers even without thinking
                 break
             has_remaining = bool(remaining.strip()) or not is_last_segment
             html_parts.append(build_thinking_block(thinking_content, message_id, has_remaining, think_idx))

From 8ac5162cbf865cf0cf1a491675998ce6bec2bd78 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 22 Apr 2026 18:52:18 -0700
Subject: [PATCH 1581/1701] Update README

---
 README.md | 33 +++++++++++++++++----------------
 1 file changed, 17 insertions(+), 16 deletions(-)

diff --git a/README.md b/README.md
index 12b26b6ef7..da13dab8f4 100644
--- a/README.md
+++ b/README.md
@@ -286,17 +286,17 @@ usage: server.py [-h] [--user-data-dir USER_DATA_DIR] [--multi-user] [--model MO
                  [--ctx-size-draft CTX_SIZE_DRAFT] [--spec-type {none,ngram-mod,ngram-simple,ngram-map-k,ngram-map-k4v,ngram-cache}] [--spec-ngram-size-n SPEC_NGRAM_SIZE_N]
                  [--spec-ngram-size-m SPEC_NGRAM_SIZE_M] [--spec-ngram-min-hits SPEC_NGRAM_MIN_HITS] [--gpu-layers N] [--cpu-moe] [--mmproj MMPROJ] [--streaming-llm] [--tensor-split TENSOR_SPLIT]
                  [--row-split] [--no-mmap] [--mlock] [--no-kv-offload] [--batch-size BATCH_SIZE] [--ubatch-size UBATCH_SIZE] [--threads THREADS] [--threads-batch THREADS_BATCH] [--numa]
-                 [--parallel PARALLEL] [--fit-target FIT_TARGET] [--extra-flags EXTRA_FLAGS] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] [--bf16]
-                 [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE]
-                 [--quant_type QUANT_TYPE] [--gpu-split GPU_SPLIT] [--enable-tp] [--tp-backend TP_BACKEND] [--cfg-cache] [--listen] [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share]
-                 [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors]
-                 [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4]
-                 [--nowebui] [--temperature N] [--dynatemp-low N] [--dynatemp-high N] [--dynatemp-exponent N] [--smoothing-factor N] [--smoothing-curve N] [--min-p N] [--top-p N] [--top-k N]
-                 [--typical-p N] [--xtc-threshold N] [--xtc-probability N] [--epsilon-cutoff N] [--eta-cutoff N] [--tfs N] [--top-a N] [--top-n-sigma N] [--adaptive-target N] [--adaptive-decay N]
-                 [--dry-multiplier N] [--dry-allowed-length N] [--dry-base N] [--repetition-penalty N] [--frequency-penalty N] [--presence-penalty N] [--encoder-repetition-penalty N]
-                 [--no-repeat-ngram-size N] [--repetition-penalty-range N] [--penalty-alpha N] [--guidance-scale N] [--mirostat-mode N] [--mirostat-tau N] [--mirostat-eta N]
-                 [--do-sample | --no-do-sample] [--dynamic-temperature | --no-dynamic-temperature] [--temperature-last | --no-temperature-last] [--sampler-priority N] [--dry-sequence-breakers N]
-                 [--enable-thinking | --no-enable-thinking] [--reasoning-effort N] [--chat-template-file CHAT_TEMPLATE_FILE]
+                 [--parallel PARALLEL] [--fit-target FIT_TARGET] [--extra-flags EXTRA_FLAGS] [--ik] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit]
+                 [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] [--load-in-4bit] [--use_double_quant]
+                 [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--gpu-split GPU_SPLIT] [--enable-tp] [--tp-backend TP_BACKEND] [--cfg-cache] [--listen] [--listen-port LISTEN_PORT]
+                 [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE]
+                 [--subpath SUBPATH] [--old-colors] [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY]
+                 [--api-enable-ipv6] [--api-disable-ipv4] [--nowebui] [--temperature N] [--dynatemp-low N] [--dynatemp-high N] [--dynatemp-exponent N] [--smoothing-factor N] [--smoothing-curve N]
+                 [--top-p N] [--top-k N] [--min-p N] [--top-n-sigma N] [--typical-p N] [--xtc-threshold N] [--xtc-probability N] [--epsilon-cutoff N] [--eta-cutoff N] [--tfs N] [--top-a N]
+                 [--adaptive-target N] [--adaptive-decay N] [--dry-multiplier N] [--dry-allowed-length N] [--dry-base N] [--repetition-penalty N] [--frequency-penalty N] [--presence-penalty N]
+                 [--encoder-repetition-penalty N] [--no-repeat-ngram-size N] [--repetition-penalty-range N] [--penalty-alpha N] [--guidance-scale N] [--mirostat-mode N] [--mirostat-tau N]
+                 [--mirostat-eta N] [--do-sample | --no-do-sample] [--dynamic-temperature | --no-dynamic-temperature] [--temperature-last | --no-temperature-last] [--sampler-priority N]
+                 [--dry-sequence-breakers N] [--enable-thinking | --no-enable-thinking] [--reasoning-effort N] [--preserve-thinking | --no-preserve-thinking] [--chat-template-file CHAT_TEMPLATE_FILE]
 
 TextGen
 
@@ -365,8 +365,8 @@ llama.cpp:
   --parallel PARALLEL                                  Number of parallel request slots. The context size is divided equally among slots. For example, to have 4 slots with 8192 context each, set
                                                        ctx_size to 32768.
   --fit-target FIT_TARGET                              Target VRAM margin per device for auto GPU layers, comma-separated list of values in MiB. A single value is broadcast across all devices.
-                                                       Default: 1024.
-  --extra-flags EXTRA_FLAGS                            Extra flags to pass to llama-server. Format: "flag1=value1,flag2,flag3=value3". Example: "override-tensor=exps=CPU"
+  --extra-flags EXTRA_FLAGS                            Extra flags to pass to llama-server. Example: "--jinja --rpc 192.168.1.100:50052"
+  --ik                                                 Use ik_llama.cpp instead of upstream llama.cpp. Requires the ik_llama_cpp_binaries package to be installed.
 
 Transformers/Accelerate:
   --cpu                                                Use the CPU to generate text. Warning: Training on CPU is extremely slow.
@@ -408,7 +408,7 @@ Gradio:
   --portable                                           Hide features not available in portable mode like training.
 
 API:
-  --api                                                Enable the API extension.
+  --api                                                Enable the API server.
   --public-api                                         Create a public URL for the API using Cloudflare.
   --public-api-id PUBLIC_API_ID                        Tunnel ID for named Cloudflare Tunnel. Use together with public-api option.
   --api-port API_PORT                                  The listening port for the API.
@@ -425,9 +425,10 @@ API generation defaults:
   --dynatemp-exponent N                                Dynamic temperature exponent
   --smoothing-factor N                                 Smoothing factor
   --smoothing-curve N                                  Smoothing curve
-  --min-p N                                            Min P
   --top-p N                                            Top P
   --top-k N                                            Top K
+  --min-p N                                            Min P
+  --top-n-sigma N                                      Top N Sigma
   --typical-p N                                        Typical P
   --xtc-threshold N                                    XTC threshold
   --xtc-probability N                                  XTC probability
@@ -435,7 +436,6 @@ API generation defaults:
   --eta-cutoff N                                       Eta cutoff
   --tfs N                                              TFS
   --top-a N                                            Top A
-  --top-n-sigma N                                      Top N Sigma
   --adaptive-target N                                  Adaptive target
   --adaptive-decay N                                   Adaptive decay
   --dry-multiplier N                                   DRY multiplier
@@ -459,6 +459,7 @@ API generation defaults:
   --dry-sequence-breakers N                            DRY sequence breakers
   --enable-thinking, --no-enable-thinking              Enable thinking
   --reasoning-effort N                                 Reasoning effort
+  --preserve-thinking, --no-preserve-thinking          Preserve thinking blocks from prior turns in the chat template
   --chat-template-file CHAT_TEMPLATE_FILE              Path to a chat template file (.jinja, .jinja2, or .yaml) to use as the default instruction template for API requests. Overrides the model's
                                                        built-in template.
 ```

From 5a9d61edc54ea7bef2c090ea0d4be14cb7cf9683 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 22 Apr 2026 18:54:41 -0700
Subject: [PATCH 1582/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 8 ++++----
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 8 ++++----
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 15 files changed, 30 insertions(+), 30 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index f3e8c3bc5c..6aba7e6656 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -42,10 +42,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/ik_llama_cpp_binaries-0.123.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/ik_llama_cpp_binaries-0.123.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.30/exllamav3-0.0.30+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.30/exllamav3-0.0.30+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 6e7ebb6467..de1a8351ed 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -38,5 +38,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index e2c57f585d..f15156db54 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -38,4 +38,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index f06d62a857..9598904e69 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -38,4 +38,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 019ff03527..461e0621d5 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -38,7 +38,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/ik_llama_cpp_binaries-0.123.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/ik_llama_cpp_binaries-0.123.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index da3e15aed9..96772c4a66 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 690bd4ce31..3b3395832e 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 99a95566ab..c2add72cbe 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -24,4 +24,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 2fc069443f..40209d07bc 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -24,4 +24,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index b79f52e603..9e76620f7b 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 3336327194..8a227f6e6c 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index 7ef6ea2e3b..a918c1b3d7 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/ik_llama_cpp_binaries-0.123.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/ik_llama_cpp_binaries-0.123.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index 2d6a6e7d4d..67def2007f 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # ik_llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/ik_llama_cpp_binaries-0.123.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/ik_llama_cpp_binaries-0.123.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index ee1c951b50..cffc3666c1 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/ik_llama_cpp_binaries-0.122.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/ik_llama_cpp_binaries-0.123.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/ik_llama_cpp_binaries-0.123.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 9bea07e695..6f95f44775 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -24,5 +24,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.122.0/llama_cpp_binaries-0.122.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.123.0/llama_cpp_binaries-0.123.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 1d1a319597553cbb78c24d1bbf39138b5a6cc927 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 23 Apr 2026 06:21:19 -0700
Subject: [PATCH 1583/1701] llama.cpp: Fix multimodal by using server's random
 media marker

---
 modules/llama_cpp_server.py | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index bb244612a9..7c4e2b90ea 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -38,6 +38,7 @@ def __init__(self, model_path, server_path=None):
         self.vocabulary_size = None
         self.n_ctx = None
         self.bos_token = ""
+        self.media_marker = "<__media__>"
         self.last_prompt_token_count = 0
 
         # Start the server
@@ -185,6 +186,9 @@ def generate_with_streaming(self, prompt, state):
             # Multimodal case
             IMAGE_TOKEN_COST_ESTIMATE = 600  # A safe, conservative estimate per image
 
+            # Translate placeholders to the server's actual media marker.
+            prompt = prompt.replace("<__media__>", self.media_marker)
+
             base64_images = [convert_pil_to_base64(img) for img in pil_images]
             payload["prompt"] = {
                 "prompt_string": prompt,
@@ -376,6 +380,12 @@ def _get_bos_token(self):
         if n_ctx:
             self.n_ctx = n_ctx
 
+        # Get the server's media marker for multimodal support.
+        # llama.cpp server generates a random marker each restart;
+        # we must use it instead of the default <__media__>.
+        if "media_marker" in response:
+            self.media_marker = response["media_marker"]
+
     def _is_port_available(self, port):
         """Check if a port is available for use."""
         with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:

From d440d837b5891db6eca36d38978a031fb361718a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 23 Apr 2026 06:25:39 -0700
Subject: [PATCH 1584/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 17 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 6aba7e6656..765c514151 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -33,8 +33,8 @@ wandb
 xformers==0.0.33.post2
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index de1a8351ed..dc81571a06 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -29,8 +29,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index f15156db54..399dcfd760 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -29,8 +29,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 9598904e69..62d9e8ed82 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -29,8 +29,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 461e0621d5..17ebd1ad19 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -29,8 +29,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 19ac51836b..6868bc6f0b 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -29,8 +29,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 96772c4a66..c08506d290 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -15,8 +15,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 3b3395832e..a376cb03f2 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -15,8 +15,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index c2add72cbe..b2c10081e9 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -15,8 +15,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 40209d07bc..ef0ccb5bbb 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -15,8 +15,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index 9e76620f7b..82d47f5a16 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -15,8 +15,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 8a227f6e6c..9a60d2ee9f 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -15,8 +15,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index a918c1b3d7..6dd65afe4c 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -15,8 +15,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index 67def2007f..de16866997 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -15,8 +15,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index cffc3666c1..1bd54e618b 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -15,8 +15,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index cafe3ceea2..ade2a6a8ce 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -15,8 +15,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 6f95f44775..11324de4b2 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -15,8 +15,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio-4.37.2+custom.19-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.19/gradio_client-1.0.2+custom.19-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From dc798aabe38ed54f544a15ba834ae9b39025633f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 23 Apr 2026 06:37:20 -0700
Subject: [PATCH 1585/1701] Pre-load MCP stdio server tools at startup

---
 server.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/server.py b/server.py
index 4fada434e9..96e6e1ab40 100644
--- a/server.py
+++ b/server.py
@@ -259,9 +259,10 @@ def create_interface():
     elif (shared.user_data_dir / 'settings.yaml').exists():
         settings_file = shared.user_data_dir / 'settings.yaml'
 
-    from modules.tool_use import has_mcp_config
+    from modules.tool_use import has_mcp_config, load_mcp_tools
     if has_mcp_config():
-        logger.warning(f"MCP stdio servers will be loaded from \"{shared.user_data_dir / 'mcp.json'}\"")
+        logger.warning(f"Loading MCP stdio servers from \"{shared.user_data_dir / 'mcp.json'}\"")
+        load_mcp_tools('')
 
     if settings_file is not None:
         logger.info(f"Loading settings from \"{settings_file}\"")

From d31662a4102cbb0cedaf572d83d900ec4c8d7885 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 23 Apr 2026 07:53:42 -0700
Subject: [PATCH 1586/1701] Revert "Pre-load MCP stdio server tools at startup"

This reverts commit dc798aabe38ed54f544a15ba834ae9b39025633f.
---
 server.py | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/server.py b/server.py
index 96e6e1ab40..4fada434e9 100644
--- a/server.py
+++ b/server.py
@@ -259,10 +259,9 @@ def create_interface():
     elif (shared.user_data_dir / 'settings.yaml').exists():
         settings_file = shared.user_data_dir / 'settings.yaml'
 
-    from modules.tool_use import has_mcp_config, load_mcp_tools
+    from modules.tool_use import has_mcp_config
     if has_mcp_config():
-        logger.warning(f"Loading MCP stdio servers from \"{shared.user_data_dir / 'mcp.json'}\"")
-        load_mcp_tools('')
+        logger.warning(f"MCP stdio servers will be loaded from \"{shared.user_data_dir / 'mcp.json'}\"")
 
     if settings_file is not None:
         logger.info(f"Loading settings from \"{settings_file}\"")

From c014efd59a01d5061e075e16613ed4fe1aa13315 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 23 Apr 2026 11:44:54 -0700
Subject: [PATCH 1587/1701] Update transformers to 5.6

---
 requirements/full/requirements.txt               | 2 +-
 requirements/full/requirements_amd.txt           | 2 +-
 requirements/full/requirements_apple_intel.txt   | 2 +-
 requirements/full/requirements_apple_silicon.txt | 2 +-
 requirements/full/requirements_cpu_only.txt      | 2 +-
 requirements/full/requirements_nowheels.txt      | 2 +-
 6 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 765c514151..00d547ce2f 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -26,7 +26,7 @@ sentencepiece
 tensorboard
 torchao==0.15.*
 trafilatura==2.0.0
-transformers==5.5.*
+transformers==5.6.*
 triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index dc81571a06..86800da3c6 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.5.*
+transformers==5.6.*
 tqdm
 trafilatura==2.0.0
 wandb
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 399dcfd760..22d8516095 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.5.*
+transformers==5.6.*
 tqdm
 trafilatura==2.0.0
 wandb
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 62d9e8ed82..98b080395b 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.5.*
+transformers==5.6.*
 tqdm
 trafilatura==2.0.0
 wandb
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 17ebd1ad19..6e6c6241d5 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.5.*
+transformers==5.6.*
 tqdm
 trafilatura==2.0.0
 wandb
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 6868bc6f0b..7616f20643 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -23,7 +23,7 @@ scipy
 sentencepiece
 tensorboard
 torchao==0.15.*
-transformers==5.5.*
+transformers==5.6.*
 tqdm
 trafilatura==2.0.0
 wandb

From 7847d7284f75a8478ac4596ea33ece41e15ddb63 Mon Sep 17 00:00:00 2001
From: oobabooga 
Date: Thu, 23 Apr 2026 16:24:33 -0300
Subject: [PATCH 1588/1701] Update the screenshots

---
 README.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index da13dab8f4..1892cb1944 100644
--- a/README.md
+++ b/README.md
@@ -27,10 +27,10 @@ Portable builds for Linux, Windows, and macOS with CUDA, Vulkan, ROCm, and CPU-o
 
 For additional backends (ExLlamaV3, Transformers), training, image generation, and extensions, see [Installation](#installation).
 
-|![Instruct mode](https://github.com/oobabooga/screenshots/raw/main/INSTRUCT-3.5.png) | ![Chat mode](https://github.com/oobabooga/screenshots/raw/main/CHAT-3.5.png) |
+|![Instruct mode](https://github.com/oobabooga/screenshots/raw/main/INSTRUCT-4-6.png) | ![Chat mode](https://github.com/oobabooga/screenshots/raw/main/CHAT-4-6.png) |
 |:---:|:---:|
 | Instruct mode | Chat mode |
-|![Notebook mode](https://github.com/oobabooga/screenshots/raw/main/DEFAULT-3.5.png) | ![Parameters](https://github.com/oobabooga/screenshots/raw/main/PARAMETERS-3.5.png) |
+|![Notebook mode](https://github.com/oobabooga/screenshots/raw/main/NOTEBOOK-4-6.png) | ![Parameters](https://github.com/oobabooga/screenshots/raw/main/PARAMETERS-4-6.png) |
 | Notebook mode | Parameters |
 
 ## Features

From 187b07a9249452005b99d2e3673a9ee82cf0c923 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 25 Apr 2026 15:14:44 -0700
Subject: [PATCH 1589/1701] Fix race condition and clean up exllamav3 module

---
 modules/exllamav3.py | 144 +++++++++++++++----------------------------
 1 file changed, 51 insertions(+), 93 deletions(-)

diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index 815ca3dd9b..a05a56ca13 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -86,15 +86,15 @@ def _iterate_loop(self):
                     self.generator.clear_queue()
                     self.has_jobs.clear()
                     continue
-            for result in results:
-                job = result["job"]
-                q = self.job_queues.get(job)
-                if q:
-                    q.put(result)
-                    if result.get("eos"):
-                        self.job_queues.pop(job, None)
-            if not self.job_queues:
-                self.has_jobs.clear()
+                for result in results:
+                    job = result["job"]
+                    q = self.job_queues.get(job)
+                    if q:
+                        q.put(result)
+                        if result.get("eos"):
+                            self.job_queues.pop(job, None)
+                if not self.job_queues:
+                    self.has_jobs.clear()
 
     def submit(self, job) -> queue.Queue:
         q = queue.Queue()
@@ -118,8 +118,6 @@ def stop(self):
 
 
 class Exllamav3Model:
-    def __init__(self):
-        pass
 
     @property
     def device(self) -> torch.device:
@@ -139,14 +137,13 @@ def from_pretrained(cls, path_to_model):
         config = Config.from_directory(str(path_to_model))
         model = Model.from_config(config)
 
-        # Calculate the closest multiple of 256 at or above the chosen value
+        # Adjust to the closest multiple of 256 at or above the chosen value
         max_tokens = shared.args.ctx_size
         if max_tokens % 256 != 0:
             adjusted_tokens = ((max_tokens // 256) + 1) * 256
             logger.warning(f"max_num_tokens must be a multiple of 256. Adjusting from {max_tokens} to {adjusted_tokens}")
             max_tokens = adjusted_tokens
 
-        # Parse cache type
         cache_type = shared.args.cache_type.lower()
         cache_kwargs = {}
         if cache_type == 'fp16':
@@ -277,11 +274,8 @@ def _process_images_for_generation(self, prompt: str, state: dict) -> Tuple[str,
         if not pil_images:
             return prompt, []
 
-        # ExLlamaV3-specific: Generate embeddings
         try:
-            # Use pre-computed embeddings if available (proper MMEmbedding lifetime)
             if 'image_embeddings' in state and state['image_embeddings']:
-                # Use existing embeddings - this preserves MMEmbedding lifetime
                 image_embeddings = state['image_embeddings']
             else:
                 # Do not reset the cache/allocator index; it causes token ID conflicts during generation.
@@ -291,7 +285,6 @@ def _process_images_for_generation(self, prompt: str, state: dict) -> Tuple[str,
                     for img in pil_images
                 ]
 
-            # ExLlamaV3-specific: Handle prompt processing with placeholders
             placeholders = [ie.text_alias for ie in image_embeddings]
 
             if '<__media__>' in prompt:
@@ -326,22 +319,19 @@ def generate_with_streaming(self, prompt, state):
         if state['temperature'] == 0:
             sampler = CustomSampler([SS_Argmax()])
         else:
-            # 1. Create a list of all active, unordered samplers
+            # 1. Collect active samplers (unordered)
             unordered_samplers = []
 
-            # Penalties
             penalty_range = state['repetition_penalty_range']
             if penalty_range <= 0:
                 penalty_range = int(10e7)  # Use large number for "full context"
             rep_decay = 0  # Not a configurable parameter
 
-            # Add penalty samplers if they are active
             if state['repetition_penalty'] != 1.0:
                 unordered_samplers.append(SS_RepP(state['repetition_penalty'], penalty_range, rep_decay))
             if state['presence_penalty'] != 0.0 or state['frequency_penalty'] != 0.0:
                 unordered_samplers.append(SS_PresFreqP(state['presence_penalty'], state['frequency_penalty'], penalty_range, rep_decay))
 
-            # Standard samplers
             if state['top_k'] > 0:
                 unordered_samplers.append(SS_TopK(state['top_k']))
             if state['top_p'] < 1.0:
@@ -349,10 +339,9 @@ def generate_with_streaming(self, prompt, state):
             if state['min_p'] > 0.0:
                 unordered_samplers.append(SS_MinP(state['min_p']))
 
-            # Temperature (SS_NoOp is returned if temp is 1.0)
             unordered_samplers.append(SS_Temperature(state['temperature']))
 
-            # 2. Define the mapping from class names to the priority list keys
+            # 2. Sort samplers by priority
             class_name_to_nickname = {
                 'SS_RepP': 'repetition_penalty',
                 'SS_PresFreqP': 'presence_frequency_penalty',
@@ -362,7 +351,6 @@ def generate_with_streaming(self, prompt, state):
                 'SS_Temperature': 'temperature',
             }
 
-            # 3. Get the priority list and handle temperature_last
             default_priority = ['repetition_penalty', 'presence_frequency_penalty', 'top_k', 'top_p', 'min_p', 'temperature']
             sampler_priority = list(state.get('sampler_priority') or default_priority)
 
@@ -374,7 +362,6 @@ def generate_with_streaming(self, prompt, state):
             # SS_PresFreqP sampler. Normalize to the combined name.
             sampler_priority = ['presence_frequency_penalty' if x in ('presence_penalty', 'frequency_penalty') else x for x in sampler_priority]
 
-            # 4. Sort the unordered list based on the priority list
             def custom_sort_key(sampler_obj):
                 class_name = sampler_obj.__class__.__name__
                 nickname = class_name_to_nickname.get(class_name)
@@ -384,7 +371,7 @@ def custom_sort_key(sampler_obj):
 
             ordered_samplers = sorted(unordered_samplers, key=custom_sort_key)
 
-            # 5. Add the final sampling stage and build the sampler
+            # 3. Add final sampling stage and build the sampler
             if state.get('adaptive_target', 0) > 0:
                 ordered_samplers.append(SS_AdaptiveP(state['adaptive_target'], state['adaptive_decay']))
             else:
@@ -392,7 +379,6 @@ def custom_sort_key(sampler_obj):
 
             sampler = CustomSampler(ordered_samplers)
 
-        # Encode prompt with embeddings (ExLlamaV3-specific)
         input_ids = self.tokenizer.encode(
             prompt,
             add_bos=state['add_bos_token'],
@@ -404,37 +390,26 @@ def custom_sort_key(sampler_obj):
 
         self._last_prompt_token_count = input_ids.shape[-1]
 
-        # Determine max_new_tokens
         if state['auto_max_new_tokens']:
             max_new_tokens = state['truncation_length'] - self._last_prompt_token_count
         else:
             max_new_tokens = state['max_new_tokens']
 
-        # Use full EOS token list from config (may contain multiple IDs)
-        stop_conditions = []
-        if not state['ban_eos_token']:
-            for eos_id in self.config.eos_token_id_list:
-                if eos_id is not None:
-                    stop_conditions.append(eos_id)
+        eos_ids = [eid for eid in self.config.eos_token_id_list if eid is not None]
+
+        stop_conditions = [] if state['ban_eos_token'] else list(eos_ids)
 
-        # Build filters for logit_bias (OpenAI API)
         filters = []
         logit_bias = state.get('logit_bias')
         if logit_bias:
             filters.append(LogitBiasFilter(self.tokenizer, logit_bias))
 
         # Suppress EOS tokens via logit bias so they are never sampled
-        if state['ban_eos_token']:
-            eos_bias = {}
-            for eos_id in self.config.eos_token_id_list:
-                if eos_id is not None:
-                    eos_bias[str(eos_id)] = float('-inf')
-            if eos_bias:
-                filters.append(LogitBiasFilter(self.tokenizer, eos_bias))
-
-        # Logprobs support (OpenAI API)
-        logprobs = state.get('logprobs', 0) or 0
-        return_top_tokens = logprobs if logprobs > 0 else 0
+        if state['ban_eos_token'] and eos_ids:
+            eos_bias = {str(eid): float('-inf') for eid in eos_ids}
+            filters.append(LogitBiasFilter(self.tokenizer, eos_bias))
+
+        return_top_tokens = max(state.get('logprobs') or 0, 0)
 
         seed = state.get('seed', -1)
         job = Job(
@@ -450,7 +425,6 @@ def custom_sort_key(sampler_obj):
             return_probs=return_top_tokens > 0,
         )
 
-        # Stream generation
         response_text = ""
         stop_event = state.get('stop_event')
         self.last_completion_probabilities = []
@@ -465,13 +439,10 @@ def custom_sort_key(sampler_obj):
                 except queue.Empty:
                     continue
                 if result is None or result.get("eos"):
-                    # Capture logprobs from the final eos result too
                     if result is not None and return_top_tokens > 0:
                         self._capture_logprobs(result)
                     break
                 chunk = result.get("text", "")
-
-                # Capture logprobs from streaming results
                 if return_top_tokens > 0:
                     self._capture_logprobs(result)
 
@@ -488,9 +459,18 @@ def _capture_logprobs(self, result):
         if top_k_tokens is None or top_k_probs is None:
             return
 
-        id_to_piece = self.tokenizer.get_id_to_piece_list(True)
-        sampled_ids = result.get("token_ids")    # (batch, seq_len) - actually sampled tokens
-        sampled_probs = result.get("token_probs")  # (batch, seq_len) - their probabilities
+        if not hasattr(self, '_id_to_piece'):
+            self._id_to_piece = self.tokenizer.get_id_to_piece_list(True)
+
+        id_to_piece = self._id_to_piece
+
+        # Bulk-convert tensors to Python lists to avoid per-element .item() calls
+        tk_tokens = top_k_tokens[0].tolist()   # (seq_len, k)
+        tk_probs = top_k_probs[0].tolist()     # (seq_len, k)
+        sampled_ids = result.get("token_ids")
+        sampled_probs = result.get("token_probs")
+        s_ids = sampled_ids[0].tolist() if sampled_ids is not None else None
+        s_probs = sampled_probs[0].tolist() if sampled_probs is not None else None
 
         def _piece(tid):
             s = id_to_piece[tid] if tid < len(id_to_piece) else f"<{tid}>"
@@ -499,24 +479,19 @@ def _piece(tid):
         def _logprob(prob):
             return math.log(prob) if prob > 0 else float("-inf")
 
-        # top_k_tokens shape: (batch, seq_len, k), top_k_probs same
-        for seq_idx in range(top_k_tokens.shape[1]):
+        for seq_idx in range(len(tk_tokens)):
             entry = {"top_logprobs": []}
-            for k_idx in range(top_k_tokens.shape[2]):
-                token_id = top_k_tokens[0, seq_idx, k_idx].item()
-                prob = top_k_probs[0, seq_idx, k_idx].item()
+            for k_idx in range(len(tk_tokens[seq_idx])):
+                token_id = tk_tokens[seq_idx][k_idx]
+                prob = tk_probs[seq_idx][k_idx]
                 entry["top_logprobs"].append({"token": _piece(token_id), "logprob": _logprob(prob)})
 
             # Record the actually sampled token at the entry level so
             # format_completion_logprobs uses it instead of top_logprobs[0]
             # (they differ with non-greedy sampling).
-            if sampled_ids is not None:
-                sid = sampled_ids[0, seq_idx].item()
-                entry["token"] = _piece(sid)
-                if sampled_probs is not None:
-                    entry["logprob"] = _logprob(sampled_probs[0, seq_idx].item())
-                else:
-                    entry["logprob"] = None
+            if s_ids is not None:
+                entry["token"] = _piece(s_ids[seq_idx])
+                entry["logprob"] = _logprob(s_probs[seq_idx]) if s_probs is not None else None
 
             self.last_completion_probabilities.append(entry)
 
@@ -532,7 +507,6 @@ def get_prompt_logits(self, input_ids):
 
         Used by prompt logprobs computation. Returns (1, seq_len, vocab) on CPU in float32.
         """
-        import torch
         input_ids_tensor = input_ids if isinstance(input_ids, torch.Tensor) else torch.tensor(input_ids, dtype=torch.long)
         input_ids_tensor = input_ids_tensor.view(1, -1).cpu()
         with torch.inference_mode():
@@ -575,45 +549,29 @@ def last_prompt_token_count(self):
     def unload(self):
         logger.info("Unloading ExLlamaV3 model components...")
 
-        if hasattr(self, 'parallel_generator') and self.parallel_generator is not None:
+        if self.parallel_generator is not None:
             try:
                 self.parallel_generator.stop()
             except Exception as e:
                 logger.warning(f"Error stopping parallel generator: {e}")
-            self.parallel_generator = None
 
-        if hasattr(self, 'vision_model') and self.vision_model is not None:
-            try:
-                del self.vision_model
-            except Exception as e:
-                logger.warning(f"Error unloading vision model: {e}")
-            self.vision_model = None
-
-        if hasattr(self, 'draft_model') and self.draft_model is not None:
+        if self.draft_model is not None:
             try:
                 self.draft_model.unload()
-                del self.draft_model
             except Exception as e:
                 logger.warning(f"Error unloading draft model: {e}")
-            self.draft_model = None
-
-        if hasattr(self, 'draft_cache') and self.draft_cache is not None:
-            self.draft_cache = None
 
-        if hasattr(self, 'model') and self.model is not None:
+        if self.model is not None:
             try:
                 self.model.unload()
-                del self.model
             except Exception as e:
                 logger.warning(f"Error unloading main model: {e}")
 
-            self.model = None
-
-        if hasattr(self, 'cache') and self.cache is not None:
-            self.cache = None
-
-        if hasattr(self, 'generator') and self.generator is not None:
-            self.generator = None
-
-        if hasattr(self, 'tokenizer') and self.tokenizer is not None:
-            self.tokenizer = None
+        self.parallel_generator = None
+        self.vision_model = None
+        self.draft_model = None
+        self.draft_cache = None
+        self.model = None
+        self.cache = None
+        self.generator = None
+        self.tokenizer = None

From 4b575a2faece27b44da07c43de6b2e545366bb53 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 26 Apr 2026 17:44:32 -0700
Subject: [PATCH 1590/1701] llama.cpp: Add --split-mode flag with tensor
 parallelism support

---
 docs/04 - Model Tab.md      | 2 +-
 modules/llama_cpp_server.py | 7 ++++---
 modules/loaders.py          | 4 ++--
 modules/shared.py           | 2 +-
 modules/ui_model_menu.py    | 4 ++--
 5 files changed, 10 insertions(+), 9 deletions(-)

diff --git a/docs/04 - Model Tab.md b/docs/04 - Model Tab.md
index ba415a93d4..6bcedceb7d 100644
--- a/docs/04 - Model Tab.md	
+++ b/docs/04 - Model Tab.md	
@@ -21,7 +21,7 @@ Example: https://huggingface.co/TheBloke/Llama-2-7b-Chat-GGUF
 * **mmproj**: Path to the mmproj file for multimodal (vision) models. This enables image understanding capabilities.
 * **streaming_llm**: Experimental feature to avoid re-evaluating the entire prompt when part of it is removed, for instance, when you hit the context length for the model in chat mode and an old message is removed.
 * **cpu**: Force a version of llama.cpp compiled without GPU acceleration to be used. Can usually be ignored. Only set this if you want to use CPU only and llama.cpp doesn't work otherwise.
-* **row_split**: Split the model by rows across GPUs. This may improve multi-gpu performance.
+* **split_mode**: How to split the model across multiple GPUs. "tensor" can make multi-GPU significantly faster.
 * **no_kv_offload**: Do not offload the KV cache to the GPU. This saves VRAM but reduces performance.
 * **no_mmap**: Loads the model into memory at once, possibly preventing I/O operations later on at the cost of a longer load time.
 * **mlock**: Force the system to keep the model in RAM rather than swapping or compressing.
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 7c4e2b90ea..0402207563 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -462,8 +462,8 @@ def _start_server(self):
             cmd += ["--numa", "distribute"]
         if shared.args.no_kv_offload:
             cmd.append("--no-kv-offload")
-        if shared.args.row_split:
-            cmd += ["--split-mode", "row"]
+        if shared.args.split_mode != "layer":
+            cmd += ["--split-mode", shared.args.split_mode]
         cache_type = "fp16"
         if shared.args.cache_type != "fp16" and shared.args.cache_type in llamacpp_valid_cache_types:
             cmd += ["--cache-type-k", shared.args.cache_type, "--cache-type-v", shared.args.cache_type]
@@ -687,6 +687,7 @@ def _patch_cmd_for_ik(cmd):
       --cache-reuse        → (removed, unsupported)
       --swa-full           → (removed, unsupported)
       --split-mode row     → --split-mode graph
+      --split-mode tensor  → --split-mode graph
     """
     # Add Hadamard KV cache rotation when using quantized cache types.
     # This significantly improves quantized cache quality (especially q4_0)
@@ -714,7 +715,7 @@ def _patch_cmd_for_ik(cmd):
             patched.append("--fit-margin")
         elif arg == "--cache-reuse":
             i += 1  # skip the value
-        elif arg == "--split-mode" and i + 1 < len(cmd) and cmd[i + 1] == "row":
+        elif arg == "--split-mode" and i + 1 < len(cmd) and cmd[i + 1] in ("row", "tensor"):
             patched += ["--split-mode", "graph"]
             i += 1  # skip the value
         elif arg == "--swa-full":
diff --git a/modules/loaders.py b/modules/loaders.py
index cc84377e60..9809fabaea 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -13,9 +13,9 @@
         'ctx_size',
         'cache_type',
         'tensor_split',
+        'split_mode',
         'extra_flags',
         'streaming_llm',
-        'row_split',
         'no_kv_offload',
         'no_mmap',
         'mlock',
@@ -321,7 +321,7 @@ def list_model_elements():
         'attn_implementation',
         'cpu',
         'disk',
-        'row_split',
+        'split_mode',
         'no_kv_offload',
         'no_mmap',
         'mlock',
diff --git a/modules/shared.py b/modules/shared.py
index 9fcf9c1f27..5f217d0b13 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -98,7 +98,7 @@
 group.add_argument('--mmproj', type=str, default=None, help='Path to the mmproj file for vision models.')
 group.add_argument('--streaming-llm', action='store_true', help='Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed.')
 group.add_argument('--tensor-split', type=str, default=None, help='Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40.')
-group.add_argument('--row-split', action='store_true', help='Split the model by rows across GPUs. This may improve multi-gpu performance.')
+group.add_argument('--split-mode', type=str, default='layer', choices=['layer', 'row', 'tensor', 'none'], help='How to split the model across multiple GPUs. "tensor" can make multi-GPU significantly faster.')
 group.add_argument('--no-mmap', action='store_true', help='Prevent mmap from being used.')
 group.add_argument('--mlock', action='store_true', help='Force the system to keep the model in RAM.')
 group.add_argument('--no-kv-offload', action='store_true', help='Do not offload the K, Q, V to the GPU. This saves VRAM but reduces performance.')
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 1dfcd5d60d..27000eab63 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -46,7 +46,7 @@ def create_ui():
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
-                            shared.gradio['fit_target'] = gr.Textbox(label='fit-target', value=shared.args.fit_target, info='Target VRAM margin per device for auto GPU layers (MiB). Comma-separated list for multiple devices.')
+                            shared.gradio['split_mode'] = gr.Dropdown(label='split-mode', choices=['layer', 'row', 'tensor', 'none'], value=shared.args.split_mode, info='How to split the model across multiple GPUs. "tensor" can make multi-GPU significantly faster.')
                             shared.gradio['tp_backend'] = gr.Dropdown(label="tp-backend", choices=['native', 'nccl'], value=shared.args.tp_backend, info='The backend for tensor parallelism.')
 
                         with gr.Column():
@@ -99,6 +99,7 @@ def create_ui():
                                 shared.gradio['threads_batch'] = gr.Slider(label="threads_batch", minimum=0, step=1, maximum=256, value=shared.args.threads_batch)
                                 shared.gradio['batch_size'] = gr.Slider(label="batch_size", minimum=1, maximum=4096, step=1, value=shared.args.batch_size)
                                 shared.gradio['ubatch_size'] = gr.Slider(label="ubatch_size", minimum=1, maximum=4096, step=1, value=shared.args.ubatch_size)
+                                shared.gradio['fit_target'] = gr.Textbox(label='fit-target', value=shared.args.fit_target, info='Target VRAM margin per device for auto GPU layers (MiB). Comma-separated list for multiple devices.')
                                 shared.gradio['tensor_split'] = gr.Textbox(label='tensor_split', info='List of proportions to split the model across multiple GPUs. Example: 60,40')
                                 shared.gradio['extra_flags'] = gr.Textbox(label='extra-flags', info='Extra flags to pass to llama-server. Example: --jinja --rpc 192.168.1.100:50052', value=shared.args.extra_flags)
                                 shared.gradio['cpu_memory'] = gr.Number(label="Maximum CPU memory in GiB. Use this for CPU offloading.", value=shared.args.cpu_memory)
@@ -109,7 +110,6 @@ def create_ui():
                                 shared.gradio['cpu'] = gr.Checkbox(label="cpu", value=shared.args.cpu, info='Use PyTorch in CPU mode.')
                                 shared.gradio['disk'] = gr.Checkbox(label="disk", value=shared.args.disk)
                                 shared.gradio['cpu_moe'] = gr.Checkbox(label="cpu-moe", value=shared.args.cpu_moe, info='Move the experts to the CPU. Saves VRAM on MoE models.')
-                                shared.gradio['row_split'] = gr.Checkbox(label="row_split", value=shared.args.row_split, info='Split the model by rows across GPUs. This may improve multi-gpu performance.')
                                 shared.gradio['no_kv_offload'] = gr.Checkbox(label="no_kv_offload", value=shared.args.no_kv_offload, info='Do not offload the K, Q, V to the GPU. This saves VRAM but reduces performance.')
                                 shared.gradio['no_mmap'] = gr.Checkbox(label="no-mmap", value=shared.args.no_mmap)
                                 shared.gradio['mlock'] = gr.Checkbox(label="mlock", value=shared.args.mlock)

From 54c4feeb4ffb546be379fa3999f0c912df8c67da Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 26 Apr 2026 18:53:02 -0700
Subject: [PATCH 1591/1701] Support .jinja/.jinja2 instruction template files
 in the UI (closes #7517)

---
 modules/api/completions.py |  7 +++---
 modules/chat.py            | 12 +++++++--
 modules/models_settings.py | 50 +++++++++++++++++++++++++-------------
 modules/training.py        | 26 +++-----------------
 modules/utils.py           | 11 ++++++---
 5 files changed, 58 insertions(+), 48 deletions(-)

diff --git a/modules/api/completions.py b/modules/api/completions.py
index 007f80a1c2..4e67a290ed 100644
--- a/modules/api/completions.py
+++ b/modules/api/completions.py
@@ -13,7 +13,7 @@
 from .typing import ToolDefinition
 from .utils import debug_msg
 from modules.tool_parsing import get_tool_call_id, parse_tool_call, detect_tool_call_format
-from modules import shared
+from modules import shared, utils
 from modules.reasoning import extract_reasoning
 from modules.chat import (
     generate_chat_prompt,
@@ -31,10 +31,9 @@
 def load_chat_template_file(filepath):
     """Load a chat template from a file path (.jinja, .jinja2, or .yaml/.yml)."""
     filepath = Path(filepath)
-    ext = filepath.suffix.lower()
     text = filepath.read_text(encoding='utf-8')
-    if ext in ['.yaml', '.yml']:
-        data = yaml.safe_load(text)
+    if filepath.suffix.lower() in utils.YAML_EXTENSIONS:
+        data = yaml.safe_load(text) or {}
         return data.get('instruction_template', '')
     return text
 
diff --git a/modules/chat.py b/modules/chat.py
index 1e01988688..a33b3321e6 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -2767,9 +2767,17 @@ def handle_save_template_click(instruction_template_str):
 
 def handle_delete_template_click(template):
     import gradio as gr
-    root = str(shared.user_data_dir / 'instruction-templates') + '/'
+    from modules.utils import TEMPLATE_EXTENSIONS
+    template_dir = shared.user_data_dir / 'instruction-templates'
+    filename = f"{template}.yaml"
+    for ext in TEMPLATE_EXTENSIONS:
+        if (template_dir / f"{template}{ext}").exists():
+            filename = f"{template}{ext}"
+            break
+
+    root = str(template_dir) + '/'
     return [
-        f"{template}.yaml",
+        filename,
         root,
         root,
         gr.update(visible=True)
diff --git a/modules/models_settings.py b/modules/models_settings.py
index 2193c1810d..53c09d07df 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -6,7 +6,7 @@
 
 import yaml
 
-from modules import loaders, metadata_gguf, shared
+from modules import loaders, metadata_gguf, shared, utils
 from modules.logging_colors import logger
 from modules.utils import resolve_model_path
 
@@ -397,28 +397,44 @@ def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type):
     return f"
            Estimated VRAM to load the model: {vram_usage:.0f} MiB
            "
 
 
-def load_instruction_template(template):
-    if template == 'None':
+def load_template_by_name(name):
+    """Find and load a single instruction template by name. Returns '' if not found."""
+    template_dir = shared.user_data_dir / 'instruction-templates'
+    for ext in utils.TEMPLATE_EXTENSIONS:
+        path = template_dir / f'{name}{ext}'
+        if path.is_file():
+            break
+    else:
         return ''
 
-    for name in (template, 'Alpaca'):
-        path = shared.user_data_dir / 'instruction-templates' / f'{name}.yaml'
-        try:
-            with open(path, 'r', encoding='utf-8') as f:
-                file_contents = f.read()
-        except FileNotFoundError:
-            if name == template:
-                logger.warning(f"Instruction template '{template}' not found, falling back to Alpaca")
-            continue
+    file_contents = path.read_text(encoding='utf-8')
+    if path.suffix in utils.JINJA_EXTENSIONS:
+        return file_contents
+
+    try:
+        data = yaml.safe_load(file_contents) or {}
+    except yaml.YAMLError:
+        logger.warning(f"Failed to parse '{path.name}' as YAML. Treating it as a raw Jinja template. Consider renaming it to '{name}.jinja'.")
+        return file_contents
 
-        break
-    else:
-        return ''
-    data = yaml.safe_load(file_contents)
     if 'instruction_template' in data:
         return data['instruction_template']
-    else:
+    elif 'turn_template' in data:
         return _jinja_template_from_old_format(data)
+    else:
+        return ''
+
+
+def load_instruction_template(template):
+    if template == 'None':
+        return ''
+
+    result = load_template_by_name(template)
+    if result:
+        return result
+
+    logger.warning(f"Instruction template '{template}' not found, falling back to Alpaca")
+    return load_template_by_name('Alpaca')
 
 
 def _jinja_template_from_old_format(params, verbose=False):
diff --git a/modules/training.py b/modules/training.py
index 3782c9fcf5..8d5619bb7c 100644
--- a/modules/training.py
+++ b/modules/training.py
@@ -14,7 +14,6 @@
 from datetime import datetime
 from pathlib import Path
 
-import yaml
 import gradio as gr
 
 from modules import shared, ui, utils
@@ -217,26 +216,8 @@ def clean_path(base_path: str, path: str):
 
 
 def get_instruction_templates():
-    path = shared.user_data_dir / 'instruction-templates'
-    names = set()
-    for ext in ['yaml', 'yml', 'jinja', 'jinja2']:
-        for f in path.glob(f'*.{ext}'):
-            names.add(f.stem)
-    return ['None', 'Chat Template'] + sorted(names, key=utils.natural_keys)
-
-
-def load_template(name):
-    """Load a Jinja2 template string from {user_data_dir}/instruction-templates/."""
-    path = shared.user_data_dir / 'instruction-templates'
-    for ext in ['jinja', 'jinja2', 'yaml', 'yml']:
-        filepath = path / f'{name}.{ext}'
-        if filepath.exists():
-            if ext in ['jinja', 'jinja2']:
-                return filepath.read_text(encoding='utf-8')
-            else:
-                data = yaml.safe_load(filepath.read_text(encoding='utf-8'))
-                return data.get('instruction_template', '')
-    return ''
+    templates = utils.get_available_instruction_templates()  # ['None', ...]
+    return [templates[0], 'Chat Template'] + templates[1:]
 
 
 def backup_adapter(input_folder):
@@ -485,7 +466,8 @@ def tokenize_text_data(data):
                 return
         else:
             # Load custom instruction template and set on tokenizer
-            template_str = load_template(format)
+            from modules.models_settings import load_template_by_name
+            template_str = load_template_by_name(format)
             if not template_str:
                 yield f"Error: could not load instruction template '{format}'."
                 return
diff --git a/modules/utils.py b/modules/utils.py
index c3cccf9be7..7bbd903791 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -247,11 +247,16 @@ def get_available_users():
     return sorted(set((k.stem for k in paths)), key=natural_keys)
 
 
+YAML_EXTENSIONS = ('.yaml', '.yml')
+JINJA_EXTENSIONS = ('.jinja', '.jinja2')
+TEMPLATE_EXTENSIONS = JINJA_EXTENSIONS + YAML_EXTENSIONS
+
+
 def get_available_instruction_templates():
-    path = str(shared.user_data_dir / "instruction-templates")
+    path = shared.user_data_dir / "instruction-templates"
     paths = []
-    if os.path.exists(path):
-        paths = (x for x in Path(path).iterdir() if x.suffix in ('.json', '.yaml', '.yml'))
+    if path.exists():
+        paths = (x for x in path.iterdir() if x.suffix in TEMPLATE_EXTENSIONS)
 
     return ['None'] + sorted(set((k.stem for k in paths)), key=natural_keys)
 

From c6a57a45727b197090d11ec4c6c28e4ce581c22d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 1 May 2026 22:03:15 -0700
Subject: [PATCH 1592/1701] Major UI overhaul

---
 css/main.css                   | 706 +++++++++++++++++++++++++++++++--
 modules/ui.py                  |  15 +-
 modules/ui_chat.py             |  14 +-
 modules/ui_default.py          |   2 +-
 modules/ui_image_generation.py |   4 +-
 modules/ui_notebook.py         |   2 +-
 modules/ui_parameters.py       |   8 +-
 7 files changed, 686 insertions(+), 65 deletions(-)

diff --git a/css/main.css b/css/main.css
index 7e5e4f29a9..e0e98916a8 100644
--- a/css/main.css
+++ b/css/main.css
@@ -1,11 +1,58 @@
 :root {
+    /* Legacy hue-named tokens (still referenced elsewhere). */
     --darker-gray: #1C1C1D;
     --dark-gray: #212125;
     --light-gray: #2C2E34;
     --light-theme-gray: #f0f3fb;
-    --border-color-dark: rgba(255, 255, 255, 0.15);
+    --border-color-dark: rgba(255, 255, 255, 0.10);
     --header-width: 112px;
     --selected-item-color-dark: #282930;
+
+    /* Role-based design tokens (light mode defaults).
+       Surfaces use a slight blue-leaning gray. The rail is one
+       step darker than the sidebar so the two read as distinct
+       elevations. */
+    --bg-body: #ffffff;
+    --bg-rail: #dbe1ec;
+    --bg-sidebar: #e8ecf3;
+    --bg-input: #f1f4f9;
+    --bg-elevated: #ffffff;
+    --bg-hover: rgba(60, 90, 160, 0.08);
+    --bg-active: rgba(60, 90, 160, 0.16);
+    --text: #1a1a1a;
+    --text-muted: #6b6b73;
+    --border: rgba(0, 0, 0, 0.10);
+    --border-soft: rgba(0, 0, 0, 0.06);
+    --accent: #4a72ff;
+
+    --radius-sm: 6px;
+    --radius-md: 8px;
+    --radius-lg: 12px;
+    --radius-xl: 16px;
+
+    --space-1: 4px;
+    --space-2: 8px;
+    --space-3: 12px;
+    --space-4: 16px;
+    --space-5: 24px;
+
+    --motion: 0.15s ease;
+    --motion-slow: 0.2s ease;
+}
+
+.dark {
+    --bg-body: #212125;
+    --bg-rail: #161617;
+    --bg-sidebar: #1a1a1c;
+    --bg-input: #2c2e34;
+    --bg-elevated: #1d1d1f;
+    --bg-hover: #2a2a2c;
+    --bg-active: #303033;
+    --text: #ececec;
+    --text-muted: #a3a3a8;
+    --border: rgba(255, 255, 255, 0.10);
+    --border-soft: rgba(255, 255, 255, 0.06);
+    --accent: #7aa2ff;
 }
 
 @font-face {
@@ -22,6 +69,13 @@
     font-style: italic;
 }
 
+body,
+.gradio-container,
+gradio-app {
+    font-family: 'Inter', 'Noto Sans', ui-sans-serif, system-ui, -apple-system,
+                 'Segoe UI', Roboto, Helvetica, Arial, sans-serif;
+}
+
 /* Hide spin buttons on number inputs (look bad on Windows) */
 input[type="number"]::-webkit-outer-spin-button,
 input[type="number"]::-webkit-inner-spin-button {
@@ -153,20 +207,23 @@ gradio-app > :first-child {
 .header_bar button {
     margin: 0;
     padding: 0.75rem;
+    border-radius: 0 !important;
+    transition: background-color var(--motion), color var(--motion);
+}
+
+.header_bar button:hover {
+    background-color: var(--bg-hover);
 }
 
 .header_bar button.selected {
     border: 0;
+    border-radius: 0;
 }
 
 .dark .header_bar {
     border: none !important;
     box-shadow: none;
-    background-color: #8080802b;
-}
-
-.header_bar button.selected {
-    border-radius: 0;
+    background-color: var(--bg-rail);
 }
 
 .textbox_default textarea {
@@ -174,19 +231,19 @@ gradio-app > :first-child {
 }
 
 .textbox_default_output textarea {
-    height: calc(100dvh - 118px);
+    height: calc(100dvh - 126px);
 }
 
 .textbox textarea {
-    height: calc(100dvh - 145px)
+    height: calc(100dvh - 153px)
 }
 
 .textbox_logits textarea {
-    height: calc(100dvh - 205px);
+    height: calc(100dvh - 213px);
 }
 
 .textbox_logits_notebook textarea {
-    height: calc(100dvh - 221px);
+    height: calc(100dvh - 229px);
 }
 
 .monospace textarea {
@@ -318,7 +375,7 @@ audio {
 }
 
 #notebook-token-counter {
-    top: calc(100dvh - 172px) !important;
+    top: calc(100dvh - 180px) !important;
 }
 
 #default-token-counter span, #notebook-token-counter span {
@@ -614,10 +671,10 @@ audio {
 }
 
 #chat-input textarea {
-    background: #f3f4f6;
+    background: var(--bg-input);
     padding: 0.675rem 2.5rem 0.6rem;
     margin-top: 0.15rem;
-    border: 1px solid var(--border-color-primary);
+    border: 1px solid var(--border);
     border-radius: 1.5rem;
     overflow-y: auto !important;
 }
@@ -866,10 +923,6 @@ audio {
     background-color: transparent;
 }
 
-.dark #show-controls:hover {
-    background-color: var(--selected-item-color-dark);
-}
-
 #show-controls label {
     display: flex;
     flex-direction: row-reverse;
@@ -1273,21 +1326,21 @@ audio {
   Dark theme
 ---------------------------------------------- */
 .dark .header_bar {
-    background-color: #1a1a1a !important;
+    background-color: var(--bg-rail) !important;
 }
 
 .dark .header_bar button.selected {
-    background: var(--selected-item-color-dark);
+    background: var(--bg-active);
 }
 
 .dark #chat-input textarea {
-    background: var(--light-gray);
-    color: white !important;
-    border-color: rgba(255, 255, 255, 0.06);
+    background: var(--bg-input);
+    color: var(--text) !important;
+    border-color: var(--border);
 }
 
 .dark #chat-input textarea::placeholder {
-    color: #9ca3af;
+    color: var(--text-muted);
 }
 
 .dark .hover-menu {
@@ -1302,19 +1355,16 @@ audio {
 
 .dark #chat-controls,
 .dark #past-chats-row {
-    background-color: var(--darker-gray);
     border: 0 !important;
     box-shadow: none;
 }
 
-.dark gradio-app .gradio-container.gradio-container-4-37-2 .contain #past-chats .selected,
 .dark gradio-app .gradio-container.gradio-container-4-37-2 .contain #past-chats label:hover {
-    background-color: var(--selected-item-color-dark) !important;
+    background-color: var(--bg-hover) !important;
 }
 
-.dark #past-chats-row,
-.dark #chat-controls {
-    background-color: var(--darker-gray);
+.dark gradio-app .gradio-container.gradio-container-4-37-2 .contain #past-chats .selected {
+    background-color: var(--bg-active) !important;
 }
 
 .dark #past-chats-toggle,
@@ -1341,25 +1391,18 @@ audio {
   Light theme
 ---------------------------------------------- */
 .header_bar {
-    background-color: #e4e8f0 !important;
+    background-color: var(--bg-rail) !important;
 }
 
 .header_bar button.selected {
-    background: #c8d8f5;
+    background: var(--bg-active);
 }
 
 #chat-controls,
 #past-chats-row {
-    background-color: var(--light-theme-gray);
-}
-
-.dark #chat-controls {
-    border-left: 1px solid rgba(255, 255, 255, 0.06);
+    background-color: var(--bg-sidebar);
 }
 
-.dark #past-chats-row {
-    border-right: 1px solid rgba(255, 255, 255, 0.06);
-}
 
 #past-chats-toggle,
 #chat-controls-toggle,
@@ -1829,23 +1872,23 @@ button:focus {
 }
 
 #character-context textarea {
-    height: calc((100vh - 350px) * 2/3) !important;
+    height: calc((100vh - 358px) * 2/3) !important;
     min-height: 90px !important;
 }
 
 #character-greeting textarea {
-    height: calc((100vh - 350px) * 1/3) !important;
+    height: calc((100vh - 358px) * 1/3) !important;
     min-height: 90px !important;
 }
 
 #user-description textarea {
-    height: calc(100vh - 334px) !important;
+    height: calc(100vh - 342px) !important;
     min-height: 90px !important;
 }
 
 #instruction-template-str textarea,
 #chat-template-str textarea {
-    height: calc(100vh - 300px) !important;
+    height: calc(100vh - 308px) !important;
     min-height: 90px !important;
 }
 
@@ -1870,7 +1913,7 @@ button:focus {
 }
 
 .dark .sidebar-vertical-separator {
-    border-bottom: 1px solid rgba(255, 255, 255, 0.06);
+    border-bottom: 1px solid var(--border-soft);
 }
 
 button#swap-height-width {
@@ -2105,3 +2148,578 @@ thead + tbody tr:first-child th {
     text-overflow: ellipsis;
     white-space: nowrap;
 }
+
+/* Mode selector — segmented control
+   Targets the gr.Radio at #chat-mode and replaces the default
+   stacked / wrap-prone radio appearance with a 3-up segmented
+   control. The actual  is hidden; the parent
+
    • +## Loading a model automatically + +To skip the Model tab on every launch, add this to `user_data/CMD_FLAGS.txt`: + +``` +--model my-model.gguf +``` + +Replace `my-model.gguf` with the name of a file in `user_data/models`. The model will load on startup. + +To pass extra flags, put each on its own line: + +``` +--model my-model.gguf +--cache-type q8_0 +``` + ## Installation For the desktop app, see the [portable builds](https://github.com/oobabooga/textgen/releases). The options below run the web UI in your browser instead. @@ -122,6 +139,9 @@ deactivate For users who need additional backends (ExLlamaV3, Transformers), training, image generation, or extensions like TTS, voice input, and translation. Requires ~10GB disk space and downloads PyTorch. +
    +Installation details + 1. Clone the repository, or [download its source code](https://github.com/oobabooga/textgen/archive/refs/heads/main.zip) and extract it. 2. Run the startup script for your OS: `start_windows.bat`, `start_linux.sh`, or `start_macos.sh`. 3. When prompted, select your GPU vendor. @@ -135,12 +155,7 @@ To update, run the update script for your OS: `update_wizard_windows.bat`, `upda To reinstall with a fresh Python environment, delete the `installer_files` folder and run the `start_` script again. -
    - -One-click installer details - - -### One-click-installer +### One-click installer details The script uses Miniforge to set up a Conda environment in the `installer_files` folder. @@ -150,13 +165,6 @@ If you ever need to install something manually in the `installer_files` environm * To install requirements for extensions, it is recommended to use the update wizard script with the "Install/update extensions requirements" option. At the end, this script will install the main requirements for the project to make sure that they take precedence in case of version conflicts. * For automated installation, you can use the `GPU_CHOICE`, `LAUNCH_AFTER_INSTALL`, and `INSTALL_EXTENSIONS` environment variables. For instance: `GPU_CHOICE=A LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh`. -
    - -
    - -Manual full installation with conda or docker - - ### Full installation with Conda #### 0. Install Conda @@ -267,12 +275,13 @@ conda activate textgen cd textgen pip install -r --upgrade ``` +
    +### Command-line flags +
    - -List of command-line flags - +Show full list ```txt usage: server.py [-h] [--user-data-dir USER_DATA_DIR] [--multi-user] [--model MODEL] [--lora LORA [LORA ...]] [--model-dir MODEL_DIR] [--lora-dir LORA_DIR] [--model-menu] [--settings SETTINGS] @@ -281,18 +290,19 @@ usage: server.py [-h] [--user-data-dir USER_DATA_DIR] [--multi-user] [--model MO [--loader LOADER] [--ctx-size N] [--cache-type N] [--model-draft MODEL_DRAFT] [--draft-max DRAFT_MAX] [--gpu-layers-draft GPU_LAYERS_DRAFT] [--device-draft DEVICE_DRAFT] [--ctx-size-draft CTX_SIZE_DRAFT] [--spec-type {none,ngram-mod,ngram-simple,ngram-map-k,ngram-map-k4v,ngram-cache}] [--spec-ngram-size-n SPEC_NGRAM_SIZE_N] [--spec-ngram-size-m SPEC_NGRAM_SIZE_M] [--spec-ngram-min-hits SPEC_NGRAM_MIN_HITS] [--gpu-layers N] [--cpu-moe] [--mmproj MMPROJ] [--streaming-llm] [--tensor-split TENSOR_SPLIT] - [--row-split] [--no-mmap] [--mlock] [--no-kv-offload] [--batch-size BATCH_SIZE] [--ubatch-size UBATCH_SIZE] [--threads THREADS] [--threads-batch THREADS_BATCH] [--numa] - [--parallel PARALLEL] [--fit-target FIT_TARGET] [--extra-flags EXTRA_FLAGS] [--ik] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] - [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] [--load-in-4bit] [--use_double_quant] - [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--gpu-split GPU_SPLIT] [--enable-tp] [--tp-backend TP_BACKEND] [--cfg-cache] [--listen] [--listen-port LISTEN_PORT] - [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] - [--subpath SUBPATH] [--old-colors] [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] [--api-key API_KEY] [--admin-key ADMIN_KEY] - [--api-enable-ipv6] [--api-disable-ipv4] [--nowebui] [--temperature N] [--dynatemp-low N] [--dynatemp-high N] [--dynatemp-exponent N] [--smoothing-factor N] [--smoothing-curve N] - [--top-p N] [--top-k N] [--min-p N] [--top-n-sigma N] [--typical-p N] [--xtc-threshold N] [--xtc-probability N] [--epsilon-cutoff N] [--eta-cutoff N] [--tfs N] [--top-a N] - [--adaptive-target N] [--adaptive-decay N] [--dry-multiplier N] [--dry-allowed-length N] [--dry-base N] [--repetition-penalty N] [--frequency-penalty N] [--presence-penalty N] - [--encoder-repetition-penalty N] [--no-repeat-ngram-size N] [--repetition-penalty-range N] [--penalty-alpha N] [--guidance-scale N] [--mirostat-mode N] [--mirostat-tau N] - [--mirostat-eta N] [--do-sample | --no-do-sample] [--dynamic-temperature | --no-dynamic-temperature] [--temperature-last | --no-temperature-last] [--sampler-priority N] - [--dry-sequence-breakers N] [--enable-thinking | --no-enable-thinking] [--reasoning-effort N] [--preserve-thinking | --no-preserve-thinking] [--chat-template-file CHAT_TEMPLATE_FILE] + [--split-mode {layer,row,tensor,none}] [--no-mmap] [--mlock] [--no-kv-offload] [--batch-size BATCH_SIZE] [--ubatch-size UBATCH_SIZE] [--threads THREADS] + [--threads-batch THREADS_BATCH] [--numa] [--parallel PARALLEL] [--fit-target FIT_TARGET] [--extra-flags EXTRA_FLAGS] [--ik] [--cpu] [--cpu-memory CPU_MEMORY] [--disk] + [--disk-cache-dir DISK_CACHE_DIR] [--load-in-8bit] [--bf16] [--no-cache] [--trust-remote-code] [--force-safetensors] [--no_use_fast] [--attn-implementation IMPLEMENTATION] + [--load-in-4bit] [--use_double_quant] [--compute_dtype COMPUTE_DTYPE] [--quant_type QUANT_TYPE] [--gpu-split GPU_SPLIT] [--enable-tp] [--tp-backend TP_BACKEND] [--cfg-cache] + [--listen] [--listen-port LISTEN_PORT] [--listen-host LISTEN_HOST] [--share] [--auto-launch] [--gradio-auth GRADIO_AUTH] [--gradio-auth-path GRADIO_AUTH_PATH] + [--ssl-keyfile SSL_KEYFILE] [--ssl-certfile SSL_CERTFILE] [--subpath SUBPATH] [--old-colors] [--portable] [--api] [--public-api] [--public-api-id PUBLIC_API_ID] [--api-port API_PORT] + [--api-key API_KEY] [--admin-key ADMIN_KEY] [--api-enable-ipv6] [--api-disable-ipv4] [--nowebui] [--temperature N] [--dynatemp-low N] [--dynatemp-high N] [--dynatemp-exponent N] + [--smoothing-factor N] [--smoothing-curve N] [--top-p N] [--top-k N] [--min-p N] [--top-n-sigma N] [--typical-p N] [--xtc-threshold N] [--xtc-probability N] [--epsilon-cutoff N] + [--eta-cutoff N] [--tfs N] [--top-a N] [--adaptive-target N] [--adaptive-decay N] [--dry-multiplier N] [--dry-allowed-length N] [--dry-base N] [--repetition-penalty N] + [--frequency-penalty N] [--presence-penalty N] [--encoder-repetition-penalty N] [--no-repeat-ngram-size N] [--repetition-penalty-range N] [--penalty-alpha N] [--guidance-scale N] + [--mirostat-mode N] [--mirostat-tau N] [--mirostat-eta N] [--do-sample | --no-do-sample] [--dynamic-temperature | --no-dynamic-temperature] + [--temperature-last | --no-temperature-last] [--sampler-priority N] [--dry-sequence-breakers N] [--enable-thinking | --no-enable-thinking] [--reasoning-effort N] + [--preserve-thinking | --no-preserve-thinking] [--chat-template-file CHAT_TEMPLATE_FILE] TextGen @@ -349,7 +359,7 @@ llama.cpp: --mmproj MMPROJ Path to the mmproj file for vision models. --streaming-llm Activate StreamingLLM to avoid re-evaluating the entire prompt when old messages are removed. --tensor-split TENSOR_SPLIT Split the model across multiple GPUs. Comma-separated list of proportions. Example: 60,40. - --row-split Split the model by rows across GPUs. This may improve multi-gpu performance. + --split-mode {layer,row,tensor,none} How to split the model across multiple GPUs. "tensor" can make multi-GPU significantly faster. --no-mmap Prevent mmap from being used. --mlock Force the system to keep the model in RAM. --no-kv-offload Do not offload the K, Q, V to the GPU. This saves VRAM but reduces performance. From 210c65aae3af26ab2d54b0f08fb5f16e21b3e352 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 5 May 2026 11:20:35 -0700 Subject: [PATCH 1622/1701] Update README --- README.md | 64 +++++++++++++++++++++++++++---------------------------- 1 file changed, 31 insertions(+), 33 deletions(-) diff --git a/README.md b/README.md index 0e7b747593..c3a6372da5 100644 --- a/README.md +++ b/README.md @@ -86,23 +86,6 @@ textgen These formats require the one-click installer (not the portable build).
    -## Loading a model automatically - -To skip the Model tab on every launch, add this to `user_data/CMD_FLAGS.txt`: - -``` ---model my-model.gguf -``` - -Replace `my-model.gguf` with the name of a file in `user_data/models`. The model will load on startup. - -To pass extra flags, put each on its own line: - -``` ---model my-model.gguf ---cache-type q8_0 -``` - ## Installation For the desktop app, see the [portable builds](https://github.com/oobabooga/textgen/releases). The options below run the web UI in your browser instead. @@ -135,35 +118,33 @@ python server.py --portable --api --auto-launch deactivate ``` -### One-click installer +### Full installation For users who need additional backends (ExLlamaV3, Transformers), training, image generation, or extensions like TTS, voice input, and translation. Requires ~10GB disk space and downloads PyTorch.
    Installation details +### One-click installer + 1. Clone the repository, or [download its source code](https://github.com/oobabooga/textgen/archive/refs/heads/main.zip) and extract it. 2. Run the startup script for your OS: `start_windows.bat`, `start_linux.sh`, or `start_macos.sh`. 3. When prompted, select your GPU vendor. 4. After installation, open `http://127.0.0.1:7860` in your browser. -To restart the web UI later, run the same `start_` script. - -You can pass command-line flags directly (e.g., `./start_linux.sh --help`), or add them to `user_data/CMD_FLAGS.txt` (e.g., `--api` to enable the API). +After installation: -To update, run the update script for your OS: `update_wizard_windows.bat`, `update_wizard_linux.sh`, or `update_wizard_macos.sh`. +* **Restart**: run the same `start_` script. +* **Pass command-line flags**: directly (e.g., `./start_linux.sh --help`), or persist them in `user_data/CMD_FLAGS.txt` (e.g., `--api` to enable the API). +* **Update**: run the update script for your OS (`update_wizard_windows.bat`, `update_wizard_linux.sh`, or `update_wizard_macos.sh`). +* **Reinstall from scratch**: delete the `installer_files` folder and run the `start_` script again. +* **Install extension requirements**: use the update wizard's "Install/update extensions requirements" option. It reinstalls the main project requirements at the end to ensure they take precedence over conflicting extension dependencies. -To reinstall with a fresh Python environment, delete the `installer_files` folder and run the `start_` script again. +Notes: -### One-click installer details - -The script uses Miniforge to set up a Conda environment in the `installer_files` folder. - -If you ever need to install something manually in the `installer_files` environment, you can launch an interactive shell using the cmd script: `cmd_linux.sh`, `cmd_windows.bat`, or `cmd_macos.sh`. - -* There is no need to run any of those scripts (`start_`, `update_wizard_`, or `cmd_`) as admin/root. -* To install requirements for extensions, it is recommended to use the update wizard script with the "Install/update extensions requirements" option. At the end, this script will install the main requirements for the project to make sure that they take precedence in case of version conflicts. -* For automated installation, you can use the `GPU_CHOICE`, `LAUNCH_AFTER_INSTALL`, and `INSTALL_EXTENSIONS` environment variables. For instance: `GPU_CHOICE=A LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh`. +* These scripts (`start_`, `update_wizard_`, `cmd_`) don't need to run as admin/root. +* For automated installation, set the `GPU_CHOICE`, `LAUNCH_AFTER_INSTALL`, and `INSTALL_EXTENSIONS` environment variables. Example: `GPU_CHOICE=A LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh`. +* Under the hood, the script uses Miniforge to set up a Conda environment in `installer_files/`. To run anything manually in this environment, launch an interactive shell using `cmd_linux.sh`, `cmd_windows.bat`, or `cmd_macos.sh`. ### Full installation with Conda @@ -278,7 +259,7 @@ pip install -r --upgrade
    -### Command-line flags +## Command-line flags
    Show full list @@ -472,6 +453,23 @@ API generation defaults:
    +## Loading a model automatically + +To skip the Model tab on every launch, add this to `user_data/CMD_FLAGS.txt`: + +``` +--model my-model.gguf +``` + +Replace `my-model.gguf` with the name of a file in `user_data/models`. The model will load on startup. + +To pass extra flags, put each on its own line: + +``` +--model my-model.gguf +--cache-type q8_0 +``` + ## Documentation https://github.com/oobabooga/textgen/wiki From cd3ac2716baa47a488731b6625be58f873285fb5 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 5 May 2026 11:27:55 -0700 Subject: [PATCH 1623/1701] Update README --- README.md | 30 ++++++++++++++---------------- 1 file changed, 14 insertions(+), 16 deletions(-) diff --git a/README.md b/README.md index c3a6372da5..a97e29476f 100644 --- a/README.md +++ b/README.md @@ -205,7 +205,7 @@ Requirements file to use: | Apple Intel | `requirements_apple_intel.txt` | | Apple Silicon | `requirements_apple_silicon.txt` | -### Start the web UI +#### 4. Start the web UI ``` conda activate textgen @@ -213,15 +213,23 @@ cd textgen python server.py ``` -Then browse to +Then browse to `http://127.0.0.1:7860`. -`http://127.0.0.1:7860` +#### Manual compilation -#### Manual install +The `requirements*.txt` files above contain wheels precompiled through GitHub Actions. To compile manually (e.g., if no wheels are available for your hardware), use `requirements_nowheels.txt` and install your desired loaders manually. -The `requirements*.txt` above contain various wheels precompiled through GitHub Actions. If you wish to compile things manually, or if you need to because no suitable wheels are available for your hardware, you can use `requirements_nowheels.txt` and then install your desired loaders manually. +#### Updating the requirements -### Alternative: Docker +From time to time, the `requirements*.txt` files change. To update: + +``` +conda activate textgen +cd textgen +pip install -r --upgrade +``` + +### Docker ``` For NVIDIA GPU: @@ -247,16 +255,6 @@ docker compose up --build * You need to have Docker Compose v2.17 or higher installed. See [this guide](https://github.com/oobabooga/textgen/wiki/09-%E2%80%90-Docker) for instructions. * For additional docker files, check out [this repository](https://github.com/Atinoda/text-generation-webui-docker). -### Updating the requirements - -From time to time, the `requirements*.txt` change. To update, use these commands: - -``` -conda activate textgen -cd textgen -pip install -r --upgrade -``` -
    ## Command-line flags From 4de6027ed3465588f40e58ed43ac8303c02743c4 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 6 May 2026 06:52:10 -0700 Subject: [PATCH 1624/1701] Fix big character picture failing to load in Electron (#7540) --- js/global_scope_js.js | 2 ++ 1 file changed, 2 insertions(+) diff --git a/js/global_scope_js.js b/js/global_scope_js.js index b7ae6b746e..bd9e6a1fa2 100644 --- a/js/global_scope_js.js +++ b/js/global_scope_js.js @@ -3,6 +3,8 @@ // ------------------------------------------------- function getProfilePictureUrl() { + const thumb = document.querySelector(".pfp_character"); + if (thumb) return thumb.src.replace("pfp_character_thumb.png", "pfp_character.png"); return "/file/user_data/cache/pfp_character.png?time=" + Date.now(); } From 71525d51a6f44df77404fb50cbeab6f08c0af0c9 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 6 May 2026 07:00:39 -0700 Subject: [PATCH 1625/1701] Fix truncation length reverting after model load on UI reload (closes #7540) --- modules/ui_model_menu.py | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py index 27000eab63..aa27adcf7b 100644 --- a/modules/ui_model_menu.py +++ b/modules/ui_model_menu.py @@ -405,11 +405,11 @@ def downloader_thread_target(): def update_truncation_length(current_length, state): if 'loader' in state: if state['loader'].lower().startswith('exllama') or state['loader'] == 'llama.cpp': - if state['ctx_size'] > 0: - return state['ctx_size'] - # ctx_size == 0 means auto: use the actual value from the server - return shared.settings['truncation_length'] + new_length = state['ctx_size'] if state['ctx_size'] > 0 else shared.settings['truncation_length'] + if not shared.args.multi_user: + shared.persistent_interface_state['truncation_length'] = new_length + return new_length return current_length From 8536d581786c13f325e8005d5753a94509935af3 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 6 May 2026 11:30:29 -0700 Subject: [PATCH 1626/1701] Update llama.cpp --- requirements/full/requirements.txt | 8 ++++---- requirements/full/requirements_amd.txt | 4 ++-- requirements/full/requirements_apple_intel.txt | 2 +- requirements/full/requirements_apple_silicon.txt | 2 +- requirements/full/requirements_cpu_only.txt | 8 ++++---- requirements/portable/requirements.txt | 4 ++-- requirements/portable/requirements_amd.txt | 4 ++-- requirements/portable/requirements_apple_intel.txt | 2 +- requirements/portable/requirements_apple_silicon.txt | 2 +- requirements/portable/requirements_cpu_only.txt | 4 ++-- requirements/portable/requirements_cuda131.txt | 4 ++-- requirements/portable/requirements_ik.txt | 4 ++-- requirements/portable/requirements_ik_cpu_only.txt | 4 ++-- requirements/portable/requirements_ik_cuda131.txt | 4 ++-- requirements/portable/requirements_vulkan.txt | 4 ++-- 15 files changed, 30 insertions(+), 30 deletions(-) diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt index 9a77872829..8a8d6d0f7f 100644 --- a/requirements/full/requirements.txt +++ b/requirements/full/requirements.txt @@ -43,10 +43,10 @@ sse-starlette==1.6.5 tiktoken # CUDA wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/ik_llama_cpp_binaries-0.124.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/ik_llama_cpp_binaries-0.124.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" https://github.com/turboderp-org/exllamav3/releases/download/v0.0.30/exllamav3-0.0.30+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13" https://github.com/turboderp-org/exllamav3/releases/download/v0.0.30/exllamav3-0.0.30+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13" https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13" diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt index b650ce49db..ba0975bb6e 100644 --- a/requirements/full/requirements_amd.txt +++ b/requirements/full/requirements_amd.txt @@ -39,5 +39,5 @@ sse-starlette==1.6.5 tiktoken # AMD wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt index 002dfee9fa..015ccb7767 100644 --- a/requirements/full/requirements_apple_intel.txt +++ b/requirements/full/requirements_apple_intel.txt @@ -39,4 +39,4 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt index 30fa95a9e7..32eb1b837c 100644 --- a/requirements/full/requirements_apple_silicon.txt +++ b/requirements/full/requirements_apple_silicon.txt @@ -39,4 +39,4 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt index 4fc3d7a7c2..196d6bbd12 100644 --- a/requirements/full/requirements_cpu_only.txt +++ b/requirements/full/requirements_cpu_only.txt @@ -39,7 +39,7 @@ sse-starlette==1.6.5 tiktoken # llama.cpp (CPU only) -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/ik_llama_cpp_binaries-0.124.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/ik_llama_cpp_binaries-0.124.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt index 0e88c45fff..4e88e6ffee 100644 --- a/requirements/portable/requirements.txt +++ b/requirements/portable/requirements.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # CUDA wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt index 9a95370b69..ba22ab94d7 100644 --- a/requirements/portable/requirements_amd.txt +++ b/requirements/portable/requirements_amd.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # AMD wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt index d341d1f9c5..b30f711665 100644 --- a/requirements/portable/requirements_apple_intel.txt +++ b/requirements/portable/requirements_apple_intel.txt @@ -25,4 +25,4 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt index ea39435df0..38a97270f8 100644 --- a/requirements/portable/requirements_apple_silicon.txt +++ b/requirements/portable/requirements_apple_silicon.txt @@ -25,4 +25,4 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt index 214580606f..21d00b727e 100644 --- a/requirements/portable/requirements_cpu_only.txt +++ b/requirements/portable/requirements_cpu_only.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # llama.cpp (CPU only) -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt index c060996f85..bc1ecb4306 100644 --- a/requirements/portable/requirements_cuda131.txt +++ b/requirements/portable/requirements_cuda131.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # CUDA wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt index eee5c31f9d..1603f5e2cd 100644 --- a/requirements/portable/requirements_ik.txt +++ b/requirements/portable/requirements_ik.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # CUDA wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/ik_llama_cpp_binaries-0.124.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/ik_llama_cpp_binaries-0.124.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt index d7caf7c21e..3f0ea6fdd2 100644 --- a/requirements/portable/requirements_ik_cpu_only.txt +++ b/requirements/portable/requirements_ik_cpu_only.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # ik_llama.cpp (CPU only) -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/ik_llama_cpp_binaries-0.124.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/ik_llama_cpp_binaries-0.124.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt index 7325c82f9f..b0bb9e076a 100644 --- a/requirements/portable/requirements_ik_cuda131.txt +++ b/requirements/portable/requirements_ik_cuda131.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # CUDA wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/ik_llama_cpp_binaries-0.124.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/ik_llama_cpp_binaries-0.124.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt index 89533d4d57..10bbcf1638 100644 --- a/requirements/portable/requirements_vulkan.txt +++ b/requirements/portable/requirements_vulkan.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # Vulkan wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.124.0/llama_cpp_binaries-0.124.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" From bf50e7e705218aaef1379d01fe419ee82bd9fa04 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 6 May 2026 20:47:23 -0700 Subject: [PATCH 1627/1701] Fix --listen mode in the new Electron launcher --- desktop/textgen.bat | 1 + desktop/textgen.sh | 2 +- 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/desktop/textgen.bat b/desktop/textgen.bat index b669805f2f..c437733a44 100644 --- a/desktop/textgen.bat +++ b/desktop/textgen.bat @@ -12,5 +12,6 @@ exit /b %errorlevel% "%APP%\portable_env\python.exe" "%APP%\server.py" --help exit /b %errorlevel% :server +cd /d "%APP%" || exit /b 1 "%APP%\portable_env\python.exe" "%APP%\server.py" --portable --api %* exit /b %errorlevel% diff --git a/desktop/textgen.sh b/desktop/textgen.sh index 66e85b7d3d..94da25e019 100644 --- a/desktop/textgen.sh +++ b/desktop/textgen.sh @@ -8,7 +8,7 @@ for arg; do exec "$PY" "$APP/server.py" --help ;; --nowebui|--listen) - exec "$PY" "$APP/server.py" --portable --api "$@" + cd "$APP" && exec "$PY" "$APP/server.py" --portable --api "$@" ;; esac done From 3e38f44a2e630cdbe52837a2bb748d63d3fcf69e Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 6 May 2026 21:06:36 -0700 Subject: [PATCH 1628/1701] Fix speculative decoding broken by llama.cpp arg rename (closes #7541) --- modules/llama_cpp_server.py | 16 ++++++++++------ 1 file changed, 10 insertions(+), 6 deletions(-) diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py index 0402207563..0ce15828b5 100644 --- a/modules/llama_cpp_server.py +++ b/modules/llama_cpp_server.py @@ -485,7 +485,7 @@ def _start_server(self): cmd += ["--model-draft", str(model_file)] if shared.args.draft_max > 0: - cmd += ["--draft-max", str(shared.args.draft_max)] + cmd += ["--spec-draft-n-max", str(shared.args.draft_max)] if shared.args.gpu_layers_draft > 0: cmd += ["--gpu-layers-draft", str(shared.args.gpu_layers_draft)] if shared.args.device_draft: @@ -494,11 +494,15 @@ def _start_server(self): cmd += ["--ctx-size-draft", str(shared.args.ctx_size_draft)] if shared.args.spec_type != 'none': cmd += ["--spec-type", shared.args.spec_type] - cmd += ["--draft-max", str(shared.args.draft_max)] - cmd += ["--draft-min", "48"] - cmd += ["--spec-ngram-size-n", str(shared.args.spec_ngram_size_n)] - cmd += ["--spec-ngram-size-m", str(shared.args.spec_ngram_size_m)] - cmd += ["--spec-ngram-min-hits", str(shared.args.spec_ngram_min_hits)] + if shared.args.spec_type == 'ngram-mod': + cmd += ["--spec-ngram-mod-n-match", str(shared.args.spec_ngram_size_n)] + cmd += ["--spec-ngram-mod-n-max", str(shared.args.draft_max)] + cmd += ["--spec-ngram-mod-n-min", str(shared.args.spec_ngram_size_m)] + elif shared.args.spec_type in ('ngram-simple', 'ngram-map-k', 'ngram-map-k4v'): + prefix = f"--spec-{shared.args.spec_type}" + cmd += [f"{prefix}-size-n", str(shared.args.spec_ngram_size_n)] + cmd += [f"{prefix}-size-m", str(shared.args.spec_ngram_size_m)] + cmd += [f"{prefix}-min-hits", str(shared.args.spec_ngram_min_hits)] cmd += ["--parallel", str(shared.args.parallel)] if shared.args.streaming_llm: cmd += ["--cache-reuse", "1"] From 4d3f355f04e4e874e76f8da056e9417948ab1444 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 6 May 2026 22:02:56 -0700 Subject: [PATCH 1629/1701] Add a `--no-electron` flag --- desktop/textgen.bat | 1 + desktop/textgen.sh | 2 +- modules/shared.py | 4 ++++ 3 files changed, 6 insertions(+), 1 deletion(-) diff --git a/desktop/textgen.bat b/desktop/textgen.bat index c437733a44..d7a170947b 100644 --- a/desktop/textgen.bat +++ b/desktop/textgen.bat @@ -5,6 +5,7 @@ for %%a in (%*) do ( if /i "%%~a"=="-h" goto :help if /i "%%~a"=="--nowebui" goto :server if /i "%%~a"=="--listen" goto :server + if /i "%%~a"=="--no-electron" goto :server ) "%APP%\electron\electron.exe" "%APP%" -- %* exit /b %errorlevel% diff --git a/desktop/textgen.sh b/desktop/textgen.sh index 94da25e019..be184d3ccc 100644 --- a/desktop/textgen.sh +++ b/desktop/textgen.sh @@ -7,7 +7,7 @@ for arg; do --help|-h) exec "$PY" "$APP/server.py" --help ;; - --nowebui|--listen) + --nowebui|--listen|--no-electron) cd "$APP" && exec "$PY" "$APP/server.py" --portable --api "$@" ;; esac diff --git a/modules/shared.py b/modules/shared.py index 5f217d0b13..d9d3719489 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -213,6 +213,10 @@ group.add_argument('--preserve-thinking', action=argparse.BooleanOptionalAction, default=False, help='Preserve thinking blocks from prior turns in the chat template') group.add_argument('--chat-template-file', type=str, default=None, help='Path to a chat template file (.jinja, .jinja2, or .yaml) to use as the default instruction template for API requests. Overrides the model\'s built-in template.') +# Electron +group = parser.add_argument_group('Electron') +group.add_argument('--no-electron', action='store_true', help='In portable builds, skip the Electron desktop window. Useful if you prefer to use the web UI in the browser.') + # Handle CMD_FLAGS.txt cmd_flags_path = user_data_dir / "CMD_FLAGS.txt" if cmd_flags_path.exists(): From bde878f0a330a8d81c71bbf883ad53849170cc47 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Wed, 6 May 2026 22:15:23 -0700 Subject: [PATCH 1630/1701] Electron: disable spellcheck --- desktop/main.js | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/desktop/main.js b/desktop/main.js index 144b0863c8..e2a05f901a 100644 --- a/desktop/main.js +++ b/desktop/main.js @@ -111,7 +111,7 @@ function createWindow(port) { ...bounds, title: TITLE, autoHideMenuBar: true, - webPreferences: { nodeIntegration: false, contextIsolation: true }, + webPreferences: { nodeIntegration: false, contextIsolation: true, spellcheck: false }, }); if (state && state.maximized) mainWindow.maximize(); mainWindow.on("page-title-updated", (e) => e.preventDefault()); From 5c0fab6e005777f7fa5e570975656d3fe44cbc62 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Thu, 7 May 2026 06:23:18 -0700 Subject: [PATCH 1631/1701] Add a scroll animation when new chat messages are sent (Gemini-like) --- js/global_scope_js.js | 19 ++++++++++++++++++- js/main.js | 3 +++ 2 files changed, 21 insertions(+), 1 deletion(-) diff --git a/js/global_scope_js.js b/js/global_scope_js.js index bd9e6a1fa2..349a05729e 100644 --- a/js/global_scope_js.js +++ b/js/global_scope_js.js @@ -273,6 +273,7 @@ function removeLastClick() { } let _scrollPending = false; +const SMOOTH_SCROLL_WINDOW_MS = 700; function autoScrollToBottom() { if (_scrollPending) return; @@ -284,7 +285,11 @@ function autoScrollToBottom() { if (chatParent) { const maxScroll = chatParent.scrollHeight - chatParent.clientHeight; if (maxScroll > 0 && chatParent.scrollTop < maxScroll - 1) { - chatParent.scrollTop = maxScroll; + if (Date.now() < window.smoothScrollUntilTs) { + chatParent.scrollTo({ top: maxScroll, behavior: "smooth" }); + } else { + chatParent.scrollTop = maxScroll; + } } } } @@ -335,6 +340,9 @@ function applyMorphdomUpdate(data) { const queryScope = target_element; + const messagesContainer = document.getElementsByClassName("messages")[0]; + const messagesCountBefore = messagesContainer ? messagesContainer.children.length : 0; + // Track open blocks and store their scroll positions const openBlocks = new Set(); const scrollPositions = {}; @@ -408,6 +416,15 @@ function applyMorphdomUpdate(data) { window.doSyntaxHighlighting(); } + // Only animate the padding jump on a fresh submission, not on chat switches or streaming chunks. + if (window.pendingGenerationStart) { + const messagesCountAfter = messagesContainer ? messagesContainer.children.length : 0; + if (messagesCountAfter > messagesCountBefore) { + window.smoothScrollUntilTs = Date.now() + SMOOTH_SCROLL_WINDOW_MS; + } + window.pendingGenerationStart = false; + } + // Auto-scroll runs both before and after padding update. // Before: so content growth isn't hidden by padding absorption. // After: so padding-added space is also scrolled into view. diff --git a/js/main.js b/js/main.js index 7d467357b4..d5420610b2 100644 --- a/js/main.js +++ b/js/main.js @@ -140,6 +140,8 @@ const targetElement = document.getElementById("chat").parentNode.parentNode.pare targetElement.classList.add("pretty_scrollbar"); targetElement.classList.add("chat-parent"); window.isScrolled = false; +window.pendingGenerationStart = false; +window.smoothScrollUntilTs = 0; let scrollTimeout; let lastScrollTop = 0; let lastScrollHeight = 0; @@ -175,6 +177,7 @@ const observer = new MutationObserver(function() { if (targetElement.classList.contains("_generating")) { document.getElementById("stop").style.display = "flex"; document.getElementById("Generate").style.display = "none"; + window.pendingGenerationStart = true; // If the user is near the bottom, ensure auto-scroll is enabled // for the new reply. This catches cases where isScrolled was // incorrectly set to true by layout shifts during page load, etc. From 88d47dc4127a9cf3c183f034b2c48ae5e4164e84 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Thu, 7 May 2026 06:34:10 -0700 Subject: [PATCH 1632/1701] Don't clear chat input when no model is loaded (closes #7542) --- modules/chat.py | 12 ++++++++---- modules/ui_chat.py | 2 ++ 2 files changed, 10 insertions(+), 4 deletions(-) diff --git a/modules/chat.py b/modules/chat.py index 531a8f9164..942b7bfdd6 100644 --- a/modules/chat.py +++ b/modules/chat.py @@ -1304,6 +1304,13 @@ def character_is_loaded(state, raise_exception=False): return True +def check_model_loaded_or_raise(): + model_is_loaded, error_message = utils.check_model_loaded() + if not model_is_loaded: + import gradio as gr + raise gr.Error(error_message) + + def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False): ''' Same as above but returns HTML for the UI. @@ -1316,10 +1323,7 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False): if not character_is_loaded(state): return - model_is_loaded, error_message = utils.check_model_loaded() - if not model_is_loaded: - import gradio as gr - raise gr.Error(error_message) + check_model_loaded_or_raise() if state['start_with'] != '' and not _continue: if regenerate: diff --git a/modules/ui_chat.py b/modules/ui_chat.py index ddd57709aa..75d37e0e52 100644 --- a/modules/ui_chat.py +++ b/modules/ui_chat.py @@ -242,6 +242,7 @@ def create_event_handlers(): shared.gradio['Generate'].click( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( + chat.check_model_loaded_or_raise, None, None, show_progress=False).success( lambda x: (x, {"text": "", "files": []}), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False).then( lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then( chat.generate_chat_reply_wrapper, gradio(inputs), gradio('display', 'history'), show_progress=False).then( @@ -250,6 +251,7 @@ def create_event_handlers(): shared.gradio['textbox'].submit( ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then( + chat.check_model_loaded_or_raise, None, None, show_progress=False).success( lambda x: (x, {"text": "", "files": []}), gradio('textbox'), gradio('Chat input', 'textbox'), show_progress=False).then( lambda: None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.add("_generating")').then( chat.generate_chat_reply_wrapper, gradio(inputs), gradio('display', 'history'), show_progress=False).then( From 206b9d21a24f031625d0c7bf6aff180ccc2ca64c Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Thu, 7 May 2026 06:48:32 -0700 Subject: [PATCH 1633/1701] Increase bottom padding of the last chat/chat-instruct message --- css/main.css | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/css/main.css b/css/main.css index c5643f0201..e6690d67f3 100644 --- a/css/main.css +++ b/css/main.css @@ -498,6 +498,10 @@ audio { padding-top: 20px; } +.chat > .messages > .message:last-child { + padding-bottom: 2rem; +} + .message-body { font-size: 16px; } From 4549fc6d2ddde7505210dedb0f1c54fcc55dd504 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Thu, 7 May 2026 08:53:40 -0700 Subject: [PATCH 1634/1701] Update llama.cpp --- requirements/full/requirements.txt | 8 ++++---- requirements/full/requirements_amd.txt | 4 ++-- requirements/full/requirements_apple_intel.txt | 2 +- requirements/full/requirements_apple_silicon.txt | 2 +- requirements/full/requirements_cpu_only.txt | 8 ++++---- requirements/portable/requirements.txt | 4 ++-- requirements/portable/requirements_amd.txt | 4 ++-- requirements/portable/requirements_apple_intel.txt | 2 +- requirements/portable/requirements_apple_silicon.txt | 2 +- requirements/portable/requirements_cpu_only.txt | 4 ++-- requirements/portable/requirements_cuda131.txt | 4 ++-- requirements/portable/requirements_ik.txt | 4 ++-- requirements/portable/requirements_ik_cpu_only.txt | 4 ++-- requirements/portable/requirements_ik_cuda131.txt | 4 ++-- requirements/portable/requirements_vulkan.txt | 4 ++-- 15 files changed, 30 insertions(+), 30 deletions(-) diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt index 8a8d6d0f7f..0f3691b57f 100644 --- a/requirements/full/requirements.txt +++ b/requirements/full/requirements.txt @@ -43,10 +43,10 @@ sse-starlette==1.6.5 tiktoken # CUDA wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" https://github.com/turboderp-org/exllamav3/releases/download/v0.0.30/exllamav3-0.0.30+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13" https://github.com/turboderp-org/exllamav3/releases/download/v0.0.30/exllamav3-0.0.30+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13" https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13" diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt index ba0975bb6e..75cb0a23d2 100644 --- a/requirements/full/requirements_amd.txt +++ b/requirements/full/requirements_amd.txt @@ -39,5 +39,5 @@ sse-starlette==1.6.5 tiktoken # AMD wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt index 015ccb7767..941ece967d 100644 --- a/requirements/full/requirements_apple_intel.txt +++ b/requirements/full/requirements_apple_intel.txt @@ -39,4 +39,4 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt index 32eb1b837c..cc604d0a37 100644 --- a/requirements/full/requirements_apple_silicon.txt +++ b/requirements/full/requirements_apple_silicon.txt @@ -39,4 +39,4 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt index 196d6bbd12..921ae8990a 100644 --- a/requirements/full/requirements_cpu_only.txt +++ b/requirements/full/requirements_cpu_only.txt @@ -39,7 +39,7 @@ sse-starlette==1.6.5 tiktoken # llama.cpp (CPU only) -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt index 4e88e6ffee..afbd6b6227 100644 --- a/requirements/portable/requirements.txt +++ b/requirements/portable/requirements.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # CUDA wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt index ba22ab94d7..30ee5d3ec3 100644 --- a/requirements/portable/requirements_amd.txt +++ b/requirements/portable/requirements_amd.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # AMD wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt index b30f711665..f59a6e18de 100644 --- a/requirements/portable/requirements_apple_intel.txt +++ b/requirements/portable/requirements_apple_intel.txt @@ -25,4 +25,4 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin" diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt index 38a97270f8..53a450658b 100644 --- a/requirements/portable/requirements_apple_silicon.txt +++ b/requirements/portable/requirements_apple_silicon.txt @@ -25,4 +25,4 @@ sse-starlette==1.6.5 tiktoken # Mac wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin" diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt index 21d00b727e..cac9961afd 100644 --- a/requirements/portable/requirements_cpu_only.txt +++ b/requirements/portable/requirements_cpu_only.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # llama.cpp (CPU only) -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt index bc1ecb4306..409cb7c547 100644 --- a/requirements/portable/requirements_cuda131.txt +++ b/requirements/portable/requirements_cuda131.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # CUDA wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt index 1603f5e2cd..8794882fd5 100644 --- a/requirements/portable/requirements_ik.txt +++ b/requirements/portable/requirements_ik.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # CUDA wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt index 3f0ea6fdd2..b0fa9c9e46 100644 --- a/requirements/portable/requirements_ik_cpu_only.txt +++ b/requirements/portable/requirements_ik_cpu_only.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # ik_llama.cpp (CPU only) -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows" diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt index b0bb9e076a..d2513b01d3 100644 --- a/requirements/portable/requirements_ik_cuda131.txt +++ b/requirements/portable/requirements_ik_cuda131.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # CUDA wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/ik_llama_cpp_binaries-0.125.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt index 10bbcf1638..0c2e86314e 100644 --- a/requirements/portable/requirements_vulkan.txt +++ b/requirements/portable/requirements_vulkan.txt @@ -25,5 +25,5 @@ sse-starlette==1.6.5 tiktoken # Vulkan wheels -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows" -https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.125.0/llama_cpp_binaries-0.125.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows" +https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" From 3ce136e918b2c49eb32da76e311bac6676702c60 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Thu, 7 May 2026 10:22:49 -0700 Subject: [PATCH 1635/1701] Update README --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index a97e29476f..40f69412d7 100644 --- a/README.md +++ b/README.md @@ -17,7 +17,7 @@ [![GitHub stars](https://img.shields.io/github/stars/oobabooga/textgen?style=for-the-badge&logo=github&logoColor=white&labelColor=black)](https://github.com/oobabooga/textgen) -[![Chat mode](https://raw.githubusercontent.com/oobabooga/screenshots/refs/heads/main/CHAT-4.7.png)](https://raw.githubusercontent.com/oobabooga/screenshots/refs/heads/main/CHAT-4.7.png) +[![Chat mode](https://raw.githubusercontent.com/oobabooga/screenshots/refs/heads/main/CHAT-4.8.png)](https://raw.githubusercontent.com/oobabooga/screenshots/refs/heads/main/CHAT-4.8.png) ## Get started in 1 minute From 3b9d0feedcf6904a5e1039c50f6e6c6cb7b7acc1 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Thu, 7 May 2026 10:27:35 -0700 Subject: [PATCH 1636/1701] Update README --- README.md | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/README.md b/README.md index 40f69412d7..693e5cd441 100644 --- a/README.md +++ b/README.md @@ -281,7 +281,7 @@ usage: server.py [-h] [--user-data-dir USER_DATA_DIR] [--multi-user] [--model MO [--frequency-penalty N] [--presence-penalty N] [--encoder-repetition-penalty N] [--no-repeat-ngram-size N] [--repetition-penalty-range N] [--penalty-alpha N] [--guidance-scale N] [--mirostat-mode N] [--mirostat-tau N] [--mirostat-eta N] [--do-sample | --no-do-sample] [--dynamic-temperature | --no-dynamic-temperature] [--temperature-last | --no-temperature-last] [--sampler-priority N] [--dry-sequence-breakers N] [--enable-thinking | --no-enable-thinking] [--reasoning-effort N] - [--preserve-thinking | --no-preserve-thinking] [--chat-template-file CHAT_TEMPLATE_FILE] + [--preserve-thinking | --no-preserve-thinking] [--chat-template-file CHAT_TEMPLATE_FILE] [--no-electron] TextGen @@ -447,6 +447,9 @@ API generation defaults: --preserve-thinking, --no-preserve-thinking Preserve thinking blocks from prior turns in the chat template --chat-template-file CHAT_TEMPLATE_FILE Path to a chat template file (.jinja, .jinja2, or .yaml) to use as the default instruction template for API requests. Overrides the model's built-in template. + +Electron: + --no-electron In portable builds, skip the Electron desktop window. Useful if you prefer to use the web UI in the browser. ``` From 95c0a261f580af0b3d2cdbac46e4b1a559394693 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Thu, 7 May 2026 14:15:09 -0700 Subject: [PATCH 1637/1701] Switch flash-attn wheels from kingbri1 to mjun0812 --- requirements/full/requirements.txt | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt index 0f3691b57f..2bf6a7a901 100644 --- a/requirements/full/requirements.txt +++ b/requirements/full/requirements.txt @@ -49,5 +49,5 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_ll https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" https://github.com/turboderp-org/exllamav3/releases/download/v0.0.30/exllamav3-0.0.30+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13" https://github.com/turboderp-org/exllamav3/releases/download/v0.0.30/exllamav3-0.0.30+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13" -https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13" -https://github.com/kingbri1/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu128torch2.9.0cxx11abiFALSE-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13" +https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.7.13/flash_attn-2.8.3+cu128torch2.9-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13" +https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.9.0/flash_attn-2.8.3+cu128torch2.9-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13" From f29e22093db9ed04ef976f1eb26d2dcf2aa3909c Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sun, 10 May 2026 21:41:05 -0700 Subject: [PATCH 1638/1701] Update README --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 693e5cd441..8ede35269f 100644 --- a/README.md +++ b/README.md @@ -83,7 +83,7 @@ textgen └── tokenizer.json ``` -These formats require the one-click installer (not the portable build). +These formats require the full installation (not the portable build). ## Installation From 9a098a9f548f85f4f5171a089a294312b866303b Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 11 May 2026 06:57:58 -0700 Subject: [PATCH 1639/1701] Update exllamav3 to 0.0.34 --- requirements/full/requirements.txt | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt index 2bf6a7a901..f71ff42c74 100644 --- a/requirements/full/requirements.txt +++ b/requirements/full/requirements.txt @@ -47,7 +47,7 @@ https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows" https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" -https://github.com/turboderp-org/exllamav3/releases/download/v0.0.30/exllamav3-0.0.30+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13" -https://github.com/turboderp-org/exllamav3/releases/download/v0.0.30/exllamav3-0.0.30+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13" +https://github.com/turboderp-org/exllamav3/releases/download/v0.0.34/exllamav3-0.0.34+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13" +https://github.com/turboderp-org/exllamav3/releases/download/v0.0.34/exllamav3-0.0.34+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13" https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.7.13/flash_attn-2.8.3+cu128torch2.9-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13" https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.9.0/flash_attn-2.8.3+cu128torch2.9-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13" From f4f556a50489ca6e56adf5d4bd06747b514fac62 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 11 May 2026 22:56:27 -0700 Subject: [PATCH 1640/1701] Add snippet support to the web_search tool (closes #7548) --- modules/web_search.py | 45 +++++++++----------------------- user_data/tools/fetch_webpage.py | 2 +- user_data/tools/web_search.py | 8 ++---- 3 files changed, 16 insertions(+), 39 deletions(-) diff --git a/modules/web_search.py b/modules/web_search.py index 5480ee49f9..8a297ce307 100644 --- a/modules/web_search.py +++ b/modules/web_search.py @@ -91,46 +91,27 @@ def perform_web_search(query, num_pages=3, max_workers=5, fetch_content=True): """Perform web search and return results, optionally with page content""" try: kwargs = {'max_results': num_pages} if num_pages is not None else {} - results = DDGS().text(query, **kwargs) + results = DDGS().text(query, backend="duckduckgo", **kwargs) - download_tasks = [(r['href'], r['title'], i) for i, r in enumerate(results)] - - search_results = [None] * len(download_tasks) # Pre-allocate to maintain order + search_results = [ + {'title': r['title'], 'url': r['href'], 'snippet': r.get('body', ''), 'content': ''} + for r in results + ] if not fetch_content: - for url, title, index in download_tasks: - search_results[index] = { - 'title': title, - 'url': url, - 'content': '' - } - return search_results - # Download pages in parallel with concurrent.futures.ThreadPoolExecutor(max_workers=max_workers) as executor: - # Submit all download tasks - future_to_task = { - executor.submit(download_web_page, task[0]): task - for task in download_tasks + future_to_index = { + executor.submit(download_web_page, r['url']): i + for i, r in enumerate(search_results) } - - # Collect results as they complete - for future in as_completed(future_to_task): - url, title, index = future_to_task[future] + for future in as_completed(future_to_index): + i = future_to_index[future] try: - content = future.result() - search_results[index] = { - 'title': title, - 'url': url, - 'content': content - } - except Exception: - search_results[index] = { - 'title': title, - 'url': url, - 'content': '' - } + search_results[i]['content'] = future.result() + except Exception as e: + logger.error(f"Error fetching {search_results[i]['url']}: {e}") return search_results diff --git a/user_data/tools/fetch_webpage.py b/user_data/tools/fetch_webpage.py index ca3e73314c..790b674ea2 100644 --- a/user_data/tools/fetch_webpage.py +++ b/user_data/tools/fetch_webpage.py @@ -23,7 +23,7 @@ def execute(arguments): if not url: return {"error": "No URL provided."} - content = download_web_page(url, include_links=True) + content = download_web_page(url) if not content or not content.strip(): return {"error": f"Failed to fetch content from {url}"} diff --git a/user_data/tools/web_search.py b/user_data/tools/web_search.py index 6c2b0f0b6f..4e0d6d2852 100644 --- a/user_data/tools/web_search.py +++ b/user_data/tools/web_search.py @@ -4,7 +4,7 @@ "type": "function", "function": { "name": "web_search", - "description": "Search the web using DuckDuckGo and return a list of result titles and URLs. Use fetch_webpage to read the contents of a specific result.", + "description": "Search the web. Returns a list of results with title, URL, and snippet (short text excerpt). The snippet often answers the query directly. Use fetch_webpage on a URL if you need the full page.", "parameters": { "type": "object", "properties": { @@ -19,9 +19,5 @@ def execute(arguments): query = arguments.get("query", "") results = perform_web_search(query, num_pages=None, fetch_content=False) - output = [] - for r in results: - if r: - output.append({"title": r["title"], "url": r["url"]}) - + output = [{"title": r["title"], "url": r["url"], "snippet": r["snippet"]} for r in results] return output if output else [{"error": "No results found."}] From b863696b47cc95f93208025324fe275676b8adb9 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Mon, 11 May 2026 22:56:50 -0700 Subject: [PATCH 1641/1701] Downgrade xformers to make exllamav3 0.0.34 work --- requirements/full/requirements.txt | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt index f71ff42c74..8beef136fe 100644 --- a/requirements/full/requirements.txt +++ b/requirements/full/requirements.txt @@ -31,7 +31,7 @@ transformers==5.6.* triton-windows==3.5.1.post24; platform_system == "Windows" tqdm wandb -xformers==0.0.33.post2 +xformers==0.0.33 # Gradio https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl From 79630e9bb1323229d78dc5b98e10ed8e22ef1935 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 12 May 2026 16:47:17 -0700 Subject: [PATCH 1642/1701] Remove `backend="duckduckgo"` from ddgs (https://github.com/oobabooga/textgen/issues/7548#issuecomment-4433225429) --- modules/web_search.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/web_search.py b/modules/web_search.py index 8a297ce307..b47b01673d 100644 --- a/modules/web_search.py +++ b/modules/web_search.py @@ -91,7 +91,7 @@ def perform_web_search(query, num_pages=3, max_workers=5, fetch_content=True): """Perform web search and return results, optionally with page content""" try: kwargs = {'max_results': num_pages} if num_pages is not None else {} - results = DDGS().text(query, backend="duckduckgo", **kwargs) + results = DDGS().text(query, **kwargs) search_results = [ {'title': r['title'], 'url': r['href'], 'snippet': r.get('body', ''), 'content': ''} From 13f5d378e570884bfda63ca8d9c50cce5a239f05 Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Tue, 12 May 2026 17:50:36 -0700 Subject: [PATCH 1643/1701] Make web_search tool call results pretty --- css/main.css | 50 +++++++++++++++++++++++++++++++++++++++ modules/html_generator.py | 45 +++++++++++++++++++++++++++++++++-- 2 files changed, 93 insertions(+), 2 deletions(-) diff --git a/css/main.css b/css/main.css index e6690d67f3..f87abbb89a 100644 --- a/css/main.css +++ b/css/main.css @@ -1525,6 +1525,56 @@ audio { font-weight: 500; } +.tool-call-spinner { + display: inline-block; + width: 12px; + height: 12px; + margin-left: 8px; + border: 2px solid rgb(0 0 0 / 15%); + border-top-color: rgb(0 0 0 / 55%); + border-radius: 50%; + animation: tool-call-spin 0.8s linear infinite; + flex-shrink: 0; +} + +@keyframes tool-call-spin { + to { transform: rotate(360deg); } +} + +.dark .tool-call-spinner { + border-color: rgb(255 255 255 / 15%); + border-top-color: rgb(255 255 255 / 65%); +} + +.web-search-results { + display: flex; + flex-direction: column; + gap: var(--space-2); +} + +.web-search-result { + padding: var(--space-3) var(--space-4); + background: var(--border-soft); + border-radius: var(--radius-sm); +} + +.web-search-title { + display: block; + font-weight: 500; + margin-bottom: var(--space-1); + text-decoration: none; +} + +.web-search-title:hover { + text-decoration: underline; +} + +.web-search-snippet { + font-size: 0.9em; + color: var(--text-muted); + line-height: 1.4; +} + .thinking-content { padding: 12px 16px; border-top: 1px solid rgb(0 0 0 / 7%); diff --git a/modules/html_generator.py b/modules/html_generator.py index e3bafda4ab..c0d2c5aa14 100644 --- a/modules/html_generator.py +++ b/modules/html_generator.py @@ -1,6 +1,7 @@ import datetime import functools import html +import json import os import re import time @@ -125,17 +126,44 @@ def extract_thinking_block(string): TOOL_APPROVAL_PENDING = '\x00approval_pending' +def _render_web_search_body(body): + """Render a web_search tool result body as structured cards. Returns None + if the body doesn't look like a valid web_search result list.""" + try: + results = json.loads(body) + except (json.JSONDecodeError, TypeError): + return None + if not isinstance(results, list) or not results: + return None + if not all(isinstance(r, dict) and 'title' in r and 'url' in r for r in results): + return None + + cards = [] + for r in results: + title = html.escape(r['title']) + url = html.escape(r['url']) + snippet = html.escape(r.get('snippet', '')) + cards.append( + f'
    ' + f'{title}' + f'
    {snippet}
    ' + f'
    ' + ) + return ''.join(cards) + + def build_tool_call_block(header, body, message_id, index): """Build HTML for a tool call accordion block.""" block_id = f"tool-call-{message_id}-{index}" if body == '...': - # Pending placeholder — no expandable body, just title with ellipsis + # Pending placeholder — tool call is in flight, body filled in later return f'''
    {tool_svg_small} - {html.escape(header)} ... + {html.escape(header)} +
    ''' @@ -155,6 +183,19 @@ def build_tool_call_block(header, body, message_id, index): ''' + if header.startswith('web_search('): + rendered = _render_web_search_body(body) + if rendered is not None: + return f''' +
    + + {tool_svg_small} + {html.escape(header)} + +
    {rendered}
    +
    + ''' + # Build a plain 
     directly to avoid highlight.js auto-detection
     escaped_body = html.escape(body)
     return f'''

From 66f01d6f208247ee47386e71f04d51116339fba4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 13 May 2026 09:50:51 -0700
Subject: [PATCH 1644/1701] Electron: Add right-click context menu for copying
 text

---
 desktop/main.js | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/desktop/main.js b/desktop/main.js
index e2a05f901a..f517cd167e 100644
--- a/desktop/main.js
+++ b/desktop/main.js
@@ -1,4 +1,4 @@
-const { app, BrowserWindow, screen } = require("electron");
+const { app, BrowserWindow, Menu, screen } = require("electron");
 const { spawn } = require("child_process");
 const path = require("path");
 const fs = require("fs");
@@ -114,6 +114,14 @@ function createWindow(port) {
     webPreferences: { nodeIntegration: false, contextIsolation: true, spellcheck: false },
   });
   if (state && state.maximized) mainWindow.maximize();
+  mainWindow.webContents.on("context-menu", (_, params) => {
+    const tmpl = [];
+    if (params.editFlags.canCut) tmpl.push({ role: "cut" });
+    if (params.editFlags.canCopy) tmpl.push({ role: "copy" });
+    if (params.editFlags.canPaste) tmpl.push({ role: "paste" });
+    if (params.editFlags.canSelectAll) tmpl.push({ type: "separator" }, { role: "selectAll" });
+    if (tmpl.length) Menu.buildFromTemplate(tmpl).popup({ window: mainWindow });
+  });
   mainWindow.on("page-title-updated", (e) => e.preventDefault());
   mainWindow.webContents.on("will-prevent-unload", (e) => e.preventDefault());
   mainWindow.on("close", saveState);

From 2c254cb80d89eccb241e78d0b2bb9aacdfd19ee1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 13 May 2026 13:17:48 -0700
Subject: [PATCH 1645/1701] docs: Small API example change

---
 docs/12 - OpenAI API.md | 1 -
 1 file changed, 1 deletion(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index 866b6432d1..81b3a8f961 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -226,7 +226,6 @@ curl -k http://127.0.0.1:5000/v1/internal/model/load \
     "model_name": "Qwen_Qwen3-0.6B-Q4_K_M.gguf",
     "args": {
       "ctx_size": 32768,
-      "flash_attn": true,
       "cache_type": "q8_0"
     }
   }'

From 47fdee9cb108bd05a7f7d79424399cf580b1ba8f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 13 May 2026 20:34:07 -0700
Subject: [PATCH 1646/1701] Electron: Add a folder picker for the models
 directory

---
 .github/workflows/build-portable-release.yml |  2 +-
 css/main.css                                 |  4 ++++
 desktop/main.js                              | 15 +++++++++++++--
 modules/shared.py                            |  1 +
 modules/ui.py                                |  6 ++++++
 modules/ui_session.py                        | 19 +++++++++++++++++++
 server.py                                    |  8 ++++++++
 7 files changed, 52 insertions(+), 3 deletions(-)

diff --git a/.github/workflows/build-portable-release.yml b/.github/workflows/build-portable-release.yml
index dcd61ae3ce..a1c5808922 100644
--- a/.github/workflows/build-portable-release.yml
+++ b/.github/workflows/build-portable-release.yml
@@ -207,7 +207,7 @@ jobs:
                 chmod +x textgen
             fi
 
-            mv desktop/main.js desktop/package.json .
+            mv desktop/main.js desktop/preload.js desktop/package.json .
             rm -rf desktop
 
             # 5c. Restructure: textgen-VERSION/{textgen, user_data/, app/}
diff --git a/css/main.css b/css/main.css
index f87abbb89a..0bd7fe3ace 100644
--- a/css/main.css
+++ b/css/main.css
@@ -145,6 +145,10 @@ div.svelte-iyf88w {
     padding: 0 !important;
 }
 
+.slim-textbox {
+    padding: 0 !important;
+}
+
 #download-label, #upload-label {
     min-height: 0
 }
diff --git a/desktop/main.js b/desktop/main.js
index f517cd167e..01d0c0a1c7 100644
--- a/desktop/main.js
+++ b/desktop/main.js
@@ -1,4 +1,4 @@
-const { app, BrowserWindow, Menu, screen } = require("electron");
+const { app, BrowserWindow, Menu, dialog, ipcMain, screen } = require("electron");
 const { spawn } = require("child_process");
 const path = require("path");
 const fs = require("fs");
@@ -111,7 +111,12 @@ function createWindow(port) {
     ...bounds,
     title: TITLE,
     autoHideMenuBar: true,
-    webPreferences: { nodeIntegration: false, contextIsolation: true, spellcheck: false },
+    webPreferences: {
+      preload: path.join(__dirname, "preload.js"),
+      nodeIntegration: false,
+      contextIsolation: true,
+      spellcheck: false,
+    },
   });
   if (state && state.maximized) mainWindow.maximize();
   mainWindow.webContents.on("context-menu", (_, params) => {
@@ -147,6 +152,11 @@ async function waitForPortAndOpen(port) {
   }, 500);
 }
 
+ipcMain.handle("pick-directory", async () => {
+  const result = await dialog.showOpenDialog(mainWindow, { properties: ["openDirectory"] });
+  return result.canceled ? null : result.filePaths[0];
+});
+
 app.whenReady().then(() => {
   serverProcess = spawn(python, ["server.py", "--portable", "--api", ...userArgs], {
     cwd: baseDir,
@@ -159,6 +169,7 @@ app.whenReady().then(() => {
       PYTHONUNBUFFERED: "1",
       FORCE_COLOR: "1",
       TERM: "xterm-256color",
+      TEXTGEN_ELECTRON: "1",
     },
   });
   if (!isWin) serverProcess.unref();
diff --git a/modules/shared.py b/modules/shared.py
index d9d3719489..9058a99b15 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -40,6 +40,7 @@
 gradio = {}
 persistent_interface_state = {}
 need_restart = False
+is_electron = os.environ.get('TEXTGEN_ELECTRON') == '1'
 
 # Parser copied from https://github.com/vladmandic/automatic
 parser = argparse.ArgumentParser(description="TextGen", conflict_handler='resolve', add_help=True, formatter_class=lambda prog: argparse.HelpFormatter(prog, max_help_position=55, indent_increment=2, width=200))
diff --git a/modules/ui.py b/modules/ui.py
index 180d171b49..835bb0c655 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -248,6 +248,9 @@ def list_interface_input_elements():
         'include_past_attachments',
     ]
 
+    if shared.is_electron:
+        elements += ['model_dir']
+
     if not shared.args.portable:
         # Image generation elements
         elements += [
@@ -515,6 +518,9 @@ def setup_auto_save():
 
     ]
 
+    if shared.is_electron:
+        change_elements += ['model_dir']
+
     if not shared.args.portable:
         # Image generation tab (ui_image_generation.py)
         change_elements += [
diff --git a/modules/ui_session.py b/modules/ui_session.py
index 3f2c8a7bec..a9bd1141af 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -1,3 +1,5 @@
+from pathlib import Path
+
 import gradio as gr
 
 from modules import shared, ui, utils
@@ -11,6 +13,10 @@ def create_ui():
             with gr.Column():
                 gr.Markdown("## Settings")
                 shared.gradio['toggle_dark_mode'] = gr.Button('Toggle light/dark theme 💡', elem_classes='refresh-button')
+                if shared.is_electron:
+                    with gr.Row():
+                        shared.gradio['model_dir'] = gr.Textbox(label='Models directory', value=shared.settings['model_dir'], scale=4, elem_classes='slim-textbox')
+                        shared.gradio['model_dir_browse'] = gr.Button('Browse', elem_classes=['refresh-button', 'refresh-button-medium'])
                 shared.gradio['show_two_notebook_columns'] = gr.Checkbox(label='Show two columns in the Notebook tab', value=shared.settings['show_two_notebook_columns'])
                 shared.gradio['paste_to_attachment'] = gr.Checkbox(label='Turn long pasted text into attachments in the Chat tab', value=shared.settings['paste_to_attachment'], elem_id='paste_to_attachment')
                 shared.gradio['include_past_attachments'] = gr.Checkbox(label='Include attachments/search results from previous messages in the chat prompt', value=shared.settings['include_past_attachments'])
@@ -36,6 +42,13 @@ def create_ui():
             lambda x: 'dark' if x == 'light' else 'light', gradio('theme_state'), gradio('theme_state')).then(
             None, None, None, js=f'() => {{{ui.dark_theme_js}; toggleDarkMode(); localStorage.setItem("theme", document.body.classList.contains("dark") ? "dark" : "light")}}')
 
+        if shared.is_electron:
+            shared.gradio['model_dir_browse'].click(
+                None, gradio('model_dir'), gradio('model_dir'),
+                js='async (current) => { const p = await window.electronAPI.pickDirectory(); return p === null ? current : p; }')
+
+            shared.gradio['model_dir'].change(apply_model_dir, gradio('model_dir'), None, show_progress=False)
+
         shared.gradio['show_two_notebook_columns'].change(
             handle_default_to_notebook_change,
             gradio('show_two_notebook_columns', 'textbox-default', 'output_textbox', 'prompt_menu-default', 'textbox-notebook', 'prompt_menu-notebook'),
@@ -86,6 +99,12 @@ def handle_default_to_notebook_change(show_two_columns, default_input, default_o
         ]
 
 
+def apply_model_dir(value):
+    shared.args.model_dir = value
+    if Path(value).is_dir():
+        shared.user_config = shared.load_user_config()
+
+
 def set_interface_arguments(extensions, bool_active):
     shared.args.extensions = extensions
 
diff --git a/server.py b/server.py
index 6ad08c64fc..1591a3abf9 100644
--- a/server.py
+++ b/server.py
@@ -261,6 +261,10 @@ def create_interface():
     if has_mcp_config():
         logger.warning(f"MCP stdio servers will be loaded from \"{shared.user_data_dir / 'mcp.json'}\"")
 
+    if shared.is_electron:
+        shared.settings['model_dir'] = shared.args.model_dir
+        shared.default_settings['model_dir'] = shared.args.model_dir
+
     if settings_file is not None:
         logger.info(f"Loading settings from \"{settings_file}\"")
         with open(settings_file, 'r', encoding='utf-8') as f:
@@ -269,6 +273,10 @@ def create_interface():
         if new_settings:
             shared.settings.update(new_settings)
 
+    if shared.is_electron:
+        shared.args.model_dir = shared.settings['model_dir']
+        shared.user_config = shared.load_user_config()
+
     # Apply CLI overrides for image model settings (CLI flags take precedence over saved settings)
     shared.apply_image_model_cli_overrides()
 

From 6a1a9591409f5265d6864e2c373d349ca5406d08 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 13 May 2026 20:37:07 -0700
Subject: [PATCH 1647/1701] Add missing file

---
 .github/workflows/build-portable-release-cuda.yml    | 2 +-
 .github/workflows/build-portable-release-ik-cuda.yml | 2 +-
 .github/workflows/build-portable-release-ik.yml      | 2 +-
 .github/workflows/build-portable-release-rocm.yml    | 2 +-
 .github/workflows/build-portable-release-vulkan.yml  | 2 +-
 desktop/preload.js                                   | 5 +++++
 6 files changed, 10 insertions(+), 5 deletions(-)
 create mode 100644 desktop/preload.js

diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index 7c4d95e2b1..12e2b16f7f 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -186,7 +186,7 @@ jobs:
                 chmod +x textgen
             fi
 
-            mv desktop/main.js desktop/package.json .
+            mv desktop/main.js desktop/preload.js desktop/package.json .
             rm -rf desktop
 
             # 5c. Restructure: textgen-VERSION/{textgen, user_data/, app/}
diff --git a/.github/workflows/build-portable-release-ik-cuda.yml b/.github/workflows/build-portable-release-ik-cuda.yml
index 2613a5c234..cce1fc61c4 100644
--- a/.github/workflows/build-portable-release-ik-cuda.yml
+++ b/.github/workflows/build-portable-release-ik-cuda.yml
@@ -186,7 +186,7 @@ jobs:
                 chmod +x textgen
             fi
 
-            mv desktop/main.js desktop/package.json .
+            mv desktop/main.js desktop/preload.js desktop/package.json .
             rm -rf desktop
 
             # 5c. Restructure: textgen-VERSION/{textgen, user_data/, app/}
diff --git a/.github/workflows/build-portable-release-ik.yml b/.github/workflows/build-portable-release-ik.yml
index 66f79711f1..75e905c0ce 100644
--- a/.github/workflows/build-portable-release-ik.yml
+++ b/.github/workflows/build-portable-release-ik.yml
@@ -181,7 +181,7 @@ jobs:
                 chmod +x textgen
             fi
 
-            mv desktop/main.js desktop/package.json .
+            mv desktop/main.js desktop/preload.js desktop/package.json .
             rm -rf desktop
 
             # 5c. Restructure: textgen-VERSION/{textgen, user_data/, app/}
diff --git a/.github/workflows/build-portable-release-rocm.yml b/.github/workflows/build-portable-release-rocm.yml
index 6a6a5a0218..189d0415d8 100644
--- a/.github/workflows/build-portable-release-rocm.yml
+++ b/.github/workflows/build-portable-release-rocm.yml
@@ -181,7 +181,7 @@ jobs:
                 chmod +x textgen
             fi
 
-            mv desktop/main.js desktop/package.json .
+            mv desktop/main.js desktop/preload.js desktop/package.json .
             rm -rf desktop
 
             # 5c. Restructure: textgen-VERSION/{textgen, user_data/, app/}
diff --git a/.github/workflows/build-portable-release-vulkan.yml b/.github/workflows/build-portable-release-vulkan.yml
index 092aeb5700..2ba49f228d 100644
--- a/.github/workflows/build-portable-release-vulkan.yml
+++ b/.github/workflows/build-portable-release-vulkan.yml
@@ -181,7 +181,7 @@ jobs:
                 chmod +x textgen
             fi
 
-            mv desktop/main.js desktop/package.json .
+            mv desktop/main.js desktop/preload.js desktop/package.json .
             rm -rf desktop
 
             # 5c. Restructure: textgen-VERSION/{textgen, user_data/, app/}
diff --git a/desktop/preload.js b/desktop/preload.js
new file mode 100644
index 0000000000..709ade730d
--- /dev/null
+++ b/desktop/preload.js
@@ -0,0 +1,5 @@
+const { contextBridge, ipcRenderer } = require("electron");
+
+contextBridge.exposeInMainWorld("electronAPI", {
+  pickDirectory: () => ipcRenderer.invoke("pick-directory"),
+});

From 66c3c49c54332fd3a007bb2af92a1c8789f30fab Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 13 May 2026 21:15:43 -0700
Subject: [PATCH 1648/1701] Treat negative ctx-size as auto

---
 modules/llama_cpp_server.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 0ce15828b5..5e92b8f5c8 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -433,6 +433,9 @@ def _start_server(self):
             "--flash-attn", "on",
         ]
 
+        if shared.args.ctx_size < 0:
+            shared.args.ctx_size = 0
+
         if shared.args.ctx_size > 0:
             cmd += ["--ctx-size", str(shared.args.ctx_size)]
         elif shared.args.gpu_layers >= 0:

From a711849fa72084f12a791cffc974ac2088a4a465 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 14 May 2026 08:01:25 -0700
Subject: [PATCH 1649/1701] Detect mmproj files in the models folder
 (simplifies LM Studio migration)

---
 modules/llama_cpp_server.py |  6 +++++-
 modules/ui_model_menu.py    |  2 +-
 modules/utils.py            | 27 ++++++++++++++++++---------
 3 files changed, 24 insertions(+), 11 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 5e92b8f5c8..525b858415 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -474,7 +474,11 @@ def _start_server(self):
         if shared.args.mmproj not in [None, 'None']:
             path = Path(shared.args.mmproj)
             if not path.exists():
-                path = shared.user_data_dir / 'mmproj' / shared.args.mmproj
+                alt = shared.user_data_dir / 'mmproj' / shared.args.mmproj
+                if alt.exists():
+                    path = alt
+                else:
+                    path = Path(shared.args.model_dir) / shared.args.mmproj
 
             if path.exists():
                 cmd += ["--mmproj", str(path)]
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index aa27adcf7b..77b5c18fc3 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -68,7 +68,7 @@ def create_ui():
                             # Multimodal
                             with gr.Accordion("Multimodal (vision)", open=False) as shared.gradio['mmproj_accordion']:
                                 with gr.Row():
-                                    shared.gradio['mmproj'] = gr.Dropdown(label="mmproj file", choices=utils.get_available_mmproj(), value=lambda: shared.args.mmproj or 'None', elem_classes='slim-dropdown', info=f'Select a file that matches your model. Must be placed in {shared.user_data_dir}/mmproj/', interactive=not mu)
+                                    shared.gradio['mmproj'] = gr.Dropdown(label="mmproj file", choices=utils.get_available_mmproj(), value=lambda: shared.args.mmproj or 'None', elem_classes='slim-dropdown', info=f'Select a file that matches your model. Lists files placed in {shared.user_data_dir}/mmproj/, plus any mmproj-*.gguf files found in your main models folder.', interactive=not mu)
                                     ui.create_refresh_button(shared.gradio['mmproj'], lambda: None, lambda: {'choices': utils.get_available_mmproj()}, 'refresh-button', interactive=not mu)
 
                             # Speculative decoding
diff --git a/modules/utils.py b/modules/utils.py
index 7bbd903791..67d42c1bf9 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -194,7 +194,8 @@ def get_available_ggufs():
 
     for dirpath, _, files in os.walk(model_dir, followlinks=True):
         for file in files:
-            if file.lower().endswith(".gguf"):
+            lower = file.lower()
+            if lower.endswith(".gguf") and not lower.startswith("mmproj"):
                 model_path = Path(dirpath) / file
                 rel_path = model_path.relative_to(model_dir)
                 model_list.append(str(rel_path))
@@ -203,16 +204,24 @@ def get_available_ggufs():
 
 
 def get_available_mmproj():
-    mmproj_dir = shared.user_data_dir / 'mmproj'
-    if not mmproj_dir.exists():
-        return ['None']
-
     mmproj_files = []
-    for item in mmproj_dir.iterdir():
-        if item.is_file() and item.suffix.lower() in ('.gguf', '.bin'):
-            mmproj_files.append(item.name)
 
-    return ['None'] + sorted(mmproj_files, key=natural_keys)
+    mmproj_dir = shared.user_data_dir / 'mmproj'
+    if mmproj_dir.exists():
+        for item in mmproj_dir.iterdir():
+            if item.is_file() and item.suffix.lower() in ('.gguf', '.bin'):
+                mmproj_files.append(item.name)
+
+    model_dir = Path(shared.args.model_dir)
+    if model_dir.exists():
+        for dirpath, _, files in os.walk(model_dir, followlinks=True):
+            for file in files:
+                lower = file.lower()
+                if lower.startswith('mmproj') and lower.endswith(('.gguf', '.bin')):
+                    rel_path = str((Path(dirpath) / file).relative_to(model_dir))
+                    mmproj_files.append(rel_path)
+
+    return ['None'] + sorted(set(mmproj_files), key=natural_keys)
 
 
 def get_available_presets():

From 630141912b95960da0e791a5ced2867f4c89fd72 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 14 May 2026 08:36:49 -0700
Subject: [PATCH 1650/1701] Fix streaming output leaking across chats (closes
 #7555)

---
 modules/chat.py | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)

diff --git a/modules/chat.py b/modules/chat.py
index 942b7bfdd6..871ff0b7e4 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -50,6 +50,10 @@
 _tool_approvals = {}
 _tool_approvals_lock = threading.Lock()
 
+# Currently-viewed chat id (single-user mode only). Used to skip streaming UI
+# updates when the user switches to a different chat mid-stream.
+viewing_unique_id = None
+
 
 def request_tool_approval(session_key, tool_name):
     """Block until the user approves/rejects a tool call.
@@ -1320,6 +1324,10 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
     using metadata['assistant_N']['tool_sequence'].
     '''
 
+    global viewing_unique_id
+    if not shared.args.multi_user:
+        viewing_unique_id = state['unique_id']
+
     if not character_is_loaded(state):
         return
 
@@ -1401,7 +1409,8 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
             if visible_prefix:
                 history['visible'][-1][1] = '\n\n'.join(visible_prefix + [_original_visible])
 
-            yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'], last_message_only=(i > 0)), history
+            if shared.args.multi_user or viewing_unique_id is None or viewing_unique_id == state['unique_id']:
+                yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'], last_message_only=(i > 0)), history
 
             if visible_prefix:
                 history['visible'][-1][1] = _original_visible
@@ -2427,6 +2436,10 @@ def handle_remove_last_click(state):
 
 
 def handle_unique_id_select(state):
+    global viewing_unique_id
+    if not shared.args.multi_user:
+        viewing_unique_id = state['unique_id']
+
     history = load_history(state['unique_id'], state['character_menu'], state['mode'])
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
 

From 58774c8cb1c8c259503be6f3cf8b9c8d5b1f2b4a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 14 May 2026 11:04:37 -0700
Subject: [PATCH 1651/1701] Also fix streaming leak across all other chat
 actions

---
 modules/chat.py | 31 +++++++++++++++++++++++++++++--
 1 file changed, 29 insertions(+), 2 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 871ff0b7e4..bb19d4399a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -1631,6 +1631,9 @@ def _cancel_remaining(from_idx):
 
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
 
+    if not shared.args.multi_user and viewing_unique_id == state['unique_id']:
+        viewing_unique_id = None
+
 
 def remove_last_message(history):
     if 'metadata' not in history:
@@ -2453,6 +2456,7 @@ def handle_unique_id_select(state):
 
 def handle_start_new_chat_click(state):
     import gradio as gr
+    global viewing_unique_id
     history = start_new_chat(state)
     histories = find_all_histories_with_first_prompts(state)
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
@@ -2461,6 +2465,8 @@ def handle_start_new_chat_click(state):
 
     if len(histories) > 0:
         past_chats_update = gr.update(choices=histories, value=histories[0][1])
+        if not shared.args.multi_user:
+            viewing_unique_id = histories[0][1]
     else:
         past_chats_update = gr.update(choices=histories)
 
@@ -2469,6 +2475,7 @@ def handle_start_new_chat_click(state):
 
 def handle_start_incognito_chat_click(state):
     import gradio as gr
+    global viewing_unique_id
     unique_id = 'incognito-' + datetime.now().strftime('%Y%m%d-%H-%M-%S')
     history = start_new_chat(state, unique_id=unique_id)
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
@@ -2477,11 +2484,14 @@ def handle_start_incognito_chat_click(state):
 
     histories = find_all_histories_with_first_prompts(state)
     past_chats_update = gr.update(choices=histories, value=unique_id)
+    if not shared.args.multi_user:
+        viewing_unique_id = unique_id
 
     return [history, html, past_chats_update]
 
 
 def handle_delete_chat_confirm_click(state):
+    global viewing_unique_id
     filtered_histories = find_all_histories_with_first_prompts(state)
     filtered_ids = [h[1] for h in filtered_histories]
 
@@ -2497,11 +2507,15 @@ def handle_delete_chat_confirm_click(state):
 
     convert_to_markdown.cache_clear()
 
+    if not shared.args.multi_user:
+        viewing_unique_id = unique_id
+
     return [history, html, unique_id]
 
 
 def handle_branch_chat_click(state):
     import gradio as gr
+    global viewing_unique_id
     branch_from_index = state['branch_index']
     if branch_from_index == -1:
         history = state['history']
@@ -2523,6 +2537,8 @@ def handle_branch_chat_click(state):
     convert_to_markdown.cache_clear()
 
     past_chats_update = gr.update(choices=histories, value=new_unique_id)
+    if not shared.args.multi_user:
+        viewing_unique_id = new_unique_id
 
     return [history, html, past_chats_update, -1]
 
@@ -2660,6 +2676,7 @@ def handle_search_chat_change(state):
 
 def handle_upload_chat_history(load_chat_history, state):
     import gradio as gr
+    global viewing_unique_id
     history = start_new_chat(state)
     history = load_history_json(load_chat_history, history)
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
@@ -2671,6 +2688,8 @@ def handle_upload_chat_history(load_chat_history, state):
 
     if len(histories) > 0:
         past_chats_update = gr.update(choices=histories, value=histories[0][1])
+        if not shared.args.multi_user:
+            viewing_unique_id = histories[0][1]
     else:
         past_chats_update = gr.update(choices=histories)
 
@@ -2683,6 +2702,7 @@ def handle_upload_chat_history(load_chat_history, state):
 
 def handle_character_menu_change(state):
     import gradio as gr
+    global viewing_unique_id
     name1, name2, picture, greeting, context = load_character(state['character_menu'], state['name1'], state['name2'])
 
     state['name1'] = name1
@@ -2698,7 +2718,10 @@ def handle_character_menu_change(state):
     convert_to_markdown.cache_clear()
 
     if len(histories) > 0:
-        past_chats_update = gr.update(choices=histories, value=loaded_unique_id or histories[0][1])
+        new_id = loaded_unique_id or histories[0][1]
+        past_chats_update = gr.update(choices=histories, value=new_id)
+        if not shared.args.multi_user:
+            viewing_unique_id = new_id
     else:
         past_chats_update = gr.update(choices=histories)
 
@@ -2734,6 +2757,7 @@ def handle_character_picture_change(picture_path):
 
 def handle_mode_change(state):
     import gradio as gr
+    global viewing_unique_id
     history, loaded_unique_id = load_latest_history(state)
     histories = find_all_histories_with_first_prompts(state)
 
@@ -2746,7 +2770,10 @@ def handle_mode_change(state):
     convert_to_markdown.cache_clear()
 
     if len(histories) > 0:
-        past_chats_update = gr.update(choices=histories, value=loaded_unique_id or histories[0][1])
+        new_id = loaded_unique_id or histories[0][1]
+        past_chats_update = gr.update(choices=histories, value=new_id)
+        if not shared.args.multi_user:
+            viewing_unique_id = new_id
     else:
         past_chats_update = gr.update(choices=histories)
 

From 1e8136dc2a6b0feb05afb91cd1e8d96b873378b0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 14 May 2026 11:21:18 -0700
Subject: [PATCH 1652/1701] Small simplifications

---
 modules/chat.py | 46 +++++++++++++++++-----------------------------
 1 file changed, 17 insertions(+), 29 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index bb19d4399a..9a86c8cdb6 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -55,6 +55,12 @@
 viewing_unique_id = None
 
 
+def set_viewing_unique_id(unique_id):
+    global viewing_unique_id
+    if not shared.args.multi_user:
+        viewing_unique_id = unique_id
+
+
 def request_tool_approval(session_key, tool_name):
     """Block until the user approves/rejects a tool call.
     Returns 'approve'|'always'|'reject', or None if generation was stopped
@@ -1324,9 +1330,7 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
     using metadata['assistant_N']['tool_sequence'].
     '''
 
-    global viewing_unique_id
-    if not shared.args.multi_user:
-        viewing_unique_id = state['unique_id']
+    set_viewing_unique_id(state['unique_id'])
 
     if not character_is_loaded(state):
         return
@@ -1631,8 +1635,8 @@ def _cancel_remaining(from_idx):
 
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
 
-    if not shared.args.multi_user and viewing_unique_id == state['unique_id']:
-        viewing_unique_id = None
+    if viewing_unique_id == state['unique_id']:
+        set_viewing_unique_id(None)
 
 
 def remove_last_message(history):
@@ -2439,9 +2443,7 @@ def handle_remove_last_click(state):
 
 
 def handle_unique_id_select(state):
-    global viewing_unique_id
-    if not shared.args.multi_user:
-        viewing_unique_id = state['unique_id']
+    set_viewing_unique_id(state['unique_id'])
 
     history = load_history(state['unique_id'], state['character_menu'], state['mode'])
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
@@ -2456,7 +2458,6 @@ def handle_unique_id_select(state):
 
 def handle_start_new_chat_click(state):
     import gradio as gr
-    global viewing_unique_id
     history = start_new_chat(state)
     histories = find_all_histories_with_first_prompts(state)
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
@@ -2465,8 +2466,7 @@ def handle_start_new_chat_click(state):
 
     if len(histories) > 0:
         past_chats_update = gr.update(choices=histories, value=histories[0][1])
-        if not shared.args.multi_user:
-            viewing_unique_id = histories[0][1]
+        set_viewing_unique_id(histories[0][1])
     else:
         past_chats_update = gr.update(choices=histories)
 
@@ -2475,7 +2475,6 @@ def handle_start_new_chat_click(state):
 
 def handle_start_incognito_chat_click(state):
     import gradio as gr
-    global viewing_unique_id
     unique_id = 'incognito-' + datetime.now().strftime('%Y%m%d-%H-%M-%S')
     history = start_new_chat(state, unique_id=unique_id)
     html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
@@ -2484,14 +2483,12 @@ def handle_start_incognito_chat_click(state):
 
     histories = find_all_histories_with_first_prompts(state)
     past_chats_update = gr.update(choices=histories, value=unique_id)
-    if not shared.args.multi_user:
-        viewing_unique_id = unique_id
+    set_viewing_unique_id(unique_id)
 
     return [history, html, past_chats_update]
 
 
 def handle_delete_chat_confirm_click(state):
-    global viewing_unique_id
     filtered_histories = find_all_histories_with_first_prompts(state)
     filtered_ids = [h[1] for h in filtered_histories]
 
@@ -2507,15 +2504,13 @@ def handle_delete_chat_confirm_click(state):
 
     convert_to_markdown.cache_clear()
 
-    if not shared.args.multi_user:
-        viewing_unique_id = unique_id
+    set_viewing_unique_id(unique_id)
 
     return [history, html, unique_id]
 
 
 def handle_branch_chat_click(state):
     import gradio as gr
-    global viewing_unique_id
     branch_from_index = state['branch_index']
     if branch_from_index == -1:
         history = state['history']
@@ -2537,8 +2532,7 @@ def handle_branch_chat_click(state):
     convert_to_markdown.cache_clear()
 
     past_chats_update = gr.update(choices=histories, value=new_unique_id)
-    if not shared.args.multi_user:
-        viewing_unique_id = new_unique_id
+    set_viewing_unique_id(new_unique_id)
 
     return [history, html, past_chats_update, -1]
 
@@ -2676,7 +2670,6 @@ def handle_search_chat_change(state):
 
 def handle_upload_chat_history(load_chat_history, state):
     import gradio as gr
-    global viewing_unique_id
     history = start_new_chat(state)
     history = load_history_json(load_chat_history, history)
     save_history(history, state['unique_id'], state['character_menu'], state['mode'])
@@ -2688,8 +2681,7 @@ def handle_upload_chat_history(load_chat_history, state):
 
     if len(histories) > 0:
         past_chats_update = gr.update(choices=histories, value=histories[0][1])
-        if not shared.args.multi_user:
-            viewing_unique_id = histories[0][1]
+        set_viewing_unique_id(histories[0][1])
     else:
         past_chats_update = gr.update(choices=histories)
 
@@ -2702,7 +2694,6 @@ def handle_upload_chat_history(load_chat_history, state):
 
 def handle_character_menu_change(state):
     import gradio as gr
-    global viewing_unique_id
     name1, name2, picture, greeting, context = load_character(state['character_menu'], state['name1'], state['name2'])
 
     state['name1'] = name1
@@ -2720,8 +2711,7 @@ def handle_character_menu_change(state):
     if len(histories) > 0:
         new_id = loaded_unique_id or histories[0][1]
         past_chats_update = gr.update(choices=histories, value=new_id)
-        if not shared.args.multi_user:
-            viewing_unique_id = new_id
+        set_viewing_unique_id(new_id)
     else:
         past_chats_update = gr.update(choices=histories)
 
@@ -2757,7 +2747,6 @@ def handle_character_picture_change(picture_path):
 
 def handle_mode_change(state):
     import gradio as gr
-    global viewing_unique_id
     history, loaded_unique_id = load_latest_history(state)
     histories = find_all_histories_with_first_prompts(state)
 
@@ -2772,8 +2761,7 @@ def handle_mode_change(state):
     if len(histories) > 0:
         new_id = loaded_unique_id or histories[0][1]
         past_chats_update = gr.update(choices=histories, value=new_id)
-        if not shared.args.multi_user:
-            viewing_unique_id = new_id
+        set_viewing_unique_id(new_id)
     else:
         past_chats_update = gr.update(choices=histories)
 

From 646f10dd22171cb05710af2480e18757e11f5639 Mon Sep 17 00:00:00 2001
From: Highlander 
Date: Fri, 15 May 2026 02:34:36 +0800
Subject: [PATCH 1653/1701] fix(win): set PYTHONUTF8 for non-ASCII locale
 Windows compatibility (#7560)

---
 desktop/textgen.bat | 1 +
 1 file changed, 1 insertion(+)

diff --git a/desktop/textgen.bat b/desktop/textgen.bat
index d7a170947b..62db7d60ba 100644
--- a/desktop/textgen.bat
+++ b/desktop/textgen.bat
@@ -1,4 +1,5 @@
 @echo off
+set PYTHONUTF8=1
 set "APP=%~dp0__APP__"
 for %%a in (%*) do (
     if /i "%%~a"=="--help" goto :help

From 0f88365b0e741550f22c3be553aa1c7f11ce6232 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Thu, 14 May 2026 19:59:56 -0700
Subject: [PATCH 1654/1701] Rename "Send dummy message/reply" to "Insert
 user/assistant message"

---
 docs/01 - Chat Tab.md | 4 ++--
 modules/ui_chat.py    | 8 ++++----
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/docs/01 - Chat Tab.md b/docs/01 - Chat Tab.md
index 5425e63acc..d2b221659a 100644
--- a/docs/01 - Chat Tab.md	
+++ b/docs/01 - Chat Tab.md	
@@ -13,8 +13,8 @@ The hover menu (☰) that appears over the chat area contains:
 * **Continue**: makes the model attempt to continue the existing reply. In some cases, the model may simply end the existing turn immediately without generating anything new, but in other cases, it may generate a longer reply.
 * **Remove last reply**: removes the last input/output pair from the history and sends your last message back into the input field.
 * **Impersonate**: makes the model generate a new message on your behalf in the input field, taking into consideration the existing chat history.
-* **Send dummy message**: adds a new message to the chat history without causing the model to generate a reply.
-* **Send dummy reply**: adds a new reply to the chat history as if the model had generated this reply. Useful in conjunction with "Send dummy message".
+* **Insert user message**: adds a new user message to the chat history without causing the model to generate a reply.
+* **Insert assistant message**: adds a new assistant message to the chat history as if the model had generated it. Useful in conjunction with "Insert user message".
 * **Send to Notebook**: sends the entire chat prompt up to now to the Notebook tab.
 * **Show controls**: checkbox that toggles the visibility of the sidebar controls (Start reply with, Mode, Chat style, etc.). Shortcut: Ctrl+S.
 
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 75d37e0e52..460226b1d6 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -73,8 +73,8 @@ def create_ui():
             shared.gradio['Continue'] = gr.Button('Continue (Alt + Enter)', elem_id='Continue')
             shared.gradio['Remove last'] = gr.Button('Remove last reply (Ctrl + Shift + Backspace)', elem_id='Remove-last')
             shared.gradio['Impersonate'] = gr.Button('Impersonate (Ctrl + Shift + M)', elem_id='Impersonate')
-            shared.gradio['Send dummy message'] = gr.Button('Send dummy message')
-            shared.gradio['Send dummy reply'] = gr.Button('Send dummy reply')
+            shared.gradio['Insert user message'] = gr.Button('Insert user message')
+            shared.gradio['Insert assistant message'] = gr.Button('Insert assistant message')
             shared.gradio['send-chat-to-notebook'] = gr.Button('Send to Notebook')
             shared.gradio['show_controls'] = gr.Checkbox(value=shared.settings['show_controls'], label='Show controls (Ctrl+S)', elem_id='show-controls')
 
@@ -280,11 +280,11 @@ def create_event_handlers():
         None, None, None, js='() => document.getElementById("chat").parentNode.parentNode.parentNode.classList.remove("_generating")').then(
         None, None, None, js=f'() => {{{ui.audio_notification_js}}}')
 
-    shared.gradio['Send dummy message'].click(
+    shared.gradio['Insert user message'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_send_dummy_message_click, gradio('textbox', 'interface_state'), gradio('history', 'display', 'textbox'), show_progress=False)
 
-    shared.gradio['Send dummy reply'].click(
+    shared.gradio['Insert assistant message'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.handle_send_dummy_reply_click, gradio('textbox', 'interface_state'), gradio('history', 'display', 'textbox'), show_progress=False)
 

From bed909d46c81fe12f2dceedaf4d1e36325dc9eb9 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 15 May 2026 06:39:30 -0700
Subject: [PATCH 1655/1701] Auto-select sibling mmproj when loading a model
 (closes #7564)

---
 modules/models_settings.py | 13 +++++++++++++
 modules/utils.py           | 31 +++++++++++++++++++++++++++++--
 2 files changed, 42 insertions(+), 2 deletions(-)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 53c09d07df..fa0a86a596 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -234,6 +234,19 @@ def apply_model_settings_to_state(model, state):
         if k in state and k != 'gpu_layers':  # Skip gpu_layers, handle separately
             state[k] = model_settings[k]
 
+    # Auto-detect a sibling mmproj when the user hasn't saved one for this model.
+    # Bare filenames (from user_data/mmproj/) persist across model switches;
+    # subfolder paths only persist while the new model lives in the same folder.
+    if state.get('loader') == 'llama.cpp' and 'mmproj' not in model_settings:
+        sibling = utils.find_sibling_mmproj(resolve_model_path(model))
+        if sibling:
+            state['mmproj'] = sibling
+        else:
+            current = state.get('mmproj')
+            if current and current != 'None' and ('/' in current or '\\' in current):
+                if Path(current).parent != Path(model).parent:
+                    state['mmproj'] = 'None'
+
     # Handle GPU layers and VRAM update for llama.cpp
     if state['loader'] == 'llama.cpp' and 'gpu_layers' in model_settings:
         gpu_layers = model_settings['gpu_layers']  # -1 (auto) by default, or user-saved value
diff --git a/modules/utils.py b/modules/utils.py
index 67d42c1bf9..c93706cdb0 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -203,6 +203,34 @@ def get_available_ggufs():
     return sorted(model_list, key=natural_keys)
 
 
+def is_mmproj_file(name):
+    lower = name.lower()
+    return lower.startswith('mmproj') and lower.endswith(('.gguf', '.bin'))
+
+
+def find_sibling_mmproj(model_path):
+    """Return an mmproj path relative to model_dir when exactly one mmproj file
+    sits next to the model in a subfolder of model_dir (any depth, not root).
+    """
+    try:
+        model_path = Path(model_path)
+        model_root = Path(shared.args.model_dir).resolve()
+        parent = model_path.parent.resolve()
+        if parent == model_root or model_root not in parent.parents:
+            return None
+
+        mmproj_candidates = [
+            entry for entry in parent.iterdir()
+            if entry.is_file() and is_mmproj_file(entry.name)
+        ]
+    except OSError:
+        return None
+
+    if len(mmproj_candidates) == 1:
+        return str(mmproj_candidates[0].relative_to(model_root))
+    return None
+
+
 def get_available_mmproj():
     mmproj_files = []
 
@@ -216,8 +244,7 @@ def get_available_mmproj():
     if model_dir.exists():
         for dirpath, _, files in os.walk(model_dir, followlinks=True):
             for file in files:
-                lower = file.lower()
-                if lower.startswith('mmproj') and lower.endswith(('.gguf', '.bin')):
+                if is_mmproj_file(file):
                     rel_path = str((Path(dirpath) / file).relative_to(model_dir))
                     mmproj_files.append(rel_path)
 

From 82d931e8ea99c00640fbc81dc06f8df1fadda850 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 15 May 2026 09:27:42 -0700
Subject: [PATCH 1656/1701] Keep web search blocks closed when user closes them
 mid-stream

---
 js/global_scope_js.js | 15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)

diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 349a05729e..3020ee6ec7 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -343,12 +343,14 @@ function applyMorphdomUpdate(data) {
   const messagesContainer = document.getElementsByClassName("messages")[0];
   const messagesCountBefore = messagesContainer ? messagesContainer.children.length : 0;
 
-  // Track open blocks and store their scroll positions
+  // Track open/closed blocks and store scroll positions for open ones
   const openBlocks = new Set();
+  const closedBlocks = new Set();
   const scrollPositions = {};
   queryScope.querySelectorAll(".thinking-block").forEach(block => {
     const blockId = block.getAttribute("data-block-id");
-    if (blockId && block.hasAttribute("open") && !block.querySelector(".tool-approval-buttons")) {
+    if (!blockId || block.querySelector(".tool-approval-buttons")) return;
+    if (block.hasAttribute("open")) {
       openBlocks.add(blockId);
       const content = block.querySelector(".thinking-content");
       if (content) {
@@ -358,6 +360,8 @@ function applyMorphdomUpdate(data) {
           isAtBottom: isAtBottom
         };
       }
+    } else {
+      closedBlocks.add(blockId);
     }
   });
 
@@ -378,13 +382,16 @@ function applyMorphdomUpdate(data) {
           }
         }
 
-        // For thinking blocks, preserve user-opened state and
-        // server-requested open (e.g. tool approval); otherwise close.
+        // For thinking blocks that already exist in the DOM, preserve the
+        // user's toggle state across streaming updates (in either direction).
+        // New blocks fall through to the server-rendered open/closed state.
         if (fromEl.classList && fromEl.classList.contains("thinking-block") &&
            toEl.classList && toEl.classList.contains("thinking-block")) {
           const blockId = toEl.getAttribute("data-block-id");
           if (blockId && openBlocks.has(blockId)) {
             toEl.setAttribute("open", "");
+          } else if (blockId && closedBlocks.has(blockId)) {
+            toEl.removeAttribute("open");
           }
         }
 

From 6033be53ac5f9f4877fd39dea8ebd30132761a0a Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 15 May 2026 09:43:22 -0700
Subject: [PATCH 1657/1701] Reorder right sidebar: Mode/Character/Chat style on
 top

---
 js/main.js         |  2 +-
 modules/ui_chat.py | 30 +++++++++++++++---------------
 2 files changed, 16 insertions(+), 16 deletions(-)

diff --git a/js/main.js b/js/main.js
index d5420610b2..9c2a60a4c1 100644
--- a/js/main.js
+++ b/js/main.js
@@ -555,7 +555,7 @@ function moveToChatTab() {
 
   const chatControlsFirstChild = document.querySelector("#chat-controls").firstElementChild;
   const newParent = chatControlsFirstChild;
-  let newPosition = newParent.children.length - 3;
+  let newPosition = 1;
 
   newParent.insertBefore(grandParent, newParent.children[newPosition]);
   document.getElementById("save-character").style.display = "none";
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 460226b1d6..b7896ecc33 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -81,23 +81,23 @@ def create_ui():
         with gr.Row(elem_id='chat-controls', elem_classes=['pretty_scrollbar']):
             with gr.Column():
                 with gr.Row():
-                    shared.gradio['start_with'] = gr.Textbox(label='Start reply with', placeholder='Sure thing!', value=shared.settings['start_with'], elem_classes=['add_scrollbar'])
+                    shared.gradio['mode'] = gr.Radio(choices=['instruct', 'chat-instruct', 'chat'], value=None, label='Mode', info='In instruct and chat-instruct modes, the template under Parameters > Instruction template is used.', elem_id='chat-mode')
 
-                show_separator, show_reasoning, show_thinking, show_preserve_thinking = utils.get_jinja_control_visibility(shared.settings.get('instruction_template_str', ''))
+                with gr.Row():
+                    shared.gradio['chat_style'] = gr.Dropdown(choices=utils.get_available_chat_styles(), label='Chat style', value=shared.settings['chat_style'], visible=shared.settings['mode'] != 'instruct')
 
-                shared.gradio['jinja_controls_separator'] = gr.HTML("
    ", visible=show_separator)
+                with gr.Row():
+                    shared.gradio['chat-instruct_command'] = gr.Textbox(value=shared.settings['chat-instruct_command'], lines=12, label='Command for chat-instruct mode', info='<|character|> and <|prompt|> get replaced with the bot name and the regular chat prompt respectively.', visible=shared.settings['mode'] == 'chat-instruct', elem_classes=['add_scrollbar'])
+
+                gr.HTML("
    ")
+
+                show_separator, show_reasoning, show_thinking, show_preserve_thinking = utils.get_jinja_control_visibility(shared.settings.get('instruction_template_str', ''))
 
                 shared.gradio['reasoning_effort'] = gr.Dropdown(value=shared.settings['reasoning_effort'], choices=['low', 'medium', 'high'], label='Reasoning effort', visible=show_reasoning)
                 shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', visible=show_thinking)
                 shared.gradio['preserve_thinking'] = gr.Checkbox(value=shared.settings['preserve_thinking'], label='Preserve thinking', visible=show_preserve_thinking)
 
-                gr.HTML("
    ")
-
-                shared.gradio['enable_web_search'] = gr.Checkbox(value=shared.settings.get('enable_web_search', False), label='Activate web search', elem_id='web-search')
-                with gr.Row(visible=shared.settings.get('enable_web_search', False)) as shared.gradio['web_search_row']:
-                    shared.gradio['web_search_pages'] = gr.Number(value=shared.settings.get('web_search_pages', 3), precision=0, label='Number of pages to download', minimum=1, maximum=10)
-
-                gr.HTML("
    ")
+                shared.gradio['jinja_controls_separator'] = gr.HTML("
    ", visible=show_separator)
 
                 from modules.tool_use import get_available_tools
                 shared.gradio['selected_tools'] = gr.CheckboxGroup(choices=get_available_tools(), value=shared.settings.get('selected_tools', []), label='Tools', info='Functions the model can call during generation.', elem_id='tools-group')
@@ -119,14 +119,14 @@ def sync_web_tools(selected):
 
                 gr.HTML("
    ")
 
-                with gr.Row():
-                    shared.gradio['mode'] = gr.Radio(choices=['instruct', 'chat-instruct', 'chat'], value=None, label='Mode', info='In instruct and chat-instruct modes, the template under Parameters > Instruction template is used.', elem_id='chat-mode')
+                shared.gradio['enable_web_search'] = gr.Checkbox(value=shared.settings.get('enable_web_search', False), label='Activate web search', elem_id='web-search')
+                with gr.Row(visible=shared.settings.get('enable_web_search', False)) as shared.gradio['web_search_row']:
+                    shared.gradio['web_search_pages'] = gr.Number(value=shared.settings.get('web_search_pages', 3), precision=0, label='Number of pages to download', minimum=1, maximum=10)
 
-                with gr.Row():
-                    shared.gradio['chat_style'] = gr.Dropdown(choices=utils.get_available_chat_styles(), label='Chat style', value=shared.settings['chat_style'], visible=shared.settings['mode'] != 'instruct')
+                gr.HTML("
    ")
 
                 with gr.Row():
-                    shared.gradio['chat-instruct_command'] = gr.Textbox(value=shared.settings['chat-instruct_command'], lines=12, label='Command for chat-instruct mode', info='<|character|> and <|prompt|> get replaced with the bot name and the regular chat prompt respectively.', visible=shared.settings['mode'] == 'chat-instruct', elem_classes=['add_scrollbar'])
+                    shared.gradio['start_with'] = gr.Textbox(label='Start reply with', placeholder='Sure thing!', value=shared.settings['start_with'], elem_classes=['add_scrollbar'])
 
                 gr.HTML("
    ")
 

From 2a27ceae9379b7d0bd6629c11291dbee0256ba7d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 15 May 2026 11:44:09 -0700
Subject: [PATCH 1658/1701] Hide reasoning and tools controls in chat mode

---
 js/main.js               |  6 ++++++
 modules/chat.py          | 11 +++++++++++
 modules/ui_chat.py       | 21 ++++++++++++---------
 modules/ui_model_menu.py |  6 ++++--
 4 files changed, 33 insertions(+), 11 deletions(-)

diff --git a/js/main.js b/js/main.js
index 9c2a60a4c1..7260ac560c 100644
--- a/js/main.js
+++ b/js/main.js
@@ -535,6 +535,12 @@ let originalIndex; // To keep track of the original position
 let movedElement;
 
 function moveToChatTab() {
+  // On first call, wait until mode is initialized so the visibility check below sees the real state
+  if (!originalParent && !document.querySelector("#chat-mode input:checked")) {
+    requestAnimationFrame(moveToChatTab);
+    return;
+  }
+
   const characterMenu = document.getElementById("character-menu");
   const grandParent = characterMenu.parentElement.parentElement;
 
diff --git a/modules/chat.py b/modules/chat.py
index 9a86c8cdb6..b31de2af10 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -2765,11 +2765,22 @@ def handle_mode_change(state):
     else:
         past_chats_update = gr.update(choices=histories)
 
+    show_separator, show_reasoning, show_thinking, show_preserve_thinking = utils.get_jinja_control_visibility(state.get('instruction_template_str', ''))
+    not_chat = state['mode'] != 'chat'
+
     return [
         history,
         html,
         gr.update(visible=state['mode'] != 'instruct'),
         gr.update(visible=state['mode'] == 'chat-instruct'),
+        gr.update(visible=not_chat),
+        gr.update(visible=show_reasoning and not_chat),
+        gr.update(visible=show_thinking and not_chat),
+        gr.update(visible=show_preserve_thinking and not_chat),
+        gr.update(visible=show_separator and not_chat),
+        gr.update(visible=not_chat),
+        gr.update(visible=not_chat),
+        gr.update(visible=not_chat),
         past_chats_update
     ]
 
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index b7896ecc33..8c01152649 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -89,18 +89,21 @@ def create_ui():
                 with gr.Row():
                     shared.gradio['chat-instruct_command'] = gr.Textbox(value=shared.settings['chat-instruct_command'], lines=12, label='Command for chat-instruct mode', info='<|character|> and <|prompt|> get replaced with the bot name and the regular chat prompt respectively.', visible=shared.settings['mode'] == 'chat-instruct', elem_classes=['add_scrollbar'])
 
-                gr.HTML("
    ")
+                # Reasoning, tools, and MCP rely on the instruction template; chat mode uses the chat template instead, so they're hidden there.
+                not_chat = shared.settings['mode'] != 'chat'
+
+                shared.gradio['tools_separator'] = gr.HTML("
    ", visible=not_chat)
 
                 show_separator, show_reasoning, show_thinking, show_preserve_thinking = utils.get_jinja_control_visibility(shared.settings.get('instruction_template_str', ''))
 
-                shared.gradio['reasoning_effort'] = gr.Dropdown(value=shared.settings['reasoning_effort'], choices=['low', 'medium', 'high'], label='Reasoning effort', visible=show_reasoning)
-                shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', visible=show_thinking)
-                shared.gradio['preserve_thinking'] = gr.Checkbox(value=shared.settings['preserve_thinking'], label='Preserve thinking', visible=show_preserve_thinking)
+                shared.gradio['reasoning_effort'] = gr.Dropdown(value=shared.settings['reasoning_effort'], choices=['low', 'medium', 'high'], label='Reasoning effort', visible=show_reasoning and not_chat)
+                shared.gradio['enable_thinking'] = gr.Checkbox(value=shared.settings['enable_thinking'], label='Enable thinking', visible=show_thinking and not_chat)
+                shared.gradio['preserve_thinking'] = gr.Checkbox(value=shared.settings['preserve_thinking'], label='Preserve thinking', visible=show_preserve_thinking and not_chat)
 
-                shared.gradio['jinja_controls_separator'] = gr.HTML("
    ", visible=show_separator)
+                shared.gradio['jinja_controls_separator'] = gr.HTML("
    ", visible=show_separator and not_chat)
 
                 from modules.tool_use import get_available_tools
-                shared.gradio['selected_tools'] = gr.CheckboxGroup(choices=get_available_tools(), value=shared.settings.get('selected_tools', []), label='Tools', info='Functions the model can call during generation.', elem_id='tools-group')
+                shared.gradio['selected_tools'] = gr.CheckboxGroup(choices=get_available_tools(), value=shared.settings.get('selected_tools', []), label='Tools', info='Functions the model can call during generation.', elem_id='tools-group', visible=not_chat)
                 shared.gradio['tools_refresh'] = gr.Button('Refresh list', elem_id='tools-refresh-btn', visible=False)
                 shared.gradio['tools_refresh'].click(fn=lambda: gr.update(choices=get_available_tools()), inputs=[], outputs=[shared.gradio['selected_tools']])
 
@@ -112,10 +115,10 @@ def sync_web_tools(selected):
 
                 shared.gradio['selected_tools'].change(fn=sync_web_tools, inputs=[shared.gradio['selected_tools']], outputs=[shared.gradio['selected_tools']], show_progress=False)
 
-                with gr.Accordion('MCP servers', open=False):
+                with gr.Accordion('MCP servers', open=False, visible=not_chat) as shared.gradio['mcp_servers_accordion']:
                     shared.gradio['mcp_servers'] = gr.Textbox(value=shared.settings.get('mcp_servers', ''), lines=3, max_lines=3, label='', info='One URL per line for HTTP servers. For headers: url,Header: value. For stdio servers, use user_data/mcp.json.', elem_classes=['add_scrollbar'])
 
-                shared.gradio['confirm_tool_calls'] = gr.Checkbox(value=shared.settings.get('confirm_tool_calls', False), label='Confirm tool calls', info='Ask for approval before executing each tool call.')
+                shared.gradio['confirm_tool_calls'] = gr.Checkbox(value=shared.settings.get('confirm_tool_calls', False), label='Confirm tool calls', info='Ask for approval before executing each tool call.', visible=not_chat)
 
                 gr.HTML("
    ")
 
@@ -352,7 +355,7 @@ def create_event_handlers():
 
     shared.gradio['mode'].change(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
-        chat.handle_mode_change, gradio('interface_state'), gradio('history', 'display', 'chat_style', 'chat-instruct_command', 'unique_id'), show_progress=False).then(
+        chat.handle_mode_change, gradio('interface_state'), gradio('history', 'display', 'chat_style', 'chat-instruct_command', 'tools_separator', 'reasoning_effort', 'enable_thinking', 'preserve_thinking', 'jinja_controls_separator', 'selected_tools', 'mcp_servers_accordion', 'confirm_tool_calls', 'unique_id'), show_progress=False).then(
         None, gradio('mode'), None, js="(mode) => {const characterContainer = document.getElementById('character-menu').parentNode.parentNode; const isInChatTab = document.querySelector('#chat-controls').contains(characterContainer); if (isInChatTab) { characterContainer.style.display = mode === 'instruct' ? 'none' : ''; } if (mode === 'instruct') document.querySelectorAll('.bigProfilePicture').forEach(el => el.remove());}")
 
     shared.gradio['chat_style'].change(chat.redraw_html, gradio(reload_arr), gradio('display'), show_progress=False)
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 77b5c18fc3..c4e2658b16 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -441,17 +441,19 @@ def handle_load_model_event_initial(model, state):
     update_model_parameters(state)  # This updates the command-line flags
 
     show_separator, _, _, _ = utils.get_jinja_control_visibility(state.get('instruction_template_str', ''))
+    not_chat = state.get('mode') != 'chat'
 
     vram_info = state.get('vram_info', "
    Estimated VRAM to load the model:
    ")
-    return output + [state] + [vram_info] + [gr.update(visible=show_separator)]
+    return output + [state] + [vram_info] + [gr.update(visible=show_separator and not_chat)]
 
 
 def handle_load_model_event_final(truncation_length, loader, state):
     truncation_length = update_truncation_length(truncation_length, state)
 
     show_separator, show_reasoning, show_thinking, show_preserve_thinking = utils.get_jinja_control_visibility(state.get('instruction_template_str', ''))
+    not_chat = state.get('mode') != 'chat'
 
-    return [truncation_length, loader, gr.update(visible=show_separator), gr.update(visible=show_reasoning), gr.update(visible=show_thinking), gr.update(visible=show_preserve_thinking)]
+    return [truncation_length, loader, gr.update(visible=show_separator and not_chat), gr.update(visible=show_reasoning and not_chat), gr.update(visible=show_thinking and not_chat), gr.update(visible=show_preserve_thinking and not_chat)]
 
 
 def handle_unload_model_click():

From 626b0899c0c596691ad2c2d37ed945b9f7171a98 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 15 May 2026 11:49:40 -0700
Subject: [PATCH 1659/1701] Polish character dropdown in chat tab

---
 js/main.js | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/js/main.js b/js/main.js
index 7260ac560c..ee4b1c3034 100644
--- a/js/main.js
+++ b/js/main.js
@@ -557,7 +557,8 @@ function moveToChatTab() {
     grandParent.style.display = "none";
   }
 
-  grandParent.children[0].style.minWidth = "100%";
+  grandParent.children[0].style.flex = "1";
+  grandParent.children[0].style.minWidth = "0";
 
   const chatControlsFirstChild = document.querySelector("#chat-controls").firstElementChild;
   const newParent = chatControlsFirstChild;
@@ -566,6 +567,9 @@ function moveToChatTab() {
   newParent.insertBefore(grandParent, newParent.children[newPosition]);
   document.getElementById("save-character").style.display = "none";
   document.getElementById("restore-character").style.display = "none";
+
+  const characterInfo = document.querySelector("#character-menu [data-testid='block-info']")?.nextElementSibling;
+  if (characterInfo) characterInfo.style.display = "none";
 }
 
 function restoreOriginalPosition() {
@@ -579,7 +583,11 @@ function restoreOriginalPosition() {
     document.getElementById("save-character").style.display = "";
     document.getElementById("restore-character").style.display = "";
     movedElement.style.display = "";
+    movedElement.children[0].style.flex = "";
     movedElement.children[0].style.minWidth = "";
+
+    const characterInfo = document.querySelector("#character-menu [data-testid='block-info']")?.nextElementSibling;
+    if (characterInfo) characterInfo.style.display = "";
   }
 }
 

From 5270943c6316028e2aeb13e84407cc424c14aa18 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 15 May 2026 11:57:23 -0700
Subject: [PATCH 1660/1701] Fix Show controls text style in hover menu

---
 css/main.css | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/css/main.css b/css/main.css
index 0bd7fe3ace..bb5b727dc7 100644
--- a/css/main.css
+++ b/css/main.css
@@ -873,9 +873,14 @@ audio {
     padding-right: 12px;
     gap: 10px;
     font-weight: 500;
+    font-size: 14px;
     color: var(--button-secondary-text-color);
 }
 
+#show-controls label span {
+    color: inherit;
+}
+
 #show-controls label input {
     margin-top: 5px;
 }

From 3c4a1402a88eb963620b1c6ec8619c8952ae96cf Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 15 May 2026 12:10:27 -0700
Subject: [PATCH 1661/1701] Soften slider and checkbox label colors in light
 theme

---
 css/main.css  | 5 +++++
 modules/ui.py | 2 +-
 2 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/css/main.css b/css/main.css
index bb5b727dc7..f290b1fd09 100644
--- a/css/main.css
+++ b/css/main.css
@@ -889,6 +889,11 @@ audio {
     opacity: 0.333;
 }
 
+/* Match checkbox label color to block title color (Gradio hardcodes body_text_color in Checkbox.svelte) */
+label:has(> input[type="checkbox"]) {
+    color: var(--button-secondary-text-color);
+}
+
 #chat-tab #chat-buttons {
     display: none !important;
 }
diff --git a/modules/ui.py b/modules/ui.py
index 835bb0c655..7e5e1cd105 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -76,7 +76,7 @@
         body_text_color='#1a1a1a',
         button_secondary_background_fill="white",
         button_secondary_border_color="var(--border-color-primary)",
-        block_title_text_color='*body_text_color',
+        block_title_text_color='*button_secondary_text_color',
         button_primary_background_fill='var(--accent, #4a72ff)',
         button_primary_background_fill_hover='#3556cc',
         button_primary_border_color='var(--accent, #4a72ff)',

From 8e34c7b1b054e8eb9526180c5c3598d3d00cab78 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 15 May 2026 12:45:05 -0700
Subject: [PATCH 1662/1701] Show live context size while generating

---
 modules/chat.py             | 20 ++++++++++++++++++++
 modules/exllamav3.py        |  5 +++++
 modules/llama_cpp_server.py | 11 +++++++----
 modules/tensorrt_llm.py     |  6 +++++-
 modules/text_generation.py  |  6 +++++-
 modules/ui_chat.py          |  3 +++
 6 files changed, 45 insertions(+), 6 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index b31de2af10..7833d58c9b 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -745,6 +745,26 @@ def count_prompt_tokens(text_input, state):
         return f"Error: {str(e)}"
 
 
+def update_token_display_from_state(state):
+    import gradio as gr
+    if shared.model is None:
+        return gr.update()
+
+    prompt_n = getattr(shared.model, 'last_prompt_token_count', None)
+    if not prompt_n:
+        return gr.update()
+
+    gen_n = getattr(shared.model, 'last_completion_token_count', 0) or 0
+    total = prompt_n + gen_n
+    max_tokens = state.get('truncation_length') or 0
+    percentage = (total / max_tokens) * 100 if max_tokens > 0 else 0
+    new_value = f"{total:,} / {max_tokens:,} tokens ({percentage:.1f}%)"
+    if new_value == getattr(shared.model, '_last_token_display', None):
+        return gr.update()
+    shared.model._last_token_display = new_value
+    return new_value
+
+
 def get_stopping_strings(state):
     renderers = []
 
diff --git a/modules/exllamav3.py b/modules/exllamav3.py
index a05a56ca13..6b5d6d99fa 100644
--- a/modules/exllamav3.py
+++ b/modules/exllamav3.py
@@ -428,6 +428,7 @@ def custom_sort_key(sampler_obj):
         response_text = ""
         stop_event = state.get('stop_event')
         self.last_completion_probabilities = []
+        self.last_completion_token_count = 0
 
         result_queue = self.parallel_generator.submit(job)
         try:
@@ -446,6 +447,10 @@ def custom_sort_key(sampler_obj):
                 if return_top_tokens > 0:
                     self._capture_logprobs(result)
 
+                step_tokens = result.get("token_ids")
+                if step_tokens is not None:
+                    self.last_completion_token_count += len(step_tokens)
+
                 if chunk:
                     response_text += chunk
                     yield response_text
diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 525b858415..7f145b17f2 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -221,6 +221,10 @@ def generate_with_streaming(self, prompt, state):
             pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(printable_payload)
             print()
 
+        full_text = ""
+        self.last_completion_probabilities = []
+        self.last_completion_token_count = 0
+
         # Make the generation request
         response = self.session.post(url, json=payload, stream=True)
         try:
@@ -230,9 +234,6 @@ def generate_with_streaming(self, prompt, state):
             else:
                 response.raise_for_status()  # Raise an exception for HTTP errors
 
-            full_text = ""
-            self.last_completion_probabilities = []
-
             # Process the streaming response
             stop_event = state.get('stop_event')
             for line in response.iter_lines():
@@ -255,14 +256,16 @@ def generate_with_streaming(self, prompt, state):
                     # Extract the token content
                     if data.get('content', ''):
                         full_text += data['content']
+                        self.last_completion_token_count += 1
                         yield full_text
 
                     # Capture logprobs if present
                     if 'completion_probabilities' in data:
                         self.last_completion_probabilities.extend(data['completion_probabilities'])
 
-                    # Check if generation is complete
                     if data.get('stop', False):
+                        # Server count includes speculative-decode tokens our per-chunk counter misses.
+                        self.last_completion_token_count = data.get('tokens_predicted', self.last_completion_token_count)
                         break
 
                 except json.JSONDecodeError as e:
diff --git a/modules/tensorrt_llm.py b/modules/tensorrt_llm.py
index ae061d0664..8ccf77e883 100644
--- a/modules/tensorrt_llm.py
+++ b/modules/tensorrt_llm.py
@@ -26,9 +26,12 @@ def from_pretrained(cls, path_to_model):
         return result
 
     def generate_with_streaming(self, prompt, state):
+        self.last_prompt_token_count = len(shared.tokenizer.encode(prompt))
+        self.last_completion_token_count = 0
+
         sampling_params = SamplingParams(
             max_tokens=state['max_new_tokens'] if not state['auto_max_new_tokens']
-                       else state['truncation_length'] - len(shared.tokenizer.encode(prompt)),
+                       else state['truncation_length'] - self.last_prompt_token_count,
             end_id=shared.tokenizer.eos_token_id,
             temperature=state['temperature'],
             top_k=state['top_k'],
@@ -53,6 +56,7 @@ def generate_with_streaming(self, prompt, state):
                 result.abort()
                 break
 
+            self.last_completion_token_count = len(output.outputs[0].token_ids)
             text_diff = output.outputs[0].text_diff
             if text_diff:
                 cumulative_reply += text_diff
diff --git a/modules/text_generation.py b/modules/text_generation.py
index da73632913..112285cdf2 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -405,6 +405,8 @@ def generate_reply_HF(question, original_question, state, stopping_strings=None,
     # Encode the input
     input_ids = encode(question, add_bos_token=state['add_bos_token'], truncation_length=get_max_prompt_length(state))
     output = input_ids[0]
+    shared.model.last_prompt_token_count = input_ids.shape[-1]
+    shared.model.last_completion_token_count = 0
     if state['auto_max_new_tokens']:
         generate_params['max_new_tokens'] = state['truncation_length'] - input_ids.shape[-1]
 
@@ -474,7 +476,8 @@ def generate_with_streaming(**kwargs):
 
             with generate_with_streaming(**generate_params) as generator:
                 cumulative_reply = ''
-                starting_from = 0 if shared.is_seq2seq else len(input_ids[0])
+                prompt_len = 0 if shared.is_seq2seq else len(input_ids[0])
+                starting_from = prompt_len
                 for output in generator:
                     if output[-1] in eos_token_ids:
                         break
@@ -485,6 +488,7 @@ def generate_with_streaming(**kwargs):
                         continue
 
                     cumulative_reply += new_content
+                    shared.model.last_completion_token_count = len(output) - prompt_len
                     starting_from = len(output)
                     yield cumulative_reply
 
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 8c01152649..2836644a1b 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -243,6 +243,9 @@ def create_event_handlers():
     # Morph HTML updates instead of updating everything
     shared.gradio['display'].change(None, gradio('display'), None, js="(data) => handleMorphdomUpdate(data)")
 
+    shared.gradio['display'].change(
+        chat.update_token_display_from_state, gradio('interface_state'), gradio('token_display'), show_progress=False)
+
     shared.gradio['Generate'].click(
         ui.gather_interface_values, gradio(shared.input_elements), gradio('interface_state')).then(
         chat.check_model_loaded_or_raise, None, None, show_progress=False).success(

From dbf922006c7886b36d29fa2726017781b34ec8b6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 15 May 2026 12:47:50 -0700
Subject: [PATCH 1663/1701] Revert "Soften slider and checkbox label colors in
 light theme"

This reverts commit 3c4a1402a88eb963620b1c6ec8619c8952ae96cf.
---
 css/main.css  | 5 -----
 modules/ui.py | 2 +-
 2 files changed, 1 insertion(+), 6 deletions(-)

diff --git a/css/main.css b/css/main.css
index f290b1fd09..bb5b727dc7 100644
--- a/css/main.css
+++ b/css/main.css
@@ -889,11 +889,6 @@ audio {
     opacity: 0.333;
 }
 
-/* Match checkbox label color to block title color (Gradio hardcodes body_text_color in Checkbox.svelte) */
-label:has(> input[type="checkbox"]) {
-    color: var(--button-secondary-text-color);
-}
-
 #chat-tab #chat-buttons {
     display: none !important;
 }
diff --git a/modules/ui.py b/modules/ui.py
index 7e5e1cd105..835bb0c655 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -76,7 +76,7 @@
         body_text_color='#1a1a1a',
         button_secondary_background_fill="white",
         button_secondary_border_color="var(--border-color-primary)",
-        block_title_text_color='*button_secondary_text_color',
+        block_title_text_color='*body_text_color',
         button_primary_background_fill='var(--accent, #4a72ff)',
         button_primary_background_fill_hover='#3556cc',
         button_primary_border_color='var(--accent, #4a72ff)',

From ceade2eb1ba3f84518076270df2240b6bbb01da0 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Fri, 15 May 2026 13:20:46 -0700
Subject: [PATCH 1664/1701] docs: reorder API examples by importance

---
 docs/12 - OpenAI API.md | 352 ++++++++++++++++++++--------------------
 1 file changed, 176 insertions(+), 176 deletions(-)

diff --git a/docs/12 - OpenAI API.md b/docs/12 - OpenAI API.md
index 81b3a8f961..d792c2cf09 100644
--- a/docs/12 - OpenAI API.md	
+++ b/docs/12 - OpenAI API.md	
@@ -23,20 +23,6 @@ For the documentation with all the endpoints, parameters and their types, consul
 
 The official examples in the [OpenAI documentation](https://platform.openai.com/docs/api-reference) should also work, and the same parameters apply (although the API here has more optional parameters).
 
-#### Completions
-
-```shell
-curl http://127.0.0.1:5000/v1/completions \
-  -H "Content-Type: application/json" \
-  -d '{
-    "prompt": "This is a cake recipe:\n\n1.",
-    "max_tokens": 512,
-    "temperature": 0.6,
-    "top_p": 0.95,
-    "top_k": 20
-  }'
-```
-
 #### Chat completions
 
 Works best with instruction-following models. If the "instruction_template" variable is not provided, it will be detected automatically from the model metadata.
@@ -57,7 +43,21 @@ curl http://127.0.0.1:5000/v1/chat/completions \
   }'
 ```
 
-#### Chat completions with characters
+#### Completions
+
+```shell
+curl http://127.0.0.1:5000/v1/completions \
+  -H "Content-Type: application/json" \
+  -d '{
+    "prompt": "This is a cake recipe:\n\n1.",
+    "max_tokens": 512,
+    "temperature": 0.6,
+    "top_p": 0.95,
+    "top_k": 20
+  }'
+```
+
+#### SSE streaming
 
 ```shell
 curl http://127.0.0.1:5000/v1/chat/completions \
@@ -66,17 +66,105 @@ curl http://127.0.0.1:5000/v1/chat/completions \
     "messages": [
       {
         "role": "user",
-        "content": "Hello! Who are you?"
+        "content": "Hello!"
       }
     ],
-    "mode": "chat-instruct",
-    "character": "Example",
     "temperature": 0.6,
     "top_p": 0.95,
-    "top_k": 20
+    "top_k": 20,
+    "stream": true
   }'
 ```
 
+#### Tool/Function calling
+
+Use a model with tool calling support (Qwen, Mistral, GPT-OSS, etc). Tools are passed via the `tools` parameter and the prompt is automatically formatted using the model's Jinja2 template.
+
+When the model decides to call a tool, the response will have `finish_reason: "tool_calls"` and a `tool_calls` array with structured function names and arguments. You then execute the tool, send the result back as a `role: "tool"` message, and continue until the model responds with `finish_reason: "stop"`.
+
+Some models call multiple tools in parallel (Qwen, Mistral), while others call one at a time (GPT-OSS). The loop below handles both styles.
+
+```python
+import json
+import requests
+
+url = "http://127.0.0.1:5000/v1/chat/completions"
+
+# Define your tools
+tools = [
+    {
+        "type": "function",
+        "function": {
+            "name": "get_weather",
+            "description": "Get the current weather for a given location",
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "location": {"type": "string", "description": "City name"},
+                },
+                "required": ["location"]
+            }
+        }
+    },
+    {
+        "type": "function",
+        "function": {
+            "name": "get_time",
+            "description": "Get the current time in a given timezone",
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "timezone": {"type": "string", "description": "IANA timezone string"},
+                },
+                "required": ["timezone"]
+            }
+        }
+    },
+]
+
+
+def execute_tool(name, arguments):
+    """Replace this with your actual tool implementations."""
+    if name == "get_weather":
+        return {"temperature": 22, "condition": "sunny", "humidity": 45}
+    elif name == "get_time":
+        return {"time": "2:30 PM", "timezone": "JST"}
+    return {"error": f"Unknown tool: {name}"}
+
+
+messages = [{"role": "user", "content": "What time is it in Tokyo and what's the weather like there?"}]
+
+# Tool-calling loop: keep going until the model gives a final answer
+for _ in range(10):
+    response = requests.post(url, json={"messages": messages, "tools": tools}).json()
+    choice = response["choices"][0]
+
+    if choice["finish_reason"] == "tool_calls":
+        # Add the assistant's response (with tool_calls) to history
+        messages.append({
+            "role": "assistant",
+            "content": choice["message"]["content"],
+            "tool_calls": choice["message"]["tool_calls"],
+        })
+
+        # Execute each tool and add results to history
+        for tool_call in choice["message"]["tool_calls"]:
+            name = tool_call["function"]["name"]
+            arguments = json.loads(tool_call["function"]["arguments"])
+            result = execute_tool(name, arguments)
+
+            print(f"Tool call: {name}({arguments}) => {result}")
+            messages.append({
+                "role": "tool",
+                "tool_call_id": tool_call["id"],
+                "content": json.dumps(result),
+            })
+    else:
+        # Final answer
+        print(f"\nAssistant: {choice['message']['content']}")
+        break
+```
+
 #### Multimodal/vision (llama.cpp and ExLlamaV3)
 
 ##### With /v1/chat/completions (recommended!)
@@ -139,6 +227,58 @@ curl http://127.0.0.1:5000/v1/completions \
 
 For base64-encoded images, just replace the inner "url" values with this format: `data:image/FORMAT;base64,BASE64_STRING` where FORMAT is the file type (png, jpeg, gif, etc.) and BASE64_STRING is your base64-encoded image data.
 
+#### List models
+
+```shell
+curl -k http://127.0.0.1:5000/v1/internal/model/list \
+  -H "Content-Type: application/json"
+```
+
+#### Load model
+
+```shell
+curl -k http://127.0.0.1:5000/v1/internal/model/load \
+  -H "Content-Type: application/json" \
+  -d '{
+    "model_name": "Qwen_Qwen3-0.6B-Q4_K_M.gguf",
+    "args": {
+      "ctx_size": 32768,
+      "cache_type": "q8_0"
+    }
+  }'
+```
+
+You can also set a default instruction template for all subsequent API requests by passing `instruction_template` (a template name from `user_data/instruction-templates/`) or `instruction_template_str` (a raw Jinja2 string):
+
+```shell
+curl -k http://127.0.0.1:5000/v1/internal/model/load \
+  -H "Content-Type: application/json" \
+  -d '{
+    "model_name": "Qwen_Qwen3-0.6B-Q4_K_M.gguf",
+    "instruction_template": "Alpaca"
+  }'
+```
+
+#### Chat completions with characters
+
+```shell
+curl http://127.0.0.1:5000/v1/chat/completions \
+  -H "Content-Type: application/json" \
+  -d '{
+    "messages": [
+      {
+        "role": "user",
+        "content": "Hello! Who are you?"
+      }
+    ],
+    "mode": "chat-instruct",
+    "character": "Example",
+    "temperature": 0.6,
+    "top_p": 0.95,
+    "top_k": 20
+  }'
+```
+
 #### Image generation
 
 ```shell
@@ -168,25 +308,6 @@ The output is a JSON object containing a `data` array. Each element has a `b64_j
 }
 ```
 
-#### SSE streaming
-
-```shell
-curl http://127.0.0.1:5000/v1/chat/completions \
-  -H "Content-Type: application/json" \
-  -d '{
-    "messages": [
-      {
-        "role": "user",
-        "content": "Hello!"
-      }
-    ],
-    "temperature": 0.6,
-    "top_p": 0.95,
-    "top_k": 20,
-    "stream": true
-  }'
-```
-
 #### Logits
 
 ```shell
@@ -210,38 +331,6 @@ curl -k http://127.0.0.1:5000/v1/internal/logits \
   }'
 ```
 
-#### List models
-
-```shell
-curl -k http://127.0.0.1:5000/v1/internal/model/list \
-  -H "Content-Type: application/json"
-```
-
-#### Load model
-
-```shell
-curl -k http://127.0.0.1:5000/v1/internal/model/load \
-  -H "Content-Type: application/json" \
-  -d '{
-    "model_name": "Qwen_Qwen3-0.6B-Q4_K_M.gguf",
-    "args": {
-      "ctx_size": 32768,
-      "cache_type": "q8_0"
-    }
-  }'
-```
-
-You can also set a default instruction template for all subsequent API requests by passing `instruction_template` (a template name from `user_data/instruction-templates/`) or `instruction_template_str` (a raw Jinja2 string):
-
-```shell
-curl -k http://127.0.0.1:5000/v1/internal/model/load \
-  -H "Content-Type: application/json" \
-  -d '{
-    "model_name": "Qwen_Qwen3-0.6B-Q4_K_M.gguf",
-    "instruction_template": "Alpaca"
-  }'
-```
-
 #### Python chat example
 
 ```python
@@ -348,35 +437,6 @@ for event in client.events():
 print()
 ```
 
-#### Python parallel requests example
-
-The API supports handling multiple requests in parallel. For ExLlamaV3, this works out of the box. For llama.cpp, you need to pass `--parallel N` to set the number of concurrent slots.
-
-```python
-import concurrent.futures
-import requests
-
-url = "http://127.0.0.1:5000/v1/chat/completions"
-prompts = [
-    "Write a haiku about the ocean.",
-    "Explain quantum computing in simple terms.",
-    "Tell me a joke about programmers.",
-]
-
-def send_request(prompt):
-    response = requests.post(url, json={
-        "messages": [{"role": "user", "content": prompt}],
-        "max_tokens": 200,
-    })
-    return response.json()["choices"][0]["message"]["content"]
-
-with concurrent.futures.ThreadPoolExecutor() as executor:
-    results = list(executor.map(send_request, prompts))
-
-for prompt, result in zip(prompts, results):
-    print(f"Q: {prompt}\nA: {result}\n")
-```
-
 #### Python example with API key
 
 Replace
@@ -398,93 +458,33 @@ headers = {
 
 in any of the examples above.
 
-#### Tool/Function calling
-
-Use a model with tool calling support (Qwen, Mistral, GPT-OSS, etc). Tools are passed via the `tools` parameter and the prompt is automatically formatted using the model's Jinja2 template.
-
-When the model decides to call a tool, the response will have `finish_reason: "tool_calls"` and a `tool_calls` array with structured function names and arguments. You then execute the tool, send the result back as a `role: "tool"` message, and continue until the model responds with `finish_reason: "stop"`.
+#### Python parallel requests example
 
-Some models call multiple tools in parallel (Qwen, Mistral), while others call one at a time (GPT-OSS). The loop below handles both styles.
+The API supports handling multiple requests in parallel. For ExLlamaV3, this works out of the box. For llama.cpp, you need to pass `--parallel N` to set the number of concurrent slots.
 
 ```python
-import json
+import concurrent.futures
 import requests
 
 url = "http://127.0.0.1:5000/v1/chat/completions"
-
-# Define your tools
-tools = [
-    {
-        "type": "function",
-        "function": {
-            "name": "get_weather",
-            "description": "Get the current weather for a given location",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "location": {"type": "string", "description": "City name"},
-                },
-                "required": ["location"]
-            }
-        }
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "get_time",
-            "description": "Get the current time in a given timezone",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "timezone": {"type": "string", "description": "IANA timezone string"},
-                },
-                "required": ["timezone"]
-            }
-        }
-    },
+prompts = [
+    "Write a haiku about the ocean.",
+    "Explain quantum computing in simple terms.",
+    "Tell me a joke about programmers.",
 ]
 
+def send_request(prompt):
+    response = requests.post(url, json={
+        "messages": [{"role": "user", "content": prompt}],
+        "max_tokens": 200,
+    })
+    return response.json()["choices"][0]["message"]["content"]
 
-def execute_tool(name, arguments):
-    """Replace this with your actual tool implementations."""
-    if name == "get_weather":
-        return {"temperature": 22, "condition": "sunny", "humidity": 45}
-    elif name == "get_time":
-        return {"time": "2:30 PM", "timezone": "JST"}
-    return {"error": f"Unknown tool: {name}"}
-
-
-messages = [{"role": "user", "content": "What time is it in Tokyo and what's the weather like there?"}]
-
-# Tool-calling loop: keep going until the model gives a final answer
-for _ in range(10):
-    response = requests.post(url, json={"messages": messages, "tools": tools}).json()
-    choice = response["choices"][0]
-
-    if choice["finish_reason"] == "tool_calls":
-        # Add the assistant's response (with tool_calls) to history
-        messages.append({
-            "role": "assistant",
-            "content": choice["message"]["content"],
-            "tool_calls": choice["message"]["tool_calls"],
-        })
-
-        # Execute each tool and add results to history
-        for tool_call in choice["message"]["tool_calls"]:
-            name = tool_call["function"]["name"]
-            arguments = json.loads(tool_call["function"]["arguments"])
-            result = execute_tool(name, arguments)
+with concurrent.futures.ThreadPoolExecutor() as executor:
+    results = list(executor.map(send_request, prompts))
 
-            print(f"Tool call: {name}({arguments}) => {result}")
-            messages.append({
-                "role": "tool",
-                "tool_call_id": tool_call["id"],
-                "content": json.dumps(result),
-            })
-    else:
-        # Final answer
-        print(f"\nAssistant: {choice['message']['content']}")
-        break
+for prompt, result in zip(prompts, results):
+    print(f"Q: {prompt}\nA: {result}\n")
 ```
 
 ### Environment variables

From 6378c5cd57c57d860ea6a9cbf6261912f40388d3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 16 May 2026 08:53:40 -0700
Subject: [PATCH 1665/1701] Electron: Add "Check for updates" button in the
 Session tab

---
 desktop/main.js       |  6 ++++-
 modules/ui_session.py | 58 +++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 63 insertions(+), 1 deletion(-)

diff --git a/desktop/main.js b/desktop/main.js
index 01d0c0a1c7..6c1ad9a62a 100644
--- a/desktop/main.js
+++ b/desktop/main.js
@@ -1,4 +1,4 @@
-const { app, BrowserWindow, Menu, dialog, ipcMain, screen } = require("electron");
+const { app, BrowserWindow, Menu, dialog, ipcMain, screen, shell } = require("electron");
 const { spawn } = require("child_process");
 const path = require("path");
 const fs = require("fs");
@@ -127,6 +127,10 @@ function createWindow(port) {
     if (params.editFlags.canSelectAll) tmpl.push({ type: "separator" }, { role: "selectAll" });
     if (tmpl.length) Menu.buildFromTemplate(tmpl).popup({ window: mainWindow });
   });
+  mainWindow.webContents.setWindowOpenHandler(({ url }) => {
+    if (/^https?:\/\//i.test(url)) shell.openExternal(url);
+    return { action: "deny" };
+  });
   mainWindow.on("page-title-updated", (e) => e.preventDefault());
   mainWindow.webContents.on("will-prevent-unload", (e) => e.preventDefault());
   mainWindow.on("close", saveState);
diff --git a/modules/ui_session.py b/modules/ui_session.py
index a9bd1141af..c7178d3f43 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -1,3 +1,6 @@
+import json
+import re
+import urllib.request
 from pathlib import Path
 
 import gradio as gr
@@ -5,9 +8,55 @@
 from modules import shared, ui, utils
 from modules.utils import gradio
 
+PORTABLE_FOLDER_RE = re.compile(r'^textgen(?:-ik)?-(\d+\.\d+(?:\.\d+)?)$')
+
+
+def detect_portable_install():
+    """Return the local version string if running from a portable build, else None."""
+    grandparent = Path(__file__).resolve().parent.parent
+    if grandparent.name != 'app':
+        return None
+
+    match = PORTABLE_FOLDER_RE.match(grandparent.parent.name)
+    if not match:
+        return None
+
+    return match.group(1)
+
+
+def check_for_updates(local_version):
+    try:
+        req = urllib.request.Request(
+            'https://api.github.com/repos/oobabooga/textgen/releases/latest',
+            headers={'Accept': 'application/vnd.github+json'},
+        )
+        with urllib.request.urlopen(req, timeout=10) as resp:
+            data = json.loads(resp.read().decode('utf-8'))
+    except Exception as e:
+        return f'
    Failed to check for updates: {e}
    '
+
+    latest = (data.get('tag_name') or '').lstrip('v')
+    published = (data.get('published_at') or '')[:10]
+    url = data.get('html_url') or 'https://github.com/oobabooga/textgen/releases/latest'
+
+    if latest and latest != local_version:
+        return (
+            f'
    '
+            f'
    Update available
    '
+            f'
      '
+            f'
    • Current version: {local_version}
      • '
+            f'
    • Latest version: {latest} (released {published})
      • '
+            f'
    '
+            f'
    Download here
    '
+            f'
    '
+        )
+
+    return f'
    Already up to date (version {local_version}).
    '
+
 
 def create_ui():
     mu = shared.args.multi_user
+    portable_version = detect_portable_install() if shared.args.portable else None
     with gr.Tab("Session", elem_id="session-tab"):
         with gr.Row():
             with gr.Column():
@@ -21,6 +70,11 @@ def create_ui():
                 shared.gradio['paste_to_attachment'] = gr.Checkbox(label='Turn long pasted text into attachments in the Chat tab', value=shared.settings['paste_to_attachment'], elem_id='paste_to_attachment')
                 shared.gradio['include_past_attachments'] = gr.Checkbox(label='Include attachments/search results from previous messages in the chat prompt', value=shared.settings['include_past_attachments'])
 
+                if portable_version:
+                    gr.Markdown("## Updates")
+                    shared.gradio['check_updates'] = gr.Button('Check for updates 🔄', elem_classes='refresh-button')
+                    shared.gradio['update_status'] = gr.HTML(value='', elem_id='update-status')
+
             with gr.Column():
                 gr.Markdown("## Extensions & flags")
                 shared.gradio['save_settings'] = gr.Button(f'Save extensions settings to {shared.user_data_dir}/settings.yaml', interactive=not mu)
@@ -42,6 +96,10 @@ def create_ui():
             lambda x: 'dark' if x == 'light' else 'light', gradio('theme_state'), gradio('theme_state')).then(
             None, None, None, js=f'() => {{{ui.dark_theme_js}; toggleDarkMode(); localStorage.setItem("theme", document.body.classList.contains("dark") ? "dark" : "light")}}')
 
+        if portable_version:
+            shared.gradio['check_updates'].click(
+                lambda: check_for_updates(portable_version), None, gradio('update_status'), show_progress=False)
+
         if shared.is_electron:
             shared.gradio['model_dir_browse'].click(
                 None, gradio('model_dir'), gradio('model_dir'),

From f327da783160d4ccadf28e76a5358ce51a4b45d1 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 16 May 2026 09:59:42 -0700
Subject: [PATCH 1666/1701] Improve the looks of the Session tab

---
 css/main.css          |  7 +++++++
 modules/ui_session.py | 10 ++++++----
 2 files changed, 13 insertions(+), 4 deletions(-)

diff --git a/css/main.css b/css/main.css
index bb5b727dc7..5ee556cedb 100644
--- a/css/main.css
+++ b/css/main.css
@@ -130,6 +130,13 @@ div.svelte-iyf88w {
     max-width: 2.2em;
 }
 
+.settings-button {
+    width: fit-content !important;
+    min-width: 0 !important;
+    max-width: 100% !important;
+    align-self: flex-start;
+}
+
 .button_nowrap {
     white-space: nowrap;
 }
diff --git a/modules/ui_session.py b/modules/ui_session.py
index c7178d3f43..546c3c6df2 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -61,24 +61,26 @@ def create_ui():
         with gr.Row():
             with gr.Column():
                 gr.Markdown("## Settings")
-                shared.gradio['toggle_dark_mode'] = gr.Button('Toggle light/dark theme 💡', elem_classes='refresh-button')
                 if shared.is_electron:
                     with gr.Row():
                         shared.gradio['model_dir'] = gr.Textbox(label='Models directory', value=shared.settings['model_dir'], scale=4, elem_classes='slim-textbox')
                         shared.gradio['model_dir_browse'] = gr.Button('Browse', elem_classes=['refresh-button', 'refresh-button-medium'])
+
+                shared.gradio['toggle_dark_mode'] = gr.Button('Toggle light/dark theme 💡', elem_classes=['refresh-button', 'settings-button'])
                 shared.gradio['show_two_notebook_columns'] = gr.Checkbox(label='Show two columns in the Notebook tab', value=shared.settings['show_two_notebook_columns'])
                 shared.gradio['paste_to_attachment'] = gr.Checkbox(label='Turn long pasted text into attachments in the Chat tab', value=shared.settings['paste_to_attachment'], elem_id='paste_to_attachment')
                 shared.gradio['include_past_attachments'] = gr.Checkbox(label='Include attachments/search results from previous messages in the chat prompt', value=shared.settings['include_past_attachments'])
 
                 if portable_version:
                     gr.Markdown("## Updates")
-                    shared.gradio['check_updates'] = gr.Button('Check for updates 🔄', elem_classes='refresh-button')
+                    shared.gradio['check_updates'] = gr.Button('Check for updates 🔄', elem_classes=['refresh-button', 'settings-button'])
                     shared.gradio['update_status'] = gr.HTML(value='', elem_id='update-status')
 
             with gr.Column():
                 gr.Markdown("## Extensions & flags")
-                shared.gradio['save_settings'] = gr.Button(f'Save extensions settings to {shared.user_data_dir}/settings.yaml', interactive=not mu)
-                shared.gradio['reset_interface'] = gr.Button("Apply flags/extensions and restart", interactive=not mu)
+                with gr.Row():
+                    shared.gradio['save_settings'] = gr.Button(f'Save extensions settings to {shared.user_data_dir}/settings.yaml', elem_classes=['refresh-button', 'settings-button'], interactive=not mu)
+                    shared.gradio['reset_interface'] = gr.Button("Apply flags/extensions and restart", elem_classes=['refresh-button', 'settings-button'], interactive=not mu)
                 with gr.Row():
                     with gr.Column():
                         shared.gradio['extensions_menu'] = gr.CheckboxGroup(choices=utils.get_available_extensions(), value=shared.args.extensions, label="Available extensions", info='Note that some of these extensions may require manually installing Python requirements through the command: pip install -r extensions/extension_name/requirements.txt', elem_classes='checkboxgroup-table')

From bf6c8cd966c4963aa5ffa4e0358dcf8869109ae4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 16 May 2026 10:23:38 -0700
Subject: [PATCH 1667/1701] Tighten spacing between dropdowns and refresh
 buttons

---
 css/main.css | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/css/main.css b/css/main.css
index 5ee556cedb..b28030fc14 100644
--- a/css/main.css
+++ b/css/main.css
@@ -122,6 +122,10 @@ div.svelte-iyf88w {
     flex: none;
 }
 
+.gradio-container .stretch:has(> .refresh-button):has([role="listbox"]) {
+    gap: 8px;
+}
+
 .refresh-button-medium {
     max-width: 4.4em;
 }

From be7f3a20f2546271a2c4f562678161c4b365fd2d Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 16 May 2026 10:48:02 -0700
Subject: [PATCH 1668/1701] Electron: Add spellcheck toggle in the Session tab
 (closes #7550)

---
 desktop/main.js       | 12 +++++++++++-
 js/main.js            | 23 +++++++++++++++++++++++
 modules/shared.py     |  1 +
 modules/ui.py         |  4 ++--
 modules/ui_session.py |  2 ++
 5 files changed, 39 insertions(+), 3 deletions(-)

diff --git a/desktop/main.js b/desktop/main.js
index 6c1ad9a62a..8815a9274c 100644
--- a/desktop/main.js
+++ b/desktop/main.js
@@ -115,12 +115,22 @@ function createWindow(port) {
       preload: path.join(__dirname, "preload.js"),
       nodeIntegration: false,
       contextIsolation: true,
-      spellcheck: false,
+      spellcheck: true,
     },
   });
   if (state && state.maximized) mainWindow.maximize();
   mainWindow.webContents.on("context-menu", (_, params) => {
     const tmpl = [];
+    if (params.misspelledWord) {
+      for (const s of params.dictionarySuggestions) {
+        tmpl.push({ label: s, click: () => mainWindow.webContents.replaceMisspelling(s) });
+      }
+      if (params.dictionarySuggestions.length) tmpl.push({ type: "separator" });
+      tmpl.push(
+        { label: "Add to dictionary", click: () => mainWindow.webContents.session.addWordToSpellCheckerDictionary(params.misspelledWord) },
+        { type: "separator" },
+      );
+    }
     if (params.editFlags.canCut) tmpl.push({ role: "cut" });
     if (params.editFlags.canCopy) tmpl.push({ role: "copy" });
     if (params.editFlags.canPaste) tmpl.push({ role: "paste" });
diff --git a/js/main.js b/js/main.js
index ee4b1c3034..8e8a93b1d9 100644
--- a/js/main.js
+++ b/js/main.js
@@ -881,6 +881,29 @@ if (document.readyState === "loading") {
   setupPasteHandler();
 }
 
+//------------------------------------------------
+// Spellcheck toggle (Electron only; checkbox is hidden in the browser)
+//------------------------------------------------
+
+function setupSpellcheckToggle() {
+  if (!window.electronAPI) return;
+  const checkbox = document.querySelector("#spellcheck input[data-testid=\"checkbox\"]");
+  if (!checkbox) {
+    setTimeout(setupSpellcheckToggle, 500);
+    return;
+  }
+
+  const apply = () => { document.body.spellcheck = checkbox.checked; };
+  apply();
+  checkbox.addEventListener("change", apply);
+}
+
+if (document.readyState === "loading") {
+  document.addEventListener("DOMContentLoaded", setupSpellcheckToggle);
+} else {
+  setupSpellcheckToggle();
+}
+
 //------------------------------------------------
 // Tooltips
 //------------------------------------------------
diff --git a/modules/shared.py b/modules/shared.py
index 9058a99b15..8fc3a20605 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -292,6 +292,7 @@
     'show_two_notebook_columns': False,
     'paste_to_attachment': False,
     'include_past_attachments': True,
+    'spellcheck': False,
 
     # Generation parameters - Curve shape
     'temperature': neutral_samplers['temperature'],
diff --git a/modules/ui.py b/modules/ui.py
index 835bb0c655..5e96958f28 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -249,7 +249,7 @@ def list_interface_input_elements():
     ]
 
     if shared.is_electron:
-        elements += ['model_dir']
+        elements += ['model_dir', 'spellcheck']
 
     if not shared.args.portable:
         # Image generation elements
@@ -519,7 +519,7 @@ def setup_auto_save():
     ]
 
     if shared.is_electron:
-        change_elements += ['model_dir']
+        change_elements += ['model_dir', 'spellcheck']
 
     if not shared.args.portable:
         # Image generation tab (ui_image_generation.py)
diff --git a/modules/ui_session.py b/modules/ui_session.py
index 546c3c6df2..6ed1a3035f 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -70,6 +70,8 @@ def create_ui():
                 shared.gradio['show_two_notebook_columns'] = gr.Checkbox(label='Show two columns in the Notebook tab', value=shared.settings['show_two_notebook_columns'])
                 shared.gradio['paste_to_attachment'] = gr.Checkbox(label='Turn long pasted text into attachments in the Chat tab', value=shared.settings['paste_to_attachment'], elem_id='paste_to_attachment')
                 shared.gradio['include_past_attachments'] = gr.Checkbox(label='Include attachments/search results from previous messages in the chat prompt', value=shared.settings['include_past_attachments'])
+                if shared.is_electron:
+                    shared.gradio['spellcheck'] = gr.Checkbox(label='Enable spellcheck in text inputs', value=shared.settings['spellcheck'], elem_id='spellcheck')
 
                 if portable_version:
                     gr.Markdown("## Updates")

From 9a396e1749c667765a4f231d816b8440a0e71c47 Mon Sep 17 00:00:00 2001
From: MeemeeLab <57629983+MeemeeLab@users.noreply.github.com>
Date: Sat, 16 May 2026 10:27:22 +0900
Subject: [PATCH 1669/1701] Fix incorrect prompts generated with continue mode

When using the continue function, add_generation_prompt was set to False,
bypassing any template logic that depends on it. This caused missing tokens
or headers that some models inject at the start of an assistant turn.
Gemma 4 was particularly sensitive to this, producing garbled output due
to the missing thought channel header (<|channel>thought\n).

Fix by popping the last incomplete assistant message and re-rendering the
prompt with add_generation_prompt=True, then appending the partial content
afterward. This ensures the prompt is structurally identical to normal
generation regardless of the model's template.

GPT-OSS and Seed-OSS thinking block handling is preserved via the existing
fake-message approach, as those models manage thinking content differently.
---
 modules/chat.py | 41 +++++++++++++++++++++++++++++------------
 1 file changed, 29 insertions(+), 12 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 7833d58c9b..e93b6f157f 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -462,6 +462,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
                 msg_dict = {"role": "assistant", "content": final_content}
                 if '<|channel|>analysis<|message|>' in assistant_msg:
                     msg_dict["thinking"] = thinking_content
+                    msg_dict["raw_content"] = assistant_msg
 
                 messages.insert(insert_pos, msg_dict)
 
@@ -485,6 +486,7 @@ def generate_chat_prompt(user_input, state, **kwargs):
                 msg_dict = {"role": "assistant", "content": final_content.strip()}
                 if thinking_content:
                     msg_dict["reasoning_content"] = thinking_content
+                    msg_dict["raw_content"] = assistant_msg
 
                 messages.insert(insert_pos, msg_dict)
 
@@ -558,17 +560,26 @@ def make_prompt(messages):
         if _continue:
             messages = copy.deepcopy(messages)
         last_message = messages[-1].copy()
+
+        # Handle partial thinking blocks (GPT-OSS and Seed-OSS)
+        content = last_message.get("content", "")
+        partial_thought = last_message.get("thinking", "") or last_message.get("reasoning_content", "")
+        gpt_oss_or_seed_oss = not content and partial_thought and partial_thought.strip()
+
         if _continue:
-            if state['mode'] == 'chat-instruct':
+            if state['mode'] == 'chat-instruct' or not gpt_oss_or_seed_oss:
                 messages = messages[:-1]
             else:
                 messages[-1]["content"] = "fake assistant message replace me"
                 messages.append({"role": "assistant", "content": "this will get deleted"})
 
         if state['mode'] != 'chat-instruct':
-            add_generation_prompt = (not _continue and not impersonate)
+            if gpt_oss_or_seed_oss:
+                add_generation_prompt = (not _continue and not impersonate)
+            else:
+                add_generation_prompt = not impersonate
         else:
-            add_generation_prompt = False
+            add_generation_prompt = not gpt_oss_or_seed_oss
 
         prompt = renderer(
             messages=messages,
@@ -586,24 +597,20 @@ def make_prompt(messages):
                 outer_messages.append({"role": "system", "content": state['custom_system_message']})
 
             outer_messages.append({"role": "user", "content": command})
-            if _continue:
+            if _continue and gpt_oss_or_seed_oss:
                 outer_messages.append(last_message.copy())
                 outer_messages[-1]["content"] = "fake assistant message replace me"
                 outer_messages.append({"role": "assistant", "content": "this will get deleted"})
 
             prompt = instruct_renderer(
                 messages=outer_messages,
-                add_generation_prompt=not _continue
+                add_generation_prompt=not gpt_oss_or_seed_oss
             )
 
         if _continue:
-            prompt = prompt.split("fake assistant message replace me", 1)[0]
-
-            content = last_message.get("content", "")
-            partial_thought = last_message.get("thinking", "") or last_message.get("reasoning_content", "")
+            if gpt_oss_or_seed_oss:
+                prompt = prompt.split("fake assistant message replace me", 1)[0]
 
-            # Handle partial thinking blocks (GPT-OSS and Seed-OSS)
-            if not content and partial_thought and partial_thought.strip():
                 search_string = partial_thought.strip()
                 index = prompt.rfind(search_string)
                 if index != -1:
@@ -613,7 +620,16 @@ def make_prompt(messages):
                     prompt += partial_thought
             else:
                 # All other cases
-                prompt += content
+                append_content = last_message.get("raw_content", "") or content
+
+                for fmt in THINKING_FORMATS:
+                    fmt_start = fmt[0]
+                    if fmt_start is not None and prompt.rstrip("\n").endswith(fmt_start) and append_content.startswith(fmt_start):
+                        # in case of thinking block (e.g. ) exist in both prompt and content
+                        append_content = append_content[len(fmt_start):]
+                        break
+
+                prompt += append_content
 
         if impersonate:
             prompt = prompt.split("fake user message replace me", 1)[0]
@@ -1363,6 +1379,7 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
             regenerate = False
 
         _continue = True
+        state["enable_thinking"] = False
         send_dummy_message(text, state)
         send_dummy_reply(state['start_with'], state)
 

From d803e29d7f580783b651dfc60275f016e733e199 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 16 May 2026 13:40:24 -0700
Subject: [PATCH 1670/1701] Fix continue-mode regressions across template
 families

---
 modules/chat.py | 64 +++++++++++++++++++++++++++++++++----------------
 1 file changed, 43 insertions(+), 21 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index e93b6f157f..28783d171a 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -490,6 +490,18 @@ def generate_chat_prompt(user_input, state, **kwargs):
 
                 messages.insert(insert_pos, msg_dict)
 
+            # End-only  format (DeepSeek V4 Pro, Qwen3-next): the opener
+            # is emitted by the template's generation prompt, so model output starts
+            # with reasoning text and uses  as a separator.
+            elif '' in assistant_msg:
+                thinking_content, final_content = assistant_msg.split('', 1)
+                msg_dict = {"role": "assistant", "content": final_content.strip()}
+                thinking_content = thinking_content.strip()
+                if thinking_content:
+                    msg_dict["reasoning_content"] = thinking_content
+                    msg_dict["raw_content"] = assistant_msg
+                messages.insert(insert_pos, msg_dict)
+
             else:
                 # Default case (used by all other models)
                 messages.insert(insert_pos, {"role": "assistant", "content": assistant_msg})
@@ -561,25 +573,24 @@ def make_prompt(messages):
             messages = copy.deepcopy(messages)
         last_message = messages[-1].copy()
 
-        # Handle partial thinking blocks (GPT-OSS and Seed-OSS)
+        # Splice partial thoughts in-place to avoid a fresh thinking block from re-rendering.
         content = last_message.get("content", "")
         partial_thought = last_message.get("thinking", "") or last_message.get("reasoning_content", "")
-        gpt_oss_or_seed_oss = not content and partial_thought and partial_thought.strip()
+        thinking_only_partial = not content and bool(partial_thought.strip())
 
         if _continue:
-            if state['mode'] == 'chat-instruct' or not gpt_oss_or_seed_oss:
+            if state['mode'] == 'chat-instruct' or not thinking_only_partial:
                 messages = messages[:-1]
             else:
                 messages[-1]["content"] = "fake assistant message replace me"
                 messages.append({"role": "assistant", "content": "this will get deleted"})
 
-        if state['mode'] != 'chat-instruct':
-            if gpt_oss_or_seed_oss:
-                add_generation_prompt = (not _continue and not impersonate)
-            else:
-                add_generation_prompt = not impersonate
+        if state['mode'] == 'chat-instruct':
+            add_generation_prompt = _continue and not thinking_only_partial
+        elif thinking_only_partial:
+            add_generation_prompt = not _continue and not impersonate
         else:
-            add_generation_prompt = not gpt_oss_or_seed_oss
+            add_generation_prompt = not impersonate
 
         prompt = renderer(
             messages=messages,
@@ -597,18 +608,18 @@ def make_prompt(messages):
                 outer_messages.append({"role": "system", "content": state['custom_system_message']})
 
             outer_messages.append({"role": "user", "content": command})
-            if _continue and gpt_oss_or_seed_oss:
+            if _continue and thinking_only_partial:
                 outer_messages.append(last_message.copy())
                 outer_messages[-1]["content"] = "fake assistant message replace me"
                 outer_messages.append({"role": "assistant", "content": "this will get deleted"})
 
             prompt = instruct_renderer(
                 messages=outer_messages,
-                add_generation_prompt=not gpt_oss_or_seed_oss
+                add_generation_prompt=not thinking_only_partial
             )
 
         if _continue:
-            if gpt_oss_or_seed_oss:
+            if thinking_only_partial:
                 prompt = prompt.split("fake assistant message replace me", 1)[0]
 
                 search_string = partial_thought.strip()
@@ -619,15 +630,27 @@ def make_prompt(messages):
                     # Fallback if search fails: just append the thought
                     prompt += partial_thought
             else:
-                # All other cases
                 append_content = last_message.get("raw_content", "") or content
-
-                for fmt in THINKING_FORMATS:
-                    fmt_start = fmt[0]
-                    if fmt_start is not None and prompt.rstrip("\n").endswith(fmt_start) and append_content.startswith(fmt_start):
-                        # in case of thinking block (e.g. ) exist in both prompt and content
-                        append_content = append_content[len(fmt_start):]
-                        break
+                prompt_tail = prompt.rstrip("\n")
+
+                for fmt_start, fmt_end, fmt_content_tag in THINKING_FORMATS:
+                    if fmt_start is None or not prompt_tail.endswith(fmt_start):
+                        continue
+                    if append_content.startswith(fmt_start):
+                        # Avoid duplicating the opener the template already emitted
+                        append_content = append_content[len(fmt_start):].lstrip("\n")
+                    elif fmt_end and fmt_end in append_content:
+                        # Content closes the block itself (DeepSeek-style separator)
+                        pass
+                    else:
+                        # Close the opened thinking block so plain content doesn't land inside it
+                        prompt += "\n" + fmt_end + (fmt_content_tag or "") + "\n\n"
+                    break
+                else:
+                    # GPT-OSS: a bare "<|start|>assistant" gen prompt needs the final-channel
+                    # marker before plain content. raw_content already includes channel framing.
+                    if not last_message.get("raw_content") and prompt_tail.endswith("<|start|>assistant"):
+                        prompt += "<|channel|>final<|message|>"
 
                 prompt += append_content
 
@@ -1379,7 +1402,6 @@ def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
             regenerate = False
 
         _continue = True
-        state["enable_thinking"] = False
         send_dummy_message(text, state)
         send_dummy_reply(state['start_with'], state)
 

From 59c67f9da9715262057eaeb94cbdfd43c76dfd53 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 16 May 2026 17:26:54 -0700
Subject: [PATCH 1671/1701] Change dependabot to target the main branch

---
 .github/dependabot.yml | 1 -
 1 file changed, 1 deletion(-)

diff --git a/.github/dependabot.yml b/.github/dependabot.yml
index 9f7c99a63b..8f0d2814e2 100644
--- a/.github/dependabot.yml
+++ b/.github/dependabot.yml
@@ -9,6 +9,5 @@ updates:
     directories:
       - "/requirements/full/"
       - "/requirements/portable/"
-    target-branch: "dev"
     schedule:
       interval: "weekly"

From aa7d6bf04818c1b8dbed5685dc18c9f427c8f963 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 16 May 2026 17:34:58 -0700
Subject: [PATCH 1672/1701] Electron: Validate model_dir path before applying
 it

---
 modules/ui_session.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_session.py b/modules/ui_session.py
index 6ed1a3035f..cb59d5ce9b 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -162,8 +162,8 @@ def handle_default_to_notebook_change(show_two_columns, default_input, default_o
 
 
 def apply_model_dir(value):
-    shared.args.model_dir = value
     if Path(value).is_dir():
+        shared.args.model_dir = value
         shared.user_config = shared.load_user_config()
 
 

From 4c16b9458f198f1d60dfc06fd8626964dfbfef35 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 16 May 2026 17:35:30 -0700
Subject: [PATCH 1673/1701] UI: Fix token count not being set in non-streaming
 mode

---
 modules/text_generation.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index 112285cdf2..f1b6a7e53f 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -460,6 +460,7 @@ def generate_reply_HF(question, original_question, state, stopping_strings=None,
                     output = output.to(device)
 
             starting_from = 0 if shared.is_seq2seq else len(input_ids[0])
+            shared.model.last_completion_token_count = len(output) - starting_from
             yield get_reply_from_output_ids(output, state, starting_from=starting_from)
 
         # Stream the reply 1 token at a time.

From dcbb323cb6c82649afdc29841938d000f8575330 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 16 May 2026 17:39:10 -0700
Subject: [PATCH 1674/1701] UI: Improve web search security by rejecting
 non-HTTP links

---
 modules/html_generator.py | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index c0d2c5aa14..17272f9690 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -141,11 +141,15 @@ def _render_web_search_body(body):
     cards = []
     for r in results:
         title = html.escape(r['title'])
-        url = html.escape(r['url'])
+        url = r['url']
         snippet = html.escape(r.get('snippet', ''))
+        if url.lower().startswith(('http://', 'https://')):
+            link = f'{title}'
+        else:
+            link = f'{title}'
         cards.append(
             f'
    '
-            f'{title}'
+            f'{link}'
             f'
    {snippet}
    '
             f'
    '
         )

From 52212f44e05c8b4108f0c1a02d000f9423e72413 Mon Sep 17 00:00:00 2001
From: Allen930311 
Date: Sun, 17 May 2026 08:43:11 +0800
Subject: [PATCH 1675/1701] fix: prevent path traversal in
 load_template_by_name (CWE-22) (#7562)

---
 modules/models_settings.py | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index fa0a86a596..7cbd9ed9dd 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -412,6 +412,11 @@ def update_gpu_layers_and_vram(loader, model, gpu_layers, ctx_size, cache_type):
 
 def load_template_by_name(name):
     """Find and load a single instruction template by name. Returns '' if not found."""
+    # Prevent path traversal: strip all directory components, keep only the filename
+    name = Path(name).name
+    if not name:
+        return ''
+
     template_dir = shared.user_data_dir / 'instruction-templates'
     for ext in utils.TEMPLATE_EXTENSIONS:
         path = template_dir / f'{name}{ext}'
@@ -420,6 +425,13 @@ def load_template_by_name(name):
     else:
         return ''
 
+    # Defense-in-depth: confirm resolved path stays inside template_dir
+    try:
+        path.resolve().relative_to(template_dir.resolve())
+    except ValueError:
+        logger.error(f'Path traversal blocked for instruction template name: {name!r}')
+        return ''
+
     file_contents = path.read_text(encoding='utf-8')
     if path.suffix in utils.JINJA_EXTENSIONS:
         return file_contents

From 959b0b90f2abadb712768f9463edb76e082aa6ee Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 16 May 2026 18:10:06 -0700
Subject: [PATCH 1676/1701] Update feature_request.md

---
 .github/ISSUE_TEMPLATE/feature_request.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/ISSUE_TEMPLATE/feature_request.md b/.github/ISSUE_TEMPLATE/feature_request.md
index b94974f865..7a0534a77d 100644
--- a/.github/ISSUE_TEMPLATE/feature_request.md
+++ b/.github/ISSUE_TEMPLATE/feature_request.md
@@ -1,6 +1,6 @@
 ---
 name: Feature request
-about: Suggest an improvement or new feature for the web UI
+about: Suggest an improvement or new feature for TextGen
 title: ''
 labels: 'enhancement'
 assignees: ''

From dbe47840cfbf24d4a8521c1103589d967681882f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 16 May 2026 18:28:15 -0700
Subject: [PATCH 1677/1701] Full version installer: track release tags instead
 of main HEAD

---
 one_click.py | 19 ++++++++++++++-----
 1 file changed, 14 insertions(+), 5 deletions(-)

diff --git a/one_click.py b/one_click.py
index 0e835dec4a..4d4cc82406 100644
--- a/one_click.py
+++ b/one_click.py
@@ -330,6 +330,10 @@ def update_requirements(initial_installation=False, pull=True):
             environment=True,
             assert_success=True
         )
+        # Land fresh installs on the latest release tag rather than bleeding-edge main.
+        latest_tag = run_cmd('git tag -l "v*" --sort=-v:refname', capture_output=True, environment=True).stdout.decode().strip().split('\n', 1)[0]
+        if latest_tag:
+            run_cmd(f"git reset --hard {latest_tag}", assert_success=True, environment=True)
 
     # Check for outdated Python version and refuse to update
     if '.'.join(map(str, sys.version_info[:2])) != PYTHON_VERSION:
@@ -375,8 +379,6 @@ def update_requirements(initial_installation=False, pull=True):
             with open(requirements_file, 'r') as f:
                 before_pull_whl_lines = [line for line in f if '.whl' in line]
 
-        print_big_message('Updating the local copy of the repository with "git pull"')
-
         # Hash files before pulling
         files_to_check = [
             'start_linux.sh', 'start_macos.sh', 'start_windows.bat', 'start_wsl.bat',
@@ -385,8 +387,15 @@ def update_requirements(initial_installation=False, pull=True):
         ]
         before_hashes = {file: calculate_file_hash(file) for file in files_to_check}
 
-        # Perform the git pull
-        run_cmd("git pull --autostash", assert_success=True, environment=True)
+        # Update to the latest release tag, but only if HEAD is an ancestor of it.
+        # This keeps users on untagged commits ahead of the last tag in place until the next release.
+        run_cmd("git fetch --tags", assert_success=True, environment=True)
+        latest_tag = run_cmd('git tag -l "v*" --sort=-v:refname', capture_output=True, environment=True).stdout.decode().strip().split('\n', 1)[0]
+        if latest_tag and run_cmd(f"git merge-base --is-ancestor HEAD {latest_tag}", capture_output=True, environment=True).returncode == 0:
+            print_big_message(f'Updating to release tag {latest_tag}.')
+            run_cmd(f"git merge --autostash --ff-only {latest_tag}", assert_success=True, environment=True)
+        else:
+            print_big_message(f'HEAD is ahead of the latest release tag ({latest_tag}). Skipping git update.')
         current_commit = get_current_commit()
 
         # Check hashes after pulling
@@ -400,7 +409,7 @@ def update_requirements(initial_installation=False, pull=True):
         # Check for changes to installer files
         for file in files_to_check:
             if before_hashes[file] != after_hashes[file]:
-                print_big_message(f"File '{file}' was updated during 'git pull'. Please run the script again.")
+                print_big_message(f"File '{file}' changed during the update. Please run the script again.")
 
                 # Save state before exiting
                 state = load_state()

From 46159d509e29af1d6e5f6adce838f0fa4537f73f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 16 May 2026 20:47:27 -0700
Subject: [PATCH 1678/1701] UI: fade in new messages, fix scroll-up jump on
 send

---
 css/main.css          | 1 +
 js/global_scope_js.js | 6 ++++++
 2 files changed, 7 insertions(+)

diff --git a/css/main.css b/css/main.css
index b28030fc14..dc2994671c 100644
--- a/css/main.css
+++ b/css/main.css
@@ -493,6 +493,7 @@ audio {
 .user-message,
 .assistant-message {
     contain: layout paint;
+    animation: fadeIn 0.2s ease-out;
 }
 
 .chat .message .timestamp {
diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 3020ee6ec7..36eef8abb4 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -342,6 +342,8 @@ function applyMorphdomUpdate(data) {
 
   const messagesContainer = document.getElementsByClassName("messages")[0];
   const messagesCountBefore = messagesContainer ? messagesContainer.children.length : 0;
+  // Survive morphdom: server HTML has no inline style.
+  const savedPaddingBottom = messagesContainer ? messagesContainer.style.paddingBottom : "";
 
   // Track open/closed blocks and store scroll positions for open ones
   const openBlocks = new Set();
@@ -418,6 +420,10 @@ function applyMorphdomUpdate(data) {
     }
   );
 
+  if (messagesContainer && savedPaddingBottom) {
+    messagesContainer.style.paddingBottom = savedPaddingBottom;
+  }
+
   // Syntax highlighting and LaTeX
   if (window.doSyntaxHighlighting) {
     window.doSyntaxHighlighting();

From 28c01b8a2ca4d5ac8843cf190b67fe5621827fe3 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sat, 16 May 2026 22:43:49 -0700
Subject: [PATCH 1679/1701] Add an info message to the "Activate web search"
 checkbox

---
 modules/ui_chat.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index 2836644a1b..8b57671694 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -122,7 +122,7 @@ def sync_web_tools(selected):
 
                 gr.HTML("
    ")
 
-                shared.gradio['enable_web_search'] = gr.Checkbox(value=shared.settings.get('enable_web_search', False), label='Activate web search', elem_id='web-search')
+                shared.gradio['enable_web_search'] = gr.Checkbox(value=shared.settings.get('enable_web_search', False), label='Activate web search', info='Fetches web search results as text attachments.', elem_id='web-search')
                 with gr.Row(visible=shared.settings.get('enable_web_search', False)) as shared.gradio['web_search_row']:
                     shared.gradio['web_search_pages'] = gr.Number(value=shared.settings.get('web_search_pages', 3), precision=0, label='Number of pages to download', minimum=1, maximum=10)
 

From b76dd13f8e4fbe5c8a66e6b5efa0ab64c4c62e3f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 May 2026 10:25:55 -0700
Subject: [PATCH 1680/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 8 ++++----
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 8 ++++----
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 15 files changed, 30 insertions(+), 30 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 8beef136fe..03ba449de5 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -43,10 +43,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.34/exllamav3-0.0.34+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.34/exllamav3-0.0.34+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.7.13/flash_attn-2.8.3+cu128torch2.9-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 75cb0a23d2..7132ad9629 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -39,5 +39,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 941ece967d..c459c66978 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -39,4 +39,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index cc604d0a37..7b202b326a 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -39,4 +39,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 921ae8990a..66dc0e374e 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -39,7 +39,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index afbd6b6227..2eb3b4971b 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 30ee5d3ec3..fb7d856f54 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index f59a6e18de..6120deb28f 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -25,4 +25,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 53a450658b..e2740558e7 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -25,4 +25,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index cac9961afd..c832aed451 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index 409cb7c547..b888fd14d2 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index 8794882fd5..cb8120108a 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index b0fa9c9e46..cda9ece06b 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # ik_llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index d2513b01d3..52899ede6e 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/ik_llama_cpp_binaries-0.126.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 0c2e86314e..38080b15bd 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.126.0/llama_cpp_binaries-0.126.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 242a2d458fb68f022725dcfdb8c9771d7d83b309 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 May 2026 11:06:16 -0700
Subject: [PATCH 1681/1701] Pin xformers to cu128

---
 requirements/full/requirements.txt | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 03ba449de5..cc33be4d5d 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -31,7 +31,8 @@ transformers==5.6.*
 triton-windows==3.5.1.post24; platform_system == "Windows"
 tqdm
 wandb
-xformers==0.0.33
+https://download.pytorch.org/whl/cu128/xformers-0.0.33-cp39-abi3-manylinux_2_28_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://download.pytorch.org/whl/cu128/xformers-0.0.33-cp39-abi3-win_amd64.whl; platform_system == "Windows"
 
 # Gradio
 https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl

From c022565b1257a9c7d9a5128c8b4c03587559cf08 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 May 2026 16:26:54 -0700
Subject: [PATCH 1682/1701] UI: Add MTP support, reorganize the speculative
 decoding menu

---
 modules/llama_cpp_server.py | 23 ++++++-------
 modules/loaders.py          |  7 ++--
 modules/shared.py           |  4 +--
 modules/ui.py               |  4 +--
 modules/ui_model_menu.py    | 64 +++++++++++++++++++++++++++----------
 5 files changed, 68 insertions(+), 34 deletions(-)

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index 7f145b17f2..f5b6e83bf4 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -485,7 +485,11 @@ def _start_server(self):
 
             if path.exists():
                 cmd += ["--mmproj", str(path)]
-        if shared.args.model_draft not in [None, 'None']:
+        spec_type = shared.args.spec_type
+        model_draft_set = shared.args.model_draft not in [None, 'None']
+        uses_draft_model_flags = spec_type in ('none', 'draft-mtp') and model_draft_set
+        uses_draft_max = uses_draft_model_flags or spec_type == 'draft-mtp'
+        if uses_draft_model_flags:
             path = resolve_model_path(shared.args.model_draft)
 
             if path.is_file():
@@ -494,22 +498,19 @@ def _start_server(self):
                 model_file = sorted(path.glob('*.gguf'))[0]
 
             cmd += ["--model-draft", str(model_file)]
-            if shared.args.draft_max > 0:
-                cmd += ["--spec-draft-n-max", str(shared.args.draft_max)]
             if shared.args.gpu_layers_draft > 0:
                 cmd += ["--gpu-layers-draft", str(shared.args.gpu_layers_draft)]
             if shared.args.device_draft:
                 cmd += ["--device-draft", shared.args.device_draft]
-            if shared.args.ctx_size_draft > 0:
-                cmd += ["--ctx-size-draft", str(shared.args.ctx_size_draft)]
-        if shared.args.spec_type != 'none':
-            cmd += ["--spec-type", shared.args.spec_type]
-            if shared.args.spec_type == 'ngram-mod':
+        if uses_draft_max and shared.args.draft_max > 0:
+            cmd += ["--spec-draft-n-max", str(shared.args.draft_max)]
+        if spec_type != 'none':
+            cmd += ["--spec-type", spec_type]
+            if spec_type == 'ngram-mod':
                 cmd += ["--spec-ngram-mod-n-match", str(shared.args.spec_ngram_size_n)]
-                cmd += ["--spec-ngram-mod-n-max", str(shared.args.draft_max)]
                 cmd += ["--spec-ngram-mod-n-min", str(shared.args.spec_ngram_size_m)]
-            elif shared.args.spec_type in ('ngram-simple', 'ngram-map-k', 'ngram-map-k4v'):
-                prefix = f"--spec-{shared.args.spec_type}"
+            elif spec_type in ('ngram-simple', 'ngram-map-k', 'ngram-map-k4v'):
+                prefix = f"--spec-{spec_type}"
                 cmd += [f"{prefix}-size-n", str(shared.args.spec_ngram_size_n)]
                 cmd += [f"{prefix}-size-m", str(shared.args.spec_ngram_size_m)]
                 cmd += [f"{prefix}-min-hits", str(shared.args.spec_ngram_min_hits)]
diff --git a/modules/loaders.py b/modules/loaders.py
index 9809fabaea..9b432e604a 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -22,12 +22,12 @@
         'numa',
         'ik',
         'parallel',
+        'draft_model_header',
         'model_draft',
+        'model_draft_refresh',
         'draft_max',
         'gpu_layers_draft',
         'device_draft',
-        'ctx_size_draft',
-        'ngram_header',
         'spec_type',
         'spec_ngram_size_n',
         'spec_ngram_size_m',
@@ -64,7 +64,9 @@
         'ctx_size',
         'cache_type',
         'gpu_split',
+        'draft_model_header',
         'model_draft',
+        'model_draft_refresh',
         'draft_max',
         'speculative_decoding_accordion',
         'enable_tp',
@@ -337,7 +339,6 @@ def list_model_elements():
         'draft_max',
         'gpu_layers_draft',
         'device_draft',
-        'ctx_size_draft',
         'spec_type',
         'spec_ngram_size_n',
         'spec_ngram_size_m',
diff --git a/modules/shared.py b/modules/shared.py
index 8fc3a20605..53754b2c53 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -83,11 +83,11 @@
 # Speculative decoding
 group = parser.add_argument_group('Speculative decoding')
 group.add_argument('--model-draft', type=str, default=None, help='Path to the draft model for speculative decoding.')
-group.add_argument('--draft-max', type=int, default=4, help='Number of tokens to draft for speculative decoding.')
+group.add_argument('--draft-max', type=int, default=3, help='Number of tokens to draft for speculative decoding.')
 group.add_argument('--gpu-layers-draft', type=int, default=256, help='Number of layers to offload to the GPU for the draft model.')
 group.add_argument('--device-draft', type=str, default=None, help='Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1')
 group.add_argument('--ctx-size-draft', type=int, default=0, help='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
-group.add_argument('--spec-type', type=str, default='none', choices=['none', 'ngram-mod', 'ngram-simple', 'ngram-map-k', 'ngram-map-k4v', 'ngram-cache'], help='Draftless speculative decoding type. Recommended: ngram-mod.')
+group.add_argument('--spec-type', type=str, default='none', choices=['none', 'draft-mtp', 'ngram-mod', 'ngram-simple', 'ngram-map-k', 'ngram-map-k4v'], help='Speculative decoding type. Recommended: draft-mtp if the main model is an MTP build, otherwise ngram-mod.')
 group.add_argument('--spec-ngram-size-n', type=int, default=24, help='N-gram lookup size for ngram speculative decoding.')
 group.add_argument('--spec-ngram-size-m', type=int, default=48, help='Draft n-gram size for ngram speculative decoding.')
 group.add_argument('--spec-ngram-min-hits', type=int, default=1, help='Minimum n-gram hits for ngram-map speculative decoding.')
diff --git a/modules/ui.py b/modules/ui.py
index 5e96958f28..5fbb7f472f 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -551,7 +551,7 @@ def setup_auto_save():
                 store_current_state_and_debounce, gradio('interface_state', 'preset_menu', 'extensions_menu', 'show_controls', 'theme_state'), None, show_progress=False)
 
 
-def create_refresh_button(refresh_component, refresh_method, refreshed_args, elem_class, interactive=True):
+def create_refresh_button(refresh_component, refresh_method, refreshed_args, elem_class, interactive=True, visible=True):
     """
     Copied from https://github.com/AUTOMATIC1111/stable-diffusion-webui
     """
@@ -564,7 +564,7 @@ def refresh():
     classes = list(elem_class) if isinstance(elem_class, (list, tuple)) else [elem_class]
     if 'refresh-icon-btn' not in classes:
         classes.append('refresh-icon-btn')
-    refresh_button = gr.Button(refresh_symbol, elem_classes=classes, interactive=interactive)
+    refresh_button = gr.Button(refresh_symbol, elem_classes=classes, interactive=interactive, visible=visible)
     refresh_button.click(
         fn=lambda: {k: tuple(v) if type(k) is list else v for k, v in refresh().items()},
         inputs=[],
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index c4e2658b16..5354fceec7 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -22,6 +22,33 @@
 )
 from modules.utils import gradio
 
+NGRAM_SIZE_TYPES = ('ngram-mod', 'ngram-simple', 'ngram-map-k', 'ngram-map-k4v')
+NGRAM_MAP_TYPES = ('ngram-simple', 'ngram-map-k', 'ngram-map-k4v')
+SPEC_TYPE_OUTPUTS = ('spec_ngram_size_n', 'spec_ngram_size_m', 'spec_ngram_min_hits', 'draft_model_header', 'model_draft', 'model_draft_refresh', 'gpu_layers_draft', 'device_draft')
+
+
+def spec_type_visibility_updates(spec_type):
+    is_ngram_size = spec_type in NGRAM_SIZE_TYPES
+    is_ngram_map = spec_type in NGRAM_MAP_TYPES
+    is_draft = spec_type in ('none', 'draft-mtp')
+    visibility = {
+        'spec_ngram_size_n': is_ngram_size,
+        'spec_ngram_size_m': is_ngram_size,
+        'spec_ngram_min_hits': is_ngram_map,
+        'draft_model_header': is_draft,
+        'model_draft': is_draft,
+        'model_draft_refresh': is_draft,
+        'gpu_layers_draft': is_draft,
+        'device_draft': is_draft,
+    }
+    return [gr.update(visible=visibility[name]) for name in SPEC_TYPE_OUTPUTS]
+
+
+def loader_spec_overlay(loader, spec_type):
+    if loader == 'llama.cpp':
+        return spec_type_visibility_updates(spec_type)
+    return [gr.update()] * len(SPEC_TYPE_OUTPUTS)
+
 
 def create_ui():
     mu = shared.args.multi_user
@@ -73,22 +100,20 @@ def create_ui():
 
                             # Speculative decoding
                             with gr.Accordion("Speculative decoding", open=False) as shared.gradio['speculative_decoding_accordion']:
-                                shared.gradio['draft_max'] = gr.Number(label="draft-max", precision=0, step=1, value=shared.args.draft_max, info='Maximum number of tokens to draft for speculative decoding. Recommended: 4 for draft model, 64 for n-gram.')
+                                shared.gradio['spec_type'] = gr.Dropdown(label="spec-type", choices=['none', 'draft-mtp', 'ngram-mod', 'ngram-simple', 'ngram-map-k', 'ngram-map-k4v'], value=shared.args.spec_type, info='Recommended: draft-mtp if the main model is an MTP build, otherwise ngram-mod.')
+                                shared.gradio['draft_max'] = gr.Number(label="draft-max", precision=0, step=1, value=shared.args.draft_max, info='Maximum number of tokens to draft per step. Recommended: 3 for draft-mtp or a draft model, 64 for n-gram.')
 
-                                gr.Markdown('#### Draft model')
-                                with gr.Row():
-                                    shared.gradio['model_draft'] = gr.Dropdown(label="model-draft", choices=['None'] + utils.get_available_models(), value=lambda: shared.args.model_draft, elem_classes='slim-dropdown', info='Draft model. Must share the same vocabulary as the main model.', interactive=not mu)
-                                    ui.create_refresh_button(shared.gradio['model_draft'], lambda: None, lambda: {'choices': ['None'] + utils.get_available_models()}, 'refresh-button', interactive=not mu)
+                                shared.gradio['spec_ngram_size_n'] = gr.Number(label="spec-ngram-size-n", precision=0, step=1, value=shared.args.spec_ngram_size_n, info='N-gram lookup size for speculative decoding.', visible=shared.args.spec_type in NGRAM_SIZE_TYPES)
+                                shared.gradio['spec_ngram_size_m'] = gr.Number(label="spec-ngram-size-m", precision=0, step=1, value=shared.args.spec_ngram_size_m, info='Draft n-gram size for speculative decoding.', visible=shared.args.spec_type in NGRAM_SIZE_TYPES)
+                                shared.gradio['spec_ngram_min_hits'] = gr.Number(label="spec-ngram-min-hits", precision=0, step=1, value=shared.args.spec_ngram_min_hits, info='Minimum n-gram hits for ngram-map speculative decoding.', visible=shared.args.spec_type in NGRAM_MAP_TYPES)
 
-                                shared.gradio['gpu_layers_draft'] = gr.Slider(label="gpu-layers-draft", minimum=0, maximum=256, value=shared.args.gpu_layers_draft, info='Number of layers to offload to the GPU for the draft model.')
-                                shared.gradio['device_draft'] = gr.Textbox(label="device-draft", value=shared.args.device_draft, info='Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1')
-                                shared.gradio['ctx_size_draft'] = gr.Number(label="ctx-size-draft", precision=0, step=256, value=shared.args.ctx_size_draft, info='Size of the prompt context for the draft model. If 0, uses the same as the main model.')
+                                shared.gradio['draft_model_header'] = gr.Markdown('#### Draft model', visible=shared.args.spec_type in ('none', 'draft-mtp'))
+                                with gr.Row():
+                                    shared.gradio['model_draft'] = gr.Dropdown(label="model-draft", choices=['None'] + utils.get_available_models(), value=lambda: shared.args.model_draft, elem_classes='slim-dropdown', info='Draft model. Must share the same vocabulary as the main model. Optional for draft-mtp (only needed when using a separate mtp-*.gguf head file).', interactive=not mu, visible=shared.args.spec_type in ('none', 'draft-mtp'))
+                                    shared.gradio['model_draft_refresh'] = ui.create_refresh_button(shared.gradio['model_draft'], lambda: None, lambda: {'choices': ['None'] + utils.get_available_models()}, 'refresh-button', interactive=not mu, visible=shared.args.spec_type in ('none', 'draft-mtp'))
 
-                                shared.gradio['ngram_header'] = gr.Markdown('#### N-gram (draftless)')
-                                shared.gradio['spec_type'] = gr.Dropdown(label="spec-type", choices=['none', 'ngram-mod', 'ngram-simple', 'ngram-map-k', 'ngram-map-k4v', 'ngram-cache'], value=shared.args.spec_type, info='Draftless speculative decoding type. Recommended: ngram-mod.')
-                                shared.gradio['spec_ngram_size_n'] = gr.Number(label="spec-ngram-size-n", precision=0, step=1, value=shared.args.spec_ngram_size_n, info='N-gram lookup size for speculative decoding.', visible=shared.args.spec_type != 'none')
-                                shared.gradio['spec_ngram_size_m'] = gr.Number(label="spec-ngram-size-m", precision=0, step=1, value=shared.args.spec_ngram_size_m, info='Draft n-gram size for speculative decoding.', visible=shared.args.spec_type != 'none')
-                                shared.gradio['spec_ngram_min_hits'] = gr.Number(label="spec-ngram-min-hits", precision=0, step=1, value=shared.args.spec_ngram_min_hits, info='Minimum n-gram hits for ngram-map speculative decoding.', visible=shared.args.spec_type != 'none')
+                                shared.gradio['gpu_layers_draft'] = gr.Slider(label="gpu-layers-draft", minimum=0, maximum=256, value=shared.args.gpu_layers_draft, info='Number of layers to offload to the GPU for the draft model.', visible=shared.args.spec_type in ('none', 'draft-mtp'))
+                                shared.gradio['device_draft'] = gr.Textbox(label="device-draft", value=shared.args.device_draft, info='Comma-separated list of devices to use for offloading the draft model. Example: CUDA0,CUDA1', visible=shared.args.spec_type in ('none', 'draft-mtp'))
 
                     gr.Markdown("## Other options")
                     with gr.Accordion("See more options", open=False):
@@ -148,7 +173,14 @@ def create_event_handlers():
     if mu:
         return
 
-    shared.gradio['loader'].change(loaders.make_loader_params_visible, gradio('loader'), gradio(loaders.get_all_params()), show_progress=False)
+    shared.gradio['loader'].change(
+        loaders.make_loader_params_visible, gradio('loader'), gradio(loaders.get_all_params()), show_progress=False
+    ).then(
+        loader_spec_overlay,
+        gradio('loader', 'spec_type'),
+        gradio(*SPEC_TYPE_OUTPUTS),
+        show_progress=False
+    )
 
     # In this event handler, the interface state is read and updated
     # with the model defaults (if any), and then the model is loaded
@@ -188,9 +220,9 @@ def create_event_handlers():
         shared.gradio['lora_menu_apply'].click(load_lora_wrapper, gradio('lora_menu'), gradio('model_status'), show_progress=False)
 
     shared.gradio['spec_type'].change(
-        lambda x: [gr.update(visible=x != 'none')] * 3,
+        spec_type_visibility_updates,
         gradio('spec_type'),
-        gradio('spec_ngram_size_n', 'spec_ngram_size_m', 'spec_ngram_min_hits'),
+        gradio(*SPEC_TYPE_OUTPUTS),
         show_progress=False
     )
 

From 9023515d348cc5bdf2dfd16b4c0a01cd5c3c7a63 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 May 2026 17:14:54 -0700
Subject: [PATCH 1683/1701] Add DGX Spark support (linux aarch64 portable
 builds)

---
 .github/workflows/build-everything-tgw.yml        |  7 +++++++
 .github/workflows/build-portable-release-cuda.yml | 10 ++++++++++
 requirements/portable/requirements_cuda131.txt    |  1 +
 3 files changed, 18 insertions(+)

diff --git a/.github/workflows/build-everything-tgw.yml b/.github/workflows/build-everything-tgw.yml
index 7d7baa6d77..904269a7df 100644
--- a/.github/workflows/build-everything-tgw.yml
+++ b/.github/workflows/build-everything-tgw.yml
@@ -27,6 +27,13 @@ jobs:
       version: ${{ inputs.version }}
       config: 'os:ubuntu-22.04'
 
+  build_release_cuda_linux_arm:
+    name: CUDA Linux ARM
+    uses: ./.github/workflows/build-portable-release-cuda.yml
+    with:
+      version: ${{ inputs.version }}
+      config: 'os:ubuntu-24.04-arm;cuda:13.1'
+
   build_release_vulkan_windows:
     name: Vulkan Windows
     uses: ./.github/workflows/build-portable-release-vulkan.yml
diff --git a/.github/workflows/build-portable-release-cuda.yml b/.github/workflows/build-portable-release-cuda.yml
index 12e2b16f7f..e6594f4369 100644
--- a/.github/workflows/build-portable-release-cuda.yml
+++ b/.github/workflows/build-portable-release-cuda.yml
@@ -120,6 +120,12 @@ jobs:
                 PIP_PATH="portable_env/python.exe -m pip"
                 PACKAGES_PATH="portable_env/Lib/site-packages"
                 rm start_linux.sh start_macos.sh
+            elif [[ "$RUNNER_ARCH" == "ARM64" ]]; then
+                PLATFORM="linux-arm64"
+                PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-aarch64-unknown-linux-gnu-install_only_stripped.tar.gz"
+                PIP_PATH="portable_env/bin/python -m pip"
+                PACKAGES_PATH="portable_env/lib/python3.13/site-packages"
+                rm start_macos.sh start_windows.bat
             else
                 PLATFORM="linux"
                 PYTHON_URL="https://github.com/astral-sh/python-build-standalone/releases/download/20260303/cpython-3.13.12+20260303-x86_64-unknown-linux-gnu-install_only_stripped.tar.gz"
@@ -165,6 +171,10 @@ jobs:
                 fi
                 ELECTRON_BIN="electron/Electron.app/Contents/MacOS/Electron"
                 rm -f start_macos.sh
+            elif [[ "$RUNNER_ARCH" == "ARM64" ]]; then
+                ELECTRON_ZIP="electron-v${ELECTRON_VERSION}-linux-arm64.zip"
+                ELECTRON_BIN="electron/electron"
+                rm -f start_linux.sh
             else
                 ELECTRON_ZIP="electron-v${ELECTRON_VERSION}-linux-x64.zip"
                 ELECTRON_BIN="electron/electron"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index b888fd14d2..ca2caa4a0e 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -27,3 +27,4 @@ tiktoken
 # CUDA wheels
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
 https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu131-py3-none-linux_aarch64.whl; platform_system == "Linux" and platform_machine == "aarch64"

From 2b07e17c522abdcb802ef1b26d567cbea0183fbc Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 May 2026 18:37:25 -0700
Subject: [PATCH 1684/1701] Show live generation speed (tokens/s) while
 generating (closes #7563)

---
 modules/chat.py | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/modules/chat.py b/modules/chat.py
index 28783d171a..a252d79d71 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -798,6 +798,23 @@ def update_token_display_from_state(state):
     max_tokens = state.get('truncation_length') or 0
     percentage = (total / max_tokens) * 100 if max_tokens > 0 else 0
     new_value = f"{total:,} / {max_tokens:,} tokens ({percentage:.1f}%)"
+
+    if gen_n > 0:
+        # A drop in gen_n means a new generation (backends reset to 0 per turn).
+        last_seen = getattr(shared.model, '_tps_last_gen_n', None)
+        if last_seen is None or gen_n < last_seen:
+            shared.model._tps_start_time = time.time()
+            shared.model._tps_baseline = gen_n
+        shared.model._tps_last_gen_n = gen_n
+
+        elapsed = time.time() - shared.model._tps_start_time
+        baseline = shared.model._tps_baseline
+        if gen_n > baseline and elapsed > 0:
+            tps = (gen_n - baseline) / elapsed
+            new_value += f"
    {gen_n:,} generated ({tps:.1f} t/s)"
+        else:
+            new_value += f"
    {gen_n:,} generated"
+
     if new_value == getattr(shared.model, '_last_token_display', None):
         return gr.update()
     shared.model._last_token_display = new_value

From fa57f838b2caafcf400b9ad8248ed013e6085fb6 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 17 May 2026 21:54:20 -0700
Subject: [PATCH 1685/1701] Electron: Refresh models dropdown when a new models
 directory is selected

---
 modules/ui_session.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/modules/ui_session.py b/modules/ui_session.py
index cb59d5ce9b..fb9bede427 100644
--- a/modules/ui_session.py
+++ b/modules/ui_session.py
@@ -109,7 +109,7 @@ def create_ui():
                 None, gradio('model_dir'), gradio('model_dir'),
                 js='async (current) => { const p = await window.electronAPI.pickDirectory(); return p === null ? current : p; }')
 
-            shared.gradio['model_dir'].change(apply_model_dir, gradio('model_dir'), None, show_progress=False)
+            shared.gradio['model_dir'].change(apply_model_dir, gradio('model_dir'), gradio('model_menu', 'model_draft'), show_progress=False)
 
         shared.gradio['show_two_notebook_columns'].change(
             handle_default_to_notebook_change,
@@ -165,6 +165,10 @@ def apply_model_dir(value):
     if Path(value).is_dir():
         shared.args.model_dir = value
         shared.user_config = shared.load_user_config()
+        models = utils.get_available_models()
+        return gr.update(choices=models), gr.update(choices=['None'] + models)
+
+    return gr.update(), gr.update()
 
 
 def set_interface_arguments(extensions, bool_active):

From 96670299ef1b3b156d5b72e01835308e058ad1f7 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 18 May 2026 06:10:25 -0700
Subject: [PATCH 1686/1701] Add an icon! (Credits: LMLocalizer on Reddit)
 https://www.reddit.com/r/Oobabooga/comments/1tffhcu/comment/omcwa31/

---
 css/icon.png    | Bin 0 -> 20970 bytes
 desktop/main.js |   5 +++++
 server.py       |   1 +
 3 files changed, 6 insertions(+)
 create mode 100644 css/icon.png

diff --git a/css/icon.png b/css/icon.png
new file mode 100644
index 0000000000000000000000000000000000000000..65b51ab38532672e7b70d7de9d93758763f1c376
GIT binary patch
literal 20970
zcmZU4Wl$Vlv~A-OEVwhcTL=%?lRE8t>NVgmqxtE4Ea4FE_lmq-96%FD^Q(oh!wkXJQT
zb>&`e|L^g?&;7p_P*DibusB{r48$ljWyn;R`IA*;LsV(qL>S^FX`~&&+HNWmVWL9b
z{Bn-`k_J3dJ{rQwI`48+c>EPPtpsW9rMQE%IfA8WwT;=dK5%LIbC{_yyUFuLg>q?r
zS2rFp(=OC9&Q#T#F_Ed)v`pjIGvka7U{>ljkj|77iISGCQxY3cwpC6r{Gh1dB_ggt
z%V?(QWTF`bRyL(2c&lcXB4$&rVO=8YrhK_=v$Q
z0V%}ODu#s$v6YD$v9PJrS$Xi7LX8Y_MVZ{06ntf^P1O8c-zR%Z7{?p2K(Poo)H3aa
zyq(O2#EZdPI$X~3d>=WrlJruYV$`*@oH{rZ!!;8vMdQ8R5a1IdBip{Uma&r2QhsNq
z43^FE#Fcozl6LjiV
zH`1~51JkJ&(I}*`8&#I`tAgaC`K+2fLO+Mae5O>4H}}j6i_V0F#sxU(xq1g!8mbww
zc_%Z9fps+UIpk;+ViLWz1jIsc7^TSgjY3$2niJKS!)l|ML}9F=!)f5@WVX*C8sQ8=
z{gF&zETTCd$b#(29V{rr^zf_wsSO<%)%*lW8$!ujQn@OkdCNb({uv*}3da59Pf!+t
z74CuW<>;ng9->m8sMHlH-|VCJ6V1sxT%$V>QjsEG=}VEBLiE|=U12L$X(vH{>Fc_Q
zSEKV-^LlLlF*oPH}GsF`8TlkTs6T4Dt1F+$}~<0VrV63cON
zoAC2GsZ-nHq>(Y>zMy!;;iT8He*A)F?2Q_fiR2g_&hL$49z25w$9b9D7+%`yI)JcU
zcK0^LV>v530H`-9$x7+^E$^DfnrHr=7mCKC?yrvzVPlJ7bHXgcDGLePvUHb!okD(P
z5vK^jWWfCV#?c~7^KGm$J35K>`|omJV`R*zcs{xRTSlWAzrV-mx~}Y8_r#0Ac1T;h
zwdauQ3Ao8!Xt-G^pFA##zQ6gm0P6uVQ_m%W<^~7ludeo=1+{U>r@GC%yYRiGT2@}I
z!+JV1RtNy;9{2==Zt8gg;`hz`6mt^zb~-~wIaL!)F233ft5I?rFNZA@J_XaAgu=bx
zw{!a|5-8`L7?f1&B;7LR)(~O(ecSUp{{fq+o>1R|=zShk{Oc|_3SIC~-4HtX&6#Vk
z1Y}Cj{LCG^8Gd>^JMkAo@&ay+5a$8$XtHc^
zbqaL7FkOms)^De5Lf_g;QG)}w;oiVTQD`zw6g(L$x$glw!)FV9KOG`=zqAiK2@JeG
z#XV2_dIUdyT@NLMO@VY~rINu<`1L4ky{7ly=bHJ$qaSW_5f?5?_w?Nd1OlG%C7Q?(
zka(}D`?%Bg6aF*zE*Y@@Uo*8(&|>YT9cegb*w7Pw{-)_jqH83?ZFF+of8$Gd`{0#x1Is=}Yn_i!Olk81>J3y4q(%#zI3VwGG%Q}o#wwy%lbH;YmbM3kx
zq7}%v&y7ML83OQ8yd|D@v;N?`2b>pq!JL*G9W`3j3Xw5~@s39D#-mi}-o{Y(np>bg
z1}Frf0hx6}UN3(XZM>93sj`)haMkb+@b~|97Bu>(6lSs7g5}e6(?=qH4
zLB;>GlGJP#AP@O^bCY9>e!M{QeB@YY)&;Q`eB1m^LtRRoTx@hcDm||3ev3`7x
zh()a%dUZQVj_tyUgk|t!2Ei03peb6W9XYYrrr_{g%C502v8yz6O6SmK;plE?xOjn9
zCiS#~C0-t>0S%aGQQD@?|4_w<8PZPl(-yQJShJf&X;C`w$HlS^=9-yz6;FD0tcSjV
zuCwXH@AghtS65qF_KvApVsx1mHjLo0K@z^~EE-omaQJ@13=U+)(lvp6qf06}>>|P2
zW|8P!2KW1Z&C;iS-Wa0uO^i)4lOuqmAk~MLZ{pCb78!BhNBeShD`Er=p;hQn7Acnc
z-ZY)yDLxbxc0?7J>rSP7owl6j7f@4GwX!Efo>n2tW}qDh*^BsMv&QNe_!4xLi_}te
z_>zLQ3KK$U;}5Qe7FR|G*8d!Tm#c;9%wV)IuRD`$Q7Kd6{8r)em~NWFXvx4s}h!`{2|gl1#%3^Ef4xxY$}2>afqwpA&Plq
z+-=nJBqrEQxFD|dP%1Ya#!28x#*TT}r$y}af`eWZj(^H`}=HIs49f{w0*zvai%nz~*
zMlknN@LcZgh&(!e`;~zs4T<|>jjIV>-75)X6Q>N}aHC5KZfcsqvzbUQgr2mvo@o3!
zi(^>VxH%D>wK>LC$eNq;w?58io$#c`WErX|I-vZ7BZe74qiQU0d%*7SN!&kN;OOac
zs`D4}FoOLCm#&L_Rv(7&+-qykLCfJWr=+GSaH#M7Ag%1=G%wpOm{(p!A;!-sC}ax
z?9Z#mE571IL_|VcZH$J#Dh9k&41U3jW*R|S*4wMut+PGX_4~d{ucq*kyLT-c
z@C8-loV+Z9mLDxGrQbj`Q@+dlvF!~Fo7nS1WxKL_;@v2Flgd3le2$}Yzke9~Kt=r{
zX8|ed7JdD<2_I=AaR3I-e%-`5buoB8xf`7fwyPbvf49AKm$=qpYYR2gQu3Hfp!oV^
z8;N$uNtQ}8?y02coxjmU&z%Y^R-=7zQ<BPgMFMUP7BjvF7dFKUrbzwlxklHd|=_P=bj#GN2mpURmR1!7Cz4cOVpA
zq92P)f+a4Et|I#S(mo!K@pfH}a`JYl=Neu*Wp<)*~;aAf@2Hp(67XgO}i>55O{M!z^-|
zTG+t&K70WTOPI_<#?6xjKjP^S&E?bV*>CqK?(Hg0DtgQ@L;Q&Mgg>5h1;~%r)9k9N
zQ*XHiC+?Jw<1ZHh=ijM$n2s0l{Le|!p06bW&0yD=Waa_>MixpGjTTz}?fbLyuqWYZ
z98MI5&Mo!c{FaA_0sHXCOU7e+0267&{kHYw#6zqQCdn_FO8AV0XY`#X<1Y1@j5Jf$
zQNrB+I@7@HPg8o~hnnt-mk4xKCUUTka+^tply{rRi0&xQ~5
zPiX`B`HeYZ&8ZX}SD(uriw!L4LB$=zw#@{0?;NoRBg@NM{{8D;Lxn7y?tT(Rm-Ipe
z-K4+$&L8*<2zKNW>@kUaLDurwQZfR4YB`Ye&aQR%n5y#0C0rorz?S>Blw518aaomz
z0kJs<#E7kp^yS0=1%?Z(-eUbW>V6Ajf1YCQz0zC~iyk$bfwuqJd3c5Gi^dh0%Jr1F
zkj14GpA;Lbo;O0%yU`ddHbS8-eufiw!7oFpCY5<;MWDqab7tuy0qkaixtGarYkcwgpS3+9
z!4CG#hj5yXi>*FL6N>p{`Rd*wEvVM?ZQQdY@J+B}-j1uAz!({_g>Onpt!i5{5=D|k
z{BR>ylhYKbO}B0uI!3&g_2~tY(n;HWCgX&+=7(71zwmZMkI446e#>?
z&7Cqi_gkU{%Thf&%DA3Lq3toKE9G18>yq+Yq~b;U#MtPG
z``vAYiDbM5L7&HYEOw*s`S+TC(hcNG=$17m+?3Q4oN@ezTHbAc=PTWTiR*<3nD$#)
zee^joxw!v<`m$Ik$H!y8`e?73cl?(pxSRj$N(F8G&`BHCw^TU(_N=g<9o8}H9{S$i
zOZL{Lj+s?FY#>tLt}Cs{LEW6Y@>f_q8I*3T*=oC!5$~MU2(iL8zN*}OKtbm1MA!ir
z3}GOi)!;*?On8%%#Gei_B*^3n3QAFGytkzWz*m$E7M-?{VE5Gz;|hF*JvREq4g#gP
zw>I%-N{K7mci4+HL8^8)279`zdbwZC1dBC~+=Gg_k~4Zd+oF}fyvZl_F@JS}3@Gnl
z0A2{{d;2?f*Pyk5GdKW7EO+z+FLC2i2VrP=Z(xfBS
z&fE_K>3S<{r;4HXDBfcJQ`o|PG$<7ofzCZrej;>`8-#i*G-k~vEhF``rpId+C$7lW*edG3bUsI7BbXR`QnwMdr
zz8av6Ya0xbX4sXB^pET+E1^aSAOn61=NG9WaChlMgV|}S6%nQ;LMP4J)e4reJ~6C2
z(U&s&8VQ2*m+l@)Ijo;$phil%{uCuCr*b3D;?vkhR4U2@1?f?uFEyHNX8XmyX9r%u
zkAzbRAm2Y_pL|yPk8nc4^On(lN0IfqP^o0h5gt2_5!G#Z`X#UQ(xmO_KgT+E=yc`d
zwlIqYLqwsUR7uuu%F>GB$alr}2aak!knuY(QtlTm~zG!F4&idY{<@Ro_9
z+i(U70MHd)NpccQ#Q3?===iYk?%~@JeZwEno9@TJA4kKjB$=o&1Gm=&b<_`RSJn&}
znVo@^F?e!Xq=9Q9Ui*m!q)&2@k+mp$e|E4eFjyR8Cr_CGyRmxYie`!VbpAtJ
z95@BSlZHka=7yxUb=#(d&wP&o52norp|(4OADdlh)y_le!?0Wf?KqRYqLxi<@2K(L
zeiWHWY}-)J2b~COh97|{VRbKcZJWI)UAN{*MCG5~dV1b^dV&XE3O`Rh6TT^Gq~1iN
zP_^19?HL&EfNl1uyb#g&-M6$-XA>)KF10ZnUrIxdEUr2#=Z=V@Ndi#4lzY7Ev)Dl!
zBpE!y3V}F)jN0KT$~o?+nXUx%@^K&D-}kK|kxHYj;AbH0gI+*@o|>9k+sY5#ktsTw
zFQrjD>{P5)wAd7lv%Zjrmw}#A85O_)YV=w*?T(%ygnruPmth#_ckhRriGt(I4ciYP
zS-k$Qp2NdreDQ&hD-Ji7d!^&7^2sjP&A(pD>X@T#o%)7`o6{<#n(tu(7Al*-6sDn4cocecv&V{g}Zg;Sw!_22w%oR>g<*IWSEBqZ
zMlA1d38VJRzNvMqm(bmNArvH)C4CT!u{+60J$P%9SA(=FunO3PCs|b;l~@T4;DS-p
zgEu8!zLS0N*;R4+1K>5llqo#>ndLucR=3ysYBPsake)<8Yx4q^=Lrh;s
z&{$)PpvJG0vzU7e`Q7R_*c3JxCUoE0Gsv#Vfy^PUx50ztO&u-Q+@ogm(5$1xc4wtS
z(V1XC)6^jIbyuj^T6O*%Gn36}wZeY+IPa*{Ub|a2cjL1)=JP>0YF}NlB}VjXe4iA~
zV(R*-1<&^isBD4}tZdvoJUootY?To--`WNW4v_$yP~m{Ww$S96OTs$d_XCpZ!4>r4
zJNCGWlbGw_ZLFfWF2LjY9$A1<&n+Ev4`t?QNKz7tXwU$IDW31_hp<9-k|l#qQ7kcJ
zes+CJ?GVqxJqQ8Xkw17q{OcM~lyUW7sp#Lbk3a!XhI6gTq5<>KSeo};HmeWp*Qo#j
zvJD1>tJr7uJUZ3y^a|1bUsCTtR~j%i_!FZSq7+dSOxta-RFB3Aence@XF1w9D-pIc
z_9LO!xjKvi-t~Xrbyh(^L5cJAC@|?`o;eP-6h_(@H{k1x!k@kCdab;_S7^VCjw5c8
za>ubZ`Baodx4praN*OoqwoIBBbzI3N3}5*zsiG_7CeVt!X5Z*V&pUlM$C=OC^WIe*x1gF;|RT8
z0Zd(#alzjLaU$%`-`$i{<_I96&|N(Kz77BmXZp9@5!12M_e6yC10vZMP`t;p$5Yt;
z{W;dmlg2jLgox)~FKZbY>+KwF0n$QSOg35&6D^d=hd_>=pEv8+qP)kVEs}Y}{JoNf
zBLo7ut0rVZ%OLwx&9RRT3_5g$???u7m4{|R|EX7
z&mmE=aTxv^8uYYm7tSgwDx70uD)cajt?j!7j102xZW6plq5yf&WrtffwJ+scMD)kQ
zbO-6KU(9)9W=RO|jOVSZKLT5Z&>eI(&?(?BK8cx4D@#+RQW&8B`GbeEE(+IC~>4
zGndbeE0=a>428>U84TavCV1XKp42J5-{Uod?`4?>%VD~RDonCZqOnHO49}ev$ed0z
zC0uZibB53uqGrmcj%Q}NAeUlfe$6DtS00lShXjYr|0)1N)WZi?tRt%TzQZaX2_8gh
z_NLJH6f)`3*5B_-si7NzW9ibHkZ9I*$)<_9h+}z&B>#WP`XJIZCV32-cE?-M37`8G
z%%PVxu;9y|nqdOvO+9#$Vm?O^UTC!_5wi83=Fe<1h>ss(Oui6)T>wAWgljw0jbH&|
zE3S_g=*h>33s(I|W?Q5BAadOwr82%@cyh-K397h1pX0|qpg0lx*izJKi5ELb^x73W
z0PSE@@*{L8>4k{AYC+zvH#ZNmx|q)@6k`A)GJ92!ij3e2pT;kDxJ>&D6v;
zHEjIRlAqC)=F)A51kg<>fQ
zKw=^d4x36oXmmr=t)&@?G9x6!Dg4JAykr+-7`hiknn}@pNbFKtP@b*Ovf1(4SL&Ko
zGiN;t6n+Op&v%#6?nmL5G7ew1$@De$k_Wmu{v`URh(Ie|nF^h(ISVJgDOybv9=Yv@
zw;nO#Pv+pquDegOmwsm0DKZbf+-+VDzU2nN7Yxk%@z^*;c9jUG>|uz_X_YLn=21_P
za%o3jz3G2jsX!0MXrvbL>E}{V4clRe{1@%@Db|)wte*S5__77e8-J*1^{W2mnmd2&k8(lv6N}RD6y({sP|s&5;Wk&J^Muk_D!b>l#H)Rq&XwGA2vUnSxloM3k>$)0Wo&ATw9Hygn
zJyw4zLI;lQ8jMm?r|0Gt5>g}i(!Bych=n~r#ImC+yVs>kZqI#=JZJ5GK4T*QVT#bY
zG;HWgS!x;aN8o*nc=b(IKOy-n1LKH;osEs{Ley`2?OF{f#g&NWIW_Gc{TJGm5UELg
ze~FiRU^#%eQzY0hbR03Gy7gC;z?9xWA@PsStT3B5J`H$8qok;@dvoJ;NYlEJerV2@`nxSVvpwQc
z=ZFqGw<5Navz}3yvm2kqP*F=zcW|km#EdvzKwUq~_ZBCSgL4>V+Hg?{f&rwEYMRa2
z0@?mmTIc7<(~0dnck!VZv*IJStVb**t-DwY2GOAXu5Eu+emRYdyfD3kmwKt+!H@1P
zMjG4*rY-5NJMYj@)bULBdtb>gpYXhKG0iA;L)|*?{!Dh>cb&bu$8q*{-dsd!W%^u8t9T-9s44c`{m9-2~#
zPMCJQ&~O=#l}?0|Rr%*Z&W>xeuQhtzZqGMN=2==}7*-erizhBp1+iNtuFPKx82oJF
zfo%?rxnO9hEAoaABCt$quixZeEQ@Rj<>dkG2FI3tk`c8
zZ5gisXLrFElsdS}Im!V^gVDO%+S)kSyW4A_uKuId1oi{kPr0q9ov?_pE{$)83V&rL
z+|m;h_0n`h#+d>b%1@2gmVK_v=k-=#Pj~fGVUNUw-LL6A8n;kN@l4u+j9S9zk)AT{
zNq5o(DvvNjg!nX_1B$or)=_~~*Nr;Ex!}Ije=X__BIiY|r2S7^K2Y6@;!QmT5{MeQ
zaPUi0ewin*#?LGk?}-vVsAxL4Y#8yh&2Ytd{%6O=dNwU%j>`lh=V=#jR1D)Bfn7Tp
z|BD8{L9RsI!3KH_%%Kl9Vn75bwIpJq|qtu-LW8-
zAht4%z>K5kwchA2&%@m_Pzj|IP;7GT`M&XN>#$v&%8%h!B<9k6
z<&}ncC4fT3E?DQJZvVl**VL46)Y|;Us<;O~25r8vfigiAbySFyGy*i(HFe?>m-4i_
zJFjPdHIS*;2~D@eQ4Zs+to{`6ncri+ZGLs}$2;+N7deQCsx
zg?+e2wX(Wdv+MT}9~#sd3iYLjKfW|-Ap|WSnO&PAzv_;KpG>${tLlshy*1390rgV9*Ab00hU)a07~
z0{Sv%SG4}hTo}miBgMR3c)CfSdpccUj%F61bEtwqq9I6M9L_wV1EzFThd|G2@ULFxssb2N19JUvmV0RJAl2St_V=JzYh
z)sIl2rw6@%Y3b=0_&(mYs<N_H%mUS?g@*b#{525dy$_2|_rT
zpwzK+!+(k)NC2zk?uLdIiOXSCE1Gc+0q`?*wBIFRr{!NKT^ns*vsj8-c|Cf?kf|*@z+4lrE*n4n
z;R6>v0qG^;l|mBYn+1db2>BukG@axm<^Va8xxJyEUk@aDIAVm!Ey&@X#ke%DhxNJb
z&wRCctgEkNT%Y{nI(j0~_4UNbFqcmk?*lwUGIpJk_n$l^FAq#MRxyunw!6Q6iVBWT`KAJkILF{H)m*}!T?lc&J>eMT<+@@P)AU6G{E+BU
z?+^8ogJId68p+x3Qxg-hqThM^v$nSCGcngZ+OU^d@3wSEaS1o@(MSD)H|Pc8
z<)!l}exYhHBQ76p{)%}id8v9OkpC?M)zoYS=?Am~+={4*DM|DhoORg4yajO}N`aP$
zorh$Y^iXoetx3QC$KaeR_rEIxEPUGko@X`UZfaiji)Nk)@=*DnPO95#lbaw(AS0We
zF`nDw*EPznwvzy7sKvy4yYe)c#{&eS$UPkzbadbtFpY5TIlk<&gqz#3*(X&SUx)j%
zmRDmnAW`NJY@O&Mg$&73+9{}wB+y41_;<~{yeS_KfZVP6Koh1DV&j6kSXX<
zSx|5(PSz#PFu~-*$KJa^mSob3{D*8S{dBaE@$w9L0hNbTxSNAfm~9(t
zw35S!ov)HFf)pM>O@;is$AIy`>|KuceM@ndKV)P|qb^?TBOLdcWf(!=rC>!&e?XjD
zNEb!sA@%MtKmtmV1dckLQW_Z@&E4zpUryk*8&6SVHh}o>iEy!qMmKI?(}AgADf-Gh
zMN~+T_rtqA)%AU0$cv$Wn*dBa#eGSsD`^EX|KI|b06t*c+oCL97hgGu{aWiYW2VH5
zP`-KYnTI%kE*
zzh>FNw)#jO#%a9yfHJmu?S>Ij4YE2^J*i(owiqBw6;DI2#-M}wkUD}%9?niOeiH!o
zC$S}cJ-?9hp4)E?Mn1C^V0o0%VLplUWAch7ZRKq`QCjw)d{D2TY|c;>6CFIU;&4c{>Ocww$N0yCi^Y5;7fER0B=z4D*eD
z!BjvC)tjk^JW54`j&_5uj<$KG*?(NDvXVB?>}%GG0*(ATJx+-tFPmrG8y>m+y(HdT
z+P#7!0pm196t*K_!PE!0y3ZA46Q{Z351S8L`e%NQg#vD~T1Qq3g4WBFd;Uhc&@e8#
zJale9Xi=msd%0+)DRPKs*hkkyvNd8Lyl%5cfp7#)Iq2JGnZ5WVbM4h>
zlFd-e1am)FLjP*XBdetKG3z$i7pfEE;?Hm*7%8HsYgl($UU8cYKTT&p-!)Z^&dAWs
zN?c@>zSk%HipwPhiwBVlep2d7O800ru^Q*lQ4S)Z$3<#+jk<(O{~@#<6Zo1?
z!s(AS9^l!EL1d|yU#94+wow6&2!xOj%kf
z`OG@38xzVTc`D5R8|@SKgFp3#K+DO9k}(~RLvcPuTs5YoePzQ+*QwG=KI0g9{T-*k
ze&oK%xDsU7P)Ylp_tf1Bhiwid1LQchQrSfcU9iE?Rg+qD;f~(McYh-~g(NUdiWs#eFSjZ?nU!
zs!)4!WV*AxzZ^ziei2ZZWYif=F4W+BZ^k4^1AAklL~Y_&v6#upl(lC|pUrq01R`VU
zNxa&^73KSnUD(UxE$=ZcARUw_AIuqsigKoi*BoTpQptE(W<7M^X;uC<`0B7%$KpG(
z;KUcU^GDf?V0UODlcYa5FD3>P3y_968?WeP1h2F9->I0rc6^p?w>%5>Ld@QFaFSmt
ztb_YjkWqUJ=~Z#KP*E?1>=KjbuusmxN=fSKUYf8I!ty6ixT
z3{-Rg0`Dm<4##L)CKPxRfd1pp{@r;JinDKUfM{0dyEg(}BWqQ|lYo2n_##2jR?Ez#
zQ}GnS_6ed1I)fto`_I*5y@j??D
zrqb4jtZ8xp1jS6ZWdP=#n-fA92_*OS%K4w>m9A1;d*Lp78$$=uS0Du3uQp#$ylpTl
z$%M#3Mw)xdxa+x1zj^j3D4?^IB(Ji!)_U>tB&>JoOFq40ej@_qX@K0iq=qS)kGXRSy+GcG(n
z-)A0c5G!qyh4BUC5x6arFXZI
zFIYRotE@$*gz^=KDk-7))cO@;U2ec$Kx0(Dz45T6J*I@3$;E4cJLg7C-_B|k<=C7g
zu`6mMz-6Y)?l}5p@3L(?g{FgtQKWwe)x)W`4?YQgjF_pgP7Y=g+D{?Lz9%EGxMBsh
zXrb)rys!Alv`rQ-BX5e{v!C;QlVX3*!&-*yJ`)X}&$Pnnhs}nLILJu#KKEBS!F~1~
z6ZT6>F1W&Y3j0U-4ty0fd@WG!)nP4-yiVW1Ot!GcZM_c5zLm@7pIoyU{KP$;CMjR1
zM?c)so)1291`Y_~k~veJLP3csp^+Bu=m8ckir1s0A;w){KNEsWMP
zNVOdH7v_{U4ZdD;+M9lU+RA_(jVrh$=*LdSpfvK&M0M<3X)
zc8{@_S%cOw6u<~Y^6t`vB1vDnX=E-{5J<6Fykj#3U_s)eV=ETG#l-n$vz=w&e>vB{
zDcJ)c%lc@pdXFV8!nPbivmv{xwQ&ulbzh!OyUp{Tki}Uam>Sy!^3v139ZGGn{(R=5
z?6M4s9LKi9Nr!L2RT{(ilD#QtJTOf&JohLyiX{KkDr!bB8<;utQi*Zb8ch0VRu7<%
zOA4TH_C^&CyFt>rGl?A53107zzd@}IM8$Wx5Bq`vBuo~
zzO&a(%S`{R=vfhv3V#e8Sjco!_s-i$tC)R4h#!rP=#6j&Y)1<_qI&;*n_&^}YPrR4
zmd9@~oOTC4l`dda>`SA^-iB17$yu&=vnf=
zdQOG!!<*dO^DsJJI#+J%>;s70bIt0-D^C(!^R-VU75F*$cTc!|@H(C^))Sg6ON&!}
z=Y42)(S_(@UcDoelG8Q6YH4tO%GMNvy1$`v0R;Vi<{|lXZsQcb*YdGcKngEdm005a
z+eq#Ip3+evMqxzP-6BxF)#I-ZUpj;ZzB4N|AzBVIam}%Vm%d}MKg=qqUH;-8|Mhi1
zgG-HC*?(E|ufl)0eF;
z+UN-X^cFge3FYZ{js~%Xzz0&Pw)?D3(mV>*B+&7Xq>dF$EhGloZq2?1>~9{;-}3$9)M5Akkn5uO^liQ8
zLb^Ze@~OA-5erNA`d!WWc;wc-@dQ+DDT|ydfo=}VMWtZ@t>_D_Kkivy&M`-
zgZhELhYX|~X{$yP_CmZ3y1Ci@Ct#M=iAInhy9PU^jC^Wf^
zh`U}AXtPf=h<|K9!Dl<8o};f$*7fw)+&q|FUmjJj5G1lte54+Xm?mlEM>4SZevL*^
zM$le=@i_jcRYjo)OZgZd&o1u4Z<@HGFF*P}{r582aoERTwV~4pA=c!(54pK{jNgvf
znb%=z5nX!!!kCNJ&>AK>f8s#2G7i*DfA8I&+cR8o=X+o`wyM!*$V8gJ?>Svm`7{uD
zZaed;;fC2DH@~efj2m&-0ce=D6hVXmgf=WBxoWCZ%l7QLo!w3t(N@C
z$}AZUR_ocZ@4Br-FOy>=!x2;GW*?e;7R)blq&{n%$iSsT=+DO-UXr<1Eeye+1_fUN
zhjJtoPq@OM+GN@u%OP@jbOx+ZF=$dObc-9wLN2B;;xc`iE-FwWBj(g0gW`EtIPLaR%H<(cx-)b>(heK
z#iJ#EJxtLS|2<~$Twef7cOabRPc^lxk_fv<|01ajpA;#DrLaHYz|iP$>7S3z=v)VC
z86+v3i{g86g*;U%35{kMG%xf2FF3=?#O;W8)W2cy4fafN&(;k?9mm$+)8@z-U)3arKRS`r~C)mU_s
zRE%L1cLT0}wg^m+A_tx)TCE_&nPWQcXw(*+sxpFqS}a%H-^hATVPmMuyiqZxE
zZ0B{>$K|uwuYW;#5IyG#`((4#|6X-z4F1zrRrOP$?q__bU-SKYbw%OTi;W_cY#5%*
znOA?LR;&hhQvAALBYxM5pT&DVh#3?5(N-1<(;VUZ9|=nC+sDh$e2~Mra$K17b9io{
z$rqG6*0V}Fnv!J>TL!k>nyODF&sSt*z3!xPFi$(r`2E}T)4ilU4zD1htl}65Wh)hl
zO{axE0WhgX*j1Fncy47}5!w-}qZnzCAXCC1p9njhd7VDc_|t&rRoq!d=Q{K;a|@fN
z7M)tqkB40_Z9iu>Yg7F@Ej;A6j?NU}Zl}4@WmUyYJ2)vgv79$bcX1`&jt;uU#Q7t2
z=V66u*3uMb9_!{qafi7($9JE~nsh){?^i+K?(l}Ul^)@EYSiT(fl4|kI07EW&Hehz
zYEzAx@$cm}r@zxG2_b7Fi1-(q%?=(g+?msif6R{tZUphJFTBxRy_(YlL|iK;-1CRN
zJW=ZVvGH)2#XZ7PE}{AQM7Ix2OJi$}J9%f1O%Lk
zb+)fg_l_%n(};
zFbPsE3xza7;($=l=);UDMvudxjgp%UaM*faYHt3kH@Se0xhv!9%?L?%TH5nGmv`9;!*VsR~?^JmrwrDcA1jkm9;
zf%y4zdScQx7ArxTP4#3@m-$bQ1SdALQq|q3k|Z7t^B=iWz=uQB=q@zs`w5guyGI$I
z(GcsrCiLyoKVD>jMi07e7PT>!s9~-xgpQ-tQg7WRm!ePALpG8k%!MsdO!4jZycrEL
zfaJ@6`UWUATexy)(wIR?iA57m2(+2B*Slz|-OtgzNhlvsNEvi$g(dI^M8#8hh>Jy@
znq#g{rWRqxs?tl4%w}jL5y$kEAUq!L>0WIYrem5sNkIVaeQKaLXWr>&C{m^kh6DL-
zU=C$RQR6EBlX+ON#-KtnO3D|JQNFUMSSS$Jo%)`jL+*Fp(P8ebdoOgO0Lg)kYiqLA%AjNmR<|T41V8cR0Z|F+D0;(S5p(+YMJdL9&
z4C+@=r>!HcIVO`1iR)VyNsygg*%^vV(%n!!WuX#S5Ie=;Y|9MQ4!IlI<;g8$dHICV
z6e`B61$}Wb
zl>5(VXcehI>7Vmr^Blu8IKJ1yi?8^H3)B`Km01fHn2nVl8wFzvP*9%}oUBlX1dD6M0!;h`QmI
zvp5JhXg&G!v!@B6hq`td8gPb0UdU>cXt~T3tFioi)Zm5qtIy3=$IidT^Rm$vjG$7g
zchylQe_lQy4b2P+#q%rChW?#6?fdPtG2DcYiSF*@5X(*`pGq2GFXB0gF{td{<^b2%
zeQU43Xk2piptN~csWM1H=|>0D4=-~laay=)(dgLmAHTctM>G^#nKMM(+VTmxSpx6V
z7+Rx8(XWS!U}msS>fkps*uq$v(6=1nG@UIQ
z+XCX~4h}XKGl=_nlo2iaXtwV&v^uyk+(tWEe=IC-+}FrFmHuTKUe`sUav!Qbn;iAL
z13zoM4H*D4kg_1-u=Yn?kzb%QM{byL&DsuGiJhR_tC+-@OUK@lpV>L#
zJMi?g$BmY_iY!9&QHFmxiq#QM_g6Sl7e~}pI@O32&1nqlh;#GsQ^xD^
zZB`2x_a`P8{D$-M&4W?*n~z3q;E^0W-J4R;ZH!PZvnXfzIAxjtYHNdKk5BYVnwI=oQ#{+7}aSu7EExsJo<&~w8
zvQc`p=#qiW`(F4;$bJoz#Dk=z%a!o}rLeFWfu^{mN)2&uKHSjo4L*~m8MPN9utTaO
zG5Vj{n=qbrs7cVaJFzeN2bKdW9*(VM_wE%ZCQ{?g`xmYAXi1OwVc>zGWq;k++1(Sw
zHcR0{^$E2oMR;I5C2)pbNJ&X_#)9Zn`Tc!!JmN2d04lNW~ryBK%Hn+om}BPS|qga0y)GE
z!Y3c#)O(H-AG`TmsIu`a*QL{is)61nqWweo1?+-i9cKwLq>*ZFH&vfH*LT^VpM+2h
zDrCM(XZsa`EGui6{4vi{;QYTbwxzB<+>OMLJQUDLa?DoVf9`eJWuCwQP34*-nNkpk
z&VUmUN{D{duWW5Q(cwQlXC@$MMGFnqo;iL+LW^vK!EPzscw)on?MIdz=}bS%Y*x&_
zq4~;Y;pgc$v*!)x!YlUjvOTwd3@@l++>qAEG})_htZ-Yf3=D@ro78nJrCh?AvlF;Y
zvp{BZFiP_`>D=Q??4Qi+3?ri^IoD4f`h5BJ9A8fU1fp4o<`fgq
z=a01>759z*6H5`a?on^Gp@;hWpMPE#i62kcSp8Z1f4mt05Z}_k>2(8ufKshCh+jsb
z?hF89jSm?s_PU*#^xKV@RHBQm{$HrCSrFhi42;w(T
z8N~w>27oV#vEsy!PQOiw{E>Wr1a1;Zim>*1<{2J*baZ@TQI?z^E(r+eHURh!Ode1h
zx4U^-LV0Bj3dpSl>rgC1tNpi@hxs`dH_!%q1AWY;k99xzAd3-c1a9sq1hEImQ;>sI9&
zIq%I>#Ag+cl={?m?N_*dlAjVEe6JRJ`^ZQ?4^G$t0cF7LF<|Fs87N4wdZ*t-03>Nv
ze04+eq=5;jd?F)77Q%<$LJpZ}UO)ZRs_gX&l3$h!&&_M(;Ji+((}!y}cG-PuMa@@W9H;+pkD>
z0RSy{bMhhyqfW!1ny<^ZwuUFhem58WQRsR7!aN_U`Xt5G1#Ee_NkEc=JtX&Jl~FI&
zq#H3hrvgIud-;wO^Ea1y@%-rbUs--6K7xCFM1BsYzsRC@BzM>aJ|w=eVw$=>CPBaeynki;
zdW#@#_%RQzlal8K0KlZ?_4!I>4z*_s-YgXc{2sOu&D^4pBuD1?Q09|{8`B-dt-&OJ
zjUIryboV&9Co2;_%sQLomN*4cMEFdFGJOp#A3FUY$Y0Tv#2RAl=h2bZfU12pTmi&?
zX0>L%DIox7UVej$cMgE$O$KPOFwnJq0nskec-UG;MsmwbPP^8H@zrtQcw*X9+wS~%
zohEIa*b?Xafk2zGN|I~xnH8e)YZ0tTZg*4Yvos$WDaJwcDM@}l@ckXzcA`5sRlHAC
zZ5(>gR9h-2k;R>g-W~wHX_pEmOY(i{TfB7T#^q^TBy~`i_kW*^%r8x@iSp!749yJy
z$cS5dLTLY@Gh9VliB_L7J8#0SAXR;BT=pRJy*Y9uLP7*T*0$)_b=J8z8fZkm0Pj;(
z@tYh0fw(&MHULP*I4%&IPsWeiQM#r9%q=EQaJ}`ayw3->`GDfeWeEokS3NY)@4J7Q5?P7p(k|EudRyqC*!_+zCMSq&a1A{?Fty!KzjJL1q+tR8&aH_
z>?4y}bO7)G0p@f~<(R
zL0VHm8f6M}*L{HIz*&sCW!38{?@Ufj-6-bfRb2YjDF6sdaJGAD5wA&Kbh5{al$f9K
z^xS6Tm?*I>!^0j^Q||X#p)Mu>Y&|pzo4XSLl$-sO;K17HeSo}SDffA94nV26JWr{6(LR7
zlzmHdkvQ+OFpmHaN*DU%yJF?b8#;IZ(WTafFTUnhO+Vs%Q*B_Ug3Fx0l5fvD&^4lr
zz7~L;4IK0rGyv0I$rnHG#1}*Gu9&<1e~5->a|Jto0sv`$r>_4t
z4?=_qeb@fhAbwIhx
zOfu{b_5YtrBu1sB;wmky$ZNao`#J$YeX#Y9
zvnQebNc?z*9L6tRl4Y3raTO>Sqr5HAWY7EIdH}?}gy4BX9pc*&`
zEkBX@U{MNtf_!UiouT8a6Lfj$`!_MeF^Nooo(f)+#T^++=o9Vk2r34f?5XG>-xsR9
z3sBwp8!x){sFl=E4Od`F?l8!5=mF)mO5M@{gG`VYr9kj@6hsfCeEQUv=N|RdX#Gj?
zG3=@c!IVF#rGi_DNlEr2IPc}Ek7s@4`NB_Go(GA~iJzspxsC;fyw_VJ
z-+^azx8#=)9GHKzYlFK3Ky1#;^=nt8B#76<)ym63Iu<|a^7xVXVT8$Uk1t7J(&V+P
zoJ(-oL#K~UxDxv0)t)`^e7Avs9g14r^(BGM0eEzM_~h~89htS_`W(ZFZ!NUnSyCn&
zFKz;Bh7{YLM?R$c%J>Kk&p(x6534@9ew_G72X%J=P#f!Hzf%oZ9X=rouZa_j0!qTK
z1AwxA-JU3gex>6Bl0y;&e)Bg)%jXxJ@kjEcpYvWH!)wj&G5~_~qyKw1VbBAxI)g7o
z-Cxa@2chTF4uERu`x}MvhcH`|8KV{8%Zo)>!EnUs%YHWo-PWS$HI)DEnt%Zq-B=a5
z5rBC->r79tSKbn8C_hL3FGSf>0mf*PU`?a3$-E(<{UYz>cDJl~Gi^O|$qB3LUkN~W
zc4ck^z#)4-zbU0gZix#3#J;?e_%t?xE}xlcBm6(PH>zz#16fRZ|I|%8I3s4`H#Rcr
zk$H9>fNmQA0ia7rS5KJ8!w&g61YDd+@>dDcudeEIYLe|v9ceC75?tZ?xZg{@Pc}Uw
zDec!78UJM`em9;zx|AHZ2Dqo!@cWT@RYJKV2I4DKUz5F<@+D&o
zmF$1f@%k%m{;mRGq^t|b;HSJYOSde+uR+t@4kVd#m>*n~{N6wRCjk6)XDtHH-g=ZDOv#YTTCICx{0Q8Mx?#&5;5Q2SJ0HAv$6028F4R=o;EqCOR_sbn&I{)SfdwZsL
zBk_|pKQA{35)`NXJdiNzNn+Z|0O$i28Nw(S|G9G)ozUmkT@TViWsjM;coh!#z1J0<
zhb)Pb{FKdI^?eF}WXx-P{(HU=49}2Ez7$k9>eOA2boCd(@iG_f?uyK&BztQ=4{xO~
zzkdPXcZ^z}KP~vuP4ZS$Zuaw=fTD!>ze4;#6#bdmv_u
zC1@^~b^R%fpF$r1z*};V_qN+szE1&AX*?>(_(60)GY4TmQ@JL@+|IDZ~>SzB|r|Gt%94}i*N8WzPs7w1b2DkK1s=YWPM
zM}9i~#a1Lccpfd7^RI4CKEB@`1pt+XS9><`nk;X1G>f<4B=Jk|MA0jkTETA5#H7lPm;GqH#ei`?Fq@{@Ac>bn8qxpCb=m>
z_FYkdOL5=2dX*`DqTv^rq+FMcQ6ImbwYMd>O8}_7ts}te=Jx^cPOI1zdaI2|PQv&(
z>Ep5wp${aN``P;21AriCYYhN!H%_KgQ;5wNEt8xm@BM?@-DLd=$IW_vsKRf!&*eAn
zqpMx02K2CnfON`ec#5T7^n3r1vpzJuBbu4rtWdn}a|b|T_fe*nYVpEhTkDAC3Mzff
z5c&kamm@##^-<05CuTAYZr+aa11jgZx`T@HA-o$iyHs!j@MX)cUJyds~7#2B5zLK#^sjap*w|
zyV_*SM=%^KDsXQ$hdz7~N52OGfc|~@87m0zpK8s4Kg7F0aQuOw_j2Nw*+i#L+OJml
z`#q%j{oS5^efqZhcwne$>s4$^{CPHB^n06eVJ67^);=DBht&fX!r%Al+vg$Wul$-t
zEQ<(o|JnG12d(9xzs1JJ9;bi5Cq11!5A}Ii3I*{G2=RYEd;NO+piqss4+TXr>9+Ri
zA-DN&{-_`BBf|i8Wv;i1y!U#1{GbCGXoR
zK?9&a`sOmEm;DSWokK@Brw`06dI1pbkAS$aGL>hpQKj^v>@=^Z^6d
z0z@2u!w*$;U|Je=^7Qr(eK((eLVj|CYDu;3km)_(0H~w_D5=1W&jSvCq5ysRJf!Me
zt_KSM<3V=Y+DWUB_20%UFVFLgt7ZC3ldvFB2&0{2U0NvC7?yYg%
zA5Z`G^1}#v`$xK$PX_~#M$p?o)jgEIqXCFW@IQlqx?KeL-$URg^7nbT|NjO8YUBP7
z_xT@2;O6rGZy<0h`TrLXxV8L*zyt0H$^YF_e$EFTTnN<0oyGr-&qD(qR0Pz<10K4)
z{A>mfA_OGTZPnjE2_CfCh{WmI@k=G>`(Qx;@%wi5_#Zq7dLX`lpnq@wC7=!`+``(fCIuQEx>px(C>b`LaMyde=`uFSCCE@=c
XkmfjW_5R(a00000NkvXXu0mjfX8~?U

literal 0
HcmV?d00001

diff --git a/desktop/main.js b/desktop/main.js
index 8815a9274c..e36b33dc3d 100644
--- a/desktop/main.js
+++ b/desktop/main.js
@@ -42,6 +42,10 @@ function resolveUserDataDir() {
 }
 const stateFile = path.join(resolveUserDataDir(), "cache", "window-state.json");
 
+// css/ sits next to main.js in portable builds, one level up in dev.
+const portableIcon = path.join(baseDir, "css", "icon.png");
+const iconPath = fs.existsSync(portableIcon) ? portableIcon : path.join(baseDir, "..", "css", "icon.png");
+
 let serverProcess = null;
 let mainWindow = null;
 let portCheckInterval = null;
@@ -110,6 +114,7 @@ function createWindow(port) {
   mainWindow = new BrowserWindow({
     ...bounds,
     title: TITLE,
+    icon: iconPath,
     autoHideMenuBar: true,
     webPreferences: {
       preload: path.join(__dirname, "preload.js"),
diff --git a/server.py b/server.py
index 1591a3abf9..74cf01009e 100644
--- a/server.py
+++ b/server.py
@@ -242,6 +242,7 @@ def create_interface():
         ssl_certfile=shared.args.ssl_certfile,
         root_path=shared.args.subpath,
         allowed_paths=allowed_paths,
+        favicon_path='css/icon.png',
     )
 
 

From 518c6961600d4f880470382411a59f389c5b3741 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 18 May 2026 06:50:23 -0700
Subject: [PATCH 1687/1701] Electron: Store app data in
 user_data/cache/electron

---
 desktop/main.js | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/desktop/main.js b/desktop/main.js
index e36b33dc3d..20b55ffd85 100644
--- a/desktop/main.js
+++ b/desktop/main.js
@@ -40,7 +40,13 @@ function resolveUserDataDir() {
   const shared = path.join(baseDir, "..", "..", "user_data");
   return fs.existsSync(shared) ? shared : path.join(baseDir, "..", "user_data");
 }
-const stateFile = path.join(resolveUserDataDir(), "cache", "window-state.json");
+const userDataDir = resolveUserDataDir();
+const stateFile = path.join(userDataDir, "cache", "window-state.json");
+
+// Redirect Electron's per-app data (cookies, GPUCache, Local Storage, etc.)
+// out of ~/.config/TextGen and into user_data/cache/electron so everything
+// the app writes lives under the project's user_data tree.
+app.setPath("userData", path.join(userDataDir, "cache", "electron"));
 
 // css/ sits next to main.js in portable builds, one level up in dev.
 const portableIcon = path.join(baseDir, "css", "icon.png");

From 895f055d3ec434ea1bc9e935b2d41cdb958684be Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 18 May 2026 07:04:58 -0700
Subject: [PATCH 1688/1701] Electron: Disable DNS-over-HTTPS probes

---
 desktop/main.js | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/desktop/main.js b/desktop/main.js
index 20b55ffd85..40303fa32a 100644
--- a/desktop/main.js
+++ b/desktop/main.js
@@ -21,6 +21,9 @@ const userArgs = dashIdx >= 0 ? argv.slice(dashIdx + 1) : argv;
 
 app.setName(TITLE);
 
+// We only load http://127.0.0.1, so skip Chromium's DNS-over-HTTPS provider probes.
+app.commandLine.appendSwitch("disable-features", "DnsOverHttps,DnsOverHttpsUpgrade");
+
 // Skip Chromium's hardware video pipeline, which probes VAAPI at startup and
 // logs a noisy version-mismatch error on systems with older libva. We don't
 // render video content anyway. (--no-sandbox / --no-zygote are passed by the

From 3a63a0eeda80c1affcdd30f64fedab7eb1dc7c5c Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Mon, 18 May 2026 20:05:12 -0700
Subject: [PATCH 1689/1701] Auto-enable MTP for MTP GGUFs

---
 modules/models_settings.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/modules/models_settings.py b/modules/models_settings.py
index 7cbd9ed9dd..4370ae9092 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -69,6 +69,8 @@ def get_model_metadata(model):
             elif k.endswith('.block_count'):
                 model_settings['gpu_layers'] = -1
                 model_settings['max_gpu_layers'] = metadata[k] + 1
+            elif k.endswith('.nextn_predict_layers') and metadata[k] > 0:
+                model_settings['spec_type'] = 'draft-mtp'
 
         if 'tokenizer.chat_template' in metadata:
             template = metadata['tokenizer.chat_template']
@@ -234,6 +236,9 @@ def apply_model_settings_to_state(model, state):
         if k in state and k != 'gpu_layers':  # Skip gpu_layers, handle separately
             state[k] = model_settings[k]
 
+    if state.get('spec_type') == 'draft-mtp' and model_settings.get('spec_type') != 'draft-mtp':
+        state['spec_type'] = 'none'
+
     # Auto-detect a sibling mmproj when the user hasn't saved one for this model.
     # Bare filenames (from user_data/mmproj/) persist across model switches;
     # subfolder paths only persist while the new model lives in the same folder.

From f5e92c6a93d22925deb901f552c6ec176a3eb36d Mon Sep 17 00:00:00 2001
From: AJ-Gazin <160578633+AJ-Gazin@users.noreply.github.com>
Date: Wed, 20 May 2026 09:32:52 -0400
Subject: [PATCH 1690/1701] Set TORCH_VERSION to 2.9.0 to match xformers
 0.0.33's torch pin (#7581)

---
 one_click.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/one_click.py b/one_click.py
index 4d4cc82406..2ed63f37b2 100644
--- a/one_click.py
+++ b/one_click.py
@@ -11,7 +11,7 @@
 import sys
 
 # Define the required versions
-TORCH_VERSION = "2.9.1"
+TORCH_VERSION = "2.9.0"
 PYTHON_VERSION = "3.13"
 LIBSTDCXX_VERSION_LINUX = "12.1.0"
 

From 163a47002b43083405c5b36d6732115ea1f703b2 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 20 May 2026 06:38:24 -0700
Subject: [PATCH 1691/1701] Fix chat deletion failing when user_data/logs is a
 symlink (closes #7579)

---
 modules/utils.py | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/modules/utils.py b/modules/utils.py
index c93706cdb0..9efb253d7a 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -28,10 +28,11 @@ def sanitize_filename(name):
 
 def _is_path_allowed(abs_path_str):
     """Check if a path is under the configured user_data directory."""
-    abs_path = Path(abs_path_str).resolve()
-    user_data_resolved = shared.user_data_dir.resolve()
+    # normpath (not resolve) preserves symlinks so a symlinked user_data/logs works.
+    abs_path = Path(os.path.normpath(os.path.abspath(abs_path_str)))
+    user_data_base = Path(os.path.normpath(os.path.abspath(shared.user_data_dir)))
     try:
-        abs_path.relative_to(user_data_resolved)
+        abs_path.relative_to(user_data_base)
         return True
     except ValueError:
         return False

From 350eb1780d71a8c30dab115c1d342da8500ac879 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 20 May 2026 06:55:50 -0700
Subject: [PATCH 1692/1701] Fix thinking channel being lost across tool-call
 turns (closes #7578)

---
 modules/chat.py | 17 ++++++++++++-----
 1 file changed, 12 insertions(+), 5 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index a252d79d71..2cb77698ef 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -30,7 +30,7 @@
 )
 from modules.image_utils import open_image_safely
 from modules.logging_colors import logger
-from modules.reasoning import THINKING_FORMATS
+from modules.reasoning import THINKING_FORMATS, extract_reasoning
 from modules.text_generation import (
     generate_reply,
     get_encoded_length,
@@ -247,11 +247,14 @@ def _expand_tool_sequence(tool_seq):
     for item in tool_seq:
         if 'tool_calls' in item:
             deserialized = _deserialize_tool_call_arguments(item['tool_calls'])
-            messages.append({
+            msg = {
                 "role": "assistant",
                 "content": item.get('content', ''),
                 "tool_calls": deserialized
-            })
+            }
+            if item.get('reasoning_content'):
+                msg['reasoning_content'] = item['reasoning_content']
+            messages.append(msg)
             for tc in item['tool_calls']:
                 tc_id = tc.get('id', '')
                 if tc_id:
@@ -1586,8 +1589,12 @@ def _render():
             tc_headers.append(f'{fn_name}({args_summary})')
 
         seq_entry = {'tool_calls': serialized}
-        if content_prefix.strip():
-            clean = _strip_channel_tokens(content_prefix)
+        reasoning, body = extract_reasoning(content_prefix)
+        reasoning = (reasoning or '').strip()
+        if reasoning:
+            seq_entry['reasoning_content'] = reasoning
+        if body.strip():
+            clean = _strip_channel_tokens(body)
             if clean:
                 seq_entry['content'] = clean
         seq.append(seq_entry)

From 2fb21dcd4e26a3065ca89d440e1afacdd2253c37 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 20 May 2026 07:12:07 -0700
Subject: [PATCH 1693/1701] Fix API model load silently dropping hyphenated arg
 keys (closes #7577)

---
 modules/api/models.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/modules/api/models.py b/modules/api/models.py
index bfcd2c31fd..379dff016c 100644
--- a/modules/api/models.py
+++ b/modules/api/models.py
@@ -64,9 +64,10 @@ def _load_model(data):
     update_model_parameters(model_settings)
 
     if args:
-        for k in args:
+        for k, v in args.items():
+            k = k.replace('-', '_')
             if k in allowed_keys and hasattr(shared.args, k):
-                setattr(shared.args, k, args[k])
+                setattr(shared.args, k, v)
 
     shared.model, shared.tokenizer = load_model(model_name)
 

From 619a2b8ee4b7e48541a9be5c07e04cd31b91b44f Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 20 May 2026 12:22:57 -0300
Subject: [PATCH 1694/1701] Restrict CORS to localhost by default to prevent
 drive-by API access

---
 modules/api/script.py | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/modules/api/script.py b/modules/api/script.py
index ceeca2dc93..b5b4dff73e 100644
--- a/modules/api/script.py
+++ b/modules/api/script.py
@@ -95,13 +95,18 @@ def __init__(self, message: str, error_type: str = "invalid_request_error", stat
 check_admin_key = [Depends(verify_admin_key)]
 check_anthropic_key = [Depends(verify_anthropic_key)]
 
-# Configure CORS settings to allow all origins, methods, and headers
+# --listen/--public-api opts into network exposure; otherwise lock to localhost.
+if shared.args.listen or shared.args.public_api:
+    cors_kwargs = {"allow_origins": ["*"]}
+else:
+    cors_kwargs = {"allow_origin_regex": r"https?://(localhost|127\.0\.0\.1)(:\d+)?"}
+
 app.add_middleware(
     CORSMiddleware,
-    allow_origins=["*"],
-    allow_credentials=True,
+    allow_credentials=False,
     allow_methods=["*"],
-    allow_headers=["*"]
+    allow_headers=["*"],
+    **cors_kwargs,
 )
 
 

From dc92ee774aa63f4a3784b83c9bebc5fb014d81d4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 20 May 2026 12:23:00 -0300
Subject: [PATCH 1695/1701] Sanitize character name in load_character to
 prevent path traversal

---
 modules/chat.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/modules/chat.py b/modules/chat.py
index 2cb77698ef..969e111379 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -2122,9 +2122,10 @@ def load_character(character, name1, name2):
     greeting_field = 'greeting'
     picture = None
 
+    safe_name = sanitize_filename(character)
     filepath = None
     for extension in ["yml", "yaml", "json"]:
-        filepath = shared.user_data_dir / 'characters' / f'{character}.{extension}'
+        filepath = shared.user_data_dir / 'characters' / f'{safe_name}.{extension}'
         if filepath.exists():
             break
 
@@ -2140,7 +2141,7 @@ def load_character(character, name1, name2):
     for path in [cache_folder / "pfp_character.png", cache_folder / "pfp_character_thumb.png"]:
         path.unlink(missing_ok=True)
 
-    picture = generate_pfp_cache(character)
+    picture = generate_pfp_cache(safe_name)
 
     # Finding the bot's name
     for k in ['name', 'bot', '<|bot|>', 'char_name']:

From d1b29f5211a3cce4ade6422f8e7796f55c174698 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 20 May 2026 08:40:51 -0700
Subject: [PATCH 1696/1701] Bind llama-server to parent process lifetime on
 Windows (closes #7574)

---
 modules/llama_cpp_server.py   |   5 ++
 modules/windows_subprocess.py | 106 ++++++++++++++++++++++++++++++++++
 2 files changed, 111 insertions(+)
 create mode 100644 modules/windows_subprocess.py

diff --git a/modules/llama_cpp_server.py b/modules/llama_cpp_server.py
index f5b6e83bf4..211a287e2a 100644
--- a/modules/llama_cpp_server.py
+++ b/modules/llama_cpp_server.py
@@ -1,3 +1,4 @@
+import atexit
 import json
 import os
 import pprint
@@ -21,6 +22,7 @@
 )
 from modules.logging_colors import logger
 from modules.utils import resolve_model_path
+from modules.windows_subprocess import bind_to_parent_lifetime
 
 llamacpp_valid_cache_types = {"fp16", "q8_0", "q4_0"}
 
@@ -576,6 +578,8 @@ def _start_server(self):
             bufsize=0,
             env=env
         )
+        bind_to_parent_lifetime(self.process.pid)
+        atexit.register(self.stop)
 
         threading.Thread(target=filter_stderr_with_progress, args=(self.process.stderr,), daemon=True).start()
 
@@ -615,6 +619,7 @@ def __del__(self):
 
     def stop(self):
         """Stop the server process."""
+        atexit.unregister(self.stop)
         if self.process:
             self.process.terminate()
             try:
diff --git a/modules/windows_subprocess.py b/modules/windows_subprocess.py
new file mode 100644
index 0000000000..8fe2c3d52e
--- /dev/null
+++ b/modules/windows_subprocess.py
@@ -0,0 +1,106 @@
+"""
+Bind child process lifetimes to the parent.
+
+On Windows, closing the console window or killing python.exe via Task Manager
+doesn't deliver SIGTERM — signal handlers and atexit don't run, so subprocess
+children are orphaned. A Job Object with JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE
+makes the kernel reap any assigned children when the parent's handle is closed.
+"""
+
+import ctypes
+import os
+
+from modules.logging_colors import logger
+
+
+_job_handle = None
+
+
+class _BasicLimitInformation(ctypes.Structure):
+    _fields_ = [
+        ('PerProcessUserTimeLimit', ctypes.c_int64),
+        ('PerJobUserTimeLimit', ctypes.c_int64),
+        ('LimitFlags', ctypes.c_uint32),
+        ('MinimumWorkingSetSize', ctypes.c_size_t),
+        ('MaximumWorkingSetSize', ctypes.c_size_t),
+        ('ActiveProcessLimit', ctypes.c_uint32),
+        ('Affinity', ctypes.c_size_t),
+        ('PriorityClass', ctypes.c_uint32),
+        ('SchedulingClass', ctypes.c_uint32),
+    ]
+
+
+class _IoCounters(ctypes.Structure):
+    _fields_ = [
+        ('ReadOperationCount', ctypes.c_uint64),
+        ('WriteOperationCount', ctypes.c_uint64),
+        ('OtherOperationCount', ctypes.c_uint64),
+        ('ReadTransferCount', ctypes.c_uint64),
+        ('WriteTransferCount', ctypes.c_uint64),
+        ('OtherTransferCount', ctypes.c_uint64),
+    ]
+
+
+class _ExtendedLimitInformation(ctypes.Structure):
+    _fields_ = [
+        ('BasicLimitInformation', _BasicLimitInformation),
+        ('IoInfo', _IoCounters),
+        ('ProcessMemoryLimit', ctypes.c_size_t),
+        ('JobMemoryLimit', ctypes.c_size_t),
+        ('PeakProcessMemoryUsed', ctypes.c_size_t),
+        ('PeakJobMemoryUsed', ctypes.c_size_t),
+    ]
+
+
+def _ensure_job():
+    global _job_handle
+    if _job_handle is not None:
+        return _job_handle
+
+    try:
+        kernel32 = ctypes.WinDLL('kernel32', use_last_error=True)
+        kernel32.CreateJobObjectW.restype = ctypes.c_void_p
+        job = kernel32.CreateJobObjectW(None, None)
+        if not job:
+            return None
+
+        info = _ExtendedLimitInformation()
+        info.BasicLimitInformation.LimitFlags = 0x2000  # JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE
+        if not kernel32.SetInformationJobObject(
+            ctypes.c_void_p(job), 9,  # JobObjectExtendedLimitInformation
+            ctypes.byref(info), ctypes.sizeof(info)
+        ):
+            kernel32.CloseHandle(ctypes.c_void_p(job))
+            return None
+
+        _job_handle = job
+        return job
+    except Exception:
+        return None
+
+
+def bind_to_parent_lifetime(pid):
+    """Bind the given child process to this process's lifetime.
+
+    When this process exits for any reason, the OS will clean up the child.
+    No-op on non-Windows or if the Job Object cannot be set up.
+    """
+    if os.name != 'nt':
+        return
+
+    job = _ensure_job()
+    if not job:
+        return
+
+    try:
+        kernel32 = ctypes.WinDLL('kernel32', use_last_error=True)
+        kernel32.OpenProcess.restype = ctypes.c_void_p
+        handle = kernel32.OpenProcess(0x0001 | 0x0100, False, pid)  # TERMINATE | SET_QUOTA
+        if not handle:
+            return
+        try:
+            kernel32.AssignProcessToJobObject(ctypes.c_void_p(job), ctypes.c_void_p(handle))
+        finally:
+            kernel32.CloseHandle(ctypes.c_void_p(handle))
+    except Exception as e:
+        logger.debug(f"Could not bind child PID {pid} to parent lifetime: {e}")

From 3cdd7e045086dc9f578679c474747f6c22fb4230 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 20 May 2026 10:14:59 -0700
Subject: [PATCH 1697/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 8 ++++----
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 8 ++++----
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 6 +++---
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 15 files changed, 31 insertions(+), 31 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index cc33be4d5d..cd998d29a2 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -44,10 +44,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.34/exllamav3-0.0.34+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.34/exllamav3-0.0.34+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.7.13/flash_attn-2.8.3+cu128torch2.9-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 7132ad9629..21675b7a6d 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -39,5 +39,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index c459c66978..41e11a2d17 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -39,4 +39,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 7b202b326a..ddc57b03b5 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -39,4 +39,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 66dc0e374e..5fda75b73b 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -39,7 +39,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 2eb3b4971b..24dcb6f804 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index fb7d856f54..4b3f185de4 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 6120deb28f..54176d49a3 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -25,4 +25,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index e2740558e7..06ad668a8f 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -25,4 +25,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index c832aed451..c108b9f841 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index ca2caa4a0e..a85ef788f0 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -25,6 +25,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+cu131-py3-none-linux_aarch64.whl; platform_system == "Linux" and platform_machine == "aarch64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cu131-py3-none-linux_aarch64.whl; platform_system == "Linux" and platform_machine == "aarch64"
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index cb8120108a..5d3b69b928 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index cda9ece06b..c753d4635b 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # ik_llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index 52899ede6e..518e8abf85 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/ik_llama_cpp_binaries-0.135.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 38080b15bd..3ed7d31bd8 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.135.0/llama_cpp_binaries-0.135.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

From 79b46b80ec7ec98141c570dbc26f867fdfc39ead Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 20 May 2026 10:27:34 -0700
Subject: [PATCH 1698/1701] Update the custom gradio wheels

---
 requirements/full/requirements.txt                   | 4 ++--
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 4 ++--
 requirements/full/requirements_apple_silicon.txt     | 4 ++--
 requirements/full/requirements_cpu_only.txt          | 4 ++--
 requirements/full/requirements_nowheels.txt          | 4 ++--
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 4 ++--
 requirements/portable/requirements_apple_silicon.txt | 4 ++--
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 4 ++--
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_nowheels.txt      | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 17 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index cd998d29a2..51a8f2c6f6 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -35,8 +35,8 @@ https://download.pytorch.org/whl/cu128/xformers-0.0.33-cp39-abi3-manylinux_2_28_
 https://download.pytorch.org/whl/cu128/xformers-0.0.33-cp39-abi3-win_amd64.whl; platform_system == "Windows"
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index 21675b7a6d..fcd204aff3 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -30,8 +30,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index 41e11a2d17..c7cdeb8f12 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -30,8 +30,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index ddc57b03b5..0d0d9781e0 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -30,8 +30,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 5fda75b73b..5e62a5e836 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -30,8 +30,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/full/requirements_nowheels.txt b/requirements/full/requirements_nowheels.txt
index 9ea38fa6f3..8f5c3d1d97 100644
--- a/requirements/full/requirements_nowheels.txt
+++ b/requirements/full/requirements_nowheels.txt
@@ -30,8 +30,8 @@ trafilatura==2.0.0
 wandb
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 24dcb6f804..4250d9a97d 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -16,8 +16,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index 4b3f185de4..c39700bb55 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -16,8 +16,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 54176d49a3..25c4ef574c 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -16,8 +16,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 06ad668a8f..6a4e316dd2 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -16,8 +16,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index c108b9f841..f7e77ca9ce 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -16,8 +16,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index a85ef788f0..ab01ee3f14 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -16,8 +16,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index 5d3b69b928..5f421eed20 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -16,8 +16,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index c753d4635b..7b92b32250 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -16,8 +16,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index 518e8abf85..302cdfb708 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -16,8 +16,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_nowheels.txt b/requirements/portable/requirements_nowheels.txt
index 2531728a20..8880d207c9 100644
--- a/requirements/portable/requirements_nowheels.txt
+++ b/requirements/portable/requirements_nowheels.txt
@@ -16,8 +16,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index 3ed7d31bd8..ab07475763 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -16,8 +16,8 @@ trafilatura==2.0.0
 tqdm
 
 # Gradio
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio-4.37.2+custom.20-py3-none-any.whl
-https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.20/gradio_client-1.0.2+custom.20-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio-4.37.2+custom.21-py3-none-any.whl
+https://github.com/oobabooga/gradio/releases/download/4.37.2-custom.21/gradio_client-1.0.2+custom.21-py3-none-any.whl
 
 # API
 flask_cloudflared==0.0.15

From f9df9be98267a79b57617d287d6d0638823116d4 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Wed, 20 May 2026 11:37:45 -0700
Subject: [PATCH 1699/1701] Minor fixes

---
 modules/html_generator.py | 2 +-
 modules/utils.py          | 3 ++-
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/modules/html_generator.py b/modules/html_generator.py
index 17272f9690..170eaddc44 100644
--- a/modules/html_generator.py
+++ b/modules/html_generator.py
@@ -142,7 +142,7 @@ def _render_web_search_body(body):
     for r in results:
         title = html.escape(r['title'])
         url = r['url']
-        snippet = html.escape(r.get('snippet', ''))
+        snippet = html.escape(r.get('snippet') or '')
         if url.lower().startswith(('http://', 'https://')):
             link = f'{title}'
         else:
diff --git a/modules/utils.py b/modules/utils.py
index 9efb253d7a..696ff1b9d9 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -211,7 +211,8 @@ def is_mmproj_file(name):
 
 def find_sibling_mmproj(model_path):
     """Return an mmproj path relative to model_dir when exactly one mmproj file
-    sits next to the model in a subfolder of model_dir (any depth, not root).
+    sits in the same folder as the model, provided that folder is a subfolder
+    of model_dir (not model_dir itself).
     """
     try:
         model_path = Path(model_path)

From fbaa8a3d908a7a897e572825557acd3ac3e80906 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 31 May 2026 19:33:40 -0700
Subject: [PATCH 1700/1701] Fix stray scrollbars and leftover spacing in
 instruct mode chat

---
 css/main.css          |  2 +-
 js/global_scope_js.js | 42 +++++++++++++++++++++++++++++++++---------
 js/main.js            |  6 ++++++
 3 files changed, 40 insertions(+), 10 deletions(-)

diff --git a/css/main.css b/css/main.css
index dc2994671c..419bc8d0d7 100644
--- a/css/main.css
+++ b/css/main.css
@@ -507,7 +507,7 @@ audio {
 .chat > .messages {
     display: flex;
     flex-direction: column;
-    min-height: calc(100vh - 225px);
+    min-height: calc(100dvh - 225px);
 }
 
 .chat > .messages > :first-child {
diff --git a/js/global_scope_js.js b/js/global_scope_js.js
index 36eef8abb4..13bbc5b1ed 100644
--- a/js/global_scope_js.js
+++ b/js/global_scope_js.js
@@ -298,18 +298,36 @@ function autoScrollToBottom() {
 
 function updateInstructPadding() {
   const chatElement = document.getElementById("chat");
-  if (chatElement && chatElement.getAttribute("data-mode") === "instruct") {
-    const messagesContainer = chatElement.querySelector(".messages");
-    const lastChild = messagesContainer?.lastElementChild;
+  const messagesContainer = chatElement?.querySelector(".messages");
+  if (!messagesContainer) return;
+
+  // The top-anchored buffer only applies in instruct mode with something to
+  // anchor against; everything else clears it, so the space can't leak across
+  // a mode switch.
+  let bufferHeight = 0;
+  if (chatElement.getAttribute("data-mode") === "instruct") {
+    const lastChild = messagesContainer.lastElementChild;
     const prevSibling = lastChild?.previousElementSibling;
     if (lastChild && prevSibling && chatElement.offsetHeight > 0) {
-      let bufferHeight = Math.max(0, Math.max(window.innerHeight - 128 - 119, window.innerHeight - prevSibling.offsetHeight - 119) - lastChild.offsetHeight);
-      if (window.innerWidth <= 924) {
-        bufferHeight = Math.max(0, bufferHeight - 32);
+      // Target the scroll container's *content* height — clientHeight minus
+      // its own vertical padding — so the buffer fills the viewport exactly
+      // instead of overshooting by that padding into a permanent scrollbar.
+      // The viewport-128 term floors the buffer so a tall previous message
+      // can't shrink it away.
+      const chatParent = document.querySelector(".chat-parent");
+      let viewport = window.innerHeight;
+      if (chatParent) {
+        const cs = getComputedStyle(chatParent);
+        viewport = chatParent.clientHeight - parseFloat(cs.paddingTop) - parseFloat(cs.paddingBottom);
       }
-      messagesContainer.style.paddingBottom = `${bufferHeight}px`;
+      bufferHeight = Math.max(0, Math.max(viewport - 128, viewport - prevSibling.offsetHeight) - lastChild.offsetHeight);
     }
   }
+
+  const next = bufferHeight ? `${bufferHeight}px` : "";
+  if (messagesContainer.style.paddingBottom !== next) {
+    messagesContainer.style.paddingBottom = next;
+  }
 }
 
 let pendingMorphdomData = null;
@@ -420,8 +438,14 @@ function applyMorphdomUpdate(data) {
     }
   );
 
-  if (messagesContainer && savedPaddingBottom) {
-    messagesContainer.style.paddingBottom = savedPaddingBottom;
+  // Re-apply the saved buffer only if the messages list still exists after
+  // morphdom. When the chat empties, morphdom repurposes that node into the
+  // welcome greeting (stripping its inline style); restoring the stale padding
+  // onto it would leave a phantom scrollbar that updateInstructPadding can't
+  // clear, since it keys off ".messages".
+  const messagesAfter = document.getElementsByClassName("messages")[0];
+  if (messagesAfter && savedPaddingBottom) {
+    messagesAfter.style.paddingBottom = savedPaddingBottom;
   }
 
   // Syntax highlighting and LaTeX
diff --git a/js/main.js b/js/main.js
index 8e8a93b1d9..1b0bcf9c40 100644
--- a/js/main.js
+++ b/js/main.js
@@ -240,6 +240,9 @@ window.doSyntaxHighlighting = function() {
                 { left: "\\(", right: "\\)", display: false },
                 { left: "\\[", right: "\\]", display: true },
               ],
+              // Render invalid LaTeX as an inline error instead of throwing,
+              // which would abort the update before paddings/scroll are fixed.
+              throwOnError: false,
             });
           }
         });
@@ -1008,6 +1011,9 @@ document.fonts.addEventListener("loadingdone", (event) => {
   // composer so the message-actions row isn't glued to it.
   function syncMargin() {
     chatParent.style.marginBottom = (chatInputRow.offsetHeight + 15) + "px";
+    // The instruct buffer is sized off chatParent.clientHeight, which the
+    // margin change above just shrank/grew, so recompute it here too.
+    window.updateInstructPadding?.();
     if (!window.isScrolled) {
       chatParent.scrollTop = chatParent.scrollHeight - chatParent.clientHeight;
     }

From ed888c71f221df552750e1834b3654abab8ae345 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Sun, 31 May 2026 22:22:00 -0700
Subject: [PATCH 1701/1701] Update llama.cpp

---
 requirements/full/requirements.txt                   | 8 ++++----
 requirements/full/requirements_amd.txt               | 4 ++--
 requirements/full/requirements_apple_intel.txt       | 2 +-
 requirements/full/requirements_apple_silicon.txt     | 2 +-
 requirements/full/requirements_cpu_only.txt          | 8 ++++----
 requirements/portable/requirements.txt               | 4 ++--
 requirements/portable/requirements_amd.txt           | 4 ++--
 requirements/portable/requirements_apple_intel.txt   | 2 +-
 requirements/portable/requirements_apple_silicon.txt | 2 +-
 requirements/portable/requirements_cpu_only.txt      | 4 ++--
 requirements/portable/requirements_cuda131.txt       | 6 +++---
 requirements/portable/requirements_ik.txt            | 4 ++--
 requirements/portable/requirements_ik_cpu_only.txt   | 4 ++--
 requirements/portable/requirements_ik_cuda131.txt    | 4 ++--
 requirements/portable/requirements_vulkan.txt        | 4 ++--
 15 files changed, 31 insertions(+), 31 deletions(-)

diff --git a/requirements/full/requirements.txt b/requirements/full/requirements.txt
index 51a8f2c6f6..7888aea368 100644
--- a/requirements/full/requirements.txt
+++ b/requirements/full/requirements.txt
@@ -44,10 +44,10 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/ik_llama_cpp_binaries-0.138.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/ik_llama_cpp_binaries-0.138.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.34/exllamav3-0.0.34+cu128.torch2.9.0-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
 https://github.com/turboderp-org/exllamav3/releases/download/v0.0.34/exllamav3-0.0.34+cu128.torch2.9.0-cp313-cp313-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.13"
 https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.7.13/flash_attn-2.8.3+cu128torch2.9-cp313-cp313-win_amd64.whl; platform_system == "Windows" and python_version == "3.13"
diff --git a/requirements/full/requirements_amd.txt b/requirements/full/requirements_amd.txt
index fcd204aff3..e62d24f34e 100644
--- a/requirements/full/requirements_amd.txt
+++ b/requirements/full/requirements_amd.txt
@@ -39,5 +39,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/full/requirements_apple_intel.txt b/requirements/full/requirements_apple_intel.txt
index c7cdeb8f12..5b1756d560 100644
--- a/requirements/full/requirements_apple_intel.txt
+++ b/requirements/full/requirements_apple_intel.txt
@@ -39,4 +39,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_apple_silicon.txt b/requirements/full/requirements_apple_silicon.txt
index 0d0d9781e0..c5022b3f7c 100644
--- a/requirements/full/requirements_apple_silicon.txt
+++ b/requirements/full/requirements_apple_silicon.txt
@@ -39,4 +39,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/full/requirements_cpu_only.txt b/requirements/full/requirements_cpu_only.txt
index 5e62a5e836..f28739b6b7 100644
--- a/requirements/full/requirements_cpu_only.txt
+++ b/requirements/full/requirements_cpu_only.txt
@@ -39,7 +39,7 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/ik_llama_cpp_binaries-0.138.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/ik_llama_cpp_binaries-0.138.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements.txt b/requirements/portable/requirements.txt
index 4250d9a97d..53b973721d 100644
--- a/requirements/portable/requirements.txt
+++ b/requirements/portable/requirements.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_amd.txt b/requirements/portable/requirements_amd.txt
index c39700bb55..4518f719ab 100644
--- a/requirements/portable/requirements_amd.txt
+++ b/requirements/portable/requirements_amd.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+rocm7.2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+rocm7.2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_apple_intel.txt b/requirements/portable/requirements_apple_intel.txt
index 25c4ef574c..15243d99f3 100644
--- a/requirements/portable/requirements_apple_intel.txt
+++ b/requirements/portable/requirements_apple_intel.txt
@@ -25,4 +25,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0-py3-none-macosx_13_0_x86_64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_apple_silicon.txt b/requirements/portable/requirements_apple_silicon.txt
index 6a4e316dd2..903c9e7aaf 100644
--- a/requirements/portable/requirements_apple_silicon.txt
+++ b/requirements/portable/requirements_apple_silicon.txt
@@ -25,4 +25,4 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0-py3-none-macosx_13_0_arm64.whl; platform_system == "Darwin"
diff --git a/requirements/portable/requirements_cpu_only.txt b/requirements/portable/requirements_cpu_only.txt
index f7e77ca9ce..a22d5a543d 100644
--- a/requirements/portable/requirements_cpu_only.txt
+++ b/requirements/portable/requirements_cpu_only.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_cuda131.txt b/requirements/portable/requirements_cuda131.txt
index ab01ee3f14..361134da2e 100644
--- a/requirements/portable/requirements_cuda131.txt
+++ b/requirements/portable/requirements_cuda131.txt
@@ -25,6 +25,6 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+cu131-py3-none-linux_aarch64.whl; platform_system == "Linux" and platform_machine == "aarch64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+cu131-py3-none-linux_aarch64.whl; platform_system == "Linux" and platform_machine == "aarch64"
diff --git a/requirements/portable/requirements_ik.txt b/requirements/portable/requirements_ik.txt
index 5f421eed20..9f9ff58106 100644
--- a/requirements/portable/requirements_ik.txt
+++ b/requirements/portable/requirements_ik.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/ik_llama_cpp_binaries-0.138.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/ik_llama_cpp_binaries-0.138.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_ik_cpu_only.txt b/requirements/portable/requirements_ik_cpu_only.txt
index 7b92b32250..359512c04a 100644
--- a/requirements/portable/requirements_ik_cpu_only.txt
+++ b/requirements/portable/requirements_ik_cpu_only.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # ik_llama.cpp (CPU only)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/ik_llama_cpp_binaries-0.138.0+cpu-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/ik_llama_cpp_binaries-0.138.0+cpu-py3-none-win_amd64.whl; platform_system == "Windows"
diff --git a/requirements/portable/requirements_ik_cuda131.txt b/requirements/portable/requirements_ik_cuda131.txt
index 302cdfb708..5e4031bc7f 100644
--- a/requirements/portable/requirements_ik_cuda131.txt
+++ b/requirements/portable/requirements_ik_cuda131.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/ik_llama_cpp_binaries-0.136.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/ik_llama_cpp_binaries-0.138.0+cu131-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/ik_llama_cpp_binaries-0.138.0+cu131-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
diff --git a/requirements/portable/requirements_vulkan.txt b/requirements/portable/requirements_vulkan.txt
index ab07475763..78e5f6e9a8 100644
--- a/requirements/portable/requirements_vulkan.txt
+++ b/requirements/portable/requirements_vulkan.txt
@@ -25,5 +25,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.136.0/llama_cpp_binaries-0.136.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.138.0/llama_cpp_binaries-0.138.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"